{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.5706134094151213,
  "eval_steps": 500,
  "global_step": 4000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0001426533523537803,
      "grad_norm": 1.1729735136032104,
      "learning_rate": 0.0,
      "loss": 1.583,
      "step": 1
    },
    {
      "epoch": 0.0002853067047075606,
      "grad_norm": 1.002526879310608,
      "learning_rate": 1.584786053882726e-08,
      "loss": 1.5026,
      "step": 2
    },
    {
      "epoch": 0.00042796005706134097,
      "grad_norm": 0.9373162984848022,
      "learning_rate": 3.169572107765452e-08,
      "loss": 1.5626,
      "step": 3
    },
    {
      "epoch": 0.0005706134094151213,
      "grad_norm": 1.0274591445922852,
      "learning_rate": 4.754358161648178e-08,
      "loss": 1.5833,
      "step": 4
    },
    {
      "epoch": 0.0007132667617689016,
      "grad_norm": 0.790949821472168,
      "learning_rate": 6.339144215530904e-08,
      "loss": 1.6648,
      "step": 5
    },
    {
      "epoch": 0.0008559201141226819,
      "grad_norm": 1.735785961151123,
      "learning_rate": 7.92393026941363e-08,
      "loss": 1.5637,
      "step": 6
    },
    {
      "epoch": 0.0009985734664764623,
      "grad_norm": 0.6724671721458435,
      "learning_rate": 9.508716323296357e-08,
      "loss": 1.601,
      "step": 7
    },
    {
      "epoch": 0.0011412268188302425,
      "grad_norm": 1.1278249025344849,
      "learning_rate": 1.1093502377179082e-07,
      "loss": 1.6493,
      "step": 8
    },
    {
      "epoch": 0.0012838801711840227,
      "grad_norm": 0.800915539264679,
      "learning_rate": 1.2678288431061808e-07,
      "loss": 1.5741,
      "step": 9
    },
    {
      "epoch": 0.0014265335235378032,
      "grad_norm": 1.5563911199569702,
      "learning_rate": 1.4263074484944535e-07,
      "loss": 1.6916,
      "step": 10
    },
    {
      "epoch": 0.0015691868758915834,
      "grad_norm": 0.9552595019340515,
      "learning_rate": 1.584786053882726e-07,
      "loss": 1.5376,
      "step": 11
    },
    {
      "epoch": 0.0017118402282453639,
      "grad_norm": 0.8556690812110901,
      "learning_rate": 1.7432646592709986e-07,
      "loss": 1.5753,
      "step": 12
    },
    {
      "epoch": 0.001854493580599144,
      "grad_norm": 1.0253041982650757,
      "learning_rate": 1.9017432646592713e-07,
      "loss": 1.6666,
      "step": 13
    },
    {
      "epoch": 0.0019971469329529245,
      "grad_norm": 0.8423359394073486,
      "learning_rate": 2.0602218700475438e-07,
      "loss": 1.4988,
      "step": 14
    },
    {
      "epoch": 0.0021398002853067048,
      "grad_norm": 1.2031570672988892,
      "learning_rate": 2.2187004754358165e-07,
      "loss": 1.6674,
      "step": 15
    },
    {
      "epoch": 0.002282453637660485,
      "grad_norm": 0.9490939974784851,
      "learning_rate": 2.3771790808240892e-07,
      "loss": 1.4538,
      "step": 16
    },
    {
      "epoch": 0.0024251069900142652,
      "grad_norm": 0.9502609968185425,
      "learning_rate": 2.5356576862123616e-07,
      "loss": 1.5736,
      "step": 17
    },
    {
      "epoch": 0.0025677603423680455,
      "grad_norm": 0.8738661408424377,
      "learning_rate": 2.694136291600634e-07,
      "loss": 1.676,
      "step": 18
    },
    {
      "epoch": 0.002710413694721826,
      "grad_norm": 0.9073160290718079,
      "learning_rate": 2.852614896988907e-07,
      "loss": 1.5503,
      "step": 19
    },
    {
      "epoch": 0.0028530670470756064,
      "grad_norm": 0.6794393062591553,
      "learning_rate": 3.0110935023771794e-07,
      "loss": 1.6698,
      "step": 20
    },
    {
      "epoch": 0.0029957203994293866,
      "grad_norm": 0.9852645397186279,
      "learning_rate": 3.169572107765452e-07,
      "loss": 1.5954,
      "step": 21
    },
    {
      "epoch": 0.003138373751783167,
      "grad_norm": 1.0426123142242432,
      "learning_rate": 3.3280507131537243e-07,
      "loss": 1.633,
      "step": 22
    },
    {
      "epoch": 0.003281027104136947,
      "grad_norm": 0.5876559615135193,
      "learning_rate": 3.486529318541997e-07,
      "loss": 1.5085,
      "step": 23
    },
    {
      "epoch": 0.0034236804564907277,
      "grad_norm": 1.1310482025146484,
      "learning_rate": 3.6450079239302697e-07,
      "loss": 1.7416,
      "step": 24
    },
    {
      "epoch": 0.003566333808844508,
      "grad_norm": 0.7944432497024536,
      "learning_rate": 3.8034865293185427e-07,
      "loss": 1.5737,
      "step": 25
    },
    {
      "epoch": 0.003708987161198288,
      "grad_norm": 1.0518521070480347,
      "learning_rate": 3.9619651347068146e-07,
      "loss": 1.6472,
      "step": 26
    },
    {
      "epoch": 0.0038516405135520684,
      "grad_norm": 0.7828543782234192,
      "learning_rate": 4.1204437400950875e-07,
      "loss": 1.5648,
      "step": 27
    },
    {
      "epoch": 0.003994293865905849,
      "grad_norm": 1.0826324224472046,
      "learning_rate": 4.27892234548336e-07,
      "loss": 1.6081,
      "step": 28
    },
    {
      "epoch": 0.004136947218259629,
      "grad_norm": 1.753553032875061,
      "learning_rate": 4.437400950871633e-07,
      "loss": 1.6047,
      "step": 29
    },
    {
      "epoch": 0.0042796005706134095,
      "grad_norm": 0.9000786542892456,
      "learning_rate": 4.5958795562599054e-07,
      "loss": 1.6023,
      "step": 30
    },
    {
      "epoch": 0.00442225392296719,
      "grad_norm": 0.694839596748352,
      "learning_rate": 4.7543581616481783e-07,
      "loss": 1.6433,
      "step": 31
    },
    {
      "epoch": 0.00456490727532097,
      "grad_norm": 1.0806282758712769,
      "learning_rate": 4.91283676703645e-07,
      "loss": 1.6082,
      "step": 32
    },
    {
      "epoch": 0.00470756062767475,
      "grad_norm": 0.9509055018424988,
      "learning_rate": 5.071315372424723e-07,
      "loss": 1.566,
      "step": 33
    },
    {
      "epoch": 0.0048502139800285305,
      "grad_norm": 0.7939223647117615,
      "learning_rate": 5.229793977812995e-07,
      "loss": 1.4584,
      "step": 34
    },
    {
      "epoch": 0.004992867332382311,
      "grad_norm": 0.8554040789604187,
      "learning_rate": 5.388272583201268e-07,
      "loss": 1.5141,
      "step": 35
    },
    {
      "epoch": 0.005135520684736091,
      "grad_norm": 1.0103391408920288,
      "learning_rate": 5.546751188589541e-07,
      "loss": 1.6457,
      "step": 36
    },
    {
      "epoch": 0.005278174037089872,
      "grad_norm": 0.9377884864807129,
      "learning_rate": 5.705229793977814e-07,
      "loss": 1.535,
      "step": 37
    },
    {
      "epoch": 0.005420827389443652,
      "grad_norm": 0.9149576425552368,
      "learning_rate": 5.863708399366086e-07,
      "loss": 1.5309,
      "step": 38
    },
    {
      "epoch": 0.0055634807417974325,
      "grad_norm": 0.98841392993927,
      "learning_rate": 6.022187004754359e-07,
      "loss": 1.4208,
      "step": 39
    },
    {
      "epoch": 0.005706134094151213,
      "grad_norm": 1.0452731847763062,
      "learning_rate": 6.180665610142631e-07,
      "loss": 1.4497,
      "step": 40
    },
    {
      "epoch": 0.005848787446504993,
      "grad_norm": 1.0485177040100098,
      "learning_rate": 6.339144215530904e-07,
      "loss": 1.6303,
      "step": 41
    },
    {
      "epoch": 0.005991440798858773,
      "grad_norm": 0.7732070684432983,
      "learning_rate": 6.497622820919177e-07,
      "loss": 1.6519,
      "step": 42
    },
    {
      "epoch": 0.006134094151212553,
      "grad_norm": 1.2481422424316406,
      "learning_rate": 6.656101426307449e-07,
      "loss": 1.5816,
      "step": 43
    },
    {
      "epoch": 0.006276747503566334,
      "grad_norm": 1.115005612373352,
      "learning_rate": 6.814580031695723e-07,
      "loss": 1.5048,
      "step": 44
    },
    {
      "epoch": 0.006419400855920114,
      "grad_norm": 0.9483717083930969,
      "learning_rate": 6.973058637083995e-07,
      "loss": 1.6211,
      "step": 45
    },
    {
      "epoch": 0.006562054208273894,
      "grad_norm": 0.8824179172515869,
      "learning_rate": 7.131537242472266e-07,
      "loss": 1.5428,
      "step": 46
    },
    {
      "epoch": 0.006704707560627674,
      "grad_norm": 1.1314977407455444,
      "learning_rate": 7.290015847860539e-07,
      "loss": 1.6307,
      "step": 47
    },
    {
      "epoch": 0.0068473609129814554,
      "grad_norm": 1.2876425981521606,
      "learning_rate": 7.448494453248812e-07,
      "loss": 1.5141,
      "step": 48
    },
    {
      "epoch": 0.006990014265335236,
      "grad_norm": 0.9325816631317139,
      "learning_rate": 7.606973058637085e-07,
      "loss": 1.5708,
      "step": 49
    },
    {
      "epoch": 0.007132667617689016,
      "grad_norm": 0.9257802367210388,
      "learning_rate": 7.765451664025357e-07,
      "loss": 1.5799,
      "step": 50
    },
    {
      "epoch": 0.007275320970042796,
      "grad_norm": 1.2095764875411987,
      "learning_rate": 7.923930269413629e-07,
      "loss": 1.6415,
      "step": 51
    },
    {
      "epoch": 0.007417974322396576,
      "grad_norm": 0.7863972783088684,
      "learning_rate": 8.082408874801903e-07,
      "loss": 1.5792,
      "step": 52
    },
    {
      "epoch": 0.007560627674750357,
      "grad_norm": 0.914885401725769,
      "learning_rate": 8.240887480190175e-07,
      "loss": 1.5083,
      "step": 53
    },
    {
      "epoch": 0.007703281027104137,
      "grad_norm": 1.314627766609192,
      "learning_rate": 8.399366085578448e-07,
      "loss": 1.5692,
      "step": 54
    },
    {
      "epoch": 0.007845934379457917,
      "grad_norm": 0.7810591459274292,
      "learning_rate": 8.55784469096672e-07,
      "loss": 1.6234,
      "step": 55
    },
    {
      "epoch": 0.007988587731811698,
      "grad_norm": 1.2098760604858398,
      "learning_rate": 8.716323296354994e-07,
      "loss": 1.5646,
      "step": 56
    },
    {
      "epoch": 0.008131241084165478,
      "grad_norm": 1.0744049549102783,
      "learning_rate": 8.874801901743266e-07,
      "loss": 1.5639,
      "step": 57
    },
    {
      "epoch": 0.008273894436519259,
      "grad_norm": 1.0017086267471313,
      "learning_rate": 9.033280507131538e-07,
      "loss": 1.6126,
      "step": 58
    },
    {
      "epoch": 0.008416547788873038,
      "grad_norm": 0.9702041745185852,
      "learning_rate": 9.191759112519811e-07,
      "loss": 1.5723,
      "step": 59
    },
    {
      "epoch": 0.008559201141226819,
      "grad_norm": 0.9367918968200684,
      "learning_rate": 9.350237717908083e-07,
      "loss": 1.5393,
      "step": 60
    },
    {
      "epoch": 0.008701854493580598,
      "grad_norm": 0.6486847996711731,
      "learning_rate": 9.508716323296357e-07,
      "loss": 1.4801,
      "step": 61
    },
    {
      "epoch": 0.00884450784593438,
      "grad_norm": 0.9823785424232483,
      "learning_rate": 9.667194928684629e-07,
      "loss": 1.3937,
      "step": 62
    },
    {
      "epoch": 0.008987161198288159,
      "grad_norm": 0.9961357712745667,
      "learning_rate": 9.8256735340729e-07,
      "loss": 1.576,
      "step": 63
    },
    {
      "epoch": 0.00912981455064194,
      "grad_norm": 0.9098517894744873,
      "learning_rate": 9.984152139461172e-07,
      "loss": 1.6427,
      "step": 64
    },
    {
      "epoch": 0.009272467902995721,
      "grad_norm": 0.8556783199310303,
      "learning_rate": 1.0142630744849446e-06,
      "loss": 1.5688,
      "step": 65
    },
    {
      "epoch": 0.0094151212553495,
      "grad_norm": 0.8681414127349854,
      "learning_rate": 1.0301109350237718e-06,
      "loss": 1.5462,
      "step": 66
    },
    {
      "epoch": 0.009557774607703282,
      "grad_norm": 1.2367175817489624,
      "learning_rate": 1.045958795562599e-06,
      "loss": 1.4963,
      "step": 67
    },
    {
      "epoch": 0.009700427960057061,
      "grad_norm": 1.0021933317184448,
      "learning_rate": 1.0618066561014264e-06,
      "loss": 1.495,
      "step": 68
    },
    {
      "epoch": 0.009843081312410842,
      "grad_norm": 1.1043912172317505,
      "learning_rate": 1.0776545166402536e-06,
      "loss": 1.6672,
      "step": 69
    },
    {
      "epoch": 0.009985734664764621,
      "grad_norm": 1.1770638227462769,
      "learning_rate": 1.093502377179081e-06,
      "loss": 1.6216,
      "step": 70
    },
    {
      "epoch": 0.010128388017118403,
      "grad_norm": 1.0524466037750244,
      "learning_rate": 1.1093502377179082e-06,
      "loss": 1.5777,
      "step": 71
    },
    {
      "epoch": 0.010271041369472182,
      "grad_norm": 0.8070217370986938,
      "learning_rate": 1.1251980982567354e-06,
      "loss": 1.4975,
      "step": 72
    },
    {
      "epoch": 0.010413694721825963,
      "grad_norm": 0.9641265869140625,
      "learning_rate": 1.1410459587955628e-06,
      "loss": 1.5548,
      "step": 73
    },
    {
      "epoch": 0.010556348074179744,
      "grad_norm": 0.6958743333816528,
      "learning_rate": 1.15689381933439e-06,
      "loss": 1.7126,
      "step": 74
    },
    {
      "epoch": 0.010699001426533523,
      "grad_norm": 0.918315589427948,
      "learning_rate": 1.1727416798732172e-06,
      "loss": 1.59,
      "step": 75
    },
    {
      "epoch": 0.010841654778887305,
      "grad_norm": 1.1720390319824219,
      "learning_rate": 1.1885895404120444e-06,
      "loss": 1.5478,
      "step": 76
    },
    {
      "epoch": 0.010984308131241084,
      "grad_norm": 1.5540131330490112,
      "learning_rate": 1.2044374009508718e-06,
      "loss": 1.6854,
      "step": 77
    },
    {
      "epoch": 0.011126961483594865,
      "grad_norm": 0.7053331732749939,
      "learning_rate": 1.220285261489699e-06,
      "loss": 1.6028,
      "step": 78
    },
    {
      "epoch": 0.011269614835948644,
      "grad_norm": 0.7510805130004883,
      "learning_rate": 1.2361331220285262e-06,
      "loss": 1.5807,
      "step": 79
    },
    {
      "epoch": 0.011412268188302425,
      "grad_norm": 0.6899828910827637,
      "learning_rate": 1.2519809825673536e-06,
      "loss": 1.4624,
      "step": 80
    },
    {
      "epoch": 0.011554921540656205,
      "grad_norm": 0.6752126216888428,
      "learning_rate": 1.2678288431061807e-06,
      "loss": 1.5698,
      "step": 81
    },
    {
      "epoch": 0.011697574893009986,
      "grad_norm": 0.8109561800956726,
      "learning_rate": 1.283676703645008e-06,
      "loss": 1.633,
      "step": 82
    },
    {
      "epoch": 0.011840228245363765,
      "grad_norm": 0.9425901770591736,
      "learning_rate": 1.2995245641838353e-06,
      "loss": 1.4105,
      "step": 83
    },
    {
      "epoch": 0.011982881597717546,
      "grad_norm": 1.0923776626586914,
      "learning_rate": 1.3153724247226625e-06,
      "loss": 1.5132,
      "step": 84
    },
    {
      "epoch": 0.012125534950071327,
      "grad_norm": 0.7857595086097717,
      "learning_rate": 1.3312202852614897e-06,
      "loss": 1.5801,
      "step": 85
    },
    {
      "epoch": 0.012268188302425107,
      "grad_norm": 0.9032093286514282,
      "learning_rate": 1.3470681458003171e-06,
      "loss": 1.5532,
      "step": 86
    },
    {
      "epoch": 0.012410841654778888,
      "grad_norm": 0.994260847568512,
      "learning_rate": 1.3629160063391445e-06,
      "loss": 1.4339,
      "step": 87
    },
    {
      "epoch": 0.012553495007132667,
      "grad_norm": 0.6411845088005066,
      "learning_rate": 1.3787638668779715e-06,
      "loss": 1.5609,
      "step": 88
    },
    {
      "epoch": 0.012696148359486448,
      "grad_norm": 0.7005854845046997,
      "learning_rate": 1.394611727416799e-06,
      "loss": 1.5113,
      "step": 89
    },
    {
      "epoch": 0.012838801711840228,
      "grad_norm": 0.9326108694076538,
      "learning_rate": 1.4104595879556259e-06,
      "loss": 1.6072,
      "step": 90
    },
    {
      "epoch": 0.012981455064194009,
      "grad_norm": 0.9695913195610046,
      "learning_rate": 1.4263074484944533e-06,
      "loss": 1.3987,
      "step": 91
    },
    {
      "epoch": 0.013124108416547788,
      "grad_norm": 1.1596570014953613,
      "learning_rate": 1.4421553090332807e-06,
      "loss": 1.5134,
      "step": 92
    },
    {
      "epoch": 0.01326676176890157,
      "grad_norm": 1.1897401809692383,
      "learning_rate": 1.4580031695721079e-06,
      "loss": 1.6665,
      "step": 93
    },
    {
      "epoch": 0.013409415121255349,
      "grad_norm": 0.829107940196991,
      "learning_rate": 1.473851030110935e-06,
      "loss": 1.5133,
      "step": 94
    },
    {
      "epoch": 0.01355206847360913,
      "grad_norm": 0.7455458045005798,
      "learning_rate": 1.4896988906497625e-06,
      "loss": 1.585,
      "step": 95
    },
    {
      "epoch": 0.013694721825962911,
      "grad_norm": 0.758978009223938,
      "learning_rate": 1.5055467511885897e-06,
      "loss": 1.45,
      "step": 96
    },
    {
      "epoch": 0.01383737517831669,
      "grad_norm": 1.5633254051208496,
      "learning_rate": 1.521394611727417e-06,
      "loss": 1.5009,
      "step": 97
    },
    {
      "epoch": 0.013980028530670471,
      "grad_norm": 0.9118133187294006,
      "learning_rate": 1.537242472266244e-06,
      "loss": 1.5366,
      "step": 98
    },
    {
      "epoch": 0.01412268188302425,
      "grad_norm": 0.877566933631897,
      "learning_rate": 1.5530903328050714e-06,
      "loss": 1.5132,
      "step": 99
    },
    {
      "epoch": 0.014265335235378032,
      "grad_norm": 0.9620270133018494,
      "learning_rate": 1.5689381933438988e-06,
      "loss": 1.5842,
      "step": 100
    },
    {
      "epoch": 0.014407988587731811,
      "grad_norm": 1.0387673377990723,
      "learning_rate": 1.5847860538827258e-06,
      "loss": 1.4847,
      "step": 101
    },
    {
      "epoch": 0.014550641940085592,
      "grad_norm": 0.9701688885688782,
      "learning_rate": 1.6006339144215532e-06,
      "loss": 1.507,
      "step": 102
    },
    {
      "epoch": 0.014693295292439372,
      "grad_norm": 0.7710080146789551,
      "learning_rate": 1.6164817749603806e-06,
      "loss": 1.4709,
      "step": 103
    },
    {
      "epoch": 0.014835948644793153,
      "grad_norm": 0.6720123291015625,
      "learning_rate": 1.6323296354992076e-06,
      "loss": 1.5219,
      "step": 104
    },
    {
      "epoch": 0.014978601997146932,
      "grad_norm": 0.9765576124191284,
      "learning_rate": 1.648177496038035e-06,
      "loss": 1.4367,
      "step": 105
    },
    {
      "epoch": 0.015121255349500713,
      "grad_norm": 0.8261208534240723,
      "learning_rate": 1.6640253565768622e-06,
      "loss": 1.5229,
      "step": 106
    },
    {
      "epoch": 0.015263908701854494,
      "grad_norm": 0.6505538821220398,
      "learning_rate": 1.6798732171156896e-06,
      "loss": 1.53,
      "step": 107
    },
    {
      "epoch": 0.015406562054208274,
      "grad_norm": 0.8936634659767151,
      "learning_rate": 1.6957210776545168e-06,
      "loss": 1.4057,
      "step": 108
    },
    {
      "epoch": 0.015549215406562055,
      "grad_norm": 0.6728257536888123,
      "learning_rate": 1.711568938193344e-06,
      "loss": 1.5097,
      "step": 109
    },
    {
      "epoch": 0.015691868758915834,
      "grad_norm": 0.8790159225463867,
      "learning_rate": 1.7274167987321714e-06,
      "loss": 1.4864,
      "step": 110
    },
    {
      "epoch": 0.015834522111269615,
      "grad_norm": 0.7423803806304932,
      "learning_rate": 1.7432646592709988e-06,
      "loss": 1.5311,
      "step": 111
    },
    {
      "epoch": 0.015977175463623396,
      "grad_norm": 0.8326552510261536,
      "learning_rate": 1.7591125198098258e-06,
      "loss": 1.4527,
      "step": 112
    },
    {
      "epoch": 0.016119828815977174,
      "grad_norm": 0.570444643497467,
      "learning_rate": 1.7749603803486532e-06,
      "loss": 1.4319,
      "step": 113
    },
    {
      "epoch": 0.016262482168330955,
      "grad_norm": 1.3569951057434082,
      "learning_rate": 1.7908082408874801e-06,
      "loss": 1.2923,
      "step": 114
    },
    {
      "epoch": 0.016405135520684736,
      "grad_norm": 0.9770404696464539,
      "learning_rate": 1.8066561014263076e-06,
      "loss": 1.5535,
      "step": 115
    },
    {
      "epoch": 0.016547788873038517,
      "grad_norm": 0.5709965825080872,
      "learning_rate": 1.822503961965135e-06,
      "loss": 1.5976,
      "step": 116
    },
    {
      "epoch": 0.0166904422253923,
      "grad_norm": 0.6833900809288025,
      "learning_rate": 1.8383518225039621e-06,
      "loss": 1.5812,
      "step": 117
    },
    {
      "epoch": 0.016833095577746076,
      "grad_norm": 0.5152895450592041,
      "learning_rate": 1.8541996830427893e-06,
      "loss": 1.5289,
      "step": 118
    },
    {
      "epoch": 0.016975748930099857,
      "grad_norm": 1.2101069688796997,
      "learning_rate": 1.8700475435816165e-06,
      "loss": 1.401,
      "step": 119
    },
    {
      "epoch": 0.017118402282453638,
      "grad_norm": 0.7375568747520447,
      "learning_rate": 1.885895404120444e-06,
      "loss": 1.4543,
      "step": 120
    },
    {
      "epoch": 0.01726105563480742,
      "grad_norm": 0.5590512156486511,
      "learning_rate": 1.9017432646592713e-06,
      "loss": 1.4852,
      "step": 121
    },
    {
      "epoch": 0.017403708987161197,
      "grad_norm": 0.8461640477180481,
      "learning_rate": 1.9175911251980985e-06,
      "loss": 1.4435,
      "step": 122
    },
    {
      "epoch": 0.017546362339514978,
      "grad_norm": 0.8407596945762634,
      "learning_rate": 1.9334389857369257e-06,
      "loss": 1.3145,
      "step": 123
    },
    {
      "epoch": 0.01768901569186876,
      "grad_norm": 0.6710478067398071,
      "learning_rate": 1.949286846275753e-06,
      "loss": 1.6102,
      "step": 124
    },
    {
      "epoch": 0.01783166904422254,
      "grad_norm": 0.5850281119346619,
      "learning_rate": 1.96513470681458e-06,
      "loss": 1.4688,
      "step": 125
    },
    {
      "epoch": 0.017974322396576318,
      "grad_norm": 0.7517263293266296,
      "learning_rate": 1.9809825673534077e-06,
      "loss": 1.3069,
      "step": 126
    },
    {
      "epoch": 0.0181169757489301,
      "grad_norm": 0.534304141998291,
      "learning_rate": 1.9968304278922345e-06,
      "loss": 1.6399,
      "step": 127
    },
    {
      "epoch": 0.01825962910128388,
      "grad_norm": 0.6381033658981323,
      "learning_rate": 2.012678288431062e-06,
      "loss": 1.4489,
      "step": 128
    },
    {
      "epoch": 0.01840228245363766,
      "grad_norm": 0.5316737294197083,
      "learning_rate": 2.0285261489698893e-06,
      "loss": 1.5069,
      "step": 129
    },
    {
      "epoch": 0.018544935805991442,
      "grad_norm": 0.5412260293960571,
      "learning_rate": 2.0443740095087165e-06,
      "loss": 1.4341,
      "step": 130
    },
    {
      "epoch": 0.01868758915834522,
      "grad_norm": 0.5599603652954102,
      "learning_rate": 2.0602218700475437e-06,
      "loss": 1.5583,
      "step": 131
    },
    {
      "epoch": 0.018830242510699,
      "grad_norm": 0.7412648797035217,
      "learning_rate": 2.0760697305863713e-06,
      "loss": 1.3646,
      "step": 132
    },
    {
      "epoch": 0.018972895863052782,
      "grad_norm": 0.6201611757278442,
      "learning_rate": 2.091917591125198e-06,
      "loss": 1.3277,
      "step": 133
    },
    {
      "epoch": 0.019115549215406563,
      "grad_norm": 0.6947835683822632,
      "learning_rate": 2.1077654516640257e-06,
      "loss": 1.2874,
      "step": 134
    },
    {
      "epoch": 0.01925820256776034,
      "grad_norm": 0.7464353442192078,
      "learning_rate": 2.123613312202853e-06,
      "loss": 1.4791,
      "step": 135
    },
    {
      "epoch": 0.019400855920114122,
      "grad_norm": 0.9278052449226379,
      "learning_rate": 2.13946117274168e-06,
      "loss": 1.3071,
      "step": 136
    },
    {
      "epoch": 0.019543509272467903,
      "grad_norm": 0.713912844657898,
      "learning_rate": 2.1553090332805072e-06,
      "loss": 1.3207,
      "step": 137
    },
    {
      "epoch": 0.019686162624821684,
      "grad_norm": 0.7252569794654846,
      "learning_rate": 2.1711568938193344e-06,
      "loss": 1.5553,
      "step": 138
    },
    {
      "epoch": 0.019828815977175465,
      "grad_norm": 0.693581759929657,
      "learning_rate": 2.187004754358162e-06,
      "loss": 1.5425,
      "step": 139
    },
    {
      "epoch": 0.019971469329529243,
      "grad_norm": 0.48828667402267456,
      "learning_rate": 2.2028526148969892e-06,
      "loss": 1.4735,
      "step": 140
    },
    {
      "epoch": 0.020114122681883024,
      "grad_norm": 0.5852355360984802,
      "learning_rate": 2.2187004754358164e-06,
      "loss": 1.4065,
      "step": 141
    },
    {
      "epoch": 0.020256776034236805,
      "grad_norm": 0.47599881887435913,
      "learning_rate": 2.2345483359746436e-06,
      "loss": 1.3987,
      "step": 142
    },
    {
      "epoch": 0.020399429386590586,
      "grad_norm": 0.4909083843231201,
      "learning_rate": 2.250396196513471e-06,
      "loss": 1.4729,
      "step": 143
    },
    {
      "epoch": 0.020542082738944364,
      "grad_norm": 0.4750116467475891,
      "learning_rate": 2.266244057052298e-06,
      "loss": 1.3937,
      "step": 144
    },
    {
      "epoch": 0.020684736091298145,
      "grad_norm": 0.48616355657577515,
      "learning_rate": 2.2820919175911256e-06,
      "loss": 1.415,
      "step": 145
    },
    {
      "epoch": 0.020827389443651926,
      "grad_norm": 0.482998788356781,
      "learning_rate": 2.2979397781299524e-06,
      "loss": 1.3461,
      "step": 146
    },
    {
      "epoch": 0.020970042796005707,
      "grad_norm": 0.600152313709259,
      "learning_rate": 2.31378763866878e-06,
      "loss": 1.5215,
      "step": 147
    },
    {
      "epoch": 0.021112696148359488,
      "grad_norm": 0.6215013861656189,
      "learning_rate": 2.329635499207607e-06,
      "loss": 1.3436,
      "step": 148
    },
    {
      "epoch": 0.021255349500713266,
      "grad_norm": 0.5478700399398804,
      "learning_rate": 2.3454833597464344e-06,
      "loss": 1.532,
      "step": 149
    },
    {
      "epoch": 0.021398002853067047,
      "grad_norm": 0.42473381757736206,
      "learning_rate": 2.3613312202852615e-06,
      "loss": 1.375,
      "step": 150
    },
    {
      "epoch": 0.021540656205420828,
      "grad_norm": 0.41281232237815857,
      "learning_rate": 2.3771790808240887e-06,
      "loss": 1.5224,
      "step": 151
    },
    {
      "epoch": 0.02168330955777461,
      "grad_norm": 0.698550820350647,
      "learning_rate": 2.3930269413629164e-06,
      "loss": 1.326,
      "step": 152
    },
    {
      "epoch": 0.021825962910128387,
      "grad_norm": 1.0649795532226562,
      "learning_rate": 2.4088748019017435e-06,
      "loss": 1.0123,
      "step": 153
    },
    {
      "epoch": 0.021968616262482168,
      "grad_norm": 0.36036765575408936,
      "learning_rate": 2.4247226624405707e-06,
      "loss": 1.3621,
      "step": 154
    },
    {
      "epoch": 0.02211126961483595,
      "grad_norm": 0.5121074914932251,
      "learning_rate": 2.440570522979398e-06,
      "loss": 1.3121,
      "step": 155
    },
    {
      "epoch": 0.02225392296718973,
      "grad_norm": 0.5649195909500122,
      "learning_rate": 2.4564183835182255e-06,
      "loss": 1.407,
      "step": 156
    },
    {
      "epoch": 0.022396576319543508,
      "grad_norm": 0.7566866278648376,
      "learning_rate": 2.4722662440570523e-06,
      "loss": 1.334,
      "step": 157
    },
    {
      "epoch": 0.02253922967189729,
      "grad_norm": 0.6937580108642578,
      "learning_rate": 2.48811410459588e-06,
      "loss": 1.2586,
      "step": 158
    },
    {
      "epoch": 0.02268188302425107,
      "grad_norm": 0.5125687122344971,
      "learning_rate": 2.503961965134707e-06,
      "loss": 1.2403,
      "step": 159
    },
    {
      "epoch": 0.02282453637660485,
      "grad_norm": 0.43598663806915283,
      "learning_rate": 2.5198098256735343e-06,
      "loss": 1.3609,
      "step": 160
    },
    {
      "epoch": 0.022967189728958632,
      "grad_norm": 0.4160390794277191,
      "learning_rate": 2.5356576862123615e-06,
      "loss": 1.3124,
      "step": 161
    },
    {
      "epoch": 0.02310984308131241,
      "grad_norm": 0.32933762669563293,
      "learning_rate": 2.551505546751189e-06,
      "loss": 1.3765,
      "step": 162
    },
    {
      "epoch": 0.02325249643366619,
      "grad_norm": 0.3887583017349243,
      "learning_rate": 2.567353407290016e-06,
      "loss": 1.3972,
      "step": 163
    },
    {
      "epoch": 0.023395149786019972,
      "grad_norm": 0.3544297218322754,
      "learning_rate": 2.583201267828843e-06,
      "loss": 1.4004,
      "step": 164
    },
    {
      "epoch": 0.023537803138373753,
      "grad_norm": 0.370754599571228,
      "learning_rate": 2.5990491283676707e-06,
      "loss": 1.3913,
      "step": 165
    },
    {
      "epoch": 0.02368045649072753,
      "grad_norm": 0.3701103925704956,
      "learning_rate": 2.614896988906498e-06,
      "loss": 1.4598,
      "step": 166
    },
    {
      "epoch": 0.02382310984308131,
      "grad_norm": 0.3261151611804962,
      "learning_rate": 2.630744849445325e-06,
      "loss": 1.4104,
      "step": 167
    },
    {
      "epoch": 0.023965763195435093,
      "grad_norm": 0.3154659569263458,
      "learning_rate": 2.6465927099841527e-06,
      "loss": 1.3251,
      "step": 168
    },
    {
      "epoch": 0.024108416547788874,
      "grad_norm": 0.435077965259552,
      "learning_rate": 2.6624405705229794e-06,
      "loss": 1.3756,
      "step": 169
    },
    {
      "epoch": 0.024251069900142655,
      "grad_norm": 0.3625625967979431,
      "learning_rate": 2.6782884310618066e-06,
      "loss": 1.5361,
      "step": 170
    },
    {
      "epoch": 0.024393723252496433,
      "grad_norm": 0.48244497179985046,
      "learning_rate": 2.6941362916006342e-06,
      "loss": 1.3244,
      "step": 171
    },
    {
      "epoch": 0.024536376604850214,
      "grad_norm": 0.7815013527870178,
      "learning_rate": 2.7099841521394614e-06,
      "loss": 1.1812,
      "step": 172
    },
    {
      "epoch": 0.024679029957203995,
      "grad_norm": 0.3500225245952606,
      "learning_rate": 2.725832012678289e-06,
      "loss": 1.359,
      "step": 173
    },
    {
      "epoch": 0.024821683309557776,
      "grad_norm": 0.34128132462501526,
      "learning_rate": 2.741679873217116e-06,
      "loss": 1.3115,
      "step": 174
    },
    {
      "epoch": 0.024964336661911554,
      "grad_norm": 0.3174647092819214,
      "learning_rate": 2.757527733755943e-06,
      "loss": 1.3462,
      "step": 175
    },
    {
      "epoch": 0.025106990014265335,
      "grad_norm": 0.40917524695396423,
      "learning_rate": 2.7733755942947706e-06,
      "loss": 1.4537,
      "step": 176
    },
    {
      "epoch": 0.025249643366619116,
      "grad_norm": 0.3159307837486267,
      "learning_rate": 2.789223454833598e-06,
      "loss": 1.3603,
      "step": 177
    },
    {
      "epoch": 0.025392296718972897,
      "grad_norm": 0.37976720929145813,
      "learning_rate": 2.805071315372425e-06,
      "loss": 1.2648,
      "step": 178
    },
    {
      "epoch": 0.025534950071326678,
      "grad_norm": 0.37940192222595215,
      "learning_rate": 2.8209191759112518e-06,
      "loss": 1.2733,
      "step": 179
    },
    {
      "epoch": 0.025677603423680456,
      "grad_norm": 0.3686927556991577,
      "learning_rate": 2.8367670364500794e-06,
      "loss": 1.3113,
      "step": 180
    },
    {
      "epoch": 0.025820256776034237,
      "grad_norm": 0.35218343138694763,
      "learning_rate": 2.8526148969889066e-06,
      "loss": 1.3315,
      "step": 181
    },
    {
      "epoch": 0.025962910128388018,
      "grad_norm": 0.3224192261695862,
      "learning_rate": 2.868462757527734e-06,
      "loss": 1.4326,
      "step": 182
    },
    {
      "epoch": 0.0261055634807418,
      "grad_norm": 0.3251970112323761,
      "learning_rate": 2.8843106180665614e-06,
      "loss": 1.3612,
      "step": 183
    },
    {
      "epoch": 0.026248216833095576,
      "grad_norm": 0.3309905529022217,
      "learning_rate": 2.900158478605388e-06,
      "loss": 1.3877,
      "step": 184
    },
    {
      "epoch": 0.026390870185449358,
      "grad_norm": 0.4436019957065582,
      "learning_rate": 2.9160063391442158e-06,
      "loss": 1.3599,
      "step": 185
    },
    {
      "epoch": 0.02653352353780314,
      "grad_norm": 0.3721274733543396,
      "learning_rate": 2.931854199683043e-06,
      "loss": 1.2321,
      "step": 186
    },
    {
      "epoch": 0.02667617689015692,
      "grad_norm": 0.3032548725605011,
      "learning_rate": 2.94770206022187e-06,
      "loss": 1.4337,
      "step": 187
    },
    {
      "epoch": 0.026818830242510697,
      "grad_norm": 0.30257293581962585,
      "learning_rate": 2.9635499207606977e-06,
      "loss": 1.409,
      "step": 188
    },
    {
      "epoch": 0.02696148359486448,
      "grad_norm": 0.3350675702095032,
      "learning_rate": 2.979397781299525e-06,
      "loss": 1.3598,
      "step": 189
    },
    {
      "epoch": 0.02710413694721826,
      "grad_norm": 0.2964542806148529,
      "learning_rate": 2.9952456418383517e-06,
      "loss": 1.2141,
      "step": 190
    },
    {
      "epoch": 0.02724679029957204,
      "grad_norm": 0.4170527458190918,
      "learning_rate": 3.0110935023771793e-06,
      "loss": 1.3052,
      "step": 191
    },
    {
      "epoch": 0.027389443651925822,
      "grad_norm": 0.4259490370750427,
      "learning_rate": 3.0269413629160065e-06,
      "loss": 1.2928,
      "step": 192
    },
    {
      "epoch": 0.0275320970042796,
      "grad_norm": 0.35218384861946106,
      "learning_rate": 3.042789223454834e-06,
      "loss": 1.386,
      "step": 193
    },
    {
      "epoch": 0.02767475035663338,
      "grad_norm": 0.3103131055831909,
      "learning_rate": 3.0586370839936613e-06,
      "loss": 1.4354,
      "step": 194
    },
    {
      "epoch": 0.02781740370898716,
      "grad_norm": 0.2982793152332306,
      "learning_rate": 3.074484944532488e-06,
      "loss": 1.4955,
      "step": 195
    },
    {
      "epoch": 0.027960057061340943,
      "grad_norm": 0.3776632249355316,
      "learning_rate": 3.0903328050713157e-06,
      "loss": 1.3391,
      "step": 196
    },
    {
      "epoch": 0.02810271041369472,
      "grad_norm": 0.32819125056266785,
      "learning_rate": 3.106180665610143e-06,
      "loss": 1.3595,
      "step": 197
    },
    {
      "epoch": 0.0282453637660485,
      "grad_norm": 0.332262247800827,
      "learning_rate": 3.12202852614897e-06,
      "loss": 1.3278,
      "step": 198
    },
    {
      "epoch": 0.028388017118402283,
      "grad_norm": 0.3423677682876587,
      "learning_rate": 3.1378763866877977e-06,
      "loss": 1.4108,
      "step": 199
    },
    {
      "epoch": 0.028530670470756064,
      "grad_norm": 0.39997029304504395,
      "learning_rate": 3.1537242472266245e-06,
      "loss": 1.1477,
      "step": 200
    },
    {
      "epoch": 0.028673323823109845,
      "grad_norm": 0.42977041006088257,
      "learning_rate": 3.1695721077654516e-06,
      "loss": 1.1345,
      "step": 201
    },
    {
      "epoch": 0.028815977175463622,
      "grad_norm": 0.31790071725845337,
      "learning_rate": 3.1854199683042793e-06,
      "loss": 1.3127,
      "step": 202
    },
    {
      "epoch": 0.028958630527817403,
      "grad_norm": 0.3349977731704712,
      "learning_rate": 3.2012678288431065e-06,
      "loss": 1.3327,
      "step": 203
    },
    {
      "epoch": 0.029101283880171185,
      "grad_norm": 0.3176394999027252,
      "learning_rate": 3.217115689381934e-06,
      "loss": 1.2639,
      "step": 204
    },
    {
      "epoch": 0.029243937232524966,
      "grad_norm": 0.3080146908760071,
      "learning_rate": 3.2329635499207613e-06,
      "loss": 1.2867,
      "step": 205
    },
    {
      "epoch": 0.029386590584878743,
      "grad_norm": 0.37435632944107056,
      "learning_rate": 3.248811410459588e-06,
      "loss": 1.2979,
      "step": 206
    },
    {
      "epoch": 0.029529243937232524,
      "grad_norm": 0.32686150074005127,
      "learning_rate": 3.2646592709984152e-06,
      "loss": 1.3988,
      "step": 207
    },
    {
      "epoch": 0.029671897289586305,
      "grad_norm": 0.3797704875469208,
      "learning_rate": 3.280507131537243e-06,
      "loss": 1.2882,
      "step": 208
    },
    {
      "epoch": 0.029814550641940087,
      "grad_norm": 0.34049201011657715,
      "learning_rate": 3.29635499207607e-06,
      "loss": 1.3673,
      "step": 209
    },
    {
      "epoch": 0.029957203994293864,
      "grad_norm": 0.3301849365234375,
      "learning_rate": 3.3122028526148976e-06,
      "loss": 1.345,
      "step": 210
    },
    {
      "epoch": 0.030099857346647645,
      "grad_norm": 0.3224700093269348,
      "learning_rate": 3.3280507131537244e-06,
      "loss": 1.3089,
      "step": 211
    },
    {
      "epoch": 0.030242510699001426,
      "grad_norm": 0.3416121006011963,
      "learning_rate": 3.3438985736925516e-06,
      "loss": 1.2731,
      "step": 212
    },
    {
      "epoch": 0.030385164051355208,
      "grad_norm": 0.3295351564884186,
      "learning_rate": 3.359746434231379e-06,
      "loss": 1.306,
      "step": 213
    },
    {
      "epoch": 0.03052781740370899,
      "grad_norm": 0.6265162229537964,
      "learning_rate": 3.3755942947702064e-06,
      "loss": 1.1073,
      "step": 214
    },
    {
      "epoch": 0.030670470756062766,
      "grad_norm": 0.3294594883918762,
      "learning_rate": 3.3914421553090336e-06,
      "loss": 1.4362,
      "step": 215
    },
    {
      "epoch": 0.030813124108416547,
      "grad_norm": 0.2984050512313843,
      "learning_rate": 3.4072900158478608e-06,
      "loss": 1.3641,
      "step": 216
    },
    {
      "epoch": 0.03095577746077033,
      "grad_norm": 0.2959788143634796,
      "learning_rate": 3.423137876386688e-06,
      "loss": 1.2501,
      "step": 217
    },
    {
      "epoch": 0.03109843081312411,
      "grad_norm": 0.2951110601425171,
      "learning_rate": 3.438985736925515e-06,
      "loss": 1.3613,
      "step": 218
    },
    {
      "epoch": 0.031241084165477887,
      "grad_norm": 0.2745145559310913,
      "learning_rate": 3.4548335974643428e-06,
      "loss": 1.4284,
      "step": 219
    },
    {
      "epoch": 0.03138373751783167,
      "grad_norm": 0.3220887780189514,
      "learning_rate": 3.47068145800317e-06,
      "loss": 1.3333,
      "step": 220
    },
    {
      "epoch": 0.031526390870185446,
      "grad_norm": 0.30729013681411743,
      "learning_rate": 3.4865293185419976e-06,
      "loss": 1.2198,
      "step": 221
    },
    {
      "epoch": 0.03166904422253923,
      "grad_norm": 0.360538512468338,
      "learning_rate": 3.5023771790808243e-06,
      "loss": 1.3019,
      "step": 222
    },
    {
      "epoch": 0.03181169757489301,
      "grad_norm": 0.30400270223617554,
      "learning_rate": 3.5182250396196515e-06,
      "loss": 1.3406,
      "step": 223
    },
    {
      "epoch": 0.03195435092724679,
      "grad_norm": 0.37592488527297974,
      "learning_rate": 3.5340729001584787e-06,
      "loss": 1.104,
      "step": 224
    },
    {
      "epoch": 0.03209700427960057,
      "grad_norm": 0.32413458824157715,
      "learning_rate": 3.5499207606973063e-06,
      "loss": 1.4291,
      "step": 225
    },
    {
      "epoch": 0.03223965763195435,
      "grad_norm": 0.335815966129303,
      "learning_rate": 3.5657686212361335e-06,
      "loss": 1.3057,
      "step": 226
    },
    {
      "epoch": 0.03238231098430813,
      "grad_norm": 0.406950980424881,
      "learning_rate": 3.5816164817749603e-06,
      "loss": 1.3551,
      "step": 227
    },
    {
      "epoch": 0.03252496433666191,
      "grad_norm": 0.30737853050231934,
      "learning_rate": 3.597464342313788e-06,
      "loss": 1.2572,
      "step": 228
    },
    {
      "epoch": 0.032667617689015695,
      "grad_norm": 0.3183289170265198,
      "learning_rate": 3.613312202852615e-06,
      "loss": 1.2256,
      "step": 229
    },
    {
      "epoch": 0.03281027104136947,
      "grad_norm": 0.2918608784675598,
      "learning_rate": 3.6291600633914427e-06,
      "loss": 1.3548,
      "step": 230
    },
    {
      "epoch": 0.03295292439372325,
      "grad_norm": 0.38249361515045166,
      "learning_rate": 3.64500792393027e-06,
      "loss": 1.3721,
      "step": 231
    },
    {
      "epoch": 0.033095577746077035,
      "grad_norm": 0.35961297154426575,
      "learning_rate": 3.6608557844690967e-06,
      "loss": 1.1631,
      "step": 232
    },
    {
      "epoch": 0.03323823109843081,
      "grad_norm": 0.32576683163642883,
      "learning_rate": 3.6767036450079243e-06,
      "loss": 1.1663,
      "step": 233
    },
    {
      "epoch": 0.0333808844507846,
      "grad_norm": 0.42287713289260864,
      "learning_rate": 3.6925515055467515e-06,
      "loss": 1.1628,
      "step": 234
    },
    {
      "epoch": 0.033523537803138374,
      "grad_norm": 0.3085416853427887,
      "learning_rate": 3.7083993660855787e-06,
      "loss": 1.3782,
      "step": 235
    },
    {
      "epoch": 0.03366619115549215,
      "grad_norm": 0.43059131503105164,
      "learning_rate": 3.7242472266244063e-06,
      "loss": 1.2775,
      "step": 236
    },
    {
      "epoch": 0.03380884450784594,
      "grad_norm": 0.294572651386261,
      "learning_rate": 3.740095087163233e-06,
      "loss": 1.1738,
      "step": 237
    },
    {
      "epoch": 0.033951497860199714,
      "grad_norm": 0.47916534543037415,
      "learning_rate": 3.7559429477020602e-06,
      "loss": 1.2438,
      "step": 238
    },
    {
      "epoch": 0.03409415121255349,
      "grad_norm": 0.5922914147377014,
      "learning_rate": 3.771790808240888e-06,
      "loss": 1.1924,
      "step": 239
    },
    {
      "epoch": 0.034236804564907276,
      "grad_norm": 0.2903591990470886,
      "learning_rate": 3.787638668779715e-06,
      "loss": 1.3334,
      "step": 240
    },
    {
      "epoch": 0.034379457917261054,
      "grad_norm": 0.3299739360809326,
      "learning_rate": 3.8034865293185427e-06,
      "loss": 1.2186,
      "step": 241
    },
    {
      "epoch": 0.03452211126961484,
      "grad_norm": 0.38692760467529297,
      "learning_rate": 3.8193343898573694e-06,
      "loss": 1.2373,
      "step": 242
    },
    {
      "epoch": 0.034664764621968616,
      "grad_norm": 0.41160643100738525,
      "learning_rate": 3.835182250396197e-06,
      "loss": 1.184,
      "step": 243
    },
    {
      "epoch": 0.034807417974322394,
      "grad_norm": 0.3497057855129242,
      "learning_rate": 3.851030110935024e-06,
      "loss": 1.1723,
      "step": 244
    },
    {
      "epoch": 0.03495007132667618,
      "grad_norm": 0.2992457151412964,
      "learning_rate": 3.866877971473851e-06,
      "loss": 1.3942,
      "step": 245
    },
    {
      "epoch": 0.035092724679029956,
      "grad_norm": 0.37590697407722473,
      "learning_rate": 3.882725832012679e-06,
      "loss": 1.259,
      "step": 246
    },
    {
      "epoch": 0.03523537803138374,
      "grad_norm": 0.387542188167572,
      "learning_rate": 3.898573692551506e-06,
      "loss": 1.1846,
      "step": 247
    },
    {
      "epoch": 0.03537803138373752,
      "grad_norm": 0.2927229702472687,
      "learning_rate": 3.9144215530903326e-06,
      "loss": 1.318,
      "step": 248
    },
    {
      "epoch": 0.035520684736091296,
      "grad_norm": 0.31897106766700745,
      "learning_rate": 3.93026941362916e-06,
      "loss": 1.3613,
      "step": 249
    },
    {
      "epoch": 0.03566333808844508,
      "grad_norm": 0.33956417441368103,
      "learning_rate": 3.946117274167988e-06,
      "loss": 1.2053,
      "step": 250
    },
    {
      "epoch": 0.03580599144079886,
      "grad_norm": 0.46036967635154724,
      "learning_rate": 3.961965134706815e-06,
      "loss": 1.2807,
      "step": 251
    },
    {
      "epoch": 0.035948644793152636,
      "grad_norm": 0.40634241700172424,
      "learning_rate": 3.977812995245642e-06,
      "loss": 1.1571,
      "step": 252
    },
    {
      "epoch": 0.03609129814550642,
      "grad_norm": 0.319195419549942,
      "learning_rate": 3.993660855784469e-06,
      "loss": 1.0505,
      "step": 253
    },
    {
      "epoch": 0.0362339514978602,
      "grad_norm": 0.31796982884407043,
      "learning_rate": 4.0095087163232966e-06,
      "loss": 1.3175,
      "step": 254
    },
    {
      "epoch": 0.03637660485021398,
      "grad_norm": 0.43286457657814026,
      "learning_rate": 4.025356576862124e-06,
      "loss": 1.1349,
      "step": 255
    },
    {
      "epoch": 0.03651925820256776,
      "grad_norm": 0.31193092465400696,
      "learning_rate": 4.041204437400951e-06,
      "loss": 1.1305,
      "step": 256
    },
    {
      "epoch": 0.03666191155492154,
      "grad_norm": 0.28341153264045715,
      "learning_rate": 4.0570522979397786e-06,
      "loss": 1.3878,
      "step": 257
    },
    {
      "epoch": 0.03680456490727532,
      "grad_norm": 0.42940574884414673,
      "learning_rate": 4.072900158478606e-06,
      "loss": 1.2368,
      "step": 258
    },
    {
      "epoch": 0.0369472182596291,
      "grad_norm": 0.31089913845062256,
      "learning_rate": 4.088748019017433e-06,
      "loss": 1.2277,
      "step": 259
    },
    {
      "epoch": 0.037089871611982884,
      "grad_norm": 0.2877831757068634,
      "learning_rate": 4.1045958795562605e-06,
      "loss": 1.3913,
      "step": 260
    },
    {
      "epoch": 0.03723252496433666,
      "grad_norm": 0.297457218170166,
      "learning_rate": 4.120443740095087e-06,
      "loss": 1.2163,
      "step": 261
    },
    {
      "epoch": 0.03737517831669044,
      "grad_norm": 0.3337115943431854,
      "learning_rate": 4.136291600633915e-06,
      "loss": 1.275,
      "step": 262
    },
    {
      "epoch": 0.037517831669044224,
      "grad_norm": 0.33439674973487854,
      "learning_rate": 4.1521394611727425e-06,
      "loss": 1.2002,
      "step": 263
    },
    {
      "epoch": 0.037660485021398,
      "grad_norm": 0.3287540674209595,
      "learning_rate": 4.167987321711569e-06,
      "loss": 1.3298,
      "step": 264
    },
    {
      "epoch": 0.037803138373751786,
      "grad_norm": 0.29057860374450684,
      "learning_rate": 4.183835182250396e-06,
      "loss": 1.3222,
      "step": 265
    },
    {
      "epoch": 0.037945791726105564,
      "grad_norm": 0.30964189767837524,
      "learning_rate": 4.199683042789224e-06,
      "loss": 1.352,
      "step": 266
    },
    {
      "epoch": 0.03808844507845934,
      "grad_norm": 0.2988928258419037,
      "learning_rate": 4.215530903328051e-06,
      "loss": 1.3347,
      "step": 267
    },
    {
      "epoch": 0.038231098430813126,
      "grad_norm": 0.40428733825683594,
      "learning_rate": 4.231378763866879e-06,
      "loss": 1.0087,
      "step": 268
    },
    {
      "epoch": 0.038373751783166904,
      "grad_norm": 0.35499265789985657,
      "learning_rate": 4.247226624405706e-06,
      "loss": 1.1082,
      "step": 269
    },
    {
      "epoch": 0.03851640513552068,
      "grad_norm": 0.3120858073234558,
      "learning_rate": 4.2630744849445325e-06,
      "loss": 1.3458,
      "step": 270
    },
    {
      "epoch": 0.038659058487874466,
      "grad_norm": 0.5083685517311096,
      "learning_rate": 4.27892234548336e-06,
      "loss": 1.2777,
      "step": 271
    },
    {
      "epoch": 0.038801711840228244,
      "grad_norm": 0.5204631686210632,
      "learning_rate": 4.294770206022188e-06,
      "loss": 1.2005,
      "step": 272
    },
    {
      "epoch": 0.03894436519258203,
      "grad_norm": 0.3237505555152893,
      "learning_rate": 4.3106180665610144e-06,
      "loss": 1.2321,
      "step": 273
    },
    {
      "epoch": 0.039087018544935806,
      "grad_norm": 0.2766949236392975,
      "learning_rate": 4.326465927099842e-06,
      "loss": 1.2984,
      "step": 274
    },
    {
      "epoch": 0.039229671897289584,
      "grad_norm": 0.41957545280456543,
      "learning_rate": 4.342313787638669e-06,
      "loss": 1.2279,
      "step": 275
    },
    {
      "epoch": 0.03937232524964337,
      "grad_norm": 0.3055225610733032,
      "learning_rate": 4.3581616481774964e-06,
      "loss": 1.2772,
      "step": 276
    },
    {
      "epoch": 0.039514978601997146,
      "grad_norm": 0.28122588992118835,
      "learning_rate": 4.374009508716324e-06,
      "loss": 1.3133,
      "step": 277
    },
    {
      "epoch": 0.03965763195435093,
      "grad_norm": 0.3901839554309845,
      "learning_rate": 4.389857369255151e-06,
      "loss": 1.1696,
      "step": 278
    },
    {
      "epoch": 0.03980028530670471,
      "grad_norm": 0.35039961338043213,
      "learning_rate": 4.4057052297939784e-06,
      "loss": 1.428,
      "step": 279
    },
    {
      "epoch": 0.039942938659058486,
      "grad_norm": 0.3964681327342987,
      "learning_rate": 4.421553090332805e-06,
      "loss": 1.2232,
      "step": 280
    },
    {
      "epoch": 0.04008559201141227,
      "grad_norm": 0.3448876142501831,
      "learning_rate": 4.437400950871633e-06,
      "loss": 1.2975,
      "step": 281
    },
    {
      "epoch": 0.04022824536376605,
      "grad_norm": 0.3363805413246155,
      "learning_rate": 4.45324881141046e-06,
      "loss": 1.3109,
      "step": 282
    },
    {
      "epoch": 0.040370898716119825,
      "grad_norm": 0.4259723722934723,
      "learning_rate": 4.469096671949287e-06,
      "loss": 1.1703,
      "step": 283
    },
    {
      "epoch": 0.04051355206847361,
      "grad_norm": 0.3295013904571533,
      "learning_rate": 4.484944532488115e-06,
      "loss": 1.1904,
      "step": 284
    },
    {
      "epoch": 0.04065620542082739,
      "grad_norm": 0.323429673910141,
      "learning_rate": 4.500792393026942e-06,
      "loss": 1.1965,
      "step": 285
    },
    {
      "epoch": 0.04079885877318117,
      "grad_norm": 0.30424764752388,
      "learning_rate": 4.516640253565769e-06,
      "loss": 1.397,
      "step": 286
    },
    {
      "epoch": 0.04094151212553495,
      "grad_norm": 0.29817402362823486,
      "learning_rate": 4.532488114104596e-06,
      "loss": 1.2242,
      "step": 287
    },
    {
      "epoch": 0.04108416547788873,
      "grad_norm": 0.3749523460865021,
      "learning_rate": 4.5483359746434236e-06,
      "loss": 1.081,
      "step": 288
    },
    {
      "epoch": 0.04122681883024251,
      "grad_norm": 0.2756361961364746,
      "learning_rate": 4.564183835182251e-06,
      "loss": 1.2206,
      "step": 289
    },
    {
      "epoch": 0.04136947218259629,
      "grad_norm": 0.34127745032310486,
      "learning_rate": 4.580031695721078e-06,
      "loss": 1.2318,
      "step": 290
    },
    {
      "epoch": 0.041512125534950074,
      "grad_norm": 0.39298638701438904,
      "learning_rate": 4.595879556259905e-06,
      "loss": 1.2457,
      "step": 291
    },
    {
      "epoch": 0.04165477888730385,
      "grad_norm": 0.3586486577987671,
      "learning_rate": 4.611727416798732e-06,
      "loss": 1.289,
      "step": 292
    },
    {
      "epoch": 0.04179743223965763,
      "grad_norm": 0.396969735622406,
      "learning_rate": 4.62757527733756e-06,
      "loss": 1.0968,
      "step": 293
    },
    {
      "epoch": 0.041940085592011414,
      "grad_norm": 0.4026542007923126,
      "learning_rate": 4.6434231378763876e-06,
      "loss": 1.2031,
      "step": 294
    },
    {
      "epoch": 0.04208273894436519,
      "grad_norm": 0.4238973557949066,
      "learning_rate": 4.659270998415214e-06,
      "loss": 1.1247,
      "step": 295
    },
    {
      "epoch": 0.042225392296718976,
      "grad_norm": 0.30522915720939636,
      "learning_rate": 4.675118858954041e-06,
      "loss": 1.2861,
      "step": 296
    },
    {
      "epoch": 0.042368045649072754,
      "grad_norm": 0.411531001329422,
      "learning_rate": 4.690966719492869e-06,
      "loss": 1.1247,
      "step": 297
    },
    {
      "epoch": 0.04251069900142653,
      "grad_norm": 0.44647908210754395,
      "learning_rate": 4.706814580031696e-06,
      "loss": 1.132,
      "step": 298
    },
    {
      "epoch": 0.042653352353780316,
      "grad_norm": 0.2880479097366333,
      "learning_rate": 4.722662440570523e-06,
      "loss": 1.3464,
      "step": 299
    },
    {
      "epoch": 0.042796005706134094,
      "grad_norm": 0.29678839445114136,
      "learning_rate": 4.738510301109351e-06,
      "loss": 1.1978,
      "step": 300
    },
    {
      "epoch": 0.04293865905848787,
      "grad_norm": 0.42006635665893555,
      "learning_rate": 4.7543581616481775e-06,
      "loss": 1.1341,
      "step": 301
    },
    {
      "epoch": 0.043081312410841656,
      "grad_norm": 0.43602508306503296,
      "learning_rate": 4.770206022187005e-06,
      "loss": 1.1031,
      "step": 302
    },
    {
      "epoch": 0.043223965763195434,
      "grad_norm": 0.4025443494319916,
      "learning_rate": 4.786053882725833e-06,
      "loss": 1.0055,
      "step": 303
    },
    {
      "epoch": 0.04336661911554922,
      "grad_norm": 0.4227970242500305,
      "learning_rate": 4.8019017432646595e-06,
      "loss": 1.1083,
      "step": 304
    },
    {
      "epoch": 0.043509272467902996,
      "grad_norm": 0.3869151473045349,
      "learning_rate": 4.817749603803487e-06,
      "loss": 1.0709,
      "step": 305
    },
    {
      "epoch": 0.04365192582025677,
      "grad_norm": 0.3243708312511444,
      "learning_rate": 4.833597464342314e-06,
      "loss": 1.3049,
      "step": 306
    },
    {
      "epoch": 0.04379457917261056,
      "grad_norm": 0.46890708804130554,
      "learning_rate": 4.8494453248811415e-06,
      "loss": 1.1072,
      "step": 307
    },
    {
      "epoch": 0.043937232524964336,
      "grad_norm": 0.3239574730396271,
      "learning_rate": 4.865293185419968e-06,
      "loss": 1.3027,
      "step": 308
    },
    {
      "epoch": 0.04407988587731812,
      "grad_norm": 0.37250688672065735,
      "learning_rate": 4.881141045958796e-06,
      "loss": 1.0762,
      "step": 309
    },
    {
      "epoch": 0.0442225392296719,
      "grad_norm": 0.38411664962768555,
      "learning_rate": 4.8969889064976235e-06,
      "loss": 1.1813,
      "step": 310
    },
    {
      "epoch": 0.044365192582025675,
      "grad_norm": 0.35546326637268066,
      "learning_rate": 4.912836767036451e-06,
      "loss": 1.1954,
      "step": 311
    },
    {
      "epoch": 0.04450784593437946,
      "grad_norm": 0.3755302131175995,
      "learning_rate": 4.928684627575278e-06,
      "loss": 1.2159,
      "step": 312
    },
    {
      "epoch": 0.04465049928673324,
      "grad_norm": 0.30921611189842224,
      "learning_rate": 4.944532488114105e-06,
      "loss": 1.2202,
      "step": 313
    },
    {
      "epoch": 0.044793152639087015,
      "grad_norm": 0.3620322644710541,
      "learning_rate": 4.960380348652932e-06,
      "loss": 1.2663,
      "step": 314
    },
    {
      "epoch": 0.0449358059914408,
      "grad_norm": 0.3114851117134094,
      "learning_rate": 4.97622820919176e-06,
      "loss": 1.27,
      "step": 315
    },
    {
      "epoch": 0.04507845934379458,
      "grad_norm": 0.38563475012779236,
      "learning_rate": 4.992076069730587e-06,
      "loss": 1.1611,
      "step": 316
    },
    {
      "epoch": 0.04522111269614836,
      "grad_norm": 0.495448499917984,
      "learning_rate": 5.007923930269414e-06,
      "loss": 1.2343,
      "step": 317
    },
    {
      "epoch": 0.04536376604850214,
      "grad_norm": 0.318773478269577,
      "learning_rate": 5.023771790808242e-06,
      "loss": 1.0744,
      "step": 318
    },
    {
      "epoch": 0.04550641940085592,
      "grad_norm": 0.3277983069419861,
      "learning_rate": 5.039619651347069e-06,
      "loss": 1.1822,
      "step": 319
    },
    {
      "epoch": 0.0456490727532097,
      "grad_norm": 0.39151665568351746,
      "learning_rate": 5.055467511885895e-06,
      "loss": 1.0839,
      "step": 320
    },
    {
      "epoch": 0.04579172610556348,
      "grad_norm": 0.4079418182373047,
      "learning_rate": 5.071315372424723e-06,
      "loss": 1.2956,
      "step": 321
    },
    {
      "epoch": 0.045934379457917264,
      "grad_norm": 0.29466429352760315,
      "learning_rate": 5.08716323296355e-06,
      "loss": 1.3297,
      "step": 322
    },
    {
      "epoch": 0.04607703281027104,
      "grad_norm": 0.7656469941139221,
      "learning_rate": 5.103011093502378e-06,
      "loss": 0.8453,
      "step": 323
    },
    {
      "epoch": 0.04621968616262482,
      "grad_norm": 0.29712972044944763,
      "learning_rate": 5.118858954041205e-06,
      "loss": 1.2395,
      "step": 324
    },
    {
      "epoch": 0.046362339514978604,
      "grad_norm": 0.4239838421344757,
      "learning_rate": 5.134706814580032e-06,
      "loss": 1.1287,
      "step": 325
    },
    {
      "epoch": 0.04650499286733238,
      "grad_norm": 0.346843421459198,
      "learning_rate": 5.150554675118859e-06,
      "loss": 1.2082,
      "step": 326
    },
    {
      "epoch": 0.046647646219686166,
      "grad_norm": 0.4217950403690338,
      "learning_rate": 5.166402535657686e-06,
      "loss": 1.1445,
      "step": 327
    },
    {
      "epoch": 0.046790299572039944,
      "grad_norm": 0.43816739320755005,
      "learning_rate": 5.182250396196515e-06,
      "loss": 1.1049,
      "step": 328
    },
    {
      "epoch": 0.04693295292439372,
      "grad_norm": 0.30863091349601746,
      "learning_rate": 5.198098256735341e-06,
      "loss": 1.193,
      "step": 329
    },
    {
      "epoch": 0.047075606276747506,
      "grad_norm": 0.3303157091140747,
      "learning_rate": 5.213946117274168e-06,
      "loss": 1.1326,
      "step": 330
    },
    {
      "epoch": 0.047218259629101283,
      "grad_norm": 0.5755234956741333,
      "learning_rate": 5.229793977812996e-06,
      "loss": 0.9575,
      "step": 331
    },
    {
      "epoch": 0.04736091298145506,
      "grad_norm": 0.4737313985824585,
      "learning_rate": 5.2456418383518225e-06,
      "loss": 1.0288,
      "step": 332
    },
    {
      "epoch": 0.047503566333808846,
      "grad_norm": 0.29919010400772095,
      "learning_rate": 5.26148969889065e-06,
      "loss": 1.2875,
      "step": 333
    },
    {
      "epoch": 0.04764621968616262,
      "grad_norm": 0.36434921622276306,
      "learning_rate": 5.277337559429478e-06,
      "loss": 1.1288,
      "step": 334
    },
    {
      "epoch": 0.04778887303851641,
      "grad_norm": 0.43472111225128174,
      "learning_rate": 5.293185419968305e-06,
      "loss": 1.0283,
      "step": 335
    },
    {
      "epoch": 0.047931526390870186,
      "grad_norm": 0.4471413791179657,
      "learning_rate": 5.309033280507132e-06,
      "loss": 1.1391,
      "step": 336
    },
    {
      "epoch": 0.04807417974322396,
      "grad_norm": 0.43796202540397644,
      "learning_rate": 5.324881141045959e-06,
      "loss": 1.0755,
      "step": 337
    },
    {
      "epoch": 0.04821683309557775,
      "grad_norm": 0.5343884825706482,
      "learning_rate": 5.3407290015847865e-06,
      "loss": 1.067,
      "step": 338
    },
    {
      "epoch": 0.048359486447931525,
      "grad_norm": 0.34646904468536377,
      "learning_rate": 5.356576862123613e-06,
      "loss": 1.1613,
      "step": 339
    },
    {
      "epoch": 0.04850213980028531,
      "grad_norm": 0.45741909742355347,
      "learning_rate": 5.372424722662442e-06,
      "loss": 1.2105,
      "step": 340
    },
    {
      "epoch": 0.04864479315263909,
      "grad_norm": 0.36324140429496765,
      "learning_rate": 5.3882725832012685e-06,
      "loss": 1.175,
      "step": 341
    },
    {
      "epoch": 0.048787446504992865,
      "grad_norm": 0.33652055263519287,
      "learning_rate": 5.404120443740095e-06,
      "loss": 1.2252,
      "step": 342
    },
    {
      "epoch": 0.04893009985734665,
      "grad_norm": 0.36475199460983276,
      "learning_rate": 5.419968304278923e-06,
      "loss": 1.2679,
      "step": 343
    },
    {
      "epoch": 0.04907275320970043,
      "grad_norm": 0.3909691870212555,
      "learning_rate": 5.43581616481775e-06,
      "loss": 1.1804,
      "step": 344
    },
    {
      "epoch": 0.049215406562054205,
      "grad_norm": 0.3171705901622772,
      "learning_rate": 5.451664025356578e-06,
      "loss": 1.1083,
      "step": 345
    },
    {
      "epoch": 0.04935805991440799,
      "grad_norm": 0.38119977712631226,
      "learning_rate": 5.467511885895405e-06,
      "loss": 1.1648,
      "step": 346
    },
    {
      "epoch": 0.04950071326676177,
      "grad_norm": 0.39248302578926086,
      "learning_rate": 5.483359746434232e-06,
      "loss": 1.0853,
      "step": 347
    },
    {
      "epoch": 0.04964336661911555,
      "grad_norm": 0.5233859419822693,
      "learning_rate": 5.499207606973059e-06,
      "loss": 1.1315,
      "step": 348
    },
    {
      "epoch": 0.04978601997146933,
      "grad_norm": 0.4257233142852783,
      "learning_rate": 5.515055467511886e-06,
      "loss": 1.2126,
      "step": 349
    },
    {
      "epoch": 0.04992867332382311,
      "grad_norm": 0.37368786334991455,
      "learning_rate": 5.5309033280507145e-06,
      "loss": 1.2713,
      "step": 350
    },
    {
      "epoch": 0.05007132667617689,
      "grad_norm": 0.3926977217197418,
      "learning_rate": 5.546751188589541e-06,
      "loss": 1.044,
      "step": 351
    },
    {
      "epoch": 0.05021398002853067,
      "grad_norm": 0.3789297938346863,
      "learning_rate": 5.562599049128368e-06,
      "loss": 1.2206,
      "step": 352
    },
    {
      "epoch": 0.050356633380884454,
      "grad_norm": 0.42507195472717285,
      "learning_rate": 5.578446909667196e-06,
      "loss": 1.1481,
      "step": 353
    },
    {
      "epoch": 0.05049928673323823,
      "grad_norm": 0.40778741240501404,
      "learning_rate": 5.594294770206022e-06,
      "loss": 1.2038,
      "step": 354
    },
    {
      "epoch": 0.05064194008559201,
      "grad_norm": 0.4026924967765808,
      "learning_rate": 5.61014263074485e-06,
      "loss": 1.0337,
      "step": 355
    },
    {
      "epoch": 0.050784593437945794,
      "grad_norm": 0.44218987226486206,
      "learning_rate": 5.625990491283677e-06,
      "loss": 1.1486,
      "step": 356
    },
    {
      "epoch": 0.05092724679029957,
      "grad_norm": 0.4402460753917694,
      "learning_rate": 5.6418383518225035e-06,
      "loss": 1.0998,
      "step": 357
    },
    {
      "epoch": 0.051069900142653356,
      "grad_norm": 0.3994021713733673,
      "learning_rate": 5.657686212361332e-06,
      "loss": 1.0053,
      "step": 358
    },
    {
      "epoch": 0.05121255349500713,
      "grad_norm": 0.3759647309780121,
      "learning_rate": 5.673534072900159e-06,
      "loss": 1.1892,
      "step": 359
    },
    {
      "epoch": 0.05135520684736091,
      "grad_norm": 0.31760871410369873,
      "learning_rate": 5.689381933438986e-06,
      "loss": 1.3337,
      "step": 360
    },
    {
      "epoch": 0.051497860199714696,
      "grad_norm": 0.378974974155426,
      "learning_rate": 5.705229793977813e-06,
      "loss": 1.1321,
      "step": 361
    },
    {
      "epoch": 0.05164051355206847,
      "grad_norm": 0.4521256387233734,
      "learning_rate": 5.72107765451664e-06,
      "loss": 1.1835,
      "step": 362
    },
    {
      "epoch": 0.05178316690442225,
      "grad_norm": 0.330822229385376,
      "learning_rate": 5.736925515055468e-06,
      "loss": 1.1133,
      "step": 363
    },
    {
      "epoch": 0.051925820256776035,
      "grad_norm": 0.37037456035614014,
      "learning_rate": 5.752773375594295e-06,
      "loss": 1.1967,
      "step": 364
    },
    {
      "epoch": 0.05206847360912981,
      "grad_norm": 0.3412037491798401,
      "learning_rate": 5.768621236133123e-06,
      "loss": 1.2311,
      "step": 365
    },
    {
      "epoch": 0.0522111269614836,
      "grad_norm": 0.5195797085762024,
      "learning_rate": 5.7844690966719495e-06,
      "loss": 1.0661,
      "step": 366
    },
    {
      "epoch": 0.052353780313837375,
      "grad_norm": 0.546709418296814,
      "learning_rate": 5.800316957210776e-06,
      "loss": 1.138,
      "step": 367
    },
    {
      "epoch": 0.05249643366619115,
      "grad_norm": 0.3824395537376404,
      "learning_rate": 5.816164817749605e-06,
      "loss": 1.1791,
      "step": 368
    },
    {
      "epoch": 0.05263908701854494,
      "grad_norm": 0.3928437829017639,
      "learning_rate": 5.8320126782884315e-06,
      "loss": 1.117,
      "step": 369
    },
    {
      "epoch": 0.052781740370898715,
      "grad_norm": 0.40824761986732483,
      "learning_rate": 5.847860538827259e-06,
      "loss": 1.1497,
      "step": 370
    },
    {
      "epoch": 0.0529243937232525,
      "grad_norm": 0.8095877766609192,
      "learning_rate": 5.863708399366086e-06,
      "loss": 1.0953,
      "step": 371
    },
    {
      "epoch": 0.05306704707560628,
      "grad_norm": 0.4009456932544708,
      "learning_rate": 5.8795562599049135e-06,
      "loss": 1.1634,
      "step": 372
    },
    {
      "epoch": 0.053209700427960055,
      "grad_norm": 0.4667862057685852,
      "learning_rate": 5.89540412044374e-06,
      "loss": 1.0981,
      "step": 373
    },
    {
      "epoch": 0.05335235378031384,
      "grad_norm": 0.44600826501846313,
      "learning_rate": 5.911251980982568e-06,
      "loss": 1.1326,
      "step": 374
    },
    {
      "epoch": 0.05349500713266762,
      "grad_norm": 0.41945433616638184,
      "learning_rate": 5.9270998415213955e-06,
      "loss": 0.9984,
      "step": 375
    },
    {
      "epoch": 0.053637660485021395,
      "grad_norm": 0.5451641082763672,
      "learning_rate": 5.942947702060222e-06,
      "loss": 1.0338,
      "step": 376
    },
    {
      "epoch": 0.05378031383737518,
      "grad_norm": 0.4329877197742462,
      "learning_rate": 5.95879556259905e-06,
      "loss": 1.1208,
      "step": 377
    },
    {
      "epoch": 0.05392296718972896,
      "grad_norm": 0.4113557040691376,
      "learning_rate": 5.974643423137877e-06,
      "loss": 1.1937,
      "step": 378
    },
    {
      "epoch": 0.05406562054208274,
      "grad_norm": 0.46751633286476135,
      "learning_rate": 5.990491283676703e-06,
      "loss": 1.0489,
      "step": 379
    },
    {
      "epoch": 0.05420827389443652,
      "grad_norm": 0.5643234848976135,
      "learning_rate": 6.006339144215532e-06,
      "loss": 1.1551,
      "step": 380
    },
    {
      "epoch": 0.0543509272467903,
      "grad_norm": 0.547410249710083,
      "learning_rate": 6.022187004754359e-06,
      "loss": 1.0762,
      "step": 381
    },
    {
      "epoch": 0.05449358059914408,
      "grad_norm": 0.5132689476013184,
      "learning_rate": 6.038034865293186e-06,
      "loss": 0.9542,
      "step": 382
    },
    {
      "epoch": 0.05463623395149786,
      "grad_norm": 0.41245096921920776,
      "learning_rate": 6.053882725832013e-06,
      "loss": 1.11,
      "step": 383
    },
    {
      "epoch": 0.054778887303851644,
      "grad_norm": 0.39710038900375366,
      "learning_rate": 6.06973058637084e-06,
      "loss": 1.101,
      "step": 384
    },
    {
      "epoch": 0.05492154065620542,
      "grad_norm": 0.3879498541355133,
      "learning_rate": 6.085578446909668e-06,
      "loss": 1.0018,
      "step": 385
    },
    {
      "epoch": 0.0550641940085592,
      "grad_norm": 0.40068891644477844,
      "learning_rate": 6.101426307448495e-06,
      "loss": 0.9925,
      "step": 386
    },
    {
      "epoch": 0.05520684736091298,
      "grad_norm": 0.3478064239025116,
      "learning_rate": 6.117274167987323e-06,
      "loss": 1.2,
      "step": 387
    },
    {
      "epoch": 0.05534950071326676,
      "grad_norm": 0.44799110293388367,
      "learning_rate": 6.133122028526149e-06,
      "loss": 1.0138,
      "step": 388
    },
    {
      "epoch": 0.05549215406562054,
      "grad_norm": 0.39405953884124756,
      "learning_rate": 6.148969889064976e-06,
      "loss": 1.1252,
      "step": 389
    },
    {
      "epoch": 0.05563480741797432,
      "grad_norm": 0.5006358623504639,
      "learning_rate": 6.164817749603804e-06,
      "loss": 1.0465,
      "step": 390
    },
    {
      "epoch": 0.0557774607703281,
      "grad_norm": 0.4405211806297302,
      "learning_rate": 6.180665610142631e-06,
      "loss": 1.1541,
      "step": 391
    },
    {
      "epoch": 0.055920114122681885,
      "grad_norm": 0.5930861234664917,
      "learning_rate": 6.196513470681459e-06,
      "loss": 1.0188,
      "step": 392
    },
    {
      "epoch": 0.05606276747503566,
      "grad_norm": 0.39982444047927856,
      "learning_rate": 6.212361331220286e-06,
      "loss": 1.0072,
      "step": 393
    },
    {
      "epoch": 0.05620542082738944,
      "grad_norm": 0.35119161009788513,
      "learning_rate": 6.2282091917591125e-06,
      "loss": 1.1994,
      "step": 394
    },
    {
      "epoch": 0.056348074179743225,
      "grad_norm": 0.5308758020401001,
      "learning_rate": 6.24405705229794e-06,
      "loss": 1.1496,
      "step": 395
    },
    {
      "epoch": 0.056490727532097,
      "grad_norm": 0.45965972542762756,
      "learning_rate": 6.259904912836767e-06,
      "loss": 1.1157,
      "step": 396
    },
    {
      "epoch": 0.05663338088445079,
      "grad_norm": 0.3781106472015381,
      "learning_rate": 6.275752773375595e-06,
      "loss": 1.1653,
      "step": 397
    },
    {
      "epoch": 0.056776034236804565,
      "grad_norm": 0.3776165544986725,
      "learning_rate": 6.291600633914422e-06,
      "loss": 1.2303,
      "step": 398
    },
    {
      "epoch": 0.05691868758915834,
      "grad_norm": 0.5492984056472778,
      "learning_rate": 6.307448494453249e-06,
      "loss": 1.1351,
      "step": 399
    },
    {
      "epoch": 0.05706134094151213,
      "grad_norm": 0.4699248671531677,
      "learning_rate": 6.3232963549920765e-06,
      "loss": 1.1055,
      "step": 400
    },
    {
      "epoch": 0.057203994293865905,
      "grad_norm": 0.4486481845378876,
      "learning_rate": 6.339144215530903e-06,
      "loss": 1.088,
      "step": 401
    },
    {
      "epoch": 0.05734664764621969,
      "grad_norm": 0.41519489884376526,
      "learning_rate": 6.354992076069732e-06,
      "loss": 1.1259,
      "step": 402
    },
    {
      "epoch": 0.05748930099857347,
      "grad_norm": 0.44372814893722534,
      "learning_rate": 6.3708399366085585e-06,
      "loss": 0.9689,
      "step": 403
    },
    {
      "epoch": 0.057631954350927245,
      "grad_norm": 0.44161999225616455,
      "learning_rate": 6.386687797147385e-06,
      "loss": 1.116,
      "step": 404
    },
    {
      "epoch": 0.05777460770328103,
      "grad_norm": 0.4586813449859619,
      "learning_rate": 6.402535657686213e-06,
      "loss": 1.0441,
      "step": 405
    },
    {
      "epoch": 0.05791726105563481,
      "grad_norm": 0.4998525381088257,
      "learning_rate": 6.41838351822504e-06,
      "loss": 1.0546,
      "step": 406
    },
    {
      "epoch": 0.058059914407988585,
      "grad_norm": 0.4752885401248932,
      "learning_rate": 6.434231378763868e-06,
      "loss": 1.074,
      "step": 407
    },
    {
      "epoch": 0.05820256776034237,
      "grad_norm": 0.5583359599113464,
      "learning_rate": 6.450079239302695e-06,
      "loss": 0.9808,
      "step": 408
    },
    {
      "epoch": 0.05834522111269615,
      "grad_norm": 0.4140510559082031,
      "learning_rate": 6.4659270998415225e-06,
      "loss": 1.1221,
      "step": 409
    },
    {
      "epoch": 0.05848787446504993,
      "grad_norm": 0.43604913353919983,
      "learning_rate": 6.481774960380349e-06,
      "loss": 1.0223,
      "step": 410
    },
    {
      "epoch": 0.05863052781740371,
      "grad_norm": 0.4573204815387726,
      "learning_rate": 6.497622820919176e-06,
      "loss": 0.9948,
      "step": 411
    },
    {
      "epoch": 0.05877318116975749,
      "grad_norm": 0.48934316635131836,
      "learning_rate": 6.513470681458004e-06,
      "loss": 1.0465,
      "step": 412
    },
    {
      "epoch": 0.05891583452211127,
      "grad_norm": 1.1828285455703735,
      "learning_rate": 6.5293185419968304e-06,
      "loss": 0.8108,
      "step": 413
    },
    {
      "epoch": 0.05905848787446505,
      "grad_norm": 0.39499032497406006,
      "learning_rate": 6.545166402535659e-06,
      "loss": 1.199,
      "step": 414
    },
    {
      "epoch": 0.05920114122681883,
      "grad_norm": 0.4509989321231842,
      "learning_rate": 6.561014263074486e-06,
      "loss": 1.0853,
      "step": 415
    },
    {
      "epoch": 0.05934379457917261,
      "grad_norm": 0.40844517946243286,
      "learning_rate": 6.5768621236133124e-06,
      "loss": 1.1005,
      "step": 416
    },
    {
      "epoch": 0.05948644793152639,
      "grad_norm": 0.3970955014228821,
      "learning_rate": 6.59270998415214e-06,
      "loss": 1.1695,
      "step": 417
    },
    {
      "epoch": 0.05962910128388017,
      "grad_norm": 0.5938369631767273,
      "learning_rate": 6.608557844690967e-06,
      "loss": 1.1761,
      "step": 418
    },
    {
      "epoch": 0.05977175463623395,
      "grad_norm": 0.4537716209888458,
      "learning_rate": 6.624405705229795e-06,
      "loss": 1.0383,
      "step": 419
    },
    {
      "epoch": 0.05991440798858773,
      "grad_norm": 0.46481889486312866,
      "learning_rate": 6.640253565768622e-06,
      "loss": 1.0384,
      "step": 420
    },
    {
      "epoch": 0.06005706134094151,
      "grad_norm": 0.4273732006549835,
      "learning_rate": 6.656101426307449e-06,
      "loss": 1.0673,
      "step": 421
    },
    {
      "epoch": 0.06019971469329529,
      "grad_norm": 0.4316958785057068,
      "learning_rate": 6.671949286846276e-06,
      "loss": 1.0772,
      "step": 422
    },
    {
      "epoch": 0.060342368045649075,
      "grad_norm": 0.44716697931289673,
      "learning_rate": 6.687797147385103e-06,
      "loss": 1.1134,
      "step": 423
    },
    {
      "epoch": 0.06048502139800285,
      "grad_norm": 0.49333369731903076,
      "learning_rate": 6.703645007923932e-06,
      "loss": 0.9782,
      "step": 424
    },
    {
      "epoch": 0.06062767475035663,
      "grad_norm": 0.531298041343689,
      "learning_rate": 6.719492868462758e-06,
      "loss": 0.9299,
      "step": 425
    },
    {
      "epoch": 0.060770328102710415,
      "grad_norm": 0.578718900680542,
      "learning_rate": 6.735340729001585e-06,
      "loss": 1.0049,
      "step": 426
    },
    {
      "epoch": 0.06091298145506419,
      "grad_norm": 0.5718407034873962,
      "learning_rate": 6.751188589540413e-06,
      "loss": 1.0456,
      "step": 427
    },
    {
      "epoch": 0.06105563480741798,
      "grad_norm": 0.40977999567985535,
      "learning_rate": 6.7670364500792396e-06,
      "loss": 1.0773,
      "step": 428
    },
    {
      "epoch": 0.061198288159771755,
      "grad_norm": 0.46405255794525146,
      "learning_rate": 6.782884310618067e-06,
      "loss": 1.0508,
      "step": 429
    },
    {
      "epoch": 0.06134094151212553,
      "grad_norm": 0.5183796286582947,
      "learning_rate": 6.798732171156894e-06,
      "loss": 1.0275,
      "step": 430
    },
    {
      "epoch": 0.06148359486447932,
      "grad_norm": 0.4515964090824127,
      "learning_rate": 6.8145800316957216e-06,
      "loss": 1.0331,
      "step": 431
    },
    {
      "epoch": 0.061626248216833095,
      "grad_norm": 0.503455638885498,
      "learning_rate": 6.830427892234549e-06,
      "loss": 1.1121,
      "step": 432
    },
    {
      "epoch": 0.06176890156918688,
      "grad_norm": 0.5511139035224915,
      "learning_rate": 6.846275752773376e-06,
      "loss": 0.9589,
      "step": 433
    },
    {
      "epoch": 0.06191155492154066,
      "grad_norm": 0.5188602209091187,
      "learning_rate": 6.8621236133122035e-06,
      "loss": 1.0141,
      "step": 434
    },
    {
      "epoch": 0.062054208273894434,
      "grad_norm": 0.46196311712265015,
      "learning_rate": 6.87797147385103e-06,
      "loss": 1.0483,
      "step": 435
    },
    {
      "epoch": 0.06219686162624822,
      "grad_norm": 0.5451324582099915,
      "learning_rate": 6.893819334389857e-06,
      "loss": 1.1733,
      "step": 436
    },
    {
      "epoch": 0.062339514978602,
      "grad_norm": 0.6649710536003113,
      "learning_rate": 6.9096671949286855e-06,
      "loss": 1.0655,
      "step": 437
    },
    {
      "epoch": 0.062482168330955774,
      "grad_norm": 0.6782967448234558,
      "learning_rate": 6.925515055467512e-06,
      "loss": 1.0939,
      "step": 438
    },
    {
      "epoch": 0.06262482168330956,
      "grad_norm": 0.5909539461135864,
      "learning_rate": 6.94136291600634e-06,
      "loss": 0.9001,
      "step": 439
    },
    {
      "epoch": 0.06276747503566334,
      "grad_norm": 0.6379039287567139,
      "learning_rate": 6.957210776545167e-06,
      "loss": 1.0132,
      "step": 440
    },
    {
      "epoch": 0.06291012838801711,
      "grad_norm": 0.5297428965568542,
      "learning_rate": 6.973058637083995e-06,
      "loss": 1.0505,
      "step": 441
    },
    {
      "epoch": 0.06305278174037089,
      "grad_norm": 0.529805600643158,
      "learning_rate": 6.988906497622822e-06,
      "loss": 0.9536,
      "step": 442
    },
    {
      "epoch": 0.06319543509272468,
      "grad_norm": 0.4905278980731964,
      "learning_rate": 7.004754358161649e-06,
      "loss": 1.025,
      "step": 443
    },
    {
      "epoch": 0.06333808844507846,
      "grad_norm": 0.45372721552848816,
      "learning_rate": 7.020602218700476e-06,
      "loss": 0.9676,
      "step": 444
    },
    {
      "epoch": 0.06348074179743224,
      "grad_norm": 0.43510833382606506,
      "learning_rate": 7.036450079239303e-06,
      "loss": 1.0614,
      "step": 445
    },
    {
      "epoch": 0.06362339514978602,
      "grad_norm": 0.4841681718826294,
      "learning_rate": 7.052297939778131e-06,
      "loss": 0.9938,
      "step": 446
    },
    {
      "epoch": 0.0637660485021398,
      "grad_norm": 0.5671361684799194,
      "learning_rate": 7.0681458003169574e-06,
      "loss": 1.0196,
      "step": 447
    },
    {
      "epoch": 0.06390870185449359,
      "grad_norm": 0.5770955085754395,
      "learning_rate": 7.083993660855785e-06,
      "loss": 0.9989,
      "step": 448
    },
    {
      "epoch": 0.06405135520684736,
      "grad_norm": 0.4592321217060089,
      "learning_rate": 7.099841521394613e-06,
      "loss": 0.9775,
      "step": 449
    },
    {
      "epoch": 0.06419400855920114,
      "grad_norm": 0.5942267775535583,
      "learning_rate": 7.1156893819334394e-06,
      "loss": 1.0381,
      "step": 450
    },
    {
      "epoch": 0.06433666191155492,
      "grad_norm": 0.531999409198761,
      "learning_rate": 7.131537242472267e-06,
      "loss": 0.9506,
      "step": 451
    },
    {
      "epoch": 0.0644793152639087,
      "grad_norm": 0.5397354960441589,
      "learning_rate": 7.147385103011094e-06,
      "loss": 0.9741,
      "step": 452
    },
    {
      "epoch": 0.06462196861626249,
      "grad_norm": 0.5519461035728455,
      "learning_rate": 7.163232963549921e-06,
      "loss": 1.0911,
      "step": 453
    },
    {
      "epoch": 0.06476462196861626,
      "grad_norm": 0.5628547668457031,
      "learning_rate": 7.179080824088749e-06,
      "loss": 1.169,
      "step": 454
    },
    {
      "epoch": 0.06490727532097004,
      "grad_norm": 0.44366130232810974,
      "learning_rate": 7.194928684627576e-06,
      "loss": 1.0363,
      "step": 455
    },
    {
      "epoch": 0.06504992867332382,
      "grad_norm": 0.48540574312210083,
      "learning_rate": 7.2107765451664034e-06,
      "loss": 0.9253,
      "step": 456
    },
    {
      "epoch": 0.0651925820256776,
      "grad_norm": 0.39855659008026123,
      "learning_rate": 7.22662440570523e-06,
      "loss": 1.0376,
      "step": 457
    },
    {
      "epoch": 0.06533523537803139,
      "grad_norm": 0.4191644489765167,
      "learning_rate": 7.242472266244057e-06,
      "loss": 0.9906,
      "step": 458
    },
    {
      "epoch": 0.06547788873038517,
      "grad_norm": 0.437701016664505,
      "learning_rate": 7.2583201267828854e-06,
      "loss": 1.0432,
      "step": 459
    },
    {
      "epoch": 0.06562054208273894,
      "grad_norm": 0.5492372512817383,
      "learning_rate": 7.274167987321712e-06,
      "loss": 1.0186,
      "step": 460
    },
    {
      "epoch": 0.06576319543509272,
      "grad_norm": 0.4746062755584717,
      "learning_rate": 7.29001584786054e-06,
      "loss": 1.0051,
      "step": 461
    },
    {
      "epoch": 0.0659058487874465,
      "grad_norm": 0.4426122009754181,
      "learning_rate": 7.305863708399367e-06,
      "loss": 0.9692,
      "step": 462
    },
    {
      "epoch": 0.06604850213980029,
      "grad_norm": 0.5334264039993286,
      "learning_rate": 7.321711568938193e-06,
      "loss": 1.0676,
      "step": 463
    },
    {
      "epoch": 0.06619115549215407,
      "grad_norm": 0.6350341439247131,
      "learning_rate": 7.337559429477021e-06,
      "loss": 1.0368,
      "step": 464
    },
    {
      "epoch": 0.06633380884450785,
      "grad_norm": 0.5225725769996643,
      "learning_rate": 7.3534072900158486e-06,
      "loss": 0.9821,
      "step": 465
    },
    {
      "epoch": 0.06647646219686162,
      "grad_norm": 0.4992254972457886,
      "learning_rate": 7.369255150554676e-06,
      "loss": 0.9978,
      "step": 466
    },
    {
      "epoch": 0.0666191155492154,
      "grad_norm": 0.4909706711769104,
      "learning_rate": 7.385103011093503e-06,
      "loss": 1.0252,
      "step": 467
    },
    {
      "epoch": 0.0667617689015692,
      "grad_norm": 0.48005351424217224,
      "learning_rate": 7.40095087163233e-06,
      "loss": 1.0211,
      "step": 468
    },
    {
      "epoch": 0.06690442225392297,
      "grad_norm": 0.40804851055145264,
      "learning_rate": 7.416798732171157e-06,
      "loss": 1.0521,
      "step": 469
    },
    {
      "epoch": 0.06704707560627675,
      "grad_norm": 0.4457440674304962,
      "learning_rate": 7.432646592709984e-06,
      "loss": 0.915,
      "step": 470
    },
    {
      "epoch": 0.06718972895863053,
      "grad_norm": 0.5630983114242554,
      "learning_rate": 7.4484944532488126e-06,
      "loss": 0.9437,
      "step": 471
    },
    {
      "epoch": 0.0673323823109843,
      "grad_norm": 0.4951111972332001,
      "learning_rate": 7.464342313787639e-06,
      "loss": 0.981,
      "step": 472
    },
    {
      "epoch": 0.06747503566333808,
      "grad_norm": 1.3637419939041138,
      "learning_rate": 7.480190174326466e-06,
      "loss": 0.8505,
      "step": 473
    },
    {
      "epoch": 0.06761768901569187,
      "grad_norm": 0.42957380414009094,
      "learning_rate": 7.496038034865294e-06,
      "loss": 1.0522,
      "step": 474
    },
    {
      "epoch": 0.06776034236804565,
      "grad_norm": 0.4950065016746521,
      "learning_rate": 7.5118858954041205e-06,
      "loss": 1.0665,
      "step": 475
    },
    {
      "epoch": 0.06790299572039943,
      "grad_norm": 0.4553673565387726,
      "learning_rate": 7.527733755942949e-06,
      "loss": 0.9747,
      "step": 476
    },
    {
      "epoch": 0.0680456490727532,
      "grad_norm": 0.6075873970985413,
      "learning_rate": 7.543581616481776e-06,
      "loss": 0.9686,
      "step": 477
    },
    {
      "epoch": 0.06818830242510698,
      "grad_norm": 0.4874385893344879,
      "learning_rate": 7.559429477020603e-06,
      "loss": 1.0179,
      "step": 478
    },
    {
      "epoch": 0.06833095577746078,
      "grad_norm": 0.44387388229370117,
      "learning_rate": 7.57527733755943e-06,
      "loss": 1.0731,
      "step": 479
    },
    {
      "epoch": 0.06847360912981455,
      "grad_norm": 0.5947974324226379,
      "learning_rate": 7.591125198098257e-06,
      "loss": 0.985,
      "step": 480
    },
    {
      "epoch": 0.06861626248216833,
      "grad_norm": 0.6631782650947571,
      "learning_rate": 7.606973058637085e-06,
      "loss": 1.0244,
      "step": 481
    },
    {
      "epoch": 0.06875891583452211,
      "grad_norm": 0.46486252546310425,
      "learning_rate": 7.622820919175912e-06,
      "loss": 0.9863,
      "step": 482
    },
    {
      "epoch": 0.06890156918687589,
      "grad_norm": 0.49008890986442566,
      "learning_rate": 7.638668779714739e-06,
      "loss": 0.9831,
      "step": 483
    },
    {
      "epoch": 0.06904422253922968,
      "grad_norm": 0.6009377837181091,
      "learning_rate": 7.654516640253566e-06,
      "loss": 0.9774,
      "step": 484
    },
    {
      "epoch": 0.06918687589158345,
      "grad_norm": 0.4419824182987213,
      "learning_rate": 7.670364500792394e-06,
      "loss": 0.9439,
      "step": 485
    },
    {
      "epoch": 0.06932952924393723,
      "grad_norm": 0.5547300577163696,
      "learning_rate": 7.686212361331222e-06,
      "loss": 0.9258,
      "step": 486
    },
    {
      "epoch": 0.06947218259629101,
      "grad_norm": 0.5227124094963074,
      "learning_rate": 7.702060221870048e-06,
      "loss": 1.0545,
      "step": 487
    },
    {
      "epoch": 0.06961483594864479,
      "grad_norm": 0.45763492584228516,
      "learning_rate": 7.717908082408875e-06,
      "loss": 0.952,
      "step": 488
    },
    {
      "epoch": 0.06975748930099858,
      "grad_norm": 0.563387393951416,
      "learning_rate": 7.733755942947703e-06,
      "loss": 0.9204,
      "step": 489
    },
    {
      "epoch": 0.06990014265335236,
      "grad_norm": 0.697047233581543,
      "learning_rate": 7.749603803486529e-06,
      "loss": 0.968,
      "step": 490
    },
    {
      "epoch": 0.07004279600570613,
      "grad_norm": 0.5165669322013855,
      "learning_rate": 7.765451664025358e-06,
      "loss": 1.0207,
      "step": 491
    },
    {
      "epoch": 0.07018544935805991,
      "grad_norm": 0.7232952117919922,
      "learning_rate": 7.781299524564184e-06,
      "loss": 0.78,
      "step": 492
    },
    {
      "epoch": 0.07032810271041369,
      "grad_norm": 0.5339188575744629,
      "learning_rate": 7.797147385103012e-06,
      "loss": 0.9616,
      "step": 493
    },
    {
      "epoch": 0.07047075606276748,
      "grad_norm": 0.4808948040008545,
      "learning_rate": 7.81299524564184e-06,
      "loss": 0.8511,
      "step": 494
    },
    {
      "epoch": 0.07061340941512126,
      "grad_norm": 0.5292257070541382,
      "learning_rate": 7.828843106180665e-06,
      "loss": 0.9998,
      "step": 495
    },
    {
      "epoch": 0.07075606276747504,
      "grad_norm": 0.5296969413757324,
      "learning_rate": 7.844690966719494e-06,
      "loss": 0.9505,
      "step": 496
    },
    {
      "epoch": 0.07089871611982881,
      "grad_norm": 0.6419568061828613,
      "learning_rate": 7.86053882725832e-06,
      "loss": 0.9637,
      "step": 497
    },
    {
      "epoch": 0.07104136947218259,
      "grad_norm": 0.900155782699585,
      "learning_rate": 7.876386687797148e-06,
      "loss": 0.771,
      "step": 498
    },
    {
      "epoch": 0.07118402282453638,
      "grad_norm": 0.48271769285202026,
      "learning_rate": 7.892234548335976e-06,
      "loss": 0.9403,
      "step": 499
    },
    {
      "epoch": 0.07132667617689016,
      "grad_norm": 0.5156475901603699,
      "learning_rate": 7.908082408874802e-06,
      "loss": 0.8973,
      "step": 500
    },
    {
      "epoch": 0.07146932952924394,
      "grad_norm": 0.5376582741737366,
      "learning_rate": 7.92393026941363e-06,
      "loss": 0.9655,
      "step": 501
    },
    {
      "epoch": 0.07161198288159772,
      "grad_norm": 0.4439854025840759,
      "learning_rate": 7.939778129952457e-06,
      "loss": 0.9807,
      "step": 502
    },
    {
      "epoch": 0.0717546362339515,
      "grad_norm": 0.537900984287262,
      "learning_rate": 7.955625990491284e-06,
      "loss": 1.0537,
      "step": 503
    },
    {
      "epoch": 0.07189728958630527,
      "grad_norm": 0.5043521523475647,
      "learning_rate": 7.971473851030112e-06,
      "loss": 0.9799,
      "step": 504
    },
    {
      "epoch": 0.07203994293865906,
      "grad_norm": 0.509040892124176,
      "learning_rate": 7.987321711568938e-06,
      "loss": 1.0014,
      "step": 505
    },
    {
      "epoch": 0.07218259629101284,
      "grad_norm": 0.5832608342170715,
      "learning_rate": 8.003169572107765e-06,
      "loss": 0.9699,
      "step": 506
    },
    {
      "epoch": 0.07232524964336662,
      "grad_norm": 0.5468351244926453,
      "learning_rate": 8.019017432646593e-06,
      "loss": 1.0328,
      "step": 507
    },
    {
      "epoch": 0.0724679029957204,
      "grad_norm": 0.6355152726173401,
      "learning_rate": 8.03486529318542e-06,
      "loss": 1.0518,
      "step": 508
    },
    {
      "epoch": 0.07261055634807417,
      "grad_norm": 0.4258562922477722,
      "learning_rate": 8.050713153724248e-06,
      "loss": 0.9437,
      "step": 509
    },
    {
      "epoch": 0.07275320970042796,
      "grad_norm": 0.514599084854126,
      "learning_rate": 8.066561014263076e-06,
      "loss": 1.0423,
      "step": 510
    },
    {
      "epoch": 0.07289586305278174,
      "grad_norm": 0.4540145993232727,
      "learning_rate": 8.082408874801902e-06,
      "loss": 0.9921,
      "step": 511
    },
    {
      "epoch": 0.07303851640513552,
      "grad_norm": 0.8550832867622375,
      "learning_rate": 8.09825673534073e-06,
      "loss": 0.8387,
      "step": 512
    },
    {
      "epoch": 0.0731811697574893,
      "grad_norm": 0.45189788937568665,
      "learning_rate": 8.114104595879557e-06,
      "loss": 1.0554,
      "step": 513
    },
    {
      "epoch": 0.07332382310984308,
      "grad_norm": 0.5321663022041321,
      "learning_rate": 8.129952456418385e-06,
      "loss": 0.8751,
      "step": 514
    },
    {
      "epoch": 0.07346647646219687,
      "grad_norm": 0.5127700567245483,
      "learning_rate": 8.145800316957212e-06,
      "loss": 0.972,
      "step": 515
    },
    {
      "epoch": 0.07360912981455064,
      "grad_norm": 1.4596971273422241,
      "learning_rate": 8.161648177496038e-06,
      "loss": 0.7484,
      "step": 516
    },
    {
      "epoch": 0.07375178316690442,
      "grad_norm": 0.5196806192398071,
      "learning_rate": 8.177496038034866e-06,
      "loss": 0.929,
      "step": 517
    },
    {
      "epoch": 0.0738944365192582,
      "grad_norm": 0.5517879724502563,
      "learning_rate": 8.193343898573693e-06,
      "loss": 0.8865,
      "step": 518
    },
    {
      "epoch": 0.07403708987161198,
      "grad_norm": 0.5994307994842529,
      "learning_rate": 8.209191759112521e-06,
      "loss": 0.8375,
      "step": 519
    },
    {
      "epoch": 0.07417974322396577,
      "grad_norm": 0.6686751246452332,
      "learning_rate": 8.225039619651349e-06,
      "loss": 0.9716,
      "step": 520
    },
    {
      "epoch": 0.07432239657631955,
      "grad_norm": 0.4750771224498749,
      "learning_rate": 8.240887480190175e-06,
      "loss": 0.9878,
      "step": 521
    },
    {
      "epoch": 0.07446504992867332,
      "grad_norm": 0.5097470283508301,
      "learning_rate": 8.256735340729002e-06,
      "loss": 1.0066,
      "step": 522
    },
    {
      "epoch": 0.0746077032810271,
      "grad_norm": 0.43969038128852844,
      "learning_rate": 8.27258320126783e-06,
      "loss": 0.9947,
      "step": 523
    },
    {
      "epoch": 0.07475035663338088,
      "grad_norm": 0.3874143064022064,
      "learning_rate": 8.288431061806656e-06,
      "loss": 1.0651,
      "step": 524
    },
    {
      "epoch": 0.07489300998573467,
      "grad_norm": 0.5227018594741821,
      "learning_rate": 8.304278922345485e-06,
      "loss": 0.9543,
      "step": 525
    },
    {
      "epoch": 0.07503566333808845,
      "grad_norm": 0.7560486197471619,
      "learning_rate": 8.320126782884311e-06,
      "loss": 1.0052,
      "step": 526
    },
    {
      "epoch": 0.07517831669044223,
      "grad_norm": 0.6004654765129089,
      "learning_rate": 8.335974643423139e-06,
      "loss": 1.0591,
      "step": 527
    },
    {
      "epoch": 0.075320970042796,
      "grad_norm": 0.4558890461921692,
      "learning_rate": 8.351822503961966e-06,
      "loss": 1.0117,
      "step": 528
    },
    {
      "epoch": 0.07546362339514978,
      "grad_norm": 0.5768234133720398,
      "learning_rate": 8.367670364500792e-06,
      "loss": 0.9102,
      "step": 529
    },
    {
      "epoch": 0.07560627674750357,
      "grad_norm": 0.5694892406463623,
      "learning_rate": 8.383518225039621e-06,
      "loss": 1.0243,
      "step": 530
    },
    {
      "epoch": 0.07574893009985735,
      "grad_norm": 0.49134907126426697,
      "learning_rate": 8.399366085578447e-06,
      "loss": 0.9375,
      "step": 531
    },
    {
      "epoch": 0.07589158345221113,
      "grad_norm": 0.5538228750228882,
      "learning_rate": 8.415213946117275e-06,
      "loss": 0.9334,
      "step": 532
    },
    {
      "epoch": 0.0760342368045649,
      "grad_norm": 0.48530998826026917,
      "learning_rate": 8.431061806656103e-06,
      "loss": 0.9813,
      "step": 533
    },
    {
      "epoch": 0.07617689015691868,
      "grad_norm": 0.4936850666999817,
      "learning_rate": 8.446909667194929e-06,
      "loss": 0.9924,
      "step": 534
    },
    {
      "epoch": 0.07631954350927246,
      "grad_norm": 0.43529391288757324,
      "learning_rate": 8.462757527733758e-06,
      "loss": 0.9429,
      "step": 535
    },
    {
      "epoch": 0.07646219686162625,
      "grad_norm": 0.48769012093544006,
      "learning_rate": 8.478605388272584e-06,
      "loss": 0.9897,
      "step": 536
    },
    {
      "epoch": 0.07660485021398003,
      "grad_norm": 0.5367408990859985,
      "learning_rate": 8.494453248811411e-06,
      "loss": 0.9396,
      "step": 537
    },
    {
      "epoch": 0.07674750356633381,
      "grad_norm": 0.5971816182136536,
      "learning_rate": 8.510301109350239e-06,
      "loss": 0.9168,
      "step": 538
    },
    {
      "epoch": 0.07689015691868759,
      "grad_norm": 0.7000044584274292,
      "learning_rate": 8.526148969889065e-06,
      "loss": 0.949,
      "step": 539
    },
    {
      "epoch": 0.07703281027104136,
      "grad_norm": 0.43918800354003906,
      "learning_rate": 8.541996830427893e-06,
      "loss": 1.0051,
      "step": 540
    },
    {
      "epoch": 0.07717546362339515,
      "grad_norm": 0.6692291498184204,
      "learning_rate": 8.55784469096672e-06,
      "loss": 0.9125,
      "step": 541
    },
    {
      "epoch": 0.07731811697574893,
      "grad_norm": 0.6336300373077393,
      "learning_rate": 8.573692551505548e-06,
      "loss": 1.036,
      "step": 542
    },
    {
      "epoch": 0.07746077032810271,
      "grad_norm": 0.47780126333236694,
      "learning_rate": 8.589540412044375e-06,
      "loss": 0.9203,
      "step": 543
    },
    {
      "epoch": 0.07760342368045649,
      "grad_norm": 0.5549113154411316,
      "learning_rate": 8.605388272583201e-06,
      "loss": 1.0006,
      "step": 544
    },
    {
      "epoch": 0.07774607703281027,
      "grad_norm": 0.6125926971435547,
      "learning_rate": 8.621236133122029e-06,
      "loss": 0.9076,
      "step": 545
    },
    {
      "epoch": 0.07788873038516406,
      "grad_norm": 0.5332480669021606,
      "learning_rate": 8.637083993660857e-06,
      "loss": 0.8981,
      "step": 546
    },
    {
      "epoch": 0.07803138373751783,
      "grad_norm": 0.5666670203208923,
      "learning_rate": 8.652931854199684e-06,
      "loss": 0.9091,
      "step": 547
    },
    {
      "epoch": 0.07817403708987161,
      "grad_norm": 0.5022792816162109,
      "learning_rate": 8.668779714738512e-06,
      "loss": 1.0507,
      "step": 548
    },
    {
      "epoch": 0.07831669044222539,
      "grad_norm": 0.5982717275619507,
      "learning_rate": 8.684627575277338e-06,
      "loss": 0.9029,
      "step": 549
    },
    {
      "epoch": 0.07845934379457917,
      "grad_norm": 0.6098309755325317,
      "learning_rate": 8.700475435816165e-06,
      "loss": 0.9194,
      "step": 550
    },
    {
      "epoch": 0.07860199714693296,
      "grad_norm": 0.5017668008804321,
      "learning_rate": 8.716323296354993e-06,
      "loss": 0.9662,
      "step": 551
    },
    {
      "epoch": 0.07874465049928674,
      "grad_norm": 0.5837364792823792,
      "learning_rate": 8.73217115689382e-06,
      "loss": 0.9717,
      "step": 552
    },
    {
      "epoch": 0.07888730385164051,
      "grad_norm": 0.4677683115005493,
      "learning_rate": 8.748019017432648e-06,
      "loss": 1.0237,
      "step": 553
    },
    {
      "epoch": 0.07902995720399429,
      "grad_norm": 0.502824604511261,
      "learning_rate": 8.763866877971474e-06,
      "loss": 0.9298,
      "step": 554
    },
    {
      "epoch": 0.07917261055634807,
      "grad_norm": 0.48983949422836304,
      "learning_rate": 8.779714738510302e-06,
      "loss": 1.0045,
      "step": 555
    },
    {
      "epoch": 0.07931526390870186,
      "grad_norm": 0.500971257686615,
      "learning_rate": 8.79556259904913e-06,
      "loss": 1.0793,
      "step": 556
    },
    {
      "epoch": 0.07945791726105564,
      "grad_norm": 0.7364229559898376,
      "learning_rate": 8.811410459587957e-06,
      "loss": 0.9338,
      "step": 557
    },
    {
      "epoch": 0.07960057061340942,
      "grad_norm": 0.5840064287185669,
      "learning_rate": 8.827258320126783e-06,
      "loss": 0.9634,
      "step": 558
    },
    {
      "epoch": 0.0797432239657632,
      "grad_norm": 0.6440963745117188,
      "learning_rate": 8.84310618066561e-06,
      "loss": 0.9646,
      "step": 559
    },
    {
      "epoch": 0.07988587731811697,
      "grad_norm": 0.4333682060241699,
      "learning_rate": 8.858954041204438e-06,
      "loss": 0.9591,
      "step": 560
    },
    {
      "epoch": 0.08002853067047076,
      "grad_norm": 0.7911944389343262,
      "learning_rate": 8.874801901743266e-06,
      "loss": 0.9921,
      "step": 561
    },
    {
      "epoch": 0.08017118402282454,
      "grad_norm": 0.5133646130561829,
      "learning_rate": 8.890649762282093e-06,
      "loss": 0.991,
      "step": 562
    },
    {
      "epoch": 0.08031383737517832,
      "grad_norm": 0.6080657243728638,
      "learning_rate": 8.90649762282092e-06,
      "loss": 0.9691,
      "step": 563
    },
    {
      "epoch": 0.0804564907275321,
      "grad_norm": 0.5872765183448792,
      "learning_rate": 8.922345483359747e-06,
      "loss": 0.957,
      "step": 564
    },
    {
      "epoch": 0.08059914407988587,
      "grad_norm": 0.5557759404182434,
      "learning_rate": 8.938193343898574e-06,
      "loss": 0.962,
      "step": 565
    },
    {
      "epoch": 0.08074179743223965,
      "grad_norm": 0.501661479473114,
      "learning_rate": 8.954041204437402e-06,
      "loss": 0.9534,
      "step": 566
    },
    {
      "epoch": 0.08088445078459344,
      "grad_norm": 0.6546953320503235,
      "learning_rate": 8.96988906497623e-06,
      "loss": 0.9015,
      "step": 567
    },
    {
      "epoch": 0.08102710413694722,
      "grad_norm": 0.601220428943634,
      "learning_rate": 8.985736925515056e-06,
      "loss": 0.8817,
      "step": 568
    },
    {
      "epoch": 0.081169757489301,
      "grad_norm": 0.5301333665847778,
      "learning_rate": 9.001584786053883e-06,
      "loss": 0.9456,
      "step": 569
    },
    {
      "epoch": 0.08131241084165478,
      "grad_norm": 0.457639217376709,
      "learning_rate": 9.01743264659271e-06,
      "loss": 0.9902,
      "step": 570
    },
    {
      "epoch": 0.08145506419400855,
      "grad_norm": 0.5956770777702332,
      "learning_rate": 9.033280507131538e-06,
      "loss": 0.9666,
      "step": 571
    },
    {
      "epoch": 0.08159771754636234,
      "grad_norm": 0.6892085671424866,
      "learning_rate": 9.049128367670366e-06,
      "loss": 0.9116,
      "step": 572
    },
    {
      "epoch": 0.08174037089871612,
      "grad_norm": 0.6338330507278442,
      "learning_rate": 9.064976228209192e-06,
      "loss": 0.9662,
      "step": 573
    },
    {
      "epoch": 0.0818830242510699,
      "grad_norm": 0.5096848011016846,
      "learning_rate": 9.08082408874802e-06,
      "loss": 1.0214,
      "step": 574
    },
    {
      "epoch": 0.08202567760342368,
      "grad_norm": 0.6919476985931396,
      "learning_rate": 9.096671949286847e-06,
      "loss": 0.9159,
      "step": 575
    },
    {
      "epoch": 0.08216833095577745,
      "grad_norm": 0.4773543179035187,
      "learning_rate": 9.112519809825675e-06,
      "loss": 0.9714,
      "step": 576
    },
    {
      "epoch": 0.08231098430813125,
      "grad_norm": 0.6229044198989868,
      "learning_rate": 9.128367670364502e-06,
      "loss": 0.9271,
      "step": 577
    },
    {
      "epoch": 0.08245363766048502,
      "grad_norm": 0.7057109475135803,
      "learning_rate": 9.144215530903328e-06,
      "loss": 0.9824,
      "step": 578
    },
    {
      "epoch": 0.0825962910128388,
      "grad_norm": 0.6915001273155212,
      "learning_rate": 9.160063391442156e-06,
      "loss": 0.8592,
      "step": 579
    },
    {
      "epoch": 0.08273894436519258,
      "grad_norm": 0.5569952130317688,
      "learning_rate": 9.175911251980984e-06,
      "loss": 0.9599,
      "step": 580
    },
    {
      "epoch": 0.08288159771754636,
      "grad_norm": 0.5352381467819214,
      "learning_rate": 9.19175911251981e-06,
      "loss": 0.879,
      "step": 581
    },
    {
      "epoch": 0.08302425106990015,
      "grad_norm": 0.43738898634910583,
      "learning_rate": 9.207606973058639e-06,
      "loss": 0.9443,
      "step": 582
    },
    {
      "epoch": 0.08316690442225393,
      "grad_norm": 1.4304313659667969,
      "learning_rate": 9.223454833597465e-06,
      "loss": 0.7674,
      "step": 583
    },
    {
      "epoch": 0.0833095577746077,
      "grad_norm": 0.5617502331733704,
      "learning_rate": 9.239302694136292e-06,
      "loss": 0.875,
      "step": 584
    },
    {
      "epoch": 0.08345221112696148,
      "grad_norm": 0.6359624266624451,
      "learning_rate": 9.25515055467512e-06,
      "loss": 1.0391,
      "step": 585
    },
    {
      "epoch": 0.08359486447931526,
      "grad_norm": 0.5497215390205383,
      "learning_rate": 9.270998415213946e-06,
      "loss": 0.9309,
      "step": 586
    },
    {
      "epoch": 0.08373751783166905,
      "grad_norm": 0.5523216724395752,
      "learning_rate": 9.286846275752775e-06,
      "loss": 0.9156,
      "step": 587
    },
    {
      "epoch": 0.08388017118402283,
      "grad_norm": 0.5747171640396118,
      "learning_rate": 9.302694136291601e-06,
      "loss": 0.8624,
      "step": 588
    },
    {
      "epoch": 0.0840228245363766,
      "grad_norm": 0.7639681100845337,
      "learning_rate": 9.318541996830429e-06,
      "loss": 0.9966,
      "step": 589
    },
    {
      "epoch": 0.08416547788873038,
      "grad_norm": 0.4723168611526489,
      "learning_rate": 9.334389857369256e-06,
      "loss": 1.0159,
      "step": 590
    },
    {
      "epoch": 0.08430813124108416,
      "grad_norm": 0.46684974431991577,
      "learning_rate": 9.350237717908082e-06,
      "loss": 1.0271,
      "step": 591
    },
    {
      "epoch": 0.08445078459343795,
      "grad_norm": 0.7794578075408936,
      "learning_rate": 9.366085578446912e-06,
      "loss": 0.9784,
      "step": 592
    },
    {
      "epoch": 0.08459343794579173,
      "grad_norm": 0.6551257371902466,
      "learning_rate": 9.381933438985737e-06,
      "loss": 1.0404,
      "step": 593
    },
    {
      "epoch": 0.08473609129814551,
      "grad_norm": 0.41763991117477417,
      "learning_rate": 9.397781299524565e-06,
      "loss": 0.9695,
      "step": 594
    },
    {
      "epoch": 0.08487874465049929,
      "grad_norm": 0.6196808815002441,
      "learning_rate": 9.413629160063393e-06,
      "loss": 0.9565,
      "step": 595
    },
    {
      "epoch": 0.08502139800285306,
      "grad_norm": 0.7163625955581665,
      "learning_rate": 9.429477020602219e-06,
      "loss": 0.9781,
      "step": 596
    },
    {
      "epoch": 0.08516405135520684,
      "grad_norm": 0.5696227550506592,
      "learning_rate": 9.445324881141046e-06,
      "loss": 1.0412,
      "step": 597
    },
    {
      "epoch": 0.08530670470756063,
      "grad_norm": 0.5615867376327515,
      "learning_rate": 9.461172741679874e-06,
      "loss": 1.0089,
      "step": 598
    },
    {
      "epoch": 0.08544935805991441,
      "grad_norm": 0.547003984451294,
      "learning_rate": 9.477020602218701e-06,
      "loss": 0.9197,
      "step": 599
    },
    {
      "epoch": 0.08559201141226819,
      "grad_norm": 0.5029258131980896,
      "learning_rate": 9.492868462757529e-06,
      "loss": 1.0293,
      "step": 600
    },
    {
      "epoch": 0.08573466476462197,
      "grad_norm": 0.5370084047317505,
      "learning_rate": 9.508716323296355e-06,
      "loss": 0.9866,
      "step": 601
    },
    {
      "epoch": 0.08587731811697574,
      "grad_norm": 0.4286113381385803,
      "learning_rate": 9.524564183835183e-06,
      "loss": 0.9768,
      "step": 602
    },
    {
      "epoch": 0.08601997146932953,
      "grad_norm": 0.6612531542778015,
      "learning_rate": 9.54041204437401e-06,
      "loss": 0.8579,
      "step": 603
    },
    {
      "epoch": 0.08616262482168331,
      "grad_norm": 0.5443513989448547,
      "learning_rate": 9.556259904912838e-06,
      "loss": 0.9375,
      "step": 604
    },
    {
      "epoch": 0.08630527817403709,
      "grad_norm": 0.5907576680183411,
      "learning_rate": 9.572107765451665e-06,
      "loss": 0.9235,
      "step": 605
    },
    {
      "epoch": 0.08644793152639087,
      "grad_norm": 0.5967205762863159,
      "learning_rate": 9.587955625990491e-06,
      "loss": 0.9973,
      "step": 606
    },
    {
      "epoch": 0.08659058487874464,
      "grad_norm": 0.8942917585372925,
      "learning_rate": 9.603803486529319e-06,
      "loss": 0.8579,
      "step": 607
    },
    {
      "epoch": 0.08673323823109844,
      "grad_norm": 0.6428115367889404,
      "learning_rate": 9.619651347068147e-06,
      "loss": 0.9871,
      "step": 608
    },
    {
      "epoch": 0.08687589158345221,
      "grad_norm": 0.4508056342601776,
      "learning_rate": 9.635499207606974e-06,
      "loss": 0.9676,
      "step": 609
    },
    {
      "epoch": 0.08701854493580599,
      "grad_norm": 0.7451004981994629,
      "learning_rate": 9.651347068145802e-06,
      "loss": 0.9257,
      "step": 610
    },
    {
      "epoch": 0.08716119828815977,
      "grad_norm": 0.6123510003089905,
      "learning_rate": 9.667194928684628e-06,
      "loss": 0.9135,
      "step": 611
    },
    {
      "epoch": 0.08730385164051355,
      "grad_norm": 0.47386524081230164,
      "learning_rate": 9.683042789223455e-06,
      "loss": 0.9185,
      "step": 612
    },
    {
      "epoch": 0.08744650499286734,
      "grad_norm": 0.6601650714874268,
      "learning_rate": 9.698890649762283e-06,
      "loss": 0.9124,
      "step": 613
    },
    {
      "epoch": 0.08758915834522112,
      "grad_norm": 0.5682607293128967,
      "learning_rate": 9.71473851030111e-06,
      "loss": 0.9254,
      "step": 614
    },
    {
      "epoch": 0.0877318116975749,
      "grad_norm": 0.5714483261108398,
      "learning_rate": 9.730586370839936e-06,
      "loss": 0.9514,
      "step": 615
    },
    {
      "epoch": 0.08787446504992867,
      "grad_norm": 0.7924468517303467,
      "learning_rate": 9.746434231378766e-06,
      "loss": 0.887,
      "step": 616
    },
    {
      "epoch": 0.08801711840228245,
      "grad_norm": 0.7420808672904968,
      "learning_rate": 9.762282091917592e-06,
      "loss": 0.9195,
      "step": 617
    },
    {
      "epoch": 0.08815977175463624,
      "grad_norm": 0.6279465556144714,
      "learning_rate": 9.77812995245642e-06,
      "loss": 0.916,
      "step": 618
    },
    {
      "epoch": 0.08830242510699002,
      "grad_norm": 0.5048345923423767,
      "learning_rate": 9.793977812995247e-06,
      "loss": 0.9049,
      "step": 619
    },
    {
      "epoch": 0.0884450784593438,
      "grad_norm": 0.5561876893043518,
      "learning_rate": 9.809825673534073e-06,
      "loss": 0.8861,
      "step": 620
    },
    {
      "epoch": 0.08858773181169757,
      "grad_norm": 0.598447859287262,
      "learning_rate": 9.825673534072902e-06,
      "loss": 0.9366,
      "step": 621
    },
    {
      "epoch": 0.08873038516405135,
      "grad_norm": 0.6633095741271973,
      "learning_rate": 9.841521394611728e-06,
      "loss": 0.887,
      "step": 622
    },
    {
      "epoch": 0.08887303851640514,
      "grad_norm": 0.7378283739089966,
      "learning_rate": 9.857369255150556e-06,
      "loss": 0.8606,
      "step": 623
    },
    {
      "epoch": 0.08901569186875892,
      "grad_norm": 0.600425660610199,
      "learning_rate": 9.873217115689383e-06,
      "loss": 0.8809,
      "step": 624
    },
    {
      "epoch": 0.0891583452211127,
      "grad_norm": 0.6276618838310242,
      "learning_rate": 9.88906497622821e-06,
      "loss": 0.8641,
      "step": 625
    },
    {
      "epoch": 0.08930099857346648,
      "grad_norm": 0.6094858646392822,
      "learning_rate": 9.904912836767039e-06,
      "loss": 0.9641,
      "step": 626
    },
    {
      "epoch": 0.08944365192582025,
      "grad_norm": 0.5553627610206604,
      "learning_rate": 9.920760697305864e-06,
      "loss": 0.8811,
      "step": 627
    },
    {
      "epoch": 0.08958630527817403,
      "grad_norm": 0.5599763989448547,
      "learning_rate": 9.936608557844692e-06,
      "loss": 0.9504,
      "step": 628
    },
    {
      "epoch": 0.08972895863052782,
      "grad_norm": 0.507016658782959,
      "learning_rate": 9.95245641838352e-06,
      "loss": 0.9171,
      "step": 629
    },
    {
      "epoch": 0.0898716119828816,
      "grad_norm": 0.74651700258255,
      "learning_rate": 9.968304278922346e-06,
      "loss": 0.9215,
      "step": 630
    },
    {
      "epoch": 0.09001426533523538,
      "grad_norm": 0.6789948344230652,
      "learning_rate": 9.984152139461173e-06,
      "loss": 0.9879,
      "step": 631
    },
    {
      "epoch": 0.09015691868758915,
      "grad_norm": 0.6133556962013245,
      "learning_rate": 1e-05,
      "loss": 0.9044,
      "step": 632
    },
    {
      "epoch": 0.09029957203994293,
      "grad_norm": 0.5088140964508057,
      "learning_rate": 9.999999940704466e-06,
      "loss": 0.9207,
      "step": 633
    },
    {
      "epoch": 0.09044222539229672,
      "grad_norm": 0.46542754769325256,
      "learning_rate": 9.999999762817858e-06,
      "loss": 1.0222,
      "step": 634
    },
    {
      "epoch": 0.0905848787446505,
      "grad_norm": 0.7682946920394897,
      "learning_rate": 9.999999466340187e-06,
      "loss": 0.8783,
      "step": 635
    },
    {
      "epoch": 0.09072753209700428,
      "grad_norm": 0.5734198689460754,
      "learning_rate": 9.999999051271458e-06,
      "loss": 0.9113,
      "step": 636
    },
    {
      "epoch": 0.09087018544935806,
      "grad_norm": 0.47362226247787476,
      "learning_rate": 9.999998517611677e-06,
      "loss": 0.9572,
      "step": 637
    },
    {
      "epoch": 0.09101283880171183,
      "grad_norm": 0.6190581321716309,
      "learning_rate": 9.999997865360862e-06,
      "loss": 0.7634,
      "step": 638
    },
    {
      "epoch": 0.09115549215406563,
      "grad_norm": 0.5904908180236816,
      "learning_rate": 9.999997094519025e-06,
      "loss": 0.8617,
      "step": 639
    },
    {
      "epoch": 0.0912981455064194,
      "grad_norm": 0.6041175723075867,
      "learning_rate": 9.999996205086187e-06,
      "loss": 0.9171,
      "step": 640
    },
    {
      "epoch": 0.09144079885877318,
      "grad_norm": 0.575161337852478,
      "learning_rate": 9.999995197062365e-06,
      "loss": 1.0264,
      "step": 641
    },
    {
      "epoch": 0.09158345221112696,
      "grad_norm": 0.5917263627052307,
      "learning_rate": 9.999994070447587e-06,
      "loss": 0.9185,
      "step": 642
    },
    {
      "epoch": 0.09172610556348074,
      "grad_norm": 0.5231802463531494,
      "learning_rate": 9.999992825241878e-06,
      "loss": 0.8656,
      "step": 643
    },
    {
      "epoch": 0.09186875891583453,
      "grad_norm": 0.6748559474945068,
      "learning_rate": 9.999991461445269e-06,
      "loss": 0.8496,
      "step": 644
    },
    {
      "epoch": 0.0920114122681883,
      "grad_norm": 0.5157588720321655,
      "learning_rate": 9.99998997905779e-06,
      "loss": 0.9527,
      "step": 645
    },
    {
      "epoch": 0.09215406562054208,
      "grad_norm": 0.7597405910491943,
      "learning_rate": 9.999988378079475e-06,
      "loss": 0.9636,
      "step": 646
    },
    {
      "epoch": 0.09229671897289586,
      "grad_norm": 0.6512150168418884,
      "learning_rate": 9.999986658510367e-06,
      "loss": 0.9696,
      "step": 647
    },
    {
      "epoch": 0.09243937232524964,
      "grad_norm": 0.5848551392555237,
      "learning_rate": 9.999984820350503e-06,
      "loss": 0.963,
      "step": 648
    },
    {
      "epoch": 0.09258202567760343,
      "grad_norm": 0.6997355818748474,
      "learning_rate": 9.999982863599927e-06,
      "loss": 0.9536,
      "step": 649
    },
    {
      "epoch": 0.09272467902995721,
      "grad_norm": 0.5355712175369263,
      "learning_rate": 9.999980788258687e-06,
      "loss": 0.8946,
      "step": 650
    },
    {
      "epoch": 0.09286733238231099,
      "grad_norm": 0.9124418497085571,
      "learning_rate": 9.999978594326831e-06,
      "loss": 0.7364,
      "step": 651
    },
    {
      "epoch": 0.09300998573466476,
      "grad_norm": 0.6042007803916931,
      "learning_rate": 9.999976281804411e-06,
      "loss": 0.8896,
      "step": 652
    },
    {
      "epoch": 0.09315263908701854,
      "grad_norm": 0.6304923295974731,
      "learning_rate": 9.999973850691483e-06,
      "loss": 0.8782,
      "step": 653
    },
    {
      "epoch": 0.09329529243937233,
      "grad_norm": 0.8566015362739563,
      "learning_rate": 9.9999713009881e-06,
      "loss": 0.8402,
      "step": 654
    },
    {
      "epoch": 0.09343794579172611,
      "grad_norm": 0.6642569303512573,
      "learning_rate": 9.99996863269433e-06,
      "loss": 0.9656,
      "step": 655
    },
    {
      "epoch": 0.09358059914407989,
      "grad_norm": 0.5763037800788879,
      "learning_rate": 9.999965845810232e-06,
      "loss": 0.9491,
      "step": 656
    },
    {
      "epoch": 0.09372325249643366,
      "grad_norm": 0.7415080666542053,
      "learning_rate": 9.999962940335873e-06,
      "loss": 0.9594,
      "step": 657
    },
    {
      "epoch": 0.09386590584878744,
      "grad_norm": 0.5235511064529419,
      "learning_rate": 9.99995991627132e-06,
      "loss": 0.9114,
      "step": 658
    },
    {
      "epoch": 0.09400855920114122,
      "grad_norm": 0.7261329293251038,
      "learning_rate": 9.999956773616647e-06,
      "loss": 0.9594,
      "step": 659
    },
    {
      "epoch": 0.09415121255349501,
      "grad_norm": 0.6547417044639587,
      "learning_rate": 9.999953512371928e-06,
      "loss": 0.8819,
      "step": 660
    },
    {
      "epoch": 0.09429386590584879,
      "grad_norm": 0.7186365723609924,
      "learning_rate": 9.99995013253724e-06,
      "loss": 0.9065,
      "step": 661
    },
    {
      "epoch": 0.09443651925820257,
      "grad_norm": 0.6450744867324829,
      "learning_rate": 9.999946634112662e-06,
      "loss": 0.8491,
      "step": 662
    },
    {
      "epoch": 0.09457917261055634,
      "grad_norm": 0.5700588226318359,
      "learning_rate": 9.99994301709828e-06,
      "loss": 0.8689,
      "step": 663
    },
    {
      "epoch": 0.09472182596291012,
      "grad_norm": 0.6099893450737,
      "learning_rate": 9.999939281494177e-06,
      "loss": 0.8701,
      "step": 664
    },
    {
      "epoch": 0.09486447931526391,
      "grad_norm": 0.4662160575389862,
      "learning_rate": 9.999935427300444e-06,
      "loss": 0.9128,
      "step": 665
    },
    {
      "epoch": 0.09500713266761769,
      "grad_norm": 0.7023544907569885,
      "learning_rate": 9.999931454517168e-06,
      "loss": 0.882,
      "step": 666
    },
    {
      "epoch": 0.09514978601997147,
      "grad_norm": 0.5410977005958557,
      "learning_rate": 9.99992736314445e-06,
      "loss": 0.9134,
      "step": 667
    },
    {
      "epoch": 0.09529243937232525,
      "grad_norm": 0.7053019404411316,
      "learning_rate": 9.999923153182382e-06,
      "loss": 0.8521,
      "step": 668
    },
    {
      "epoch": 0.09543509272467902,
      "grad_norm": 0.4959987998008728,
      "learning_rate": 9.999918824631065e-06,
      "loss": 0.9275,
      "step": 669
    },
    {
      "epoch": 0.09557774607703282,
      "grad_norm": 0.6551489233970642,
      "learning_rate": 9.999914377490601e-06,
      "loss": 0.9987,
      "step": 670
    },
    {
      "epoch": 0.0957203994293866,
      "grad_norm": 0.43556663393974304,
      "learning_rate": 9.999909811761098e-06,
      "loss": 1.0103,
      "step": 671
    },
    {
      "epoch": 0.09586305278174037,
      "grad_norm": 0.5840201377868652,
      "learning_rate": 9.999905127442662e-06,
      "loss": 0.927,
      "step": 672
    },
    {
      "epoch": 0.09600570613409415,
      "grad_norm": 0.6692790985107422,
      "learning_rate": 9.999900324535404e-06,
      "loss": 0.9702,
      "step": 673
    },
    {
      "epoch": 0.09614835948644793,
      "grad_norm": 0.7664057016372681,
      "learning_rate": 9.999895403039439e-06,
      "loss": 0.923,
      "step": 674
    },
    {
      "epoch": 0.09629101283880172,
      "grad_norm": 0.5922096371650696,
      "learning_rate": 9.999890362954883e-06,
      "loss": 0.9974,
      "step": 675
    },
    {
      "epoch": 0.0964336661911555,
      "grad_norm": 0.6685681939125061,
      "learning_rate": 9.999885204281858e-06,
      "loss": 0.8513,
      "step": 676
    },
    {
      "epoch": 0.09657631954350927,
      "grad_norm": 0.7253565192222595,
      "learning_rate": 9.999879927020482e-06,
      "loss": 0.7313,
      "step": 677
    },
    {
      "epoch": 0.09671897289586305,
      "grad_norm": 0.48530927300453186,
      "learning_rate": 9.999874531170883e-06,
      "loss": 0.8826,
      "step": 678
    },
    {
      "epoch": 0.09686162624821683,
      "grad_norm": 0.6055790781974792,
      "learning_rate": 9.99986901673319e-06,
      "loss": 0.9327,
      "step": 679
    },
    {
      "epoch": 0.09700427960057062,
      "grad_norm": 0.6805019974708557,
      "learning_rate": 9.99986338370753e-06,
      "loss": 0.9407,
      "step": 680
    },
    {
      "epoch": 0.0971469329529244,
      "grad_norm": 0.5405174493789673,
      "learning_rate": 9.999857632094039e-06,
      "loss": 0.9735,
      "step": 681
    },
    {
      "epoch": 0.09728958630527818,
      "grad_norm": 0.6433508992195129,
      "learning_rate": 9.999851761892854e-06,
      "loss": 0.955,
      "step": 682
    },
    {
      "epoch": 0.09743223965763195,
      "grad_norm": 0.6980995535850525,
      "learning_rate": 9.999845773104113e-06,
      "loss": 0.7192,
      "step": 683
    },
    {
      "epoch": 0.09757489300998573,
      "grad_norm": 0.5696197152137756,
      "learning_rate": 9.99983966572796e-06,
      "loss": 0.917,
      "step": 684
    },
    {
      "epoch": 0.09771754636233952,
      "grad_norm": 0.7470669150352478,
      "learning_rate": 9.999833439764536e-06,
      "loss": 0.9455,
      "step": 685
    },
    {
      "epoch": 0.0978601997146933,
      "grad_norm": 0.6613849997520447,
      "learning_rate": 9.999827095213994e-06,
      "loss": 0.9069,
      "step": 686
    },
    {
      "epoch": 0.09800285306704708,
      "grad_norm": 0.7304914593696594,
      "learning_rate": 9.999820632076478e-06,
      "loss": 0.8804,
      "step": 687
    },
    {
      "epoch": 0.09814550641940085,
      "grad_norm": 0.47209638357162476,
      "learning_rate": 9.999814050352148e-06,
      "loss": 0.991,
      "step": 688
    },
    {
      "epoch": 0.09828815977175463,
      "grad_norm": 0.5041055083274841,
      "learning_rate": 9.999807350041156e-06,
      "loss": 0.9164,
      "step": 689
    },
    {
      "epoch": 0.09843081312410841,
      "grad_norm": 0.7846484780311584,
      "learning_rate": 9.99980053114366e-06,
      "loss": 0.8503,
      "step": 690
    },
    {
      "epoch": 0.0985734664764622,
      "grad_norm": 0.5664672255516052,
      "learning_rate": 9.999793593659825e-06,
      "loss": 0.9555,
      "step": 691
    },
    {
      "epoch": 0.09871611982881598,
      "grad_norm": 0.5747510194778442,
      "learning_rate": 9.999786537589815e-06,
      "loss": 0.892,
      "step": 692
    },
    {
      "epoch": 0.09885877318116976,
      "grad_norm": 0.7297970056533813,
      "learning_rate": 9.999779362933795e-06,
      "loss": 0.896,
      "step": 693
    },
    {
      "epoch": 0.09900142653352353,
      "grad_norm": 0.7442668676376343,
      "learning_rate": 9.999772069691938e-06,
      "loss": 0.8587,
      "step": 694
    },
    {
      "epoch": 0.09914407988587731,
      "grad_norm": 0.623637318611145,
      "learning_rate": 9.999764657864415e-06,
      "loss": 0.9335,
      "step": 695
    },
    {
      "epoch": 0.0992867332382311,
      "grad_norm": 0.7037668824195862,
      "learning_rate": 9.999757127451402e-06,
      "loss": 0.8621,
      "step": 696
    },
    {
      "epoch": 0.09942938659058488,
      "grad_norm": 0.5125731825828552,
      "learning_rate": 9.999749478453078e-06,
      "loss": 0.9325,
      "step": 697
    },
    {
      "epoch": 0.09957203994293866,
      "grad_norm": 0.8168233036994934,
      "learning_rate": 9.999741710869623e-06,
      "loss": 0.9487,
      "step": 698
    },
    {
      "epoch": 0.09971469329529244,
      "grad_norm": 0.6448724865913391,
      "learning_rate": 9.999733824701226e-06,
      "loss": 0.8703,
      "step": 699
    },
    {
      "epoch": 0.09985734664764621,
      "grad_norm": 0.7140878438949585,
      "learning_rate": 9.999725819948068e-06,
      "loss": 0.8607,
      "step": 700
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6680331230163574,
      "learning_rate": 9.999717696610343e-06,
      "loss": 0.9373,
      "step": 701
    },
    {
      "epoch": 0.10014265335235378,
      "grad_norm": 0.642981767654419,
      "learning_rate": 9.999709454688243e-06,
      "loss": 0.9686,
      "step": 702
    },
    {
      "epoch": 0.10028530670470756,
      "grad_norm": 0.7402138710021973,
      "learning_rate": 9.999701094181962e-06,
      "loss": 0.8575,
      "step": 703
    },
    {
      "epoch": 0.10042796005706134,
      "grad_norm": 0.6488072872161865,
      "learning_rate": 9.999692615091699e-06,
      "loss": 0.892,
      "step": 704
    },
    {
      "epoch": 0.10057061340941512,
      "grad_norm": 0.5455636978149414,
      "learning_rate": 9.999684017417653e-06,
      "loss": 0.8922,
      "step": 705
    },
    {
      "epoch": 0.10071326676176891,
      "grad_norm": 0.6167206764221191,
      "learning_rate": 9.999675301160032e-06,
      "loss": 0.9615,
      "step": 706
    },
    {
      "epoch": 0.10085592011412269,
      "grad_norm": 0.6116539239883423,
      "learning_rate": 9.999666466319041e-06,
      "loss": 0.9432,
      "step": 707
    },
    {
      "epoch": 0.10099857346647646,
      "grad_norm": 0.6285659074783325,
      "learning_rate": 9.99965751289489e-06,
      "loss": 0.8619,
      "step": 708
    },
    {
      "epoch": 0.10114122681883024,
      "grad_norm": 0.664391815662384,
      "learning_rate": 9.999648440887789e-06,
      "loss": 0.8443,
      "step": 709
    },
    {
      "epoch": 0.10128388017118402,
      "grad_norm": 0.7672937512397766,
      "learning_rate": 9.999639250297956e-06,
      "loss": 0.9184,
      "step": 710
    },
    {
      "epoch": 0.10142653352353781,
      "grad_norm": 0.8006978034973145,
      "learning_rate": 9.999629941125608e-06,
      "loss": 0.8876,
      "step": 711
    },
    {
      "epoch": 0.10156918687589159,
      "grad_norm": 0.6475245356559753,
      "learning_rate": 9.999620513370965e-06,
      "loss": 0.9251,
      "step": 712
    },
    {
      "epoch": 0.10171184022824536,
      "grad_norm": 0.720442533493042,
      "learning_rate": 9.999610967034252e-06,
      "loss": 0.9017,
      "step": 713
    },
    {
      "epoch": 0.10185449358059914,
      "grad_norm": 0.6661159992218018,
      "learning_rate": 9.999601302115693e-06,
      "loss": 0.8709,
      "step": 714
    },
    {
      "epoch": 0.10199714693295292,
      "grad_norm": 0.6637430191040039,
      "learning_rate": 9.999591518615521e-06,
      "loss": 0.9481,
      "step": 715
    },
    {
      "epoch": 0.10213980028530671,
      "grad_norm": 0.5911180377006531,
      "learning_rate": 9.999581616533964e-06,
      "loss": 0.9107,
      "step": 716
    },
    {
      "epoch": 0.10228245363766049,
      "grad_norm": 0.6676122546195984,
      "learning_rate": 9.99957159587126e-06,
      "loss": 0.9506,
      "step": 717
    },
    {
      "epoch": 0.10242510699001427,
      "grad_norm": 0.4627430737018585,
      "learning_rate": 9.999561456627645e-06,
      "loss": 0.9372,
      "step": 718
    },
    {
      "epoch": 0.10256776034236804,
      "grad_norm": 0.7318198680877686,
      "learning_rate": 9.999551198803362e-06,
      "loss": 0.8507,
      "step": 719
    },
    {
      "epoch": 0.10271041369472182,
      "grad_norm": 0.8023821115493774,
      "learning_rate": 9.99954082239865e-06,
      "loss": 0.8392,
      "step": 720
    },
    {
      "epoch": 0.1028530670470756,
      "grad_norm": 0.6835203766822815,
      "learning_rate": 9.999530327413757e-06,
      "loss": 0.8794,
      "step": 721
    },
    {
      "epoch": 0.10299572039942939,
      "grad_norm": 0.47203439474105835,
      "learning_rate": 9.999519713848934e-06,
      "loss": 1.0079,
      "step": 722
    },
    {
      "epoch": 0.10313837375178317,
      "grad_norm": 0.5977882742881775,
      "learning_rate": 9.999508981704432e-06,
      "loss": 0.8756,
      "step": 723
    },
    {
      "epoch": 0.10328102710413695,
      "grad_norm": 0.6931073069572449,
      "learning_rate": 9.999498130980502e-06,
      "loss": 0.7312,
      "step": 724
    },
    {
      "epoch": 0.10342368045649072,
      "grad_norm": 0.7802338004112244,
      "learning_rate": 9.999487161677404e-06,
      "loss": 0.8725,
      "step": 725
    },
    {
      "epoch": 0.1035663338088445,
      "grad_norm": 0.9021378755569458,
      "learning_rate": 9.9994760737954e-06,
      "loss": 0.8778,
      "step": 726
    },
    {
      "epoch": 0.1037089871611983,
      "grad_norm": 0.4828585982322693,
      "learning_rate": 9.999464867334751e-06,
      "loss": 0.8971,
      "step": 727
    },
    {
      "epoch": 0.10385164051355207,
      "grad_norm": 0.4677823781967163,
      "learning_rate": 9.99945354229572e-06,
      "loss": 0.9673,
      "step": 728
    },
    {
      "epoch": 0.10399429386590585,
      "grad_norm": 0.4684864580631256,
      "learning_rate": 9.999442098678582e-06,
      "loss": 0.9235,
      "step": 729
    },
    {
      "epoch": 0.10413694721825963,
      "grad_norm": 0.7469426989555359,
      "learning_rate": 9.999430536483603e-06,
      "loss": 0.9443,
      "step": 730
    },
    {
      "epoch": 0.1042796005706134,
      "grad_norm": 0.6579768061637878,
      "learning_rate": 9.999418855711061e-06,
      "loss": 0.9426,
      "step": 731
    },
    {
      "epoch": 0.1044222539229672,
      "grad_norm": 0.5234336853027344,
      "learning_rate": 9.99940705636123e-06,
      "loss": 0.9752,
      "step": 732
    },
    {
      "epoch": 0.10456490727532097,
      "grad_norm": 0.8590401411056519,
      "learning_rate": 9.999395138434392e-06,
      "loss": 0.8162,
      "step": 733
    },
    {
      "epoch": 0.10470756062767475,
      "grad_norm": 0.8299201130867004,
      "learning_rate": 9.999383101930826e-06,
      "loss": 0.8756,
      "step": 734
    },
    {
      "epoch": 0.10485021398002853,
      "grad_norm": 0.7054989337921143,
      "learning_rate": 9.999370946850824e-06,
      "loss": 0.964,
      "step": 735
    },
    {
      "epoch": 0.1049928673323823,
      "grad_norm": 0.8901962637901306,
      "learning_rate": 9.999358673194667e-06,
      "loss": 0.8405,
      "step": 736
    },
    {
      "epoch": 0.1051355206847361,
      "grad_norm": 0.8690089583396912,
      "learning_rate": 9.999346280962652e-06,
      "loss": 0.8672,
      "step": 737
    },
    {
      "epoch": 0.10527817403708987,
      "grad_norm": 0.7922226190567017,
      "learning_rate": 9.999333770155071e-06,
      "loss": 0.8528,
      "step": 738
    },
    {
      "epoch": 0.10542082738944365,
      "grad_norm": 0.5827857255935669,
      "learning_rate": 9.999321140772217e-06,
      "loss": 0.9091,
      "step": 739
    },
    {
      "epoch": 0.10556348074179743,
      "grad_norm": 0.924829363822937,
      "learning_rate": 9.999308392814397e-06,
      "loss": 0.9159,
      "step": 740
    },
    {
      "epoch": 0.10570613409415121,
      "grad_norm": 0.6799506545066833,
      "learning_rate": 9.999295526281906e-06,
      "loss": 0.9056,
      "step": 741
    },
    {
      "epoch": 0.105848787446505,
      "grad_norm": 0.6216866374015808,
      "learning_rate": 9.999282541175056e-06,
      "loss": 0.8959,
      "step": 742
    },
    {
      "epoch": 0.10599144079885878,
      "grad_norm": 0.5728886723518372,
      "learning_rate": 9.999269437494148e-06,
      "loss": 0.9833,
      "step": 743
    },
    {
      "epoch": 0.10613409415121255,
      "grad_norm": 0.7292998433113098,
      "learning_rate": 9.999256215239497e-06,
      "loss": 0.8613,
      "step": 744
    },
    {
      "epoch": 0.10627674750356633,
      "grad_norm": 0.6396505832672119,
      "learning_rate": 9.999242874411416e-06,
      "loss": 0.9325,
      "step": 745
    },
    {
      "epoch": 0.10641940085592011,
      "grad_norm": 0.7370311617851257,
      "learning_rate": 9.999229415010222e-06,
      "loss": 0.8855,
      "step": 746
    },
    {
      "epoch": 0.10656205420827389,
      "grad_norm": 0.5834444761276245,
      "learning_rate": 9.99921583703623e-06,
      "loss": 0.9271,
      "step": 747
    },
    {
      "epoch": 0.10670470756062768,
      "grad_norm": 0.49565738439559937,
      "learning_rate": 9.99920214048977e-06,
      "loss": 0.9065,
      "step": 748
    },
    {
      "epoch": 0.10684736091298146,
      "grad_norm": 0.7068101763725281,
      "learning_rate": 9.99918832537116e-06,
      "loss": 0.8554,
      "step": 749
    },
    {
      "epoch": 0.10699001426533523,
      "grad_norm": 0.881374716758728,
      "learning_rate": 9.99917439168073e-06,
      "loss": 0.9012,
      "step": 750
    },
    {
      "epoch": 0.10713266761768901,
      "grad_norm": 0.6914541721343994,
      "learning_rate": 9.999160339418812e-06,
      "loss": 0.9369,
      "step": 751
    },
    {
      "epoch": 0.10727532097004279,
      "grad_norm": 0.6502618789672852,
      "learning_rate": 9.999146168585736e-06,
      "loss": 0.8951,
      "step": 752
    },
    {
      "epoch": 0.10741797432239658,
      "grad_norm": 0.48643290996551514,
      "learning_rate": 9.999131879181841e-06,
      "loss": 0.886,
      "step": 753
    },
    {
      "epoch": 0.10756062767475036,
      "grad_norm": 0.5962386727333069,
      "learning_rate": 9.999117471207465e-06,
      "loss": 0.8911,
      "step": 754
    },
    {
      "epoch": 0.10770328102710414,
      "grad_norm": 0.4888778626918793,
      "learning_rate": 9.999102944662948e-06,
      "loss": 0.9166,
      "step": 755
    },
    {
      "epoch": 0.10784593437945791,
      "grad_norm": 0.6880698204040527,
      "learning_rate": 9.999088299548636e-06,
      "loss": 0.8551,
      "step": 756
    },
    {
      "epoch": 0.10798858773181169,
      "grad_norm": 0.8405009508132935,
      "learning_rate": 9.999073535864879e-06,
      "loss": 0.8704,
      "step": 757
    },
    {
      "epoch": 0.10813124108416548,
      "grad_norm": 0.6738948225975037,
      "learning_rate": 9.999058653612022e-06,
      "loss": 0.9412,
      "step": 758
    },
    {
      "epoch": 0.10827389443651926,
      "grad_norm": 0.688207745552063,
      "learning_rate": 9.999043652790421e-06,
      "loss": 0.9065,
      "step": 759
    },
    {
      "epoch": 0.10841654778887304,
      "grad_norm": 0.6640740633010864,
      "learning_rate": 9.999028533400432e-06,
      "loss": 0.9356,
      "step": 760
    },
    {
      "epoch": 0.10855920114122682,
      "grad_norm": 0.6320343017578125,
      "learning_rate": 9.999013295442413e-06,
      "loss": 0.8912,
      "step": 761
    },
    {
      "epoch": 0.1087018544935806,
      "grad_norm": 0.8238005042076111,
      "learning_rate": 9.998997938916725e-06,
      "loss": 0.8738,
      "step": 762
    },
    {
      "epoch": 0.10884450784593439,
      "grad_norm": 0.9503881931304932,
      "learning_rate": 9.998982463823734e-06,
      "loss": 0.9157,
      "step": 763
    },
    {
      "epoch": 0.10898716119828816,
      "grad_norm": 0.7548708319664001,
      "learning_rate": 9.998966870163805e-06,
      "loss": 0.8612,
      "step": 764
    },
    {
      "epoch": 0.10912981455064194,
      "grad_norm": 0.8359969854354858,
      "learning_rate": 9.998951157937307e-06,
      "loss": 0.931,
      "step": 765
    },
    {
      "epoch": 0.10927246790299572,
      "grad_norm": 0.6833460330963135,
      "learning_rate": 9.998935327144617e-06,
      "loss": 0.9083,
      "step": 766
    },
    {
      "epoch": 0.1094151212553495,
      "grad_norm": 0.5777496695518494,
      "learning_rate": 9.998919377786105e-06,
      "loss": 0.9209,
      "step": 767
    },
    {
      "epoch": 0.10955777460770329,
      "grad_norm": 0.5251607894897461,
      "learning_rate": 9.998903309862153e-06,
      "loss": 0.9073,
      "step": 768
    },
    {
      "epoch": 0.10970042796005706,
      "grad_norm": 0.6294717788696289,
      "learning_rate": 9.99888712337314e-06,
      "loss": 0.8563,
      "step": 769
    },
    {
      "epoch": 0.10984308131241084,
      "grad_norm": 0.5857126116752625,
      "learning_rate": 9.998870818319453e-06,
      "loss": 0.9388,
      "step": 770
    },
    {
      "epoch": 0.10998573466476462,
      "grad_norm": 0.8108024001121521,
      "learning_rate": 9.998854394701476e-06,
      "loss": 0.9339,
      "step": 771
    },
    {
      "epoch": 0.1101283880171184,
      "grad_norm": 0.5285236835479736,
      "learning_rate": 9.9988378525196e-06,
      "loss": 0.9615,
      "step": 772
    },
    {
      "epoch": 0.11027104136947219,
      "grad_norm": 0.46636226773262024,
      "learning_rate": 9.998821191774215e-06,
      "loss": 0.9249,
      "step": 773
    },
    {
      "epoch": 0.11041369472182597,
      "grad_norm": 0.8498694896697998,
      "learning_rate": 9.998804412465717e-06,
      "loss": 0.8918,
      "step": 774
    },
    {
      "epoch": 0.11055634807417974,
      "grad_norm": 0.5153661370277405,
      "learning_rate": 9.998787514594505e-06,
      "loss": 0.8741,
      "step": 775
    },
    {
      "epoch": 0.11069900142653352,
      "grad_norm": 0.7411801815032959,
      "learning_rate": 9.99877049816098e-06,
      "loss": 0.6312,
      "step": 776
    },
    {
      "epoch": 0.1108416547788873,
      "grad_norm": 0.6182443499565125,
      "learning_rate": 9.998753363165547e-06,
      "loss": 0.9343,
      "step": 777
    },
    {
      "epoch": 0.11098430813124108,
      "grad_norm": 0.9486439228057861,
      "learning_rate": 9.99873610960861e-06,
      "loss": 0.9334,
      "step": 778
    },
    {
      "epoch": 0.11112696148359487,
      "grad_norm": 0.5551772117614746,
      "learning_rate": 9.998718737490576e-06,
      "loss": 0.8739,
      "step": 779
    },
    {
      "epoch": 0.11126961483594865,
      "grad_norm": 0.7792413234710693,
      "learning_rate": 9.998701246811863e-06,
      "loss": 0.8978,
      "step": 780
    },
    {
      "epoch": 0.11141226818830242,
      "grad_norm": 0.6563642024993896,
      "learning_rate": 9.998683637572882e-06,
      "loss": 0.8819,
      "step": 781
    },
    {
      "epoch": 0.1115549215406562,
      "grad_norm": 0.7804805040359497,
      "learning_rate": 9.99866590977405e-06,
      "loss": 0.8101,
      "step": 782
    },
    {
      "epoch": 0.11169757489300998,
      "grad_norm": 0.4641774594783783,
      "learning_rate": 9.998648063415787e-06,
      "loss": 0.9002,
      "step": 783
    },
    {
      "epoch": 0.11184022824536377,
      "grad_norm": 0.5388142466545105,
      "learning_rate": 9.998630098498522e-06,
      "loss": 0.9923,
      "step": 784
    },
    {
      "epoch": 0.11198288159771755,
      "grad_norm": 0.752376139163971,
      "learning_rate": 9.998612015022675e-06,
      "loss": 0.9021,
      "step": 785
    },
    {
      "epoch": 0.11212553495007133,
      "grad_norm": 0.781582236289978,
      "learning_rate": 9.99859381298868e-06,
      "loss": 0.8599,
      "step": 786
    },
    {
      "epoch": 0.1122681883024251,
      "grad_norm": 0.569164514541626,
      "learning_rate": 9.998575492396962e-06,
      "loss": 0.9035,
      "step": 787
    },
    {
      "epoch": 0.11241084165477888,
      "grad_norm": 0.6883033514022827,
      "learning_rate": 9.99855705324796e-06,
      "loss": 1.0016,
      "step": 788
    },
    {
      "epoch": 0.11255349500713267,
      "grad_norm": 0.5926066637039185,
      "learning_rate": 9.998538495542111e-06,
      "loss": 0.8653,
      "step": 789
    },
    {
      "epoch": 0.11269614835948645,
      "grad_norm": 0.6186153888702393,
      "learning_rate": 9.998519819279855e-06,
      "loss": 0.9434,
      "step": 790
    },
    {
      "epoch": 0.11283880171184023,
      "grad_norm": 0.877582848072052,
      "learning_rate": 9.998501024461636e-06,
      "loss": 0.9036,
      "step": 791
    },
    {
      "epoch": 0.112981455064194,
      "grad_norm": 0.6416645050048828,
      "learning_rate": 9.998482111087897e-06,
      "loss": 0.6692,
      "step": 792
    },
    {
      "epoch": 0.11312410841654778,
      "grad_norm": 0.7448471188545227,
      "learning_rate": 9.99846307915909e-06,
      "loss": 0.9356,
      "step": 793
    },
    {
      "epoch": 0.11326676176890157,
      "grad_norm": 0.6980072259902954,
      "learning_rate": 9.998443928675661e-06,
      "loss": 0.8437,
      "step": 794
    },
    {
      "epoch": 0.11340941512125535,
      "grad_norm": 0.4740276038646698,
      "learning_rate": 9.998424659638071e-06,
      "loss": 0.8907,
      "step": 795
    },
    {
      "epoch": 0.11355206847360913,
      "grad_norm": 0.7384071350097656,
      "learning_rate": 9.998405272046772e-06,
      "loss": 0.9034,
      "step": 796
    },
    {
      "epoch": 0.11369472182596291,
      "grad_norm": 0.7091321349143982,
      "learning_rate": 9.998385765902225e-06,
      "loss": 0.8201,
      "step": 797
    },
    {
      "epoch": 0.11383737517831669,
      "grad_norm": 0.6403769850730896,
      "learning_rate": 9.998366141204896e-06,
      "loss": 0.9436,
      "step": 798
    },
    {
      "epoch": 0.11398002853067048,
      "grad_norm": 0.6469430923461914,
      "learning_rate": 9.998346397955245e-06,
      "loss": 0.8494,
      "step": 799
    },
    {
      "epoch": 0.11412268188302425,
      "grad_norm": 0.8421128392219543,
      "learning_rate": 9.998326536153746e-06,
      "loss": 0.8924,
      "step": 800
    },
    {
      "epoch": 0.11426533523537803,
      "grad_norm": 0.9186739325523376,
      "learning_rate": 9.998306555800865e-06,
      "loss": 0.8441,
      "step": 801
    },
    {
      "epoch": 0.11440798858773181,
      "grad_norm": 0.8992595672607422,
      "learning_rate": 9.998286456897077e-06,
      "loss": 0.8433,
      "step": 802
    },
    {
      "epoch": 0.11455064194008559,
      "grad_norm": 0.7035727500915527,
      "learning_rate": 9.998266239442862e-06,
      "loss": 0.8591,
      "step": 803
    },
    {
      "epoch": 0.11469329529243938,
      "grad_norm": 0.7800397276878357,
      "learning_rate": 9.998245903438697e-06,
      "loss": 0.8754,
      "step": 804
    },
    {
      "epoch": 0.11483594864479316,
      "grad_norm": 0.8508110642433167,
      "learning_rate": 9.998225448885063e-06,
      "loss": 0.9149,
      "step": 805
    },
    {
      "epoch": 0.11497860199714693,
      "grad_norm": 0.6382063031196594,
      "learning_rate": 9.998204875782448e-06,
      "loss": 0.9411,
      "step": 806
    },
    {
      "epoch": 0.11512125534950071,
      "grad_norm": 0.6287840008735657,
      "learning_rate": 9.998184184131339e-06,
      "loss": 0.9203,
      "step": 807
    },
    {
      "epoch": 0.11526390870185449,
      "grad_norm": 0.7639631628990173,
      "learning_rate": 9.998163373932224e-06,
      "loss": 0.886,
      "step": 808
    },
    {
      "epoch": 0.11540656205420827,
      "grad_norm": 0.6414012908935547,
      "learning_rate": 9.9981424451856e-06,
      "loss": 0.9347,
      "step": 809
    },
    {
      "epoch": 0.11554921540656206,
      "grad_norm": 0.5841327905654907,
      "learning_rate": 9.998121397891964e-06,
      "loss": 0.8779,
      "step": 810
    },
    {
      "epoch": 0.11569186875891584,
      "grad_norm": 0.6971140503883362,
      "learning_rate": 9.998100232051813e-06,
      "loss": 0.8971,
      "step": 811
    },
    {
      "epoch": 0.11583452211126961,
      "grad_norm": 0.5263389348983765,
      "learning_rate": 9.998078947665648e-06,
      "loss": 0.9519,
      "step": 812
    },
    {
      "epoch": 0.11597717546362339,
      "grad_norm": 0.5685816407203674,
      "learning_rate": 9.998057544733979e-06,
      "loss": 0.9713,
      "step": 813
    },
    {
      "epoch": 0.11611982881597717,
      "grad_norm": 0.6715201139450073,
      "learning_rate": 9.998036023257307e-06,
      "loss": 0.887,
      "step": 814
    },
    {
      "epoch": 0.11626248216833096,
      "grad_norm": 0.7427307963371277,
      "learning_rate": 9.998014383236147e-06,
      "loss": 0.862,
      "step": 815
    },
    {
      "epoch": 0.11640513552068474,
      "grad_norm": 0.8541010618209839,
      "learning_rate": 9.99799262467101e-06,
      "loss": 0.9105,
      "step": 816
    },
    {
      "epoch": 0.11654778887303852,
      "grad_norm": 0.7574614882469177,
      "learning_rate": 9.997970747562413e-06,
      "loss": 0.865,
      "step": 817
    },
    {
      "epoch": 0.1166904422253923,
      "grad_norm": 0.6815717220306396,
      "learning_rate": 9.997948751910875e-06,
      "loss": 0.8968,
      "step": 818
    },
    {
      "epoch": 0.11683309557774607,
      "grad_norm": 0.8007264137268066,
      "learning_rate": 9.997926637716917e-06,
      "loss": 0.9665,
      "step": 819
    },
    {
      "epoch": 0.11697574893009986,
      "grad_norm": 0.8848460912704468,
      "learning_rate": 9.997904404981064e-06,
      "loss": 0.885,
      "step": 820
    },
    {
      "epoch": 0.11711840228245364,
      "grad_norm": 0.8217048645019531,
      "learning_rate": 9.997882053703844e-06,
      "loss": 0.8322,
      "step": 821
    },
    {
      "epoch": 0.11726105563480742,
      "grad_norm": 0.6951943039894104,
      "learning_rate": 9.997859583885786e-06,
      "loss": 0.952,
      "step": 822
    },
    {
      "epoch": 0.1174037089871612,
      "grad_norm": 0.7835627198219299,
      "learning_rate": 9.997836995527423e-06,
      "loss": 0.9116,
      "step": 823
    },
    {
      "epoch": 0.11754636233951497,
      "grad_norm": 0.8103483319282532,
      "learning_rate": 9.997814288629293e-06,
      "loss": 0.8805,
      "step": 824
    },
    {
      "epoch": 0.11768901569186876,
      "grad_norm": 0.6608843207359314,
      "learning_rate": 9.99779146319193e-06,
      "loss": 0.9667,
      "step": 825
    },
    {
      "epoch": 0.11783166904422254,
      "grad_norm": 0.7242885231971741,
      "learning_rate": 9.99776851921588e-06,
      "loss": 0.9672,
      "step": 826
    },
    {
      "epoch": 0.11797432239657632,
      "grad_norm": 1.1274234056472778,
      "learning_rate": 9.997745456701684e-06,
      "loss": 0.9623,
      "step": 827
    },
    {
      "epoch": 0.1181169757489301,
      "grad_norm": 0.7334691882133484,
      "learning_rate": 9.99772227564989e-06,
      "loss": 0.905,
      "step": 828
    },
    {
      "epoch": 0.11825962910128388,
      "grad_norm": 0.7089143991470337,
      "learning_rate": 9.99769897606105e-06,
      "loss": 0.8703,
      "step": 829
    },
    {
      "epoch": 0.11840228245363767,
      "grad_norm": 0.8237111568450928,
      "learning_rate": 9.997675557935714e-06,
      "loss": 0.9014,
      "step": 830
    },
    {
      "epoch": 0.11854493580599144,
      "grad_norm": 0.5905559659004211,
      "learning_rate": 9.997652021274438e-06,
      "loss": 0.9067,
      "step": 831
    },
    {
      "epoch": 0.11868758915834522,
      "grad_norm": 0.6693427562713623,
      "learning_rate": 9.997628366077781e-06,
      "loss": 0.884,
      "step": 832
    },
    {
      "epoch": 0.118830242510699,
      "grad_norm": 0.6466344594955444,
      "learning_rate": 9.997604592346301e-06,
      "loss": 0.9198,
      "step": 833
    },
    {
      "epoch": 0.11897289586305278,
      "grad_norm": 0.8904848694801331,
      "learning_rate": 9.997580700080567e-06,
      "loss": 0.8998,
      "step": 834
    },
    {
      "epoch": 0.11911554921540657,
      "grad_norm": 0.6356571912765503,
      "learning_rate": 9.997556689281142e-06,
      "loss": 0.9558,
      "step": 835
    },
    {
      "epoch": 0.11925820256776035,
      "grad_norm": 0.8112600445747375,
      "learning_rate": 9.997532559948597e-06,
      "loss": 0.819,
      "step": 836
    },
    {
      "epoch": 0.11940085592011412,
      "grad_norm": 0.5827550292015076,
      "learning_rate": 9.997508312083503e-06,
      "loss": 0.9217,
      "step": 837
    },
    {
      "epoch": 0.1195435092724679,
      "grad_norm": 0.6937353014945984,
      "learning_rate": 9.997483945686435e-06,
      "loss": 0.8083,
      "step": 838
    },
    {
      "epoch": 0.11968616262482168,
      "grad_norm": 0.8962831497192383,
      "learning_rate": 9.997459460757973e-06,
      "loss": 0.866,
      "step": 839
    },
    {
      "epoch": 0.11982881597717546,
      "grad_norm": 0.5882939696311951,
      "learning_rate": 9.997434857298697e-06,
      "loss": 0.8634,
      "step": 840
    },
    {
      "epoch": 0.11997146932952925,
      "grad_norm": 0.9035172462463379,
      "learning_rate": 9.99741013530919e-06,
      "loss": 0.8731,
      "step": 841
    },
    {
      "epoch": 0.12011412268188303,
      "grad_norm": 0.6360054016113281,
      "learning_rate": 9.997385294790038e-06,
      "loss": 0.9279,
      "step": 842
    },
    {
      "epoch": 0.1202567760342368,
      "grad_norm": 0.6242084503173828,
      "learning_rate": 9.99736033574183e-06,
      "loss": 0.9538,
      "step": 843
    },
    {
      "epoch": 0.12039942938659058,
      "grad_norm": 0.7019553780555725,
      "learning_rate": 9.99733525816516e-06,
      "loss": 0.8733,
      "step": 844
    },
    {
      "epoch": 0.12054208273894436,
      "grad_norm": 0.6724885702133179,
      "learning_rate": 9.99731006206062e-06,
      "loss": 0.8644,
      "step": 845
    },
    {
      "epoch": 0.12068473609129815,
      "grad_norm": 0.6924759745597839,
      "learning_rate": 9.997284747428811e-06,
      "loss": 0.9115,
      "step": 846
    },
    {
      "epoch": 0.12082738944365193,
      "grad_norm": 0.7439209818840027,
      "learning_rate": 9.997259314270331e-06,
      "loss": 0.8206,
      "step": 847
    },
    {
      "epoch": 0.1209700427960057,
      "grad_norm": 0.8042154312133789,
      "learning_rate": 9.997233762585783e-06,
      "loss": 0.8647,
      "step": 848
    },
    {
      "epoch": 0.12111269614835948,
      "grad_norm": 0.7187075614929199,
      "learning_rate": 9.997208092375773e-06,
      "loss": 0.8607,
      "step": 849
    },
    {
      "epoch": 0.12125534950071326,
      "grad_norm": 0.8251458406448364,
      "learning_rate": 9.997182303640912e-06,
      "loss": 0.8745,
      "step": 850
    },
    {
      "epoch": 0.12139800285306705,
      "grad_norm": 0.8559502959251404,
      "learning_rate": 9.99715639638181e-06,
      "loss": 0.8801,
      "step": 851
    },
    {
      "epoch": 0.12154065620542083,
      "grad_norm": 0.7346495985984802,
      "learning_rate": 9.997130370599081e-06,
      "loss": 0.8799,
      "step": 852
    },
    {
      "epoch": 0.12168330955777461,
      "grad_norm": 0.5646926760673523,
      "learning_rate": 9.997104226293343e-06,
      "loss": 0.897,
      "step": 853
    },
    {
      "epoch": 0.12182596291012839,
      "grad_norm": 0.7479763031005859,
      "learning_rate": 9.997077963465219e-06,
      "loss": 0.8199,
      "step": 854
    },
    {
      "epoch": 0.12196861626248216,
      "grad_norm": 0.5906651616096497,
      "learning_rate": 9.997051582115325e-06,
      "loss": 0.8739,
      "step": 855
    },
    {
      "epoch": 0.12211126961483595,
      "grad_norm": 1.1264022588729858,
      "learning_rate": 9.997025082244291e-06,
      "loss": 0.9145,
      "step": 856
    },
    {
      "epoch": 0.12225392296718973,
      "grad_norm": 0.6673288941383362,
      "learning_rate": 9.996998463852747e-06,
      "loss": 0.9036,
      "step": 857
    },
    {
      "epoch": 0.12239657631954351,
      "grad_norm": 0.6579235792160034,
      "learning_rate": 9.996971726941322e-06,
      "loss": 0.9104,
      "step": 858
    },
    {
      "epoch": 0.12253922967189729,
      "grad_norm": 0.8649102449417114,
      "learning_rate": 9.996944871510652e-06,
      "loss": 0.9032,
      "step": 859
    },
    {
      "epoch": 0.12268188302425106,
      "grad_norm": 0.6878856420516968,
      "learning_rate": 9.996917897561372e-06,
      "loss": 0.8311,
      "step": 860
    },
    {
      "epoch": 0.12282453637660486,
      "grad_norm": 0.756231963634491,
      "learning_rate": 9.996890805094122e-06,
      "loss": 0.8804,
      "step": 861
    },
    {
      "epoch": 0.12296718972895863,
      "grad_norm": 0.7084113359451294,
      "learning_rate": 9.996863594109544e-06,
      "loss": 0.9322,
      "step": 862
    },
    {
      "epoch": 0.12310984308131241,
      "grad_norm": 0.517266035079956,
      "learning_rate": 9.996836264608287e-06,
      "loss": 0.8073,
      "step": 863
    },
    {
      "epoch": 0.12325249643366619,
      "grad_norm": 0.8390178084373474,
      "learning_rate": 9.996808816590996e-06,
      "loss": 0.698,
      "step": 864
    },
    {
      "epoch": 0.12339514978601997,
      "grad_norm": 0.865898072719574,
      "learning_rate": 9.996781250058321e-06,
      "loss": 0.8309,
      "step": 865
    },
    {
      "epoch": 0.12353780313837376,
      "grad_norm": 0.654003381729126,
      "learning_rate": 9.996753565010919e-06,
      "loss": 0.9114,
      "step": 866
    },
    {
      "epoch": 0.12368045649072754,
      "grad_norm": 0.7689074873924255,
      "learning_rate": 9.996725761449443e-06,
      "loss": 0.9499,
      "step": 867
    },
    {
      "epoch": 0.12382310984308131,
      "grad_norm": 0.7274837493896484,
      "learning_rate": 9.996697839374559e-06,
      "loss": 0.8318,
      "step": 868
    },
    {
      "epoch": 0.12396576319543509,
      "grad_norm": 0.6424981355667114,
      "learning_rate": 9.996669798786922e-06,
      "loss": 0.8216,
      "step": 869
    },
    {
      "epoch": 0.12410841654778887,
      "grad_norm": 0.7490353584289551,
      "learning_rate": 9.996641639687199e-06,
      "loss": 0.8917,
      "step": 870
    },
    {
      "epoch": 0.12425106990014265,
      "grad_norm": 0.544717013835907,
      "learning_rate": 9.99661336207606e-06,
      "loss": 0.9394,
      "step": 871
    },
    {
      "epoch": 0.12439372325249644,
      "grad_norm": 0.6359512805938721,
      "learning_rate": 9.996584965954176e-06,
      "loss": 0.8508,
      "step": 872
    },
    {
      "epoch": 0.12453637660485022,
      "grad_norm": 0.6323925256729126,
      "learning_rate": 9.996556451322217e-06,
      "loss": 0.8678,
      "step": 873
    },
    {
      "epoch": 0.124679029957204,
      "grad_norm": 0.6573577523231506,
      "learning_rate": 9.99652781818086e-06,
      "loss": 0.8533,
      "step": 874
    },
    {
      "epoch": 0.12482168330955777,
      "grad_norm": 0.6162577271461487,
      "learning_rate": 9.99649906653079e-06,
      "loss": 0.9112,
      "step": 875
    },
    {
      "epoch": 0.12496433666191155,
      "grad_norm": 1.4229447841644287,
      "learning_rate": 9.99647019637268e-06,
      "loss": 0.6754,
      "step": 876
    },
    {
      "epoch": 0.12510699001426534,
      "grad_norm": 0.7424920797348022,
      "learning_rate": 9.996441207707222e-06,
      "loss": 0.8453,
      "step": 877
    },
    {
      "epoch": 0.12524964336661912,
      "grad_norm": 0.7387217879295349,
      "learning_rate": 9.9964121005351e-06,
      "loss": 0.8343,
      "step": 878
    },
    {
      "epoch": 0.1253922967189729,
      "grad_norm": 0.9214134812355042,
      "learning_rate": 9.996382874857004e-06,
      "loss": 0.8657,
      "step": 879
    },
    {
      "epoch": 0.12553495007132667,
      "grad_norm": 0.5441204309463501,
      "learning_rate": 9.99635353067363e-06,
      "loss": 0.9288,
      "step": 880
    },
    {
      "epoch": 0.12567760342368045,
      "grad_norm": 0.7430056929588318,
      "learning_rate": 9.996324067985672e-06,
      "loss": 0.849,
      "step": 881
    },
    {
      "epoch": 0.12582025677603423,
      "grad_norm": 0.9778558611869812,
      "learning_rate": 9.996294486793827e-06,
      "loss": 0.8932,
      "step": 882
    },
    {
      "epoch": 0.125962910128388,
      "grad_norm": 0.7526888847351074,
      "learning_rate": 9.996264787098801e-06,
      "loss": 0.8981,
      "step": 883
    },
    {
      "epoch": 0.12610556348074178,
      "grad_norm": 0.7484474182128906,
      "learning_rate": 9.996234968901295e-06,
      "loss": 0.9127,
      "step": 884
    },
    {
      "epoch": 0.1262482168330956,
      "grad_norm": 0.6596251726150513,
      "learning_rate": 9.996205032202017e-06,
      "loss": 0.8431,
      "step": 885
    },
    {
      "epoch": 0.12639087018544937,
      "grad_norm": 0.9939471483230591,
      "learning_rate": 9.996174977001677e-06,
      "loss": 0.914,
      "step": 886
    },
    {
      "epoch": 0.12653352353780314,
      "grad_norm": 0.7965244054794312,
      "learning_rate": 9.996144803300991e-06,
      "loss": 0.8923,
      "step": 887
    },
    {
      "epoch": 0.12667617689015692,
      "grad_norm": 0.807079017162323,
      "learning_rate": 9.99611451110067e-06,
      "loss": 0.827,
      "step": 888
    },
    {
      "epoch": 0.1268188302425107,
      "grad_norm": 0.5425110459327698,
      "learning_rate": 9.996084100401435e-06,
      "loss": 0.9275,
      "step": 889
    },
    {
      "epoch": 0.12696148359486448,
      "grad_norm": 0.5462783575057983,
      "learning_rate": 9.996053571204005e-06,
      "loss": 0.882,
      "step": 890
    },
    {
      "epoch": 0.12710413694721825,
      "grad_norm": 0.7539640665054321,
      "learning_rate": 9.996022923509108e-06,
      "loss": 0.8529,
      "step": 891
    },
    {
      "epoch": 0.12724679029957203,
      "grad_norm": 0.5665767788887024,
      "learning_rate": 9.995992157317466e-06,
      "loss": 0.9015,
      "step": 892
    },
    {
      "epoch": 0.1273894436519258,
      "grad_norm": 0.8361479640007019,
      "learning_rate": 9.995961272629815e-06,
      "loss": 0.8317,
      "step": 893
    },
    {
      "epoch": 0.1275320970042796,
      "grad_norm": 0.8227171897888184,
      "learning_rate": 9.99593026944688e-06,
      "loss": 0.7751,
      "step": 894
    },
    {
      "epoch": 0.1276747503566334,
      "grad_norm": 0.5182986259460449,
      "learning_rate": 9.995899147769403e-06,
      "loss": 0.9356,
      "step": 895
    },
    {
      "epoch": 0.12781740370898717,
      "grad_norm": 1.1745651960372925,
      "learning_rate": 9.995867907598119e-06,
      "loss": 0.7538,
      "step": 896
    },
    {
      "epoch": 0.12796005706134095,
      "grad_norm": 0.7005908489227295,
      "learning_rate": 9.995836548933768e-06,
      "loss": 0.8818,
      "step": 897
    },
    {
      "epoch": 0.12810271041369473,
      "grad_norm": 0.7603804469108582,
      "learning_rate": 9.995805071777096e-06,
      "loss": 0.8509,
      "step": 898
    },
    {
      "epoch": 0.1282453637660485,
      "grad_norm": 0.8213330507278442,
      "learning_rate": 9.995773476128849e-06,
      "loss": 0.8355,
      "step": 899
    },
    {
      "epoch": 0.12838801711840228,
      "grad_norm": 0.7150991559028625,
      "learning_rate": 9.995741761989778e-06,
      "loss": 0.8585,
      "step": 900
    },
    {
      "epoch": 0.12853067047075606,
      "grad_norm": 0.7070020437240601,
      "learning_rate": 9.995709929360629e-06,
      "loss": 0.9078,
      "step": 901
    },
    {
      "epoch": 0.12867332382310984,
      "grad_norm": 0.715682327747345,
      "learning_rate": 9.995677978242164e-06,
      "loss": 0.9014,
      "step": 902
    },
    {
      "epoch": 0.12881597717546361,
      "grad_norm": 0.6890527606010437,
      "learning_rate": 9.995645908635139e-06,
      "loss": 0.8958,
      "step": 903
    },
    {
      "epoch": 0.1289586305278174,
      "grad_norm": 0.6184905767440796,
      "learning_rate": 9.995613720540312e-06,
      "loss": 0.9359,
      "step": 904
    },
    {
      "epoch": 0.1291012838801712,
      "grad_norm": 0.7181702852249146,
      "learning_rate": 9.995581413958448e-06,
      "loss": 0.9108,
      "step": 905
    },
    {
      "epoch": 0.12924393723252497,
      "grad_norm": 0.7546802163124084,
      "learning_rate": 9.995548988890316e-06,
      "loss": 0.8612,
      "step": 906
    },
    {
      "epoch": 0.12938659058487875,
      "grad_norm": 0.6117148995399475,
      "learning_rate": 9.99551644533668e-06,
      "loss": 0.8909,
      "step": 907
    },
    {
      "epoch": 0.12952924393723253,
      "grad_norm": 0.7722803950309753,
      "learning_rate": 9.995483783298315e-06,
      "loss": 0.9538,
      "step": 908
    },
    {
      "epoch": 0.1296718972895863,
      "grad_norm": 0.9300944805145264,
      "learning_rate": 9.995451002775995e-06,
      "loss": 0.8414,
      "step": 909
    },
    {
      "epoch": 0.12981455064194009,
      "grad_norm": 0.7007274031639099,
      "learning_rate": 9.9954181037705e-06,
      "loss": 0.8163,
      "step": 910
    },
    {
      "epoch": 0.12995720399429386,
      "grad_norm": 0.7200602889060974,
      "learning_rate": 9.995385086282604e-06,
      "loss": 0.8925,
      "step": 911
    },
    {
      "epoch": 0.13009985734664764,
      "grad_norm": 0.6431737542152405,
      "learning_rate": 9.995351950313098e-06,
      "loss": 0.8477,
      "step": 912
    },
    {
      "epoch": 0.13024251069900142,
      "grad_norm": 0.7531431317329407,
      "learning_rate": 9.995318695862762e-06,
      "loss": 0.8907,
      "step": 913
    },
    {
      "epoch": 0.1303851640513552,
      "grad_norm": 0.7862675189971924,
      "learning_rate": 9.995285322932385e-06,
      "loss": 0.8865,
      "step": 914
    },
    {
      "epoch": 0.13052781740370897,
      "grad_norm": 0.8877861499786377,
      "learning_rate": 9.995251831522763e-06,
      "loss": 0.8969,
      "step": 915
    },
    {
      "epoch": 0.13067047075606278,
      "grad_norm": 0.5905489325523376,
      "learning_rate": 9.995218221634686e-06,
      "loss": 0.8297,
      "step": 916
    },
    {
      "epoch": 0.13081312410841656,
      "grad_norm": 0.9023591876029968,
      "learning_rate": 9.995184493268954e-06,
      "loss": 0.8982,
      "step": 917
    },
    {
      "epoch": 0.13095577746077033,
      "grad_norm": 0.7605781555175781,
      "learning_rate": 9.995150646426364e-06,
      "loss": 0.8962,
      "step": 918
    },
    {
      "epoch": 0.1310984308131241,
      "grad_norm": 0.7599273920059204,
      "learning_rate": 9.995116681107721e-06,
      "loss": 0.8883,
      "step": 919
    },
    {
      "epoch": 0.1312410841654779,
      "grad_norm": 0.7208825945854187,
      "learning_rate": 9.995082597313832e-06,
      "loss": 0.8444,
      "step": 920
    },
    {
      "epoch": 0.13138373751783167,
      "grad_norm": 1.0031217336654663,
      "learning_rate": 9.995048395045502e-06,
      "loss": 0.9012,
      "step": 921
    },
    {
      "epoch": 0.13152639087018544,
      "grad_norm": 0.8403066992759705,
      "learning_rate": 9.995014074303544e-06,
      "loss": 0.8789,
      "step": 922
    },
    {
      "epoch": 0.13166904422253922,
      "grad_norm": 0.7000038027763367,
      "learning_rate": 9.994979635088772e-06,
      "loss": 0.9594,
      "step": 923
    },
    {
      "epoch": 0.131811697574893,
      "grad_norm": 0.6286900043487549,
      "learning_rate": 9.994945077402002e-06,
      "loss": 0.866,
      "step": 924
    },
    {
      "epoch": 0.13195435092724678,
      "grad_norm": 0.7327531576156616,
      "learning_rate": 9.994910401244056e-06,
      "loss": 0.8621,
      "step": 925
    },
    {
      "epoch": 0.13209700427960058,
      "grad_norm": 0.7145300507545471,
      "learning_rate": 9.994875606615752e-06,
      "loss": 0.9447,
      "step": 926
    },
    {
      "epoch": 0.13223965763195436,
      "grad_norm": 0.7216164469718933,
      "learning_rate": 9.99484069351792e-06,
      "loss": 0.8226,
      "step": 927
    },
    {
      "epoch": 0.13238231098430814,
      "grad_norm": 0.6584069728851318,
      "learning_rate": 9.994805661951386e-06,
      "loss": 0.9572,
      "step": 928
    },
    {
      "epoch": 0.13252496433666192,
      "grad_norm": 0.6539363265037537,
      "learning_rate": 9.99477051191698e-06,
      "loss": 0.8985,
      "step": 929
    },
    {
      "epoch": 0.1326676176890157,
      "grad_norm": 0.9867337346076965,
      "learning_rate": 9.994735243415536e-06,
      "loss": 0.8692,
      "step": 930
    },
    {
      "epoch": 0.13281027104136947,
      "grad_norm": 0.7150183320045471,
      "learning_rate": 9.994699856447893e-06,
      "loss": 0.8236,
      "step": 931
    },
    {
      "epoch": 0.13295292439372325,
      "grad_norm": 0.6724671125411987,
      "learning_rate": 9.994664351014886e-06,
      "loss": 0.8764,
      "step": 932
    },
    {
      "epoch": 0.13309557774607703,
      "grad_norm": 0.7004656791687012,
      "learning_rate": 9.994628727117362e-06,
      "loss": 0.87,
      "step": 933
    },
    {
      "epoch": 0.1332382310984308,
      "grad_norm": 0.8211258053779602,
      "learning_rate": 9.994592984756163e-06,
      "loss": 0.8604,
      "step": 934
    },
    {
      "epoch": 0.13338088445078458,
      "grad_norm": 0.8809522390365601,
      "learning_rate": 9.994557123932138e-06,
      "loss": 0.9265,
      "step": 935
    },
    {
      "epoch": 0.1335235378031384,
      "grad_norm": 0.7160531282424927,
      "learning_rate": 9.994521144646136e-06,
      "loss": 0.835,
      "step": 936
    },
    {
      "epoch": 0.13366619115549216,
      "grad_norm": 0.5444583296775818,
      "learning_rate": 9.99448504689901e-06,
      "loss": 0.8967,
      "step": 937
    },
    {
      "epoch": 0.13380884450784594,
      "grad_norm": 0.8648520112037659,
      "learning_rate": 9.994448830691618e-06,
      "loss": 0.8299,
      "step": 938
    },
    {
      "epoch": 0.13395149786019972,
      "grad_norm": 0.7733383178710938,
      "learning_rate": 9.99441249602482e-06,
      "loss": 0.9028,
      "step": 939
    },
    {
      "epoch": 0.1340941512125535,
      "grad_norm": 0.7461035251617432,
      "learning_rate": 9.994376042899474e-06,
      "loss": 0.9103,
      "step": 940
    },
    {
      "epoch": 0.13423680456490727,
      "grad_norm": 0.6235612630844116,
      "learning_rate": 9.994339471316447e-06,
      "loss": 0.8323,
      "step": 941
    },
    {
      "epoch": 0.13437945791726105,
      "grad_norm": 0.6510120034217834,
      "learning_rate": 9.994302781276608e-06,
      "loss": 0.9043,
      "step": 942
    },
    {
      "epoch": 0.13452211126961483,
      "grad_norm": 0.6366427540779114,
      "learning_rate": 9.994265972780824e-06,
      "loss": 0.8177,
      "step": 943
    },
    {
      "epoch": 0.1346647646219686,
      "grad_norm": 0.8007756471633911,
      "learning_rate": 9.994229045829969e-06,
      "loss": 0.8732,
      "step": 944
    },
    {
      "epoch": 0.13480741797432239,
      "grad_norm": 0.7733418345451355,
      "learning_rate": 9.99419200042492e-06,
      "loss": 0.8649,
      "step": 945
    },
    {
      "epoch": 0.13495007132667616,
      "grad_norm": 0.6091267466545105,
      "learning_rate": 9.994154836566555e-06,
      "loss": 0.8779,
      "step": 946
    },
    {
      "epoch": 0.13509272467902997,
      "grad_norm": 0.7535198926925659,
      "learning_rate": 9.994117554255756e-06,
      "loss": 0.9231,
      "step": 947
    },
    {
      "epoch": 0.13523537803138375,
      "grad_norm": 0.7609281539916992,
      "learning_rate": 9.994080153493405e-06,
      "loss": 0.873,
      "step": 948
    },
    {
      "epoch": 0.13537803138373752,
      "grad_norm": 0.7612059712409973,
      "learning_rate": 9.994042634280391e-06,
      "loss": 0.9069,
      "step": 949
    },
    {
      "epoch": 0.1355206847360913,
      "grad_norm": 0.6364650726318359,
      "learning_rate": 9.994004996617604e-06,
      "loss": 0.8902,
      "step": 950
    },
    {
      "epoch": 0.13566333808844508,
      "grad_norm": 0.6534590721130371,
      "learning_rate": 9.993967240505938e-06,
      "loss": 0.9005,
      "step": 951
    },
    {
      "epoch": 0.13580599144079886,
      "grad_norm": 0.8126868605613708,
      "learning_rate": 9.993929365946284e-06,
      "loss": 0.8455,
      "step": 952
    },
    {
      "epoch": 0.13594864479315263,
      "grad_norm": 0.7154679894447327,
      "learning_rate": 9.993891372939545e-06,
      "loss": 0.9016,
      "step": 953
    },
    {
      "epoch": 0.1360912981455064,
      "grad_norm": 0.740562379360199,
      "learning_rate": 9.993853261486619e-06,
      "loss": 0.9093,
      "step": 954
    },
    {
      "epoch": 0.1362339514978602,
      "grad_norm": 0.9111164808273315,
      "learning_rate": 9.993815031588412e-06,
      "loss": 0.8247,
      "step": 955
    },
    {
      "epoch": 0.13637660485021397,
      "grad_norm": 0.5779091119766235,
      "learning_rate": 9.993776683245829e-06,
      "loss": 0.9569,
      "step": 956
    },
    {
      "epoch": 0.13651925820256777,
      "grad_norm": 0.7303023338317871,
      "learning_rate": 9.993738216459782e-06,
      "loss": 0.8136,
      "step": 957
    },
    {
      "epoch": 0.13666191155492155,
      "grad_norm": 1.7917770147323608,
      "learning_rate": 9.993699631231181e-06,
      "loss": 0.7486,
      "step": 958
    },
    {
      "epoch": 0.13680456490727533,
      "grad_norm": 0.9355977177619934,
      "learning_rate": 9.993660927560942e-06,
      "loss": 0.8275,
      "step": 959
    },
    {
      "epoch": 0.1369472182596291,
      "grad_norm": 0.7837207913398743,
      "learning_rate": 9.993622105449983e-06,
      "loss": 0.8468,
      "step": 960
    },
    {
      "epoch": 0.13708987161198288,
      "grad_norm": 0.595391035079956,
      "learning_rate": 9.993583164899224e-06,
      "loss": 0.9133,
      "step": 961
    },
    {
      "epoch": 0.13723252496433666,
      "grad_norm": 0.687847375869751,
      "learning_rate": 9.99354410590959e-06,
      "loss": 0.8575,
      "step": 962
    },
    {
      "epoch": 0.13737517831669044,
      "grad_norm": 0.6320686340332031,
      "learning_rate": 9.993504928482007e-06,
      "loss": 0.9417,
      "step": 963
    },
    {
      "epoch": 0.13751783166904422,
      "grad_norm": 0.743326723575592,
      "learning_rate": 9.993465632617405e-06,
      "loss": 0.8574,
      "step": 964
    },
    {
      "epoch": 0.137660485021398,
      "grad_norm": 0.6819217801094055,
      "learning_rate": 9.993426218316714e-06,
      "loss": 0.9254,
      "step": 965
    },
    {
      "epoch": 0.13780313837375177,
      "grad_norm": 0.7288452982902527,
      "learning_rate": 9.99338668558087e-06,
      "loss": 0.9224,
      "step": 966
    },
    {
      "epoch": 0.13794579172610558,
      "grad_norm": 0.8626713156700134,
      "learning_rate": 9.99334703441081e-06,
      "loss": 0.9685,
      "step": 967
    },
    {
      "epoch": 0.13808844507845935,
      "grad_norm": 0.7050989270210266,
      "learning_rate": 9.993307264807475e-06,
      "loss": 0.8818,
      "step": 968
    },
    {
      "epoch": 0.13823109843081313,
      "grad_norm": 0.7689258456230164,
      "learning_rate": 9.99326737677181e-06,
      "loss": 0.8551,
      "step": 969
    },
    {
      "epoch": 0.1383737517831669,
      "grad_norm": 0.5554561018943787,
      "learning_rate": 9.993227370304758e-06,
      "loss": 0.9251,
      "step": 970
    },
    {
      "epoch": 0.1385164051355207,
      "grad_norm": 0.6589494943618774,
      "learning_rate": 9.99318724540727e-06,
      "loss": 0.8533,
      "step": 971
    },
    {
      "epoch": 0.13865905848787446,
      "grad_norm": 0.6229296922683716,
      "learning_rate": 9.993147002080294e-06,
      "loss": 0.8369,
      "step": 972
    },
    {
      "epoch": 0.13880171184022824,
      "grad_norm": 0.8061829805374146,
      "learning_rate": 9.99310664032479e-06,
      "loss": 0.8777,
      "step": 973
    },
    {
      "epoch": 0.13894436519258202,
      "grad_norm": 0.7398107051849365,
      "learning_rate": 9.993066160141714e-06,
      "loss": 0.9083,
      "step": 974
    },
    {
      "epoch": 0.1390870185449358,
      "grad_norm": 0.9920467734336853,
      "learning_rate": 9.993025561532021e-06,
      "loss": 0.8731,
      "step": 975
    },
    {
      "epoch": 0.13922967189728958,
      "grad_norm": 0.6123038530349731,
      "learning_rate": 9.99298484449668e-06,
      "loss": 0.9163,
      "step": 976
    },
    {
      "epoch": 0.13937232524964335,
      "grad_norm": 1.1445646286010742,
      "learning_rate": 9.992944009036657e-06,
      "loss": 0.9156,
      "step": 977
    },
    {
      "epoch": 0.13951497860199716,
      "grad_norm": 0.573752224445343,
      "learning_rate": 9.992903055152914e-06,
      "loss": 0.8817,
      "step": 978
    },
    {
      "epoch": 0.13965763195435094,
      "grad_norm": 0.8756287693977356,
      "learning_rate": 9.992861982846427e-06,
      "loss": 0.8271,
      "step": 979
    },
    {
      "epoch": 0.1398002853067047,
      "grad_norm": 0.7713871002197266,
      "learning_rate": 9.992820792118172e-06,
      "loss": 0.821,
      "step": 980
    },
    {
      "epoch": 0.1399429386590585,
      "grad_norm": 0.7339078187942505,
      "learning_rate": 9.992779482969121e-06,
      "loss": 0.8016,
      "step": 981
    },
    {
      "epoch": 0.14008559201141227,
      "grad_norm": 0.691035270690918,
      "learning_rate": 9.992738055400257e-06,
      "loss": 0.8973,
      "step": 982
    },
    {
      "epoch": 0.14022824536376605,
      "grad_norm": 0.6231393814086914,
      "learning_rate": 9.992696509412563e-06,
      "loss": 0.8909,
      "step": 983
    },
    {
      "epoch": 0.14037089871611982,
      "grad_norm": 0.8889601826667786,
      "learning_rate": 9.992654845007022e-06,
      "loss": 0.8557,
      "step": 984
    },
    {
      "epoch": 0.1405135520684736,
      "grad_norm": 0.7168399691581726,
      "learning_rate": 9.992613062184623e-06,
      "loss": 0.8655,
      "step": 985
    },
    {
      "epoch": 0.14065620542082738,
      "grad_norm": 0.7748574018478394,
      "learning_rate": 9.992571160946359e-06,
      "loss": 0.816,
      "step": 986
    },
    {
      "epoch": 0.14079885877318116,
      "grad_norm": 0.7910947799682617,
      "learning_rate": 9.992529141293221e-06,
      "loss": 0.7946,
      "step": 987
    },
    {
      "epoch": 0.14094151212553496,
      "grad_norm": 0.6379979252815247,
      "learning_rate": 9.992487003226209e-06,
      "loss": 0.9722,
      "step": 988
    },
    {
      "epoch": 0.14108416547788874,
      "grad_norm": 0.7884982228279114,
      "learning_rate": 9.992444746746316e-06,
      "loss": 0.7898,
      "step": 989
    },
    {
      "epoch": 0.14122681883024252,
      "grad_norm": 0.8776577115058899,
      "learning_rate": 9.992402371854553e-06,
      "loss": 0.8405,
      "step": 990
    },
    {
      "epoch": 0.1413694721825963,
      "grad_norm": 0.9063935279846191,
      "learning_rate": 9.99235987855192e-06,
      "loss": 0.8733,
      "step": 991
    },
    {
      "epoch": 0.14151212553495007,
      "grad_norm": 0.6572399139404297,
      "learning_rate": 9.992317266839425e-06,
      "loss": 0.873,
      "step": 992
    },
    {
      "epoch": 0.14165477888730385,
      "grad_norm": 0.8871137499809265,
      "learning_rate": 9.992274536718078e-06,
      "loss": 0.8343,
      "step": 993
    },
    {
      "epoch": 0.14179743223965763,
      "grad_norm": 0.8485704064369202,
      "learning_rate": 9.992231688188895e-06,
      "loss": 0.8819,
      "step": 994
    },
    {
      "epoch": 0.1419400855920114,
      "grad_norm": 0.7479352355003357,
      "learning_rate": 9.992188721252891e-06,
      "loss": 0.8971,
      "step": 995
    },
    {
      "epoch": 0.14208273894436518,
      "grad_norm": 1.3619558811187744,
      "learning_rate": 9.992145635911085e-06,
      "loss": 0.8704,
      "step": 996
    },
    {
      "epoch": 0.14222539229671896,
      "grad_norm": 0.6050323247909546,
      "learning_rate": 9.9921024321645e-06,
      "loss": 0.9826,
      "step": 997
    },
    {
      "epoch": 0.14236804564907277,
      "grad_norm": 0.5990591645240784,
      "learning_rate": 9.992059110014157e-06,
      "loss": 0.9134,
      "step": 998
    },
    {
      "epoch": 0.14251069900142654,
      "grad_norm": 0.9036572575569153,
      "learning_rate": 9.992015669461089e-06,
      "loss": 0.863,
      "step": 999
    },
    {
      "epoch": 0.14265335235378032,
      "grad_norm": 0.8414834141731262,
      "learning_rate": 9.991972110506322e-06,
      "loss": 0.8551,
      "step": 1000
    },
    {
      "epoch": 0.1427960057061341,
      "grad_norm": 0.9207097291946411,
      "learning_rate": 9.99192843315089e-06,
      "loss": 0.9099,
      "step": 1001
    },
    {
      "epoch": 0.14293865905848788,
      "grad_norm": 0.6639807820320129,
      "learning_rate": 9.991884637395833e-06,
      "loss": 0.9566,
      "step": 1002
    },
    {
      "epoch": 0.14308131241084165,
      "grad_norm": 0.6359516382217407,
      "learning_rate": 9.991840723242183e-06,
      "loss": 0.8945,
      "step": 1003
    },
    {
      "epoch": 0.14322396576319543,
      "grad_norm": 0.9046689867973328,
      "learning_rate": 9.991796690690988e-06,
      "loss": 0.8852,
      "step": 1004
    },
    {
      "epoch": 0.1433666191155492,
      "grad_norm": 0.7262295484542847,
      "learning_rate": 9.991752539743287e-06,
      "loss": 0.8947,
      "step": 1005
    },
    {
      "epoch": 0.143509272467903,
      "grad_norm": 0.7294202446937561,
      "learning_rate": 9.991708270400128e-06,
      "loss": 0.8629,
      "step": 1006
    },
    {
      "epoch": 0.14365192582025677,
      "grad_norm": 0.6945996284484863,
      "learning_rate": 9.991663882662565e-06,
      "loss": 0.8053,
      "step": 1007
    },
    {
      "epoch": 0.14379457917261054,
      "grad_norm": 0.6925103664398193,
      "learning_rate": 9.991619376531649e-06,
      "loss": 0.8881,
      "step": 1008
    },
    {
      "epoch": 0.14393723252496435,
      "grad_norm": 0.7418951988220215,
      "learning_rate": 9.991574752008436e-06,
      "loss": 0.8712,
      "step": 1009
    },
    {
      "epoch": 0.14407988587731813,
      "grad_norm": 0.6167254447937012,
      "learning_rate": 9.99153000909398e-06,
      "loss": 0.8878,
      "step": 1010
    },
    {
      "epoch": 0.1442225392296719,
      "grad_norm": 0.7526018619537354,
      "learning_rate": 9.991485147789348e-06,
      "loss": 0.9019,
      "step": 1011
    },
    {
      "epoch": 0.14436519258202568,
      "grad_norm": 0.9956358075141907,
      "learning_rate": 9.9914401680956e-06,
      "loss": 0.9458,
      "step": 1012
    },
    {
      "epoch": 0.14450784593437946,
      "grad_norm": 0.837101936340332,
      "learning_rate": 9.991395070013806e-06,
      "loss": 0.8684,
      "step": 1013
    },
    {
      "epoch": 0.14465049928673324,
      "grad_norm": 0.9686935544013977,
      "learning_rate": 9.99134985354503e-06,
      "loss": 0.8609,
      "step": 1014
    },
    {
      "epoch": 0.144793152639087,
      "grad_norm": 0.7613106369972229,
      "learning_rate": 9.991304518690354e-06,
      "loss": 0.8575,
      "step": 1015
    },
    {
      "epoch": 0.1449358059914408,
      "grad_norm": 0.6044681668281555,
      "learning_rate": 9.991259065450846e-06,
      "loss": 0.9365,
      "step": 1016
    },
    {
      "epoch": 0.14507845934379457,
      "grad_norm": 0.8422667384147644,
      "learning_rate": 9.991213493827585e-06,
      "loss": 0.8833,
      "step": 1017
    },
    {
      "epoch": 0.14522111269614835,
      "grad_norm": 0.6289754509925842,
      "learning_rate": 9.991167803821653e-06,
      "loss": 0.8492,
      "step": 1018
    },
    {
      "epoch": 0.14536376604850215,
      "grad_norm": 0.797286331653595,
      "learning_rate": 9.991121995434135e-06,
      "loss": 0.911,
      "step": 1019
    },
    {
      "epoch": 0.14550641940085593,
      "grad_norm": 0.8525795340538025,
      "learning_rate": 9.991076068666114e-06,
      "loss": 0.8438,
      "step": 1020
    },
    {
      "epoch": 0.1456490727532097,
      "grad_norm": 0.840726912021637,
      "learning_rate": 9.991030023518681e-06,
      "loss": 0.8567,
      "step": 1021
    },
    {
      "epoch": 0.14579172610556348,
      "grad_norm": 0.615422248840332,
      "learning_rate": 9.99098385999293e-06,
      "loss": 0.8845,
      "step": 1022
    },
    {
      "epoch": 0.14593437945791726,
      "grad_norm": 0.93955397605896,
      "learning_rate": 9.990937578089953e-06,
      "loss": 0.7803,
      "step": 1023
    },
    {
      "epoch": 0.14607703281027104,
      "grad_norm": 0.8013250827789307,
      "learning_rate": 9.99089117781085e-06,
      "loss": 0.8792,
      "step": 1024
    },
    {
      "epoch": 0.14621968616262482,
      "grad_norm": 0.5773342847824097,
      "learning_rate": 9.99084465915672e-06,
      "loss": 0.9155,
      "step": 1025
    },
    {
      "epoch": 0.1463623395149786,
      "grad_norm": 0.6571844816207886,
      "learning_rate": 9.990798022128667e-06,
      "loss": 0.8368,
      "step": 1026
    },
    {
      "epoch": 0.14650499286733237,
      "grad_norm": 0.9351130723953247,
      "learning_rate": 9.990751266727795e-06,
      "loss": 0.897,
      "step": 1027
    },
    {
      "epoch": 0.14664764621968615,
      "grad_norm": 0.7844663262367249,
      "learning_rate": 9.990704392955218e-06,
      "loss": 0.8556,
      "step": 1028
    },
    {
      "epoch": 0.14679029957203996,
      "grad_norm": 0.7472745776176453,
      "learning_rate": 9.990657400812043e-06,
      "loss": 0.8071,
      "step": 1029
    },
    {
      "epoch": 0.14693295292439373,
      "grad_norm": 0.8757637739181519,
      "learning_rate": 9.990610290299387e-06,
      "loss": 0.7922,
      "step": 1030
    },
    {
      "epoch": 0.1470756062767475,
      "grad_norm": 0.7582662105560303,
      "learning_rate": 9.990563061418367e-06,
      "loss": 0.8197,
      "step": 1031
    },
    {
      "epoch": 0.1472182596291013,
      "grad_norm": 0.6933606863021851,
      "learning_rate": 9.990515714170103e-06,
      "loss": 0.9047,
      "step": 1032
    },
    {
      "epoch": 0.14736091298145507,
      "grad_norm": 0.7019277811050415,
      "learning_rate": 9.990468248555716e-06,
      "loss": 0.8288,
      "step": 1033
    },
    {
      "epoch": 0.14750356633380884,
      "grad_norm": 0.7585964798927307,
      "learning_rate": 9.990420664576334e-06,
      "loss": 0.768,
      "step": 1034
    },
    {
      "epoch": 0.14764621968616262,
      "grad_norm": 0.7656587958335876,
      "learning_rate": 9.990372962233087e-06,
      "loss": 0.8458,
      "step": 1035
    },
    {
      "epoch": 0.1477888730385164,
      "grad_norm": 1.2133119106292725,
      "learning_rate": 9.990325141527105e-06,
      "loss": 0.6152,
      "step": 1036
    },
    {
      "epoch": 0.14793152639087018,
      "grad_norm": 0.7640325427055359,
      "learning_rate": 9.99027720245952e-06,
      "loss": 0.9362,
      "step": 1037
    },
    {
      "epoch": 0.14807417974322395,
      "grad_norm": 0.8579018712043762,
      "learning_rate": 9.990229145031472e-06,
      "loss": 0.8926,
      "step": 1038
    },
    {
      "epoch": 0.14821683309557773,
      "grad_norm": 0.9569811224937439,
      "learning_rate": 9.9901809692441e-06,
      "loss": 0.9036,
      "step": 1039
    },
    {
      "epoch": 0.14835948644793154,
      "grad_norm": 0.637883722782135,
      "learning_rate": 9.990132675098546e-06,
      "loss": 0.8715,
      "step": 1040
    },
    {
      "epoch": 0.14850213980028532,
      "grad_norm": 0.6271453499794006,
      "learning_rate": 9.990084262595954e-06,
      "loss": 0.8285,
      "step": 1041
    },
    {
      "epoch": 0.1486447931526391,
      "grad_norm": 0.6584393978118896,
      "learning_rate": 9.990035731737477e-06,
      "loss": 0.9456,
      "step": 1042
    },
    {
      "epoch": 0.14878744650499287,
      "grad_norm": 0.9535266160964966,
      "learning_rate": 9.989987082524262e-06,
      "loss": 0.8898,
      "step": 1043
    },
    {
      "epoch": 0.14893009985734665,
      "grad_norm": 0.6312132477760315,
      "learning_rate": 9.989938314957463e-06,
      "loss": 0.8398,
      "step": 1044
    },
    {
      "epoch": 0.14907275320970043,
      "grad_norm": 0.8063013553619385,
      "learning_rate": 9.989889429038238e-06,
      "loss": 0.5944,
      "step": 1045
    },
    {
      "epoch": 0.1492154065620542,
      "grad_norm": 0.7557687759399414,
      "learning_rate": 9.989840424767748e-06,
      "loss": 0.8615,
      "step": 1046
    },
    {
      "epoch": 0.14935805991440798,
      "grad_norm": 0.6279237866401672,
      "learning_rate": 9.989791302147151e-06,
      "loss": 0.8655,
      "step": 1047
    },
    {
      "epoch": 0.14950071326676176,
      "grad_norm": 1.0212706327438354,
      "learning_rate": 9.989742061177617e-06,
      "loss": 0.7489,
      "step": 1048
    },
    {
      "epoch": 0.14964336661911554,
      "grad_norm": 0.9332009553909302,
      "learning_rate": 9.98969270186031e-06,
      "loss": 0.829,
      "step": 1049
    },
    {
      "epoch": 0.14978601997146934,
      "grad_norm": 0.7611649036407471,
      "learning_rate": 9.989643224196403e-06,
      "loss": 0.8425,
      "step": 1050
    },
    {
      "epoch": 0.14992867332382312,
      "grad_norm": 0.7248733043670654,
      "learning_rate": 9.989593628187068e-06,
      "loss": 0.9409,
      "step": 1051
    },
    {
      "epoch": 0.1500713266761769,
      "grad_norm": 0.5785107612609863,
      "learning_rate": 9.989543913833483e-06,
      "loss": 0.8661,
      "step": 1052
    },
    {
      "epoch": 0.15021398002853067,
      "grad_norm": 0.8986565470695496,
      "learning_rate": 9.989494081136825e-06,
      "loss": 0.899,
      "step": 1053
    },
    {
      "epoch": 0.15035663338088445,
      "grad_norm": 0.7995362281799316,
      "learning_rate": 9.989444130098279e-06,
      "loss": 0.8898,
      "step": 1054
    },
    {
      "epoch": 0.15049928673323823,
      "grad_norm": 0.915816068649292,
      "learning_rate": 9.989394060719026e-06,
      "loss": 0.8574,
      "step": 1055
    },
    {
      "epoch": 0.150641940085592,
      "grad_norm": 0.7998825907707214,
      "learning_rate": 9.989343873000257e-06,
      "loss": 0.9059,
      "step": 1056
    },
    {
      "epoch": 0.15078459343794579,
      "grad_norm": 0.6868506669998169,
      "learning_rate": 9.98929356694316e-06,
      "loss": 0.9177,
      "step": 1057
    },
    {
      "epoch": 0.15092724679029956,
      "grad_norm": 0.9999756813049316,
      "learning_rate": 9.98924314254893e-06,
      "loss": 0.9093,
      "step": 1058
    },
    {
      "epoch": 0.15106990014265334,
      "grad_norm": 0.7811575531959534,
      "learning_rate": 9.98919259981876e-06,
      "loss": 0.8138,
      "step": 1059
    },
    {
      "epoch": 0.15121255349500715,
      "grad_norm": 0.911644697189331,
      "learning_rate": 9.989141938753852e-06,
      "loss": 0.8323,
      "step": 1060
    },
    {
      "epoch": 0.15135520684736092,
      "grad_norm": 0.6191381812095642,
      "learning_rate": 9.989091159355406e-06,
      "loss": 0.8473,
      "step": 1061
    },
    {
      "epoch": 0.1514978601997147,
      "grad_norm": 0.7394691705703735,
      "learning_rate": 9.989040261624628e-06,
      "loss": 0.7634,
      "step": 1062
    },
    {
      "epoch": 0.15164051355206848,
      "grad_norm": 0.9278765916824341,
      "learning_rate": 9.988989245562722e-06,
      "loss": 0.8995,
      "step": 1063
    },
    {
      "epoch": 0.15178316690442226,
      "grad_norm": 0.7579810619354248,
      "learning_rate": 9.988938111170902e-06,
      "loss": 0.8339,
      "step": 1064
    },
    {
      "epoch": 0.15192582025677603,
      "grad_norm": 0.8946094512939453,
      "learning_rate": 9.988886858450377e-06,
      "loss": 0.8315,
      "step": 1065
    },
    {
      "epoch": 0.1520684736091298,
      "grad_norm": 0.678996205329895,
      "learning_rate": 9.988835487402366e-06,
      "loss": 0.8308,
      "step": 1066
    },
    {
      "epoch": 0.1522111269614836,
      "grad_norm": 0.724155604839325,
      "learning_rate": 9.988783998028085e-06,
      "loss": 0.8547,
      "step": 1067
    },
    {
      "epoch": 0.15235378031383737,
      "grad_norm": 1.0426677465438843,
      "learning_rate": 9.988732390328754e-06,
      "loss": 0.9468,
      "step": 1068
    },
    {
      "epoch": 0.15249643366619114,
      "grad_norm": 0.7471747994422913,
      "learning_rate": 9.988680664305603e-06,
      "loss": 0.813,
      "step": 1069
    },
    {
      "epoch": 0.15263908701854492,
      "grad_norm": 0.9394763112068176,
      "learning_rate": 9.988628819959853e-06,
      "loss": 0.8757,
      "step": 1070
    },
    {
      "epoch": 0.15278174037089873,
      "grad_norm": 0.6399518251419067,
      "learning_rate": 9.988576857292736e-06,
      "loss": 0.884,
      "step": 1071
    },
    {
      "epoch": 0.1529243937232525,
      "grad_norm": 0.8174779415130615,
      "learning_rate": 9.988524776305483e-06,
      "loss": 0.8657,
      "step": 1072
    },
    {
      "epoch": 0.15306704707560628,
      "grad_norm": 0.6559473872184753,
      "learning_rate": 9.988472576999331e-06,
      "loss": 0.8526,
      "step": 1073
    },
    {
      "epoch": 0.15320970042796006,
      "grad_norm": 0.7370951771736145,
      "learning_rate": 9.988420259375518e-06,
      "loss": 0.8755,
      "step": 1074
    },
    {
      "epoch": 0.15335235378031384,
      "grad_norm": 0.8102855682373047,
      "learning_rate": 9.988367823435282e-06,
      "loss": 0.8673,
      "step": 1075
    },
    {
      "epoch": 0.15349500713266762,
      "grad_norm": 0.7813428044319153,
      "learning_rate": 9.98831526917987e-06,
      "loss": 0.9025,
      "step": 1076
    },
    {
      "epoch": 0.1536376604850214,
      "grad_norm": 0.8604279160499573,
      "learning_rate": 9.98826259661053e-06,
      "loss": 0.8536,
      "step": 1077
    },
    {
      "epoch": 0.15378031383737517,
      "grad_norm": 0.8180389404296875,
      "learning_rate": 9.988209805728505e-06,
      "loss": 0.8953,
      "step": 1078
    },
    {
      "epoch": 0.15392296718972895,
      "grad_norm": 1.1215965747833252,
      "learning_rate": 9.988156896535055e-06,
      "loss": 0.89,
      "step": 1079
    },
    {
      "epoch": 0.15406562054208273,
      "grad_norm": 0.8152418732643127,
      "learning_rate": 9.988103869031428e-06,
      "loss": 0.8455,
      "step": 1080
    },
    {
      "epoch": 0.15420827389443653,
      "grad_norm": 0.6798477172851562,
      "learning_rate": 9.988050723218885e-06,
      "loss": 0.9168,
      "step": 1081
    },
    {
      "epoch": 0.1543509272467903,
      "grad_norm": 0.7677843570709229,
      "learning_rate": 9.987997459098686e-06,
      "loss": 0.8213,
      "step": 1082
    },
    {
      "epoch": 0.1544935805991441,
      "grad_norm": 0.7342092990875244,
      "learning_rate": 9.987944076672096e-06,
      "loss": 0.9088,
      "step": 1083
    },
    {
      "epoch": 0.15463623395149786,
      "grad_norm": 0.8838322162628174,
      "learning_rate": 9.987890575940378e-06,
      "loss": 0.8295,
      "step": 1084
    },
    {
      "epoch": 0.15477888730385164,
      "grad_norm": 1.0102605819702148,
      "learning_rate": 9.987836956904802e-06,
      "loss": 0.9076,
      "step": 1085
    },
    {
      "epoch": 0.15492154065620542,
      "grad_norm": 0.6762107610702515,
      "learning_rate": 9.98778321956664e-06,
      "loss": 0.7223,
      "step": 1086
    },
    {
      "epoch": 0.1550641940085592,
      "grad_norm": 0.7655817866325378,
      "learning_rate": 9.987729363927167e-06,
      "loss": 0.8291,
      "step": 1087
    },
    {
      "epoch": 0.15520684736091298,
      "grad_norm": 0.7236725687980652,
      "learning_rate": 9.987675389987661e-06,
      "loss": 0.7924,
      "step": 1088
    },
    {
      "epoch": 0.15534950071326675,
      "grad_norm": 0.7709295749664307,
      "learning_rate": 9.987621297749401e-06,
      "loss": 0.7729,
      "step": 1089
    },
    {
      "epoch": 0.15549215406562053,
      "grad_norm": 0.9965909123420715,
      "learning_rate": 9.98756708721367e-06,
      "loss": 0.832,
      "step": 1090
    },
    {
      "epoch": 0.15563480741797434,
      "grad_norm": 0.7601058483123779,
      "learning_rate": 9.987512758381754e-06,
      "loss": 0.8576,
      "step": 1091
    },
    {
      "epoch": 0.1557774607703281,
      "grad_norm": 0.6895081996917725,
      "learning_rate": 9.987458311254942e-06,
      "loss": 0.8564,
      "step": 1092
    },
    {
      "epoch": 0.1559201141226819,
      "grad_norm": 0.9341307878494263,
      "learning_rate": 9.987403745834525e-06,
      "loss": 0.8855,
      "step": 1093
    },
    {
      "epoch": 0.15606276747503567,
      "grad_norm": 0.9359080791473389,
      "learning_rate": 9.987349062121797e-06,
      "loss": 0.9022,
      "step": 1094
    },
    {
      "epoch": 0.15620542082738945,
      "grad_norm": 0.8535083532333374,
      "learning_rate": 9.987294260118055e-06,
      "loss": 0.8289,
      "step": 1095
    },
    {
      "epoch": 0.15634807417974322,
      "grad_norm": 0.711827278137207,
      "learning_rate": 9.987239339824599e-06,
      "loss": 0.8588,
      "step": 1096
    },
    {
      "epoch": 0.156490727532097,
      "grad_norm": 1.1219594478607178,
      "learning_rate": 9.98718430124273e-06,
      "loss": 0.8032,
      "step": 1097
    },
    {
      "epoch": 0.15663338088445078,
      "grad_norm": 1.6010338068008423,
      "learning_rate": 9.987129144373758e-06,
      "loss": 0.7201,
      "step": 1098
    },
    {
      "epoch": 0.15677603423680456,
      "grad_norm": 0.6423503756523132,
      "learning_rate": 9.987073869218986e-06,
      "loss": 0.8264,
      "step": 1099
    },
    {
      "epoch": 0.15691868758915833,
      "grad_norm": 0.7870016694068909,
      "learning_rate": 9.987018475779728e-06,
      "loss": 0.8288,
      "step": 1100
    },
    {
      "epoch": 0.1570613409415121,
      "grad_norm": 0.9856181144714355,
      "learning_rate": 9.986962964057297e-06,
      "loss": 0.7654,
      "step": 1101
    },
    {
      "epoch": 0.15720399429386592,
      "grad_norm": 0.7096058130264282,
      "learning_rate": 9.986907334053011e-06,
      "loss": 0.8829,
      "step": 1102
    },
    {
      "epoch": 0.1573466476462197,
      "grad_norm": 0.8059448003768921,
      "learning_rate": 9.986851585768186e-06,
      "loss": 0.9302,
      "step": 1103
    },
    {
      "epoch": 0.15748930099857347,
      "grad_norm": 0.7682087421417236,
      "learning_rate": 9.986795719204148e-06,
      "loss": 0.9289,
      "step": 1104
    },
    {
      "epoch": 0.15763195435092725,
      "grad_norm": 0.9392029047012329,
      "learning_rate": 9.986739734362221e-06,
      "loss": 0.7944,
      "step": 1105
    },
    {
      "epoch": 0.15777460770328103,
      "grad_norm": 0.8738077282905579,
      "learning_rate": 9.98668363124373e-06,
      "loss": 0.8127,
      "step": 1106
    },
    {
      "epoch": 0.1579172610556348,
      "grad_norm": 0.8040480613708496,
      "learning_rate": 9.98662740985001e-06,
      "loss": 0.8424,
      "step": 1107
    },
    {
      "epoch": 0.15805991440798858,
      "grad_norm": 0.7068692445755005,
      "learning_rate": 9.986571070182391e-06,
      "loss": 0.9402,
      "step": 1108
    },
    {
      "epoch": 0.15820256776034236,
      "grad_norm": 0.8845683932304382,
      "learning_rate": 9.986514612242213e-06,
      "loss": 0.8926,
      "step": 1109
    },
    {
      "epoch": 0.15834522111269614,
      "grad_norm": 0.6887304186820984,
      "learning_rate": 9.98645803603081e-06,
      "loss": 0.907,
      "step": 1110
    },
    {
      "epoch": 0.15848787446504992,
      "grad_norm": 0.6196120381355286,
      "learning_rate": 9.986401341549528e-06,
      "loss": 0.9172,
      "step": 1111
    },
    {
      "epoch": 0.15863052781740372,
      "grad_norm": 0.7921424508094788,
      "learning_rate": 9.986344528799711e-06,
      "loss": 0.9042,
      "step": 1112
    },
    {
      "epoch": 0.1587731811697575,
      "grad_norm": 0.7048509120941162,
      "learning_rate": 9.986287597782707e-06,
      "loss": 0.7891,
      "step": 1113
    },
    {
      "epoch": 0.15891583452211128,
      "grad_norm": 0.7620043158531189,
      "learning_rate": 9.986230548499863e-06,
      "loss": 0.8985,
      "step": 1114
    },
    {
      "epoch": 0.15905848787446505,
      "grad_norm": 0.9016017913818359,
      "learning_rate": 9.986173380952535e-06,
      "loss": 0.8309,
      "step": 1115
    },
    {
      "epoch": 0.15920114122681883,
      "grad_norm": 0.6812255382537842,
      "learning_rate": 9.986116095142078e-06,
      "loss": 0.8712,
      "step": 1116
    },
    {
      "epoch": 0.1593437945791726,
      "grad_norm": 0.6163586974143982,
      "learning_rate": 9.98605869106985e-06,
      "loss": 0.9178,
      "step": 1117
    },
    {
      "epoch": 0.1594864479315264,
      "grad_norm": 0.5008699297904968,
      "learning_rate": 9.986001168737217e-06,
      "loss": 0.8856,
      "step": 1118
    },
    {
      "epoch": 0.15962910128388016,
      "grad_norm": 0.851180911064148,
      "learning_rate": 9.985943528145537e-06,
      "loss": 0.8541,
      "step": 1119
    },
    {
      "epoch": 0.15977175463623394,
      "grad_norm": 0.7396800518035889,
      "learning_rate": 9.98588576929618e-06,
      "loss": 0.8499,
      "step": 1120
    },
    {
      "epoch": 0.15991440798858772,
      "grad_norm": 0.7640107870101929,
      "learning_rate": 9.985827892190516e-06,
      "loss": 0.8381,
      "step": 1121
    },
    {
      "epoch": 0.16005706134094153,
      "grad_norm": 0.8326088786125183,
      "learning_rate": 9.985769896829918e-06,
      "loss": 0.8361,
      "step": 1122
    },
    {
      "epoch": 0.1601997146932953,
      "grad_norm": 0.8211906552314758,
      "learning_rate": 9.98571178321576e-06,
      "loss": 0.8448,
      "step": 1123
    },
    {
      "epoch": 0.16034236804564908,
      "grad_norm": 0.8785040974617004,
      "learning_rate": 9.985653551349425e-06,
      "loss": 0.779,
      "step": 1124
    },
    {
      "epoch": 0.16048502139800286,
      "grad_norm": 0.8220649361610413,
      "learning_rate": 9.985595201232288e-06,
      "loss": 0.9088,
      "step": 1125
    },
    {
      "epoch": 0.16062767475035664,
      "grad_norm": 0.7736160159111023,
      "learning_rate": 9.985536732865735e-06,
      "loss": 0.8144,
      "step": 1126
    },
    {
      "epoch": 0.1607703281027104,
      "grad_norm": 1.540403962135315,
      "learning_rate": 9.985478146251156e-06,
      "loss": 0.7456,
      "step": 1127
    },
    {
      "epoch": 0.1609129814550642,
      "grad_norm": 0.7961999177932739,
      "learning_rate": 9.985419441389936e-06,
      "loss": 0.7858,
      "step": 1128
    },
    {
      "epoch": 0.16105563480741797,
      "grad_norm": 0.7917633652687073,
      "learning_rate": 9.985360618283468e-06,
      "loss": 0.868,
      "step": 1129
    },
    {
      "epoch": 0.16119828815977175,
      "grad_norm": 0.9145285487174988,
      "learning_rate": 9.985301676933151e-06,
      "loss": 0.843,
      "step": 1130
    },
    {
      "epoch": 0.16134094151212552,
      "grad_norm": 0.6547228097915649,
      "learning_rate": 9.98524261734038e-06,
      "loss": 0.8247,
      "step": 1131
    },
    {
      "epoch": 0.1614835948644793,
      "grad_norm": 0.7892211675643921,
      "learning_rate": 9.985183439506556e-06,
      "loss": 0.8616,
      "step": 1132
    },
    {
      "epoch": 0.1616262482168331,
      "grad_norm": 0.9836748838424683,
      "learning_rate": 9.985124143433082e-06,
      "loss": 0.8685,
      "step": 1133
    },
    {
      "epoch": 0.16176890156918688,
      "grad_norm": 0.8149711489677429,
      "learning_rate": 9.985064729121367e-06,
      "loss": 0.8477,
      "step": 1134
    },
    {
      "epoch": 0.16191155492154066,
      "grad_norm": 0.6759414076805115,
      "learning_rate": 9.985005196572817e-06,
      "loss": 0.8915,
      "step": 1135
    },
    {
      "epoch": 0.16205420827389444,
      "grad_norm": 0.6080157160758972,
      "learning_rate": 9.984945545788847e-06,
      "loss": 0.8656,
      "step": 1136
    },
    {
      "epoch": 0.16219686162624822,
      "grad_norm": 0.7513832449913025,
      "learning_rate": 9.984885776770869e-06,
      "loss": 0.9318,
      "step": 1137
    },
    {
      "epoch": 0.162339514978602,
      "grad_norm": 0.9032502770423889,
      "learning_rate": 9.984825889520303e-06,
      "loss": 0.865,
      "step": 1138
    },
    {
      "epoch": 0.16248216833095577,
      "grad_norm": 0.6458960771560669,
      "learning_rate": 9.984765884038567e-06,
      "loss": 0.8007,
      "step": 1139
    },
    {
      "epoch": 0.16262482168330955,
      "grad_norm": 0.8249204158782959,
      "learning_rate": 9.984705760327085e-06,
      "loss": 0.8908,
      "step": 1140
    },
    {
      "epoch": 0.16276747503566333,
      "grad_norm": 0.8001787662506104,
      "learning_rate": 9.984645518387287e-06,
      "loss": 0.8545,
      "step": 1141
    },
    {
      "epoch": 0.1629101283880171,
      "grad_norm": 0.8849978446960449,
      "learning_rate": 9.984585158220594e-06,
      "loss": 0.8599,
      "step": 1142
    },
    {
      "epoch": 0.1630527817403709,
      "grad_norm": 0.7225953936576843,
      "learning_rate": 9.984524679828445e-06,
      "loss": 0.8637,
      "step": 1143
    },
    {
      "epoch": 0.1631954350927247,
      "grad_norm": 0.5402420163154602,
      "learning_rate": 9.98446408321227e-06,
      "loss": 0.8936,
      "step": 1144
    },
    {
      "epoch": 0.16333808844507847,
      "grad_norm": 0.7665300369262695,
      "learning_rate": 9.98440336837351e-06,
      "loss": 0.912,
      "step": 1145
    },
    {
      "epoch": 0.16348074179743224,
      "grad_norm": 0.9873577356338501,
      "learning_rate": 9.9843425353136e-06,
      "loss": 0.8323,
      "step": 1146
    },
    {
      "epoch": 0.16362339514978602,
      "grad_norm": 0.7201932072639465,
      "learning_rate": 9.984281584033988e-06,
      "loss": 0.8669,
      "step": 1147
    },
    {
      "epoch": 0.1637660485021398,
      "grad_norm": 0.6797975897789001,
      "learning_rate": 9.984220514536115e-06,
      "loss": 0.8417,
      "step": 1148
    },
    {
      "epoch": 0.16390870185449358,
      "grad_norm": 0.8092601895332336,
      "learning_rate": 9.984159326821433e-06,
      "loss": 0.8525,
      "step": 1149
    },
    {
      "epoch": 0.16405135520684735,
      "grad_norm": 0.7208330035209656,
      "learning_rate": 9.984098020891391e-06,
      "loss": 0.8427,
      "step": 1150
    },
    {
      "epoch": 0.16419400855920113,
      "grad_norm": 0.8303775787353516,
      "learning_rate": 9.984036596747445e-06,
      "loss": 0.8823,
      "step": 1151
    },
    {
      "epoch": 0.1643366619115549,
      "grad_norm": 0.7441378831863403,
      "learning_rate": 9.983975054391051e-06,
      "loss": 0.8433,
      "step": 1152
    },
    {
      "epoch": 0.16447931526390872,
      "grad_norm": 0.9276571273803711,
      "learning_rate": 9.98391339382367e-06,
      "loss": 0.8364,
      "step": 1153
    },
    {
      "epoch": 0.1646219686162625,
      "grad_norm": 0.7798731923103333,
      "learning_rate": 9.98385161504676e-06,
      "loss": 0.8763,
      "step": 1154
    },
    {
      "epoch": 0.16476462196861627,
      "grad_norm": 1.1109129190444946,
      "learning_rate": 9.983789718061792e-06,
      "loss": 0.8132,
      "step": 1155
    },
    {
      "epoch": 0.16490727532097005,
      "grad_norm": 1.026613712310791,
      "learning_rate": 9.98372770287023e-06,
      "loss": 0.7514,
      "step": 1156
    },
    {
      "epoch": 0.16504992867332383,
      "grad_norm": 0.674289882183075,
      "learning_rate": 9.983665569473547e-06,
      "loss": 0.8774,
      "step": 1157
    },
    {
      "epoch": 0.1651925820256776,
      "grad_norm": 0.9314978718757629,
      "learning_rate": 9.983603317873217e-06,
      "loss": 0.8595,
      "step": 1158
    },
    {
      "epoch": 0.16533523537803138,
      "grad_norm": 0.7667367458343506,
      "learning_rate": 9.983540948070713e-06,
      "loss": 0.7934,
      "step": 1159
    },
    {
      "epoch": 0.16547788873038516,
      "grad_norm": 0.7203590869903564,
      "learning_rate": 9.983478460067519e-06,
      "loss": 0.8936,
      "step": 1160
    },
    {
      "epoch": 0.16562054208273894,
      "grad_norm": 0.768721878528595,
      "learning_rate": 9.983415853865113e-06,
      "loss": 0.8781,
      "step": 1161
    },
    {
      "epoch": 0.16576319543509271,
      "grad_norm": 0.9745950698852539,
      "learning_rate": 9.983353129464983e-06,
      "loss": 0.8604,
      "step": 1162
    },
    {
      "epoch": 0.1659058487874465,
      "grad_norm": 0.8448209762573242,
      "learning_rate": 9.983290286868617e-06,
      "loss": 0.8615,
      "step": 1163
    },
    {
      "epoch": 0.1660485021398003,
      "grad_norm": 0.893592894077301,
      "learning_rate": 9.983227326077503e-06,
      "loss": 0.8517,
      "step": 1164
    },
    {
      "epoch": 0.16619115549215407,
      "grad_norm": 0.7238365411758423,
      "learning_rate": 9.983164247093134e-06,
      "loss": 0.851,
      "step": 1165
    },
    {
      "epoch": 0.16633380884450785,
      "grad_norm": 0.8795214891433716,
      "learning_rate": 9.983101049917007e-06,
      "loss": 0.8641,
      "step": 1166
    },
    {
      "epoch": 0.16647646219686163,
      "grad_norm": 0.6775270104408264,
      "learning_rate": 9.983037734550625e-06,
      "loss": 0.8664,
      "step": 1167
    },
    {
      "epoch": 0.1666191155492154,
      "grad_norm": 0.6358814239501953,
      "learning_rate": 9.982974300995483e-06,
      "loss": 0.8293,
      "step": 1168
    },
    {
      "epoch": 0.16676176890156919,
      "grad_norm": 0.75082927942276,
      "learning_rate": 9.982910749253091e-06,
      "loss": 0.8579,
      "step": 1169
    },
    {
      "epoch": 0.16690442225392296,
      "grad_norm": 1.2580338716506958,
      "learning_rate": 9.982847079324953e-06,
      "loss": 0.8421,
      "step": 1170
    },
    {
      "epoch": 0.16704707560627674,
      "grad_norm": 0.7623533010482788,
      "learning_rate": 9.98278329121258e-06,
      "loss": 0.9265,
      "step": 1171
    },
    {
      "epoch": 0.16718972895863052,
      "grad_norm": 0.7251095175743103,
      "learning_rate": 9.982719384917486e-06,
      "loss": 0.793,
      "step": 1172
    },
    {
      "epoch": 0.1673323823109843,
      "grad_norm": 0.7233574986457825,
      "learning_rate": 9.982655360441185e-06,
      "loss": 0.9149,
      "step": 1173
    },
    {
      "epoch": 0.1674750356633381,
      "grad_norm": 0.8108822703361511,
      "learning_rate": 9.982591217785195e-06,
      "loss": 0.8378,
      "step": 1174
    },
    {
      "epoch": 0.16761768901569188,
      "grad_norm": 0.8185499906539917,
      "learning_rate": 9.982526956951041e-06,
      "loss": 0.6309,
      "step": 1175
    },
    {
      "epoch": 0.16776034236804566,
      "grad_norm": 0.872106671333313,
      "learning_rate": 9.982462577940244e-06,
      "loss": 0.8082,
      "step": 1176
    },
    {
      "epoch": 0.16790299572039943,
      "grad_norm": 0.8142444491386414,
      "learning_rate": 9.982398080754331e-06,
      "loss": 0.8214,
      "step": 1177
    },
    {
      "epoch": 0.1680456490727532,
      "grad_norm": 0.7037302851676941,
      "learning_rate": 9.982333465394834e-06,
      "loss": 0.8348,
      "step": 1178
    },
    {
      "epoch": 0.168188302425107,
      "grad_norm": 0.7830038666725159,
      "learning_rate": 9.982268731863285e-06,
      "loss": 0.9163,
      "step": 1179
    },
    {
      "epoch": 0.16833095577746077,
      "grad_norm": 0.6541464328765869,
      "learning_rate": 9.982203880161216e-06,
      "loss": 0.8602,
      "step": 1180
    },
    {
      "epoch": 0.16847360912981454,
      "grad_norm": 0.6037623286247253,
      "learning_rate": 9.98213891029017e-06,
      "loss": 0.8497,
      "step": 1181
    },
    {
      "epoch": 0.16861626248216832,
      "grad_norm": 0.9497793316841125,
      "learning_rate": 9.982073822251685e-06,
      "loss": 0.8877,
      "step": 1182
    },
    {
      "epoch": 0.1687589158345221,
      "grad_norm": 0.9584270119667053,
      "learning_rate": 9.982008616047305e-06,
      "loss": 0.9903,
      "step": 1183
    },
    {
      "epoch": 0.1689015691868759,
      "grad_norm": 0.9811310768127441,
      "learning_rate": 9.981943291678578e-06,
      "loss": 0.8805,
      "step": 1184
    },
    {
      "epoch": 0.16904422253922968,
      "grad_norm": 0.7309616208076477,
      "learning_rate": 9.98187784914705e-06,
      "loss": 0.8533,
      "step": 1185
    },
    {
      "epoch": 0.16918687589158346,
      "grad_norm": 0.5927829146385193,
      "learning_rate": 9.981812288454278e-06,
      "loss": 0.83,
      "step": 1186
    },
    {
      "epoch": 0.16932952924393724,
      "grad_norm": 0.7374027371406555,
      "learning_rate": 9.981746609601815e-06,
      "loss": 0.8102,
      "step": 1187
    },
    {
      "epoch": 0.16947218259629102,
      "grad_norm": 0.8296719193458557,
      "learning_rate": 9.981680812591215e-06,
      "loss": 0.9044,
      "step": 1188
    },
    {
      "epoch": 0.1696148359486448,
      "grad_norm": 1.2708629369735718,
      "learning_rate": 9.981614897424046e-06,
      "loss": 0.8514,
      "step": 1189
    },
    {
      "epoch": 0.16975748930099857,
      "grad_norm": 0.6136476397514343,
      "learning_rate": 9.981548864101864e-06,
      "loss": 0.8612,
      "step": 1190
    },
    {
      "epoch": 0.16990014265335235,
      "grad_norm": 0.7049573659896851,
      "learning_rate": 9.98148271262624e-06,
      "loss": 0.8801,
      "step": 1191
    },
    {
      "epoch": 0.17004279600570613,
      "grad_norm": 0.8933731317520142,
      "learning_rate": 9.98141644299874e-06,
      "loss": 0.8331,
      "step": 1192
    },
    {
      "epoch": 0.1701854493580599,
      "grad_norm": 0.5952308177947998,
      "learning_rate": 9.98135005522094e-06,
      "loss": 0.8901,
      "step": 1193
    },
    {
      "epoch": 0.17032810271041368,
      "grad_norm": 1.0774075984954834,
      "learning_rate": 9.981283549294411e-06,
      "loss": 0.9354,
      "step": 1194
    },
    {
      "epoch": 0.1704707560627675,
      "grad_norm": 0.8799782991409302,
      "learning_rate": 9.981216925220731e-06,
      "loss": 0.8744,
      "step": 1195
    },
    {
      "epoch": 0.17061340941512126,
      "grad_norm": 0.7206964492797852,
      "learning_rate": 9.98115018300148e-06,
      "loss": 0.8734,
      "step": 1196
    },
    {
      "epoch": 0.17075606276747504,
      "grad_norm": 0.7009869813919067,
      "learning_rate": 9.981083322638243e-06,
      "loss": 0.8943,
      "step": 1197
    },
    {
      "epoch": 0.17089871611982882,
      "grad_norm": 0.7496351599693298,
      "learning_rate": 9.981016344132604e-06,
      "loss": 0.8585,
      "step": 1198
    },
    {
      "epoch": 0.1710413694721826,
      "grad_norm": 0.9334327578544617,
      "learning_rate": 9.98094924748615e-06,
      "loss": 0.8503,
      "step": 1199
    },
    {
      "epoch": 0.17118402282453637,
      "grad_norm": 0.7665506601333618,
      "learning_rate": 9.980882032700476e-06,
      "loss": 0.8383,
      "step": 1200
    },
    {
      "epoch": 0.17132667617689015,
      "grad_norm": 0.7536643147468567,
      "learning_rate": 9.980814699777173e-06,
      "loss": 0.882,
      "step": 1201
    },
    {
      "epoch": 0.17146932952924393,
      "grad_norm": 0.7478814125061035,
      "learning_rate": 9.980747248717841e-06,
      "loss": 0.8675,
      "step": 1202
    },
    {
      "epoch": 0.1716119828815977,
      "grad_norm": 0.7154534459114075,
      "learning_rate": 9.98067967952408e-06,
      "loss": 0.8799,
      "step": 1203
    },
    {
      "epoch": 0.17175463623395149,
      "grad_norm": 0.6845526695251465,
      "learning_rate": 9.980611992197488e-06,
      "loss": 0.8574,
      "step": 1204
    },
    {
      "epoch": 0.1718972895863053,
      "grad_norm": 0.7099165916442871,
      "learning_rate": 9.980544186739674e-06,
      "loss": 0.8832,
      "step": 1205
    },
    {
      "epoch": 0.17203994293865907,
      "grad_norm": 0.7951145172119141,
      "learning_rate": 9.980476263152249e-06,
      "loss": 0.845,
      "step": 1206
    },
    {
      "epoch": 0.17218259629101285,
      "grad_norm": 0.8383887410163879,
      "learning_rate": 9.980408221436817e-06,
      "loss": 0.782,
      "step": 1207
    },
    {
      "epoch": 0.17232524964336662,
      "grad_norm": 0.7200697660446167,
      "learning_rate": 9.980340061594995e-06,
      "loss": 0.8029,
      "step": 1208
    },
    {
      "epoch": 0.1724679029957204,
      "grad_norm": 0.5835481882095337,
      "learning_rate": 9.980271783628402e-06,
      "loss": 0.8718,
      "step": 1209
    },
    {
      "epoch": 0.17261055634807418,
      "grad_norm": 0.6431670188903809,
      "learning_rate": 9.980203387538656e-06,
      "loss": 0.867,
      "step": 1210
    },
    {
      "epoch": 0.17275320970042796,
      "grad_norm": 1.0457934141159058,
      "learning_rate": 9.980134873327378e-06,
      "loss": 0.8991,
      "step": 1211
    },
    {
      "epoch": 0.17289586305278173,
      "grad_norm": 0.7832580208778381,
      "learning_rate": 9.980066240996193e-06,
      "loss": 0.8591,
      "step": 1212
    },
    {
      "epoch": 0.1730385164051355,
      "grad_norm": 0.7906302809715271,
      "learning_rate": 9.979997490546732e-06,
      "loss": 0.8512,
      "step": 1213
    },
    {
      "epoch": 0.1731811697574893,
      "grad_norm": 1.0404938459396362,
      "learning_rate": 9.97992862198062e-06,
      "loss": 0.7457,
      "step": 1214
    },
    {
      "epoch": 0.1733238231098431,
      "grad_norm": 1.0030831098556519,
      "learning_rate": 9.979859635299498e-06,
      "loss": 0.847,
      "step": 1215
    },
    {
      "epoch": 0.17346647646219687,
      "grad_norm": 0.7839489579200745,
      "learning_rate": 9.979790530504995e-06,
      "loss": 0.8379,
      "step": 1216
    },
    {
      "epoch": 0.17360912981455065,
      "grad_norm": 0.9872682690620422,
      "learning_rate": 9.979721307598754e-06,
      "loss": 0.9056,
      "step": 1217
    },
    {
      "epoch": 0.17375178316690443,
      "grad_norm": 0.8990773558616638,
      "learning_rate": 9.979651966582415e-06,
      "loss": 0.7916,
      "step": 1218
    },
    {
      "epoch": 0.1738944365192582,
      "grad_norm": 0.7062201499938965,
      "learning_rate": 9.979582507457626e-06,
      "loss": 0.8749,
      "step": 1219
    },
    {
      "epoch": 0.17403708987161198,
      "grad_norm": 1.092830777168274,
      "learning_rate": 9.979512930226029e-06,
      "loss": 0.7154,
      "step": 1220
    },
    {
      "epoch": 0.17417974322396576,
      "grad_norm": 0.804801881313324,
      "learning_rate": 9.97944323488928e-06,
      "loss": 0.8728,
      "step": 1221
    },
    {
      "epoch": 0.17432239657631954,
      "grad_norm": 0.8164383172988892,
      "learning_rate": 9.979373421449026e-06,
      "loss": 0.8472,
      "step": 1222
    },
    {
      "epoch": 0.17446504992867332,
      "grad_norm": 1.0955476760864258,
      "learning_rate": 9.979303489906929e-06,
      "loss": 0.9077,
      "step": 1223
    },
    {
      "epoch": 0.1746077032810271,
      "grad_norm": 0.8254255652427673,
      "learning_rate": 9.979233440264644e-06,
      "loss": 0.8198,
      "step": 1224
    },
    {
      "epoch": 0.17475035663338087,
      "grad_norm": 1.022617220878601,
      "learning_rate": 9.979163272523833e-06,
      "loss": 0.7953,
      "step": 1225
    },
    {
      "epoch": 0.17489300998573468,
      "grad_norm": 0.8240652680397034,
      "learning_rate": 9.97909298668616e-06,
      "loss": 0.8761,
      "step": 1226
    },
    {
      "epoch": 0.17503566333808845,
      "grad_norm": 0.6608521342277527,
      "learning_rate": 9.979022582753293e-06,
      "loss": 0.89,
      "step": 1227
    },
    {
      "epoch": 0.17517831669044223,
      "grad_norm": 0.7590407133102417,
      "learning_rate": 9.978952060726901e-06,
      "loss": 0.8207,
      "step": 1228
    },
    {
      "epoch": 0.175320970042796,
      "grad_norm": 1.3209112882614136,
      "learning_rate": 9.978881420608657e-06,
      "loss": 0.891,
      "step": 1229
    },
    {
      "epoch": 0.1754636233951498,
      "grad_norm": 0.5832023620605469,
      "learning_rate": 9.978810662400238e-06,
      "loss": 0.8418,
      "step": 1230
    },
    {
      "epoch": 0.17560627674750356,
      "grad_norm": 0.7879243493080139,
      "learning_rate": 9.97873978610332e-06,
      "loss": 0.7728,
      "step": 1231
    },
    {
      "epoch": 0.17574893009985734,
      "grad_norm": 1.1560922861099243,
      "learning_rate": 9.978668791719584e-06,
      "loss": 0.8481,
      "step": 1232
    },
    {
      "epoch": 0.17589158345221112,
      "grad_norm": 0.8897849321365356,
      "learning_rate": 9.978597679250716e-06,
      "loss": 0.7139,
      "step": 1233
    },
    {
      "epoch": 0.1760342368045649,
      "grad_norm": 0.9120283722877502,
      "learning_rate": 9.9785264486984e-06,
      "loss": 0.8597,
      "step": 1234
    },
    {
      "epoch": 0.17617689015691868,
      "grad_norm": 0.8726806640625,
      "learning_rate": 9.978455100064328e-06,
      "loss": 0.8664,
      "step": 1235
    },
    {
      "epoch": 0.17631954350927248,
      "grad_norm": 0.7392376661300659,
      "learning_rate": 9.97838363335019e-06,
      "loss": 0.8109,
      "step": 1236
    },
    {
      "epoch": 0.17646219686162626,
      "grad_norm": 0.8633730411529541,
      "learning_rate": 9.978312048557683e-06,
      "loss": 0.7605,
      "step": 1237
    },
    {
      "epoch": 0.17660485021398004,
      "grad_norm": 0.8165681958198547,
      "learning_rate": 9.978240345688503e-06,
      "loss": 0.9173,
      "step": 1238
    },
    {
      "epoch": 0.1767475035663338,
      "grad_norm": 0.9438103437423706,
      "learning_rate": 9.978168524744352e-06,
      "loss": 0.8134,
      "step": 1239
    },
    {
      "epoch": 0.1768901569186876,
      "grad_norm": 0.758353054523468,
      "learning_rate": 9.978096585726935e-06,
      "loss": 0.8101,
      "step": 1240
    },
    {
      "epoch": 0.17703281027104137,
      "grad_norm": 0.5737704634666443,
      "learning_rate": 9.978024528637954e-06,
      "loss": 0.8293,
      "step": 1241
    },
    {
      "epoch": 0.17717546362339515,
      "grad_norm": 0.6982561945915222,
      "learning_rate": 9.977952353479121e-06,
      "loss": 0.883,
      "step": 1242
    },
    {
      "epoch": 0.17731811697574892,
      "grad_norm": 0.7997564673423767,
      "learning_rate": 9.977880060252147e-06,
      "loss": 0.8539,
      "step": 1243
    },
    {
      "epoch": 0.1774607703281027,
      "grad_norm": 1.4520130157470703,
      "learning_rate": 9.977807648958748e-06,
      "loss": 0.848,
      "step": 1244
    },
    {
      "epoch": 0.17760342368045648,
      "grad_norm": 0.6465441584587097,
      "learning_rate": 9.97773511960064e-06,
      "loss": 0.8627,
      "step": 1245
    },
    {
      "epoch": 0.17774607703281028,
      "grad_norm": 0.7276219129562378,
      "learning_rate": 9.977662472179545e-06,
      "loss": 0.8171,
      "step": 1246
    },
    {
      "epoch": 0.17788873038516406,
      "grad_norm": 0.6480956673622131,
      "learning_rate": 9.977589706697183e-06,
      "loss": 0.8385,
      "step": 1247
    },
    {
      "epoch": 0.17803138373751784,
      "grad_norm": 0.8526870608329773,
      "learning_rate": 9.977516823155283e-06,
      "loss": 0.7951,
      "step": 1248
    },
    {
      "epoch": 0.17817403708987162,
      "grad_norm": 0.8076682686805725,
      "learning_rate": 9.977443821555571e-06,
      "loss": 0.793,
      "step": 1249
    },
    {
      "epoch": 0.1783166904422254,
      "grad_norm": 0.9861093759536743,
      "learning_rate": 9.977370701899781e-06,
      "loss": 0.793,
      "step": 1250
    },
    {
      "epoch": 0.17845934379457917,
      "grad_norm": 0.989619791507721,
      "learning_rate": 9.977297464189646e-06,
      "loss": 0.8507,
      "step": 1251
    },
    {
      "epoch": 0.17860199714693295,
      "grad_norm": 0.8125431537628174,
      "learning_rate": 9.977224108426901e-06,
      "loss": 0.8499,
      "step": 1252
    },
    {
      "epoch": 0.17874465049928673,
      "grad_norm": 0.8821991086006165,
      "learning_rate": 9.977150634613291e-06,
      "loss": 0.8917,
      "step": 1253
    },
    {
      "epoch": 0.1788873038516405,
      "grad_norm": 0.7301605939865112,
      "learning_rate": 9.977077042750554e-06,
      "loss": 0.88,
      "step": 1254
    },
    {
      "epoch": 0.17902995720399428,
      "grad_norm": 0.7050444483757019,
      "learning_rate": 9.977003332840437e-06,
      "loss": 0.8706,
      "step": 1255
    },
    {
      "epoch": 0.17917261055634806,
      "grad_norm": 0.873283863067627,
      "learning_rate": 9.97692950488469e-06,
      "loss": 0.7479,
      "step": 1256
    },
    {
      "epoch": 0.17931526390870187,
      "grad_norm": 0.8825015425682068,
      "learning_rate": 9.97685555888506e-06,
      "loss": 0.727,
      "step": 1257
    },
    {
      "epoch": 0.17945791726105564,
      "grad_norm": 0.7452796101570129,
      "learning_rate": 9.976781494843306e-06,
      "loss": 0.6812,
      "step": 1258
    },
    {
      "epoch": 0.17960057061340942,
      "grad_norm": 1.0616790056228638,
      "learning_rate": 9.976707312761181e-06,
      "loss": 0.8564,
      "step": 1259
    },
    {
      "epoch": 0.1797432239657632,
      "grad_norm": 0.7580656409263611,
      "learning_rate": 9.976633012640444e-06,
      "loss": 0.8373,
      "step": 1260
    },
    {
      "epoch": 0.17988587731811698,
      "grad_norm": 0.6261967420578003,
      "learning_rate": 9.97655859448286e-06,
      "loss": 0.8865,
      "step": 1261
    },
    {
      "epoch": 0.18002853067047075,
      "grad_norm": 0.9243199825286865,
      "learning_rate": 9.976484058290191e-06,
      "loss": 0.825,
      "step": 1262
    },
    {
      "epoch": 0.18017118402282453,
      "grad_norm": 0.8551122546195984,
      "learning_rate": 9.97640940406421e-06,
      "loss": 0.8567,
      "step": 1263
    },
    {
      "epoch": 0.1803138373751783,
      "grad_norm": 0.8754786252975464,
      "learning_rate": 9.976334631806684e-06,
      "loss": 0.8059,
      "step": 1264
    },
    {
      "epoch": 0.1804564907275321,
      "grad_norm": 0.8937332630157471,
      "learning_rate": 9.976259741519385e-06,
      "loss": 0.899,
      "step": 1265
    },
    {
      "epoch": 0.18059914407988586,
      "grad_norm": 0.9022212624549866,
      "learning_rate": 9.97618473320409e-06,
      "loss": 0.8596,
      "step": 1266
    },
    {
      "epoch": 0.18074179743223967,
      "grad_norm": 0.7678316831588745,
      "learning_rate": 9.976109606862581e-06,
      "loss": 0.8117,
      "step": 1267
    },
    {
      "epoch": 0.18088445078459345,
      "grad_norm": 0.8664880990982056,
      "learning_rate": 9.976034362496639e-06,
      "loss": 0.903,
      "step": 1268
    },
    {
      "epoch": 0.18102710413694723,
      "grad_norm": 0.9873309135437012,
      "learning_rate": 9.975959000108046e-06,
      "loss": 0.8767,
      "step": 1269
    },
    {
      "epoch": 0.181169757489301,
      "grad_norm": 0.983790934085846,
      "learning_rate": 9.975883519698592e-06,
      "loss": 0.8354,
      "step": 1270
    },
    {
      "epoch": 0.18131241084165478,
      "grad_norm": 0.7086114287376404,
      "learning_rate": 9.975807921270063e-06,
      "loss": 0.9251,
      "step": 1271
    },
    {
      "epoch": 0.18145506419400856,
      "grad_norm": 0.8843178153038025,
      "learning_rate": 9.97573220482426e-06,
      "loss": 0.8446,
      "step": 1272
    },
    {
      "epoch": 0.18159771754636234,
      "grad_norm": 0.95987868309021,
      "learning_rate": 9.975656370362972e-06,
      "loss": 0.875,
      "step": 1273
    },
    {
      "epoch": 0.1817403708987161,
      "grad_norm": 0.6299657821655273,
      "learning_rate": 9.975580417888e-06,
      "loss": 0.8612,
      "step": 1274
    },
    {
      "epoch": 0.1818830242510699,
      "grad_norm": 0.8124523758888245,
      "learning_rate": 9.975504347401147e-06,
      "loss": 0.7919,
      "step": 1275
    },
    {
      "epoch": 0.18202567760342367,
      "grad_norm": 0.6986494660377502,
      "learning_rate": 9.975428158904215e-06,
      "loss": 0.7638,
      "step": 1276
    },
    {
      "epoch": 0.18216833095577747,
      "grad_norm": 0.7109574675559998,
      "learning_rate": 9.975351852399011e-06,
      "loss": 0.7871,
      "step": 1277
    },
    {
      "epoch": 0.18231098430813125,
      "grad_norm": 0.7800731062889099,
      "learning_rate": 9.975275427887346e-06,
      "loss": 0.7934,
      "step": 1278
    },
    {
      "epoch": 0.18245363766048503,
      "grad_norm": 0.9133411645889282,
      "learning_rate": 9.97519888537103e-06,
      "loss": 0.8887,
      "step": 1279
    },
    {
      "epoch": 0.1825962910128388,
      "grad_norm": 0.7395647168159485,
      "learning_rate": 9.975122224851884e-06,
      "loss": 0.8354,
      "step": 1280
    },
    {
      "epoch": 0.18273894436519258,
      "grad_norm": 1.1085537672042847,
      "learning_rate": 9.975045446331723e-06,
      "loss": 0.8239,
      "step": 1281
    },
    {
      "epoch": 0.18288159771754636,
      "grad_norm": 0.7853597402572632,
      "learning_rate": 9.974968549812367e-06,
      "loss": 0.8743,
      "step": 1282
    },
    {
      "epoch": 0.18302425106990014,
      "grad_norm": 0.9312048554420471,
      "learning_rate": 9.97489153529564e-06,
      "loss": 0.8069,
      "step": 1283
    },
    {
      "epoch": 0.18316690442225392,
      "grad_norm": 0.7590904235839844,
      "learning_rate": 9.97481440278337e-06,
      "loss": 0.7983,
      "step": 1284
    },
    {
      "epoch": 0.1833095577746077,
      "grad_norm": 0.7187520861625671,
      "learning_rate": 9.974737152277385e-06,
      "loss": 0.8311,
      "step": 1285
    },
    {
      "epoch": 0.18345221112696147,
      "grad_norm": 0.6758371591567993,
      "learning_rate": 9.97465978377952e-06,
      "loss": 0.8674,
      "step": 1286
    },
    {
      "epoch": 0.18359486447931525,
      "grad_norm": 0.8807397484779358,
      "learning_rate": 9.974582297291607e-06,
      "loss": 0.941,
      "step": 1287
    },
    {
      "epoch": 0.18373751783166906,
      "grad_norm": 0.850386381149292,
      "learning_rate": 9.974504692815485e-06,
      "loss": 0.7757,
      "step": 1288
    },
    {
      "epoch": 0.18388017118402283,
      "grad_norm": 0.9141903519630432,
      "learning_rate": 9.974426970352996e-06,
      "loss": 0.8307,
      "step": 1289
    },
    {
      "epoch": 0.1840228245363766,
      "grad_norm": 0.645063579082489,
      "learning_rate": 9.97434912990598e-06,
      "loss": 0.937,
      "step": 1290
    },
    {
      "epoch": 0.1841654778887304,
      "grad_norm": 0.5873847603797913,
      "learning_rate": 9.974271171476287e-06,
      "loss": 0.9047,
      "step": 1291
    },
    {
      "epoch": 0.18430813124108417,
      "grad_norm": 0.9669590592384338,
      "learning_rate": 9.974193095065763e-06,
      "loss": 0.9448,
      "step": 1292
    },
    {
      "epoch": 0.18445078459343794,
      "grad_norm": 0.9505781531333923,
      "learning_rate": 9.974114900676262e-06,
      "loss": 0.9263,
      "step": 1293
    },
    {
      "epoch": 0.18459343794579172,
      "grad_norm": 0.8218867778778076,
      "learning_rate": 9.974036588309639e-06,
      "loss": 0.8104,
      "step": 1294
    },
    {
      "epoch": 0.1847360912981455,
      "grad_norm": 0.7694748044013977,
      "learning_rate": 9.97395815796775e-06,
      "loss": 0.8938,
      "step": 1295
    },
    {
      "epoch": 0.18487874465049928,
      "grad_norm": 0.6638206839561462,
      "learning_rate": 9.973879609652452e-06,
      "loss": 0.9507,
      "step": 1296
    },
    {
      "epoch": 0.18502139800285305,
      "grad_norm": 0.8577884435653687,
      "learning_rate": 9.973800943365616e-06,
      "loss": 0.9323,
      "step": 1297
    },
    {
      "epoch": 0.18516405135520686,
      "grad_norm": 0.6918472647666931,
      "learning_rate": 9.9737221591091e-06,
      "loss": 0.8634,
      "step": 1298
    },
    {
      "epoch": 0.18530670470756064,
      "grad_norm": 1.036751389503479,
      "learning_rate": 9.97364325688478e-06,
      "loss": 0.8545,
      "step": 1299
    },
    {
      "epoch": 0.18544935805991442,
      "grad_norm": 0.6787410378456116,
      "learning_rate": 9.97356423669452e-06,
      "loss": 0.7864,
      "step": 1300
    },
    {
      "epoch": 0.1855920114122682,
      "grad_norm": 0.6573331952095032,
      "learning_rate": 9.973485098540199e-06,
      "loss": 0.8529,
      "step": 1301
    },
    {
      "epoch": 0.18573466476462197,
      "grad_norm": 1.0073238611221313,
      "learning_rate": 9.973405842423691e-06,
      "loss": 0.7799,
      "step": 1302
    },
    {
      "epoch": 0.18587731811697575,
      "grad_norm": 0.9594725370407104,
      "learning_rate": 9.97332646834688e-06,
      "loss": 0.8208,
      "step": 1303
    },
    {
      "epoch": 0.18601997146932953,
      "grad_norm": 0.7488198280334473,
      "learning_rate": 9.973246976311643e-06,
      "loss": 0.8644,
      "step": 1304
    },
    {
      "epoch": 0.1861626248216833,
      "grad_norm": 0.7605316638946533,
      "learning_rate": 9.973167366319872e-06,
      "loss": 0.8014,
      "step": 1305
    },
    {
      "epoch": 0.18630527817403708,
      "grad_norm": 0.9765550494194031,
      "learning_rate": 9.973087638373449e-06,
      "loss": 0.8,
      "step": 1306
    },
    {
      "epoch": 0.18644793152639086,
      "grad_norm": 0.945305347442627,
      "learning_rate": 9.97300779247427e-06,
      "loss": 0.8289,
      "step": 1307
    },
    {
      "epoch": 0.18659058487874466,
      "grad_norm": 1.0048795938491821,
      "learning_rate": 9.972927828624224e-06,
      "loss": 0.8447,
      "step": 1308
    },
    {
      "epoch": 0.18673323823109844,
      "grad_norm": 0.7458885312080383,
      "learning_rate": 9.972847746825211e-06,
      "loss": 0.6681,
      "step": 1309
    },
    {
      "epoch": 0.18687589158345222,
      "grad_norm": 0.7100270986557007,
      "learning_rate": 9.97276754707913e-06,
      "loss": 0.8355,
      "step": 1310
    },
    {
      "epoch": 0.187018544935806,
      "grad_norm": 0.8413822650909424,
      "learning_rate": 9.972687229387882e-06,
      "loss": 0.8886,
      "step": 1311
    },
    {
      "epoch": 0.18716119828815977,
      "grad_norm": 0.8513115048408508,
      "learning_rate": 9.972606793753374e-06,
      "loss": 0.7423,
      "step": 1312
    },
    {
      "epoch": 0.18730385164051355,
      "grad_norm": 0.6329718232154846,
      "learning_rate": 9.972526240177511e-06,
      "loss": 0.8155,
      "step": 1313
    },
    {
      "epoch": 0.18744650499286733,
      "grad_norm": 0.8879488706588745,
      "learning_rate": 9.972445568662206e-06,
      "loss": 0.8847,
      "step": 1314
    },
    {
      "epoch": 0.1875891583452211,
      "grad_norm": 0.7198883891105652,
      "learning_rate": 9.972364779209372e-06,
      "loss": 0.6882,
      "step": 1315
    },
    {
      "epoch": 0.18773181169757489,
      "grad_norm": 0.6940578818321228,
      "learning_rate": 9.972283871820922e-06,
      "loss": 0.8243,
      "step": 1316
    },
    {
      "epoch": 0.18787446504992866,
      "grad_norm": 0.7445089221000671,
      "learning_rate": 9.972202846498779e-06,
      "loss": 0.9295,
      "step": 1317
    },
    {
      "epoch": 0.18801711840228244,
      "grad_norm": 0.7449865937232971,
      "learning_rate": 9.972121703244864e-06,
      "loss": 0.8018,
      "step": 1318
    },
    {
      "epoch": 0.18815977175463625,
      "grad_norm": 0.7003435492515564,
      "learning_rate": 9.972040442061101e-06,
      "loss": 0.8368,
      "step": 1319
    },
    {
      "epoch": 0.18830242510699002,
      "grad_norm": 0.8622342944145203,
      "learning_rate": 9.971959062949417e-06,
      "loss": 0.5611,
      "step": 1320
    },
    {
      "epoch": 0.1884450784593438,
      "grad_norm": 0.7828042507171631,
      "learning_rate": 9.971877565911742e-06,
      "loss": 0.8166,
      "step": 1321
    },
    {
      "epoch": 0.18858773181169758,
      "grad_norm": 0.7658116817474365,
      "learning_rate": 9.97179595095001e-06,
      "loss": 0.8256,
      "step": 1322
    },
    {
      "epoch": 0.18873038516405136,
      "grad_norm": 0.892788290977478,
      "learning_rate": 9.971714218066155e-06,
      "loss": 0.7364,
      "step": 1323
    },
    {
      "epoch": 0.18887303851640513,
      "grad_norm": 0.8484310507774353,
      "learning_rate": 9.97163236726212e-06,
      "loss": 0.7845,
      "step": 1324
    },
    {
      "epoch": 0.1890156918687589,
      "grad_norm": 0.6448402404785156,
      "learning_rate": 9.97155039853984e-06,
      "loss": 0.7648,
      "step": 1325
    },
    {
      "epoch": 0.1891583452211127,
      "grad_norm": 0.6692468523979187,
      "learning_rate": 9.971468311901262e-06,
      "loss": 0.8491,
      "step": 1326
    },
    {
      "epoch": 0.18930099857346647,
      "grad_norm": 0.6772980690002441,
      "learning_rate": 9.971386107348333e-06,
      "loss": 0.873,
      "step": 1327
    },
    {
      "epoch": 0.18944365192582024,
      "grad_norm": 0.790951669216156,
      "learning_rate": 9.971303784883006e-06,
      "loss": 0.8977,
      "step": 1328
    },
    {
      "epoch": 0.18958630527817405,
      "grad_norm": 0.7230355739593506,
      "learning_rate": 9.971221344507229e-06,
      "loss": 0.8403,
      "step": 1329
    },
    {
      "epoch": 0.18972895863052783,
      "grad_norm": 0.8040663003921509,
      "learning_rate": 9.971138786222959e-06,
      "loss": 0.8679,
      "step": 1330
    },
    {
      "epoch": 0.1898716119828816,
      "grad_norm": 0.8062121272087097,
      "learning_rate": 9.971056110032153e-06,
      "loss": 0.887,
      "step": 1331
    },
    {
      "epoch": 0.19001426533523538,
      "grad_norm": 0.7967968583106995,
      "learning_rate": 9.970973315936773e-06,
      "loss": 0.8603,
      "step": 1332
    },
    {
      "epoch": 0.19015691868758916,
      "grad_norm": 0.7940622568130493,
      "learning_rate": 9.970890403938783e-06,
      "loss": 0.8229,
      "step": 1333
    },
    {
      "epoch": 0.19029957203994294,
      "grad_norm": 0.7876176238059998,
      "learning_rate": 9.97080737404015e-06,
      "loss": 0.8662,
      "step": 1334
    },
    {
      "epoch": 0.19044222539229672,
      "grad_norm": 0.7090651988983154,
      "learning_rate": 9.970724226242842e-06,
      "loss": 0.7787,
      "step": 1335
    },
    {
      "epoch": 0.1905848787446505,
      "grad_norm": 0.8402417898178101,
      "learning_rate": 9.97064096054883e-06,
      "loss": 0.9188,
      "step": 1336
    },
    {
      "epoch": 0.19072753209700427,
      "grad_norm": 0.7670844197273254,
      "learning_rate": 9.970557576960093e-06,
      "loss": 0.8744,
      "step": 1337
    },
    {
      "epoch": 0.19087018544935805,
      "grad_norm": 0.8511255383491516,
      "learning_rate": 9.970474075478604e-06,
      "loss": 0.7396,
      "step": 1338
    },
    {
      "epoch": 0.19101283880171185,
      "grad_norm": 0.7267671227455139,
      "learning_rate": 9.970390456106349e-06,
      "loss": 0.8837,
      "step": 1339
    },
    {
      "epoch": 0.19115549215406563,
      "grad_norm": 0.8868377804756165,
      "learning_rate": 9.970306718845303e-06,
      "loss": 0.8495,
      "step": 1340
    },
    {
      "epoch": 0.1912981455064194,
      "grad_norm": 0.7415902018547058,
      "learning_rate": 9.970222863697462e-06,
      "loss": 0.8479,
      "step": 1341
    },
    {
      "epoch": 0.1914407988587732,
      "grad_norm": 0.8380827903747559,
      "learning_rate": 9.970138890664807e-06,
      "loss": 0.9269,
      "step": 1342
    },
    {
      "epoch": 0.19158345221112696,
      "grad_norm": 0.7889651656150818,
      "learning_rate": 9.970054799749334e-06,
      "loss": 0.8834,
      "step": 1343
    },
    {
      "epoch": 0.19172610556348074,
      "grad_norm": 0.8100181818008423,
      "learning_rate": 9.969970590953036e-06,
      "loss": 0.7753,
      "step": 1344
    },
    {
      "epoch": 0.19186875891583452,
      "grad_norm": 0.7502317428588867,
      "learning_rate": 9.969886264277911e-06,
      "loss": 0.8223,
      "step": 1345
    },
    {
      "epoch": 0.1920114122681883,
      "grad_norm": 0.6847222447395325,
      "learning_rate": 9.969801819725959e-06,
      "loss": 0.8619,
      "step": 1346
    },
    {
      "epoch": 0.19215406562054207,
      "grad_norm": 0.7199435830116272,
      "learning_rate": 9.96971725729918e-06,
      "loss": 0.8235,
      "step": 1347
    },
    {
      "epoch": 0.19229671897289585,
      "grad_norm": 0.9098808765411377,
      "learning_rate": 9.969632576999583e-06,
      "loss": 0.8578,
      "step": 1348
    },
    {
      "epoch": 0.19243937232524963,
      "grad_norm": 0.9622191786766052,
      "learning_rate": 9.969547778829177e-06,
      "loss": 0.8435,
      "step": 1349
    },
    {
      "epoch": 0.19258202567760344,
      "grad_norm": 1.048285961151123,
      "learning_rate": 9.969462862789972e-06,
      "loss": 0.8485,
      "step": 1350
    },
    {
      "epoch": 0.1927246790299572,
      "grad_norm": 0.9849112629890442,
      "learning_rate": 9.969377828883979e-06,
      "loss": 0.8363,
      "step": 1351
    },
    {
      "epoch": 0.192867332382311,
      "grad_norm": 0.828313410282135,
      "learning_rate": 9.96929267711322e-06,
      "loss": 0.7776,
      "step": 1352
    },
    {
      "epoch": 0.19300998573466477,
      "grad_norm": 0.676027774810791,
      "learning_rate": 9.969207407479713e-06,
      "loss": 0.8087,
      "step": 1353
    },
    {
      "epoch": 0.19315263908701855,
      "grad_norm": 0.7559381723403931,
      "learning_rate": 9.969122019985477e-06,
      "loss": 0.807,
      "step": 1354
    },
    {
      "epoch": 0.19329529243937232,
      "grad_norm": 0.8531244397163391,
      "learning_rate": 9.969036514632544e-06,
      "loss": 0.7793,
      "step": 1355
    },
    {
      "epoch": 0.1934379457917261,
      "grad_norm": 0.7253252267837524,
      "learning_rate": 9.968950891422934e-06,
      "loss": 0.9126,
      "step": 1356
    },
    {
      "epoch": 0.19358059914407988,
      "grad_norm": 0.6927995085716248,
      "learning_rate": 9.968865150358685e-06,
      "loss": 0.897,
      "step": 1357
    },
    {
      "epoch": 0.19372325249643366,
      "grad_norm": 1.0707788467407227,
      "learning_rate": 9.968779291441827e-06,
      "loss": 0.8988,
      "step": 1358
    },
    {
      "epoch": 0.19386590584878743,
      "grad_norm": 1.0267117023468018,
      "learning_rate": 9.968693314674396e-06,
      "loss": 0.8216,
      "step": 1359
    },
    {
      "epoch": 0.19400855920114124,
      "grad_norm": 0.8915116190910339,
      "learning_rate": 9.968607220058434e-06,
      "loss": 0.9453,
      "step": 1360
    },
    {
      "epoch": 0.19415121255349502,
      "grad_norm": 0.7114331126213074,
      "learning_rate": 9.96852100759598e-06,
      "loss": 0.8684,
      "step": 1361
    },
    {
      "epoch": 0.1942938659058488,
      "grad_norm": 0.5982599854469299,
      "learning_rate": 9.968434677289078e-06,
      "loss": 0.9415,
      "step": 1362
    },
    {
      "epoch": 0.19443651925820257,
      "grad_norm": 0.7128475308418274,
      "learning_rate": 9.96834822913978e-06,
      "loss": 0.8216,
      "step": 1363
    },
    {
      "epoch": 0.19457917261055635,
      "grad_norm": 0.7566629648208618,
      "learning_rate": 9.968261663150133e-06,
      "loss": 0.8028,
      "step": 1364
    },
    {
      "epoch": 0.19472182596291013,
      "grad_norm": 0.9583125710487366,
      "learning_rate": 9.968174979322193e-06,
      "loss": 0.8147,
      "step": 1365
    },
    {
      "epoch": 0.1948644793152639,
      "grad_norm": 0.9913781881332397,
      "learning_rate": 9.968088177658013e-06,
      "loss": 0.8716,
      "step": 1366
    },
    {
      "epoch": 0.19500713266761768,
      "grad_norm": 0.7165933847427368,
      "learning_rate": 9.968001258159652e-06,
      "loss": 0.8776,
      "step": 1367
    },
    {
      "epoch": 0.19514978601997146,
      "grad_norm": 0.5812830328941345,
      "learning_rate": 9.967914220829173e-06,
      "loss": 0.8651,
      "step": 1368
    },
    {
      "epoch": 0.19529243937232524,
      "grad_norm": 0.7699722051620483,
      "learning_rate": 9.96782706566864e-06,
      "loss": 0.8094,
      "step": 1369
    },
    {
      "epoch": 0.19543509272467904,
      "grad_norm": 0.9172202348709106,
      "learning_rate": 9.967739792680122e-06,
      "loss": 0.9267,
      "step": 1370
    },
    {
      "epoch": 0.19557774607703282,
      "grad_norm": 0.6696454882621765,
      "learning_rate": 9.967652401865685e-06,
      "loss": 0.8736,
      "step": 1371
    },
    {
      "epoch": 0.1957203994293866,
      "grad_norm": 0.7624047994613647,
      "learning_rate": 9.967564893227404e-06,
      "loss": 0.7729,
      "step": 1372
    },
    {
      "epoch": 0.19586305278174038,
      "grad_norm": 0.6636433601379395,
      "learning_rate": 9.967477266767354e-06,
      "loss": 0.9278,
      "step": 1373
    },
    {
      "epoch": 0.19600570613409415,
      "grad_norm": 0.969912052154541,
      "learning_rate": 9.967389522487615e-06,
      "loss": 0.8405,
      "step": 1374
    },
    {
      "epoch": 0.19614835948644793,
      "grad_norm": 0.5688678026199341,
      "learning_rate": 9.967301660390267e-06,
      "loss": 0.8659,
      "step": 1375
    },
    {
      "epoch": 0.1962910128388017,
      "grad_norm": 1.0910042524337769,
      "learning_rate": 9.967213680477392e-06,
      "loss": 0.915,
      "step": 1376
    },
    {
      "epoch": 0.1964336661911555,
      "grad_norm": 0.6769224405288696,
      "learning_rate": 9.967125582751078e-06,
      "loss": 0.8701,
      "step": 1377
    },
    {
      "epoch": 0.19657631954350926,
      "grad_norm": 0.7530673146247864,
      "learning_rate": 9.967037367213417e-06,
      "loss": 0.8574,
      "step": 1378
    },
    {
      "epoch": 0.19671897289586304,
      "grad_norm": 0.9435679912567139,
      "learning_rate": 9.966949033866498e-06,
      "loss": 0.7807,
      "step": 1379
    },
    {
      "epoch": 0.19686162624821682,
      "grad_norm": 0.7873449921607971,
      "learning_rate": 9.966860582712418e-06,
      "loss": 0.835,
      "step": 1380
    },
    {
      "epoch": 0.19700427960057063,
      "grad_norm": 0.8070048093795776,
      "learning_rate": 9.966772013753275e-06,
      "loss": 0.8098,
      "step": 1381
    },
    {
      "epoch": 0.1971469329529244,
      "grad_norm": 0.7220003604888916,
      "learning_rate": 9.96668332699117e-06,
      "loss": 0.8181,
      "step": 1382
    },
    {
      "epoch": 0.19728958630527818,
      "grad_norm": 1.0374494791030884,
      "learning_rate": 9.966594522428203e-06,
      "loss": 0.8704,
      "step": 1383
    },
    {
      "epoch": 0.19743223965763196,
      "grad_norm": 0.9598643183708191,
      "learning_rate": 9.966505600066482e-06,
      "loss": 0.6479,
      "step": 1384
    },
    {
      "epoch": 0.19757489300998574,
      "grad_norm": 0.7491325736045837,
      "learning_rate": 9.96641655990812e-06,
      "loss": 0.7976,
      "step": 1385
    },
    {
      "epoch": 0.1977175463623395,
      "grad_norm": 0.8149396777153015,
      "learning_rate": 9.966327401955225e-06,
      "loss": 0.6355,
      "step": 1386
    },
    {
      "epoch": 0.1978601997146933,
      "grad_norm": 1.5392215251922607,
      "learning_rate": 9.96623812620991e-06,
      "loss": 0.6817,
      "step": 1387
    },
    {
      "epoch": 0.19800285306704707,
      "grad_norm": 1.1809272766113281,
      "learning_rate": 9.966148732674296e-06,
      "loss": 0.82,
      "step": 1388
    },
    {
      "epoch": 0.19814550641940085,
      "grad_norm": 0.83845055103302,
      "learning_rate": 9.966059221350504e-06,
      "loss": 0.8338,
      "step": 1389
    },
    {
      "epoch": 0.19828815977175462,
      "grad_norm": 1.3236448764801025,
      "learning_rate": 9.965969592240654e-06,
      "loss": 0.6959,
      "step": 1390
    },
    {
      "epoch": 0.19843081312410843,
      "grad_norm": 0.8994632363319397,
      "learning_rate": 9.965879845346874e-06,
      "loss": 0.7831,
      "step": 1391
    },
    {
      "epoch": 0.1985734664764622,
      "grad_norm": 0.6459245085716248,
      "learning_rate": 9.96578998067129e-06,
      "loss": 0.9102,
      "step": 1392
    },
    {
      "epoch": 0.19871611982881598,
      "grad_norm": 1.321109652519226,
      "learning_rate": 9.965699998216035e-06,
      "loss": 0.6065,
      "step": 1393
    },
    {
      "epoch": 0.19885877318116976,
      "grad_norm": 0.7358494400978088,
      "learning_rate": 9.965609897983244e-06,
      "loss": 0.6254,
      "step": 1394
    },
    {
      "epoch": 0.19900142653352354,
      "grad_norm": 0.7518607974052429,
      "learning_rate": 9.965519679975053e-06,
      "loss": 0.8559,
      "step": 1395
    },
    {
      "epoch": 0.19914407988587732,
      "grad_norm": 0.7181278467178345,
      "learning_rate": 9.965429344193604e-06,
      "loss": 0.9398,
      "step": 1396
    },
    {
      "epoch": 0.1992867332382311,
      "grad_norm": 0.9121167063713074,
      "learning_rate": 9.965338890641036e-06,
      "loss": 0.7441,
      "step": 1397
    },
    {
      "epoch": 0.19942938659058487,
      "grad_norm": 0.9675090312957764,
      "learning_rate": 9.965248319319495e-06,
      "loss": 0.8634,
      "step": 1398
    },
    {
      "epoch": 0.19957203994293865,
      "grad_norm": 0.7653679847717285,
      "learning_rate": 9.965157630231132e-06,
      "loss": 0.7931,
      "step": 1399
    },
    {
      "epoch": 0.19971469329529243,
      "grad_norm": 0.6233118772506714,
      "learning_rate": 9.965066823378096e-06,
      "loss": 0.8327,
      "step": 1400
    },
    {
      "epoch": 0.19985734664764623,
      "grad_norm": 0.9585648775100708,
      "learning_rate": 9.964975898762542e-06,
      "loss": 0.8547,
      "step": 1401
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9791272878646851,
      "learning_rate": 9.964884856386625e-06,
      "loss": 0.9014,
      "step": 1402
    },
    {
      "epoch": 0.2001426533523538,
      "grad_norm": 0.7262995839118958,
      "learning_rate": 9.964793696252505e-06,
      "loss": 0.7906,
      "step": 1403
    },
    {
      "epoch": 0.20028530670470757,
      "grad_norm": 0.6719980239868164,
      "learning_rate": 9.964702418362344e-06,
      "loss": 0.8392,
      "step": 1404
    },
    {
      "epoch": 0.20042796005706134,
      "grad_norm": 0.7025762796401978,
      "learning_rate": 9.964611022718307e-06,
      "loss": 0.866,
      "step": 1405
    },
    {
      "epoch": 0.20057061340941512,
      "grad_norm": 1.1558644771575928,
      "learning_rate": 9.964519509322562e-06,
      "loss": 0.8687,
      "step": 1406
    },
    {
      "epoch": 0.2007132667617689,
      "grad_norm": 1.01312255859375,
      "learning_rate": 9.96442787817728e-06,
      "loss": 0.8252,
      "step": 1407
    },
    {
      "epoch": 0.20085592011412268,
      "grad_norm": 0.7223882079124451,
      "learning_rate": 9.964336129284633e-06,
      "loss": 0.84,
      "step": 1408
    },
    {
      "epoch": 0.20099857346647645,
      "grad_norm": 0.9957675337791443,
      "learning_rate": 9.964244262646798e-06,
      "loss": 0.851,
      "step": 1409
    },
    {
      "epoch": 0.20114122681883023,
      "grad_norm": 0.6859878301620483,
      "learning_rate": 9.964152278265953e-06,
      "loss": 0.8173,
      "step": 1410
    },
    {
      "epoch": 0.201283880171184,
      "grad_norm": 1.0454378128051758,
      "learning_rate": 9.964060176144281e-06,
      "loss": 0.8231,
      "step": 1411
    },
    {
      "epoch": 0.20142653352353782,
      "grad_norm": 0.7367770671844482,
      "learning_rate": 9.963967956283966e-06,
      "loss": 0.788,
      "step": 1412
    },
    {
      "epoch": 0.2015691868758916,
      "grad_norm": 0.8545896410942078,
      "learning_rate": 9.963875618687196e-06,
      "loss": 0.8883,
      "step": 1413
    },
    {
      "epoch": 0.20171184022824537,
      "grad_norm": 0.8085616827011108,
      "learning_rate": 9.96378316335616e-06,
      "loss": 0.8329,
      "step": 1414
    },
    {
      "epoch": 0.20185449358059915,
      "grad_norm": 0.8312434554100037,
      "learning_rate": 9.963690590293051e-06,
      "loss": 0.8483,
      "step": 1415
    },
    {
      "epoch": 0.20199714693295293,
      "grad_norm": 0.769288182258606,
      "learning_rate": 9.963597899500066e-06,
      "loss": 0.8054,
      "step": 1416
    },
    {
      "epoch": 0.2021398002853067,
      "grad_norm": 0.7340685129165649,
      "learning_rate": 9.963505090979401e-06,
      "loss": 0.9286,
      "step": 1417
    },
    {
      "epoch": 0.20228245363766048,
      "grad_norm": 0.7468265891075134,
      "learning_rate": 9.96341216473326e-06,
      "loss": 0.8526,
      "step": 1418
    },
    {
      "epoch": 0.20242510699001426,
      "grad_norm": 0.8425663113594055,
      "learning_rate": 9.963319120763844e-06,
      "loss": 0.8726,
      "step": 1419
    },
    {
      "epoch": 0.20256776034236804,
      "grad_norm": 0.7844623327255249,
      "learning_rate": 9.963225959073361e-06,
      "loss": 0.8478,
      "step": 1420
    },
    {
      "epoch": 0.2027104136947218,
      "grad_norm": 0.7162002325057983,
      "learning_rate": 9.963132679664023e-06,
      "loss": 0.8952,
      "step": 1421
    },
    {
      "epoch": 0.20285306704707562,
      "grad_norm": 0.9700891971588135,
      "learning_rate": 9.963039282538039e-06,
      "loss": 0.8846,
      "step": 1422
    },
    {
      "epoch": 0.2029957203994294,
      "grad_norm": 0.9550398588180542,
      "learning_rate": 9.962945767697628e-06,
      "loss": 0.7338,
      "step": 1423
    },
    {
      "epoch": 0.20313837375178317,
      "grad_norm": 0.8337945342063904,
      "learning_rate": 9.962852135145002e-06,
      "loss": 0.8405,
      "step": 1424
    },
    {
      "epoch": 0.20328102710413695,
      "grad_norm": 0.7780858278274536,
      "learning_rate": 9.962758384882388e-06,
      "loss": 0.9113,
      "step": 1425
    },
    {
      "epoch": 0.20342368045649073,
      "grad_norm": 0.8404639363288879,
      "learning_rate": 9.962664516912005e-06,
      "loss": 0.6913,
      "step": 1426
    },
    {
      "epoch": 0.2035663338088445,
      "grad_norm": 0.7688642144203186,
      "learning_rate": 9.962570531236086e-06,
      "loss": 0.7987,
      "step": 1427
    },
    {
      "epoch": 0.20370898716119828,
      "grad_norm": 0.5986928939819336,
      "learning_rate": 9.962476427856853e-06,
      "loss": 0.8337,
      "step": 1428
    },
    {
      "epoch": 0.20385164051355206,
      "grad_norm": 0.835395097732544,
      "learning_rate": 9.96238220677654e-06,
      "loss": 0.8206,
      "step": 1429
    },
    {
      "epoch": 0.20399429386590584,
      "grad_norm": 0.6732452511787415,
      "learning_rate": 9.962287867997383e-06,
      "loss": 0.8274,
      "step": 1430
    },
    {
      "epoch": 0.20413694721825962,
      "grad_norm": 0.7343263626098633,
      "learning_rate": 9.96219341152162e-06,
      "loss": 0.8371,
      "step": 1431
    },
    {
      "epoch": 0.20427960057061342,
      "grad_norm": 1.052815318107605,
      "learning_rate": 9.962098837351488e-06,
      "loss": 0.8115,
      "step": 1432
    },
    {
      "epoch": 0.2044222539229672,
      "grad_norm": 0.7229878306388855,
      "learning_rate": 9.962004145489235e-06,
      "loss": 0.8195,
      "step": 1433
    },
    {
      "epoch": 0.20456490727532098,
      "grad_norm": 0.7342201471328735,
      "learning_rate": 9.961909335937102e-06,
      "loss": 0.877,
      "step": 1434
    },
    {
      "epoch": 0.20470756062767476,
      "grad_norm": 0.8799934983253479,
      "learning_rate": 9.961814408697341e-06,
      "loss": 0.8691,
      "step": 1435
    },
    {
      "epoch": 0.20485021398002853,
      "grad_norm": 0.9665185213088989,
      "learning_rate": 9.961719363772203e-06,
      "loss": 0.8859,
      "step": 1436
    },
    {
      "epoch": 0.2049928673323823,
      "grad_norm": 0.7919800281524658,
      "learning_rate": 9.961624201163942e-06,
      "loss": 0.8623,
      "step": 1437
    },
    {
      "epoch": 0.2051355206847361,
      "grad_norm": 0.5769456624984741,
      "learning_rate": 9.961528920874814e-06,
      "loss": 0.6311,
      "step": 1438
    },
    {
      "epoch": 0.20527817403708987,
      "grad_norm": 0.7586618065834045,
      "learning_rate": 9.96143352290708e-06,
      "loss": 0.8375,
      "step": 1439
    },
    {
      "epoch": 0.20542082738944364,
      "grad_norm": 0.8899186253547668,
      "learning_rate": 9.961338007263004e-06,
      "loss": 0.851,
      "step": 1440
    },
    {
      "epoch": 0.20556348074179742,
      "grad_norm": 0.8149307370185852,
      "learning_rate": 9.961242373944848e-06,
      "loss": 0.8635,
      "step": 1441
    },
    {
      "epoch": 0.2057061340941512,
      "grad_norm": 0.8376900553703308,
      "learning_rate": 9.961146622954883e-06,
      "loss": 0.8212,
      "step": 1442
    },
    {
      "epoch": 0.205848787446505,
      "grad_norm": 0.757046639919281,
      "learning_rate": 9.961050754295379e-06,
      "loss": 0.9017,
      "step": 1443
    },
    {
      "epoch": 0.20599144079885878,
      "grad_norm": 0.992950439453125,
      "learning_rate": 9.960954767968611e-06,
      "loss": 0.8446,
      "step": 1444
    },
    {
      "epoch": 0.20613409415121256,
      "grad_norm": 0.8884807229042053,
      "learning_rate": 9.960858663976854e-06,
      "loss": 0.7902,
      "step": 1445
    },
    {
      "epoch": 0.20627674750356634,
      "grad_norm": 1.0772405862808228,
      "learning_rate": 9.960762442322386e-06,
      "loss": 0.8405,
      "step": 1446
    },
    {
      "epoch": 0.20641940085592012,
      "grad_norm": 0.7249343395233154,
      "learning_rate": 9.960666103007495e-06,
      "loss": 0.8426,
      "step": 1447
    },
    {
      "epoch": 0.2065620542082739,
      "grad_norm": 1.0311321020126343,
      "learning_rate": 9.960569646034461e-06,
      "loss": 0.8344,
      "step": 1448
    },
    {
      "epoch": 0.20670470756062767,
      "grad_norm": 0.8216068744659424,
      "learning_rate": 9.960473071405574e-06,
      "loss": 0.8188,
      "step": 1449
    },
    {
      "epoch": 0.20684736091298145,
      "grad_norm": 0.920533299446106,
      "learning_rate": 9.960376379123123e-06,
      "loss": 0.8147,
      "step": 1450
    },
    {
      "epoch": 0.20699001426533523,
      "grad_norm": 0.7541977167129517,
      "learning_rate": 9.9602795691894e-06,
      "loss": 0.8744,
      "step": 1451
    },
    {
      "epoch": 0.207132667617689,
      "grad_norm": 0.9311749339103699,
      "learning_rate": 9.960182641606704e-06,
      "loss": 0.831,
      "step": 1452
    },
    {
      "epoch": 0.2072753209700428,
      "grad_norm": 0.6928288340568542,
      "learning_rate": 9.960085596377335e-06,
      "loss": 0.8744,
      "step": 1453
    },
    {
      "epoch": 0.2074179743223966,
      "grad_norm": 0.8880890607833862,
      "learning_rate": 9.959988433503592e-06,
      "loss": 0.8748,
      "step": 1454
    },
    {
      "epoch": 0.20756062767475036,
      "grad_norm": 0.7945384979248047,
      "learning_rate": 9.959891152987779e-06,
      "loss": 0.8836,
      "step": 1455
    },
    {
      "epoch": 0.20770328102710414,
      "grad_norm": 0.8306334018707275,
      "learning_rate": 9.959793754832208e-06,
      "loss": 0.8725,
      "step": 1456
    },
    {
      "epoch": 0.20784593437945792,
      "grad_norm": 0.8966955542564392,
      "learning_rate": 9.959696239039182e-06,
      "loss": 0.8302,
      "step": 1457
    },
    {
      "epoch": 0.2079885877318117,
      "grad_norm": 0.7303590774536133,
      "learning_rate": 9.95959860561102e-06,
      "loss": 0.8507,
      "step": 1458
    },
    {
      "epoch": 0.20813124108416547,
      "grad_norm": 0.7616184949874878,
      "learning_rate": 9.959500854550036e-06,
      "loss": 0.8586,
      "step": 1459
    },
    {
      "epoch": 0.20827389443651925,
      "grad_norm": 0.6741976141929626,
      "learning_rate": 9.959402985858548e-06,
      "loss": 0.9009,
      "step": 1460
    },
    {
      "epoch": 0.20841654778887303,
      "grad_norm": 0.7569184899330139,
      "learning_rate": 9.959304999538875e-06,
      "loss": 0.8256,
      "step": 1461
    },
    {
      "epoch": 0.2085592011412268,
      "grad_norm": 0.7997366189956665,
      "learning_rate": 9.959206895593345e-06,
      "loss": 0.8549,
      "step": 1462
    },
    {
      "epoch": 0.20870185449358059,
      "grad_norm": 0.6974430084228516,
      "learning_rate": 9.959108674024282e-06,
      "loss": 0.8359,
      "step": 1463
    },
    {
      "epoch": 0.2088445078459344,
      "grad_norm": 0.6249332427978516,
      "learning_rate": 9.959010334834018e-06,
      "loss": 0.7799,
      "step": 1464
    },
    {
      "epoch": 0.20898716119828817,
      "grad_norm": 0.8389965891838074,
      "learning_rate": 9.958911878024884e-06,
      "loss": 0.8686,
      "step": 1465
    },
    {
      "epoch": 0.20912981455064195,
      "grad_norm": 0.9963505268096924,
      "learning_rate": 9.958813303599215e-06,
      "loss": 0.8453,
      "step": 1466
    },
    {
      "epoch": 0.20927246790299572,
      "grad_norm": 0.7172727584838867,
      "learning_rate": 9.95871461155935e-06,
      "loss": 0.836,
      "step": 1467
    },
    {
      "epoch": 0.2094151212553495,
      "grad_norm": 0.9143319129943848,
      "learning_rate": 9.958615801907629e-06,
      "loss": 0.8144,
      "step": 1468
    },
    {
      "epoch": 0.20955777460770328,
      "grad_norm": 0.8972535133361816,
      "learning_rate": 9.958516874646396e-06,
      "loss": 0.7733,
      "step": 1469
    },
    {
      "epoch": 0.20970042796005706,
      "grad_norm": 0.9460464715957642,
      "learning_rate": 9.958417829777996e-06,
      "loss": 0.8378,
      "step": 1470
    },
    {
      "epoch": 0.20984308131241083,
      "grad_norm": 0.8458881974220276,
      "learning_rate": 9.95831866730478e-06,
      "loss": 0.8638,
      "step": 1471
    },
    {
      "epoch": 0.2099857346647646,
      "grad_norm": 0.7687568664550781,
      "learning_rate": 9.958219387229099e-06,
      "loss": 0.7991,
      "step": 1472
    },
    {
      "epoch": 0.2101283880171184,
      "grad_norm": 0.6833989024162292,
      "learning_rate": 9.958119989553309e-06,
      "loss": 0.8146,
      "step": 1473
    },
    {
      "epoch": 0.2102710413694722,
      "grad_norm": 0.9045029282569885,
      "learning_rate": 9.958020474279766e-06,
      "loss": 0.7656,
      "step": 1474
    },
    {
      "epoch": 0.21041369472182597,
      "grad_norm": 0.8134523630142212,
      "learning_rate": 9.95792084141083e-06,
      "loss": 0.7935,
      "step": 1475
    },
    {
      "epoch": 0.21055634807417975,
      "grad_norm": 0.8164616227149963,
      "learning_rate": 9.957821090948867e-06,
      "loss": 0.8838,
      "step": 1476
    },
    {
      "epoch": 0.21069900142653353,
      "grad_norm": 0.8561461567878723,
      "learning_rate": 9.95772122289624e-06,
      "loss": 0.7915,
      "step": 1477
    },
    {
      "epoch": 0.2108416547788873,
      "grad_norm": 0.897421658039093,
      "learning_rate": 9.957621237255318e-06,
      "loss": 0.8708,
      "step": 1478
    },
    {
      "epoch": 0.21098430813124108,
      "grad_norm": 0.9836632609367371,
      "learning_rate": 9.957521134028474e-06,
      "loss": 0.8615,
      "step": 1479
    },
    {
      "epoch": 0.21112696148359486,
      "grad_norm": 0.9291009306907654,
      "learning_rate": 9.95742091321808e-06,
      "loss": 0.8834,
      "step": 1480
    },
    {
      "epoch": 0.21126961483594864,
      "grad_norm": 0.7535540461540222,
      "learning_rate": 9.957320574826516e-06,
      "loss": 0.8303,
      "step": 1481
    },
    {
      "epoch": 0.21141226818830242,
      "grad_norm": 0.8194931745529175,
      "learning_rate": 9.95722011885616e-06,
      "loss": 0.7933,
      "step": 1482
    },
    {
      "epoch": 0.2115549215406562,
      "grad_norm": 1.0039008855819702,
      "learning_rate": 9.957119545309393e-06,
      "loss": 0.9125,
      "step": 1483
    },
    {
      "epoch": 0.21169757489301,
      "grad_norm": 0.6903035640716553,
      "learning_rate": 9.957018854188604e-06,
      "loss": 0.8154,
      "step": 1484
    },
    {
      "epoch": 0.21184022824536378,
      "grad_norm": 0.8587132692337036,
      "learning_rate": 9.95691804549618e-06,
      "loss": 0.8082,
      "step": 1485
    },
    {
      "epoch": 0.21198288159771755,
      "grad_norm": 0.8282580971717834,
      "learning_rate": 9.956817119234511e-06,
      "loss": 0.858,
      "step": 1486
    },
    {
      "epoch": 0.21212553495007133,
      "grad_norm": 0.7019696235656738,
      "learning_rate": 9.956716075405993e-06,
      "loss": 0.748,
      "step": 1487
    },
    {
      "epoch": 0.2122681883024251,
      "grad_norm": 0.5958597660064697,
      "learning_rate": 9.956614914013018e-06,
      "loss": 0.8647,
      "step": 1488
    },
    {
      "epoch": 0.2124108416547789,
      "grad_norm": 1.059517502784729,
      "learning_rate": 9.95651363505799e-06,
      "loss": 0.7933,
      "step": 1489
    },
    {
      "epoch": 0.21255349500713266,
      "grad_norm": 0.877007007598877,
      "learning_rate": 9.956412238543309e-06,
      "loss": 0.8232,
      "step": 1490
    },
    {
      "epoch": 0.21269614835948644,
      "grad_norm": 0.6188616156578064,
      "learning_rate": 9.95631072447138e-06,
      "loss": 0.8703,
      "step": 1491
    },
    {
      "epoch": 0.21283880171184022,
      "grad_norm": 0.9746018648147583,
      "learning_rate": 9.95620909284461e-06,
      "loss": 0.802,
      "step": 1492
    },
    {
      "epoch": 0.212981455064194,
      "grad_norm": 0.7111904621124268,
      "learning_rate": 9.956107343665412e-06,
      "loss": 0.8449,
      "step": 1493
    },
    {
      "epoch": 0.21312410841654778,
      "grad_norm": 0.7560496926307678,
      "learning_rate": 9.956005476936198e-06,
      "loss": 0.8304,
      "step": 1494
    },
    {
      "epoch": 0.21326676176890158,
      "grad_norm": 0.9106425642967224,
      "learning_rate": 9.955903492659386e-06,
      "loss": 0.8314,
      "step": 1495
    },
    {
      "epoch": 0.21340941512125536,
      "grad_norm": 0.7547789216041565,
      "learning_rate": 9.95580139083739e-06,
      "loss": 0.7854,
      "step": 1496
    },
    {
      "epoch": 0.21355206847360914,
      "grad_norm": 0.8750001192092896,
      "learning_rate": 9.955699171472637e-06,
      "loss": 0.7986,
      "step": 1497
    },
    {
      "epoch": 0.2136947218259629,
      "grad_norm": 0.8077685236930847,
      "learning_rate": 9.955596834567547e-06,
      "loss": 0.8956,
      "step": 1498
    },
    {
      "epoch": 0.2138373751783167,
      "grad_norm": 0.8945642113685608,
      "learning_rate": 9.955494380124551e-06,
      "loss": 0.827,
      "step": 1499
    },
    {
      "epoch": 0.21398002853067047,
      "grad_norm": 0.706613302230835,
      "learning_rate": 9.955391808146075e-06,
      "loss": 0.7894,
      "step": 1500
    },
    {
      "epoch": 0.21412268188302425,
      "grad_norm": 0.974545419216156,
      "learning_rate": 9.955289118634556e-06,
      "loss": 0.8162,
      "step": 1501
    },
    {
      "epoch": 0.21426533523537802,
      "grad_norm": 0.7902306914329529,
      "learning_rate": 9.95518631159243e-06,
      "loss": 0.879,
      "step": 1502
    },
    {
      "epoch": 0.2144079885877318,
      "grad_norm": 0.6855301856994629,
      "learning_rate": 9.95508338702213e-06,
      "loss": 0.8494,
      "step": 1503
    },
    {
      "epoch": 0.21455064194008558,
      "grad_norm": 0.7092685699462891,
      "learning_rate": 9.9549803449261e-06,
      "loss": 0.8121,
      "step": 1504
    },
    {
      "epoch": 0.21469329529243938,
      "grad_norm": 0.7201098799705505,
      "learning_rate": 9.954877185306784e-06,
      "loss": 0.8713,
      "step": 1505
    },
    {
      "epoch": 0.21483594864479316,
      "grad_norm": 0.7282015681266785,
      "learning_rate": 9.954773908166631e-06,
      "loss": 0.785,
      "step": 1506
    },
    {
      "epoch": 0.21497860199714694,
      "grad_norm": 1.1726269721984863,
      "learning_rate": 9.954670513508089e-06,
      "loss": 0.8062,
      "step": 1507
    },
    {
      "epoch": 0.21512125534950072,
      "grad_norm": 0.9354681968688965,
      "learning_rate": 9.954567001333609e-06,
      "loss": 0.7863,
      "step": 1508
    },
    {
      "epoch": 0.2152639087018545,
      "grad_norm": 0.7831276059150696,
      "learning_rate": 9.954463371645646e-06,
      "loss": 0.8388,
      "step": 1509
    },
    {
      "epoch": 0.21540656205420827,
      "grad_norm": 0.7156667113304138,
      "learning_rate": 9.954359624446662e-06,
      "loss": 0.7901,
      "step": 1510
    },
    {
      "epoch": 0.21554921540656205,
      "grad_norm": 0.5895144939422607,
      "learning_rate": 9.954255759739113e-06,
      "loss": 0.834,
      "step": 1511
    },
    {
      "epoch": 0.21569186875891583,
      "grad_norm": 0.894581139087677,
      "learning_rate": 9.954151777525464e-06,
      "loss": 0.8422,
      "step": 1512
    },
    {
      "epoch": 0.2158345221112696,
      "grad_norm": 0.8653997778892517,
      "learning_rate": 9.95404767780818e-06,
      "loss": 0.7984,
      "step": 1513
    },
    {
      "epoch": 0.21597717546362338,
      "grad_norm": 0.8264660239219666,
      "learning_rate": 9.953943460589734e-06,
      "loss": 0.8289,
      "step": 1514
    },
    {
      "epoch": 0.2161198288159772,
      "grad_norm": 0.8366053104400635,
      "learning_rate": 9.953839125872595e-06,
      "loss": 0.8601,
      "step": 1515
    },
    {
      "epoch": 0.21626248216833097,
      "grad_norm": 0.8729448914527893,
      "learning_rate": 9.953734673659239e-06,
      "loss": 0.8096,
      "step": 1516
    },
    {
      "epoch": 0.21640513552068474,
      "grad_norm": 0.7833065986633301,
      "learning_rate": 9.953630103952142e-06,
      "loss": 0.7769,
      "step": 1517
    },
    {
      "epoch": 0.21654778887303852,
      "grad_norm": 0.9304572343826294,
      "learning_rate": 9.953525416753782e-06,
      "loss": 0.8985,
      "step": 1518
    },
    {
      "epoch": 0.2166904422253923,
      "grad_norm": 1.3142472505569458,
      "learning_rate": 9.953420612066646e-06,
      "loss": 0.8003,
      "step": 1519
    },
    {
      "epoch": 0.21683309557774608,
      "grad_norm": 0.8854172229766846,
      "learning_rate": 9.953315689893219e-06,
      "loss": 0.9071,
      "step": 1520
    },
    {
      "epoch": 0.21697574893009985,
      "grad_norm": 0.7279800176620483,
      "learning_rate": 9.95321065023599e-06,
      "loss": 0.8095,
      "step": 1521
    },
    {
      "epoch": 0.21711840228245363,
      "grad_norm": 1.1312992572784424,
      "learning_rate": 9.953105493097449e-06,
      "loss": 0.8317,
      "step": 1522
    },
    {
      "epoch": 0.2172610556348074,
      "grad_norm": 0.7317725419998169,
      "learning_rate": 9.953000218480088e-06,
      "loss": 0.8973,
      "step": 1523
    },
    {
      "epoch": 0.2174037089871612,
      "grad_norm": 0.9556543231010437,
      "learning_rate": 9.952894826386408e-06,
      "loss": 0.7908,
      "step": 1524
    },
    {
      "epoch": 0.21754636233951496,
      "grad_norm": 0.9414548873901367,
      "learning_rate": 9.952789316818907e-06,
      "loss": 0.8429,
      "step": 1525
    },
    {
      "epoch": 0.21768901569186877,
      "grad_norm": 0.7839816808700562,
      "learning_rate": 9.952683689780089e-06,
      "loss": 0.8439,
      "step": 1526
    },
    {
      "epoch": 0.21783166904422255,
      "grad_norm": 0.7676011919975281,
      "learning_rate": 9.952577945272458e-06,
      "loss": 0.8922,
      "step": 1527
    },
    {
      "epoch": 0.21797432239657633,
      "grad_norm": 0.738705575466156,
      "learning_rate": 9.95247208329852e-06,
      "loss": 0.8221,
      "step": 1528
    },
    {
      "epoch": 0.2181169757489301,
      "grad_norm": 0.76386958360672,
      "learning_rate": 9.952366103860788e-06,
      "loss": 0.8387,
      "step": 1529
    },
    {
      "epoch": 0.21825962910128388,
      "grad_norm": 0.8681210279464722,
      "learning_rate": 9.952260006961776e-06,
      "loss": 0.7862,
      "step": 1530
    },
    {
      "epoch": 0.21840228245363766,
      "grad_norm": 1.1066902875900269,
      "learning_rate": 9.952153792604e-06,
      "loss": 0.8484,
      "step": 1531
    },
    {
      "epoch": 0.21854493580599144,
      "grad_norm": 1.1809475421905518,
      "learning_rate": 9.95204746078998e-06,
      "loss": 0.8923,
      "step": 1532
    },
    {
      "epoch": 0.2186875891583452,
      "grad_norm": 0.951051652431488,
      "learning_rate": 9.951941011522234e-06,
      "loss": 0.7666,
      "step": 1533
    },
    {
      "epoch": 0.218830242510699,
      "grad_norm": 0.8252032399177551,
      "learning_rate": 9.951834444803294e-06,
      "loss": 0.7735,
      "step": 1534
    },
    {
      "epoch": 0.21897289586305277,
      "grad_norm": 3.1314423084259033,
      "learning_rate": 9.95172776063568e-06,
      "loss": 0.7101,
      "step": 1535
    },
    {
      "epoch": 0.21911554921540657,
      "grad_norm": 0.9142853617668152,
      "learning_rate": 9.951620959021929e-06,
      "loss": 0.8895,
      "step": 1536
    },
    {
      "epoch": 0.21925820256776035,
      "grad_norm": 0.841295063495636,
      "learning_rate": 9.951514039964569e-06,
      "loss": 0.7787,
      "step": 1537
    },
    {
      "epoch": 0.21940085592011413,
      "grad_norm": 0.7550581097602844,
      "learning_rate": 9.951407003466139e-06,
      "loss": 0.8533,
      "step": 1538
    },
    {
      "epoch": 0.2195435092724679,
      "grad_norm": 1.0160503387451172,
      "learning_rate": 9.951299849529175e-06,
      "loss": 0.8078,
      "step": 1539
    },
    {
      "epoch": 0.21968616262482168,
      "grad_norm": 1.1737887859344482,
      "learning_rate": 9.951192578156221e-06,
      "loss": 0.7924,
      "step": 1540
    },
    {
      "epoch": 0.21982881597717546,
      "grad_norm": 1.1103118658065796,
      "learning_rate": 9.95108518934982e-06,
      "loss": 0.8444,
      "step": 1541
    },
    {
      "epoch": 0.21997146932952924,
      "grad_norm": 0.9257894158363342,
      "learning_rate": 9.950977683112521e-06,
      "loss": 0.8423,
      "step": 1542
    },
    {
      "epoch": 0.22011412268188302,
      "grad_norm": 0.7128218412399292,
      "learning_rate": 9.950870059446872e-06,
      "loss": 0.8094,
      "step": 1543
    },
    {
      "epoch": 0.2202567760342368,
      "grad_norm": 0.754286527633667,
      "learning_rate": 9.950762318355425e-06,
      "loss": 0.8264,
      "step": 1544
    },
    {
      "epoch": 0.22039942938659057,
      "grad_norm": 0.8896986246109009,
      "learning_rate": 9.950654459840736e-06,
      "loss": 0.7568,
      "step": 1545
    },
    {
      "epoch": 0.22054208273894438,
      "grad_norm": 0.9886698126792908,
      "learning_rate": 9.950546483905366e-06,
      "loss": 0.8754,
      "step": 1546
    },
    {
      "epoch": 0.22068473609129816,
      "grad_norm": 0.8639637231826782,
      "learning_rate": 9.95043839055187e-06,
      "loss": 0.8645,
      "step": 1547
    },
    {
      "epoch": 0.22082738944365193,
      "grad_norm": 0.5904603004455566,
      "learning_rate": 9.950330179782818e-06,
      "loss": 0.9219,
      "step": 1548
    },
    {
      "epoch": 0.2209700427960057,
      "grad_norm": 0.6694537997245789,
      "learning_rate": 9.950221851600775e-06,
      "loss": 0.8739,
      "step": 1549
    },
    {
      "epoch": 0.2211126961483595,
      "grad_norm": 0.6146013140678406,
      "learning_rate": 9.950113406008306e-06,
      "loss": 0.9186,
      "step": 1550
    },
    {
      "epoch": 0.22125534950071327,
      "grad_norm": 0.778023898601532,
      "learning_rate": 9.95000484300799e-06,
      "loss": 0.8382,
      "step": 1551
    },
    {
      "epoch": 0.22139800285306704,
      "grad_norm": 0.7983495593070984,
      "learning_rate": 9.949896162602396e-06,
      "loss": 0.8371,
      "step": 1552
    },
    {
      "epoch": 0.22154065620542082,
      "grad_norm": 0.8418344259262085,
      "learning_rate": 9.949787364794104e-06,
      "loss": 0.81,
      "step": 1553
    },
    {
      "epoch": 0.2216833095577746,
      "grad_norm": 0.8970357775688171,
      "learning_rate": 9.949678449585695e-06,
      "loss": 0.8177,
      "step": 1554
    },
    {
      "epoch": 0.22182596291012838,
      "grad_norm": 0.8553009033203125,
      "learning_rate": 9.949569416979753e-06,
      "loss": 0.7695,
      "step": 1555
    },
    {
      "epoch": 0.22196861626248215,
      "grad_norm": 1.215833067893982,
      "learning_rate": 9.949460266978862e-06,
      "loss": 0.808,
      "step": 1556
    },
    {
      "epoch": 0.22211126961483596,
      "grad_norm": 0.6654186844825745,
      "learning_rate": 9.949350999585612e-06,
      "loss": 0.7478,
      "step": 1557
    },
    {
      "epoch": 0.22225392296718974,
      "grad_norm": 0.6524320244789124,
      "learning_rate": 9.949241614802593e-06,
      "loss": 0.8465,
      "step": 1558
    },
    {
      "epoch": 0.22239657631954352,
      "grad_norm": 0.9813492298126221,
      "learning_rate": 9.949132112632402e-06,
      "loss": 0.81,
      "step": 1559
    },
    {
      "epoch": 0.2225392296718973,
      "grad_norm": 0.8177356123924255,
      "learning_rate": 9.949022493077636e-06,
      "loss": 0.5872,
      "step": 1560
    },
    {
      "epoch": 0.22268188302425107,
      "grad_norm": 0.9326146245002747,
      "learning_rate": 9.948912756140891e-06,
      "loss": 0.8172,
      "step": 1561
    },
    {
      "epoch": 0.22282453637660485,
      "grad_norm": 0.7023347616195679,
      "learning_rate": 9.948802901824775e-06,
      "loss": 0.8085,
      "step": 1562
    },
    {
      "epoch": 0.22296718972895863,
      "grad_norm": 0.7878661155700684,
      "learning_rate": 9.948692930131892e-06,
      "loss": 0.9046,
      "step": 1563
    },
    {
      "epoch": 0.2231098430813124,
      "grad_norm": 0.7527204751968384,
      "learning_rate": 9.948582841064847e-06,
      "loss": 0.7797,
      "step": 1564
    },
    {
      "epoch": 0.22325249643366618,
      "grad_norm": 0.8323975205421448,
      "learning_rate": 9.948472634626256e-06,
      "loss": 0.8538,
      "step": 1565
    },
    {
      "epoch": 0.22339514978601996,
      "grad_norm": 0.8328871726989746,
      "learning_rate": 9.948362310818729e-06,
      "loss": 0.8325,
      "step": 1566
    },
    {
      "epoch": 0.22353780313837376,
      "grad_norm": 0.8960254788398743,
      "learning_rate": 9.948251869644885e-06,
      "loss": 0.7934,
      "step": 1567
    },
    {
      "epoch": 0.22368045649072754,
      "grad_norm": 0.619554340839386,
      "learning_rate": 9.948141311107344e-06,
      "loss": 0.8542,
      "step": 1568
    },
    {
      "epoch": 0.22382310984308132,
      "grad_norm": 0.8335708379745483,
      "learning_rate": 9.948030635208726e-06,
      "loss": 0.8629,
      "step": 1569
    },
    {
      "epoch": 0.2239657631954351,
      "grad_norm": 0.7417574524879456,
      "learning_rate": 9.947919841951656e-06,
      "loss": 0.7854,
      "step": 1570
    },
    {
      "epoch": 0.22410841654778887,
      "grad_norm": 0.7167844772338867,
      "learning_rate": 9.947808931338763e-06,
      "loss": 0.8039,
      "step": 1571
    },
    {
      "epoch": 0.22425106990014265,
      "grad_norm": 1.0231636762619019,
      "learning_rate": 9.947697903372678e-06,
      "loss": 0.8038,
      "step": 1572
    },
    {
      "epoch": 0.22439372325249643,
      "grad_norm": 0.8088961243629456,
      "learning_rate": 9.947586758056035e-06,
      "loss": 0.8621,
      "step": 1573
    },
    {
      "epoch": 0.2245363766048502,
      "grad_norm": 0.9731947183609009,
      "learning_rate": 9.947475495391467e-06,
      "loss": 0.8177,
      "step": 1574
    },
    {
      "epoch": 0.22467902995720399,
      "grad_norm": 0.7506547570228577,
      "learning_rate": 9.947364115381617e-06,
      "loss": 0.7207,
      "step": 1575
    },
    {
      "epoch": 0.22482168330955776,
      "grad_norm": 0.7788878679275513,
      "learning_rate": 9.947252618029124e-06,
      "loss": 0.857,
      "step": 1576
    },
    {
      "epoch": 0.22496433666191157,
      "grad_norm": 0.6956376433372498,
      "learning_rate": 9.947141003336633e-06,
      "loss": 0.8099,
      "step": 1577
    },
    {
      "epoch": 0.22510699001426535,
      "grad_norm": 0.593755841255188,
      "learning_rate": 9.947029271306792e-06,
      "loss": 0.8418,
      "step": 1578
    },
    {
      "epoch": 0.22524964336661912,
      "grad_norm": 0.7121747732162476,
      "learning_rate": 9.946917421942249e-06,
      "loss": 0.8181,
      "step": 1579
    },
    {
      "epoch": 0.2253922967189729,
      "grad_norm": 0.9468618035316467,
      "learning_rate": 9.94680545524566e-06,
      "loss": 0.8404,
      "step": 1580
    },
    {
      "epoch": 0.22553495007132668,
      "grad_norm": 0.9029089212417603,
      "learning_rate": 9.946693371219679e-06,
      "loss": 0.7842,
      "step": 1581
    },
    {
      "epoch": 0.22567760342368046,
      "grad_norm": 0.9820320010185242,
      "learning_rate": 9.946581169866964e-06,
      "loss": 0.8875,
      "step": 1582
    },
    {
      "epoch": 0.22582025677603423,
      "grad_norm": 0.8099811673164368,
      "learning_rate": 9.946468851190178e-06,
      "loss": 0.8181,
      "step": 1583
    },
    {
      "epoch": 0.225962910128388,
      "grad_norm": 0.773884654045105,
      "learning_rate": 9.946356415191983e-06,
      "loss": 0.8079,
      "step": 1584
    },
    {
      "epoch": 0.2261055634807418,
      "grad_norm": 0.8076888918876648,
      "learning_rate": 9.946243861875046e-06,
      "loss": 0.9412,
      "step": 1585
    },
    {
      "epoch": 0.22624821683309557,
      "grad_norm": 0.7969920635223389,
      "learning_rate": 9.946131191242038e-06,
      "loss": 0.8288,
      "step": 1586
    },
    {
      "epoch": 0.22639087018544934,
      "grad_norm": 0.7188319563865662,
      "learning_rate": 9.946018403295629e-06,
      "loss": 0.8562,
      "step": 1587
    },
    {
      "epoch": 0.22653352353780315,
      "grad_norm": 0.8297858238220215,
      "learning_rate": 9.945905498038498e-06,
      "loss": 0.7975,
      "step": 1588
    },
    {
      "epoch": 0.22667617689015693,
      "grad_norm": 1.1273237466812134,
      "learning_rate": 9.945792475473319e-06,
      "loss": 0.8782,
      "step": 1589
    },
    {
      "epoch": 0.2268188302425107,
      "grad_norm": 0.9453109502792358,
      "learning_rate": 9.945679335602775e-06,
      "loss": 0.8654,
      "step": 1590
    },
    {
      "epoch": 0.22696148359486448,
      "grad_norm": 0.8723772168159485,
      "learning_rate": 9.945566078429547e-06,
      "loss": 0.7964,
      "step": 1591
    },
    {
      "epoch": 0.22710413694721826,
      "grad_norm": 0.9568871855735779,
      "learning_rate": 9.945452703956325e-06,
      "loss": 0.8647,
      "step": 1592
    },
    {
      "epoch": 0.22724679029957204,
      "grad_norm": 0.8982163667678833,
      "learning_rate": 9.945339212185793e-06,
      "loss": 0.8167,
      "step": 1593
    },
    {
      "epoch": 0.22738944365192582,
      "grad_norm": 0.8818013072013855,
      "learning_rate": 9.94522560312065e-06,
      "loss": 0.8144,
      "step": 1594
    },
    {
      "epoch": 0.2275320970042796,
      "grad_norm": 0.9404963850975037,
      "learning_rate": 9.945111876763584e-06,
      "loss": 0.8257,
      "step": 1595
    },
    {
      "epoch": 0.22767475035663337,
      "grad_norm": 0.8494290113449097,
      "learning_rate": 9.944998033117294e-06,
      "loss": 0.8062,
      "step": 1596
    },
    {
      "epoch": 0.22781740370898715,
      "grad_norm": 0.8072165846824646,
      "learning_rate": 9.944884072184481e-06,
      "loss": 0.8291,
      "step": 1597
    },
    {
      "epoch": 0.22796005706134095,
      "grad_norm": 0.7694734334945679,
      "learning_rate": 9.944769993967851e-06,
      "loss": 0.7759,
      "step": 1598
    },
    {
      "epoch": 0.22810271041369473,
      "grad_norm": 0.719517171382904,
      "learning_rate": 9.944655798470104e-06,
      "loss": 0.8932,
      "step": 1599
    },
    {
      "epoch": 0.2282453637660485,
      "grad_norm": 0.9067398905754089,
      "learning_rate": 9.944541485693952e-06,
      "loss": 0.8331,
      "step": 1600
    },
    {
      "epoch": 0.2283880171184023,
      "grad_norm": 0.8119226098060608,
      "learning_rate": 9.944427055642106e-06,
      "loss": 0.7957,
      "step": 1601
    },
    {
      "epoch": 0.22853067047075606,
      "grad_norm": 0.6486713290214539,
      "learning_rate": 9.944312508317277e-06,
      "loss": 0.8689,
      "step": 1602
    },
    {
      "epoch": 0.22867332382310984,
      "grad_norm": 0.8576393127441406,
      "learning_rate": 9.944197843722186e-06,
      "loss": 0.7606,
      "step": 1603
    },
    {
      "epoch": 0.22881597717546362,
      "grad_norm": 0.8671183586120605,
      "learning_rate": 9.944083061859552e-06,
      "loss": 0.8407,
      "step": 1604
    },
    {
      "epoch": 0.2289586305278174,
      "grad_norm": 0.6724821329116821,
      "learning_rate": 9.943968162732096e-06,
      "loss": 0.8406,
      "step": 1605
    },
    {
      "epoch": 0.22910128388017117,
      "grad_norm": 0.8926640748977661,
      "learning_rate": 9.943853146342543e-06,
      "loss": 0.6138,
      "step": 1606
    },
    {
      "epoch": 0.22924393723252495,
      "grad_norm": 0.7364065051078796,
      "learning_rate": 9.94373801269362e-06,
      "loss": 0.7925,
      "step": 1607
    },
    {
      "epoch": 0.22938659058487876,
      "grad_norm": 1.13555109500885,
      "learning_rate": 9.943622761788062e-06,
      "loss": 0.8352,
      "step": 1608
    },
    {
      "epoch": 0.22952924393723254,
      "grad_norm": 0.7744009494781494,
      "learning_rate": 9.943507393628599e-06,
      "loss": 0.8385,
      "step": 1609
    },
    {
      "epoch": 0.2296718972895863,
      "grad_norm": 1.1767977476119995,
      "learning_rate": 9.94339190821797e-06,
      "loss": 0.9137,
      "step": 1610
    },
    {
      "epoch": 0.2298145506419401,
      "grad_norm": 0.6439663767814636,
      "learning_rate": 9.94327630555891e-06,
      "loss": 0.8694,
      "step": 1611
    },
    {
      "epoch": 0.22995720399429387,
      "grad_norm": 0.8223626613616943,
      "learning_rate": 9.943160585654164e-06,
      "loss": 0.8267,
      "step": 1612
    },
    {
      "epoch": 0.23009985734664765,
      "grad_norm": 1.0448628664016724,
      "learning_rate": 9.943044748506476e-06,
      "loss": 0.8596,
      "step": 1613
    },
    {
      "epoch": 0.23024251069900142,
      "grad_norm": 0.8901304006576538,
      "learning_rate": 9.942928794118593e-06,
      "loss": 0.8276,
      "step": 1614
    },
    {
      "epoch": 0.2303851640513552,
      "grad_norm": 1.0466147661209106,
      "learning_rate": 9.942812722493266e-06,
      "loss": 0.8693,
      "step": 1615
    },
    {
      "epoch": 0.23052781740370898,
      "grad_norm": 0.9215507507324219,
      "learning_rate": 9.942696533633248e-06,
      "loss": 0.8539,
      "step": 1616
    },
    {
      "epoch": 0.23067047075606276,
      "grad_norm": 0.9924200177192688,
      "learning_rate": 9.942580227541294e-06,
      "loss": 0.767,
      "step": 1617
    },
    {
      "epoch": 0.23081312410841653,
      "grad_norm": 0.6106994152069092,
      "learning_rate": 9.942463804220164e-06,
      "loss": 0.8978,
      "step": 1618
    },
    {
      "epoch": 0.23095577746077034,
      "grad_norm": 0.9730753302574158,
      "learning_rate": 9.942347263672617e-06,
      "loss": 0.849,
      "step": 1619
    },
    {
      "epoch": 0.23109843081312412,
      "grad_norm": 0.801425039768219,
      "learning_rate": 9.94223060590142e-06,
      "loss": 0.8151,
      "step": 1620
    },
    {
      "epoch": 0.2312410841654779,
      "grad_norm": 0.7202739119529724,
      "learning_rate": 9.942113830909336e-06,
      "loss": 0.5747,
      "step": 1621
    },
    {
      "epoch": 0.23138373751783167,
      "grad_norm": 1.0737241506576538,
      "learning_rate": 9.941996938699139e-06,
      "loss": 0.8335,
      "step": 1622
    },
    {
      "epoch": 0.23152639087018545,
      "grad_norm": 0.8841755390167236,
      "learning_rate": 9.941879929273598e-06,
      "loss": 0.8232,
      "step": 1623
    },
    {
      "epoch": 0.23166904422253923,
      "grad_norm": 0.9682257771492004,
      "learning_rate": 9.941762802635491e-06,
      "loss": 0.8213,
      "step": 1624
    },
    {
      "epoch": 0.231811697574893,
      "grad_norm": 0.8073592185974121,
      "learning_rate": 9.941645558787594e-06,
      "loss": 0.8897,
      "step": 1625
    },
    {
      "epoch": 0.23195435092724678,
      "grad_norm": 0.8076974153518677,
      "learning_rate": 9.94152819773269e-06,
      "loss": 0.8805,
      "step": 1626
    },
    {
      "epoch": 0.23209700427960056,
      "grad_norm": 0.8395135998725891,
      "learning_rate": 9.941410719473562e-06,
      "loss": 0.7626,
      "step": 1627
    },
    {
      "epoch": 0.23223965763195434,
      "grad_norm": 1.0759040117263794,
      "learning_rate": 9.941293124012993e-06,
      "loss": 0.8684,
      "step": 1628
    },
    {
      "epoch": 0.23238231098430814,
      "grad_norm": 0.6490672826766968,
      "learning_rate": 9.941175411353776e-06,
      "loss": 0.8351,
      "step": 1629
    },
    {
      "epoch": 0.23252496433666192,
      "grad_norm": 0.6349145770072937,
      "learning_rate": 9.9410575814987e-06,
      "loss": 0.802,
      "step": 1630
    },
    {
      "epoch": 0.2326676176890157,
      "grad_norm": 0.7153817415237427,
      "learning_rate": 9.940939634450564e-06,
      "loss": 0.8002,
      "step": 1631
    },
    {
      "epoch": 0.23281027104136948,
      "grad_norm": 0.7590317726135254,
      "learning_rate": 9.940821570212162e-06,
      "loss": 0.6527,
      "step": 1632
    },
    {
      "epoch": 0.23295292439372325,
      "grad_norm": 0.8898860812187195,
      "learning_rate": 9.940703388786296e-06,
      "loss": 0.8587,
      "step": 1633
    },
    {
      "epoch": 0.23309557774607703,
      "grad_norm": 0.8333155512809753,
      "learning_rate": 9.940585090175768e-06,
      "loss": 0.8109,
      "step": 1634
    },
    {
      "epoch": 0.2332382310984308,
      "grad_norm": 0.9909689426422119,
      "learning_rate": 9.940466674383383e-06,
      "loss": 0.8695,
      "step": 1635
    },
    {
      "epoch": 0.2333808844507846,
      "grad_norm": 0.8332314491271973,
      "learning_rate": 9.94034814141195e-06,
      "loss": 0.779,
      "step": 1636
    },
    {
      "epoch": 0.23352353780313836,
      "grad_norm": 0.7979727983474731,
      "learning_rate": 9.940229491264284e-06,
      "loss": 0.8014,
      "step": 1637
    },
    {
      "epoch": 0.23366619115549214,
      "grad_norm": 0.7950722575187683,
      "learning_rate": 9.940110723943196e-06,
      "loss": 0.8589,
      "step": 1638
    },
    {
      "epoch": 0.23380884450784595,
      "grad_norm": 0.6917372941970825,
      "learning_rate": 9.939991839451501e-06,
      "loss": 0.8161,
      "step": 1639
    },
    {
      "epoch": 0.23395149786019973,
      "grad_norm": 0.7580235600471497,
      "learning_rate": 9.939872837792022e-06,
      "loss": 0.7317,
      "step": 1640
    },
    {
      "epoch": 0.2340941512125535,
      "grad_norm": 0.8008021116256714,
      "learning_rate": 9.93975371896758e-06,
      "loss": 0.7995,
      "step": 1641
    },
    {
      "epoch": 0.23423680456490728,
      "grad_norm": 0.6910989284515381,
      "learning_rate": 9.939634482981e-06,
      "loss": 0.8452,
      "step": 1642
    },
    {
      "epoch": 0.23437945791726106,
      "grad_norm": 1.1207400560379028,
      "learning_rate": 9.939515129835112e-06,
      "loss": 0.9101,
      "step": 1643
    },
    {
      "epoch": 0.23452211126961484,
      "grad_norm": 0.8197427988052368,
      "learning_rate": 9.939395659532746e-06,
      "loss": 0.88,
      "step": 1644
    },
    {
      "epoch": 0.2346647646219686,
      "grad_norm": 0.7048259377479553,
      "learning_rate": 9.939276072076733e-06,
      "loss": 0.8642,
      "step": 1645
    },
    {
      "epoch": 0.2348074179743224,
      "grad_norm": 0.7667585015296936,
      "learning_rate": 9.939156367469914e-06,
      "loss": 0.882,
      "step": 1646
    },
    {
      "epoch": 0.23495007132667617,
      "grad_norm": 0.9176719188690186,
      "learning_rate": 9.939036545715126e-06,
      "loss": 0.7242,
      "step": 1647
    },
    {
      "epoch": 0.23509272467902995,
      "grad_norm": 0.8437342643737793,
      "learning_rate": 9.93891660681521e-06,
      "loss": 0.8303,
      "step": 1648
    },
    {
      "epoch": 0.23523537803138372,
      "grad_norm": 0.8201102614402771,
      "learning_rate": 9.93879655077301e-06,
      "loss": 0.8857,
      "step": 1649
    },
    {
      "epoch": 0.23537803138373753,
      "grad_norm": 0.7154242396354675,
      "learning_rate": 9.938676377591377e-06,
      "loss": 0.8175,
      "step": 1650
    },
    {
      "epoch": 0.2355206847360913,
      "grad_norm": 1.0373033285140991,
      "learning_rate": 9.938556087273158e-06,
      "loss": 0.8078,
      "step": 1651
    },
    {
      "epoch": 0.23566333808844508,
      "grad_norm": 1.3327196836471558,
      "learning_rate": 9.93843567982121e-06,
      "loss": 0.8027,
      "step": 1652
    },
    {
      "epoch": 0.23580599144079886,
      "grad_norm": 0.9266972541809082,
      "learning_rate": 9.938315155238385e-06,
      "loss": 0.7856,
      "step": 1653
    },
    {
      "epoch": 0.23594864479315264,
      "grad_norm": 0.7412891387939453,
      "learning_rate": 9.938194513527542e-06,
      "loss": 0.8634,
      "step": 1654
    },
    {
      "epoch": 0.23609129814550642,
      "grad_norm": 0.7532863020896912,
      "learning_rate": 9.938073754691543e-06,
      "loss": 0.8435,
      "step": 1655
    },
    {
      "epoch": 0.2362339514978602,
      "grad_norm": 0.7971359491348267,
      "learning_rate": 9.937952878733254e-06,
      "loss": 0.8465,
      "step": 1656
    },
    {
      "epoch": 0.23637660485021397,
      "grad_norm": 1.2778613567352295,
      "learning_rate": 9.93783188565554e-06,
      "loss": 0.6068,
      "step": 1657
    },
    {
      "epoch": 0.23651925820256775,
      "grad_norm": 0.9497107267379761,
      "learning_rate": 9.93771077546127e-06,
      "loss": 0.812,
      "step": 1658
    },
    {
      "epoch": 0.23666191155492153,
      "grad_norm": 1.138228178024292,
      "learning_rate": 9.937589548153319e-06,
      "loss": 0.8334,
      "step": 1659
    },
    {
      "epoch": 0.23680456490727533,
      "grad_norm": 0.8257390260696411,
      "learning_rate": 9.93746820373456e-06,
      "loss": 0.8706,
      "step": 1660
    },
    {
      "epoch": 0.2369472182596291,
      "grad_norm": 0.710342288017273,
      "learning_rate": 9.937346742207875e-06,
      "loss": 0.8297,
      "step": 1661
    },
    {
      "epoch": 0.2370898716119829,
      "grad_norm": 0.6259229779243469,
      "learning_rate": 9.937225163576138e-06,
      "loss": 0.8604,
      "step": 1662
    },
    {
      "epoch": 0.23723252496433667,
      "grad_norm": 0.8494049310684204,
      "learning_rate": 9.937103467842238e-06,
      "loss": 0.7666,
      "step": 1663
    },
    {
      "epoch": 0.23737517831669044,
      "grad_norm": 0.8034436702728271,
      "learning_rate": 9.93698165500906e-06,
      "loss": 0.8688,
      "step": 1664
    },
    {
      "epoch": 0.23751783166904422,
      "grad_norm": 0.8645340204238892,
      "learning_rate": 9.936859725079492e-06,
      "loss": 0.8275,
      "step": 1665
    },
    {
      "epoch": 0.237660485021398,
      "grad_norm": 0.828236997127533,
      "learning_rate": 9.936737678056429e-06,
      "loss": 0.8357,
      "step": 1666
    },
    {
      "epoch": 0.23780313837375178,
      "grad_norm": 0.7412002086639404,
      "learning_rate": 9.936615513942763e-06,
      "loss": 0.8065,
      "step": 1667
    },
    {
      "epoch": 0.23794579172610555,
      "grad_norm": 0.8817216753959656,
      "learning_rate": 9.936493232741394e-06,
      "loss": 0.8028,
      "step": 1668
    },
    {
      "epoch": 0.23808844507845933,
      "grad_norm": 1.2394442558288574,
      "learning_rate": 9.936370834455218e-06,
      "loss": 0.625,
      "step": 1669
    },
    {
      "epoch": 0.23823109843081314,
      "grad_norm": 0.8922938108444214,
      "learning_rate": 9.936248319087141e-06,
      "loss": 0.8644,
      "step": 1670
    },
    {
      "epoch": 0.23837375178316692,
      "grad_norm": 0.7142878174781799,
      "learning_rate": 9.93612568664007e-06,
      "loss": 0.8089,
      "step": 1671
    },
    {
      "epoch": 0.2385164051355207,
      "grad_norm": 0.9219820499420166,
      "learning_rate": 9.936002937116912e-06,
      "loss": 0.8657,
      "step": 1672
    },
    {
      "epoch": 0.23865905848787447,
      "grad_norm": 1.1039023399353027,
      "learning_rate": 9.935880070520578e-06,
      "loss": 0.7808,
      "step": 1673
    },
    {
      "epoch": 0.23880171184022825,
      "grad_norm": 0.8792162537574768,
      "learning_rate": 9.935757086853984e-06,
      "loss": 0.8895,
      "step": 1674
    },
    {
      "epoch": 0.23894436519258203,
      "grad_norm": 0.7179603576660156,
      "learning_rate": 9.935633986120044e-06,
      "loss": 0.8454,
      "step": 1675
    },
    {
      "epoch": 0.2390870185449358,
      "grad_norm": 0.7662439346313477,
      "learning_rate": 9.93551076832168e-06,
      "loss": 0.8214,
      "step": 1676
    },
    {
      "epoch": 0.23922967189728958,
      "grad_norm": 0.9854037761688232,
      "learning_rate": 9.935387433461814e-06,
      "loss": 0.8049,
      "step": 1677
    },
    {
      "epoch": 0.23937232524964336,
      "grad_norm": 0.7705009579658508,
      "learning_rate": 9.93526398154337e-06,
      "loss": 0.8203,
      "step": 1678
    },
    {
      "epoch": 0.23951497860199714,
      "grad_norm": 0.9047114253044128,
      "learning_rate": 9.93514041256928e-06,
      "loss": 0.8151,
      "step": 1679
    },
    {
      "epoch": 0.2396576319543509,
      "grad_norm": 0.9852571487426758,
      "learning_rate": 9.935016726542471e-06,
      "loss": 0.7873,
      "step": 1680
    },
    {
      "epoch": 0.23980028530670472,
      "grad_norm": 0.8604314923286438,
      "learning_rate": 9.934892923465875e-06,
      "loss": 0.7685,
      "step": 1681
    },
    {
      "epoch": 0.2399429386590585,
      "grad_norm": 1.0661243200302124,
      "learning_rate": 9.934769003342435e-06,
      "loss": 0.8301,
      "step": 1682
    },
    {
      "epoch": 0.24008559201141227,
      "grad_norm": 0.8258073925971985,
      "learning_rate": 9.934644966175084e-06,
      "loss": 0.8213,
      "step": 1683
    },
    {
      "epoch": 0.24022824536376605,
      "grad_norm": 1.0695422887802124,
      "learning_rate": 9.934520811966768e-06,
      "loss": 0.8046,
      "step": 1684
    },
    {
      "epoch": 0.24037089871611983,
      "grad_norm": 0.8595890402793884,
      "learning_rate": 9.934396540720428e-06,
      "loss": 0.8753,
      "step": 1685
    },
    {
      "epoch": 0.2405135520684736,
      "grad_norm": 0.8812291622161865,
      "learning_rate": 9.934272152439014e-06,
      "loss": 0.7167,
      "step": 1686
    },
    {
      "epoch": 0.24065620542082738,
      "grad_norm": 1.1692529916763306,
      "learning_rate": 9.934147647125476e-06,
      "loss": 0.8211,
      "step": 1687
    },
    {
      "epoch": 0.24079885877318116,
      "grad_norm": 0.8988098502159119,
      "learning_rate": 9.934023024782767e-06,
      "loss": 0.7913,
      "step": 1688
    },
    {
      "epoch": 0.24094151212553494,
      "grad_norm": 0.8145775198936462,
      "learning_rate": 9.933898285413843e-06,
      "loss": 0.8902,
      "step": 1689
    },
    {
      "epoch": 0.24108416547788872,
      "grad_norm": 0.7871213555335999,
      "learning_rate": 9.93377342902166e-06,
      "loss": 0.8396,
      "step": 1690
    },
    {
      "epoch": 0.24122681883024252,
      "grad_norm": 0.8794150948524475,
      "learning_rate": 9.933648455609184e-06,
      "loss": 0.8457,
      "step": 1691
    },
    {
      "epoch": 0.2413694721825963,
      "grad_norm": 0.8141103982925415,
      "learning_rate": 9.933523365179377e-06,
      "loss": 0.8583,
      "step": 1692
    },
    {
      "epoch": 0.24151212553495008,
      "grad_norm": 0.8373968005180359,
      "learning_rate": 9.933398157735204e-06,
      "loss": 0.7936,
      "step": 1693
    },
    {
      "epoch": 0.24165477888730386,
      "grad_norm": 0.7925145626068115,
      "learning_rate": 9.933272833279635e-06,
      "loss": 0.785,
      "step": 1694
    },
    {
      "epoch": 0.24179743223965763,
      "grad_norm": 0.830041229724884,
      "learning_rate": 9.933147391815646e-06,
      "loss": 0.8212,
      "step": 1695
    },
    {
      "epoch": 0.2419400855920114,
      "grad_norm": 0.9906054139137268,
      "learning_rate": 9.93302183334621e-06,
      "loss": 0.7736,
      "step": 1696
    },
    {
      "epoch": 0.2420827389443652,
      "grad_norm": 0.7971070408821106,
      "learning_rate": 9.932896157874304e-06,
      "loss": 0.8356,
      "step": 1697
    },
    {
      "epoch": 0.24222539229671897,
      "grad_norm": 0.9696570634841919,
      "learning_rate": 9.932770365402909e-06,
      "loss": 0.7921,
      "step": 1698
    },
    {
      "epoch": 0.24236804564907274,
      "grad_norm": 0.7731328010559082,
      "learning_rate": 9.93264445593501e-06,
      "loss": 0.8394,
      "step": 1699
    },
    {
      "epoch": 0.24251069900142652,
      "grad_norm": 0.7872111797332764,
      "learning_rate": 9.932518429473595e-06,
      "loss": 0.7934,
      "step": 1700
    },
    {
      "epoch": 0.24265335235378033,
      "grad_norm": 0.7657778263092041,
      "learning_rate": 9.932392286021649e-06,
      "loss": 0.8636,
      "step": 1701
    },
    {
      "epoch": 0.2427960057061341,
      "grad_norm": 0.8215643167495728,
      "learning_rate": 9.932266025582165e-06,
      "loss": 0.773,
      "step": 1702
    },
    {
      "epoch": 0.24293865905848788,
      "grad_norm": 0.9458875060081482,
      "learning_rate": 9.932139648158142e-06,
      "loss": 0.7793,
      "step": 1703
    },
    {
      "epoch": 0.24308131241084166,
      "grad_norm": 0.8223559856414795,
      "learning_rate": 9.93201315375257e-06,
      "loss": 0.7971,
      "step": 1704
    },
    {
      "epoch": 0.24322396576319544,
      "grad_norm": 1.04073965549469,
      "learning_rate": 9.931886542368454e-06,
      "loss": 0.773,
      "step": 1705
    },
    {
      "epoch": 0.24336661911554922,
      "grad_norm": 0.8274113535881042,
      "learning_rate": 9.931759814008796e-06,
      "loss": 0.7855,
      "step": 1706
    },
    {
      "epoch": 0.243509272467903,
      "grad_norm": 0.8987480401992798,
      "learning_rate": 9.931632968676603e-06,
      "loss": 0.8289,
      "step": 1707
    },
    {
      "epoch": 0.24365192582025677,
      "grad_norm": 0.7026833295822144,
      "learning_rate": 9.931506006374883e-06,
      "loss": 0.8352,
      "step": 1708
    },
    {
      "epoch": 0.24379457917261055,
      "grad_norm": 0.7738713622093201,
      "learning_rate": 9.931378927106645e-06,
      "loss": 0.7855,
      "step": 1709
    },
    {
      "epoch": 0.24393723252496433,
      "grad_norm": 0.9599910974502563,
      "learning_rate": 9.931251730874905e-06,
      "loss": 0.7995,
      "step": 1710
    },
    {
      "epoch": 0.2440798858773181,
      "grad_norm": 0.9438477754592896,
      "learning_rate": 9.931124417682681e-06,
      "loss": 0.8559,
      "step": 1711
    },
    {
      "epoch": 0.2442225392296719,
      "grad_norm": 0.789526104927063,
      "learning_rate": 9.93099698753299e-06,
      "loss": 0.7975,
      "step": 1712
    },
    {
      "epoch": 0.2443651925820257,
      "grad_norm": 0.7984293103218079,
      "learning_rate": 9.930869440428857e-06,
      "loss": 0.8323,
      "step": 1713
    },
    {
      "epoch": 0.24450784593437946,
      "grad_norm": 1.0544899702072144,
      "learning_rate": 9.930741776373305e-06,
      "loss": 0.8537,
      "step": 1714
    },
    {
      "epoch": 0.24465049928673324,
      "grad_norm": 0.9012507796287537,
      "learning_rate": 9.930613995369362e-06,
      "loss": 0.7284,
      "step": 1715
    },
    {
      "epoch": 0.24479315263908702,
      "grad_norm": 1.0507197380065918,
      "learning_rate": 9.93048609742006e-06,
      "loss": 0.5911,
      "step": 1716
    },
    {
      "epoch": 0.2449358059914408,
      "grad_norm": 0.7745269536972046,
      "learning_rate": 9.930358082528433e-06,
      "loss": 0.7643,
      "step": 1717
    },
    {
      "epoch": 0.24507845934379457,
      "grad_norm": 0.8316447734832764,
      "learning_rate": 9.930229950697518e-06,
      "loss": 0.8697,
      "step": 1718
    },
    {
      "epoch": 0.24522111269614835,
      "grad_norm": 0.8305345773696899,
      "learning_rate": 9.93010170193035e-06,
      "loss": 0.778,
      "step": 1719
    },
    {
      "epoch": 0.24536376604850213,
      "grad_norm": 0.9902516007423401,
      "learning_rate": 9.929973336229975e-06,
      "loss": 0.8369,
      "step": 1720
    },
    {
      "epoch": 0.2455064194008559,
      "grad_norm": 0.809398353099823,
      "learning_rate": 9.929844853599434e-06,
      "loss": 0.7423,
      "step": 1721
    },
    {
      "epoch": 0.2456490727532097,
      "grad_norm": 1.2137434482574463,
      "learning_rate": 9.929716254041779e-06,
      "loss": 0.8325,
      "step": 1722
    },
    {
      "epoch": 0.2457917261055635,
      "grad_norm": 0.9980698227882385,
      "learning_rate": 9.929587537560055e-06,
      "loss": 0.8153,
      "step": 1723
    },
    {
      "epoch": 0.24593437945791727,
      "grad_norm": 1.2068803310394287,
      "learning_rate": 9.929458704157319e-06,
      "loss": 0.7866,
      "step": 1724
    },
    {
      "epoch": 0.24607703281027105,
      "grad_norm": 0.8856320977210999,
      "learning_rate": 9.929329753836624e-06,
      "loss": 0.8647,
      "step": 1725
    },
    {
      "epoch": 0.24621968616262482,
      "grad_norm": 0.7726538777351379,
      "learning_rate": 9.92920068660103e-06,
      "loss": 0.835,
      "step": 1726
    },
    {
      "epoch": 0.2463623395149786,
      "grad_norm": 0.9689743518829346,
      "learning_rate": 9.929071502453599e-06,
      "loss": 0.847,
      "step": 1727
    },
    {
      "epoch": 0.24650499286733238,
      "grad_norm": 0.8871773481369019,
      "learning_rate": 9.928942201397391e-06,
      "loss": 0.796,
      "step": 1728
    },
    {
      "epoch": 0.24664764621968616,
      "grad_norm": 0.8159291744232178,
      "learning_rate": 9.928812783435477e-06,
      "loss": 0.8607,
      "step": 1729
    },
    {
      "epoch": 0.24679029957203993,
      "grad_norm": 0.8086974620819092,
      "learning_rate": 9.928683248570928e-06,
      "loss": 0.8934,
      "step": 1730
    },
    {
      "epoch": 0.2469329529243937,
      "grad_norm": 0.9735966324806213,
      "learning_rate": 9.928553596806809e-06,
      "loss": 0.848,
      "step": 1731
    },
    {
      "epoch": 0.24707560627674752,
      "grad_norm": 0.688496470451355,
      "learning_rate": 9.928423828146202e-06,
      "loss": 0.7779,
      "step": 1732
    },
    {
      "epoch": 0.2472182596291013,
      "grad_norm": 0.8273606896400452,
      "learning_rate": 9.928293942592182e-06,
      "loss": 0.8101,
      "step": 1733
    },
    {
      "epoch": 0.24736091298145507,
      "grad_norm": 0.8033018708229065,
      "learning_rate": 9.92816394014783e-06,
      "loss": 0.8352,
      "step": 1734
    },
    {
      "epoch": 0.24750356633380885,
      "grad_norm": 0.8431486487388611,
      "learning_rate": 9.928033820816231e-06,
      "loss": 0.7885,
      "step": 1735
    },
    {
      "epoch": 0.24764621968616263,
      "grad_norm": 1.2040809392929077,
      "learning_rate": 9.927903584600468e-06,
      "loss": 0.8404,
      "step": 1736
    },
    {
      "epoch": 0.2477888730385164,
      "grad_norm": 0.8894022703170776,
      "learning_rate": 9.927773231503633e-06,
      "loss": 0.7737,
      "step": 1737
    },
    {
      "epoch": 0.24793152639087018,
      "grad_norm": 0.703839123249054,
      "learning_rate": 9.927642761528815e-06,
      "loss": 0.8407,
      "step": 1738
    },
    {
      "epoch": 0.24807417974322396,
      "grad_norm": 0.8526538014411926,
      "learning_rate": 9.927512174679111e-06,
      "loss": 0.8415,
      "step": 1739
    },
    {
      "epoch": 0.24821683309557774,
      "grad_norm": 1.0738188028335571,
      "learning_rate": 9.927381470957617e-06,
      "loss": 0.8345,
      "step": 1740
    },
    {
      "epoch": 0.24835948644793152,
      "grad_norm": 1.0603893995285034,
      "learning_rate": 9.927250650367435e-06,
      "loss": 0.7458,
      "step": 1741
    },
    {
      "epoch": 0.2485021398002853,
      "grad_norm": 0.9537172913551331,
      "learning_rate": 9.927119712911664e-06,
      "loss": 0.7977,
      "step": 1742
    },
    {
      "epoch": 0.2486447931526391,
      "grad_norm": 1.0061439275741577,
      "learning_rate": 9.926988658593413e-06,
      "loss": 0.746,
      "step": 1743
    },
    {
      "epoch": 0.24878744650499288,
      "grad_norm": 0.6799601912498474,
      "learning_rate": 9.92685748741579e-06,
      "loss": 0.9128,
      "step": 1744
    },
    {
      "epoch": 0.24893009985734665,
      "grad_norm": 0.8178061246871948,
      "learning_rate": 9.926726199381902e-06,
      "loss": 0.7994,
      "step": 1745
    },
    {
      "epoch": 0.24907275320970043,
      "grad_norm": 0.756705641746521,
      "learning_rate": 9.926594794494867e-06,
      "loss": 0.9008,
      "step": 1746
    },
    {
      "epoch": 0.2492154065620542,
      "grad_norm": 0.9939935803413391,
      "learning_rate": 9.926463272757805e-06,
      "loss": 0.7959,
      "step": 1747
    },
    {
      "epoch": 0.249358059914408,
      "grad_norm": 1.0848501920700073,
      "learning_rate": 9.926331634173827e-06,
      "loss": 0.7923,
      "step": 1748
    },
    {
      "epoch": 0.24950071326676176,
      "grad_norm": 1.0696003437042236,
      "learning_rate": 9.926199878746061e-06,
      "loss": 0.8095,
      "step": 1749
    },
    {
      "epoch": 0.24964336661911554,
      "grad_norm": 0.775341272354126,
      "learning_rate": 9.92606800647763e-06,
      "loss": 0.8407,
      "step": 1750
    },
    {
      "epoch": 0.24978601997146932,
      "grad_norm": 0.8625017404556274,
      "learning_rate": 9.925936017371664e-06,
      "loss": 0.8632,
      "step": 1751
    },
    {
      "epoch": 0.2499286733238231,
      "grad_norm": 1.0839468240737915,
      "learning_rate": 9.925803911431292e-06,
      "loss": 0.7061,
      "step": 1752
    },
    {
      "epoch": 0.2500713266761769,
      "grad_norm": 0.9675854444503784,
      "learning_rate": 9.925671688659646e-06,
      "loss": 0.8347,
      "step": 1753
    },
    {
      "epoch": 0.2502139800285307,
      "grad_norm": 0.6596314907073975,
      "learning_rate": 9.925539349059864e-06,
      "loss": 0.7942,
      "step": 1754
    },
    {
      "epoch": 0.25035663338088443,
      "grad_norm": 0.8776452541351318,
      "learning_rate": 9.925406892635083e-06,
      "loss": 0.7596,
      "step": 1755
    },
    {
      "epoch": 0.25049928673323824,
      "grad_norm": 0.9428473114967346,
      "learning_rate": 9.925274319388447e-06,
      "loss": 0.7694,
      "step": 1756
    },
    {
      "epoch": 0.250641940085592,
      "grad_norm": 1.1374789476394653,
      "learning_rate": 9.9251416293231e-06,
      "loss": 0.7954,
      "step": 1757
    },
    {
      "epoch": 0.2507845934379458,
      "grad_norm": 1.1859643459320068,
      "learning_rate": 9.925008822442187e-06,
      "loss": 0.8431,
      "step": 1758
    },
    {
      "epoch": 0.2509272467902996,
      "grad_norm": 0.7116193771362305,
      "learning_rate": 9.924875898748861e-06,
      "loss": 0.8564,
      "step": 1759
    },
    {
      "epoch": 0.25106990014265335,
      "grad_norm": 0.8621867299079895,
      "learning_rate": 9.924742858246271e-06,
      "loss": 0.7682,
      "step": 1760
    },
    {
      "epoch": 0.25121255349500715,
      "grad_norm": 0.7891177535057068,
      "learning_rate": 9.924609700937575e-06,
      "loss": 0.8185,
      "step": 1761
    },
    {
      "epoch": 0.2513552068473609,
      "grad_norm": 0.8107175827026367,
      "learning_rate": 9.92447642682593e-06,
      "loss": 0.7584,
      "step": 1762
    },
    {
      "epoch": 0.2514978601997147,
      "grad_norm": 0.7248867154121399,
      "learning_rate": 9.9243430359145e-06,
      "loss": 0.8432,
      "step": 1763
    },
    {
      "epoch": 0.25164051355206846,
      "grad_norm": 0.9015504121780396,
      "learning_rate": 9.924209528206446e-06,
      "loss": 0.7818,
      "step": 1764
    },
    {
      "epoch": 0.25178316690442226,
      "grad_norm": 0.8606206774711609,
      "learning_rate": 9.924075903704935e-06,
      "loss": 0.821,
      "step": 1765
    },
    {
      "epoch": 0.251925820256776,
      "grad_norm": 0.7115620374679565,
      "learning_rate": 9.923942162413136e-06,
      "loss": 0.8123,
      "step": 1766
    },
    {
      "epoch": 0.2520684736091298,
      "grad_norm": 0.8173115849494934,
      "learning_rate": 9.923808304334222e-06,
      "loss": 0.8171,
      "step": 1767
    },
    {
      "epoch": 0.25221112696148357,
      "grad_norm": 0.7100493311882019,
      "learning_rate": 9.923674329471366e-06,
      "loss": 0.8178,
      "step": 1768
    },
    {
      "epoch": 0.2523537803138374,
      "grad_norm": 0.7646355032920837,
      "learning_rate": 9.92354023782775e-06,
      "loss": 0.815,
      "step": 1769
    },
    {
      "epoch": 0.2524964336661912,
      "grad_norm": 0.9536492228507996,
      "learning_rate": 9.92340602940655e-06,
      "loss": 0.8257,
      "step": 1770
    },
    {
      "epoch": 0.25263908701854493,
      "grad_norm": 0.9343258142471313,
      "learning_rate": 9.92327170421095e-06,
      "loss": 0.7676,
      "step": 1771
    },
    {
      "epoch": 0.25278174037089873,
      "grad_norm": 1.1464265584945679,
      "learning_rate": 9.923137262244138e-06,
      "loss": 0.8456,
      "step": 1772
    },
    {
      "epoch": 0.2529243937232525,
      "grad_norm": 0.8079214096069336,
      "learning_rate": 9.923002703509301e-06,
      "loss": 0.8753,
      "step": 1773
    },
    {
      "epoch": 0.2530670470756063,
      "grad_norm": 0.778343677520752,
      "learning_rate": 9.92286802800963e-06,
      "loss": 0.8405,
      "step": 1774
    },
    {
      "epoch": 0.25320970042796004,
      "grad_norm": 0.937049925327301,
      "learning_rate": 9.92273323574832e-06,
      "loss": 0.8797,
      "step": 1775
    },
    {
      "epoch": 0.25335235378031384,
      "grad_norm": 1.1116619110107422,
      "learning_rate": 9.92259832672857e-06,
      "loss": 0.7696,
      "step": 1776
    },
    {
      "epoch": 0.2534950071326676,
      "grad_norm": 0.7162361145019531,
      "learning_rate": 9.922463300953578e-06,
      "loss": 0.8254,
      "step": 1777
    },
    {
      "epoch": 0.2536376604850214,
      "grad_norm": 0.7013463973999023,
      "learning_rate": 9.922328158426545e-06,
      "loss": 0.7982,
      "step": 1778
    },
    {
      "epoch": 0.2537803138373752,
      "grad_norm": 1.1140834093093872,
      "learning_rate": 9.922192899150679e-06,
      "loss": 0.8328,
      "step": 1779
    },
    {
      "epoch": 0.25392296718972895,
      "grad_norm": 1.1142244338989258,
      "learning_rate": 9.922057523129187e-06,
      "loss": 0.8104,
      "step": 1780
    },
    {
      "epoch": 0.25406562054208276,
      "grad_norm": 0.9946165680885315,
      "learning_rate": 9.921922030365278e-06,
      "loss": 0.7862,
      "step": 1781
    },
    {
      "epoch": 0.2542082738944365,
      "grad_norm": 0.701593816280365,
      "learning_rate": 9.92178642086217e-06,
      "loss": 0.8262,
      "step": 1782
    },
    {
      "epoch": 0.2543509272467903,
      "grad_norm": 1.4632997512817383,
      "learning_rate": 9.921650694623077e-06,
      "loss": 0.6542,
      "step": 1783
    },
    {
      "epoch": 0.25449358059914406,
      "grad_norm": 0.7976811528205872,
      "learning_rate": 9.921514851651216e-06,
      "loss": 0.8723,
      "step": 1784
    },
    {
      "epoch": 0.25463623395149787,
      "grad_norm": 0.7965141534805298,
      "learning_rate": 9.921378891949813e-06,
      "loss": 0.8243,
      "step": 1785
    },
    {
      "epoch": 0.2547788873038516,
      "grad_norm": 1.0921205282211304,
      "learning_rate": 9.921242815522089e-06,
      "loss": 0.7583,
      "step": 1786
    },
    {
      "epoch": 0.2549215406562054,
      "grad_norm": 0.7497575879096985,
      "learning_rate": 9.921106622371275e-06,
      "loss": 0.8354,
      "step": 1787
    },
    {
      "epoch": 0.2550641940085592,
      "grad_norm": 1.1817249059677124,
      "learning_rate": 9.9209703125006e-06,
      "loss": 0.6963,
      "step": 1788
    },
    {
      "epoch": 0.255206847360913,
      "grad_norm": 0.8144768476486206,
      "learning_rate": 9.920833885913295e-06,
      "loss": 0.8296,
      "step": 1789
    },
    {
      "epoch": 0.2553495007132668,
      "grad_norm": 0.8028969764709473,
      "learning_rate": 9.920697342612598e-06,
      "loss": 0.7478,
      "step": 1790
    },
    {
      "epoch": 0.25549215406562054,
      "grad_norm": 0.7523711919784546,
      "learning_rate": 9.920560682601746e-06,
      "loss": 0.7942,
      "step": 1791
    },
    {
      "epoch": 0.25563480741797434,
      "grad_norm": 0.7429263591766357,
      "learning_rate": 9.920423905883982e-06,
      "loss": 0.8674,
      "step": 1792
    },
    {
      "epoch": 0.2557774607703281,
      "grad_norm": 0.8647870421409607,
      "learning_rate": 9.92028701246255e-06,
      "loss": 0.8352,
      "step": 1793
    },
    {
      "epoch": 0.2559201141226819,
      "grad_norm": 0.9710491895675659,
      "learning_rate": 9.920150002340696e-06,
      "loss": 0.8285,
      "step": 1794
    },
    {
      "epoch": 0.25606276747503565,
      "grad_norm": 0.8176320791244507,
      "learning_rate": 9.92001287552167e-06,
      "loss": 0.7461,
      "step": 1795
    },
    {
      "epoch": 0.25620542082738945,
      "grad_norm": 1.0631942749023438,
      "learning_rate": 9.919875632008724e-06,
      "loss": 0.7482,
      "step": 1796
    },
    {
      "epoch": 0.2563480741797432,
      "grad_norm": 1.1244279146194458,
      "learning_rate": 9.919738271805112e-06,
      "loss": 0.7229,
      "step": 1797
    },
    {
      "epoch": 0.256490727532097,
      "grad_norm": 0.9116760492324829,
      "learning_rate": 9.919600794914094e-06,
      "loss": 0.8452,
      "step": 1798
    },
    {
      "epoch": 0.25663338088445076,
      "grad_norm": 0.9942961931228638,
      "learning_rate": 9.919463201338933e-06,
      "loss": 0.7275,
      "step": 1799
    },
    {
      "epoch": 0.25677603423680456,
      "grad_norm": 0.9490411877632141,
      "learning_rate": 9.919325491082888e-06,
      "loss": 0.909,
      "step": 1800
    },
    {
      "epoch": 0.25691868758915837,
      "grad_norm": 0.9126991033554077,
      "learning_rate": 9.919187664149226e-06,
      "loss": 0.8144,
      "step": 1801
    },
    {
      "epoch": 0.2570613409415121,
      "grad_norm": 0.8840201497077942,
      "learning_rate": 9.919049720541216e-06,
      "loss": 0.832,
      "step": 1802
    },
    {
      "epoch": 0.2572039942938659,
      "grad_norm": 0.9308497905731201,
      "learning_rate": 9.91891166026213e-06,
      "loss": 0.8694,
      "step": 1803
    },
    {
      "epoch": 0.2573466476462197,
      "grad_norm": 0.8584593534469604,
      "learning_rate": 9.918773483315246e-06,
      "loss": 0.8431,
      "step": 1804
    },
    {
      "epoch": 0.2574893009985735,
      "grad_norm": 1.003719449043274,
      "learning_rate": 9.918635189703838e-06,
      "loss": 0.8484,
      "step": 1805
    },
    {
      "epoch": 0.25763195435092723,
      "grad_norm": 0.8581739068031311,
      "learning_rate": 9.918496779431186e-06,
      "loss": 0.7958,
      "step": 1806
    },
    {
      "epoch": 0.25777460770328103,
      "grad_norm": 1.4660345315933228,
      "learning_rate": 9.918358252500572e-06,
      "loss": 0.7827,
      "step": 1807
    },
    {
      "epoch": 0.2579172610556348,
      "grad_norm": 0.6751096248626709,
      "learning_rate": 9.918219608915285e-06,
      "loss": 0.7967,
      "step": 1808
    },
    {
      "epoch": 0.2580599144079886,
      "grad_norm": 0.9812304377555847,
      "learning_rate": 9.918080848678611e-06,
      "loss": 0.803,
      "step": 1809
    },
    {
      "epoch": 0.2582025677603424,
      "grad_norm": 0.9044556617736816,
      "learning_rate": 9.91794197179384e-06,
      "loss": 0.8133,
      "step": 1810
    },
    {
      "epoch": 0.25834522111269614,
      "grad_norm": 0.9294826984405518,
      "learning_rate": 9.917802978264269e-06,
      "loss": 0.7915,
      "step": 1811
    },
    {
      "epoch": 0.25848787446504995,
      "grad_norm": 0.8307023048400879,
      "learning_rate": 9.917663868093192e-06,
      "loss": 0.8041,
      "step": 1812
    },
    {
      "epoch": 0.2586305278174037,
      "grad_norm": 0.684723973274231,
      "learning_rate": 9.917524641283911e-06,
      "loss": 0.7893,
      "step": 1813
    },
    {
      "epoch": 0.2587731811697575,
      "grad_norm": 0.8535851240158081,
      "learning_rate": 9.917385297839725e-06,
      "loss": 0.7872,
      "step": 1814
    },
    {
      "epoch": 0.25891583452211125,
      "grad_norm": 1.284294605255127,
      "learning_rate": 9.917245837763942e-06,
      "loss": 0.658,
      "step": 1815
    },
    {
      "epoch": 0.25905848787446506,
      "grad_norm": 0.8843228816986084,
      "learning_rate": 9.917106261059867e-06,
      "loss": 0.8177,
      "step": 1816
    },
    {
      "epoch": 0.2592011412268188,
      "grad_norm": 0.8427625894546509,
      "learning_rate": 9.916966567730815e-06,
      "loss": 0.7868,
      "step": 1817
    },
    {
      "epoch": 0.2593437945791726,
      "grad_norm": 0.790692925453186,
      "learning_rate": 9.916826757780094e-06,
      "loss": 0.7719,
      "step": 1818
    },
    {
      "epoch": 0.25948644793152637,
      "grad_norm": 0.9854372143745422,
      "learning_rate": 9.916686831211022e-06,
      "loss": 0.8206,
      "step": 1819
    },
    {
      "epoch": 0.25962910128388017,
      "grad_norm": 0.8204353451728821,
      "learning_rate": 9.91654678802692e-06,
      "loss": 0.8109,
      "step": 1820
    },
    {
      "epoch": 0.259771754636234,
      "grad_norm": 0.9865830540657043,
      "learning_rate": 9.916406628231105e-06,
      "loss": 0.7548,
      "step": 1821
    },
    {
      "epoch": 0.2599144079885877,
      "grad_norm": 0.8347112536430359,
      "learning_rate": 9.916266351826907e-06,
      "loss": 0.8322,
      "step": 1822
    },
    {
      "epoch": 0.26005706134094153,
      "grad_norm": 0.9512296915054321,
      "learning_rate": 9.916125958817649e-06,
      "loss": 0.7825,
      "step": 1823
    },
    {
      "epoch": 0.2601997146932953,
      "grad_norm": 1.1798293590545654,
      "learning_rate": 9.91598544920666e-06,
      "loss": 0.8193,
      "step": 1824
    },
    {
      "epoch": 0.2603423680456491,
      "grad_norm": 0.9011582136154175,
      "learning_rate": 9.915844822997276e-06,
      "loss": 0.7892,
      "step": 1825
    },
    {
      "epoch": 0.26048502139800284,
      "grad_norm": 0.6729601621627808,
      "learning_rate": 9.915704080192832e-06,
      "loss": 0.8465,
      "step": 1826
    },
    {
      "epoch": 0.26062767475035664,
      "grad_norm": 0.8563498854637146,
      "learning_rate": 9.915563220796665e-06,
      "loss": 0.7662,
      "step": 1827
    },
    {
      "epoch": 0.2607703281027104,
      "grad_norm": 0.7469445466995239,
      "learning_rate": 9.915422244812115e-06,
      "loss": 0.8521,
      "step": 1828
    },
    {
      "epoch": 0.2609129814550642,
      "grad_norm": 0.9483319520950317,
      "learning_rate": 9.915281152242528e-06,
      "loss": 0.8422,
      "step": 1829
    },
    {
      "epoch": 0.26105563480741795,
      "grad_norm": 0.7566171288490295,
      "learning_rate": 9.915139943091248e-06,
      "loss": 0.8238,
      "step": 1830
    },
    {
      "epoch": 0.26119828815977175,
      "grad_norm": 0.6985406279563904,
      "learning_rate": 9.914998617361626e-06,
      "loss": 0.786,
      "step": 1831
    },
    {
      "epoch": 0.26134094151212556,
      "grad_norm": 0.8736724853515625,
      "learning_rate": 9.914857175057015e-06,
      "loss": 0.8642,
      "step": 1832
    },
    {
      "epoch": 0.2614835948644793,
      "grad_norm": 0.8644083142280579,
      "learning_rate": 9.914715616180768e-06,
      "loss": 0.7676,
      "step": 1833
    },
    {
      "epoch": 0.2616262482168331,
      "grad_norm": 0.7934670448303223,
      "learning_rate": 9.914573940736241e-06,
      "loss": 0.8021,
      "step": 1834
    },
    {
      "epoch": 0.26176890156918686,
      "grad_norm": 0.8259593844413757,
      "learning_rate": 9.914432148726799e-06,
      "loss": 0.8874,
      "step": 1835
    },
    {
      "epoch": 0.26191155492154067,
      "grad_norm": 0.8342422842979431,
      "learning_rate": 9.9142902401558e-06,
      "loss": 0.8138,
      "step": 1836
    },
    {
      "epoch": 0.2620542082738944,
      "grad_norm": 0.6923991441726685,
      "learning_rate": 9.914148215026612e-06,
      "loss": 0.8096,
      "step": 1837
    },
    {
      "epoch": 0.2621968616262482,
      "grad_norm": 0.7407599687576294,
      "learning_rate": 9.914006073342604e-06,
      "loss": 0.874,
      "step": 1838
    },
    {
      "epoch": 0.262339514978602,
      "grad_norm": 0.984772264957428,
      "learning_rate": 9.913863815107146e-06,
      "loss": 0.821,
      "step": 1839
    },
    {
      "epoch": 0.2624821683309558,
      "grad_norm": 1.0290112495422363,
      "learning_rate": 9.913721440323613e-06,
      "loss": 0.7882,
      "step": 1840
    },
    {
      "epoch": 0.2626248216833096,
      "grad_norm": 0.8359882235527039,
      "learning_rate": 9.913578948995385e-06,
      "loss": 0.8609,
      "step": 1841
    },
    {
      "epoch": 0.26276747503566333,
      "grad_norm": 1.0276292562484741,
      "learning_rate": 9.913436341125834e-06,
      "loss": 0.8341,
      "step": 1842
    },
    {
      "epoch": 0.26291012838801714,
      "grad_norm": 0.8084568381309509,
      "learning_rate": 9.91329361671835e-06,
      "loss": 0.7654,
      "step": 1843
    },
    {
      "epoch": 0.2630527817403709,
      "grad_norm": 0.9717106819152832,
      "learning_rate": 9.913150775776314e-06,
      "loss": 0.7615,
      "step": 1844
    },
    {
      "epoch": 0.2631954350927247,
      "grad_norm": 1.142538070678711,
      "learning_rate": 9.913007818303115e-06,
      "loss": 0.8357,
      "step": 1845
    },
    {
      "epoch": 0.26333808844507844,
      "grad_norm": 0.9166866540908813,
      "learning_rate": 9.912864744302145e-06,
      "loss": 0.7173,
      "step": 1846
    },
    {
      "epoch": 0.26348074179743225,
      "grad_norm": 1.0183794498443604,
      "learning_rate": 9.912721553776793e-06,
      "loss": 0.7675,
      "step": 1847
    },
    {
      "epoch": 0.263623395149786,
      "grad_norm": 0.7247642874717712,
      "learning_rate": 9.912578246730462e-06,
      "loss": 0.8502,
      "step": 1848
    },
    {
      "epoch": 0.2637660485021398,
      "grad_norm": 0.8331905603408813,
      "learning_rate": 9.912434823166545e-06,
      "loss": 0.7808,
      "step": 1849
    },
    {
      "epoch": 0.26390870185449355,
      "grad_norm": 0.9375711679458618,
      "learning_rate": 9.912291283088447e-06,
      "loss": 0.8247,
      "step": 1850
    },
    {
      "epoch": 0.26405135520684736,
      "grad_norm": 0.7520945072174072,
      "learning_rate": 9.912147626499573e-06,
      "loss": 0.8518,
      "step": 1851
    },
    {
      "epoch": 0.26419400855920117,
      "grad_norm": 0.8544954061508179,
      "learning_rate": 9.912003853403327e-06,
      "loss": 0.7918,
      "step": 1852
    },
    {
      "epoch": 0.2643366619115549,
      "grad_norm": 0.8004382848739624,
      "learning_rate": 9.911859963803122e-06,
      "loss": 0.9079,
      "step": 1853
    },
    {
      "epoch": 0.2644793152639087,
      "grad_norm": 0.9976580142974854,
      "learning_rate": 9.91171595770237e-06,
      "loss": 0.9102,
      "step": 1854
    },
    {
      "epoch": 0.26462196861626247,
      "grad_norm": 0.8790906667709351,
      "learning_rate": 9.911571835104485e-06,
      "loss": 0.8299,
      "step": 1855
    },
    {
      "epoch": 0.2647646219686163,
      "grad_norm": 0.6801099181175232,
      "learning_rate": 9.911427596012889e-06,
      "loss": 0.846,
      "step": 1856
    },
    {
      "epoch": 0.26490727532097,
      "grad_norm": 1.0817416906356812,
      "learning_rate": 9.911283240430998e-06,
      "loss": 0.775,
      "step": 1857
    },
    {
      "epoch": 0.26504992867332383,
      "grad_norm": 0.8968457579612732,
      "learning_rate": 9.911138768362241e-06,
      "loss": 0.7453,
      "step": 1858
    },
    {
      "epoch": 0.2651925820256776,
      "grad_norm": 0.7213981747627258,
      "learning_rate": 9.91099417981004e-06,
      "loss": 0.8386,
      "step": 1859
    },
    {
      "epoch": 0.2653352353780314,
      "grad_norm": 0.918195903301239,
      "learning_rate": 9.910849474777828e-06,
      "loss": 0.8895,
      "step": 1860
    },
    {
      "epoch": 0.26547788873038514,
      "grad_norm": 0.8049584627151489,
      "learning_rate": 9.910704653269037e-06,
      "loss": 0.8433,
      "step": 1861
    },
    {
      "epoch": 0.26562054208273894,
      "grad_norm": 0.7233972549438477,
      "learning_rate": 9.910559715287099e-06,
      "loss": 0.8634,
      "step": 1862
    },
    {
      "epoch": 0.26576319543509275,
      "grad_norm": 0.7243645191192627,
      "learning_rate": 9.910414660835453e-06,
      "loss": 0.8012,
      "step": 1863
    },
    {
      "epoch": 0.2659058487874465,
      "grad_norm": 0.7203574776649475,
      "learning_rate": 9.91026948991754e-06,
      "loss": 0.7966,
      "step": 1864
    },
    {
      "epoch": 0.2660485021398003,
      "grad_norm": 0.7382854223251343,
      "learning_rate": 9.910124202536803e-06,
      "loss": 0.8403,
      "step": 1865
    },
    {
      "epoch": 0.26619115549215405,
      "grad_norm": 1.0563346147537231,
      "learning_rate": 9.90997879869669e-06,
      "loss": 0.9433,
      "step": 1866
    },
    {
      "epoch": 0.26633380884450786,
      "grad_norm": 0.8045656681060791,
      "learning_rate": 9.909833278400648e-06,
      "loss": 0.8359,
      "step": 1867
    },
    {
      "epoch": 0.2664764621968616,
      "grad_norm": 0.7768740057945251,
      "learning_rate": 9.909687641652125e-06,
      "loss": 0.7659,
      "step": 1868
    },
    {
      "epoch": 0.2666191155492154,
      "grad_norm": 0.7277352213859558,
      "learning_rate": 9.909541888454579e-06,
      "loss": 0.8444,
      "step": 1869
    },
    {
      "epoch": 0.26676176890156916,
      "grad_norm": 0.7429260015487671,
      "learning_rate": 9.909396018811468e-06,
      "loss": 0.821,
      "step": 1870
    },
    {
      "epoch": 0.26690442225392297,
      "grad_norm": 0.8484064340591431,
      "learning_rate": 9.909250032726248e-06,
      "loss": 0.7954,
      "step": 1871
    },
    {
      "epoch": 0.2670470756062768,
      "grad_norm": 0.7717201113700867,
      "learning_rate": 9.909103930202387e-06,
      "loss": 0.7197,
      "step": 1872
    },
    {
      "epoch": 0.2671897289586305,
      "grad_norm": 0.7083487510681152,
      "learning_rate": 9.908957711243344e-06,
      "loss": 0.8331,
      "step": 1873
    },
    {
      "epoch": 0.26733238231098433,
      "grad_norm": 1.0699069499969482,
      "learning_rate": 9.90881137585259e-06,
      "loss": 0.7806,
      "step": 1874
    },
    {
      "epoch": 0.2674750356633381,
      "grad_norm": 0.7462451457977295,
      "learning_rate": 9.908664924033597e-06,
      "loss": 0.8331,
      "step": 1875
    },
    {
      "epoch": 0.2676176890156919,
      "grad_norm": 1.1656798124313354,
      "learning_rate": 9.908518355789838e-06,
      "loss": 0.8717,
      "step": 1876
    },
    {
      "epoch": 0.26776034236804563,
      "grad_norm": 0.9811933040618896,
      "learning_rate": 9.908371671124788e-06,
      "loss": 0.7728,
      "step": 1877
    },
    {
      "epoch": 0.26790299572039944,
      "grad_norm": 0.8236185908317566,
      "learning_rate": 9.908224870041927e-06,
      "loss": 0.7903,
      "step": 1878
    },
    {
      "epoch": 0.2680456490727532,
      "grad_norm": 0.9544426202774048,
      "learning_rate": 9.908077952544738e-06,
      "loss": 0.7954,
      "step": 1879
    },
    {
      "epoch": 0.268188302425107,
      "grad_norm": 0.7393714189529419,
      "learning_rate": 9.9079309186367e-06,
      "loss": 0.7989,
      "step": 1880
    },
    {
      "epoch": 0.26833095577746074,
      "grad_norm": 0.9020375609397888,
      "learning_rate": 9.907783768321308e-06,
      "loss": 0.8449,
      "step": 1881
    },
    {
      "epoch": 0.26847360912981455,
      "grad_norm": 0.8916321992874146,
      "learning_rate": 9.907636501602049e-06,
      "loss": 0.8259,
      "step": 1882
    },
    {
      "epoch": 0.26861626248216836,
      "grad_norm": 0.878184974193573,
      "learning_rate": 9.907489118482416e-06,
      "loss": 0.8694,
      "step": 1883
    },
    {
      "epoch": 0.2687589158345221,
      "grad_norm": 0.8452358841896057,
      "learning_rate": 9.907341618965904e-06,
      "loss": 0.8481,
      "step": 1884
    },
    {
      "epoch": 0.2689015691868759,
      "grad_norm": 0.7845661640167236,
      "learning_rate": 9.907194003056011e-06,
      "loss": 0.8263,
      "step": 1885
    },
    {
      "epoch": 0.26904422253922966,
      "grad_norm": 0.8402097821235657,
      "learning_rate": 9.90704627075624e-06,
      "loss": 0.8312,
      "step": 1886
    },
    {
      "epoch": 0.26918687589158347,
      "grad_norm": 0.8752978444099426,
      "learning_rate": 9.906898422070095e-06,
      "loss": 0.7898,
      "step": 1887
    },
    {
      "epoch": 0.2693295292439372,
      "grad_norm": 0.793667733669281,
      "learning_rate": 9.90675045700108e-06,
      "loss": 0.7785,
      "step": 1888
    },
    {
      "epoch": 0.269472182596291,
      "grad_norm": 0.8000465035438538,
      "learning_rate": 9.906602375552706e-06,
      "loss": 0.7866,
      "step": 1889
    },
    {
      "epoch": 0.26961483594864477,
      "grad_norm": 0.9185377955436707,
      "learning_rate": 9.906454177728486e-06,
      "loss": 0.8018,
      "step": 1890
    },
    {
      "epoch": 0.2697574893009986,
      "grad_norm": 0.8185838460922241,
      "learning_rate": 9.906305863531933e-06,
      "loss": 0.8672,
      "step": 1891
    },
    {
      "epoch": 0.2699001426533523,
      "grad_norm": 0.7808707356452942,
      "learning_rate": 9.906157432966569e-06,
      "loss": 0.7791,
      "step": 1892
    },
    {
      "epoch": 0.27004279600570613,
      "grad_norm": 0.9372024536132812,
      "learning_rate": 9.906008886035909e-06,
      "loss": 0.8064,
      "step": 1893
    },
    {
      "epoch": 0.27018544935805994,
      "grad_norm": 0.8064765930175781,
      "learning_rate": 9.90586022274348e-06,
      "loss": 0.8222,
      "step": 1894
    },
    {
      "epoch": 0.2703281027104137,
      "grad_norm": 0.7457480430603027,
      "learning_rate": 9.905711443092809e-06,
      "loss": 0.8121,
      "step": 1895
    },
    {
      "epoch": 0.2704707560627675,
      "grad_norm": 0.8527641296386719,
      "learning_rate": 9.90556254708742e-06,
      "loss": 0.7355,
      "step": 1896
    },
    {
      "epoch": 0.27061340941512124,
      "grad_norm": 1.4741696119308472,
      "learning_rate": 9.905413534730848e-06,
      "loss": 0.6868,
      "step": 1897
    },
    {
      "epoch": 0.27075606276747505,
      "grad_norm": 0.8204596638679504,
      "learning_rate": 9.905264406026627e-06,
      "loss": 0.7794,
      "step": 1898
    },
    {
      "epoch": 0.2708987161198288,
      "grad_norm": 0.9594339728355408,
      "learning_rate": 9.905115160978292e-06,
      "loss": 0.774,
      "step": 1899
    },
    {
      "epoch": 0.2710413694721826,
      "grad_norm": 1.2117671966552734,
      "learning_rate": 9.904965799589385e-06,
      "loss": 0.8822,
      "step": 1900
    },
    {
      "epoch": 0.27118402282453635,
      "grad_norm": 0.9087314605712891,
      "learning_rate": 9.904816321863449e-06,
      "loss": 0.7369,
      "step": 1901
    },
    {
      "epoch": 0.27132667617689016,
      "grad_norm": 0.9013188481330872,
      "learning_rate": 9.90466672780403e-06,
      "loss": 0.8629,
      "step": 1902
    },
    {
      "epoch": 0.27146932952924396,
      "grad_norm": 0.8729089498519897,
      "learning_rate": 9.904517017414672e-06,
      "loss": 0.7628,
      "step": 1903
    },
    {
      "epoch": 0.2716119828815977,
      "grad_norm": 1.22653329372406,
      "learning_rate": 9.904367190698929e-06,
      "loss": 0.8104,
      "step": 1904
    },
    {
      "epoch": 0.2717546362339515,
      "grad_norm": 0.7489973306655884,
      "learning_rate": 9.904217247660352e-06,
      "loss": 0.5759,
      "step": 1905
    },
    {
      "epoch": 0.27189728958630527,
      "grad_norm": 0.950798511505127,
      "learning_rate": 9.904067188302502e-06,
      "loss": 0.8215,
      "step": 1906
    },
    {
      "epoch": 0.2720399429386591,
      "grad_norm": 1.006636619567871,
      "learning_rate": 9.903917012628936e-06,
      "loss": 0.8021,
      "step": 1907
    },
    {
      "epoch": 0.2721825962910128,
      "grad_norm": 0.9519075751304626,
      "learning_rate": 9.903766720643213e-06,
      "loss": 0.805,
      "step": 1908
    },
    {
      "epoch": 0.27232524964336663,
      "grad_norm": 1.1060423851013184,
      "learning_rate": 9.903616312348903e-06,
      "loss": 0.7663,
      "step": 1909
    },
    {
      "epoch": 0.2724679029957204,
      "grad_norm": 0.922545850276947,
      "learning_rate": 9.903465787749567e-06,
      "loss": 0.7222,
      "step": 1910
    },
    {
      "epoch": 0.2726105563480742,
      "grad_norm": 0.8590786457061768,
      "learning_rate": 9.903315146848782e-06,
      "loss": 0.8197,
      "step": 1911
    },
    {
      "epoch": 0.27275320970042793,
      "grad_norm": 0.8543412089347839,
      "learning_rate": 9.903164389650115e-06,
      "loss": 0.7675,
      "step": 1912
    },
    {
      "epoch": 0.27289586305278174,
      "grad_norm": 0.8302007913589478,
      "learning_rate": 9.903013516157145e-06,
      "loss": 0.8347,
      "step": 1913
    },
    {
      "epoch": 0.27303851640513555,
      "grad_norm": 0.9529709815979004,
      "learning_rate": 9.902862526373452e-06,
      "loss": 0.7625,
      "step": 1914
    },
    {
      "epoch": 0.2731811697574893,
      "grad_norm": 0.6725876927375793,
      "learning_rate": 9.902711420302611e-06,
      "loss": 0.815,
      "step": 1915
    },
    {
      "epoch": 0.2733238231098431,
      "grad_norm": 0.9992535710334778,
      "learning_rate": 9.902560197948213e-06,
      "loss": 0.7819,
      "step": 1916
    },
    {
      "epoch": 0.27346647646219685,
      "grad_norm": 0.8791998624801636,
      "learning_rate": 9.90240885931384e-06,
      "loss": 0.786,
      "step": 1917
    },
    {
      "epoch": 0.27360912981455066,
      "grad_norm": 1.16233229637146,
      "learning_rate": 9.902257404403084e-06,
      "loss": 0.8775,
      "step": 1918
    },
    {
      "epoch": 0.2737517831669044,
      "grad_norm": 0.6956872940063477,
      "learning_rate": 9.902105833219536e-06,
      "loss": 0.8513,
      "step": 1919
    },
    {
      "epoch": 0.2738944365192582,
      "grad_norm": 1.01936936378479,
      "learning_rate": 9.901954145766791e-06,
      "loss": 0.7573,
      "step": 1920
    },
    {
      "epoch": 0.27403708987161196,
      "grad_norm": 0.8860127925872803,
      "learning_rate": 9.901802342048448e-06,
      "loss": 0.7638,
      "step": 1921
    },
    {
      "epoch": 0.27417974322396577,
      "grad_norm": 0.9512662887573242,
      "learning_rate": 9.901650422068105e-06,
      "loss": 0.6495,
      "step": 1922
    },
    {
      "epoch": 0.2743223965763195,
      "grad_norm": 0.6039310693740845,
      "learning_rate": 9.901498385829368e-06,
      "loss": 0.8469,
      "step": 1923
    },
    {
      "epoch": 0.2744650499286733,
      "grad_norm": 1.309650182723999,
      "learning_rate": 9.901346233335842e-06,
      "loss": 0.8151,
      "step": 1924
    },
    {
      "epoch": 0.2746077032810271,
      "grad_norm": 0.9727128744125366,
      "learning_rate": 9.901193964591136e-06,
      "loss": 0.7634,
      "step": 1925
    },
    {
      "epoch": 0.2747503566333809,
      "grad_norm": 0.9597587585449219,
      "learning_rate": 9.901041579598862e-06,
      "loss": 0.7804,
      "step": 1926
    },
    {
      "epoch": 0.2748930099857347,
      "grad_norm": 0.8195070028305054,
      "learning_rate": 9.900889078362632e-06,
      "loss": 0.856,
      "step": 1927
    },
    {
      "epoch": 0.27503566333808843,
      "grad_norm": 0.9469627141952515,
      "learning_rate": 9.900736460886065e-06,
      "loss": 0.8192,
      "step": 1928
    },
    {
      "epoch": 0.27517831669044224,
      "grad_norm": 0.9985857009887695,
      "learning_rate": 9.90058372717278e-06,
      "loss": 0.8751,
      "step": 1929
    },
    {
      "epoch": 0.275320970042796,
      "grad_norm": 0.9491239190101624,
      "learning_rate": 9.900430877226402e-06,
      "loss": 0.8496,
      "step": 1930
    },
    {
      "epoch": 0.2754636233951498,
      "grad_norm": 0.864153265953064,
      "learning_rate": 9.900277911050552e-06,
      "loss": 0.7982,
      "step": 1931
    },
    {
      "epoch": 0.27560627674750354,
      "grad_norm": 0.7806166410446167,
      "learning_rate": 9.90012482864886e-06,
      "loss": 0.731,
      "step": 1932
    },
    {
      "epoch": 0.27574893009985735,
      "grad_norm": 0.7389256954193115,
      "learning_rate": 9.899971630024959e-06,
      "loss": 0.8548,
      "step": 1933
    },
    {
      "epoch": 0.27589158345221115,
      "grad_norm": 0.8455584645271301,
      "learning_rate": 9.89981831518248e-06,
      "loss": 0.8258,
      "step": 1934
    },
    {
      "epoch": 0.2760342368045649,
      "grad_norm": 0.6914425492286682,
      "learning_rate": 9.89966488412506e-06,
      "loss": 0.7724,
      "step": 1935
    },
    {
      "epoch": 0.2761768901569187,
      "grad_norm": 0.7934995889663696,
      "learning_rate": 9.899511336856339e-06,
      "loss": 0.8076,
      "step": 1936
    },
    {
      "epoch": 0.27631954350927246,
      "grad_norm": 0.9315785765647888,
      "learning_rate": 9.899357673379958e-06,
      "loss": 0.8478,
      "step": 1937
    },
    {
      "epoch": 0.27646219686162626,
      "grad_norm": 1.0555291175842285,
      "learning_rate": 9.899203893699562e-06,
      "loss": 0.8919,
      "step": 1938
    },
    {
      "epoch": 0.27660485021398,
      "grad_norm": 0.7730499505996704,
      "learning_rate": 9.899049997818797e-06,
      "loss": 0.7527,
      "step": 1939
    },
    {
      "epoch": 0.2767475035663338,
      "grad_norm": 0.9190470576286316,
      "learning_rate": 9.898895985741315e-06,
      "loss": 0.9001,
      "step": 1940
    },
    {
      "epoch": 0.27689015691868757,
      "grad_norm": 0.8603836297988892,
      "learning_rate": 9.898741857470768e-06,
      "loss": 0.7746,
      "step": 1941
    },
    {
      "epoch": 0.2770328102710414,
      "grad_norm": 1.4602895975112915,
      "learning_rate": 9.898587613010813e-06,
      "loss": 0.7827,
      "step": 1942
    },
    {
      "epoch": 0.2771754636233951,
      "grad_norm": 0.5700140595436096,
      "learning_rate": 9.898433252365107e-06,
      "loss": 0.828,
      "step": 1943
    },
    {
      "epoch": 0.27731811697574893,
      "grad_norm": 0.8412007689476013,
      "learning_rate": 9.898278775537311e-06,
      "loss": 0.5979,
      "step": 1944
    },
    {
      "epoch": 0.27746077032810273,
      "grad_norm": 0.9965437054634094,
      "learning_rate": 9.898124182531089e-06,
      "loss": 0.8555,
      "step": 1945
    },
    {
      "epoch": 0.2776034236804565,
      "grad_norm": 0.7343485951423645,
      "learning_rate": 9.897969473350108e-06,
      "loss": 0.8199,
      "step": 1946
    },
    {
      "epoch": 0.2777460770328103,
      "grad_norm": 0.7718836069107056,
      "learning_rate": 9.897814647998037e-06,
      "loss": 0.8313,
      "step": 1947
    },
    {
      "epoch": 0.27788873038516404,
      "grad_norm": 0.9867026209831238,
      "learning_rate": 9.89765970647855e-06,
      "loss": 0.8149,
      "step": 1948
    },
    {
      "epoch": 0.27803138373751785,
      "grad_norm": 0.8841606974601746,
      "learning_rate": 9.89750464879532e-06,
      "loss": 0.8281,
      "step": 1949
    },
    {
      "epoch": 0.2781740370898716,
      "grad_norm": 1.020338535308838,
      "learning_rate": 9.897349474952024e-06,
      "loss": 0.7385,
      "step": 1950
    },
    {
      "epoch": 0.2783166904422254,
      "grad_norm": 1.148627758026123,
      "learning_rate": 9.897194184952345e-06,
      "loss": 0.7497,
      "step": 1951
    },
    {
      "epoch": 0.27845934379457915,
      "grad_norm": 0.8143110275268555,
      "learning_rate": 9.897038778799965e-06,
      "loss": 0.829,
      "step": 1952
    },
    {
      "epoch": 0.27860199714693296,
      "grad_norm": 0.8986217975616455,
      "learning_rate": 9.896883256498568e-06,
      "loss": 0.8975,
      "step": 1953
    },
    {
      "epoch": 0.2787446504992867,
      "grad_norm": 0.9541431069374084,
      "learning_rate": 9.896727618051847e-06,
      "loss": 0.9014,
      "step": 1954
    },
    {
      "epoch": 0.2788873038516405,
      "grad_norm": 1.1523237228393555,
      "learning_rate": 9.89657186346349e-06,
      "loss": 0.8349,
      "step": 1955
    },
    {
      "epoch": 0.2790299572039943,
      "grad_norm": 0.7698736190795898,
      "learning_rate": 9.896415992737192e-06,
      "loss": 0.8688,
      "step": 1956
    },
    {
      "epoch": 0.27917261055634807,
      "grad_norm": 0.8719567060470581,
      "learning_rate": 9.89626000587665e-06,
      "loss": 0.8121,
      "step": 1957
    },
    {
      "epoch": 0.27931526390870187,
      "grad_norm": 0.6933151483535767,
      "learning_rate": 9.896103902885566e-06,
      "loss": 0.7388,
      "step": 1958
    },
    {
      "epoch": 0.2794579172610556,
      "grad_norm": 1.1415010690689087,
      "learning_rate": 9.895947683767638e-06,
      "loss": 0.8419,
      "step": 1959
    },
    {
      "epoch": 0.2796005706134094,
      "grad_norm": 0.9674553871154785,
      "learning_rate": 9.895791348526573e-06,
      "loss": 0.7627,
      "step": 1960
    },
    {
      "epoch": 0.2797432239657632,
      "grad_norm": 1.1427204608917236,
      "learning_rate": 9.895634897166082e-06,
      "loss": 0.7922,
      "step": 1961
    },
    {
      "epoch": 0.279885877318117,
      "grad_norm": 0.7016680240631104,
      "learning_rate": 9.895478329689872e-06,
      "loss": 0.8672,
      "step": 1962
    },
    {
      "epoch": 0.28002853067047073,
      "grad_norm": 0.9272727966308594,
      "learning_rate": 9.895321646101659e-06,
      "loss": 0.8336,
      "step": 1963
    },
    {
      "epoch": 0.28017118402282454,
      "grad_norm": 0.9848592281341553,
      "learning_rate": 9.895164846405157e-06,
      "loss": 0.7641,
      "step": 1964
    },
    {
      "epoch": 0.28031383737517834,
      "grad_norm": 0.9903753399848938,
      "learning_rate": 9.895007930604088e-06,
      "loss": 0.824,
      "step": 1965
    },
    {
      "epoch": 0.2804564907275321,
      "grad_norm": 0.7130590081214905,
      "learning_rate": 9.89485089870217e-06,
      "loss": 0.7927,
      "step": 1966
    },
    {
      "epoch": 0.2805991440798859,
      "grad_norm": 0.8927415609359741,
      "learning_rate": 9.894693750703129e-06,
      "loss": 0.8597,
      "step": 1967
    },
    {
      "epoch": 0.28074179743223965,
      "grad_norm": 2.3921425342559814,
      "learning_rate": 9.894536486610694e-06,
      "loss": 0.8199,
      "step": 1968
    },
    {
      "epoch": 0.28088445078459345,
      "grad_norm": 0.7472104430198669,
      "learning_rate": 9.894379106428593e-06,
      "loss": 0.8225,
      "step": 1969
    },
    {
      "epoch": 0.2810271041369472,
      "grad_norm": 0.6429457664489746,
      "learning_rate": 9.89422161016056e-06,
      "loss": 0.8998,
      "step": 1970
    },
    {
      "epoch": 0.281169757489301,
      "grad_norm": 0.8767111897468567,
      "learning_rate": 9.89406399781033e-06,
      "loss": 0.8298,
      "step": 1971
    },
    {
      "epoch": 0.28131241084165476,
      "grad_norm": 0.7979419231414795,
      "learning_rate": 9.893906269381641e-06,
      "loss": 0.8413,
      "step": 1972
    },
    {
      "epoch": 0.28145506419400856,
      "grad_norm": 1.0332896709442139,
      "learning_rate": 9.893748424878236e-06,
      "loss": 0.7598,
      "step": 1973
    },
    {
      "epoch": 0.2815977175463623,
      "grad_norm": 0.7872117757797241,
      "learning_rate": 9.893590464303855e-06,
      "loss": 0.8619,
      "step": 1974
    },
    {
      "epoch": 0.2817403708987161,
      "grad_norm": 0.7206314206123352,
      "learning_rate": 9.893432387662247e-06,
      "loss": 0.7714,
      "step": 1975
    },
    {
      "epoch": 0.2818830242510699,
      "grad_norm": 0.8268548250198364,
      "learning_rate": 9.893274194957163e-06,
      "loss": 0.7982,
      "step": 1976
    },
    {
      "epoch": 0.2820256776034237,
      "grad_norm": 0.8085371851921082,
      "learning_rate": 9.89311588619235e-06,
      "loss": 0.7757,
      "step": 1977
    },
    {
      "epoch": 0.2821683309557775,
      "grad_norm": 0.7373887896537781,
      "learning_rate": 9.892957461371568e-06,
      "loss": 0.8767,
      "step": 1978
    },
    {
      "epoch": 0.28231098430813123,
      "grad_norm": 0.8956614136695862,
      "learning_rate": 9.892798920498572e-06,
      "loss": 0.8047,
      "step": 1979
    },
    {
      "epoch": 0.28245363766048504,
      "grad_norm": 1.2321044206619263,
      "learning_rate": 9.892640263577123e-06,
      "loss": 0.8781,
      "step": 1980
    },
    {
      "epoch": 0.2825962910128388,
      "grad_norm": 0.9143946766853333,
      "learning_rate": 9.892481490610984e-06,
      "loss": 0.886,
      "step": 1981
    },
    {
      "epoch": 0.2827389443651926,
      "grad_norm": 0.7610822916030884,
      "learning_rate": 9.892322601603918e-06,
      "loss": 0.8833,
      "step": 1982
    },
    {
      "epoch": 0.28288159771754634,
      "grad_norm": 1.0275628566741943,
      "learning_rate": 9.892163596559698e-06,
      "loss": 0.7262,
      "step": 1983
    },
    {
      "epoch": 0.28302425106990015,
      "grad_norm": 0.9696669578552246,
      "learning_rate": 9.892004475482094e-06,
      "loss": 0.8268,
      "step": 1984
    },
    {
      "epoch": 0.2831669044222539,
      "grad_norm": 0.9769879579544067,
      "learning_rate": 9.89184523837488e-06,
      "loss": 0.7619,
      "step": 1985
    },
    {
      "epoch": 0.2833095577746077,
      "grad_norm": 1.043142318725586,
      "learning_rate": 9.89168588524183e-06,
      "loss": 0.7586,
      "step": 1986
    },
    {
      "epoch": 0.2834522111269615,
      "grad_norm": 0.8820614218711853,
      "learning_rate": 9.891526416086727e-06,
      "loss": 0.8276,
      "step": 1987
    },
    {
      "epoch": 0.28359486447931526,
      "grad_norm": 0.8670064806938171,
      "learning_rate": 9.891366830913353e-06,
      "loss": 0.7495,
      "step": 1988
    },
    {
      "epoch": 0.28373751783166906,
      "grad_norm": 0.752273678779602,
      "learning_rate": 9.891207129725493e-06,
      "loss": 0.8117,
      "step": 1989
    },
    {
      "epoch": 0.2838801711840228,
      "grad_norm": 1.0996286869049072,
      "learning_rate": 9.891047312526932e-06,
      "loss": 0.7631,
      "step": 1990
    },
    {
      "epoch": 0.2840228245363766,
      "grad_norm": 0.7424052953720093,
      "learning_rate": 9.890887379321464e-06,
      "loss": 0.7772,
      "step": 1991
    },
    {
      "epoch": 0.28416547788873037,
      "grad_norm": 0.7052674293518066,
      "learning_rate": 9.89072733011288e-06,
      "loss": 0.8142,
      "step": 1992
    },
    {
      "epoch": 0.28430813124108417,
      "grad_norm": 1.076757550239563,
      "learning_rate": 9.890567164904978e-06,
      "loss": 0.8167,
      "step": 1993
    },
    {
      "epoch": 0.2844507845934379,
      "grad_norm": 0.8586769104003906,
      "learning_rate": 9.890406883701555e-06,
      "loss": 0.8052,
      "step": 1994
    },
    {
      "epoch": 0.2845934379457917,
      "grad_norm": 0.8514905571937561,
      "learning_rate": 9.890246486506416e-06,
      "loss": 0.7627,
      "step": 1995
    },
    {
      "epoch": 0.28473609129814553,
      "grad_norm": 0.7438399791717529,
      "learning_rate": 9.89008597332336e-06,
      "loss": 0.8691,
      "step": 1996
    },
    {
      "epoch": 0.2848787446504993,
      "grad_norm": 1.0306975841522217,
      "learning_rate": 9.889925344156198e-06,
      "loss": 0.8418,
      "step": 1997
    },
    {
      "epoch": 0.2850213980028531,
      "grad_norm": 0.9475598335266113,
      "learning_rate": 9.88976459900874e-06,
      "loss": 0.8198,
      "step": 1998
    },
    {
      "epoch": 0.28516405135520684,
      "grad_norm": 0.9006056785583496,
      "learning_rate": 9.889603737884796e-06,
      "loss": 0.5998,
      "step": 1999
    },
    {
      "epoch": 0.28530670470756064,
      "grad_norm": 1.1654584407806396,
      "learning_rate": 9.889442760788183e-06,
      "loss": 0.764,
      "step": 2000
    },
    {
      "epoch": 0.2854493580599144,
      "grad_norm": 0.7571951150894165,
      "learning_rate": 9.889281667722718e-06,
      "loss": 0.7987,
      "step": 2001
    },
    {
      "epoch": 0.2855920114122682,
      "grad_norm": 0.707734227180481,
      "learning_rate": 9.889120458692223e-06,
      "loss": 0.8486,
      "step": 2002
    },
    {
      "epoch": 0.28573466476462195,
      "grad_norm": 0.9231438040733337,
      "learning_rate": 9.888959133700523e-06,
      "loss": 0.8625,
      "step": 2003
    },
    {
      "epoch": 0.28587731811697575,
      "grad_norm": 0.84060138463974,
      "learning_rate": 9.888797692751442e-06,
      "loss": 0.7805,
      "step": 2004
    },
    {
      "epoch": 0.2860199714693295,
      "grad_norm": 0.8617420196533203,
      "learning_rate": 9.888636135848808e-06,
      "loss": 0.7294,
      "step": 2005
    },
    {
      "epoch": 0.2861626248216833,
      "grad_norm": 0.6491578221321106,
      "learning_rate": 9.888474462996457e-06,
      "loss": 0.7926,
      "step": 2006
    },
    {
      "epoch": 0.2863052781740371,
      "grad_norm": 0.7339638471603394,
      "learning_rate": 9.888312674198218e-06,
      "loss": 0.7536,
      "step": 2007
    },
    {
      "epoch": 0.28644793152639086,
      "grad_norm": 0.7412121295928955,
      "learning_rate": 9.888150769457934e-06,
      "loss": 0.8336,
      "step": 2008
    },
    {
      "epoch": 0.28659058487874467,
      "grad_norm": 0.9829639196395874,
      "learning_rate": 9.887988748779441e-06,
      "loss": 0.7488,
      "step": 2009
    },
    {
      "epoch": 0.2867332382310984,
      "grad_norm": 0.7689664363861084,
      "learning_rate": 9.887826612166584e-06,
      "loss": 0.7866,
      "step": 2010
    },
    {
      "epoch": 0.2868758915834522,
      "grad_norm": 0.8491288423538208,
      "learning_rate": 9.887664359623208e-06,
      "loss": 0.8923,
      "step": 2011
    },
    {
      "epoch": 0.287018544935806,
      "grad_norm": 0.9316936731338501,
      "learning_rate": 9.887501991153161e-06,
      "loss": 0.7523,
      "step": 2012
    },
    {
      "epoch": 0.2871611982881598,
      "grad_norm": 0.9495995044708252,
      "learning_rate": 9.887339506760294e-06,
      "loss": 0.821,
      "step": 2013
    },
    {
      "epoch": 0.28730385164051353,
      "grad_norm": 0.8646326065063477,
      "learning_rate": 9.887176906448463e-06,
      "loss": 0.9199,
      "step": 2014
    },
    {
      "epoch": 0.28744650499286734,
      "grad_norm": 1.0679668188095093,
      "learning_rate": 9.887014190221522e-06,
      "loss": 0.889,
      "step": 2015
    },
    {
      "epoch": 0.2875891583452211,
      "grad_norm": 1.015106439590454,
      "learning_rate": 9.886851358083332e-06,
      "loss": 0.8016,
      "step": 2016
    },
    {
      "epoch": 0.2877318116975749,
      "grad_norm": 0.628006637096405,
      "learning_rate": 9.886688410037752e-06,
      "loss": 0.8931,
      "step": 2017
    },
    {
      "epoch": 0.2878744650499287,
      "grad_norm": 0.9774754643440247,
      "learning_rate": 9.88652534608865e-06,
      "loss": 0.8701,
      "step": 2018
    },
    {
      "epoch": 0.28801711840228245,
      "grad_norm": 0.8394613862037659,
      "learning_rate": 9.886362166239894e-06,
      "loss": 0.8705,
      "step": 2019
    },
    {
      "epoch": 0.28815977175463625,
      "grad_norm": 0.8570374846458435,
      "learning_rate": 9.886198870495351e-06,
      "loss": 0.8012,
      "step": 2020
    },
    {
      "epoch": 0.28830242510699,
      "grad_norm": 0.9115515351295471,
      "learning_rate": 9.886035458858899e-06,
      "loss": 0.7363,
      "step": 2021
    },
    {
      "epoch": 0.2884450784593438,
      "grad_norm": 0.9202782511711121,
      "learning_rate": 9.885871931334408e-06,
      "loss": 0.8111,
      "step": 2022
    },
    {
      "epoch": 0.28858773181169756,
      "grad_norm": 1.2247858047485352,
      "learning_rate": 9.88570828792576e-06,
      "loss": 0.8151,
      "step": 2023
    },
    {
      "epoch": 0.28873038516405136,
      "grad_norm": 0.7569736242294312,
      "learning_rate": 9.885544528636836e-06,
      "loss": 0.932,
      "step": 2024
    },
    {
      "epoch": 0.2888730385164051,
      "grad_norm": 0.7694206237792969,
      "learning_rate": 9.885380653471521e-06,
      "loss": 0.6508,
      "step": 2025
    },
    {
      "epoch": 0.2890156918687589,
      "grad_norm": 0.9679473638534546,
      "learning_rate": 9.8852166624337e-06,
      "loss": 0.8412,
      "step": 2026
    },
    {
      "epoch": 0.2891583452211127,
      "grad_norm": 0.6804773211479187,
      "learning_rate": 9.885052555527263e-06,
      "loss": 0.7427,
      "step": 2027
    },
    {
      "epoch": 0.2893009985734665,
      "grad_norm": 0.8567250967025757,
      "learning_rate": 9.884888332756104e-06,
      "loss": 0.8438,
      "step": 2028
    },
    {
      "epoch": 0.2894436519258203,
      "grad_norm": 0.7586188912391663,
      "learning_rate": 9.884723994124117e-06,
      "loss": 0.8462,
      "step": 2029
    },
    {
      "epoch": 0.289586305278174,
      "grad_norm": 1.1931333541870117,
      "learning_rate": 9.884559539635197e-06,
      "loss": 0.8473,
      "step": 2030
    },
    {
      "epoch": 0.28972895863052783,
      "grad_norm": 0.8863710165023804,
      "learning_rate": 9.88439496929325e-06,
      "loss": 0.7864,
      "step": 2031
    },
    {
      "epoch": 0.2898716119828816,
      "grad_norm": 0.7390526533126831,
      "learning_rate": 9.884230283102173e-06,
      "loss": 0.8177,
      "step": 2032
    },
    {
      "epoch": 0.2900142653352354,
      "grad_norm": 0.8629305958747864,
      "learning_rate": 9.884065481065878e-06,
      "loss": 0.811,
      "step": 2033
    },
    {
      "epoch": 0.29015691868758914,
      "grad_norm": 0.7086455821990967,
      "learning_rate": 9.88390056318827e-06,
      "loss": 0.7686,
      "step": 2034
    },
    {
      "epoch": 0.29029957203994294,
      "grad_norm": 0.7655584216117859,
      "learning_rate": 9.883735529473262e-06,
      "loss": 0.8167,
      "step": 2035
    },
    {
      "epoch": 0.2904422253922967,
      "grad_norm": 0.7016885280609131,
      "learning_rate": 9.88357037992477e-06,
      "loss": 0.8214,
      "step": 2036
    },
    {
      "epoch": 0.2905848787446505,
      "grad_norm": 1.3929604291915894,
      "learning_rate": 9.883405114546706e-06,
      "loss": 0.7529,
      "step": 2037
    },
    {
      "epoch": 0.2907275320970043,
      "grad_norm": 0.8756687641143799,
      "learning_rate": 9.883239733342995e-06,
      "loss": 0.7332,
      "step": 2038
    },
    {
      "epoch": 0.29087018544935805,
      "grad_norm": 0.6817427277565002,
      "learning_rate": 9.883074236317558e-06,
      "loss": 0.6934,
      "step": 2039
    },
    {
      "epoch": 0.29101283880171186,
      "grad_norm": 0.7502633333206177,
      "learning_rate": 9.882908623474318e-06,
      "loss": 0.834,
      "step": 2040
    },
    {
      "epoch": 0.2911554921540656,
      "grad_norm": 0.8368335962295532,
      "learning_rate": 9.882742894817207e-06,
      "loss": 0.8301,
      "step": 2041
    },
    {
      "epoch": 0.2912981455064194,
      "grad_norm": 0.8467952609062195,
      "learning_rate": 9.882577050350152e-06,
      "loss": 0.8217,
      "step": 2042
    },
    {
      "epoch": 0.29144079885877316,
      "grad_norm": 0.9308727979660034,
      "learning_rate": 9.88241109007709e-06,
      "loss": 0.8206,
      "step": 2043
    },
    {
      "epoch": 0.29158345221112697,
      "grad_norm": 1.0542320013046265,
      "learning_rate": 9.882245014001954e-06,
      "loss": 0.8099,
      "step": 2044
    },
    {
      "epoch": 0.2917261055634807,
      "grad_norm": 0.8059763312339783,
      "learning_rate": 9.882078822128687e-06,
      "loss": 0.8226,
      "step": 2045
    },
    {
      "epoch": 0.2918687589158345,
      "grad_norm": 0.7922718524932861,
      "learning_rate": 9.881912514461224e-06,
      "loss": 0.8056,
      "step": 2046
    },
    {
      "epoch": 0.2920114122681883,
      "grad_norm": 0.7705415487289429,
      "learning_rate": 9.881746091003517e-06,
      "loss": 0.9385,
      "step": 2047
    },
    {
      "epoch": 0.2921540656205421,
      "grad_norm": 0.7232141494750977,
      "learning_rate": 9.88157955175951e-06,
      "loss": 0.7884,
      "step": 2048
    },
    {
      "epoch": 0.2922967189728959,
      "grad_norm": 0.9239014983177185,
      "learning_rate": 9.881412896733152e-06,
      "loss": 0.7982,
      "step": 2049
    },
    {
      "epoch": 0.29243937232524964,
      "grad_norm": 0.6806542277336121,
      "learning_rate": 9.881246125928397e-06,
      "loss": 0.7991,
      "step": 2050
    },
    {
      "epoch": 0.29258202567760344,
      "grad_norm": 0.918830156326294,
      "learning_rate": 9.8810792393492e-06,
      "loss": 0.7707,
      "step": 2051
    },
    {
      "epoch": 0.2927246790299572,
      "grad_norm": 0.7483209371566772,
      "learning_rate": 9.88091223699952e-06,
      "loss": 0.7601,
      "step": 2052
    },
    {
      "epoch": 0.292867332382311,
      "grad_norm": 1.057376742362976,
      "learning_rate": 9.880745118883319e-06,
      "loss": 0.8039,
      "step": 2053
    },
    {
      "epoch": 0.29300998573466475,
      "grad_norm": 0.7470461130142212,
      "learning_rate": 9.880577885004557e-06,
      "loss": 0.6418,
      "step": 2054
    },
    {
      "epoch": 0.29315263908701855,
      "grad_norm": 0.7077125310897827,
      "learning_rate": 9.880410535367205e-06,
      "loss": 0.7487,
      "step": 2055
    },
    {
      "epoch": 0.2932952924393723,
      "grad_norm": 0.7707588076591492,
      "learning_rate": 9.880243069975228e-06,
      "loss": 0.7815,
      "step": 2056
    },
    {
      "epoch": 0.2934379457917261,
      "grad_norm": 0.9563851356506348,
      "learning_rate": 9.880075488832601e-06,
      "loss": 0.8199,
      "step": 2057
    },
    {
      "epoch": 0.2935805991440799,
      "grad_norm": 0.7423125505447388,
      "learning_rate": 9.879907791943297e-06,
      "loss": 0.8316,
      "step": 2058
    },
    {
      "epoch": 0.29372325249643366,
      "grad_norm": 0.6795817613601685,
      "learning_rate": 9.879739979311296e-06,
      "loss": 0.8132,
      "step": 2059
    },
    {
      "epoch": 0.29386590584878747,
      "grad_norm": 0.6662577986717224,
      "learning_rate": 9.879572050940576e-06,
      "loss": 0.849,
      "step": 2060
    },
    {
      "epoch": 0.2940085592011412,
      "grad_norm": 0.789661169052124,
      "learning_rate": 9.879404006835119e-06,
      "loss": 0.8125,
      "step": 2061
    },
    {
      "epoch": 0.294151212553495,
      "grad_norm": 0.59214186668396,
      "learning_rate": 9.879235846998913e-06,
      "loss": 0.8108,
      "step": 2062
    },
    {
      "epoch": 0.2942938659058488,
      "grad_norm": 0.7700336575508118,
      "learning_rate": 9.879067571435946e-06,
      "loss": 0.7625,
      "step": 2063
    },
    {
      "epoch": 0.2944365192582026,
      "grad_norm": 0.5752390623092651,
      "learning_rate": 9.878899180150208e-06,
      "loss": 0.6208,
      "step": 2064
    },
    {
      "epoch": 0.29457917261055633,
      "grad_norm": 0.6733284592628479,
      "learning_rate": 9.878730673145695e-06,
      "loss": 0.7889,
      "step": 2065
    },
    {
      "epoch": 0.29472182596291013,
      "grad_norm": 0.8462895750999451,
      "learning_rate": 9.8785620504264e-06,
      "loss": 0.8705,
      "step": 2066
    },
    {
      "epoch": 0.2948644793152639,
      "grad_norm": 0.6242477297782898,
      "learning_rate": 9.878393311996329e-06,
      "loss": 0.7492,
      "step": 2067
    },
    {
      "epoch": 0.2950071326676177,
      "grad_norm": 0.8804550170898438,
      "learning_rate": 9.878224457859476e-06,
      "loss": 0.8494,
      "step": 2068
    },
    {
      "epoch": 0.2951497860199715,
      "grad_norm": 0.8771013617515564,
      "learning_rate": 9.878055488019852e-06,
      "loss": 0.8028,
      "step": 2069
    },
    {
      "epoch": 0.29529243937232524,
      "grad_norm": 0.6493806838989258,
      "learning_rate": 9.877886402481461e-06,
      "loss": 0.5711,
      "step": 2070
    },
    {
      "epoch": 0.29543509272467905,
      "grad_norm": 0.872032105922699,
      "learning_rate": 9.877717201248317e-06,
      "loss": 0.8921,
      "step": 2071
    },
    {
      "epoch": 0.2955777460770328,
      "grad_norm": 0.9797448515892029,
      "learning_rate": 9.87754788432443e-06,
      "loss": 0.8849,
      "step": 2072
    },
    {
      "epoch": 0.2957203994293866,
      "grad_norm": 0.676609992980957,
      "learning_rate": 9.877378451713817e-06,
      "loss": 0.7689,
      "step": 2073
    },
    {
      "epoch": 0.29586305278174035,
      "grad_norm": 0.8016397953033447,
      "learning_rate": 9.877208903420496e-06,
      "loss": 0.8762,
      "step": 2074
    },
    {
      "epoch": 0.29600570613409416,
      "grad_norm": 0.8251256346702576,
      "learning_rate": 9.87703923944849e-06,
      "loss": 0.7959,
      "step": 2075
    },
    {
      "epoch": 0.2961483594864479,
      "grad_norm": 0.9070316553115845,
      "learning_rate": 9.876869459801822e-06,
      "loss": 0.7998,
      "step": 2076
    },
    {
      "epoch": 0.2962910128388017,
      "grad_norm": 0.8603582382202148,
      "learning_rate": 9.87669956448452e-06,
      "loss": 0.792,
      "step": 2077
    },
    {
      "epoch": 0.29643366619115546,
      "grad_norm": 0.7488096356391907,
      "learning_rate": 9.876529553500612e-06,
      "loss": 0.8162,
      "step": 2078
    },
    {
      "epoch": 0.29657631954350927,
      "grad_norm": 0.9013305306434631,
      "learning_rate": 9.87635942685413e-06,
      "loss": 0.7686,
      "step": 2079
    },
    {
      "epoch": 0.2967189728958631,
      "grad_norm": 0.8539659976959229,
      "learning_rate": 9.876189184549111e-06,
      "loss": 0.9297,
      "step": 2080
    },
    {
      "epoch": 0.2968616262482168,
      "grad_norm": 0.807090699672699,
      "learning_rate": 9.876018826589593e-06,
      "loss": 0.8406,
      "step": 2081
    },
    {
      "epoch": 0.29700427960057063,
      "grad_norm": 0.898745059967041,
      "learning_rate": 9.875848352979613e-06,
      "loss": 0.8019,
      "step": 2082
    },
    {
      "epoch": 0.2971469329529244,
      "grad_norm": 1.1248681545257568,
      "learning_rate": 9.87567776372322e-06,
      "loss": 0.7014,
      "step": 2083
    },
    {
      "epoch": 0.2972895863052782,
      "grad_norm": 1.222901463508606,
      "learning_rate": 9.875507058824454e-06,
      "loss": 0.8078,
      "step": 2084
    },
    {
      "epoch": 0.29743223965763194,
      "grad_norm": 0.9312527775764465,
      "learning_rate": 9.875336238287368e-06,
      "loss": 0.7726,
      "step": 2085
    },
    {
      "epoch": 0.29757489300998574,
      "grad_norm": 0.9272825121879578,
      "learning_rate": 9.875165302116013e-06,
      "loss": 0.8807,
      "step": 2086
    },
    {
      "epoch": 0.2977175463623395,
      "grad_norm": 0.8349441885948181,
      "learning_rate": 9.874994250314443e-06,
      "loss": 0.7877,
      "step": 2087
    },
    {
      "epoch": 0.2978601997146933,
      "grad_norm": 1.006554126739502,
      "learning_rate": 9.874823082886713e-06,
      "loss": 0.8353,
      "step": 2088
    },
    {
      "epoch": 0.2980028530670471,
      "grad_norm": 0.79835444688797,
      "learning_rate": 9.874651799836884e-06,
      "loss": 0.726,
      "step": 2089
    },
    {
      "epoch": 0.29814550641940085,
      "grad_norm": 0.8016711473464966,
      "learning_rate": 9.874480401169022e-06,
      "loss": 0.7513,
      "step": 2090
    },
    {
      "epoch": 0.29828815977175466,
      "grad_norm": 0.8496073484420776,
      "learning_rate": 9.874308886887187e-06,
      "loss": 0.8491,
      "step": 2091
    },
    {
      "epoch": 0.2984308131241084,
      "grad_norm": 1.7348815202713013,
      "learning_rate": 9.874137256995451e-06,
      "loss": 0.8721,
      "step": 2092
    },
    {
      "epoch": 0.2985734664764622,
      "grad_norm": 0.922148585319519,
      "learning_rate": 9.87396551149788e-06,
      "loss": 0.8141,
      "step": 2093
    },
    {
      "epoch": 0.29871611982881596,
      "grad_norm": 0.9337180852890015,
      "learning_rate": 9.873793650398553e-06,
      "loss": 0.8335,
      "step": 2094
    },
    {
      "epoch": 0.29885877318116977,
      "grad_norm": 0.7504223585128784,
      "learning_rate": 9.873621673701543e-06,
      "loss": 0.8182,
      "step": 2095
    },
    {
      "epoch": 0.2990014265335235,
      "grad_norm": 0.8032134771347046,
      "learning_rate": 9.873449581410932e-06,
      "loss": 0.7608,
      "step": 2096
    },
    {
      "epoch": 0.2991440798858773,
      "grad_norm": 0.9668902158737183,
      "learning_rate": 9.873277373530796e-06,
      "loss": 0.8535,
      "step": 2097
    },
    {
      "epoch": 0.2992867332382311,
      "grad_norm": 0.9308344721794128,
      "learning_rate": 9.873105050065225e-06,
      "loss": 0.8363,
      "step": 2098
    },
    {
      "epoch": 0.2994293865905849,
      "grad_norm": 0.7573476433753967,
      "learning_rate": 9.872932611018303e-06,
      "loss": 0.9092,
      "step": 2099
    },
    {
      "epoch": 0.2995720399429387,
      "grad_norm": 0.9825876355171204,
      "learning_rate": 9.872760056394122e-06,
      "loss": 0.8367,
      "step": 2100
    },
    {
      "epoch": 0.29971469329529243,
      "grad_norm": 0.8236121535301208,
      "learning_rate": 9.872587386196773e-06,
      "loss": 0.7682,
      "step": 2101
    },
    {
      "epoch": 0.29985734664764624,
      "grad_norm": 0.9628751277923584,
      "learning_rate": 9.872414600430354e-06,
      "loss": 0.7783,
      "step": 2102
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1155561208724976,
      "learning_rate": 9.872241699098959e-06,
      "loss": 0.8202,
      "step": 2103
    },
    {
      "epoch": 0.3001426533523538,
      "grad_norm": 0.7287495136260986,
      "learning_rate": 9.872068682206693e-06,
      "loss": 0.7811,
      "step": 2104
    },
    {
      "epoch": 0.30028530670470754,
      "grad_norm": 0.8614897727966309,
      "learning_rate": 9.871895549757655e-06,
      "loss": 0.8026,
      "step": 2105
    },
    {
      "epoch": 0.30042796005706135,
      "grad_norm": 0.823110818862915,
      "learning_rate": 9.871722301755958e-06,
      "loss": 0.7913,
      "step": 2106
    },
    {
      "epoch": 0.3005706134094151,
      "grad_norm": 0.8629475235939026,
      "learning_rate": 9.871548938205707e-06,
      "loss": 0.8101,
      "step": 2107
    },
    {
      "epoch": 0.3007132667617689,
      "grad_norm": 1.211033582687378,
      "learning_rate": 9.871375459111012e-06,
      "loss": 0.818,
      "step": 2108
    },
    {
      "epoch": 0.30085592011412265,
      "grad_norm": 0.8554418087005615,
      "learning_rate": 9.87120186447599e-06,
      "loss": 0.7734,
      "step": 2109
    },
    {
      "epoch": 0.30099857346647646,
      "grad_norm": 0.8902543187141418,
      "learning_rate": 9.87102815430476e-06,
      "loss": 0.8581,
      "step": 2110
    },
    {
      "epoch": 0.30114122681883027,
      "grad_norm": 0.727468729019165,
      "learning_rate": 9.870854328601441e-06,
      "loss": 0.7895,
      "step": 2111
    },
    {
      "epoch": 0.301283880171184,
      "grad_norm": 0.7078068256378174,
      "learning_rate": 9.870680387370153e-06,
      "loss": 0.8017,
      "step": 2112
    },
    {
      "epoch": 0.3014265335235378,
      "grad_norm": 0.863304078578949,
      "learning_rate": 9.870506330615026e-06,
      "loss": 0.7368,
      "step": 2113
    },
    {
      "epoch": 0.30156918687589157,
      "grad_norm": 0.9752305746078491,
      "learning_rate": 9.870332158340184e-06,
      "loss": 0.8393,
      "step": 2114
    },
    {
      "epoch": 0.3017118402282454,
      "grad_norm": 0.7903264760971069,
      "learning_rate": 9.870157870549761e-06,
      "loss": 0.773,
      "step": 2115
    },
    {
      "epoch": 0.3018544935805991,
      "grad_norm": 0.6576414108276367,
      "learning_rate": 9.86998346724789e-06,
      "loss": 0.7978,
      "step": 2116
    },
    {
      "epoch": 0.30199714693295293,
      "grad_norm": 0.9541743397712708,
      "learning_rate": 9.869808948438709e-06,
      "loss": 0.8005,
      "step": 2117
    },
    {
      "epoch": 0.3021398002853067,
      "grad_norm": 0.753960371017456,
      "learning_rate": 9.869634314126355e-06,
      "loss": 0.8049,
      "step": 2118
    },
    {
      "epoch": 0.3022824536376605,
      "grad_norm": 1.614227533340454,
      "learning_rate": 9.86945956431497e-06,
      "loss": 0.5986,
      "step": 2119
    },
    {
      "epoch": 0.3024251069900143,
      "grad_norm": 0.9586261510848999,
      "learning_rate": 9.8692846990087e-06,
      "loss": 0.7981,
      "step": 2120
    },
    {
      "epoch": 0.30256776034236804,
      "grad_norm": 0.9887835383415222,
      "learning_rate": 9.86910971821169e-06,
      "loss": 0.8577,
      "step": 2121
    },
    {
      "epoch": 0.30271041369472185,
      "grad_norm": 0.793765127658844,
      "learning_rate": 9.868934621928096e-06,
      "loss": 0.8871,
      "step": 2122
    },
    {
      "epoch": 0.3028530670470756,
      "grad_norm": 0.9992631077766418,
      "learning_rate": 9.868759410162065e-06,
      "loss": 0.8226,
      "step": 2123
    },
    {
      "epoch": 0.3029957203994294,
      "grad_norm": 1.7083792686462402,
      "learning_rate": 9.868584082917756e-06,
      "loss": 0.623,
      "step": 2124
    },
    {
      "epoch": 0.30313837375178315,
      "grad_norm": 0.9900110960006714,
      "learning_rate": 9.868408640199325e-06,
      "loss": 0.8066,
      "step": 2125
    },
    {
      "epoch": 0.30328102710413696,
      "grad_norm": 0.8131747841835022,
      "learning_rate": 9.868233082010936e-06,
      "loss": 0.7903,
      "step": 2126
    },
    {
      "epoch": 0.3034236804564907,
      "grad_norm": 0.9614596366882324,
      "learning_rate": 9.868057408356751e-06,
      "loss": 0.7603,
      "step": 2127
    },
    {
      "epoch": 0.3035663338088445,
      "grad_norm": 0.8352845907211304,
      "learning_rate": 9.867881619240938e-06,
      "loss": 0.7485,
      "step": 2128
    },
    {
      "epoch": 0.30370898716119826,
      "grad_norm": 0.7611213326454163,
      "learning_rate": 9.867705714667663e-06,
      "loss": 0.7751,
      "step": 2129
    },
    {
      "epoch": 0.30385164051355207,
      "grad_norm": 1.0592976808547974,
      "learning_rate": 9.867529694641104e-06,
      "loss": 0.8413,
      "step": 2130
    },
    {
      "epoch": 0.3039942938659059,
      "grad_norm": 1.0875227451324463,
      "learning_rate": 9.867353559165432e-06,
      "loss": 0.8408,
      "step": 2131
    },
    {
      "epoch": 0.3041369472182596,
      "grad_norm": 1.003643274307251,
      "learning_rate": 9.867177308244823e-06,
      "loss": 0.7836,
      "step": 2132
    },
    {
      "epoch": 0.30427960057061343,
      "grad_norm": 0.9505198001861572,
      "learning_rate": 9.86700094188346e-06,
      "loss": 0.8237,
      "step": 2133
    },
    {
      "epoch": 0.3044222539229672,
      "grad_norm": 0.8064021468162537,
      "learning_rate": 9.866824460085526e-06,
      "loss": 0.9115,
      "step": 2134
    },
    {
      "epoch": 0.304564907275321,
      "grad_norm": 1.0296398401260376,
      "learning_rate": 9.86664786285521e-06,
      "loss": 0.8384,
      "step": 2135
    },
    {
      "epoch": 0.30470756062767473,
      "grad_norm": 0.6528288722038269,
      "learning_rate": 9.866471150196692e-06,
      "loss": 0.8159,
      "step": 2136
    },
    {
      "epoch": 0.30485021398002854,
      "grad_norm": 0.863062858581543,
      "learning_rate": 9.866294322114171e-06,
      "loss": 0.7495,
      "step": 2137
    },
    {
      "epoch": 0.3049928673323823,
      "grad_norm": 0.9573654532432556,
      "learning_rate": 9.866117378611839e-06,
      "loss": 0.7603,
      "step": 2138
    },
    {
      "epoch": 0.3051355206847361,
      "grad_norm": 1.066430687904358,
      "learning_rate": 9.865940319693891e-06,
      "loss": 0.7633,
      "step": 2139
    },
    {
      "epoch": 0.30527817403708984,
      "grad_norm": 1.0387983322143555,
      "learning_rate": 9.865763145364528e-06,
      "loss": 0.7622,
      "step": 2140
    },
    {
      "epoch": 0.30542082738944365,
      "grad_norm": 0.6708153486251831,
      "learning_rate": 9.865585855627951e-06,
      "loss": 0.8167,
      "step": 2141
    },
    {
      "epoch": 0.30556348074179746,
      "grad_norm": 0.8030664324760437,
      "learning_rate": 9.865408450488367e-06,
      "loss": 0.8367,
      "step": 2142
    },
    {
      "epoch": 0.3057061340941512,
      "grad_norm": 0.8071599006652832,
      "learning_rate": 9.865230929949984e-06,
      "loss": 0.7612,
      "step": 2143
    },
    {
      "epoch": 0.305848787446505,
      "grad_norm": 0.9778062701225281,
      "learning_rate": 9.86505329401701e-06,
      "loss": 0.7897,
      "step": 2144
    },
    {
      "epoch": 0.30599144079885876,
      "grad_norm": 0.8493444323539734,
      "learning_rate": 9.86487554269366e-06,
      "loss": 0.8974,
      "step": 2145
    },
    {
      "epoch": 0.30613409415121257,
      "grad_norm": 1.0544459819793701,
      "learning_rate": 9.864697675984148e-06,
      "loss": 0.7673,
      "step": 2146
    },
    {
      "epoch": 0.3062767475035663,
      "grad_norm": 1.2514063119888306,
      "learning_rate": 9.864519693892694e-06,
      "loss": 0.7424,
      "step": 2147
    },
    {
      "epoch": 0.3064194008559201,
      "grad_norm": 0.8082605004310608,
      "learning_rate": 9.864341596423522e-06,
      "loss": 0.8396,
      "step": 2148
    },
    {
      "epoch": 0.30656205420827387,
      "grad_norm": 1.1238821744918823,
      "learning_rate": 9.864163383580852e-06,
      "loss": 0.7987,
      "step": 2149
    },
    {
      "epoch": 0.3067047075606277,
      "grad_norm": 0.7748532295227051,
      "learning_rate": 9.86398505536891e-06,
      "loss": 0.786,
      "step": 2150
    },
    {
      "epoch": 0.3068473609129815,
      "grad_norm": 0.7004071474075317,
      "learning_rate": 9.863806611791932e-06,
      "loss": 0.738,
      "step": 2151
    },
    {
      "epoch": 0.30699001426533523,
      "grad_norm": 0.6278806328773499,
      "learning_rate": 9.863628052854144e-06,
      "loss": 0.8497,
      "step": 2152
    },
    {
      "epoch": 0.30713266761768904,
      "grad_norm": 0.8097479343414307,
      "learning_rate": 9.863449378559785e-06,
      "loss": 0.8619,
      "step": 2153
    },
    {
      "epoch": 0.3072753209700428,
      "grad_norm": 0.9160128831863403,
      "learning_rate": 9.863270588913093e-06,
      "loss": 0.8547,
      "step": 2154
    },
    {
      "epoch": 0.3074179743223966,
      "grad_norm": 0.6985733509063721,
      "learning_rate": 9.863091683918304e-06,
      "loss": 0.7888,
      "step": 2155
    },
    {
      "epoch": 0.30756062767475034,
      "grad_norm": 0.727873682975769,
      "learning_rate": 9.862912663579664e-06,
      "loss": 0.8047,
      "step": 2156
    },
    {
      "epoch": 0.30770328102710415,
      "grad_norm": 0.8142025470733643,
      "learning_rate": 9.86273352790142e-06,
      "loss": 0.8039,
      "step": 2157
    },
    {
      "epoch": 0.3078459343794579,
      "grad_norm": 1.0665960311889648,
      "learning_rate": 9.862554276887822e-06,
      "loss": 0.7953,
      "step": 2158
    },
    {
      "epoch": 0.3079885877318117,
      "grad_norm": 1.0293890237808228,
      "learning_rate": 9.862374910543118e-06,
      "loss": 0.8261,
      "step": 2159
    },
    {
      "epoch": 0.30813124108416545,
      "grad_norm": 0.6995319128036499,
      "learning_rate": 9.862195428871565e-06,
      "loss": 0.8047,
      "step": 2160
    },
    {
      "epoch": 0.30827389443651926,
      "grad_norm": 0.8701078295707703,
      "learning_rate": 9.862015831877418e-06,
      "loss": 0.8245,
      "step": 2161
    },
    {
      "epoch": 0.30841654778887306,
      "grad_norm": 1.0178847312927246,
      "learning_rate": 9.861836119564938e-06,
      "loss": 0.8391,
      "step": 2162
    },
    {
      "epoch": 0.3085592011412268,
      "grad_norm": 0.9585567116737366,
      "learning_rate": 9.861656291938386e-06,
      "loss": 0.801,
      "step": 2163
    },
    {
      "epoch": 0.3087018544935806,
      "grad_norm": 0.9947155714035034,
      "learning_rate": 9.861476349002031e-06,
      "loss": 0.7549,
      "step": 2164
    },
    {
      "epoch": 0.30884450784593437,
      "grad_norm": 0.8896161317825317,
      "learning_rate": 9.861296290760135e-06,
      "loss": 0.7609,
      "step": 2165
    },
    {
      "epoch": 0.3089871611982882,
      "grad_norm": 0.9120628237724304,
      "learning_rate": 9.861116117216973e-06,
      "loss": 0.7965,
      "step": 2166
    },
    {
      "epoch": 0.3091298145506419,
      "grad_norm": 0.8187360763549805,
      "learning_rate": 9.860935828376818e-06,
      "loss": 0.8245,
      "step": 2167
    },
    {
      "epoch": 0.30927246790299573,
      "grad_norm": 0.8231696486473083,
      "learning_rate": 9.860755424243944e-06,
      "loss": 0.8937,
      "step": 2168
    },
    {
      "epoch": 0.3094151212553495,
      "grad_norm": 0.7503544092178345,
      "learning_rate": 9.860574904822631e-06,
      "loss": 0.7331,
      "step": 2169
    },
    {
      "epoch": 0.3095577746077033,
      "grad_norm": 1.016889214515686,
      "learning_rate": 9.860394270117164e-06,
      "loss": 0.8046,
      "step": 2170
    },
    {
      "epoch": 0.30970042796005703,
      "grad_norm": 0.9495335817337036,
      "learning_rate": 9.860213520131821e-06,
      "loss": 0.82,
      "step": 2171
    },
    {
      "epoch": 0.30984308131241084,
      "grad_norm": 0.6853278875350952,
      "learning_rate": 9.860032654870892e-06,
      "loss": 0.7896,
      "step": 2172
    },
    {
      "epoch": 0.30998573466476464,
      "grad_norm": 0.9106058478355408,
      "learning_rate": 9.859851674338668e-06,
      "loss": 0.7421,
      "step": 2173
    },
    {
      "epoch": 0.3101283880171184,
      "grad_norm": 0.7071598172187805,
      "learning_rate": 9.859670578539442e-06,
      "loss": 0.7729,
      "step": 2174
    },
    {
      "epoch": 0.3102710413694722,
      "grad_norm": 0.7470977902412415,
      "learning_rate": 9.859489367477507e-06,
      "loss": 0.8296,
      "step": 2175
    },
    {
      "epoch": 0.31041369472182595,
      "grad_norm": 0.8395113945007324,
      "learning_rate": 9.859308041157159e-06,
      "loss": 0.6945,
      "step": 2176
    },
    {
      "epoch": 0.31055634807417976,
      "grad_norm": 0.840278685092926,
      "learning_rate": 9.859126599582705e-06,
      "loss": 0.7228,
      "step": 2177
    },
    {
      "epoch": 0.3106990014265335,
      "grad_norm": 0.9928528070449829,
      "learning_rate": 9.858945042758443e-06,
      "loss": 0.7163,
      "step": 2178
    },
    {
      "epoch": 0.3108416547788873,
      "grad_norm": 0.8427954316139221,
      "learning_rate": 9.858763370688682e-06,
      "loss": 0.7255,
      "step": 2179
    },
    {
      "epoch": 0.31098430813124106,
      "grad_norm": 0.932619035243988,
      "learning_rate": 9.85858158337773e-06,
      "loss": 0.7165,
      "step": 2180
    },
    {
      "epoch": 0.31112696148359487,
      "grad_norm": 1.118070125579834,
      "learning_rate": 9.8583996808299e-06,
      "loss": 0.7315,
      "step": 2181
    },
    {
      "epoch": 0.31126961483594867,
      "grad_norm": 0.8051701784133911,
      "learning_rate": 9.858217663049503e-06,
      "loss": 0.789,
      "step": 2182
    },
    {
      "epoch": 0.3114122681883024,
      "grad_norm": 0.7789623141288757,
      "learning_rate": 9.85803553004086e-06,
      "loss": 0.8121,
      "step": 2183
    },
    {
      "epoch": 0.3115549215406562,
      "grad_norm": 0.7830784916877747,
      "learning_rate": 9.85785328180829e-06,
      "loss": 0.6264,
      "step": 2184
    },
    {
      "epoch": 0.31169757489301,
      "grad_norm": 1.0730844736099243,
      "learning_rate": 9.857670918356113e-06,
      "loss": 0.7213,
      "step": 2185
    },
    {
      "epoch": 0.3118402282453638,
      "grad_norm": 0.7656449675559998,
      "learning_rate": 9.857488439688657e-06,
      "loss": 0.7442,
      "step": 2186
    },
    {
      "epoch": 0.31198288159771753,
      "grad_norm": 0.8415981531143188,
      "learning_rate": 9.85730584581025e-06,
      "loss": 0.7284,
      "step": 2187
    },
    {
      "epoch": 0.31212553495007134,
      "grad_norm": 1.2432276010513306,
      "learning_rate": 9.85712313672522e-06,
      "loss": 0.8322,
      "step": 2188
    },
    {
      "epoch": 0.3122681883024251,
      "grad_norm": 0.7802530527114868,
      "learning_rate": 9.856940312437904e-06,
      "loss": 0.8186,
      "step": 2189
    },
    {
      "epoch": 0.3124108416547789,
      "grad_norm": 0.894239604473114,
      "learning_rate": 9.856757372952636e-06,
      "loss": 0.8504,
      "step": 2190
    },
    {
      "epoch": 0.31255349500713264,
      "grad_norm": 1.0458093881607056,
      "learning_rate": 9.856574318273757e-06,
      "loss": 0.7908,
      "step": 2191
    },
    {
      "epoch": 0.31269614835948645,
      "grad_norm": 1.034009575843811,
      "learning_rate": 9.856391148405607e-06,
      "loss": 0.8138,
      "step": 2192
    },
    {
      "epoch": 0.31283880171184025,
      "grad_norm": 1.2163127660751343,
      "learning_rate": 9.856207863352531e-06,
      "loss": 0.7842,
      "step": 2193
    },
    {
      "epoch": 0.312981455064194,
      "grad_norm": 1.059219479560852,
      "learning_rate": 9.856024463118876e-06,
      "loss": 0.8056,
      "step": 2194
    },
    {
      "epoch": 0.3131241084165478,
      "grad_norm": 0.8998891115188599,
      "learning_rate": 9.855840947708994e-06,
      "loss": 0.7759,
      "step": 2195
    },
    {
      "epoch": 0.31326676176890156,
      "grad_norm": 0.8348072171211243,
      "learning_rate": 9.855657317127233e-06,
      "loss": 0.7015,
      "step": 2196
    },
    {
      "epoch": 0.31340941512125536,
      "grad_norm": 0.8206185102462769,
      "learning_rate": 9.855473571377954e-06,
      "loss": 0.8026,
      "step": 2197
    },
    {
      "epoch": 0.3135520684736091,
      "grad_norm": 0.7508227825164795,
      "learning_rate": 9.85528971046551e-06,
      "loss": 0.8614,
      "step": 2198
    },
    {
      "epoch": 0.3136947218259629,
      "grad_norm": 0.7386098504066467,
      "learning_rate": 9.855105734394266e-06,
      "loss": 0.8698,
      "step": 2199
    },
    {
      "epoch": 0.31383737517831667,
      "grad_norm": 1.059203028678894,
      "learning_rate": 9.854921643168582e-06,
      "loss": 0.7743,
      "step": 2200
    },
    {
      "epoch": 0.3139800285306705,
      "grad_norm": 0.9634552001953125,
      "learning_rate": 9.854737436792828e-06,
      "loss": 0.8486,
      "step": 2201
    },
    {
      "epoch": 0.3141226818830242,
      "grad_norm": 1.0923370122909546,
      "learning_rate": 9.854553115271369e-06,
      "loss": 0.8711,
      "step": 2202
    },
    {
      "epoch": 0.31426533523537803,
      "grad_norm": 0.7884952425956726,
      "learning_rate": 9.85436867860858e-06,
      "loss": 0.8044,
      "step": 2203
    },
    {
      "epoch": 0.31440798858773183,
      "grad_norm": 0.7849683165550232,
      "learning_rate": 9.854184126808834e-06,
      "loss": 0.8388,
      "step": 2204
    },
    {
      "epoch": 0.3145506419400856,
      "grad_norm": 0.9549512267112732,
      "learning_rate": 9.85399945987651e-06,
      "loss": 0.7388,
      "step": 2205
    },
    {
      "epoch": 0.3146932952924394,
      "grad_norm": 0.8654511570930481,
      "learning_rate": 9.853814677815985e-06,
      "loss": 0.7575,
      "step": 2206
    },
    {
      "epoch": 0.31483594864479314,
      "grad_norm": 0.8052588105201721,
      "learning_rate": 9.853629780631643e-06,
      "loss": 0.5873,
      "step": 2207
    },
    {
      "epoch": 0.31497860199714695,
      "grad_norm": 0.7848721146583557,
      "learning_rate": 9.85344476832787e-06,
      "loss": 0.8867,
      "step": 2208
    },
    {
      "epoch": 0.3151212553495007,
      "grad_norm": 1.3158105611801147,
      "learning_rate": 9.853259640909057e-06,
      "loss": 0.9041,
      "step": 2209
    },
    {
      "epoch": 0.3152639087018545,
      "grad_norm": 0.9283496737480164,
      "learning_rate": 9.85307439837959e-06,
      "loss": 0.749,
      "step": 2210
    },
    {
      "epoch": 0.31540656205420825,
      "grad_norm": 0.6851430535316467,
      "learning_rate": 9.852889040743862e-06,
      "loss": 0.838,
      "step": 2211
    },
    {
      "epoch": 0.31554921540656206,
      "grad_norm": 0.7198887467384338,
      "learning_rate": 9.852703568006274e-06,
      "loss": 0.7585,
      "step": 2212
    },
    {
      "epoch": 0.31569186875891586,
      "grad_norm": 0.9755715131759644,
      "learning_rate": 9.852517980171223e-06,
      "loss": 0.8576,
      "step": 2213
    },
    {
      "epoch": 0.3158345221112696,
      "grad_norm": 0.8580317497253418,
      "learning_rate": 9.852332277243111e-06,
      "loss": 0.7675,
      "step": 2214
    },
    {
      "epoch": 0.3159771754636234,
      "grad_norm": 1.0044978857040405,
      "learning_rate": 9.85214645922634e-06,
      "loss": 0.7174,
      "step": 2215
    },
    {
      "epoch": 0.31611982881597717,
      "grad_norm": 0.9367561936378479,
      "learning_rate": 9.851960526125322e-06,
      "loss": 0.8766,
      "step": 2216
    },
    {
      "epoch": 0.31626248216833097,
      "grad_norm": 1.1179993152618408,
      "learning_rate": 9.851774477944463e-06,
      "loss": 0.7772,
      "step": 2217
    },
    {
      "epoch": 0.3164051355206847,
      "grad_norm": 0.7806481719017029,
      "learning_rate": 9.851588314688177e-06,
      "loss": 0.7549,
      "step": 2218
    },
    {
      "epoch": 0.3165477888730385,
      "grad_norm": 0.8973761796951294,
      "learning_rate": 9.85140203636088e-06,
      "loss": 0.8134,
      "step": 2219
    },
    {
      "epoch": 0.3166904422253923,
      "grad_norm": 1.1041462421417236,
      "learning_rate": 9.851215642966992e-06,
      "loss": 0.7921,
      "step": 2220
    },
    {
      "epoch": 0.3168330955777461,
      "grad_norm": 0.960660457611084,
      "learning_rate": 9.851029134510928e-06,
      "loss": 0.7575,
      "step": 2221
    },
    {
      "epoch": 0.31697574893009983,
      "grad_norm": 0.6323800086975098,
      "learning_rate": 9.850842510997119e-06,
      "loss": 0.8378,
      "step": 2222
    },
    {
      "epoch": 0.31711840228245364,
      "grad_norm": 0.9967565536499023,
      "learning_rate": 9.850655772429984e-06,
      "loss": 0.7803,
      "step": 2223
    },
    {
      "epoch": 0.31726105563480744,
      "grad_norm": 0.8638782501220703,
      "learning_rate": 9.85046891881396e-06,
      "loss": 0.7756,
      "step": 2224
    },
    {
      "epoch": 0.3174037089871612,
      "grad_norm": 0.8508341312408447,
      "learning_rate": 9.850281950153473e-06,
      "loss": 0.785,
      "step": 2225
    },
    {
      "epoch": 0.317546362339515,
      "grad_norm": 0.9450670480728149,
      "learning_rate": 9.850094866452959e-06,
      "loss": 0.7804,
      "step": 2226
    },
    {
      "epoch": 0.31768901569186875,
      "grad_norm": 0.8388874530792236,
      "learning_rate": 9.849907667716856e-06,
      "loss": 0.744,
      "step": 2227
    },
    {
      "epoch": 0.31783166904422255,
      "grad_norm": 0.9147560000419617,
      "learning_rate": 9.849720353949603e-06,
      "loss": 0.8408,
      "step": 2228
    },
    {
      "epoch": 0.3179743223965763,
      "grad_norm": 0.7567501068115234,
      "learning_rate": 9.849532925155643e-06,
      "loss": 0.8426,
      "step": 2229
    },
    {
      "epoch": 0.3181169757489301,
      "grad_norm": 0.7739532589912415,
      "learning_rate": 9.849345381339424e-06,
      "loss": 0.8377,
      "step": 2230
    },
    {
      "epoch": 0.31825962910128386,
      "grad_norm": 0.761940062046051,
      "learning_rate": 9.849157722505388e-06,
      "loss": 0.8111,
      "step": 2231
    },
    {
      "epoch": 0.31840228245363766,
      "grad_norm": 0.780726969242096,
      "learning_rate": 9.848969948657994e-06,
      "loss": 0.7938,
      "step": 2232
    },
    {
      "epoch": 0.3185449358059914,
      "grad_norm": 0.7162595987319946,
      "learning_rate": 9.848782059801692e-06,
      "loss": 0.7784,
      "step": 2233
    },
    {
      "epoch": 0.3186875891583452,
      "grad_norm": 0.9884735345840454,
      "learning_rate": 9.848594055940935e-06,
      "loss": 0.8687,
      "step": 2234
    },
    {
      "epoch": 0.318830242510699,
      "grad_norm": 1.2923800945281982,
      "learning_rate": 9.848405937080187e-06,
      "loss": 0.7915,
      "step": 2235
    },
    {
      "epoch": 0.3189728958630528,
      "grad_norm": 0.8164573311805725,
      "learning_rate": 9.848217703223908e-06,
      "loss": 0.7613,
      "step": 2236
    },
    {
      "epoch": 0.3191155492154066,
      "grad_norm": 0.8309138417243958,
      "learning_rate": 9.848029354376561e-06,
      "loss": 0.7831,
      "step": 2237
    },
    {
      "epoch": 0.31925820256776033,
      "grad_norm": 0.8395116925239563,
      "learning_rate": 9.847840890542618e-06,
      "loss": 0.8752,
      "step": 2238
    },
    {
      "epoch": 0.31940085592011414,
      "grad_norm": 0.9086002707481384,
      "learning_rate": 9.847652311726542e-06,
      "loss": 0.7986,
      "step": 2239
    },
    {
      "epoch": 0.3195435092724679,
      "grad_norm": 0.9900307059288025,
      "learning_rate": 9.847463617932811e-06,
      "loss": 0.7143,
      "step": 2240
    },
    {
      "epoch": 0.3196861626248217,
      "grad_norm": 0.9896170496940613,
      "learning_rate": 9.847274809165901e-06,
      "loss": 0.7538,
      "step": 2241
    },
    {
      "epoch": 0.31982881597717544,
      "grad_norm": 0.8203100562095642,
      "learning_rate": 9.847085885430286e-06,
      "loss": 0.7835,
      "step": 2242
    },
    {
      "epoch": 0.31997146932952925,
      "grad_norm": 1.2037122249603271,
      "learning_rate": 9.846896846730451e-06,
      "loss": 0.7972,
      "step": 2243
    },
    {
      "epoch": 0.32011412268188305,
      "grad_norm": 0.7327616810798645,
      "learning_rate": 9.846707693070876e-06,
      "loss": 0.837,
      "step": 2244
    },
    {
      "epoch": 0.3202567760342368,
      "grad_norm": 0.7550442814826965,
      "learning_rate": 9.84651842445605e-06,
      "loss": 0.8301,
      "step": 2245
    },
    {
      "epoch": 0.3203994293865906,
      "grad_norm": 0.841128408908844,
      "learning_rate": 9.846329040890463e-06,
      "loss": 0.8046,
      "step": 2246
    },
    {
      "epoch": 0.32054208273894436,
      "grad_norm": 0.874862790107727,
      "learning_rate": 9.846139542378604e-06,
      "loss": 0.7318,
      "step": 2247
    },
    {
      "epoch": 0.32068473609129816,
      "grad_norm": 0.6857011914253235,
      "learning_rate": 9.84594992892497e-06,
      "loss": 0.8781,
      "step": 2248
    },
    {
      "epoch": 0.3208273894436519,
      "grad_norm": 0.7826842069625854,
      "learning_rate": 9.845760200534055e-06,
      "loss": 0.8683,
      "step": 2249
    },
    {
      "epoch": 0.3209700427960057,
      "grad_norm": 1.3429893255233765,
      "learning_rate": 9.845570357210362e-06,
      "loss": 0.6449,
      "step": 2250
    },
    {
      "epoch": 0.32111269614835947,
      "grad_norm": 0.9500937461853027,
      "learning_rate": 9.845380398958392e-06,
      "loss": 0.7185,
      "step": 2251
    },
    {
      "epoch": 0.32125534950071327,
      "grad_norm": 0.804557740688324,
      "learning_rate": 9.845190325782654e-06,
      "loss": 0.7884,
      "step": 2252
    },
    {
      "epoch": 0.321398002853067,
      "grad_norm": 0.7487125992774963,
      "learning_rate": 9.845000137687652e-06,
      "loss": 0.8521,
      "step": 2253
    },
    {
      "epoch": 0.3215406562054208,
      "grad_norm": 0.8841039538383484,
      "learning_rate": 9.844809834677898e-06,
      "loss": 0.7946,
      "step": 2254
    },
    {
      "epoch": 0.32168330955777463,
      "grad_norm": 0.6417714357376099,
      "learning_rate": 9.844619416757907e-06,
      "loss": 0.797,
      "step": 2255
    },
    {
      "epoch": 0.3218259629101284,
      "grad_norm": 1.0088717937469482,
      "learning_rate": 9.844428883932193e-06,
      "loss": 0.8008,
      "step": 2256
    },
    {
      "epoch": 0.3219686162624822,
      "grad_norm": 1.0375267267227173,
      "learning_rate": 9.844238236205278e-06,
      "loss": 0.8408,
      "step": 2257
    },
    {
      "epoch": 0.32211126961483594,
      "grad_norm": 0.770523726940155,
      "learning_rate": 9.844047473581682e-06,
      "loss": 0.7569,
      "step": 2258
    },
    {
      "epoch": 0.32225392296718974,
      "grad_norm": 1.4878278970718384,
      "learning_rate": 9.84385659606593e-06,
      "loss": 0.8058,
      "step": 2259
    },
    {
      "epoch": 0.3223965763195435,
      "grad_norm": 1.0476981401443481,
      "learning_rate": 9.84366560366255e-06,
      "loss": 0.9196,
      "step": 2260
    },
    {
      "epoch": 0.3225392296718973,
      "grad_norm": 1.2552111148834229,
      "learning_rate": 9.843474496376071e-06,
      "loss": 0.7388,
      "step": 2261
    },
    {
      "epoch": 0.32268188302425105,
      "grad_norm": 0.8368445038795471,
      "learning_rate": 9.843283274211026e-06,
      "loss": 0.7715,
      "step": 2262
    },
    {
      "epoch": 0.32282453637660485,
      "grad_norm": 1.0327579975128174,
      "learning_rate": 9.84309193717195e-06,
      "loss": 0.7277,
      "step": 2263
    },
    {
      "epoch": 0.3229671897289586,
      "grad_norm": 0.6204162836074829,
      "learning_rate": 9.84290048526338e-06,
      "loss": 0.8427,
      "step": 2264
    },
    {
      "epoch": 0.3231098430813124,
      "grad_norm": 1.0847949981689453,
      "learning_rate": 9.842708918489861e-06,
      "loss": 0.7236,
      "step": 2265
    },
    {
      "epoch": 0.3232524964336662,
      "grad_norm": 1.2121959924697876,
      "learning_rate": 9.842517236855932e-06,
      "loss": 0.7765,
      "step": 2266
    },
    {
      "epoch": 0.32339514978601996,
      "grad_norm": 0.678694486618042,
      "learning_rate": 9.842325440366143e-06,
      "loss": 0.7895,
      "step": 2267
    },
    {
      "epoch": 0.32353780313837377,
      "grad_norm": 1.0767388343811035,
      "learning_rate": 9.842133529025043e-06,
      "loss": 0.7751,
      "step": 2268
    },
    {
      "epoch": 0.3236804564907275,
      "grad_norm": 0.8977599143981934,
      "learning_rate": 9.841941502837177e-06,
      "loss": 0.7958,
      "step": 2269
    },
    {
      "epoch": 0.3238231098430813,
      "grad_norm": 1.0656965970993042,
      "learning_rate": 9.84174936180711e-06,
      "loss": 0.8319,
      "step": 2270
    },
    {
      "epoch": 0.3239657631954351,
      "grad_norm": 1.094441294670105,
      "learning_rate": 9.841557105939391e-06,
      "loss": 0.775,
      "step": 2271
    },
    {
      "epoch": 0.3241084165477889,
      "grad_norm": 0.715997576713562,
      "learning_rate": 9.841364735238586e-06,
      "loss": 0.7562,
      "step": 2272
    },
    {
      "epoch": 0.32425106990014263,
      "grad_norm": 0.9680311679840088,
      "learning_rate": 9.841172249709252e-06,
      "loss": 0.7118,
      "step": 2273
    },
    {
      "epoch": 0.32439372325249644,
      "grad_norm": 0.9165420532226562,
      "learning_rate": 9.84097964935596e-06,
      "loss": 0.7522,
      "step": 2274
    },
    {
      "epoch": 0.32453637660485024,
      "grad_norm": 0.9857007265090942,
      "learning_rate": 9.840786934183273e-06,
      "loss": 0.6922,
      "step": 2275
    },
    {
      "epoch": 0.324679029957204,
      "grad_norm": 0.7574713826179504,
      "learning_rate": 9.840594104195765e-06,
      "loss": 0.8032,
      "step": 2276
    },
    {
      "epoch": 0.3248216833095578,
      "grad_norm": 0.7682188153266907,
      "learning_rate": 9.840401159398007e-06,
      "loss": 0.8562,
      "step": 2277
    },
    {
      "epoch": 0.32496433666191155,
      "grad_norm": 0.9269386529922485,
      "learning_rate": 9.840208099794579e-06,
      "loss": 0.7957,
      "step": 2278
    },
    {
      "epoch": 0.32510699001426535,
      "grad_norm": 0.9559633135795593,
      "learning_rate": 9.840014925390057e-06,
      "loss": 0.7634,
      "step": 2279
    },
    {
      "epoch": 0.3252496433666191,
      "grad_norm": 1.1983888149261475,
      "learning_rate": 9.839821636189024e-06,
      "loss": 0.7992,
      "step": 2280
    },
    {
      "epoch": 0.3253922967189729,
      "grad_norm": 0.7660537362098694,
      "learning_rate": 9.839628232196064e-06,
      "loss": 0.8107,
      "step": 2281
    },
    {
      "epoch": 0.32553495007132666,
      "grad_norm": 0.7760874032974243,
      "learning_rate": 9.839434713415765e-06,
      "loss": 0.7389,
      "step": 2282
    },
    {
      "epoch": 0.32567760342368046,
      "grad_norm": 1.4310803413391113,
      "learning_rate": 9.839241079852718e-06,
      "loss": 0.7812,
      "step": 2283
    },
    {
      "epoch": 0.3258202567760342,
      "grad_norm": 1.1486469507217407,
      "learning_rate": 9.839047331511511e-06,
      "loss": 0.7388,
      "step": 2284
    },
    {
      "epoch": 0.325962910128388,
      "grad_norm": 0.944904625415802,
      "learning_rate": 9.838853468396744e-06,
      "loss": 0.823,
      "step": 2285
    },
    {
      "epoch": 0.3261055634807418,
      "grad_norm": 0.762138843536377,
      "learning_rate": 9.838659490513015e-06,
      "loss": 0.8101,
      "step": 2286
    },
    {
      "epoch": 0.3262482168330956,
      "grad_norm": 0.6126900315284729,
      "learning_rate": 9.838465397864921e-06,
      "loss": 0.851,
      "step": 2287
    },
    {
      "epoch": 0.3263908701854494,
      "grad_norm": 0.7258740067481995,
      "learning_rate": 9.838271190457067e-06,
      "loss": 0.8152,
      "step": 2288
    },
    {
      "epoch": 0.3265335235378031,
      "grad_norm": 0.9702327847480774,
      "learning_rate": 9.838076868294061e-06,
      "loss": 0.816,
      "step": 2289
    },
    {
      "epoch": 0.32667617689015693,
      "grad_norm": 0.762395441532135,
      "learning_rate": 9.837882431380514e-06,
      "loss": 0.8324,
      "step": 2290
    },
    {
      "epoch": 0.3268188302425107,
      "grad_norm": 0.8828544020652771,
      "learning_rate": 9.837687879721031e-06,
      "loss": 0.7704,
      "step": 2291
    },
    {
      "epoch": 0.3269614835948645,
      "grad_norm": 0.9014937281608582,
      "learning_rate": 9.837493213320233e-06,
      "loss": 0.6042,
      "step": 2292
    },
    {
      "epoch": 0.32710413694721824,
      "grad_norm": 0.8077760934829712,
      "learning_rate": 9.837298432182731e-06,
      "loss": 0.8175,
      "step": 2293
    },
    {
      "epoch": 0.32724679029957204,
      "grad_norm": 1.1762315034866333,
      "learning_rate": 9.83710353631315e-06,
      "loss": 0.9323,
      "step": 2294
    },
    {
      "epoch": 0.3273894436519258,
      "grad_norm": 1.0175790786743164,
      "learning_rate": 9.83690852571611e-06,
      "loss": 0.8173,
      "step": 2295
    },
    {
      "epoch": 0.3275320970042796,
      "grad_norm": 0.9739888906478882,
      "learning_rate": 9.836713400396239e-06,
      "loss": 0.8183,
      "step": 2296
    },
    {
      "epoch": 0.3276747503566334,
      "grad_norm": 0.8701342940330505,
      "learning_rate": 9.836518160358162e-06,
      "loss": 0.8104,
      "step": 2297
    },
    {
      "epoch": 0.32781740370898715,
      "grad_norm": 0.6412081718444824,
      "learning_rate": 9.83632280560651e-06,
      "loss": 0.7953,
      "step": 2298
    },
    {
      "epoch": 0.32796005706134096,
      "grad_norm": 0.854129433631897,
      "learning_rate": 9.836127336145919e-06,
      "loss": 0.7707,
      "step": 2299
    },
    {
      "epoch": 0.3281027104136947,
      "grad_norm": 0.8388704657554626,
      "learning_rate": 9.835931751981022e-06,
      "loss": 0.7845,
      "step": 2300
    },
    {
      "epoch": 0.3282453637660485,
      "grad_norm": 0.7747339010238647,
      "learning_rate": 9.83573605311646e-06,
      "loss": 0.8175,
      "step": 2301
    },
    {
      "epoch": 0.32838801711840226,
      "grad_norm": 0.6816387176513672,
      "learning_rate": 9.835540239556876e-06,
      "loss": 0.7926,
      "step": 2302
    },
    {
      "epoch": 0.32853067047075607,
      "grad_norm": 0.9259821176528931,
      "learning_rate": 9.83534431130691e-06,
      "loss": 0.7238,
      "step": 2303
    },
    {
      "epoch": 0.3286733238231098,
      "grad_norm": 0.8266289830207825,
      "learning_rate": 9.83514826837121e-06,
      "loss": 0.842,
      "step": 2304
    },
    {
      "epoch": 0.3288159771754636,
      "grad_norm": 0.8679077625274658,
      "learning_rate": 9.83495211075443e-06,
      "loss": 0.7786,
      "step": 2305
    },
    {
      "epoch": 0.32895863052781743,
      "grad_norm": 1.1242318153381348,
      "learning_rate": 9.834755838461219e-06,
      "loss": 0.7827,
      "step": 2306
    },
    {
      "epoch": 0.3291012838801712,
      "grad_norm": 0.7000129222869873,
      "learning_rate": 9.834559451496234e-06,
      "loss": 0.8138,
      "step": 2307
    },
    {
      "epoch": 0.329243937232525,
      "grad_norm": 0.8433248400688171,
      "learning_rate": 9.834362949864129e-06,
      "loss": 0.779,
      "step": 2308
    },
    {
      "epoch": 0.32938659058487874,
      "grad_norm": 1.2266490459442139,
      "learning_rate": 9.834166333569569e-06,
      "loss": 0.8196,
      "step": 2309
    },
    {
      "epoch": 0.32952924393723254,
      "grad_norm": 0.9142070412635803,
      "learning_rate": 9.833969602617218e-06,
      "loss": 0.55,
      "step": 2310
    },
    {
      "epoch": 0.3296718972895863,
      "grad_norm": 0.6996576189994812,
      "learning_rate": 9.833772757011736e-06,
      "loss": 0.8136,
      "step": 2311
    },
    {
      "epoch": 0.3298145506419401,
      "grad_norm": 1.056432843208313,
      "learning_rate": 9.8335757967578e-06,
      "loss": 0.7629,
      "step": 2312
    },
    {
      "epoch": 0.32995720399429385,
      "grad_norm": 0.7544792890548706,
      "learning_rate": 9.833378721860075e-06,
      "loss": 0.7414,
      "step": 2313
    },
    {
      "epoch": 0.33009985734664765,
      "grad_norm": 0.9668774604797363,
      "learning_rate": 9.833181532323237e-06,
      "loss": 0.6001,
      "step": 2314
    },
    {
      "epoch": 0.3302425106990014,
      "grad_norm": 0.7365550398826599,
      "learning_rate": 9.832984228151965e-06,
      "loss": 0.7766,
      "step": 2315
    },
    {
      "epoch": 0.3303851640513552,
      "grad_norm": 0.9288215637207031,
      "learning_rate": 9.832786809350938e-06,
      "loss": 0.7828,
      "step": 2316
    },
    {
      "epoch": 0.330527817403709,
      "grad_norm": 0.7639610767364502,
      "learning_rate": 9.832589275924838e-06,
      "loss": 0.7315,
      "step": 2317
    },
    {
      "epoch": 0.33067047075606276,
      "grad_norm": 1.0120751857757568,
      "learning_rate": 9.832391627878349e-06,
      "loss": 0.7244,
      "step": 2318
    },
    {
      "epoch": 0.33081312410841657,
      "grad_norm": 0.9512212872505188,
      "learning_rate": 9.832193865216159e-06,
      "loss": 0.7466,
      "step": 2319
    },
    {
      "epoch": 0.3309557774607703,
      "grad_norm": 1.036668300628662,
      "learning_rate": 9.831995987942961e-06,
      "loss": 0.8128,
      "step": 2320
    },
    {
      "epoch": 0.3310984308131241,
      "grad_norm": 0.7296528220176697,
      "learning_rate": 9.831797996063446e-06,
      "loss": 0.7596,
      "step": 2321
    },
    {
      "epoch": 0.3312410841654779,
      "grad_norm": 0.8461437821388245,
      "learning_rate": 9.83159988958231e-06,
      "loss": 0.732,
      "step": 2322
    },
    {
      "epoch": 0.3313837375178317,
      "grad_norm": 0.9580225348472595,
      "learning_rate": 9.831401668504252e-06,
      "loss": 0.7528,
      "step": 2323
    },
    {
      "epoch": 0.33152639087018543,
      "grad_norm": 0.9040142893791199,
      "learning_rate": 9.831203332833977e-06,
      "loss": 0.7836,
      "step": 2324
    },
    {
      "epoch": 0.33166904422253923,
      "grad_norm": 0.7880234718322754,
      "learning_rate": 9.831004882576183e-06,
      "loss": 0.7709,
      "step": 2325
    },
    {
      "epoch": 0.331811697574893,
      "grad_norm": 0.8258834481239319,
      "learning_rate": 9.830806317735581e-06,
      "loss": 0.7916,
      "step": 2326
    },
    {
      "epoch": 0.3319543509272468,
      "grad_norm": 0.7732114195823669,
      "learning_rate": 9.83060763831688e-06,
      "loss": 0.7807,
      "step": 2327
    },
    {
      "epoch": 0.3320970042796006,
      "grad_norm": 0.9464761018753052,
      "learning_rate": 9.83040884432479e-06,
      "loss": 0.837,
      "step": 2328
    },
    {
      "epoch": 0.33223965763195434,
      "grad_norm": 0.8679453730583191,
      "learning_rate": 9.83020993576403e-06,
      "loss": 0.8117,
      "step": 2329
    },
    {
      "epoch": 0.33238231098430815,
      "grad_norm": 0.8833606839179993,
      "learning_rate": 9.830010912639315e-06,
      "loss": 0.8299,
      "step": 2330
    },
    {
      "epoch": 0.3325249643366619,
      "grad_norm": 1.0865508317947388,
      "learning_rate": 9.829811774955367e-06,
      "loss": 0.7529,
      "step": 2331
    },
    {
      "epoch": 0.3326676176890157,
      "grad_norm": 0.9653570652008057,
      "learning_rate": 9.829612522716908e-06,
      "loss": 0.8046,
      "step": 2332
    },
    {
      "epoch": 0.33281027104136945,
      "grad_norm": 0.8964740633964539,
      "learning_rate": 9.829413155928662e-06,
      "loss": 0.7528,
      "step": 2333
    },
    {
      "epoch": 0.33295292439372326,
      "grad_norm": 0.8548893928527832,
      "learning_rate": 9.829213674595362e-06,
      "loss": 0.7959,
      "step": 2334
    },
    {
      "epoch": 0.333095577746077,
      "grad_norm": 1.0186488628387451,
      "learning_rate": 9.829014078721736e-06,
      "loss": 0.7133,
      "step": 2335
    },
    {
      "epoch": 0.3332382310984308,
      "grad_norm": 0.8716478943824768,
      "learning_rate": 9.82881436831252e-06,
      "loss": 0.7676,
      "step": 2336
    },
    {
      "epoch": 0.3333808844507846,
      "grad_norm": 0.900202751159668,
      "learning_rate": 9.82861454337245e-06,
      "loss": 0.8241,
      "step": 2337
    },
    {
      "epoch": 0.33352353780313837,
      "grad_norm": 0.6762422919273376,
      "learning_rate": 9.828414603906265e-06,
      "loss": 0.7335,
      "step": 2338
    },
    {
      "epoch": 0.3336661911554922,
      "grad_norm": 0.850338339805603,
      "learning_rate": 9.82821454991871e-06,
      "loss": 0.867,
      "step": 2339
    },
    {
      "epoch": 0.3338088445078459,
      "grad_norm": 1.1082850694656372,
      "learning_rate": 9.828014381414525e-06,
      "loss": 0.755,
      "step": 2340
    },
    {
      "epoch": 0.33395149786019973,
      "grad_norm": 0.6684832572937012,
      "learning_rate": 9.82781409839846e-06,
      "loss": 0.8119,
      "step": 2341
    },
    {
      "epoch": 0.3340941512125535,
      "grad_norm": 0.7720212936401367,
      "learning_rate": 9.827613700875267e-06,
      "loss": 0.8271,
      "step": 2342
    },
    {
      "epoch": 0.3342368045649073,
      "grad_norm": 0.7583309412002563,
      "learning_rate": 9.827413188849698e-06,
      "loss": 0.758,
      "step": 2343
    },
    {
      "epoch": 0.33437945791726104,
      "grad_norm": 0.9926135540008545,
      "learning_rate": 9.827212562326508e-06,
      "loss": 0.8526,
      "step": 2344
    },
    {
      "epoch": 0.33452211126961484,
      "grad_norm": 0.7970448732376099,
      "learning_rate": 9.827011821310455e-06,
      "loss": 0.8293,
      "step": 2345
    },
    {
      "epoch": 0.3346647646219686,
      "grad_norm": 0.8557180762290955,
      "learning_rate": 9.826810965806302e-06,
      "loss": 0.7789,
      "step": 2346
    },
    {
      "epoch": 0.3348074179743224,
      "grad_norm": 0.8372285962104797,
      "learning_rate": 9.826609995818812e-06,
      "loss": 0.6468,
      "step": 2347
    },
    {
      "epoch": 0.3349500713266762,
      "grad_norm": 0.9167457222938538,
      "learning_rate": 9.826408911352753e-06,
      "loss": 0.8412,
      "step": 2348
    },
    {
      "epoch": 0.33509272467902995,
      "grad_norm": 1.0763708353042603,
      "learning_rate": 9.826207712412891e-06,
      "loss": 0.7799,
      "step": 2349
    },
    {
      "epoch": 0.33523537803138376,
      "grad_norm": 0.9208558201789856,
      "learning_rate": 9.826006399004003e-06,
      "loss": 0.8056,
      "step": 2350
    },
    {
      "epoch": 0.3353780313837375,
      "grad_norm": 0.7773422002792358,
      "learning_rate": 9.82580497113086e-06,
      "loss": 0.7237,
      "step": 2351
    },
    {
      "epoch": 0.3355206847360913,
      "grad_norm": 0.7527308464050293,
      "learning_rate": 9.82560342879824e-06,
      "loss": 0.8651,
      "step": 2352
    },
    {
      "epoch": 0.33566333808844506,
      "grad_norm": 0.9323861598968506,
      "learning_rate": 9.825401772010923e-06,
      "loss": 0.7686,
      "step": 2353
    },
    {
      "epoch": 0.33580599144079887,
      "grad_norm": 0.8846133947372437,
      "learning_rate": 9.825200000773694e-06,
      "loss": 0.7928,
      "step": 2354
    },
    {
      "epoch": 0.3359486447931526,
      "grad_norm": 0.8226498961448669,
      "learning_rate": 9.824998115091336e-06,
      "loss": 0.7947,
      "step": 2355
    },
    {
      "epoch": 0.3360912981455064,
      "grad_norm": 0.9621587991714478,
      "learning_rate": 9.82479611496864e-06,
      "loss": 0.783,
      "step": 2356
    },
    {
      "epoch": 0.3362339514978602,
      "grad_norm": 0.9527766108512878,
      "learning_rate": 9.824594000410395e-06,
      "loss": 0.8012,
      "step": 2357
    },
    {
      "epoch": 0.336376604850214,
      "grad_norm": 1.204533576965332,
      "learning_rate": 9.824391771421398e-06,
      "loss": 0.8526,
      "step": 2358
    },
    {
      "epoch": 0.3365192582025678,
      "grad_norm": 0.8632824420928955,
      "learning_rate": 9.824189428006441e-06,
      "loss": 0.8141,
      "step": 2359
    },
    {
      "epoch": 0.33666191155492153,
      "grad_norm": 1.0136874914169312,
      "learning_rate": 9.823986970170324e-06,
      "loss": 0.8182,
      "step": 2360
    },
    {
      "epoch": 0.33680456490727534,
      "grad_norm": 1.1227818727493286,
      "learning_rate": 9.823784397917852e-06,
      "loss": 0.7657,
      "step": 2361
    },
    {
      "epoch": 0.3369472182596291,
      "grad_norm": 0.689673662185669,
      "learning_rate": 9.823581711253828e-06,
      "loss": 0.7461,
      "step": 2362
    },
    {
      "epoch": 0.3370898716119829,
      "grad_norm": 1.087032675743103,
      "learning_rate": 9.82337891018306e-06,
      "loss": 0.7345,
      "step": 2363
    },
    {
      "epoch": 0.33723252496433664,
      "grad_norm": 0.8440565466880798,
      "learning_rate": 9.823175994710356e-06,
      "loss": 0.7913,
      "step": 2364
    },
    {
      "epoch": 0.33737517831669045,
      "grad_norm": 1.036074161529541,
      "learning_rate": 9.82297296484053e-06,
      "loss": 0.8048,
      "step": 2365
    },
    {
      "epoch": 0.3375178316690442,
      "grad_norm": 1.0963290929794312,
      "learning_rate": 9.8227698205784e-06,
      "loss": 0.7909,
      "step": 2366
    },
    {
      "epoch": 0.337660485021398,
      "grad_norm": 0.7091878652572632,
      "learning_rate": 9.82256656192878e-06,
      "loss": 0.8131,
      "step": 2367
    },
    {
      "epoch": 0.3378031383737518,
      "grad_norm": 0.752960741519928,
      "learning_rate": 9.822363188896492e-06,
      "loss": 0.7583,
      "step": 2368
    },
    {
      "epoch": 0.33794579172610556,
      "grad_norm": 0.8184833526611328,
      "learning_rate": 9.822159701486363e-06,
      "loss": 0.8417,
      "step": 2369
    },
    {
      "epoch": 0.33808844507845937,
      "grad_norm": 0.7302771210670471,
      "learning_rate": 9.821956099703213e-06,
      "loss": 0.7717,
      "step": 2370
    },
    {
      "epoch": 0.3382310984308131,
      "grad_norm": 1.1408780813217163,
      "learning_rate": 9.821752383551879e-06,
      "loss": 0.7666,
      "step": 2371
    },
    {
      "epoch": 0.3383737517831669,
      "grad_norm": 1.0522292852401733,
      "learning_rate": 9.821548553037185e-06,
      "loss": 0.8067,
      "step": 2372
    },
    {
      "epoch": 0.33851640513552067,
      "grad_norm": 0.8145163059234619,
      "learning_rate": 9.821344608163972e-06,
      "loss": 0.8059,
      "step": 2373
    },
    {
      "epoch": 0.3386590584878745,
      "grad_norm": 0.6594700813293457,
      "learning_rate": 9.821140548937073e-06,
      "loss": 0.7328,
      "step": 2374
    },
    {
      "epoch": 0.3388017118402282,
      "grad_norm": 0.9611747860908508,
      "learning_rate": 9.82093637536133e-06,
      "loss": 0.8003,
      "step": 2375
    },
    {
      "epoch": 0.33894436519258203,
      "grad_norm": 1.3086268901824951,
      "learning_rate": 9.820732087441586e-06,
      "loss": 0.7714,
      "step": 2376
    },
    {
      "epoch": 0.3390870185449358,
      "grad_norm": 0.7010059356689453,
      "learning_rate": 9.820527685182684e-06,
      "loss": 0.8639,
      "step": 2377
    },
    {
      "epoch": 0.3392296718972896,
      "grad_norm": 0.9378232359886169,
      "learning_rate": 9.820323168589473e-06,
      "loss": 0.7348,
      "step": 2378
    },
    {
      "epoch": 0.3393723252496434,
      "grad_norm": 0.8938297629356384,
      "learning_rate": 9.820118537666805e-06,
      "loss": 0.8376,
      "step": 2379
    },
    {
      "epoch": 0.33951497860199714,
      "grad_norm": 0.7937347292900085,
      "learning_rate": 9.819913792419532e-06,
      "loss": 0.8924,
      "step": 2380
    },
    {
      "epoch": 0.33965763195435095,
      "grad_norm": 0.9030845165252686,
      "learning_rate": 9.819708932852512e-06,
      "loss": 0.8003,
      "step": 2381
    },
    {
      "epoch": 0.3398002853067047,
      "grad_norm": 0.7395643591880798,
      "learning_rate": 9.8195039589706e-06,
      "loss": 0.8144,
      "step": 2382
    },
    {
      "epoch": 0.3399429386590585,
      "grad_norm": 1.0145503282546997,
      "learning_rate": 9.819298870778663e-06,
      "loss": 0.77,
      "step": 2383
    },
    {
      "epoch": 0.34008559201141225,
      "grad_norm": 0.884653627872467,
      "learning_rate": 9.819093668281561e-06,
      "loss": 0.8517,
      "step": 2384
    },
    {
      "epoch": 0.34022824536376606,
      "grad_norm": 0.9484686851501465,
      "learning_rate": 9.818888351484164e-06,
      "loss": 0.7931,
      "step": 2385
    },
    {
      "epoch": 0.3403708987161198,
      "grad_norm": 0.7944343090057373,
      "learning_rate": 9.818682920391337e-06,
      "loss": 0.7936,
      "step": 2386
    },
    {
      "epoch": 0.3405135520684736,
      "grad_norm": 1.1869202852249146,
      "learning_rate": 9.81847737500796e-06,
      "loss": 0.7966,
      "step": 2387
    },
    {
      "epoch": 0.34065620542082736,
      "grad_norm": 0.7878841161727905,
      "learning_rate": 9.818271715338903e-06,
      "loss": 0.7684,
      "step": 2388
    },
    {
      "epoch": 0.34079885877318117,
      "grad_norm": 0.9743000268936157,
      "learning_rate": 9.818065941389044e-06,
      "loss": 0.7204,
      "step": 2389
    },
    {
      "epoch": 0.340941512125535,
      "grad_norm": 0.7342555522918701,
      "learning_rate": 9.817860053163266e-06,
      "loss": 0.8814,
      "step": 2390
    },
    {
      "epoch": 0.3410841654778887,
      "grad_norm": 0.9856871962547302,
      "learning_rate": 9.81765405066645e-06,
      "loss": 0.8494,
      "step": 2391
    },
    {
      "epoch": 0.34122681883024253,
      "grad_norm": 0.9224755764007568,
      "learning_rate": 9.817447933903481e-06,
      "loss": 0.7714,
      "step": 2392
    },
    {
      "epoch": 0.3413694721825963,
      "grad_norm": 0.8648902177810669,
      "learning_rate": 9.817241702879253e-06,
      "loss": 0.8198,
      "step": 2393
    },
    {
      "epoch": 0.3415121255349501,
      "grad_norm": 0.7871795892715454,
      "learning_rate": 9.817035357598652e-06,
      "loss": 0.7751,
      "step": 2394
    },
    {
      "epoch": 0.34165477888730383,
      "grad_norm": 0.8133195042610168,
      "learning_rate": 9.816828898066574e-06,
      "loss": 0.8081,
      "step": 2395
    },
    {
      "epoch": 0.34179743223965764,
      "grad_norm": 0.8734281659126282,
      "learning_rate": 9.816622324287916e-06,
      "loss": 0.8192,
      "step": 2396
    },
    {
      "epoch": 0.3419400855920114,
      "grad_norm": 0.8564184308052063,
      "learning_rate": 9.816415636267578e-06,
      "loss": 0.7833,
      "step": 2397
    },
    {
      "epoch": 0.3420827389443652,
      "grad_norm": 1.0076076984405518,
      "learning_rate": 9.81620883401046e-06,
      "loss": 0.8053,
      "step": 2398
    },
    {
      "epoch": 0.342225392296719,
      "grad_norm": 0.8750330805778503,
      "learning_rate": 9.816001917521472e-06,
      "loss": 0.7704,
      "step": 2399
    },
    {
      "epoch": 0.34236804564907275,
      "grad_norm": 0.8701121807098389,
      "learning_rate": 9.815794886805517e-06,
      "loss": 0.7769,
      "step": 2400
    },
    {
      "epoch": 0.34251069900142656,
      "grad_norm": 0.8542219400405884,
      "learning_rate": 9.815587741867507e-06,
      "loss": 0.8211,
      "step": 2401
    },
    {
      "epoch": 0.3426533523537803,
      "grad_norm": 1.0593914985656738,
      "learning_rate": 9.815380482712353e-06,
      "loss": 0.7406,
      "step": 2402
    },
    {
      "epoch": 0.3427960057061341,
      "grad_norm": 1.03482985496521,
      "learning_rate": 9.815173109344976e-06,
      "loss": 0.7539,
      "step": 2403
    },
    {
      "epoch": 0.34293865905848786,
      "grad_norm": 0.9401360154151917,
      "learning_rate": 9.814965621770289e-06,
      "loss": 0.7953,
      "step": 2404
    },
    {
      "epoch": 0.34308131241084167,
      "grad_norm": 1.158489465713501,
      "learning_rate": 9.814758019993216e-06,
      "loss": 0.7696,
      "step": 2405
    },
    {
      "epoch": 0.3432239657631954,
      "grad_norm": 1.2007076740264893,
      "learning_rate": 9.814550304018682e-06,
      "loss": 0.8182,
      "step": 2406
    },
    {
      "epoch": 0.3433666191155492,
      "grad_norm": 1.037165880203247,
      "learning_rate": 9.81434247385161e-06,
      "loss": 0.8141,
      "step": 2407
    },
    {
      "epoch": 0.34350927246790297,
      "grad_norm": 1.1913652420043945,
      "learning_rate": 9.814134529496934e-06,
      "loss": 0.7943,
      "step": 2408
    },
    {
      "epoch": 0.3436519258202568,
      "grad_norm": 1.300634741783142,
      "learning_rate": 9.813926470959582e-06,
      "loss": 0.6194,
      "step": 2409
    },
    {
      "epoch": 0.3437945791726106,
      "grad_norm": 0.9805374145507812,
      "learning_rate": 9.81371829824449e-06,
      "loss": 0.7247,
      "step": 2410
    },
    {
      "epoch": 0.34393723252496433,
      "grad_norm": 1.0163896083831787,
      "learning_rate": 9.813510011356597e-06,
      "loss": 0.8524,
      "step": 2411
    },
    {
      "epoch": 0.34407988587731814,
      "grad_norm": 0.9004878997802734,
      "learning_rate": 9.813301610300841e-06,
      "loss": 0.8224,
      "step": 2412
    },
    {
      "epoch": 0.3442225392296719,
      "grad_norm": 0.8286635875701904,
      "learning_rate": 9.813093095082167e-06,
      "loss": 0.8188,
      "step": 2413
    },
    {
      "epoch": 0.3443651925820257,
      "grad_norm": 0.9868021011352539,
      "learning_rate": 9.81288446570552e-06,
      "loss": 0.8024,
      "step": 2414
    },
    {
      "epoch": 0.34450784593437944,
      "grad_norm": 0.8571799993515015,
      "learning_rate": 9.812675722175846e-06,
      "loss": 0.7445,
      "step": 2415
    },
    {
      "epoch": 0.34465049928673325,
      "grad_norm": 0.7464162111282349,
      "learning_rate": 9.812466864498097e-06,
      "loss": 0.7484,
      "step": 2416
    },
    {
      "epoch": 0.344793152639087,
      "grad_norm": 1.1445914506912231,
      "learning_rate": 9.81225789267723e-06,
      "loss": 0.8077,
      "step": 2417
    },
    {
      "epoch": 0.3449358059914408,
      "grad_norm": 1.1138544082641602,
      "learning_rate": 9.812048806718199e-06,
      "loss": 0.8036,
      "step": 2418
    },
    {
      "epoch": 0.34507845934379455,
      "grad_norm": 1.2120366096496582,
      "learning_rate": 9.811839606625962e-06,
      "loss": 0.7882,
      "step": 2419
    },
    {
      "epoch": 0.34522111269614836,
      "grad_norm": 0.9743795990943909,
      "learning_rate": 9.811630292405484e-06,
      "loss": 0.7773,
      "step": 2420
    },
    {
      "epoch": 0.34536376604850216,
      "grad_norm": 0.919040322303772,
      "learning_rate": 9.811420864061725e-06,
      "loss": 0.7424,
      "step": 2421
    },
    {
      "epoch": 0.3455064194008559,
      "grad_norm": 0.8817082643508911,
      "learning_rate": 9.811211321599657e-06,
      "loss": 0.8003,
      "step": 2422
    },
    {
      "epoch": 0.3456490727532097,
      "grad_norm": 1.0432809591293335,
      "learning_rate": 9.811001665024246e-06,
      "loss": 0.7844,
      "step": 2423
    },
    {
      "epoch": 0.34579172610556347,
      "grad_norm": 1.0473785400390625,
      "learning_rate": 9.810791894340467e-06,
      "loss": 0.7574,
      "step": 2424
    },
    {
      "epoch": 0.3459343794579173,
      "grad_norm": 0.7565572261810303,
      "learning_rate": 9.810582009553296e-06,
      "loss": 0.7903,
      "step": 2425
    },
    {
      "epoch": 0.346077032810271,
      "grad_norm": 1.2424083948135376,
      "learning_rate": 9.81037201066771e-06,
      "loss": 0.8094,
      "step": 2426
    },
    {
      "epoch": 0.34621968616262483,
      "grad_norm": 0.8809653520584106,
      "learning_rate": 9.810161897688688e-06,
      "loss": 0.739,
      "step": 2427
    },
    {
      "epoch": 0.3463623395149786,
      "grad_norm": 1.0089491605758667,
      "learning_rate": 9.809951670621218e-06,
      "loss": 0.751,
      "step": 2428
    },
    {
      "epoch": 0.3465049928673324,
      "grad_norm": 0.7561847567558289,
      "learning_rate": 9.809741329470281e-06,
      "loss": 0.8513,
      "step": 2429
    },
    {
      "epoch": 0.3466476462196862,
      "grad_norm": 0.7132732272148132,
      "learning_rate": 9.809530874240869e-06,
      "loss": 0.7771,
      "step": 2430
    },
    {
      "epoch": 0.34679029957203994,
      "grad_norm": 1.2511911392211914,
      "learning_rate": 9.809320304937974e-06,
      "loss": 0.8489,
      "step": 2431
    },
    {
      "epoch": 0.34693295292439374,
      "grad_norm": 0.8901185989379883,
      "learning_rate": 9.809109621566589e-06,
      "loss": 0.7419,
      "step": 2432
    },
    {
      "epoch": 0.3470756062767475,
      "grad_norm": 2.0431883335113525,
      "learning_rate": 9.808898824131712e-06,
      "loss": 0.6619,
      "step": 2433
    },
    {
      "epoch": 0.3472182596291013,
      "grad_norm": 0.8219075798988342,
      "learning_rate": 9.808687912638342e-06,
      "loss": 0.8378,
      "step": 2434
    },
    {
      "epoch": 0.34736091298145505,
      "grad_norm": 1.001102328300476,
      "learning_rate": 9.80847688709148e-06,
      "loss": 0.8029,
      "step": 2435
    },
    {
      "epoch": 0.34750356633380886,
      "grad_norm": 0.7215900421142578,
      "learning_rate": 9.808265747496135e-06,
      "loss": 0.8174,
      "step": 2436
    },
    {
      "epoch": 0.3476462196861626,
      "grad_norm": 0.9470601081848145,
      "learning_rate": 9.808054493857311e-06,
      "loss": 0.7696,
      "step": 2437
    },
    {
      "epoch": 0.3477888730385164,
      "grad_norm": 0.902509331703186,
      "learning_rate": 9.807843126180022e-06,
      "loss": 0.7846,
      "step": 2438
    },
    {
      "epoch": 0.34793152639087016,
      "grad_norm": 0.8843024969100952,
      "learning_rate": 9.807631644469277e-06,
      "loss": 0.8151,
      "step": 2439
    },
    {
      "epoch": 0.34807417974322397,
      "grad_norm": 0.8945369124412537,
      "learning_rate": 9.807420048730095e-06,
      "loss": 0.8141,
      "step": 2440
    },
    {
      "epoch": 0.34821683309557777,
      "grad_norm": 0.9173300862312317,
      "learning_rate": 9.807208338967495e-06,
      "loss": 0.8004,
      "step": 2441
    },
    {
      "epoch": 0.3483594864479315,
      "grad_norm": 0.9097233414649963,
      "learning_rate": 9.806996515186498e-06,
      "loss": 0.8155,
      "step": 2442
    },
    {
      "epoch": 0.3485021398002853,
      "grad_norm": 0.7454264760017395,
      "learning_rate": 9.806784577392125e-06,
      "loss": 0.8061,
      "step": 2443
    },
    {
      "epoch": 0.3486447931526391,
      "grad_norm": 0.9167298674583435,
      "learning_rate": 9.80657252558941e-06,
      "loss": 0.7576,
      "step": 2444
    },
    {
      "epoch": 0.3487874465049929,
      "grad_norm": 0.8795915842056274,
      "learning_rate": 9.806360359783374e-06,
      "loss": 0.7774,
      "step": 2445
    },
    {
      "epoch": 0.34893009985734663,
      "grad_norm": 0.7418687343597412,
      "learning_rate": 9.806148079979052e-06,
      "loss": 0.7697,
      "step": 2446
    },
    {
      "epoch": 0.34907275320970044,
      "grad_norm": 1.1968201398849487,
      "learning_rate": 9.805935686181483e-06,
      "loss": 0.7797,
      "step": 2447
    },
    {
      "epoch": 0.3492154065620542,
      "grad_norm": 0.8256352543830872,
      "learning_rate": 9.805723178395699e-06,
      "loss": 0.7957,
      "step": 2448
    },
    {
      "epoch": 0.349358059914408,
      "grad_norm": 0.9797470569610596,
      "learning_rate": 9.805510556626744e-06,
      "loss": 0.719,
      "step": 2449
    },
    {
      "epoch": 0.34950071326676174,
      "grad_norm": 0.8725929856300354,
      "learning_rate": 9.80529782087966e-06,
      "loss": 0.8111,
      "step": 2450
    },
    {
      "epoch": 0.34964336661911555,
      "grad_norm": 0.9257572293281555,
      "learning_rate": 9.805084971159493e-06,
      "loss": 0.7767,
      "step": 2451
    },
    {
      "epoch": 0.34978601997146935,
      "grad_norm": 0.8392630815505981,
      "learning_rate": 9.80487200747129e-06,
      "loss": 0.8188,
      "step": 2452
    },
    {
      "epoch": 0.3499286733238231,
      "grad_norm": 0.9308575987815857,
      "learning_rate": 9.804658929820104e-06,
      "loss": 0.748,
      "step": 2453
    },
    {
      "epoch": 0.3500713266761769,
      "grad_norm": 0.8789147734642029,
      "learning_rate": 9.804445738210988e-06,
      "loss": 0.8509,
      "step": 2454
    },
    {
      "epoch": 0.35021398002853066,
      "grad_norm": 0.9652302861213684,
      "learning_rate": 9.804232432648997e-06,
      "loss": 0.7711,
      "step": 2455
    },
    {
      "epoch": 0.35035663338088446,
      "grad_norm": 1.293696641921997,
      "learning_rate": 9.804019013139192e-06,
      "loss": 0.8026,
      "step": 2456
    },
    {
      "epoch": 0.3504992867332382,
      "grad_norm": 0.6802603602409363,
      "learning_rate": 9.803805479686635e-06,
      "loss": 0.7896,
      "step": 2457
    },
    {
      "epoch": 0.350641940085592,
      "grad_norm": 0.9308311939239502,
      "learning_rate": 9.80359183229639e-06,
      "loss": 0.7302,
      "step": 2458
    },
    {
      "epoch": 0.35078459343794577,
      "grad_norm": 0.8198142647743225,
      "learning_rate": 9.803378070973524e-06,
      "loss": 0.7897,
      "step": 2459
    },
    {
      "epoch": 0.3509272467902996,
      "grad_norm": 1.0778833627700806,
      "learning_rate": 9.80316419572311e-06,
      "loss": 0.8393,
      "step": 2460
    },
    {
      "epoch": 0.3510699001426534,
      "grad_norm": 0.6993482708930969,
      "learning_rate": 9.802950206550215e-06,
      "loss": 0.7964,
      "step": 2461
    },
    {
      "epoch": 0.35121255349500713,
      "grad_norm": 0.9422694444656372,
      "learning_rate": 9.80273610345992e-06,
      "loss": 0.6952,
      "step": 2462
    },
    {
      "epoch": 0.35135520684736093,
      "grad_norm": 1.0927560329437256,
      "learning_rate": 9.802521886457298e-06,
      "loss": 0.9076,
      "step": 2463
    },
    {
      "epoch": 0.3514978601997147,
      "grad_norm": 0.8735848069190979,
      "learning_rate": 9.802307555547436e-06,
      "loss": 0.7146,
      "step": 2464
    },
    {
      "epoch": 0.3516405135520685,
      "grad_norm": 1.0025066137313843,
      "learning_rate": 9.802093110735414e-06,
      "loss": 0.7766,
      "step": 2465
    },
    {
      "epoch": 0.35178316690442224,
      "grad_norm": 0.7589578628540039,
      "learning_rate": 9.801878552026318e-06,
      "loss": 0.8558,
      "step": 2466
    },
    {
      "epoch": 0.35192582025677605,
      "grad_norm": 0.7957066297531128,
      "learning_rate": 9.801663879425236e-06,
      "loss": 0.7767,
      "step": 2467
    },
    {
      "epoch": 0.3520684736091298,
      "grad_norm": 2.2545599937438965,
      "learning_rate": 9.801449092937263e-06,
      "loss": 0.644,
      "step": 2468
    },
    {
      "epoch": 0.3522111269614836,
      "grad_norm": 0.9092872738838196,
      "learning_rate": 9.80123419256749e-06,
      "loss": 0.7203,
      "step": 2469
    },
    {
      "epoch": 0.35235378031383735,
      "grad_norm": 1.0055960416793823,
      "learning_rate": 9.801019178321017e-06,
      "loss": 0.7836,
      "step": 2470
    },
    {
      "epoch": 0.35249643366619116,
      "grad_norm": 3.424903631210327,
      "learning_rate": 9.800804050202942e-06,
      "loss": 0.6291,
      "step": 2471
    },
    {
      "epoch": 0.35263908701854496,
      "grad_norm": 1.0196425914764404,
      "learning_rate": 9.800588808218368e-06,
      "loss": 0.7185,
      "step": 2472
    },
    {
      "epoch": 0.3527817403708987,
      "grad_norm": 0.8520036935806274,
      "learning_rate": 9.800373452372399e-06,
      "loss": 0.6775,
      "step": 2473
    },
    {
      "epoch": 0.3529243937232525,
      "grad_norm": 0.9552729725837708,
      "learning_rate": 9.800157982670145e-06,
      "loss": 0.7737,
      "step": 2474
    },
    {
      "epoch": 0.35306704707560627,
      "grad_norm": 0.7399102449417114,
      "learning_rate": 9.799942399116713e-06,
      "loss": 0.7775,
      "step": 2475
    },
    {
      "epoch": 0.35320970042796007,
      "grad_norm": 1.059720754623413,
      "learning_rate": 9.79972670171722e-06,
      "loss": 0.8431,
      "step": 2476
    },
    {
      "epoch": 0.3533523537803138,
      "grad_norm": 1.0841524600982666,
      "learning_rate": 9.799510890476782e-06,
      "loss": 0.7776,
      "step": 2477
    },
    {
      "epoch": 0.3534950071326676,
      "grad_norm": 0.7432464361190796,
      "learning_rate": 9.799294965400514e-06,
      "loss": 0.8011,
      "step": 2478
    },
    {
      "epoch": 0.3536376604850214,
      "grad_norm": 0.9959025979042053,
      "learning_rate": 9.799078926493539e-06,
      "loss": 0.7847,
      "step": 2479
    },
    {
      "epoch": 0.3537803138373752,
      "grad_norm": 1.0784518718719482,
      "learning_rate": 9.798862773760984e-06,
      "loss": 0.7722,
      "step": 2480
    },
    {
      "epoch": 0.35392296718972893,
      "grad_norm": 0.867393970489502,
      "learning_rate": 9.798646507207971e-06,
      "loss": 0.8095,
      "step": 2481
    },
    {
      "epoch": 0.35406562054208274,
      "grad_norm": 0.7331669330596924,
      "learning_rate": 9.798430126839633e-06,
      "loss": 0.805,
      "step": 2482
    },
    {
      "epoch": 0.35420827389443654,
      "grad_norm": 1.0410361289978027,
      "learning_rate": 9.798213632661102e-06,
      "loss": 0.8331,
      "step": 2483
    },
    {
      "epoch": 0.3543509272467903,
      "grad_norm": 1.0381349325180054,
      "learning_rate": 9.79799702467751e-06,
      "loss": 0.8218,
      "step": 2484
    },
    {
      "epoch": 0.3544935805991441,
      "grad_norm": 0.6517478823661804,
      "learning_rate": 9.797780302893998e-06,
      "loss": 0.7942,
      "step": 2485
    },
    {
      "epoch": 0.35463623395149785,
      "grad_norm": 0.8073909282684326,
      "learning_rate": 9.797563467315703e-06,
      "loss": 0.8717,
      "step": 2486
    },
    {
      "epoch": 0.35477888730385165,
      "grad_norm": 0.8968905806541443,
      "learning_rate": 9.797346517947772e-06,
      "loss": 0.8036,
      "step": 2487
    },
    {
      "epoch": 0.3549215406562054,
      "grad_norm": 0.7728807330131531,
      "learning_rate": 9.797129454795346e-06,
      "loss": 0.8009,
      "step": 2488
    },
    {
      "epoch": 0.3550641940085592,
      "grad_norm": 0.7919692993164062,
      "learning_rate": 9.796912277863577e-06,
      "loss": 0.8218,
      "step": 2489
    },
    {
      "epoch": 0.35520684736091296,
      "grad_norm": 0.9538540244102478,
      "learning_rate": 9.796694987157614e-06,
      "loss": 0.8023,
      "step": 2490
    },
    {
      "epoch": 0.35534950071326676,
      "grad_norm": 0.9211897850036621,
      "learning_rate": 9.796477582682611e-06,
      "loss": 0.7562,
      "step": 2491
    },
    {
      "epoch": 0.35549215406562057,
      "grad_norm": 0.9030013084411621,
      "learning_rate": 9.796260064443724e-06,
      "loss": 0.8064,
      "step": 2492
    },
    {
      "epoch": 0.3556348074179743,
      "grad_norm": 1.1324703693389893,
      "learning_rate": 9.796042432446114e-06,
      "loss": 0.8613,
      "step": 2493
    },
    {
      "epoch": 0.3557774607703281,
      "grad_norm": 0.8738144636154175,
      "learning_rate": 9.795824686694943e-06,
      "loss": 0.7492,
      "step": 2494
    },
    {
      "epoch": 0.3559201141226819,
      "grad_norm": 0.70047926902771,
      "learning_rate": 9.795606827195373e-06,
      "loss": 0.5744,
      "step": 2495
    },
    {
      "epoch": 0.3560627674750357,
      "grad_norm": 0.8143078088760376,
      "learning_rate": 9.795388853952573e-06,
      "loss": 0.8359,
      "step": 2496
    },
    {
      "epoch": 0.35620542082738943,
      "grad_norm": 1.0857017040252686,
      "learning_rate": 9.795170766971712e-06,
      "loss": 0.8786,
      "step": 2497
    },
    {
      "epoch": 0.35634807417974323,
      "grad_norm": 0.7692033648490906,
      "learning_rate": 9.794952566257963e-06,
      "loss": 0.7636,
      "step": 2498
    },
    {
      "epoch": 0.356490727532097,
      "grad_norm": 0.6650281548500061,
      "learning_rate": 9.794734251816503e-06,
      "loss": 0.878,
      "step": 2499
    },
    {
      "epoch": 0.3566333808844508,
      "grad_norm": 0.8415030837059021,
      "learning_rate": 9.794515823652507e-06,
      "loss": 0.8108,
      "step": 2500
    },
    {
      "epoch": 0.35677603423680454,
      "grad_norm": 0.9431477189064026,
      "learning_rate": 9.794297281771158e-06,
      "loss": 0.879,
      "step": 2501
    },
    {
      "epoch": 0.35691868758915835,
      "grad_norm": 0.7415030002593994,
      "learning_rate": 9.794078626177639e-06,
      "loss": 0.9159,
      "step": 2502
    },
    {
      "epoch": 0.35706134094151215,
      "grad_norm": 0.9204234480857849,
      "learning_rate": 9.793859856877134e-06,
      "loss": 0.7233,
      "step": 2503
    },
    {
      "epoch": 0.3572039942938659,
      "grad_norm": 0.7926920652389526,
      "learning_rate": 9.793640973874837e-06,
      "loss": 0.7959,
      "step": 2504
    },
    {
      "epoch": 0.3573466476462197,
      "grad_norm": 0.8344845175743103,
      "learning_rate": 9.793421977175934e-06,
      "loss": 0.8314,
      "step": 2505
    },
    {
      "epoch": 0.35748930099857346,
      "grad_norm": 0.7620161175727844,
      "learning_rate": 9.79320286678562e-06,
      "loss": 0.7839,
      "step": 2506
    },
    {
      "epoch": 0.35763195435092726,
      "grad_norm": 0.9139984250068665,
      "learning_rate": 9.792983642709096e-06,
      "loss": 0.7797,
      "step": 2507
    },
    {
      "epoch": 0.357774607703281,
      "grad_norm": 0.6945261359214783,
      "learning_rate": 9.792764304951557e-06,
      "loss": 0.7363,
      "step": 2508
    },
    {
      "epoch": 0.3579172610556348,
      "grad_norm": 0.8204848170280457,
      "learning_rate": 9.792544853518206e-06,
      "loss": 0.7733,
      "step": 2509
    },
    {
      "epoch": 0.35805991440798857,
      "grad_norm": 0.8774577379226685,
      "learning_rate": 9.792325288414252e-06,
      "loss": 0.7407,
      "step": 2510
    },
    {
      "epoch": 0.35820256776034237,
      "grad_norm": 1.1311724185943604,
      "learning_rate": 9.792105609644899e-06,
      "loss": 0.8133,
      "step": 2511
    },
    {
      "epoch": 0.3583452211126961,
      "grad_norm": 0.7154818177223206,
      "learning_rate": 9.791885817215357e-06,
      "loss": 0.777,
      "step": 2512
    },
    {
      "epoch": 0.3584878744650499,
      "grad_norm": 1.033969521522522,
      "learning_rate": 9.791665911130841e-06,
      "loss": 0.8339,
      "step": 2513
    },
    {
      "epoch": 0.35863052781740373,
      "grad_norm": 1.0158448219299316,
      "learning_rate": 9.791445891396568e-06,
      "loss": 0.8523,
      "step": 2514
    },
    {
      "epoch": 0.3587731811697575,
      "grad_norm": 0.899519145488739,
      "learning_rate": 9.791225758017752e-06,
      "loss": 0.7123,
      "step": 2515
    },
    {
      "epoch": 0.3589158345221113,
      "grad_norm": 0.9491230845451355,
      "learning_rate": 9.791005510999619e-06,
      "loss": 0.787,
      "step": 2516
    },
    {
      "epoch": 0.35905848787446504,
      "grad_norm": 0.6641083359718323,
      "learning_rate": 9.790785150347388e-06,
      "loss": 0.7788,
      "step": 2517
    },
    {
      "epoch": 0.35920114122681884,
      "grad_norm": 0.931577205657959,
      "learning_rate": 9.79056467606629e-06,
      "loss": 0.8076,
      "step": 2518
    },
    {
      "epoch": 0.3593437945791726,
      "grad_norm": 0.8346139192581177,
      "learning_rate": 9.790344088161551e-06,
      "loss": 0.7443,
      "step": 2519
    },
    {
      "epoch": 0.3594864479315264,
      "grad_norm": 1.0931133031845093,
      "learning_rate": 9.790123386638405e-06,
      "loss": 0.8247,
      "step": 2520
    },
    {
      "epoch": 0.35962910128388015,
      "grad_norm": 0.7270719408988953,
      "learning_rate": 9.789902571502086e-06,
      "loss": 0.8063,
      "step": 2521
    },
    {
      "epoch": 0.35977175463623395,
      "grad_norm": 0.8199694752693176,
      "learning_rate": 9.789681642757832e-06,
      "loss": 0.8354,
      "step": 2522
    },
    {
      "epoch": 0.35991440798858776,
      "grad_norm": 1.01858389377594,
      "learning_rate": 9.789460600410882e-06,
      "loss": 0.8221,
      "step": 2523
    },
    {
      "epoch": 0.3600570613409415,
      "grad_norm": 0.8585506081581116,
      "learning_rate": 9.789239444466478e-06,
      "loss": 0.8108,
      "step": 2524
    },
    {
      "epoch": 0.3601997146932953,
      "grad_norm": 1.1112741231918335,
      "learning_rate": 9.789018174929868e-06,
      "loss": 0.7597,
      "step": 2525
    },
    {
      "epoch": 0.36034236804564906,
      "grad_norm": 1.0793445110321045,
      "learning_rate": 9.788796791806298e-06,
      "loss": 0.7678,
      "step": 2526
    },
    {
      "epoch": 0.36048502139800287,
      "grad_norm": 0.8803876042366028,
      "learning_rate": 9.78857529510102e-06,
      "loss": 0.8029,
      "step": 2527
    },
    {
      "epoch": 0.3606276747503566,
      "grad_norm": 0.7571743130683899,
      "learning_rate": 9.788353684819288e-06,
      "loss": 0.5882,
      "step": 2528
    },
    {
      "epoch": 0.3607703281027104,
      "grad_norm": 0.9352931380271912,
      "learning_rate": 9.788131960966353e-06,
      "loss": 0.7699,
      "step": 2529
    },
    {
      "epoch": 0.3609129814550642,
      "grad_norm": 0.7634739875793457,
      "learning_rate": 9.787910123547481e-06,
      "loss": 0.841,
      "step": 2530
    },
    {
      "epoch": 0.361055634807418,
      "grad_norm": 0.8426674604415894,
      "learning_rate": 9.787688172567932e-06,
      "loss": 0.8585,
      "step": 2531
    },
    {
      "epoch": 0.36119828815977173,
      "grad_norm": 0.8924207091331482,
      "learning_rate": 9.787466108032968e-06,
      "loss": 0.7654,
      "step": 2532
    },
    {
      "epoch": 0.36134094151212554,
      "grad_norm": 0.8272600769996643,
      "learning_rate": 9.787243929947857e-06,
      "loss": 0.7827,
      "step": 2533
    },
    {
      "epoch": 0.36148359486447934,
      "grad_norm": 0.8443830013275146,
      "learning_rate": 9.787021638317868e-06,
      "loss": 0.775,
      "step": 2534
    },
    {
      "epoch": 0.3616262482168331,
      "grad_norm": 1.0147038698196411,
      "learning_rate": 9.786799233148273e-06,
      "loss": 0.7492,
      "step": 2535
    },
    {
      "epoch": 0.3617689015691869,
      "grad_norm": 0.9985344409942627,
      "learning_rate": 9.786576714444349e-06,
      "loss": 0.8384,
      "step": 2536
    },
    {
      "epoch": 0.36191155492154065,
      "grad_norm": 0.573411762714386,
      "learning_rate": 9.786354082211374e-06,
      "loss": 0.7608,
      "step": 2537
    },
    {
      "epoch": 0.36205420827389445,
      "grad_norm": 0.9318847060203552,
      "learning_rate": 9.786131336454625e-06,
      "loss": 0.7311,
      "step": 2538
    },
    {
      "epoch": 0.3621968616262482,
      "grad_norm": 0.8152886033058167,
      "learning_rate": 9.78590847717939e-06,
      "loss": 0.7587,
      "step": 2539
    },
    {
      "epoch": 0.362339514978602,
      "grad_norm": 0.9994775056838989,
      "learning_rate": 9.78568550439095e-06,
      "loss": 0.8209,
      "step": 2540
    },
    {
      "epoch": 0.36248216833095576,
      "grad_norm": 0.8203306794166565,
      "learning_rate": 9.785462418094595e-06,
      "loss": 0.7314,
      "step": 2541
    },
    {
      "epoch": 0.36262482168330956,
      "grad_norm": 0.8252012729644775,
      "learning_rate": 9.785239218295618e-06,
      "loss": 0.8093,
      "step": 2542
    },
    {
      "epoch": 0.3627674750356633,
      "grad_norm": 0.8270312547683716,
      "learning_rate": 9.785015904999311e-06,
      "loss": 0.8546,
      "step": 2543
    },
    {
      "epoch": 0.3629101283880171,
      "grad_norm": 1.0600306987762451,
      "learning_rate": 9.784792478210971e-06,
      "loss": 0.8662,
      "step": 2544
    },
    {
      "epoch": 0.3630527817403709,
      "grad_norm": 0.9592757821083069,
      "learning_rate": 9.7845689379359e-06,
      "loss": 0.798,
      "step": 2545
    },
    {
      "epoch": 0.36319543509272467,
      "grad_norm": 0.9203529953956604,
      "learning_rate": 9.784345284179397e-06,
      "loss": 0.7938,
      "step": 2546
    },
    {
      "epoch": 0.3633380884450785,
      "grad_norm": 1.1838785409927368,
      "learning_rate": 9.784121516946765e-06,
      "loss": 0.8015,
      "step": 2547
    },
    {
      "epoch": 0.3634807417974322,
      "grad_norm": 1.0804641246795654,
      "learning_rate": 9.783897636243316e-06,
      "loss": 0.851,
      "step": 2548
    },
    {
      "epoch": 0.36362339514978603,
      "grad_norm": 0.8632106781005859,
      "learning_rate": 9.783673642074358e-06,
      "loss": 0.7753,
      "step": 2549
    },
    {
      "epoch": 0.3637660485021398,
      "grad_norm": 1.0359077453613281,
      "learning_rate": 9.783449534445202e-06,
      "loss": 0.7797,
      "step": 2550
    },
    {
      "epoch": 0.3639087018544936,
      "grad_norm": 0.8755848407745361,
      "learning_rate": 9.783225313361165e-06,
      "loss": 0.8103,
      "step": 2551
    },
    {
      "epoch": 0.36405135520684734,
      "grad_norm": 0.9089139699935913,
      "learning_rate": 9.783000978827565e-06,
      "loss": 0.8141,
      "step": 2552
    },
    {
      "epoch": 0.36419400855920114,
      "grad_norm": 1.0342538356781006,
      "learning_rate": 9.782776530849722e-06,
      "loss": 0.8129,
      "step": 2553
    },
    {
      "epoch": 0.36433666191155495,
      "grad_norm": 0.70041823387146,
      "learning_rate": 9.782551969432963e-06,
      "loss": 0.808,
      "step": 2554
    },
    {
      "epoch": 0.3644793152639087,
      "grad_norm": 0.9141285419464111,
      "learning_rate": 9.78232729458261e-06,
      "loss": 0.7329,
      "step": 2555
    },
    {
      "epoch": 0.3646219686162625,
      "grad_norm": 0.8585652709007263,
      "learning_rate": 9.782102506303993e-06,
      "loss": 0.7318,
      "step": 2556
    },
    {
      "epoch": 0.36476462196861625,
      "grad_norm": 1.0369824171066284,
      "learning_rate": 9.781877604602444e-06,
      "loss": 0.8093,
      "step": 2557
    },
    {
      "epoch": 0.36490727532097006,
      "grad_norm": 0.9943699240684509,
      "learning_rate": 9.781652589483296e-06,
      "loss": 0.8641,
      "step": 2558
    },
    {
      "epoch": 0.3650499286733238,
      "grad_norm": 0.9613951444625854,
      "learning_rate": 9.78142746095189e-06,
      "loss": 0.7307,
      "step": 2559
    },
    {
      "epoch": 0.3651925820256776,
      "grad_norm": 0.8950900435447693,
      "learning_rate": 9.781202219013561e-06,
      "loss": 0.8279,
      "step": 2560
    },
    {
      "epoch": 0.36533523537803136,
      "grad_norm": 1.0290483236312866,
      "learning_rate": 9.780976863673654e-06,
      "loss": 0.7668,
      "step": 2561
    },
    {
      "epoch": 0.36547788873038517,
      "grad_norm": 0.8867672085762024,
      "learning_rate": 9.780751394937513e-06,
      "loss": 0.797,
      "step": 2562
    },
    {
      "epoch": 0.3656205420827389,
      "grad_norm": 0.7999266982078552,
      "learning_rate": 9.780525812810486e-06,
      "loss": 0.5798,
      "step": 2563
    },
    {
      "epoch": 0.3657631954350927,
      "grad_norm": 0.9021796584129333,
      "learning_rate": 9.780300117297922e-06,
      "loss": 0.8297,
      "step": 2564
    },
    {
      "epoch": 0.36590584878744653,
      "grad_norm": 1.075351357460022,
      "learning_rate": 9.78007430840518e-06,
      "loss": 0.7587,
      "step": 2565
    },
    {
      "epoch": 0.3660485021398003,
      "grad_norm": 0.8415443897247314,
      "learning_rate": 9.779848386137606e-06,
      "loss": 0.6022,
      "step": 2566
    },
    {
      "epoch": 0.3661911554921541,
      "grad_norm": 1.243349313735962,
      "learning_rate": 9.779622350500565e-06,
      "loss": 0.7213,
      "step": 2567
    },
    {
      "epoch": 0.36633380884450784,
      "grad_norm": 1.0616198778152466,
      "learning_rate": 9.779396201499418e-06,
      "loss": 0.8023,
      "step": 2568
    },
    {
      "epoch": 0.36647646219686164,
      "grad_norm": 0.7718018889427185,
      "learning_rate": 9.779169939139526e-06,
      "loss": 0.7595,
      "step": 2569
    },
    {
      "epoch": 0.3666191155492154,
      "grad_norm": 0.8942586779594421,
      "learning_rate": 9.77894356342626e-06,
      "loss": 0.7716,
      "step": 2570
    },
    {
      "epoch": 0.3667617689015692,
      "grad_norm": 0.9758250117301941,
      "learning_rate": 9.778717074364985e-06,
      "loss": 0.8131,
      "step": 2571
    },
    {
      "epoch": 0.36690442225392295,
      "grad_norm": 1.077924370765686,
      "learning_rate": 9.778490471961074e-06,
      "loss": 0.7641,
      "step": 2572
    },
    {
      "epoch": 0.36704707560627675,
      "grad_norm": 0.9291079044342041,
      "learning_rate": 9.778263756219902e-06,
      "loss": 0.7663,
      "step": 2573
    },
    {
      "epoch": 0.3671897289586305,
      "grad_norm": 1.079791784286499,
      "learning_rate": 9.778036927146845e-06,
      "loss": 0.8218,
      "step": 2574
    },
    {
      "epoch": 0.3673323823109843,
      "grad_norm": 1.0044819116592407,
      "learning_rate": 9.777809984747287e-06,
      "loss": 0.8751,
      "step": 2575
    },
    {
      "epoch": 0.3674750356633381,
      "grad_norm": 0.775492250919342,
      "learning_rate": 9.777582929026606e-06,
      "loss": 0.7893,
      "step": 2576
    },
    {
      "epoch": 0.36761768901569186,
      "grad_norm": 0.8879029750823975,
      "learning_rate": 9.77735575999019e-06,
      "loss": 0.8414,
      "step": 2577
    },
    {
      "epoch": 0.36776034236804567,
      "grad_norm": 0.9367338418960571,
      "learning_rate": 9.777128477643426e-06,
      "loss": 0.6126,
      "step": 2578
    },
    {
      "epoch": 0.3679029957203994,
      "grad_norm": 0.752149224281311,
      "learning_rate": 9.776901081991706e-06,
      "loss": 0.7874,
      "step": 2579
    },
    {
      "epoch": 0.3680456490727532,
      "grad_norm": 0.821786642074585,
      "learning_rate": 9.77667357304042e-06,
      "loss": 0.8036,
      "step": 2580
    },
    {
      "epoch": 0.368188302425107,
      "grad_norm": 0.8069832921028137,
      "learning_rate": 9.776445950794968e-06,
      "loss": 0.7852,
      "step": 2581
    },
    {
      "epoch": 0.3683309557774608,
      "grad_norm": 0.7493656873703003,
      "learning_rate": 9.776218215260749e-06,
      "loss": 0.7966,
      "step": 2582
    },
    {
      "epoch": 0.3684736091298145,
      "grad_norm": 0.9002495408058167,
      "learning_rate": 9.775990366443162e-06,
      "loss": 0.8047,
      "step": 2583
    },
    {
      "epoch": 0.36861626248216833,
      "grad_norm": 0.8920180201530457,
      "learning_rate": 9.775762404347611e-06,
      "loss": 0.815,
      "step": 2584
    },
    {
      "epoch": 0.36875891583452214,
      "grad_norm": 1.1317116022109985,
      "learning_rate": 9.775534328979504e-06,
      "loss": 0.8338,
      "step": 2585
    },
    {
      "epoch": 0.3689015691868759,
      "grad_norm": 0.9028584957122803,
      "learning_rate": 9.775306140344251e-06,
      "loss": 0.8188,
      "step": 2586
    },
    {
      "epoch": 0.3690442225392297,
      "grad_norm": 0.7854993343353271,
      "learning_rate": 9.775077838447261e-06,
      "loss": 0.5499,
      "step": 2587
    },
    {
      "epoch": 0.36918687589158344,
      "grad_norm": 0.7696592807769775,
      "learning_rate": 9.774849423293955e-06,
      "loss": 0.7132,
      "step": 2588
    },
    {
      "epoch": 0.36932952924393725,
      "grad_norm": 0.849781334400177,
      "learning_rate": 9.774620894889747e-06,
      "loss": 0.7801,
      "step": 2589
    },
    {
      "epoch": 0.369472182596291,
      "grad_norm": 1.1491278409957886,
      "learning_rate": 9.774392253240056e-06,
      "loss": 0.7375,
      "step": 2590
    },
    {
      "epoch": 0.3696148359486448,
      "grad_norm": 1.0900774002075195,
      "learning_rate": 9.774163498350307e-06,
      "loss": 0.8474,
      "step": 2591
    },
    {
      "epoch": 0.36975748930099855,
      "grad_norm": 1.0792137384414673,
      "learning_rate": 9.773934630225926e-06,
      "loss": 0.8593,
      "step": 2592
    },
    {
      "epoch": 0.36990014265335236,
      "grad_norm": 0.7018675804138184,
      "learning_rate": 9.773705648872338e-06,
      "loss": 0.817,
      "step": 2593
    },
    {
      "epoch": 0.3700427960057061,
      "grad_norm": 0.7739388942718506,
      "learning_rate": 9.773476554294978e-06,
      "loss": 0.8053,
      "step": 2594
    },
    {
      "epoch": 0.3701854493580599,
      "grad_norm": 0.8324184417724609,
      "learning_rate": 9.77324734649928e-06,
      "loss": 0.8171,
      "step": 2595
    },
    {
      "epoch": 0.3703281027104137,
      "grad_norm": 0.9940540790557861,
      "learning_rate": 9.773018025490674e-06,
      "loss": 0.8009,
      "step": 2596
    },
    {
      "epoch": 0.37047075606276747,
      "grad_norm": 1.1315734386444092,
      "learning_rate": 9.772788591274606e-06,
      "loss": 0.8332,
      "step": 2597
    },
    {
      "epoch": 0.3706134094151213,
      "grad_norm": 1.3557701110839844,
      "learning_rate": 9.772559043856518e-06,
      "loss": 0.7393,
      "step": 2598
    },
    {
      "epoch": 0.370756062767475,
      "grad_norm": 0.7162796258926392,
      "learning_rate": 9.772329383241848e-06,
      "loss": 0.7788,
      "step": 2599
    },
    {
      "epoch": 0.37089871611982883,
      "grad_norm": 1.10598886013031,
      "learning_rate": 9.77209960943605e-06,
      "loss": 0.7347,
      "step": 2600
    },
    {
      "epoch": 0.3710413694721826,
      "grad_norm": 0.6513818502426147,
      "learning_rate": 9.771869722444569e-06,
      "loss": 0.8699,
      "step": 2601
    },
    {
      "epoch": 0.3711840228245364,
      "grad_norm": 1.0559954643249512,
      "learning_rate": 9.77163972227286e-06,
      "loss": 0.7199,
      "step": 2602
    },
    {
      "epoch": 0.37132667617689014,
      "grad_norm": 1.0768969058990479,
      "learning_rate": 9.771409608926377e-06,
      "loss": 0.8428,
      "step": 2603
    },
    {
      "epoch": 0.37146932952924394,
      "grad_norm": 0.8907505869865417,
      "learning_rate": 9.771179382410578e-06,
      "loss": 0.8004,
      "step": 2604
    },
    {
      "epoch": 0.3716119828815977,
      "grad_norm": 0.9419316649436951,
      "learning_rate": 9.770949042730924e-06,
      "loss": 0.7885,
      "step": 2605
    },
    {
      "epoch": 0.3717546362339515,
      "grad_norm": 1.0965865850448608,
      "learning_rate": 9.77071858989288e-06,
      "loss": 0.7895,
      "step": 2606
    },
    {
      "epoch": 0.3718972895863053,
      "grad_norm": 0.9528594613075256,
      "learning_rate": 9.770488023901907e-06,
      "loss": 0.8403,
      "step": 2607
    },
    {
      "epoch": 0.37203994293865905,
      "grad_norm": 0.8624246716499329,
      "learning_rate": 9.77025734476348e-06,
      "loss": 0.9,
      "step": 2608
    },
    {
      "epoch": 0.37218259629101286,
      "grad_norm": 0.8880956172943115,
      "learning_rate": 9.770026552483065e-06,
      "loss": 0.7662,
      "step": 2609
    },
    {
      "epoch": 0.3723252496433666,
      "grad_norm": 0.949665367603302,
      "learning_rate": 9.76979564706614e-06,
      "loss": 0.7712,
      "step": 2610
    },
    {
      "epoch": 0.3724679029957204,
      "grad_norm": 0.8791137933731079,
      "learning_rate": 9.769564628518178e-06,
      "loss": 0.8127,
      "step": 2611
    },
    {
      "epoch": 0.37261055634807416,
      "grad_norm": 1.0559605360031128,
      "learning_rate": 9.76933349684466e-06,
      "loss": 0.8561,
      "step": 2612
    },
    {
      "epoch": 0.37275320970042797,
      "grad_norm": 0.8810771703720093,
      "learning_rate": 9.769102252051069e-06,
      "loss": 0.8494,
      "step": 2613
    },
    {
      "epoch": 0.3728958630527817,
      "grad_norm": 0.7815802693367004,
      "learning_rate": 9.768870894142887e-06,
      "loss": 0.7677,
      "step": 2614
    },
    {
      "epoch": 0.3730385164051355,
      "grad_norm": 0.8313312530517578,
      "learning_rate": 9.768639423125605e-06,
      "loss": 0.7337,
      "step": 2615
    },
    {
      "epoch": 0.37318116975748933,
      "grad_norm": 0.9385679960250854,
      "learning_rate": 9.768407839004712e-06,
      "loss": 0.7887,
      "step": 2616
    },
    {
      "epoch": 0.3733238231098431,
      "grad_norm": 1.0283102989196777,
      "learning_rate": 9.768176141785699e-06,
      "loss": 0.7511,
      "step": 2617
    },
    {
      "epoch": 0.3734664764621969,
      "grad_norm": 0.753254234790802,
      "learning_rate": 9.767944331474062e-06,
      "loss": 0.7493,
      "step": 2618
    },
    {
      "epoch": 0.37360912981455063,
      "grad_norm": 1.1009119749069214,
      "learning_rate": 9.7677124080753e-06,
      "loss": 0.5995,
      "step": 2619
    },
    {
      "epoch": 0.37375178316690444,
      "grad_norm": 0.7960112690925598,
      "learning_rate": 9.767480371594914e-06,
      "loss": 0.8729,
      "step": 2620
    },
    {
      "epoch": 0.3738944365192582,
      "grad_norm": 0.9416414499282837,
      "learning_rate": 9.767248222038406e-06,
      "loss": 0.8489,
      "step": 2621
    },
    {
      "epoch": 0.374037089871612,
      "grad_norm": 0.7720905542373657,
      "learning_rate": 9.767015959411283e-06,
      "loss": 0.8232,
      "step": 2622
    },
    {
      "epoch": 0.37417974322396574,
      "grad_norm": 0.8114632964134216,
      "learning_rate": 9.766783583719054e-06,
      "loss": 0.7892,
      "step": 2623
    },
    {
      "epoch": 0.37432239657631955,
      "grad_norm": 1.4543877840042114,
      "learning_rate": 9.76655109496723e-06,
      "loss": 0.8484,
      "step": 2624
    },
    {
      "epoch": 0.3744650499286733,
      "grad_norm": 0.9627675414085388,
      "learning_rate": 9.766318493161326e-06,
      "loss": 0.7684,
      "step": 2625
    },
    {
      "epoch": 0.3746077032810271,
      "grad_norm": 0.6586639285087585,
      "learning_rate": 9.766085778306859e-06,
      "loss": 0.7958,
      "step": 2626
    },
    {
      "epoch": 0.3747503566333809,
      "grad_norm": 1.0314810276031494,
      "learning_rate": 9.765852950409349e-06,
      "loss": 0.6011,
      "step": 2627
    },
    {
      "epoch": 0.37489300998573466,
      "grad_norm": 0.8590237498283386,
      "learning_rate": 9.765620009474317e-06,
      "loss": 0.8111,
      "step": 2628
    },
    {
      "epoch": 0.37503566333808847,
      "grad_norm": 0.8130232095718384,
      "learning_rate": 9.765386955507288e-06,
      "loss": 0.5973,
      "step": 2629
    },
    {
      "epoch": 0.3751783166904422,
      "grad_norm": 0.921552300453186,
      "learning_rate": 9.76515378851379e-06,
      "loss": 0.7221,
      "step": 2630
    },
    {
      "epoch": 0.375320970042796,
      "grad_norm": 0.9046935439109802,
      "learning_rate": 9.764920508499352e-06,
      "loss": 0.7321,
      "step": 2631
    },
    {
      "epoch": 0.37546362339514977,
      "grad_norm": 1.291777491569519,
      "learning_rate": 9.76468711546951e-06,
      "loss": 0.8003,
      "step": 2632
    },
    {
      "epoch": 0.3756062767475036,
      "grad_norm": 0.8085049390792847,
      "learning_rate": 9.764453609429798e-06,
      "loss": 0.903,
      "step": 2633
    },
    {
      "epoch": 0.3757489300998573,
      "grad_norm": 0.9415332078933716,
      "learning_rate": 9.764219990385753e-06,
      "loss": 0.758,
      "step": 2634
    },
    {
      "epoch": 0.37589158345221113,
      "grad_norm": 1.023079752922058,
      "learning_rate": 9.763986258342918e-06,
      "loss": 0.8447,
      "step": 2635
    },
    {
      "epoch": 0.3760342368045649,
      "grad_norm": 1.039222240447998,
      "learning_rate": 9.763752413306836e-06,
      "loss": 0.7931,
      "step": 2636
    },
    {
      "epoch": 0.3761768901569187,
      "grad_norm": 1.2192070484161377,
      "learning_rate": 9.763518455283055e-06,
      "loss": 0.8556,
      "step": 2637
    },
    {
      "epoch": 0.3763195435092725,
      "grad_norm": 0.8125724196434021,
      "learning_rate": 9.76328438427712e-06,
      "loss": 0.7761,
      "step": 2638
    },
    {
      "epoch": 0.37646219686162624,
      "grad_norm": 0.7936736345291138,
      "learning_rate": 9.763050200294586e-06,
      "loss": 0.7777,
      "step": 2639
    },
    {
      "epoch": 0.37660485021398005,
      "grad_norm": 0.6694625020027161,
      "learning_rate": 9.762815903341006e-06,
      "loss": 0.8426,
      "step": 2640
    },
    {
      "epoch": 0.3767475035663338,
      "grad_norm": 0.9329331517219543,
      "learning_rate": 9.762581493421938e-06,
      "loss": 0.8454,
      "step": 2641
    },
    {
      "epoch": 0.3768901569186876,
      "grad_norm": 0.8289737105369568,
      "learning_rate": 9.762346970542943e-06,
      "loss": 0.7621,
      "step": 2642
    },
    {
      "epoch": 0.37703281027104135,
      "grad_norm": 1.041685938835144,
      "learning_rate": 9.76211233470958e-06,
      "loss": 0.8076,
      "step": 2643
    },
    {
      "epoch": 0.37717546362339516,
      "grad_norm": 0.8847928643226624,
      "learning_rate": 9.761877585927419e-06,
      "loss": 0.8006,
      "step": 2644
    },
    {
      "epoch": 0.3773181169757489,
      "grad_norm": 0.7176461219787598,
      "learning_rate": 9.761642724202023e-06,
      "loss": 0.8136,
      "step": 2645
    },
    {
      "epoch": 0.3774607703281027,
      "grad_norm": 0.8413593769073486,
      "learning_rate": 9.761407749538964e-06,
      "loss": 0.7785,
      "step": 2646
    },
    {
      "epoch": 0.3776034236804565,
      "grad_norm": 0.7981699705123901,
      "learning_rate": 9.761172661943816e-06,
      "loss": 0.5767,
      "step": 2647
    },
    {
      "epoch": 0.37774607703281027,
      "grad_norm": 0.9135536551475525,
      "learning_rate": 9.760937461422155e-06,
      "loss": 0.5791,
      "step": 2648
    },
    {
      "epoch": 0.3778887303851641,
      "grad_norm": 0.8779171705245972,
      "learning_rate": 9.76070214797956e-06,
      "loss": 0.7945,
      "step": 2649
    },
    {
      "epoch": 0.3780313837375178,
      "grad_norm": 0.8615351319313049,
      "learning_rate": 9.76046672162161e-06,
      "loss": 0.783,
      "step": 2650
    },
    {
      "epoch": 0.37817403708987163,
      "grad_norm": 0.7770572304725647,
      "learning_rate": 9.76023118235389e-06,
      "loss": 0.7985,
      "step": 2651
    },
    {
      "epoch": 0.3783166904422254,
      "grad_norm": 0.7578920722007751,
      "learning_rate": 9.759995530181987e-06,
      "loss": 0.823,
      "step": 2652
    },
    {
      "epoch": 0.3784593437945792,
      "grad_norm": 0.8214627504348755,
      "learning_rate": 9.75975976511149e-06,
      "loss": 0.8105,
      "step": 2653
    },
    {
      "epoch": 0.37860199714693293,
      "grad_norm": 0.8335827589035034,
      "learning_rate": 9.759523887147992e-06,
      "loss": 0.7065,
      "step": 2654
    },
    {
      "epoch": 0.37874465049928674,
      "grad_norm": 0.9503043293952942,
      "learning_rate": 9.759287896297084e-06,
      "loss": 0.7974,
      "step": 2655
    },
    {
      "epoch": 0.3788873038516405,
      "grad_norm": 0.8033483028411865,
      "learning_rate": 9.759051792564369e-06,
      "loss": 0.8391,
      "step": 2656
    },
    {
      "epoch": 0.3790299572039943,
      "grad_norm": 1.2023918628692627,
      "learning_rate": 9.758815575955441e-06,
      "loss": 0.7311,
      "step": 2657
    },
    {
      "epoch": 0.3791726105563481,
      "grad_norm": 0.8341216444969177,
      "learning_rate": 9.758579246475906e-06,
      "loss": 0.7213,
      "step": 2658
    },
    {
      "epoch": 0.37931526390870185,
      "grad_norm": 0.8582043051719666,
      "learning_rate": 9.758342804131368e-06,
      "loss": 0.7988,
      "step": 2659
    },
    {
      "epoch": 0.37945791726105566,
      "grad_norm": 0.7988712787628174,
      "learning_rate": 9.758106248927439e-06,
      "loss": 0.8425,
      "step": 2660
    },
    {
      "epoch": 0.3796005706134094,
      "grad_norm": 0.7101856470108032,
      "learning_rate": 9.757869580869724e-06,
      "loss": 0.8193,
      "step": 2661
    },
    {
      "epoch": 0.3797432239657632,
      "grad_norm": 0.9709107875823975,
      "learning_rate": 9.757632799963838e-06,
      "loss": 0.7668,
      "step": 2662
    },
    {
      "epoch": 0.37988587731811696,
      "grad_norm": 0.7359352707862854,
      "learning_rate": 9.757395906215399e-06,
      "loss": 0.7309,
      "step": 2663
    },
    {
      "epoch": 0.38002853067047077,
      "grad_norm": 0.9795485138893127,
      "learning_rate": 9.757158899630024e-06,
      "loss": 0.745,
      "step": 2664
    },
    {
      "epoch": 0.3801711840228245,
      "grad_norm": 0.9430841207504272,
      "learning_rate": 9.756921780213334e-06,
      "loss": 0.7417,
      "step": 2665
    },
    {
      "epoch": 0.3803138373751783,
      "grad_norm": 0.9501784443855286,
      "learning_rate": 9.756684547970954e-06,
      "loss": 0.847,
      "step": 2666
    },
    {
      "epoch": 0.38045649072753207,
      "grad_norm": 0.9335957765579224,
      "learning_rate": 9.756447202908512e-06,
      "loss": 0.7392,
      "step": 2667
    },
    {
      "epoch": 0.3805991440798859,
      "grad_norm": 1.1261742115020752,
      "learning_rate": 9.756209745031635e-06,
      "loss": 0.7525,
      "step": 2668
    },
    {
      "epoch": 0.3807417974322397,
      "grad_norm": 0.8628441095352173,
      "learning_rate": 9.755972174345955e-06,
      "loss": 0.7191,
      "step": 2669
    },
    {
      "epoch": 0.38088445078459343,
      "grad_norm": 0.9380318522453308,
      "learning_rate": 9.75573449085711e-06,
      "loss": 0.7961,
      "step": 2670
    },
    {
      "epoch": 0.38102710413694724,
      "grad_norm": 0.9635017514228821,
      "learning_rate": 9.755496694570734e-06,
      "loss": 0.7904,
      "step": 2671
    },
    {
      "epoch": 0.381169757489301,
      "grad_norm": 0.7903203368186951,
      "learning_rate": 9.755258785492468e-06,
      "loss": 0.8209,
      "step": 2672
    },
    {
      "epoch": 0.3813124108416548,
      "grad_norm": 0.7948763966560364,
      "learning_rate": 9.755020763627956e-06,
      "loss": 0.7637,
      "step": 2673
    },
    {
      "epoch": 0.38145506419400854,
      "grad_norm": 0.7615258097648621,
      "learning_rate": 9.754782628982842e-06,
      "loss": 0.8453,
      "step": 2674
    },
    {
      "epoch": 0.38159771754636235,
      "grad_norm": 0.7442870736122131,
      "learning_rate": 9.754544381562774e-06,
      "loss": 0.7853,
      "step": 2675
    },
    {
      "epoch": 0.3817403708987161,
      "grad_norm": 0.9709129333496094,
      "learning_rate": 9.754306021373405e-06,
      "loss": 0.7171,
      "step": 2676
    },
    {
      "epoch": 0.3818830242510699,
      "grad_norm": 0.9641637802124023,
      "learning_rate": 9.754067548420386e-06,
      "loss": 0.7876,
      "step": 2677
    },
    {
      "epoch": 0.3820256776034237,
      "grad_norm": 0.933236837387085,
      "learning_rate": 9.753828962709374e-06,
      "loss": 0.7901,
      "step": 2678
    },
    {
      "epoch": 0.38216833095577746,
      "grad_norm": 1.2068079710006714,
      "learning_rate": 9.753590264246028e-06,
      "loss": 0.6204,
      "step": 2679
    },
    {
      "epoch": 0.38231098430813126,
      "grad_norm": 1.1676217317581177,
      "learning_rate": 9.75335145303601e-06,
      "loss": 0.7691,
      "step": 2680
    },
    {
      "epoch": 0.382453637660485,
      "grad_norm": 1.1245152950286865,
      "learning_rate": 9.753112529084983e-06,
      "loss": 0.7144,
      "step": 2681
    },
    {
      "epoch": 0.3825962910128388,
      "grad_norm": 0.9518791437149048,
      "learning_rate": 9.752873492398615e-06,
      "loss": 0.7337,
      "step": 2682
    },
    {
      "epoch": 0.38273894436519257,
      "grad_norm": 0.7627285122871399,
      "learning_rate": 9.752634342982575e-06,
      "loss": 0.7944,
      "step": 2683
    },
    {
      "epoch": 0.3828815977175464,
      "grad_norm": 0.8641947507858276,
      "learning_rate": 9.752395080842533e-06,
      "loss": 0.808,
      "step": 2684
    },
    {
      "epoch": 0.3830242510699001,
      "grad_norm": 0.9172922372817993,
      "learning_rate": 9.75215570598417e-06,
      "loss": 0.8174,
      "step": 2685
    },
    {
      "epoch": 0.38316690442225393,
      "grad_norm": 0.8682686686515808,
      "learning_rate": 9.751916218413157e-06,
      "loss": 0.8089,
      "step": 2686
    },
    {
      "epoch": 0.3833095577746077,
      "grad_norm": 0.7675897479057312,
      "learning_rate": 9.751676618135178e-06,
      "loss": 0.849,
      "step": 2687
    },
    {
      "epoch": 0.3834522111269615,
      "grad_norm": 1.2480727434158325,
      "learning_rate": 9.751436905155914e-06,
      "loss": 0.7802,
      "step": 2688
    },
    {
      "epoch": 0.3835948644793153,
      "grad_norm": 0.8453992605209351,
      "learning_rate": 9.751197079481051e-06,
      "loss": 0.7772,
      "step": 2689
    },
    {
      "epoch": 0.38373751783166904,
      "grad_norm": 0.7082547545433044,
      "learning_rate": 9.750957141116278e-06,
      "loss": 0.8644,
      "step": 2690
    },
    {
      "epoch": 0.38388017118402284,
      "grad_norm": 0.8735644817352295,
      "learning_rate": 9.750717090067286e-06,
      "loss": 0.7443,
      "step": 2691
    },
    {
      "epoch": 0.3840228245363766,
      "grad_norm": 1.0152959823608398,
      "learning_rate": 9.750476926339767e-06,
      "loss": 0.7828,
      "step": 2692
    },
    {
      "epoch": 0.3841654778887304,
      "grad_norm": 0.832862377166748,
      "learning_rate": 9.750236649939418e-06,
      "loss": 0.8786,
      "step": 2693
    },
    {
      "epoch": 0.38430813124108415,
      "grad_norm": 0.797649085521698,
      "learning_rate": 9.749996260871938e-06,
      "loss": 0.8077,
      "step": 2694
    },
    {
      "epoch": 0.38445078459343796,
      "grad_norm": 1.1882747411727905,
      "learning_rate": 9.74975575914303e-06,
      "loss": 0.7217,
      "step": 2695
    },
    {
      "epoch": 0.3845934379457917,
      "grad_norm": 1.260117769241333,
      "learning_rate": 9.749515144758396e-06,
      "loss": 0.8008,
      "step": 2696
    },
    {
      "epoch": 0.3847360912981455,
      "grad_norm": 0.7837929129600525,
      "learning_rate": 9.749274417723745e-06,
      "loss": 0.8494,
      "step": 2697
    },
    {
      "epoch": 0.38487874465049926,
      "grad_norm": 0.8835393190383911,
      "learning_rate": 9.749033578044785e-06,
      "loss": 0.8154,
      "step": 2698
    },
    {
      "epoch": 0.38502139800285307,
      "grad_norm": 1.013067364692688,
      "learning_rate": 9.74879262572723e-06,
      "loss": 0.7286,
      "step": 2699
    },
    {
      "epoch": 0.38516405135520687,
      "grad_norm": 1.0670238733291626,
      "learning_rate": 9.748551560776792e-06,
      "loss": 0.8268,
      "step": 2700
    },
    {
      "epoch": 0.3853067047075606,
      "grad_norm": 0.8018626570701599,
      "learning_rate": 9.748310383199191e-06,
      "loss": 0.7929,
      "step": 2701
    },
    {
      "epoch": 0.3854493580599144,
      "grad_norm": 0.9058294892311096,
      "learning_rate": 9.748069093000148e-06,
      "loss": 0.8724,
      "step": 2702
    },
    {
      "epoch": 0.3855920114122682,
      "grad_norm": 0.7368719577789307,
      "learning_rate": 9.747827690185384e-06,
      "loss": 0.8173,
      "step": 2703
    },
    {
      "epoch": 0.385734664764622,
      "grad_norm": 0.90807044506073,
      "learning_rate": 9.747586174760627e-06,
      "loss": 0.7342,
      "step": 2704
    },
    {
      "epoch": 0.38587731811697573,
      "grad_norm": 0.9778223037719727,
      "learning_rate": 9.747344546731601e-06,
      "loss": 0.7719,
      "step": 2705
    },
    {
      "epoch": 0.38601997146932954,
      "grad_norm": 0.815638542175293,
      "learning_rate": 9.747102806104042e-06,
      "loss": 0.7554,
      "step": 2706
    },
    {
      "epoch": 0.3861626248216833,
      "grad_norm": 1.021000862121582,
      "learning_rate": 9.746860952883679e-06,
      "loss": 0.7847,
      "step": 2707
    },
    {
      "epoch": 0.3863052781740371,
      "grad_norm": 1.0167509317398071,
      "learning_rate": 9.746618987076252e-06,
      "loss": 0.7935,
      "step": 2708
    },
    {
      "epoch": 0.3864479315263909,
      "grad_norm": 0.9476686716079712,
      "learning_rate": 9.746376908687498e-06,
      "loss": 0.8703,
      "step": 2709
    },
    {
      "epoch": 0.38659058487874465,
      "grad_norm": 0.7579607963562012,
      "learning_rate": 9.74613471772316e-06,
      "loss": 0.828,
      "step": 2710
    },
    {
      "epoch": 0.38673323823109845,
      "grad_norm": 1.3265131711959839,
      "learning_rate": 9.745892414188981e-06,
      "loss": 0.7895,
      "step": 2711
    },
    {
      "epoch": 0.3868758915834522,
      "grad_norm": 0.8370159864425659,
      "learning_rate": 9.745649998090709e-06,
      "loss": 0.8543,
      "step": 2712
    },
    {
      "epoch": 0.387018544935806,
      "grad_norm": 0.772868812084198,
      "learning_rate": 9.745407469434095e-06,
      "loss": 0.8196,
      "step": 2713
    },
    {
      "epoch": 0.38716119828815976,
      "grad_norm": 0.8357599973678589,
      "learning_rate": 9.745164828224888e-06,
      "loss": 0.7744,
      "step": 2714
    },
    {
      "epoch": 0.38730385164051356,
      "grad_norm": 0.7198459506034851,
      "learning_rate": 9.744922074468845e-06,
      "loss": 0.8051,
      "step": 2715
    },
    {
      "epoch": 0.3874465049928673,
      "grad_norm": 0.9756195545196533,
      "learning_rate": 9.744679208171724e-06,
      "loss": 0.6199,
      "step": 2716
    },
    {
      "epoch": 0.3875891583452211,
      "grad_norm": 0.8039839863777161,
      "learning_rate": 9.744436229339284e-06,
      "loss": 0.7723,
      "step": 2717
    },
    {
      "epoch": 0.38773181169757487,
      "grad_norm": 0.776809573173523,
      "learning_rate": 9.74419313797729e-06,
      "loss": 0.8159,
      "step": 2718
    },
    {
      "epoch": 0.3878744650499287,
      "grad_norm": 0.7821508049964905,
      "learning_rate": 9.743949934091506e-06,
      "loss": 0.8218,
      "step": 2719
    },
    {
      "epoch": 0.3880171184022825,
      "grad_norm": 0.6585720181465149,
      "learning_rate": 9.7437066176877e-06,
      "loss": 0.7731,
      "step": 2720
    },
    {
      "epoch": 0.38815977175463623,
      "grad_norm": 1.144324541091919,
      "learning_rate": 9.743463188771644e-06,
      "loss": 0.8415,
      "step": 2721
    },
    {
      "epoch": 0.38830242510699003,
      "grad_norm": 1.0549962520599365,
      "learning_rate": 9.743219647349111e-06,
      "loss": 0.7864,
      "step": 2722
    },
    {
      "epoch": 0.3884450784593438,
      "grad_norm": 0.744493842124939,
      "learning_rate": 9.74297599342588e-06,
      "loss": 0.7736,
      "step": 2723
    },
    {
      "epoch": 0.3885877318116976,
      "grad_norm": 1.04491126537323,
      "learning_rate": 9.742732227007727e-06,
      "loss": 0.7707,
      "step": 2724
    },
    {
      "epoch": 0.38873038516405134,
      "grad_norm": 0.9046798348426819,
      "learning_rate": 9.742488348100435e-06,
      "loss": 0.7321,
      "step": 2725
    },
    {
      "epoch": 0.38887303851640515,
      "grad_norm": 0.7387887835502625,
      "learning_rate": 9.742244356709787e-06,
      "loss": 0.7697,
      "step": 2726
    },
    {
      "epoch": 0.3890156918687589,
      "grad_norm": 0.9351281523704529,
      "learning_rate": 9.742000252841572e-06,
      "loss": 0.8096,
      "step": 2727
    },
    {
      "epoch": 0.3891583452211127,
      "grad_norm": 0.7774584889411926,
      "learning_rate": 9.741756036501579e-06,
      "loss": 0.8076,
      "step": 2728
    },
    {
      "epoch": 0.38930099857346645,
      "grad_norm": 0.6976808905601501,
      "learning_rate": 9.7415117076956e-06,
      "loss": 0.7464,
      "step": 2729
    },
    {
      "epoch": 0.38944365192582026,
      "grad_norm": 0.7944491505622864,
      "learning_rate": 9.74126726642943e-06,
      "loss": 0.8093,
      "step": 2730
    },
    {
      "epoch": 0.38958630527817406,
      "grad_norm": 0.9049003720283508,
      "learning_rate": 9.741022712708867e-06,
      "loss": 0.7647,
      "step": 2731
    },
    {
      "epoch": 0.3897289586305278,
      "grad_norm": 0.7585650086402893,
      "learning_rate": 9.740778046539712e-06,
      "loss": 0.7726,
      "step": 2732
    },
    {
      "epoch": 0.3898716119828816,
      "grad_norm": 0.8783090710639954,
      "learning_rate": 9.740533267927766e-06,
      "loss": 0.7521,
      "step": 2733
    },
    {
      "epoch": 0.39001426533523537,
      "grad_norm": 0.8248961567878723,
      "learning_rate": 9.740288376878837e-06,
      "loss": 0.7884,
      "step": 2734
    },
    {
      "epoch": 0.39015691868758917,
      "grad_norm": 0.8012792468070984,
      "learning_rate": 9.740043373398733e-06,
      "loss": 0.6581,
      "step": 2735
    },
    {
      "epoch": 0.3902995720399429,
      "grad_norm": 0.9968196153640747,
      "learning_rate": 9.739798257493263e-06,
      "loss": 0.7685,
      "step": 2736
    },
    {
      "epoch": 0.3904422253922967,
      "grad_norm": 0.7691065073013306,
      "learning_rate": 9.739553029168242e-06,
      "loss": 0.6316,
      "step": 2737
    },
    {
      "epoch": 0.3905848787446505,
      "grad_norm": 0.9571718573570251,
      "learning_rate": 9.73930768842949e-06,
      "loss": 0.7773,
      "step": 2738
    },
    {
      "epoch": 0.3907275320970043,
      "grad_norm": 0.7709513306617737,
      "learning_rate": 9.73906223528282e-06,
      "loss": 0.8136,
      "step": 2739
    },
    {
      "epoch": 0.3908701854493581,
      "grad_norm": 0.6877098083496094,
      "learning_rate": 9.738816669734055e-06,
      "loss": 0.834,
      "step": 2740
    },
    {
      "epoch": 0.39101283880171184,
      "grad_norm": 0.9988148212432861,
      "learning_rate": 9.738570991789023e-06,
      "loss": 0.8033,
      "step": 2741
    },
    {
      "epoch": 0.39115549215406564,
      "grad_norm": 0.8886873722076416,
      "learning_rate": 9.738325201453547e-06,
      "loss": 0.6928,
      "step": 2742
    },
    {
      "epoch": 0.3912981455064194,
      "grad_norm": 0.84345543384552,
      "learning_rate": 9.73807929873346e-06,
      "loss": 0.7314,
      "step": 2743
    },
    {
      "epoch": 0.3914407988587732,
      "grad_norm": 1.0821406841278076,
      "learning_rate": 9.737833283634592e-06,
      "loss": 0.7557,
      "step": 2744
    },
    {
      "epoch": 0.39158345221112695,
      "grad_norm": 0.9236408472061157,
      "learning_rate": 9.73758715616278e-06,
      "loss": 0.8094,
      "step": 2745
    },
    {
      "epoch": 0.39172610556348075,
      "grad_norm": 0.9534152150154114,
      "learning_rate": 9.73734091632386e-06,
      "loss": 0.8072,
      "step": 2746
    },
    {
      "epoch": 0.3918687589158345,
      "grad_norm": 0.7680412530899048,
      "learning_rate": 9.737094564123672e-06,
      "loss": 0.7875,
      "step": 2747
    },
    {
      "epoch": 0.3920114122681883,
      "grad_norm": 0.999618411064148,
      "learning_rate": 9.736848099568062e-06,
      "loss": 0.8146,
      "step": 2748
    },
    {
      "epoch": 0.39215406562054206,
      "grad_norm": 0.8808778524398804,
      "learning_rate": 9.736601522662871e-06,
      "loss": 0.819,
      "step": 2749
    },
    {
      "epoch": 0.39229671897289586,
      "grad_norm": 1.0400785207748413,
      "learning_rate": 9.736354833413953e-06,
      "loss": 0.8888,
      "step": 2750
    },
    {
      "epoch": 0.39243937232524967,
      "grad_norm": 0.7799466252326965,
      "learning_rate": 9.736108031827155e-06,
      "loss": 0.777,
      "step": 2751
    },
    {
      "epoch": 0.3925820256776034,
      "grad_norm": 0.8672760128974915,
      "learning_rate": 9.735861117908329e-06,
      "loss": 0.7409,
      "step": 2752
    },
    {
      "epoch": 0.3927246790299572,
      "grad_norm": 0.9319669604301453,
      "learning_rate": 9.735614091663336e-06,
      "loss": 0.8232,
      "step": 2753
    },
    {
      "epoch": 0.392867332382311,
      "grad_norm": 0.6662679314613342,
      "learning_rate": 9.735366953098035e-06,
      "loss": 0.6303,
      "step": 2754
    },
    {
      "epoch": 0.3930099857346648,
      "grad_norm": 1.0762574672698975,
      "learning_rate": 9.735119702218285e-06,
      "loss": 0.8742,
      "step": 2755
    },
    {
      "epoch": 0.39315263908701853,
      "grad_norm": 1.2987182140350342,
      "learning_rate": 9.73487233902995e-06,
      "loss": 0.7481,
      "step": 2756
    },
    {
      "epoch": 0.39329529243937233,
      "grad_norm": 1.0922690629959106,
      "learning_rate": 9.7346248635389e-06,
      "loss": 0.7634,
      "step": 2757
    },
    {
      "epoch": 0.3934379457917261,
      "grad_norm": 0.8129608035087585,
      "learning_rate": 9.734377275751002e-06,
      "loss": 0.7855,
      "step": 2758
    },
    {
      "epoch": 0.3935805991440799,
      "grad_norm": 0.7861506342887878,
      "learning_rate": 9.734129575672127e-06,
      "loss": 0.8174,
      "step": 2759
    },
    {
      "epoch": 0.39372325249643364,
      "grad_norm": 1.0716252326965332,
      "learning_rate": 9.733881763308155e-06,
      "loss": 0.7856,
      "step": 2760
    },
    {
      "epoch": 0.39386590584878745,
      "grad_norm": 0.7261438965797424,
      "learning_rate": 9.73363383866496e-06,
      "loss": 0.7678,
      "step": 2761
    },
    {
      "epoch": 0.39400855920114125,
      "grad_norm": 0.9121649265289307,
      "learning_rate": 9.733385801748423e-06,
      "loss": 0.7824,
      "step": 2762
    },
    {
      "epoch": 0.394151212553495,
      "grad_norm": 0.8819028735160828,
      "learning_rate": 9.733137652564428e-06,
      "loss": 0.7833,
      "step": 2763
    },
    {
      "epoch": 0.3942938659058488,
      "grad_norm": 0.7690829038619995,
      "learning_rate": 9.732889391118859e-06,
      "loss": 0.8084,
      "step": 2764
    },
    {
      "epoch": 0.39443651925820256,
      "grad_norm": 0.8435876369476318,
      "learning_rate": 9.732641017417606e-06,
      "loss": 0.7105,
      "step": 2765
    },
    {
      "epoch": 0.39457917261055636,
      "grad_norm": 0.7878632545471191,
      "learning_rate": 9.732392531466559e-06,
      "loss": 0.7949,
      "step": 2766
    },
    {
      "epoch": 0.3947218259629101,
      "grad_norm": 0.9608743190765381,
      "learning_rate": 9.732143933271612e-06,
      "loss": 0.7771,
      "step": 2767
    },
    {
      "epoch": 0.3948644793152639,
      "grad_norm": 0.8073573112487793,
      "learning_rate": 9.731895222838661e-06,
      "loss": 0.7732,
      "step": 2768
    },
    {
      "epoch": 0.39500713266761767,
      "grad_norm": 0.9152679443359375,
      "learning_rate": 9.731646400173605e-06,
      "loss": 0.7618,
      "step": 2769
    },
    {
      "epoch": 0.39514978601997147,
      "grad_norm": 0.7204840779304504,
      "learning_rate": 9.731397465282343e-06,
      "loss": 0.8244,
      "step": 2770
    },
    {
      "epoch": 0.3952924393723253,
      "grad_norm": 0.8135253190994263,
      "learning_rate": 9.731148418170786e-06,
      "loss": 0.7748,
      "step": 2771
    },
    {
      "epoch": 0.395435092724679,
      "grad_norm": 0.9012436866760254,
      "learning_rate": 9.730899258844835e-06,
      "loss": 0.8807,
      "step": 2772
    },
    {
      "epoch": 0.39557774607703283,
      "grad_norm": 1.1430065631866455,
      "learning_rate": 9.730649987310403e-06,
      "loss": 0.847,
      "step": 2773
    },
    {
      "epoch": 0.3957203994293866,
      "grad_norm": 0.9346035122871399,
      "learning_rate": 9.7304006035734e-06,
      "loss": 0.7522,
      "step": 2774
    },
    {
      "epoch": 0.3958630527817404,
      "grad_norm": 0.8465891480445862,
      "learning_rate": 9.73015110763974e-06,
      "loss": 0.7143,
      "step": 2775
    },
    {
      "epoch": 0.39600570613409414,
      "grad_norm": 0.7051974534988403,
      "learning_rate": 9.729901499515345e-06,
      "loss": 0.728,
      "step": 2776
    },
    {
      "epoch": 0.39614835948644794,
      "grad_norm": 0.8334681987762451,
      "learning_rate": 9.729651779206132e-06,
      "loss": 0.7767,
      "step": 2777
    },
    {
      "epoch": 0.3962910128388017,
      "grad_norm": 0.7618001103401184,
      "learning_rate": 9.729401946718025e-06,
      "loss": 0.8218,
      "step": 2778
    },
    {
      "epoch": 0.3964336661911555,
      "grad_norm": 0.8659477233886719,
      "learning_rate": 9.729152002056947e-06,
      "loss": 0.7719,
      "step": 2779
    },
    {
      "epoch": 0.39657631954350925,
      "grad_norm": 1.1242305040359497,
      "learning_rate": 9.72890194522883e-06,
      "loss": 0.7171,
      "step": 2780
    },
    {
      "epoch": 0.39671897289586305,
      "grad_norm": 0.9327376484870911,
      "learning_rate": 9.728651776239604e-06,
      "loss": 0.796,
      "step": 2781
    },
    {
      "epoch": 0.39686162624821686,
      "grad_norm": 0.9061946272850037,
      "learning_rate": 9.7284014950952e-06,
      "loss": 0.7779,
      "step": 2782
    },
    {
      "epoch": 0.3970042796005706,
      "grad_norm": 0.9660446047782898,
      "learning_rate": 9.728151101801558e-06,
      "loss": 0.8504,
      "step": 2783
    },
    {
      "epoch": 0.3971469329529244,
      "grad_norm": 0.721226692199707,
      "learning_rate": 9.727900596364613e-06,
      "loss": 0.7862,
      "step": 2784
    },
    {
      "epoch": 0.39728958630527816,
      "grad_norm": 0.7617622017860413,
      "learning_rate": 9.72764997879031e-06,
      "loss": 0.7985,
      "step": 2785
    },
    {
      "epoch": 0.39743223965763197,
      "grad_norm": 0.9045830368995667,
      "learning_rate": 9.72739924908459e-06,
      "loss": 0.8475,
      "step": 2786
    },
    {
      "epoch": 0.3975748930099857,
      "grad_norm": 0.9054057002067566,
      "learning_rate": 9.727148407253402e-06,
      "loss": 0.7966,
      "step": 2787
    },
    {
      "epoch": 0.3977175463623395,
      "grad_norm": 0.901283860206604,
      "learning_rate": 9.726897453302697e-06,
      "loss": 0.8035,
      "step": 2788
    },
    {
      "epoch": 0.3978601997146933,
      "grad_norm": 1.1419939994812012,
      "learning_rate": 9.726646387238424e-06,
      "loss": 0.7672,
      "step": 2789
    },
    {
      "epoch": 0.3980028530670471,
      "grad_norm": 1.0149915218353271,
      "learning_rate": 9.726395209066539e-06,
      "loss": 0.7542,
      "step": 2790
    },
    {
      "epoch": 0.39814550641940083,
      "grad_norm": 0.8135383725166321,
      "learning_rate": 9.726143918793e-06,
      "loss": 0.8171,
      "step": 2791
    },
    {
      "epoch": 0.39828815977175464,
      "grad_norm": 0.847780704498291,
      "learning_rate": 9.725892516423765e-06,
      "loss": 0.8331,
      "step": 2792
    },
    {
      "epoch": 0.39843081312410844,
      "grad_norm": 0.7311589121818542,
      "learning_rate": 9.7256410019648e-06,
      "loss": 0.7983,
      "step": 2793
    },
    {
      "epoch": 0.3985734664764622,
      "grad_norm": 0.7674535512924194,
      "learning_rate": 9.72538937542207e-06,
      "loss": 0.8564,
      "step": 2794
    },
    {
      "epoch": 0.398716119828816,
      "grad_norm": 0.6326112747192383,
      "learning_rate": 9.72513763680154e-06,
      "loss": 0.8182,
      "step": 2795
    },
    {
      "epoch": 0.39885877318116975,
      "grad_norm": 0.7226607203483582,
      "learning_rate": 9.724885786109186e-06,
      "loss": 0.8457,
      "step": 2796
    },
    {
      "epoch": 0.39900142653352355,
      "grad_norm": 1.111894965171814,
      "learning_rate": 9.724633823350975e-06,
      "loss": 0.6933,
      "step": 2797
    },
    {
      "epoch": 0.3991440798858773,
      "grad_norm": 0.9274622201919556,
      "learning_rate": 9.724381748532888e-06,
      "loss": 0.7499,
      "step": 2798
    },
    {
      "epoch": 0.3992867332382311,
      "grad_norm": 0.9712232351303101,
      "learning_rate": 9.724129561660901e-06,
      "loss": 0.7916,
      "step": 2799
    },
    {
      "epoch": 0.39942938659058486,
      "grad_norm": 0.7862420082092285,
      "learning_rate": 9.723877262741e-06,
      "loss": 0.7479,
      "step": 2800
    },
    {
      "epoch": 0.39957203994293866,
      "grad_norm": 1.0056772232055664,
      "learning_rate": 9.723624851779163e-06,
      "loss": 0.7359,
      "step": 2801
    },
    {
      "epoch": 0.39971469329529247,
      "grad_norm": 0.9222816228866577,
      "learning_rate": 9.72337232878138e-06,
      "loss": 0.7021,
      "step": 2802
    },
    {
      "epoch": 0.3998573466476462,
      "grad_norm": 0.8763768672943115,
      "learning_rate": 9.72311969375364e-06,
      "loss": 0.775,
      "step": 2803
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8317567706108093,
      "learning_rate": 9.722866946701937e-06,
      "loss": 0.7815,
      "step": 2804
    },
    {
      "epoch": 0.40014265335235377,
      "grad_norm": 0.8371714353561401,
      "learning_rate": 9.72261408763226e-06,
      "loss": 0.7607,
      "step": 2805
    },
    {
      "epoch": 0.4002853067047076,
      "grad_norm": 0.9584025144577026,
      "learning_rate": 9.722361116550614e-06,
      "loss": 0.7464,
      "step": 2806
    },
    {
      "epoch": 0.4004279600570613,
      "grad_norm": 0.9503485560417175,
      "learning_rate": 9.722108033462992e-06,
      "loss": 0.8436,
      "step": 2807
    },
    {
      "epoch": 0.40057061340941513,
      "grad_norm": 1.078795075416565,
      "learning_rate": 9.721854838375402e-06,
      "loss": 0.7928,
      "step": 2808
    },
    {
      "epoch": 0.4007132667617689,
      "grad_norm": 1.1382116079330444,
      "learning_rate": 9.721601531293845e-06,
      "loss": 0.7984,
      "step": 2809
    },
    {
      "epoch": 0.4008559201141227,
      "grad_norm": 0.9619671106338501,
      "learning_rate": 9.721348112224331e-06,
      "loss": 0.7672,
      "step": 2810
    },
    {
      "epoch": 0.40099857346647644,
      "grad_norm": 1.0031604766845703,
      "learning_rate": 9.721094581172872e-06,
      "loss": 0.7621,
      "step": 2811
    },
    {
      "epoch": 0.40114122681883024,
      "grad_norm": 0.9520820379257202,
      "learning_rate": 9.720840938145479e-06,
      "loss": 0.7674,
      "step": 2812
    },
    {
      "epoch": 0.40128388017118405,
      "grad_norm": 0.9491026997566223,
      "learning_rate": 9.720587183148168e-06,
      "loss": 0.8088,
      "step": 2813
    },
    {
      "epoch": 0.4014265335235378,
      "grad_norm": 0.7598925828933716,
      "learning_rate": 9.720333316186961e-06,
      "loss": 0.6926,
      "step": 2814
    },
    {
      "epoch": 0.4015691868758916,
      "grad_norm": 0.8415775299072266,
      "learning_rate": 9.720079337267877e-06,
      "loss": 0.8191,
      "step": 2815
    },
    {
      "epoch": 0.40171184022824535,
      "grad_norm": 0.8067095875740051,
      "learning_rate": 9.719825246396939e-06,
      "loss": 0.7903,
      "step": 2816
    },
    {
      "epoch": 0.40185449358059916,
      "grad_norm": 0.6660135388374329,
      "learning_rate": 9.719571043580174e-06,
      "loss": 0.834,
      "step": 2817
    },
    {
      "epoch": 0.4019971469329529,
      "grad_norm": 0.6990180611610413,
      "learning_rate": 9.719316728823612e-06,
      "loss": 0.7858,
      "step": 2818
    },
    {
      "epoch": 0.4021398002853067,
      "grad_norm": 0.8663573265075684,
      "learning_rate": 9.719062302133284e-06,
      "loss": 0.7581,
      "step": 2819
    },
    {
      "epoch": 0.40228245363766046,
      "grad_norm": 1.0680193901062012,
      "learning_rate": 9.718807763515226e-06,
      "loss": 0.7723,
      "step": 2820
    },
    {
      "epoch": 0.40242510699001427,
      "grad_norm": 0.8193547129631042,
      "learning_rate": 9.718553112975473e-06,
      "loss": 0.7844,
      "step": 2821
    },
    {
      "epoch": 0.402567760342368,
      "grad_norm": 1.0157464742660522,
      "learning_rate": 9.718298350520066e-06,
      "loss": 0.8367,
      "step": 2822
    },
    {
      "epoch": 0.4027104136947218,
      "grad_norm": 0.8064296841621399,
      "learning_rate": 9.718043476155048e-06,
      "loss": 0.7501,
      "step": 2823
    },
    {
      "epoch": 0.40285306704707563,
      "grad_norm": 0.8575136065483093,
      "learning_rate": 9.717788489886464e-06,
      "loss": 0.7637,
      "step": 2824
    },
    {
      "epoch": 0.4029957203994294,
      "grad_norm": 0.7274131774902344,
      "learning_rate": 9.717533391720361e-06,
      "loss": 0.8096,
      "step": 2825
    },
    {
      "epoch": 0.4031383737517832,
      "grad_norm": 0.8452036380767822,
      "learning_rate": 9.717278181662791e-06,
      "loss": 0.771,
      "step": 2826
    },
    {
      "epoch": 0.40328102710413694,
      "grad_norm": 0.8290517330169678,
      "learning_rate": 9.717022859719807e-06,
      "loss": 0.5842,
      "step": 2827
    },
    {
      "epoch": 0.40342368045649074,
      "grad_norm": 0.7403361797332764,
      "learning_rate": 9.716767425897463e-06,
      "loss": 0.7776,
      "step": 2828
    },
    {
      "epoch": 0.4035663338088445,
      "grad_norm": 0.9250881671905518,
      "learning_rate": 9.716511880201818e-06,
      "loss": 0.7167,
      "step": 2829
    },
    {
      "epoch": 0.4037089871611983,
      "grad_norm": 0.8735909461975098,
      "learning_rate": 9.716256222638934e-06,
      "loss": 0.8424,
      "step": 2830
    },
    {
      "epoch": 0.40385164051355205,
      "grad_norm": 1.0311628580093384,
      "learning_rate": 9.716000453214873e-06,
      "loss": 0.7285,
      "step": 2831
    },
    {
      "epoch": 0.40399429386590585,
      "grad_norm": 0.7494921088218689,
      "learning_rate": 9.715744571935704e-06,
      "loss": 0.7569,
      "step": 2832
    },
    {
      "epoch": 0.40413694721825966,
      "grad_norm": 0.7012854814529419,
      "learning_rate": 9.715488578807495e-06,
      "loss": 0.8128,
      "step": 2833
    },
    {
      "epoch": 0.4042796005706134,
      "grad_norm": 0.7978854775428772,
      "learning_rate": 9.715232473836316e-06,
      "loss": 0.7329,
      "step": 2834
    },
    {
      "epoch": 0.4044222539229672,
      "grad_norm": 0.7102094292640686,
      "learning_rate": 9.714976257028243e-06,
      "loss": 0.7961,
      "step": 2835
    },
    {
      "epoch": 0.40456490727532096,
      "grad_norm": 1.024623990058899,
      "learning_rate": 9.714719928389353e-06,
      "loss": 0.801,
      "step": 2836
    },
    {
      "epoch": 0.40470756062767477,
      "grad_norm": 0.8503227829933167,
      "learning_rate": 9.714463487925727e-06,
      "loss": 0.722,
      "step": 2837
    },
    {
      "epoch": 0.4048502139800285,
      "grad_norm": 0.7975144386291504,
      "learning_rate": 9.714206935643445e-06,
      "loss": 0.8133,
      "step": 2838
    },
    {
      "epoch": 0.4049928673323823,
      "grad_norm": 1.0284011363983154,
      "learning_rate": 9.71395027154859e-06,
      "loss": 0.7676,
      "step": 2839
    },
    {
      "epoch": 0.4051355206847361,
      "grad_norm": 0.8952131867408752,
      "learning_rate": 9.713693495647256e-06,
      "loss": 0.7489,
      "step": 2840
    },
    {
      "epoch": 0.4052781740370899,
      "grad_norm": 1.1101491451263428,
      "learning_rate": 9.713436607945527e-06,
      "loss": 0.7814,
      "step": 2841
    },
    {
      "epoch": 0.4054208273894436,
      "grad_norm": 0.8903674483299255,
      "learning_rate": 9.713179608449499e-06,
      "loss": 0.7911,
      "step": 2842
    },
    {
      "epoch": 0.40556348074179743,
      "grad_norm": 0.6775222420692444,
      "learning_rate": 9.712922497165267e-06,
      "loss": 0.6458,
      "step": 2843
    },
    {
      "epoch": 0.40570613409415124,
      "grad_norm": 0.799577534198761,
      "learning_rate": 9.71266527409893e-06,
      "loss": 0.8535,
      "step": 2844
    },
    {
      "epoch": 0.405848787446505,
      "grad_norm": 0.7899248600006104,
      "learning_rate": 9.712407939256588e-06,
      "loss": 0.8031,
      "step": 2845
    },
    {
      "epoch": 0.4059914407988588,
      "grad_norm": 0.713304877281189,
      "learning_rate": 9.712150492644345e-06,
      "loss": 0.7477,
      "step": 2846
    },
    {
      "epoch": 0.40613409415121254,
      "grad_norm": 0.9414926767349243,
      "learning_rate": 9.711892934268305e-06,
      "loss": 0.7355,
      "step": 2847
    },
    {
      "epoch": 0.40627674750356635,
      "grad_norm": 0.9517255425453186,
      "learning_rate": 9.71163526413458e-06,
      "loss": 0.8135,
      "step": 2848
    },
    {
      "epoch": 0.4064194008559201,
      "grad_norm": 0.8743422031402588,
      "learning_rate": 9.71137748224928e-06,
      "loss": 0.774,
      "step": 2849
    },
    {
      "epoch": 0.4065620542082739,
      "grad_norm": 0.7215398550033569,
      "learning_rate": 9.711119588618519e-06,
      "loss": 0.7994,
      "step": 2850
    },
    {
      "epoch": 0.40670470756062765,
      "grad_norm": 0.8900471925735474,
      "learning_rate": 9.710861583248414e-06,
      "loss": 0.8477,
      "step": 2851
    },
    {
      "epoch": 0.40684736091298146,
      "grad_norm": 0.9419020414352417,
      "learning_rate": 9.710603466145085e-06,
      "loss": 0.7946,
      "step": 2852
    },
    {
      "epoch": 0.4069900142653352,
      "grad_norm": 0.9247307777404785,
      "learning_rate": 9.710345237314653e-06,
      "loss": 0.8304,
      "step": 2853
    },
    {
      "epoch": 0.407132667617689,
      "grad_norm": 1.071629285812378,
      "learning_rate": 9.710086896763245e-06,
      "loss": 0.7594,
      "step": 2854
    },
    {
      "epoch": 0.4072753209700428,
      "grad_norm": 0.9002863168716431,
      "learning_rate": 9.709828444496985e-06,
      "loss": 0.7731,
      "step": 2855
    },
    {
      "epoch": 0.40741797432239657,
      "grad_norm": 0.7038922309875488,
      "learning_rate": 9.709569880522004e-06,
      "loss": 0.741,
      "step": 2856
    },
    {
      "epoch": 0.4075606276747504,
      "grad_norm": 0.6473375558853149,
      "learning_rate": 9.709311204844436e-06,
      "loss": 0.809,
      "step": 2857
    },
    {
      "epoch": 0.4077032810271041,
      "grad_norm": 0.7838351130485535,
      "learning_rate": 9.709052417470415e-06,
      "loss": 0.7669,
      "step": 2858
    },
    {
      "epoch": 0.40784593437945793,
      "grad_norm": 0.7748342156410217,
      "learning_rate": 9.708793518406082e-06,
      "loss": 0.8423,
      "step": 2859
    },
    {
      "epoch": 0.4079885877318117,
      "grad_norm": 0.7048565745353699,
      "learning_rate": 9.708534507657572e-06,
      "loss": 0.7845,
      "step": 2860
    },
    {
      "epoch": 0.4081312410841655,
      "grad_norm": 0.7788856029510498,
      "learning_rate": 9.708275385231035e-06,
      "loss": 0.7807,
      "step": 2861
    },
    {
      "epoch": 0.40827389443651924,
      "grad_norm": 0.9324522614479065,
      "learning_rate": 9.70801615113261e-06,
      "loss": 0.7685,
      "step": 2862
    },
    {
      "epoch": 0.40841654778887304,
      "grad_norm": 0.9006614089012146,
      "learning_rate": 9.70775680536845e-06,
      "loss": 0.7371,
      "step": 2863
    },
    {
      "epoch": 0.40855920114122685,
      "grad_norm": 1.0853517055511475,
      "learning_rate": 9.707497347944707e-06,
      "loss": 0.7871,
      "step": 2864
    },
    {
      "epoch": 0.4087018544935806,
      "grad_norm": 0.8286964893341064,
      "learning_rate": 9.707237778867532e-06,
      "loss": 0.7706,
      "step": 2865
    },
    {
      "epoch": 0.4088445078459344,
      "grad_norm": 0.9044437408447266,
      "learning_rate": 9.706978098143082e-06,
      "loss": 0.8755,
      "step": 2866
    },
    {
      "epoch": 0.40898716119828815,
      "grad_norm": 0.9929351806640625,
      "learning_rate": 9.706718305777519e-06,
      "loss": 0.6942,
      "step": 2867
    },
    {
      "epoch": 0.40912981455064196,
      "grad_norm": 0.9444536566734314,
      "learning_rate": 9.706458401777e-06,
      "loss": 0.7041,
      "step": 2868
    },
    {
      "epoch": 0.4092724679029957,
      "grad_norm": 0.8999462723731995,
      "learning_rate": 9.706198386147694e-06,
      "loss": 0.8094,
      "step": 2869
    },
    {
      "epoch": 0.4094151212553495,
      "grad_norm": 1.2214356660842896,
      "learning_rate": 9.705938258895765e-06,
      "loss": 0.5336,
      "step": 2870
    },
    {
      "epoch": 0.40955777460770326,
      "grad_norm": 1.0949814319610596,
      "learning_rate": 9.705678020027385e-06,
      "loss": 0.7205,
      "step": 2871
    },
    {
      "epoch": 0.40970042796005707,
      "grad_norm": 1.0050899982452393,
      "learning_rate": 9.705417669548723e-06,
      "loss": 0.7636,
      "step": 2872
    },
    {
      "epoch": 0.4098430813124108,
      "grad_norm": 0.8458533883094788,
      "learning_rate": 9.705157207465959e-06,
      "loss": 0.8918,
      "step": 2873
    },
    {
      "epoch": 0.4099857346647646,
      "grad_norm": 0.9296154379844666,
      "learning_rate": 9.704896633785266e-06,
      "loss": 0.745,
      "step": 2874
    },
    {
      "epoch": 0.41012838801711843,
      "grad_norm": 0.8902323842048645,
      "learning_rate": 9.704635948512828e-06,
      "loss": 0.8048,
      "step": 2875
    },
    {
      "epoch": 0.4102710413694722,
      "grad_norm": 0.9437968134880066,
      "learning_rate": 9.704375151654825e-06,
      "loss": 0.8345,
      "step": 2876
    },
    {
      "epoch": 0.410413694721826,
      "grad_norm": 0.8752991557121277,
      "learning_rate": 9.704114243217444e-06,
      "loss": 0.7575,
      "step": 2877
    },
    {
      "epoch": 0.41055634807417973,
      "grad_norm": 0.9018489122390747,
      "learning_rate": 9.703853223206875e-06,
      "loss": 0.7553,
      "step": 2878
    },
    {
      "epoch": 0.41069900142653354,
      "grad_norm": 0.9929783940315247,
      "learning_rate": 9.703592091629306e-06,
      "loss": 0.8109,
      "step": 2879
    },
    {
      "epoch": 0.4108416547788873,
      "grad_norm": 0.7652303576469421,
      "learning_rate": 9.703330848490933e-06,
      "loss": 0.7772,
      "step": 2880
    },
    {
      "epoch": 0.4109843081312411,
      "grad_norm": 0.8961701393127441,
      "learning_rate": 9.70306949379795e-06,
      "loss": 0.7696,
      "step": 2881
    },
    {
      "epoch": 0.41112696148359484,
      "grad_norm": 0.8824974298477173,
      "learning_rate": 9.702808027556557e-06,
      "loss": 0.7934,
      "step": 2882
    },
    {
      "epoch": 0.41126961483594865,
      "grad_norm": 0.6955432295799255,
      "learning_rate": 9.702546449772955e-06,
      "loss": 0.7195,
      "step": 2883
    },
    {
      "epoch": 0.4114122681883024,
      "grad_norm": 0.7119831442832947,
      "learning_rate": 9.702284760453349e-06,
      "loss": 0.7584,
      "step": 2884
    },
    {
      "epoch": 0.4115549215406562,
      "grad_norm": 0.8682061433792114,
      "learning_rate": 9.702022959603944e-06,
      "loss": 0.8205,
      "step": 2885
    },
    {
      "epoch": 0.41169757489301,
      "grad_norm": 0.9031049609184265,
      "learning_rate": 9.701761047230953e-06,
      "loss": 0.8606,
      "step": 2886
    },
    {
      "epoch": 0.41184022824536376,
      "grad_norm": 0.8140887022018433,
      "learning_rate": 9.701499023340587e-06,
      "loss": 0.7225,
      "step": 2887
    },
    {
      "epoch": 0.41198288159771757,
      "grad_norm": 0.9121723175048828,
      "learning_rate": 9.701236887939056e-06,
      "loss": 0.7841,
      "step": 2888
    },
    {
      "epoch": 0.4121255349500713,
      "grad_norm": 0.8388721942901611,
      "learning_rate": 9.700974641032583e-06,
      "loss": 0.7966,
      "step": 2889
    },
    {
      "epoch": 0.4122681883024251,
      "grad_norm": 0.8396943807601929,
      "learning_rate": 9.700712282627388e-06,
      "loss": 0.7714,
      "step": 2890
    },
    {
      "epoch": 0.41241084165477887,
      "grad_norm": 0.9269974827766418,
      "learning_rate": 9.700449812729689e-06,
      "loss": 0.7563,
      "step": 2891
    },
    {
      "epoch": 0.4125534950071327,
      "grad_norm": 0.7480577826499939,
      "learning_rate": 9.700187231345715e-06,
      "loss": 0.7538,
      "step": 2892
    },
    {
      "epoch": 0.4126961483594864,
      "grad_norm": 1.004860758781433,
      "learning_rate": 9.699924538481691e-06,
      "loss": 0.7932,
      "step": 2893
    },
    {
      "epoch": 0.41283880171184023,
      "grad_norm": 0.9278278350830078,
      "learning_rate": 9.699661734143853e-06,
      "loss": 0.7935,
      "step": 2894
    },
    {
      "epoch": 0.41298145506419404,
      "grad_norm": 0.9548095464706421,
      "learning_rate": 9.699398818338427e-06,
      "loss": 0.7767,
      "step": 2895
    },
    {
      "epoch": 0.4131241084165478,
      "grad_norm": 0.8039991855621338,
      "learning_rate": 9.699135791071656e-06,
      "loss": 0.8061,
      "step": 2896
    },
    {
      "epoch": 0.4132667617689016,
      "grad_norm": 0.9833992123603821,
      "learning_rate": 9.698872652349775e-06,
      "loss": 0.694,
      "step": 2897
    },
    {
      "epoch": 0.41340941512125534,
      "grad_norm": 0.8005412817001343,
      "learning_rate": 9.698609402179024e-06,
      "loss": 0.7802,
      "step": 2898
    },
    {
      "epoch": 0.41355206847360915,
      "grad_norm": 0.7481583952903748,
      "learning_rate": 9.698346040565649e-06,
      "loss": 0.7829,
      "step": 2899
    },
    {
      "epoch": 0.4136947218259629,
      "grad_norm": 1.1110355854034424,
      "learning_rate": 9.698082567515897e-06,
      "loss": 0.7476,
      "step": 2900
    },
    {
      "epoch": 0.4138373751783167,
      "grad_norm": 0.7226589322090149,
      "learning_rate": 9.697818983036013e-06,
      "loss": 0.8259,
      "step": 2901
    },
    {
      "epoch": 0.41398002853067045,
      "grad_norm": 0.8115856647491455,
      "learning_rate": 9.697555287132253e-06,
      "loss": 0.612,
      "step": 2902
    },
    {
      "epoch": 0.41412268188302426,
      "grad_norm": 0.9504525661468506,
      "learning_rate": 9.697291479810873e-06,
      "loss": 0.7367,
      "step": 2903
    },
    {
      "epoch": 0.414265335235378,
      "grad_norm": 0.8218483328819275,
      "learning_rate": 9.697027561078124e-06,
      "loss": 0.7726,
      "step": 2904
    },
    {
      "epoch": 0.4144079885877318,
      "grad_norm": 1.4041001796722412,
      "learning_rate": 9.696763530940269e-06,
      "loss": 0.7757,
      "step": 2905
    },
    {
      "epoch": 0.4145506419400856,
      "grad_norm": 0.8205307722091675,
      "learning_rate": 9.69649938940357e-06,
      "loss": 0.7886,
      "step": 2906
    },
    {
      "epoch": 0.41469329529243937,
      "grad_norm": 0.7615723013877869,
      "learning_rate": 9.696235136474292e-06,
      "loss": 0.7337,
      "step": 2907
    },
    {
      "epoch": 0.4148359486447932,
      "grad_norm": 1.0919002294540405,
      "learning_rate": 9.695970772158704e-06,
      "loss": 0.8087,
      "step": 2908
    },
    {
      "epoch": 0.4149786019971469,
      "grad_norm": 0.8195972442626953,
      "learning_rate": 9.695706296463074e-06,
      "loss": 0.7811,
      "step": 2909
    },
    {
      "epoch": 0.41512125534950073,
      "grad_norm": 0.7052676677703857,
      "learning_rate": 9.695441709393675e-06,
      "loss": 0.8215,
      "step": 2910
    },
    {
      "epoch": 0.4152639087018545,
      "grad_norm": 0.7744655609130859,
      "learning_rate": 9.695177010956785e-06,
      "loss": 0.7793,
      "step": 2911
    },
    {
      "epoch": 0.4154065620542083,
      "grad_norm": 0.7613050937652588,
      "learning_rate": 9.694912201158679e-06,
      "loss": 0.7613,
      "step": 2912
    },
    {
      "epoch": 0.41554921540656203,
      "grad_norm": 0.8056584596633911,
      "learning_rate": 9.69464728000564e-06,
      "loss": 0.8121,
      "step": 2913
    },
    {
      "epoch": 0.41569186875891584,
      "grad_norm": 0.7558019757270813,
      "learning_rate": 9.694382247503951e-06,
      "loss": 0.7907,
      "step": 2914
    },
    {
      "epoch": 0.4158345221112696,
      "grad_norm": 1.5257713794708252,
      "learning_rate": 9.694117103659896e-06,
      "loss": 0.7828,
      "step": 2915
    },
    {
      "epoch": 0.4159771754636234,
      "grad_norm": 1.0058388710021973,
      "learning_rate": 9.693851848479769e-06,
      "loss": 0.7927,
      "step": 2916
    },
    {
      "epoch": 0.4161198288159772,
      "grad_norm": 0.7403674721717834,
      "learning_rate": 9.693586481969856e-06,
      "loss": 0.6133,
      "step": 2917
    },
    {
      "epoch": 0.41626248216833095,
      "grad_norm": 0.8431648015975952,
      "learning_rate": 9.693321004136451e-06,
      "loss": 0.7607,
      "step": 2918
    },
    {
      "epoch": 0.41640513552068475,
      "grad_norm": 1.0459867715835571,
      "learning_rate": 9.693055414985856e-06,
      "loss": 0.8597,
      "step": 2919
    },
    {
      "epoch": 0.4165477888730385,
      "grad_norm": 0.7903814911842346,
      "learning_rate": 9.692789714524366e-06,
      "loss": 0.7889,
      "step": 2920
    },
    {
      "epoch": 0.4166904422253923,
      "grad_norm": 1.0505493879318237,
      "learning_rate": 9.692523902758285e-06,
      "loss": 0.7474,
      "step": 2921
    },
    {
      "epoch": 0.41683309557774606,
      "grad_norm": 0.8948885798454285,
      "learning_rate": 9.692257979693914e-06,
      "loss": 0.7673,
      "step": 2922
    },
    {
      "epoch": 0.41697574893009987,
      "grad_norm": 0.8693767189979553,
      "learning_rate": 9.691991945337565e-06,
      "loss": 0.7114,
      "step": 2923
    },
    {
      "epoch": 0.4171184022824536,
      "grad_norm": 0.9207613468170166,
      "learning_rate": 9.691725799695544e-06,
      "loss": 0.7066,
      "step": 2924
    },
    {
      "epoch": 0.4172610556348074,
      "grad_norm": 0.9942516088485718,
      "learning_rate": 9.691459542774164e-06,
      "loss": 0.7637,
      "step": 2925
    },
    {
      "epoch": 0.41740370898716117,
      "grad_norm": 0.8703949451446533,
      "learning_rate": 9.691193174579744e-06,
      "loss": 0.71,
      "step": 2926
    },
    {
      "epoch": 0.417546362339515,
      "grad_norm": 0.7398990392684937,
      "learning_rate": 9.690926695118598e-06,
      "loss": 0.7718,
      "step": 2927
    },
    {
      "epoch": 0.4176890156918688,
      "grad_norm": 0.8628445863723755,
      "learning_rate": 9.690660104397047e-06,
      "loss": 0.7644,
      "step": 2928
    },
    {
      "epoch": 0.41783166904422253,
      "grad_norm": 0.995631992816925,
      "learning_rate": 9.690393402421413e-06,
      "loss": 0.8154,
      "step": 2929
    },
    {
      "epoch": 0.41797432239657634,
      "grad_norm": 0.7129629850387573,
      "learning_rate": 9.690126589198026e-06,
      "loss": 0.7785,
      "step": 2930
    },
    {
      "epoch": 0.4181169757489301,
      "grad_norm": 0.8479319214820862,
      "learning_rate": 9.68985966473321e-06,
      "loss": 0.7697,
      "step": 2931
    },
    {
      "epoch": 0.4182596291012839,
      "grad_norm": 0.8483423590660095,
      "learning_rate": 9.689592629033298e-06,
      "loss": 0.8224,
      "step": 2932
    },
    {
      "epoch": 0.41840228245363764,
      "grad_norm": 0.6983876824378967,
      "learning_rate": 9.689325482104623e-06,
      "loss": 0.7851,
      "step": 2933
    },
    {
      "epoch": 0.41854493580599145,
      "grad_norm": 0.9249723553657532,
      "learning_rate": 9.689058223953521e-06,
      "loss": 0.7409,
      "step": 2934
    },
    {
      "epoch": 0.4186875891583452,
      "grad_norm": 0.8699818253517151,
      "learning_rate": 9.688790854586332e-06,
      "loss": 0.8245,
      "step": 2935
    },
    {
      "epoch": 0.418830242510699,
      "grad_norm": 0.8217840790748596,
      "learning_rate": 9.688523374009395e-06,
      "loss": 0.8005,
      "step": 2936
    },
    {
      "epoch": 0.4189728958630528,
      "grad_norm": 0.8918729424476624,
      "learning_rate": 9.688255782229057e-06,
      "loss": 0.8313,
      "step": 2937
    },
    {
      "epoch": 0.41911554921540656,
      "grad_norm": 0.8414826393127441,
      "learning_rate": 9.687988079251664e-06,
      "loss": 0.8073,
      "step": 2938
    },
    {
      "epoch": 0.41925820256776036,
      "grad_norm": 1.081383228302002,
      "learning_rate": 9.687720265083564e-06,
      "loss": 0.7288,
      "step": 2939
    },
    {
      "epoch": 0.4194008559201141,
      "grad_norm": 1.0577263832092285,
      "learning_rate": 9.687452339731112e-06,
      "loss": 0.8177,
      "step": 2940
    },
    {
      "epoch": 0.4195435092724679,
      "grad_norm": 0.9822176098823547,
      "learning_rate": 9.687184303200658e-06,
      "loss": 0.7495,
      "step": 2941
    },
    {
      "epoch": 0.41968616262482167,
      "grad_norm": 1.1737984418869019,
      "learning_rate": 9.686916155498565e-06,
      "loss": 0.753,
      "step": 2942
    },
    {
      "epoch": 0.4198288159771755,
      "grad_norm": 1.0080480575561523,
      "learning_rate": 9.68664789663119e-06,
      "loss": 0.7267,
      "step": 2943
    },
    {
      "epoch": 0.4199714693295292,
      "grad_norm": 0.8572803139686584,
      "learning_rate": 9.686379526604895e-06,
      "loss": 0.7757,
      "step": 2944
    },
    {
      "epoch": 0.42011412268188303,
      "grad_norm": 0.9193053841590881,
      "learning_rate": 9.686111045426045e-06,
      "loss": 0.7465,
      "step": 2945
    },
    {
      "epoch": 0.4202567760342368,
      "grad_norm": 0.9724590182304382,
      "learning_rate": 9.685842453101011e-06,
      "loss": 0.764,
      "step": 2946
    },
    {
      "epoch": 0.4203994293865906,
      "grad_norm": 0.866526186466217,
      "learning_rate": 9.685573749636161e-06,
      "loss": 0.7693,
      "step": 2947
    },
    {
      "epoch": 0.4205420827389444,
      "grad_norm": 0.8589038848876953,
      "learning_rate": 9.68530493503787e-06,
      "loss": 0.7453,
      "step": 2948
    },
    {
      "epoch": 0.42068473609129814,
      "grad_norm": 0.7004780173301697,
      "learning_rate": 9.68503600931251e-06,
      "loss": 0.8181,
      "step": 2949
    },
    {
      "epoch": 0.42082738944365194,
      "grad_norm": 0.8782320618629456,
      "learning_rate": 9.684766972466463e-06,
      "loss": 0.814,
      "step": 2950
    },
    {
      "epoch": 0.4209700427960057,
      "grad_norm": 0.9447126388549805,
      "learning_rate": 9.68449782450611e-06,
      "loss": 0.791,
      "step": 2951
    },
    {
      "epoch": 0.4211126961483595,
      "grad_norm": 1.1747344732284546,
      "learning_rate": 9.684228565437833e-06,
      "loss": 0.8835,
      "step": 2952
    },
    {
      "epoch": 0.42125534950071325,
      "grad_norm": 0.7942746877670288,
      "learning_rate": 9.683959195268018e-06,
      "loss": 0.8682,
      "step": 2953
    },
    {
      "epoch": 0.42139800285306706,
      "grad_norm": 0.8650901317596436,
      "learning_rate": 9.683689714003055e-06,
      "loss": 0.8234,
      "step": 2954
    },
    {
      "epoch": 0.4215406562054208,
      "grad_norm": 0.8994263410568237,
      "learning_rate": 9.683420121649336e-06,
      "loss": 0.7916,
      "step": 2955
    },
    {
      "epoch": 0.4216833095577746,
      "grad_norm": 1.0481706857681274,
      "learning_rate": 9.683150418213257e-06,
      "loss": 0.6965,
      "step": 2956
    },
    {
      "epoch": 0.42182596291012836,
      "grad_norm": 0.9313482046127319,
      "learning_rate": 9.68288060370121e-06,
      "loss": 0.8502,
      "step": 2957
    },
    {
      "epoch": 0.42196861626248217,
      "grad_norm": 0.9151977896690369,
      "learning_rate": 9.682610678119599e-06,
      "loss": 0.7831,
      "step": 2958
    },
    {
      "epoch": 0.42211126961483597,
      "grad_norm": 0.6517190337181091,
      "learning_rate": 9.682340641474824e-06,
      "loss": 0.7718,
      "step": 2959
    },
    {
      "epoch": 0.4222539229671897,
      "grad_norm": 0.7682685256004333,
      "learning_rate": 9.68207049377329e-06,
      "loss": 0.786,
      "step": 2960
    },
    {
      "epoch": 0.4223965763195435,
      "grad_norm": 0.8396185040473938,
      "learning_rate": 9.681800235021407e-06,
      "loss": 0.7308,
      "step": 2961
    },
    {
      "epoch": 0.4225392296718973,
      "grad_norm": 1.1531387567520142,
      "learning_rate": 9.68152986522558e-06,
      "loss": 0.8158,
      "step": 2962
    },
    {
      "epoch": 0.4226818830242511,
      "grad_norm": 0.820608377456665,
      "learning_rate": 9.681259384392225e-06,
      "loss": 0.787,
      "step": 2963
    },
    {
      "epoch": 0.42282453637660483,
      "grad_norm": 0.9447684288024902,
      "learning_rate": 9.680988792527758e-06,
      "loss": 0.8109,
      "step": 2964
    },
    {
      "epoch": 0.42296718972895864,
      "grad_norm": 1.0396854877471924,
      "learning_rate": 9.680718089638594e-06,
      "loss": 0.8199,
      "step": 2965
    },
    {
      "epoch": 0.4231098430813124,
      "grad_norm": 0.913512647151947,
      "learning_rate": 9.680447275731156e-06,
      "loss": 0.7568,
      "step": 2966
    },
    {
      "epoch": 0.4232524964336662,
      "grad_norm": 0.9895064830780029,
      "learning_rate": 9.680176350811867e-06,
      "loss": 0.8481,
      "step": 2967
    },
    {
      "epoch": 0.42339514978602,
      "grad_norm": 0.8953961133956909,
      "learning_rate": 9.679905314887152e-06,
      "loss": 0.8444,
      "step": 2968
    },
    {
      "epoch": 0.42353780313837375,
      "grad_norm": 0.88303142786026,
      "learning_rate": 9.67963416796344e-06,
      "loss": 0.7658,
      "step": 2969
    },
    {
      "epoch": 0.42368045649072755,
      "grad_norm": 1.0362825393676758,
      "learning_rate": 9.67936291004716e-06,
      "loss": 0.8038,
      "step": 2970
    },
    {
      "epoch": 0.4238231098430813,
      "grad_norm": 0.9750179052352905,
      "learning_rate": 9.67909154114475e-06,
      "loss": 0.8254,
      "step": 2971
    },
    {
      "epoch": 0.4239657631954351,
      "grad_norm": 1.0256450176239014,
      "learning_rate": 9.678820061262643e-06,
      "loss": 0.7433,
      "step": 2972
    },
    {
      "epoch": 0.42410841654778886,
      "grad_norm": 0.7564718723297119,
      "learning_rate": 9.678548470407281e-06,
      "loss": 0.8129,
      "step": 2973
    },
    {
      "epoch": 0.42425106990014266,
      "grad_norm": 1.1093556880950928,
      "learning_rate": 9.678276768585102e-06,
      "loss": 0.7977,
      "step": 2974
    },
    {
      "epoch": 0.4243937232524964,
      "grad_norm": 0.9291697144508362,
      "learning_rate": 9.678004955802552e-06,
      "loss": 0.7628,
      "step": 2975
    },
    {
      "epoch": 0.4245363766048502,
      "grad_norm": 0.9266040325164795,
      "learning_rate": 9.67773303206608e-06,
      "loss": 0.8135,
      "step": 2976
    },
    {
      "epoch": 0.42467902995720397,
      "grad_norm": 0.8013978600502014,
      "learning_rate": 9.677460997382131e-06,
      "loss": 0.7634,
      "step": 2977
    },
    {
      "epoch": 0.4248216833095578,
      "grad_norm": 0.8356562852859497,
      "learning_rate": 9.677188851757162e-06,
      "loss": 0.7233,
      "step": 2978
    },
    {
      "epoch": 0.4249643366619116,
      "grad_norm": 0.9609328508377075,
      "learning_rate": 9.676916595197624e-06,
      "loss": 0.7878,
      "step": 2979
    },
    {
      "epoch": 0.42510699001426533,
      "grad_norm": 0.7560855746269226,
      "learning_rate": 9.676644227709975e-06,
      "loss": 0.8273,
      "step": 2980
    },
    {
      "epoch": 0.42524964336661913,
      "grad_norm": 0.8932199478149414,
      "learning_rate": 9.676371749300678e-06,
      "loss": 0.8321,
      "step": 2981
    },
    {
      "epoch": 0.4253922967189729,
      "grad_norm": 0.8932722806930542,
      "learning_rate": 9.676099159976193e-06,
      "loss": 0.7603,
      "step": 2982
    },
    {
      "epoch": 0.4255349500713267,
      "grad_norm": 1.1066131591796875,
      "learning_rate": 9.675826459742984e-06,
      "loss": 0.8208,
      "step": 2983
    },
    {
      "epoch": 0.42567760342368044,
      "grad_norm": 0.8241662979125977,
      "learning_rate": 9.675553648607525e-06,
      "loss": 0.8354,
      "step": 2984
    },
    {
      "epoch": 0.42582025677603424,
      "grad_norm": 0.87542325258255,
      "learning_rate": 9.675280726576282e-06,
      "loss": 0.769,
      "step": 2985
    },
    {
      "epoch": 0.425962910128388,
      "grad_norm": 0.7184815406799316,
      "learning_rate": 9.675007693655727e-06,
      "loss": 0.7796,
      "step": 2986
    },
    {
      "epoch": 0.4261055634807418,
      "grad_norm": 0.8368867635726929,
      "learning_rate": 9.674734549852338e-06,
      "loss": 0.7569,
      "step": 2987
    },
    {
      "epoch": 0.42624821683309555,
      "grad_norm": 0.8585637807846069,
      "learning_rate": 9.674461295172593e-06,
      "loss": 0.7845,
      "step": 2988
    },
    {
      "epoch": 0.42639087018544936,
      "grad_norm": 1.1723369359970093,
      "learning_rate": 9.674187929622973e-06,
      "loss": 0.749,
      "step": 2989
    },
    {
      "epoch": 0.42653352353780316,
      "grad_norm": 0.8972938656806946,
      "learning_rate": 9.673914453209963e-06,
      "loss": 0.7623,
      "step": 2990
    },
    {
      "epoch": 0.4266761768901569,
      "grad_norm": 0.7342985272407532,
      "learning_rate": 9.673640865940049e-06,
      "loss": 0.648,
      "step": 2991
    },
    {
      "epoch": 0.4268188302425107,
      "grad_norm": 0.852169394493103,
      "learning_rate": 9.673367167819718e-06,
      "loss": 0.817,
      "step": 2992
    },
    {
      "epoch": 0.42696148359486447,
      "grad_norm": 0.9238767623901367,
      "learning_rate": 9.673093358855464e-06,
      "loss": 0.7707,
      "step": 2993
    },
    {
      "epoch": 0.42710413694721827,
      "grad_norm": 1.145296335220337,
      "learning_rate": 9.67281943905378e-06,
      "loss": 0.7956,
      "step": 2994
    },
    {
      "epoch": 0.427246790299572,
      "grad_norm": 0.7868570685386658,
      "learning_rate": 9.672545408421163e-06,
      "loss": 0.7725,
      "step": 2995
    },
    {
      "epoch": 0.4273894436519258,
      "grad_norm": 0.813515841960907,
      "learning_rate": 9.672271266964114e-06,
      "loss": 0.7757,
      "step": 2996
    },
    {
      "epoch": 0.4275320970042796,
      "grad_norm": 0.8881155848503113,
      "learning_rate": 9.671997014689132e-06,
      "loss": 0.6963,
      "step": 2997
    },
    {
      "epoch": 0.4276747503566334,
      "grad_norm": 0.9524695873260498,
      "learning_rate": 9.671722651602723e-06,
      "loss": 0.8461,
      "step": 2998
    },
    {
      "epoch": 0.4278174037089872,
      "grad_norm": 0.9918982982635498,
      "learning_rate": 9.671448177711397e-06,
      "loss": 0.7447,
      "step": 2999
    },
    {
      "epoch": 0.42796005706134094,
      "grad_norm": 1.121673345565796,
      "learning_rate": 9.67117359302166e-06,
      "loss": 0.757,
      "step": 3000
    },
    {
      "epoch": 0.42810271041369474,
      "grad_norm": 0.9098827242851257,
      "learning_rate": 9.670898897540028e-06,
      "loss": 0.7916,
      "step": 3001
    },
    {
      "epoch": 0.4282453637660485,
      "grad_norm": 1.1406325101852417,
      "learning_rate": 9.670624091273015e-06,
      "loss": 0.6309,
      "step": 3002
    },
    {
      "epoch": 0.4283880171184023,
      "grad_norm": 0.8661484718322754,
      "learning_rate": 9.67034917422714e-06,
      "loss": 0.7726,
      "step": 3003
    },
    {
      "epoch": 0.42853067047075605,
      "grad_norm": 1.3654043674468994,
      "learning_rate": 9.670074146408921e-06,
      "loss": 0.7237,
      "step": 3004
    },
    {
      "epoch": 0.42867332382310985,
      "grad_norm": 0.8865222334861755,
      "learning_rate": 9.669799007824883e-06,
      "loss": 0.7603,
      "step": 3005
    },
    {
      "epoch": 0.4288159771754636,
      "grad_norm": 0.8929343223571777,
      "learning_rate": 9.669523758481552e-06,
      "loss": 0.718,
      "step": 3006
    },
    {
      "epoch": 0.4289586305278174,
      "grad_norm": 0.904663622379303,
      "learning_rate": 9.669248398385455e-06,
      "loss": 0.6883,
      "step": 3007
    },
    {
      "epoch": 0.42910128388017116,
      "grad_norm": 0.8316035866737366,
      "learning_rate": 9.668972927543124e-06,
      "loss": 0.7782,
      "step": 3008
    },
    {
      "epoch": 0.42924393723252496,
      "grad_norm": 1.2531659603118896,
      "learning_rate": 9.668697345961093e-06,
      "loss": 0.8153,
      "step": 3009
    },
    {
      "epoch": 0.42938659058487877,
      "grad_norm": 0.7125189900398254,
      "learning_rate": 9.668421653645897e-06,
      "loss": 0.7405,
      "step": 3010
    },
    {
      "epoch": 0.4295292439372325,
      "grad_norm": 0.9366601705551147,
      "learning_rate": 9.668145850604077e-06,
      "loss": 0.8485,
      "step": 3011
    },
    {
      "epoch": 0.4296718972895863,
      "grad_norm": 1.0495223999023438,
      "learning_rate": 9.667869936842173e-06,
      "loss": 0.6174,
      "step": 3012
    },
    {
      "epoch": 0.4298145506419401,
      "grad_norm": 0.9196107983589172,
      "learning_rate": 9.66759391236673e-06,
      "loss": 0.8012,
      "step": 3013
    },
    {
      "epoch": 0.4299572039942939,
      "grad_norm": 1.0751285552978516,
      "learning_rate": 9.667317777184293e-06,
      "loss": 0.7651,
      "step": 3014
    },
    {
      "epoch": 0.43009985734664763,
      "grad_norm": 1.247733235359192,
      "learning_rate": 9.667041531301412e-06,
      "loss": 0.7583,
      "step": 3015
    },
    {
      "epoch": 0.43024251069900143,
      "grad_norm": 1.0579503774642944,
      "learning_rate": 9.666765174724641e-06,
      "loss": 0.7364,
      "step": 3016
    },
    {
      "epoch": 0.4303851640513552,
      "grad_norm": 0.7801899909973145,
      "learning_rate": 9.666488707460536e-06,
      "loss": 0.7744,
      "step": 3017
    },
    {
      "epoch": 0.430527817403709,
      "grad_norm": 1.1810188293457031,
      "learning_rate": 9.66621212951565e-06,
      "loss": 0.8798,
      "step": 3018
    },
    {
      "epoch": 0.43067047075606274,
      "grad_norm": 0.9429445862770081,
      "learning_rate": 9.665935440896544e-06,
      "loss": 0.8093,
      "step": 3019
    },
    {
      "epoch": 0.43081312410841655,
      "grad_norm": 0.8942497372627258,
      "learning_rate": 9.665658641609781e-06,
      "loss": 0.7485,
      "step": 3020
    },
    {
      "epoch": 0.43095577746077035,
      "grad_norm": 0.9179964661598206,
      "learning_rate": 9.665381731661927e-06,
      "loss": 0.8204,
      "step": 3021
    },
    {
      "epoch": 0.4310984308131241,
      "grad_norm": 0.8470039963722229,
      "learning_rate": 9.66510471105955e-06,
      "loss": 0.7596,
      "step": 3022
    },
    {
      "epoch": 0.4312410841654779,
      "grad_norm": 1.2013462781906128,
      "learning_rate": 9.664827579809218e-06,
      "loss": 0.7973,
      "step": 3023
    },
    {
      "epoch": 0.43138373751783166,
      "grad_norm": 1.0752782821655273,
      "learning_rate": 9.664550337917508e-06,
      "loss": 0.7089,
      "step": 3024
    },
    {
      "epoch": 0.43152639087018546,
      "grad_norm": 0.9532427191734314,
      "learning_rate": 9.664272985390992e-06,
      "loss": 0.7647,
      "step": 3025
    },
    {
      "epoch": 0.4316690442225392,
      "grad_norm": 1.0374795198440552,
      "learning_rate": 9.66399552223625e-06,
      "loss": 0.7886,
      "step": 3026
    },
    {
      "epoch": 0.431811697574893,
      "grad_norm": 0.8640323877334595,
      "learning_rate": 9.663717948459862e-06,
      "loss": 0.7641,
      "step": 3027
    },
    {
      "epoch": 0.43195435092724677,
      "grad_norm": 0.9225583672523499,
      "learning_rate": 9.663440264068416e-06,
      "loss": 0.8084,
      "step": 3028
    },
    {
      "epoch": 0.43209700427960057,
      "grad_norm": 1.0404295921325684,
      "learning_rate": 9.663162469068491e-06,
      "loss": 0.7242,
      "step": 3029
    },
    {
      "epoch": 0.4322396576319544,
      "grad_norm": 0.7711402773857117,
      "learning_rate": 9.66288456346668e-06,
      "loss": 0.746,
      "step": 3030
    },
    {
      "epoch": 0.4323823109843081,
      "grad_norm": 0.9285078644752502,
      "learning_rate": 9.662606547269574e-06,
      "loss": 0.7943,
      "step": 3031
    },
    {
      "epoch": 0.43252496433666193,
      "grad_norm": 0.7582229375839233,
      "learning_rate": 9.662328420483766e-06,
      "loss": 0.7592,
      "step": 3032
    },
    {
      "epoch": 0.4326676176890157,
      "grad_norm": 1.0217987298965454,
      "learning_rate": 9.662050183115853e-06,
      "loss": 0.7232,
      "step": 3033
    },
    {
      "epoch": 0.4328102710413695,
      "grad_norm": 0.9547514319419861,
      "learning_rate": 9.661771835172438e-06,
      "loss": 0.7271,
      "step": 3034
    },
    {
      "epoch": 0.43295292439372324,
      "grad_norm": 0.6787132024765015,
      "learning_rate": 9.661493376660117e-06,
      "loss": 0.7799,
      "step": 3035
    },
    {
      "epoch": 0.43309557774607704,
      "grad_norm": 0.8108693957328796,
      "learning_rate": 9.661214807585498e-06,
      "loss": 0.7701,
      "step": 3036
    },
    {
      "epoch": 0.4332382310984308,
      "grad_norm": 1.157433271408081,
      "learning_rate": 9.660936127955185e-06,
      "loss": 0.803,
      "step": 3037
    },
    {
      "epoch": 0.4333808844507846,
      "grad_norm": 0.7043929696083069,
      "learning_rate": 9.660657337775792e-06,
      "loss": 0.7616,
      "step": 3038
    },
    {
      "epoch": 0.43352353780313835,
      "grad_norm": 0.9452171325683594,
      "learning_rate": 9.66037843705393e-06,
      "loss": 0.759,
      "step": 3039
    },
    {
      "epoch": 0.43366619115549215,
      "grad_norm": 0.9995279908180237,
      "learning_rate": 9.660099425796212e-06,
      "loss": 0.8103,
      "step": 3040
    },
    {
      "epoch": 0.43380884450784596,
      "grad_norm": 0.9892847537994385,
      "learning_rate": 9.659820304009259e-06,
      "loss": 0.8005,
      "step": 3041
    },
    {
      "epoch": 0.4339514978601997,
      "grad_norm": 0.7565533518791199,
      "learning_rate": 9.659541071699689e-06,
      "loss": 0.7973,
      "step": 3042
    },
    {
      "epoch": 0.4340941512125535,
      "grad_norm": 0.8700818419456482,
      "learning_rate": 9.659261728874124e-06,
      "loss": 0.8226,
      "step": 3043
    },
    {
      "epoch": 0.43423680456490726,
      "grad_norm": 0.9775557518005371,
      "learning_rate": 9.658982275539192e-06,
      "loss": 0.8062,
      "step": 3044
    },
    {
      "epoch": 0.43437945791726107,
      "grad_norm": 0.7708172798156738,
      "learning_rate": 9.65870271170152e-06,
      "loss": 0.8032,
      "step": 3045
    },
    {
      "epoch": 0.4345221112696148,
      "grad_norm": 1.0917460918426514,
      "learning_rate": 9.65842303736774e-06,
      "loss": 0.7665,
      "step": 3046
    },
    {
      "epoch": 0.4346647646219686,
      "grad_norm": 0.7986547946929932,
      "learning_rate": 9.65814325254448e-06,
      "loss": 0.7897,
      "step": 3047
    },
    {
      "epoch": 0.4348074179743224,
      "grad_norm": 0.7651693820953369,
      "learning_rate": 9.657863357238383e-06,
      "loss": 0.8331,
      "step": 3048
    },
    {
      "epoch": 0.4349500713266762,
      "grad_norm": 0.9188210964202881,
      "learning_rate": 9.657583351456087e-06,
      "loss": 0.8155,
      "step": 3049
    },
    {
      "epoch": 0.43509272467902993,
      "grad_norm": 1.0141257047653198,
      "learning_rate": 9.657303235204225e-06,
      "loss": 0.8118,
      "step": 3050
    },
    {
      "epoch": 0.43523537803138374,
      "grad_norm": 0.786700963973999,
      "learning_rate": 9.657023008489453e-06,
      "loss": 0.7622,
      "step": 3051
    },
    {
      "epoch": 0.43537803138373754,
      "grad_norm": 0.846847653388977,
      "learning_rate": 9.656742671318407e-06,
      "loss": 0.7399,
      "step": 3052
    },
    {
      "epoch": 0.4355206847360913,
      "grad_norm": 0.9744015336036682,
      "learning_rate": 9.656462223697743e-06,
      "loss": 0.8088,
      "step": 3053
    },
    {
      "epoch": 0.4356633380884451,
      "grad_norm": 1.167312741279602,
      "learning_rate": 9.656181665634108e-06,
      "loss": 0.5903,
      "step": 3054
    },
    {
      "epoch": 0.43580599144079885,
      "grad_norm": 0.9063330888748169,
      "learning_rate": 9.65590099713416e-06,
      "loss": 0.8071,
      "step": 3055
    },
    {
      "epoch": 0.43594864479315265,
      "grad_norm": 1.0339521169662476,
      "learning_rate": 9.655620218204553e-06,
      "loss": 0.7904,
      "step": 3056
    },
    {
      "epoch": 0.4360912981455064,
      "grad_norm": 0.7241429686546326,
      "learning_rate": 9.65533932885195e-06,
      "loss": 0.7982,
      "step": 3057
    },
    {
      "epoch": 0.4362339514978602,
      "grad_norm": 0.8558134436607361,
      "learning_rate": 9.65505832908301e-06,
      "loss": 0.8195,
      "step": 3058
    },
    {
      "epoch": 0.43637660485021396,
      "grad_norm": 0.9787079095840454,
      "learning_rate": 9.654777218904398e-06,
      "loss": 0.7268,
      "step": 3059
    },
    {
      "epoch": 0.43651925820256776,
      "grad_norm": 0.8959437012672424,
      "learning_rate": 9.654495998322785e-06,
      "loss": 0.8182,
      "step": 3060
    },
    {
      "epoch": 0.43666191155492157,
      "grad_norm": 0.842130720615387,
      "learning_rate": 9.654214667344836e-06,
      "loss": 0.8604,
      "step": 3061
    },
    {
      "epoch": 0.4368045649072753,
      "grad_norm": 0.924280047416687,
      "learning_rate": 9.653933225977228e-06,
      "loss": 0.7298,
      "step": 3062
    },
    {
      "epoch": 0.4369472182596291,
      "grad_norm": 1.0871145725250244,
      "learning_rate": 9.653651674226633e-06,
      "loss": 0.8445,
      "step": 3063
    },
    {
      "epoch": 0.43708987161198287,
      "grad_norm": 1.035857915878296,
      "learning_rate": 9.653370012099731e-06,
      "loss": 0.8529,
      "step": 3064
    },
    {
      "epoch": 0.4372325249643367,
      "grad_norm": 0.8568934202194214,
      "learning_rate": 9.653088239603203e-06,
      "loss": 0.7676,
      "step": 3065
    },
    {
      "epoch": 0.4373751783166904,
      "grad_norm": 1.0932868719100952,
      "learning_rate": 9.65280635674373e-06,
      "loss": 0.8164,
      "step": 3066
    },
    {
      "epoch": 0.43751783166904423,
      "grad_norm": 0.7900270819664001,
      "learning_rate": 9.652524363527998e-06,
      "loss": 0.7831,
      "step": 3067
    },
    {
      "epoch": 0.437660485021398,
      "grad_norm": 0.7176871299743652,
      "learning_rate": 9.652242259962696e-06,
      "loss": 0.7843,
      "step": 3068
    },
    {
      "epoch": 0.4378031383737518,
      "grad_norm": 0.677873432636261,
      "learning_rate": 9.651960046054515e-06,
      "loss": 0.8556,
      "step": 3069
    },
    {
      "epoch": 0.43794579172610554,
      "grad_norm": 0.9160250425338745,
      "learning_rate": 9.65167772181015e-06,
      "loss": 0.7274,
      "step": 3070
    },
    {
      "epoch": 0.43808844507845934,
      "grad_norm": 0.6624865531921387,
      "learning_rate": 9.651395287236296e-06,
      "loss": 0.7651,
      "step": 3071
    },
    {
      "epoch": 0.43823109843081315,
      "grad_norm": 1.0936319828033447,
      "learning_rate": 9.651112742339652e-06,
      "loss": 0.7677,
      "step": 3072
    },
    {
      "epoch": 0.4383737517831669,
      "grad_norm": 0.7446964979171753,
      "learning_rate": 9.650830087126918e-06,
      "loss": 0.8186,
      "step": 3073
    },
    {
      "epoch": 0.4385164051355207,
      "grad_norm": 0.8047989010810852,
      "learning_rate": 9.6505473216048e-06,
      "loss": 0.7856,
      "step": 3074
    },
    {
      "epoch": 0.43865905848787445,
      "grad_norm": 0.8886755704879761,
      "learning_rate": 9.650264445780004e-06,
      "loss": 0.687,
      "step": 3075
    },
    {
      "epoch": 0.43880171184022826,
      "grad_norm": 0.8686378598213196,
      "learning_rate": 9.64998145965924e-06,
      "loss": 0.7465,
      "step": 3076
    },
    {
      "epoch": 0.438944365192582,
      "grad_norm": 0.746656596660614,
      "learning_rate": 9.649698363249218e-06,
      "loss": 0.8124,
      "step": 3077
    },
    {
      "epoch": 0.4390870185449358,
      "grad_norm": 0.9459176659584045,
      "learning_rate": 9.649415156556655e-06,
      "loss": 0.6717,
      "step": 3078
    },
    {
      "epoch": 0.43922967189728956,
      "grad_norm": 1.091217041015625,
      "learning_rate": 9.649131839588267e-06,
      "loss": 0.8181,
      "step": 3079
    },
    {
      "epoch": 0.43937232524964337,
      "grad_norm": 0.8017871975898743,
      "learning_rate": 9.648848412350771e-06,
      "loss": 0.7779,
      "step": 3080
    },
    {
      "epoch": 0.4395149786019971,
      "grad_norm": 0.6447712182998657,
      "learning_rate": 9.648564874850896e-06,
      "loss": 0.5722,
      "step": 3081
    },
    {
      "epoch": 0.4396576319543509,
      "grad_norm": 0.8549794554710388,
      "learning_rate": 9.648281227095361e-06,
      "loss": 0.7929,
      "step": 3082
    },
    {
      "epoch": 0.43980028530670473,
      "grad_norm": 0.8107016682624817,
      "learning_rate": 9.647997469090897e-06,
      "loss": 0.8119,
      "step": 3083
    },
    {
      "epoch": 0.4399429386590585,
      "grad_norm": 1.0441479682922363,
      "learning_rate": 9.64771360084423e-06,
      "loss": 0.583,
      "step": 3084
    },
    {
      "epoch": 0.4400855920114123,
      "grad_norm": 0.9004194140434265,
      "learning_rate": 9.6474296223621e-06,
      "loss": 0.7648,
      "step": 3085
    },
    {
      "epoch": 0.44022824536376604,
      "grad_norm": 1.1389235258102417,
      "learning_rate": 9.647145533651236e-06,
      "loss": 0.58,
      "step": 3086
    },
    {
      "epoch": 0.44037089871611984,
      "grad_norm": 0.9744426012039185,
      "learning_rate": 9.646861334718378e-06,
      "loss": 0.7146,
      "step": 3087
    },
    {
      "epoch": 0.4405135520684736,
      "grad_norm": 0.8823460340499878,
      "learning_rate": 9.646577025570267e-06,
      "loss": 0.8082,
      "step": 3088
    },
    {
      "epoch": 0.4406562054208274,
      "grad_norm": 0.8189143538475037,
      "learning_rate": 9.646292606213648e-06,
      "loss": 0.7566,
      "step": 3089
    },
    {
      "epoch": 0.44079885877318115,
      "grad_norm": 0.8790888786315918,
      "learning_rate": 9.646008076655264e-06,
      "loss": 0.7985,
      "step": 3090
    },
    {
      "epoch": 0.44094151212553495,
      "grad_norm": 0.896385908126831,
      "learning_rate": 9.645723436901865e-06,
      "loss": 0.7181,
      "step": 3091
    },
    {
      "epoch": 0.44108416547788876,
      "grad_norm": 0.8352712392807007,
      "learning_rate": 9.645438686960202e-06,
      "loss": 0.8581,
      "step": 3092
    },
    {
      "epoch": 0.4412268188302425,
      "grad_norm": 0.9876428842544556,
      "learning_rate": 9.645153826837029e-06,
      "loss": 0.7531,
      "step": 3093
    },
    {
      "epoch": 0.4413694721825963,
      "grad_norm": 0.9629712104797363,
      "learning_rate": 9.644868856539102e-06,
      "loss": 0.7679,
      "step": 3094
    },
    {
      "epoch": 0.44151212553495006,
      "grad_norm": 0.8564965724945068,
      "learning_rate": 9.64458377607318e-06,
      "loss": 0.7726,
      "step": 3095
    },
    {
      "epoch": 0.44165477888730387,
      "grad_norm": 0.964055061340332,
      "learning_rate": 9.644298585446025e-06,
      "loss": 0.722,
      "step": 3096
    },
    {
      "epoch": 0.4417974322396576,
      "grad_norm": 1.0898507833480835,
      "learning_rate": 9.6440132846644e-06,
      "loss": 0.8075,
      "step": 3097
    },
    {
      "epoch": 0.4419400855920114,
      "grad_norm": 0.8865700960159302,
      "learning_rate": 9.643727873735074e-06,
      "loss": 0.75,
      "step": 3098
    },
    {
      "epoch": 0.4420827389443652,
      "grad_norm": 0.9955110549926758,
      "learning_rate": 9.643442352664815e-06,
      "loss": 0.7279,
      "step": 3099
    },
    {
      "epoch": 0.442225392296719,
      "grad_norm": 0.8635203242301941,
      "learning_rate": 9.643156721460393e-06,
      "loss": 0.7342,
      "step": 3100
    },
    {
      "epoch": 0.4423680456490727,
      "grad_norm": 0.7685550451278687,
      "learning_rate": 9.642870980128588e-06,
      "loss": 0.824,
      "step": 3101
    },
    {
      "epoch": 0.44251069900142653,
      "grad_norm": 1.062356948852539,
      "learning_rate": 9.642585128676172e-06,
      "loss": 0.7089,
      "step": 3102
    },
    {
      "epoch": 0.44265335235378034,
      "grad_norm": 0.958548903465271,
      "learning_rate": 9.642299167109929e-06,
      "loss": 0.8108,
      "step": 3103
    },
    {
      "epoch": 0.4427960057061341,
      "grad_norm": 1.234354019165039,
      "learning_rate": 9.642013095436638e-06,
      "loss": 0.8021,
      "step": 3104
    },
    {
      "epoch": 0.4429386590584879,
      "grad_norm": 0.7994375824928284,
      "learning_rate": 9.641726913663085e-06,
      "loss": 0.7735,
      "step": 3105
    },
    {
      "epoch": 0.44308131241084164,
      "grad_norm": 0.8465738296508789,
      "learning_rate": 9.641440621796061e-06,
      "loss": 0.7941,
      "step": 3106
    },
    {
      "epoch": 0.44322396576319545,
      "grad_norm": 0.9791854619979858,
      "learning_rate": 9.64115421984235e-06,
      "loss": 0.8316,
      "step": 3107
    },
    {
      "epoch": 0.4433666191155492,
      "grad_norm": 0.735660195350647,
      "learning_rate": 9.640867707808751e-06,
      "loss": 0.7989,
      "step": 3108
    },
    {
      "epoch": 0.443509272467903,
      "grad_norm": 0.7800251841545105,
      "learning_rate": 9.640581085702059e-06,
      "loss": 0.7541,
      "step": 3109
    },
    {
      "epoch": 0.44365192582025675,
      "grad_norm": 1.094420075416565,
      "learning_rate": 9.640294353529067e-06,
      "loss": 0.7858,
      "step": 3110
    },
    {
      "epoch": 0.44379457917261056,
      "grad_norm": 1.032516360282898,
      "learning_rate": 9.64000751129658e-06,
      "loss": 0.7806,
      "step": 3111
    },
    {
      "epoch": 0.4439372325249643,
      "grad_norm": 0.7042543292045593,
      "learning_rate": 9.639720559011401e-06,
      "loss": 0.7736,
      "step": 3112
    },
    {
      "epoch": 0.4440798858773181,
      "grad_norm": 0.821504533290863,
      "learning_rate": 9.639433496680337e-06,
      "loss": 0.8189,
      "step": 3113
    },
    {
      "epoch": 0.4442225392296719,
      "grad_norm": 0.7929046154022217,
      "learning_rate": 9.639146324310192e-06,
      "loss": 0.7493,
      "step": 3114
    },
    {
      "epoch": 0.44436519258202567,
      "grad_norm": 0.8610069155693054,
      "learning_rate": 9.638859041907781e-06,
      "loss": 0.7907,
      "step": 3115
    },
    {
      "epoch": 0.4445078459343795,
      "grad_norm": 1.0897170305252075,
      "learning_rate": 9.638571649479919e-06,
      "loss": 0.7437,
      "step": 3116
    },
    {
      "epoch": 0.4446504992867332,
      "grad_norm": 0.998016893863678,
      "learning_rate": 9.63828414703342e-06,
      "loss": 0.6941,
      "step": 3117
    },
    {
      "epoch": 0.44479315263908703,
      "grad_norm": 0.7024821639060974,
      "learning_rate": 9.637996534575103e-06,
      "loss": 0.8237,
      "step": 3118
    },
    {
      "epoch": 0.4449358059914408,
      "grad_norm": 0.9574607610702515,
      "learning_rate": 9.63770881211179e-06,
      "loss": 0.7375,
      "step": 3119
    },
    {
      "epoch": 0.4450784593437946,
      "grad_norm": 0.8610005974769592,
      "learning_rate": 9.637420979650307e-06,
      "loss": 0.7895,
      "step": 3120
    },
    {
      "epoch": 0.44522111269614834,
      "grad_norm": 0.8034377694129944,
      "learning_rate": 9.637133037197479e-06,
      "loss": 0.8204,
      "step": 3121
    },
    {
      "epoch": 0.44536376604850214,
      "grad_norm": 0.7363777160644531,
      "learning_rate": 9.636844984760134e-06,
      "loss": 0.8248,
      "step": 3122
    },
    {
      "epoch": 0.44550641940085595,
      "grad_norm": 1.3569444417953491,
      "learning_rate": 9.63655682234511e-06,
      "loss": 0.802,
      "step": 3123
    },
    {
      "epoch": 0.4456490727532097,
      "grad_norm": 0.9390245079994202,
      "learning_rate": 9.636268549959233e-06,
      "loss": 0.7577,
      "step": 3124
    },
    {
      "epoch": 0.4457917261055635,
      "grad_norm": 0.8765220046043396,
      "learning_rate": 9.635980167609347e-06,
      "loss": 0.8271,
      "step": 3125
    },
    {
      "epoch": 0.44593437945791725,
      "grad_norm": 0.693592369556427,
      "learning_rate": 9.63569167530229e-06,
      "loss": 0.7494,
      "step": 3126
    },
    {
      "epoch": 0.44607703281027106,
      "grad_norm": 0.8288514018058777,
      "learning_rate": 9.635403073044905e-06,
      "loss": 0.825,
      "step": 3127
    },
    {
      "epoch": 0.4462196861626248,
      "grad_norm": 0.8633185625076294,
      "learning_rate": 9.635114360844035e-06,
      "loss": 0.7698,
      "step": 3128
    },
    {
      "epoch": 0.4463623395149786,
      "grad_norm": 0.9909041523933411,
      "learning_rate": 9.634825538706532e-06,
      "loss": 0.7964,
      "step": 3129
    },
    {
      "epoch": 0.44650499286733236,
      "grad_norm": 1.0626932382583618,
      "learning_rate": 9.634536606639243e-06,
      "loss": 0.7989,
      "step": 3130
    },
    {
      "epoch": 0.44664764621968617,
      "grad_norm": 0.6464933753013611,
      "learning_rate": 9.634247564649019e-06,
      "loss": 0.8357,
      "step": 3131
    },
    {
      "epoch": 0.4467902995720399,
      "grad_norm": 0.9832266569137573,
      "learning_rate": 9.63395841274272e-06,
      "loss": 0.7892,
      "step": 3132
    },
    {
      "epoch": 0.4469329529243937,
      "grad_norm": 0.9554382562637329,
      "learning_rate": 9.633669150927201e-06,
      "loss": 0.7939,
      "step": 3133
    },
    {
      "epoch": 0.44707560627674753,
      "grad_norm": 0.8843060731887817,
      "learning_rate": 9.633379779209326e-06,
      "loss": 0.8062,
      "step": 3134
    },
    {
      "epoch": 0.4472182596291013,
      "grad_norm": 0.9719558358192444,
      "learning_rate": 9.633090297595956e-06,
      "loss": 0.7762,
      "step": 3135
    },
    {
      "epoch": 0.4473609129814551,
      "grad_norm": 0.9557838439941406,
      "learning_rate": 9.632800706093956e-06,
      "loss": 0.8144,
      "step": 3136
    },
    {
      "epoch": 0.44750356633380883,
      "grad_norm": 1.2489206790924072,
      "learning_rate": 9.6325110047102e-06,
      "loss": 0.7602,
      "step": 3137
    },
    {
      "epoch": 0.44764621968616264,
      "grad_norm": 0.7784707546234131,
      "learning_rate": 9.63222119345155e-06,
      "loss": 0.7544,
      "step": 3138
    },
    {
      "epoch": 0.4477888730385164,
      "grad_norm": 0.7684941291809082,
      "learning_rate": 9.631931272324889e-06,
      "loss": 0.8124,
      "step": 3139
    },
    {
      "epoch": 0.4479315263908702,
      "grad_norm": 1.026983380317688,
      "learning_rate": 9.631641241337088e-06,
      "loss": 0.7457,
      "step": 3140
    },
    {
      "epoch": 0.44807417974322394,
      "grad_norm": 0.898593008518219,
      "learning_rate": 9.631351100495028e-06,
      "loss": 0.7497,
      "step": 3141
    },
    {
      "epoch": 0.44821683309557775,
      "grad_norm": 0.8573411107063293,
      "learning_rate": 9.63106084980559e-06,
      "loss": 0.788,
      "step": 3142
    },
    {
      "epoch": 0.4483594864479315,
      "grad_norm": 1.0061452388763428,
      "learning_rate": 9.630770489275657e-06,
      "loss": 0.8013,
      "step": 3143
    },
    {
      "epoch": 0.4485021398002853,
      "grad_norm": 1.1630775928497314,
      "learning_rate": 9.630480018912119e-06,
      "loss": 0.7759,
      "step": 3144
    },
    {
      "epoch": 0.4486447931526391,
      "grad_norm": 0.8170731067657471,
      "learning_rate": 9.630189438721862e-06,
      "loss": 0.7687,
      "step": 3145
    },
    {
      "epoch": 0.44878744650499286,
      "grad_norm": 0.9104035496711731,
      "learning_rate": 9.62989874871178e-06,
      "loss": 0.7778,
      "step": 3146
    },
    {
      "epoch": 0.44893009985734667,
      "grad_norm": 0.9890031814575195,
      "learning_rate": 9.629607948888769e-06,
      "loss": 0.7919,
      "step": 3147
    },
    {
      "epoch": 0.4490727532097004,
      "grad_norm": 0.8792649507522583,
      "learning_rate": 9.629317039259722e-06,
      "loss": 0.8439,
      "step": 3148
    },
    {
      "epoch": 0.4492154065620542,
      "grad_norm": 0.7971907258033752,
      "learning_rate": 9.629026019831545e-06,
      "loss": 0.7737,
      "step": 3149
    },
    {
      "epoch": 0.44935805991440797,
      "grad_norm": 1.1017431020736694,
      "learning_rate": 9.628734890611133e-06,
      "loss": 0.7723,
      "step": 3150
    },
    {
      "epoch": 0.4495007132667618,
      "grad_norm": 0.8510836362838745,
      "learning_rate": 9.628443651605397e-06,
      "loss": 0.6911,
      "step": 3151
    },
    {
      "epoch": 0.4496433666191155,
      "grad_norm": 0.6971721053123474,
      "learning_rate": 9.628152302821242e-06,
      "loss": 0.7196,
      "step": 3152
    },
    {
      "epoch": 0.44978601997146933,
      "grad_norm": 0.9653314352035522,
      "learning_rate": 9.62786084426558e-06,
      "loss": 0.7898,
      "step": 3153
    },
    {
      "epoch": 0.44992867332382314,
      "grad_norm": 1.0043598413467407,
      "learning_rate": 9.62756927594532e-06,
      "loss": 0.6401,
      "step": 3154
    },
    {
      "epoch": 0.4500713266761769,
      "grad_norm": 1.0867184400558472,
      "learning_rate": 9.627277597867383e-06,
      "loss": 0.7209,
      "step": 3155
    },
    {
      "epoch": 0.4502139800285307,
      "grad_norm": 0.8640044331550598,
      "learning_rate": 9.626985810038685e-06,
      "loss": 0.7667,
      "step": 3156
    },
    {
      "epoch": 0.45035663338088444,
      "grad_norm": 0.6708057522773743,
      "learning_rate": 9.626693912466142e-06,
      "loss": 0.7246,
      "step": 3157
    },
    {
      "epoch": 0.45049928673323825,
      "grad_norm": 0.9218789935112,
      "learning_rate": 9.626401905156686e-06,
      "loss": 0.7633,
      "step": 3158
    },
    {
      "epoch": 0.450641940085592,
      "grad_norm": 0.7658910751342773,
      "learning_rate": 9.626109788117236e-06,
      "loss": 0.7817,
      "step": 3159
    },
    {
      "epoch": 0.4507845934379458,
      "grad_norm": 0.9468095898628235,
      "learning_rate": 9.625817561354723e-06,
      "loss": 0.7421,
      "step": 3160
    },
    {
      "epoch": 0.45092724679029955,
      "grad_norm": 0.753380298614502,
      "learning_rate": 9.625525224876076e-06,
      "loss": 0.7626,
      "step": 3161
    },
    {
      "epoch": 0.45106990014265336,
      "grad_norm": 0.9808356165885925,
      "learning_rate": 9.625232778688232e-06,
      "loss": 0.7686,
      "step": 3162
    },
    {
      "epoch": 0.4512125534950071,
      "grad_norm": 0.953310489654541,
      "learning_rate": 9.624940222798126e-06,
      "loss": 0.7573,
      "step": 3163
    },
    {
      "epoch": 0.4513552068473609,
      "grad_norm": 1.0124878883361816,
      "learning_rate": 9.624647557212696e-06,
      "loss": 0.7834,
      "step": 3164
    },
    {
      "epoch": 0.4514978601997147,
      "grad_norm": 0.9995359182357788,
      "learning_rate": 9.624354781938885e-06,
      "loss": 0.764,
      "step": 3165
    },
    {
      "epoch": 0.45164051355206847,
      "grad_norm": 1.0033024549484253,
      "learning_rate": 9.624061896983637e-06,
      "loss": 0.7917,
      "step": 3166
    },
    {
      "epoch": 0.4517831669044223,
      "grad_norm": 1.0837607383728027,
      "learning_rate": 9.623768902353896e-06,
      "loss": 0.7303,
      "step": 3167
    },
    {
      "epoch": 0.451925820256776,
      "grad_norm": 1.1728651523590088,
      "learning_rate": 9.623475798056615e-06,
      "loss": 0.8236,
      "step": 3168
    },
    {
      "epoch": 0.45206847360912983,
      "grad_norm": 0.7436525821685791,
      "learning_rate": 9.623182584098744e-06,
      "loss": 0.8001,
      "step": 3169
    },
    {
      "epoch": 0.4522111269614836,
      "grad_norm": 0.9537336826324463,
      "learning_rate": 9.622889260487236e-06,
      "loss": 0.7598,
      "step": 3170
    },
    {
      "epoch": 0.4523537803138374,
      "grad_norm": 1.2444756031036377,
      "learning_rate": 9.622595827229052e-06,
      "loss": 0.7903,
      "step": 3171
    },
    {
      "epoch": 0.45249643366619113,
      "grad_norm": 0.8312756419181824,
      "learning_rate": 9.622302284331149e-06,
      "loss": 0.8249,
      "step": 3172
    },
    {
      "epoch": 0.45263908701854494,
      "grad_norm": 0.6423664093017578,
      "learning_rate": 9.62200863180049e-06,
      "loss": 0.8271,
      "step": 3173
    },
    {
      "epoch": 0.4527817403708987,
      "grad_norm": 0.815904974937439,
      "learning_rate": 9.621714869644039e-06,
      "loss": 0.7393,
      "step": 3174
    },
    {
      "epoch": 0.4529243937232525,
      "grad_norm": 0.9608733654022217,
      "learning_rate": 9.621420997868765e-06,
      "loss": 0.7226,
      "step": 3175
    },
    {
      "epoch": 0.4530670470756063,
      "grad_norm": 1.05320405960083,
      "learning_rate": 9.621127016481637e-06,
      "loss": 0.7528,
      "step": 3176
    },
    {
      "epoch": 0.45320970042796005,
      "grad_norm": 1.0415343046188354,
      "learning_rate": 9.620832925489628e-06,
      "loss": 0.8105,
      "step": 3177
    },
    {
      "epoch": 0.45335235378031385,
      "grad_norm": 0.6745753884315491,
      "learning_rate": 9.620538724899713e-06,
      "loss": 0.7592,
      "step": 3178
    },
    {
      "epoch": 0.4534950071326676,
      "grad_norm": 0.7603384852409363,
      "learning_rate": 9.620244414718873e-06,
      "loss": 0.7451,
      "step": 3179
    },
    {
      "epoch": 0.4536376604850214,
      "grad_norm": 1.0224984884262085,
      "learning_rate": 9.619949994954083e-06,
      "loss": 0.8243,
      "step": 3180
    },
    {
      "epoch": 0.45378031383737516,
      "grad_norm": 1.0537947416305542,
      "learning_rate": 9.61965546561233e-06,
      "loss": 0.7938,
      "step": 3181
    },
    {
      "epoch": 0.45392296718972897,
      "grad_norm": 1.0355217456817627,
      "learning_rate": 9.6193608267006e-06,
      "loss": 0.7635,
      "step": 3182
    },
    {
      "epoch": 0.4540656205420827,
      "grad_norm": 1.05327308177948,
      "learning_rate": 9.619066078225879e-06,
      "loss": 0.7105,
      "step": 3183
    },
    {
      "epoch": 0.4542082738944365,
      "grad_norm": 0.8358237743377686,
      "learning_rate": 9.618771220195158e-06,
      "loss": 0.7776,
      "step": 3184
    },
    {
      "epoch": 0.4543509272467903,
      "grad_norm": 0.8938114047050476,
      "learning_rate": 9.618476252615432e-06,
      "loss": 0.8192,
      "step": 3185
    },
    {
      "epoch": 0.4544935805991441,
      "grad_norm": 0.8601402044296265,
      "learning_rate": 9.618181175493695e-06,
      "loss": 0.7887,
      "step": 3186
    },
    {
      "epoch": 0.4546362339514979,
      "grad_norm": 0.9431390166282654,
      "learning_rate": 9.617885988836951e-06,
      "loss": 0.772,
      "step": 3187
    },
    {
      "epoch": 0.45477888730385163,
      "grad_norm": 1.0409581661224365,
      "learning_rate": 9.617590692652194e-06,
      "loss": 0.8162,
      "step": 3188
    },
    {
      "epoch": 0.45492154065620544,
      "grad_norm": 1.045816421508789,
      "learning_rate": 9.617295286946433e-06,
      "loss": 0.8299,
      "step": 3189
    },
    {
      "epoch": 0.4550641940085592,
      "grad_norm": 1.2533879280090332,
      "learning_rate": 9.616999771726672e-06,
      "loss": 0.7699,
      "step": 3190
    },
    {
      "epoch": 0.455206847360913,
      "grad_norm": 0.8418307304382324,
      "learning_rate": 9.616704146999922e-06,
      "loss": 0.8287,
      "step": 3191
    },
    {
      "epoch": 0.45534950071326674,
      "grad_norm": 0.9584277868270874,
      "learning_rate": 9.616408412773192e-06,
      "loss": 0.7418,
      "step": 3192
    },
    {
      "epoch": 0.45549215406562055,
      "grad_norm": 0.9167273044586182,
      "learning_rate": 9.6161125690535e-06,
      "loss": 0.8158,
      "step": 3193
    },
    {
      "epoch": 0.4556348074179743,
      "grad_norm": 1.0698728561401367,
      "learning_rate": 9.61581661584786e-06,
      "loss": 0.8805,
      "step": 3194
    },
    {
      "epoch": 0.4557774607703281,
      "grad_norm": 1.150191068649292,
      "learning_rate": 9.615520553163294e-06,
      "loss": 0.8711,
      "step": 3195
    },
    {
      "epoch": 0.4559201141226819,
      "grad_norm": 0.6739160418510437,
      "learning_rate": 9.61522438100682e-06,
      "loss": 0.8376,
      "step": 3196
    },
    {
      "epoch": 0.45606276747503566,
      "grad_norm": 0.6858443021774292,
      "learning_rate": 9.614928099385467e-06,
      "loss": 0.8262,
      "step": 3197
    },
    {
      "epoch": 0.45620542082738946,
      "grad_norm": 0.7538442611694336,
      "learning_rate": 9.61463170830626e-06,
      "loss": 0.7978,
      "step": 3198
    },
    {
      "epoch": 0.4563480741797432,
      "grad_norm": 0.9261438250541687,
      "learning_rate": 9.61433520777623e-06,
      "loss": 0.7524,
      "step": 3199
    },
    {
      "epoch": 0.456490727532097,
      "grad_norm": 0.9965091347694397,
      "learning_rate": 9.614038597802405e-06,
      "loss": 0.7287,
      "step": 3200
    },
    {
      "epoch": 0.45663338088445077,
      "grad_norm": 1.0924919843673706,
      "learning_rate": 9.613741878391826e-06,
      "loss": 0.7976,
      "step": 3201
    },
    {
      "epoch": 0.4567760342368046,
      "grad_norm": 0.9510096907615662,
      "learning_rate": 9.613445049551529e-06,
      "loss": 0.8361,
      "step": 3202
    },
    {
      "epoch": 0.4569186875891583,
      "grad_norm": 0.7954645752906799,
      "learning_rate": 9.613148111288555e-06,
      "loss": 0.7906,
      "step": 3203
    },
    {
      "epoch": 0.45706134094151213,
      "grad_norm": 0.7574803233146667,
      "learning_rate": 9.612851063609943e-06,
      "loss": 0.765,
      "step": 3204
    },
    {
      "epoch": 0.4572039942938659,
      "grad_norm": 0.9515501260757446,
      "learning_rate": 9.61255390652274e-06,
      "loss": 0.6484,
      "step": 3205
    },
    {
      "epoch": 0.4573466476462197,
      "grad_norm": 0.872578501701355,
      "learning_rate": 9.612256640034e-06,
      "loss": 0.7244,
      "step": 3206
    },
    {
      "epoch": 0.4574893009985735,
      "grad_norm": 0.9286817312240601,
      "learning_rate": 9.611959264150763e-06,
      "loss": 0.8157,
      "step": 3207
    },
    {
      "epoch": 0.45763195435092724,
      "grad_norm": 0.8575893044471741,
      "learning_rate": 9.61166177888009e-06,
      "loss": 0.7894,
      "step": 3208
    },
    {
      "epoch": 0.45777460770328104,
      "grad_norm": 0.7181274890899658,
      "learning_rate": 9.611364184229037e-06,
      "loss": 0.7959,
      "step": 3209
    },
    {
      "epoch": 0.4579172610556348,
      "grad_norm": 0.9846453070640564,
      "learning_rate": 9.611066480204657e-06,
      "loss": 0.5517,
      "step": 3210
    },
    {
      "epoch": 0.4580599144079886,
      "grad_norm": 0.7997971177101135,
      "learning_rate": 9.610768666814015e-06,
      "loss": 0.798,
      "step": 3211
    },
    {
      "epoch": 0.45820256776034235,
      "grad_norm": 0.9456349015235901,
      "learning_rate": 9.610470744064175e-06,
      "loss": 0.6774,
      "step": 3212
    },
    {
      "epoch": 0.45834522111269616,
      "grad_norm": 0.8151498436927795,
      "learning_rate": 9.6101727119622e-06,
      "loss": 0.8158,
      "step": 3213
    },
    {
      "epoch": 0.4584878744650499,
      "grad_norm": 0.8408619165420532,
      "learning_rate": 9.60987457051516e-06,
      "loss": 0.8121,
      "step": 3214
    },
    {
      "epoch": 0.4586305278174037,
      "grad_norm": 0.9264407753944397,
      "learning_rate": 9.60957631973013e-06,
      "loss": 0.7625,
      "step": 3215
    },
    {
      "epoch": 0.4587731811697575,
      "grad_norm": 0.7770088911056519,
      "learning_rate": 9.60927795961418e-06,
      "loss": 0.819,
      "step": 3216
    },
    {
      "epoch": 0.45891583452211127,
      "grad_norm": 0.9818166494369507,
      "learning_rate": 9.608979490174385e-06,
      "loss": 0.5445,
      "step": 3217
    },
    {
      "epoch": 0.45905848787446507,
      "grad_norm": 0.9573136568069458,
      "learning_rate": 9.608680911417831e-06,
      "loss": 0.8056,
      "step": 3218
    },
    {
      "epoch": 0.4592011412268188,
      "grad_norm": 0.9539743065834045,
      "learning_rate": 9.608382223351594e-06,
      "loss": 0.7475,
      "step": 3219
    },
    {
      "epoch": 0.4593437945791726,
      "grad_norm": 0.7882934212684631,
      "learning_rate": 9.608083425982758e-06,
      "loss": 0.7953,
      "step": 3220
    },
    {
      "epoch": 0.4594864479315264,
      "grad_norm": 0.8050908446311951,
      "learning_rate": 9.607784519318413e-06,
      "loss": 0.7532,
      "step": 3221
    },
    {
      "epoch": 0.4596291012838802,
      "grad_norm": 0.9157688021659851,
      "learning_rate": 9.607485503365648e-06,
      "loss": 0.7805,
      "step": 3222
    },
    {
      "epoch": 0.45977175463623393,
      "grad_norm": 0.7888590693473816,
      "learning_rate": 9.607186378131554e-06,
      "loss": 0.7634,
      "step": 3223
    },
    {
      "epoch": 0.45991440798858774,
      "grad_norm": 1.0131776332855225,
      "learning_rate": 9.606887143623227e-06,
      "loss": 0.7819,
      "step": 3224
    },
    {
      "epoch": 0.4600570613409415,
      "grad_norm": 0.8839631676673889,
      "learning_rate": 9.606587799847762e-06,
      "loss": 0.7212,
      "step": 3225
    },
    {
      "epoch": 0.4601997146932953,
      "grad_norm": 0.9736989140510559,
      "learning_rate": 9.60628834681226e-06,
      "loss": 0.8043,
      "step": 3226
    },
    {
      "epoch": 0.4603423680456491,
      "grad_norm": 0.964718759059906,
      "learning_rate": 9.605988784523825e-06,
      "loss": 0.7951,
      "step": 3227
    },
    {
      "epoch": 0.46048502139800285,
      "grad_norm": 0.682881236076355,
      "learning_rate": 9.605689112989562e-06,
      "loss": 0.8274,
      "step": 3228
    },
    {
      "epoch": 0.46062767475035665,
      "grad_norm": 0.9523836374282837,
      "learning_rate": 9.605389332216577e-06,
      "loss": 0.8222,
      "step": 3229
    },
    {
      "epoch": 0.4607703281027104,
      "grad_norm": 0.7938910126686096,
      "learning_rate": 9.605089442211979e-06,
      "loss": 0.7983,
      "step": 3230
    },
    {
      "epoch": 0.4609129814550642,
      "grad_norm": 0.8781543970108032,
      "learning_rate": 9.604789442982885e-06,
      "loss": 0.8498,
      "step": 3231
    },
    {
      "epoch": 0.46105563480741796,
      "grad_norm": 0.9280606508255005,
      "learning_rate": 9.604489334536407e-06,
      "loss": 0.7853,
      "step": 3232
    },
    {
      "epoch": 0.46119828815977176,
      "grad_norm": 0.8369792103767395,
      "learning_rate": 9.604189116879666e-06,
      "loss": 0.7916,
      "step": 3233
    },
    {
      "epoch": 0.4613409415121255,
      "grad_norm": 0.9160106182098389,
      "learning_rate": 9.603888790019778e-06,
      "loss": 0.8411,
      "step": 3234
    },
    {
      "epoch": 0.4614835948644793,
      "grad_norm": 0.8717945218086243,
      "learning_rate": 9.60358835396387e-06,
      "loss": 0.7309,
      "step": 3235
    },
    {
      "epoch": 0.46162624821683307,
      "grad_norm": 0.9075588583946228,
      "learning_rate": 9.603287808719068e-06,
      "loss": 0.8055,
      "step": 3236
    },
    {
      "epoch": 0.4617689015691869,
      "grad_norm": 1.030797004699707,
      "learning_rate": 9.602987154292498e-06,
      "loss": 0.7871,
      "step": 3237
    },
    {
      "epoch": 0.4619115549215407,
      "grad_norm": 0.858007550239563,
      "learning_rate": 9.602686390691294e-06,
      "loss": 0.8354,
      "step": 3238
    },
    {
      "epoch": 0.46205420827389443,
      "grad_norm": 1.1889032125473022,
      "learning_rate": 9.602385517922587e-06,
      "loss": 0.7111,
      "step": 3239
    },
    {
      "epoch": 0.46219686162624823,
      "grad_norm": 1.2265406847000122,
      "learning_rate": 9.602084535993515e-06,
      "loss": 0.7946,
      "step": 3240
    },
    {
      "epoch": 0.462339514978602,
      "grad_norm": 0.9264851212501526,
      "learning_rate": 9.601783444911215e-06,
      "loss": 0.7307,
      "step": 3241
    },
    {
      "epoch": 0.4624821683309558,
      "grad_norm": 0.9731734991073608,
      "learning_rate": 9.601482244682828e-06,
      "loss": 0.832,
      "step": 3242
    },
    {
      "epoch": 0.46262482168330954,
      "grad_norm": 0.6344369649887085,
      "learning_rate": 9.601180935315499e-06,
      "loss": 0.7934,
      "step": 3243
    },
    {
      "epoch": 0.46276747503566334,
      "grad_norm": 0.7999869585037231,
      "learning_rate": 9.600879516816376e-06,
      "loss": 0.7569,
      "step": 3244
    },
    {
      "epoch": 0.4629101283880171,
      "grad_norm": 1.0258768796920776,
      "learning_rate": 9.600577989192607e-06,
      "loss": 0.7703,
      "step": 3245
    },
    {
      "epoch": 0.4630527817403709,
      "grad_norm": 1.1087878942489624,
      "learning_rate": 9.600276352451342e-06,
      "loss": 0.84,
      "step": 3246
    },
    {
      "epoch": 0.4631954350927247,
      "grad_norm": 0.8236591815948486,
      "learning_rate": 9.599974606599737e-06,
      "loss": 0.7499,
      "step": 3247
    },
    {
      "epoch": 0.46333808844507846,
      "grad_norm": 0.7220486402511597,
      "learning_rate": 9.59967275164495e-06,
      "loss": 0.7741,
      "step": 3248
    },
    {
      "epoch": 0.46348074179743226,
      "grad_norm": 0.7836496233940125,
      "learning_rate": 9.599370787594136e-06,
      "loss": 0.7621,
      "step": 3249
    },
    {
      "epoch": 0.463623395149786,
      "grad_norm": 0.9032376408576965,
      "learning_rate": 9.599068714454463e-06,
      "loss": 0.7961,
      "step": 3250
    },
    {
      "epoch": 0.4637660485021398,
      "grad_norm": 0.8954452872276306,
      "learning_rate": 9.59876653223309e-06,
      "loss": 0.7963,
      "step": 3251
    },
    {
      "epoch": 0.46390870185449357,
      "grad_norm": 1.0254467725753784,
      "learning_rate": 9.598464240937188e-06,
      "loss": 0.7492,
      "step": 3252
    },
    {
      "epoch": 0.46405135520684737,
      "grad_norm": 1.0668203830718994,
      "learning_rate": 9.598161840573927e-06,
      "loss": 0.7895,
      "step": 3253
    },
    {
      "epoch": 0.4641940085592011,
      "grad_norm": 0.978826105594635,
      "learning_rate": 9.597859331150476e-06,
      "loss": 0.6915,
      "step": 3254
    },
    {
      "epoch": 0.4643366619115549,
      "grad_norm": 0.8736929893493652,
      "learning_rate": 9.597556712674014e-06,
      "loss": 0.7851,
      "step": 3255
    },
    {
      "epoch": 0.4644793152639087,
      "grad_norm": 0.8989753723144531,
      "learning_rate": 9.597253985151715e-06,
      "loss": 0.8459,
      "step": 3256
    },
    {
      "epoch": 0.4646219686162625,
      "grad_norm": 0.9451749324798584,
      "learning_rate": 9.596951148590762e-06,
      "loss": 0.7292,
      "step": 3257
    },
    {
      "epoch": 0.4647646219686163,
      "grad_norm": 1.0433512926101685,
      "learning_rate": 9.596648202998336e-06,
      "loss": 0.8053,
      "step": 3258
    },
    {
      "epoch": 0.46490727532097004,
      "grad_norm": 1.5700123310089111,
      "learning_rate": 9.596345148381622e-06,
      "loss": 0.7484,
      "step": 3259
    },
    {
      "epoch": 0.46504992867332384,
      "grad_norm": 1.0634270906448364,
      "learning_rate": 9.596041984747812e-06,
      "loss": 0.766,
      "step": 3260
    },
    {
      "epoch": 0.4651925820256776,
      "grad_norm": 0.9275653958320618,
      "learning_rate": 9.59573871210409e-06,
      "loss": 0.8348,
      "step": 3261
    },
    {
      "epoch": 0.4653352353780314,
      "grad_norm": 1.2164363861083984,
      "learning_rate": 9.595435330457655e-06,
      "loss": 0.7802,
      "step": 3262
    },
    {
      "epoch": 0.46547788873038515,
      "grad_norm": 1.0342893600463867,
      "learning_rate": 9.595131839815698e-06,
      "loss": 0.5786,
      "step": 3263
    },
    {
      "epoch": 0.46562054208273895,
      "grad_norm": 0.9382723569869995,
      "learning_rate": 9.59482824018542e-06,
      "loss": 0.7375,
      "step": 3264
    },
    {
      "epoch": 0.4657631954350927,
      "grad_norm": 0.8210813403129578,
      "learning_rate": 9.59452453157402e-06,
      "loss": 0.792,
      "step": 3265
    },
    {
      "epoch": 0.4659058487874465,
      "grad_norm": 0.7499663829803467,
      "learning_rate": 9.594220713988705e-06,
      "loss": 0.7837,
      "step": 3266
    },
    {
      "epoch": 0.46604850213980026,
      "grad_norm": 0.7613505125045776,
      "learning_rate": 9.593916787436677e-06,
      "loss": 0.8459,
      "step": 3267
    },
    {
      "epoch": 0.46619115549215406,
      "grad_norm": 0.8401245474815369,
      "learning_rate": 9.593612751925146e-06,
      "loss": 0.7263,
      "step": 3268
    },
    {
      "epoch": 0.46633380884450787,
      "grad_norm": 0.7662743330001831,
      "learning_rate": 9.593308607461323e-06,
      "loss": 0.8032,
      "step": 3269
    },
    {
      "epoch": 0.4664764621968616,
      "grad_norm": 0.8141566514968872,
      "learning_rate": 9.593004354052424e-06,
      "loss": 0.8167,
      "step": 3270
    },
    {
      "epoch": 0.4666191155492154,
      "grad_norm": 1.0903931856155396,
      "learning_rate": 9.592699991705662e-06,
      "loss": 0.7851,
      "step": 3271
    },
    {
      "epoch": 0.4667617689015692,
      "grad_norm": 0.8852677941322327,
      "learning_rate": 9.592395520428257e-06,
      "loss": 0.8324,
      "step": 3272
    },
    {
      "epoch": 0.466904422253923,
      "grad_norm": 0.8141461610794067,
      "learning_rate": 9.592090940227432e-06,
      "loss": 0.7859,
      "step": 3273
    },
    {
      "epoch": 0.46704707560627673,
      "grad_norm": 1.0055336952209473,
      "learning_rate": 9.59178625111041e-06,
      "loss": 0.7143,
      "step": 3274
    },
    {
      "epoch": 0.46718972895863053,
      "grad_norm": 0.8097848296165466,
      "learning_rate": 9.591481453084416e-06,
      "loss": 0.8142,
      "step": 3275
    },
    {
      "epoch": 0.4673323823109843,
      "grad_norm": 1.0817996263504028,
      "learning_rate": 9.591176546156683e-06,
      "loss": 0.7397,
      "step": 3276
    },
    {
      "epoch": 0.4674750356633381,
      "grad_norm": 0.7397019267082214,
      "learning_rate": 9.59087153033444e-06,
      "loss": 0.7348,
      "step": 3277
    },
    {
      "epoch": 0.4676176890156919,
      "grad_norm": 0.7519397735595703,
      "learning_rate": 9.590566405624923e-06,
      "loss": 0.7753,
      "step": 3278
    },
    {
      "epoch": 0.46776034236804565,
      "grad_norm": 0.8674522638320923,
      "learning_rate": 9.590261172035368e-06,
      "loss": 0.8022,
      "step": 3279
    },
    {
      "epoch": 0.46790299572039945,
      "grad_norm": 0.8686073422431946,
      "learning_rate": 9.589955829573015e-06,
      "loss": 0.7183,
      "step": 3280
    },
    {
      "epoch": 0.4680456490727532,
      "grad_norm": 0.8603644967079163,
      "learning_rate": 9.589650378245106e-06,
      "loss": 0.8503,
      "step": 3281
    },
    {
      "epoch": 0.468188302425107,
      "grad_norm": 0.938601553440094,
      "learning_rate": 9.589344818058884e-06,
      "loss": 0.719,
      "step": 3282
    },
    {
      "epoch": 0.46833095577746076,
      "grad_norm": 1.1432517766952515,
      "learning_rate": 9.589039149021599e-06,
      "loss": 0.7682,
      "step": 3283
    },
    {
      "epoch": 0.46847360912981456,
      "grad_norm": 1.1235514879226685,
      "learning_rate": 9.588733371140502e-06,
      "loss": 0.7667,
      "step": 3284
    },
    {
      "epoch": 0.4686162624821683,
      "grad_norm": 1.0099997520446777,
      "learning_rate": 9.588427484422842e-06,
      "loss": 0.5631,
      "step": 3285
    },
    {
      "epoch": 0.4687589158345221,
      "grad_norm": 0.8651304841041565,
      "learning_rate": 9.588121488875875e-06,
      "loss": 0.7448,
      "step": 3286
    },
    {
      "epoch": 0.46890156918687587,
      "grad_norm": 0.8427119851112366,
      "learning_rate": 9.58781538450686e-06,
      "loss": 0.814,
      "step": 3287
    },
    {
      "epoch": 0.46904422253922967,
      "grad_norm": 0.8180068731307983,
      "learning_rate": 9.587509171323058e-06,
      "loss": 0.7244,
      "step": 3288
    },
    {
      "epoch": 0.4691868758915835,
      "grad_norm": 0.9020771980285645,
      "learning_rate": 9.587202849331728e-06,
      "loss": 0.8689,
      "step": 3289
    },
    {
      "epoch": 0.4693295292439372,
      "grad_norm": 0.8076386451721191,
      "learning_rate": 9.586896418540138e-06,
      "loss": 0.7931,
      "step": 3290
    },
    {
      "epoch": 0.46947218259629103,
      "grad_norm": 0.9270254373550415,
      "learning_rate": 9.586589878955558e-06,
      "loss": 0.8838,
      "step": 3291
    },
    {
      "epoch": 0.4696148359486448,
      "grad_norm": 0.95148104429245,
      "learning_rate": 9.586283230585255e-06,
      "loss": 0.76,
      "step": 3292
    },
    {
      "epoch": 0.4697574893009986,
      "grad_norm": 0.777870774269104,
      "learning_rate": 9.585976473436503e-06,
      "loss": 0.7963,
      "step": 3293
    },
    {
      "epoch": 0.46990014265335234,
      "grad_norm": 0.8370087146759033,
      "learning_rate": 9.585669607516579e-06,
      "loss": 0.8036,
      "step": 3294
    },
    {
      "epoch": 0.47004279600570614,
      "grad_norm": 1.1503015756607056,
      "learning_rate": 9.58536263283276e-06,
      "loss": 0.7816,
      "step": 3295
    },
    {
      "epoch": 0.4701854493580599,
      "grad_norm": 0.6871997117996216,
      "learning_rate": 9.585055549392328e-06,
      "loss": 0.7205,
      "step": 3296
    },
    {
      "epoch": 0.4703281027104137,
      "grad_norm": 1.048008918762207,
      "learning_rate": 9.584748357202565e-06,
      "loss": 0.7118,
      "step": 3297
    },
    {
      "epoch": 0.47047075606276745,
      "grad_norm": 1.0687289237976074,
      "learning_rate": 9.584441056270758e-06,
      "loss": 0.7412,
      "step": 3298
    },
    {
      "epoch": 0.47061340941512125,
      "grad_norm": 1.0124335289001465,
      "learning_rate": 9.584133646604198e-06,
      "loss": 0.8091,
      "step": 3299
    },
    {
      "epoch": 0.47075606276747506,
      "grad_norm": 0.9034653306007385,
      "learning_rate": 9.583826128210172e-06,
      "loss": 0.7489,
      "step": 3300
    },
    {
      "epoch": 0.4708987161198288,
      "grad_norm": 1.1596565246582031,
      "learning_rate": 9.583518501095976e-06,
      "loss": 0.5105,
      "step": 3301
    },
    {
      "epoch": 0.4710413694721826,
      "grad_norm": 0.9735150337219238,
      "learning_rate": 9.583210765268905e-06,
      "loss": 0.8305,
      "step": 3302
    },
    {
      "epoch": 0.47118402282453636,
      "grad_norm": 1.0492440462112427,
      "learning_rate": 9.582902920736259e-06,
      "loss": 0.7761,
      "step": 3303
    },
    {
      "epoch": 0.47132667617689017,
      "grad_norm": 0.6919054388999939,
      "learning_rate": 9.582594967505341e-06,
      "loss": 0.7489,
      "step": 3304
    },
    {
      "epoch": 0.4714693295292439,
      "grad_norm": 0.9516935348510742,
      "learning_rate": 9.582286905583452e-06,
      "loss": 0.7201,
      "step": 3305
    },
    {
      "epoch": 0.4716119828815977,
      "grad_norm": 0.9665992259979248,
      "learning_rate": 9.581978734977902e-06,
      "loss": 0.7759,
      "step": 3306
    },
    {
      "epoch": 0.4717546362339515,
      "grad_norm": 2.512998580932617,
      "learning_rate": 9.581670455695995e-06,
      "loss": 0.7929,
      "step": 3307
    },
    {
      "epoch": 0.4718972895863053,
      "grad_norm": 1.007515788078308,
      "learning_rate": 9.58136206774505e-06,
      "loss": 0.742,
      "step": 3308
    },
    {
      "epoch": 0.4720399429386591,
      "grad_norm": 0.7736210227012634,
      "learning_rate": 9.581053571132377e-06,
      "loss": 0.7686,
      "step": 3309
    },
    {
      "epoch": 0.47218259629101283,
      "grad_norm": 0.8426063060760498,
      "learning_rate": 9.580744965865292e-06,
      "loss": 0.8,
      "step": 3310
    },
    {
      "epoch": 0.47232524964336664,
      "grad_norm": 0.924156665802002,
      "learning_rate": 9.580436251951118e-06,
      "loss": 0.8075,
      "step": 3311
    },
    {
      "epoch": 0.4724679029957204,
      "grad_norm": 0.8497976660728455,
      "learning_rate": 9.580127429397173e-06,
      "loss": 0.7329,
      "step": 3312
    },
    {
      "epoch": 0.4726105563480742,
      "grad_norm": 1.2462340593338013,
      "learning_rate": 9.579818498210787e-06,
      "loss": 0.714,
      "step": 3313
    },
    {
      "epoch": 0.47275320970042795,
      "grad_norm": 0.9014277458190918,
      "learning_rate": 9.579509458399283e-06,
      "loss": 0.7491,
      "step": 3314
    },
    {
      "epoch": 0.47289586305278175,
      "grad_norm": 0.9916141033172607,
      "learning_rate": 9.579200309969992e-06,
      "loss": 0.8,
      "step": 3315
    },
    {
      "epoch": 0.4730385164051355,
      "grad_norm": 0.6881513595581055,
      "learning_rate": 9.578891052930248e-06,
      "loss": 0.7914,
      "step": 3316
    },
    {
      "epoch": 0.4731811697574893,
      "grad_norm": 1.0062134265899658,
      "learning_rate": 9.578581687287383e-06,
      "loss": 0.7581,
      "step": 3317
    },
    {
      "epoch": 0.47332382310984306,
      "grad_norm": 0.8403618335723877,
      "learning_rate": 9.578272213048738e-06,
      "loss": 0.7786,
      "step": 3318
    },
    {
      "epoch": 0.47346647646219686,
      "grad_norm": 1.5289185047149658,
      "learning_rate": 9.57796263022165e-06,
      "loss": 0.758,
      "step": 3319
    },
    {
      "epoch": 0.47360912981455067,
      "grad_norm": 0.7625239491462708,
      "learning_rate": 9.577652938813465e-06,
      "loss": 0.7897,
      "step": 3320
    },
    {
      "epoch": 0.4737517831669044,
      "grad_norm": 0.8987564444541931,
      "learning_rate": 9.577343138831526e-06,
      "loss": 0.7817,
      "step": 3321
    },
    {
      "epoch": 0.4738944365192582,
      "grad_norm": 1.0212286710739136,
      "learning_rate": 9.577033230283181e-06,
      "loss": 0.7974,
      "step": 3322
    },
    {
      "epoch": 0.47403708987161197,
      "grad_norm": 1.2200279235839844,
      "learning_rate": 9.576723213175781e-06,
      "loss": 0.6999,
      "step": 3323
    },
    {
      "epoch": 0.4741797432239658,
      "grad_norm": 1.008220911026001,
      "learning_rate": 9.57641308751668e-06,
      "loss": 0.691,
      "step": 3324
    },
    {
      "epoch": 0.4743223965763195,
      "grad_norm": 0.7976358532905579,
      "learning_rate": 9.57610285331323e-06,
      "loss": 0.8719,
      "step": 3325
    },
    {
      "epoch": 0.47446504992867333,
      "grad_norm": 1.099548578262329,
      "learning_rate": 9.575792510572796e-06,
      "loss": 0.7651,
      "step": 3326
    },
    {
      "epoch": 0.4746077032810271,
      "grad_norm": 1.777839183807373,
      "learning_rate": 9.575482059302732e-06,
      "loss": 0.7474,
      "step": 3327
    },
    {
      "epoch": 0.4747503566333809,
      "grad_norm": 0.9914703965187073,
      "learning_rate": 9.575171499510405e-06,
      "loss": 0.754,
      "step": 3328
    },
    {
      "epoch": 0.47489300998573464,
      "grad_norm": 1.0722758769989014,
      "learning_rate": 9.57486083120318e-06,
      "loss": 0.7997,
      "step": 3329
    },
    {
      "epoch": 0.47503566333808844,
      "grad_norm": 1.1728448867797852,
      "learning_rate": 9.574550054388426e-06,
      "loss": 0.7775,
      "step": 3330
    },
    {
      "epoch": 0.47517831669044225,
      "grad_norm": 0.8477550148963928,
      "learning_rate": 9.574239169073511e-06,
      "loss": 0.8113,
      "step": 3331
    },
    {
      "epoch": 0.475320970042796,
      "grad_norm": 0.9855422377586365,
      "learning_rate": 9.573928175265814e-06,
      "loss": 0.7778,
      "step": 3332
    },
    {
      "epoch": 0.4754636233951498,
      "grad_norm": 0.9079716801643372,
      "learning_rate": 9.573617072972707e-06,
      "loss": 0.8002,
      "step": 3333
    },
    {
      "epoch": 0.47560627674750355,
      "grad_norm": 0.7468597888946533,
      "learning_rate": 9.57330586220157e-06,
      "loss": 0.7523,
      "step": 3334
    },
    {
      "epoch": 0.47574893009985736,
      "grad_norm": 0.6399210095405579,
      "learning_rate": 9.572994542959784e-06,
      "loss": 0.779,
      "step": 3335
    },
    {
      "epoch": 0.4758915834522111,
      "grad_norm": 0.8479769229888916,
      "learning_rate": 9.572683115254736e-06,
      "loss": 0.7492,
      "step": 3336
    },
    {
      "epoch": 0.4760342368045649,
      "grad_norm": 0.9519271850585938,
      "learning_rate": 9.572371579093806e-06,
      "loss": 0.6451,
      "step": 3337
    },
    {
      "epoch": 0.47617689015691866,
      "grad_norm": 0.8485866189002991,
      "learning_rate": 9.57205993448439e-06,
      "loss": 0.7952,
      "step": 3338
    },
    {
      "epoch": 0.47631954350927247,
      "grad_norm": 1.2046514749526978,
      "learning_rate": 9.571748181433876e-06,
      "loss": 0.7759,
      "step": 3339
    },
    {
      "epoch": 0.4764621968616263,
      "grad_norm": 0.8765907883644104,
      "learning_rate": 9.57143631994966e-06,
      "loss": 0.7768,
      "step": 3340
    },
    {
      "epoch": 0.47660485021398,
      "grad_norm": 1.1514155864715576,
      "learning_rate": 9.571124350039136e-06,
      "loss": 0.782,
      "step": 3341
    },
    {
      "epoch": 0.47674750356633383,
      "grad_norm": 1.0372592210769653,
      "learning_rate": 9.570812271709707e-06,
      "loss": 0.7029,
      "step": 3342
    },
    {
      "epoch": 0.4768901569186876,
      "grad_norm": 0.9821612238883972,
      "learning_rate": 9.57050008496877e-06,
      "loss": 0.7207,
      "step": 3343
    },
    {
      "epoch": 0.4770328102710414,
      "grad_norm": 0.8796873092651367,
      "learning_rate": 9.570187789823735e-06,
      "loss": 0.6802,
      "step": 3344
    },
    {
      "epoch": 0.47717546362339514,
      "grad_norm": 1.1821767091751099,
      "learning_rate": 9.569875386282006e-06,
      "loss": 0.7817,
      "step": 3345
    },
    {
      "epoch": 0.47731811697574894,
      "grad_norm": 0.9369605183601379,
      "learning_rate": 9.569562874350993e-06,
      "loss": 0.7882,
      "step": 3346
    },
    {
      "epoch": 0.4774607703281027,
      "grad_norm": 0.941922664642334,
      "learning_rate": 9.56925025403811e-06,
      "loss": 0.7018,
      "step": 3347
    },
    {
      "epoch": 0.4776034236804565,
      "grad_norm": 1.0803322792053223,
      "learning_rate": 9.568937525350768e-06,
      "loss": 0.7262,
      "step": 3348
    },
    {
      "epoch": 0.47774607703281025,
      "grad_norm": 1.0246617794036865,
      "learning_rate": 9.568624688296387e-06,
      "loss": 0.7835,
      "step": 3349
    },
    {
      "epoch": 0.47788873038516405,
      "grad_norm": 0.8225551247596741,
      "learning_rate": 9.568311742882387e-06,
      "loss": 0.7754,
      "step": 3350
    },
    {
      "epoch": 0.47803138373751786,
      "grad_norm": 0.9982939958572388,
      "learning_rate": 9.567998689116192e-06,
      "loss": 0.7233,
      "step": 3351
    },
    {
      "epoch": 0.4781740370898716,
      "grad_norm": 0.9623139500617981,
      "learning_rate": 9.567685527005223e-06,
      "loss": 0.6851,
      "step": 3352
    },
    {
      "epoch": 0.4783166904422254,
      "grad_norm": 0.7305071353912354,
      "learning_rate": 9.56737225655691e-06,
      "loss": 0.731,
      "step": 3353
    },
    {
      "epoch": 0.47845934379457916,
      "grad_norm": 0.9447537660598755,
      "learning_rate": 9.567058877778683e-06,
      "loss": 0.8088,
      "step": 3354
    },
    {
      "epoch": 0.47860199714693297,
      "grad_norm": 0.8206272721290588,
      "learning_rate": 9.566745390677976e-06,
      "loss": 0.7427,
      "step": 3355
    },
    {
      "epoch": 0.4787446504992867,
      "grad_norm": 0.7819055318832397,
      "learning_rate": 9.566431795262223e-06,
      "loss": 0.7698,
      "step": 3356
    },
    {
      "epoch": 0.4788873038516405,
      "grad_norm": 0.9388282895088196,
      "learning_rate": 9.566118091538861e-06,
      "loss": 0.7823,
      "step": 3357
    },
    {
      "epoch": 0.47902995720399427,
      "grad_norm": 0.9281818270683289,
      "learning_rate": 9.565804279515333e-06,
      "loss": 0.7414,
      "step": 3358
    },
    {
      "epoch": 0.4791726105563481,
      "grad_norm": 0.9480597972869873,
      "learning_rate": 9.56549035919908e-06,
      "loss": 0.7342,
      "step": 3359
    },
    {
      "epoch": 0.4793152639087018,
      "grad_norm": 0.9322537779808044,
      "learning_rate": 9.565176330597548e-06,
      "loss": 0.7337,
      "step": 3360
    },
    {
      "epoch": 0.47945791726105563,
      "grad_norm": 0.8991124629974365,
      "learning_rate": 9.564862193718187e-06,
      "loss": 0.765,
      "step": 3361
    },
    {
      "epoch": 0.47960057061340944,
      "grad_norm": 0.9824044108390808,
      "learning_rate": 9.564547948568443e-06,
      "loss": 0.8108,
      "step": 3362
    },
    {
      "epoch": 0.4797432239657632,
      "grad_norm": 1.0409741401672363,
      "learning_rate": 9.564233595155777e-06,
      "loss": 0.7842,
      "step": 3363
    },
    {
      "epoch": 0.479885877318117,
      "grad_norm": 1.0946305990219116,
      "learning_rate": 9.563919133487638e-06,
      "loss": 0.8092,
      "step": 3364
    },
    {
      "epoch": 0.48002853067047074,
      "grad_norm": 0.9016760587692261,
      "learning_rate": 9.563604563571486e-06,
      "loss": 0.7146,
      "step": 3365
    },
    {
      "epoch": 0.48017118402282455,
      "grad_norm": 1.192918062210083,
      "learning_rate": 9.563289885414784e-06,
      "loss": 0.7887,
      "step": 3366
    },
    {
      "epoch": 0.4803138373751783,
      "grad_norm": 0.9595857262611389,
      "learning_rate": 9.562975099024995e-06,
      "loss": 0.7346,
      "step": 3367
    },
    {
      "epoch": 0.4804564907275321,
      "grad_norm": 0.8157551288604736,
      "learning_rate": 9.562660204409584e-06,
      "loss": 0.7408,
      "step": 3368
    },
    {
      "epoch": 0.48059914407988585,
      "grad_norm": 0.8149961233139038,
      "learning_rate": 9.562345201576021e-06,
      "loss": 0.7228,
      "step": 3369
    },
    {
      "epoch": 0.48074179743223966,
      "grad_norm": 1.0081431865692139,
      "learning_rate": 9.562030090531777e-06,
      "loss": 0.8398,
      "step": 3370
    },
    {
      "epoch": 0.48088445078459346,
      "grad_norm": 0.8619146347045898,
      "learning_rate": 9.561714871284326e-06,
      "loss": 0.7842,
      "step": 3371
    },
    {
      "epoch": 0.4810271041369472,
      "grad_norm": 0.8915643095970154,
      "learning_rate": 9.561399543841143e-06,
      "loss": 0.7844,
      "step": 3372
    },
    {
      "epoch": 0.481169757489301,
      "grad_norm": 0.8182870745658875,
      "learning_rate": 9.561084108209708e-06,
      "loss": 0.7867,
      "step": 3373
    },
    {
      "epoch": 0.48131241084165477,
      "grad_norm": 0.6612070798873901,
      "learning_rate": 9.560768564397505e-06,
      "loss": 0.789,
      "step": 3374
    },
    {
      "epoch": 0.4814550641940086,
      "grad_norm": 0.961587131023407,
      "learning_rate": 9.560452912412013e-06,
      "loss": 0.7723,
      "step": 3375
    },
    {
      "epoch": 0.4815977175463623,
      "grad_norm": 0.8533160090446472,
      "learning_rate": 9.560137152260723e-06,
      "loss": 0.7668,
      "step": 3376
    },
    {
      "epoch": 0.48174037089871613,
      "grad_norm": 1.1560102701187134,
      "learning_rate": 9.559821283951123e-06,
      "loss": 0.7964,
      "step": 3377
    },
    {
      "epoch": 0.4818830242510699,
      "grad_norm": 0.9823300838470459,
      "learning_rate": 9.559505307490704e-06,
      "loss": 0.7446,
      "step": 3378
    },
    {
      "epoch": 0.4820256776034237,
      "grad_norm": 1.1701860427856445,
      "learning_rate": 9.559189222886962e-06,
      "loss": 0.5649,
      "step": 3379
    },
    {
      "epoch": 0.48216833095577744,
      "grad_norm": 1.1225320100784302,
      "learning_rate": 9.558873030147392e-06,
      "loss": 0.7186,
      "step": 3380
    },
    {
      "epoch": 0.48231098430813124,
      "grad_norm": 0.9208173155784607,
      "learning_rate": 9.558556729279496e-06,
      "loss": 0.7182,
      "step": 3381
    },
    {
      "epoch": 0.48245363766048505,
      "grad_norm": 0.740606963634491,
      "learning_rate": 9.558240320290773e-06,
      "loss": 0.7868,
      "step": 3382
    },
    {
      "epoch": 0.4825962910128388,
      "grad_norm": 1.0487650632858276,
      "learning_rate": 9.55792380318873e-06,
      "loss": 0.7207,
      "step": 3383
    },
    {
      "epoch": 0.4827389443651926,
      "grad_norm": 0.9889148473739624,
      "learning_rate": 9.557607177980872e-06,
      "loss": 0.7738,
      "step": 3384
    },
    {
      "epoch": 0.48288159771754635,
      "grad_norm": 1.0722016096115112,
      "learning_rate": 9.557290444674711e-06,
      "loss": 0.7778,
      "step": 3385
    },
    {
      "epoch": 0.48302425106990016,
      "grad_norm": 0.7276331186294556,
      "learning_rate": 9.556973603277761e-06,
      "loss": 0.8651,
      "step": 3386
    },
    {
      "epoch": 0.4831669044222539,
      "grad_norm": 0.8589654564857483,
      "learning_rate": 9.556656653797531e-06,
      "loss": 0.7953,
      "step": 3387
    },
    {
      "epoch": 0.4833095577746077,
      "grad_norm": 0.8468416929244995,
      "learning_rate": 9.556339596241544e-06,
      "loss": 0.8578,
      "step": 3388
    },
    {
      "epoch": 0.48345221112696146,
      "grad_norm": 0.8124231696128845,
      "learning_rate": 9.556022430617318e-06,
      "loss": 0.8363,
      "step": 3389
    },
    {
      "epoch": 0.48359486447931527,
      "grad_norm": 1.1093146800994873,
      "learning_rate": 9.555705156932375e-06,
      "loss": 0.8176,
      "step": 3390
    },
    {
      "epoch": 0.483737517831669,
      "grad_norm": 1.0626344680786133,
      "learning_rate": 9.555387775194243e-06,
      "loss": 0.7326,
      "step": 3391
    },
    {
      "epoch": 0.4838801711840228,
      "grad_norm": 1.0361603498458862,
      "learning_rate": 9.555070285410444e-06,
      "loss": 0.8014,
      "step": 3392
    },
    {
      "epoch": 0.48402282453637663,
      "grad_norm": 0.8652250170707703,
      "learning_rate": 9.554752687588515e-06,
      "loss": 0.8119,
      "step": 3393
    },
    {
      "epoch": 0.4841654778887304,
      "grad_norm": 1.2987350225448608,
      "learning_rate": 9.554434981735984e-06,
      "loss": 0.803,
      "step": 3394
    },
    {
      "epoch": 0.4843081312410842,
      "grad_norm": 0.9913198351860046,
      "learning_rate": 9.554117167860389e-06,
      "loss": 0.8752,
      "step": 3395
    },
    {
      "epoch": 0.48445078459343793,
      "grad_norm": 0.9127089977264404,
      "learning_rate": 9.553799245969266e-06,
      "loss": 0.7903,
      "step": 3396
    },
    {
      "epoch": 0.48459343794579174,
      "grad_norm": 0.8692578077316284,
      "learning_rate": 9.553481216070157e-06,
      "loss": 0.7779,
      "step": 3397
    },
    {
      "epoch": 0.4847360912981455,
      "grad_norm": 1.1671979427337646,
      "learning_rate": 9.553163078170605e-06,
      "loss": 0.7761,
      "step": 3398
    },
    {
      "epoch": 0.4848787446504993,
      "grad_norm": 0.7351517081260681,
      "learning_rate": 9.552844832278155e-06,
      "loss": 0.6904,
      "step": 3399
    },
    {
      "epoch": 0.48502139800285304,
      "grad_norm": 1.8013851642608643,
      "learning_rate": 9.552526478400356e-06,
      "loss": 0.8007,
      "step": 3400
    },
    {
      "epoch": 0.48516405135520685,
      "grad_norm": 0.8513445258140564,
      "learning_rate": 9.552208016544759e-06,
      "loss": 0.7916,
      "step": 3401
    },
    {
      "epoch": 0.48530670470756065,
      "grad_norm": 0.854399561882019,
      "learning_rate": 9.551889446718916e-06,
      "loss": 0.8007,
      "step": 3402
    },
    {
      "epoch": 0.4854493580599144,
      "grad_norm": 0.8587920069694519,
      "learning_rate": 9.551570768930384e-06,
      "loss": 0.821,
      "step": 3403
    },
    {
      "epoch": 0.4855920114122682,
      "grad_norm": 0.8655844926834106,
      "learning_rate": 9.55125198318672e-06,
      "loss": 0.7874,
      "step": 3404
    },
    {
      "epoch": 0.48573466476462196,
      "grad_norm": 0.920585572719574,
      "learning_rate": 9.550933089495486e-06,
      "loss": 0.755,
      "step": 3405
    },
    {
      "epoch": 0.48587731811697576,
      "grad_norm": 0.9027933478355408,
      "learning_rate": 9.550614087864247e-06,
      "loss": 0.8007,
      "step": 3406
    },
    {
      "epoch": 0.4860199714693295,
      "grad_norm": 0.7437239289283752,
      "learning_rate": 9.550294978300568e-06,
      "loss": 0.7513,
      "step": 3407
    },
    {
      "epoch": 0.4861626248216833,
      "grad_norm": 0.748741090297699,
      "learning_rate": 9.549975760812019e-06,
      "loss": 0.838,
      "step": 3408
    },
    {
      "epoch": 0.48630527817403707,
      "grad_norm": 1.9705638885498047,
      "learning_rate": 9.549656435406167e-06,
      "loss": 0.6532,
      "step": 3409
    },
    {
      "epoch": 0.4864479315263909,
      "grad_norm": 1.141339659690857,
      "learning_rate": 9.549337002090589e-06,
      "loss": 0.7726,
      "step": 3410
    },
    {
      "epoch": 0.4865905848787446,
      "grad_norm": 1.154482364654541,
      "learning_rate": 9.549017460872862e-06,
      "loss": 0.7307,
      "step": 3411
    },
    {
      "epoch": 0.48673323823109843,
      "grad_norm": 1.2299119234085083,
      "learning_rate": 9.548697811760565e-06,
      "loss": 0.7401,
      "step": 3412
    },
    {
      "epoch": 0.48687589158345224,
      "grad_norm": 0.8068072199821472,
      "learning_rate": 9.548378054761278e-06,
      "loss": 0.8125,
      "step": 3413
    },
    {
      "epoch": 0.487018544935806,
      "grad_norm": 0.8187466263771057,
      "learning_rate": 9.548058189882584e-06,
      "loss": 0.7122,
      "step": 3414
    },
    {
      "epoch": 0.4871611982881598,
      "grad_norm": 0.8442869782447815,
      "learning_rate": 9.547738217132072e-06,
      "loss": 0.7748,
      "step": 3415
    },
    {
      "epoch": 0.48730385164051354,
      "grad_norm": 1.218238115310669,
      "learning_rate": 9.54741813651733e-06,
      "loss": 0.8156,
      "step": 3416
    },
    {
      "epoch": 0.48744650499286735,
      "grad_norm": 0.8840314745903015,
      "learning_rate": 9.547097948045952e-06,
      "loss": 0.7733,
      "step": 3417
    },
    {
      "epoch": 0.4875891583452211,
      "grad_norm": 0.8033078908920288,
      "learning_rate": 9.546777651725529e-06,
      "loss": 0.7879,
      "step": 3418
    },
    {
      "epoch": 0.4877318116975749,
      "grad_norm": 0.8911908864974976,
      "learning_rate": 9.54645724756366e-06,
      "loss": 0.8573,
      "step": 3419
    },
    {
      "epoch": 0.48787446504992865,
      "grad_norm": 0.7573607563972473,
      "learning_rate": 9.546136735567943e-06,
      "loss": 0.7739,
      "step": 3420
    },
    {
      "epoch": 0.48801711840228246,
      "grad_norm": 0.965168833732605,
      "learning_rate": 9.545816115745981e-06,
      "loss": 0.769,
      "step": 3421
    },
    {
      "epoch": 0.4881597717546362,
      "grad_norm": 1.0998084545135498,
      "learning_rate": 9.545495388105379e-06,
      "loss": 0.7601,
      "step": 3422
    },
    {
      "epoch": 0.48830242510699,
      "grad_norm": 0.8415771126747131,
      "learning_rate": 9.545174552653741e-06,
      "loss": 0.8746,
      "step": 3423
    },
    {
      "epoch": 0.4884450784593438,
      "grad_norm": 0.9667220115661621,
      "learning_rate": 9.544853609398681e-06,
      "loss": 0.7261,
      "step": 3424
    },
    {
      "epoch": 0.48858773181169757,
      "grad_norm": 0.8491535782814026,
      "learning_rate": 9.544532558347809e-06,
      "loss": 0.7778,
      "step": 3425
    },
    {
      "epoch": 0.4887303851640514,
      "grad_norm": 0.842559814453125,
      "learning_rate": 9.544211399508739e-06,
      "loss": 0.7902,
      "step": 3426
    },
    {
      "epoch": 0.4888730385164051,
      "grad_norm": 0.9630987644195557,
      "learning_rate": 9.54389013288909e-06,
      "loss": 0.7747,
      "step": 3427
    },
    {
      "epoch": 0.48901569186875893,
      "grad_norm": 0.8068737387657166,
      "learning_rate": 9.543568758496482e-06,
      "loss": 0.8007,
      "step": 3428
    },
    {
      "epoch": 0.4891583452211127,
      "grad_norm": 0.7649833559989929,
      "learning_rate": 9.543247276338535e-06,
      "loss": 0.6022,
      "step": 3429
    },
    {
      "epoch": 0.4893009985734665,
      "grad_norm": 0.8590714335441589,
      "learning_rate": 9.542925686422874e-06,
      "loss": 0.821,
      "step": 3430
    },
    {
      "epoch": 0.48944365192582023,
      "grad_norm": 1.0582678318023682,
      "learning_rate": 9.54260398875713e-06,
      "loss": 0.7332,
      "step": 3431
    },
    {
      "epoch": 0.48958630527817404,
      "grad_norm": 0.8106250762939453,
      "learning_rate": 9.542282183348932e-06,
      "loss": 0.7699,
      "step": 3432
    },
    {
      "epoch": 0.48972895863052784,
      "grad_norm": 0.8199784159660339,
      "learning_rate": 9.54196027020591e-06,
      "loss": 0.7216,
      "step": 3433
    },
    {
      "epoch": 0.4898716119828816,
      "grad_norm": 1.0330818891525269,
      "learning_rate": 9.541638249335702e-06,
      "loss": 0.7559,
      "step": 3434
    },
    {
      "epoch": 0.4900142653352354,
      "grad_norm": 0.7265938520431519,
      "learning_rate": 9.541316120745944e-06,
      "loss": 0.8366,
      "step": 3435
    },
    {
      "epoch": 0.49015691868758915,
      "grad_norm": 0.9404028654098511,
      "learning_rate": 9.540993884444278e-06,
      "loss": 0.7518,
      "step": 3436
    },
    {
      "epoch": 0.49029957203994295,
      "grad_norm": 1.1536147594451904,
      "learning_rate": 9.540671540438346e-06,
      "loss": 0.825,
      "step": 3437
    },
    {
      "epoch": 0.4904422253922967,
      "grad_norm": 0.9902913570404053,
      "learning_rate": 9.540349088735795e-06,
      "loss": 0.8027,
      "step": 3438
    },
    {
      "epoch": 0.4905848787446505,
      "grad_norm": 1.057051658630371,
      "learning_rate": 9.540026529344268e-06,
      "loss": 0.6917,
      "step": 3439
    },
    {
      "epoch": 0.49072753209700426,
      "grad_norm": 0.8220142126083374,
      "learning_rate": 9.53970386227142e-06,
      "loss": 0.7244,
      "step": 3440
    },
    {
      "epoch": 0.49087018544935807,
      "grad_norm": 1.011712908744812,
      "learning_rate": 9.539381087524904e-06,
      "loss": 0.7658,
      "step": 3441
    },
    {
      "epoch": 0.4910128388017118,
      "grad_norm": 0.8901670575141907,
      "learning_rate": 9.539058205112372e-06,
      "loss": 0.6595,
      "step": 3442
    },
    {
      "epoch": 0.4911554921540656,
      "grad_norm": 0.752519965171814,
      "learning_rate": 9.538735215041489e-06,
      "loss": 0.8126,
      "step": 3443
    },
    {
      "epoch": 0.4912981455064194,
      "grad_norm": 1.0619455575942993,
      "learning_rate": 9.538412117319907e-06,
      "loss": 0.6629,
      "step": 3444
    },
    {
      "epoch": 0.4914407988587732,
      "grad_norm": 0.7560083866119385,
      "learning_rate": 9.538088911955297e-06,
      "loss": 0.5751,
      "step": 3445
    },
    {
      "epoch": 0.491583452211127,
      "grad_norm": 0.9296385645866394,
      "learning_rate": 9.53776559895532e-06,
      "loss": 0.8717,
      "step": 3446
    },
    {
      "epoch": 0.49172610556348073,
      "grad_norm": 0.8130275011062622,
      "learning_rate": 9.537442178327647e-06,
      "loss": 0.7586,
      "step": 3447
    },
    {
      "epoch": 0.49186875891583454,
      "grad_norm": 0.7646908164024353,
      "learning_rate": 9.537118650079947e-06,
      "loss": 0.8079,
      "step": 3448
    },
    {
      "epoch": 0.4920114122681883,
      "grad_norm": 0.8075964450836182,
      "learning_rate": 9.536795014219895e-06,
      "loss": 0.7877,
      "step": 3449
    },
    {
      "epoch": 0.4921540656205421,
      "grad_norm": 0.9516100287437439,
      "learning_rate": 9.536471270755167e-06,
      "loss": 0.8018,
      "step": 3450
    },
    {
      "epoch": 0.49229671897289584,
      "grad_norm": 0.942426323890686,
      "learning_rate": 9.536147419693442e-06,
      "loss": 0.6944,
      "step": 3451
    },
    {
      "epoch": 0.49243937232524965,
      "grad_norm": 1.0228946208953857,
      "learning_rate": 9.535823461042399e-06,
      "loss": 0.7664,
      "step": 3452
    },
    {
      "epoch": 0.4925820256776034,
      "grad_norm": 0.8058091998100281,
      "learning_rate": 9.535499394809723e-06,
      "loss": 0.7978,
      "step": 3453
    },
    {
      "epoch": 0.4927246790299572,
      "grad_norm": 0.7258778214454651,
      "learning_rate": 9.535175221003103e-06,
      "loss": 0.7752,
      "step": 3454
    },
    {
      "epoch": 0.492867332382311,
      "grad_norm": 1.1583764553070068,
      "learning_rate": 9.534850939630224e-06,
      "loss": 0.8383,
      "step": 3455
    },
    {
      "epoch": 0.49300998573466476,
      "grad_norm": 1.035607099533081,
      "learning_rate": 9.534526550698777e-06,
      "loss": 0.5475,
      "step": 3456
    },
    {
      "epoch": 0.49315263908701856,
      "grad_norm": 0.8877325057983398,
      "learning_rate": 9.53420205421646e-06,
      "loss": 0.7747,
      "step": 3457
    },
    {
      "epoch": 0.4932952924393723,
      "grad_norm": 0.7912925481796265,
      "learning_rate": 9.533877450190965e-06,
      "loss": 0.8278,
      "step": 3458
    },
    {
      "epoch": 0.4934379457917261,
      "grad_norm": 0.8658379912376404,
      "learning_rate": 9.533552738629993e-06,
      "loss": 0.7532,
      "step": 3459
    },
    {
      "epoch": 0.49358059914407987,
      "grad_norm": 0.8575937747955322,
      "learning_rate": 9.533227919541247e-06,
      "loss": 0.8228,
      "step": 3460
    },
    {
      "epoch": 0.4937232524964337,
      "grad_norm": 0.762713611125946,
      "learning_rate": 9.53290299293243e-06,
      "loss": 0.7547,
      "step": 3461
    },
    {
      "epoch": 0.4938659058487874,
      "grad_norm": 0.9738590121269226,
      "learning_rate": 9.532577958811247e-06,
      "loss": 0.7268,
      "step": 3462
    },
    {
      "epoch": 0.49400855920114123,
      "grad_norm": 0.7226889729499817,
      "learning_rate": 9.532252817185408e-06,
      "loss": 0.7789,
      "step": 3463
    },
    {
      "epoch": 0.49415121255349503,
      "grad_norm": 1.1516462564468384,
      "learning_rate": 9.531927568062627e-06,
      "loss": 0.7941,
      "step": 3464
    },
    {
      "epoch": 0.4942938659058488,
      "grad_norm": 1.071286916732788,
      "learning_rate": 9.531602211450617e-06,
      "loss": 0.7673,
      "step": 3465
    },
    {
      "epoch": 0.4944365192582026,
      "grad_norm": 1.0745704174041748,
      "learning_rate": 9.531276747357094e-06,
      "loss": 0.711,
      "step": 3466
    },
    {
      "epoch": 0.49457917261055634,
      "grad_norm": 0.6707049608230591,
      "learning_rate": 9.530951175789777e-06,
      "loss": 0.7474,
      "step": 3467
    },
    {
      "epoch": 0.49472182596291014,
      "grad_norm": 0.8591291308403015,
      "learning_rate": 9.530625496756392e-06,
      "loss": 0.7541,
      "step": 3468
    },
    {
      "epoch": 0.4948644793152639,
      "grad_norm": 0.8556170463562012,
      "learning_rate": 9.530299710264658e-06,
      "loss": 0.8524,
      "step": 3469
    },
    {
      "epoch": 0.4950071326676177,
      "grad_norm": 0.8599547743797302,
      "learning_rate": 9.529973816322304e-06,
      "loss": 0.8055,
      "step": 3470
    },
    {
      "epoch": 0.49514978601997145,
      "grad_norm": 0.9279347658157349,
      "learning_rate": 9.529647814937063e-06,
      "loss": 0.8195,
      "step": 3471
    },
    {
      "epoch": 0.49529243937232525,
      "grad_norm": 0.9515174627304077,
      "learning_rate": 9.529321706116663e-06,
      "loss": 0.7372,
      "step": 3472
    },
    {
      "epoch": 0.495435092724679,
      "grad_norm": 0.949888288974762,
      "learning_rate": 9.52899548986884e-06,
      "loss": 0.7677,
      "step": 3473
    },
    {
      "epoch": 0.4955777460770328,
      "grad_norm": 1.099470615386963,
      "learning_rate": 9.528669166201331e-06,
      "loss": 0.8114,
      "step": 3474
    },
    {
      "epoch": 0.4957203994293866,
      "grad_norm": 1.121291160583496,
      "learning_rate": 9.528342735121878e-06,
      "loss": 0.7872,
      "step": 3475
    },
    {
      "epoch": 0.49586305278174037,
      "grad_norm": 0.6462776064872742,
      "learning_rate": 9.52801619663822e-06,
      "loss": 0.7725,
      "step": 3476
    },
    {
      "epoch": 0.49600570613409417,
      "grad_norm": 0.8585007190704346,
      "learning_rate": 9.527689550758105e-06,
      "loss": 0.813,
      "step": 3477
    },
    {
      "epoch": 0.4961483594864479,
      "grad_norm": 1.0574462413787842,
      "learning_rate": 9.527362797489278e-06,
      "loss": 0.6989,
      "step": 3478
    },
    {
      "epoch": 0.4962910128388017,
      "grad_norm": 0.8546985387802124,
      "learning_rate": 9.52703593683949e-06,
      "loss": 0.791,
      "step": 3479
    },
    {
      "epoch": 0.4964336661911555,
      "grad_norm": 0.8837389945983887,
      "learning_rate": 9.526708968816493e-06,
      "loss": 0.693,
      "step": 3480
    },
    {
      "epoch": 0.4965763195435093,
      "grad_norm": 0.866013765335083,
      "learning_rate": 9.526381893428042e-06,
      "loss": 0.5031,
      "step": 3481
    },
    {
      "epoch": 0.49671897289586303,
      "grad_norm": 0.9066146016120911,
      "learning_rate": 9.526054710681897e-06,
      "loss": 0.7991,
      "step": 3482
    },
    {
      "epoch": 0.49686162624821684,
      "grad_norm": 0.8528669476509094,
      "learning_rate": 9.525727420585817e-06,
      "loss": 0.7852,
      "step": 3483
    },
    {
      "epoch": 0.4970042796005706,
      "grad_norm": 1.0253227949142456,
      "learning_rate": 9.525400023147563e-06,
      "loss": 0.7319,
      "step": 3484
    },
    {
      "epoch": 0.4971469329529244,
      "grad_norm": 0.9645678400993347,
      "learning_rate": 9.525072518374902e-06,
      "loss": 0.6882,
      "step": 3485
    },
    {
      "epoch": 0.4972895863052782,
      "grad_norm": 0.9075619578361511,
      "learning_rate": 9.5247449062756e-06,
      "loss": 0.7304,
      "step": 3486
    },
    {
      "epoch": 0.49743223965763195,
      "grad_norm": 0.8408627510070801,
      "learning_rate": 9.524417186857431e-06,
      "loss": 0.7668,
      "step": 3487
    },
    {
      "epoch": 0.49757489300998575,
      "grad_norm": 0.7215490341186523,
      "learning_rate": 9.524089360128164e-06,
      "loss": 0.7671,
      "step": 3488
    },
    {
      "epoch": 0.4977175463623395,
      "grad_norm": 0.8745356798171997,
      "learning_rate": 9.523761426095576e-06,
      "loss": 0.7881,
      "step": 3489
    },
    {
      "epoch": 0.4978601997146933,
      "grad_norm": 0.7559221982955933,
      "learning_rate": 9.523433384767445e-06,
      "loss": 0.7475,
      "step": 3490
    },
    {
      "epoch": 0.49800285306704706,
      "grad_norm": 1.106828212738037,
      "learning_rate": 9.523105236151551e-06,
      "loss": 0.8218,
      "step": 3491
    },
    {
      "epoch": 0.49814550641940086,
      "grad_norm": 0.991148054599762,
      "learning_rate": 9.522776980255679e-06,
      "loss": 0.7087,
      "step": 3492
    },
    {
      "epoch": 0.4982881597717546,
      "grad_norm": 1.0358657836914062,
      "learning_rate": 9.522448617087614e-06,
      "loss": 0.7952,
      "step": 3493
    },
    {
      "epoch": 0.4984308131241084,
      "grad_norm": 0.9941891431808472,
      "learning_rate": 9.522120146655143e-06,
      "loss": 0.8071,
      "step": 3494
    },
    {
      "epoch": 0.4985734664764622,
      "grad_norm": 0.8804892897605896,
      "learning_rate": 9.521791568966057e-06,
      "loss": 0.7422,
      "step": 3495
    },
    {
      "epoch": 0.498716119828816,
      "grad_norm": 0.957059919834137,
      "learning_rate": 9.52146288402815e-06,
      "loss": 0.7304,
      "step": 3496
    },
    {
      "epoch": 0.4988587731811698,
      "grad_norm": 1.0978821516036987,
      "learning_rate": 9.521134091849216e-06,
      "loss": 0.7141,
      "step": 3497
    },
    {
      "epoch": 0.49900142653352353,
      "grad_norm": 0.6598830223083496,
      "learning_rate": 9.520805192437055e-06,
      "loss": 0.8085,
      "step": 3498
    },
    {
      "epoch": 0.49914407988587733,
      "grad_norm": 1.1140689849853516,
      "learning_rate": 9.52047618579947e-06,
      "loss": 0.8339,
      "step": 3499
    },
    {
      "epoch": 0.4992867332382311,
      "grad_norm": 0.9824712872505188,
      "learning_rate": 9.520147071944259e-06,
      "loss": 0.7018,
      "step": 3500
    },
    {
      "epoch": 0.4994293865905849,
      "grad_norm": 0.9506702423095703,
      "learning_rate": 9.519817850879232e-06,
      "loss": 0.8059,
      "step": 3501
    },
    {
      "epoch": 0.49957203994293864,
      "grad_norm": 0.8838631510734558,
      "learning_rate": 9.519488522612198e-06,
      "loss": 0.82,
      "step": 3502
    },
    {
      "epoch": 0.49971469329529244,
      "grad_norm": 0.8757986426353455,
      "learning_rate": 9.519159087150967e-06,
      "loss": 0.7687,
      "step": 3503
    },
    {
      "epoch": 0.4998573466476462,
      "grad_norm": 0.865254819393158,
      "learning_rate": 9.51882954450335e-06,
      "loss": 0.8303,
      "step": 3504
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8819034099578857,
      "learning_rate": 9.518499894677168e-06,
      "loss": 0.729,
      "step": 3505
    },
    {
      "epoch": 0.5001426533523537,
      "grad_norm": 0.8352157473564148,
      "learning_rate": 9.518170137680235e-06,
      "loss": 0.8139,
      "step": 3506
    },
    {
      "epoch": 0.5002853067047076,
      "grad_norm": 0.8177391886711121,
      "learning_rate": 9.517840273520376e-06,
      "loss": 0.7847,
      "step": 3507
    },
    {
      "epoch": 0.5004279600570614,
      "grad_norm": 0.8699647784233093,
      "learning_rate": 9.517510302205412e-06,
      "loss": 0.8454,
      "step": 3508
    },
    {
      "epoch": 0.5005706134094151,
      "grad_norm": 0.888346254825592,
      "learning_rate": 9.51718022374317e-06,
      "loss": 0.8272,
      "step": 3509
    },
    {
      "epoch": 0.5007132667617689,
      "grad_norm": 1.0211471319198608,
      "learning_rate": 9.51685003814148e-06,
      "loss": 0.7535,
      "step": 3510
    },
    {
      "epoch": 0.5008559201141227,
      "grad_norm": 0.8559572100639343,
      "learning_rate": 9.516519745408175e-06,
      "loss": 0.8029,
      "step": 3511
    },
    {
      "epoch": 0.5009985734664765,
      "grad_norm": 0.8854065537452698,
      "learning_rate": 9.516189345551086e-06,
      "loss": 0.732,
      "step": 3512
    },
    {
      "epoch": 0.5011412268188302,
      "grad_norm": 0.8586737513542175,
      "learning_rate": 9.515858838578049e-06,
      "loss": 0.7802,
      "step": 3513
    },
    {
      "epoch": 0.501283880171184,
      "grad_norm": 1.0144084692001343,
      "learning_rate": 9.515528224496903e-06,
      "loss": 0.719,
      "step": 3514
    },
    {
      "epoch": 0.5014265335235378,
      "grad_norm": 0.9962849617004395,
      "learning_rate": 9.515197503315493e-06,
      "loss": 0.7886,
      "step": 3515
    },
    {
      "epoch": 0.5015691868758916,
      "grad_norm": 0.7915270328521729,
      "learning_rate": 9.514866675041664e-06,
      "loss": 0.6798,
      "step": 3516
    },
    {
      "epoch": 0.5017118402282453,
      "grad_norm": 0.8684111833572388,
      "learning_rate": 9.514535739683255e-06,
      "loss": 0.7718,
      "step": 3517
    },
    {
      "epoch": 0.5018544935805992,
      "grad_norm": 0.8695011734962463,
      "learning_rate": 9.514204697248122e-06,
      "loss": 0.7509,
      "step": 3518
    },
    {
      "epoch": 0.5019971469329529,
      "grad_norm": 0.866762101650238,
      "learning_rate": 9.513873547744113e-06,
      "loss": 0.8408,
      "step": 3519
    },
    {
      "epoch": 0.5021398002853067,
      "grad_norm": 0.7866602540016174,
      "learning_rate": 9.513542291179085e-06,
      "loss": 0.7516,
      "step": 3520
    },
    {
      "epoch": 0.5022824536376604,
      "grad_norm": 0.6946073770523071,
      "learning_rate": 9.513210927560894e-06,
      "loss": 0.7936,
      "step": 3521
    },
    {
      "epoch": 0.5024251069900143,
      "grad_norm": 0.8390775322914124,
      "learning_rate": 9.5128794568974e-06,
      "loss": 0.7568,
      "step": 3522
    },
    {
      "epoch": 0.502567760342368,
      "grad_norm": 1.0733027458190918,
      "learning_rate": 9.512547879196461e-06,
      "loss": 0.8082,
      "step": 3523
    },
    {
      "epoch": 0.5027104136947218,
      "grad_norm": 1.0854607820510864,
      "learning_rate": 9.512216194465946e-06,
      "loss": 0.7777,
      "step": 3524
    },
    {
      "epoch": 0.5028530670470756,
      "grad_norm": 0.8890421390533447,
      "learning_rate": 9.511884402713721e-06,
      "loss": 0.7799,
      "step": 3525
    },
    {
      "epoch": 0.5029957203994294,
      "grad_norm": 0.7463951706886292,
      "learning_rate": 9.511552503947654e-06,
      "loss": 0.73,
      "step": 3526
    },
    {
      "epoch": 0.5031383737517832,
      "grad_norm": 1.0551022291183472,
      "learning_rate": 9.511220498175619e-06,
      "loss": 0.7731,
      "step": 3527
    },
    {
      "epoch": 0.5032810271041369,
      "grad_norm": 1.0738459825515747,
      "learning_rate": 9.510888385405488e-06,
      "loss": 0.708,
      "step": 3528
    },
    {
      "epoch": 0.5034236804564908,
      "grad_norm": 1.0715283155441284,
      "learning_rate": 9.510556165645141e-06,
      "loss": 0.7213,
      "step": 3529
    },
    {
      "epoch": 0.5035663338088445,
      "grad_norm": 1.0010348558425903,
      "learning_rate": 9.510223838902455e-06,
      "loss": 0.7985,
      "step": 3530
    },
    {
      "epoch": 0.5037089871611983,
      "grad_norm": 1.141632080078125,
      "learning_rate": 9.509891405185314e-06,
      "loss": 0.739,
      "step": 3531
    },
    {
      "epoch": 0.503851640513552,
      "grad_norm": 0.7817306518554688,
      "learning_rate": 9.509558864501602e-06,
      "loss": 0.7156,
      "step": 3532
    },
    {
      "epoch": 0.5039942938659059,
      "grad_norm": 0.8662512302398682,
      "learning_rate": 9.509226216859208e-06,
      "loss": 0.7869,
      "step": 3533
    },
    {
      "epoch": 0.5041369472182596,
      "grad_norm": 1.1924934387207031,
      "learning_rate": 9.508893462266017e-06,
      "loss": 0.792,
      "step": 3534
    },
    {
      "epoch": 0.5042796005706134,
      "grad_norm": 0.9089195728302002,
      "learning_rate": 9.508560600729928e-06,
      "loss": 0.7608,
      "step": 3535
    },
    {
      "epoch": 0.5044222539229671,
      "grad_norm": 0.7966616749763489,
      "learning_rate": 9.508227632258831e-06,
      "loss": 0.755,
      "step": 3536
    },
    {
      "epoch": 0.504564907275321,
      "grad_norm": 0.9615272283554077,
      "learning_rate": 9.507894556860625e-06,
      "loss": 0.7772,
      "step": 3537
    },
    {
      "epoch": 0.5047075606276747,
      "grad_norm": 0.8201649785041809,
      "learning_rate": 9.50756137454321e-06,
      "loss": 0.7975,
      "step": 3538
    },
    {
      "epoch": 0.5048502139800285,
      "grad_norm": 0.8268119692802429,
      "learning_rate": 9.507228085314488e-06,
      "loss": 0.7017,
      "step": 3539
    },
    {
      "epoch": 0.5049928673323824,
      "grad_norm": 0.7588708996772766,
      "learning_rate": 9.506894689182366e-06,
      "loss": 0.5902,
      "step": 3540
    },
    {
      "epoch": 0.5051355206847361,
      "grad_norm": 0.9902154207229614,
      "learning_rate": 9.506561186154748e-06,
      "loss": 0.8953,
      "step": 3541
    },
    {
      "epoch": 0.5052781740370899,
      "grad_norm": 1.1865127086639404,
      "learning_rate": 9.506227576239546e-06,
      "loss": 0.7981,
      "step": 3542
    },
    {
      "epoch": 0.5054208273894436,
      "grad_norm": 0.9655767679214478,
      "learning_rate": 9.505893859444674e-06,
      "loss": 0.8221,
      "step": 3543
    },
    {
      "epoch": 0.5055634807417975,
      "grad_norm": 0.9593498706817627,
      "learning_rate": 9.505560035778045e-06,
      "loss": 0.5432,
      "step": 3544
    },
    {
      "epoch": 0.5057061340941512,
      "grad_norm": 0.7647463083267212,
      "learning_rate": 9.505226105247579e-06,
      "loss": 0.5542,
      "step": 3545
    },
    {
      "epoch": 0.505848787446505,
      "grad_norm": 0.8450886607170105,
      "learning_rate": 9.504892067861194e-06,
      "loss": 0.8534,
      "step": 3546
    },
    {
      "epoch": 0.5059914407988588,
      "grad_norm": 0.8688698410987854,
      "learning_rate": 9.50455792362681e-06,
      "loss": 0.7664,
      "step": 3547
    },
    {
      "epoch": 0.5061340941512126,
      "grad_norm": 0.9295186996459961,
      "learning_rate": 9.504223672552362e-06,
      "loss": 0.7595,
      "step": 3548
    },
    {
      "epoch": 0.5062767475035663,
      "grad_norm": 0.8820104002952576,
      "learning_rate": 9.50388931464577e-06,
      "loss": 0.8455,
      "step": 3549
    },
    {
      "epoch": 0.5064194008559201,
      "grad_norm": 1.1296863555908203,
      "learning_rate": 9.503554849914962e-06,
      "loss": 0.8242,
      "step": 3550
    },
    {
      "epoch": 0.5065620542082739,
      "grad_norm": 0.9853180050849915,
      "learning_rate": 9.50322027836788e-06,
      "loss": 0.7073,
      "step": 3551
    },
    {
      "epoch": 0.5067047075606277,
      "grad_norm": 0.9048812389373779,
      "learning_rate": 9.50288560001245e-06,
      "loss": 0.7121,
      "step": 3552
    },
    {
      "epoch": 0.5068473609129814,
      "grad_norm": 1.1925512552261353,
      "learning_rate": 9.502550814856616e-06,
      "loss": 0.7647,
      "step": 3553
    },
    {
      "epoch": 0.5069900142653352,
      "grad_norm": 0.823168933391571,
      "learning_rate": 9.502215922908319e-06,
      "loss": 0.7714,
      "step": 3554
    },
    {
      "epoch": 0.507132667617689,
      "grad_norm": 0.7568991184234619,
      "learning_rate": 9.501880924175499e-06,
      "loss": 0.8182,
      "step": 3555
    },
    {
      "epoch": 0.5072753209700428,
      "grad_norm": 0.8724026083946228,
      "learning_rate": 9.501545818666102e-06,
      "loss": 0.8022,
      "step": 3556
    },
    {
      "epoch": 0.5074179743223965,
      "grad_norm": 0.9513503909111023,
      "learning_rate": 9.501210606388075e-06,
      "loss": 0.7868,
      "step": 3557
    },
    {
      "epoch": 0.5075606276747504,
      "grad_norm": 0.8349027037620544,
      "learning_rate": 9.500875287349374e-06,
      "loss": 0.7011,
      "step": 3558
    },
    {
      "epoch": 0.5077032810271042,
      "grad_norm": 0.7522615194320679,
      "learning_rate": 9.500539861557944e-06,
      "loss": 0.6727,
      "step": 3559
    },
    {
      "epoch": 0.5078459343794579,
      "grad_norm": 0.9592998027801514,
      "learning_rate": 9.50020432902175e-06,
      "loss": 0.739,
      "step": 3560
    },
    {
      "epoch": 0.5079885877318117,
      "grad_norm": 0.7955388426780701,
      "learning_rate": 9.499868689748742e-06,
      "loss": 0.7906,
      "step": 3561
    },
    {
      "epoch": 0.5081312410841655,
      "grad_norm": 0.8330178260803223,
      "learning_rate": 9.499532943746885e-06,
      "loss": 0.7143,
      "step": 3562
    },
    {
      "epoch": 0.5082738944365193,
      "grad_norm": 0.9068897366523743,
      "learning_rate": 9.49919709102414e-06,
      "loss": 0.7876,
      "step": 3563
    },
    {
      "epoch": 0.508416547788873,
      "grad_norm": 1.0191758871078491,
      "learning_rate": 9.498861131588476e-06,
      "loss": 0.6956,
      "step": 3564
    },
    {
      "epoch": 0.5085592011412268,
      "grad_norm": 0.9252503514289856,
      "learning_rate": 9.49852506544786e-06,
      "loss": 0.7659,
      "step": 3565
    },
    {
      "epoch": 0.5087018544935806,
      "grad_norm": 1.0246816873550415,
      "learning_rate": 9.498188892610261e-06,
      "loss": 0.6951,
      "step": 3566
    },
    {
      "epoch": 0.5088445078459344,
      "grad_norm": 0.9582048058509827,
      "learning_rate": 9.497852613083653e-06,
      "loss": 0.8055,
      "step": 3567
    },
    {
      "epoch": 0.5089871611982881,
      "grad_norm": 0.7913559675216675,
      "learning_rate": 9.497516226876015e-06,
      "loss": 0.774,
      "step": 3568
    },
    {
      "epoch": 0.509129814550642,
      "grad_norm": 0.9547677636146545,
      "learning_rate": 9.497179733995323e-06,
      "loss": 0.7667,
      "step": 3569
    },
    {
      "epoch": 0.5092724679029957,
      "grad_norm": 0.9399577975273132,
      "learning_rate": 9.496843134449558e-06,
      "loss": 0.7585,
      "step": 3570
    },
    {
      "epoch": 0.5094151212553495,
      "grad_norm": 1.0156217813491821,
      "learning_rate": 9.496506428246703e-06,
      "loss": 0.6632,
      "step": 3571
    },
    {
      "epoch": 0.5095577746077032,
      "grad_norm": 0.8402301073074341,
      "learning_rate": 9.496169615394746e-06,
      "loss": 0.7184,
      "step": 3572
    },
    {
      "epoch": 0.5097004279600571,
      "grad_norm": 0.8508307337760925,
      "learning_rate": 9.495832695901675e-06,
      "loss": 0.7316,
      "step": 3573
    },
    {
      "epoch": 0.5098430813124109,
      "grad_norm": 0.780446469783783,
      "learning_rate": 9.495495669775481e-06,
      "loss": 0.7526,
      "step": 3574
    },
    {
      "epoch": 0.5099857346647646,
      "grad_norm": 0.9137591123580933,
      "learning_rate": 9.495158537024156e-06,
      "loss": 0.7703,
      "step": 3575
    },
    {
      "epoch": 0.5101283880171184,
      "grad_norm": 1.0749053955078125,
      "learning_rate": 9.494821297655699e-06,
      "loss": 0.7643,
      "step": 3576
    },
    {
      "epoch": 0.5102710413694722,
      "grad_norm": 0.8332220911979675,
      "learning_rate": 9.494483951678105e-06,
      "loss": 0.8825,
      "step": 3577
    },
    {
      "epoch": 0.510413694721826,
      "grad_norm": 0.8721501231193542,
      "learning_rate": 9.494146499099379e-06,
      "loss": 0.8774,
      "step": 3578
    },
    {
      "epoch": 0.5105563480741797,
      "grad_norm": 1.0534876585006714,
      "learning_rate": 9.493808939927522e-06,
      "loss": 0.7787,
      "step": 3579
    },
    {
      "epoch": 0.5106990014265336,
      "grad_norm": 1.0937044620513916,
      "learning_rate": 9.493471274170542e-06,
      "loss": 0.7382,
      "step": 3580
    },
    {
      "epoch": 0.5108416547788873,
      "grad_norm": 0.758887767791748,
      "learning_rate": 9.49313350183645e-06,
      "loss": 0.7667,
      "step": 3581
    },
    {
      "epoch": 0.5109843081312411,
      "grad_norm": 0.8439927697181702,
      "learning_rate": 9.492795622933252e-06,
      "loss": 0.777,
      "step": 3582
    },
    {
      "epoch": 0.5111269614835948,
      "grad_norm": 0.9568290710449219,
      "learning_rate": 9.492457637468966e-06,
      "loss": 0.7759,
      "step": 3583
    },
    {
      "epoch": 0.5112696148359487,
      "grad_norm": 1.1437251567840576,
      "learning_rate": 9.492119545451605e-06,
      "loss": 0.7935,
      "step": 3584
    },
    {
      "epoch": 0.5114122681883024,
      "grad_norm": 1.3443694114685059,
      "learning_rate": 9.491781346889192e-06,
      "loss": 0.774,
      "step": 3585
    },
    {
      "epoch": 0.5115549215406562,
      "grad_norm": 0.8938597440719604,
      "learning_rate": 9.491443041789746e-06,
      "loss": 0.7622,
      "step": 3586
    },
    {
      "epoch": 0.5116975748930099,
      "grad_norm": 1.1223803758621216,
      "learning_rate": 9.49110463016129e-06,
      "loss": 0.5727,
      "step": 3587
    },
    {
      "epoch": 0.5118402282453638,
      "grad_norm": 0.9260916113853455,
      "learning_rate": 9.490766112011853e-06,
      "loss": 0.778,
      "step": 3588
    },
    {
      "epoch": 0.5119828815977175,
      "grad_norm": 0.8162046670913696,
      "learning_rate": 9.490427487349465e-06,
      "loss": 0.7675,
      "step": 3589
    },
    {
      "epoch": 0.5121255349500713,
      "grad_norm": 0.7428349852561951,
      "learning_rate": 9.490088756182153e-06,
      "loss": 0.7292,
      "step": 3590
    },
    {
      "epoch": 0.5122681883024252,
      "grad_norm": 0.9647871255874634,
      "learning_rate": 9.489749918517955e-06,
      "loss": 0.7386,
      "step": 3591
    },
    {
      "epoch": 0.5124108416547789,
      "grad_norm": 0.8613197803497314,
      "learning_rate": 9.489410974364905e-06,
      "loss": 0.8379,
      "step": 3592
    },
    {
      "epoch": 0.5125534950071327,
      "grad_norm": 0.9482429027557373,
      "learning_rate": 9.489071923731043e-06,
      "loss": 0.7613,
      "step": 3593
    },
    {
      "epoch": 0.5126961483594864,
      "grad_norm": 1.1268008947372437,
      "learning_rate": 9.488732766624413e-06,
      "loss": 0.8024,
      "step": 3594
    },
    {
      "epoch": 0.5128388017118403,
      "grad_norm": 0.9092463254928589,
      "learning_rate": 9.488393503053057e-06,
      "loss": 0.7992,
      "step": 3595
    },
    {
      "epoch": 0.512981455064194,
      "grad_norm": 0.8599215149879456,
      "learning_rate": 9.488054133025022e-06,
      "loss": 0.7407,
      "step": 3596
    },
    {
      "epoch": 0.5131241084165478,
      "grad_norm": 0.8078734874725342,
      "learning_rate": 9.487714656548357e-06,
      "loss": 0.7776,
      "step": 3597
    },
    {
      "epoch": 0.5132667617689015,
      "grad_norm": 0.9748033881187439,
      "learning_rate": 9.487375073631114e-06,
      "loss": 0.801,
      "step": 3598
    },
    {
      "epoch": 0.5134094151212554,
      "grad_norm": 1.144802212715149,
      "learning_rate": 9.487035384281348e-06,
      "loss": 0.7511,
      "step": 3599
    },
    {
      "epoch": 0.5135520684736091,
      "grad_norm": 0.9737942218780518,
      "learning_rate": 9.486695588507114e-06,
      "loss": 0.7913,
      "step": 3600
    },
    {
      "epoch": 0.5136947218259629,
      "grad_norm": 0.8674310445785522,
      "learning_rate": 9.486355686316473e-06,
      "loss": 0.7976,
      "step": 3601
    },
    {
      "epoch": 0.5138373751783167,
      "grad_norm": 0.994288980960846,
      "learning_rate": 9.486015677717488e-06,
      "loss": 0.8288,
      "step": 3602
    },
    {
      "epoch": 0.5139800285306705,
      "grad_norm": 0.9822586178779602,
      "learning_rate": 9.48567556271822e-06,
      "loss": 0.5385,
      "step": 3603
    },
    {
      "epoch": 0.5141226818830242,
      "grad_norm": 0.8938551545143127,
      "learning_rate": 9.485335341326739e-06,
      "loss": 0.8137,
      "step": 3604
    },
    {
      "epoch": 0.514265335235378,
      "grad_norm": 0.8361135721206665,
      "learning_rate": 9.484995013551112e-06,
      "loss": 0.7933,
      "step": 3605
    },
    {
      "epoch": 0.5144079885877318,
      "grad_norm": 0.9457839727401733,
      "learning_rate": 9.484654579399411e-06,
      "loss": 0.7884,
      "step": 3606
    },
    {
      "epoch": 0.5145506419400856,
      "grad_norm": 0.9015101194381714,
      "learning_rate": 9.484314038879713e-06,
      "loss": 0.7497,
      "step": 3607
    },
    {
      "epoch": 0.5146932952924393,
      "grad_norm": 0.8200323581695557,
      "learning_rate": 9.483973392000094e-06,
      "loss": 0.8567,
      "step": 3608
    },
    {
      "epoch": 0.5148359486447932,
      "grad_norm": 0.7484765648841858,
      "learning_rate": 9.483632638768634e-06,
      "loss": 0.7913,
      "step": 3609
    },
    {
      "epoch": 0.514978601997147,
      "grad_norm": 0.9521093368530273,
      "learning_rate": 9.483291779193413e-06,
      "loss": 0.6828,
      "step": 3610
    },
    {
      "epoch": 0.5151212553495007,
      "grad_norm": 1.0979682207107544,
      "learning_rate": 9.482950813282518e-06,
      "loss": 0.7313,
      "step": 3611
    },
    {
      "epoch": 0.5152639087018545,
      "grad_norm": 0.8850939869880676,
      "learning_rate": 9.482609741044034e-06,
      "loss": 0.7149,
      "step": 3612
    },
    {
      "epoch": 0.5154065620542083,
      "grad_norm": 1.010465383529663,
      "learning_rate": 9.48226856248605e-06,
      "loss": 0.7398,
      "step": 3613
    },
    {
      "epoch": 0.5155492154065621,
      "grad_norm": 0.921806812286377,
      "learning_rate": 9.481927277616663e-06,
      "loss": 0.6896,
      "step": 3614
    },
    {
      "epoch": 0.5156918687589158,
      "grad_norm": 0.9806258678436279,
      "learning_rate": 9.481585886443962e-06,
      "loss": 0.7171,
      "step": 3615
    },
    {
      "epoch": 0.5158345221112696,
      "grad_norm": 0.9127873182296753,
      "learning_rate": 9.481244388976048e-06,
      "loss": 0.7802,
      "step": 3616
    },
    {
      "epoch": 0.5159771754636234,
      "grad_norm": 1.0865702629089355,
      "learning_rate": 9.480902785221018e-06,
      "loss": 0.7494,
      "step": 3617
    },
    {
      "epoch": 0.5161198288159772,
      "grad_norm": 1.1731276512145996,
      "learning_rate": 9.480561075186976e-06,
      "loss": 0.8649,
      "step": 3618
    },
    {
      "epoch": 0.5162624821683309,
      "grad_norm": 0.8446781635284424,
      "learning_rate": 9.480219258882027e-06,
      "loss": 0.7662,
      "step": 3619
    },
    {
      "epoch": 0.5164051355206848,
      "grad_norm": 0.8926602005958557,
      "learning_rate": 9.479877336314277e-06,
      "loss": 0.8907,
      "step": 3620
    },
    {
      "epoch": 0.5165477888730385,
      "grad_norm": 0.9072000980377197,
      "learning_rate": 9.479535307491836e-06,
      "loss": 0.7547,
      "step": 3621
    },
    {
      "epoch": 0.5166904422253923,
      "grad_norm": 0.8798696994781494,
      "learning_rate": 9.479193172422818e-06,
      "loss": 0.7849,
      "step": 3622
    },
    {
      "epoch": 0.516833095577746,
      "grad_norm": 1.029890537261963,
      "learning_rate": 9.478850931115335e-06,
      "loss": 0.7595,
      "step": 3623
    },
    {
      "epoch": 0.5169757489300999,
      "grad_norm": 0.9974260926246643,
      "learning_rate": 9.478508583577506e-06,
      "loss": 0.7973,
      "step": 3624
    },
    {
      "epoch": 0.5171184022824536,
      "grad_norm": 0.672558069229126,
      "learning_rate": 9.478166129817449e-06,
      "loss": 0.7986,
      "step": 3625
    },
    {
      "epoch": 0.5172610556348074,
      "grad_norm": 0.910285472869873,
      "learning_rate": 9.477823569843291e-06,
      "loss": 0.7627,
      "step": 3626
    },
    {
      "epoch": 0.5174037089871611,
      "grad_norm": 0.7758559584617615,
      "learning_rate": 9.477480903663152e-06,
      "loss": 0.7544,
      "step": 3627
    },
    {
      "epoch": 0.517546362339515,
      "grad_norm": 1.2214888334274292,
      "learning_rate": 9.477138131285164e-06,
      "loss": 0.8018,
      "step": 3628
    },
    {
      "epoch": 0.5176890156918688,
      "grad_norm": 0.8100362420082092,
      "learning_rate": 9.476795252717452e-06,
      "loss": 0.7418,
      "step": 3629
    },
    {
      "epoch": 0.5178316690442225,
      "grad_norm": 0.8331372141838074,
      "learning_rate": 9.476452267968151e-06,
      "loss": 0.7488,
      "step": 3630
    },
    {
      "epoch": 0.5179743223965764,
      "grad_norm": 0.8265190720558167,
      "learning_rate": 9.476109177045395e-06,
      "loss": 0.8099,
      "step": 3631
    },
    {
      "epoch": 0.5181169757489301,
      "grad_norm": 0.8415448069572449,
      "learning_rate": 9.475765979957325e-06,
      "loss": 0.7815,
      "step": 3632
    },
    {
      "epoch": 0.5182596291012839,
      "grad_norm": 0.7178728580474854,
      "learning_rate": 9.475422676712076e-06,
      "loss": 0.8397,
      "step": 3633
    },
    {
      "epoch": 0.5184022824536376,
      "grad_norm": 0.9677311778068542,
      "learning_rate": 9.475079267317795e-06,
      "loss": 0.7822,
      "step": 3634
    },
    {
      "epoch": 0.5185449358059915,
      "grad_norm": 1.090571641921997,
      "learning_rate": 9.474735751782624e-06,
      "loss": 0.788,
      "step": 3635
    },
    {
      "epoch": 0.5186875891583452,
      "grad_norm": 0.9308180212974548,
      "learning_rate": 9.474392130114711e-06,
      "loss": 0.7907,
      "step": 3636
    },
    {
      "epoch": 0.518830242510699,
      "grad_norm": 0.8341872692108154,
      "learning_rate": 9.474048402322207e-06,
      "loss": 0.6871,
      "step": 3637
    },
    {
      "epoch": 0.5189728958630527,
      "grad_norm": 0.9667650461196899,
      "learning_rate": 9.473704568413265e-06,
      "loss": 0.7448,
      "step": 3638
    },
    {
      "epoch": 0.5191155492154066,
      "grad_norm": 0.9378452897071838,
      "learning_rate": 9.473360628396038e-06,
      "loss": 0.814,
      "step": 3639
    },
    {
      "epoch": 0.5192582025677603,
      "grad_norm": 0.9220533967018127,
      "learning_rate": 9.473016582278687e-06,
      "loss": 0.7933,
      "step": 3640
    },
    {
      "epoch": 0.5194008559201141,
      "grad_norm": 0.6695970892906189,
      "learning_rate": 9.472672430069368e-06,
      "loss": 0.8177,
      "step": 3641
    },
    {
      "epoch": 0.519543509272468,
      "grad_norm": 1.0739068984985352,
      "learning_rate": 9.472328171776246e-06,
      "loss": 0.8237,
      "step": 3642
    },
    {
      "epoch": 0.5196861626248217,
      "grad_norm": 0.6684971451759338,
      "learning_rate": 9.471983807407488e-06,
      "loss": 0.8188,
      "step": 3643
    },
    {
      "epoch": 0.5198288159771755,
      "grad_norm": 0.7271811366081238,
      "learning_rate": 9.471639336971258e-06,
      "loss": 0.8048,
      "step": 3644
    },
    {
      "epoch": 0.5199714693295292,
      "grad_norm": 0.9835609197616577,
      "learning_rate": 9.471294760475728e-06,
      "loss": 0.8445,
      "step": 3645
    },
    {
      "epoch": 0.5201141226818831,
      "grad_norm": 0.6809604167938232,
      "learning_rate": 9.47095007792907e-06,
      "loss": 0.8051,
      "step": 3646
    },
    {
      "epoch": 0.5202567760342368,
      "grad_norm": 1.2839874029159546,
      "learning_rate": 9.470605289339462e-06,
      "loss": 0.5462,
      "step": 3647
    },
    {
      "epoch": 0.5203994293865906,
      "grad_norm": 0.7694956064224243,
      "learning_rate": 9.470260394715079e-06,
      "loss": 0.7472,
      "step": 3648
    },
    {
      "epoch": 0.5205420827389443,
      "grad_norm": 0.8619051575660706,
      "learning_rate": 9.469915394064101e-06,
      "loss": 0.7096,
      "step": 3649
    },
    {
      "epoch": 0.5206847360912982,
      "grad_norm": 1.0383868217468262,
      "learning_rate": 9.469570287394713e-06,
      "loss": 0.7074,
      "step": 3650
    },
    {
      "epoch": 0.5208273894436519,
      "grad_norm": 0.8686227798461914,
      "learning_rate": 9.469225074715098e-06,
      "loss": 0.7352,
      "step": 3651
    },
    {
      "epoch": 0.5209700427960057,
      "grad_norm": 0.8512762784957886,
      "learning_rate": 9.468879756033446e-06,
      "loss": 0.737,
      "step": 3652
    },
    {
      "epoch": 0.5211126961483595,
      "grad_norm": 0.7683886885643005,
      "learning_rate": 9.468534331357946e-06,
      "loss": 0.773,
      "step": 3653
    },
    {
      "epoch": 0.5212553495007133,
      "grad_norm": 1.0536611080169678,
      "learning_rate": 9.46818880069679e-06,
      "loss": 0.7079,
      "step": 3654
    },
    {
      "epoch": 0.521398002853067,
      "grad_norm": 0.8594650626182556,
      "learning_rate": 9.467843164058176e-06,
      "loss": 0.7836,
      "step": 3655
    },
    {
      "epoch": 0.5215406562054208,
      "grad_norm": 0.8585334420204163,
      "learning_rate": 9.4674974214503e-06,
      "loss": 0.7949,
      "step": 3656
    },
    {
      "epoch": 0.5216833095577746,
      "grad_norm": 0.9271411299705505,
      "learning_rate": 9.467151572881363e-06,
      "loss": 0.8577,
      "step": 3657
    },
    {
      "epoch": 0.5218259629101284,
      "grad_norm": 0.7891442179679871,
      "learning_rate": 9.466805618359567e-06,
      "loss": 0.7561,
      "step": 3658
    },
    {
      "epoch": 0.5219686162624821,
      "grad_norm": 0.9075076580047607,
      "learning_rate": 9.466459557893119e-06,
      "loss": 0.8193,
      "step": 3659
    },
    {
      "epoch": 0.5221112696148359,
      "grad_norm": 1.1432865858078003,
      "learning_rate": 9.466113391490225e-06,
      "loss": 0.7082,
      "step": 3660
    },
    {
      "epoch": 0.5222539229671898,
      "grad_norm": 0.8940004110336304,
      "learning_rate": 9.465767119159097e-06,
      "loss": 0.7857,
      "step": 3661
    },
    {
      "epoch": 0.5223965763195435,
      "grad_norm": 1.145477056503296,
      "learning_rate": 9.465420740907948e-06,
      "loss": 0.7813,
      "step": 3662
    },
    {
      "epoch": 0.5225392296718973,
      "grad_norm": 0.771138072013855,
      "learning_rate": 9.465074256744993e-06,
      "loss": 0.7888,
      "step": 3663
    },
    {
      "epoch": 0.5226818830242511,
      "grad_norm": 1.0088539123535156,
      "learning_rate": 9.464727666678448e-06,
      "loss": 0.732,
      "step": 3664
    },
    {
      "epoch": 0.5228245363766049,
      "grad_norm": 0.7582293748855591,
      "learning_rate": 9.464380970716538e-06,
      "loss": 0.8644,
      "step": 3665
    },
    {
      "epoch": 0.5229671897289586,
      "grad_norm": 1.0350309610366821,
      "learning_rate": 9.464034168867483e-06,
      "loss": 0.72,
      "step": 3666
    },
    {
      "epoch": 0.5231098430813124,
      "grad_norm": 0.9733157157897949,
      "learning_rate": 9.463687261139507e-06,
      "loss": 0.8307,
      "step": 3667
    },
    {
      "epoch": 0.5232524964336662,
      "grad_norm": 1.0166035890579224,
      "learning_rate": 9.463340247540841e-06,
      "loss": 0.7685,
      "step": 3668
    },
    {
      "epoch": 0.52339514978602,
      "grad_norm": 1.02400803565979,
      "learning_rate": 9.462993128079716e-06,
      "loss": 0.8393,
      "step": 3669
    },
    {
      "epoch": 0.5235378031383737,
      "grad_norm": 0.7539787888526917,
      "learning_rate": 9.462645902764362e-06,
      "loss": 0.766,
      "step": 3670
    },
    {
      "epoch": 0.5236804564907276,
      "grad_norm": 0.8282262086868286,
      "learning_rate": 9.462298571603017e-06,
      "loss": 0.7855,
      "step": 3671
    },
    {
      "epoch": 0.5238231098430813,
      "grad_norm": 0.8732316493988037,
      "learning_rate": 9.461951134603918e-06,
      "loss": 0.792,
      "step": 3672
    },
    {
      "epoch": 0.5239657631954351,
      "grad_norm": 0.9385262131690979,
      "learning_rate": 9.461603591775304e-06,
      "loss": 0.7031,
      "step": 3673
    },
    {
      "epoch": 0.5241084165477888,
      "grad_norm": 0.8600476384162903,
      "learning_rate": 9.461255943125422e-06,
      "loss": 0.7815,
      "step": 3674
    },
    {
      "epoch": 0.5242510699001427,
      "grad_norm": 0.8772199749946594,
      "learning_rate": 9.460908188662516e-06,
      "loss": 0.6744,
      "step": 3675
    },
    {
      "epoch": 0.5243937232524964,
      "grad_norm": 0.7366943955421448,
      "learning_rate": 9.460560328394832e-06,
      "loss": 0.7759,
      "step": 3676
    },
    {
      "epoch": 0.5245363766048502,
      "grad_norm": 1.5101613998413086,
      "learning_rate": 9.460212362330623e-06,
      "loss": 0.8221,
      "step": 3677
    },
    {
      "epoch": 0.524679029957204,
      "grad_norm": 1.0198205709457397,
      "learning_rate": 9.45986429047814e-06,
      "loss": 0.7409,
      "step": 3678
    },
    {
      "epoch": 0.5248216833095578,
      "grad_norm": 0.8249444365501404,
      "learning_rate": 9.459516112845643e-06,
      "loss": 0.8117,
      "step": 3679
    },
    {
      "epoch": 0.5249643366619116,
      "grad_norm": 0.9865692257881165,
      "learning_rate": 9.459167829441385e-06,
      "loss": 0.7935,
      "step": 3680
    },
    {
      "epoch": 0.5251069900142653,
      "grad_norm": 0.984447181224823,
      "learning_rate": 9.45881944027363e-06,
      "loss": 0.7764,
      "step": 3681
    },
    {
      "epoch": 0.5252496433666192,
      "grad_norm": 0.8019534349441528,
      "learning_rate": 9.45847094535064e-06,
      "loss": 0.8343,
      "step": 3682
    },
    {
      "epoch": 0.5253922967189729,
      "grad_norm": 0.9061942100524902,
      "learning_rate": 9.45812234468068e-06,
      "loss": 0.791,
      "step": 3683
    },
    {
      "epoch": 0.5255349500713267,
      "grad_norm": 1.1901942491531372,
      "learning_rate": 9.457773638272018e-06,
      "loss": 0.7438,
      "step": 3684
    },
    {
      "epoch": 0.5256776034236804,
      "grad_norm": 0.6756222248077393,
      "learning_rate": 9.457424826132925e-06,
      "loss": 0.7347,
      "step": 3685
    },
    {
      "epoch": 0.5258202567760343,
      "grad_norm": 0.8619942665100098,
      "learning_rate": 9.457075908271677e-06,
      "loss": 0.8252,
      "step": 3686
    },
    {
      "epoch": 0.525962910128388,
      "grad_norm": 0.8412014245986938,
      "learning_rate": 9.456726884696548e-06,
      "loss": 0.7471,
      "step": 3687
    },
    {
      "epoch": 0.5261055634807418,
      "grad_norm": 0.8969194293022156,
      "learning_rate": 9.456377755415814e-06,
      "loss": 0.7682,
      "step": 3688
    },
    {
      "epoch": 0.5262482168330955,
      "grad_norm": 0.9360155463218689,
      "learning_rate": 9.456028520437758e-06,
      "loss": 0.7758,
      "step": 3689
    },
    {
      "epoch": 0.5263908701854494,
      "grad_norm": 0.9619754552841187,
      "learning_rate": 9.455679179770662e-06,
      "loss": 0.7107,
      "step": 3690
    },
    {
      "epoch": 0.5265335235378031,
      "grad_norm": 1.0945255756378174,
      "learning_rate": 9.455329733422814e-06,
      "loss": 0.7382,
      "step": 3691
    },
    {
      "epoch": 0.5266761768901569,
      "grad_norm": 0.9404793977737427,
      "learning_rate": 9.454980181402499e-06,
      "loss": 0.7799,
      "step": 3692
    },
    {
      "epoch": 0.5268188302425107,
      "grad_norm": 1.1391710042953491,
      "learning_rate": 9.45463052371801e-06,
      "loss": 0.7617,
      "step": 3693
    },
    {
      "epoch": 0.5269614835948645,
      "grad_norm": 1.1695657968521118,
      "learning_rate": 9.45428076037764e-06,
      "loss": 0.7959,
      "step": 3694
    },
    {
      "epoch": 0.5271041369472182,
      "grad_norm": 0.9902835488319397,
      "learning_rate": 9.453930891389684e-06,
      "loss": 0.6934,
      "step": 3695
    },
    {
      "epoch": 0.527246790299572,
      "grad_norm": 1.1658791303634644,
      "learning_rate": 9.45358091676244e-06,
      "loss": 0.7592,
      "step": 3696
    },
    {
      "epoch": 0.5273894436519259,
      "grad_norm": 0.8045031428337097,
      "learning_rate": 9.45323083650421e-06,
      "loss": 0.8133,
      "step": 3697
    },
    {
      "epoch": 0.5275320970042796,
      "grad_norm": 0.930515706539154,
      "learning_rate": 9.452880650623297e-06,
      "loss": 0.7203,
      "step": 3698
    },
    {
      "epoch": 0.5276747503566334,
      "grad_norm": 1.0128291845321655,
      "learning_rate": 9.452530359128007e-06,
      "loss": 0.7574,
      "step": 3699
    },
    {
      "epoch": 0.5278174037089871,
      "grad_norm": 0.7646418809890747,
      "learning_rate": 9.452179962026647e-06,
      "loss": 0.7076,
      "step": 3700
    },
    {
      "epoch": 0.527960057061341,
      "grad_norm": 0.9137625098228455,
      "learning_rate": 9.451829459327528e-06,
      "loss": 0.5396,
      "step": 3701
    },
    {
      "epoch": 0.5281027104136947,
      "grad_norm": 0.9639657139778137,
      "learning_rate": 9.451478851038965e-06,
      "loss": 0.7742,
      "step": 3702
    },
    {
      "epoch": 0.5282453637660485,
      "grad_norm": 1.0482797622680664,
      "learning_rate": 9.451128137169273e-06,
      "loss": 0.7571,
      "step": 3703
    },
    {
      "epoch": 0.5283880171184023,
      "grad_norm": 1.0665918588638306,
      "learning_rate": 9.450777317726768e-06,
      "loss": 0.5858,
      "step": 3704
    },
    {
      "epoch": 0.5285306704707561,
      "grad_norm": 0.9101869463920593,
      "learning_rate": 9.450426392719773e-06,
      "loss": 0.7022,
      "step": 3705
    },
    {
      "epoch": 0.5286733238231098,
      "grad_norm": 0.9322646856307983,
      "learning_rate": 9.450075362156614e-06,
      "loss": 0.7925,
      "step": 3706
    },
    {
      "epoch": 0.5288159771754636,
      "grad_norm": 0.9340315461158752,
      "learning_rate": 9.449724226045612e-06,
      "loss": 0.6898,
      "step": 3707
    },
    {
      "epoch": 0.5289586305278174,
      "grad_norm": 0.9280468821525574,
      "learning_rate": 9.449372984395096e-06,
      "loss": 0.799,
      "step": 3708
    },
    {
      "epoch": 0.5291012838801712,
      "grad_norm": 0.8312258124351501,
      "learning_rate": 9.4490216372134e-06,
      "loss": 0.7917,
      "step": 3709
    },
    {
      "epoch": 0.5292439372325249,
      "grad_norm": 0.7946302890777588,
      "learning_rate": 9.448670184508854e-06,
      "loss": 0.8267,
      "step": 3710
    },
    {
      "epoch": 0.5293865905848787,
      "grad_norm": 0.8247292637825012,
      "learning_rate": 9.448318626289796e-06,
      "loss": 0.865,
      "step": 3711
    },
    {
      "epoch": 0.5295292439372326,
      "grad_norm": 0.7541313171386719,
      "learning_rate": 9.447966962564564e-06,
      "loss": 0.7116,
      "step": 3712
    },
    {
      "epoch": 0.5296718972895863,
      "grad_norm": 0.9139978289604187,
      "learning_rate": 9.447615193341498e-06,
      "loss": 0.7635,
      "step": 3713
    },
    {
      "epoch": 0.52981455064194,
      "grad_norm": 0.7234423160552979,
      "learning_rate": 9.447263318628942e-06,
      "loss": 0.5395,
      "step": 3714
    },
    {
      "epoch": 0.5299572039942939,
      "grad_norm": 0.8707560300827026,
      "learning_rate": 9.44691133843524e-06,
      "loss": 0.6405,
      "step": 3715
    },
    {
      "epoch": 0.5300998573466477,
      "grad_norm": 0.6593236327171326,
      "learning_rate": 9.446559252768745e-06,
      "loss": 0.7982,
      "step": 3716
    },
    {
      "epoch": 0.5302425106990014,
      "grad_norm": 0.9893619418144226,
      "learning_rate": 9.446207061637802e-06,
      "loss": 0.8119,
      "step": 3717
    },
    {
      "epoch": 0.5303851640513552,
      "grad_norm": 0.8062154650688171,
      "learning_rate": 9.445854765050768e-06,
      "loss": 0.7776,
      "step": 3718
    },
    {
      "epoch": 0.530527817403709,
      "grad_norm": 0.7974979877471924,
      "learning_rate": 9.445502363015999e-06,
      "loss": 0.8263,
      "step": 3719
    },
    {
      "epoch": 0.5306704707560628,
      "grad_norm": 0.8471797704696655,
      "learning_rate": 9.44514985554185e-06,
      "loss": 0.7055,
      "step": 3720
    },
    {
      "epoch": 0.5308131241084165,
      "grad_norm": 0.9826592206954956,
      "learning_rate": 9.444797242636686e-06,
      "loss": 0.7937,
      "step": 3721
    },
    {
      "epoch": 0.5309557774607703,
      "grad_norm": 0.8173101544380188,
      "learning_rate": 9.444444524308869e-06,
      "loss": 0.7614,
      "step": 3722
    },
    {
      "epoch": 0.5310984308131241,
      "grad_norm": 0.8010444641113281,
      "learning_rate": 9.444091700566761e-06,
      "loss": 0.7716,
      "step": 3723
    },
    {
      "epoch": 0.5312410841654779,
      "grad_norm": 0.9752053022384644,
      "learning_rate": 9.443738771418738e-06,
      "loss": 0.7182,
      "step": 3724
    },
    {
      "epoch": 0.5313837375178316,
      "grad_norm": 0.9180756211280823,
      "learning_rate": 9.443385736873166e-06,
      "loss": 0.7389,
      "step": 3725
    },
    {
      "epoch": 0.5315263908701855,
      "grad_norm": 0.9787564873695374,
      "learning_rate": 9.443032596938416e-06,
      "loss": 0.7067,
      "step": 3726
    },
    {
      "epoch": 0.5316690442225392,
      "grad_norm": 1.2496058940887451,
      "learning_rate": 9.442679351622868e-06,
      "loss": 0.7819,
      "step": 3727
    },
    {
      "epoch": 0.531811697574893,
      "grad_norm": 1.0999810695648193,
      "learning_rate": 9.442326000934901e-06,
      "loss": 0.6903,
      "step": 3728
    },
    {
      "epoch": 0.5319543509272467,
      "grad_norm": 0.9185510277748108,
      "learning_rate": 9.441972544882892e-06,
      "loss": 0.7864,
      "step": 3729
    },
    {
      "epoch": 0.5320970042796006,
      "grad_norm": 1.1549924612045288,
      "learning_rate": 9.441618983475226e-06,
      "loss": 0.7518,
      "step": 3730
    },
    {
      "epoch": 0.5322396576319544,
      "grad_norm": 0.7317920923233032,
      "learning_rate": 9.44126531672029e-06,
      "loss": 0.7407,
      "step": 3731
    },
    {
      "epoch": 0.5323823109843081,
      "grad_norm": 0.9729320406913757,
      "learning_rate": 9.44091154462647e-06,
      "loss": 0.8163,
      "step": 3732
    },
    {
      "epoch": 0.532524964336662,
      "grad_norm": 0.8830687999725342,
      "learning_rate": 9.440557667202159e-06,
      "loss": 0.7556,
      "step": 3733
    },
    {
      "epoch": 0.5326676176890157,
      "grad_norm": 1.000052809715271,
      "learning_rate": 9.44020368445575e-06,
      "loss": 0.5106,
      "step": 3734
    },
    {
      "epoch": 0.5328102710413695,
      "grad_norm": 0.710538923740387,
      "learning_rate": 9.43984959639564e-06,
      "loss": 0.7229,
      "step": 3735
    },
    {
      "epoch": 0.5329529243937232,
      "grad_norm": 1.032488465309143,
      "learning_rate": 9.439495403030224e-06,
      "loss": 0.7088,
      "step": 3736
    },
    {
      "epoch": 0.5330955777460771,
      "grad_norm": 1.1073076725006104,
      "learning_rate": 9.439141104367906e-06,
      "loss": 0.7845,
      "step": 3737
    },
    {
      "epoch": 0.5332382310984308,
      "grad_norm": 1.2522554397583008,
      "learning_rate": 9.438786700417086e-06,
      "loss": 0.8004,
      "step": 3738
    },
    {
      "epoch": 0.5333808844507846,
      "grad_norm": 0.892382800579071,
      "learning_rate": 9.438432191186173e-06,
      "loss": 0.7946,
      "step": 3739
    },
    {
      "epoch": 0.5335235378031383,
      "grad_norm": 0.9575504660606384,
      "learning_rate": 9.438077576683575e-06,
      "loss": 0.7192,
      "step": 3740
    },
    {
      "epoch": 0.5336661911554922,
      "grad_norm": 1.0557219982147217,
      "learning_rate": 9.437722856917701e-06,
      "loss": 0.7357,
      "step": 3741
    },
    {
      "epoch": 0.5338088445078459,
      "grad_norm": 0.9808925986289978,
      "learning_rate": 9.437368031896965e-06,
      "loss": 0.8335,
      "step": 3742
    },
    {
      "epoch": 0.5339514978601997,
      "grad_norm": 0.8837986588478088,
      "learning_rate": 9.437013101629785e-06,
      "loss": 0.7742,
      "step": 3743
    },
    {
      "epoch": 0.5340941512125535,
      "grad_norm": 0.9830921292304993,
      "learning_rate": 9.436658066124575e-06,
      "loss": 0.8071,
      "step": 3744
    },
    {
      "epoch": 0.5342368045649073,
      "grad_norm": 0.9636303186416626,
      "learning_rate": 9.43630292538976e-06,
      "loss": 0.7329,
      "step": 3745
    },
    {
      "epoch": 0.534379457917261,
      "grad_norm": 0.7886168956756592,
      "learning_rate": 9.435947679433761e-06,
      "loss": 0.7752,
      "step": 3746
    },
    {
      "epoch": 0.5345221112696148,
      "grad_norm": 1.0022714138031006,
      "learning_rate": 9.435592328265003e-06,
      "loss": 0.7812,
      "step": 3747
    },
    {
      "epoch": 0.5346647646219687,
      "grad_norm": 1.0230801105499268,
      "learning_rate": 9.435236871891917e-06,
      "loss": 0.8012,
      "step": 3748
    },
    {
      "epoch": 0.5348074179743224,
      "grad_norm": 0.7865760326385498,
      "learning_rate": 9.434881310322932e-06,
      "loss": 0.7669,
      "step": 3749
    },
    {
      "epoch": 0.5349500713266762,
      "grad_norm": 0.8765584826469421,
      "learning_rate": 9.434525643566481e-06,
      "loss": 0.8231,
      "step": 3750
    },
    {
      "epoch": 0.5350927246790299,
      "grad_norm": 1.0778067111968994,
      "learning_rate": 9.434169871631003e-06,
      "loss": 0.7504,
      "step": 3751
    },
    {
      "epoch": 0.5352353780313838,
      "grad_norm": 0.9879114627838135,
      "learning_rate": 9.433813994524933e-06,
      "loss": 0.7609,
      "step": 3752
    },
    {
      "epoch": 0.5353780313837375,
      "grad_norm": 0.7823513746261597,
      "learning_rate": 9.43345801225671e-06,
      "loss": 0.8272,
      "step": 3753
    },
    {
      "epoch": 0.5355206847360913,
      "grad_norm": 0.9532626271247864,
      "learning_rate": 9.433101924834782e-06,
      "loss": 0.74,
      "step": 3754
    },
    {
      "epoch": 0.5356633380884451,
      "grad_norm": 0.7085595726966858,
      "learning_rate": 9.432745732267593e-06,
      "loss": 0.7988,
      "step": 3755
    },
    {
      "epoch": 0.5358059914407989,
      "grad_norm": 0.8867372870445251,
      "learning_rate": 9.432389434563589e-06,
      "loss": 0.7765,
      "step": 3756
    },
    {
      "epoch": 0.5359486447931526,
      "grad_norm": 0.8110182285308838,
      "learning_rate": 9.432033031731222e-06,
      "loss": 0.7357,
      "step": 3757
    },
    {
      "epoch": 0.5360912981455064,
      "grad_norm": 0.8507335782051086,
      "learning_rate": 9.431676523778948e-06,
      "loss": 0.75,
      "step": 3758
    },
    {
      "epoch": 0.5362339514978602,
      "grad_norm": 0.9301802515983582,
      "learning_rate": 9.43131991071522e-06,
      "loss": 0.7358,
      "step": 3759
    },
    {
      "epoch": 0.536376604850214,
      "grad_norm": 0.8076273798942566,
      "learning_rate": 9.430963192548497e-06,
      "loss": 0.7378,
      "step": 3760
    },
    {
      "epoch": 0.5365192582025677,
      "grad_norm": 0.9314234852790833,
      "learning_rate": 9.430606369287238e-06,
      "loss": 0.7996,
      "step": 3761
    },
    {
      "epoch": 0.5366619115549215,
      "grad_norm": 0.9800743460655212,
      "learning_rate": 9.43024944093991e-06,
      "loss": 0.759,
      "step": 3762
    },
    {
      "epoch": 0.5368045649072754,
      "grad_norm": 0.8995145559310913,
      "learning_rate": 9.429892407514973e-06,
      "loss": 0.753,
      "step": 3763
    },
    {
      "epoch": 0.5369472182596291,
      "grad_norm": 0.880490779876709,
      "learning_rate": 9.4295352690209e-06,
      "loss": 0.7698,
      "step": 3764
    },
    {
      "epoch": 0.5370898716119828,
      "grad_norm": 0.9529716968536377,
      "learning_rate": 9.429178025466162e-06,
      "loss": 0.7406,
      "step": 3765
    },
    {
      "epoch": 0.5372325249643367,
      "grad_norm": 0.8968760371208191,
      "learning_rate": 9.428820676859229e-06,
      "loss": 0.7051,
      "step": 3766
    },
    {
      "epoch": 0.5373751783166905,
      "grad_norm": 0.8870418667793274,
      "learning_rate": 9.428463223208579e-06,
      "loss": 0.7241,
      "step": 3767
    },
    {
      "epoch": 0.5375178316690442,
      "grad_norm": 0.8251546621322632,
      "learning_rate": 9.428105664522688e-06,
      "loss": 0.7874,
      "step": 3768
    },
    {
      "epoch": 0.537660485021398,
      "grad_norm": 0.9303982853889465,
      "learning_rate": 9.427748000810037e-06,
      "loss": 0.7694,
      "step": 3769
    },
    {
      "epoch": 0.5378031383737518,
      "grad_norm": 0.8409507274627686,
      "learning_rate": 9.427390232079113e-06,
      "loss": 0.5782,
      "step": 3770
    },
    {
      "epoch": 0.5379457917261056,
      "grad_norm": 0.9384737610816956,
      "learning_rate": 9.427032358338397e-06,
      "loss": 0.6343,
      "step": 3771
    },
    {
      "epoch": 0.5380884450784593,
      "grad_norm": 0.7768946290016174,
      "learning_rate": 9.426674379596378e-06,
      "loss": 0.5803,
      "step": 3772
    },
    {
      "epoch": 0.5382310984308131,
      "grad_norm": 0.9682494401931763,
      "learning_rate": 9.426316295861551e-06,
      "loss": 0.7583,
      "step": 3773
    },
    {
      "epoch": 0.5383737517831669,
      "grad_norm": 1.0426535606384277,
      "learning_rate": 9.425958107142403e-06,
      "loss": 0.6997,
      "step": 3774
    },
    {
      "epoch": 0.5385164051355207,
      "grad_norm": 1.061850666999817,
      "learning_rate": 9.42559981344743e-06,
      "loss": 0.7664,
      "step": 3775
    },
    {
      "epoch": 0.5386590584878744,
      "grad_norm": 1.269931435585022,
      "learning_rate": 9.425241414785136e-06,
      "loss": 0.6823,
      "step": 3776
    },
    {
      "epoch": 0.5388017118402283,
      "grad_norm": 0.648510754108429,
      "learning_rate": 9.424882911164017e-06,
      "loss": 0.5664,
      "step": 3777
    },
    {
      "epoch": 0.538944365192582,
      "grad_norm": 0.9643648862838745,
      "learning_rate": 9.424524302592575e-06,
      "loss": 0.7396,
      "step": 3778
    },
    {
      "epoch": 0.5390870185449358,
      "grad_norm": 0.9548050761222839,
      "learning_rate": 9.424165589079319e-06,
      "loss": 0.7476,
      "step": 3779
    },
    {
      "epoch": 0.5392296718972895,
      "grad_norm": 1.1638333797454834,
      "learning_rate": 9.423806770632755e-06,
      "loss": 0.8011,
      "step": 3780
    },
    {
      "epoch": 0.5393723252496434,
      "grad_norm": 0.7156169414520264,
      "learning_rate": 9.423447847261394e-06,
      "loss": 0.5816,
      "step": 3781
    },
    {
      "epoch": 0.5395149786019972,
      "grad_norm": 0.8134765028953552,
      "learning_rate": 9.42308881897375e-06,
      "loss": 0.7886,
      "step": 3782
    },
    {
      "epoch": 0.5396576319543509,
      "grad_norm": 0.7660357356071472,
      "learning_rate": 9.422729685778336e-06,
      "loss": 0.5448,
      "step": 3783
    },
    {
      "epoch": 0.5398002853067047,
      "grad_norm": 0.6888023018836975,
      "learning_rate": 9.422370447683673e-06,
      "loss": 0.748,
      "step": 3784
    },
    {
      "epoch": 0.5399429386590585,
      "grad_norm": 0.6250286102294922,
      "learning_rate": 9.422011104698279e-06,
      "loss": 0.623,
      "step": 3785
    },
    {
      "epoch": 0.5400855920114123,
      "grad_norm": 0.8474677801132202,
      "learning_rate": 9.42165165683068e-06,
      "loss": 0.7005,
      "step": 3786
    },
    {
      "epoch": 0.540228245363766,
      "grad_norm": 0.7340326905250549,
      "learning_rate": 9.421292104089396e-06,
      "loss": 0.8158,
      "step": 3787
    },
    {
      "epoch": 0.5403708987161199,
      "grad_norm": 0.8920254707336426,
      "learning_rate": 9.42093244648296e-06,
      "loss": 0.7517,
      "step": 3788
    },
    {
      "epoch": 0.5405135520684736,
      "grad_norm": 1.0397709608078003,
      "learning_rate": 9.420572684019902e-06,
      "loss": 0.7797,
      "step": 3789
    },
    {
      "epoch": 0.5406562054208274,
      "grad_norm": 0.9333937764167786,
      "learning_rate": 9.420212816708752e-06,
      "loss": 0.6948,
      "step": 3790
    },
    {
      "epoch": 0.5407988587731811,
      "grad_norm": 0.6868544816970825,
      "learning_rate": 9.419852844558049e-06,
      "loss": 0.583,
      "step": 3791
    },
    {
      "epoch": 0.540941512125535,
      "grad_norm": 0.8502908945083618,
      "learning_rate": 9.419492767576329e-06,
      "loss": 0.855,
      "step": 3792
    },
    {
      "epoch": 0.5410841654778887,
      "grad_norm": 0.9610970616340637,
      "learning_rate": 9.41913258577213e-06,
      "loss": 0.7259,
      "step": 3793
    },
    {
      "epoch": 0.5412268188302425,
      "grad_norm": 0.7129231691360474,
      "learning_rate": 9.418772299153999e-06,
      "loss": 0.8318,
      "step": 3794
    },
    {
      "epoch": 0.5413694721825963,
      "grad_norm": 1.0146222114562988,
      "learning_rate": 9.41841190773048e-06,
      "loss": 0.7347,
      "step": 3795
    },
    {
      "epoch": 0.5415121255349501,
      "grad_norm": 0.7784373760223389,
      "learning_rate": 9.41805141151012e-06,
      "loss": 0.7789,
      "step": 3796
    },
    {
      "epoch": 0.5416547788873038,
      "grad_norm": 0.9648737907409668,
      "learning_rate": 9.417690810501469e-06,
      "loss": 0.7374,
      "step": 3797
    },
    {
      "epoch": 0.5417974322396576,
      "grad_norm": 1.2017556428909302,
      "learning_rate": 9.417330104713084e-06,
      "loss": 0.7079,
      "step": 3798
    },
    {
      "epoch": 0.5419400855920115,
      "grad_norm": 1.0219480991363525,
      "learning_rate": 9.416969294153512e-06,
      "loss": 0.771,
      "step": 3799
    },
    {
      "epoch": 0.5420827389443652,
      "grad_norm": 0.9495611786842346,
      "learning_rate": 9.41660837883132e-06,
      "loss": 0.6917,
      "step": 3800
    },
    {
      "epoch": 0.542225392296719,
      "grad_norm": 0.9101681709289551,
      "learning_rate": 9.416247358755062e-06,
      "loss": 0.7347,
      "step": 3801
    },
    {
      "epoch": 0.5423680456490727,
      "grad_norm": 0.7405933737754822,
      "learning_rate": 9.415886233933303e-06,
      "loss": 0.7821,
      "step": 3802
    },
    {
      "epoch": 0.5425106990014266,
      "grad_norm": 1.2204949855804443,
      "learning_rate": 9.415525004374608e-06,
      "loss": 0.7268,
      "step": 3803
    },
    {
      "epoch": 0.5426533523537803,
      "grad_norm": 0.7763726115226746,
      "learning_rate": 9.415163670087544e-06,
      "loss": 0.765,
      "step": 3804
    },
    {
      "epoch": 0.5427960057061341,
      "grad_norm": 0.8272026777267456,
      "learning_rate": 9.414802231080683e-06,
      "loss": 0.7719,
      "step": 3805
    },
    {
      "epoch": 0.5429386590584879,
      "grad_norm": 1.023071050643921,
      "learning_rate": 9.414440687362599e-06,
      "loss": 0.7672,
      "step": 3806
    },
    {
      "epoch": 0.5430813124108417,
      "grad_norm": 0.8470318913459778,
      "learning_rate": 9.414079038941861e-06,
      "loss": 0.7465,
      "step": 3807
    },
    {
      "epoch": 0.5432239657631954,
      "grad_norm": 1.1065614223480225,
      "learning_rate": 9.41371728582705e-06,
      "loss": 0.7848,
      "step": 3808
    },
    {
      "epoch": 0.5433666191155492,
      "grad_norm": 0.8630528450012207,
      "learning_rate": 9.413355428026748e-06,
      "loss": 0.8045,
      "step": 3809
    },
    {
      "epoch": 0.543509272467903,
      "grad_norm": 1.003273606300354,
      "learning_rate": 9.412993465549539e-06,
      "loss": 0.7217,
      "step": 3810
    },
    {
      "epoch": 0.5436519258202568,
      "grad_norm": 0.8895302414894104,
      "learning_rate": 9.412631398404002e-06,
      "loss": 0.7901,
      "step": 3811
    },
    {
      "epoch": 0.5437945791726105,
      "grad_norm": 0.960581362247467,
      "learning_rate": 9.412269226598728e-06,
      "loss": 0.7509,
      "step": 3812
    },
    {
      "epoch": 0.5439372325249643,
      "grad_norm": 0.9232653975486755,
      "learning_rate": 9.411906950142307e-06,
      "loss": 0.7023,
      "step": 3813
    },
    {
      "epoch": 0.5440798858773181,
      "grad_norm": 1.0568187236785889,
      "learning_rate": 9.411544569043332e-06,
      "loss": 0.8439,
      "step": 3814
    },
    {
      "epoch": 0.5442225392296719,
      "grad_norm": 0.8565939664840698,
      "learning_rate": 9.411182083310398e-06,
      "loss": 0.8,
      "step": 3815
    },
    {
      "epoch": 0.5443651925820256,
      "grad_norm": 1.2284659147262573,
      "learning_rate": 9.410819492952101e-06,
      "loss": 0.736,
      "step": 3816
    },
    {
      "epoch": 0.5445078459343795,
      "grad_norm": 0.8112896680831909,
      "learning_rate": 9.410456797977043e-06,
      "loss": 0.7771,
      "step": 3817
    },
    {
      "epoch": 0.5446504992867333,
      "grad_norm": 0.8988441824913025,
      "learning_rate": 9.410093998393826e-06,
      "loss": 0.7399,
      "step": 3818
    },
    {
      "epoch": 0.544793152639087,
      "grad_norm": 0.7892339825630188,
      "learning_rate": 9.409731094211054e-06,
      "loss": 0.8353,
      "step": 3819
    },
    {
      "epoch": 0.5449358059914408,
      "grad_norm": 0.909959614276886,
      "learning_rate": 9.409368085437334e-06,
      "loss": 0.8046,
      "step": 3820
    },
    {
      "epoch": 0.5450784593437946,
      "grad_norm": 0.9480772614479065,
      "learning_rate": 9.409004972081277e-06,
      "loss": 0.7268,
      "step": 3821
    },
    {
      "epoch": 0.5452211126961484,
      "grad_norm": 0.7678883075714111,
      "learning_rate": 9.408641754151496e-06,
      "loss": 0.7189,
      "step": 3822
    },
    {
      "epoch": 0.5453637660485021,
      "grad_norm": 1.0699303150177002,
      "learning_rate": 9.408278431656605e-06,
      "loss": 0.6943,
      "step": 3823
    },
    {
      "epoch": 0.5455064194008559,
      "grad_norm": 0.8879801034927368,
      "learning_rate": 9.407915004605221e-06,
      "loss": 0.7314,
      "step": 3824
    },
    {
      "epoch": 0.5456490727532097,
      "grad_norm": 0.8187634944915771,
      "learning_rate": 9.407551473005965e-06,
      "loss": 0.7599,
      "step": 3825
    },
    {
      "epoch": 0.5457917261055635,
      "grad_norm": 0.8397625684738159,
      "learning_rate": 9.407187836867459e-06,
      "loss": 0.7384,
      "step": 3826
    },
    {
      "epoch": 0.5459343794579172,
      "grad_norm": 0.8377321362495422,
      "learning_rate": 9.406824096198326e-06,
      "loss": 0.7193,
      "step": 3827
    },
    {
      "epoch": 0.5460770328102711,
      "grad_norm": 1.0204477310180664,
      "learning_rate": 9.406460251007195e-06,
      "loss": 0.7563,
      "step": 3828
    },
    {
      "epoch": 0.5462196861626248,
      "grad_norm": 0.7482069730758667,
      "learning_rate": 9.406096301302696e-06,
      "loss": 0.7271,
      "step": 3829
    },
    {
      "epoch": 0.5463623395149786,
      "grad_norm": 0.8671506643295288,
      "learning_rate": 9.405732247093463e-06,
      "loss": 0.7212,
      "step": 3830
    },
    {
      "epoch": 0.5465049928673323,
      "grad_norm": 1.0759758949279785,
      "learning_rate": 9.405368088388124e-06,
      "loss": 0.7894,
      "step": 3831
    },
    {
      "epoch": 0.5466476462196862,
      "grad_norm": 0.7334364056587219,
      "learning_rate": 9.405003825195324e-06,
      "loss": 0.749,
      "step": 3832
    },
    {
      "epoch": 0.54679029957204,
      "grad_norm": 0.9418511390686035,
      "learning_rate": 9.404639457523698e-06,
      "loss": 0.7525,
      "step": 3833
    },
    {
      "epoch": 0.5469329529243937,
      "grad_norm": 0.9294775128364563,
      "learning_rate": 9.404274985381889e-06,
      "loss": 0.8074,
      "step": 3834
    },
    {
      "epoch": 0.5470756062767475,
      "grad_norm": 0.82530277967453,
      "learning_rate": 9.403910408778543e-06,
      "loss": 0.7286,
      "step": 3835
    },
    {
      "epoch": 0.5472182596291013,
      "grad_norm": 0.9933231472969055,
      "learning_rate": 9.403545727722306e-06,
      "loss": 0.7873,
      "step": 3836
    },
    {
      "epoch": 0.5473609129814551,
      "grad_norm": 1.3327282667160034,
      "learning_rate": 9.403180942221827e-06,
      "loss": 0.7544,
      "step": 3837
    },
    {
      "epoch": 0.5475035663338088,
      "grad_norm": 1.029980182647705,
      "learning_rate": 9.40281605228576e-06,
      "loss": 0.7961,
      "step": 3838
    },
    {
      "epoch": 0.5476462196861627,
      "grad_norm": 0.9801850318908691,
      "learning_rate": 9.402451057922757e-06,
      "loss": 0.7422,
      "step": 3839
    },
    {
      "epoch": 0.5477888730385164,
      "grad_norm": 0.7895873188972473,
      "learning_rate": 9.402085959141476e-06,
      "loss": 0.7254,
      "step": 3840
    },
    {
      "epoch": 0.5479315263908702,
      "grad_norm": 0.9539476633071899,
      "learning_rate": 9.401720755950578e-06,
      "loss": 0.7911,
      "step": 3841
    },
    {
      "epoch": 0.5480741797432239,
      "grad_norm": 0.7882383465766907,
      "learning_rate": 9.401355448358725e-06,
      "loss": 0.7592,
      "step": 3842
    },
    {
      "epoch": 0.5482168330955778,
      "grad_norm": 1.0241360664367676,
      "learning_rate": 9.40099003637458e-06,
      "loss": 0.7755,
      "step": 3843
    },
    {
      "epoch": 0.5483594864479315,
      "grad_norm": 0.914991557598114,
      "learning_rate": 9.40062452000681e-06,
      "loss": 0.8418,
      "step": 3844
    },
    {
      "epoch": 0.5485021398002853,
      "grad_norm": 0.9055288434028625,
      "learning_rate": 9.400258899264083e-06,
      "loss": 0.7925,
      "step": 3845
    },
    {
      "epoch": 0.548644793152639,
      "grad_norm": 0.8815552592277527,
      "learning_rate": 9.399893174155072e-06,
      "loss": 0.5296,
      "step": 3846
    },
    {
      "epoch": 0.5487874465049929,
      "grad_norm": 0.7472600340843201,
      "learning_rate": 9.399527344688454e-06,
      "loss": 0.5569,
      "step": 3847
    },
    {
      "epoch": 0.5489300998573466,
      "grad_norm": 0.852327287197113,
      "learning_rate": 9.399161410872904e-06,
      "loss": 0.746,
      "step": 3848
    },
    {
      "epoch": 0.5490727532097004,
      "grad_norm": 0.9859546422958374,
      "learning_rate": 9.3987953727171e-06,
      "loss": 0.7886,
      "step": 3849
    },
    {
      "epoch": 0.5492154065620543,
      "grad_norm": 0.8234342336654663,
      "learning_rate": 9.398429230229724e-06,
      "loss": 0.7614,
      "step": 3850
    },
    {
      "epoch": 0.549358059914408,
      "grad_norm": 1.2982840538024902,
      "learning_rate": 9.39806298341946e-06,
      "loss": 0.8429,
      "step": 3851
    },
    {
      "epoch": 0.5495007132667618,
      "grad_norm": 0.8056015372276306,
      "learning_rate": 9.397696632294998e-06,
      "loss": 0.7934,
      "step": 3852
    },
    {
      "epoch": 0.5496433666191155,
      "grad_norm": 0.6674420237541199,
      "learning_rate": 9.397330176865026e-06,
      "loss": 0.7803,
      "step": 3853
    },
    {
      "epoch": 0.5497860199714694,
      "grad_norm": 1.0118069648742676,
      "learning_rate": 9.396963617138231e-06,
      "loss": 0.8304,
      "step": 3854
    },
    {
      "epoch": 0.5499286733238231,
      "grad_norm": 0.822985827922821,
      "learning_rate": 9.396596953123312e-06,
      "loss": 0.7801,
      "step": 3855
    },
    {
      "epoch": 0.5500713266761769,
      "grad_norm": 0.9515111446380615,
      "learning_rate": 9.396230184828964e-06,
      "loss": 0.5683,
      "step": 3856
    },
    {
      "epoch": 0.5502139800285307,
      "grad_norm": 0.9106077551841736,
      "learning_rate": 9.395863312263888e-06,
      "loss": 0.7109,
      "step": 3857
    },
    {
      "epoch": 0.5503566333808845,
      "grad_norm": 1.0948283672332764,
      "learning_rate": 9.395496335436783e-06,
      "loss": 0.8231,
      "step": 3858
    },
    {
      "epoch": 0.5504992867332382,
      "grad_norm": 0.7987005114555359,
      "learning_rate": 9.395129254356354e-06,
      "loss": 0.7439,
      "step": 3859
    },
    {
      "epoch": 0.550641940085592,
      "grad_norm": 1.1406909227371216,
      "learning_rate": 9.394762069031306e-06,
      "loss": 0.7732,
      "step": 3860
    },
    {
      "epoch": 0.5507845934379458,
      "grad_norm": 0.9241793155670166,
      "learning_rate": 9.394394779470351e-06,
      "loss": 0.7821,
      "step": 3861
    },
    {
      "epoch": 0.5509272467902996,
      "grad_norm": 0.9340644478797913,
      "learning_rate": 9.394027385682198e-06,
      "loss": 0.7734,
      "step": 3862
    },
    {
      "epoch": 0.5510699001426533,
      "grad_norm": 1.0453873872756958,
      "learning_rate": 9.393659887675564e-06,
      "loss": 0.7709,
      "step": 3863
    },
    {
      "epoch": 0.5512125534950071,
      "grad_norm": 1.141420841217041,
      "learning_rate": 9.393292285459162e-06,
      "loss": 0.7037,
      "step": 3864
    },
    {
      "epoch": 0.551355206847361,
      "grad_norm": 0.9163342714309692,
      "learning_rate": 9.392924579041711e-06,
      "loss": 0.7508,
      "step": 3865
    },
    {
      "epoch": 0.5514978601997147,
      "grad_norm": 1.008906364440918,
      "learning_rate": 9.392556768431936e-06,
      "loss": 0.7192,
      "step": 3866
    },
    {
      "epoch": 0.5516405135520684,
      "grad_norm": 1.2465540170669556,
      "learning_rate": 9.392188853638556e-06,
      "loss": 0.8364,
      "step": 3867
    },
    {
      "epoch": 0.5517831669044223,
      "grad_norm": 1.1515864133834839,
      "learning_rate": 9.3918208346703e-06,
      "loss": 0.741,
      "step": 3868
    },
    {
      "epoch": 0.5519258202567761,
      "grad_norm": 0.8714399337768555,
      "learning_rate": 9.391452711535897e-06,
      "loss": 0.7673,
      "step": 3869
    },
    {
      "epoch": 0.5520684736091298,
      "grad_norm": 0.8498027920722961,
      "learning_rate": 9.391084484244077e-06,
      "loss": 0.7968,
      "step": 3870
    },
    {
      "epoch": 0.5522111269614836,
      "grad_norm": 0.9492782354354858,
      "learning_rate": 9.390716152803573e-06,
      "loss": 0.7765,
      "step": 3871
    },
    {
      "epoch": 0.5523537803138374,
      "grad_norm": 1.3390405178070068,
      "learning_rate": 9.390347717223124e-06,
      "loss": 0.5709,
      "step": 3872
    },
    {
      "epoch": 0.5524964336661912,
      "grad_norm": 0.7939488887786865,
      "learning_rate": 9.389979177511467e-06,
      "loss": 0.7349,
      "step": 3873
    },
    {
      "epoch": 0.5526390870185449,
      "grad_norm": 0.9034905433654785,
      "learning_rate": 9.389610533677342e-06,
      "loss": 0.6985,
      "step": 3874
    },
    {
      "epoch": 0.5527817403708987,
      "grad_norm": 1.0521996021270752,
      "learning_rate": 9.389241785729495e-06,
      "loss": 0.7328,
      "step": 3875
    },
    {
      "epoch": 0.5529243937232525,
      "grad_norm": 1.0269242525100708,
      "learning_rate": 9.388872933676672e-06,
      "loss": 0.674,
      "step": 3876
    },
    {
      "epoch": 0.5530670470756063,
      "grad_norm": 0.7578728795051575,
      "learning_rate": 9.388503977527617e-06,
      "loss": 0.8702,
      "step": 3877
    },
    {
      "epoch": 0.55320970042796,
      "grad_norm": 0.9115679264068604,
      "learning_rate": 9.388134917291085e-06,
      "loss": 0.7844,
      "step": 3878
    },
    {
      "epoch": 0.5533523537803139,
      "grad_norm": 0.7785453200340271,
      "learning_rate": 9.38776575297583e-06,
      "loss": 0.762,
      "step": 3879
    },
    {
      "epoch": 0.5534950071326676,
      "grad_norm": 0.7482083439826965,
      "learning_rate": 9.387396484590607e-06,
      "loss": 0.7261,
      "step": 3880
    },
    {
      "epoch": 0.5536376604850214,
      "grad_norm": 0.86270672082901,
      "learning_rate": 9.387027112144171e-06,
      "loss": 0.5208,
      "step": 3881
    },
    {
      "epoch": 0.5537803138373751,
      "grad_norm": 1.0379152297973633,
      "learning_rate": 9.38665763564529e-06,
      "loss": 0.7124,
      "step": 3882
    },
    {
      "epoch": 0.553922967189729,
      "grad_norm": 1.0672602653503418,
      "learning_rate": 9.38628805510272e-06,
      "loss": 0.7421,
      "step": 3883
    },
    {
      "epoch": 0.5540656205420827,
      "grad_norm": 0.8030416369438171,
      "learning_rate": 9.38591837052523e-06,
      "loss": 0.8006,
      "step": 3884
    },
    {
      "epoch": 0.5542082738944365,
      "grad_norm": 0.8115381002426147,
      "learning_rate": 9.385548581921589e-06,
      "loss": 0.783,
      "step": 3885
    },
    {
      "epoch": 0.5543509272467902,
      "grad_norm": 1.0033485889434814,
      "learning_rate": 9.385178689300567e-06,
      "loss": 0.7728,
      "step": 3886
    },
    {
      "epoch": 0.5544935805991441,
      "grad_norm": 1.0693013668060303,
      "learning_rate": 9.384808692670937e-06,
      "loss": 0.7733,
      "step": 3887
    },
    {
      "epoch": 0.5546362339514979,
      "grad_norm": 0.7807943820953369,
      "learning_rate": 9.384438592041476e-06,
      "loss": 0.8106,
      "step": 3888
    },
    {
      "epoch": 0.5547788873038516,
      "grad_norm": 1.0082643032073975,
      "learning_rate": 9.384068387420959e-06,
      "loss": 0.7763,
      "step": 3889
    },
    {
      "epoch": 0.5549215406562055,
      "grad_norm": 0.6661744713783264,
      "learning_rate": 9.38369807881817e-06,
      "loss": 0.7905,
      "step": 3890
    },
    {
      "epoch": 0.5550641940085592,
      "grad_norm": 0.7957794070243835,
      "learning_rate": 9.38332766624189e-06,
      "loss": 0.7971,
      "step": 3891
    },
    {
      "epoch": 0.555206847360913,
      "grad_norm": 0.9141960740089417,
      "learning_rate": 9.382957149700904e-06,
      "loss": 0.764,
      "step": 3892
    },
    {
      "epoch": 0.5553495007132667,
      "grad_norm": 0.9018802642822266,
      "learning_rate": 9.382586529204002e-06,
      "loss": 0.7772,
      "step": 3893
    },
    {
      "epoch": 0.5554921540656206,
      "grad_norm": 1.0242016315460205,
      "learning_rate": 9.382215804759975e-06,
      "loss": 0.8302,
      "step": 3894
    },
    {
      "epoch": 0.5556348074179743,
      "grad_norm": 0.8564112782478333,
      "learning_rate": 9.381844976377612e-06,
      "loss": 0.7643,
      "step": 3895
    },
    {
      "epoch": 0.5557774607703281,
      "grad_norm": 0.82747882604599,
      "learning_rate": 9.381474044065712e-06,
      "loss": 0.7983,
      "step": 3896
    },
    {
      "epoch": 0.5559201141226818,
      "grad_norm": 0.9846975207328796,
      "learning_rate": 9.381103007833072e-06,
      "loss": 0.6618,
      "step": 3897
    },
    {
      "epoch": 0.5560627674750357,
      "grad_norm": 1.0486607551574707,
      "learning_rate": 9.38073186768849e-06,
      "loss": 0.7422,
      "step": 3898
    },
    {
      "epoch": 0.5562054208273894,
      "grad_norm": 1.1494112014770508,
      "learning_rate": 9.380360623640774e-06,
      "loss": 0.7019,
      "step": 3899
    },
    {
      "epoch": 0.5563480741797432,
      "grad_norm": 0.8423972129821777,
      "learning_rate": 9.379989275698724e-06,
      "loss": 0.8188,
      "step": 3900
    },
    {
      "epoch": 0.556490727532097,
      "grad_norm": 0.9076462984085083,
      "learning_rate": 9.37961782387115e-06,
      "loss": 0.7645,
      "step": 3901
    },
    {
      "epoch": 0.5566333808844508,
      "grad_norm": 0.8854547739028931,
      "learning_rate": 9.379246268166863e-06,
      "loss": 0.6709,
      "step": 3902
    },
    {
      "epoch": 0.5567760342368046,
      "grad_norm": 0.769031822681427,
      "learning_rate": 9.378874608594673e-06,
      "loss": 0.7791,
      "step": 3903
    },
    {
      "epoch": 0.5569186875891583,
      "grad_norm": 0.9481874108314514,
      "learning_rate": 9.378502845163397e-06,
      "loss": 0.8044,
      "step": 3904
    },
    {
      "epoch": 0.5570613409415122,
      "grad_norm": 0.7037337422370911,
      "learning_rate": 9.378130977881855e-06,
      "loss": 0.7142,
      "step": 3905
    },
    {
      "epoch": 0.5572039942938659,
      "grad_norm": 0.8134806156158447,
      "learning_rate": 9.37775900675886e-06,
      "loss": 0.8321,
      "step": 3906
    },
    {
      "epoch": 0.5573466476462197,
      "grad_norm": 1.2003941535949707,
      "learning_rate": 9.37738693180324e-06,
      "loss": 0.83,
      "step": 3907
    },
    {
      "epoch": 0.5574893009985734,
      "grad_norm": 1.061151385307312,
      "learning_rate": 9.37701475302382e-06,
      "loss": 0.7487,
      "step": 3908
    },
    {
      "epoch": 0.5576319543509273,
      "grad_norm": 0.9353809356689453,
      "learning_rate": 9.376642470429425e-06,
      "loss": 0.7537,
      "step": 3909
    },
    {
      "epoch": 0.557774607703281,
      "grad_norm": 0.8664660453796387,
      "learning_rate": 9.376270084028887e-06,
      "loss": 0.7445,
      "step": 3910
    },
    {
      "epoch": 0.5579172610556348,
      "grad_norm": 0.9144179821014404,
      "learning_rate": 9.375897593831036e-06,
      "loss": 0.7737,
      "step": 3911
    },
    {
      "epoch": 0.5580599144079886,
      "grad_norm": 1.072312593460083,
      "learning_rate": 9.375524999844709e-06,
      "loss": 0.7481,
      "step": 3912
    },
    {
      "epoch": 0.5582025677603424,
      "grad_norm": 0.8142551779747009,
      "learning_rate": 9.375152302078742e-06,
      "loss": 0.7438,
      "step": 3913
    },
    {
      "epoch": 0.5583452211126961,
      "grad_norm": 0.9728956818580627,
      "learning_rate": 9.374779500541975e-06,
      "loss": 0.765,
      "step": 3914
    },
    {
      "epoch": 0.5584878744650499,
      "grad_norm": 0.8081833720207214,
      "learning_rate": 9.374406595243252e-06,
      "loss": 0.8573,
      "step": 3915
    },
    {
      "epoch": 0.5586305278174037,
      "grad_norm": 0.7585574984550476,
      "learning_rate": 9.374033586191413e-06,
      "loss": 0.8176,
      "step": 3916
    },
    {
      "epoch": 0.5587731811697575,
      "grad_norm": 0.8608925938606262,
      "learning_rate": 9.37366047339531e-06,
      "loss": 0.8215,
      "step": 3917
    },
    {
      "epoch": 0.5589158345221112,
      "grad_norm": 0.8893569707870483,
      "learning_rate": 9.373287256863792e-06,
      "loss": 0.7927,
      "step": 3918
    },
    {
      "epoch": 0.5590584878744651,
      "grad_norm": 0.7753645181655884,
      "learning_rate": 9.372913936605708e-06,
      "loss": 0.795,
      "step": 3919
    },
    {
      "epoch": 0.5592011412268189,
      "grad_norm": 0.8030126094818115,
      "learning_rate": 9.372540512629913e-06,
      "loss": 0.7629,
      "step": 3920
    },
    {
      "epoch": 0.5593437945791726,
      "grad_norm": 0.6363753080368042,
      "learning_rate": 9.372166984945266e-06,
      "loss": 0.7646,
      "step": 3921
    },
    {
      "epoch": 0.5594864479315264,
      "grad_norm": 0.6828857064247131,
      "learning_rate": 9.371793353560627e-06,
      "loss": 0.7824,
      "step": 3922
    },
    {
      "epoch": 0.5596291012838802,
      "grad_norm": 0.7620823979377747,
      "learning_rate": 9.371419618484854e-06,
      "loss": 0.8032,
      "step": 3923
    },
    {
      "epoch": 0.559771754636234,
      "grad_norm": 0.9093039631843567,
      "learning_rate": 9.371045779726815e-06,
      "loss": 0.6905,
      "step": 3924
    },
    {
      "epoch": 0.5599144079885877,
      "grad_norm": 0.8852125406265259,
      "learning_rate": 9.370671837295376e-06,
      "loss": 0.7307,
      "step": 3925
    },
    {
      "epoch": 0.5600570613409415,
      "grad_norm": 1.2022708654403687,
      "learning_rate": 9.370297791199406e-06,
      "loss": 0.7569,
      "step": 3926
    },
    {
      "epoch": 0.5601997146932953,
      "grad_norm": 1.2121343612670898,
      "learning_rate": 9.369923641447774e-06,
      "loss": 0.8059,
      "step": 3927
    },
    {
      "epoch": 0.5603423680456491,
      "grad_norm": 0.9693889617919922,
      "learning_rate": 9.369549388049359e-06,
      "loss": 0.671,
      "step": 3928
    },
    {
      "epoch": 0.5604850213980028,
      "grad_norm": 0.8566369414329529,
      "learning_rate": 9.369175031013034e-06,
      "loss": 0.7901,
      "step": 3929
    },
    {
      "epoch": 0.5606276747503567,
      "grad_norm": 0.6705704927444458,
      "learning_rate": 9.368800570347681e-06,
      "loss": 0.81,
      "step": 3930
    },
    {
      "epoch": 0.5607703281027104,
      "grad_norm": 0.7827990055084229,
      "learning_rate": 9.368426006062179e-06,
      "loss": 0.8014,
      "step": 3931
    },
    {
      "epoch": 0.5609129814550642,
      "grad_norm": 1.008968472480774,
      "learning_rate": 9.36805133816541e-06,
      "loss": 0.7946,
      "step": 3932
    },
    {
      "epoch": 0.5610556348074179,
      "grad_norm": 1.0732449293136597,
      "learning_rate": 9.367676566666268e-06,
      "loss": 0.7743,
      "step": 3933
    },
    {
      "epoch": 0.5611982881597718,
      "grad_norm": 0.844943106174469,
      "learning_rate": 9.367301691573633e-06,
      "loss": 0.8017,
      "step": 3934
    },
    {
      "epoch": 0.5613409415121255,
      "grad_norm": 1.220983862876892,
      "learning_rate": 9.366926712896404e-06,
      "loss": 0.7263,
      "step": 3935
    },
    {
      "epoch": 0.5614835948644793,
      "grad_norm": 0.9163217544555664,
      "learning_rate": 9.36655163064347e-06,
      "loss": 0.7281,
      "step": 3936
    },
    {
      "epoch": 0.561626248216833,
      "grad_norm": 0.798708975315094,
      "learning_rate": 9.36617644482373e-06,
      "loss": 0.7522,
      "step": 3937
    },
    {
      "epoch": 0.5617689015691869,
      "grad_norm": 0.7758632898330688,
      "learning_rate": 9.365801155446079e-06,
      "loss": 0.8327,
      "step": 3938
    },
    {
      "epoch": 0.5619115549215407,
      "grad_norm": 0.8243889212608337,
      "learning_rate": 9.365425762519422e-06,
      "loss": 0.8003,
      "step": 3939
    },
    {
      "epoch": 0.5620542082738944,
      "grad_norm": 0.8462749123573303,
      "learning_rate": 9.36505026605266e-06,
      "loss": 0.7252,
      "step": 3940
    },
    {
      "epoch": 0.5621968616262483,
      "grad_norm": 1.0904769897460938,
      "learning_rate": 9.364674666054701e-06,
      "loss": 0.7509,
      "step": 3941
    },
    {
      "epoch": 0.562339514978602,
      "grad_norm": 0.9569264054298401,
      "learning_rate": 9.364298962534455e-06,
      "loss": 0.7646,
      "step": 3942
    },
    {
      "epoch": 0.5624821683309558,
      "grad_norm": 0.7271814346313477,
      "learning_rate": 9.363923155500829e-06,
      "loss": 0.7848,
      "step": 3943
    },
    {
      "epoch": 0.5626248216833095,
      "grad_norm": 0.8120964169502258,
      "learning_rate": 9.363547244962738e-06,
      "loss": 0.7717,
      "step": 3944
    },
    {
      "epoch": 0.5627674750356634,
      "grad_norm": 0.8208606243133545,
      "learning_rate": 9.3631712309291e-06,
      "loss": 0.8162,
      "step": 3945
    },
    {
      "epoch": 0.5629101283880171,
      "grad_norm": 1.1177812814712524,
      "learning_rate": 9.36279511340883e-06,
      "loss": 0.7225,
      "step": 3946
    },
    {
      "epoch": 0.5630527817403709,
      "grad_norm": 1.286088466644287,
      "learning_rate": 9.362418892410851e-06,
      "loss": 0.8166,
      "step": 3947
    },
    {
      "epoch": 0.5631954350927246,
      "grad_norm": 1.0582689046859741,
      "learning_rate": 9.362042567944088e-06,
      "loss": 0.7698,
      "step": 3948
    },
    {
      "epoch": 0.5633380884450785,
      "grad_norm": 0.9759199619293213,
      "learning_rate": 9.361666140017463e-06,
      "loss": 0.7137,
      "step": 3949
    },
    {
      "epoch": 0.5634807417974322,
      "grad_norm": 0.8342788219451904,
      "learning_rate": 9.361289608639905e-06,
      "loss": 0.7432,
      "step": 3950
    },
    {
      "epoch": 0.563623395149786,
      "grad_norm": 1.1413004398345947,
      "learning_rate": 9.360912973820347e-06,
      "loss": 0.7713,
      "step": 3951
    },
    {
      "epoch": 0.5637660485021398,
      "grad_norm": 0.8673157691955566,
      "learning_rate": 9.36053623556772e-06,
      "loss": 0.7325,
      "step": 3952
    },
    {
      "epoch": 0.5639087018544936,
      "grad_norm": 0.8513365983963013,
      "learning_rate": 9.360159393890958e-06,
      "loss": 0.7634,
      "step": 3953
    },
    {
      "epoch": 0.5640513552068473,
      "grad_norm": 0.646003007888794,
      "learning_rate": 9.359782448799005e-06,
      "loss": 0.5906,
      "step": 3954
    },
    {
      "epoch": 0.5641940085592011,
      "grad_norm": 1.680774450302124,
      "learning_rate": 9.359405400300794e-06,
      "loss": 0.7783,
      "step": 3955
    },
    {
      "epoch": 0.564336661911555,
      "grad_norm": 1.0468223094940186,
      "learning_rate": 9.359028248405273e-06,
      "loss": 0.7025,
      "step": 3956
    },
    {
      "epoch": 0.5644793152639087,
      "grad_norm": 0.8599289059638977,
      "learning_rate": 9.358650993121386e-06,
      "loss": 0.7677,
      "step": 3957
    },
    {
      "epoch": 0.5646219686162625,
      "grad_norm": 0.9311829805374146,
      "learning_rate": 9.358273634458082e-06,
      "loss": 0.658,
      "step": 3958
    },
    {
      "epoch": 0.5647646219686162,
      "grad_norm": 0.9008866548538208,
      "learning_rate": 9.357896172424308e-06,
      "loss": 0.7866,
      "step": 3959
    },
    {
      "epoch": 0.5649072753209701,
      "grad_norm": 0.9304162263870239,
      "learning_rate": 9.357518607029019e-06,
      "loss": 0.5638,
      "step": 3960
    },
    {
      "epoch": 0.5650499286733238,
      "grad_norm": 1.1516731977462769,
      "learning_rate": 9.35714093828117e-06,
      "loss": 0.8124,
      "step": 3961
    },
    {
      "epoch": 0.5651925820256776,
      "grad_norm": 0.9910885095596313,
      "learning_rate": 9.35676316618972e-06,
      "loss": 0.8032,
      "step": 3962
    },
    {
      "epoch": 0.5653352353780314,
      "grad_norm": 0.8922032713890076,
      "learning_rate": 9.356385290763625e-06,
      "loss": 0.8418,
      "step": 3963
    },
    {
      "epoch": 0.5654778887303852,
      "grad_norm": 1.1360137462615967,
      "learning_rate": 9.356007312011853e-06,
      "loss": 0.6794,
      "step": 3964
    },
    {
      "epoch": 0.5656205420827389,
      "grad_norm": 0.9588967561721802,
      "learning_rate": 9.355629229943363e-06,
      "loss": 0.7752,
      "step": 3965
    },
    {
      "epoch": 0.5657631954350927,
      "grad_norm": 0.768655002117157,
      "learning_rate": 9.355251044567129e-06,
      "loss": 0.7379,
      "step": 3966
    },
    {
      "epoch": 0.5659058487874465,
      "grad_norm": 0.9083778858184814,
      "learning_rate": 9.354872755892115e-06,
      "loss": 0.7248,
      "step": 3967
    },
    {
      "epoch": 0.5660485021398003,
      "grad_norm": 0.8241592049598694,
      "learning_rate": 9.354494363927296e-06,
      "loss": 0.5281,
      "step": 3968
    },
    {
      "epoch": 0.566191155492154,
      "grad_norm": 1.0365681648254395,
      "learning_rate": 9.354115868681648e-06,
      "loss": 0.6871,
      "step": 3969
    },
    {
      "epoch": 0.5663338088445078,
      "grad_norm": 0.8775922060012817,
      "learning_rate": 9.353737270164145e-06,
      "loss": 0.7885,
      "step": 3970
    },
    {
      "epoch": 0.5664764621968617,
      "grad_norm": 1.0082521438598633,
      "learning_rate": 9.35335856838377e-06,
      "loss": 0.8228,
      "step": 3971
    },
    {
      "epoch": 0.5666191155492154,
      "grad_norm": 0.7796494960784912,
      "learning_rate": 9.352979763349501e-06,
      "loss": 0.7185,
      "step": 3972
    },
    {
      "epoch": 0.5667617689015692,
      "grad_norm": 0.9608703255653381,
      "learning_rate": 9.352600855070327e-06,
      "loss": 0.7932,
      "step": 3973
    },
    {
      "epoch": 0.566904422253923,
      "grad_norm": 1.2445639371871948,
      "learning_rate": 9.352221843555234e-06,
      "loss": 0.727,
      "step": 3974
    },
    {
      "epoch": 0.5670470756062768,
      "grad_norm": 0.9000669121742249,
      "learning_rate": 9.35184272881321e-06,
      "loss": 0.7399,
      "step": 3975
    },
    {
      "epoch": 0.5671897289586305,
      "grad_norm": 0.7240111827850342,
      "learning_rate": 9.351463510853245e-06,
      "loss": 0.7932,
      "step": 3976
    },
    {
      "epoch": 0.5673323823109843,
      "grad_norm": 1.207418441772461,
      "learning_rate": 9.351084189684339e-06,
      "loss": 0.7482,
      "step": 3977
    },
    {
      "epoch": 0.5674750356633381,
      "grad_norm": 1.0589617490768433,
      "learning_rate": 9.350704765315485e-06,
      "loss": 0.8294,
      "step": 3978
    },
    {
      "epoch": 0.5676176890156919,
      "grad_norm": 0.7942510843276978,
      "learning_rate": 9.350325237755683e-06,
      "loss": 0.7877,
      "step": 3979
    },
    {
      "epoch": 0.5677603423680456,
      "grad_norm": 0.8113898634910583,
      "learning_rate": 9.349945607013935e-06,
      "loss": 0.7244,
      "step": 3980
    },
    {
      "epoch": 0.5679029957203995,
      "grad_norm": 0.8928048014640808,
      "learning_rate": 9.349565873099245e-06,
      "loss": 0.7469,
      "step": 3981
    },
    {
      "epoch": 0.5680456490727532,
      "grad_norm": 0.7207472920417786,
      "learning_rate": 9.349186036020618e-06,
      "loss": 0.828,
      "step": 3982
    },
    {
      "epoch": 0.568188302425107,
      "grad_norm": 0.8237086534500122,
      "learning_rate": 9.348806095787065e-06,
      "loss": 0.758,
      "step": 3983
    },
    {
      "epoch": 0.5683309557774607,
      "grad_norm": 0.7653695940971375,
      "learning_rate": 9.3484260524076e-06,
      "loss": 0.5715,
      "step": 3984
    },
    {
      "epoch": 0.5684736091298146,
      "grad_norm": 0.8891986608505249,
      "learning_rate": 9.348045905891231e-06,
      "loss": 0.7624,
      "step": 3985
    },
    {
      "epoch": 0.5686162624821683,
      "grad_norm": 0.7860222458839417,
      "learning_rate": 9.347665656246978e-06,
      "loss": 0.745,
      "step": 3986
    },
    {
      "epoch": 0.5687589158345221,
      "grad_norm": 0.8255372047424316,
      "learning_rate": 9.34728530348386e-06,
      "loss": 0.739,
      "step": 3987
    },
    {
      "epoch": 0.5689015691868758,
      "grad_norm": 0.8702134490013123,
      "learning_rate": 9.346904847610898e-06,
      "loss": 0.7413,
      "step": 3988
    },
    {
      "epoch": 0.5690442225392297,
      "grad_norm": 1.1986943483352661,
      "learning_rate": 9.346524288637114e-06,
      "loss": 0.7627,
      "step": 3989
    },
    {
      "epoch": 0.5691868758915835,
      "grad_norm": 0.8165330290794373,
      "learning_rate": 9.346143626571536e-06,
      "loss": 0.8448,
      "step": 3990
    },
    {
      "epoch": 0.5693295292439372,
      "grad_norm": 0.842220664024353,
      "learning_rate": 9.34576286142319e-06,
      "loss": 0.7451,
      "step": 3991
    },
    {
      "epoch": 0.5694721825962911,
      "grad_norm": 1.1866034269332886,
      "learning_rate": 9.345381993201113e-06,
      "loss": 0.5583,
      "step": 3992
    },
    {
      "epoch": 0.5696148359486448,
      "grad_norm": 1.0934287309646606,
      "learning_rate": 9.345001021914334e-06,
      "loss": 0.6854,
      "step": 3993
    },
    {
      "epoch": 0.5697574893009986,
      "grad_norm": 0.9536330103874207,
      "learning_rate": 9.344619947571888e-06,
      "loss": 0.804,
      "step": 3994
    },
    {
      "epoch": 0.5699001426533523,
      "grad_norm": 0.8321189284324646,
      "learning_rate": 9.344238770182817e-06,
      "loss": 0.7551,
      "step": 3995
    },
    {
      "epoch": 0.5700427960057062,
      "grad_norm": 1.0529017448425293,
      "learning_rate": 9.343857489756155e-06,
      "loss": 0.7332,
      "step": 3996
    },
    {
      "epoch": 0.5701854493580599,
      "grad_norm": 0.9437790513038635,
      "learning_rate": 9.343476106300957e-06,
      "loss": 0.8217,
      "step": 3997
    },
    {
      "epoch": 0.5703281027104137,
      "grad_norm": 0.6984540224075317,
      "learning_rate": 9.343094619826257e-06,
      "loss": 0.7697,
      "step": 3998
    },
    {
      "epoch": 0.5704707560627674,
      "grad_norm": 0.7660011053085327,
      "learning_rate": 9.34271303034111e-06,
      "loss": 0.7877,
      "step": 3999
    },
    {
      "epoch": 0.5706134094151213,
      "grad_norm": 0.8685745000839233,
      "learning_rate": 9.342331337854565e-06,
      "loss": 0.7815,
      "step": 4000
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 21030,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 5.319886361723929e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}