{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.995008319467554,
  "eval_steps": 500,
  "global_step": 1350,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0022185246810870773,
      "grad_norm": 0.03917507827281952,
      "learning_rate": 0.0,
      "loss": 0.6701,
      "step": 1
    },
    {
      "epoch": 0.004437049362174155,
      "grad_norm": 0.042684901505708694,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.5712,
      "step": 2
    },
    {
      "epoch": 0.0066555740432612314,
      "grad_norm": 0.05378508195281029,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.6799,
      "step": 3
    },
    {
      "epoch": 0.00887409872434831,
      "grad_norm": 0.038782719522714615,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.5101,
      "step": 4
    },
    {
      "epoch": 0.011092623405435386,
      "grad_norm": 0.04402119293808937,
      "learning_rate": 8.888888888888888e-06,
      "loss": 0.562,
      "step": 5
    },
    {
      "epoch": 0.013311148086522463,
      "grad_norm": 0.043714866042137146,
      "learning_rate": 1.111111111111111e-05,
      "loss": 0.5273,
      "step": 6
    },
    {
      "epoch": 0.01552967276760954,
      "grad_norm": 0.04812934249639511,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.4713,
      "step": 7
    },
    {
      "epoch": 0.01774819744869662,
      "grad_norm": 0.055056456476449966,
      "learning_rate": 1.5555555555555555e-05,
      "loss": 0.6859,
      "step": 8
    },
    {
      "epoch": 0.019966722129783693,
      "grad_norm": 0.10950233787298203,
      "learning_rate": 1.7777777777777777e-05,
      "loss": 0.5307,
      "step": 9
    },
    {
      "epoch": 0.022185246810870772,
      "grad_norm": 0.032932959496974945,
      "learning_rate": 1.9999999999999998e-05,
      "loss": 0.7209,
      "step": 10
    },
    {
      "epoch": 0.024403771491957847,
      "grad_norm": 0.1700541377067566,
      "learning_rate": 2.222222222222222e-05,
      "loss": 0.896,
      "step": 11
    },
    {
      "epoch": 0.026622296173044926,
      "grad_norm": 0.05196228623390198,
      "learning_rate": 2.4444444444444445e-05,
      "loss": 0.5375,
      "step": 12
    },
    {
      "epoch": 0.028840820854132,
      "grad_norm": 0.054002657532691956,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 0.6672,
      "step": 13
    },
    {
      "epoch": 0.03105934553521908,
      "grad_norm": 0.052467066794633865,
      "learning_rate": 2.8888888888888885e-05,
      "loss": 0.817,
      "step": 14
    },
    {
      "epoch": 0.033277870216306155,
      "grad_norm": 0.04975535720586777,
      "learning_rate": 3.111111111111111e-05,
      "loss": 0.5462,
      "step": 15
    },
    {
      "epoch": 0.03549639489739324,
      "grad_norm": 0.04383036494255066,
      "learning_rate": 3.333333333333333e-05,
      "loss": 0.6632,
      "step": 16
    },
    {
      "epoch": 0.03771491957848031,
      "grad_norm": 0.05149222910404205,
      "learning_rate": 3.555555555555555e-05,
      "loss": 0.6606,
      "step": 17
    },
    {
      "epoch": 0.03993344425956739,
      "grad_norm": 0.07624880224466324,
      "learning_rate": 3.777777777777777e-05,
      "loss": 0.5515,
      "step": 18
    },
    {
      "epoch": 0.04215196894065446,
      "grad_norm": 0.035006165504455566,
      "learning_rate": 3.9999999999999996e-05,
      "loss": 0.527,
      "step": 19
    },
    {
      "epoch": 0.044370493621741544,
      "grad_norm": 0.04498816654086113,
      "learning_rate": 4.222222222222222e-05,
      "loss": 0.5872,
      "step": 20
    },
    {
      "epoch": 0.04658901830282862,
      "grad_norm": 0.04130326211452484,
      "learning_rate": 4.444444444444444e-05,
      "loss": 0.6591,
      "step": 21
    },
    {
      "epoch": 0.048807542983915694,
      "grad_norm": 0.03968038409948349,
      "learning_rate": 4.6666666666666665e-05,
      "loss": 0.6512,
      "step": 22
    },
    {
      "epoch": 0.051026067665002776,
      "grad_norm": 0.0656369999051094,
      "learning_rate": 4.888888888888889e-05,
      "loss": 0.7001,
      "step": 23
    },
    {
      "epoch": 0.05324459234608985,
      "grad_norm": 0.033299434930086136,
      "learning_rate": 5.111111111111111e-05,
      "loss": 0.5187,
      "step": 24
    },
    {
      "epoch": 0.05546311702717693,
      "grad_norm": 0.056177277117967606,
      "learning_rate": 5.333333333333333e-05,
      "loss": 0.7107,
      "step": 25
    },
    {
      "epoch": 0.057681641708264,
      "grad_norm": 0.060889244079589844,
      "learning_rate": 5.5555555555555545e-05,
      "loss": 0.6376,
      "step": 26
    },
    {
      "epoch": 0.059900166389351084,
      "grad_norm": 0.042255859822034836,
      "learning_rate": 5.777777777777777e-05,
      "loss": 0.7609,
      "step": 27
    },
    {
      "epoch": 0.06211869107043816,
      "grad_norm": 0.03910442814230919,
      "learning_rate": 5.9999999999999995e-05,
      "loss": 0.5018,
      "step": 28
    },
    {
      "epoch": 0.06433721575152523,
      "grad_norm": 0.04364166408777237,
      "learning_rate": 6.222222222222222e-05,
      "loss": 0.7034,
      "step": 29
    },
    {
      "epoch": 0.06655574043261231,
      "grad_norm": 0.05427662655711174,
      "learning_rate": 6.444444444444444e-05,
      "loss": 0.5801,
      "step": 30
    },
    {
      "epoch": 0.06877426511369938,
      "grad_norm": 0.035717662423849106,
      "learning_rate": 6.666666666666666e-05,
      "loss": 0.5902,
      "step": 31
    },
    {
      "epoch": 0.07099278979478647,
      "grad_norm": 0.05804882198572159,
      "learning_rate": 6.888888888888888e-05,
      "loss": 0.7236,
      "step": 32
    },
    {
      "epoch": 0.07321131447587355,
      "grad_norm": 0.05491464212536812,
      "learning_rate": 7.11111111111111e-05,
      "loss": 0.6458,
      "step": 33
    },
    {
      "epoch": 0.07542983915696062,
      "grad_norm": 0.0616191141307354,
      "learning_rate": 7.333333333333332e-05,
      "loss": 0.6746,
      "step": 34
    },
    {
      "epoch": 0.0776483638380477,
      "grad_norm": 0.055012296885252,
      "learning_rate": 7.555555555555554e-05,
      "loss": 0.5886,
      "step": 35
    },
    {
      "epoch": 0.07986688851913477,
      "grad_norm": 0.03640875592827797,
      "learning_rate": 7.777777777777777e-05,
      "loss": 0.6063,
      "step": 36
    },
    {
      "epoch": 0.08208541320022185,
      "grad_norm": 0.05145784467458725,
      "learning_rate": 7.999999999999999e-05,
      "loss": 0.7056,
      "step": 37
    },
    {
      "epoch": 0.08430393788130892,
      "grad_norm": 0.05369505286216736,
      "learning_rate": 8.222222222222222e-05,
      "loss": 0.7136,
      "step": 38
    },
    {
      "epoch": 0.08652246256239601,
      "grad_norm": 0.13857048749923706,
      "learning_rate": 8.444444444444444e-05,
      "loss": 0.7172,
      "step": 39
    },
    {
      "epoch": 0.08874098724348309,
      "grad_norm": 0.04781906306743622,
      "learning_rate": 8.666666666666665e-05,
      "loss": 0.6697,
      "step": 40
    },
    {
      "epoch": 0.09095951192457016,
      "grad_norm": 0.04735789820551872,
      "learning_rate": 8.888888888888888e-05,
      "loss": 0.584,
      "step": 41
    },
    {
      "epoch": 0.09317803660565724,
      "grad_norm": 0.05417654290795326,
      "learning_rate": 9.11111111111111e-05,
      "loss": 0.605,
      "step": 42
    },
    {
      "epoch": 0.09539656128674431,
      "grad_norm": 0.09498804062604904,
      "learning_rate": 9.333333333333333e-05,
      "loss": 0.7129,
      "step": 43
    },
    {
      "epoch": 0.09761508596783139,
      "grad_norm": 0.03670273721218109,
      "learning_rate": 9.555555555555555e-05,
      "loss": 0.6376,
      "step": 44
    },
    {
      "epoch": 0.09983361064891846,
      "grad_norm": 0.06774840503931046,
      "learning_rate": 9.777777777777778e-05,
      "loss": 0.6658,
      "step": 45
    },
    {
      "epoch": 0.10205213533000555,
      "grad_norm": 0.09117970615625381,
      "learning_rate": 9.999999999999999e-05,
      "loss": 0.6101,
      "step": 46
    },
    {
      "epoch": 0.10427066001109263,
      "grad_norm": 0.061939775943756104,
      "learning_rate": 0.00010222222222222222,
      "loss": 0.7753,
      "step": 47
    },
    {
      "epoch": 0.1064891846921797,
      "grad_norm": 0.05575374886393547,
      "learning_rate": 0.00010444444444444444,
      "loss": 0.6227,
      "step": 48
    },
    {
      "epoch": 0.10870770937326678,
      "grad_norm": 0.0446038581430912,
      "learning_rate": 0.00010666666666666667,
      "loss": 0.756,
      "step": 49
    },
    {
      "epoch": 0.11092623405435385,
      "grad_norm": 0.06581006199121475,
      "learning_rate": 0.00010888888888888889,
      "loss": 0.6761,
      "step": 50
    },
    {
      "epoch": 0.11314475873544093,
      "grad_norm": 0.04741488769650459,
      "learning_rate": 0.00011111111111111109,
      "loss": 0.6872,
      "step": 51
    },
    {
      "epoch": 0.115363283416528,
      "grad_norm": 0.03598475456237793,
      "learning_rate": 0.00011333333333333331,
      "loss": 0.7238,
      "step": 52
    },
    {
      "epoch": 0.11758180809761509,
      "grad_norm": 0.05006685480475426,
      "learning_rate": 0.00011555555555555554,
      "loss": 0.7773,
      "step": 53
    },
    {
      "epoch": 0.11980033277870217,
      "grad_norm": 0.05483674257993698,
      "learning_rate": 0.00011777777777777776,
      "loss": 0.7814,
      "step": 54
    },
    {
      "epoch": 0.12201885745978924,
      "grad_norm": 0.03535892069339752,
      "learning_rate": 0.00011999999999999999,
      "loss": 0.4962,
      "step": 55
    },
    {
      "epoch": 0.12423738214087632,
      "grad_norm": 0.040345534682273865,
      "learning_rate": 0.00012222222222222221,
      "loss": 0.5804,
      "step": 56
    },
    {
      "epoch": 0.1264559068219634,
      "grad_norm": 0.05404891446232796,
      "learning_rate": 0.00012444444444444444,
      "loss": 0.5652,
      "step": 57
    },
    {
      "epoch": 0.12867443150305047,
      "grad_norm": 0.05984383448958397,
      "learning_rate": 0.00012666666666666666,
      "loss": 0.6575,
      "step": 58
    },
    {
      "epoch": 0.13089295618413754,
      "grad_norm": 0.06058034300804138,
      "learning_rate": 0.0001288888888888889,
      "loss": 0.5918,
      "step": 59
    },
    {
      "epoch": 0.13311148086522462,
      "grad_norm": 0.034835513681173325,
      "learning_rate": 0.00013111111111111111,
      "loss": 0.4576,
      "step": 60
    },
    {
      "epoch": 0.1353300055463117,
      "grad_norm": 0.04703117534518242,
      "learning_rate": 0.0001333333333333333,
      "loss": 0.6652,
      "step": 61
    },
    {
      "epoch": 0.13754853022739877,
      "grad_norm": 0.0401008166372776,
      "learning_rate": 0.00013555555555555554,
      "loss": 0.6052,
      "step": 62
    },
    {
      "epoch": 0.13976705490848584,
      "grad_norm": 0.038222476840019226,
      "learning_rate": 0.00013777777777777776,
      "loss": 0.6258,
      "step": 63
    },
    {
      "epoch": 0.14198557958957295,
      "grad_norm": 0.042367398738861084,
      "learning_rate": 0.00014,
      "loss": 0.5761,
      "step": 64
    },
    {
      "epoch": 0.14420410427066002,
      "grad_norm": 0.07417109608650208,
      "learning_rate": 0.0001422222222222222,
      "loss": 0.7253,
      "step": 65
    },
    {
      "epoch": 0.1464226289517471,
      "grad_norm": 0.03332237899303436,
      "learning_rate": 0.0001444444444444444,
      "loss": 0.578,
      "step": 66
    },
    {
      "epoch": 0.14864115363283417,
      "grad_norm": 0.04727284982800484,
      "learning_rate": 0.00014666666666666664,
      "loss": 0.6481,
      "step": 67
    },
    {
      "epoch": 0.15085967831392125,
      "grad_norm": 0.033440254628658295,
      "learning_rate": 0.00014888888888888886,
      "loss": 0.4466,
      "step": 68
    },
    {
      "epoch": 0.15307820299500832,
      "grad_norm": 0.03731625899672508,
      "learning_rate": 0.00015111111111111109,
      "loss": 0.6424,
      "step": 69
    },
    {
      "epoch": 0.1552967276760954,
      "grad_norm": 0.05849206820130348,
      "learning_rate": 0.0001533333333333333,
      "loss": 0.6161,
      "step": 70
    },
    {
      "epoch": 0.15751525235718247,
      "grad_norm": 0.04216286540031433,
      "learning_rate": 0.00015555555555555554,
      "loss": 0.5506,
      "step": 71
    },
    {
      "epoch": 0.15973377703826955,
      "grad_norm": 0.059060655534267426,
      "learning_rate": 0.00015777777777777776,
      "loss": 0.6234,
      "step": 72
    },
    {
      "epoch": 0.16195230171935662,
      "grad_norm": 0.04549403488636017,
      "learning_rate": 0.00015999999999999999,
      "loss": 0.5948,
      "step": 73
    },
    {
      "epoch": 0.1641708264004437,
      "grad_norm": 0.0403430350124836,
      "learning_rate": 0.0001622222222222222,
      "loss": 0.5449,
      "step": 74
    },
    {
      "epoch": 0.16638935108153077,
      "grad_norm": 0.06389044225215912,
      "learning_rate": 0.00016444444444444444,
      "loss": 0.6324,
      "step": 75
    },
    {
      "epoch": 0.16860787576261785,
      "grad_norm": 0.03827684372663498,
      "learning_rate": 0.00016666666666666666,
      "loss": 0.8163,
      "step": 76
    },
    {
      "epoch": 0.17082640044370492,
      "grad_norm": 0.054330311715602875,
      "learning_rate": 0.00016888888888888889,
      "loss": 0.6964,
      "step": 77
    },
    {
      "epoch": 0.17304492512479203,
      "grad_norm": 0.03159399330615997,
      "learning_rate": 0.0001711111111111111,
      "loss": 0.5382,
      "step": 78
    },
    {
      "epoch": 0.1752634498058791,
      "grad_norm": 0.07049114257097244,
      "learning_rate": 0.0001733333333333333,
      "loss": 0.6753,
      "step": 79
    },
    {
      "epoch": 0.17748197448696618,
      "grad_norm": 0.037046972662210464,
      "learning_rate": 0.00017555555555555553,
      "loss": 0.586,
      "step": 80
    },
    {
      "epoch": 0.17970049916805325,
      "grad_norm": 0.039180293679237366,
      "learning_rate": 0.00017777777777777776,
      "loss": 0.6636,
      "step": 81
    },
    {
      "epoch": 0.18191902384914033,
      "grad_norm": 0.030743183568120003,
      "learning_rate": 0.00017999999999999998,
      "loss": 0.5547,
      "step": 82
    },
    {
      "epoch": 0.1841375485302274,
      "grad_norm": 0.06636806577444077,
      "learning_rate": 0.0001822222222222222,
      "loss": 0.685,
      "step": 83
    },
    {
      "epoch": 0.18635607321131448,
      "grad_norm": 0.04130888357758522,
      "learning_rate": 0.00018444444444444443,
      "loss": 0.5107,
      "step": 84
    },
    {
      "epoch": 0.18857459789240155,
      "grad_norm": 0.1490316540002823,
      "learning_rate": 0.00018666666666666666,
      "loss": 0.7012,
      "step": 85
    },
    {
      "epoch": 0.19079312257348863,
      "grad_norm": 0.041374169290065765,
      "learning_rate": 0.00018888888888888888,
      "loss": 0.5545,
      "step": 86
    },
    {
      "epoch": 0.1930116472545757,
      "grad_norm": 0.05077047646045685,
      "learning_rate": 0.0001911111111111111,
      "loss": 0.659,
      "step": 87
    },
    {
      "epoch": 0.19523017193566278,
      "grad_norm": 0.055017534643411636,
      "learning_rate": 0.00019333333333333333,
      "loss": 0.6515,
      "step": 88
    },
    {
      "epoch": 0.19744869661674985,
      "grad_norm": 0.05250139161944389,
      "learning_rate": 0.00019555555555555556,
      "loss": 0.6305,
      "step": 89
    },
    {
      "epoch": 0.19966722129783693,
      "grad_norm": 0.05778712406754494,
      "learning_rate": 0.00019777777777777776,
      "loss": 0.6275,
      "step": 90
    },
    {
      "epoch": 0.201885745978924,
      "grad_norm": 0.05375876650214195,
      "learning_rate": 0.00019999999999999998,
      "loss": 0.8601,
      "step": 91
    },
    {
      "epoch": 0.2041042706600111,
      "grad_norm": 0.037869423627853394,
      "learning_rate": 0.0002022222222222222,
      "loss": 0.6284,
      "step": 92
    },
    {
      "epoch": 0.20632279534109818,
      "grad_norm": 0.05282152071595192,
      "learning_rate": 0.00020444444444444443,
      "loss": 0.731,
      "step": 93
    },
    {
      "epoch": 0.20854132002218526,
      "grad_norm": 0.058124564588069916,
      "learning_rate": 0.00020666666666666666,
      "loss": 0.6407,
      "step": 94
    },
    {
      "epoch": 0.21075984470327233,
      "grad_norm": 0.05066797509789467,
      "learning_rate": 0.00020888888888888888,
      "loss": 0.6865,
      "step": 95
    },
    {
      "epoch": 0.2129783693843594,
      "grad_norm": 0.043611034750938416,
      "learning_rate": 0.0002111111111111111,
      "loss": 0.7964,
      "step": 96
    },
    {
      "epoch": 0.21519689406544648,
      "grad_norm": 0.04645869880914688,
      "learning_rate": 0.00021333333333333333,
      "loss": 0.5719,
      "step": 97
    },
    {
      "epoch": 0.21741541874653356,
      "grad_norm": 0.031185291707515717,
      "learning_rate": 0.00021555555555555556,
      "loss": 0.5696,
      "step": 98
    },
    {
      "epoch": 0.21963394342762063,
      "grad_norm": 0.05939400568604469,
      "learning_rate": 0.00021777777777777778,
      "loss": 0.7267,
      "step": 99
    },
    {
      "epoch": 0.2218524681087077,
      "grad_norm": 0.0459572859108448,
      "learning_rate": 0.00021999999999999995,
      "loss": 0.7824,
      "step": 100
    },
    {
      "epoch": 0.22407099278979478,
      "grad_norm": 0.047330647706985474,
      "learning_rate": 0.00022222222222222218,
      "loss": 0.5683,
      "step": 101
    },
    {
      "epoch": 0.22628951747088186,
      "grad_norm": 0.03356607258319855,
      "learning_rate": 0.0002244444444444444,
      "loss": 0.6214,
      "step": 102
    },
    {
      "epoch": 0.22850804215196893,
      "grad_norm": 0.03941280394792557,
      "learning_rate": 0.00022666666666666663,
      "loss": 0.644,
      "step": 103
    },
    {
      "epoch": 0.230726566833056,
      "grad_norm": 0.05826086178421974,
      "learning_rate": 0.00022888888888888885,
      "loss": 0.6373,
      "step": 104
    },
    {
      "epoch": 0.23294509151414308,
      "grad_norm": 0.03979550674557686,
      "learning_rate": 0.00023111111111111108,
      "loss": 0.6275,
      "step": 105
    },
    {
      "epoch": 0.23516361619523019,
      "grad_norm": 0.05471813678741455,
      "learning_rate": 0.0002333333333333333,
      "loss": 0.5513,
      "step": 106
    },
    {
      "epoch": 0.23738214087631726,
      "grad_norm": 0.04059416800737381,
      "learning_rate": 0.00023555555555555553,
      "loss": 0.6003,
      "step": 107
    },
    {
      "epoch": 0.23960066555740434,
      "grad_norm": 0.03678559511899948,
      "learning_rate": 0.00023777777777777775,
      "loss": 0.6231,
      "step": 108
    },
    {
      "epoch": 0.2418191902384914,
      "grad_norm": 0.049639247357845306,
      "learning_rate": 0.00023999999999999998,
      "loss": 0.6794,
      "step": 109
    },
    {
      "epoch": 0.24403771491957849,
      "grad_norm": 0.04981416463851929,
      "learning_rate": 0.0002422222222222222,
      "loss": 0.6818,
      "step": 110
    },
    {
      "epoch": 0.24625623960066556,
      "grad_norm": 0.04143702611327171,
      "learning_rate": 0.00024444444444444443,
      "loss": 0.4649,
      "step": 111
    },
    {
      "epoch": 0.24847476428175264,
      "grad_norm": 0.05574282631278038,
      "learning_rate": 0.0002466666666666666,
      "loss": 0.5888,
      "step": 112
    },
    {
      "epoch": 0.2506932889628397,
      "grad_norm": 0.04851828143000603,
      "learning_rate": 0.0002488888888888889,
      "loss": 0.6548,
      "step": 113
    },
    {
      "epoch": 0.2529118136439268,
      "grad_norm": 0.044172223657369614,
      "learning_rate": 0.0002511111111111111,
      "loss": 0.7709,
      "step": 114
    },
    {
      "epoch": 0.25513033832501386,
      "grad_norm": 0.03317507356405258,
      "learning_rate": 0.00025333333333333333,
      "loss": 0.6008,
      "step": 115
    },
    {
      "epoch": 0.25734886300610094,
      "grad_norm": 0.07023453712463379,
      "learning_rate": 0.00025555555555555553,
      "loss": 0.7429,
      "step": 116
    },
    {
      "epoch": 0.259567387687188,
      "grad_norm": 0.069769948720932,
      "learning_rate": 0.0002577777777777778,
      "loss": 0.6759,
      "step": 117
    },
    {
      "epoch": 0.2617859123682751,
      "grad_norm": 0.037692226469516754,
      "learning_rate": 0.00026,
      "loss": 0.5962,
      "step": 118
    },
    {
      "epoch": 0.26400443704936216,
      "grad_norm": 0.04492403194308281,
      "learning_rate": 0.00026222222222222223,
      "loss": 0.7153,
      "step": 119
    },
    {
      "epoch": 0.26622296173044924,
      "grad_norm": 0.05199439823627472,
      "learning_rate": 0.00026444444444444443,
      "loss": 0.8855,
      "step": 120
    },
    {
      "epoch": 0.2684414864115363,
      "grad_norm": 0.051766347140073776,
      "learning_rate": 0.0002666666666666666,
      "loss": 0.613,
      "step": 121
    },
    {
      "epoch": 0.2706600110926234,
      "grad_norm": 0.03881412371993065,
      "learning_rate": 0.0002688888888888889,
      "loss": 0.6853,
      "step": 122
    },
    {
      "epoch": 0.27287853577371046,
      "grad_norm": 0.0362153984606266,
      "learning_rate": 0.0002711111111111111,
      "loss": 0.5035,
      "step": 123
    },
    {
      "epoch": 0.27509706045479754,
      "grad_norm": 0.036816734820604324,
      "learning_rate": 0.00027333333333333333,
      "loss": 0.6543,
      "step": 124
    },
    {
      "epoch": 0.2773155851358846,
      "grad_norm": 0.043365586549043655,
      "learning_rate": 0.0002755555555555555,
      "loss": 0.6407,
      "step": 125
    },
    {
      "epoch": 0.2795341098169717,
      "grad_norm": 0.0375325009226799,
      "learning_rate": 0.0002777777777777778,
      "loss": 0.507,
      "step": 126
    },
    {
      "epoch": 0.2817526344980588,
      "grad_norm": 0.03831303119659424,
      "learning_rate": 0.00028,
      "loss": 0.5423,
      "step": 127
    },
    {
      "epoch": 0.2839711591791459,
      "grad_norm": 0.02818150259554386,
      "learning_rate": 0.00028222222222222223,
      "loss": 0.5521,
      "step": 128
    },
    {
      "epoch": 0.28618968386023297,
      "grad_norm": 0.038800742477178574,
      "learning_rate": 0.0002844444444444444,
      "loss": 0.6802,
      "step": 129
    },
    {
      "epoch": 0.28840820854132004,
      "grad_norm": 0.035592708736658096,
      "learning_rate": 0.0002866666666666667,
      "loss": 0.624,
      "step": 130
    },
    {
      "epoch": 0.2906267332224071,
      "grad_norm": 0.07881312072277069,
      "learning_rate": 0.0002888888888888888,
      "loss": 0.6601,
      "step": 131
    },
    {
      "epoch": 0.2928452579034942,
      "grad_norm": 0.04960599169135094,
      "learning_rate": 0.0002911111111111111,
      "loss": 0.4682,
      "step": 132
    },
    {
      "epoch": 0.29506378258458127,
      "grad_norm": 0.04417654871940613,
      "learning_rate": 0.00029333333333333327,
      "loss": 0.6912,
      "step": 133
    },
    {
      "epoch": 0.29728230726566834,
      "grad_norm": 0.038360413163900375,
      "learning_rate": 0.0002955555555555555,
      "loss": 0.5476,
      "step": 134
    },
    {
      "epoch": 0.2995008319467554,
      "grad_norm": 0.0784471184015274,
      "learning_rate": 0.0002977777777777777,
      "loss": 0.5421,
      "step": 135
    },
    {
      "epoch": 0.3017193566278425,
      "grad_norm": 0.06291938573122025,
      "learning_rate": 0.0003,
      "loss": 0.5011,
      "step": 136
    },
    {
      "epoch": 0.30393788130892957,
      "grad_norm": 0.04521437734365463,
      "learning_rate": 0.00029975308641975307,
      "loss": 0.7747,
      "step": 137
    },
    {
      "epoch": 0.30615640599001664,
      "grad_norm": 0.04320581629872322,
      "learning_rate": 0.00029950617283950616,
      "loss": 0.6061,
      "step": 138
    },
    {
      "epoch": 0.3083749306711037,
      "grad_norm": 0.04171106964349747,
      "learning_rate": 0.00029925925925925926,
      "loss": 0.563,
      "step": 139
    },
    {
      "epoch": 0.3105934553521908,
      "grad_norm": 0.04506370425224304,
      "learning_rate": 0.0002990123456790123,
      "loss": 0.6295,
      "step": 140
    },
    {
      "epoch": 0.31281198003327787,
      "grad_norm": 0.08872256428003311,
      "learning_rate": 0.0002987654320987654,
      "loss": 0.585,
      "step": 141
    },
    {
      "epoch": 0.31503050471436495,
      "grad_norm": 0.03989364206790924,
      "learning_rate": 0.0002985185185185185,
      "loss": 0.5998,
      "step": 142
    },
    {
      "epoch": 0.317249029395452,
      "grad_norm": 0.04013008251786232,
      "learning_rate": 0.0002982716049382716,
      "loss": 0.6724,
      "step": 143
    },
    {
      "epoch": 0.3194675540765391,
      "grad_norm": 0.04329978674650192,
      "learning_rate": 0.0002980246913580247,
      "loss": 0.51,
      "step": 144
    },
    {
      "epoch": 0.32168607875762617,
      "grad_norm": 0.044073592871427536,
      "learning_rate": 0.0002977777777777777,
      "loss": 0.5728,
      "step": 145
    },
    {
      "epoch": 0.32390460343871325,
      "grad_norm": 0.04309270530939102,
      "learning_rate": 0.00029753086419753087,
      "loss": 0.6055,
      "step": 146
    },
    {
      "epoch": 0.3261231281198003,
      "grad_norm": 0.04368756338953972,
      "learning_rate": 0.0002972839506172839,
      "loss": 0.4279,
      "step": 147
    },
    {
      "epoch": 0.3283416528008874,
      "grad_norm": 0.031211353838443756,
      "learning_rate": 0.000297037037037037,
      "loss": 0.5294,
      "step": 148
    },
    {
      "epoch": 0.33056017748197447,
      "grad_norm": 0.03931813687086105,
      "learning_rate": 0.0002967901234567901,
      "loss": 0.62,
      "step": 149
    },
    {
      "epoch": 0.33277870216306155,
      "grad_norm": 0.039884429425001144,
      "learning_rate": 0.0002965432098765432,
      "loss": 0.606,
      "step": 150
    },
    {
      "epoch": 0.3349972268441486,
      "grad_norm": 0.04382513463497162,
      "learning_rate": 0.00029629629629629624,
      "loss": 0.5634,
      "step": 151
    },
    {
      "epoch": 0.3372157515252357,
      "grad_norm": 0.07473330944776535,
      "learning_rate": 0.00029604938271604933,
      "loss": 0.5901,
      "step": 152
    },
    {
      "epoch": 0.33943427620632277,
      "grad_norm": 0.04941010847687721,
      "learning_rate": 0.0002958024691358025,
      "loss": 0.6385,
      "step": 153
    },
    {
      "epoch": 0.34165280088740985,
      "grad_norm": 0.04138413816690445,
      "learning_rate": 0.0002955555555555555,
      "loss": 0.6586,
      "step": 154
    },
    {
      "epoch": 0.343871325568497,
      "grad_norm": 0.04140457138419151,
      "learning_rate": 0.0002953086419753086,
      "loss": 0.667,
      "step": 155
    },
    {
      "epoch": 0.34608985024958405,
      "grad_norm": 0.042775873094797134,
      "learning_rate": 0.0002950617283950617,
      "loss": 0.5238,
      "step": 156
    },
    {
      "epoch": 0.3483083749306711,
      "grad_norm": 0.0553482249379158,
      "learning_rate": 0.0002948148148148148,
      "loss": 0.6493,
      "step": 157
    },
    {
      "epoch": 0.3505268996117582,
      "grad_norm": 0.05174122378230095,
      "learning_rate": 0.00029456790123456785,
      "loss": 0.6222,
      "step": 158
    },
    {
      "epoch": 0.3527454242928453,
      "grad_norm": 0.044321026653051376,
      "learning_rate": 0.00029432098765432095,
      "loss": 0.4924,
      "step": 159
    },
    {
      "epoch": 0.35496394897393235,
      "grad_norm": 0.04143316298723221,
      "learning_rate": 0.00029407407407407404,
      "loss": 0.4963,
      "step": 160
    },
    {
      "epoch": 0.35718247365501943,
      "grad_norm": 0.036954838782548904,
      "learning_rate": 0.00029382716049382714,
      "loss": 0.5638,
      "step": 161
    },
    {
      "epoch": 0.3594009983361065,
      "grad_norm": 0.051504503935575485,
      "learning_rate": 0.00029358024691358023,
      "loss": 0.6538,
      "step": 162
    },
    {
      "epoch": 0.3616195230171936,
      "grad_norm": 0.03560314327478409,
      "learning_rate": 0.00029333333333333327,
      "loss": 0.5898,
      "step": 163
    },
    {
      "epoch": 0.36383804769828065,
      "grad_norm": 0.03258368372917175,
      "learning_rate": 0.0002930864197530864,
      "loss": 0.5363,
      "step": 164
    },
    {
      "epoch": 0.36605657237936773,
      "grad_norm": 0.031641826033592224,
      "learning_rate": 0.00029283950617283946,
      "loss": 0.5203,
      "step": 165
    },
    {
      "epoch": 0.3682750970604548,
      "grad_norm": 0.04365066438913345,
      "learning_rate": 0.00029259259259259256,
      "loss": 0.6559,
      "step": 166
    },
    {
      "epoch": 0.3704936217415419,
      "grad_norm": 0.04250362142920494,
      "learning_rate": 0.00029234567901234565,
      "loss": 0.5871,
      "step": 167
    },
    {
      "epoch": 0.37271214642262895,
      "grad_norm": 0.027095040306448936,
      "learning_rate": 0.00029209876543209875,
      "loss": 0.5425,
      "step": 168
    },
    {
      "epoch": 0.37493067110371603,
      "grad_norm": 0.04397342726588249,
      "learning_rate": 0.00029185185185185184,
      "loss": 0.4722,
      "step": 169
    },
    {
      "epoch": 0.3771491957848031,
      "grad_norm": 0.04245046526193619,
      "learning_rate": 0.0002916049382716049,
      "loss": 0.628,
      "step": 170
    },
    {
      "epoch": 0.3793677204658902,
      "grad_norm": 0.05846960470080376,
      "learning_rate": 0.000291358024691358,
      "loss": 0.6854,
      "step": 171
    },
    {
      "epoch": 0.38158624514697725,
      "grad_norm": 0.04455876722931862,
      "learning_rate": 0.0002911111111111111,
      "loss": 0.5489,
      "step": 172
    },
    {
      "epoch": 0.38380476982806433,
      "grad_norm": 0.05556606501340866,
      "learning_rate": 0.00029086419753086417,
      "loss": 0.645,
      "step": 173
    },
    {
      "epoch": 0.3860232945091514,
      "grad_norm": 0.046242229640483856,
      "learning_rate": 0.00029061728395061726,
      "loss": 0.6639,
      "step": 174
    },
    {
      "epoch": 0.3882418191902385,
      "grad_norm": 0.06647568196058273,
      "learning_rate": 0.00029037037037037036,
      "loss": 0.5309,
      "step": 175
    },
    {
      "epoch": 0.39046034387132555,
      "grad_norm": 0.0378749780356884,
      "learning_rate": 0.00029012345679012345,
      "loss": 0.5801,
      "step": 176
    },
    {
      "epoch": 0.39267886855241263,
      "grad_norm": 0.04971344769001007,
      "learning_rate": 0.0002898765432098765,
      "loss": 0.6388,
      "step": 177
    },
    {
      "epoch": 0.3948973932334997,
      "grad_norm": 0.037592899054288864,
      "learning_rate": 0.0002896296296296296,
      "loss": 0.5621,
      "step": 178
    },
    {
      "epoch": 0.3971159179145868,
      "grad_norm": 0.045531805604696274,
      "learning_rate": 0.0002893827160493827,
      "loss": 0.6523,
      "step": 179
    },
    {
      "epoch": 0.39933444259567386,
      "grad_norm": 0.04259159043431282,
      "learning_rate": 0.0002891358024691358,
      "loss": 0.6095,
      "step": 180
    },
    {
      "epoch": 0.40155296727676093,
      "grad_norm": 0.04374194145202637,
      "learning_rate": 0.0002888888888888888,
      "loss": 0.637,
      "step": 181
    },
    {
      "epoch": 0.403771491957848,
      "grad_norm": 0.035119350999593735,
      "learning_rate": 0.00028864197530864197,
      "loss": 0.5115,
      "step": 182
    },
    {
      "epoch": 0.4059900166389351,
      "grad_norm": 0.04536384716629982,
      "learning_rate": 0.00028839506172839507,
      "loss": 0.546,
      "step": 183
    },
    {
      "epoch": 0.4082085413200222,
      "grad_norm": 0.04097611457109451,
      "learning_rate": 0.0002881481481481481,
      "loss": 0.6053,
      "step": 184
    },
    {
      "epoch": 0.4104270660011093,
      "grad_norm": 0.03778279200196266,
      "learning_rate": 0.0002879012345679012,
      "loss": 0.5192,
      "step": 185
    },
    {
      "epoch": 0.41264559068219636,
      "grad_norm": 0.05951259285211563,
      "learning_rate": 0.0002876543209876543,
      "loss": 0.6754,
      "step": 186
    },
    {
      "epoch": 0.41486411536328344,
      "grad_norm": 0.06113960221409798,
      "learning_rate": 0.0002874074074074074,
      "loss": 0.6089,
      "step": 187
    },
    {
      "epoch": 0.4170826400443705,
      "grad_norm": 0.03457357734441757,
      "learning_rate": 0.00028716049382716043,
      "loss": 0.5609,
      "step": 188
    },
    {
      "epoch": 0.4193011647254576,
      "grad_norm": 0.05214797332882881,
      "learning_rate": 0.00028691358024691353,
      "loss": 0.742,
      "step": 189
    },
    {
      "epoch": 0.42151968940654466,
      "grad_norm": 0.03830960392951965,
      "learning_rate": 0.0002866666666666667,
      "loss": 0.5586,
      "step": 190
    },
    {
      "epoch": 0.42373821408763174,
      "grad_norm": 0.04538080468773842,
      "learning_rate": 0.0002864197530864197,
      "loss": 0.5173,
      "step": 191
    },
    {
      "epoch": 0.4259567387687188,
      "grad_norm": 0.04081344977021217,
      "learning_rate": 0.0002861728395061728,
      "loss": 0.658,
      "step": 192
    },
    {
      "epoch": 0.4281752634498059,
      "grad_norm": 0.031216373667120934,
      "learning_rate": 0.0002859259259259259,
      "loss": 0.6022,
      "step": 193
    },
    {
      "epoch": 0.43039378813089296,
      "grad_norm": 0.04306355491280556,
      "learning_rate": 0.000285679012345679,
      "loss": 0.4758,
      "step": 194
    },
    {
      "epoch": 0.43261231281198004,
      "grad_norm": 0.050251398235559464,
      "learning_rate": 0.00028543209876543204,
      "loss": 0.5195,
      "step": 195
    },
    {
      "epoch": 0.4348308374930671,
      "grad_norm": 0.05923117324709892,
      "learning_rate": 0.00028518518518518514,
      "loss": 0.7711,
      "step": 196
    },
    {
      "epoch": 0.4370493621741542,
      "grad_norm": 0.038104962557554245,
      "learning_rate": 0.00028493827160493824,
      "loss": 0.6618,
      "step": 197
    },
    {
      "epoch": 0.43926788685524126,
      "grad_norm": 0.03521755710244179,
      "learning_rate": 0.00028469135802469133,
      "loss": 0.5884,
      "step": 198
    },
    {
      "epoch": 0.44148641153632834,
      "grad_norm": 0.048643674701452255,
      "learning_rate": 0.0002844444444444444,
      "loss": 0.6507,
      "step": 199
    },
    {
      "epoch": 0.4437049362174154,
      "grad_norm": 0.034321535378694534,
      "learning_rate": 0.0002841975308641975,
      "loss": 0.6179,
      "step": 200
    },
    {
      "epoch": 0.4459234608985025,
      "grad_norm": 0.03651395067572594,
      "learning_rate": 0.0002839506172839506,
      "loss": 0.6,
      "step": 201
    },
    {
      "epoch": 0.44814198557958956,
      "grad_norm": 0.03452093154191971,
      "learning_rate": 0.00028370370370370366,
      "loss": 0.4804,
      "step": 202
    },
    {
      "epoch": 0.45036051026067664,
      "grad_norm": 0.03303656727075577,
      "learning_rate": 0.00028345679012345675,
      "loss": 0.5927,
      "step": 203
    },
    {
      "epoch": 0.4525790349417637,
      "grad_norm": 0.0902390256524086,
      "learning_rate": 0.00028320987654320985,
      "loss": 0.8343,
      "step": 204
    },
    {
      "epoch": 0.4547975596228508,
      "grad_norm": 0.05407838895916939,
      "learning_rate": 0.00028296296296296294,
      "loss": 0.7052,
      "step": 205
    },
    {
      "epoch": 0.45701608430393786,
      "grad_norm": 0.0356922410428524,
      "learning_rate": 0.00028271604938271604,
      "loss": 0.5915,
      "step": 206
    },
    {
      "epoch": 0.45923460898502494,
      "grad_norm": 0.035561855882406235,
      "learning_rate": 0.0002824691358024691,
      "loss": 0.7087,
      "step": 207
    },
    {
      "epoch": 0.461453133666112,
      "grad_norm": 0.03673597797751427,
      "learning_rate": 0.00028222222222222223,
      "loss": 0.5699,
      "step": 208
    },
    {
      "epoch": 0.4636716583471991,
      "grad_norm": 0.05469575524330139,
      "learning_rate": 0.00028197530864197527,
      "loss": 0.4981,
      "step": 209
    },
    {
      "epoch": 0.46589018302828616,
      "grad_norm": 0.041431788355112076,
      "learning_rate": 0.00028172839506172836,
      "loss": 0.6146,
      "step": 210
    },
    {
      "epoch": 0.46810870770937324,
      "grad_norm": 0.03300826996564865,
      "learning_rate": 0.00028148148148148146,
      "loss": 0.5066,
      "step": 211
    },
    {
      "epoch": 0.47032723239046037,
      "grad_norm": 0.04210164025425911,
      "learning_rate": 0.00028123456790123455,
      "loss": 0.6146,
      "step": 212
    },
    {
      "epoch": 0.47254575707154745,
      "grad_norm": 0.05362192541360855,
      "learning_rate": 0.00028098765432098765,
      "loss": 0.5987,
      "step": 213
    },
    {
      "epoch": 0.4747642817526345,
      "grad_norm": 0.0536411814391613,
      "learning_rate": 0.0002807407407407407,
      "loss": 0.5951,
      "step": 214
    },
    {
      "epoch": 0.4769828064337216,
      "grad_norm": 0.04033851623535156,
      "learning_rate": 0.0002804938271604938,
      "loss": 0.5977,
      "step": 215
    },
    {
      "epoch": 0.47920133111480867,
      "grad_norm": 0.059326376765966415,
      "learning_rate": 0.0002802469135802469,
      "loss": 0.6277,
      "step": 216
    },
    {
      "epoch": 0.48141985579589575,
      "grad_norm": 0.04036042466759682,
      "learning_rate": 0.00028,
      "loss": 0.561,
      "step": 217
    },
    {
      "epoch": 0.4836383804769828,
      "grad_norm": 0.04038504138588905,
      "learning_rate": 0.00027975308641975307,
      "loss": 0.5951,
      "step": 218
    },
    {
      "epoch": 0.4858569051580699,
      "grad_norm": 0.04149752855300903,
      "learning_rate": 0.00027950617283950617,
      "loss": 0.5525,
      "step": 219
    },
    {
      "epoch": 0.48807542983915697,
      "grad_norm": 0.04323798045516014,
      "learning_rate": 0.00027925925925925926,
      "loss": 0.5487,
      "step": 220
    },
    {
      "epoch": 0.49029395452024405,
      "grad_norm": 0.04178311303257942,
      "learning_rate": 0.0002790123456790123,
      "loss": 0.6673,
      "step": 221
    },
    {
      "epoch": 0.4925124792013311,
      "grad_norm": 0.04074716940522194,
      "learning_rate": 0.0002787654320987654,
      "loss": 0.6175,
      "step": 222
    },
    {
      "epoch": 0.4947310038824182,
      "grad_norm": 0.038529422134160995,
      "learning_rate": 0.0002785185185185185,
      "loss": 0.7108,
      "step": 223
    },
    {
      "epoch": 0.49694952856350527,
      "grad_norm": 0.03892909735441208,
      "learning_rate": 0.0002782716049382716,
      "loss": 0.6796,
      "step": 224
    },
    {
      "epoch": 0.49916805324459235,
      "grad_norm": 0.05137438699603081,
      "learning_rate": 0.00027802469135802463,
      "loss": 0.5532,
      "step": 225
    },
    {
      "epoch": 0.5013865779256794,
      "grad_norm": 0.05634597688913345,
      "learning_rate": 0.0002777777777777778,
      "loss": 0.7506,
      "step": 226
    },
    {
      "epoch": 0.5036051026067665,
      "grad_norm": 0.043151311576366425,
      "learning_rate": 0.00027753086419753087,
      "loss": 0.8033,
      "step": 227
    },
    {
      "epoch": 0.5058236272878536,
      "grad_norm": 0.038264982402324677,
      "learning_rate": 0.0002772839506172839,
      "loss": 0.5555,
      "step": 228
    },
    {
      "epoch": 0.5080421519689406,
      "grad_norm": 0.03774739429354668,
      "learning_rate": 0.000277037037037037,
      "loss": 0.6582,
      "step": 229
    },
    {
      "epoch": 0.5102606766500277,
      "grad_norm": 0.03672982379794121,
      "learning_rate": 0.0002767901234567901,
      "loss": 0.6092,
      "step": 230
    },
    {
      "epoch": 0.5124792013311148,
      "grad_norm": 0.06741472333669662,
      "learning_rate": 0.0002765432098765432,
      "loss": 0.4723,
      "step": 231
    },
    {
      "epoch": 0.5146977260122019,
      "grad_norm": 0.03806981071829796,
      "learning_rate": 0.00027629629629629624,
      "loss": 0.6143,
      "step": 232
    },
    {
      "epoch": 0.516916250693289,
      "grad_norm": 0.03872114419937134,
      "learning_rate": 0.00027604938271604934,
      "loss": 0.6412,
      "step": 233
    },
    {
      "epoch": 0.519134775374376,
      "grad_norm": 0.05645802617073059,
      "learning_rate": 0.00027580246913580243,
      "loss": 0.5221,
      "step": 234
    },
    {
      "epoch": 0.5213533000554631,
      "grad_norm": 0.04578409343957901,
      "learning_rate": 0.0002755555555555555,
      "loss": 0.504,
      "step": 235
    },
    {
      "epoch": 0.5235718247365502,
      "grad_norm": 0.03882138058543205,
      "learning_rate": 0.0002753086419753086,
      "loss": 0.6059,
      "step": 236
    },
    {
      "epoch": 0.5257903494176372,
      "grad_norm": 0.03885430470108986,
      "learning_rate": 0.0002750617283950617,
      "loss": 0.6048,
      "step": 237
    },
    {
      "epoch": 0.5280088740987243,
      "grad_norm": 0.04190083593130112,
      "learning_rate": 0.0002748148148148148,
      "loss": 0.6753,
      "step": 238
    },
    {
      "epoch": 0.5302273987798114,
      "grad_norm": 0.04791463539004326,
      "learning_rate": 0.00027456790123456785,
      "loss": 0.5314,
      "step": 239
    },
    {
      "epoch": 0.5324459234608985,
      "grad_norm": 0.036376819014549255,
      "learning_rate": 0.00027432098765432095,
      "loss": 0.4866,
      "step": 240
    },
    {
      "epoch": 0.5346644481419855,
      "grad_norm": 0.05267614126205444,
      "learning_rate": 0.00027407407407407404,
      "loss": 0.5953,
      "step": 241
    },
    {
      "epoch": 0.5368829728230726,
      "grad_norm": 0.04082438722252846,
      "learning_rate": 0.00027382716049382714,
      "loss": 0.4982,
      "step": 242
    },
    {
      "epoch": 0.5391014975041597,
      "grad_norm": 0.032857056707143784,
      "learning_rate": 0.00027358024691358023,
      "loss": 0.4982,
      "step": 243
    },
    {
      "epoch": 0.5413200221852468,
      "grad_norm": 0.031236371025443077,
      "learning_rate": 0.00027333333333333333,
      "loss": 0.5473,
      "step": 244
    },
    {
      "epoch": 0.5435385468663338,
      "grad_norm": 0.051693420857191086,
      "learning_rate": 0.0002730864197530864,
      "loss": 0.6378,
      "step": 245
    },
    {
      "epoch": 0.5457570715474209,
      "grad_norm": 0.0477912575006485,
      "learning_rate": 0.00027283950617283946,
      "loss": 0.4608,
      "step": 246
    },
    {
      "epoch": 0.547975596228508,
      "grad_norm": 0.03433907404541969,
      "learning_rate": 0.00027259259259259256,
      "loss": 0.6146,
      "step": 247
    },
    {
      "epoch": 0.5501941209095951,
      "grad_norm": 0.05301147699356079,
      "learning_rate": 0.00027234567901234565,
      "loss": 0.4906,
      "step": 248
    },
    {
      "epoch": 0.5524126455906821,
      "grad_norm": 0.03305695950984955,
      "learning_rate": 0.00027209876543209875,
      "loss": 0.5339,
      "step": 249
    },
    {
      "epoch": 0.5546311702717692,
      "grad_norm": 0.037618931382894516,
      "learning_rate": 0.00027185185185185184,
      "loss": 0.5238,
      "step": 250
    },
    {
      "epoch": 0.5568496949528563,
      "grad_norm": 0.049261171370744705,
      "learning_rate": 0.0002716049382716049,
      "loss": 0.6533,
      "step": 251
    },
    {
      "epoch": 0.5590682196339434,
      "grad_norm": 0.040279682725667953,
      "learning_rate": 0.00027135802469135803,
      "loss": 0.551,
      "step": 252
    },
    {
      "epoch": 0.5612867443150305,
      "grad_norm": 0.03231072425842285,
      "learning_rate": 0.0002711111111111111,
      "loss": 0.5796,
      "step": 253
    },
    {
      "epoch": 0.5635052689961176,
      "grad_norm": 0.05573466792702675,
      "learning_rate": 0.00027086419753086417,
      "loss": 0.6975,
      "step": 254
    },
    {
      "epoch": 0.5657237936772047,
      "grad_norm": 0.032696038484573364,
      "learning_rate": 0.00027061728395061727,
      "loss": 0.525,
      "step": 255
    },
    {
      "epoch": 0.5679423183582918,
      "grad_norm": 0.03475712984800339,
      "learning_rate": 0.00027037037037037036,
      "loss": 0.5355,
      "step": 256
    },
    {
      "epoch": 0.5701608430393789,
      "grad_norm": 0.03692876175045967,
      "learning_rate": 0.00027012345679012346,
      "loss": 0.597,
      "step": 257
    },
    {
      "epoch": 0.5723793677204659,
      "grad_norm": 0.05656394362449646,
      "learning_rate": 0.0002698765432098765,
      "loss": 0.6815,
      "step": 258
    },
    {
      "epoch": 0.574597892401553,
      "grad_norm": 0.044353850185871124,
      "learning_rate": 0.0002696296296296296,
      "loss": 0.5652,
      "step": 259
    },
    {
      "epoch": 0.5768164170826401,
      "grad_norm": 0.03201902657747269,
      "learning_rate": 0.0002693827160493827,
      "loss": 0.6032,
      "step": 260
    },
    {
      "epoch": 0.5790349417637272,
      "grad_norm": 0.04966944456100464,
      "learning_rate": 0.0002691358024691358,
      "loss": 0.6842,
      "step": 261
    },
    {
      "epoch": 0.5812534664448142,
      "grad_norm": 0.03357621282339096,
      "learning_rate": 0.0002688888888888889,
      "loss": 0.4461,
      "step": 262
    },
    {
      "epoch": 0.5834719911259013,
      "grad_norm": 0.036126840859651566,
      "learning_rate": 0.00026864197530864197,
      "loss": 0.5739,
      "step": 263
    },
    {
      "epoch": 0.5856905158069884,
      "grad_norm": 0.049908597022295,
      "learning_rate": 0.000268395061728395,
      "loss": 0.513,
      "step": 264
    },
    {
      "epoch": 0.5879090404880755,
      "grad_norm": 0.03935200721025467,
      "learning_rate": 0.0002681481481481481,
      "loss": 0.5388,
      "step": 265
    },
    {
      "epoch": 0.5901275651691625,
      "grad_norm": 0.0488467663526535,
      "learning_rate": 0.0002679012345679012,
      "loss": 0.6747,
      "step": 266
    },
    {
      "epoch": 0.5923460898502496,
      "grad_norm": 0.04177908971905708,
      "learning_rate": 0.0002676543209876543,
      "loss": 0.6088,
      "step": 267
    },
    {
      "epoch": 0.5945646145313367,
      "grad_norm": 0.042733579874038696,
      "learning_rate": 0.0002674074074074074,
      "loss": 0.6457,
      "step": 268
    },
    {
      "epoch": 0.5967831392124238,
      "grad_norm": 0.05746898055076599,
      "learning_rate": 0.00026716049382716043,
      "loss": 0.6461,
      "step": 269
    },
    {
      "epoch": 0.5990016638935108,
      "grad_norm": 0.03865082561969757,
      "learning_rate": 0.0002669135802469136,
      "loss": 0.6324,
      "step": 270
    },
    {
      "epoch": 0.6012201885745979,
      "grad_norm": 0.05101585388183594,
      "learning_rate": 0.0002666666666666666,
      "loss": 0.6481,
      "step": 271
    },
    {
      "epoch": 0.603438713255685,
      "grad_norm": 0.03375884145498276,
      "learning_rate": 0.0002664197530864197,
      "loss": 0.442,
      "step": 272
    },
    {
      "epoch": 0.6056572379367721,
      "grad_norm": 0.04302699491381645,
      "learning_rate": 0.0002661728395061728,
      "loss": 0.6234,
      "step": 273
    },
    {
      "epoch": 0.6078757626178591,
      "grad_norm": 0.03469923883676529,
      "learning_rate": 0.0002659259259259259,
      "loss": 0.7142,
      "step": 274
    },
    {
      "epoch": 0.6100942872989462,
      "grad_norm": 0.03414471819996834,
      "learning_rate": 0.000265679012345679,
      "loss": 0.4289,
      "step": 275
    },
    {
      "epoch": 0.6123128119800333,
      "grad_norm": 0.04485994949936867,
      "learning_rate": 0.00026543209876543205,
      "loss": 0.6242,
      "step": 276
    },
    {
      "epoch": 0.6145313366611204,
      "grad_norm": 0.028806578367948532,
      "learning_rate": 0.00026518518518518514,
      "loss": 0.5999,
      "step": 277
    },
    {
      "epoch": 0.6167498613422074,
      "grad_norm": 0.041319966316223145,
      "learning_rate": 0.00026493827160493824,
      "loss": 0.5275,
      "step": 278
    },
    {
      "epoch": 0.6189683860232945,
      "grad_norm": 0.0336199514567852,
      "learning_rate": 0.00026469135802469133,
      "loss": 0.5035,
      "step": 279
    },
    {
      "epoch": 0.6211869107043816,
      "grad_norm": 0.03850705176591873,
      "learning_rate": 0.00026444444444444443,
      "loss": 0.4755,
      "step": 280
    },
    {
      "epoch": 0.6234054353854687,
      "grad_norm": 0.04337645322084427,
      "learning_rate": 0.0002641975308641975,
      "loss": 0.5238,
      "step": 281
    },
    {
      "epoch": 0.6256239600665557,
      "grad_norm": 0.04332086071372032,
      "learning_rate": 0.0002639506172839506,
      "loss": 0.524,
      "step": 282
    },
    {
      "epoch": 0.6278424847476428,
      "grad_norm": 0.02760210633277893,
      "learning_rate": 0.00026370370370370366,
      "loss": 0.4148,
      "step": 283
    },
    {
      "epoch": 0.6300610094287299,
      "grad_norm": 0.03623291850090027,
      "learning_rate": 0.00026345679012345675,
      "loss": 0.5096,
      "step": 284
    },
    {
      "epoch": 0.632279534109817,
      "grad_norm": 0.04340188205242157,
      "learning_rate": 0.00026320987654320985,
      "loss": 0.6064,
      "step": 285
    },
    {
      "epoch": 0.634498058790904,
      "grad_norm": 0.04238671064376831,
      "learning_rate": 0.00026296296296296294,
      "loss": 0.5688,
      "step": 286
    },
    {
      "epoch": 0.6367165834719911,
      "grad_norm": 0.035364728420972824,
      "learning_rate": 0.00026271604938271604,
      "loss": 0.571,
      "step": 287
    },
    {
      "epoch": 0.6389351081530782,
      "grad_norm": 0.07423172891139984,
      "learning_rate": 0.00026246913580246913,
      "loss": 0.5429,
      "step": 288
    },
    {
      "epoch": 0.6411536328341653,
      "grad_norm": 0.03342824801802635,
      "learning_rate": 0.00026222222222222223,
      "loss": 0.6512,
      "step": 289
    },
    {
      "epoch": 0.6433721575152523,
      "grad_norm": 0.08727965503931046,
      "learning_rate": 0.00026197530864197527,
      "loss": 0.6664,
      "step": 290
    },
    {
      "epoch": 0.6455906821963394,
      "grad_norm": 0.054186541587114334,
      "learning_rate": 0.00026172839506172837,
      "loss": 0.6868,
      "step": 291
    },
    {
      "epoch": 0.6478092068774265,
      "grad_norm": 0.03236742317676544,
      "learning_rate": 0.00026148148148148146,
      "loss": 0.6117,
      "step": 292
    },
    {
      "epoch": 0.6500277315585136,
      "grad_norm": 0.032983675599098206,
      "learning_rate": 0.00026123456790123456,
      "loss": 0.5955,
      "step": 293
    },
    {
      "epoch": 0.6522462562396006,
      "grad_norm": 0.0363464280962944,
      "learning_rate": 0.0002609876543209876,
      "loss": 0.5326,
      "step": 294
    },
    {
      "epoch": 0.6544647809206877,
      "grad_norm": 0.041482843458652496,
      "learning_rate": 0.0002607407407407407,
      "loss": 0.455,
      "step": 295
    },
    {
      "epoch": 0.6566833056017748,
      "grad_norm": 0.046176567673683167,
      "learning_rate": 0.00026049382716049384,
      "loss": 0.6812,
      "step": 296
    },
    {
      "epoch": 0.6589018302828619,
      "grad_norm": 0.03784364461898804,
      "learning_rate": 0.0002602469135802469,
      "loss": 0.4871,
      "step": 297
    },
    {
      "epoch": 0.6611203549639489,
      "grad_norm": 0.029925156384706497,
      "learning_rate": 0.00026,
      "loss": 0.5121,
      "step": 298
    },
    {
      "epoch": 0.663338879645036,
      "grad_norm": 0.0403217189013958,
      "learning_rate": 0.00025975308641975307,
      "loss": 0.7325,
      "step": 299
    },
    {
      "epoch": 0.6655574043261231,
      "grad_norm": 0.04283583164215088,
      "learning_rate": 0.00025950617283950617,
      "loss": 0.6667,
      "step": 300
    },
    {
      "epoch": 0.6677759290072102,
      "grad_norm": 0.048798151314258575,
      "learning_rate": 0.0002592592592592592,
      "loss": 0.5183,
      "step": 301
    },
    {
      "epoch": 0.6699944536882972,
      "grad_norm": 0.04228866472840309,
      "learning_rate": 0.0002590123456790123,
      "loss": 0.632,
      "step": 302
    },
    {
      "epoch": 0.6722129783693843,
      "grad_norm": 0.03235164284706116,
      "learning_rate": 0.0002587654320987654,
      "loss": 0.502,
      "step": 303
    },
    {
      "epoch": 0.6744315030504714,
      "grad_norm": 0.037360530346632004,
      "learning_rate": 0.0002585185185185185,
      "loss": 0.4523,
      "step": 304
    },
    {
      "epoch": 0.6766500277315585,
      "grad_norm": 0.03517124056816101,
      "learning_rate": 0.0002582716049382716,
      "loss": 0.5807,
      "step": 305
    },
    {
      "epoch": 0.6788685524126455,
      "grad_norm": 0.04381927102804184,
      "learning_rate": 0.0002580246913580247,
      "loss": 0.4927,
      "step": 306
    },
    {
      "epoch": 0.6810870770937326,
      "grad_norm": 0.04483035206794739,
      "learning_rate": 0.0002577777777777778,
      "loss": 0.6275,
      "step": 307
    },
    {
      "epoch": 0.6833056017748197,
      "grad_norm": 0.04361843690276146,
      "learning_rate": 0.0002575308641975308,
      "loss": 0.6428,
      "step": 308
    },
    {
      "epoch": 0.6855241264559068,
      "grad_norm": 0.040627527981996536,
      "learning_rate": 0.0002572839506172839,
      "loss": 0.5228,
      "step": 309
    },
    {
      "epoch": 0.687742651136994,
      "grad_norm": 0.040374595671892166,
      "learning_rate": 0.000257037037037037,
      "loss": 0.494,
      "step": 310
    },
    {
      "epoch": 0.689961175818081,
      "grad_norm": 0.04171556979417801,
      "learning_rate": 0.0002567901234567901,
      "loss": 0.6831,
      "step": 311
    },
    {
      "epoch": 0.6921797004991681,
      "grad_norm": 0.038119278848171234,
      "learning_rate": 0.0002565432098765432,
      "loss": 0.5827,
      "step": 312
    },
    {
      "epoch": 0.6943982251802552,
      "grad_norm": 0.03851034492254257,
      "learning_rate": 0.00025629629629629624,
      "loss": 0.5368,
      "step": 313
    },
    {
      "epoch": 0.6966167498613423,
      "grad_norm": 0.04358214512467384,
      "learning_rate": 0.0002560493827160494,
      "loss": 0.6102,
      "step": 314
    },
    {
      "epoch": 0.6988352745424293,
      "grad_norm": 0.03442758694291115,
      "learning_rate": 0.00025580246913580243,
      "loss": 0.5603,
      "step": 315
    },
    {
      "epoch": 0.7010537992235164,
      "grad_norm": 0.035294756293296814,
      "learning_rate": 0.00025555555555555553,
      "loss": 0.51,
      "step": 316
    },
    {
      "epoch": 0.7032723239046035,
      "grad_norm": 0.03068569488823414,
      "learning_rate": 0.0002553086419753086,
      "loss": 0.4247,
      "step": 317
    },
    {
      "epoch": 0.7054908485856906,
      "grad_norm": 0.04096401110291481,
      "learning_rate": 0.0002550617283950617,
      "loss": 0.4086,
      "step": 318
    },
    {
      "epoch": 0.7077093732667776,
      "grad_norm": 0.03940800949931145,
      "learning_rate": 0.0002548148148148148,
      "loss": 0.6127,
      "step": 319
    },
    {
      "epoch": 0.7099278979478647,
      "grad_norm": 0.046139538288116455,
      "learning_rate": 0.00025456790123456785,
      "loss": 0.6575,
      "step": 320
    },
    {
      "epoch": 0.7121464226289518,
      "grad_norm": 0.030355270951986313,
      "learning_rate": 0.00025432098765432095,
      "loss": 0.5003,
      "step": 321
    },
    {
      "epoch": 0.7143649473100389,
      "grad_norm": 0.04158418998122215,
      "learning_rate": 0.00025407407407407404,
      "loss": 0.6059,
      "step": 322
    },
    {
      "epoch": 0.7165834719911259,
      "grad_norm": 0.041633863002061844,
      "learning_rate": 0.00025382716049382714,
      "loss": 0.5293,
      "step": 323
    },
    {
      "epoch": 0.718801996672213,
      "grad_norm": 0.057849619537591934,
      "learning_rate": 0.00025358024691358023,
      "loss": 0.5765,
      "step": 324
    },
    {
      "epoch": 0.7210205213533001,
      "grad_norm": 0.034538384526968,
      "learning_rate": 0.00025333333333333333,
      "loss": 0.4692,
      "step": 325
    },
    {
      "epoch": 0.7232390460343872,
      "grad_norm": 0.03561645746231079,
      "learning_rate": 0.0002530864197530864,
      "loss": 0.5848,
      "step": 326
    },
    {
      "epoch": 0.7254575707154742,
      "grad_norm": 0.03822767361998558,
      "learning_rate": 0.00025283950617283947,
      "loss": 0.6313,
      "step": 327
    },
    {
      "epoch": 0.7276760953965613,
      "grad_norm": 0.04086768254637718,
      "learning_rate": 0.00025259259259259256,
      "loss": 0.6092,
      "step": 328
    },
    {
      "epoch": 0.7298946200776484,
      "grad_norm": 0.04021381959319115,
      "learning_rate": 0.00025234567901234566,
      "loss": 0.4819,
      "step": 329
    },
    {
      "epoch": 0.7321131447587355,
      "grad_norm": 0.047175463289022446,
      "learning_rate": 0.00025209876543209875,
      "loss": 0.539,
      "step": 330
    },
    {
      "epoch": 0.7343316694398225,
      "grad_norm": 0.03492865711450577,
      "learning_rate": 0.0002518518518518518,
      "loss": 0.7038,
      "step": 331
    },
    {
      "epoch": 0.7365501941209096,
      "grad_norm": 0.03879347816109657,
      "learning_rate": 0.00025160493827160494,
      "loss": 0.594,
      "step": 332
    },
    {
      "epoch": 0.7387687188019967,
      "grad_norm": 0.03739601746201515,
      "learning_rate": 0.00025135802469135804,
      "loss": 0.4747,
      "step": 333
    },
    {
      "epoch": 0.7409872434830838,
      "grad_norm": 0.0405651293694973,
      "learning_rate": 0.0002511111111111111,
      "loss": 0.6137,
      "step": 334
    },
    {
      "epoch": 0.7432057681641708,
      "grad_norm": 0.046391360461711884,
      "learning_rate": 0.00025086419753086417,
      "loss": 0.6375,
      "step": 335
    },
    {
      "epoch": 0.7454242928452579,
      "grad_norm": 0.03695761412382126,
      "learning_rate": 0.00025061728395061727,
      "loss": 0.609,
      "step": 336
    },
    {
      "epoch": 0.747642817526345,
      "grad_norm": 0.04839314520359039,
      "learning_rate": 0.00025037037037037036,
      "loss": 0.6489,
      "step": 337
    },
    {
      "epoch": 0.7498613422074321,
      "grad_norm": 0.0365786999464035,
      "learning_rate": 0.0002501234567901234,
      "loss": 0.628,
      "step": 338
    },
    {
      "epoch": 0.7520798668885191,
      "grad_norm": 0.05373973399400711,
      "learning_rate": 0.0002498765432098765,
      "loss": 0.6232,
      "step": 339
    },
    {
      "epoch": 0.7542983915696062,
      "grad_norm": 0.041798315942287445,
      "learning_rate": 0.00024962962962962965,
      "loss": 0.6676,
      "step": 340
    },
    {
      "epoch": 0.7565169162506933,
      "grad_norm": 0.059344369918107986,
      "learning_rate": 0.0002493827160493827,
      "loss": 0.5558,
      "step": 341
    },
    {
      "epoch": 0.7587354409317804,
      "grad_norm": 0.042829856276512146,
      "learning_rate": 0.0002491358024691358,
      "loss": 0.437,
      "step": 342
    },
    {
      "epoch": 0.7609539656128674,
      "grad_norm": 0.03888658434152603,
      "learning_rate": 0.0002488888888888889,
      "loss": 0.4491,
      "step": 343
    },
    {
      "epoch": 0.7631724902939545,
      "grad_norm": 0.04601922631263733,
      "learning_rate": 0.000248641975308642,
      "loss": 0.6218,
      "step": 344
    },
    {
      "epoch": 0.7653910149750416,
      "grad_norm": 0.054116930812597275,
      "learning_rate": 0.000248395061728395,
      "loss": 0.6244,
      "step": 345
    },
    {
      "epoch": 0.7676095396561287,
      "grad_norm": 0.03604884818196297,
      "learning_rate": 0.0002481481481481481,
      "loss": 0.5723,
      "step": 346
    },
    {
      "epoch": 0.7698280643372157,
      "grad_norm": 0.03583709895610809,
      "learning_rate": 0.0002479012345679012,
      "loss": 0.501,
      "step": 347
    },
    {
      "epoch": 0.7720465890183028,
      "grad_norm": 0.04517853260040283,
      "learning_rate": 0.0002476543209876543,
      "loss": 0.5385,
      "step": 348
    },
    {
      "epoch": 0.7742651136993899,
      "grad_norm": 0.03331049531698227,
      "learning_rate": 0.0002474074074074074,
      "loss": 0.4984,
      "step": 349
    },
    {
      "epoch": 0.776483638380477,
      "grad_norm": 0.13833943009376526,
      "learning_rate": 0.0002471604938271605,
      "loss": 0.7639,
      "step": 350
    },
    {
      "epoch": 0.778702163061564,
      "grad_norm": 0.0343298576772213,
      "learning_rate": 0.0002469135802469136,
      "loss": 0.4547,
      "step": 351
    },
    {
      "epoch": 0.7809206877426511,
      "grad_norm": 0.03798506781458855,
      "learning_rate": 0.0002466666666666666,
      "loss": 0.5778,
      "step": 352
    },
    {
      "epoch": 0.7831392124237382,
      "grad_norm": 0.03674226626753807,
      "learning_rate": 0.0002464197530864197,
      "loss": 0.5771,
      "step": 353
    },
    {
      "epoch": 0.7853577371048253,
      "grad_norm": 0.05443619191646576,
      "learning_rate": 0.0002461728395061728,
      "loss": 0.5674,
      "step": 354
    },
    {
      "epoch": 0.7875762617859123,
      "grad_norm": 0.060172948986291885,
      "learning_rate": 0.0002459259259259259,
      "loss": 0.7703,
      "step": 355
    },
    {
      "epoch": 0.7897947864669994,
      "grad_norm": 0.038806743919849396,
      "learning_rate": 0.000245679012345679,
      "loss": 0.625,
      "step": 356
    },
    {
      "epoch": 0.7920133111480865,
      "grad_norm": 0.04248486086726189,
      "learning_rate": 0.00024543209876543205,
      "loss": 0.6148,
      "step": 357
    },
    {
      "epoch": 0.7942318358291736,
      "grad_norm": 0.04469778388738632,
      "learning_rate": 0.0002451851851851852,
      "loss": 0.5559,
      "step": 358
    },
    {
      "epoch": 0.7964503605102606,
      "grad_norm": 0.042841680347919464,
      "learning_rate": 0.00024493827160493824,
      "loss": 0.5551,
      "step": 359
    },
    {
      "epoch": 0.7986688851913477,
      "grad_norm": 0.04645748436450958,
      "learning_rate": 0.00024469135802469133,
      "loss": 0.6842,
      "step": 360
    },
    {
      "epoch": 0.8008874098724348,
      "grad_norm": 0.03864132985472679,
      "learning_rate": 0.00024444444444444443,
      "loss": 0.578,
      "step": 361
    },
    {
      "epoch": 0.8031059345535219,
      "grad_norm": 0.04636433348059654,
      "learning_rate": 0.0002441975308641975,
      "loss": 0.4613,
      "step": 362
    },
    {
      "epoch": 0.8053244592346089,
      "grad_norm": 0.034862831234931946,
      "learning_rate": 0.00024395061728395062,
      "loss": 0.5521,
      "step": 363
    },
    {
      "epoch": 0.807542983915696,
      "grad_norm": 0.08976205438375473,
      "learning_rate": 0.0002437037037037037,
      "loss": 0.7389,
      "step": 364
    },
    {
      "epoch": 0.8097615085967831,
      "grad_norm": 0.04147373139858246,
      "learning_rate": 0.00024345679012345678,
      "loss": 0.7361,
      "step": 365
    },
    {
      "epoch": 0.8119800332778702,
      "grad_norm": 0.04085776209831238,
      "learning_rate": 0.00024320987654320985,
      "loss": 0.4503,
      "step": 366
    },
    {
      "epoch": 0.8141985579589573,
      "grad_norm": 0.052295979112386703,
      "learning_rate": 0.00024296296296296295,
      "loss": 0.7464,
      "step": 367
    },
    {
      "epoch": 0.8164170826400444,
      "grad_norm": 0.03480244055390358,
      "learning_rate": 0.00024271604938271601,
      "loss": 0.5633,
      "step": 368
    },
    {
      "epoch": 0.8186356073211315,
      "grad_norm": 0.05535735934972763,
      "learning_rate": 0.0002424691358024691,
      "loss": 0.6954,
      "step": 369
    },
    {
      "epoch": 0.8208541320022186,
      "grad_norm": 0.03231757879257202,
      "learning_rate": 0.0002422222222222222,
      "loss": 0.6278,
      "step": 370
    },
    {
      "epoch": 0.8230726566833056,
      "grad_norm": 0.03065868653357029,
      "learning_rate": 0.00024197530864197527,
      "loss": 0.5191,
      "step": 371
    },
    {
      "epoch": 0.8252911813643927,
      "grad_norm": 0.04809517413377762,
      "learning_rate": 0.0002417283950617284,
      "loss": 0.6115,
      "step": 372
    },
    {
      "epoch": 0.8275097060454798,
      "grad_norm": 0.044137705117464066,
      "learning_rate": 0.00024148148148148146,
      "loss": 0.5499,
      "step": 373
    },
    {
      "epoch": 0.8297282307265669,
      "grad_norm": 0.04987265169620514,
      "learning_rate": 0.00024123456790123456,
      "loss": 0.5585,
      "step": 374
    },
    {
      "epoch": 0.831946755407654,
      "grad_norm": 0.031602151691913605,
      "learning_rate": 0.00024098765432098763,
      "loss": 0.474,
      "step": 375
    },
    {
      "epoch": 0.834165280088741,
      "grad_norm": 0.03339662030339241,
      "learning_rate": 0.00024074074074074072,
      "loss": 0.6836,
      "step": 376
    },
    {
      "epoch": 0.8363838047698281,
      "grad_norm": 0.043529074639081955,
      "learning_rate": 0.0002404938271604938,
      "loss": 0.6452,
      "step": 377
    },
    {
      "epoch": 0.8386023294509152,
      "grad_norm": 0.052328046411275864,
      "learning_rate": 0.00024024691358024688,
      "loss": 0.8168,
      "step": 378
    },
    {
      "epoch": 0.8408208541320022,
      "grad_norm": 0.04468843713402748,
      "learning_rate": 0.00023999999999999998,
      "loss": 0.5779,
      "step": 379
    },
    {
      "epoch": 0.8430393788130893,
      "grad_norm": 0.053680796176195145,
      "learning_rate": 0.00023975308641975305,
      "loss": 0.5029,
      "step": 380
    },
    {
      "epoch": 0.8452579034941764,
      "grad_norm": 0.0473027266561985,
      "learning_rate": 0.00023950617283950617,
      "loss": 0.5874,
      "step": 381
    },
    {
      "epoch": 0.8474764281752635,
      "grad_norm": 0.05462586507201195,
      "learning_rate": 0.00023925925925925924,
      "loss": 0.5289,
      "step": 382
    },
    {
      "epoch": 0.8496949528563505,
      "grad_norm": 0.033070605248212814,
      "learning_rate": 0.00023901234567901233,
      "loss": 0.5906,
      "step": 383
    },
    {
      "epoch": 0.8519134775374376,
      "grad_norm": 0.04231147840619087,
      "learning_rate": 0.0002387654320987654,
      "loss": 0.4241,
      "step": 384
    },
    {
      "epoch": 0.8541320022185247,
      "grad_norm": 0.03161492198705673,
      "learning_rate": 0.0002385185185185185,
      "loss": 0.5004,
      "step": 385
    },
    {
      "epoch": 0.8563505268996118,
      "grad_norm": 0.08939296752214432,
      "learning_rate": 0.0002382716049382716,
      "loss": 0.5706,
      "step": 386
    },
    {
      "epoch": 0.8585690515806988,
      "grad_norm": 0.03739146143198013,
      "learning_rate": 0.00023802469135802466,
      "loss": 0.843,
      "step": 387
    },
    {
      "epoch": 0.8607875762617859,
      "grad_norm": 0.03274370729923248,
      "learning_rate": 0.00023777777777777775,
      "loss": 0.5002,
      "step": 388
    },
    {
      "epoch": 0.863006100942873,
      "grad_norm": 0.04806267097592354,
      "learning_rate": 0.00023753086419753082,
      "loss": 0.5362,
      "step": 389
    },
    {
      "epoch": 0.8652246256239601,
      "grad_norm": 0.04218921810388565,
      "learning_rate": 0.00023728395061728394,
      "loss": 0.6074,
      "step": 390
    },
    {
      "epoch": 0.8674431503050472,
      "grad_norm": 0.03663468733429909,
      "learning_rate": 0.000237037037037037,
      "loss": 0.6034,
      "step": 391
    },
    {
      "epoch": 0.8696616749861342,
      "grad_norm": 0.030716918408870697,
      "learning_rate": 0.0002367901234567901,
      "loss": 0.4713,
      "step": 392
    },
    {
      "epoch": 0.8718801996672213,
      "grad_norm": 0.03910070285201073,
      "learning_rate": 0.0002365432098765432,
      "loss": 0.5223,
      "step": 393
    },
    {
      "epoch": 0.8740987243483084,
      "grad_norm": 0.03394321724772453,
      "learning_rate": 0.00023629629629629627,
      "loss": 0.5342,
      "step": 394
    },
    {
      "epoch": 0.8763172490293955,
      "grad_norm": 0.040968649089336395,
      "learning_rate": 0.00023604938271604937,
      "loss": 0.5139,
      "step": 395
    },
    {
      "epoch": 0.8785357737104825,
      "grad_norm": 0.033750783652067184,
      "learning_rate": 0.00023580246913580243,
      "loss": 0.547,
      "step": 396
    },
    {
      "epoch": 0.8807542983915696,
      "grad_norm": 0.06549157202243805,
      "learning_rate": 0.00023555555555555553,
      "loss": 0.6372,
      "step": 397
    },
    {
      "epoch": 0.8829728230726567,
      "grad_norm": 0.03847658261656761,
      "learning_rate": 0.0002353086419753086,
      "loss": 0.6176,
      "step": 398
    },
    {
      "epoch": 0.8851913477537438,
      "grad_norm": 0.03911906108260155,
      "learning_rate": 0.00023506172839506172,
      "loss": 0.7058,
      "step": 399
    },
    {
      "epoch": 0.8874098724348308,
      "grad_norm": 0.040336448699235916,
      "learning_rate": 0.0002348148148148148,
      "loss": 0.5785,
      "step": 400
    },
    {
      "epoch": 0.8896283971159179,
      "grad_norm": 0.03645087406039238,
      "learning_rate": 0.00023456790123456788,
      "loss": 0.4801,
      "step": 401
    },
    {
      "epoch": 0.891846921797005,
      "grad_norm": 0.04698243737220764,
      "learning_rate": 0.00023432098765432098,
      "loss": 0.5597,
      "step": 402
    },
    {
      "epoch": 0.894065446478092,
      "grad_norm": 0.04328956454992294,
      "learning_rate": 0.00023407407407407405,
      "loss": 0.5218,
      "step": 403
    },
    {
      "epoch": 0.8962839711591791,
      "grad_norm": 0.03931349888443947,
      "learning_rate": 0.00023382716049382714,
      "loss": 0.5575,
      "step": 404
    },
    {
      "epoch": 0.8985024958402662,
      "grad_norm": 0.03999287262558937,
      "learning_rate": 0.0002335802469135802,
      "loss": 0.5811,
      "step": 405
    },
    {
      "epoch": 0.9007210205213533,
      "grad_norm": 0.031024063006043434,
      "learning_rate": 0.0002333333333333333,
      "loss": 0.3933,
      "step": 406
    },
    {
      "epoch": 0.9029395452024404,
      "grad_norm": 0.04489195719361305,
      "learning_rate": 0.00023308641975308637,
      "loss": 0.582,
      "step": 407
    },
    {
      "epoch": 0.9051580698835274,
      "grad_norm": 0.04660533368587494,
      "learning_rate": 0.0002328395061728395,
      "loss": 0.8031,
      "step": 408
    },
    {
      "epoch": 0.9073765945646145,
      "grad_norm": 0.034962672740221024,
      "learning_rate": 0.0002325925925925926,
      "loss": 0.6419,
      "step": 409
    },
    {
      "epoch": 0.9095951192457016,
      "grad_norm": 0.03762185573577881,
      "learning_rate": 0.00023234567901234566,
      "loss": 0.6109,
      "step": 410
    },
    {
      "epoch": 0.9118136439267887,
      "grad_norm": 0.05127018690109253,
      "learning_rate": 0.00023209876543209875,
      "loss": 0.5504,
      "step": 411
    },
    {
      "epoch": 0.9140321686078757,
      "grad_norm": 0.04802126809954643,
      "learning_rate": 0.00023185185185185182,
      "loss": 0.5988,
      "step": 412
    },
    {
      "epoch": 0.9162506932889628,
      "grad_norm": 0.04899923875927925,
      "learning_rate": 0.00023160493827160492,
      "loss": 0.4469,
      "step": 413
    },
    {
      "epoch": 0.9184692179700499,
      "grad_norm": 0.03725762292742729,
      "learning_rate": 0.00023135802469135798,
      "loss": 0.5779,
      "step": 414
    },
    {
      "epoch": 0.920687742651137,
      "grad_norm": 0.0317765548825264,
      "learning_rate": 0.00023111111111111108,
      "loss": 0.5131,
      "step": 415
    },
    {
      "epoch": 0.922906267332224,
      "grad_norm": 0.03806503489613533,
      "learning_rate": 0.0002308641975308642,
      "loss": 0.5754,
      "step": 416
    },
    {
      "epoch": 0.9251247920133111,
      "grad_norm": 0.056015126407146454,
      "learning_rate": 0.00023061728395061727,
      "loss": 0.5678,
      "step": 417
    },
    {
      "epoch": 0.9273433166943982,
      "grad_norm": 0.05800500884652138,
      "learning_rate": 0.00023037037037037036,
      "loss": 0.6307,
      "step": 418
    },
    {
      "epoch": 0.9295618413754853,
      "grad_norm": 0.056719861924648285,
      "learning_rate": 0.00023012345679012343,
      "loss": 0.6126,
      "step": 419
    },
    {
      "epoch": 0.9317803660565723,
      "grad_norm": 0.036754243075847626,
      "learning_rate": 0.00022987654320987653,
      "loss": 0.5456,
      "step": 420
    },
    {
      "epoch": 0.9339988907376594,
      "grad_norm": 0.07285641878843307,
      "learning_rate": 0.0002296296296296296,
      "loss": 0.5645,
      "step": 421
    },
    {
      "epoch": 0.9362174154187465,
      "grad_norm": 0.03208046779036522,
      "learning_rate": 0.0002293827160493827,
      "loss": 0.6952,
      "step": 422
    },
    {
      "epoch": 0.9384359400998337,
      "grad_norm": 0.04581914842128754,
      "learning_rate": 0.00022913580246913579,
      "loss": 0.5942,
      "step": 423
    },
    {
      "epoch": 0.9406544647809207,
      "grad_norm": 0.036682769656181335,
      "learning_rate": 0.00022888888888888885,
      "loss": 0.6186,
      "step": 424
    },
    {
      "epoch": 0.9428729894620078,
      "grad_norm": 0.03525863215327263,
      "learning_rate": 0.00022864197530864198,
      "loss": 0.636,
      "step": 425
    },
    {
      "epoch": 0.9450915141430949,
      "grad_norm": 0.05726366862654686,
      "learning_rate": 0.00022839506172839504,
      "loss": 0.6599,
      "step": 426
    },
    {
      "epoch": 0.947310038824182,
      "grad_norm": 0.03583671525120735,
      "learning_rate": 0.00022814814814814814,
      "loss": 0.5547,
      "step": 427
    },
    {
      "epoch": 0.949528563505269,
      "grad_norm": 0.040966715663671494,
      "learning_rate": 0.0002279012345679012,
      "loss": 0.6016,
      "step": 428
    },
    {
      "epoch": 0.9517470881863561,
      "grad_norm": 0.03824619948863983,
      "learning_rate": 0.0002276543209876543,
      "loss": 0.5923,
      "step": 429
    },
    {
      "epoch": 0.9539656128674432,
      "grad_norm": 0.03335024043917656,
      "learning_rate": 0.00022740740740740737,
      "loss": 0.4582,
      "step": 430
    },
    {
      "epoch": 0.9561841375485303,
      "grad_norm": 0.045678116381168365,
      "learning_rate": 0.00022716049382716047,
      "loss": 0.5915,
      "step": 431
    },
    {
      "epoch": 0.9584026622296173,
      "grad_norm": 0.05383066087961197,
      "learning_rate": 0.00022691358024691356,
      "loss": 0.6695,
      "step": 432
    },
    {
      "epoch": 0.9606211869107044,
      "grad_norm": 0.04088721424341202,
      "learning_rate": 0.00022666666666666663,
      "loss": 0.6278,
      "step": 433
    },
    {
      "epoch": 0.9628397115917915,
      "grad_norm": 0.053601253777742386,
      "learning_rate": 0.00022641975308641975,
      "loss": 0.7239,
      "step": 434
    },
    {
      "epoch": 0.9650582362728786,
      "grad_norm": 0.045154303312301636,
      "learning_rate": 0.00022617283950617282,
      "loss": 0.6115,
      "step": 435
    },
    {
      "epoch": 0.9672767609539656,
      "grad_norm": 0.033632971346378326,
      "learning_rate": 0.00022592592592592591,
      "loss": 0.5753,
      "step": 436
    },
    {
      "epoch": 0.9694952856350527,
      "grad_norm": 0.033672161400318146,
      "learning_rate": 0.00022567901234567898,
      "loss": 0.5517,
      "step": 437
    },
    {
      "epoch": 0.9717138103161398,
      "grad_norm": 0.036619942635297775,
      "learning_rate": 0.00022543209876543208,
      "loss": 0.4632,
      "step": 438
    },
    {
      "epoch": 0.9739323349972269,
      "grad_norm": 0.029008660465478897,
      "learning_rate": 0.00022518518518518517,
      "loss": 0.5606,
      "step": 439
    },
    {
      "epoch": 0.9761508596783139,
      "grad_norm": 0.04997819662094116,
      "learning_rate": 0.00022493827160493824,
      "loss": 0.7404,
      "step": 440
    },
    {
      "epoch": 0.978369384359401,
      "grad_norm": 0.038672931492328644,
      "learning_rate": 0.00022469135802469134,
      "loss": 0.5614,
      "step": 441
    },
    {
      "epoch": 0.9805879090404881,
      "grad_norm": 0.04046705737709999,
      "learning_rate": 0.0002244444444444444,
      "loss": 0.5154,
      "step": 442
    },
    {
      "epoch": 0.9828064337215752,
      "grad_norm": 0.04479946196079254,
      "learning_rate": 0.00022419753086419753,
      "loss": 0.7625,
      "step": 443
    },
    {
      "epoch": 0.9850249584026622,
      "grad_norm": 0.04047355800867081,
      "learning_rate": 0.0002239506172839506,
      "loss": 0.5068,
      "step": 444
    },
    {
      "epoch": 0.9872434830837493,
      "grad_norm": 0.02731996215879917,
      "learning_rate": 0.0002237037037037037,
      "loss": 0.5364,
      "step": 445
    },
    {
      "epoch": 0.9894620077648364,
      "grad_norm": 0.047373272478580475,
      "learning_rate": 0.00022345679012345678,
      "loss": 0.6522,
      "step": 446
    },
    {
      "epoch": 0.9916805324459235,
      "grad_norm": 0.046231307089328766,
      "learning_rate": 0.00022320987654320985,
      "loss": 0.4836,
      "step": 447
    },
    {
      "epoch": 0.9938990571270105,
      "grad_norm": 0.03921741992235184,
      "learning_rate": 0.00022296296296296295,
      "loss": 0.5549,
      "step": 448
    },
    {
      "epoch": 0.9961175818080976,
      "grad_norm": 0.06438081711530685,
      "learning_rate": 0.00022271604938271602,
      "loss": 0.7062,
      "step": 449
    },
    {
      "epoch": 0.9983361064891847,
      "grad_norm": 0.04100671038031578,
      "learning_rate": 0.0002224691358024691,
      "loss": 0.608,
      "step": 450
    },
    {
      "epoch": 1.0005546311702718,
      "grad_norm": 0.03336994722485542,
      "learning_rate": 0.00022222222222222218,
      "loss": 0.7356,
      "step": 451
    },
    {
      "epoch": 1.0027731558513588,
      "grad_norm": 0.042254332453012466,
      "learning_rate": 0.0002219753086419753,
      "loss": 0.6674,
      "step": 452
    },
    {
      "epoch": 1.004991680532446,
      "grad_norm": 0.04811901971697807,
      "learning_rate": 0.0002217283950617284,
      "loss": 0.5671,
      "step": 453
    },
    {
      "epoch": 1.007210205213533,
      "grad_norm": 0.028940865769982338,
      "learning_rate": 0.00022148148148148146,
      "loss": 0.4807,
      "step": 454
    },
    {
      "epoch": 1.00942872989462,
      "grad_norm": 0.03878101333975792,
      "learning_rate": 0.00022123456790123456,
      "loss": 0.4682,
      "step": 455
    },
    {
      "epoch": 1.0116472545757071,
      "grad_norm": 0.040350984781980515,
      "learning_rate": 0.00022098765432098763,
      "loss": 0.6472,
      "step": 456
    },
    {
      "epoch": 1.0138657792567942,
      "grad_norm": 0.060598164796829224,
      "learning_rate": 0.00022074074074074072,
      "loss": 0.4783,
      "step": 457
    },
    {
      "epoch": 1.0160843039378813,
      "grad_norm": 0.030727174133062363,
      "learning_rate": 0.0002204938271604938,
      "loss": 0.4676,
      "step": 458
    },
    {
      "epoch": 1.0183028286189684,
      "grad_norm": 0.0507417730987072,
      "learning_rate": 0.00022024691358024689,
      "loss": 0.5272,
      "step": 459
    },
    {
      "epoch": 1.0205213533000554,
      "grad_norm": 0.032012972980737686,
      "learning_rate": 0.00021999999999999995,
      "loss": 0.444,
      "step": 460
    },
    {
      "epoch": 1.0227398779811425,
      "grad_norm": 0.05119691789150238,
      "learning_rate": 0.00021975308641975308,
      "loss": 0.4803,
      "step": 461
    },
    {
      "epoch": 1.0249584026622296,
      "grad_norm": 0.07903239130973816,
      "learning_rate": 0.00021950617283950617,
      "loss": 0.6335,
      "step": 462
    },
    {
      "epoch": 1.0271769273433167,
      "grad_norm": 0.0410827212035656,
      "learning_rate": 0.00021925925925925924,
      "loss": 0.4738,
      "step": 463
    },
    {
      "epoch": 1.0293954520244037,
      "grad_norm": 0.03638838604092598,
      "learning_rate": 0.00021901234567901233,
      "loss": 0.5126,
      "step": 464
    },
    {
      "epoch": 1.0316139767054908,
      "grad_norm": 0.04378284513950348,
      "learning_rate": 0.0002187654320987654,
      "loss": 0.5453,
      "step": 465
    },
    {
      "epoch": 1.033832501386578,
      "grad_norm": 0.04775509238243103,
      "learning_rate": 0.0002185185185185185,
      "loss": 0.5637,
      "step": 466
    },
    {
      "epoch": 1.036051026067665,
      "grad_norm": 0.05937077850103378,
      "learning_rate": 0.00021827160493827157,
      "loss": 0.6605,
      "step": 467
    },
    {
      "epoch": 1.038269550748752,
      "grad_norm": 0.059186168015003204,
      "learning_rate": 0.00021802469135802466,
      "loss": 0.713,
      "step": 468
    },
    {
      "epoch": 1.0404880754298391,
      "grad_norm": 0.035836439579725266,
      "learning_rate": 0.00021777777777777778,
      "loss": 0.5354,
      "step": 469
    },
    {
      "epoch": 1.0427066001109262,
      "grad_norm": 0.05742331221699715,
      "learning_rate": 0.00021753086419753085,
      "loss": 0.542,
      "step": 470
    },
    {
      "epoch": 1.0449251247920133,
      "grad_norm": 0.03966318070888519,
      "learning_rate": 0.00021728395061728395,
      "loss": 0.5614,
      "step": 471
    },
    {
      "epoch": 1.0471436494731003,
      "grad_norm": 0.04096744582056999,
      "learning_rate": 0.00021703703703703701,
      "loss": 0.6608,
      "step": 472
    },
    {
      "epoch": 1.0493621741541874,
      "grad_norm": 0.036844652146101,
      "learning_rate": 0.0002167901234567901,
      "loss": 0.5633,
      "step": 473
    },
    {
      "epoch": 1.0515806988352745,
      "grad_norm": 0.029021376743912697,
      "learning_rate": 0.00021654320987654318,
      "loss": 0.6102,
      "step": 474
    },
    {
      "epoch": 1.0537992235163616,
      "grad_norm": 0.045701682567596436,
      "learning_rate": 0.00021629629629629627,
      "loss": 0.5606,
      "step": 475
    },
    {
      "epoch": 1.0560177481974486,
      "grad_norm": 0.043971139937639236,
      "learning_rate": 0.00021604938271604937,
      "loss": 0.5941,
      "step": 476
    },
    {
      "epoch": 1.0582362728785357,
      "grad_norm": 0.0386858806014061,
      "learning_rate": 0.00021580246913580244,
      "loss": 0.406,
      "step": 477
    },
    {
      "epoch": 1.0604547975596228,
      "grad_norm": 0.03465273231267929,
      "learning_rate": 0.00021555555555555556,
      "loss": 0.4744,
      "step": 478
    },
    {
      "epoch": 1.0626733222407099,
      "grad_norm": 0.039155833423137665,
      "learning_rate": 0.00021530864197530863,
      "loss": 0.5333,
      "step": 479
    },
    {
      "epoch": 1.064891846921797,
      "grad_norm": 0.054260145872831345,
      "learning_rate": 0.00021506172839506172,
      "loss": 0.6998,
      "step": 480
    },
    {
      "epoch": 1.067110371602884,
      "grad_norm": 0.04836370050907135,
      "learning_rate": 0.0002148148148148148,
      "loss": 0.5366,
      "step": 481
    },
    {
      "epoch": 1.069328896283971,
      "grad_norm": 0.03994794562458992,
      "learning_rate": 0.00021456790123456788,
      "loss": 0.6373,
      "step": 482
    },
    {
      "epoch": 1.0715474209650582,
      "grad_norm": 0.04445124417543411,
      "learning_rate": 0.00021432098765432095,
      "loss": 0.7588,
      "step": 483
    },
    {
      "epoch": 1.0737659456461452,
      "grad_norm": 0.03997334465384483,
      "learning_rate": 0.00021407407407407405,
      "loss": 0.6162,
      "step": 484
    },
    {
      "epoch": 1.0759844703272323,
      "grad_norm": 0.046485934406518936,
      "learning_rate": 0.00021382716049382714,
      "loss": 0.4609,
      "step": 485
    },
    {
      "epoch": 1.0782029950083194,
      "grad_norm": 0.03483448550105095,
      "learning_rate": 0.0002135802469135802,
      "loss": 0.5076,
      "step": 486
    },
    {
      "epoch": 1.0804215196894065,
      "grad_norm": 0.05512923747301102,
      "learning_rate": 0.00021333333333333333,
      "loss": 0.4532,
      "step": 487
    },
    {
      "epoch": 1.0826400443704935,
      "grad_norm": 0.038199152797460556,
      "learning_rate": 0.0002130864197530864,
      "loss": 0.6257,
      "step": 488
    },
    {
      "epoch": 1.0848585690515806,
      "grad_norm": 0.04597720876336098,
      "learning_rate": 0.0002128395061728395,
      "loss": 0.8345,
      "step": 489
    },
    {
      "epoch": 1.0870770937326677,
      "grad_norm": 0.03937354311347008,
      "learning_rate": 0.00021259259259259256,
      "loss": 0.6976,
      "step": 490
    },
    {
      "epoch": 1.0892956184137548,
      "grad_norm": 0.05166208744049072,
      "learning_rate": 0.00021234567901234566,
      "loss": 0.8028,
      "step": 491
    },
    {
      "epoch": 1.0915141430948418,
      "grad_norm": 0.0348212756216526,
      "learning_rate": 0.00021209876543209875,
      "loss": 0.5251,
      "step": 492
    },
    {
      "epoch": 1.093732667775929,
      "grad_norm": 0.044654812663793564,
      "learning_rate": 0.00021185185185185182,
      "loss": 0.7057,
      "step": 493
    },
    {
      "epoch": 1.095951192457016,
      "grad_norm": 0.04829816147685051,
      "learning_rate": 0.00021160493827160492,
      "loss": 0.59,
      "step": 494
    },
    {
      "epoch": 1.098169717138103,
      "grad_norm": 0.036866530776023865,
      "learning_rate": 0.00021135802469135799,
      "loss": 0.4581,
      "step": 495
    },
    {
      "epoch": 1.1003882418191901,
      "grad_norm": 0.03212961181998253,
      "learning_rate": 0.0002111111111111111,
      "loss": 0.6414,
      "step": 496
    },
    {
      "epoch": 1.1026067665002772,
      "grad_norm": 0.03309328854084015,
      "learning_rate": 0.00021086419753086418,
      "loss": 0.6068,
      "step": 497
    },
    {
      "epoch": 1.1048252911813643,
      "grad_norm": 0.04961882904171944,
      "learning_rate": 0.00021061728395061727,
      "loss": 0.5441,
      "step": 498
    },
    {
      "epoch": 1.1070438158624514,
      "grad_norm": 0.04178699851036072,
      "learning_rate": 0.00021037037037037037,
      "loss": 0.5958,
      "step": 499
    },
    {
      "epoch": 1.1092623405435384,
      "grad_norm": 0.03949996829032898,
      "learning_rate": 0.00021012345679012343,
      "loss": 0.6053,
      "step": 500
    },
    {
      "epoch": 1.1114808652246255,
      "grad_norm": 0.03820941597223282,
      "learning_rate": 0.00020987654320987653,
      "loss": 0.5294,
      "step": 501
    },
    {
      "epoch": 1.1136993899057126,
      "grad_norm": 0.036799024790525436,
      "learning_rate": 0.0002096296296296296,
      "loss": 0.5239,
      "step": 502
    },
    {
      "epoch": 1.1159179145867997,
      "grad_norm": 0.03496834635734558,
      "learning_rate": 0.0002093827160493827,
      "loss": 0.5839,
      "step": 503
    },
    {
      "epoch": 1.1181364392678868,
      "grad_norm": 0.05472426861524582,
      "learning_rate": 0.00020913580246913576,
      "loss": 0.6243,
      "step": 504
    },
    {
      "epoch": 1.1203549639489738,
      "grad_norm": 0.045006245374679565,
      "learning_rate": 0.00020888888888888888,
      "loss": 0.5251,
      "step": 505
    },
    {
      "epoch": 1.122573488630061,
      "grad_norm": 0.037893157452344894,
      "learning_rate": 0.00020864197530864198,
      "loss": 0.5127,
      "step": 506
    },
    {
      "epoch": 1.124792013311148,
      "grad_norm": 0.03406621888279915,
      "learning_rate": 0.00020839506172839505,
      "loss": 0.603,
      "step": 507
    },
    {
      "epoch": 1.127010537992235,
      "grad_norm": 0.04484511539340019,
      "learning_rate": 0.00020814814814814814,
      "loss": 0.5207,
      "step": 508
    },
    {
      "epoch": 1.1292290626733221,
      "grad_norm": 0.032776378095149994,
      "learning_rate": 0.0002079012345679012,
      "loss": 0.4762,
      "step": 509
    },
    {
      "epoch": 1.1314475873544092,
      "grad_norm": 0.03771619871258736,
      "learning_rate": 0.0002076543209876543,
      "loss": 0.6488,
      "step": 510
    },
    {
      "epoch": 1.1336661120354963,
      "grad_norm": 0.04020845144987106,
      "learning_rate": 0.00020740740740740737,
      "loss": 0.6278,
      "step": 511
    },
    {
      "epoch": 1.1358846367165834,
      "grad_norm": 0.041059646755456924,
      "learning_rate": 0.00020716049382716047,
      "loss": 0.7609,
      "step": 512
    },
    {
      "epoch": 1.1381031613976704,
      "grad_norm": 0.04396531358361244,
      "learning_rate": 0.00020691358024691354,
      "loss": 0.6534,
      "step": 513
    },
    {
      "epoch": 1.1403216860787575,
      "grad_norm": 0.056344207376241684,
      "learning_rate": 0.00020666666666666666,
      "loss": 0.4579,
      "step": 514
    },
    {
      "epoch": 1.1425402107598448,
      "grad_norm": 0.04609009250998497,
      "learning_rate": 0.00020641975308641975,
      "loss": 0.6226,
      "step": 515
    },
    {
      "epoch": 1.1447587354409319,
      "grad_norm": 0.03476858139038086,
      "learning_rate": 0.00020617283950617282,
      "loss": 0.6247,
      "step": 516
    },
    {
      "epoch": 1.146977260122019,
      "grad_norm": 0.0459267683327198,
      "learning_rate": 0.00020592592592592592,
      "loss": 0.621,
      "step": 517
    },
    {
      "epoch": 1.149195784803106,
      "grad_norm": 0.0718795508146286,
      "learning_rate": 0.00020567901234567898,
      "loss": 0.5952,
      "step": 518
    },
    {
      "epoch": 1.151414309484193,
      "grad_norm": 0.04744618013501167,
      "learning_rate": 0.00020543209876543208,
      "loss": 0.6774,
      "step": 519
    },
    {
      "epoch": 1.1536328341652802,
      "grad_norm": 0.04409843310713768,
      "learning_rate": 0.00020518518518518515,
      "loss": 0.4977,
      "step": 520
    },
    {
      "epoch": 1.1558513588463672,
      "grad_norm": 0.03489145264029503,
      "learning_rate": 0.00020493827160493824,
      "loss": 0.5668,
      "step": 521
    },
    {
      "epoch": 1.1580698835274543,
      "grad_norm": 0.0285625159740448,
      "learning_rate": 0.00020469135802469136,
      "loss": 0.573,
      "step": 522
    },
    {
      "epoch": 1.1602884082085414,
      "grad_norm": 0.031186986714601517,
      "learning_rate": 0.00020444444444444443,
      "loss": 0.5487,
      "step": 523
    },
    {
      "epoch": 1.1625069328896285,
      "grad_norm": 0.039960771799087524,
      "learning_rate": 0.00020419753086419753,
      "loss": 0.5053,
      "step": 524
    },
    {
      "epoch": 1.1647254575707155,
      "grad_norm": 0.05363314971327782,
      "learning_rate": 0.0002039506172839506,
      "loss": 0.646,
      "step": 525
    },
    {
      "epoch": 1.1669439822518026,
      "grad_norm": 0.035462070256471634,
      "learning_rate": 0.0002037037037037037,
      "loss": 0.6553,
      "step": 526
    },
    {
      "epoch": 1.1691625069328897,
      "grad_norm": 0.03960970789194107,
      "learning_rate": 0.00020345679012345676,
      "loss": 0.4943,
      "step": 527
    },
    {
      "epoch": 1.1713810316139768,
      "grad_norm": 0.036298226565122604,
      "learning_rate": 0.00020320987654320985,
      "loss": 0.5337,
      "step": 528
    },
    {
      "epoch": 1.1735995562950639,
      "grad_norm": 0.049671195447444916,
      "learning_rate": 0.00020296296296296295,
      "loss": 0.69,
      "step": 529
    },
    {
      "epoch": 1.175818080976151,
      "grad_norm": 0.03985781967639923,
      "learning_rate": 0.00020271604938271602,
      "loss": 0.5198,
      "step": 530
    },
    {
      "epoch": 1.178036605657238,
      "grad_norm": 0.06500115990638733,
      "learning_rate": 0.00020246913580246914,
      "loss": 0.6391,
      "step": 531
    },
    {
      "epoch": 1.180255130338325,
      "grad_norm": 0.03536362946033478,
      "learning_rate": 0.0002022222222222222,
      "loss": 0.5512,
      "step": 532
    },
    {
      "epoch": 1.1824736550194122,
      "grad_norm": 0.07266086339950562,
      "learning_rate": 0.0002019753086419753,
      "loss": 0.591,
      "step": 533
    },
    {
      "epoch": 1.1846921797004992,
      "grad_norm": 0.033611539751291275,
      "learning_rate": 0.00020172839506172837,
      "loss": 0.444,
      "step": 534
    },
    {
      "epoch": 1.1869107043815863,
      "grad_norm": 0.041430432349443436,
      "learning_rate": 0.00020148148148148147,
      "loss": 0.5663,
      "step": 535
    },
    {
      "epoch": 1.1891292290626734,
      "grad_norm": 0.058304205536842346,
      "learning_rate": 0.00020123456790123456,
      "loss": 0.5342,
      "step": 536
    },
    {
      "epoch": 1.1913477537437605,
      "grad_norm": 0.04225342720746994,
      "learning_rate": 0.00020098765432098763,
      "loss": 0.5614,
      "step": 537
    },
    {
      "epoch": 1.1935662784248475,
      "grad_norm": 0.05314590036869049,
      "learning_rate": 0.00020074074074074072,
      "loss": 0.6382,
      "step": 538
    },
    {
      "epoch": 1.1957848031059346,
      "grad_norm": 0.03678717836737633,
      "learning_rate": 0.0002004938271604938,
      "loss": 0.5096,
      "step": 539
    },
    {
      "epoch": 1.1980033277870217,
      "grad_norm": 0.06726421415805817,
      "learning_rate": 0.00020024691358024691,
      "loss": 0.5216,
      "step": 540
    },
    {
      "epoch": 1.2002218524681088,
      "grad_norm": 0.029924508184194565,
      "learning_rate": 0.00019999999999999998,
      "loss": 0.5521,
      "step": 541
    },
    {
      "epoch": 1.2024403771491958,
      "grad_norm": 0.0466361902654171,
      "learning_rate": 0.00019975308641975308,
      "loss": 0.4977,
      "step": 542
    },
    {
      "epoch": 1.204658901830283,
      "grad_norm": 0.03543318435549736,
      "learning_rate": 0.00019950617283950615,
      "loss": 0.5964,
      "step": 543
    },
    {
      "epoch": 1.20687742651137,
      "grad_norm": 0.036266129463911057,
      "learning_rate": 0.00019925925925925924,
      "loss": 0.4635,
      "step": 544
    },
    {
      "epoch": 1.209095951192457,
      "grad_norm": 0.03439534083008766,
      "learning_rate": 0.00019901234567901234,
      "loss": 0.5015,
      "step": 545
    },
    {
      "epoch": 1.2113144758735441,
      "grad_norm": 0.04214196279644966,
      "learning_rate": 0.0001987654320987654,
      "loss": 0.4431,
      "step": 546
    },
    {
      "epoch": 1.2135330005546312,
      "grad_norm": 0.03417308256030083,
      "learning_rate": 0.0001985185185185185,
      "loss": 0.5755,
      "step": 547
    },
    {
      "epoch": 1.2157515252357183,
      "grad_norm": 0.04507344588637352,
      "learning_rate": 0.00019827160493827157,
      "loss": 0.5425,
      "step": 548
    },
    {
      "epoch": 1.2179700499168054,
      "grad_norm": 0.03655485808849335,
      "learning_rate": 0.0001980246913580247,
      "loss": 0.6912,
      "step": 549
    },
    {
      "epoch": 1.2201885745978924,
      "grad_norm": 0.054391272366046906,
      "learning_rate": 0.00019777777777777776,
      "loss": 0.707,
      "step": 550
    },
    {
      "epoch": 1.2224070992789795,
      "grad_norm": 0.041762325912714005,
      "learning_rate": 0.00019753086419753085,
      "loss": 0.6314,
      "step": 551
    },
    {
      "epoch": 1.2246256239600666,
      "grad_norm": 0.05875431001186371,
      "learning_rate": 0.00019728395061728395,
      "loss": 0.5954,
      "step": 552
    },
    {
      "epoch": 1.2268441486411537,
      "grad_norm": 0.038574665784835815,
      "learning_rate": 0.00019703703703703702,
      "loss": 0.5575,
      "step": 553
    },
    {
      "epoch": 1.2290626733222407,
      "grad_norm": 0.037381257861852646,
      "learning_rate": 0.0001967901234567901,
      "loss": 0.5782,
      "step": 554
    },
    {
      "epoch": 1.2312811980033278,
      "grad_norm": 0.03367564454674721,
      "learning_rate": 0.00019654320987654318,
      "loss": 0.6493,
      "step": 555
    },
    {
      "epoch": 1.2334997226844149,
      "grad_norm": 0.04225460812449455,
      "learning_rate": 0.00019629629629629627,
      "loss": 0.6393,
      "step": 556
    },
    {
      "epoch": 1.235718247365502,
      "grad_norm": 0.03949972242116928,
      "learning_rate": 0.00019604938271604934,
      "loss": 0.553,
      "step": 557
    },
    {
      "epoch": 1.237936772046589,
      "grad_norm": 0.041419077664613724,
      "learning_rate": 0.00019580246913580246,
      "loss": 0.5249,
      "step": 558
    },
    {
      "epoch": 1.240155296727676,
      "grad_norm": 0.0346052385866642,
      "learning_rate": 0.00019555555555555556,
      "loss": 0.5429,
      "step": 559
    },
    {
      "epoch": 1.2423738214087632,
      "grad_norm": 0.0350908488035202,
      "learning_rate": 0.00019530864197530863,
      "loss": 0.5526,
      "step": 560
    },
    {
      "epoch": 1.2445923460898503,
      "grad_norm": 0.04003891348838806,
      "learning_rate": 0.00019506172839506172,
      "loss": 0.5082,
      "step": 561
    },
    {
      "epoch": 1.2468108707709373,
      "grad_norm": 0.04278858006000519,
      "learning_rate": 0.0001948148148148148,
      "loss": 0.5861,
      "step": 562
    },
    {
      "epoch": 1.2490293954520244,
      "grad_norm": 0.04367564246058464,
      "learning_rate": 0.00019456790123456789,
      "loss": 0.696,
      "step": 563
    },
    {
      "epoch": 1.2512479201331115,
      "grad_norm": 0.04128764197230339,
      "learning_rate": 0.00019432098765432095,
      "loss": 0.6366,
      "step": 564
    },
    {
      "epoch": 1.2534664448141986,
      "grad_norm": 0.034572355449199677,
      "learning_rate": 0.00019407407407407405,
      "loss": 0.5275,
      "step": 565
    },
    {
      "epoch": 1.2556849694952856,
      "grad_norm": 0.041921231895685196,
      "learning_rate": 0.00019382716049382712,
      "loss": 0.7072,
      "step": 566
    },
    {
      "epoch": 1.2579034941763727,
      "grad_norm": 0.03814862668514252,
      "learning_rate": 0.00019358024691358024,
      "loss": 0.5898,
      "step": 567
    },
    {
      "epoch": 1.2601220188574598,
      "grad_norm": 0.028689106926321983,
      "learning_rate": 0.00019333333333333333,
      "loss": 0.5996,
      "step": 568
    },
    {
      "epoch": 1.2623405435385469,
      "grad_norm": 0.026630034670233727,
      "learning_rate": 0.0001930864197530864,
      "loss": 0.5065,
      "step": 569
    },
    {
      "epoch": 1.264559068219634,
      "grad_norm": 0.03232064098119736,
      "learning_rate": 0.0001928395061728395,
      "loss": 0.5044,
      "step": 570
    },
    {
      "epoch": 1.266777592900721,
      "grad_norm": 0.05055594816803932,
      "learning_rate": 0.00019259259259259257,
      "loss": 0.5354,
      "step": 571
    },
    {
      "epoch": 1.268996117581808,
      "grad_norm": 0.044294700026512146,
      "learning_rate": 0.00019234567901234566,
      "loss": 0.6332,
      "step": 572
    },
    {
      "epoch": 1.2712146422628952,
      "grad_norm": 0.03314494714140892,
      "learning_rate": 0.00019209876543209873,
      "loss": 0.6271,
      "step": 573
    },
    {
      "epoch": 1.2734331669439822,
      "grad_norm": 0.03905435651540756,
      "learning_rate": 0.00019185185185185182,
      "loss": 0.5521,
      "step": 574
    },
    {
      "epoch": 1.2756516916250693,
      "grad_norm": 0.03912075236439705,
      "learning_rate": 0.00019160493827160495,
      "loss": 0.7129,
      "step": 575
    },
    {
      "epoch": 1.2778702163061564,
      "grad_norm": 0.03727858141064644,
      "learning_rate": 0.00019135802469135801,
      "loss": 0.5062,
      "step": 576
    },
    {
      "epoch": 1.2800887409872435,
      "grad_norm": 0.039887934923172,
      "learning_rate": 0.0001911111111111111,
      "loss": 0.6117,
      "step": 577
    },
    {
      "epoch": 1.2823072656683305,
      "grad_norm": 0.03077658824622631,
      "learning_rate": 0.00019086419753086418,
      "loss": 0.5851,
      "step": 578
    },
    {
      "epoch": 1.2845257903494176,
      "grad_norm": 0.028013156726956367,
      "learning_rate": 0.00019061728395061727,
      "loss": 0.474,
      "step": 579
    },
    {
      "epoch": 1.2867443150305047,
      "grad_norm": 0.046493593603372574,
      "learning_rate": 0.00019037037037037034,
      "loss": 0.7635,
      "step": 580
    },
    {
      "epoch": 1.2889628397115918,
      "grad_norm": 0.0416119322180748,
      "learning_rate": 0.00019012345679012344,
      "loss": 0.7192,
      "step": 581
    },
    {
      "epoch": 1.2911813643926788,
      "grad_norm": 0.03868315368890762,
      "learning_rate": 0.00018987654320987653,
      "loss": 0.5112,
      "step": 582
    },
    {
      "epoch": 1.293399889073766,
      "grad_norm": 0.043333400040864944,
      "learning_rate": 0.0001896296296296296,
      "loss": 0.4866,
      "step": 583
    },
    {
      "epoch": 1.295618413754853,
      "grad_norm": 0.03921723738312721,
      "learning_rate": 0.00018938271604938272,
      "loss": 0.5842,
      "step": 584
    },
    {
      "epoch": 1.29783693843594,
      "grad_norm": 0.04876575991511345,
      "learning_rate": 0.0001891358024691358,
      "loss": 0.6397,
      "step": 585
    },
    {
      "epoch": 1.3000554631170271,
      "grad_norm": 0.03612883761525154,
      "learning_rate": 0.00018888888888888888,
      "loss": 0.495,
      "step": 586
    },
    {
      "epoch": 1.3022739877981142,
      "grad_norm": 0.04147052392363548,
      "learning_rate": 0.00018864197530864195,
      "loss": 0.4576,
      "step": 587
    },
    {
      "epoch": 1.3044925124792013,
      "grad_norm": 0.04152943566441536,
      "learning_rate": 0.00018839506172839505,
      "loss": 0.6216,
      "step": 588
    },
    {
      "epoch": 1.3067110371602884,
      "grad_norm": 0.03779630735516548,
      "learning_rate": 0.00018814814814814814,
      "loss": 0.6051,
      "step": 589
    },
    {
      "epoch": 1.3089295618413754,
      "grad_norm": 0.04005211964249611,
      "learning_rate": 0.0001879012345679012,
      "loss": 0.5851,
      "step": 590
    },
    {
      "epoch": 1.3111480865224625,
      "grad_norm": 0.03717253729701042,
      "learning_rate": 0.0001876543209876543,
      "loss": 0.3887,
      "step": 591
    },
    {
      "epoch": 1.3133666112035496,
      "grad_norm": 0.05021635815501213,
      "learning_rate": 0.00018740740740740737,
      "loss": 0.6944,
      "step": 592
    },
    {
      "epoch": 1.3155851358846367,
      "grad_norm": 0.03149298205971718,
      "learning_rate": 0.0001871604938271605,
      "loss": 0.5085,
      "step": 593
    },
    {
      "epoch": 1.3178036605657237,
      "grad_norm": 0.03708035871386528,
      "learning_rate": 0.00018691358024691356,
      "loss": 0.7213,
      "step": 594
    },
    {
      "epoch": 1.3200221852468108,
      "grad_norm": 0.035653404891490936,
      "learning_rate": 0.00018666666666666666,
      "loss": 0.7949,
      "step": 595
    },
    {
      "epoch": 1.3222407099278979,
      "grad_norm": 0.03299754858016968,
      "learning_rate": 0.00018641975308641973,
      "loss": 0.4761,
      "step": 596
    },
    {
      "epoch": 1.324459234608985,
      "grad_norm": 0.038019005209207535,
      "learning_rate": 0.00018617283950617282,
      "loss": 0.464,
      "step": 597
    },
    {
      "epoch": 1.326677759290072,
      "grad_norm": 0.04103780910372734,
      "learning_rate": 0.00018592592592592592,
      "loss": 0.5642,
      "step": 598
    },
    {
      "epoch": 1.328896283971159,
      "grad_norm": 0.03484074026346207,
      "learning_rate": 0.00018567901234567899,
      "loss": 0.402,
      "step": 599
    },
    {
      "epoch": 1.3311148086522462,
      "grad_norm": 0.03696504235267639,
      "learning_rate": 0.00018543209876543208,
      "loss": 0.6102,
      "step": 600
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 0.038149431347846985,
      "learning_rate": 0.00018518518518518515,
      "loss": 0.6516,
      "step": 601
    },
    {
      "epoch": 1.3355518580144203,
      "grad_norm": 0.039646465331315994,
      "learning_rate": 0.00018493827160493827,
      "loss": 0.6509,
      "step": 602
    },
    {
      "epoch": 1.3377703826955074,
      "grad_norm": 0.03054131381213665,
      "learning_rate": 0.00018469135802469134,
      "loss": 0.5153,
      "step": 603
    },
    {
      "epoch": 1.3399889073765945,
      "grad_norm": 0.09024311602115631,
      "learning_rate": 0.00018444444444444443,
      "loss": 0.5716,
      "step": 604
    },
    {
      "epoch": 1.3422074320576816,
      "grad_norm": 0.042115092277526855,
      "learning_rate": 0.00018419753086419753,
      "loss": 0.6098,
      "step": 605
    },
    {
      "epoch": 1.3444259567387689,
      "grad_norm": 0.03492320328950882,
      "learning_rate": 0.0001839506172839506,
      "loss": 0.6803,
      "step": 606
    },
    {
      "epoch": 1.346644481419856,
      "grad_norm": 0.037822604179382324,
      "learning_rate": 0.0001837037037037037,
      "loss": 0.6823,
      "step": 607
    },
    {
      "epoch": 1.348863006100943,
      "grad_norm": 0.03751061484217644,
      "learning_rate": 0.00018345679012345676,
      "loss": 0.6306,
      "step": 608
    },
    {
      "epoch": 1.35108153078203,
      "grad_norm": 0.037911493331193924,
      "learning_rate": 0.00018320987654320986,
      "loss": 0.706,
      "step": 609
    },
    {
      "epoch": 1.3533000554631172,
      "grad_norm": 0.03935304656624794,
      "learning_rate": 0.00018296296296296292,
      "loss": 0.55,
      "step": 610
    },
    {
      "epoch": 1.3555185801442042,
      "grad_norm": 0.03231262043118477,
      "learning_rate": 0.00018271604938271605,
      "loss": 0.4242,
      "step": 611
    },
    {
      "epoch": 1.3577371048252913,
      "grad_norm": 0.05424240604043007,
      "learning_rate": 0.00018246913580246914,
      "loss": 0.591,
      "step": 612
    },
    {
      "epoch": 1.3599556295063784,
      "grad_norm": 0.03460954874753952,
      "learning_rate": 0.0001822222222222222,
      "loss": 0.6022,
      "step": 613
    },
    {
      "epoch": 1.3621741541874655,
      "grad_norm": 0.039916615933179855,
      "learning_rate": 0.0001819753086419753,
      "loss": 0.5771,
      "step": 614
    },
    {
      "epoch": 1.3643926788685525,
      "grad_norm": 0.03581291809678078,
      "learning_rate": 0.00018172839506172837,
      "loss": 0.5222,
      "step": 615
    },
    {
      "epoch": 1.3666112035496396,
      "grad_norm": 0.0306674987077713,
      "learning_rate": 0.00018148148148148147,
      "loss": 0.4792,
      "step": 616
    },
    {
      "epoch": 1.3688297282307267,
      "grad_norm": 0.046209827065467834,
      "learning_rate": 0.00018123456790123454,
      "loss": 0.7055,
      "step": 617
    },
    {
      "epoch": 1.3710482529118138,
      "grad_norm": 0.03650926798582077,
      "learning_rate": 0.00018098765432098763,
      "loss": 0.5798,
      "step": 618
    },
    {
      "epoch": 1.3732667775929008,
      "grad_norm": 0.04206550866365433,
      "learning_rate": 0.00018074074074074075,
      "loss": 0.5845,
      "step": 619
    },
    {
      "epoch": 1.375485302273988,
      "grad_norm": 0.0507446825504303,
      "learning_rate": 0.00018049382716049382,
      "loss": 0.4947,
      "step": 620
    },
    {
      "epoch": 1.377703826955075,
      "grad_norm": 0.042161911725997925,
      "learning_rate": 0.00018024691358024692,
      "loss": 0.684,
      "step": 621
    },
    {
      "epoch": 1.379922351636162,
      "grad_norm": 0.031945694237947464,
      "learning_rate": 0.00017999999999999998,
      "loss": 0.4159,
      "step": 622
    },
    {
      "epoch": 1.3821408763172491,
      "grad_norm": 0.06357549875974655,
      "learning_rate": 0.00017975308641975308,
      "loss": 0.6854,
      "step": 623
    },
    {
      "epoch": 1.3843594009983362,
      "grad_norm": 0.03368821740150452,
      "learning_rate": 0.00017950617283950615,
      "loss": 0.5792,
      "step": 624
    },
    {
      "epoch": 1.3865779256794233,
      "grad_norm": 0.04134785756468773,
      "learning_rate": 0.00017925925925925924,
      "loss": 0.5494,
      "step": 625
    },
    {
      "epoch": 1.3887964503605104,
      "grad_norm": 0.03704426810145378,
      "learning_rate": 0.0001790123456790123,
      "loss": 0.426,
      "step": 626
    },
    {
      "epoch": 1.3910149750415974,
      "grad_norm": 0.040071915835142136,
      "learning_rate": 0.0001787654320987654,
      "loss": 0.566,
      "step": 627
    },
    {
      "epoch": 1.3932334997226845,
      "grad_norm": 0.033120330423116684,
      "learning_rate": 0.00017851851851851853,
      "loss": 0.4868,
      "step": 628
    },
    {
      "epoch": 1.3954520244037716,
      "grad_norm": 0.04258786514401436,
      "learning_rate": 0.0001782716049382716,
      "loss": 0.5492,
      "step": 629
    },
    {
      "epoch": 1.3976705490848587,
      "grad_norm": 0.03773237392306328,
      "learning_rate": 0.0001780246913580247,
      "loss": 0.5344,
      "step": 630
    },
    {
      "epoch": 1.3998890737659457,
      "grad_norm": 0.04928630217909813,
      "learning_rate": 0.00017777777777777776,
      "loss": 0.5443,
      "step": 631
    },
    {
      "epoch": 1.4021075984470328,
      "grad_norm": 0.03370613232254982,
      "learning_rate": 0.00017753086419753085,
      "loss": 0.5036,
      "step": 632
    },
    {
      "epoch": 1.4043261231281199,
      "grad_norm": 0.03447793796658516,
      "learning_rate": 0.00017728395061728392,
      "loss": 0.5709,
      "step": 633
    },
    {
      "epoch": 1.406544647809207,
      "grad_norm": 0.04332738742232323,
      "learning_rate": 0.00017703703703703702,
      "loss": 0.5897,
      "step": 634
    },
    {
      "epoch": 1.408763172490294,
      "grad_norm": 0.04380321502685547,
      "learning_rate": 0.0001767901234567901,
      "loss": 0.5111,
      "step": 635
    },
    {
      "epoch": 1.410981697171381,
      "grad_norm": 0.03381207585334778,
      "learning_rate": 0.00017654320987654318,
      "loss": 0.5243,
      "step": 636
    },
    {
      "epoch": 1.4132002218524682,
      "grad_norm": 0.03360865265130997,
      "learning_rate": 0.0001762962962962963,
      "loss": 0.5388,
      "step": 637
    },
    {
      "epoch": 1.4154187465335553,
      "grad_norm": 0.043903663754463196,
      "learning_rate": 0.00017604938271604937,
      "loss": 0.6242,
      "step": 638
    },
    {
      "epoch": 1.4176372712146423,
      "grad_norm": 0.050343822687864304,
      "learning_rate": 0.00017580246913580247,
      "loss": 0.6073,
      "step": 639
    },
    {
      "epoch": 1.4198557958957294,
      "grad_norm": 0.039043642580509186,
      "learning_rate": 0.00017555555555555553,
      "loss": 0.5383,
      "step": 640
    },
    {
      "epoch": 1.4220743205768165,
      "grad_norm": 0.04155135154724121,
      "learning_rate": 0.00017530864197530863,
      "loss": 0.5437,
      "step": 641
    },
    {
      "epoch": 1.4242928452579036,
      "grad_norm": 0.029898790642619133,
      "learning_rate": 0.00017506172839506172,
      "loss": 0.5032,
      "step": 642
    },
    {
      "epoch": 1.4265113699389906,
      "grad_norm": 0.04658302292227745,
      "learning_rate": 0.0001748148148148148,
      "loss": 0.4633,
      "step": 643
    },
    {
      "epoch": 1.4287298946200777,
      "grad_norm": 0.05075034126639366,
      "learning_rate": 0.0001745679012345679,
      "loss": 0.6942,
      "step": 644
    },
    {
      "epoch": 1.4309484193011648,
      "grad_norm": 0.030335120856761932,
      "learning_rate": 0.00017432098765432096,
      "loss": 0.497,
      "step": 645
    },
    {
      "epoch": 1.4331669439822519,
      "grad_norm": 0.038972605019807816,
      "learning_rate": 0.00017407407407407408,
      "loss": 0.5233,
      "step": 646
    },
    {
      "epoch": 1.435385468663339,
      "grad_norm": 0.0385974645614624,
      "learning_rate": 0.00017382716049382715,
      "loss": 0.5021,
      "step": 647
    },
    {
      "epoch": 1.437603993344426,
      "grad_norm": 0.03921416401863098,
      "learning_rate": 0.00017358024691358024,
      "loss": 0.4811,
      "step": 648
    },
    {
      "epoch": 1.439822518025513,
      "grad_norm": 0.028492935001850128,
      "learning_rate": 0.0001733333333333333,
      "loss": 0.5077,
      "step": 649
    },
    {
      "epoch": 1.4420410427066002,
      "grad_norm": 0.05210812762379646,
      "learning_rate": 0.0001730864197530864,
      "loss": 0.4959,
      "step": 650
    },
    {
      "epoch": 1.4442595673876872,
      "grad_norm": 0.04232173040509224,
      "learning_rate": 0.0001728395061728395,
      "loss": 0.5735,
      "step": 651
    },
    {
      "epoch": 1.4464780920687743,
      "grad_norm": 0.06512024253606796,
      "learning_rate": 0.00017259259259259257,
      "loss": 0.5996,
      "step": 652
    },
    {
      "epoch": 1.4486966167498614,
      "grad_norm": 0.053563982248306274,
      "learning_rate": 0.00017234567901234566,
      "loss": 0.4739,
      "step": 653
    },
    {
      "epoch": 1.4509151414309485,
      "grad_norm": 0.04886218532919884,
      "learning_rate": 0.00017209876543209873,
      "loss": 0.5051,
      "step": 654
    },
    {
      "epoch": 1.4531336661120355,
      "grad_norm": 0.0432475209236145,
      "learning_rate": 0.00017185185185185185,
      "loss": 0.5803,
      "step": 655
    },
    {
      "epoch": 1.4553521907931226,
      "grad_norm": 0.04271094128489494,
      "learning_rate": 0.00017160493827160492,
      "loss": 0.5722,
      "step": 656
    },
    {
      "epoch": 1.4575707154742097,
      "grad_norm": 0.036052532494068146,
      "learning_rate": 0.00017135802469135802,
      "loss": 0.5,
      "step": 657
    },
    {
      "epoch": 1.4597892401552968,
      "grad_norm": 0.0418868213891983,
      "learning_rate": 0.0001711111111111111,
      "loss": 0.5842,
      "step": 658
    },
    {
      "epoch": 1.4620077648363838,
      "grad_norm": 0.04189569875597954,
      "learning_rate": 0.00017086419753086418,
      "loss": 0.7651,
      "step": 659
    },
    {
      "epoch": 1.464226289517471,
      "grad_norm": 0.035675376653671265,
      "learning_rate": 0.00017061728395061727,
      "loss": 0.491,
      "step": 660
    },
    {
      "epoch": 1.466444814198558,
      "grad_norm": 0.041652366518974304,
      "learning_rate": 0.00017037037037037034,
      "loss": 0.7407,
      "step": 661
    },
    {
      "epoch": 1.468663338879645,
      "grad_norm": 0.035895105451345444,
      "learning_rate": 0.00017012345679012344,
      "loss": 0.6354,
      "step": 662
    },
    {
      "epoch": 1.4708818635607321,
      "grad_norm": 0.029283588752150536,
      "learning_rate": 0.0001698765432098765,
      "loss": 0.5629,
      "step": 663
    },
    {
      "epoch": 1.4731003882418192,
      "grad_norm": 0.048661746084690094,
      "learning_rate": 0.00016962962962962963,
      "loss": 0.5636,
      "step": 664
    },
    {
      "epoch": 1.4753189129229063,
      "grad_norm": 0.03794539347290993,
      "learning_rate": 0.00016938271604938272,
      "loss": 0.6974,
      "step": 665
    },
    {
      "epoch": 1.4775374376039934,
      "grad_norm": 0.04186217486858368,
      "learning_rate": 0.0001691358024691358,
      "loss": 0.7059,
      "step": 666
    },
    {
      "epoch": 1.4797559622850804,
      "grad_norm": 0.03493627533316612,
      "learning_rate": 0.00016888888888888889,
      "loss": 0.5659,
      "step": 667
    },
    {
      "epoch": 1.4819744869661675,
      "grad_norm": 0.056807760149240494,
      "learning_rate": 0.00016864197530864195,
      "loss": 0.4351,
      "step": 668
    },
    {
      "epoch": 1.4841930116472546,
      "grad_norm": 0.03597313165664673,
      "learning_rate": 0.00016839506172839505,
      "loss": 0.6373,
      "step": 669
    },
    {
      "epoch": 1.4864115363283417,
      "grad_norm": 0.04453812912106514,
      "learning_rate": 0.00016814814814814812,
      "loss": 0.5673,
      "step": 670
    },
    {
      "epoch": 1.4886300610094287,
      "grad_norm": 0.033846888691186905,
      "learning_rate": 0.0001679012345679012,
      "loss": 0.5438,
      "step": 671
    },
    {
      "epoch": 1.4908485856905158,
      "grad_norm": 0.0398983359336853,
      "learning_rate": 0.00016765432098765433,
      "loss": 0.5755,
      "step": 672
    },
    {
      "epoch": 1.493067110371603,
      "grad_norm": 0.03347493335604668,
      "learning_rate": 0.0001674074074074074,
      "loss": 0.5137,
      "step": 673
    },
    {
      "epoch": 1.49528563505269,
      "grad_norm": 0.04694952443242073,
      "learning_rate": 0.0001671604938271605,
      "loss": 0.5395,
      "step": 674
    },
    {
      "epoch": 1.497504159733777,
      "grad_norm": 0.032678864896297455,
      "learning_rate": 0.00016691358024691357,
      "loss": 0.4676,
      "step": 675
    },
    {
      "epoch": 1.4997226844148641,
      "grad_norm": 0.03940966725349426,
      "learning_rate": 0.00016666666666666666,
      "loss": 0.5123,
      "step": 676
    },
    {
      "epoch": 1.5019412090959512,
      "grad_norm": 0.04400421679019928,
      "learning_rate": 0.00016641975308641973,
      "loss": 0.6755,
      "step": 677
    },
    {
      "epoch": 1.5041597337770383,
      "grad_norm": 0.0585503950715065,
      "learning_rate": 0.00016617283950617282,
      "loss": 0.4613,
      "step": 678
    },
    {
      "epoch": 1.5063782584581253,
      "grad_norm": 0.03227279335260391,
      "learning_rate": 0.0001659259259259259,
      "loss": 0.5427,
      "step": 679
    },
    {
      "epoch": 1.5085967831392124,
      "grad_norm": 0.0362737663090229,
      "learning_rate": 0.000165679012345679,
      "loss": 0.4902,
      "step": 680
    },
    {
      "epoch": 1.5108153078202995,
      "grad_norm": 0.05084448307752609,
      "learning_rate": 0.0001654320987654321,
      "loss": 0.6439,
      "step": 681
    },
    {
      "epoch": 1.5130338325013866,
      "grad_norm": 0.07740761339664459,
      "learning_rate": 0.00016518518518518518,
      "loss": 0.8097,
      "step": 682
    },
    {
      "epoch": 1.5152523571824736,
      "grad_norm": 0.03728952258825302,
      "learning_rate": 0.00016493827160493827,
      "loss": 0.5665,
      "step": 683
    },
    {
      "epoch": 1.5174708818635607,
      "grad_norm": 0.048041023313999176,
      "learning_rate": 0.00016469135802469134,
      "loss": 0.566,
      "step": 684
    },
    {
      "epoch": 1.5196894065446478,
      "grad_norm": 0.04466738551855087,
      "learning_rate": 0.00016444444444444444,
      "loss": 0.6274,
      "step": 685
    },
    {
      "epoch": 1.5219079312257349,
      "grad_norm": 0.04854423552751541,
      "learning_rate": 0.0001641975308641975,
      "loss": 0.493,
      "step": 686
    },
    {
      "epoch": 1.524126455906822,
      "grad_norm": 0.09303252398967743,
      "learning_rate": 0.0001639506172839506,
      "loss": 0.6256,
      "step": 687
    },
    {
      "epoch": 1.526344980587909,
      "grad_norm": 0.03179485723376274,
      "learning_rate": 0.0001637037037037037,
      "loss": 0.6953,
      "step": 688
    },
    {
      "epoch": 1.528563505268996,
      "grad_norm": 0.044354408979415894,
      "learning_rate": 0.00016345679012345676,
      "loss": 0.4731,
      "step": 689
    },
    {
      "epoch": 1.5307820299500832,
      "grad_norm": 0.033977191895246506,
      "learning_rate": 0.00016320987654320988,
      "loss": 0.5118,
      "step": 690
    },
    {
      "epoch": 1.5330005546311702,
      "grad_norm": 0.03163310885429382,
      "learning_rate": 0.00016296296296296295,
      "loss": 0.5309,
      "step": 691
    },
    {
      "epoch": 1.5352190793122573,
      "grad_norm": 0.04574287310242653,
      "learning_rate": 0.00016271604938271605,
      "loss": 0.5859,
      "step": 692
    },
    {
      "epoch": 1.5374376039933444,
      "grad_norm": 0.047205254435539246,
      "learning_rate": 0.00016246913580246912,
      "loss": 0.6051,
      "step": 693
    },
    {
      "epoch": 1.5396561286744315,
      "grad_norm": 0.045713067054748535,
      "learning_rate": 0.0001622222222222222,
      "loss": 0.6695,
      "step": 694
    },
    {
      "epoch": 1.5418746533555185,
      "grad_norm": 0.04430000111460686,
      "learning_rate": 0.0001619753086419753,
      "loss": 0.5849,
      "step": 695
    },
    {
      "epoch": 1.5440931780366056,
      "grad_norm": 0.03729617968201637,
      "learning_rate": 0.00016172839506172837,
      "loss": 0.6315,
      "step": 696
    },
    {
      "epoch": 1.5463117027176927,
      "grad_norm": 0.041310861706733704,
      "learning_rate": 0.00016148148148148147,
      "loss": 0.5469,
      "step": 697
    },
    {
      "epoch": 1.5485302273987798,
      "grad_norm": 0.04393087700009346,
      "learning_rate": 0.00016123456790123454,
      "loss": 0.4856,
      "step": 698
    },
    {
      "epoch": 1.5507487520798668,
      "grad_norm": 0.039317574352025986,
      "learning_rate": 0.00016098765432098766,
      "loss": 0.6654,
      "step": 699
    },
    {
      "epoch": 1.552967276760954,
      "grad_norm": 0.04437847062945366,
      "learning_rate": 0.00016074074074074073,
      "loss": 0.5399,
      "step": 700
    },
    {
      "epoch": 1.555185801442041,
      "grad_norm": 0.0373670794069767,
      "learning_rate": 0.00016049382716049382,
      "loss": 0.6119,
      "step": 701
    },
    {
      "epoch": 1.557404326123128,
      "grad_norm": 0.05090388283133507,
      "learning_rate": 0.00016024691358024692,
      "loss": 0.665,
      "step": 702
    },
    {
      "epoch": 1.5596228508042151,
      "grad_norm": 0.05446712672710419,
      "learning_rate": 0.00015999999999999999,
      "loss": 0.7716,
      "step": 703
    },
    {
      "epoch": 1.5618413754853022,
      "grad_norm": 0.03216097876429558,
      "learning_rate": 0.00015975308641975308,
      "loss": 0.6072,
      "step": 704
    },
    {
      "epoch": 1.5640599001663893,
      "grad_norm": 0.05534995347261429,
      "learning_rate": 0.00015950617283950615,
      "loss": 0.6326,
      "step": 705
    },
    {
      "epoch": 1.5662784248474764,
      "grad_norm": 0.04208800569176674,
      "learning_rate": 0.00015925925925925924,
      "loss": 0.5404,
      "step": 706
    },
    {
      "epoch": 1.5684969495285634,
      "grad_norm": 0.03433000296354294,
      "learning_rate": 0.0001590123456790123,
      "loss": 0.5142,
      "step": 707
    },
    {
      "epoch": 1.5707154742096505,
      "grad_norm": 0.032243676483631134,
      "learning_rate": 0.00015876543209876543,
      "loss": 0.4539,
      "step": 708
    },
    {
      "epoch": 1.5729339988907376,
      "grad_norm": 0.030815979465842247,
      "learning_rate": 0.0001585185185185185,
      "loss": 0.559,
      "step": 709
    },
    {
      "epoch": 1.5751525235718247,
      "grad_norm": 0.02867242507636547,
      "learning_rate": 0.0001582716049382716,
      "loss": 0.4648,
      "step": 710
    },
    {
      "epoch": 1.5773710482529117,
      "grad_norm": 0.05748599395155907,
      "learning_rate": 0.0001580246913580247,
      "loss": 0.687,
      "step": 711
    },
    {
      "epoch": 1.5795895729339988,
      "grad_norm": 0.03895612061023712,
      "learning_rate": 0.00015777777777777776,
      "loss": 0.5169,
      "step": 712
    },
    {
      "epoch": 1.581808097615086,
      "grad_norm": 0.03753532096743584,
      "learning_rate": 0.00015753086419753086,
      "loss": 0.4211,
      "step": 713
    },
    {
      "epoch": 1.584026622296173,
      "grad_norm": 0.04172753170132637,
      "learning_rate": 0.00015728395061728392,
      "loss": 0.6811,
      "step": 714
    },
    {
      "epoch": 1.58624514697726,
      "grad_norm": 0.035896431654691696,
      "learning_rate": 0.00015703703703703702,
      "loss": 0.4545,
      "step": 715
    },
    {
      "epoch": 1.5884636716583471,
      "grad_norm": 0.03902221471071243,
      "learning_rate": 0.0001567901234567901,
      "loss": 0.6657,
      "step": 716
    },
    {
      "epoch": 1.5906821963394342,
      "grad_norm": 0.03759798780083656,
      "learning_rate": 0.0001565432098765432,
      "loss": 0.4691,
      "step": 717
    },
    {
      "epoch": 1.5929007210205213,
      "grad_norm": 0.06004739925265312,
      "learning_rate": 0.0001562962962962963,
      "loss": 0.5924,
      "step": 718
    },
    {
      "epoch": 1.5951192457016083,
      "grad_norm": 0.040592193603515625,
      "learning_rate": 0.00015604938271604937,
      "loss": 0.5513,
      "step": 719
    },
    {
      "epoch": 1.5973377703826954,
      "grad_norm": 0.04156576469540596,
      "learning_rate": 0.00015580246913580247,
      "loss": 0.6789,
      "step": 720
    },
    {
      "epoch": 1.5995562950637825,
      "grad_norm": 0.030493153259158134,
      "learning_rate": 0.00015555555555555554,
      "loss": 0.6086,
      "step": 721
    },
    {
      "epoch": 1.6017748197448696,
      "grad_norm": 0.03438849747180939,
      "learning_rate": 0.00015530864197530863,
      "loss": 0.5663,
      "step": 722
    },
    {
      "epoch": 1.6039933444259566,
      "grad_norm": 0.05278339982032776,
      "learning_rate": 0.0001550617283950617,
      "loss": 0.441,
      "step": 723
    },
    {
      "epoch": 1.6062118691070437,
      "grad_norm": 0.029958384111523628,
      "learning_rate": 0.0001548148148148148,
      "loss": 0.7471,
      "step": 724
    },
    {
      "epoch": 1.6084303937881308,
      "grad_norm": 0.05187821760773659,
      "learning_rate": 0.00015456790123456792,
      "loss": 0.4969,
      "step": 725
    },
    {
      "epoch": 1.6106489184692179,
      "grad_norm": 0.05061693489551544,
      "learning_rate": 0.00015432098765432098,
      "loss": 0.6028,
      "step": 726
    },
    {
      "epoch": 1.612867443150305,
      "grad_norm": 0.04307565093040466,
      "learning_rate": 0.00015407407407407408,
      "loss": 0.5968,
      "step": 727
    },
    {
      "epoch": 1.615085967831392,
      "grad_norm": 0.05185339227318764,
      "learning_rate": 0.00015382716049382715,
      "loss": 0.6989,
      "step": 728
    },
    {
      "epoch": 1.617304492512479,
      "grad_norm": 0.045695144683122635,
      "learning_rate": 0.00015358024691358024,
      "loss": 0.5909,
      "step": 729
    },
    {
      "epoch": 1.6195230171935662,
      "grad_norm": 0.04648905619978905,
      "learning_rate": 0.0001533333333333333,
      "loss": 0.6153,
      "step": 730
    },
    {
      "epoch": 1.6217415418746532,
      "grad_norm": 0.03722282871603966,
      "learning_rate": 0.0001530864197530864,
      "loss": 0.524,
      "step": 731
    },
    {
      "epoch": 1.6239600665557403,
      "grad_norm": 0.0514126792550087,
      "learning_rate": 0.00015283950617283947,
      "loss": 0.6728,
      "step": 732
    },
    {
      "epoch": 1.6261785912368274,
      "grad_norm": 0.029080573469400406,
      "learning_rate": 0.00015259259259259257,
      "loss": 0.4363,
      "step": 733
    },
    {
      "epoch": 1.6283971159179145,
      "grad_norm": 0.036240361630916595,
      "learning_rate": 0.0001523456790123457,
      "loss": 0.537,
      "step": 734
    },
    {
      "epoch": 1.6306156405990015,
      "grad_norm": 0.0410381555557251,
      "learning_rate": 0.00015209876543209873,
      "loss": 0.5463,
      "step": 735
    },
    {
      "epoch": 1.6328341652800886,
      "grad_norm": 0.037101421505212784,
      "learning_rate": 0.00015185185185185185,
      "loss": 0.5325,
      "step": 736
    },
    {
      "epoch": 1.6350526899611757,
      "grad_norm": 0.03693998605012894,
      "learning_rate": 0.00015160493827160492,
      "loss": 0.5493,
      "step": 737
    },
    {
      "epoch": 1.6372712146422628,
      "grad_norm": 0.0391085259616375,
      "learning_rate": 0.00015135802469135802,
      "loss": 0.5404,
      "step": 738
    },
    {
      "epoch": 1.6394897393233498,
      "grad_norm": 0.034098658710718155,
      "learning_rate": 0.00015111111111111109,
      "loss": 0.5559,
      "step": 739
    },
    {
      "epoch": 1.641708264004437,
      "grad_norm": 0.04981136694550514,
      "learning_rate": 0.00015086419753086418,
      "loss": 0.6026,
      "step": 740
    },
    {
      "epoch": 1.643926788685524,
      "grad_norm": 0.05441569909453392,
      "learning_rate": 0.00015061728395061728,
      "loss": 0.6457,
      "step": 741
    },
    {
      "epoch": 1.646145313366611,
      "grad_norm": 0.055073998868465424,
      "learning_rate": 0.00015037037037037034,
      "loss": 0.4159,
      "step": 742
    },
    {
      "epoch": 1.6483638380476981,
      "grad_norm": 0.047073401510715485,
      "learning_rate": 0.00015012345679012347,
      "loss": 0.6426,
      "step": 743
    },
    {
      "epoch": 1.6505823627287852,
      "grad_norm": 0.039165280759334564,
      "learning_rate": 0.00014987654320987653,
      "loss": 0.5794,
      "step": 744
    },
    {
      "epoch": 1.6528008874098723,
      "grad_norm": 0.037176188081502914,
      "learning_rate": 0.00014962962962962963,
      "loss": 0.5738,
      "step": 745
    },
    {
      "epoch": 1.6550194120909594,
      "grad_norm": 0.03942460939288139,
      "learning_rate": 0.0001493827160493827,
      "loss": 0.613,
      "step": 746
    },
    {
      "epoch": 1.6572379367720464,
      "grad_norm": 0.03945528343319893,
      "learning_rate": 0.0001491358024691358,
      "loss": 0.6426,
      "step": 747
    },
    {
      "epoch": 1.6594564614531335,
      "grad_norm": 0.031806714832782745,
      "learning_rate": 0.00014888888888888886,
      "loss": 0.5084,
      "step": 748
    },
    {
      "epoch": 1.6616749861342206,
      "grad_norm": 0.03320614993572235,
      "learning_rate": 0.00014864197530864196,
      "loss": 0.4921,
      "step": 749
    },
    {
      "epoch": 1.6638935108153077,
      "grad_norm": 0.034095048904418945,
      "learning_rate": 0.00014839506172839505,
      "loss": 0.3673,
      "step": 750
    },
    {
      "epoch": 1.6661120354963947,
      "grad_norm": 0.02911224775016308,
      "learning_rate": 0.00014814814814814812,
      "loss": 0.4515,
      "step": 751
    },
    {
      "epoch": 1.6683305601774818,
      "grad_norm": 0.04327123239636421,
      "learning_rate": 0.00014790123456790124,
      "loss": 0.4842,
      "step": 752
    },
    {
      "epoch": 1.670549084858569,
      "grad_norm": 0.04450168088078499,
      "learning_rate": 0.0001476543209876543,
      "loss": 0.5607,
      "step": 753
    },
    {
      "epoch": 1.672767609539656,
      "grad_norm": 0.04193771257996559,
      "learning_rate": 0.0001474074074074074,
      "loss": 0.5795,
      "step": 754
    },
    {
      "epoch": 1.674986134220743,
      "grad_norm": 0.028406308963894844,
      "learning_rate": 0.00014716049382716047,
      "loss": 0.5976,
      "step": 755
    },
    {
      "epoch": 1.6772046589018301,
      "grad_norm": 0.03857554867863655,
      "learning_rate": 0.00014691358024691357,
      "loss": 0.6989,
      "step": 756
    },
    {
      "epoch": 1.6794231835829172,
      "grad_norm": 0.03644504398107529,
      "learning_rate": 0.00014666666666666664,
      "loss": 0.5958,
      "step": 757
    },
    {
      "epoch": 1.6816417082640043,
      "grad_norm": 0.032135676592588425,
      "learning_rate": 0.00014641975308641973,
      "loss": 0.5634,
      "step": 758
    },
    {
      "epoch": 1.6838602329450914,
      "grad_norm": 0.04024998098611832,
      "learning_rate": 0.00014617283950617283,
      "loss": 0.5769,
      "step": 759
    },
    {
      "epoch": 1.6860787576261784,
      "grad_norm": 0.042579203844070435,
      "learning_rate": 0.00014592592592592592,
      "loss": 0.6049,
      "step": 760
    },
    {
      "epoch": 1.6882972823072657,
      "grad_norm": 0.03473803773522377,
      "learning_rate": 0.000145679012345679,
      "loss": 0.577,
      "step": 761
    },
    {
      "epoch": 1.6905158069883528,
      "grad_norm": 0.039196133613586426,
      "learning_rate": 0.00014543209876543208,
      "loss": 0.5652,
      "step": 762
    },
    {
      "epoch": 1.6927343316694399,
      "grad_norm": 0.05407264828681946,
      "learning_rate": 0.00014518518518518518,
      "loss": 0.9248,
      "step": 763
    },
    {
      "epoch": 1.694952856350527,
      "grad_norm": 0.06167978420853615,
      "learning_rate": 0.00014493827160493825,
      "loss": 0.5647,
      "step": 764
    },
    {
      "epoch": 1.697171381031614,
      "grad_norm": 0.1038186326622963,
      "learning_rate": 0.00014469135802469134,
      "loss": 0.7033,
      "step": 765
    },
    {
      "epoch": 1.699389905712701,
      "grad_norm": 0.046323493123054504,
      "learning_rate": 0.0001444444444444444,
      "loss": 0.5275,
      "step": 766
    },
    {
      "epoch": 1.7016084303937882,
      "grad_norm": 0.04297422245144844,
      "learning_rate": 0.00014419753086419753,
      "loss": 0.7985,
      "step": 767
    },
    {
      "epoch": 1.7038269550748752,
      "grad_norm": 0.03176214545965195,
      "learning_rate": 0.0001439506172839506,
      "loss": 0.4424,
      "step": 768
    },
    {
      "epoch": 1.7060454797559623,
      "grad_norm": 0.059254687279462814,
      "learning_rate": 0.0001437037037037037,
      "loss": 0.5678,
      "step": 769
    },
    {
      "epoch": 1.7082640044370494,
      "grad_norm": 0.07520011812448502,
      "learning_rate": 0.00014345679012345676,
      "loss": 0.677,
      "step": 770
    },
    {
      "epoch": 1.7104825291181365,
      "grad_norm": 0.032213859260082245,
      "learning_rate": 0.00014320987654320986,
      "loss": 0.519,
      "step": 771
    },
    {
      "epoch": 1.7127010537992235,
      "grad_norm": 0.04368875175714493,
      "learning_rate": 0.00014296296296296295,
      "loss": 0.5381,
      "step": 772
    },
    {
      "epoch": 1.7149195784803106,
      "grad_norm": 0.046515095978975296,
      "learning_rate": 0.00014271604938271602,
      "loss": 0.6386,
      "step": 773
    },
    {
      "epoch": 1.7171381031613977,
      "grad_norm": 0.04264412820339203,
      "learning_rate": 0.00014246913580246912,
      "loss": 0.6471,
      "step": 774
    },
    {
      "epoch": 1.7193566278424848,
      "grad_norm": 0.047533467411994934,
      "learning_rate": 0.0001422222222222222,
      "loss": 0.4663,
      "step": 775
    },
    {
      "epoch": 1.7215751525235718,
      "grad_norm": 0.038209568709135056,
      "learning_rate": 0.0001419753086419753,
      "loss": 0.6115,
      "step": 776
    },
    {
      "epoch": 1.723793677204659,
      "grad_norm": 0.040848638862371445,
      "learning_rate": 0.00014172839506172838,
      "loss": 0.6812,
      "step": 777
    },
    {
      "epoch": 1.726012201885746,
      "grad_norm": 0.04021172225475311,
      "learning_rate": 0.00014148148148148147,
      "loss": 0.64,
      "step": 778
    },
    {
      "epoch": 1.728230726566833,
      "grad_norm": 0.03432880714535713,
      "learning_rate": 0.00014123456790123454,
      "loss": 0.558,
      "step": 779
    },
    {
      "epoch": 1.7304492512479202,
      "grad_norm": 0.03905004635453224,
      "learning_rate": 0.00014098765432098763,
      "loss": 0.6104,
      "step": 780
    },
    {
      "epoch": 1.7326677759290072,
      "grad_norm": 0.03439386934041977,
      "learning_rate": 0.00014074074074074073,
      "loss": 0.53,
      "step": 781
    },
    {
      "epoch": 1.7348863006100943,
      "grad_norm": 0.06457950174808502,
      "learning_rate": 0.00014049382716049382,
      "loss": 0.574,
      "step": 782
    },
    {
      "epoch": 1.7371048252911814,
      "grad_norm": 0.03952830284833908,
      "learning_rate": 0.0001402469135802469,
      "loss": 0.6007,
      "step": 783
    },
    {
      "epoch": 1.7393233499722685,
      "grad_norm": 0.043567102402448654,
      "learning_rate": 0.00014,
      "loss": 0.539,
      "step": 784
    },
    {
      "epoch": 1.7415418746533555,
      "grad_norm": 0.03959580883383751,
      "learning_rate": 0.00013975308641975308,
      "loss": 0.5983,
      "step": 785
    },
    {
      "epoch": 1.7437603993344426,
      "grad_norm": 0.03956570103764534,
      "learning_rate": 0.00013950617283950615,
      "loss": 0.605,
      "step": 786
    },
    {
      "epoch": 1.7459789240155297,
      "grad_norm": 0.03941044956445694,
      "learning_rate": 0.00013925925925925925,
      "loss": 0.4739,
      "step": 787
    },
    {
      "epoch": 1.7481974486966168,
      "grad_norm": 0.04601512849330902,
      "learning_rate": 0.00013901234567901231,
      "loss": 0.7618,
      "step": 788
    },
    {
      "epoch": 1.7504159733777038,
      "grad_norm": 0.04309384897351265,
      "learning_rate": 0.00013876543209876544,
      "loss": 0.5319,
      "step": 789
    },
    {
      "epoch": 1.752634498058791,
      "grad_norm": 0.039698418229818344,
      "learning_rate": 0.0001385185185185185,
      "loss": 0.488,
      "step": 790
    },
    {
      "epoch": 1.754853022739878,
      "grad_norm": 0.03702308610081673,
      "learning_rate": 0.0001382716049382716,
      "loss": 0.5841,
      "step": 791
    },
    {
      "epoch": 1.757071547420965,
      "grad_norm": 0.03526584804058075,
      "learning_rate": 0.00013802469135802467,
      "loss": 0.4679,
      "step": 792
    },
    {
      "epoch": 1.7592900721020521,
      "grad_norm": 0.05473826825618744,
      "learning_rate": 0.00013777777777777776,
      "loss": 0.4533,
      "step": 793
    },
    {
      "epoch": 1.7615085967831392,
      "grad_norm": 0.04165370762348175,
      "learning_rate": 0.00013753086419753086,
      "loss": 0.5831,
      "step": 794
    },
    {
      "epoch": 1.7637271214642263,
      "grad_norm": 0.044653844088315964,
      "learning_rate": 0.00013728395061728393,
      "loss": 0.6227,
      "step": 795
    },
    {
      "epoch": 1.7659456461453134,
      "grad_norm": 0.0428091399371624,
      "learning_rate": 0.00013703703703703702,
      "loss": 0.5819,
      "step": 796
    },
    {
      "epoch": 1.7681641708264004,
      "grad_norm": 0.03761467710137367,
      "learning_rate": 0.00013679012345679012,
      "loss": 0.583,
      "step": 797
    },
    {
      "epoch": 1.7703826955074875,
      "grad_norm": 0.037230607122182846,
      "learning_rate": 0.0001365432098765432,
      "loss": 0.5185,
      "step": 798
    },
    {
      "epoch": 1.7726012201885746,
      "grad_norm": 0.04164591059088707,
      "learning_rate": 0.00013629629629629628,
      "loss": 0.6262,
      "step": 799
    },
    {
      "epoch": 1.7748197448696617,
      "grad_norm": 0.03944012150168419,
      "learning_rate": 0.00013604938271604937,
      "loss": 0.5155,
      "step": 800
    },
    {
      "epoch": 1.7770382695507487,
      "grad_norm": 0.037940919399261475,
      "learning_rate": 0.00013580246913580244,
      "loss": 0.4909,
      "step": 801
    },
    {
      "epoch": 1.7792567942318358,
      "grad_norm": 0.0380069874227047,
      "learning_rate": 0.00013555555555555554,
      "loss": 0.5103,
      "step": 802
    },
    {
      "epoch": 1.7814753189129229,
      "grad_norm": 0.03154885396361351,
      "learning_rate": 0.00013530864197530863,
      "loss": 0.4727,
      "step": 803
    },
    {
      "epoch": 1.78369384359401,
      "grad_norm": 0.053600605577230453,
      "learning_rate": 0.00013506172839506173,
      "loss": 0.617,
      "step": 804
    },
    {
      "epoch": 1.785912368275097,
      "grad_norm": 0.03860149160027504,
      "learning_rate": 0.0001348148148148148,
      "loss": 0.4682,
      "step": 805
    },
    {
      "epoch": 1.788130892956184,
      "grad_norm": 0.03296085074543953,
      "learning_rate": 0.0001345679012345679,
      "loss": 0.5297,
      "step": 806
    },
    {
      "epoch": 1.7903494176372712,
      "grad_norm": 0.04759255051612854,
      "learning_rate": 0.00013432098765432099,
      "loss": 0.5755,
      "step": 807
    },
    {
      "epoch": 1.7925679423183583,
      "grad_norm": 0.06548307090997696,
      "learning_rate": 0.00013407407407407405,
      "loss": 0.6593,
      "step": 808
    },
    {
      "epoch": 1.7947864669994453,
      "grad_norm": 0.02739465795457363,
      "learning_rate": 0.00013382716049382715,
      "loss": 0.4612,
      "step": 809
    },
    {
      "epoch": 1.7970049916805324,
      "grad_norm": 0.03148824721574783,
      "learning_rate": 0.00013358024691358022,
      "loss": 0.5242,
      "step": 810
    },
    {
      "epoch": 1.7992235163616195,
      "grad_norm": 0.03582452982664108,
      "learning_rate": 0.0001333333333333333,
      "loss": 0.4903,
      "step": 811
    },
    {
      "epoch": 1.8014420410427066,
      "grad_norm": 0.03661932051181793,
      "learning_rate": 0.0001330864197530864,
      "loss": 0.5826,
      "step": 812
    },
    {
      "epoch": 1.8036605657237936,
      "grad_norm": 0.04068029299378395,
      "learning_rate": 0.0001328395061728395,
      "loss": 0.5919,
      "step": 813
    },
    {
      "epoch": 1.8058790904048807,
      "grad_norm": 0.04375843703746796,
      "learning_rate": 0.00013259259259259257,
      "loss": 0.5576,
      "step": 814
    },
    {
      "epoch": 1.8080976150859678,
      "grad_norm": 0.0557732954621315,
      "learning_rate": 0.00013234567901234567,
      "loss": 0.5186,
      "step": 815
    },
    {
      "epoch": 1.8103161397670549,
      "grad_norm": 0.07507844269275665,
      "learning_rate": 0.00013209876543209876,
      "loss": 0.7415,
      "step": 816
    },
    {
      "epoch": 1.8125346644481422,
      "grad_norm": 0.03939047455787659,
      "learning_rate": 0.00013185185185185183,
      "loss": 0.5406,
      "step": 817
    },
    {
      "epoch": 1.8147531891292292,
      "grad_norm": 0.032194025814533234,
      "learning_rate": 0.00013160493827160492,
      "loss": 0.5922,
      "step": 818
    },
    {
      "epoch": 1.8169717138103163,
      "grad_norm": 0.04659094288945198,
      "learning_rate": 0.00013135802469135802,
      "loss": 0.5689,
      "step": 819
    },
    {
      "epoch": 1.8191902384914034,
      "grad_norm": 0.04984493926167488,
      "learning_rate": 0.00013111111111111111,
      "loss": 0.533,
      "step": 820
    },
    {
      "epoch": 1.8214087631724905,
      "grad_norm": 0.042220309376716614,
      "learning_rate": 0.00013086419753086418,
      "loss": 0.5932,
      "step": 821
    },
    {
      "epoch": 1.8236272878535775,
      "grad_norm": 0.05591040104627609,
      "learning_rate": 0.00013061728395061728,
      "loss": 0.6132,
      "step": 822
    },
    {
      "epoch": 1.8258458125346646,
      "grad_norm": 0.037706296890974045,
      "learning_rate": 0.00013037037037037035,
      "loss": 0.5186,
      "step": 823
    },
    {
      "epoch": 1.8280643372157517,
      "grad_norm": 0.05157928168773651,
      "learning_rate": 0.00013012345679012344,
      "loss": 0.6994,
      "step": 824
    },
    {
      "epoch": 1.8302828618968388,
      "grad_norm": 0.038450270891189575,
      "learning_rate": 0.00012987654320987654,
      "loss": 0.5838,
      "step": 825
    },
    {
      "epoch": 1.8325013865779258,
      "grad_norm": 0.062297165393829346,
      "learning_rate": 0.0001296296296296296,
      "loss": 0.6127,
      "step": 826
    },
    {
      "epoch": 1.834719911259013,
      "grad_norm": 0.03308014199137688,
      "learning_rate": 0.0001293827160493827,
      "loss": 0.6331,
      "step": 827
    },
    {
      "epoch": 1.8369384359401,
      "grad_norm": 0.04162484407424927,
      "learning_rate": 0.0001291358024691358,
      "loss": 0.5124,
      "step": 828
    },
    {
      "epoch": 1.839156960621187,
      "grad_norm": 0.031125960871577263,
      "learning_rate": 0.0001288888888888889,
      "loss": 0.5942,
      "step": 829
    },
    {
      "epoch": 1.8413754853022741,
      "grad_norm": 0.07014033198356628,
      "learning_rate": 0.00012864197530864196,
      "loss": 0.6086,
      "step": 830
    },
    {
      "epoch": 1.8435940099833612,
      "grad_norm": 0.07441037893295288,
      "learning_rate": 0.00012839506172839505,
      "loss": 0.5575,
      "step": 831
    },
    {
      "epoch": 1.8458125346644483,
      "grad_norm": 0.04153788462281227,
      "learning_rate": 0.00012814814814814812,
      "loss": 0.5741,
      "step": 832
    },
    {
      "epoch": 1.8480310593455354,
      "grad_norm": 0.05977976694703102,
      "learning_rate": 0.00012790123456790122,
      "loss": 0.6735,
      "step": 833
    },
    {
      "epoch": 1.8502495840266224,
      "grad_norm": 0.030887646600604057,
      "learning_rate": 0.0001276543209876543,
      "loss": 0.465,
      "step": 834
    },
    {
      "epoch": 1.8524681087077095,
      "grad_norm": 0.044465914368629456,
      "learning_rate": 0.0001274074074074074,
      "loss": 0.5505,
      "step": 835
    },
    {
      "epoch": 1.8546866333887966,
      "grad_norm": 0.04347898066043854,
      "learning_rate": 0.00012716049382716047,
      "loss": 0.6784,
      "step": 836
    },
    {
      "epoch": 1.8569051580698837,
      "grad_norm": 0.0450383685529232,
      "learning_rate": 0.00012691358024691357,
      "loss": 0.6173,
      "step": 837
    },
    {
      "epoch": 1.8591236827509707,
      "grad_norm": 0.05687909945845604,
      "learning_rate": 0.00012666666666666666,
      "loss": 0.5455,
      "step": 838
    },
    {
      "epoch": 1.8613422074320578,
      "grad_norm": 0.039995402097702026,
      "learning_rate": 0.00012641975308641973,
      "loss": 0.4694,
      "step": 839
    },
    {
      "epoch": 1.8635607321131449,
      "grad_norm": 0.0334894135594368,
      "learning_rate": 0.00012617283950617283,
      "loss": 0.5635,
      "step": 840
    },
    {
      "epoch": 1.865779256794232,
      "grad_norm": 0.0635187029838562,
      "learning_rate": 0.0001259259259259259,
      "loss": 0.7049,
      "step": 841
    },
    {
      "epoch": 1.867997781475319,
      "grad_norm": 0.04256037250161171,
      "learning_rate": 0.00012567901234567902,
      "loss": 0.634,
      "step": 842
    },
    {
      "epoch": 1.870216306156406,
      "grad_norm": 0.05096891522407532,
      "learning_rate": 0.00012543209876543209,
      "loss": 0.5249,
      "step": 843
    },
    {
      "epoch": 1.8724348308374932,
      "grad_norm": 0.041665419936180115,
      "learning_rate": 0.00012518518518518518,
      "loss": 0.6832,
      "step": 844
    },
    {
      "epoch": 1.8746533555185803,
      "grad_norm": 0.03602922707796097,
      "learning_rate": 0.00012493827160493825,
      "loss": 0.5072,
      "step": 845
    },
    {
      "epoch": 1.8768718801996673,
      "grad_norm": 0.039854612201452255,
      "learning_rate": 0.00012469135802469134,
      "loss": 0.6509,
      "step": 846
    },
    {
      "epoch": 1.8790904048807544,
      "grad_norm": 0.035419754683971405,
      "learning_rate": 0.00012444444444444444,
      "loss": 0.5361,
      "step": 847
    },
    {
      "epoch": 1.8813089295618415,
      "grad_norm": 0.03483320772647858,
      "learning_rate": 0.0001241975308641975,
      "loss": 0.6028,
      "step": 848
    },
    {
      "epoch": 1.8835274542429286,
      "grad_norm": 0.04165695235133171,
      "learning_rate": 0.0001239506172839506,
      "loss": 0.5274,
      "step": 849
    },
    {
      "epoch": 1.8857459789240156,
      "grad_norm": 0.04272043704986572,
      "learning_rate": 0.0001237037037037037,
      "loss": 0.5866,
      "step": 850
    },
    {
      "epoch": 1.8879645036051027,
      "grad_norm": 0.038002751767635345,
      "learning_rate": 0.0001234567901234568,
      "loss": 0.584,
      "step": 851
    },
    {
      "epoch": 1.8901830282861898,
      "grad_norm": 0.033742114901542664,
      "learning_rate": 0.00012320987654320986,
      "loss": 0.5509,
      "step": 852
    },
    {
      "epoch": 1.8924015529672769,
      "grad_norm": 0.03713928908109665,
      "learning_rate": 0.00012296296296296296,
      "loss": 0.439,
      "step": 853
    },
    {
      "epoch": 1.894620077648364,
      "grad_norm": 0.03454671427607536,
      "learning_rate": 0.00012271604938271602,
      "loss": 0.5287,
      "step": 854
    },
    {
      "epoch": 1.896838602329451,
      "grad_norm": 0.029390433803200722,
      "learning_rate": 0.00012246913580246912,
      "loss": 0.7221,
      "step": 855
    },
    {
      "epoch": 1.899057127010538,
      "grad_norm": 0.027915360406041145,
      "learning_rate": 0.00012222222222222221,
      "loss": 0.4638,
      "step": 856
    },
    {
      "epoch": 1.9012756516916252,
      "grad_norm": 0.03753940388560295,
      "learning_rate": 0.00012197530864197531,
      "loss": 0.422,
      "step": 857
    },
    {
      "epoch": 1.9034941763727122,
      "grad_norm": 0.05374818667769432,
      "learning_rate": 0.00012172839506172839,
      "loss": 0.6717,
      "step": 858
    },
    {
      "epoch": 1.9057127010537993,
      "grad_norm": 0.05799092724919319,
      "learning_rate": 0.00012148148148148147,
      "loss": 0.7482,
      "step": 859
    },
    {
      "epoch": 1.9079312257348864,
      "grad_norm": 0.04836909845471382,
      "learning_rate": 0.00012123456790123455,
      "loss": 0.5809,
      "step": 860
    },
    {
      "epoch": 1.9101497504159735,
      "grad_norm": 0.033289309591054916,
      "learning_rate": 0.00012098765432098764,
      "loss": 0.7623,
      "step": 861
    },
    {
      "epoch": 1.9123682750970605,
      "grad_norm": 0.033976126462221146,
      "learning_rate": 0.00012074074074074073,
      "loss": 0.4701,
      "step": 862
    },
    {
      "epoch": 1.9145867997781476,
      "grad_norm": 0.03435344994068146,
      "learning_rate": 0.00012049382716049381,
      "loss": 0.4677,
      "step": 863
    },
    {
      "epoch": 1.9168053244592347,
      "grad_norm": 0.042621590197086334,
      "learning_rate": 0.0001202469135802469,
      "loss": 0.5382,
      "step": 864
    },
    {
      "epoch": 1.9190238491403218,
      "grad_norm": 0.039394307881593704,
      "learning_rate": 0.00011999999999999999,
      "loss": 0.547,
      "step": 865
    },
    {
      "epoch": 1.9212423738214088,
      "grad_norm": 0.04855889454483986,
      "learning_rate": 0.00011975308641975308,
      "loss": 0.5583,
      "step": 866
    },
    {
      "epoch": 1.923460898502496,
      "grad_norm": 0.04434354975819588,
      "learning_rate": 0.00011950617283950617,
      "loss": 0.4947,
      "step": 867
    },
    {
      "epoch": 1.925679423183583,
      "grad_norm": 0.04484400153160095,
      "learning_rate": 0.00011925925925925925,
      "loss": 0.7012,
      "step": 868
    },
    {
      "epoch": 1.92789794786467,
      "grad_norm": 0.059671830385923386,
      "learning_rate": 0.00011901234567901233,
      "loss": 0.5806,
      "step": 869
    },
    {
      "epoch": 1.9301164725457571,
      "grad_norm": 0.05093313008546829,
      "learning_rate": 0.00011876543209876541,
      "loss": 0.7626,
      "step": 870
    },
    {
      "epoch": 1.9323349972268442,
      "grad_norm": 0.038524579256772995,
      "learning_rate": 0.0001185185185185185,
      "loss": 0.5775,
      "step": 871
    },
    {
      "epoch": 1.9345535219079313,
      "grad_norm": 0.04435029998421669,
      "learning_rate": 0.0001182716049382716,
      "loss": 0.6085,
      "step": 872
    },
    {
      "epoch": 1.9367720465890184,
      "grad_norm": 0.03293889760971069,
      "learning_rate": 0.00011802469135802468,
      "loss": 0.5176,
      "step": 873
    },
    {
      "epoch": 1.9389905712701054,
      "grad_norm": 0.03898153826594353,
      "learning_rate": 0.00011777777777777776,
      "loss": 0.5794,
      "step": 874
    },
    {
      "epoch": 1.9412090959511925,
      "grad_norm": 0.03818023204803467,
      "learning_rate": 0.00011753086419753086,
      "loss": 0.6088,
      "step": 875
    },
    {
      "epoch": 1.9434276206322796,
      "grad_norm": 0.05293511971831322,
      "learning_rate": 0.00011728395061728394,
      "loss": 0.5501,
      "step": 876
    },
    {
      "epoch": 1.9456461453133667,
      "grad_norm": 0.04615112766623497,
      "learning_rate": 0.00011703703703703702,
      "loss": 0.4523,
      "step": 877
    },
    {
      "epoch": 1.9478646699944537,
      "grad_norm": 0.03888378664851189,
      "learning_rate": 0.0001167901234567901,
      "loss": 0.5268,
      "step": 878
    },
    {
      "epoch": 1.9500831946755408,
      "grad_norm": 0.039171457290649414,
      "learning_rate": 0.00011654320987654319,
      "loss": 0.5982,
      "step": 879
    },
    {
      "epoch": 1.9523017193566279,
      "grad_norm": 0.031206833198666573,
      "learning_rate": 0.0001162962962962963,
      "loss": 0.4954,
      "step": 880
    },
    {
      "epoch": 1.954520244037715,
      "grad_norm": 0.037594735622406006,
      "learning_rate": 0.00011604938271604938,
      "loss": 0.6268,
      "step": 881
    },
    {
      "epoch": 1.956738768718802,
      "grad_norm": 0.030708303675055504,
      "learning_rate": 0.00011580246913580246,
      "loss": 0.5756,
      "step": 882
    },
    {
      "epoch": 1.958957293399889,
      "grad_norm": 0.04101470485329628,
      "learning_rate": 0.00011555555555555554,
      "loss": 0.667,
      "step": 883
    },
    {
      "epoch": 1.9611758180809762,
      "grad_norm": 0.04550621658563614,
      "learning_rate": 0.00011530864197530863,
      "loss": 0.5717,
      "step": 884
    },
    {
      "epoch": 1.9633943427620633,
      "grad_norm": 0.03729967772960663,
      "learning_rate": 0.00011506172839506172,
      "loss": 0.6258,
      "step": 885
    },
    {
      "epoch": 1.9656128674431503,
      "grad_norm": 0.046293359249830246,
      "learning_rate": 0.0001148148148148148,
      "loss": 0.5932,
      "step": 886
    },
    {
      "epoch": 1.9678313921242374,
      "grad_norm": 0.05053786560893059,
      "learning_rate": 0.00011456790123456789,
      "loss": 0.5735,
      "step": 887
    },
    {
      "epoch": 1.9700499168053245,
      "grad_norm": 0.04987156391143799,
      "learning_rate": 0.00011432098765432099,
      "loss": 0.5786,
      "step": 888
    },
    {
      "epoch": 1.9722684414864116,
      "grad_norm": 0.03204355761408806,
      "learning_rate": 0.00011407407407407407,
      "loss": 0.5158,
      "step": 889
    },
    {
      "epoch": 1.9744869661674986,
      "grad_norm": 0.03077753260731697,
      "learning_rate": 0.00011382716049382715,
      "loss": 0.4855,
      "step": 890
    },
    {
      "epoch": 1.9767054908485857,
      "grad_norm": 0.04180914908647537,
      "learning_rate": 0.00011358024691358023,
      "loss": 0.6378,
      "step": 891
    },
    {
      "epoch": 1.9789240155296728,
      "grad_norm": 0.0315224714577198,
      "learning_rate": 0.00011333333333333331,
      "loss": 0.4457,
      "step": 892
    },
    {
      "epoch": 1.9811425402107599,
      "grad_norm": 0.052665095776319504,
      "learning_rate": 0.00011308641975308641,
      "loss": 0.5204,
      "step": 893
    },
    {
      "epoch": 1.983361064891847,
      "grad_norm": 0.033638037741184235,
      "learning_rate": 0.00011283950617283949,
      "loss": 0.5009,
      "step": 894
    },
    {
      "epoch": 1.985579589572934,
      "grad_norm": 0.03907589241862297,
      "learning_rate": 0.00011259259259259259,
      "loss": 0.6065,
      "step": 895
    },
    {
      "epoch": 1.987798114254021,
      "grad_norm": 0.057822152972221375,
      "learning_rate": 0.00011234567901234567,
      "loss": 0.5652,
      "step": 896
    },
    {
      "epoch": 1.9900166389351082,
      "grad_norm": 0.036670468747615814,
      "learning_rate": 0.00011209876543209876,
      "loss": 0.482,
      "step": 897
    },
    {
      "epoch": 1.9922351636161952,
      "grad_norm": 0.03933965414762497,
      "learning_rate": 0.00011185185185185184,
      "loss": 0.467,
      "step": 898
    },
    {
      "epoch": 1.9944536882972823,
      "grad_norm": 0.03263339772820473,
      "learning_rate": 0.00011160493827160493,
      "loss": 0.5597,
      "step": 899
    },
    {
      "epoch": 1.9966722129783694,
      "grad_norm": 0.04834209755063057,
      "learning_rate": 0.00011135802469135801,
      "loss": 0.6337,
      "step": 900
    },
    {
      "epoch": 1.9988907376594565,
      "grad_norm": 0.03267941251397133,
      "learning_rate": 0.00011111111111111109,
      "loss": 0.5184,
      "step": 901
    },
    {
      "epoch": 2.0011092623405435,
      "grad_norm": 0.0361943319439888,
      "learning_rate": 0.0001108641975308642,
      "loss": 0.6063,
      "step": 902
    },
    {
      "epoch": 2.0033277870216306,
      "grad_norm": 0.036976832896471024,
      "learning_rate": 0.00011061728395061728,
      "loss": 0.4793,
      "step": 903
    },
    {
      "epoch": 2.0055463117027177,
      "grad_norm": 0.05037384480237961,
      "learning_rate": 0.00011037037037037036,
      "loss": 0.6375,
      "step": 904
    },
    {
      "epoch": 2.0077648363838048,
      "grad_norm": 0.03725269064307213,
      "learning_rate": 0.00011012345679012344,
      "loss": 0.6251,
      "step": 905
    },
    {
      "epoch": 2.009983361064892,
      "grad_norm": 0.057177457958459854,
      "learning_rate": 0.00010987654320987654,
      "loss": 0.5192,
      "step": 906
    },
    {
      "epoch": 2.012201885745979,
      "grad_norm": 0.03307095915079117,
      "learning_rate": 0.00010962962962962962,
      "loss": 0.5717,
      "step": 907
    },
    {
      "epoch": 2.014420410427066,
      "grad_norm": 0.03307688981294632,
      "learning_rate": 0.0001093827160493827,
      "loss": 0.5823,
      "step": 908
    },
    {
      "epoch": 2.016638935108153,
      "grad_norm": 0.042366255074739456,
      "learning_rate": 0.00010913580246913578,
      "loss": 0.6722,
      "step": 909
    },
    {
      "epoch": 2.01885745978924,
      "grad_norm": 0.042373593896627426,
      "learning_rate": 0.00010888888888888889,
      "loss": 0.6113,
      "step": 910
    },
    {
      "epoch": 2.021075984470327,
      "grad_norm": 0.03368338942527771,
      "learning_rate": 0.00010864197530864197,
      "loss": 0.5809,
      "step": 911
    },
    {
      "epoch": 2.0232945091514143,
      "grad_norm": 0.04419630393385887,
      "learning_rate": 0.00010839506172839505,
      "loss": 0.5323,
      "step": 912
    },
    {
      "epoch": 2.0255130338325014,
      "grad_norm": 0.05008912831544876,
      "learning_rate": 0.00010814814814814814,
      "loss": 0.4526,
      "step": 913
    },
    {
      "epoch": 2.0277315585135884,
      "grad_norm": 0.03352295234799385,
      "learning_rate": 0.00010790123456790122,
      "loss": 0.6005,
      "step": 914
    },
    {
      "epoch": 2.0299500831946755,
      "grad_norm": 0.06918387115001678,
      "learning_rate": 0.00010765432098765431,
      "loss": 0.5343,
      "step": 915
    },
    {
      "epoch": 2.0321686078757626,
      "grad_norm": 0.03278977796435356,
      "learning_rate": 0.0001074074074074074,
      "loss": 0.4362,
      "step": 916
    },
    {
      "epoch": 2.0343871325568497,
      "grad_norm": 0.039134204387664795,
      "learning_rate": 0.00010716049382716048,
      "loss": 0.5218,
      "step": 917
    },
    {
      "epoch": 2.0366056572379367,
      "grad_norm": 0.038427144289016724,
      "learning_rate": 0.00010691358024691357,
      "loss": 0.5071,
      "step": 918
    },
    {
      "epoch": 2.038824181919024,
      "grad_norm": 0.03353983163833618,
      "learning_rate": 0.00010666666666666667,
      "loss": 0.6886,
      "step": 919
    },
    {
      "epoch": 2.041042706600111,
      "grad_norm": 0.03024151176214218,
      "learning_rate": 0.00010641975308641975,
      "loss": 0.4486,
      "step": 920
    },
    {
      "epoch": 2.043261231281198,
      "grad_norm": 0.028950698673725128,
      "learning_rate": 0.00010617283950617283,
      "loss": 0.5078,
      "step": 921
    },
    {
      "epoch": 2.045479755962285,
      "grad_norm": 0.04098600521683693,
      "learning_rate": 0.00010592592592592591,
      "loss": 0.5424,
      "step": 922
    },
    {
      "epoch": 2.047698280643372,
      "grad_norm": 0.034262459725141525,
      "learning_rate": 0.00010567901234567899,
      "loss": 0.4674,
      "step": 923
    },
    {
      "epoch": 2.049916805324459,
      "grad_norm": 0.03452077880501747,
      "learning_rate": 0.00010543209876543209,
      "loss": 0.4565,
      "step": 924
    },
    {
      "epoch": 2.0521353300055463,
      "grad_norm": 0.03236804157495499,
      "learning_rate": 0.00010518518518518518,
      "loss": 0.4929,
      "step": 925
    },
    {
      "epoch": 2.0543538546866333,
      "grad_norm": 0.041670091450214386,
      "learning_rate": 0.00010493827160493826,
      "loss": 0.5934,
      "step": 926
    },
    {
      "epoch": 2.0565723793677204,
      "grad_norm": 0.035300642251968384,
      "learning_rate": 0.00010469135802469135,
      "loss": 0.5981,
      "step": 927
    },
    {
      "epoch": 2.0587909040488075,
      "grad_norm": 0.0449208989739418,
      "learning_rate": 0.00010444444444444444,
      "loss": 0.4521,
      "step": 928
    },
    {
      "epoch": 2.0610094287298946,
      "grad_norm": 0.03894314542412758,
      "learning_rate": 0.00010419753086419752,
      "loss": 0.6671,
      "step": 929
    },
    {
      "epoch": 2.0632279534109816,
      "grad_norm": 0.03621317818760872,
      "learning_rate": 0.0001039506172839506,
      "loss": 0.5376,
      "step": 930
    },
    {
      "epoch": 2.0654464780920687,
      "grad_norm": 0.033455200493335724,
      "learning_rate": 0.00010370370370370369,
      "loss": 0.713,
      "step": 931
    },
    {
      "epoch": 2.067665002773156,
      "grad_norm": 0.032819852232933044,
      "learning_rate": 0.00010345679012345677,
      "loss": 0.4664,
      "step": 932
    },
    {
      "epoch": 2.069883527454243,
      "grad_norm": 0.03853554651141167,
      "learning_rate": 0.00010320987654320988,
      "loss": 0.6311,
      "step": 933
    },
    {
      "epoch": 2.07210205213533,
      "grad_norm": 0.04115855321288109,
      "learning_rate": 0.00010296296296296296,
      "loss": 0.6775,
      "step": 934
    },
    {
      "epoch": 2.074320576816417,
      "grad_norm": 0.04069429636001587,
      "learning_rate": 0.00010271604938271604,
      "loss": 0.557,
      "step": 935
    },
    {
      "epoch": 2.076539101497504,
      "grad_norm": 0.04103720933198929,
      "learning_rate": 0.00010246913580246912,
      "loss": 0.7299,
      "step": 936
    },
    {
      "epoch": 2.078757626178591,
      "grad_norm": 0.03932051360607147,
      "learning_rate": 0.00010222222222222222,
      "loss": 0.5947,
      "step": 937
    },
    {
      "epoch": 2.0809761508596782,
      "grad_norm": 0.05004004389047623,
      "learning_rate": 0.0001019753086419753,
      "loss": 0.4227,
      "step": 938
    },
    {
      "epoch": 2.0831946755407653,
      "grad_norm": 0.04372744634747505,
      "learning_rate": 0.00010172839506172838,
      "loss": 0.5291,
      "step": 939
    },
    {
      "epoch": 2.0854132002218524,
      "grad_norm": 0.032593049108982086,
      "learning_rate": 0.00010148148148148147,
      "loss": 0.4796,
      "step": 940
    },
    {
      "epoch": 2.0876317249029395,
      "grad_norm": 0.03912177309393883,
      "learning_rate": 0.00010123456790123457,
      "loss": 0.6587,
      "step": 941
    },
    {
      "epoch": 2.0898502495840265,
      "grad_norm": 0.038519419729709625,
      "learning_rate": 0.00010098765432098765,
      "loss": 0.6196,
      "step": 942
    },
    {
      "epoch": 2.0920687742651136,
      "grad_norm": 0.04508130997419357,
      "learning_rate": 0.00010074074074074073,
      "loss": 0.5826,
      "step": 943
    },
    {
      "epoch": 2.0942872989462007,
      "grad_norm": 0.03528081253170967,
      "learning_rate": 0.00010049382716049381,
      "loss": 0.5277,
      "step": 944
    },
    {
      "epoch": 2.0965058236272878,
      "grad_norm": 0.03180696815252304,
      "learning_rate": 0.0001002469135802469,
      "loss": 0.706,
      "step": 945
    },
    {
      "epoch": 2.098724348308375,
      "grad_norm": 0.05823463574051857,
      "learning_rate": 9.999999999999999e-05,
      "loss": 0.6364,
      "step": 946
    },
    {
      "epoch": 2.100942872989462,
      "grad_norm": 0.04195134714245796,
      "learning_rate": 9.975308641975307e-05,
      "loss": 0.5328,
      "step": 947
    },
    {
      "epoch": 2.103161397670549,
      "grad_norm": 0.05784667283296585,
      "learning_rate": 9.950617283950617e-05,
      "loss": 0.5445,
      "step": 948
    },
    {
      "epoch": 2.105379922351636,
      "grad_norm": 0.02850363217294216,
      "learning_rate": 9.925925925925925e-05,
      "loss": 0.4965,
      "step": 949
    },
    {
      "epoch": 2.107598447032723,
      "grad_norm": 0.042600154876708984,
      "learning_rate": 9.901234567901234e-05,
      "loss": 0.5782,
      "step": 950
    },
    {
      "epoch": 2.10981697171381,
      "grad_norm": 0.03764021769165993,
      "learning_rate": 9.876543209876543e-05,
      "loss": 0.5927,
      "step": 951
    },
    {
      "epoch": 2.1120354963948973,
      "grad_norm": 0.03132227808237076,
      "learning_rate": 9.851851851851851e-05,
      "loss": 0.4786,
      "step": 952
    },
    {
      "epoch": 2.1142540210759844,
      "grad_norm": 0.0473855622112751,
      "learning_rate": 9.827160493827159e-05,
      "loss": 0.571,
      "step": 953
    },
    {
      "epoch": 2.1164725457570714,
      "grad_norm": 0.04276057332754135,
      "learning_rate": 9.802469135802467e-05,
      "loss": 0.6177,
      "step": 954
    },
    {
      "epoch": 2.1186910704381585,
      "grad_norm": 0.038058385252952576,
      "learning_rate": 9.777777777777778e-05,
      "loss": 0.4823,
      "step": 955
    },
    {
      "epoch": 2.1209095951192456,
      "grad_norm": 0.04257947951555252,
      "learning_rate": 9.753086419753086e-05,
      "loss": 0.6113,
      "step": 956
    },
    {
      "epoch": 2.1231281198003327,
      "grad_norm": 0.03807676210999489,
      "learning_rate": 9.728395061728394e-05,
      "loss": 0.6813,
      "step": 957
    },
    {
      "epoch": 2.1253466444814197,
      "grad_norm": 0.042694754898548126,
      "learning_rate": 9.703703703703702e-05,
      "loss": 0.619,
      "step": 958
    },
    {
      "epoch": 2.127565169162507,
      "grad_norm": 0.04667281359434128,
      "learning_rate": 9.679012345679012e-05,
      "loss": 0.5401,
      "step": 959
    },
    {
      "epoch": 2.129783693843594,
      "grad_norm": 0.08750811964273453,
      "learning_rate": 9.65432098765432e-05,
      "loss": 0.8228,
      "step": 960
    },
    {
      "epoch": 2.132002218524681,
      "grad_norm": 0.03858141228556633,
      "learning_rate": 9.629629629629628e-05,
      "loss": 0.4523,
      "step": 961
    },
    {
      "epoch": 2.134220743205768,
      "grad_norm": 0.05716320499777794,
      "learning_rate": 9.604938271604936e-05,
      "loss": 0.8255,
      "step": 962
    },
    {
      "epoch": 2.136439267886855,
      "grad_norm": 0.04648544266819954,
      "learning_rate": 9.580246913580247e-05,
      "loss": 0.6095,
      "step": 963
    },
    {
      "epoch": 2.138657792567942,
      "grad_norm": 0.040696803480386734,
      "learning_rate": 9.555555555555555e-05,
      "loss": 0.5849,
      "step": 964
    },
    {
      "epoch": 2.1408763172490293,
      "grad_norm": 0.04951886460185051,
      "learning_rate": 9.530864197530864e-05,
      "loss": 0.64,
      "step": 965
    },
    {
      "epoch": 2.1430948419301163,
      "grad_norm": 0.039406560361385345,
      "learning_rate": 9.506172839506172e-05,
      "loss": 0.649,
      "step": 966
    },
    {
      "epoch": 2.1453133666112034,
      "grad_norm": 0.04234379902482033,
      "learning_rate": 9.48148148148148e-05,
      "loss": 0.7074,
      "step": 967
    },
    {
      "epoch": 2.1475318912922905,
      "grad_norm": 0.029455428943037987,
      "learning_rate": 9.45679012345679e-05,
      "loss": 0.4806,
      "step": 968
    },
    {
      "epoch": 2.1497504159733776,
      "grad_norm": 0.03162490949034691,
      "learning_rate": 9.432098765432098e-05,
      "loss": 0.6173,
      "step": 969
    },
    {
      "epoch": 2.1519689406544646,
      "grad_norm": 0.040464457124471664,
      "learning_rate": 9.407407407407407e-05,
      "loss": 0.6578,
      "step": 970
    },
    {
      "epoch": 2.1541874653355517,
      "grad_norm": 0.07601195573806763,
      "learning_rate": 9.382716049382715e-05,
      "loss": 0.3986,
      "step": 971
    },
    {
      "epoch": 2.156405990016639,
      "grad_norm": 0.03190547972917557,
      "learning_rate": 9.358024691358025e-05,
      "loss": 0.4303,
      "step": 972
    },
    {
      "epoch": 2.158624514697726,
      "grad_norm": 0.03842867165803909,
      "learning_rate": 9.333333333333333e-05,
      "loss": 0.5898,
      "step": 973
    },
    {
      "epoch": 2.160843039378813,
      "grad_norm": 0.041078660637140274,
      "learning_rate": 9.308641975308641e-05,
      "loss": 0.5944,
      "step": 974
    },
    {
      "epoch": 2.1630615640599,
      "grad_norm": 0.03528919070959091,
      "learning_rate": 9.283950617283949e-05,
      "loss": 0.5488,
      "step": 975
    },
    {
      "epoch": 2.165280088740987,
      "grad_norm": 0.033999886363744736,
      "learning_rate": 9.259259259259257e-05,
      "loss": 0.5435,
      "step": 976
    },
    {
      "epoch": 2.167498613422074,
      "grad_norm": 0.026123588904738426,
      "learning_rate": 9.234567901234567e-05,
      "loss": 0.3573,
      "step": 977
    },
    {
      "epoch": 2.1697171381031612,
      "grad_norm": 0.03814035281538963,
      "learning_rate": 9.209876543209876e-05,
      "loss": 0.5351,
      "step": 978
    },
    {
      "epoch": 2.1719356627842483,
      "grad_norm": 0.03665050491690636,
      "learning_rate": 9.185185185185185e-05,
      "loss": 0.8903,
      "step": 979
    },
    {
      "epoch": 2.1741541874653354,
      "grad_norm": 0.0444774366915226,
      "learning_rate": 9.160493827160493e-05,
      "loss": 0.6411,
      "step": 980
    },
    {
      "epoch": 2.1763727121464225,
      "grad_norm": 0.05196886882185936,
      "learning_rate": 9.135802469135802e-05,
      "loss": 0.4207,
      "step": 981
    },
    {
      "epoch": 2.1785912368275095,
      "grad_norm": 0.045661017298698425,
      "learning_rate": 9.11111111111111e-05,
      "loss": 0.4935,
      "step": 982
    },
    {
      "epoch": 2.1808097615085966,
      "grad_norm": 0.04910513013601303,
      "learning_rate": 9.086419753086419e-05,
      "loss": 0.7257,
      "step": 983
    },
    {
      "epoch": 2.1830282861896837,
      "grad_norm": 0.029208136722445488,
      "learning_rate": 9.061728395061727e-05,
      "loss": 0.5016,
      "step": 984
    },
    {
      "epoch": 2.1852468108707708,
      "grad_norm": 0.04599707946181297,
      "learning_rate": 9.037037037037038e-05,
      "loss": 0.5385,
      "step": 985
    },
    {
      "epoch": 2.187465335551858,
      "grad_norm": 0.029431896284222603,
      "learning_rate": 9.012345679012346e-05,
      "loss": 0.5463,
      "step": 986
    },
    {
      "epoch": 2.189683860232945,
      "grad_norm": 0.04243699088692665,
      "learning_rate": 8.987654320987654e-05,
      "loss": 0.3747,
      "step": 987
    },
    {
      "epoch": 2.191902384914032,
      "grad_norm": 0.05023906007409096,
      "learning_rate": 8.962962962962962e-05,
      "loss": 0.535,
      "step": 988
    },
    {
      "epoch": 2.194120909595119,
      "grad_norm": 0.036512281745672226,
      "learning_rate": 8.93827160493827e-05,
      "loss": 0.4797,
      "step": 989
    },
    {
      "epoch": 2.196339434276206,
      "grad_norm": 0.03909103572368622,
      "learning_rate": 8.91358024691358e-05,
      "loss": 0.5269,
      "step": 990
    },
    {
      "epoch": 2.1985579589572932,
      "grad_norm": 0.03427448496222496,
      "learning_rate": 8.888888888888888e-05,
      "loss": 0.5317,
      "step": 991
    },
    {
      "epoch": 2.2007764836383803,
      "grad_norm": 0.03474743291735649,
      "learning_rate": 8.864197530864196e-05,
      "loss": 0.4512,
      "step": 992
    },
    {
      "epoch": 2.2029950083194674,
      "grad_norm": 0.04623313993215561,
      "learning_rate": 8.839506172839506e-05,
      "loss": 0.9486,
      "step": 993
    },
    {
      "epoch": 2.2052135330005544,
      "grad_norm": 0.03590194508433342,
      "learning_rate": 8.814814814814815e-05,
      "loss": 0.4372,
      "step": 994
    },
    {
      "epoch": 2.2074320576816415,
      "grad_norm": 0.036686982959508896,
      "learning_rate": 8.790123456790123e-05,
      "loss": 0.4039,
      "step": 995
    },
    {
      "epoch": 2.2096505823627286,
      "grad_norm": 0.05053959786891937,
      "learning_rate": 8.765432098765431e-05,
      "loss": 0.6943,
      "step": 996
    },
    {
      "epoch": 2.2118691070438157,
      "grad_norm": 0.04072181135416031,
      "learning_rate": 8.74074074074074e-05,
      "loss": 0.4963,
      "step": 997
    },
    {
      "epoch": 2.2140876317249027,
      "grad_norm": 0.034398168325424194,
      "learning_rate": 8.716049382716048e-05,
      "loss": 0.6391,
      "step": 998
    },
    {
      "epoch": 2.21630615640599,
      "grad_norm": 0.048181578516960144,
      "learning_rate": 8.691358024691357e-05,
      "loss": 0.4571,
      "step": 999
    },
    {
      "epoch": 2.218524681087077,
      "grad_norm": 0.045372527092695236,
      "learning_rate": 8.666666666666665e-05,
      "loss": 0.5471,
      "step": 1000
    },
    {
      "epoch": 2.220743205768164,
      "grad_norm": 0.048364363610744476,
      "learning_rate": 8.641975308641975e-05,
      "loss": 0.6015,
      "step": 1001
    },
    {
      "epoch": 2.222961730449251,
      "grad_norm": 0.03488036245107651,
      "learning_rate": 8.617283950617283e-05,
      "loss": 0.5931,
      "step": 1002
    },
    {
      "epoch": 2.225180255130338,
      "grad_norm": 0.03259008750319481,
      "learning_rate": 8.592592592592593e-05,
      "loss": 0.5002,
      "step": 1003
    },
    {
      "epoch": 2.227398779811425,
      "grad_norm": 0.03562018647789955,
      "learning_rate": 8.567901234567901e-05,
      "loss": 0.5042,
      "step": 1004
    },
    {
      "epoch": 2.2296173044925123,
      "grad_norm": 0.044877056032419205,
      "learning_rate": 8.543209876543209e-05,
      "loss": 0.4835,
      "step": 1005
    },
    {
      "epoch": 2.2318358291735994,
      "grad_norm": 0.03873593360185623,
      "learning_rate": 8.518518518518517e-05,
      "loss": 0.5849,
      "step": 1006
    },
    {
      "epoch": 2.2340543538546864,
      "grad_norm": 0.04477420449256897,
      "learning_rate": 8.493827160493825e-05,
      "loss": 0.4843,
      "step": 1007
    },
    {
      "epoch": 2.2362728785357735,
      "grad_norm": 0.06584437191486359,
      "learning_rate": 8.469135802469136e-05,
      "loss": 0.6388,
      "step": 1008
    },
    {
      "epoch": 2.2384914032168606,
      "grad_norm": 0.034934524446725845,
      "learning_rate": 8.444444444444444e-05,
      "loss": 0.5825,
      "step": 1009
    },
    {
      "epoch": 2.2407099278979477,
      "grad_norm": 0.034885428845882416,
      "learning_rate": 8.419753086419752e-05,
      "loss": 0.4441,
      "step": 1010
    },
    {
      "epoch": 2.2429284525790347,
      "grad_norm": 0.0357072651386261,
      "learning_rate": 8.39506172839506e-05,
      "loss": 0.5164,
      "step": 1011
    },
    {
      "epoch": 2.245146977260122,
      "grad_norm": 0.033950936049222946,
      "learning_rate": 8.37037037037037e-05,
      "loss": 0.4651,
      "step": 1012
    },
    {
      "epoch": 2.247365501941209,
      "grad_norm": 0.04866531118750572,
      "learning_rate": 8.345679012345678e-05,
      "loss": 0.5609,
      "step": 1013
    },
    {
      "epoch": 2.249584026622296,
      "grad_norm": 0.04033484682440758,
      "learning_rate": 8.320987654320986e-05,
      "loss": 0.6572,
      "step": 1014
    },
    {
      "epoch": 2.251802551303383,
      "grad_norm": 0.034960586577653885,
      "learning_rate": 8.296296296296295e-05,
      "loss": 0.6525,
      "step": 1015
    },
    {
      "epoch": 2.25402107598447,
      "grad_norm": 0.032728198915719986,
      "learning_rate": 8.271604938271605e-05,
      "loss": 0.61,
      "step": 1016
    },
    {
      "epoch": 2.256239600665557,
      "grad_norm": 0.038337185978889465,
      "learning_rate": 8.246913580246914e-05,
      "loss": 0.7172,
      "step": 1017
    },
    {
      "epoch": 2.2584581253466443,
      "grad_norm": 0.05141346529126167,
      "learning_rate": 8.222222222222222e-05,
      "loss": 0.7304,
      "step": 1018
    },
    {
      "epoch": 2.2606766500277313,
      "grad_norm": 0.03697873279452324,
      "learning_rate": 8.19753086419753e-05,
      "loss": 0.6693,
      "step": 1019
    },
    {
      "epoch": 2.2628951747088184,
      "grad_norm": 0.050354860723018646,
      "learning_rate": 8.172839506172838e-05,
      "loss": 0.6583,
      "step": 1020
    },
    {
      "epoch": 2.2651136993899055,
      "grad_norm": 0.045183874666690826,
      "learning_rate": 8.148148148148148e-05,
      "loss": 0.6456,
      "step": 1021
    },
    {
      "epoch": 2.2673322240709926,
      "grad_norm": 0.05711262300610542,
      "learning_rate": 8.123456790123456e-05,
      "loss": 0.5281,
      "step": 1022
    },
    {
      "epoch": 2.2695507487520796,
      "grad_norm": 0.037834323942661285,
      "learning_rate": 8.098765432098765e-05,
      "loss": 0.6225,
      "step": 1023
    },
    {
      "epoch": 2.2717692734331667,
      "grad_norm": 0.04976195842027664,
      "learning_rate": 8.074074074074073e-05,
      "loss": 0.654,
      "step": 1024
    },
    {
      "epoch": 2.2739877981142538,
      "grad_norm": 0.02972816862165928,
      "learning_rate": 8.049382716049383e-05,
      "loss": 0.5934,
      "step": 1025
    },
    {
      "epoch": 2.276206322795341,
      "grad_norm": 0.05596752092242241,
      "learning_rate": 8.024691358024691e-05,
      "loss": 0.5753,
      "step": 1026
    },
    {
      "epoch": 2.278424847476428,
      "grad_norm": 0.0589897558093071,
      "learning_rate": 7.999999999999999e-05,
      "loss": 0.5368,
      "step": 1027
    },
    {
      "epoch": 2.280643372157515,
      "grad_norm": 0.04768100380897522,
      "learning_rate": 7.975308641975307e-05,
      "loss": 0.4786,
      "step": 1028
    },
    {
      "epoch": 2.2828618968386025,
      "grad_norm": 0.03086143359541893,
      "learning_rate": 7.950617283950616e-05,
      "loss": 0.5503,
      "step": 1029
    },
    {
      "epoch": 2.2850804215196896,
      "grad_norm": 0.047776252031326294,
      "learning_rate": 7.925925925925925e-05,
      "loss": 0.6438,
      "step": 1030
    },
    {
      "epoch": 2.2872989462007767,
      "grad_norm": 0.049423348158597946,
      "learning_rate": 7.901234567901235e-05,
      "loss": 0.6942,
      "step": 1031
    },
    {
      "epoch": 2.2895174708818637,
      "grad_norm": 0.031016642227768898,
      "learning_rate": 7.876543209876543e-05,
      "loss": 0.5977,
      "step": 1032
    },
    {
      "epoch": 2.291735995562951,
      "grad_norm": 0.04408005252480507,
      "learning_rate": 7.851851851851851e-05,
      "loss": 0.475,
      "step": 1033
    },
    {
      "epoch": 2.293954520244038,
      "grad_norm": 0.0563134029507637,
      "learning_rate": 7.82716049382716e-05,
      "loss": 0.6261,
      "step": 1034
    },
    {
      "epoch": 2.296173044925125,
      "grad_norm": 0.028872035443782806,
      "learning_rate": 7.802469135802469e-05,
      "loss": 0.4325,
      "step": 1035
    },
    {
      "epoch": 2.298391569606212,
      "grad_norm": 0.03898025304079056,
      "learning_rate": 7.777777777777777e-05,
      "loss": 0.5559,
      "step": 1036
    },
    {
      "epoch": 2.300610094287299,
      "grad_norm": 0.035244155675172806,
      "learning_rate": 7.753086419753085e-05,
      "loss": 0.4557,
      "step": 1037
    },
    {
      "epoch": 2.302828618968386,
      "grad_norm": 0.027673883363604546,
      "learning_rate": 7.728395061728396e-05,
      "loss": 0.623,
      "step": 1038
    },
    {
      "epoch": 2.3050471436494733,
      "grad_norm": 0.04300675913691521,
      "learning_rate": 7.703703703703704e-05,
      "loss": 0.5802,
      "step": 1039
    },
    {
      "epoch": 2.3072656683305603,
      "grad_norm": 0.030332626774907112,
      "learning_rate": 7.679012345679012e-05,
      "loss": 0.5501,
      "step": 1040
    },
    {
      "epoch": 2.3094841930116474,
      "grad_norm": 0.04027707502245903,
      "learning_rate": 7.65432098765432e-05,
      "loss": 0.6418,
      "step": 1041
    },
    {
      "epoch": 2.3117027176927345,
      "grad_norm": 0.053012680262327194,
      "learning_rate": 7.629629629629628e-05,
      "loss": 0.5119,
      "step": 1042
    },
    {
      "epoch": 2.3139212423738216,
      "grad_norm": 0.055856458842754364,
      "learning_rate": 7.604938271604937e-05,
      "loss": 0.5898,
      "step": 1043
    },
    {
      "epoch": 2.3161397670549086,
      "grad_norm": 0.03730011731386185,
      "learning_rate": 7.580246913580246e-05,
      "loss": 0.547,
      "step": 1044
    },
    {
      "epoch": 2.3183582917359957,
      "grad_norm": 0.06076604872941971,
      "learning_rate": 7.555555555555554e-05,
      "loss": 0.6016,
      "step": 1045
    },
    {
      "epoch": 2.320576816417083,
      "grad_norm": 0.03502635657787323,
      "learning_rate": 7.530864197530864e-05,
      "loss": 0.6611,
      "step": 1046
    },
    {
      "epoch": 2.32279534109817,
      "grad_norm": 0.03913009166717529,
      "learning_rate": 7.506172839506173e-05,
      "loss": 0.4809,
      "step": 1047
    },
    {
      "epoch": 2.325013865779257,
      "grad_norm": 0.04104864224791527,
      "learning_rate": 7.481481481481481e-05,
      "loss": 0.5208,
      "step": 1048
    },
    {
      "epoch": 2.327232390460344,
      "grad_norm": 0.06915964931249619,
      "learning_rate": 7.45679012345679e-05,
      "loss": 0.6388,
      "step": 1049
    },
    {
      "epoch": 2.329450915141431,
      "grad_norm": 0.05258198082447052,
      "learning_rate": 7.432098765432098e-05,
      "loss": 0.6024,
      "step": 1050
    },
    {
      "epoch": 2.331669439822518,
      "grad_norm": 0.03615143895149231,
      "learning_rate": 7.407407407407406e-05,
      "loss": 0.5823,
      "step": 1051
    },
    {
      "epoch": 2.3338879645036053,
      "grad_norm": 0.03468695655465126,
      "learning_rate": 7.382716049382715e-05,
      "loss": 0.4891,
      "step": 1052
    },
    {
      "epoch": 2.3361064891846923,
      "grad_norm": 0.11859353631734848,
      "learning_rate": 7.358024691358024e-05,
      "loss": 0.6896,
      "step": 1053
    },
    {
      "epoch": 2.3383250138657794,
      "grad_norm": 0.04108664393424988,
      "learning_rate": 7.333333333333332e-05,
      "loss": 0.6401,
      "step": 1054
    },
    {
      "epoch": 2.3405435385468665,
      "grad_norm": 0.04725251719355583,
      "learning_rate": 7.308641975308641e-05,
      "loss": 0.6168,
      "step": 1055
    },
    {
      "epoch": 2.3427620632279536,
      "grad_norm": 0.04292666167020798,
      "learning_rate": 7.28395061728395e-05,
      "loss": 0.6383,
      "step": 1056
    },
    {
      "epoch": 2.3449805879090406,
      "grad_norm": 0.04848600551486015,
      "learning_rate": 7.259259259259259e-05,
      "loss": 0.5481,
      "step": 1057
    },
    {
      "epoch": 2.3471991125901277,
      "grad_norm": 0.043581489473581314,
      "learning_rate": 7.234567901234567e-05,
      "loss": 0.572,
      "step": 1058
    },
    {
      "epoch": 2.3494176372712148,
      "grad_norm": 0.03410588204860687,
      "learning_rate": 7.209876543209877e-05,
      "loss": 0.5713,
      "step": 1059
    },
    {
      "epoch": 2.351636161952302,
      "grad_norm": 0.04346185922622681,
      "learning_rate": 7.185185185185185e-05,
      "loss": 0.4952,
      "step": 1060
    },
    {
      "epoch": 2.353854686633389,
      "grad_norm": 0.04467775672674179,
      "learning_rate": 7.160493827160493e-05,
      "loss": 0.5052,
      "step": 1061
    },
    {
      "epoch": 2.356073211314476,
      "grad_norm": 0.029043078422546387,
      "learning_rate": 7.135802469135801e-05,
      "loss": 0.5508,
      "step": 1062
    },
    {
      "epoch": 2.358291735995563,
      "grad_norm": 0.03268744796514511,
      "learning_rate": 7.11111111111111e-05,
      "loss": 0.4578,
      "step": 1063
    },
    {
      "epoch": 2.36051026067665,
      "grad_norm": 0.04449523985385895,
      "learning_rate": 7.086419753086419e-05,
      "loss": 0.6145,
      "step": 1064
    },
    {
      "epoch": 2.3627287853577372,
      "grad_norm": 0.05389757081866264,
      "learning_rate": 7.061728395061727e-05,
      "loss": 0.4872,
      "step": 1065
    },
    {
      "epoch": 2.3649473100388243,
      "grad_norm": 0.027174999937415123,
      "learning_rate": 7.037037037037036e-05,
      "loss": 0.503,
      "step": 1066
    },
    {
      "epoch": 2.3671658347199114,
      "grad_norm": 0.058143775910139084,
      "learning_rate": 7.012345679012345e-05,
      "loss": 0.6817,
      "step": 1067
    },
    {
      "epoch": 2.3693843594009985,
      "grad_norm": 0.04235734045505524,
      "learning_rate": 6.987654320987654e-05,
      "loss": 0.7351,
      "step": 1068
    },
    {
      "epoch": 2.3716028840820855,
      "grad_norm": 0.028445202857255936,
      "learning_rate": 6.962962962962962e-05,
      "loss": 0.4253,
      "step": 1069
    },
    {
      "epoch": 2.3738214087631726,
      "grad_norm": 0.049056969583034515,
      "learning_rate": 6.938271604938272e-05,
      "loss": 0.5024,
      "step": 1070
    },
    {
      "epoch": 2.3760399334442597,
      "grad_norm": 0.0438736192882061,
      "learning_rate": 6.91358024691358e-05,
      "loss": 0.6601,
      "step": 1071
    },
    {
      "epoch": 2.3782584581253468,
      "grad_norm": 0.03774663433432579,
      "learning_rate": 6.888888888888888e-05,
      "loss": 0.627,
      "step": 1072
    },
    {
      "epoch": 2.380476982806434,
      "grad_norm": 0.03760320320725441,
      "learning_rate": 6.864197530864196e-05,
      "loss": 0.6227,
      "step": 1073
    },
    {
      "epoch": 2.382695507487521,
      "grad_norm": 0.040387723594903946,
      "learning_rate": 6.839506172839506e-05,
      "loss": 0.6069,
      "step": 1074
    },
    {
      "epoch": 2.384914032168608,
      "grad_norm": 0.05957155302166939,
      "learning_rate": 6.814814814814814e-05,
      "loss": 0.4694,
      "step": 1075
    },
    {
      "epoch": 2.387132556849695,
      "grad_norm": 0.04312063381075859,
      "learning_rate": 6.790123456790122e-05,
      "loss": 0.4833,
      "step": 1076
    },
    {
      "epoch": 2.389351081530782,
      "grad_norm": 0.03880666196346283,
      "learning_rate": 6.765432098765432e-05,
      "loss": 0.7542,
      "step": 1077
    },
    {
      "epoch": 2.391569606211869,
      "grad_norm": 0.08743049204349518,
      "learning_rate": 6.74074074074074e-05,
      "loss": 0.8163,
      "step": 1078
    },
    {
      "epoch": 2.3937881308929563,
      "grad_norm": 0.04956267401576042,
      "learning_rate": 6.716049382716049e-05,
      "loss": 0.6242,
      "step": 1079
    },
    {
      "epoch": 2.3960066555740434,
      "grad_norm": 0.03280191868543625,
      "learning_rate": 6.691358024691357e-05,
      "loss": 0.5347,
      "step": 1080
    },
    {
      "epoch": 2.3982251802551304,
      "grad_norm": 0.04548501595854759,
      "learning_rate": 6.666666666666666e-05,
      "loss": 0.5618,
      "step": 1081
    },
    {
      "epoch": 2.4004437049362175,
      "grad_norm": 0.04287225380539894,
      "learning_rate": 6.641975308641975e-05,
      "loss": 0.5083,
      "step": 1082
    },
    {
      "epoch": 2.4026622296173046,
      "grad_norm": 0.06229160353541374,
      "learning_rate": 6.617283950617283e-05,
      "loss": 0.4835,
      "step": 1083
    },
    {
      "epoch": 2.4048807542983917,
      "grad_norm": 0.03640734776854515,
      "learning_rate": 6.592592592592591e-05,
      "loss": 0.5271,
      "step": 1084
    },
    {
      "epoch": 2.4070992789794787,
      "grad_norm": 0.034687500447034836,
      "learning_rate": 6.567901234567901e-05,
      "loss": 0.4938,
      "step": 1085
    },
    {
      "epoch": 2.409317803660566,
      "grad_norm": 0.034138698130846024,
      "learning_rate": 6.543209876543209e-05,
      "loss": 0.5023,
      "step": 1086
    },
    {
      "epoch": 2.411536328341653,
      "grad_norm": 0.03097585029900074,
      "learning_rate": 6.518518518518517e-05,
      "loss": 0.4946,
      "step": 1087
    },
    {
      "epoch": 2.41375485302274,
      "grad_norm": 0.0352029986679554,
      "learning_rate": 6.493827160493827e-05,
      "loss": 0.4329,
      "step": 1088
    },
    {
      "epoch": 2.415973377703827,
      "grad_norm": 0.03877712041139603,
      "learning_rate": 6.469135802469135e-05,
      "loss": 0.451,
      "step": 1089
    },
    {
      "epoch": 2.418191902384914,
      "grad_norm": 0.03704957291483879,
      "learning_rate": 6.444444444444444e-05,
      "loss": 0.5767,
      "step": 1090
    },
    {
      "epoch": 2.420410427066001,
      "grad_norm": 0.029981613159179688,
      "learning_rate": 6.419753086419753e-05,
      "loss": 0.5614,
      "step": 1091
    },
    {
      "epoch": 2.4226289517470883,
      "grad_norm": 0.04133082926273346,
      "learning_rate": 6.395061728395061e-05,
      "loss": 0.452,
      "step": 1092
    },
    {
      "epoch": 2.4248474764281753,
      "grad_norm": 0.041626300662755966,
      "learning_rate": 6.37037037037037e-05,
      "loss": 0.5541,
      "step": 1093
    },
    {
      "epoch": 2.4270660011092624,
      "grad_norm": 0.04535186290740967,
      "learning_rate": 6.345679012345678e-05,
      "loss": 0.7066,
      "step": 1094
    },
    {
      "epoch": 2.4292845257903495,
      "grad_norm": 0.031195247545838356,
      "learning_rate": 6.320987654320987e-05,
      "loss": 0.4482,
      "step": 1095
    },
    {
      "epoch": 2.4315030504714366,
      "grad_norm": 0.035386569797992706,
      "learning_rate": 6.296296296296295e-05,
      "loss": 0.5623,
      "step": 1096
    },
    {
      "epoch": 2.4337215751525236,
      "grad_norm": 0.037585604935884476,
      "learning_rate": 6.271604938271604e-05,
      "loss": 0.5472,
      "step": 1097
    },
    {
      "epoch": 2.4359400998336107,
      "grad_norm": 0.046316951513290405,
      "learning_rate": 6.246913580246912e-05,
      "loss": 0.5966,
      "step": 1098
    },
    {
      "epoch": 2.438158624514698,
      "grad_norm": 0.030985720455646515,
      "learning_rate": 6.222222222222222e-05,
      "loss": 0.5107,
      "step": 1099
    },
    {
      "epoch": 2.440377149195785,
      "grad_norm": 0.030455226078629494,
      "learning_rate": 6.19753086419753e-05,
      "loss": 0.5047,
      "step": 1100
    },
    {
      "epoch": 2.442595673876872,
      "grad_norm": 0.03762245178222656,
      "learning_rate": 6.17283950617284e-05,
      "loss": 0.5545,
      "step": 1101
    },
    {
      "epoch": 2.444814198557959,
      "grad_norm": 0.03886816278100014,
      "learning_rate": 6.148148148148148e-05,
      "loss": 0.5211,
      "step": 1102
    },
    {
      "epoch": 2.447032723239046,
      "grad_norm": 0.03680301457643509,
      "learning_rate": 6.123456790123456e-05,
      "loss": 0.554,
      "step": 1103
    },
    {
      "epoch": 2.449251247920133,
      "grad_norm": 0.03106062114238739,
      "learning_rate": 6.0987654320987655e-05,
      "loss": 0.4282,
      "step": 1104
    },
    {
      "epoch": 2.4514697726012202,
      "grad_norm": 0.028792664408683777,
      "learning_rate": 6.0740740740740736e-05,
      "loss": 0.4822,
      "step": 1105
    },
    {
      "epoch": 2.4536882972823073,
      "grad_norm": 0.03883747011423111,
      "learning_rate": 6.049382716049382e-05,
      "loss": 0.5379,
      "step": 1106
    },
    {
      "epoch": 2.4559068219633944,
      "grad_norm": 0.05315295606851578,
      "learning_rate": 6.0246913580246906e-05,
      "loss": 0.5054,
      "step": 1107
    },
    {
      "epoch": 2.4581253466444815,
      "grad_norm": 0.02992473728954792,
      "learning_rate": 5.9999999999999995e-05,
      "loss": 0.4588,
      "step": 1108
    },
    {
      "epoch": 2.4603438713255685,
      "grad_norm": 0.042305413633584976,
      "learning_rate": 5.975308641975308e-05,
      "loss": 0.437,
      "step": 1109
    },
    {
      "epoch": 2.4625623960066556,
      "grad_norm": 0.03690609335899353,
      "learning_rate": 5.9506172839506165e-05,
      "loss": 0.5212,
      "step": 1110
    },
    {
      "epoch": 2.4647809206877427,
      "grad_norm": 0.03595404326915741,
      "learning_rate": 5.925925925925925e-05,
      "loss": 0.5126,
      "step": 1111
    },
    {
      "epoch": 2.4669994453688298,
      "grad_norm": 0.034664545208215714,
      "learning_rate": 5.901234567901234e-05,
      "loss": 0.4478,
      "step": 1112
    },
    {
      "epoch": 2.469217970049917,
      "grad_norm": 0.051099810749292374,
      "learning_rate": 5.876543209876543e-05,
      "loss": 0.5352,
      "step": 1113
    },
    {
      "epoch": 2.471436494731004,
      "grad_norm": 0.053003400564193726,
      "learning_rate": 5.851851851851851e-05,
      "loss": 0.5214,
      "step": 1114
    },
    {
      "epoch": 2.473655019412091,
      "grad_norm": 0.03135830909013748,
      "learning_rate": 5.827160493827159e-05,
      "loss": 0.4924,
      "step": 1115
    },
    {
      "epoch": 2.475873544093178,
      "grad_norm": 0.055121198296546936,
      "learning_rate": 5.802469135802469e-05,
      "loss": 0.7461,
      "step": 1116
    },
    {
      "epoch": 2.478092068774265,
      "grad_norm": 0.04110425338149071,
      "learning_rate": 5.777777777777777e-05,
      "loss": 0.5285,
      "step": 1117
    },
    {
      "epoch": 2.480310593455352,
      "grad_norm": 0.03775554522871971,
      "learning_rate": 5.753086419753086e-05,
      "loss": 0.5189,
      "step": 1118
    },
    {
      "epoch": 2.4825291181364393,
      "grad_norm": 0.038824744522571564,
      "learning_rate": 5.7283950617283946e-05,
      "loss": 0.5546,
      "step": 1119
    },
    {
      "epoch": 2.4847476428175264,
      "grad_norm": 0.03548000380396843,
      "learning_rate": 5.7037037037037035e-05,
      "loss": 0.5579,
      "step": 1120
    },
    {
      "epoch": 2.4869661674986134,
      "grad_norm": 0.043440185487270355,
      "learning_rate": 5.6790123456790116e-05,
      "loss": 0.5139,
      "step": 1121
    },
    {
      "epoch": 2.4891846921797005,
      "grad_norm": 0.03753809258341789,
      "learning_rate": 5.6543209876543205e-05,
      "loss": 0.6366,
      "step": 1122
    },
    {
      "epoch": 2.4914032168607876,
      "grad_norm": 0.043515726923942566,
      "learning_rate": 5.629629629629629e-05,
      "loss": 0.5962,
      "step": 1123
    },
    {
      "epoch": 2.4936217415418747,
      "grad_norm": 0.042143840342760086,
      "learning_rate": 5.604938271604938e-05,
      "loss": 0.6313,
      "step": 1124
    },
    {
      "epoch": 2.4958402662229617,
      "grad_norm": 0.04426976665854454,
      "learning_rate": 5.580246913580246e-05,
      "loss": 0.6124,
      "step": 1125
    },
    {
      "epoch": 2.498058790904049,
      "grad_norm": 0.03353346884250641,
      "learning_rate": 5.5555555555555545e-05,
      "loss": 0.5632,
      "step": 1126
    },
    {
      "epoch": 2.500277315585136,
      "grad_norm": 0.04805271327495575,
      "learning_rate": 5.530864197530864e-05,
      "loss": 0.4622,
      "step": 1127
    },
    {
      "epoch": 2.502495840266223,
      "grad_norm": 0.0457775704562664,
      "learning_rate": 5.506172839506172e-05,
      "loss": 0.5526,
      "step": 1128
    },
    {
      "epoch": 2.50471436494731,
      "grad_norm": 0.04630595073103905,
      "learning_rate": 5.481481481481481e-05,
      "loss": 0.5918,
      "step": 1129
    },
    {
      "epoch": 2.506932889628397,
      "grad_norm": 0.05171075835824013,
      "learning_rate": 5.456790123456789e-05,
      "loss": 0.4981,
      "step": 1130
    },
    {
      "epoch": 2.509151414309484,
      "grad_norm": 0.05739015340805054,
      "learning_rate": 5.4320987654320986e-05,
      "loss": 0.5526,
      "step": 1131
    },
    {
      "epoch": 2.5113699389905713,
      "grad_norm": 0.04030559957027435,
      "learning_rate": 5.407407407407407e-05,
      "loss": 0.6513,
      "step": 1132
    },
    {
      "epoch": 2.5135884636716583,
      "grad_norm": 0.033779703080654144,
      "learning_rate": 5.3827160493827156e-05,
      "loss": 0.4821,
      "step": 1133
    },
    {
      "epoch": 2.5158069883527454,
      "grad_norm": 0.034196723252534866,
      "learning_rate": 5.358024691358024e-05,
      "loss": 0.645,
      "step": 1134
    },
    {
      "epoch": 2.5180255130338325,
      "grad_norm": 0.04460705444216728,
      "learning_rate": 5.333333333333333e-05,
      "loss": 0.4124,
      "step": 1135
    },
    {
      "epoch": 2.5202440377149196,
      "grad_norm": 0.05757799744606018,
      "learning_rate": 5.3086419753086415e-05,
      "loss": 0.6566,
      "step": 1136
    },
    {
      "epoch": 2.5224625623960066,
      "grad_norm": 0.043432097882032394,
      "learning_rate": 5.2839506172839496e-05,
      "loss": 0.5655,
      "step": 1137
    },
    {
      "epoch": 2.5246810870770937,
      "grad_norm": 0.04850088432431221,
      "learning_rate": 5.259259259259259e-05,
      "loss": 0.4939,
      "step": 1138
    },
    {
      "epoch": 2.526899611758181,
      "grad_norm": 0.06233000382781029,
      "learning_rate": 5.234567901234567e-05,
      "loss": 0.7906,
      "step": 1139
    },
    {
      "epoch": 2.529118136439268,
      "grad_norm": 0.036589570343494415,
      "learning_rate": 5.209876543209876e-05,
      "loss": 0.5209,
      "step": 1140
    },
    {
      "epoch": 2.531336661120355,
      "grad_norm": 0.047147925943136215,
      "learning_rate": 5.185185185185184e-05,
      "loss": 0.4743,
      "step": 1141
    },
    {
      "epoch": 2.533555185801442,
      "grad_norm": 0.035007718950510025,
      "learning_rate": 5.160493827160494e-05,
      "loss": 0.6089,
      "step": 1142
    },
    {
      "epoch": 2.535773710482529,
      "grad_norm": 0.07631321251392365,
      "learning_rate": 5.135802469135802e-05,
      "loss": 0.5873,
      "step": 1143
    },
    {
      "epoch": 2.537992235163616,
      "grad_norm": 0.0632716566324234,
      "learning_rate": 5.111111111111111e-05,
      "loss": 0.9179,
      "step": 1144
    },
    {
      "epoch": 2.5402107598447032,
      "grad_norm": 0.03311275318264961,
      "learning_rate": 5.086419753086419e-05,
      "loss": 0.4848,
      "step": 1145
    },
    {
      "epoch": 2.5424292845257903,
      "grad_norm": 0.041001636534929276,
      "learning_rate": 5.0617283950617285e-05,
      "loss": 0.5409,
      "step": 1146
    },
    {
      "epoch": 2.5446478092068774,
      "grad_norm": 0.05013132840394974,
      "learning_rate": 5.0370370370370366e-05,
      "loss": 0.7448,
      "step": 1147
    },
    {
      "epoch": 2.5468663338879645,
      "grad_norm": 0.05755642428994179,
      "learning_rate": 5.012345679012345e-05,
      "loss": 0.6266,
      "step": 1148
    },
    {
      "epoch": 2.5490848585690515,
      "grad_norm": 0.04079374298453331,
      "learning_rate": 4.9876543209876536e-05,
      "loss": 0.5769,
      "step": 1149
    },
    {
      "epoch": 2.5513033832501386,
      "grad_norm": 0.056092873215675354,
      "learning_rate": 4.9629629629629625e-05,
      "loss": 0.5703,
      "step": 1150
    },
    {
      "epoch": 2.5535219079312257,
      "grad_norm": 0.04066433757543564,
      "learning_rate": 4.938271604938271e-05,
      "loss": 0.576,
      "step": 1151
    },
    {
      "epoch": 2.5557404326123128,
      "grad_norm": 0.03990304470062256,
      "learning_rate": 4.9135802469135795e-05,
      "loss": 0.5985,
      "step": 1152
    },
    {
      "epoch": 2.5579589572934,
      "grad_norm": 0.03486698120832443,
      "learning_rate": 4.888888888888889e-05,
      "loss": 0.48,
      "step": 1153
    },
    {
      "epoch": 2.560177481974487,
      "grad_norm": 0.03874533623456955,
      "learning_rate": 4.864197530864197e-05,
      "loss": 0.5586,
      "step": 1154
    },
    {
      "epoch": 2.562396006655574,
      "grad_norm": 0.039097052067518234,
      "learning_rate": 4.839506172839506e-05,
      "loss": 0.6205,
      "step": 1155
    },
    {
      "epoch": 2.564614531336661,
      "grad_norm": 0.038373593240976334,
      "learning_rate": 4.814814814814814e-05,
      "loss": 0.5313,
      "step": 1156
    },
    {
      "epoch": 2.566833056017748,
      "grad_norm": 0.0414130799472332,
      "learning_rate": 4.7901234567901237e-05,
      "loss": 0.4286,
      "step": 1157
    },
    {
      "epoch": 2.569051580698835,
      "grad_norm": 0.038319267332553864,
      "learning_rate": 4.765432098765432e-05,
      "loss": 0.6032,
      "step": 1158
    },
    {
      "epoch": 2.5712701053799223,
      "grad_norm": 0.03815809264779091,
      "learning_rate": 4.74074074074074e-05,
      "loss": 0.5636,
      "step": 1159
    },
    {
      "epoch": 2.5734886300610094,
      "grad_norm": 0.046863071620464325,
      "learning_rate": 4.716049382716049e-05,
      "loss": 0.5796,
      "step": 1160
    },
    {
      "epoch": 2.5757071547420964,
      "grad_norm": 0.0393722839653492,
      "learning_rate": 4.6913580246913576e-05,
      "loss": 0.569,
      "step": 1161
    },
    {
      "epoch": 2.5779256794231835,
      "grad_norm": 0.058395564556121826,
      "learning_rate": 4.6666666666666665e-05,
      "loss": 0.542,
      "step": 1162
    },
    {
      "epoch": 2.5801442041042706,
      "grad_norm": 0.03891870379447937,
      "learning_rate": 4.6419753086419746e-05,
      "loss": 0.6801,
      "step": 1163
    },
    {
      "epoch": 2.5823627287853577,
      "grad_norm": 0.038542989641427994,
      "learning_rate": 4.6172839506172835e-05,
      "loss": 0.5931,
      "step": 1164
    },
    {
      "epoch": 2.5845812534664447,
      "grad_norm": 0.03438922017812729,
      "learning_rate": 4.592592592592592e-05,
      "loss": 0.4305,
      "step": 1165
    },
    {
      "epoch": 2.586799778147532,
      "grad_norm": 0.030714431777596474,
      "learning_rate": 4.567901234567901e-05,
      "loss": 0.4047,
      "step": 1166
    },
    {
      "epoch": 2.589018302828619,
      "grad_norm": 0.06754889339208603,
      "learning_rate": 4.543209876543209e-05,
      "loss": 0.6723,
      "step": 1167
    },
    {
      "epoch": 2.591236827509706,
      "grad_norm": 0.04761623963713646,
      "learning_rate": 4.518518518518519e-05,
      "loss": 0.5861,
      "step": 1168
    },
    {
      "epoch": 2.593455352190793,
      "grad_norm": 0.0397719107568264,
      "learning_rate": 4.493827160493827e-05,
      "loss": 0.4886,
      "step": 1169
    },
    {
      "epoch": 2.59567387687188,
      "grad_norm": 0.03461003676056862,
      "learning_rate": 4.469135802469135e-05,
      "loss": 0.5463,
      "step": 1170
    },
    {
      "epoch": 2.597892401552967,
      "grad_norm": 0.052945639938116074,
      "learning_rate": 4.444444444444444e-05,
      "loss": 0.5637,
      "step": 1171
    },
    {
      "epoch": 2.6001109262340543,
      "grad_norm": 0.03932298347353935,
      "learning_rate": 4.419753086419753e-05,
      "loss": 0.6147,
      "step": 1172
    },
    {
      "epoch": 2.6023294509151413,
      "grad_norm": 0.04640820994973183,
      "learning_rate": 4.3950617283950617e-05,
      "loss": 0.5989,
      "step": 1173
    },
    {
      "epoch": 2.6045479755962284,
      "grad_norm": 0.029170742258429527,
      "learning_rate": 4.37037037037037e-05,
      "loss": 0.4385,
      "step": 1174
    },
    {
      "epoch": 2.6067665002773155,
      "grad_norm": 0.05705037713050842,
      "learning_rate": 4.3456790123456786e-05,
      "loss": 0.5423,
      "step": 1175
    },
    {
      "epoch": 2.6089850249584026,
      "grad_norm": 0.02691579796373844,
      "learning_rate": 4.3209876543209875e-05,
      "loss": 0.4904,
      "step": 1176
    },
    {
      "epoch": 2.6112035496394896,
      "grad_norm": 0.043135866522789,
      "learning_rate": 4.296296296296296e-05,
      "loss": 0.5772,
      "step": 1177
    },
    {
      "epoch": 2.6134220743205767,
      "grad_norm": 0.037349145859479904,
      "learning_rate": 4.2716049382716045e-05,
      "loss": 0.6537,
      "step": 1178
    },
    {
      "epoch": 2.615640599001664,
      "grad_norm": 0.041416384279727936,
      "learning_rate": 4.2469135802469126e-05,
      "loss": 0.6434,
      "step": 1179
    },
    {
      "epoch": 2.617859123682751,
      "grad_norm": 0.034590769559144974,
      "learning_rate": 4.222222222222222e-05,
      "loss": 0.5598,
      "step": 1180
    },
    {
      "epoch": 2.620077648363838,
      "grad_norm": 0.03367970511317253,
      "learning_rate": 4.19753086419753e-05,
      "loss": 0.5317,
      "step": 1181
    },
    {
      "epoch": 2.622296173044925,
      "grad_norm": 0.03658882528543472,
      "learning_rate": 4.172839506172839e-05,
      "loss": 0.5939,
      "step": 1182
    },
    {
      "epoch": 2.624514697726012,
      "grad_norm": 0.05713294818997383,
      "learning_rate": 4.148148148148147e-05,
      "loss": 0.3915,
      "step": 1183
    },
    {
      "epoch": 2.626733222407099,
      "grad_norm": 0.04519135504961014,
      "learning_rate": 4.123456790123457e-05,
      "loss": 0.5955,
      "step": 1184
    },
    {
      "epoch": 2.6289517470881862,
      "grad_norm": 0.040452953428030014,
      "learning_rate": 4.098765432098765e-05,
      "loss": 0.6471,
      "step": 1185
    },
    {
      "epoch": 2.6311702717692733,
      "grad_norm": 0.03603832796216011,
      "learning_rate": 4.074074074074074e-05,
      "loss": 0.4489,
      "step": 1186
    },
    {
      "epoch": 2.6333887964503604,
      "grad_norm": 0.042187247425317764,
      "learning_rate": 4.0493827160493827e-05,
      "loss": 0.7088,
      "step": 1187
    },
    {
      "epoch": 2.6356073211314475,
      "grad_norm": 0.03868582472205162,
      "learning_rate": 4.0246913580246915e-05,
      "loss": 0.4084,
      "step": 1188
    },
    {
      "epoch": 2.6378258458125345,
      "grad_norm": 0.04304947331547737,
      "learning_rate": 3.9999999999999996e-05,
      "loss": 0.5073,
      "step": 1189
    },
    {
      "epoch": 2.6400443704936216,
      "grad_norm": 0.027072343975305557,
      "learning_rate": 3.975308641975308e-05,
      "loss": 0.4715,
      "step": 1190
    },
    {
      "epoch": 2.6422628951747087,
      "grad_norm": 0.0381648950278759,
      "learning_rate": 3.950617283950617e-05,
      "loss": 0.4893,
      "step": 1191
    },
    {
      "epoch": 2.6444814198557958,
      "grad_norm": 0.029268911108374596,
      "learning_rate": 3.9259259259259255e-05,
      "loss": 0.5604,
      "step": 1192
    },
    {
      "epoch": 2.646699944536883,
      "grad_norm": 0.03340864181518555,
      "learning_rate": 3.901234567901234e-05,
      "loss": 0.6684,
      "step": 1193
    },
    {
      "epoch": 2.64891846921797,
      "grad_norm": 0.05228373408317566,
      "learning_rate": 3.8765432098765425e-05,
      "loss": 0.5795,
      "step": 1194
    },
    {
      "epoch": 2.651136993899057,
      "grad_norm": 0.04547541216015816,
      "learning_rate": 3.851851851851852e-05,
      "loss": 0.4949,
      "step": 1195
    },
    {
      "epoch": 2.653355518580144,
      "grad_norm": 0.05488264933228493,
      "learning_rate": 3.82716049382716e-05,
      "loss": 0.4614,
      "step": 1196
    },
    {
      "epoch": 2.655574043261231,
      "grad_norm": 0.03899548947811127,
      "learning_rate": 3.802469135802468e-05,
      "loss": 0.4785,
      "step": 1197
    },
    {
      "epoch": 2.657792567942318,
      "grad_norm": 0.033592723309993744,
      "learning_rate": 3.777777777777777e-05,
      "loss": 0.5079,
      "step": 1198
    },
    {
      "epoch": 2.6600110926234053,
      "grad_norm": 0.040112171322107315,
      "learning_rate": 3.7530864197530867e-05,
      "loss": 0.5791,
      "step": 1199
    },
    {
      "epoch": 2.6622296173044924,
      "grad_norm": 0.044878143817186356,
      "learning_rate": 3.728395061728395e-05,
      "loss": 0.66,
      "step": 1200
    },
    {
      "epoch": 2.6644481419855794,
      "grad_norm": 0.03919333219528198,
      "learning_rate": 3.703703703703703e-05,
      "loss": 0.5341,
      "step": 1201
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 0.03717885911464691,
      "learning_rate": 3.679012345679012e-05,
      "loss": 0.6037,
      "step": 1202
    },
    {
      "epoch": 2.6688851913477536,
      "grad_norm": 0.03957350179553032,
      "learning_rate": 3.6543209876543207e-05,
      "loss": 0.7132,
      "step": 1203
    },
    {
      "epoch": 2.6711037160288407,
      "grad_norm": 0.08652543276548386,
      "learning_rate": 3.6296296296296295e-05,
      "loss": 0.7888,
      "step": 1204
    },
    {
      "epoch": 2.6733222407099277,
      "grad_norm": 0.03165268525481224,
      "learning_rate": 3.604938271604938e-05,
      "loss": 0.5051,
      "step": 1205
    },
    {
      "epoch": 2.675540765391015,
      "grad_norm": 0.04151745140552521,
      "learning_rate": 3.5802469135802465e-05,
      "loss": 0.594,
      "step": 1206
    },
    {
      "epoch": 2.677759290072102,
      "grad_norm": 0.04961137846112251,
      "learning_rate": 3.555555555555555e-05,
      "loss": 0.5146,
      "step": 1207
    },
    {
      "epoch": 2.679977814753189,
      "grad_norm": 0.04734780266880989,
      "learning_rate": 3.5308641975308635e-05,
      "loss": 0.6483,
      "step": 1208
    },
    {
      "epoch": 2.682196339434276,
      "grad_norm": 0.05196371674537659,
      "learning_rate": 3.506172839506172e-05,
      "loss": 0.5253,
      "step": 1209
    },
    {
      "epoch": 2.684414864115363,
      "grad_norm": 0.04378390312194824,
      "learning_rate": 3.481481481481481e-05,
      "loss": 0.5246,
      "step": 1210
    },
    {
      "epoch": 2.68663338879645,
      "grad_norm": 0.04174394905567169,
      "learning_rate": 3.45679012345679e-05,
      "loss": 0.5522,
      "step": 1211
    },
    {
      "epoch": 2.6888519134775377,
      "grad_norm": 0.03972980007529259,
      "learning_rate": 3.432098765432098e-05,
      "loss": 0.6415,
      "step": 1212
    },
    {
      "epoch": 2.691070438158625,
      "grad_norm": 0.036516617983579636,
      "learning_rate": 3.407407407407407e-05,
      "loss": 0.5385,
      "step": 1213
    },
    {
      "epoch": 2.693288962839712,
      "grad_norm": 0.03818022832274437,
      "learning_rate": 3.382716049382716e-05,
      "loss": 0.6238,
      "step": 1214
    },
    {
      "epoch": 2.695507487520799,
      "grad_norm": 0.045793261379003525,
      "learning_rate": 3.3580246913580247e-05,
      "loss": 0.5282,
      "step": 1215
    },
    {
      "epoch": 2.697726012201886,
      "grad_norm": 0.048041049391031265,
      "learning_rate": 3.333333333333333e-05,
      "loss": 0.6338,
      "step": 1216
    },
    {
      "epoch": 2.699944536882973,
      "grad_norm": 0.03621658310294151,
      "learning_rate": 3.3086419753086417e-05,
      "loss": 0.5706,
      "step": 1217
    },
    {
      "epoch": 2.70216306156406,
      "grad_norm": 0.039707403630018234,
      "learning_rate": 3.2839506172839505e-05,
      "loss": 0.5561,
      "step": 1218
    },
    {
      "epoch": 2.7043815862451472,
      "grad_norm": 0.04496249929070473,
      "learning_rate": 3.2592592592592586e-05,
      "loss": 0.6184,
      "step": 1219
    },
    {
      "epoch": 2.7066001109262343,
      "grad_norm": 0.07033520936965942,
      "learning_rate": 3.2345679012345675e-05,
      "loss": 0.6511,
      "step": 1220
    },
    {
      "epoch": 2.7088186356073214,
      "grad_norm": 0.040686819702386856,
      "learning_rate": 3.209876543209876e-05,
      "loss": 0.5091,
      "step": 1221
    },
    {
      "epoch": 2.7110371602884085,
      "grad_norm": 0.03303593397140503,
      "learning_rate": 3.185185185185185e-05,
      "loss": 0.5987,
      "step": 1222
    },
    {
      "epoch": 2.7132556849694955,
      "grad_norm": 0.04120054841041565,
      "learning_rate": 3.160493827160493e-05,
      "loss": 0.6219,
      "step": 1223
    },
    {
      "epoch": 2.7154742096505826,
      "grad_norm": 0.07147549092769623,
      "learning_rate": 3.135802469135802e-05,
      "loss": 0.6832,
      "step": 1224
    },
    {
      "epoch": 2.7176927343316697,
      "grad_norm": 0.059234362095594406,
      "learning_rate": 3.111111111111111e-05,
      "loss": 0.6537,
      "step": 1225
    },
    {
      "epoch": 2.7199112590127568,
      "grad_norm": 0.0356709361076355,
      "learning_rate": 3.08641975308642e-05,
      "loss": 0.5405,
      "step": 1226
    },
    {
      "epoch": 2.722129783693844,
      "grad_norm": 0.04336230084300041,
      "learning_rate": 3.061728395061728e-05,
      "loss": 0.6178,
      "step": 1227
    },
    {
      "epoch": 2.724348308374931,
      "grad_norm": 0.03438165783882141,
      "learning_rate": 3.0370370370370368e-05,
      "loss": 0.5554,
      "step": 1228
    },
    {
      "epoch": 2.726566833056018,
      "grad_norm": 0.043426480144262314,
      "learning_rate": 3.0123456790123453e-05,
      "loss": 0.5662,
      "step": 1229
    },
    {
      "epoch": 2.728785357737105,
      "grad_norm": 0.03418657183647156,
      "learning_rate": 2.987654320987654e-05,
      "loss": 0.4815,
      "step": 1230
    },
    {
      "epoch": 2.731003882418192,
      "grad_norm": 0.03722352162003517,
      "learning_rate": 2.9629629629629627e-05,
      "loss": 0.6191,
      "step": 1231
    },
    {
      "epoch": 2.733222407099279,
      "grad_norm": 0.04361360892653465,
      "learning_rate": 2.9382716049382715e-05,
      "loss": 0.4106,
      "step": 1232
    },
    {
      "epoch": 2.7354409317803663,
      "grad_norm": 0.04802580177783966,
      "learning_rate": 2.9135802469135796e-05,
      "loss": 0.5152,
      "step": 1233
    },
    {
      "epoch": 2.7376594564614534,
      "grad_norm": 0.036394596099853516,
      "learning_rate": 2.8888888888888885e-05,
      "loss": 0.5319,
      "step": 1234
    },
    {
      "epoch": 2.7398779811425404,
      "grad_norm": 0.03709585964679718,
      "learning_rate": 2.8641975308641973e-05,
      "loss": 0.4367,
      "step": 1235
    },
    {
      "epoch": 2.7420965058236275,
      "grad_norm": 0.031694021075963974,
      "learning_rate": 2.8395061728395058e-05,
      "loss": 0.5327,
      "step": 1236
    },
    {
      "epoch": 2.7443150305047146,
      "grad_norm": 0.028433453291654587,
      "learning_rate": 2.8148148148148147e-05,
      "loss": 0.5729,
      "step": 1237
    },
    {
      "epoch": 2.7465335551858017,
      "grad_norm": 0.0583704374730587,
      "learning_rate": 2.790123456790123e-05,
      "loss": 0.6225,
      "step": 1238
    },
    {
      "epoch": 2.7487520798668887,
      "grad_norm": 0.06915649771690369,
      "learning_rate": 2.765432098765432e-05,
      "loss": 0.5313,
      "step": 1239
    },
    {
      "epoch": 2.750970604547976,
      "grad_norm": 0.0422714427113533,
      "learning_rate": 2.7407407407407405e-05,
      "loss": 0.6739,
      "step": 1240
    },
    {
      "epoch": 2.753189129229063,
      "grad_norm": 0.055159103125333786,
      "learning_rate": 2.7160493827160493e-05,
      "loss": 0.5713,
      "step": 1241
    },
    {
      "epoch": 2.75540765391015,
      "grad_norm": 0.036662377417087555,
      "learning_rate": 2.6913580246913578e-05,
      "loss": 0.4886,
      "step": 1242
    },
    {
      "epoch": 2.757626178591237,
      "grad_norm": 0.032279785722494125,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 0.5354,
      "step": 1243
    },
    {
      "epoch": 2.759844703272324,
      "grad_norm": 0.04173227772116661,
      "learning_rate": 2.6419753086419748e-05,
      "loss": 0.6649,
      "step": 1244
    },
    {
      "epoch": 2.762063227953411,
      "grad_norm": 0.035207826644182205,
      "learning_rate": 2.6172839506172837e-05,
      "loss": 0.6485,
      "step": 1245
    },
    {
      "epoch": 2.7642817526344983,
      "grad_norm": 0.035952359437942505,
      "learning_rate": 2.592592592592592e-05,
      "loss": 0.5305,
      "step": 1246
    },
    {
      "epoch": 2.7665002773155853,
      "grad_norm": 0.04481656849384308,
      "learning_rate": 2.567901234567901e-05,
      "loss": 0.5936,
      "step": 1247
    },
    {
      "epoch": 2.7687188019966724,
      "grad_norm": 0.034188538789749146,
      "learning_rate": 2.5432098765432095e-05,
      "loss": 0.4816,
      "step": 1248
    },
    {
      "epoch": 2.7709373266777595,
      "grad_norm": 0.031913045793771744,
      "learning_rate": 2.5185185185185183e-05,
      "loss": 0.6905,
      "step": 1249
    },
    {
      "epoch": 2.7731558513588466,
      "grad_norm": 0.03986291214823723,
      "learning_rate": 2.4938271604938268e-05,
      "loss": 0.5626,
      "step": 1250
    },
    {
      "epoch": 2.7753743760399336,
      "grad_norm": 0.04012264683842659,
      "learning_rate": 2.4691358024691357e-05,
      "loss": 0.6864,
      "step": 1251
    },
    {
      "epoch": 2.7775929007210207,
      "grad_norm": 0.03735329583287239,
      "learning_rate": 2.4444444444444445e-05,
      "loss": 0.5426,
      "step": 1252
    },
    {
      "epoch": 2.779811425402108,
      "grad_norm": 0.04680916666984558,
      "learning_rate": 2.419753086419753e-05,
      "loss": 0.7803,
      "step": 1253
    },
    {
      "epoch": 2.782029950083195,
      "grad_norm": 0.04490986093878746,
      "learning_rate": 2.3950617283950618e-05,
      "loss": 0.507,
      "step": 1254
    },
    {
      "epoch": 2.784248474764282,
      "grad_norm": 0.035756371915340424,
      "learning_rate": 2.37037037037037e-05,
      "loss": 0.5256,
      "step": 1255
    },
    {
      "epoch": 2.786466999445369,
      "grad_norm": 0.043469302356243134,
      "learning_rate": 2.3456790123456788e-05,
      "loss": 0.5944,
      "step": 1256
    },
    {
      "epoch": 2.788685524126456,
      "grad_norm": 0.04514380171895027,
      "learning_rate": 2.3209876543209873e-05,
      "loss": 0.6799,
      "step": 1257
    },
    {
      "epoch": 2.790904048807543,
      "grad_norm": 0.03310393914580345,
      "learning_rate": 2.296296296296296e-05,
      "loss": 0.5333,
      "step": 1258
    },
    {
      "epoch": 2.7931225734886302,
      "grad_norm": 0.03845299780368805,
      "learning_rate": 2.2716049382716047e-05,
      "loss": 0.7571,
      "step": 1259
    },
    {
      "epoch": 2.7953410981697173,
      "grad_norm": 0.041685521602630615,
      "learning_rate": 2.2469135802469135e-05,
      "loss": 0.6139,
      "step": 1260
    },
    {
      "epoch": 2.7975596228508044,
      "grad_norm": 0.04806831106543541,
      "learning_rate": 2.222222222222222e-05,
      "loss": 0.4286,
      "step": 1261
    },
    {
      "epoch": 2.7997781475318915,
      "grad_norm": 0.04432718828320503,
      "learning_rate": 2.1975308641975308e-05,
      "loss": 0.6249,
      "step": 1262
    },
    {
      "epoch": 2.8019966722129785,
      "grad_norm": 0.03560454770922661,
      "learning_rate": 2.1728395061728393e-05,
      "loss": 0.457,
      "step": 1263
    },
    {
      "epoch": 2.8042151968940656,
      "grad_norm": 0.05061856284737587,
      "learning_rate": 2.148148148148148e-05,
      "loss": 0.6198,
      "step": 1264
    },
    {
      "epoch": 2.8064337215751527,
      "grad_norm": 0.04333832859992981,
      "learning_rate": 2.1234567901234563e-05,
      "loss": 0.7272,
      "step": 1265
    },
    {
      "epoch": 2.8086522462562398,
      "grad_norm": 0.03778650611639023,
      "learning_rate": 2.098765432098765e-05,
      "loss": 0.5102,
      "step": 1266
    },
    {
      "epoch": 2.810870770937327,
      "grad_norm": 0.03803731128573418,
      "learning_rate": 2.0740740740740737e-05,
      "loss": 0.682,
      "step": 1267
    },
    {
      "epoch": 2.813089295618414,
      "grad_norm": 0.05231521278619766,
      "learning_rate": 2.0493827160493825e-05,
      "loss": 0.5173,
      "step": 1268
    },
    {
      "epoch": 2.815307820299501,
      "grad_norm": 0.031249944120645523,
      "learning_rate": 2.0246913580246913e-05,
      "loss": 0.4752,
      "step": 1269
    },
    {
      "epoch": 2.817526344980588,
      "grad_norm": 0.035584524273872375,
      "learning_rate": 1.9999999999999998e-05,
      "loss": 0.4976,
      "step": 1270
    },
    {
      "epoch": 2.819744869661675,
      "grad_norm": 0.045795049518346786,
      "learning_rate": 1.9753086419753087e-05,
      "loss": 0.5069,
      "step": 1271
    },
    {
      "epoch": 2.821963394342762,
      "grad_norm": 0.03974984213709831,
      "learning_rate": 1.950617283950617e-05,
      "loss": 0.564,
      "step": 1272
    },
    {
      "epoch": 2.8241819190238493,
      "grad_norm": 0.03639185056090355,
      "learning_rate": 1.925925925925926e-05,
      "loss": 0.5429,
      "step": 1273
    },
    {
      "epoch": 2.8264004437049364,
      "grad_norm": 0.03716005012392998,
      "learning_rate": 1.901234567901234e-05,
      "loss": 0.5674,
      "step": 1274
    },
    {
      "epoch": 2.8286189683860234,
      "grad_norm": 0.043952859938144684,
      "learning_rate": 1.8765432098765433e-05,
      "loss": 0.6989,
      "step": 1275
    },
    {
      "epoch": 2.8308374930671105,
      "grad_norm": 0.03944656252861023,
      "learning_rate": 1.8518518518518515e-05,
      "loss": 0.5181,
      "step": 1276
    },
    {
      "epoch": 2.8330560177481976,
      "grad_norm": 0.04158734530210495,
      "learning_rate": 1.8271604938271603e-05,
      "loss": 0.6728,
      "step": 1277
    },
    {
      "epoch": 2.8352745424292847,
      "grad_norm": 0.03405851498246193,
      "learning_rate": 1.802469135802469e-05,
      "loss": 0.4614,
      "step": 1278
    },
    {
      "epoch": 2.8374930671103717,
      "grad_norm": 0.039757147431373596,
      "learning_rate": 1.7777777777777777e-05,
      "loss": 0.5336,
      "step": 1279
    },
    {
      "epoch": 2.839711591791459,
      "grad_norm": 0.043418340384960175,
      "learning_rate": 1.753086419753086e-05,
      "loss": 0.5586,
      "step": 1280
    },
    {
      "epoch": 2.841930116472546,
      "grad_norm": 0.032932594418525696,
      "learning_rate": 1.728395061728395e-05,
      "loss": 0.4361,
      "step": 1281
    },
    {
      "epoch": 2.844148641153633,
      "grad_norm": 0.047009628266096115,
      "learning_rate": 1.7037037037037035e-05,
      "loss": 0.7304,
      "step": 1282
    },
    {
      "epoch": 2.84636716583472,
      "grad_norm": 0.02955682761967182,
      "learning_rate": 1.6790123456790123e-05,
      "loss": 0.5012,
      "step": 1283
    },
    {
      "epoch": 2.848585690515807,
      "grad_norm": 0.03543225675821304,
      "learning_rate": 1.6543209876543208e-05,
      "loss": 0.4364,
      "step": 1284
    },
    {
      "epoch": 2.850804215196894,
      "grad_norm": 0.03977542743086815,
      "learning_rate": 1.6296296296296293e-05,
      "loss": 0.5637,
      "step": 1285
    },
    {
      "epoch": 2.8530227398779813,
      "grad_norm": 0.033268384635448456,
      "learning_rate": 1.604938271604938e-05,
      "loss": 0.4943,
      "step": 1286
    },
    {
      "epoch": 2.8552412645590683,
      "grad_norm": 0.045146726071834564,
      "learning_rate": 1.5802469135802467e-05,
      "loss": 0.3815,
      "step": 1287
    },
    {
      "epoch": 2.8574597892401554,
      "grad_norm": 0.0371917188167572,
      "learning_rate": 1.5555555555555555e-05,
      "loss": 0.525,
      "step": 1288
    },
    {
      "epoch": 2.8596783139212425,
      "grad_norm": 0.06794360280036926,
      "learning_rate": 1.530864197530864e-05,
      "loss": 0.6057,
      "step": 1289
    },
    {
      "epoch": 2.8618968386023296,
      "grad_norm": 0.039147377014160156,
      "learning_rate": 1.5061728395061727e-05,
      "loss": 0.4858,
      "step": 1290
    },
    {
      "epoch": 2.8641153632834166,
      "grad_norm": 0.03104197047650814,
      "learning_rate": 1.4814814814814813e-05,
      "loss": 0.5267,
      "step": 1291
    },
    {
      "epoch": 2.8663338879645037,
      "grad_norm": 0.05600616708397865,
      "learning_rate": 1.4567901234567898e-05,
      "loss": 0.516,
      "step": 1292
    },
    {
      "epoch": 2.868552412645591,
      "grad_norm": 0.03996296226978302,
      "learning_rate": 1.4320987654320987e-05,
      "loss": 0.7136,
      "step": 1293
    },
    {
      "epoch": 2.870770937326678,
      "grad_norm": 0.04207553714513779,
      "learning_rate": 1.4074074074074073e-05,
      "loss": 0.6354,
      "step": 1294
    },
    {
      "epoch": 2.872989462007765,
      "grad_norm": 0.03918815776705742,
      "learning_rate": 1.382716049382716e-05,
      "loss": 0.6534,
      "step": 1295
    },
    {
      "epoch": 2.875207986688852,
      "grad_norm": 0.03098967857658863,
      "learning_rate": 1.3580246913580247e-05,
      "loss": 0.5297,
      "step": 1296
    },
    {
      "epoch": 2.877426511369939,
      "grad_norm": 0.03798598051071167,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.5395,
      "step": 1297
    },
    {
      "epoch": 2.879645036051026,
      "grad_norm": 0.047824420034885406,
      "learning_rate": 1.3086419753086418e-05,
      "loss": 0.6588,
      "step": 1298
    },
    {
      "epoch": 2.8818635607321132,
      "grad_norm": 0.04483456164598465,
      "learning_rate": 1.2839506172839505e-05,
      "loss": 0.7132,
      "step": 1299
    },
    {
      "epoch": 2.8840820854132003,
      "grad_norm": 0.037899795919656754,
      "learning_rate": 1.2592592592592592e-05,
      "loss": 0.5008,
      "step": 1300
    },
    {
      "epoch": 2.8863006100942874,
      "grad_norm": 0.0634029284119606,
      "learning_rate": 1.2345679012345678e-05,
      "loss": 0.6288,
      "step": 1301
    },
    {
      "epoch": 2.8885191347753745,
      "grad_norm": 0.05174478515982628,
      "learning_rate": 1.2098765432098765e-05,
      "loss": 0.4801,
      "step": 1302
    },
    {
      "epoch": 2.8907376594564616,
      "grad_norm": 0.06148293986916542,
      "learning_rate": 1.185185185185185e-05,
      "loss": 0.6061,
      "step": 1303
    },
    {
      "epoch": 2.8929561841375486,
      "grad_norm": 0.04003925994038582,
      "learning_rate": 1.1604938271604937e-05,
      "loss": 0.5741,
      "step": 1304
    },
    {
      "epoch": 2.8951747088186357,
      "grad_norm": 0.06649940460920334,
      "learning_rate": 1.1358024691358023e-05,
      "loss": 0.5138,
      "step": 1305
    },
    {
      "epoch": 2.8973932334997228,
      "grad_norm": 0.03758198767900467,
      "learning_rate": 1.111111111111111e-05,
      "loss": 0.5517,
      "step": 1306
    },
    {
      "epoch": 2.89961175818081,
      "grad_norm": 0.0445292592048645,
      "learning_rate": 1.0864197530864197e-05,
      "loss": 0.6089,
      "step": 1307
    },
    {
      "epoch": 2.901830282861897,
      "grad_norm": 0.037991348654031754,
      "learning_rate": 1.0617283950617282e-05,
      "loss": 0.5828,
      "step": 1308
    },
    {
      "epoch": 2.904048807542984,
      "grad_norm": 0.04191242903470993,
      "learning_rate": 1.0370370370370368e-05,
      "loss": 0.5354,
      "step": 1309
    },
    {
      "epoch": 2.906267332224071,
      "grad_norm": 0.031815435737371445,
      "learning_rate": 1.0123456790123457e-05,
      "loss": 0.3407,
      "step": 1310
    },
    {
      "epoch": 2.908485856905158,
      "grad_norm": 0.029780181124806404,
      "learning_rate": 9.876543209876543e-06,
      "loss": 0.5733,
      "step": 1311
    },
    {
      "epoch": 2.9107043815862452,
      "grad_norm": 0.03866731375455856,
      "learning_rate": 9.62962962962963e-06,
      "loss": 0.5853,
      "step": 1312
    },
    {
      "epoch": 2.9129229062673323,
      "grad_norm": 0.04620196670293808,
      "learning_rate": 9.382716049382717e-06,
      "loss": 0.5234,
      "step": 1313
    },
    {
      "epoch": 2.9151414309484194,
      "grad_norm": 0.03367392346262932,
      "learning_rate": 9.135802469135802e-06,
      "loss": 0.5396,
      "step": 1314
    },
    {
      "epoch": 2.9173599556295065,
      "grad_norm": 0.03621271252632141,
      "learning_rate": 8.888888888888888e-06,
      "loss": 0.5319,
      "step": 1315
    },
    {
      "epoch": 2.9195784803105935,
      "grad_norm": 0.03952901437878609,
      "learning_rate": 8.641975308641975e-06,
      "loss": 0.7952,
      "step": 1316
    },
    {
      "epoch": 2.9217970049916806,
      "grad_norm": 0.04182921722531319,
      "learning_rate": 8.395061728395062e-06,
      "loss": 0.6203,
      "step": 1317
    },
    {
      "epoch": 2.9240155296727677,
      "grad_norm": 0.038099873811006546,
      "learning_rate": 8.148148148148147e-06,
      "loss": 0.5498,
      "step": 1318
    },
    {
      "epoch": 2.9262340543538548,
      "grad_norm": 0.03311851620674133,
      "learning_rate": 7.901234567901233e-06,
      "loss": 0.5367,
      "step": 1319
    },
    {
      "epoch": 2.928452579034942,
      "grad_norm": 0.04026024416089058,
      "learning_rate": 7.65432098765432e-06,
      "loss": 0.6115,
      "step": 1320
    },
    {
      "epoch": 2.930671103716029,
      "grad_norm": 0.03772692382335663,
      "learning_rate": 7.407407407407407e-06,
      "loss": 0.6864,
      "step": 1321
    },
    {
      "epoch": 2.932889628397116,
      "grad_norm": 0.03806360810995102,
      "learning_rate": 7.160493827160493e-06,
      "loss": 0.5139,
      "step": 1322
    },
    {
      "epoch": 2.935108153078203,
      "grad_norm": 0.040801092982292175,
      "learning_rate": 6.91358024691358e-06,
      "loss": 0.5379,
      "step": 1323
    },
    {
      "epoch": 2.93732667775929,
      "grad_norm": 0.04487309604883194,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.5284,
      "step": 1324
    },
    {
      "epoch": 2.939545202440377,
      "grad_norm": 0.043474651873111725,
      "learning_rate": 6.4197530864197525e-06,
      "loss": 0.6452,
      "step": 1325
    },
    {
      "epoch": 2.9417637271214643,
      "grad_norm": 0.03828020393848419,
      "learning_rate": 6.172839506172839e-06,
      "loss": 0.5881,
      "step": 1326
    },
    {
      "epoch": 2.9439822518025514,
      "grad_norm": 0.03561675548553467,
      "learning_rate": 5.925925925925925e-06,
      "loss": 0.6173,
      "step": 1327
    },
    {
      "epoch": 2.9462007764836384,
      "grad_norm": 0.04336895793676376,
      "learning_rate": 5.679012345679012e-06,
      "loss": 0.6707,
      "step": 1328
    },
    {
      "epoch": 2.9484193011647255,
      "grad_norm": 0.037055496126413345,
      "learning_rate": 5.432098765432098e-06,
      "loss": 0.4827,
      "step": 1329
    },
    {
      "epoch": 2.9506378258458126,
      "grad_norm": 0.03776615485548973,
      "learning_rate": 5.185185185185184e-06,
      "loss": 0.6016,
      "step": 1330
    },
    {
      "epoch": 2.9528563505268997,
      "grad_norm": 0.048225317150354385,
      "learning_rate": 4.938271604938272e-06,
      "loss": 0.6564,
      "step": 1331
    },
    {
      "epoch": 2.9550748752079867,
      "grad_norm": 0.02783077023923397,
      "learning_rate": 4.691358024691358e-06,
      "loss": 0.5677,
      "step": 1332
    },
    {
      "epoch": 2.957293399889074,
      "grad_norm": 0.048876676708459854,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.5308,
      "step": 1333
    },
    {
      "epoch": 2.959511924570161,
      "grad_norm": 0.0488305501639843,
      "learning_rate": 4.197530864197531e-06,
      "loss": 0.6081,
      "step": 1334
    },
    {
      "epoch": 2.961730449251248,
      "grad_norm": 0.036671049892902374,
      "learning_rate": 3.950617283950617e-06,
      "loss": 0.6789,
      "step": 1335
    },
    {
      "epoch": 2.963948973932335,
      "grad_norm": 0.05885665863752365,
      "learning_rate": 3.7037037037037033e-06,
      "loss": 0.7269,
      "step": 1336
    },
    {
      "epoch": 2.966167498613422,
      "grad_norm": 0.05364936590194702,
      "learning_rate": 3.45679012345679e-06,
      "loss": 0.5398,
      "step": 1337
    },
    {
      "epoch": 2.968386023294509,
      "grad_norm": 0.03259783983230591,
      "learning_rate": 3.2098765432098762e-06,
      "loss": 0.5127,
      "step": 1338
    },
    {
      "epoch": 2.9706045479755963,
      "grad_norm": 0.03313614800572395,
      "learning_rate": 2.9629629629629625e-06,
      "loss": 0.4651,
      "step": 1339
    },
    {
      "epoch": 2.9728230726566833,
      "grad_norm": 0.03485209122300148,
      "learning_rate": 2.716049382716049e-06,
      "loss": 0.6406,
      "step": 1340
    },
    {
      "epoch": 2.9750415973377704,
      "grad_norm": 0.038500651717185974,
      "learning_rate": 2.469135802469136e-06,
      "loss": 0.5802,
      "step": 1341
    },
    {
      "epoch": 2.9772601220188575,
      "grad_norm": 0.11935567855834961,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.6649,
      "step": 1342
    },
    {
      "epoch": 2.9794786466999446,
      "grad_norm": 0.035361435264348984,
      "learning_rate": 1.9753086419753083e-06,
      "loss": 0.5334,
      "step": 1343
    },
    {
      "epoch": 2.9816971713810316,
      "grad_norm": 0.05473244935274124,
      "learning_rate": 1.728395061728395e-06,
      "loss": 0.6039,
      "step": 1344
    },
    {
      "epoch": 2.9839156960621187,
      "grad_norm": 0.032312896102666855,
      "learning_rate": 1.4814814814814812e-06,
      "loss": 0.7357,
      "step": 1345
    },
    {
      "epoch": 2.986134220743206,
      "grad_norm": 0.0421266183257103,
      "learning_rate": 1.234567901234568e-06,
      "loss": 0.7709,
      "step": 1346
    },
    {
      "epoch": 2.988352745424293,
      "grad_norm": 0.04149707779288292,
      "learning_rate": 9.876543209876542e-07,
      "loss": 0.4991,
      "step": 1347
    },
    {
      "epoch": 2.99057127010538,
      "grad_norm": 0.03665245696902275,
      "learning_rate": 7.407407407407406e-07,
      "loss": 0.5044,
      "step": 1348
    },
    {
      "epoch": 2.992789794786467,
      "grad_norm": 0.04145513102412224,
      "learning_rate": 4.938271604938271e-07,
      "loss": 0.5788,
      "step": 1349
    },
    {
      "epoch": 2.995008319467554,
      "grad_norm": 0.032134927809238434,
      "learning_rate": 2.4691358024691354e-07,
      "loss": 0.4827,
      "step": 1350
    },
    {
      "epoch": 2.995008319467554,
      "step": 1350,
      "total_flos": 1.2007874157241958e+18,
      "train_loss": 0.5807985235143591,
      "train_runtime": 6839.0426,
      "train_samples_per_second": 3.163,
      "train_steps_per_second": 0.197
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 1350,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.2007874157241958e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}