{
  "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": 0.09311392903327942,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.8259,
      "step": 1
    },
    {
      "epoch": 0.004437049362174155,
      "grad_norm": 0.17367631196975708,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.7533,
      "step": 2
    },
    {
      "epoch": 0.0066555740432612314,
      "grad_norm": 0.12261653691530228,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.8634,
      "step": 3
    },
    {
      "epoch": 0.00887409872434831,
      "grad_norm": 0.09639201313257217,
      "learning_rate": 8.888888888888888e-06,
      "loss": 0.7027,
      "step": 4
    },
    {
      "epoch": 0.011092623405435386,
      "grad_norm": 0.07629800587892532,
      "learning_rate": 1.111111111111111e-05,
      "loss": 0.7436,
      "step": 5
    },
    {
      "epoch": 0.013311148086522463,
      "grad_norm": 0.11611945182085037,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.7038,
      "step": 6
    },
    {
      "epoch": 0.01552967276760954,
      "grad_norm": 0.11175328493118286,
      "learning_rate": 1.5555555555555555e-05,
      "loss": 0.7394,
      "step": 7
    },
    {
      "epoch": 0.01774819744869662,
      "grad_norm": 0.0947323590517044,
      "learning_rate": 1.7777777777777777e-05,
      "loss": 0.9079,
      "step": 8
    },
    {
      "epoch": 0.019966722129783693,
      "grad_norm": 0.23429566621780396,
      "learning_rate": 1.9999999999999998e-05,
      "loss": 0.8004,
      "step": 9
    },
    {
      "epoch": 0.022185246810870772,
      "grad_norm": 0.07250005751848221,
      "learning_rate": 2.222222222222222e-05,
      "loss": 0.9161,
      "step": 10
    },
    {
      "epoch": 0.024403771491957847,
      "grad_norm": 0.31661713123321533,
      "learning_rate": 2.4444444444444445e-05,
      "loss": 1.1644,
      "step": 11
    },
    {
      "epoch": 0.026622296173044926,
      "grad_norm": 0.09892977774143219,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 0.7022,
      "step": 12
    },
    {
      "epoch": 0.028840820854132,
      "grad_norm": 0.10269516706466675,
      "learning_rate": 2.8888888888888885e-05,
      "loss": 0.8657,
      "step": 13
    },
    {
      "epoch": 0.03105934553521908,
      "grad_norm": 0.07999370247125626,
      "learning_rate": 3.111111111111111e-05,
      "loss": 1.0626,
      "step": 14
    },
    {
      "epoch": 0.033277870216306155,
      "grad_norm": 0.15100766718387604,
      "learning_rate": 3.333333333333333e-05,
      "loss": 0.8593,
      "step": 15
    },
    {
      "epoch": 0.03549639489739324,
      "grad_norm": 0.11366188526153564,
      "learning_rate": 3.555555555555555e-05,
      "loss": 0.8863,
      "step": 16
    },
    {
      "epoch": 0.03771491957848031,
      "grad_norm": 0.11790313571691513,
      "learning_rate": 3.777777777777777e-05,
      "loss": 0.8245,
      "step": 17
    },
    {
      "epoch": 0.03993344425956739,
      "grad_norm": 0.08872196078300476,
      "learning_rate": 3.9999999999999996e-05,
      "loss": 0.7463,
      "step": 18
    },
    {
      "epoch": 0.04215196894065446,
      "grad_norm": 0.07815517485141754,
      "learning_rate": 4.222222222222222e-05,
      "loss": 0.7349,
      "step": 19
    },
    {
      "epoch": 0.044370493621741544,
      "grad_norm": 0.10559378564357758,
      "learning_rate": 4.444444444444444e-05,
      "loss": 0.7921,
      "step": 20
    },
    {
      "epoch": 0.04658901830282862,
      "grad_norm": 0.11112574487924576,
      "learning_rate": 4.6666666666666665e-05,
      "loss": 0.8868,
      "step": 21
    },
    {
      "epoch": 0.048807542983915694,
      "grad_norm": 0.0983230248093605,
      "learning_rate": 4.888888888888889e-05,
      "loss": 0.8287,
      "step": 22
    },
    {
      "epoch": 0.051026067665002776,
      "grad_norm": 0.12154821306467056,
      "learning_rate": 5.111111111111111e-05,
      "loss": 0.9546,
      "step": 23
    },
    {
      "epoch": 0.05324459234608985,
      "grad_norm": 0.07469753921031952,
      "learning_rate": 5.333333333333333e-05,
      "loss": 0.7113,
      "step": 24
    },
    {
      "epoch": 0.05546311702717693,
      "grad_norm": 0.1166706383228302,
      "learning_rate": 5.5555555555555545e-05,
      "loss": 0.935,
      "step": 25
    },
    {
      "epoch": 0.057681641708264,
      "grad_norm": 0.1292790174484253,
      "learning_rate": 5.777777777777777e-05,
      "loss": 0.8035,
      "step": 26
    },
    {
      "epoch": 0.059900166389351084,
      "grad_norm": 0.0873345136642456,
      "learning_rate": 5.9999999999999995e-05,
      "loss": 0.9862,
      "step": 27
    },
    {
      "epoch": 0.06211869107043816,
      "grad_norm": 0.10106363147497177,
      "learning_rate": 6.222222222222222e-05,
      "loss": 0.7299,
      "step": 28
    },
    {
      "epoch": 0.06433721575152523,
      "grad_norm": 0.09163547307252884,
      "learning_rate": 6.444444444444444e-05,
      "loss": 0.8663,
      "step": 29
    },
    {
      "epoch": 0.06655574043261231,
      "grad_norm": 0.09624188393354416,
      "learning_rate": 6.666666666666666e-05,
      "loss": 0.7739,
      "step": 30
    },
    {
      "epoch": 0.06877426511369938,
      "grad_norm": 0.0785500630736351,
      "learning_rate": 6.888888888888888e-05,
      "loss": 0.7997,
      "step": 31
    },
    {
      "epoch": 0.07099278979478647,
      "grad_norm": 0.10194975137710571,
      "learning_rate": 7.11111111111111e-05,
      "loss": 0.9465,
      "step": 32
    },
    {
      "epoch": 0.07321131447587355,
      "grad_norm": 0.0985187217593193,
      "learning_rate": 7.333333333333332e-05,
      "loss": 0.8434,
      "step": 33
    },
    {
      "epoch": 0.07542983915696062,
      "grad_norm": 0.11365296691656113,
      "learning_rate": 7.555555555555554e-05,
      "loss": 0.8904,
      "step": 34
    },
    {
      "epoch": 0.0776483638380477,
      "grad_norm": 0.12622281908988953,
      "learning_rate": 7.777777777777777e-05,
      "loss": 0.7564,
      "step": 35
    },
    {
      "epoch": 0.07986688851913477,
      "grad_norm": 0.10349198430776596,
      "learning_rate": 7.999999999999999e-05,
      "loss": 0.8167,
      "step": 36
    },
    {
      "epoch": 0.08208541320022185,
      "grad_norm": 0.11963187903165817,
      "learning_rate": 8.222222222222222e-05,
      "loss": 0.9434,
      "step": 37
    },
    {
      "epoch": 0.08430393788130892,
      "grad_norm": 0.11631946265697479,
      "learning_rate": 8.444444444444444e-05,
      "loss": 0.9016,
      "step": 38
    },
    {
      "epoch": 0.08652246256239601,
      "grad_norm": 0.2723287343978882,
      "learning_rate": 8.666666666666665e-05,
      "loss": 0.8333,
      "step": 39
    },
    {
      "epoch": 0.08874098724348309,
      "grad_norm": 0.09371933341026306,
      "learning_rate": 8.888888888888888e-05,
      "loss": 0.881,
      "step": 40
    },
    {
      "epoch": 0.09095951192457016,
      "grad_norm": 0.09889836609363556,
      "learning_rate": 9.11111111111111e-05,
      "loss": 0.8617,
      "step": 41
    },
    {
      "epoch": 0.09317803660565724,
      "grad_norm": 0.10903526097536087,
      "learning_rate": 9.333333333333333e-05,
      "loss": 0.8075,
      "step": 42
    },
    {
      "epoch": 0.09539656128674431,
      "grad_norm": 0.0968666598200798,
      "learning_rate": 9.555555555555555e-05,
      "loss": 0.9012,
      "step": 43
    },
    {
      "epoch": 0.09761508596783139,
      "grad_norm": 0.08768698573112488,
      "learning_rate": 9.777777777777778e-05,
      "loss": 0.879,
      "step": 44
    },
    {
      "epoch": 0.09983361064891846,
      "grad_norm": 0.1483982503414154,
      "learning_rate": 9.999999999999999e-05,
      "loss": 0.9299,
      "step": 45
    },
    {
      "epoch": 0.10205213533000555,
      "grad_norm": 0.10396942496299744,
      "learning_rate": 0.00010222222222222222,
      "loss": 0.8794,
      "step": 46
    },
    {
      "epoch": 0.10427066001109263,
      "grad_norm": 0.13117918372154236,
      "learning_rate": 0.00010444444444444444,
      "loss": 0.9461,
      "step": 47
    },
    {
      "epoch": 0.1064891846921797,
      "grad_norm": 0.12005843967199326,
      "learning_rate": 0.00010666666666666667,
      "loss": 0.7959,
      "step": 48
    },
    {
      "epoch": 0.10870770937326678,
      "grad_norm": 0.09258624911308289,
      "learning_rate": 0.00010888888888888889,
      "loss": 0.9379,
      "step": 49
    },
    {
      "epoch": 0.11092623405435385,
      "grad_norm": 0.10733049362897873,
      "learning_rate": 0.00011111111111111109,
      "loss": 0.8854,
      "step": 50
    },
    {
      "epoch": 0.11314475873544093,
      "grad_norm": 0.09922249615192413,
      "learning_rate": 0.00011333333333333331,
      "loss": 0.8724,
      "step": 51
    },
    {
      "epoch": 0.115363283416528,
      "grad_norm": 0.0882197842001915,
      "learning_rate": 0.00011555555555555554,
      "loss": 0.8966,
      "step": 52
    },
    {
      "epoch": 0.11758180809761509,
      "grad_norm": 0.10572140663862228,
      "learning_rate": 0.00011777777777777776,
      "loss": 0.9801,
      "step": 53
    },
    {
      "epoch": 0.11980033277870217,
      "grad_norm": 0.10663247108459473,
      "learning_rate": 0.00011999999999999999,
      "loss": 1.0039,
      "step": 54
    },
    {
      "epoch": 0.12201885745978924,
      "grad_norm": 0.08858172595500946,
      "learning_rate": 0.00012222222222222221,
      "loss": 0.6706,
      "step": 55
    },
    {
      "epoch": 0.12423738214087632,
      "grad_norm": 0.10027160495519638,
      "learning_rate": 0.00012444444444444444,
      "loss": 0.7864,
      "step": 56
    },
    {
      "epoch": 0.1264559068219634,
      "grad_norm": 0.08747286349534988,
      "learning_rate": 0.00012666666666666666,
      "loss": 0.7417,
      "step": 57
    },
    {
      "epoch": 0.12867443150305047,
      "grad_norm": 0.14820325374603271,
      "learning_rate": 0.0001288888888888889,
      "loss": 0.8917,
      "step": 58
    },
    {
      "epoch": 0.13089295618413754,
      "grad_norm": 0.1078650951385498,
      "learning_rate": 0.00013111111111111111,
      "loss": 0.786,
      "step": 59
    },
    {
      "epoch": 0.13311148086522462,
      "grad_norm": 0.08175945281982422,
      "learning_rate": 0.0001333333333333333,
      "loss": 0.6256,
      "step": 60
    },
    {
      "epoch": 0.1353300055463117,
      "grad_norm": 0.08706164360046387,
      "learning_rate": 0.00013555555555555554,
      "loss": 0.8981,
      "step": 61
    },
    {
      "epoch": 0.13754853022739877,
      "grad_norm": 0.1044289618730545,
      "learning_rate": 0.00013777777777777776,
      "loss": 0.8133,
      "step": 62
    },
    {
      "epoch": 0.13976705490848584,
      "grad_norm": 0.07393615692853928,
      "learning_rate": 0.00014,
      "loss": 0.7962,
      "step": 63
    },
    {
      "epoch": 0.14198557958957295,
      "grad_norm": 0.0890442356467247,
      "learning_rate": 0.0001422222222222222,
      "loss": 0.7844,
      "step": 64
    },
    {
      "epoch": 0.14420410427066002,
      "grad_norm": 0.1089281514286995,
      "learning_rate": 0.0001444444444444444,
      "loss": 0.909,
      "step": 65
    },
    {
      "epoch": 0.1464226289517471,
      "grad_norm": 0.07143402099609375,
      "learning_rate": 0.00014666666666666664,
      "loss": 0.7802,
      "step": 66
    },
    {
      "epoch": 0.14864115363283417,
      "grad_norm": 0.09307138621807098,
      "learning_rate": 0.00014888888888888886,
      "loss": 0.859,
      "step": 67
    },
    {
      "epoch": 0.15085967831392125,
      "grad_norm": 0.07586538046598434,
      "learning_rate": 0.00015111111111111109,
      "loss": 0.5943,
      "step": 68
    },
    {
      "epoch": 0.15307820299500832,
      "grad_norm": 0.09674172103404999,
      "learning_rate": 0.0001533333333333333,
      "loss": 0.7963,
      "step": 69
    },
    {
      "epoch": 0.1552967276760954,
      "grad_norm": 0.11833146214485168,
      "learning_rate": 0.00015555555555555554,
      "loss": 0.8705,
      "step": 70
    },
    {
      "epoch": 0.15751525235718247,
      "grad_norm": 0.08720320463180542,
      "learning_rate": 0.00015777777777777776,
      "loss": 0.7562,
      "step": 71
    },
    {
      "epoch": 0.15973377703826955,
      "grad_norm": 0.1389217972755432,
      "learning_rate": 0.00015999999999999999,
      "loss": 0.849,
      "step": 72
    },
    {
      "epoch": 0.16195230171935662,
      "grad_norm": 0.10427886247634888,
      "learning_rate": 0.0001622222222222222,
      "loss": 0.7749,
      "step": 73
    },
    {
      "epoch": 0.1641708264004437,
      "grad_norm": 0.07968995720148087,
      "learning_rate": 0.00016444444444444444,
      "loss": 0.7719,
      "step": 74
    },
    {
      "epoch": 0.16638935108153077,
      "grad_norm": 0.108415886759758,
      "learning_rate": 0.00016666666666666666,
      "loss": 0.8033,
      "step": 75
    },
    {
      "epoch": 0.16860787576261785,
      "grad_norm": 0.09191995859146118,
      "learning_rate": 0.00016888888888888889,
      "loss": 1.055,
      "step": 76
    },
    {
      "epoch": 0.17082640044370492,
      "grad_norm": 0.08926394581794739,
      "learning_rate": 0.0001711111111111111,
      "loss": 0.8719,
      "step": 77
    },
    {
      "epoch": 0.17304492512479203,
      "grad_norm": 0.0706048533320427,
      "learning_rate": 0.0001733333333333333,
      "loss": 0.737,
      "step": 78
    },
    {
      "epoch": 0.1752634498058791,
      "grad_norm": 0.09676831215620041,
      "learning_rate": 0.00017555555555555553,
      "loss": 0.9335,
      "step": 79
    },
    {
      "epoch": 0.17748197448696618,
      "grad_norm": 0.08885685354471207,
      "learning_rate": 0.00017777777777777776,
      "loss": 0.8137,
      "step": 80
    },
    {
      "epoch": 0.17970049916805325,
      "grad_norm": 0.08725814521312714,
      "learning_rate": 0.00017999999999999998,
      "loss": 0.8672,
      "step": 81
    },
    {
      "epoch": 0.18191902384914033,
      "grad_norm": 0.07098839432001114,
      "learning_rate": 0.0001822222222222222,
      "loss": 0.7363,
      "step": 82
    },
    {
      "epoch": 0.1841375485302274,
      "grad_norm": 0.11859265714883804,
      "learning_rate": 0.00018444444444444443,
      "loss": 0.9679,
      "step": 83
    },
    {
      "epoch": 0.18635607321131448,
      "grad_norm": 0.08069822937250137,
      "learning_rate": 0.00018666666666666666,
      "loss": 0.7235,
      "step": 84
    },
    {
      "epoch": 0.18857459789240155,
      "grad_norm": 0.2394903600215912,
      "learning_rate": 0.00018888888888888888,
      "loss": 0.9318,
      "step": 85
    },
    {
      "epoch": 0.19079312257348863,
      "grad_norm": 0.09466389566659927,
      "learning_rate": 0.0001911111111111111,
      "loss": 0.7504,
      "step": 86
    },
    {
      "epoch": 0.1930116472545757,
      "grad_norm": 0.08041612058877945,
      "learning_rate": 0.00019333333333333333,
      "loss": 0.805,
      "step": 87
    },
    {
      "epoch": 0.19523017193566278,
      "grad_norm": 0.10023625940084457,
      "learning_rate": 0.00019555555555555556,
      "loss": 0.8203,
      "step": 88
    },
    {
      "epoch": 0.19744869661674985,
      "grad_norm": 0.1126708909869194,
      "learning_rate": 0.00019777777777777776,
      "loss": 0.856,
      "step": 89
    },
    {
      "epoch": 0.19966722129783693,
      "grad_norm": 0.09988386929035187,
      "learning_rate": 0.00019999999999999998,
      "loss": 0.896,
      "step": 90
    },
    {
      "epoch": 0.201885745978924,
      "grad_norm": 0.11388827860355377,
      "learning_rate": 0.0002022222222222222,
      "loss": 1.0702,
      "step": 91
    },
    {
      "epoch": 0.2041042706600111,
      "grad_norm": 0.09045268595218658,
      "learning_rate": 0.00020444444444444443,
      "loss": 0.8529,
      "step": 92
    },
    {
      "epoch": 0.20632279534109818,
      "grad_norm": 0.07464978843927383,
      "learning_rate": 0.00020666666666666666,
      "loss": 0.9016,
      "step": 93
    },
    {
      "epoch": 0.20854132002218526,
      "grad_norm": 0.07931617647409439,
      "learning_rate": 0.00020888888888888888,
      "loss": 0.9406,
      "step": 94
    },
    {
      "epoch": 0.21075984470327233,
      "grad_norm": 0.11883913725614548,
      "learning_rate": 0.0002111111111111111,
      "loss": 0.8949,
      "step": 95
    },
    {
      "epoch": 0.2129783693843594,
      "grad_norm": 0.09190099686384201,
      "learning_rate": 0.00021333333333333333,
      "loss": 1.0113,
      "step": 96
    },
    {
      "epoch": 0.21519689406544648,
      "grad_norm": 0.09972347319126129,
      "learning_rate": 0.00021555555555555556,
      "loss": 0.7492,
      "step": 97
    },
    {
      "epoch": 0.21741541874653356,
      "grad_norm": 0.07947293668985367,
      "learning_rate": 0.00021777777777777778,
      "loss": 0.7641,
      "step": 98
    },
    {
      "epoch": 0.21963394342762063,
      "grad_norm": 0.08855797350406647,
      "learning_rate": 0.00021999999999999995,
      "loss": 0.9462,
      "step": 99
    },
    {
      "epoch": 0.2218524681087077,
      "grad_norm": 0.11437303572893143,
      "learning_rate": 0.00022222222222222218,
      "loss": 0.9529,
      "step": 100
    },
    {
      "epoch": 0.22407099278979478,
      "grad_norm": 0.11550132930278778,
      "learning_rate": 0.0002244444444444444,
      "loss": 0.7178,
      "step": 101
    },
    {
      "epoch": 0.22628951747088186,
      "grad_norm": 0.08455786854028702,
      "learning_rate": 0.00022666666666666663,
      "loss": 0.7973,
      "step": 102
    },
    {
      "epoch": 0.22850804215196893,
      "grad_norm": 0.10519180446863174,
      "learning_rate": 0.00022888888888888885,
      "loss": 0.8187,
      "step": 103
    },
    {
      "epoch": 0.230726566833056,
      "grad_norm": 0.13653722405433655,
      "learning_rate": 0.00023111111111111108,
      "loss": 0.8384,
      "step": 104
    },
    {
      "epoch": 0.23294509151414308,
      "grad_norm": 0.07808375358581543,
      "learning_rate": 0.0002333333333333333,
      "loss": 0.7744,
      "step": 105
    },
    {
      "epoch": 0.23516361619523019,
      "grad_norm": 0.1049056127667427,
      "learning_rate": 0.00023555555555555553,
      "loss": 0.7438,
      "step": 106
    },
    {
      "epoch": 0.23738214087631726,
      "grad_norm": 0.0943755954504013,
      "learning_rate": 0.00023777777777777775,
      "loss": 0.7704,
      "step": 107
    },
    {
      "epoch": 0.23960066555740434,
      "grad_norm": 0.1024504229426384,
      "learning_rate": 0.00023999999999999998,
      "loss": 0.8376,
      "step": 108
    },
    {
      "epoch": 0.2418191902384914,
      "grad_norm": 0.08710005134344101,
      "learning_rate": 0.0002422222222222222,
      "loss": 0.9358,
      "step": 109
    },
    {
      "epoch": 0.24403771491957849,
      "grad_norm": 0.08408643305301666,
      "learning_rate": 0.00024444444444444443,
      "loss": 0.8839,
      "step": 110
    },
    {
      "epoch": 0.24625623960066556,
      "grad_norm": 0.07873968780040741,
      "learning_rate": 0.0002466666666666666,
      "loss": 0.6693,
      "step": 111
    },
    {
      "epoch": 0.24847476428175264,
      "grad_norm": 0.11656618118286133,
      "learning_rate": 0.0002488888888888889,
      "loss": 0.7551,
      "step": 112
    },
    {
      "epoch": 0.2506932889628397,
      "grad_norm": 0.11387936025857925,
      "learning_rate": 0.0002511111111111111,
      "loss": 0.8611,
      "step": 113
    },
    {
      "epoch": 0.2529118136439268,
      "grad_norm": 0.09405595064163208,
      "learning_rate": 0.00025333333333333333,
      "loss": 0.9272,
      "step": 114
    },
    {
      "epoch": 0.25513033832501386,
      "grad_norm": 0.06243271008133888,
      "learning_rate": 0.00025555555555555553,
      "loss": 0.7602,
      "step": 115
    },
    {
      "epoch": 0.25734886300610094,
      "grad_norm": 0.1395256668329239,
      "learning_rate": 0.0002577777777777778,
      "loss": 0.937,
      "step": 116
    },
    {
      "epoch": 0.259567387687188,
      "grad_norm": 0.13361899554729462,
      "learning_rate": 0.00026,
      "loss": 0.8351,
      "step": 117
    },
    {
      "epoch": 0.2617859123682751,
      "grad_norm": 0.09131667017936707,
      "learning_rate": 0.00026222222222222223,
      "loss": 0.78,
      "step": 118
    },
    {
      "epoch": 0.26400443704936216,
      "grad_norm": 0.09531912207603455,
      "learning_rate": 0.00026444444444444443,
      "loss": 0.9263,
      "step": 119
    },
    {
      "epoch": 0.26622296173044924,
      "grad_norm": 0.11719240248203278,
      "learning_rate": 0.0002666666666666666,
      "loss": 1.1007,
      "step": 120
    },
    {
      "epoch": 0.2684414864115363,
      "grad_norm": 0.09963520616292953,
      "learning_rate": 0.0002688888888888889,
      "loss": 0.8582,
      "step": 121
    },
    {
      "epoch": 0.2706600110926234,
      "grad_norm": 0.10103681683540344,
      "learning_rate": 0.0002711111111111111,
      "loss": 0.9504,
      "step": 122
    },
    {
      "epoch": 0.27287853577371046,
      "grad_norm": 0.084881491959095,
      "learning_rate": 0.00027333333333333333,
      "loss": 0.6794,
      "step": 123
    },
    {
      "epoch": 0.27509706045479754,
      "grad_norm": 0.07723584771156311,
      "learning_rate": 0.0002755555555555555,
      "loss": 0.8862,
      "step": 124
    },
    {
      "epoch": 0.2773155851358846,
      "grad_norm": 0.09102759510278702,
      "learning_rate": 0.0002777777777777778,
      "loss": 0.8913,
      "step": 125
    },
    {
      "epoch": 0.2795341098169717,
      "grad_norm": 0.09926798194646835,
      "learning_rate": 0.00028,
      "loss": 0.6647,
      "step": 126
    },
    {
      "epoch": 0.2817526344980588,
      "grad_norm": 0.08959900587797165,
      "learning_rate": 0.00028222222222222223,
      "loss": 0.7343,
      "step": 127
    },
    {
      "epoch": 0.2839711591791459,
      "grad_norm": 0.072727270424366,
      "learning_rate": 0.0002844444444444444,
      "loss": 0.7734,
      "step": 128
    },
    {
      "epoch": 0.28618968386023297,
      "grad_norm": 0.07037808746099472,
      "learning_rate": 0.0002866666666666667,
      "loss": 0.8324,
      "step": 129
    },
    {
      "epoch": 0.28840820854132004,
      "grad_norm": 0.07432516664266586,
      "learning_rate": 0.0002888888888888888,
      "loss": 0.8081,
      "step": 130
    },
    {
      "epoch": 0.2906267332224071,
      "grad_norm": 0.10861325263977051,
      "learning_rate": 0.0002911111111111111,
      "loss": 0.9382,
      "step": 131
    },
    {
      "epoch": 0.2928452579034942,
      "grad_norm": 0.1049029752612114,
      "learning_rate": 0.00029333333333333327,
      "loss": 0.6703,
      "step": 132
    },
    {
      "epoch": 0.29506378258458127,
      "grad_norm": 0.09810491651296616,
      "learning_rate": 0.0002955555555555555,
      "loss": 0.9657,
      "step": 133
    },
    {
      "epoch": 0.29728230726566834,
      "grad_norm": 0.07675714790821075,
      "learning_rate": 0.0002977777777777777,
      "loss": 0.766,
      "step": 134
    },
    {
      "epoch": 0.2995008319467554,
      "grad_norm": 0.1567544937133789,
      "learning_rate": 0.0003,
      "loss": 0.7325,
      "step": 135
    },
    {
      "epoch": 0.3017193566278425,
      "grad_norm": 0.09014812856912613,
      "learning_rate": 0.00029975308641975307,
      "loss": 0.7128,
      "step": 136
    },
    {
      "epoch": 0.30393788130892957,
      "grad_norm": 0.09727205336093903,
      "learning_rate": 0.00029950617283950616,
      "loss": 0.984,
      "step": 137
    },
    {
      "epoch": 0.30615640599001664,
      "grad_norm": 0.09497466683387756,
      "learning_rate": 0.00029925925925925926,
      "loss": 0.814,
      "step": 138
    },
    {
      "epoch": 0.3083749306711037,
      "grad_norm": 0.09212425351142883,
      "learning_rate": 0.0002990123456790123,
      "loss": 0.7385,
      "step": 139
    },
    {
      "epoch": 0.3105934553521908,
      "grad_norm": 0.08676060289144516,
      "learning_rate": 0.0002987654320987654,
      "loss": 0.8301,
      "step": 140
    },
    {
      "epoch": 0.31281198003327787,
      "grad_norm": 0.10276718437671661,
      "learning_rate": 0.0002985185185185185,
      "loss": 0.6957,
      "step": 141
    },
    {
      "epoch": 0.31503050471436495,
      "grad_norm": 0.08550626784563065,
      "learning_rate": 0.0002982716049382716,
      "loss": 0.8013,
      "step": 142
    },
    {
      "epoch": 0.317249029395452,
      "grad_norm": 0.09392906725406647,
      "learning_rate": 0.0002980246913580247,
      "loss": 0.8495,
      "step": 143
    },
    {
      "epoch": 0.3194675540765391,
      "grad_norm": 0.09811647981405258,
      "learning_rate": 0.0002977777777777777,
      "loss": 0.7348,
      "step": 144
    },
    {
      "epoch": 0.32168607875762617,
      "grad_norm": 0.09334246069192886,
      "learning_rate": 0.00029753086419753087,
      "loss": 0.7606,
      "step": 145
    },
    {
      "epoch": 0.32390460343871325,
      "grad_norm": 0.10100342333316803,
      "learning_rate": 0.0002972839506172839,
      "loss": 0.7488,
      "step": 146
    },
    {
      "epoch": 0.3261231281198003,
      "grad_norm": 0.07997012883424759,
      "learning_rate": 0.000297037037037037,
      "loss": 0.6174,
      "step": 147
    },
    {
      "epoch": 0.3283416528008874,
      "grad_norm": 0.07816442847251892,
      "learning_rate": 0.0002967901234567901,
      "loss": 0.7379,
      "step": 148
    },
    {
      "epoch": 0.33056017748197447,
      "grad_norm": 0.07927139848470688,
      "learning_rate": 0.0002965432098765432,
      "loss": 0.8353,
      "step": 149
    },
    {
      "epoch": 0.33277870216306155,
      "grad_norm": 0.10762617737054825,
      "learning_rate": 0.00029629629629629624,
      "loss": 0.8179,
      "step": 150
    },
    {
      "epoch": 0.3349972268441486,
      "grad_norm": 0.0797627717256546,
      "learning_rate": 0.00029604938271604933,
      "loss": 0.7837,
      "step": 151
    },
    {
      "epoch": 0.3372157515252357,
      "grad_norm": 0.1313309669494629,
      "learning_rate": 0.0002958024691358025,
      "loss": 0.9239,
      "step": 152
    },
    {
      "epoch": 0.33943427620632277,
      "grad_norm": 0.1220150887966156,
      "learning_rate": 0.0002955555555555555,
      "loss": 0.8998,
      "step": 153
    },
    {
      "epoch": 0.34165280088740985,
      "grad_norm": 0.09039723128080368,
      "learning_rate": 0.0002953086419753086,
      "loss": 0.8919,
      "step": 154
    },
    {
      "epoch": 0.343871325568497,
      "grad_norm": 0.11235062777996063,
      "learning_rate": 0.0002950617283950617,
      "loss": 0.8849,
      "step": 155
    },
    {
      "epoch": 0.34608985024958405,
      "grad_norm": 0.07742858678102493,
      "learning_rate": 0.0002948148148148148,
      "loss": 0.9297,
      "step": 156
    },
    {
      "epoch": 0.3483083749306711,
      "grad_norm": 0.15448038280010223,
      "learning_rate": 0.00029456790123456785,
      "loss": 0.8243,
      "step": 157
    },
    {
      "epoch": 0.3505268996117582,
      "grad_norm": 0.08074051141738892,
      "learning_rate": 0.00029432098765432095,
      "loss": 0.8444,
      "step": 158
    },
    {
      "epoch": 0.3527454242928453,
      "grad_norm": 0.07954861223697662,
      "learning_rate": 0.00029407407407407404,
      "loss": 0.6593,
      "step": 159
    },
    {
      "epoch": 0.35496394897393235,
      "grad_norm": 0.09272817522287369,
      "learning_rate": 0.00029382716049382714,
      "loss": 0.6942,
      "step": 160
    },
    {
      "epoch": 0.35718247365501943,
      "grad_norm": 0.08088618516921997,
      "learning_rate": 0.00029358024691358023,
      "loss": 0.7825,
      "step": 161
    },
    {
      "epoch": 0.3594009983361065,
      "grad_norm": 0.0957372784614563,
      "learning_rate": 0.00029333333333333327,
      "loss": 0.8973,
      "step": 162
    },
    {
      "epoch": 0.3616195230171936,
      "grad_norm": 0.09212084114551544,
      "learning_rate": 0.0002930864197530864,
      "loss": 0.779,
      "step": 163
    },
    {
      "epoch": 0.36383804769828065,
      "grad_norm": 0.0695093646645546,
      "learning_rate": 0.00029283950617283946,
      "loss": 0.6965,
      "step": 164
    },
    {
      "epoch": 0.36605657237936773,
      "grad_norm": 0.07858648896217346,
      "learning_rate": 0.00029259259259259256,
      "loss": 0.6933,
      "step": 165
    },
    {
      "epoch": 0.3682750970604548,
      "grad_norm": 0.09808747470378876,
      "learning_rate": 0.00029234567901234565,
      "loss": 0.9202,
      "step": 166
    },
    {
      "epoch": 0.3704936217415419,
      "grad_norm": 0.09445574134588242,
      "learning_rate": 0.00029209876543209875,
      "loss": 0.8204,
      "step": 167
    },
    {
      "epoch": 0.37271214642262895,
      "grad_norm": 0.06962455809116364,
      "learning_rate": 0.00029185185185185184,
      "loss": 0.7347,
      "step": 168
    },
    {
      "epoch": 0.37493067110371603,
      "grad_norm": 0.08811014890670776,
      "learning_rate": 0.0002916049382716049,
      "loss": 0.6481,
      "step": 169
    },
    {
      "epoch": 0.3771491957848031,
      "grad_norm": 0.0889468789100647,
      "learning_rate": 0.000291358024691358,
      "loss": 0.8307,
      "step": 170
    },
    {
      "epoch": 0.3793677204658902,
      "grad_norm": 0.13039159774780273,
      "learning_rate": 0.0002911111111111111,
      "loss": 0.8903,
      "step": 171
    },
    {
      "epoch": 0.38158624514697725,
      "grad_norm": 0.09932645410299301,
      "learning_rate": 0.00029086419753086417,
      "loss": 0.7281,
      "step": 172
    },
    {
      "epoch": 0.38380476982806433,
      "grad_norm": 0.08021000027656555,
      "learning_rate": 0.00029061728395061726,
      "loss": 0.8244,
      "step": 173
    },
    {
      "epoch": 0.3860232945091514,
      "grad_norm": 0.0873599424958229,
      "learning_rate": 0.00029037037037037036,
      "loss": 0.9276,
      "step": 174
    },
    {
      "epoch": 0.3882418191902385,
      "grad_norm": 0.09124568104743958,
      "learning_rate": 0.00029012345679012345,
      "loss": 0.8127,
      "step": 175
    },
    {
      "epoch": 0.39046034387132555,
      "grad_norm": 0.08030134439468384,
      "learning_rate": 0.0002898765432098765,
      "loss": 0.7939,
      "step": 176
    },
    {
      "epoch": 0.39267886855241263,
      "grad_norm": 0.10353704541921616,
      "learning_rate": 0.0002896296296296296,
      "loss": 0.8653,
      "step": 177
    },
    {
      "epoch": 0.3948973932334997,
      "grad_norm": 0.0833268091082573,
      "learning_rate": 0.0002893827160493827,
      "loss": 0.731,
      "step": 178
    },
    {
      "epoch": 0.3971159179145868,
      "grad_norm": 0.06832997500896454,
      "learning_rate": 0.0002891358024691358,
      "loss": 0.9195,
      "step": 179
    },
    {
      "epoch": 0.39933444259567386,
      "grad_norm": 0.08206876367330551,
      "learning_rate": 0.0002888888888888888,
      "loss": 0.7706,
      "step": 180
    },
    {
      "epoch": 0.40155296727676093,
      "grad_norm": 0.10592249780893326,
      "learning_rate": 0.00028864197530864197,
      "loss": 0.8889,
      "step": 181
    },
    {
      "epoch": 0.403771491957848,
      "grad_norm": 0.10580477118492126,
      "learning_rate": 0.00028839506172839507,
      "loss": 0.6867,
      "step": 182
    },
    {
      "epoch": 0.4059900166389351,
      "grad_norm": 0.10561928898096085,
      "learning_rate": 0.0002881481481481481,
      "loss": 0.751,
      "step": 183
    },
    {
      "epoch": 0.4082085413200222,
      "grad_norm": 0.08269621431827545,
      "learning_rate": 0.0002879012345679012,
      "loss": 0.8422,
      "step": 184
    },
    {
      "epoch": 0.4104270660011093,
      "grad_norm": 0.0739932656288147,
      "learning_rate": 0.0002876543209876543,
      "loss": 0.6544,
      "step": 185
    },
    {
      "epoch": 0.41264559068219636,
      "grad_norm": 0.1316475123167038,
      "learning_rate": 0.0002874074074074074,
      "loss": 0.8873,
      "step": 186
    },
    {
      "epoch": 0.41486411536328344,
      "grad_norm": 0.13711637258529663,
      "learning_rate": 0.00028716049382716043,
      "loss": 0.8353,
      "step": 187
    },
    {
      "epoch": 0.4170826400443705,
      "grad_norm": 0.08551054447889328,
      "learning_rate": 0.00028691358024691353,
      "loss": 0.7651,
      "step": 188
    },
    {
      "epoch": 0.4193011647254576,
      "grad_norm": 0.10682949423789978,
      "learning_rate": 0.0002866666666666667,
      "loss": 1.0562,
      "step": 189
    },
    {
      "epoch": 0.42151968940654466,
      "grad_norm": 0.08605273813009262,
      "learning_rate": 0.0002864197530864197,
      "loss": 0.7269,
      "step": 190
    },
    {
      "epoch": 0.42373821408763174,
      "grad_norm": 0.10960530489683151,
      "learning_rate": 0.0002861728395061728,
      "loss": 0.6448,
      "step": 191
    },
    {
      "epoch": 0.4259567387687188,
      "grad_norm": 0.08502240478992462,
      "learning_rate": 0.0002859259259259259,
      "loss": 0.8436,
      "step": 192
    },
    {
      "epoch": 0.4281752634498059,
      "grad_norm": 0.06750859320163727,
      "learning_rate": 0.000285679012345679,
      "loss": 0.7524,
      "step": 193
    },
    {
      "epoch": 0.43039378813089296,
      "grad_norm": 0.10397569090127945,
      "learning_rate": 0.00028543209876543204,
      "loss": 0.6604,
      "step": 194
    },
    {
      "epoch": 0.43261231281198004,
      "grad_norm": 0.11751885712146759,
      "learning_rate": 0.00028518518518518514,
      "loss": 0.7321,
      "step": 195
    },
    {
      "epoch": 0.4348308374930671,
      "grad_norm": 0.09464067220687866,
      "learning_rate": 0.00028493827160493824,
      "loss": 1.0175,
      "step": 196
    },
    {
      "epoch": 0.4370493621741542,
      "grad_norm": 0.09692603349685669,
      "learning_rate": 0.00028469135802469133,
      "loss": 0.8274,
      "step": 197
    },
    {
      "epoch": 0.43926788685524126,
      "grad_norm": 0.09040063619613647,
      "learning_rate": 0.0002844444444444444,
      "loss": 0.8045,
      "step": 198
    },
    {
      "epoch": 0.44148641153632834,
      "grad_norm": 0.09752462059259415,
      "learning_rate": 0.0002841975308641975,
      "loss": 0.8737,
      "step": 199
    },
    {
      "epoch": 0.4437049362174154,
      "grad_norm": 0.07462728768587112,
      "learning_rate": 0.0002839506172839506,
      "loss": 0.7814,
      "step": 200
    },
    {
      "epoch": 0.4459234608985025,
      "grad_norm": 0.09206775575876236,
      "learning_rate": 0.00028370370370370366,
      "loss": 0.7737,
      "step": 201
    },
    {
      "epoch": 0.44814198557958956,
      "grad_norm": 0.08017104864120483,
      "learning_rate": 0.00028345679012345675,
      "loss": 0.6827,
      "step": 202
    },
    {
      "epoch": 0.45036051026067664,
      "grad_norm": 0.094303660094738,
      "learning_rate": 0.00028320987654320985,
      "loss": 0.8232,
      "step": 203
    },
    {
      "epoch": 0.4525790349417637,
      "grad_norm": 0.09772761911153793,
      "learning_rate": 0.00028296296296296294,
      "loss": 1.0997,
      "step": 204
    },
    {
      "epoch": 0.4547975596228508,
      "grad_norm": 0.16040252149105072,
      "learning_rate": 0.00028271604938271604,
      "loss": 0.9595,
      "step": 205
    },
    {
      "epoch": 0.45701608430393786,
      "grad_norm": 0.08671429753303528,
      "learning_rate": 0.0002824691358024691,
      "loss": 0.7408,
      "step": 206
    },
    {
      "epoch": 0.45923460898502494,
      "grad_norm": 0.0676223635673523,
      "learning_rate": 0.00028222222222222223,
      "loss": 0.8749,
      "step": 207
    },
    {
      "epoch": 0.461453133666112,
      "grad_norm": 0.08092256635427475,
      "learning_rate": 0.00028197530864197527,
      "loss": 0.7467,
      "step": 208
    },
    {
      "epoch": 0.4636716583471991,
      "grad_norm": 0.13109372556209564,
      "learning_rate": 0.00028172839506172836,
      "loss": 0.6758,
      "step": 209
    },
    {
      "epoch": 0.46589018302828616,
      "grad_norm": 0.08598847687244415,
      "learning_rate": 0.00028148148148148146,
      "loss": 0.8271,
      "step": 210
    },
    {
      "epoch": 0.46810870770937324,
      "grad_norm": 0.06640809774398804,
      "learning_rate": 0.00028123456790123455,
      "loss": 0.7,
      "step": 211
    },
    {
      "epoch": 0.47032723239046037,
      "grad_norm": 0.08522395044565201,
      "learning_rate": 0.00028098765432098765,
      "loss": 0.8102,
      "step": 212
    },
    {
      "epoch": 0.47254575707154745,
      "grad_norm": 0.08832666277885437,
      "learning_rate": 0.0002807407407407407,
      "loss": 0.8195,
      "step": 213
    },
    {
      "epoch": 0.4747642817526345,
      "grad_norm": 0.10240650922060013,
      "learning_rate": 0.0002804938271604938,
      "loss": 0.8069,
      "step": 214
    },
    {
      "epoch": 0.4769828064337216,
      "grad_norm": 0.16724427044391632,
      "learning_rate": 0.0002802469135802469,
      "loss": 0.763,
      "step": 215
    },
    {
      "epoch": 0.47920133111480867,
      "grad_norm": 0.1139446347951889,
      "learning_rate": 0.00028,
      "loss": 0.819,
      "step": 216
    },
    {
      "epoch": 0.48141985579589575,
      "grad_norm": 0.08204243332147598,
      "learning_rate": 0.00027975308641975307,
      "loss": 0.7591,
      "step": 217
    },
    {
      "epoch": 0.4836383804769828,
      "grad_norm": 0.08862531185150146,
      "learning_rate": 0.00027950617283950617,
      "loss": 0.7196,
      "step": 218
    },
    {
      "epoch": 0.4858569051580699,
      "grad_norm": 0.0840478166937828,
      "learning_rate": 0.00027925925925925926,
      "loss": 0.7027,
      "step": 219
    },
    {
      "epoch": 0.48807542983915697,
      "grad_norm": 0.07281976193189621,
      "learning_rate": 0.0002790123456790123,
      "loss": 0.7365,
      "step": 220
    },
    {
      "epoch": 0.49029395452024405,
      "grad_norm": 0.08599011600017548,
      "learning_rate": 0.0002787654320987654,
      "loss": 0.87,
      "step": 221
    },
    {
      "epoch": 0.4925124792013311,
      "grad_norm": 0.09043736010789871,
      "learning_rate": 0.0002785185185185185,
      "loss": 0.8237,
      "step": 222
    },
    {
      "epoch": 0.4947310038824182,
      "grad_norm": 0.068006731569767,
      "learning_rate": 0.0002782716049382716,
      "loss": 0.8547,
      "step": 223
    },
    {
      "epoch": 0.49694952856350527,
      "grad_norm": 0.08268806338310242,
      "learning_rate": 0.00027802469135802463,
      "loss": 0.8657,
      "step": 224
    },
    {
      "epoch": 0.49916805324459235,
      "grad_norm": 0.1173485815525055,
      "learning_rate": 0.0002777777777777778,
      "loss": 0.7301,
      "step": 225
    },
    {
      "epoch": 0.5013865779256794,
      "grad_norm": 0.1006660908460617,
      "learning_rate": 0.00027753086419753087,
      "loss": 0.9353,
      "step": 226
    },
    {
      "epoch": 0.5036051026067665,
      "grad_norm": 0.10356108844280243,
      "learning_rate": 0.0002772839506172839,
      "loss": 0.9999,
      "step": 227
    },
    {
      "epoch": 0.5058236272878536,
      "grad_norm": 0.06711801886558533,
      "learning_rate": 0.000277037037037037,
      "loss": 0.7482,
      "step": 228
    },
    {
      "epoch": 0.5080421519689406,
      "grad_norm": 0.09008052200078964,
      "learning_rate": 0.0002767901234567901,
      "loss": 0.8489,
      "step": 229
    },
    {
      "epoch": 0.5102606766500277,
      "grad_norm": 0.07854015380144119,
      "learning_rate": 0.0002765432098765432,
      "loss": 0.7951,
      "step": 230
    },
    {
      "epoch": 0.5124792013311148,
      "grad_norm": 0.13175378739833832,
      "learning_rate": 0.00027629629629629624,
      "loss": 0.6806,
      "step": 231
    },
    {
      "epoch": 0.5146977260122019,
      "grad_norm": 0.09410752356052399,
      "learning_rate": 0.00027604938271604934,
      "loss": 0.8309,
      "step": 232
    },
    {
      "epoch": 0.516916250693289,
      "grad_norm": 0.08800879120826721,
      "learning_rate": 0.00027580246913580243,
      "loss": 0.8433,
      "step": 233
    },
    {
      "epoch": 0.519134775374376,
      "grad_norm": 0.0872422531247139,
      "learning_rate": 0.0002755555555555555,
      "loss": 0.732,
      "step": 234
    },
    {
      "epoch": 0.5213533000554631,
      "grad_norm": 0.11477510631084442,
      "learning_rate": 0.0002753086419753086,
      "loss": 0.6452,
      "step": 235
    },
    {
      "epoch": 0.5235718247365502,
      "grad_norm": 0.08534936606884003,
      "learning_rate": 0.0002750617283950617,
      "loss": 0.8184,
      "step": 236
    },
    {
      "epoch": 0.5257903494176372,
      "grad_norm": 0.08374905586242676,
      "learning_rate": 0.0002748148148148148,
      "loss": 0.8148,
      "step": 237
    },
    {
      "epoch": 0.5280088740987243,
      "grad_norm": 0.09268096834421158,
      "learning_rate": 0.00027456790123456785,
      "loss": 0.8585,
      "step": 238
    },
    {
      "epoch": 0.5302273987798114,
      "grad_norm": 0.0944875031709671,
      "learning_rate": 0.00027432098765432095,
      "loss": 0.7124,
      "step": 239
    },
    {
      "epoch": 0.5324459234608985,
      "grad_norm": 0.07623323798179626,
      "learning_rate": 0.00027407407407407404,
      "loss": 0.6666,
      "step": 240
    },
    {
      "epoch": 0.5346644481419855,
      "grad_norm": 0.08309529721736908,
      "learning_rate": 0.00027382716049382714,
      "loss": 0.8188,
      "step": 241
    },
    {
      "epoch": 0.5368829728230726,
      "grad_norm": 0.08109453320503235,
      "learning_rate": 0.00027358024691358023,
      "loss": 0.7225,
      "step": 242
    },
    {
      "epoch": 0.5391014975041597,
      "grad_norm": 0.0769454687833786,
      "learning_rate": 0.00027333333333333333,
      "loss": 0.6795,
      "step": 243
    },
    {
      "epoch": 0.5413200221852468,
      "grad_norm": 0.08318306505680084,
      "learning_rate": 0.0002730864197530864,
      "loss": 0.7225,
      "step": 244
    },
    {
      "epoch": 0.5435385468663338,
      "grad_norm": 0.08993634581565857,
      "learning_rate": 0.00027283950617283946,
      "loss": 0.7906,
      "step": 245
    },
    {
      "epoch": 0.5457570715474209,
      "grad_norm": 0.10697447508573532,
      "learning_rate": 0.00027259259259259256,
      "loss": 0.6816,
      "step": 246
    },
    {
      "epoch": 0.547975596228508,
      "grad_norm": 0.06921118497848511,
      "learning_rate": 0.00027234567901234565,
      "loss": 0.776,
      "step": 247
    },
    {
      "epoch": 0.5501941209095951,
      "grad_norm": 0.08231881260871887,
      "learning_rate": 0.00027209876543209875,
      "loss": 0.6573,
      "step": 248
    },
    {
      "epoch": 0.5524126455906821,
      "grad_norm": 0.06871269643306732,
      "learning_rate": 0.00027185185185185184,
      "loss": 0.6738,
      "step": 249
    },
    {
      "epoch": 0.5546311702717692,
      "grad_norm": 0.08890746533870697,
      "learning_rate": 0.0002716049382716049,
      "loss": 0.6784,
      "step": 250
    },
    {
      "epoch": 0.5568496949528563,
      "grad_norm": 0.10056876391172409,
      "learning_rate": 0.00027135802469135803,
      "loss": 0.8627,
      "step": 251
    },
    {
      "epoch": 0.5590682196339434,
      "grad_norm": 0.09984014183282852,
      "learning_rate": 0.0002711111111111111,
      "loss": 0.7072,
      "step": 252
    },
    {
      "epoch": 0.5612867443150305,
      "grad_norm": 0.07467436790466309,
      "learning_rate": 0.00027086419753086417,
      "loss": 0.7884,
      "step": 253
    },
    {
      "epoch": 0.5635052689961176,
      "grad_norm": 0.08617124706506729,
      "learning_rate": 0.00027061728395061727,
      "loss": 0.8914,
      "step": 254
    },
    {
      "epoch": 0.5657237936772047,
      "grad_norm": 0.0683143213391304,
      "learning_rate": 0.00027037037037037036,
      "loss": 0.7005,
      "step": 255
    },
    {
      "epoch": 0.5679423183582918,
      "grad_norm": 0.08523419499397278,
      "learning_rate": 0.00027012345679012346,
      "loss": 0.6864,
      "step": 256
    },
    {
      "epoch": 0.5701608430393789,
      "grad_norm": 0.09049207717180252,
      "learning_rate": 0.0002698765432098765,
      "loss": 0.7468,
      "step": 257
    },
    {
      "epoch": 0.5723793677204659,
      "grad_norm": 0.11385578662157059,
      "learning_rate": 0.0002696296296296296,
      "loss": 0.8582,
      "step": 258
    },
    {
      "epoch": 0.574597892401553,
      "grad_norm": 0.0704026147723198,
      "learning_rate": 0.0002693827160493827,
      "loss": 0.7709,
      "step": 259
    },
    {
      "epoch": 0.5768164170826401,
      "grad_norm": 0.06082779914140701,
      "learning_rate": 0.0002691358024691358,
      "loss": 0.7866,
      "step": 260
    },
    {
      "epoch": 0.5790349417637272,
      "grad_norm": 0.1361791342496872,
      "learning_rate": 0.0002688888888888889,
      "loss": 0.9895,
      "step": 261
    },
    {
      "epoch": 0.5812534664448142,
      "grad_norm": 0.07748665660619736,
      "learning_rate": 0.00026864197530864197,
      "loss": 0.602,
      "step": 262
    },
    {
      "epoch": 0.5834719911259013,
      "grad_norm": 0.08939169347286224,
      "learning_rate": 0.000268395061728395,
      "loss": 0.7325,
      "step": 263
    },
    {
      "epoch": 0.5856905158069884,
      "grad_norm": 0.12888678908348083,
      "learning_rate": 0.0002681481481481481,
      "loss": 0.7063,
      "step": 264
    },
    {
      "epoch": 0.5879090404880755,
      "grad_norm": 0.0812646746635437,
      "learning_rate": 0.0002679012345679012,
      "loss": 0.7479,
      "step": 265
    },
    {
      "epoch": 0.5901275651691625,
      "grad_norm": 0.0976184830069542,
      "learning_rate": 0.0002676543209876543,
      "loss": 0.8715,
      "step": 266
    },
    {
      "epoch": 0.5923460898502496,
      "grad_norm": 0.08552854508161545,
      "learning_rate": 0.0002674074074074074,
      "loss": 0.7335,
      "step": 267
    },
    {
      "epoch": 0.5945646145313367,
      "grad_norm": 0.08157272636890411,
      "learning_rate": 0.00026716049382716043,
      "loss": 0.9132,
      "step": 268
    },
    {
      "epoch": 0.5967831392124238,
      "grad_norm": 0.0835312157869339,
      "learning_rate": 0.0002669135802469136,
      "loss": 0.9108,
      "step": 269
    },
    {
      "epoch": 0.5990016638935108,
      "grad_norm": 0.09799845516681671,
      "learning_rate": 0.0002666666666666666,
      "loss": 0.7954,
      "step": 270
    },
    {
      "epoch": 0.6012201885745979,
      "grad_norm": 0.1178496852517128,
      "learning_rate": 0.0002664197530864197,
      "loss": 0.9214,
      "step": 271
    },
    {
      "epoch": 0.603438713255685,
      "grad_norm": 0.08210403472185135,
      "learning_rate": 0.0002661728395061728,
      "loss": 0.5819,
      "step": 272
    },
    {
      "epoch": 0.6056572379367721,
      "grad_norm": 0.08564604073762894,
      "learning_rate": 0.0002659259259259259,
      "loss": 0.8272,
      "step": 273
    },
    {
      "epoch": 0.6078757626178591,
      "grad_norm": 0.07128099352121353,
      "learning_rate": 0.000265679012345679,
      "loss": 0.8959,
      "step": 274
    },
    {
      "epoch": 0.6100942872989462,
      "grad_norm": 0.057627636939287186,
      "learning_rate": 0.00026543209876543205,
      "loss": 0.5865,
      "step": 275
    },
    {
      "epoch": 0.6123128119800333,
      "grad_norm": 0.06789959222078323,
      "learning_rate": 0.00026518518518518514,
      "loss": 0.7992,
      "step": 276
    },
    {
      "epoch": 0.6145313366611204,
      "grad_norm": 0.08316505700349808,
      "learning_rate": 0.00026493827160493824,
      "loss": 0.7939,
      "step": 277
    },
    {
      "epoch": 0.6167498613422074,
      "grad_norm": 0.08976881206035614,
      "learning_rate": 0.00026469135802469133,
      "loss": 0.7022,
      "step": 278
    },
    {
      "epoch": 0.6189683860232945,
      "grad_norm": 0.06505075842142105,
      "learning_rate": 0.00026444444444444443,
      "loss": 0.7061,
      "step": 279
    },
    {
      "epoch": 0.6211869107043816,
      "grad_norm": 0.0766439363360405,
      "learning_rate": 0.0002641975308641975,
      "loss": 0.6642,
      "step": 280
    },
    {
      "epoch": 0.6234054353854687,
      "grad_norm": 0.08418625593185425,
      "learning_rate": 0.0002639506172839506,
      "loss": 0.7111,
      "step": 281
    },
    {
      "epoch": 0.6256239600665557,
      "grad_norm": 0.09453851729631424,
      "learning_rate": 0.00026370370370370366,
      "loss": 0.6864,
      "step": 282
    },
    {
      "epoch": 0.6278424847476428,
      "grad_norm": 0.06271867454051971,
      "learning_rate": 0.00026345679012345675,
      "loss": 0.5599,
      "step": 283
    },
    {
      "epoch": 0.6300610094287299,
      "grad_norm": 0.09013552218675613,
      "learning_rate": 0.00026320987654320985,
      "loss": 0.7017,
      "step": 284
    },
    {
      "epoch": 0.632279534109817,
      "grad_norm": 0.07347094267606735,
      "learning_rate": 0.00026296296296296294,
      "loss": 0.7372,
      "step": 285
    },
    {
      "epoch": 0.634498058790904,
      "grad_norm": 0.09328638017177582,
      "learning_rate": 0.00026271604938271604,
      "loss": 0.7583,
      "step": 286
    },
    {
      "epoch": 0.6367165834719911,
      "grad_norm": 0.07239331305027008,
      "learning_rate": 0.00026246913580246913,
      "loss": 0.7858,
      "step": 287
    },
    {
      "epoch": 0.6389351081530782,
      "grad_norm": 0.15532372891902924,
      "learning_rate": 0.00026222222222222223,
      "loss": 0.7438,
      "step": 288
    },
    {
      "epoch": 0.6411536328341653,
      "grad_norm": 0.061730869114398956,
      "learning_rate": 0.00026197530864197527,
      "loss": 0.8308,
      "step": 289
    },
    {
      "epoch": 0.6433721575152523,
      "grad_norm": 0.19147923588752747,
      "learning_rate": 0.00026172839506172837,
      "loss": 0.8325,
      "step": 290
    },
    {
      "epoch": 0.6455906821963394,
      "grad_norm": 0.08945952355861664,
      "learning_rate": 0.00026148148148148146,
      "loss": 0.874,
      "step": 291
    },
    {
      "epoch": 0.6478092068774265,
      "grad_norm": 0.0672970786690712,
      "learning_rate": 0.00026123456790123456,
      "loss": 0.846,
      "step": 292
    },
    {
      "epoch": 0.6500277315585136,
      "grad_norm": 0.08254943042993546,
      "learning_rate": 0.0002609876543209876,
      "loss": 0.7758,
      "step": 293
    },
    {
      "epoch": 0.6522462562396006,
      "grad_norm": 0.08533348143100739,
      "learning_rate": 0.0002607407407407407,
      "loss": 0.7121,
      "step": 294
    },
    {
      "epoch": 0.6544647809206877,
      "grad_norm": 0.06619753688573837,
      "learning_rate": 0.00026049382716049384,
      "loss": 0.5883,
      "step": 295
    },
    {
      "epoch": 0.6566833056017748,
      "grad_norm": 0.09811896085739136,
      "learning_rate": 0.0002602469135802469,
      "loss": 0.8716,
      "step": 296
    },
    {
      "epoch": 0.6589018302828619,
      "grad_norm": 0.08157233893871307,
      "learning_rate": 0.00026,
      "loss": 0.681,
      "step": 297
    },
    {
      "epoch": 0.6611203549639489,
      "grad_norm": 0.06866102665662766,
      "learning_rate": 0.00025975308641975307,
      "loss": 0.709,
      "step": 298
    },
    {
      "epoch": 0.663338879645036,
      "grad_norm": 0.07522811740636826,
      "learning_rate": 0.00025950617283950617,
      "loss": 0.9192,
      "step": 299
    },
    {
      "epoch": 0.6655574043261231,
      "grad_norm": 0.08763494342565536,
      "learning_rate": 0.0002592592592592592,
      "loss": 0.9086,
      "step": 300
    },
    {
      "epoch": 0.6677759290072102,
      "grad_norm": 0.09220443665981293,
      "learning_rate": 0.0002590123456790123,
      "loss": 0.6612,
      "step": 301
    },
    {
      "epoch": 0.6699944536882972,
      "grad_norm": 0.09781794995069504,
      "learning_rate": 0.0002587654320987654,
      "loss": 0.8268,
      "step": 302
    },
    {
      "epoch": 0.6722129783693843,
      "grad_norm": 0.05783013626933098,
      "learning_rate": 0.0002585185185185185,
      "loss": 0.6705,
      "step": 303
    },
    {
      "epoch": 0.6744315030504714,
      "grad_norm": 0.09064796566963196,
      "learning_rate": 0.0002582716049382716,
      "loss": 0.6042,
      "step": 304
    },
    {
      "epoch": 0.6766500277315585,
      "grad_norm": 0.07068487256765366,
      "learning_rate": 0.0002580246913580247,
      "loss": 0.7798,
      "step": 305
    },
    {
      "epoch": 0.6788685524126455,
      "grad_norm": 0.08207566291093826,
      "learning_rate": 0.0002577777777777778,
      "loss": 0.6835,
      "step": 306
    },
    {
      "epoch": 0.6810870770937326,
      "grad_norm": 0.08798323571681976,
      "learning_rate": 0.0002575308641975308,
      "loss": 0.8103,
      "step": 307
    },
    {
      "epoch": 0.6833056017748197,
      "grad_norm": 0.08952302485704422,
      "learning_rate": 0.0002572839506172839,
      "loss": 0.8225,
      "step": 308
    },
    {
      "epoch": 0.6855241264559068,
      "grad_norm": 0.07416049391031265,
      "learning_rate": 0.000257037037037037,
      "loss": 0.7032,
      "step": 309
    },
    {
      "epoch": 0.687742651136994,
      "grad_norm": 0.0987696722149849,
      "learning_rate": 0.0002567901234567901,
      "loss": 0.6657,
      "step": 310
    },
    {
      "epoch": 0.689961175818081,
      "grad_norm": 0.09498900175094604,
      "learning_rate": 0.0002565432098765432,
      "loss": 0.8443,
      "step": 311
    },
    {
      "epoch": 0.6921797004991681,
      "grad_norm": 0.08986473828554153,
      "learning_rate": 0.00025629629629629624,
      "loss": 0.7994,
      "step": 312
    },
    {
      "epoch": 0.6943982251802552,
      "grad_norm": 0.08918307721614838,
      "learning_rate": 0.0002560493827160494,
      "loss": 0.7129,
      "step": 313
    },
    {
      "epoch": 0.6966167498613423,
      "grad_norm": 0.07563033699989319,
      "learning_rate": 0.00025580246913580243,
      "loss": 0.7947,
      "step": 314
    },
    {
      "epoch": 0.6988352745424293,
      "grad_norm": 0.09090328961610794,
      "learning_rate": 0.00025555555555555553,
      "loss": 0.7476,
      "step": 315
    },
    {
      "epoch": 0.7010537992235164,
      "grad_norm": 0.08217328041791916,
      "learning_rate": 0.0002553086419753086,
      "loss": 0.697,
      "step": 316
    },
    {
      "epoch": 0.7032723239046035,
      "grad_norm": 0.0727621540427208,
      "learning_rate": 0.0002550617283950617,
      "loss": 0.578,
      "step": 317
    },
    {
      "epoch": 0.7054908485856906,
      "grad_norm": 0.16869592666625977,
      "learning_rate": 0.0002548148148148148,
      "loss": 0.5622,
      "step": 318
    },
    {
      "epoch": 0.7077093732667776,
      "grad_norm": 0.0858042761683464,
      "learning_rate": 0.00025456790123456785,
      "loss": 0.8177,
      "step": 319
    },
    {
      "epoch": 0.7099278979478647,
      "grad_norm": 0.12341685593128204,
      "learning_rate": 0.00025432098765432095,
      "loss": 0.8944,
      "step": 320
    },
    {
      "epoch": 0.7121464226289518,
      "grad_norm": 0.07295197248458862,
      "learning_rate": 0.00025407407407407404,
      "loss": 0.6783,
      "step": 321
    },
    {
      "epoch": 0.7143649473100389,
      "grad_norm": 0.11438731849193573,
      "learning_rate": 0.00025382716049382714,
      "loss": 0.8269,
      "step": 322
    },
    {
      "epoch": 0.7165834719911259,
      "grad_norm": 0.07610272616147995,
      "learning_rate": 0.00025358024691358023,
      "loss": 0.737,
      "step": 323
    },
    {
      "epoch": 0.718801996672213,
      "grad_norm": 0.11154204607009888,
      "learning_rate": 0.00025333333333333333,
      "loss": 0.8182,
      "step": 324
    },
    {
      "epoch": 0.7210205213533001,
      "grad_norm": 0.09946271032094955,
      "learning_rate": 0.0002530864197530864,
      "loss": 0.6215,
      "step": 325
    },
    {
      "epoch": 0.7232390460343872,
      "grad_norm": 0.07620283961296082,
      "learning_rate": 0.00025283950617283947,
      "loss": 0.7683,
      "step": 326
    },
    {
      "epoch": 0.7254575707154742,
      "grad_norm": 0.07863817363977432,
      "learning_rate": 0.00025259259259259256,
      "loss": 0.8469,
      "step": 327
    },
    {
      "epoch": 0.7276760953965613,
      "grad_norm": 0.06656937301158905,
      "learning_rate": 0.00025234567901234566,
      "loss": 0.8157,
      "step": 328
    },
    {
      "epoch": 0.7298946200776484,
      "grad_norm": 0.10073520988225937,
      "learning_rate": 0.00025209876543209875,
      "loss": 0.788,
      "step": 329
    },
    {
      "epoch": 0.7321131447587355,
      "grad_norm": 0.08716189861297607,
      "learning_rate": 0.0002518518518518518,
      "loss": 0.7849,
      "step": 330
    },
    {
      "epoch": 0.7343316694398225,
      "grad_norm": 0.08480710536241531,
      "learning_rate": 0.00025160493827160494,
      "loss": 0.8544,
      "step": 331
    },
    {
      "epoch": 0.7365501941209096,
      "grad_norm": 0.07336384803056717,
      "learning_rate": 0.00025135802469135804,
      "loss": 0.7745,
      "step": 332
    },
    {
      "epoch": 0.7387687188019967,
      "grad_norm": 0.07170961797237396,
      "learning_rate": 0.0002511111111111111,
      "loss": 0.5935,
      "step": 333
    },
    {
      "epoch": 0.7409872434830838,
      "grad_norm": 0.0806935504078865,
      "learning_rate": 0.00025086419753086417,
      "loss": 0.7879,
      "step": 334
    },
    {
      "epoch": 0.7432057681641708,
      "grad_norm": 0.10466580837965012,
      "learning_rate": 0.00025061728395061727,
      "loss": 0.7908,
      "step": 335
    },
    {
      "epoch": 0.7454242928452579,
      "grad_norm": 0.10991580039262772,
      "learning_rate": 0.00025037037037037036,
      "loss": 0.8305,
      "step": 336
    },
    {
      "epoch": 0.747642817526345,
      "grad_norm": 0.09212615340948105,
      "learning_rate": 0.0002501234567901234,
      "loss": 0.8294,
      "step": 337
    },
    {
      "epoch": 0.7498613422074321,
      "grad_norm": 0.07880492508411407,
      "learning_rate": 0.0002498765432098765,
      "loss": 0.817,
      "step": 338
    },
    {
      "epoch": 0.7520798668885191,
      "grad_norm": 0.12379463016986847,
      "learning_rate": 0.00024962962962962965,
      "loss": 0.8894,
      "step": 339
    },
    {
      "epoch": 0.7542983915696062,
      "grad_norm": 0.08559797704219818,
      "learning_rate": 0.0002493827160493827,
      "loss": 0.8761,
      "step": 340
    },
    {
      "epoch": 0.7565169162506933,
      "grad_norm": 0.12532523274421692,
      "learning_rate": 0.0002491358024691358,
      "loss": 0.7979,
      "step": 341
    },
    {
      "epoch": 0.7587354409317804,
      "grad_norm": 0.1087912768125534,
      "learning_rate": 0.0002488888888888889,
      "loss": 0.6133,
      "step": 342
    },
    {
      "epoch": 0.7609539656128674,
      "grad_norm": 0.07987014949321747,
      "learning_rate": 0.000248641975308642,
      "loss": 0.6192,
      "step": 343
    },
    {
      "epoch": 0.7631724902939545,
      "grad_norm": 0.1035771444439888,
      "learning_rate": 0.000248395061728395,
      "loss": 0.8269,
      "step": 344
    },
    {
      "epoch": 0.7653910149750416,
      "grad_norm": 0.10207753628492355,
      "learning_rate": 0.0002481481481481481,
      "loss": 0.8905,
      "step": 345
    },
    {
      "epoch": 0.7676095396561287,
      "grad_norm": 0.06909305602312088,
      "learning_rate": 0.0002479012345679012,
      "loss": 0.7482,
      "step": 346
    },
    {
      "epoch": 0.7698280643372157,
      "grad_norm": 0.0900287851691246,
      "learning_rate": 0.0002476543209876543,
      "loss": 0.6367,
      "step": 347
    },
    {
      "epoch": 0.7720465890183028,
      "grad_norm": 0.08618215471506119,
      "learning_rate": 0.0002474074074074074,
      "loss": 0.7544,
      "step": 348
    },
    {
      "epoch": 0.7742651136993899,
      "grad_norm": 0.0944581925868988,
      "learning_rate": 0.0002471604938271605,
      "loss": 0.6732,
      "step": 349
    },
    {
      "epoch": 0.776483638380477,
      "grad_norm": 0.1118108481168747,
      "learning_rate": 0.0002469135802469136,
      "loss": 0.9529,
      "step": 350
    },
    {
      "epoch": 0.778702163061564,
      "grad_norm": 0.06773846596479416,
      "learning_rate": 0.0002466666666666666,
      "loss": 0.6073,
      "step": 351
    },
    {
      "epoch": 0.7809206877426511,
      "grad_norm": 0.07912863790988922,
      "learning_rate": 0.0002464197530864197,
      "loss": 0.778,
      "step": 352
    },
    {
      "epoch": 0.7831392124237382,
      "grad_norm": 0.08606277406215668,
      "learning_rate": 0.0002461728395061728,
      "loss": 0.7932,
      "step": 353
    },
    {
      "epoch": 0.7853577371048253,
      "grad_norm": 0.07674917578697205,
      "learning_rate": 0.0002459259259259259,
      "loss": 0.7772,
      "step": 354
    },
    {
      "epoch": 0.7875762617859123,
      "grad_norm": 0.11481207609176636,
      "learning_rate": 0.000245679012345679,
      "loss": 0.9676,
      "step": 355
    },
    {
      "epoch": 0.7897947864669994,
      "grad_norm": 0.07774189114570618,
      "learning_rate": 0.00024543209876543205,
      "loss": 0.8794,
      "step": 356
    },
    {
      "epoch": 0.7920133111480865,
      "grad_norm": 0.0882255956530571,
      "learning_rate": 0.0002451851851851852,
      "loss": 0.7855,
      "step": 357
    },
    {
      "epoch": 0.7942318358291736,
      "grad_norm": 0.09668004512786865,
      "learning_rate": 0.00024493827160493824,
      "loss": 0.742,
      "step": 358
    },
    {
      "epoch": 0.7964503605102606,
      "grad_norm": 0.08986759185791016,
      "learning_rate": 0.00024469135802469133,
      "loss": 0.7285,
      "step": 359
    },
    {
      "epoch": 0.7986688851913477,
      "grad_norm": 0.10049987584352493,
      "learning_rate": 0.00024444444444444443,
      "loss": 0.8916,
      "step": 360
    },
    {
      "epoch": 0.8008874098724348,
      "grad_norm": 0.09034445136785507,
      "learning_rate": 0.0002441975308641975,
      "loss": 0.7283,
      "step": 361
    },
    {
      "epoch": 0.8031059345535219,
      "grad_norm": 0.08775437623262405,
      "learning_rate": 0.00024395061728395062,
      "loss": 0.7361,
      "step": 362
    },
    {
      "epoch": 0.8053244592346089,
      "grad_norm": 0.08264368027448654,
      "learning_rate": 0.0002437037037037037,
      "loss": 0.7476,
      "step": 363
    },
    {
      "epoch": 0.807542983915696,
      "grad_norm": 0.08713068813085556,
      "learning_rate": 0.00024345679012345678,
      "loss": 0.9554,
      "step": 364
    },
    {
      "epoch": 0.8097615085967831,
      "grad_norm": 0.11728395521640778,
      "learning_rate": 0.00024320987654320985,
      "loss": 0.9754,
      "step": 365
    },
    {
      "epoch": 0.8119800332778702,
      "grad_norm": 0.09372711181640625,
      "learning_rate": 0.00024296296296296295,
      "loss": 0.6328,
      "step": 366
    },
    {
      "epoch": 0.8141985579589573,
      "grad_norm": 0.08802945911884308,
      "learning_rate": 0.00024271604938271601,
      "loss": 0.9541,
      "step": 367
    },
    {
      "epoch": 0.8164170826400444,
      "grad_norm": 0.08314149081707001,
      "learning_rate": 0.0002424691358024691,
      "loss": 0.738,
      "step": 368
    },
    {
      "epoch": 0.8186356073211315,
      "grad_norm": 0.10593748837709427,
      "learning_rate": 0.0002422222222222222,
      "loss": 0.8809,
      "step": 369
    },
    {
      "epoch": 0.8208541320022186,
      "grad_norm": 0.07615520805120468,
      "learning_rate": 0.00024197530864197527,
      "loss": 0.8418,
      "step": 370
    },
    {
      "epoch": 0.8230726566833056,
      "grad_norm": 0.07385171949863434,
      "learning_rate": 0.0002417283950617284,
      "loss": 0.6752,
      "step": 371
    },
    {
      "epoch": 0.8252911813643927,
      "grad_norm": 0.07140367478132248,
      "learning_rate": 0.00024148148148148146,
      "loss": 0.7766,
      "step": 372
    },
    {
      "epoch": 0.8275097060454798,
      "grad_norm": 0.07949560880661011,
      "learning_rate": 0.00024123456790123456,
      "loss": 0.74,
      "step": 373
    },
    {
      "epoch": 0.8297282307265669,
      "grad_norm": 0.13998445868492126,
      "learning_rate": 0.00024098765432098763,
      "loss": 0.7416,
      "step": 374
    },
    {
      "epoch": 0.831946755407654,
      "grad_norm": 0.07700325548648834,
      "learning_rate": 0.00024074074074074072,
      "loss": 0.6729,
      "step": 375
    },
    {
      "epoch": 0.834165280088741,
      "grad_norm": 0.07196114212274551,
      "learning_rate": 0.0002404938271604938,
      "loss": 0.8827,
      "step": 376
    },
    {
      "epoch": 0.8363838047698281,
      "grad_norm": 0.0900164321064949,
      "learning_rate": 0.00024024691358024688,
      "loss": 0.8736,
      "step": 377
    },
    {
      "epoch": 0.8386023294509152,
      "grad_norm": 0.07657673954963684,
      "learning_rate": 0.00023999999999999998,
      "loss": 1.0052,
      "step": 378
    },
    {
      "epoch": 0.8408208541320022,
      "grad_norm": 0.07959526777267456,
      "learning_rate": 0.00023975308641975305,
      "loss": 0.8283,
      "step": 379
    },
    {
      "epoch": 0.8430393788130893,
      "grad_norm": 0.0992848202586174,
      "learning_rate": 0.00023950617283950617,
      "loss": 0.8152,
      "step": 380
    },
    {
      "epoch": 0.8452579034941764,
      "grad_norm": 0.11339849978685379,
      "learning_rate": 0.00023925925925925924,
      "loss": 0.7626,
      "step": 381
    },
    {
      "epoch": 0.8474764281752635,
      "grad_norm": 0.07468780130147934,
      "learning_rate": 0.00023901234567901233,
      "loss": 0.7058,
      "step": 382
    },
    {
      "epoch": 0.8496949528563505,
      "grad_norm": 0.07049230486154556,
      "learning_rate": 0.0002387654320987654,
      "loss": 0.7721,
      "step": 383
    },
    {
      "epoch": 0.8519134775374376,
      "grad_norm": 0.0936582088470459,
      "learning_rate": 0.0002385185185185185,
      "loss": 0.6427,
      "step": 384
    },
    {
      "epoch": 0.8541320022185247,
      "grad_norm": 0.08805491030216217,
      "learning_rate": 0.0002382716049382716,
      "loss": 0.7135,
      "step": 385
    },
    {
      "epoch": 0.8563505268996118,
      "grad_norm": 0.11405949294567108,
      "learning_rate": 0.00023802469135802466,
      "loss": 0.7361,
      "step": 386
    },
    {
      "epoch": 0.8585690515806988,
      "grad_norm": 0.07151123881340027,
      "learning_rate": 0.00023777777777777775,
      "loss": 1.012,
      "step": 387
    },
    {
      "epoch": 0.8607875762617859,
      "grad_norm": 0.07353276759386063,
      "learning_rate": 0.00023753086419753082,
      "loss": 0.7498,
      "step": 388
    },
    {
      "epoch": 0.863006100942873,
      "grad_norm": 0.09952674806118011,
      "learning_rate": 0.00023728395061728394,
      "loss": 0.7852,
      "step": 389
    },
    {
      "epoch": 0.8652246256239601,
      "grad_norm": 0.07856793701648712,
      "learning_rate": 0.000237037037037037,
      "loss": 0.8363,
      "step": 390
    },
    {
      "epoch": 0.8674431503050472,
      "grad_norm": 0.09657839685678482,
      "learning_rate": 0.0002367901234567901,
      "loss": 0.8087,
      "step": 391
    },
    {
      "epoch": 0.8696616749861342,
      "grad_norm": 0.07158765196800232,
      "learning_rate": 0.0002365432098765432,
      "loss": 0.6897,
      "step": 392
    },
    {
      "epoch": 0.8718801996672213,
      "grad_norm": 0.07419024407863617,
      "learning_rate": 0.00023629629629629627,
      "loss": 0.7043,
      "step": 393
    },
    {
      "epoch": 0.8740987243483084,
      "grad_norm": 0.07447310537099838,
      "learning_rate": 0.00023604938271604937,
      "loss": 0.7516,
      "step": 394
    },
    {
      "epoch": 0.8763172490293955,
      "grad_norm": 0.07784067094326019,
      "learning_rate": 0.00023580246913580243,
      "loss": 0.69,
      "step": 395
    },
    {
      "epoch": 0.8785357737104825,
      "grad_norm": 0.06102798506617546,
      "learning_rate": 0.00023555555555555553,
      "loss": 0.7216,
      "step": 396
    },
    {
      "epoch": 0.8807542983915696,
      "grad_norm": 0.09555430710315704,
      "learning_rate": 0.0002353086419753086,
      "loss": 0.845,
      "step": 397
    },
    {
      "epoch": 0.8829728230726567,
      "grad_norm": 0.07931692898273468,
      "learning_rate": 0.00023506172839506172,
      "loss": 0.8646,
      "step": 398
    },
    {
      "epoch": 0.8851913477537438,
      "grad_norm": 0.09660515934228897,
      "learning_rate": 0.0002348148148148148,
      "loss": 0.9534,
      "step": 399
    },
    {
      "epoch": 0.8874098724348308,
      "grad_norm": 0.08203554898500443,
      "learning_rate": 0.00023456790123456788,
      "loss": 0.8135,
      "step": 400
    },
    {
      "epoch": 0.8896283971159179,
      "grad_norm": 0.1045071929693222,
      "learning_rate": 0.00023432098765432098,
      "loss": 0.6675,
      "step": 401
    },
    {
      "epoch": 0.891846921797005,
      "grad_norm": 0.10470564663410187,
      "learning_rate": 0.00023407407407407405,
      "loss": 0.7607,
      "step": 402
    },
    {
      "epoch": 0.894065446478092,
      "grad_norm": 0.08998801559209824,
      "learning_rate": 0.00023382716049382714,
      "loss": 0.7263,
      "step": 403
    },
    {
      "epoch": 0.8962839711591791,
      "grad_norm": 0.09407757222652435,
      "learning_rate": 0.0002335802469135802,
      "loss": 0.752,
      "step": 404
    },
    {
      "epoch": 0.8985024958402662,
      "grad_norm": 0.08354225754737854,
      "learning_rate": 0.0002333333333333333,
      "loss": 0.7577,
      "step": 405
    },
    {
      "epoch": 0.9007210205213533,
      "grad_norm": 0.09638811647891998,
      "learning_rate": 0.00023308641975308637,
      "loss": 0.5404,
      "step": 406
    },
    {
      "epoch": 0.9029395452024404,
      "grad_norm": 0.08965232223272324,
      "learning_rate": 0.0002328395061728395,
      "loss": 0.7662,
      "step": 407
    },
    {
      "epoch": 0.9051580698835274,
      "grad_norm": 0.10293035209178925,
      "learning_rate": 0.0002325925925925926,
      "loss": 1.0167,
      "step": 408
    },
    {
      "epoch": 0.9073765945646145,
      "grad_norm": 0.06421936303377151,
      "learning_rate": 0.00023234567901234566,
      "loss": 0.868,
      "step": 409
    },
    {
      "epoch": 0.9095951192457016,
      "grad_norm": 0.08666825294494629,
      "learning_rate": 0.00023209876543209875,
      "loss": 0.8389,
      "step": 410
    },
    {
      "epoch": 0.9118136439267887,
      "grad_norm": 0.08242958039045334,
      "learning_rate": 0.00023185185185185182,
      "loss": 0.7518,
      "step": 411
    },
    {
      "epoch": 0.9140321686078757,
      "grad_norm": 0.09477598965167999,
      "learning_rate": 0.00023160493827160492,
      "loss": 0.7693,
      "step": 412
    },
    {
      "epoch": 0.9162506932889628,
      "grad_norm": 0.07810801267623901,
      "learning_rate": 0.00023135802469135798,
      "loss": 0.6555,
      "step": 413
    },
    {
      "epoch": 0.9184692179700499,
      "grad_norm": 0.07353605329990387,
      "learning_rate": 0.00023111111111111108,
      "loss": 0.7397,
      "step": 414
    },
    {
      "epoch": 0.920687742651137,
      "grad_norm": 0.0822383239865303,
      "learning_rate": 0.0002308641975308642,
      "loss": 0.6878,
      "step": 415
    },
    {
      "epoch": 0.922906267332224,
      "grad_norm": 0.07441329210996628,
      "learning_rate": 0.00023061728395061727,
      "loss": 0.7104,
      "step": 416
    },
    {
      "epoch": 0.9251247920133111,
      "grad_norm": 0.10996097326278687,
      "learning_rate": 0.00023037037037037036,
      "loss": 0.7683,
      "step": 417
    },
    {
      "epoch": 0.9273433166943982,
      "grad_norm": 0.10795944929122925,
      "learning_rate": 0.00023012345679012343,
      "loss": 0.8506,
      "step": 418
    },
    {
      "epoch": 0.9295618413754853,
      "grad_norm": 0.09849577397108078,
      "learning_rate": 0.00022987654320987653,
      "loss": 0.7984,
      "step": 419
    },
    {
      "epoch": 0.9317803660565723,
      "grad_norm": 0.08883882313966751,
      "learning_rate": 0.0002296296296296296,
      "loss": 0.6746,
      "step": 420
    },
    {
      "epoch": 0.9339988907376594,
      "grad_norm": 0.09640064090490341,
      "learning_rate": 0.0002293827160493827,
      "loss": 0.7668,
      "step": 421
    },
    {
      "epoch": 0.9362174154187465,
      "grad_norm": 0.08142515271902084,
      "learning_rate": 0.00022913580246913579,
      "loss": 0.8916,
      "step": 422
    },
    {
      "epoch": 0.9384359400998337,
      "grad_norm": 0.08883035182952881,
      "learning_rate": 0.00022888888888888885,
      "loss": 0.9111,
      "step": 423
    },
    {
      "epoch": 0.9406544647809207,
      "grad_norm": 0.10725768655538559,
      "learning_rate": 0.00022864197530864198,
      "loss": 0.8343,
      "step": 424
    },
    {
      "epoch": 0.9428729894620078,
      "grad_norm": 0.08553190529346466,
      "learning_rate": 0.00022839506172839504,
      "loss": 0.8316,
      "step": 425
    },
    {
      "epoch": 0.9450915141430949,
      "grad_norm": 0.10945054888725281,
      "learning_rate": 0.00022814814814814814,
      "loss": 0.873,
      "step": 426
    },
    {
      "epoch": 0.947310038824182,
      "grad_norm": 0.07965762913227081,
      "learning_rate": 0.0002279012345679012,
      "loss": 0.7444,
      "step": 427
    },
    {
      "epoch": 0.949528563505269,
      "grad_norm": 0.08350668102502823,
      "learning_rate": 0.0002276543209876543,
      "loss": 0.8357,
      "step": 428
    },
    {
      "epoch": 0.9517470881863561,
      "grad_norm": 0.08606898784637451,
      "learning_rate": 0.00022740740740740737,
      "loss": 0.7964,
      "step": 429
    },
    {
      "epoch": 0.9539656128674432,
      "grad_norm": 0.08270793408155441,
      "learning_rate": 0.00022716049382716047,
      "loss": 0.5742,
      "step": 430
    },
    {
      "epoch": 0.9561841375485303,
      "grad_norm": 0.08705919235944748,
      "learning_rate": 0.00022691358024691356,
      "loss": 0.8154,
      "step": 431
    },
    {
      "epoch": 0.9584026622296173,
      "grad_norm": 0.09269081056118011,
      "learning_rate": 0.00022666666666666663,
      "loss": 0.8276,
      "step": 432
    },
    {
      "epoch": 0.9606211869107044,
      "grad_norm": 0.09241532534360886,
      "learning_rate": 0.00022641975308641975,
      "loss": 0.8533,
      "step": 433
    },
    {
      "epoch": 0.9628397115917915,
      "grad_norm": 0.11687024682760239,
      "learning_rate": 0.00022617283950617282,
      "loss": 0.9646,
      "step": 434
    },
    {
      "epoch": 0.9650582362728786,
      "grad_norm": 0.09747391939163208,
      "learning_rate": 0.00022592592592592591,
      "loss": 0.7906,
      "step": 435
    },
    {
      "epoch": 0.9672767609539656,
      "grad_norm": 0.07556334137916565,
      "learning_rate": 0.00022567901234567898,
      "loss": 0.7522,
      "step": 436
    },
    {
      "epoch": 0.9694952856350527,
      "grad_norm": 0.07642029970884323,
      "learning_rate": 0.00022543209876543208,
      "loss": 0.7044,
      "step": 437
    },
    {
      "epoch": 0.9717138103161398,
      "grad_norm": 0.09152086079120636,
      "learning_rate": 0.00022518518518518517,
      "loss": 0.6303,
      "step": 438
    },
    {
      "epoch": 0.9739323349972269,
      "grad_norm": 0.0725337415933609,
      "learning_rate": 0.00022493827160493824,
      "loss": 0.7266,
      "step": 439
    },
    {
      "epoch": 0.9761508596783139,
      "grad_norm": 0.0947161614894867,
      "learning_rate": 0.00022469135802469134,
      "loss": 0.9671,
      "step": 440
    },
    {
      "epoch": 0.978369384359401,
      "grad_norm": 0.07334205508232117,
      "learning_rate": 0.0002244444444444444,
      "loss": 0.7741,
      "step": 441
    },
    {
      "epoch": 0.9805879090404881,
      "grad_norm": 0.09519793093204498,
      "learning_rate": 0.00022419753086419753,
      "loss": 0.7048,
      "step": 442
    },
    {
      "epoch": 0.9828064337215752,
      "grad_norm": 0.09177181124687195,
      "learning_rate": 0.0002239506172839506,
      "loss": 0.9525,
      "step": 443
    },
    {
      "epoch": 0.9850249584026622,
      "grad_norm": 0.08252163976430893,
      "learning_rate": 0.0002237037037037037,
      "loss": 0.7138,
      "step": 444
    },
    {
      "epoch": 0.9872434830837493,
      "grad_norm": 0.061253275722265244,
      "learning_rate": 0.00022345679012345678,
      "loss": 0.6955,
      "step": 445
    },
    {
      "epoch": 0.9894620077648364,
      "grad_norm": 0.10505226254463196,
      "learning_rate": 0.00022320987654320985,
      "loss": 0.8501,
      "step": 446
    },
    {
      "epoch": 0.9916805324459235,
      "grad_norm": 0.08966520428657532,
      "learning_rate": 0.00022296296296296295,
      "loss": 0.6376,
      "step": 447
    },
    {
      "epoch": 0.9938990571270105,
      "grad_norm": 0.08384251594543457,
      "learning_rate": 0.00022271604938271602,
      "loss": 0.7839,
      "step": 448
    },
    {
      "epoch": 0.9961175818080976,
      "grad_norm": 0.1037226989865303,
      "learning_rate": 0.0002224691358024691,
      "loss": 0.8876,
      "step": 449
    },
    {
      "epoch": 0.9983361064891847,
      "grad_norm": 0.08799834549427032,
      "learning_rate": 0.00022222222222222218,
      "loss": 0.8144,
      "step": 450
    },
    {
      "epoch": 1.0005546311702718,
      "grad_norm": 0.08214768767356873,
      "learning_rate": 0.0002219753086419753,
      "loss": 0.9403,
      "step": 451
    },
    {
      "epoch": 1.0027731558513588,
      "grad_norm": 0.08205091208219528,
      "learning_rate": 0.0002217283950617284,
      "loss": 0.8868,
      "step": 452
    },
    {
      "epoch": 1.004991680532446,
      "grad_norm": 0.09211612492799759,
      "learning_rate": 0.00022148148148148146,
      "loss": 0.7995,
      "step": 453
    },
    {
      "epoch": 1.007210205213533,
      "grad_norm": 0.06985071301460266,
      "learning_rate": 0.00022123456790123456,
      "loss": 0.6385,
      "step": 454
    },
    {
      "epoch": 1.00942872989462,
      "grad_norm": 0.09531697630882263,
      "learning_rate": 0.00022098765432098763,
      "loss": 0.6668,
      "step": 455
    },
    {
      "epoch": 1.0116472545757071,
      "grad_norm": 0.07812660187482834,
      "learning_rate": 0.00022074074074074072,
      "loss": 0.8309,
      "step": 456
    },
    {
      "epoch": 1.0138657792567942,
      "grad_norm": 0.09489808976650238,
      "learning_rate": 0.0002204938271604938,
      "loss": 0.6647,
      "step": 457
    },
    {
      "epoch": 1.0160843039378813,
      "grad_norm": 0.06921736150979996,
      "learning_rate": 0.00022024691358024689,
      "loss": 0.6243,
      "step": 458
    },
    {
      "epoch": 1.0183028286189684,
      "grad_norm": 0.07702562212944031,
      "learning_rate": 0.00021999999999999995,
      "loss": 0.6768,
      "step": 459
    },
    {
      "epoch": 1.0205213533000554,
      "grad_norm": 0.08165331184864044,
      "learning_rate": 0.00021975308641975308,
      "loss": 0.5771,
      "step": 460
    },
    {
      "epoch": 1.0227398779811425,
      "grad_norm": 0.13993725180625916,
      "learning_rate": 0.00021950617283950617,
      "loss": 0.6883,
      "step": 461
    },
    {
      "epoch": 1.0249584026622296,
      "grad_norm": 0.0766364187002182,
      "learning_rate": 0.00021925925925925924,
      "loss": 0.8097,
      "step": 462
    },
    {
      "epoch": 1.0271769273433167,
      "grad_norm": 0.07085635513067245,
      "learning_rate": 0.00021901234567901233,
      "loss": 0.6492,
      "step": 463
    },
    {
      "epoch": 1.0293954520244037,
      "grad_norm": 0.09685762226581573,
      "learning_rate": 0.0002187654320987654,
      "loss": 0.6885,
      "step": 464
    },
    {
      "epoch": 1.0316139767054908,
      "grad_norm": 0.10411633551120758,
      "learning_rate": 0.0002185185185185185,
      "loss": 0.7533,
      "step": 465
    },
    {
      "epoch": 1.033832501386578,
      "grad_norm": 0.08149150758981705,
      "learning_rate": 0.00021827160493827157,
      "loss": 0.7784,
      "step": 466
    },
    {
      "epoch": 1.036051026067665,
      "grad_norm": 0.09520027786493301,
      "learning_rate": 0.00021802469135802466,
      "loss": 0.8495,
      "step": 467
    },
    {
      "epoch": 1.038269550748752,
      "grad_norm": 0.12128206342458725,
      "learning_rate": 0.00021777777777777778,
      "loss": 0.9329,
      "step": 468
    },
    {
      "epoch": 1.0404880754298391,
      "grad_norm": 0.07489380240440369,
      "learning_rate": 0.00021753086419753085,
      "loss": 0.7492,
      "step": 469
    },
    {
      "epoch": 1.0427066001109262,
      "grad_norm": 0.08513535559177399,
      "learning_rate": 0.00021728395061728395,
      "loss": 0.79,
      "step": 470
    },
    {
      "epoch": 1.0449251247920133,
      "grad_norm": 0.08387118577957153,
      "learning_rate": 0.00021703703703703701,
      "loss": 0.7469,
      "step": 471
    },
    {
      "epoch": 1.0471436494731003,
      "grad_norm": 0.09435407817363739,
      "learning_rate": 0.0002167901234567901,
      "loss": 0.9106,
      "step": 472
    },
    {
      "epoch": 1.0493621741541874,
      "grad_norm": 0.07460123300552368,
      "learning_rate": 0.00021654320987654318,
      "loss": 0.7546,
      "step": 473
    },
    {
      "epoch": 1.0515806988352745,
      "grad_norm": 0.07005149126052856,
      "learning_rate": 0.00021629629629629627,
      "loss": 0.8144,
      "step": 474
    },
    {
      "epoch": 1.0537992235163616,
      "grad_norm": 0.08547985553741455,
      "learning_rate": 0.00021604938271604937,
      "loss": 0.7236,
      "step": 475
    },
    {
      "epoch": 1.0560177481974486,
      "grad_norm": 0.08737259358167648,
      "learning_rate": 0.00021580246913580244,
      "loss": 0.7826,
      "step": 476
    },
    {
      "epoch": 1.0582362728785357,
      "grad_norm": 0.09947231411933899,
      "learning_rate": 0.00021555555555555556,
      "loss": 0.5397,
      "step": 477
    },
    {
      "epoch": 1.0604547975596228,
      "grad_norm": 0.09093943238258362,
      "learning_rate": 0.00021530864197530863,
      "loss": 0.643,
      "step": 478
    },
    {
      "epoch": 1.0626733222407099,
      "grad_norm": 0.09002818912267685,
      "learning_rate": 0.00021506172839506172,
      "loss": 0.7213,
      "step": 479
    },
    {
      "epoch": 1.064891846921797,
      "grad_norm": 0.10396579653024673,
      "learning_rate": 0.0002148148148148148,
      "loss": 0.9969,
      "step": 480
    },
    {
      "epoch": 1.067110371602884,
      "grad_norm": 0.10313493758440018,
      "learning_rate": 0.00021456790123456788,
      "loss": 0.7281,
      "step": 481
    },
    {
      "epoch": 1.069328896283971,
      "grad_norm": 0.08581659197807312,
      "learning_rate": 0.00021432098765432095,
      "loss": 0.8188,
      "step": 482
    },
    {
      "epoch": 1.0715474209650582,
      "grad_norm": 0.12499208003282547,
      "learning_rate": 0.00021407407407407405,
      "loss": 0.9716,
      "step": 483
    },
    {
      "epoch": 1.0737659456461452,
      "grad_norm": 0.10099252313375473,
      "learning_rate": 0.00021382716049382714,
      "loss": 0.8649,
      "step": 484
    },
    {
      "epoch": 1.0759844703272323,
      "grad_norm": 0.08392249792814255,
      "learning_rate": 0.0002135802469135802,
      "loss": 0.6128,
      "step": 485
    },
    {
      "epoch": 1.0782029950083194,
      "grad_norm": 0.0751982256770134,
      "learning_rate": 0.00021333333333333333,
      "loss": 0.7425,
      "step": 486
    },
    {
      "epoch": 1.0804215196894065,
      "grad_norm": 0.1223888024687767,
      "learning_rate": 0.0002130864197530864,
      "loss": 0.6822,
      "step": 487
    },
    {
      "epoch": 1.0826400443704935,
      "grad_norm": 0.0934222862124443,
      "learning_rate": 0.0002128395061728395,
      "loss": 0.8696,
      "step": 488
    },
    {
      "epoch": 1.0848585690515806,
      "grad_norm": 0.08371588587760925,
      "learning_rate": 0.00021259259259259256,
      "loss": 1.0198,
      "step": 489
    },
    {
      "epoch": 1.0870770937326677,
      "grad_norm": 0.06690515577793121,
      "learning_rate": 0.00021234567901234566,
      "loss": 0.9174,
      "step": 490
    },
    {
      "epoch": 1.0892956184137548,
      "grad_norm": 0.08754364401102066,
      "learning_rate": 0.00021209876543209875,
      "loss": 1.0026,
      "step": 491
    },
    {
      "epoch": 1.0915141430948418,
      "grad_norm": 0.060238420963287354,
      "learning_rate": 0.00021185185185185182,
      "loss": 0.7119,
      "step": 492
    },
    {
      "epoch": 1.093732667775929,
      "grad_norm": 0.07675990462303162,
      "learning_rate": 0.00021160493827160492,
      "loss": 0.8764,
      "step": 493
    },
    {
      "epoch": 1.095951192457016,
      "grad_norm": 0.09489665925502777,
      "learning_rate": 0.00021135802469135799,
      "loss": 0.803,
      "step": 494
    },
    {
      "epoch": 1.098169717138103,
      "grad_norm": 0.08652761578559875,
      "learning_rate": 0.0002111111111111111,
      "loss": 0.599,
      "step": 495
    },
    {
      "epoch": 1.1003882418191901,
      "grad_norm": 0.08433463424444199,
      "learning_rate": 0.00021086419753086418,
      "loss": 0.8595,
      "step": 496
    },
    {
      "epoch": 1.1026067665002772,
      "grad_norm": 0.10147581249475479,
      "learning_rate": 0.00021061728395061727,
      "loss": 0.8305,
      "step": 497
    },
    {
      "epoch": 1.1048252911813643,
      "grad_norm": 0.08841203153133392,
      "learning_rate": 0.00021037037037037037,
      "loss": 0.7517,
      "step": 498
    },
    {
      "epoch": 1.1070438158624514,
      "grad_norm": 0.0752338394522667,
      "learning_rate": 0.00021012345679012343,
      "loss": 0.7756,
      "step": 499
    },
    {
      "epoch": 1.1092623405435384,
      "grad_norm": 0.07579973340034485,
      "learning_rate": 0.00020987654320987653,
      "loss": 0.776,
      "step": 500
    },
    {
      "epoch": 1.1114808652246255,
      "grad_norm": 0.08451181650161743,
      "learning_rate": 0.0002096296296296296,
      "loss": 0.7178,
      "step": 501
    },
    {
      "epoch": 1.1136993899057126,
      "grad_norm": 0.08528746664524078,
      "learning_rate": 0.0002093827160493827,
      "loss": 0.7242,
      "step": 502
    },
    {
      "epoch": 1.1159179145867997,
      "grad_norm": 0.06420214474201202,
      "learning_rate": 0.00020913580246913576,
      "loss": 0.8042,
      "step": 503
    },
    {
      "epoch": 1.1181364392678868,
      "grad_norm": 0.0957309901714325,
      "learning_rate": 0.00020888888888888888,
      "loss": 0.7836,
      "step": 504
    },
    {
      "epoch": 1.1203549639489738,
      "grad_norm": 0.07760097831487656,
      "learning_rate": 0.00020864197530864198,
      "loss": 0.7295,
      "step": 505
    },
    {
      "epoch": 1.122573488630061,
      "grad_norm": 0.09017680585384369,
      "learning_rate": 0.00020839506172839505,
      "loss": 0.7201,
      "step": 506
    },
    {
      "epoch": 1.124792013311148,
      "grad_norm": 0.09066541492938995,
      "learning_rate": 0.00020814814814814814,
      "loss": 0.8548,
      "step": 507
    },
    {
      "epoch": 1.127010537992235,
      "grad_norm": 0.08926627039909363,
      "learning_rate": 0.0002079012345679012,
      "loss": 0.7296,
      "step": 508
    },
    {
      "epoch": 1.1292290626733221,
      "grad_norm": 0.06617864966392517,
      "learning_rate": 0.0002076543209876543,
      "loss": 0.6473,
      "step": 509
    },
    {
      "epoch": 1.1314475873544092,
      "grad_norm": 0.06825745105743408,
      "learning_rate": 0.00020740740740740737,
      "loss": 0.8032,
      "step": 510
    },
    {
      "epoch": 1.1336661120354963,
      "grad_norm": 0.0760616660118103,
      "learning_rate": 0.00020716049382716047,
      "loss": 0.871,
      "step": 511
    },
    {
      "epoch": 1.1358846367165834,
      "grad_norm": 0.09925301373004913,
      "learning_rate": 0.00020691358024691354,
      "loss": 0.9739,
      "step": 512
    },
    {
      "epoch": 1.1381031613976704,
      "grad_norm": 0.08360178768634796,
      "learning_rate": 0.00020666666666666666,
      "loss": 0.8955,
      "step": 513
    },
    {
      "epoch": 1.1403216860787575,
      "grad_norm": 0.11236846446990967,
      "learning_rate": 0.00020641975308641975,
      "loss": 0.5983,
      "step": 514
    },
    {
      "epoch": 1.1425402107598448,
      "grad_norm": 0.10900504142045975,
      "learning_rate": 0.00020617283950617282,
      "loss": 0.8787,
      "step": 515
    },
    {
      "epoch": 1.1447587354409319,
      "grad_norm": 0.09850780665874481,
      "learning_rate": 0.00020592592592592592,
      "loss": 0.8976,
      "step": 516
    },
    {
      "epoch": 1.146977260122019,
      "grad_norm": 0.09892156720161438,
      "learning_rate": 0.00020567901234567898,
      "loss": 0.824,
      "step": 517
    },
    {
      "epoch": 1.149195784803106,
      "grad_norm": 0.13722093403339386,
      "learning_rate": 0.00020543209876543208,
      "loss": 0.8148,
      "step": 518
    },
    {
      "epoch": 1.151414309484193,
      "grad_norm": 0.10007547587156296,
      "learning_rate": 0.00020518518518518515,
      "loss": 0.9508,
      "step": 519
    },
    {
      "epoch": 1.1536328341652802,
      "grad_norm": 0.07496339827775955,
      "learning_rate": 0.00020493827160493824,
      "loss": 0.6836,
      "step": 520
    },
    {
      "epoch": 1.1558513588463672,
      "grad_norm": 0.06336478143930435,
      "learning_rate": 0.00020469135802469136,
      "loss": 0.7254,
      "step": 521
    },
    {
      "epoch": 1.1580698835274543,
      "grad_norm": 0.06988365203142166,
      "learning_rate": 0.00020444444444444443,
      "loss": 0.7424,
      "step": 522
    },
    {
      "epoch": 1.1602884082085414,
      "grad_norm": 0.08677584677934647,
      "learning_rate": 0.00020419753086419753,
      "loss": 0.7353,
      "step": 523
    },
    {
      "epoch": 1.1625069328896285,
      "grad_norm": 0.08813011646270752,
      "learning_rate": 0.0002039506172839506,
      "loss": 0.7104,
      "step": 524
    },
    {
      "epoch": 1.1647254575707155,
      "grad_norm": 0.10195858776569366,
      "learning_rate": 0.0002037037037037037,
      "loss": 0.9307,
      "step": 525
    },
    {
      "epoch": 1.1669439822518026,
      "grad_norm": 0.08053350448608398,
      "learning_rate": 0.00020345679012345676,
      "loss": 0.8199,
      "step": 526
    },
    {
      "epoch": 1.1691625069328897,
      "grad_norm": 0.07883057743310928,
      "learning_rate": 0.00020320987654320985,
      "loss": 0.6413,
      "step": 527
    },
    {
      "epoch": 1.1713810316139768,
      "grad_norm": 0.08071167021989822,
      "learning_rate": 0.00020296296296296295,
      "loss": 0.7096,
      "step": 528
    },
    {
      "epoch": 1.1735995562950639,
      "grad_norm": 0.09079280495643616,
      "learning_rate": 0.00020271604938271602,
      "loss": 0.8851,
      "step": 529
    },
    {
      "epoch": 1.175818080976151,
      "grad_norm": 0.08630382269620895,
      "learning_rate": 0.00020246913580246914,
      "loss": 0.7138,
      "step": 530
    },
    {
      "epoch": 1.178036605657238,
      "grad_norm": 0.12275026738643646,
      "learning_rate": 0.0002022222222222222,
      "loss": 0.8525,
      "step": 531
    },
    {
      "epoch": 1.180255130338325,
      "grad_norm": 0.09046459197998047,
      "learning_rate": 0.0002019753086419753,
      "loss": 0.7934,
      "step": 532
    },
    {
      "epoch": 1.1824736550194122,
      "grad_norm": 0.1475738286972046,
      "learning_rate": 0.00020172839506172837,
      "loss": 0.8072,
      "step": 533
    },
    {
      "epoch": 1.1846921797004992,
      "grad_norm": 0.07192722707986832,
      "learning_rate": 0.00020148148148148147,
      "loss": 0.6021,
      "step": 534
    },
    {
      "epoch": 1.1869107043815863,
      "grad_norm": 0.08052520453929901,
      "learning_rate": 0.00020123456790123456,
      "loss": 0.7612,
      "step": 535
    },
    {
      "epoch": 1.1891292290626734,
      "grad_norm": 0.116862453520298,
      "learning_rate": 0.00020098765432098763,
      "loss": 0.7581,
      "step": 536
    },
    {
      "epoch": 1.1913477537437605,
      "grad_norm": 0.08883317559957504,
      "learning_rate": 0.00020074074074074072,
      "loss": 0.6925,
      "step": 537
    },
    {
      "epoch": 1.1935662784248475,
      "grad_norm": 0.10259954631328583,
      "learning_rate": 0.0002004938271604938,
      "loss": 0.795,
      "step": 538
    },
    {
      "epoch": 1.1957848031059346,
      "grad_norm": 0.07333339005708694,
      "learning_rate": 0.00020024691358024691,
      "loss": 0.6726,
      "step": 539
    },
    {
      "epoch": 1.1980033277870217,
      "grad_norm": 0.1296006143093109,
      "learning_rate": 0.00019999999999999998,
      "loss": 0.7814,
      "step": 540
    },
    {
      "epoch": 1.2002218524681088,
      "grad_norm": 0.08468916267156601,
      "learning_rate": 0.00019975308641975308,
      "loss": 0.7194,
      "step": 541
    },
    {
      "epoch": 1.2024403771491958,
      "grad_norm": 0.07875210791826248,
      "learning_rate": 0.00019950617283950615,
      "loss": 0.7045,
      "step": 542
    },
    {
      "epoch": 1.204658901830283,
      "grad_norm": 0.07139656692743301,
      "learning_rate": 0.00019925925925925924,
      "loss": 0.7766,
      "step": 543
    },
    {
      "epoch": 1.20687742651137,
      "grad_norm": 0.07260648161172867,
      "learning_rate": 0.00019901234567901234,
      "loss": 0.6642,
      "step": 544
    },
    {
      "epoch": 1.209095951192457,
      "grad_norm": 0.08689554780721664,
      "learning_rate": 0.0001987654320987654,
      "loss": 0.6459,
      "step": 545
    },
    {
      "epoch": 1.2113144758735441,
      "grad_norm": 0.08016476035118103,
      "learning_rate": 0.0001985185185185185,
      "loss": 0.6167,
      "step": 546
    },
    {
      "epoch": 1.2135330005546312,
      "grad_norm": 0.08668354153633118,
      "learning_rate": 0.00019827160493827157,
      "loss": 0.7636,
      "step": 547
    },
    {
      "epoch": 1.2157515252357183,
      "grad_norm": 0.11029621958732605,
      "learning_rate": 0.0001980246913580247,
      "loss": 0.7179,
      "step": 548
    },
    {
      "epoch": 1.2179700499168054,
      "grad_norm": 0.09763128310441971,
      "learning_rate": 0.00019777777777777776,
      "loss": 0.9222,
      "step": 549
    },
    {
      "epoch": 1.2201885745978924,
      "grad_norm": 0.08694533258676529,
      "learning_rate": 0.00019753086419753085,
      "loss": 0.877,
      "step": 550
    },
    {
      "epoch": 1.2224070992789795,
      "grad_norm": 0.09525983035564423,
      "learning_rate": 0.00019728395061728395,
      "loss": 0.8251,
      "step": 551
    },
    {
      "epoch": 1.2246256239600666,
      "grad_norm": 0.08505046367645264,
      "learning_rate": 0.00019703703703703702,
      "loss": 0.858,
      "step": 552
    },
    {
      "epoch": 1.2268441486411537,
      "grad_norm": 0.07911120355129242,
      "learning_rate": 0.0001967901234567901,
      "loss": 0.7367,
      "step": 553
    },
    {
      "epoch": 1.2290626733222407,
      "grad_norm": 0.07007575780153275,
      "learning_rate": 0.00019654320987654318,
      "loss": 0.7523,
      "step": 554
    },
    {
      "epoch": 1.2312811980033278,
      "grad_norm": 0.07355703413486481,
      "learning_rate": 0.00019629629629629627,
      "loss": 0.8293,
      "step": 555
    },
    {
      "epoch": 1.2334997226844149,
      "grad_norm": 0.0939091220498085,
      "learning_rate": 0.00019604938271604934,
      "loss": 0.7906,
      "step": 556
    },
    {
      "epoch": 1.235718247365502,
      "grad_norm": 0.07832761853933334,
      "learning_rate": 0.00019580246913580246,
      "loss": 0.7207,
      "step": 557
    },
    {
      "epoch": 1.237936772046589,
      "grad_norm": 0.08007757365703583,
      "learning_rate": 0.00019555555555555556,
      "loss": 0.6667,
      "step": 558
    },
    {
      "epoch": 1.240155296727676,
      "grad_norm": 0.07047320902347565,
      "learning_rate": 0.00019530864197530863,
      "loss": 0.7208,
      "step": 559
    },
    {
      "epoch": 1.2423738214087632,
      "grad_norm": 0.0791020616889,
      "learning_rate": 0.00019506172839506172,
      "loss": 0.7504,
      "step": 560
    },
    {
      "epoch": 1.2445923460898503,
      "grad_norm": 0.08235763758420944,
      "learning_rate": 0.0001948148148148148,
      "loss": 0.668,
      "step": 561
    },
    {
      "epoch": 1.2468108707709373,
      "grad_norm": 0.08907868713140488,
      "learning_rate": 0.00019456790123456789,
      "loss": 0.8249,
      "step": 562
    },
    {
      "epoch": 1.2490293954520244,
      "grad_norm": 0.0859551951289177,
      "learning_rate": 0.00019432098765432095,
      "loss": 0.8899,
      "step": 563
    },
    {
      "epoch": 1.2512479201331115,
      "grad_norm": 0.07931114733219147,
      "learning_rate": 0.00019407407407407405,
      "loss": 0.8384,
      "step": 564
    },
    {
      "epoch": 1.2534664448141986,
      "grad_norm": 0.07584373652935028,
      "learning_rate": 0.00019382716049382712,
      "loss": 0.7117,
      "step": 565
    },
    {
      "epoch": 1.2556849694952856,
      "grad_norm": 0.08035347610712051,
      "learning_rate": 0.00019358024691358024,
      "loss": 0.8867,
      "step": 566
    },
    {
      "epoch": 1.2579034941763727,
      "grad_norm": 0.0658009797334671,
      "learning_rate": 0.00019333333333333333,
      "loss": 0.8065,
      "step": 567
    },
    {
      "epoch": 1.2601220188574598,
      "grad_norm": 0.06503137946128845,
      "learning_rate": 0.0001930864197530864,
      "loss": 0.7693,
      "step": 568
    },
    {
      "epoch": 1.2623405435385469,
      "grad_norm": 0.07313820719718933,
      "learning_rate": 0.0001928395061728395,
      "loss": 0.6746,
      "step": 569
    },
    {
      "epoch": 1.264559068219634,
      "grad_norm": 0.07270326465368271,
      "learning_rate": 0.00019259259259259257,
      "loss": 0.6401,
      "step": 570
    },
    {
      "epoch": 1.266777592900721,
      "grad_norm": 0.07033909112215042,
      "learning_rate": 0.00019234567901234566,
      "loss": 0.7466,
      "step": 571
    },
    {
      "epoch": 1.268996117581808,
      "grad_norm": 0.10364579409360886,
      "learning_rate": 0.00019209876543209873,
      "loss": 0.8721,
      "step": 572
    },
    {
      "epoch": 1.2712146422628952,
      "grad_norm": 0.08161678165197372,
      "learning_rate": 0.00019185185185185182,
      "loss": 0.8161,
      "step": 573
    },
    {
      "epoch": 1.2734331669439822,
      "grad_norm": 0.07219507545232773,
      "learning_rate": 0.00019160493827160495,
      "loss": 0.7538,
      "step": 574
    },
    {
      "epoch": 1.2756516916250693,
      "grad_norm": 0.09501083195209503,
      "learning_rate": 0.00019135802469135801,
      "loss": 0.9324,
      "step": 575
    },
    {
      "epoch": 1.2778702163061564,
      "grad_norm": 0.08709080517292023,
      "learning_rate": 0.0001911111111111111,
      "loss": 0.6872,
      "step": 576
    },
    {
      "epoch": 1.2800887409872435,
      "grad_norm": 0.14032532274723053,
      "learning_rate": 0.00019086419753086418,
      "loss": 0.793,
      "step": 577
    },
    {
      "epoch": 1.2823072656683305,
      "grad_norm": 0.06909395009279251,
      "learning_rate": 0.00019061728395061727,
      "loss": 0.7651,
      "step": 578
    },
    {
      "epoch": 1.2845257903494176,
      "grad_norm": 0.06692535430192947,
      "learning_rate": 0.00019037037037037034,
      "loss": 0.6404,
      "step": 579
    },
    {
      "epoch": 1.2867443150305047,
      "grad_norm": 0.12628145515918732,
      "learning_rate": 0.00019012345679012344,
      "loss": 1.0179,
      "step": 580
    },
    {
      "epoch": 1.2889628397115918,
      "grad_norm": 0.07603675872087479,
      "learning_rate": 0.00018987654320987653,
      "loss": 0.9125,
      "step": 581
    },
    {
      "epoch": 1.2911813643926788,
      "grad_norm": 0.0757095217704773,
      "learning_rate": 0.0001896296296296296,
      "loss": 0.6998,
      "step": 582
    },
    {
      "epoch": 1.293399889073766,
      "grad_norm": 0.11450722068548203,
      "learning_rate": 0.00018938271604938272,
      "loss": 0.6532,
      "step": 583
    },
    {
      "epoch": 1.295618413754853,
      "grad_norm": 0.09681407362222672,
      "learning_rate": 0.0001891358024691358,
      "loss": 0.7827,
      "step": 584
    },
    {
      "epoch": 1.29783693843594,
      "grad_norm": 0.09956950694322586,
      "learning_rate": 0.00018888888888888888,
      "loss": 0.8763,
      "step": 585
    },
    {
      "epoch": 1.3000554631170271,
      "grad_norm": 0.09230848401784897,
      "learning_rate": 0.00018864197530864195,
      "loss": 0.632,
      "step": 586
    },
    {
      "epoch": 1.3022739877981142,
      "grad_norm": 0.08826889097690582,
      "learning_rate": 0.00018839506172839505,
      "loss": 0.6883,
      "step": 587
    },
    {
      "epoch": 1.3044925124792013,
      "grad_norm": 0.07408338040113449,
      "learning_rate": 0.00018814814814814814,
      "loss": 0.8311,
      "step": 588
    },
    {
      "epoch": 1.3067110371602884,
      "grad_norm": 0.07717019319534302,
      "learning_rate": 0.0001879012345679012,
      "loss": 0.7705,
      "step": 589
    },
    {
      "epoch": 1.3089295618413754,
      "grad_norm": 0.08360118418931961,
      "learning_rate": 0.0001876543209876543,
      "loss": 0.7703,
      "step": 590
    },
    {
      "epoch": 1.3111480865224625,
      "grad_norm": 0.07447011023759842,
      "learning_rate": 0.00018740740740740737,
      "loss": 0.589,
      "step": 591
    },
    {
      "epoch": 1.3133666112035496,
      "grad_norm": 0.10814107209444046,
      "learning_rate": 0.0001871604938271605,
      "loss": 0.8853,
      "step": 592
    },
    {
      "epoch": 1.3155851358846367,
      "grad_norm": 0.07028859853744507,
      "learning_rate": 0.00018691358024691356,
      "loss": 0.6983,
      "step": 593
    },
    {
      "epoch": 1.3178036605657237,
      "grad_norm": 0.08274299651384354,
      "learning_rate": 0.00018666666666666666,
      "loss": 0.9347,
      "step": 594
    },
    {
      "epoch": 1.3200221852468108,
      "grad_norm": 0.07571959495544434,
      "learning_rate": 0.00018641975308641973,
      "loss": 1.0269,
      "step": 595
    },
    {
      "epoch": 1.3222407099278979,
      "grad_norm": 0.0801168903708458,
      "learning_rate": 0.00018617283950617282,
      "loss": 0.6709,
      "step": 596
    },
    {
      "epoch": 1.324459234608985,
      "grad_norm": 0.08135650306940079,
      "learning_rate": 0.00018592592592592592,
      "loss": 0.6377,
      "step": 597
    },
    {
      "epoch": 1.326677759290072,
      "grad_norm": 0.09267068654298782,
      "learning_rate": 0.00018567901234567899,
      "loss": 0.7807,
      "step": 598
    },
    {
      "epoch": 1.328896283971159,
      "grad_norm": 0.09667902439832687,
      "learning_rate": 0.00018543209876543208,
      "loss": 0.5625,
      "step": 599
    },
    {
      "epoch": 1.3311148086522462,
      "grad_norm": 0.08911213278770447,
      "learning_rate": 0.00018518518518518515,
      "loss": 0.7811,
      "step": 600
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 0.07871294766664505,
      "learning_rate": 0.00018493827160493827,
      "loss": 0.8825,
      "step": 601
    },
    {
      "epoch": 1.3355518580144203,
      "grad_norm": 0.08177413046360016,
      "learning_rate": 0.00018469135802469134,
      "loss": 0.8492,
      "step": 602
    },
    {
      "epoch": 1.3377703826955074,
      "grad_norm": 0.07946442067623138,
      "learning_rate": 0.00018444444444444443,
      "loss": 0.7,
      "step": 603
    },
    {
      "epoch": 1.3399889073765945,
      "grad_norm": 0.07931774854660034,
      "learning_rate": 0.00018419753086419753,
      "loss": 0.8063,
      "step": 604
    },
    {
      "epoch": 1.3422074320576816,
      "grad_norm": 0.09007471054792404,
      "learning_rate": 0.0001839506172839506,
      "loss": 0.7702,
      "step": 605
    },
    {
      "epoch": 1.3444259567387689,
      "grad_norm": 0.07740438729524612,
      "learning_rate": 0.0001837037037037037,
      "loss": 0.8966,
      "step": 606
    },
    {
      "epoch": 1.346644481419856,
      "grad_norm": 0.08975096791982651,
      "learning_rate": 0.00018345679012345676,
      "loss": 0.8945,
      "step": 607
    },
    {
      "epoch": 1.348863006100943,
      "grad_norm": 0.07721004635095596,
      "learning_rate": 0.00018320987654320986,
      "loss": 0.8346,
      "step": 608
    },
    {
      "epoch": 1.35108153078203,
      "grad_norm": 0.07194160670042038,
      "learning_rate": 0.00018296296296296292,
      "loss": 0.9206,
      "step": 609
    },
    {
      "epoch": 1.3533000554631172,
      "grad_norm": 0.07245586812496185,
      "learning_rate": 0.00018271604938271605,
      "loss": 0.7583,
      "step": 610
    },
    {
      "epoch": 1.3555185801442042,
      "grad_norm": 0.08178890496492386,
      "learning_rate": 0.00018246913580246914,
      "loss": 0.6029,
      "step": 611
    },
    {
      "epoch": 1.3577371048252913,
      "grad_norm": 0.10096482187509537,
      "learning_rate": 0.0001822222222222222,
      "loss": 0.7751,
      "step": 612
    },
    {
      "epoch": 1.3599556295063784,
      "grad_norm": 0.07438313961029053,
      "learning_rate": 0.0001819753086419753,
      "loss": 0.7404,
      "step": 613
    },
    {
      "epoch": 1.3621741541874655,
      "grad_norm": 0.08168497681617737,
      "learning_rate": 0.00018172839506172837,
      "loss": 0.7094,
      "step": 614
    },
    {
      "epoch": 1.3643926788685525,
      "grad_norm": 0.08591803163290024,
      "learning_rate": 0.00018148148148148147,
      "loss": 0.6998,
      "step": 615
    },
    {
      "epoch": 1.3666112035496396,
      "grad_norm": 0.062350161373615265,
      "learning_rate": 0.00018123456790123454,
      "loss": 0.6505,
      "step": 616
    },
    {
      "epoch": 1.3688297282307267,
      "grad_norm": 0.09299323707818985,
      "learning_rate": 0.00018098765432098763,
      "loss": 0.9099,
      "step": 617
    },
    {
      "epoch": 1.3710482529118138,
      "grad_norm": 0.06593653559684753,
      "learning_rate": 0.00018074074074074075,
      "loss": 0.7662,
      "step": 618
    },
    {
      "epoch": 1.3732667775929008,
      "grad_norm": 0.08937115967273712,
      "learning_rate": 0.00018049382716049382,
      "loss": 0.7536,
      "step": 619
    },
    {
      "epoch": 1.375485302273988,
      "grad_norm": 0.10412895679473877,
      "learning_rate": 0.00018024691358024692,
      "loss": 0.7212,
      "step": 620
    },
    {
      "epoch": 1.377703826955075,
      "grad_norm": 0.0757964625954628,
      "learning_rate": 0.00017999999999999998,
      "loss": 0.8893,
      "step": 621
    },
    {
      "epoch": 1.379922351636162,
      "grad_norm": 0.07701815664768219,
      "learning_rate": 0.00017975308641975308,
      "loss": 0.5999,
      "step": 622
    },
    {
      "epoch": 1.3821408763172491,
      "grad_norm": 0.09901320189237595,
      "learning_rate": 0.00017950617283950615,
      "loss": 0.8765,
      "step": 623
    },
    {
      "epoch": 1.3843594009983362,
      "grad_norm": 0.09571735560894012,
      "learning_rate": 0.00017925925925925924,
      "loss": 0.7835,
      "step": 624
    },
    {
      "epoch": 1.3865779256794233,
      "grad_norm": 0.09918275475502014,
      "learning_rate": 0.0001790123456790123,
      "loss": 0.7431,
      "step": 625
    },
    {
      "epoch": 1.3887964503605104,
      "grad_norm": 0.07726182788610458,
      "learning_rate": 0.0001787654320987654,
      "loss": 0.5649,
      "step": 626
    },
    {
      "epoch": 1.3910149750415974,
      "grad_norm": 0.07171943038702011,
      "learning_rate": 0.00017851851851851853,
      "loss": 0.7212,
      "step": 627
    },
    {
      "epoch": 1.3932334997226845,
      "grad_norm": 0.09367648512125015,
      "learning_rate": 0.0001782716049382716,
      "loss": 0.6101,
      "step": 628
    },
    {
      "epoch": 1.3954520244037716,
      "grad_norm": 0.0912347212433815,
      "learning_rate": 0.0001780246913580247,
      "loss": 0.7766,
      "step": 629
    },
    {
      "epoch": 1.3976705490848587,
      "grad_norm": 0.08907873183488846,
      "learning_rate": 0.00017777777777777776,
      "loss": 0.7346,
      "step": 630
    },
    {
      "epoch": 1.3998890737659457,
      "grad_norm": 0.09368817508220673,
      "learning_rate": 0.00017753086419753085,
      "loss": 0.733,
      "step": 631
    },
    {
      "epoch": 1.4021075984470328,
      "grad_norm": 0.06311885267496109,
      "learning_rate": 0.00017728395061728392,
      "loss": 0.6271,
      "step": 632
    },
    {
      "epoch": 1.4043261231281199,
      "grad_norm": 0.06488551199436188,
      "learning_rate": 0.00017703703703703702,
      "loss": 0.7547,
      "step": 633
    },
    {
      "epoch": 1.406544647809207,
      "grad_norm": 0.09249793738126755,
      "learning_rate": 0.0001767901234567901,
      "loss": 0.743,
      "step": 634
    },
    {
      "epoch": 1.408763172490294,
      "grad_norm": 0.1041136160492897,
      "learning_rate": 0.00017654320987654318,
      "loss": 0.7546,
      "step": 635
    },
    {
      "epoch": 1.410981697171381,
      "grad_norm": 0.07278372347354889,
      "learning_rate": 0.0001762962962962963,
      "loss": 0.7456,
      "step": 636
    },
    {
      "epoch": 1.4132002218524682,
      "grad_norm": 0.0902879536151886,
      "learning_rate": 0.00017604938271604937,
      "loss": 0.6801,
      "step": 637
    },
    {
      "epoch": 1.4154187465335553,
      "grad_norm": 0.08684354275465012,
      "learning_rate": 0.00017580246913580247,
      "loss": 0.8289,
      "step": 638
    },
    {
      "epoch": 1.4176372712146423,
      "grad_norm": 0.08778969198465347,
      "learning_rate": 0.00017555555555555553,
      "loss": 0.8292,
      "step": 639
    },
    {
      "epoch": 1.4198557958957294,
      "grad_norm": 0.09461043030023575,
      "learning_rate": 0.00017530864197530863,
      "loss": 0.7495,
      "step": 640
    },
    {
      "epoch": 1.4220743205768165,
      "grad_norm": 0.07748755067586899,
      "learning_rate": 0.00017506172839506172,
      "loss": 0.7729,
      "step": 641
    },
    {
      "epoch": 1.4242928452579036,
      "grad_norm": 0.0893358513712883,
      "learning_rate": 0.0001748148148148148,
      "loss": 0.6794,
      "step": 642
    },
    {
      "epoch": 1.4265113699389906,
      "grad_norm": 0.09026248008012772,
      "learning_rate": 0.0001745679012345679,
      "loss": 0.5771,
      "step": 643
    },
    {
      "epoch": 1.4287298946200777,
      "grad_norm": 0.11601400375366211,
      "learning_rate": 0.00017432098765432096,
      "loss": 0.8817,
      "step": 644
    },
    {
      "epoch": 1.4309484193011648,
      "grad_norm": 0.0648113414645195,
      "learning_rate": 0.00017407407407407408,
      "loss": 0.7011,
      "step": 645
    },
    {
      "epoch": 1.4331669439822519,
      "grad_norm": 0.08075165748596191,
      "learning_rate": 0.00017382716049382715,
      "loss": 0.7144,
      "step": 646
    },
    {
      "epoch": 1.435385468663339,
      "grad_norm": 0.09320619702339172,
      "learning_rate": 0.00017358024691358024,
      "loss": 0.7234,
      "step": 647
    },
    {
      "epoch": 1.437603993344426,
      "grad_norm": 0.08322218805551529,
      "learning_rate": 0.0001733333333333333,
      "loss": 0.6561,
      "step": 648
    },
    {
      "epoch": 1.439822518025513,
      "grad_norm": 0.0657225251197815,
      "learning_rate": 0.0001730864197530864,
      "loss": 0.751,
      "step": 649
    },
    {
      "epoch": 1.4420410427066002,
      "grad_norm": 0.10518378764390945,
      "learning_rate": 0.0001728395061728395,
      "loss": 0.7667,
      "step": 650
    },
    {
      "epoch": 1.4442595673876872,
      "grad_norm": 0.10373567044734955,
      "learning_rate": 0.00017259259259259257,
      "loss": 0.8128,
      "step": 651
    },
    {
      "epoch": 1.4464780920687743,
      "grad_norm": 0.10150046646595001,
      "learning_rate": 0.00017234567901234566,
      "loss": 0.7802,
      "step": 652
    },
    {
      "epoch": 1.4486966167498614,
      "grad_norm": 0.11277901381254196,
      "learning_rate": 0.00017209876543209873,
      "loss": 0.6195,
      "step": 653
    },
    {
      "epoch": 1.4509151414309485,
      "grad_norm": 0.08926472067832947,
      "learning_rate": 0.00017185185185185185,
      "loss": 0.7193,
      "step": 654
    },
    {
      "epoch": 1.4531336661120355,
      "grad_norm": 0.07237178832292557,
      "learning_rate": 0.00017160493827160492,
      "loss": 0.7808,
      "step": 655
    },
    {
      "epoch": 1.4553521907931226,
      "grad_norm": 0.08401527255773544,
      "learning_rate": 0.00017135802469135802,
      "loss": 0.7445,
      "step": 656
    },
    {
      "epoch": 1.4575707154742097,
      "grad_norm": 0.06503847241401672,
      "learning_rate": 0.0001711111111111111,
      "loss": 0.6357,
      "step": 657
    },
    {
      "epoch": 1.4597892401552968,
      "grad_norm": 0.0855102464556694,
      "learning_rate": 0.00017086419753086418,
      "loss": 0.7553,
      "step": 658
    },
    {
      "epoch": 1.4620077648363838,
      "grad_norm": 0.09509997814893723,
      "learning_rate": 0.00017061728395061727,
      "loss": 0.9862,
      "step": 659
    },
    {
      "epoch": 1.464226289517471,
      "grad_norm": 0.08729561418294907,
      "learning_rate": 0.00017037037037037034,
      "loss": 0.6191,
      "step": 660
    },
    {
      "epoch": 1.466444814198558,
      "grad_norm": 0.0847608745098114,
      "learning_rate": 0.00017012345679012344,
      "loss": 0.9921,
      "step": 661
    },
    {
      "epoch": 1.468663338879645,
      "grad_norm": 0.11522991210222244,
      "learning_rate": 0.0001698765432098765,
      "loss": 0.877,
      "step": 662
    },
    {
      "epoch": 1.4708818635607321,
      "grad_norm": 0.066033273935318,
      "learning_rate": 0.00016962962962962963,
      "loss": 0.7129,
      "step": 663
    },
    {
      "epoch": 1.4731003882418192,
      "grad_norm": 0.12610451877117157,
      "learning_rate": 0.00016938271604938272,
      "loss": 0.8226,
      "step": 664
    },
    {
      "epoch": 1.4753189129229063,
      "grad_norm": 0.09242416173219681,
      "learning_rate": 0.0001691358024691358,
      "loss": 0.8341,
      "step": 665
    },
    {
      "epoch": 1.4775374376039934,
      "grad_norm": 0.08232415467500687,
      "learning_rate": 0.00016888888888888889,
      "loss": 0.965,
      "step": 666
    },
    {
      "epoch": 1.4797559622850804,
      "grad_norm": 0.08571493625640869,
      "learning_rate": 0.00016864197530864195,
      "loss": 0.7516,
      "step": 667
    },
    {
      "epoch": 1.4819744869661675,
      "grad_norm": 0.09496626257896423,
      "learning_rate": 0.00016839506172839505,
      "loss": 0.6747,
      "step": 668
    },
    {
      "epoch": 1.4841930116472546,
      "grad_norm": 0.0819113627076149,
      "learning_rate": 0.00016814814814814812,
      "loss": 0.8087,
      "step": 669
    },
    {
      "epoch": 1.4864115363283417,
      "grad_norm": 0.08605154603719711,
      "learning_rate": 0.0001679012345679012,
      "loss": 0.7654,
      "step": 670
    },
    {
      "epoch": 1.4886300610094287,
      "grad_norm": 0.07348262518644333,
      "learning_rate": 0.00016765432098765433,
      "loss": 0.7096,
      "step": 671
    },
    {
      "epoch": 1.4908485856905158,
      "grad_norm": 0.0751531571149826,
      "learning_rate": 0.0001674074074074074,
      "loss": 0.7247,
      "step": 672
    },
    {
      "epoch": 1.493067110371603,
      "grad_norm": 0.06946388632059097,
      "learning_rate": 0.0001671604938271605,
      "loss": 0.6839,
      "step": 673
    },
    {
      "epoch": 1.49528563505269,
      "grad_norm": 0.08002138882875443,
      "learning_rate": 0.00016691358024691357,
      "loss": 0.8001,
      "step": 674
    },
    {
      "epoch": 1.497504159733777,
      "grad_norm": 0.07802995294332504,
      "learning_rate": 0.00016666666666666666,
      "loss": 0.6281,
      "step": 675
    },
    {
      "epoch": 1.4997226844148641,
      "grad_norm": 0.08003750443458557,
      "learning_rate": 0.00016641975308641973,
      "loss": 0.7073,
      "step": 676
    },
    {
      "epoch": 1.5019412090959512,
      "grad_norm": 0.11089735478162766,
      "learning_rate": 0.00016617283950617282,
      "loss": 0.9149,
      "step": 677
    },
    {
      "epoch": 1.5041597337770383,
      "grad_norm": 0.10394146293401718,
      "learning_rate": 0.0001659259259259259,
      "loss": 0.6011,
      "step": 678
    },
    {
      "epoch": 1.5063782584581253,
      "grad_norm": 0.0678648054599762,
      "learning_rate": 0.000165679012345679,
      "loss": 0.7586,
      "step": 679
    },
    {
      "epoch": 1.5085967831392124,
      "grad_norm": 0.09287431836128235,
      "learning_rate": 0.0001654320987654321,
      "loss": 0.7414,
      "step": 680
    },
    {
      "epoch": 1.5108153078202995,
      "grad_norm": 0.08245453238487244,
      "learning_rate": 0.00016518518518518518,
      "loss": 0.8707,
      "step": 681
    },
    {
      "epoch": 1.5130338325013866,
      "grad_norm": 0.1531284749507904,
      "learning_rate": 0.00016493827160493827,
      "loss": 1.0187,
      "step": 682
    },
    {
      "epoch": 1.5152523571824736,
      "grad_norm": 0.07590752840042114,
      "learning_rate": 0.00016469135802469134,
      "loss": 0.7658,
      "step": 683
    },
    {
      "epoch": 1.5174708818635607,
      "grad_norm": 0.09907389432191849,
      "learning_rate": 0.00016444444444444444,
      "loss": 0.7871,
      "step": 684
    },
    {
      "epoch": 1.5196894065446478,
      "grad_norm": 0.08066434413194656,
      "learning_rate": 0.0001641975308641975,
      "loss": 0.7803,
      "step": 685
    },
    {
      "epoch": 1.5219079312257349,
      "grad_norm": 0.10961002111434937,
      "learning_rate": 0.0001639506172839506,
      "loss": 0.6947,
      "step": 686
    },
    {
      "epoch": 1.524126455906822,
      "grad_norm": 0.1419977992773056,
      "learning_rate": 0.0001637037037037037,
      "loss": 0.8577,
      "step": 687
    },
    {
      "epoch": 1.526344980587909,
      "grad_norm": 0.06945803761482239,
      "learning_rate": 0.00016345679012345676,
      "loss": 0.8783,
      "step": 688
    },
    {
      "epoch": 1.528563505268996,
      "grad_norm": 0.06904809176921844,
      "learning_rate": 0.00016320987654320988,
      "loss": 0.6837,
      "step": 689
    },
    {
      "epoch": 1.5307820299500832,
      "grad_norm": 0.09190991520881653,
      "learning_rate": 0.00016296296296296295,
      "loss": 0.6728,
      "step": 690
    },
    {
      "epoch": 1.5330005546311702,
      "grad_norm": 0.07218042761087418,
      "learning_rate": 0.00016271604938271605,
      "loss": 0.7445,
      "step": 691
    },
    {
      "epoch": 1.5352190793122573,
      "grad_norm": 0.08163771033287048,
      "learning_rate": 0.00016246913580246912,
      "loss": 0.8921,
      "step": 692
    },
    {
      "epoch": 1.5374376039933444,
      "grad_norm": 0.0839741975069046,
      "learning_rate": 0.0001622222222222222,
      "loss": 0.8418,
      "step": 693
    },
    {
      "epoch": 1.5396561286744315,
      "grad_norm": 0.09347599744796753,
      "learning_rate": 0.0001619753086419753,
      "loss": 0.8232,
      "step": 694
    },
    {
      "epoch": 1.5418746533555185,
      "grad_norm": 0.08164405822753906,
      "learning_rate": 0.00016172839506172837,
      "loss": 0.7315,
      "step": 695
    },
    {
      "epoch": 1.5440931780366056,
      "grad_norm": 0.09851223975419998,
      "learning_rate": 0.00016148148148148147,
      "loss": 0.8172,
      "step": 696
    },
    {
      "epoch": 1.5463117027176927,
      "grad_norm": 0.08209551125764847,
      "learning_rate": 0.00016123456790123454,
      "loss": 0.7401,
      "step": 697
    },
    {
      "epoch": 1.5485302273987798,
      "grad_norm": 0.08361760526895523,
      "learning_rate": 0.00016098765432098766,
      "loss": 0.7165,
      "step": 698
    },
    {
      "epoch": 1.5507487520798668,
      "grad_norm": 0.10286468267440796,
      "learning_rate": 0.00016074074074074073,
      "loss": 0.8366,
      "step": 699
    },
    {
      "epoch": 1.552967276760954,
      "grad_norm": 0.08021558821201324,
      "learning_rate": 0.00016049382716049382,
      "loss": 0.7026,
      "step": 700
    },
    {
      "epoch": 1.555185801442041,
      "grad_norm": 0.08461610227823257,
      "learning_rate": 0.00016024691358024692,
      "loss": 0.9098,
      "step": 701
    },
    {
      "epoch": 1.557404326123128,
      "grad_norm": 0.126322403550148,
      "learning_rate": 0.00015999999999999999,
      "loss": 0.8596,
      "step": 702
    },
    {
      "epoch": 1.5596228508042151,
      "grad_norm": 0.09586157649755478,
      "learning_rate": 0.00015975308641975308,
      "loss": 1.1198,
      "step": 703
    },
    {
      "epoch": 1.5618413754853022,
      "grad_norm": 0.07655948400497437,
      "learning_rate": 0.00015950617283950615,
      "loss": 0.7665,
      "step": 704
    },
    {
      "epoch": 1.5640599001663893,
      "grad_norm": 0.12633992731571198,
      "learning_rate": 0.00015925925925925924,
      "loss": 0.8389,
      "step": 705
    },
    {
      "epoch": 1.5662784248474764,
      "grad_norm": 0.07662942260503769,
      "learning_rate": 0.0001590123456790123,
      "loss": 0.7053,
      "step": 706
    },
    {
      "epoch": 1.5684969495285634,
      "grad_norm": 0.07341480255126953,
      "learning_rate": 0.00015876543209876543,
      "loss": 0.6781,
      "step": 707
    },
    {
      "epoch": 1.5707154742096505,
      "grad_norm": 0.06333422660827637,
      "learning_rate": 0.0001585185185185185,
      "loss": 0.6113,
      "step": 708
    },
    {
      "epoch": 1.5729339988907376,
      "grad_norm": 0.0694422721862793,
      "learning_rate": 0.0001582716049382716,
      "loss": 0.7617,
      "step": 709
    },
    {
      "epoch": 1.5751525235718247,
      "grad_norm": 0.05946282669901848,
      "learning_rate": 0.0001580246913580247,
      "loss": 0.6378,
      "step": 710
    },
    {
      "epoch": 1.5773710482529117,
      "grad_norm": 0.09391656517982483,
      "learning_rate": 0.00015777777777777776,
      "loss": 1.0933,
      "step": 711
    },
    {
      "epoch": 1.5795895729339988,
      "grad_norm": 0.08855625987052917,
      "learning_rate": 0.00015753086419753086,
      "loss": 0.7236,
      "step": 712
    },
    {
      "epoch": 1.581808097615086,
      "grad_norm": 0.10264936834573746,
      "learning_rate": 0.00015728395061728392,
      "loss": 0.5678,
      "step": 713
    },
    {
      "epoch": 1.584026622296173,
      "grad_norm": 0.08650382608175278,
      "learning_rate": 0.00015703703703703702,
      "loss": 0.891,
      "step": 714
    },
    {
      "epoch": 1.58624514697726,
      "grad_norm": 0.08085893094539642,
      "learning_rate": 0.0001567901234567901,
      "loss": 0.6829,
      "step": 715
    },
    {
      "epoch": 1.5884636716583471,
      "grad_norm": 0.09839639812707901,
      "learning_rate": 0.0001565432098765432,
      "loss": 0.8828,
      "step": 716
    },
    {
      "epoch": 1.5906821963394342,
      "grad_norm": 0.07129570096731186,
      "learning_rate": 0.0001562962962962963,
      "loss": 0.6563,
      "step": 717
    },
    {
      "epoch": 1.5929007210205213,
      "grad_norm": 0.10907716304063797,
      "learning_rate": 0.00015604938271604937,
      "loss": 0.7401,
      "step": 718
    },
    {
      "epoch": 1.5951192457016083,
      "grad_norm": 0.08592294156551361,
      "learning_rate": 0.00015580246913580247,
      "loss": 0.7372,
      "step": 719
    },
    {
      "epoch": 1.5973377703826954,
      "grad_norm": 0.08678999543190002,
      "learning_rate": 0.00015555555555555554,
      "loss": 0.86,
      "step": 720
    },
    {
      "epoch": 1.5995562950637825,
      "grad_norm": 0.07518002390861511,
      "learning_rate": 0.00015530864197530863,
      "loss": 0.7836,
      "step": 721
    },
    {
      "epoch": 1.6017748197448696,
      "grad_norm": 0.09843295812606812,
      "learning_rate": 0.0001550617283950617,
      "loss": 0.7549,
      "step": 722
    },
    {
      "epoch": 1.6039933444259566,
      "grad_norm": 0.10787026584148407,
      "learning_rate": 0.0001548148148148148,
      "loss": 0.5988,
      "step": 723
    },
    {
      "epoch": 1.6062118691070437,
      "grad_norm": 0.0600600391626358,
      "learning_rate": 0.00015456790123456792,
      "loss": 0.9047,
      "step": 724
    },
    {
      "epoch": 1.6084303937881308,
      "grad_norm": 0.0900898203253746,
      "learning_rate": 0.00015432098765432098,
      "loss": 0.6973,
      "step": 725
    },
    {
      "epoch": 1.6106489184692179,
      "grad_norm": 0.15169657766819,
      "learning_rate": 0.00015407407407407408,
      "loss": 0.8364,
      "step": 726
    },
    {
      "epoch": 1.612867443150305,
      "grad_norm": 0.12756381928920746,
      "learning_rate": 0.00015382716049382715,
      "loss": 0.8311,
      "step": 727
    },
    {
      "epoch": 1.615085967831392,
      "grad_norm": 0.10192670673131943,
      "learning_rate": 0.00015358024691358024,
      "loss": 0.8872,
      "step": 728
    },
    {
      "epoch": 1.617304492512479,
      "grad_norm": 0.0872483104467392,
      "learning_rate": 0.0001533333333333333,
      "loss": 0.8127,
      "step": 729
    },
    {
      "epoch": 1.6195230171935662,
      "grad_norm": 0.08723791688680649,
      "learning_rate": 0.0001530864197530864,
      "loss": 0.7803,
      "step": 730
    },
    {
      "epoch": 1.6217415418746532,
      "grad_norm": 0.06951799243688583,
      "learning_rate": 0.00015283950617283947,
      "loss": 0.7196,
      "step": 731
    },
    {
      "epoch": 1.6239600665557403,
      "grad_norm": 0.08409649133682251,
      "learning_rate": 0.00015259259259259257,
      "loss": 0.8482,
      "step": 732
    },
    {
      "epoch": 1.6261785912368274,
      "grad_norm": 0.06577211618423462,
      "learning_rate": 0.0001523456790123457,
      "loss": 0.599,
      "step": 733
    },
    {
      "epoch": 1.6283971159179145,
      "grad_norm": 0.08126416802406311,
      "learning_rate": 0.00015209876543209873,
      "loss": 0.6919,
      "step": 734
    },
    {
      "epoch": 1.6306156405990015,
      "grad_norm": 0.06883861869573593,
      "learning_rate": 0.00015185185185185185,
      "loss": 0.7611,
      "step": 735
    },
    {
      "epoch": 1.6328341652800886,
      "grad_norm": 0.10178206861019135,
      "learning_rate": 0.00015160493827160492,
      "loss": 0.7179,
      "step": 736
    },
    {
      "epoch": 1.6350526899611757,
      "grad_norm": 0.07397965341806412,
      "learning_rate": 0.00015135802469135802,
      "loss": 0.6954,
      "step": 737
    },
    {
      "epoch": 1.6372712146422628,
      "grad_norm": 0.08134113997220993,
      "learning_rate": 0.00015111111111111109,
      "loss": 0.7033,
      "step": 738
    },
    {
      "epoch": 1.6394897393233498,
      "grad_norm": 0.07948186248540878,
      "learning_rate": 0.00015086419753086418,
      "loss": 0.7725,
      "step": 739
    },
    {
      "epoch": 1.641708264004437,
      "grad_norm": 0.09669387340545654,
      "learning_rate": 0.00015061728395061728,
      "loss": 0.8167,
      "step": 740
    },
    {
      "epoch": 1.643926788685524,
      "grad_norm": 0.08828137069940567,
      "learning_rate": 0.00015037037037037034,
      "loss": 0.9298,
      "step": 741
    },
    {
      "epoch": 1.646145313366611,
      "grad_norm": 0.07561568915843964,
      "learning_rate": 0.00015012345679012347,
      "loss": 0.6471,
      "step": 742
    },
    {
      "epoch": 1.6483638380476981,
      "grad_norm": 0.155600443482399,
      "learning_rate": 0.00014987654320987653,
      "loss": 0.8467,
      "step": 743
    },
    {
      "epoch": 1.6505823627287852,
      "grad_norm": 0.08624161779880524,
      "learning_rate": 0.00014962962962962963,
      "loss": 0.7436,
      "step": 744
    },
    {
      "epoch": 1.6528008874098723,
      "grad_norm": 0.07259858399629593,
      "learning_rate": 0.0001493827160493827,
      "loss": 0.703,
      "step": 745
    },
    {
      "epoch": 1.6550194120909594,
      "grad_norm": 0.1027446761727333,
      "learning_rate": 0.0001491358024691358,
      "loss": 0.8024,
      "step": 746
    },
    {
      "epoch": 1.6572379367720464,
      "grad_norm": 0.08585378527641296,
      "learning_rate": 0.00014888888888888886,
      "loss": 0.8242,
      "step": 747
    },
    {
      "epoch": 1.6594564614531335,
      "grad_norm": 0.06362909823656082,
      "learning_rate": 0.00014864197530864196,
      "loss": 0.6891,
      "step": 748
    },
    {
      "epoch": 1.6616749861342206,
      "grad_norm": 0.08205240964889526,
      "learning_rate": 0.00014839506172839505,
      "loss": 0.6813,
      "step": 749
    },
    {
      "epoch": 1.6638935108153077,
      "grad_norm": 0.07566703855991364,
      "learning_rate": 0.00014814814814814812,
      "loss": 0.5107,
      "step": 750
    },
    {
      "epoch": 1.6661120354963947,
      "grad_norm": 0.07456362992525101,
      "learning_rate": 0.00014790123456790124,
      "loss": 0.6031,
      "step": 751
    },
    {
      "epoch": 1.6683305601774818,
      "grad_norm": 0.09219841659069061,
      "learning_rate": 0.0001476543209876543,
      "loss": 0.6426,
      "step": 752
    },
    {
      "epoch": 1.670549084858569,
      "grad_norm": 0.0928010642528534,
      "learning_rate": 0.0001474074074074074,
      "loss": 0.7122,
      "step": 753
    },
    {
      "epoch": 1.672767609539656,
      "grad_norm": 0.11300668865442276,
      "learning_rate": 0.00014716049382716047,
      "loss": 0.7999,
      "step": 754
    },
    {
      "epoch": 1.674986134220743,
      "grad_norm": 0.0735291913151741,
      "learning_rate": 0.00014691358024691357,
      "loss": 0.7845,
      "step": 755
    },
    {
      "epoch": 1.6772046589018301,
      "grad_norm": 0.0811205580830574,
      "learning_rate": 0.00014666666666666664,
      "loss": 0.8559,
      "step": 756
    },
    {
      "epoch": 1.6794231835829172,
      "grad_norm": 0.08844750374555588,
      "learning_rate": 0.00014641975308641973,
      "loss": 0.7839,
      "step": 757
    },
    {
      "epoch": 1.6816417082640043,
      "grad_norm": 0.07183957099914551,
      "learning_rate": 0.00014617283950617283,
      "loss": 0.7291,
      "step": 758
    },
    {
      "epoch": 1.6838602329450914,
      "grad_norm": 0.087847501039505,
      "learning_rate": 0.00014592592592592592,
      "loss": 0.7709,
      "step": 759
    },
    {
      "epoch": 1.6860787576261784,
      "grad_norm": 0.10700446367263794,
      "learning_rate": 0.000145679012345679,
      "loss": 0.8311,
      "step": 760
    },
    {
      "epoch": 1.6882972823072657,
      "grad_norm": 0.07175292819738388,
      "learning_rate": 0.00014543209876543208,
      "loss": 0.7852,
      "step": 761
    },
    {
      "epoch": 1.6905158069883528,
      "grad_norm": 0.09171941876411438,
      "learning_rate": 0.00014518518518518518,
      "loss": 0.7456,
      "step": 762
    },
    {
      "epoch": 1.6927343316694399,
      "grad_norm": 0.08928991109132767,
      "learning_rate": 0.00014493827160493825,
      "loss": 1.1414,
      "step": 763
    },
    {
      "epoch": 1.694952856350527,
      "grad_norm": 0.10604675859212875,
      "learning_rate": 0.00014469135802469134,
      "loss": 0.7842,
      "step": 764
    },
    {
      "epoch": 1.697171381031614,
      "grad_norm": 0.09763672947883606,
      "learning_rate": 0.0001444444444444444,
      "loss": 0.9452,
      "step": 765
    },
    {
      "epoch": 1.699389905712701,
      "grad_norm": 0.06990678608417511,
      "learning_rate": 0.00014419753086419753,
      "loss": 0.6906,
      "step": 766
    },
    {
      "epoch": 1.7016084303937882,
      "grad_norm": 0.09231503307819366,
      "learning_rate": 0.0001439506172839506,
      "loss": 0.9911,
      "step": 767
    },
    {
      "epoch": 1.7038269550748752,
      "grad_norm": 0.06659533083438873,
      "learning_rate": 0.0001437037037037037,
      "loss": 0.6095,
      "step": 768
    },
    {
      "epoch": 1.7060454797559623,
      "grad_norm": 0.1096971407532692,
      "learning_rate": 0.00014345679012345676,
      "loss": 0.7804,
      "step": 769
    },
    {
      "epoch": 1.7082640044370494,
      "grad_norm": 0.14034387469291687,
      "learning_rate": 0.00014320987654320986,
      "loss": 0.8909,
      "step": 770
    },
    {
      "epoch": 1.7104825291181365,
      "grad_norm": 0.06680990010499954,
      "learning_rate": 0.00014296296296296295,
      "loss": 0.6683,
      "step": 771
    },
    {
      "epoch": 1.7127010537992235,
      "grad_norm": 0.10662411898374557,
      "learning_rate": 0.00014271604938271602,
      "loss": 0.74,
      "step": 772
    },
    {
      "epoch": 1.7149195784803106,
      "grad_norm": 0.07847341895103455,
      "learning_rate": 0.00014246913580246912,
      "loss": 0.8337,
      "step": 773
    },
    {
      "epoch": 1.7171381031613977,
      "grad_norm": 0.07129880785942078,
      "learning_rate": 0.0001422222222222222,
      "loss": 0.8822,
      "step": 774
    },
    {
      "epoch": 1.7193566278424848,
      "grad_norm": 0.08240088820457458,
      "learning_rate": 0.0001419753086419753,
      "loss": 0.6587,
      "step": 775
    },
    {
      "epoch": 1.7215751525235718,
      "grad_norm": 0.07805512100458145,
      "learning_rate": 0.00014172839506172838,
      "loss": 0.7681,
      "step": 776
    },
    {
      "epoch": 1.723793677204659,
      "grad_norm": 0.09040573984384537,
      "learning_rate": 0.00014148148148148147,
      "loss": 0.9023,
      "step": 777
    },
    {
      "epoch": 1.726012201885746,
      "grad_norm": 0.09727460891008377,
      "learning_rate": 0.00014123456790123454,
      "loss": 0.8576,
      "step": 778
    },
    {
      "epoch": 1.728230726566833,
      "grad_norm": 0.07889188826084137,
      "learning_rate": 0.00014098765432098763,
      "loss": 0.8274,
      "step": 779
    },
    {
      "epoch": 1.7304492512479202,
      "grad_norm": 0.0736161321401596,
      "learning_rate": 0.00014074074074074073,
      "loss": 0.8024,
      "step": 780
    },
    {
      "epoch": 1.7326677759290072,
      "grad_norm": 0.07309345155954361,
      "learning_rate": 0.00014049382716049382,
      "loss": 0.6963,
      "step": 781
    },
    {
      "epoch": 1.7348863006100943,
      "grad_norm": 0.07392074912786484,
      "learning_rate": 0.0001402469135802469,
      "loss": 0.7634,
      "step": 782
    },
    {
      "epoch": 1.7371048252911814,
      "grad_norm": 0.08625944703817368,
      "learning_rate": 0.00014,
      "loss": 0.79,
      "step": 783
    },
    {
      "epoch": 1.7393233499722685,
      "grad_norm": 0.08888623118400574,
      "learning_rate": 0.00013975308641975308,
      "loss": 0.7403,
      "step": 784
    },
    {
      "epoch": 1.7415418746533555,
      "grad_norm": 0.08069104701280594,
      "learning_rate": 0.00013950617283950615,
      "loss": 0.7572,
      "step": 785
    },
    {
      "epoch": 1.7437603993344426,
      "grad_norm": 0.07853289693593979,
      "learning_rate": 0.00013925925925925925,
      "loss": 0.8291,
      "step": 786
    },
    {
      "epoch": 1.7459789240155297,
      "grad_norm": 0.08607794344425201,
      "learning_rate": 0.00013901234567901231,
      "loss": 0.6614,
      "step": 787
    },
    {
      "epoch": 1.7481974486966168,
      "grad_norm": 0.12110010534524918,
      "learning_rate": 0.00013876543209876544,
      "loss": 0.9438,
      "step": 788
    },
    {
      "epoch": 1.7504159733777038,
      "grad_norm": 0.06850749254226685,
      "learning_rate": 0.0001385185185185185,
      "loss": 0.6999,
      "step": 789
    },
    {
      "epoch": 1.752634498058791,
      "grad_norm": 0.09467557817697525,
      "learning_rate": 0.0001382716049382716,
      "loss": 0.6847,
      "step": 790
    },
    {
      "epoch": 1.754853022739878,
      "grad_norm": 0.0860932469367981,
      "learning_rate": 0.00013802469135802467,
      "loss": 0.7933,
      "step": 791
    },
    {
      "epoch": 1.757071547420965,
      "grad_norm": 0.07519947737455368,
      "learning_rate": 0.00013777777777777776,
      "loss": 0.6173,
      "step": 792
    },
    {
      "epoch": 1.7592900721020521,
      "grad_norm": 0.0802871510386467,
      "learning_rate": 0.00013753086419753086,
      "loss": 0.6573,
      "step": 793
    },
    {
      "epoch": 1.7615085967831392,
      "grad_norm": 0.09792372584342957,
      "learning_rate": 0.00013728395061728393,
      "loss": 0.7854,
      "step": 794
    },
    {
      "epoch": 1.7637271214642263,
      "grad_norm": 0.08905304223299026,
      "learning_rate": 0.00013703703703703702,
      "loss": 0.7873,
      "step": 795
    },
    {
      "epoch": 1.7659456461453134,
      "grad_norm": 0.08164618909358978,
      "learning_rate": 0.00013679012345679012,
      "loss": 0.8546,
      "step": 796
    },
    {
      "epoch": 1.7681641708264004,
      "grad_norm": 0.07909978181123734,
      "learning_rate": 0.0001365432098765432,
      "loss": 0.7505,
      "step": 797
    },
    {
      "epoch": 1.7703826955074875,
      "grad_norm": 0.07303280383348465,
      "learning_rate": 0.00013629629629629628,
      "loss": 0.7225,
      "step": 798
    },
    {
      "epoch": 1.7726012201885746,
      "grad_norm": 0.07738049328327179,
      "learning_rate": 0.00013604938271604937,
      "loss": 0.7751,
      "step": 799
    },
    {
      "epoch": 1.7748197448696617,
      "grad_norm": 0.08515508472919464,
      "learning_rate": 0.00013580246913580244,
      "loss": 0.7279,
      "step": 800
    },
    {
      "epoch": 1.7770382695507487,
      "grad_norm": 0.07488072663545609,
      "learning_rate": 0.00013555555555555554,
      "loss": 0.6963,
      "step": 801
    },
    {
      "epoch": 1.7792567942318358,
      "grad_norm": 0.08700244873762131,
      "learning_rate": 0.00013530864197530863,
      "loss": 0.6965,
      "step": 802
    },
    {
      "epoch": 1.7814753189129229,
      "grad_norm": 0.07716389745473862,
      "learning_rate": 0.00013506172839506173,
      "loss": 0.6722,
      "step": 803
    },
    {
      "epoch": 1.78369384359401,
      "grad_norm": 0.11183466762304306,
      "learning_rate": 0.0001348148148148148,
      "loss": 0.7615,
      "step": 804
    },
    {
      "epoch": 1.785912368275097,
      "grad_norm": 0.10434123128652573,
      "learning_rate": 0.0001345679012345679,
      "loss": 0.6339,
      "step": 805
    },
    {
      "epoch": 1.788130892956184,
      "grad_norm": 0.07478269934654236,
      "learning_rate": 0.00013432098765432099,
      "loss": 0.7107,
      "step": 806
    },
    {
      "epoch": 1.7903494176372712,
      "grad_norm": 0.10989516228437424,
      "learning_rate": 0.00013407407407407405,
      "loss": 0.7511,
      "step": 807
    },
    {
      "epoch": 1.7925679423183583,
      "grad_norm": 0.09178129583597183,
      "learning_rate": 0.00013382716049382715,
      "loss": 0.8628,
      "step": 808
    },
    {
      "epoch": 1.7947864669994453,
      "grad_norm": 0.08923018723726273,
      "learning_rate": 0.00013358024691358022,
      "loss": 0.6664,
      "step": 809
    },
    {
      "epoch": 1.7970049916805324,
      "grad_norm": 0.07539831846952438,
      "learning_rate": 0.0001333333333333333,
      "loss": 0.6674,
      "step": 810
    },
    {
      "epoch": 1.7992235163616195,
      "grad_norm": 0.0740896686911583,
      "learning_rate": 0.0001330864197530864,
      "loss": 0.6744,
      "step": 811
    },
    {
      "epoch": 1.8014420410427066,
      "grad_norm": 0.0719057023525238,
      "learning_rate": 0.0001328395061728395,
      "loss": 0.7474,
      "step": 812
    },
    {
      "epoch": 1.8036605657237936,
      "grad_norm": 0.06794360280036926,
      "learning_rate": 0.00013259259259259257,
      "loss": 0.7647,
      "step": 813
    },
    {
      "epoch": 1.8058790904048807,
      "grad_norm": 0.09509988129138947,
      "learning_rate": 0.00013234567901234567,
      "loss": 0.7377,
      "step": 814
    },
    {
      "epoch": 1.8080976150859678,
      "grad_norm": 0.0927988737821579,
      "learning_rate": 0.00013209876543209876,
      "loss": 0.6708,
      "step": 815
    },
    {
      "epoch": 1.8103161397670549,
      "grad_norm": 0.11492742598056793,
      "learning_rate": 0.00013185185185185183,
      "loss": 0.9936,
      "step": 816
    },
    {
      "epoch": 1.8125346644481422,
      "grad_norm": 0.09261582046747208,
      "learning_rate": 0.00013160493827160492,
      "loss": 0.753,
      "step": 817
    },
    {
      "epoch": 1.8147531891292292,
      "grad_norm": 0.07379738986492157,
      "learning_rate": 0.00013135802469135802,
      "loss": 0.7491,
      "step": 818
    },
    {
      "epoch": 1.8169717138103163,
      "grad_norm": 0.09494208544492722,
      "learning_rate": 0.00013111111111111111,
      "loss": 0.8415,
      "step": 819
    },
    {
      "epoch": 1.8191902384914034,
      "grad_norm": 0.09548774361610413,
      "learning_rate": 0.00013086419753086418,
      "loss": 0.8235,
      "step": 820
    },
    {
      "epoch": 1.8214087631724905,
      "grad_norm": 0.08251963555812836,
      "learning_rate": 0.00013061728395061728,
      "loss": 0.8295,
      "step": 821
    },
    {
      "epoch": 1.8236272878535775,
      "grad_norm": 0.10742012411355972,
      "learning_rate": 0.00013037037037037035,
      "loss": 0.8717,
      "step": 822
    },
    {
      "epoch": 1.8258458125346646,
      "grad_norm": 0.07618577778339386,
      "learning_rate": 0.00013012345679012344,
      "loss": 0.6623,
      "step": 823
    },
    {
      "epoch": 1.8280643372157517,
      "grad_norm": 0.09946849197149277,
      "learning_rate": 0.00012987654320987654,
      "loss": 0.9236,
      "step": 824
    },
    {
      "epoch": 1.8302828618968388,
      "grad_norm": 0.09540962427854538,
      "learning_rate": 0.0001296296296296296,
      "loss": 0.8319,
      "step": 825
    },
    {
      "epoch": 1.8325013865779258,
      "grad_norm": 0.09057582169771194,
      "learning_rate": 0.0001293827160493827,
      "loss": 0.892,
      "step": 826
    },
    {
      "epoch": 1.834719911259013,
      "grad_norm": 0.0668749213218689,
      "learning_rate": 0.0001291358024691358,
      "loss": 0.8425,
      "step": 827
    },
    {
      "epoch": 1.8369384359401,
      "grad_norm": 0.08880408853292465,
      "learning_rate": 0.0001288888888888889,
      "loss": 0.6818,
      "step": 828
    },
    {
      "epoch": 1.839156960621187,
      "grad_norm": 0.07133720815181732,
      "learning_rate": 0.00012864197530864196,
      "loss": 0.7923,
      "step": 829
    },
    {
      "epoch": 1.8413754853022741,
      "grad_norm": 0.10232459008693695,
      "learning_rate": 0.00012839506172839505,
      "loss": 0.7818,
      "step": 830
    },
    {
      "epoch": 1.8435940099833612,
      "grad_norm": 0.13368907570838928,
      "learning_rate": 0.00012814814814814812,
      "loss": 0.8435,
      "step": 831
    },
    {
      "epoch": 1.8458125346644483,
      "grad_norm": 0.07957194745540619,
      "learning_rate": 0.00012790123456790122,
      "loss": 0.8184,
      "step": 832
    },
    {
      "epoch": 1.8480310593455354,
      "grad_norm": 0.12279932200908661,
      "learning_rate": 0.0001276543209876543,
      "loss": 0.8706,
      "step": 833
    },
    {
      "epoch": 1.8502495840266224,
      "grad_norm": 0.06368744373321533,
      "learning_rate": 0.0001274074074074074,
      "loss": 0.6582,
      "step": 834
    },
    {
      "epoch": 1.8524681087077095,
      "grad_norm": 0.09765467792749405,
      "learning_rate": 0.00012716049382716047,
      "loss": 0.7949,
      "step": 835
    },
    {
      "epoch": 1.8546866333887966,
      "grad_norm": 0.08187538385391235,
      "learning_rate": 0.00012691358024691357,
      "loss": 0.9288,
      "step": 836
    },
    {
      "epoch": 1.8569051580698837,
      "grad_norm": 0.11214959621429443,
      "learning_rate": 0.00012666666666666666,
      "loss": 0.8867,
      "step": 837
    },
    {
      "epoch": 1.8591236827509707,
      "grad_norm": 0.09884259849786758,
      "learning_rate": 0.00012641975308641973,
      "loss": 0.751,
      "step": 838
    },
    {
      "epoch": 1.8613422074320578,
      "grad_norm": 0.10395622253417969,
      "learning_rate": 0.00012617283950617283,
      "loss": 0.6337,
      "step": 839
    },
    {
      "epoch": 1.8635607321131449,
      "grad_norm": 0.0749327763915062,
      "learning_rate": 0.0001259259259259259,
      "loss": 0.7225,
      "step": 840
    },
    {
      "epoch": 1.865779256794232,
      "grad_norm": 0.10912591964006424,
      "learning_rate": 0.00012567901234567902,
      "loss": 0.897,
      "step": 841
    },
    {
      "epoch": 1.867997781475319,
      "grad_norm": 0.08895165473222733,
      "learning_rate": 0.00012543209876543209,
      "loss": 0.8544,
      "step": 842
    },
    {
      "epoch": 1.870216306156406,
      "grad_norm": 0.11437104642391205,
      "learning_rate": 0.00012518518518518518,
      "loss": 0.7248,
      "step": 843
    },
    {
      "epoch": 1.8724348308374932,
      "grad_norm": 0.10211485624313354,
      "learning_rate": 0.00012493827160493825,
      "loss": 0.9415,
      "step": 844
    },
    {
      "epoch": 1.8746533555185803,
      "grad_norm": 0.07184267044067383,
      "learning_rate": 0.00012469135802469134,
      "loss": 0.7057,
      "step": 845
    },
    {
      "epoch": 1.8768718801996673,
      "grad_norm": 0.08426348865032196,
      "learning_rate": 0.00012444444444444444,
      "loss": 0.8352,
      "step": 846
    },
    {
      "epoch": 1.8790904048807544,
      "grad_norm": 0.07707478106021881,
      "learning_rate": 0.0001241975308641975,
      "loss": 0.722,
      "step": 847
    },
    {
      "epoch": 1.8813089295618415,
      "grad_norm": 0.0818880945444107,
      "learning_rate": 0.0001239506172839506,
      "loss": 0.7279,
      "step": 848
    },
    {
      "epoch": 1.8835274542429286,
      "grad_norm": 0.10362354665994644,
      "learning_rate": 0.0001237037037037037,
      "loss": 0.7468,
      "step": 849
    },
    {
      "epoch": 1.8857459789240156,
      "grad_norm": 0.08486620336771011,
      "learning_rate": 0.0001234567901234568,
      "loss": 0.792,
      "step": 850
    },
    {
      "epoch": 1.8879645036051027,
      "grad_norm": 0.07412763684988022,
      "learning_rate": 0.00012320987654320986,
      "loss": 0.762,
      "step": 851
    },
    {
      "epoch": 1.8901830282861898,
      "grad_norm": 0.07106179744005203,
      "learning_rate": 0.00012296296296296296,
      "loss": 0.7235,
      "step": 852
    },
    {
      "epoch": 1.8924015529672769,
      "grad_norm": 0.08058220148086548,
      "learning_rate": 0.00012271604938271602,
      "loss": 0.5891,
      "step": 853
    },
    {
      "epoch": 1.894620077648364,
      "grad_norm": 0.08627761900424957,
      "learning_rate": 0.00012246913580246912,
      "loss": 0.7186,
      "step": 854
    },
    {
      "epoch": 1.896838602329451,
      "grad_norm": 0.06967978924512863,
      "learning_rate": 0.00012222222222222221,
      "loss": 1.0381,
      "step": 855
    },
    {
      "epoch": 1.899057127010538,
      "grad_norm": 0.06846122443675995,
      "learning_rate": 0.00012197530864197531,
      "loss": 0.6236,
      "step": 856
    },
    {
      "epoch": 1.9012756516916252,
      "grad_norm": 0.09386609494686127,
      "learning_rate": 0.00012172839506172839,
      "loss": 0.6358,
      "step": 857
    },
    {
      "epoch": 1.9034941763727122,
      "grad_norm": 0.09192738682031631,
      "learning_rate": 0.00012148148148148147,
      "loss": 0.8997,
      "step": 858
    },
    {
      "epoch": 1.9057127010537993,
      "grad_norm": 0.1191820576786995,
      "learning_rate": 0.00012123456790123455,
      "loss": 0.9219,
      "step": 859
    },
    {
      "epoch": 1.9079312257348864,
      "grad_norm": 0.09395468235015869,
      "learning_rate": 0.00012098765432098764,
      "loss": 0.7924,
      "step": 860
    },
    {
      "epoch": 1.9101497504159735,
      "grad_norm": 0.06420453637838364,
      "learning_rate": 0.00012074074074074073,
      "loss": 0.9414,
      "step": 861
    },
    {
      "epoch": 1.9123682750970605,
      "grad_norm": 0.08387985825538635,
      "learning_rate": 0.00012049382716049381,
      "loss": 0.6279,
      "step": 862
    },
    {
      "epoch": 1.9145867997781476,
      "grad_norm": 0.08457259088754654,
      "learning_rate": 0.0001202469135802469,
      "loss": 0.6241,
      "step": 863
    },
    {
      "epoch": 1.9168053244592347,
      "grad_norm": 0.0935431569814682,
      "learning_rate": 0.00011999999999999999,
      "loss": 0.7878,
      "step": 864
    },
    {
      "epoch": 1.9190238491403218,
      "grad_norm": 0.07624135911464691,
      "learning_rate": 0.00011975308641975308,
      "loss": 0.7203,
      "step": 865
    },
    {
      "epoch": 1.9212423738214088,
      "grad_norm": 0.09238835424184799,
      "learning_rate": 0.00011950617283950617,
      "loss": 0.7342,
      "step": 866
    },
    {
      "epoch": 1.923460898502496,
      "grad_norm": 0.09092478454113007,
      "learning_rate": 0.00011925925925925925,
      "loss": 0.6432,
      "step": 867
    },
    {
      "epoch": 1.925679423183583,
      "grad_norm": 0.09091051667928696,
      "learning_rate": 0.00011901234567901233,
      "loss": 0.9473,
      "step": 868
    },
    {
      "epoch": 1.92789794786467,
      "grad_norm": 0.12042125314474106,
      "learning_rate": 0.00011876543209876541,
      "loss": 0.8331,
      "step": 869
    },
    {
      "epoch": 1.9301164725457571,
      "grad_norm": 0.0887221097946167,
      "learning_rate": 0.0001185185185185185,
      "loss": 0.9984,
      "step": 870
    },
    {
      "epoch": 1.9323349972268442,
      "grad_norm": 0.08738864213228226,
      "learning_rate": 0.0001182716049382716,
      "loss": 0.7843,
      "step": 871
    },
    {
      "epoch": 1.9345535219079313,
      "grad_norm": 0.09144040942192078,
      "learning_rate": 0.00011802469135802468,
      "loss": 0.8014,
      "step": 872
    },
    {
      "epoch": 1.9367720465890184,
      "grad_norm": 0.08861452341079712,
      "learning_rate": 0.00011777777777777776,
      "loss": 0.6684,
      "step": 873
    },
    {
      "epoch": 1.9389905712701054,
      "grad_norm": 0.09785947948694229,
      "learning_rate": 0.00011753086419753086,
      "loss": 0.7912,
      "step": 874
    },
    {
      "epoch": 1.9412090959511925,
      "grad_norm": 0.06605258584022522,
      "learning_rate": 0.00011728395061728394,
      "loss": 0.8642,
      "step": 875
    },
    {
      "epoch": 1.9434276206322796,
      "grad_norm": 0.08670786023139954,
      "learning_rate": 0.00011703703703703702,
      "loss": 0.7335,
      "step": 876
    },
    {
      "epoch": 1.9456461453133667,
      "grad_norm": 0.07528499513864517,
      "learning_rate": 0.0001167901234567901,
      "loss": 0.614,
      "step": 877
    },
    {
      "epoch": 1.9478646699944537,
      "grad_norm": 0.07546906173229218,
      "learning_rate": 0.00011654320987654319,
      "loss": 0.6974,
      "step": 878
    },
    {
      "epoch": 1.9500831946755408,
      "grad_norm": 0.08600408583879471,
      "learning_rate": 0.0001162962962962963,
      "loss": 0.7846,
      "step": 879
    },
    {
      "epoch": 1.9523017193566279,
      "grad_norm": 0.07703296840190887,
      "learning_rate": 0.00011604938271604938,
      "loss": 0.6807,
      "step": 880
    },
    {
      "epoch": 1.954520244037715,
      "grad_norm": 0.09106189012527466,
      "learning_rate": 0.00011580246913580246,
      "loss": 0.7902,
      "step": 881
    },
    {
      "epoch": 1.956738768718802,
      "grad_norm": 0.0718742311000824,
      "learning_rate": 0.00011555555555555554,
      "loss": 0.7718,
      "step": 882
    },
    {
      "epoch": 1.958957293399889,
      "grad_norm": 0.07902219146490097,
      "learning_rate": 0.00011530864197530863,
      "loss": 0.8776,
      "step": 883
    },
    {
      "epoch": 1.9611758180809762,
      "grad_norm": 0.09435555338859558,
      "learning_rate": 0.00011506172839506172,
      "loss": 0.7986,
      "step": 884
    },
    {
      "epoch": 1.9633943427620633,
      "grad_norm": 0.08558067679405212,
      "learning_rate": 0.0001148148148148148,
      "loss": 0.766,
      "step": 885
    },
    {
      "epoch": 1.9656128674431503,
      "grad_norm": 0.08999361842870712,
      "learning_rate": 0.00011456790123456789,
      "loss": 0.8277,
      "step": 886
    },
    {
      "epoch": 1.9678313921242374,
      "grad_norm": 0.09770625084638596,
      "learning_rate": 0.00011432098765432099,
      "loss": 0.8792,
      "step": 887
    },
    {
      "epoch": 1.9700499168053245,
      "grad_norm": 0.10224196314811707,
      "learning_rate": 0.00011407407407407407,
      "loss": 0.75,
      "step": 888
    },
    {
      "epoch": 1.9722684414864116,
      "grad_norm": 0.073021799325943,
      "learning_rate": 0.00011382716049382715,
      "loss": 0.6696,
      "step": 889
    },
    {
      "epoch": 1.9744869661674986,
      "grad_norm": 0.07111401855945587,
      "learning_rate": 0.00011358024691358023,
      "loss": 0.6426,
      "step": 890
    },
    {
      "epoch": 1.9767054908485857,
      "grad_norm": 0.15025469660758972,
      "learning_rate": 0.00011333333333333331,
      "loss": 0.9393,
      "step": 891
    },
    {
      "epoch": 1.9789240155296728,
      "grad_norm": 0.07760697603225708,
      "learning_rate": 0.00011308641975308641,
      "loss": 0.6009,
      "step": 892
    },
    {
      "epoch": 1.9811425402107599,
      "grad_norm": 0.0910193994641304,
      "learning_rate": 0.00011283950617283949,
      "loss": 0.7265,
      "step": 893
    },
    {
      "epoch": 1.983361064891847,
      "grad_norm": 0.07187125831842422,
      "learning_rate": 0.00011259259259259259,
      "loss": 0.6645,
      "step": 894
    },
    {
      "epoch": 1.985579589572934,
      "grad_norm": 0.09415319561958313,
      "learning_rate": 0.00011234567901234567,
      "loss": 0.8346,
      "step": 895
    },
    {
      "epoch": 1.987798114254021,
      "grad_norm": 0.08764591068029404,
      "learning_rate": 0.00011209876543209876,
      "loss": 0.7551,
      "step": 896
    },
    {
      "epoch": 1.9900166389351082,
      "grad_norm": 0.07882042229175568,
      "learning_rate": 0.00011185185185185184,
      "loss": 0.636,
      "step": 897
    },
    {
      "epoch": 1.9922351636161952,
      "grad_norm": 0.08374660462141037,
      "learning_rate": 0.00011160493827160493,
      "loss": 0.5942,
      "step": 898
    },
    {
      "epoch": 1.9944536882972823,
      "grad_norm": 0.06693168729543686,
      "learning_rate": 0.00011135802469135801,
      "loss": 0.7606,
      "step": 899
    },
    {
      "epoch": 1.9966722129783694,
      "grad_norm": 0.08837968856096268,
      "learning_rate": 0.00011111111111111109,
      "loss": 0.802,
      "step": 900
    },
    {
      "epoch": 1.9988907376594565,
      "grad_norm": 0.09674722701311111,
      "learning_rate": 0.0001108641975308642,
      "loss": 0.6848,
      "step": 901
    },
    {
      "epoch": 2.0011092623405435,
      "grad_norm": 0.06837698817253113,
      "learning_rate": 0.00011061728395061728,
      "loss": 0.7487,
      "step": 902
    },
    {
      "epoch": 2.0033277870216306,
      "grad_norm": 0.08834054321050644,
      "learning_rate": 0.00011037037037037036,
      "loss": 0.6675,
      "step": 903
    },
    {
      "epoch": 2.0055463117027177,
      "grad_norm": 0.10914549231529236,
      "learning_rate": 0.00011012345679012344,
      "loss": 0.8158,
      "step": 904
    },
    {
      "epoch": 2.0077648363838048,
      "grad_norm": 0.08255346119403839,
      "learning_rate": 0.00010987654320987654,
      "loss": 0.8264,
      "step": 905
    },
    {
      "epoch": 2.009983361064892,
      "grad_norm": 0.11768186092376709,
      "learning_rate": 0.00010962962962962962,
      "loss": 0.7322,
      "step": 906
    },
    {
      "epoch": 2.012201885745979,
      "grad_norm": 0.07289319485425949,
      "learning_rate": 0.0001093827160493827,
      "loss": 0.7154,
      "step": 907
    },
    {
      "epoch": 2.014420410427066,
      "grad_norm": 0.08181645721197128,
      "learning_rate": 0.00010913580246913578,
      "loss": 0.7839,
      "step": 908
    },
    {
      "epoch": 2.016638935108153,
      "grad_norm": 0.08076876401901245,
      "learning_rate": 0.00010888888888888889,
      "loss": 0.8674,
      "step": 909
    },
    {
      "epoch": 2.01885745978924,
      "grad_norm": 0.07641995698213577,
      "learning_rate": 0.00010864197530864197,
      "loss": 0.7721,
      "step": 910
    },
    {
      "epoch": 2.021075984470327,
      "grad_norm": 0.06739487498998642,
      "learning_rate": 0.00010839506172839505,
      "loss": 0.7823,
      "step": 911
    },
    {
      "epoch": 2.0232945091514143,
      "grad_norm": 0.08827894926071167,
      "learning_rate": 0.00010814814814814814,
      "loss": 0.7059,
      "step": 912
    },
    {
      "epoch": 2.0255130338325014,
      "grad_norm": 0.09905824065208435,
      "learning_rate": 0.00010790123456790122,
      "loss": 0.6577,
      "step": 913
    },
    {
      "epoch": 2.0277315585135884,
      "grad_norm": 0.07477370649576187,
      "learning_rate": 0.00010765432098765431,
      "loss": 0.7629,
      "step": 914
    },
    {
      "epoch": 2.0299500831946755,
      "grad_norm": 0.108803391456604,
      "learning_rate": 0.0001074074074074074,
      "loss": 0.7901,
      "step": 915
    },
    {
      "epoch": 2.0321686078757626,
      "grad_norm": 0.06811602413654327,
      "learning_rate": 0.00010716049382716048,
      "loss": 0.631,
      "step": 916
    },
    {
      "epoch": 2.0343871325568497,
      "grad_norm": 0.08448217064142227,
      "learning_rate": 0.00010691358024691357,
      "loss": 0.7035,
      "step": 917
    },
    {
      "epoch": 2.0366056572379367,
      "grad_norm": 0.08425777405500412,
      "learning_rate": 0.00010666666666666667,
      "loss": 0.6851,
      "step": 918
    },
    {
      "epoch": 2.038824181919024,
      "grad_norm": 0.08387485891580582,
      "learning_rate": 0.00010641975308641975,
      "loss": 0.8783,
      "step": 919
    },
    {
      "epoch": 2.041042706600111,
      "grad_norm": 0.07551386952400208,
      "learning_rate": 0.00010617283950617283,
      "loss": 0.6191,
      "step": 920
    },
    {
      "epoch": 2.043261231281198,
      "grad_norm": 0.07384245842695236,
      "learning_rate": 0.00010592592592592591,
      "loss": 0.654,
      "step": 921
    },
    {
      "epoch": 2.045479755962285,
      "grad_norm": 0.0785510390996933,
      "learning_rate": 0.00010567901234567899,
      "loss": 0.8227,
      "step": 922
    },
    {
      "epoch": 2.047698280643372,
      "grad_norm": 0.07068431377410889,
      "learning_rate": 0.00010543209876543209,
      "loss": 0.6417,
      "step": 923
    },
    {
      "epoch": 2.049916805324459,
      "grad_norm": 0.0701022669672966,
      "learning_rate": 0.00010518518518518518,
      "loss": 0.5913,
      "step": 924
    },
    {
      "epoch": 2.0521353300055463,
      "grad_norm": 0.06226358935236931,
      "learning_rate": 0.00010493827160493826,
      "loss": 0.6633,
      "step": 925
    },
    {
      "epoch": 2.0543538546866333,
      "grad_norm": 0.06851407885551453,
      "learning_rate": 0.00010469135802469135,
      "loss": 0.8161,
      "step": 926
    },
    {
      "epoch": 2.0565723793677204,
      "grad_norm": 0.06912954896688461,
      "learning_rate": 0.00010444444444444444,
      "loss": 0.7952,
      "step": 927
    },
    {
      "epoch": 2.0587909040488075,
      "grad_norm": 0.09206000715494156,
      "learning_rate": 0.00010419753086419752,
      "loss": 0.6219,
      "step": 928
    },
    {
      "epoch": 2.0610094287298946,
      "grad_norm": 0.08452209830284119,
      "learning_rate": 0.0001039506172839506,
      "loss": 0.858,
      "step": 929
    },
    {
      "epoch": 2.0632279534109816,
      "grad_norm": 0.07242301851511002,
      "learning_rate": 0.00010370370370370369,
      "loss": 0.7245,
      "step": 930
    },
    {
      "epoch": 2.0654464780920687,
      "grad_norm": 0.07564006000757217,
      "learning_rate": 0.00010345679012345677,
      "loss": 0.8432,
      "step": 931
    },
    {
      "epoch": 2.067665002773156,
      "grad_norm": 0.08008022606372833,
      "learning_rate": 0.00010320987654320988,
      "loss": 0.6292,
      "step": 932
    },
    {
      "epoch": 2.069883527454243,
      "grad_norm": 0.07764486223459244,
      "learning_rate": 0.00010296296296296296,
      "loss": 0.8579,
      "step": 933
    },
    {
      "epoch": 2.07210205213533,
      "grad_norm": 0.09481363743543625,
      "learning_rate": 0.00010271604938271604,
      "loss": 0.8864,
      "step": 934
    },
    {
      "epoch": 2.074320576816417,
      "grad_norm": 0.1099577248096466,
      "learning_rate": 0.00010246913580246912,
      "loss": 0.7115,
      "step": 935
    },
    {
      "epoch": 2.076539101497504,
      "grad_norm": 0.0744839757680893,
      "learning_rate": 0.00010222222222222222,
      "loss": 1.0108,
      "step": 936
    },
    {
      "epoch": 2.078757626178591,
      "grad_norm": 0.12358307093381882,
      "learning_rate": 0.0001019753086419753,
      "loss": 0.8417,
      "step": 937
    },
    {
      "epoch": 2.0809761508596782,
      "grad_norm": 0.08985494822263718,
      "learning_rate": 0.00010172839506172838,
      "loss": 0.6279,
      "step": 938
    },
    {
      "epoch": 2.0831946755407653,
      "grad_norm": 0.12909828126430511,
      "learning_rate": 0.00010148148148148147,
      "loss": 0.707,
      "step": 939
    },
    {
      "epoch": 2.0854132002218524,
      "grad_norm": 0.07320055365562439,
      "learning_rate": 0.00010123456790123457,
      "loss": 0.6747,
      "step": 940
    },
    {
      "epoch": 2.0876317249029395,
      "grad_norm": 0.08926205337047577,
      "learning_rate": 0.00010098765432098765,
      "loss": 0.8614,
      "step": 941
    },
    {
      "epoch": 2.0898502495840265,
      "grad_norm": 0.075705386698246,
      "learning_rate": 0.00010074074074074073,
      "loss": 0.8383,
      "step": 942
    },
    {
      "epoch": 2.0920687742651136,
      "grad_norm": 0.09681301563978195,
      "learning_rate": 0.00010049382716049381,
      "loss": 0.7514,
      "step": 943
    },
    {
      "epoch": 2.0942872989462007,
      "grad_norm": 0.07854966074228287,
      "learning_rate": 0.0001002469135802469,
      "loss": 0.6538,
      "step": 944
    },
    {
      "epoch": 2.0965058236272878,
      "grad_norm": 0.06550772488117218,
      "learning_rate": 9.999999999999999e-05,
      "loss": 0.9014,
      "step": 945
    },
    {
      "epoch": 2.098724348308375,
      "grad_norm": 0.09532083570957184,
      "learning_rate": 9.975308641975307e-05,
      "loss": 0.7991,
      "step": 946
    },
    {
      "epoch": 2.100942872989462,
      "grad_norm": 0.1048618033528328,
      "learning_rate": 9.950617283950617e-05,
      "loss": 0.7772,
      "step": 947
    },
    {
      "epoch": 2.103161397670549,
      "grad_norm": 0.1006808876991272,
      "learning_rate": 9.925925925925925e-05,
      "loss": 0.7753,
      "step": 948
    },
    {
      "epoch": 2.105379922351636,
      "grad_norm": 0.07441078871488571,
      "learning_rate": 9.901234567901234e-05,
      "loss": 0.7097,
      "step": 949
    },
    {
      "epoch": 2.107598447032723,
      "grad_norm": 0.08280546963214874,
      "learning_rate": 9.876543209876543e-05,
      "loss": 0.796,
      "step": 950
    },
    {
      "epoch": 2.10981697171381,
      "grad_norm": 0.10022372007369995,
      "learning_rate": 9.851851851851851e-05,
      "loss": 0.7997,
      "step": 951
    },
    {
      "epoch": 2.1120354963948973,
      "grad_norm": 0.07938052713871002,
      "learning_rate": 9.827160493827159e-05,
      "loss": 0.6655,
      "step": 952
    },
    {
      "epoch": 2.1142540210759844,
      "grad_norm": 0.09219428896903992,
      "learning_rate": 9.802469135802467e-05,
      "loss": 0.8008,
      "step": 953
    },
    {
      "epoch": 2.1164725457570714,
      "grad_norm": 0.09468308091163635,
      "learning_rate": 9.777777777777778e-05,
      "loss": 0.8239,
      "step": 954
    },
    {
      "epoch": 2.1186910704381585,
      "grad_norm": 0.07892098277807236,
      "learning_rate": 9.753086419753086e-05,
      "loss": 0.6642,
      "step": 955
    },
    {
      "epoch": 2.1209095951192456,
      "grad_norm": 0.08519411087036133,
      "learning_rate": 9.728395061728394e-05,
      "loss": 0.8452,
      "step": 956
    },
    {
      "epoch": 2.1231281198003327,
      "grad_norm": 0.09923538565635681,
      "learning_rate": 9.703703703703702e-05,
      "loss": 0.9251,
      "step": 957
    },
    {
      "epoch": 2.1253466444814197,
      "grad_norm": 0.06991422921419144,
      "learning_rate": 9.679012345679012e-05,
      "loss": 0.7693,
      "step": 958
    },
    {
      "epoch": 2.127565169162507,
      "grad_norm": 0.11210955679416656,
      "learning_rate": 9.65432098765432e-05,
      "loss": 0.7492,
      "step": 959
    },
    {
      "epoch": 2.129783693843594,
      "grad_norm": 0.08031900227069855,
      "learning_rate": 9.629629629629628e-05,
      "loss": 1.044,
      "step": 960
    },
    {
      "epoch": 2.132002218524681,
      "grad_norm": 0.07291650772094727,
      "learning_rate": 9.604938271604936e-05,
      "loss": 0.6802,
      "step": 961
    },
    {
      "epoch": 2.134220743205768,
      "grad_norm": 0.1030220314860344,
      "learning_rate": 9.580246913580247e-05,
      "loss": 1.0644,
      "step": 962
    },
    {
      "epoch": 2.136439267886855,
      "grad_norm": 0.10331052541732788,
      "learning_rate": 9.555555555555555e-05,
      "loss": 0.8429,
      "step": 963
    },
    {
      "epoch": 2.138657792567942,
      "grad_norm": 0.09096255153417587,
      "learning_rate": 9.530864197530864e-05,
      "loss": 0.7889,
      "step": 964
    },
    {
      "epoch": 2.1408763172490293,
      "grad_norm": 0.10837976634502411,
      "learning_rate": 9.506172839506172e-05,
      "loss": 0.8398,
      "step": 965
    },
    {
      "epoch": 2.1430948419301163,
      "grad_norm": 0.0779571384191513,
      "learning_rate": 9.48148148148148e-05,
      "loss": 0.822,
      "step": 966
    },
    {
      "epoch": 2.1453133666112034,
      "grad_norm": 0.09430311620235443,
      "learning_rate": 9.45679012345679e-05,
      "loss": 0.8965,
      "step": 967
    },
    {
      "epoch": 2.1475318912922905,
      "grad_norm": 0.06892245262861252,
      "learning_rate": 9.432098765432098e-05,
      "loss": 0.6527,
      "step": 968
    },
    {
      "epoch": 2.1497504159733776,
      "grad_norm": 0.08991076052188873,
      "learning_rate": 9.407407407407407e-05,
      "loss": 0.83,
      "step": 969
    },
    {
      "epoch": 2.1519689406544646,
      "grad_norm": 0.08407233655452728,
      "learning_rate": 9.382716049382715e-05,
      "loss": 0.8234,
      "step": 970
    },
    {
      "epoch": 2.1541874653355517,
      "grad_norm": 0.1319512128829956,
      "learning_rate": 9.358024691358025e-05,
      "loss": 0.5938,
      "step": 971
    },
    {
      "epoch": 2.156405990016639,
      "grad_norm": 0.07449740916490555,
      "learning_rate": 9.333333333333333e-05,
      "loss": 0.6004,
      "step": 972
    },
    {
      "epoch": 2.158624514697726,
      "grad_norm": 0.06854834407567978,
      "learning_rate": 9.308641975308641e-05,
      "loss": 0.7437,
      "step": 973
    },
    {
      "epoch": 2.160843039378813,
      "grad_norm": 0.06925453990697861,
      "learning_rate": 9.283950617283949e-05,
      "loss": 0.7961,
      "step": 974
    },
    {
      "epoch": 2.1630615640599,
      "grad_norm": 0.05940806493163109,
      "learning_rate": 9.259259259259257e-05,
      "loss": 0.7187,
      "step": 975
    },
    {
      "epoch": 2.165280088740987,
      "grad_norm": 0.07671648263931274,
      "learning_rate": 9.234567901234567e-05,
      "loss": 0.7082,
      "step": 976
    },
    {
      "epoch": 2.167498613422074,
      "grad_norm": 0.06501585990190506,
      "learning_rate": 9.209876543209876e-05,
      "loss": 0.4825,
      "step": 977
    },
    {
      "epoch": 2.1697171381031612,
      "grad_norm": 0.06785466521978378,
      "learning_rate": 9.185185185185185e-05,
      "loss": 0.6892,
      "step": 978
    },
    {
      "epoch": 2.1719356627842483,
      "grad_norm": 0.07500135898590088,
      "learning_rate": 9.160493827160493e-05,
      "loss": 1.078,
      "step": 979
    },
    {
      "epoch": 2.1741541874653354,
      "grad_norm": 0.09660045057535172,
      "learning_rate": 9.135802469135802e-05,
      "loss": 0.8714,
      "step": 980
    },
    {
      "epoch": 2.1763727121464225,
      "grad_norm": 0.09202384948730469,
      "learning_rate": 9.11111111111111e-05,
      "loss": 0.6773,
      "step": 981
    },
    {
      "epoch": 2.1785912368275095,
      "grad_norm": 0.08338309824466705,
      "learning_rate": 9.086419753086419e-05,
      "loss": 0.6466,
      "step": 982
    },
    {
      "epoch": 2.1808097615085966,
      "grad_norm": 0.10309063643217087,
      "learning_rate": 9.061728395061727e-05,
      "loss": 1.043,
      "step": 983
    },
    {
      "epoch": 2.1830282861896837,
      "grad_norm": 0.06168477609753609,
      "learning_rate": 9.037037037037038e-05,
      "loss": 0.6372,
      "step": 984
    },
    {
      "epoch": 2.1852468108707708,
      "grad_norm": 0.09867622703313828,
      "learning_rate": 9.012345679012346e-05,
      "loss": 0.7783,
      "step": 985
    },
    {
      "epoch": 2.187465335551858,
      "grad_norm": 0.0775771364569664,
      "learning_rate": 8.987654320987654e-05,
      "loss": 0.7654,
      "step": 986
    },
    {
      "epoch": 2.189683860232945,
      "grad_norm": 0.10528204590082169,
      "learning_rate": 8.962962962962962e-05,
      "loss": 0.5047,
      "step": 987
    },
    {
      "epoch": 2.191902384914032,
      "grad_norm": 0.07828556001186371,
      "learning_rate": 8.93827160493827e-05,
      "loss": 0.7214,
      "step": 988
    },
    {
      "epoch": 2.194120909595119,
      "grad_norm": 0.0828702375292778,
      "learning_rate": 8.91358024691358e-05,
      "loss": 0.6582,
      "step": 989
    },
    {
      "epoch": 2.196339434276206,
      "grad_norm": 0.08468307554721832,
      "learning_rate": 8.888888888888888e-05,
      "loss": 0.6903,
      "step": 990
    },
    {
      "epoch": 2.1985579589572932,
      "grad_norm": 0.0865941271185875,
      "learning_rate": 8.864197530864196e-05,
      "loss": 0.6952,
      "step": 991
    },
    {
      "epoch": 2.2007764836383803,
      "grad_norm": 0.070749431848526,
      "learning_rate": 8.839506172839506e-05,
      "loss": 0.6385,
      "step": 992
    },
    {
      "epoch": 2.2029950083194674,
      "grad_norm": 0.08216385543346405,
      "learning_rate": 8.814814814814815e-05,
      "loss": 1.1873,
      "step": 993
    },
    {
      "epoch": 2.2052135330005544,
      "grad_norm": 0.07159837335348129,
      "learning_rate": 8.790123456790123e-05,
      "loss": 0.6112,
      "step": 994
    },
    {
      "epoch": 2.2074320576816415,
      "grad_norm": 0.09212449938058853,
      "learning_rate": 8.765432098765431e-05,
      "loss": 0.5982,
      "step": 995
    },
    {
      "epoch": 2.2096505823627286,
      "grad_norm": 0.12332265079021454,
      "learning_rate": 8.74074074074074e-05,
      "loss": 0.9327,
      "step": 996
    },
    {
      "epoch": 2.2118691070438157,
      "grad_norm": 0.08651484549045563,
      "learning_rate": 8.716049382716048e-05,
      "loss": 0.6593,
      "step": 997
    },
    {
      "epoch": 2.2140876317249027,
      "grad_norm": 0.09251416474580765,
      "learning_rate": 8.691358024691357e-05,
      "loss": 0.7987,
      "step": 998
    },
    {
      "epoch": 2.21630615640599,
      "grad_norm": 0.08525733649730682,
      "learning_rate": 8.666666666666665e-05,
      "loss": 0.5614,
      "step": 999
    },
    {
      "epoch": 2.218524681087077,
      "grad_norm": 0.1015748381614685,
      "learning_rate": 8.641975308641975e-05,
      "loss": 0.7813,
      "step": 1000
    },
    {
      "epoch": 2.220743205768164,
      "grad_norm": 0.10311701893806458,
      "learning_rate": 8.617283950617283e-05,
      "loss": 0.7992,
      "step": 1001
    },
    {
      "epoch": 2.222961730449251,
      "grad_norm": 0.07063712924718857,
      "learning_rate": 8.592592592592593e-05,
      "loss": 0.7691,
      "step": 1002
    },
    {
      "epoch": 2.225180255130338,
      "grad_norm": 0.06508079916238785,
      "learning_rate": 8.567901234567901e-05,
      "loss": 0.6525,
      "step": 1003
    },
    {
      "epoch": 2.227398779811425,
      "grad_norm": 0.09282784909009933,
      "learning_rate": 8.543209876543209e-05,
      "loss": 0.6801,
      "step": 1004
    },
    {
      "epoch": 2.2296173044925123,
      "grad_norm": 0.10547066479921341,
      "learning_rate": 8.518518518518517e-05,
      "loss": 0.7518,
      "step": 1005
    },
    {
      "epoch": 2.2318358291735994,
      "grad_norm": 0.08423610776662827,
      "learning_rate": 8.493827160493825e-05,
      "loss": 0.8348,
      "step": 1006
    },
    {
      "epoch": 2.2340543538546864,
      "grad_norm": 0.10541067272424698,
      "learning_rate": 8.469135802469136e-05,
      "loss": 0.6505,
      "step": 1007
    },
    {
      "epoch": 2.2362728785357735,
      "grad_norm": 0.09162668883800507,
      "learning_rate": 8.444444444444444e-05,
      "loss": 0.8474,
      "step": 1008
    },
    {
      "epoch": 2.2384914032168606,
      "grad_norm": 0.0661270022392273,
      "learning_rate": 8.419753086419752e-05,
      "loss": 0.7677,
      "step": 1009
    },
    {
      "epoch": 2.2407099278979477,
      "grad_norm": 0.07047272473573685,
      "learning_rate": 8.39506172839506e-05,
      "loss": 0.6498,
      "step": 1010
    },
    {
      "epoch": 2.2429284525790347,
      "grad_norm": 0.08262878656387329,
      "learning_rate": 8.37037037037037e-05,
      "loss": 0.67,
      "step": 1011
    },
    {
      "epoch": 2.245146977260122,
      "grad_norm": 0.0813671424984932,
      "learning_rate": 8.345679012345678e-05,
      "loss": 0.6182,
      "step": 1012
    },
    {
      "epoch": 2.247365501941209,
      "grad_norm": 0.08854053169488907,
      "learning_rate": 8.320987654320986e-05,
      "loss": 0.7642,
      "step": 1013
    },
    {
      "epoch": 2.249584026622296,
      "grad_norm": 0.07848663628101349,
      "learning_rate": 8.296296296296295e-05,
      "loss": 0.9043,
      "step": 1014
    },
    {
      "epoch": 2.251802551303383,
      "grad_norm": 0.0837557390332222,
      "learning_rate": 8.271604938271605e-05,
      "loss": 0.8882,
      "step": 1015
    },
    {
      "epoch": 2.25402107598447,
      "grad_norm": 0.0727316290140152,
      "learning_rate": 8.246913580246914e-05,
      "loss": 0.8159,
      "step": 1016
    },
    {
      "epoch": 2.256239600665557,
      "grad_norm": 0.0704192966222763,
      "learning_rate": 8.222222222222222e-05,
      "loss": 0.9112,
      "step": 1017
    },
    {
      "epoch": 2.2584581253466443,
      "grad_norm": 0.0913740023970604,
      "learning_rate": 8.19753086419753e-05,
      "loss": 0.9644,
      "step": 1018
    },
    {
      "epoch": 2.2606766500277313,
      "grad_norm": 0.07699087262153625,
      "learning_rate": 8.172839506172838e-05,
      "loss": 0.8758,
      "step": 1019
    },
    {
      "epoch": 2.2628951747088184,
      "grad_norm": 0.10226847976446152,
      "learning_rate": 8.148148148148148e-05,
      "loss": 0.8811,
      "step": 1020
    },
    {
      "epoch": 2.2651136993899055,
      "grad_norm": 0.1038108691573143,
      "learning_rate": 8.123456790123456e-05,
      "loss": 0.8378,
      "step": 1021
    },
    {
      "epoch": 2.2673322240709926,
      "grad_norm": 0.09710036218166351,
      "learning_rate": 8.098765432098765e-05,
      "loss": 0.7596,
      "step": 1022
    },
    {
      "epoch": 2.2695507487520796,
      "grad_norm": 0.08344046771526337,
      "learning_rate": 8.074074074074073e-05,
      "loss": 0.8215,
      "step": 1023
    },
    {
      "epoch": 2.2717692734331667,
      "grad_norm": 0.08593735843896866,
      "learning_rate": 8.049382716049383e-05,
      "loss": 0.8451,
      "step": 1024
    },
    {
      "epoch": 2.2739877981142538,
      "grad_norm": 0.06035824492573738,
      "learning_rate": 8.024691358024691e-05,
      "loss": 0.7184,
      "step": 1025
    },
    {
      "epoch": 2.276206322795341,
      "grad_norm": 0.08864479511976242,
      "learning_rate": 7.999999999999999e-05,
      "loss": 0.8628,
      "step": 1026
    },
    {
      "epoch": 2.278424847476428,
      "grad_norm": 0.11448577046394348,
      "learning_rate": 7.975308641975307e-05,
      "loss": 0.7714,
      "step": 1027
    },
    {
      "epoch": 2.280643372157515,
      "grad_norm": 0.097632996737957,
      "learning_rate": 7.950617283950616e-05,
      "loss": 0.666,
      "step": 1028
    },
    {
      "epoch": 2.2828618968386025,
      "grad_norm": 0.07832937687635422,
      "learning_rate": 7.925925925925925e-05,
      "loss": 0.711,
      "step": 1029
    },
    {
      "epoch": 2.2850804215196896,
      "grad_norm": 0.0968175008893013,
      "learning_rate": 7.901234567901235e-05,
      "loss": 0.9498,
      "step": 1030
    },
    {
      "epoch": 2.2872989462007767,
      "grad_norm": 0.08657008409500122,
      "learning_rate": 7.876543209876543e-05,
      "loss": 0.8749,
      "step": 1031
    },
    {
      "epoch": 2.2895174708818637,
      "grad_norm": 0.07002072781324387,
      "learning_rate": 7.851851851851851e-05,
      "loss": 0.7941,
      "step": 1032
    },
    {
      "epoch": 2.291735995562951,
      "grad_norm": 0.0886651873588562,
      "learning_rate": 7.82716049382716e-05,
      "loss": 0.6292,
      "step": 1033
    },
    {
      "epoch": 2.293954520244038,
      "grad_norm": 0.08392421901226044,
      "learning_rate": 7.802469135802469e-05,
      "loss": 0.9023,
      "step": 1034
    },
    {
      "epoch": 2.296173044925125,
      "grad_norm": 0.07099150866270065,
      "learning_rate": 7.777777777777777e-05,
      "loss": 0.6068,
      "step": 1035
    },
    {
      "epoch": 2.298391569606212,
      "grad_norm": 0.0901346206665039,
      "learning_rate": 7.753086419753085e-05,
      "loss": 0.74,
      "step": 1036
    },
    {
      "epoch": 2.300610094287299,
      "grad_norm": 0.07137968391180038,
      "learning_rate": 7.728395061728396e-05,
      "loss": 0.6497,
      "step": 1037
    },
    {
      "epoch": 2.302828618968386,
      "grad_norm": 0.060215216130018234,
      "learning_rate": 7.703703703703704e-05,
      "loss": 0.8158,
      "step": 1038
    },
    {
      "epoch": 2.3050471436494733,
      "grad_norm": 0.07373488694429398,
      "learning_rate": 7.679012345679012e-05,
      "loss": 0.709,
      "step": 1039
    },
    {
      "epoch": 2.3072656683305603,
      "grad_norm": 0.09427979588508606,
      "learning_rate": 7.65432098765432e-05,
      "loss": 0.689,
      "step": 1040
    },
    {
      "epoch": 2.3094841930116474,
      "grad_norm": 0.07258276641368866,
      "learning_rate": 7.629629629629628e-05,
      "loss": 0.913,
      "step": 1041
    },
    {
      "epoch": 2.3117027176927345,
      "grad_norm": 0.08229970186948776,
      "learning_rate": 7.604938271604937e-05,
      "loss": 0.7388,
      "step": 1042
    },
    {
      "epoch": 2.3139212423738216,
      "grad_norm": 0.09133493900299072,
      "learning_rate": 7.580246913580246e-05,
      "loss": 0.7412,
      "step": 1043
    },
    {
      "epoch": 2.3161397670549086,
      "grad_norm": 0.0668063834309578,
      "learning_rate": 7.555555555555554e-05,
      "loss": 0.6906,
      "step": 1044
    },
    {
      "epoch": 2.3183582917359957,
      "grad_norm": 0.13640889525413513,
      "learning_rate": 7.530864197530864e-05,
      "loss": 0.8094,
      "step": 1045
    },
    {
      "epoch": 2.320576816417083,
      "grad_norm": 0.07781457901000977,
      "learning_rate": 7.506172839506173e-05,
      "loss": 0.8931,
      "step": 1046
    },
    {
      "epoch": 2.32279534109817,
      "grad_norm": 0.09848855435848236,
      "learning_rate": 7.481481481481481e-05,
      "loss": 0.6915,
      "step": 1047
    },
    {
      "epoch": 2.325013865779257,
      "grad_norm": 0.07313980907201767,
      "learning_rate": 7.45679012345679e-05,
      "loss": 0.7047,
      "step": 1048
    },
    {
      "epoch": 2.327232390460344,
      "grad_norm": 0.13725405931472778,
      "learning_rate": 7.432098765432098e-05,
      "loss": 0.8477,
      "step": 1049
    },
    {
      "epoch": 2.329450915141431,
      "grad_norm": 0.0820774957537651,
      "learning_rate": 7.407407407407406e-05,
      "loss": 0.7881,
      "step": 1050
    },
    {
      "epoch": 2.331669439822518,
      "grad_norm": 0.07415801286697388,
      "learning_rate": 7.382716049382715e-05,
      "loss": 0.7455,
      "step": 1051
    },
    {
      "epoch": 2.3338879645036053,
      "grad_norm": 0.07658666372299194,
      "learning_rate": 7.358024691358024e-05,
      "loss": 0.6171,
      "step": 1052
    },
    {
      "epoch": 2.3361064891846923,
      "grad_norm": 0.08392451703548431,
      "learning_rate": 7.333333333333332e-05,
      "loss": 0.9515,
      "step": 1053
    },
    {
      "epoch": 2.3383250138657794,
      "grad_norm": 0.08976923674345016,
      "learning_rate": 7.308641975308641e-05,
      "loss": 0.7804,
      "step": 1054
    },
    {
      "epoch": 2.3405435385468665,
      "grad_norm": 0.09554589539766312,
      "learning_rate": 7.28395061728395e-05,
      "loss": 0.7953,
      "step": 1055
    },
    {
      "epoch": 2.3427620632279536,
      "grad_norm": 0.06892561912536621,
      "learning_rate": 7.259259259259259e-05,
      "loss": 0.7954,
      "step": 1056
    },
    {
      "epoch": 2.3449805879090406,
      "grad_norm": 0.09343255311250687,
      "learning_rate": 7.234567901234567e-05,
      "loss": 0.7436,
      "step": 1057
    },
    {
      "epoch": 2.3471991125901277,
      "grad_norm": 0.08782149851322174,
      "learning_rate": 7.209876543209877e-05,
      "loss": 0.7925,
      "step": 1058
    },
    {
      "epoch": 2.3494176372712148,
      "grad_norm": 0.08092944324016571,
      "learning_rate": 7.185185185185185e-05,
      "loss": 0.7464,
      "step": 1059
    },
    {
      "epoch": 2.351636161952302,
      "grad_norm": 0.0828278511762619,
      "learning_rate": 7.160493827160493e-05,
      "loss": 0.6895,
      "step": 1060
    },
    {
      "epoch": 2.353854686633389,
      "grad_norm": 0.11581443250179291,
      "learning_rate": 7.135802469135801e-05,
      "loss": 0.6967,
      "step": 1061
    },
    {
      "epoch": 2.356073211314476,
      "grad_norm": 0.06735804677009583,
      "learning_rate": 7.11111111111111e-05,
      "loss": 0.7096,
      "step": 1062
    },
    {
      "epoch": 2.358291735995563,
      "grad_norm": 0.07137575000524521,
      "learning_rate": 7.086419753086419e-05,
      "loss": 0.6754,
      "step": 1063
    },
    {
      "epoch": 2.36051026067665,
      "grad_norm": 0.10158606618642807,
      "learning_rate": 7.061728395061727e-05,
      "loss": 0.8366,
      "step": 1064
    },
    {
      "epoch": 2.3627287853577372,
      "grad_norm": 0.07599575817584991,
      "learning_rate": 7.037037037037036e-05,
      "loss": 0.6758,
      "step": 1065
    },
    {
      "epoch": 2.3649473100388243,
      "grad_norm": 0.06353987753391266,
      "learning_rate": 7.012345679012345e-05,
      "loss": 0.6568,
      "step": 1066
    },
    {
      "epoch": 2.3671658347199114,
      "grad_norm": 0.10119745880365372,
      "learning_rate": 6.987654320987654e-05,
      "loss": 0.9085,
      "step": 1067
    },
    {
      "epoch": 2.3693843594009985,
      "grad_norm": 0.08269741386175156,
      "learning_rate": 6.962962962962962e-05,
      "loss": 0.9581,
      "step": 1068
    },
    {
      "epoch": 2.3716028840820855,
      "grad_norm": 0.06470052897930145,
      "learning_rate": 6.938271604938272e-05,
      "loss": 0.5513,
      "step": 1069
    },
    {
      "epoch": 2.3738214087631726,
      "grad_norm": 0.08577947318553925,
      "learning_rate": 6.91358024691358e-05,
      "loss": 0.6965,
      "step": 1070
    },
    {
      "epoch": 2.3760399334442597,
      "grad_norm": 0.09072365611791611,
      "learning_rate": 6.888888888888888e-05,
      "loss": 0.924,
      "step": 1071
    },
    {
      "epoch": 2.3782584581253468,
      "grad_norm": 0.061750300228595734,
      "learning_rate": 6.864197530864196e-05,
      "loss": 0.8078,
      "step": 1072
    },
    {
      "epoch": 2.380476982806434,
      "grad_norm": 0.08044982701539993,
      "learning_rate": 6.839506172839506e-05,
      "loss": 0.8525,
      "step": 1073
    },
    {
      "epoch": 2.382695507487521,
      "grad_norm": 0.075640469789505,
      "learning_rate": 6.814814814814814e-05,
      "loss": 0.8906,
      "step": 1074
    },
    {
      "epoch": 2.384914032168608,
      "grad_norm": 0.08767558634281158,
      "learning_rate": 6.790123456790122e-05,
      "loss": 0.6719,
      "step": 1075
    },
    {
      "epoch": 2.387132556849695,
      "grad_norm": 0.11405991017818451,
      "learning_rate": 6.765432098765432e-05,
      "loss": 0.7389,
      "step": 1076
    },
    {
      "epoch": 2.389351081530782,
      "grad_norm": 0.08391735702753067,
      "learning_rate": 6.74074074074074e-05,
      "loss": 0.9655,
      "step": 1077
    },
    {
      "epoch": 2.391569606211869,
      "grad_norm": 0.08459236472845078,
      "learning_rate": 6.716049382716049e-05,
      "loss": 0.9931,
      "step": 1078
    },
    {
      "epoch": 2.3937881308929563,
      "grad_norm": 0.11496855318546295,
      "learning_rate": 6.691358024691357e-05,
      "loss": 0.7821,
      "step": 1079
    },
    {
      "epoch": 2.3960066555740434,
      "grad_norm": 0.09169444441795349,
      "learning_rate": 6.666666666666666e-05,
      "loss": 0.6858,
      "step": 1080
    },
    {
      "epoch": 2.3982251802551304,
      "grad_norm": 0.08257808536291122,
      "learning_rate": 6.641975308641975e-05,
      "loss": 0.7982,
      "step": 1081
    },
    {
      "epoch": 2.4004437049362175,
      "grad_norm": 0.07889552414417267,
      "learning_rate": 6.617283950617283e-05,
      "loss": 0.6804,
      "step": 1082
    },
    {
      "epoch": 2.4026622296173046,
      "grad_norm": 0.06720870733261108,
      "learning_rate": 6.592592592592591e-05,
      "loss": 0.7291,
      "step": 1083
    },
    {
      "epoch": 2.4048807542983917,
      "grad_norm": 0.08785364031791687,
      "learning_rate": 6.567901234567901e-05,
      "loss": 0.7121,
      "step": 1084
    },
    {
      "epoch": 2.4070992789794787,
      "grad_norm": 0.07742965966463089,
      "learning_rate": 6.543209876543209e-05,
      "loss": 0.6073,
      "step": 1085
    },
    {
      "epoch": 2.409317803660566,
      "grad_norm": 0.07000548392534256,
      "learning_rate": 6.518518518518517e-05,
      "loss": 0.6456,
      "step": 1086
    },
    {
      "epoch": 2.411536328341653,
      "grad_norm": 0.09608156979084015,
      "learning_rate": 6.493827160493827e-05,
      "loss": 0.6672,
      "step": 1087
    },
    {
      "epoch": 2.41375485302274,
      "grad_norm": 0.10120079666376114,
      "learning_rate": 6.469135802469135e-05,
      "loss": 0.6855,
      "step": 1088
    },
    {
      "epoch": 2.415973377703827,
      "grad_norm": 0.07225026190280914,
      "learning_rate": 6.444444444444444e-05,
      "loss": 0.6607,
      "step": 1089
    },
    {
      "epoch": 2.418191902384914,
      "grad_norm": 0.08440103381872177,
      "learning_rate": 6.419753086419753e-05,
      "loss": 0.7397,
      "step": 1090
    },
    {
      "epoch": 2.420410427066001,
      "grad_norm": 0.07493556290864944,
      "learning_rate": 6.395061728395061e-05,
      "loss": 0.7419,
      "step": 1091
    },
    {
      "epoch": 2.4226289517470883,
      "grad_norm": 0.08209782093763351,
      "learning_rate": 6.37037037037037e-05,
      "loss": 0.6463,
      "step": 1092
    },
    {
      "epoch": 2.4248474764281753,
      "grad_norm": 0.08170017600059509,
      "learning_rate": 6.345679012345678e-05,
      "loss": 0.7178,
      "step": 1093
    },
    {
      "epoch": 2.4270660011092624,
      "grad_norm": 0.09155097603797913,
      "learning_rate": 6.320987654320987e-05,
      "loss": 0.9165,
      "step": 1094
    },
    {
      "epoch": 2.4292845257903495,
      "grad_norm": 0.07250907272100449,
      "learning_rate": 6.296296296296295e-05,
      "loss": 0.6021,
      "step": 1095
    },
    {
      "epoch": 2.4315030504714366,
      "grad_norm": 0.07573800534009933,
      "learning_rate": 6.271604938271604e-05,
      "loss": 0.7827,
      "step": 1096
    },
    {
      "epoch": 2.4337215751525236,
      "grad_norm": 0.08711908757686615,
      "learning_rate": 6.246913580246912e-05,
      "loss": 0.7233,
      "step": 1097
    },
    {
      "epoch": 2.4359400998336107,
      "grad_norm": 0.07622338086366653,
      "learning_rate": 6.222222222222222e-05,
      "loss": 0.7323,
      "step": 1098
    },
    {
      "epoch": 2.438158624514698,
      "grad_norm": 0.08349321037530899,
      "learning_rate": 6.19753086419753e-05,
      "loss": 0.6496,
      "step": 1099
    },
    {
      "epoch": 2.440377149195785,
      "grad_norm": 0.06779665499925613,
      "learning_rate": 6.17283950617284e-05,
      "loss": 0.6497,
      "step": 1100
    },
    {
      "epoch": 2.442595673876872,
      "grad_norm": 0.07832888513803482,
      "learning_rate": 6.148148148148148e-05,
      "loss": 0.7358,
      "step": 1101
    },
    {
      "epoch": 2.444814198557959,
      "grad_norm": 0.07454412430524826,
      "learning_rate": 6.123456790123456e-05,
      "loss": 0.7402,
      "step": 1102
    },
    {
      "epoch": 2.447032723239046,
      "grad_norm": 0.07566478848457336,
      "learning_rate": 6.0987654320987655e-05,
      "loss": 0.8014,
      "step": 1103
    },
    {
      "epoch": 2.449251247920133,
      "grad_norm": 0.07449538260698318,
      "learning_rate": 6.0740740740740736e-05,
      "loss": 0.5709,
      "step": 1104
    },
    {
      "epoch": 2.4514697726012202,
      "grad_norm": 0.06790108978748322,
      "learning_rate": 6.049382716049382e-05,
      "loss": 0.6242,
      "step": 1105
    },
    {
      "epoch": 2.4536882972823073,
      "grad_norm": 0.08497802168130875,
      "learning_rate": 6.0246913580246906e-05,
      "loss": 0.6833,
      "step": 1106
    },
    {
      "epoch": 2.4559068219633944,
      "grad_norm": 0.10092911869287491,
      "learning_rate": 5.9999999999999995e-05,
      "loss": 0.7926,
      "step": 1107
    },
    {
      "epoch": 2.4581253466444815,
      "grad_norm": 0.06707913428544998,
      "learning_rate": 5.975308641975308e-05,
      "loss": 0.6341,
      "step": 1108
    },
    {
      "epoch": 2.4603438713255685,
      "grad_norm": 0.11700978875160217,
      "learning_rate": 5.9506172839506165e-05,
      "loss": 0.546,
      "step": 1109
    },
    {
      "epoch": 2.4625623960066556,
      "grad_norm": 0.09489001333713531,
      "learning_rate": 5.925925925925925e-05,
      "loss": 0.7302,
      "step": 1110
    },
    {
      "epoch": 2.4647809206877427,
      "grad_norm": 0.0832776203751564,
      "learning_rate": 5.901234567901234e-05,
      "loss": 0.6558,
      "step": 1111
    },
    {
      "epoch": 2.4669994453688298,
      "grad_norm": 0.06944943964481354,
      "learning_rate": 5.876543209876543e-05,
      "loss": 0.633,
      "step": 1112
    },
    {
      "epoch": 2.469217970049917,
      "grad_norm": 0.09864591807126999,
      "learning_rate": 5.851851851851851e-05,
      "loss": 0.7404,
      "step": 1113
    },
    {
      "epoch": 2.471436494731004,
      "grad_norm": 0.07334929704666138,
      "learning_rate": 5.827160493827159e-05,
      "loss": 0.7179,
      "step": 1114
    },
    {
      "epoch": 2.473655019412091,
      "grad_norm": 0.08204148709774017,
      "learning_rate": 5.802469135802469e-05,
      "loss": 0.6398,
      "step": 1115
    },
    {
      "epoch": 2.475873544093178,
      "grad_norm": 0.12183347344398499,
      "learning_rate": 5.777777777777777e-05,
      "loss": 1.0156,
      "step": 1116
    },
    {
      "epoch": 2.478092068774265,
      "grad_norm": 0.08889223635196686,
      "learning_rate": 5.753086419753086e-05,
      "loss": 0.7459,
      "step": 1117
    },
    {
      "epoch": 2.480310593455352,
      "grad_norm": 0.09667259454727173,
      "learning_rate": 5.7283950617283946e-05,
      "loss": 0.6785,
      "step": 1118
    },
    {
      "epoch": 2.4825291181364393,
      "grad_norm": 0.09557890892028809,
      "learning_rate": 5.7037037037037035e-05,
      "loss": 0.7397,
      "step": 1119
    },
    {
      "epoch": 2.4847476428175264,
      "grad_norm": 0.06919311732053757,
      "learning_rate": 5.6790123456790116e-05,
      "loss": 0.7535,
      "step": 1120
    },
    {
      "epoch": 2.4869661674986134,
      "grad_norm": 0.09841174632310867,
      "learning_rate": 5.6543209876543205e-05,
      "loss": 0.7438,
      "step": 1121
    },
    {
      "epoch": 2.4891846921797005,
      "grad_norm": 0.07316316664218903,
      "learning_rate": 5.629629629629629e-05,
      "loss": 0.8427,
      "step": 1122
    },
    {
      "epoch": 2.4914032168607876,
      "grad_norm": 0.09466356784105301,
      "learning_rate": 5.604938271604938e-05,
      "loss": 0.6953,
      "step": 1123
    },
    {
      "epoch": 2.4936217415418747,
      "grad_norm": 0.07418382912874222,
      "learning_rate": 5.580246913580246e-05,
      "loss": 0.815,
      "step": 1124
    },
    {
      "epoch": 2.4958402662229617,
      "grad_norm": 0.06377240270376205,
      "learning_rate": 5.5555555555555545e-05,
      "loss": 0.8553,
      "step": 1125
    },
    {
      "epoch": 2.498058790904049,
      "grad_norm": 0.07324185967445374,
      "learning_rate": 5.530864197530864e-05,
      "loss": 0.7625,
      "step": 1126
    },
    {
      "epoch": 2.500277315585136,
      "grad_norm": 0.06149562448263168,
      "learning_rate": 5.506172839506172e-05,
      "loss": 0.6032,
      "step": 1127
    },
    {
      "epoch": 2.502495840266223,
      "grad_norm": 0.06575816124677658,
      "learning_rate": 5.481481481481481e-05,
      "loss": 0.7118,
      "step": 1128
    },
    {
      "epoch": 2.50471436494731,
      "grad_norm": 0.09046422690153122,
      "learning_rate": 5.456790123456789e-05,
      "loss": 0.8168,
      "step": 1129
    },
    {
      "epoch": 2.506932889628397,
      "grad_norm": 0.10097432136535645,
      "learning_rate": 5.4320987654320986e-05,
      "loss": 0.7642,
      "step": 1130
    },
    {
      "epoch": 2.509151414309484,
      "grad_norm": 0.08827638626098633,
      "learning_rate": 5.407407407407407e-05,
      "loss": 0.7275,
      "step": 1131
    },
    {
      "epoch": 2.5113699389905713,
      "grad_norm": 0.07644348591566086,
      "learning_rate": 5.3827160493827156e-05,
      "loss": 0.8351,
      "step": 1132
    },
    {
      "epoch": 2.5135884636716583,
      "grad_norm": 0.0825062021613121,
      "learning_rate": 5.358024691358024e-05,
      "loss": 0.6557,
      "step": 1133
    },
    {
      "epoch": 2.5158069883527454,
      "grad_norm": 0.0865861251950264,
      "learning_rate": 5.333333333333333e-05,
      "loss": 0.8823,
      "step": 1134
    },
    {
      "epoch": 2.5180255130338325,
      "grad_norm": 0.09801134467124939,
      "learning_rate": 5.3086419753086415e-05,
      "loss": 0.6393,
      "step": 1135
    },
    {
      "epoch": 2.5202440377149196,
      "grad_norm": 0.12184236943721771,
      "learning_rate": 5.2839506172839496e-05,
      "loss": 0.8695,
      "step": 1136
    },
    {
      "epoch": 2.5224625623960066,
      "grad_norm": 0.07654005289077759,
      "learning_rate": 5.259259259259259e-05,
      "loss": 0.7807,
      "step": 1137
    },
    {
      "epoch": 2.5246810870770937,
      "grad_norm": 0.09142210334539413,
      "learning_rate": 5.234567901234567e-05,
      "loss": 0.7467,
      "step": 1138
    },
    {
      "epoch": 2.526899611758181,
      "grad_norm": 0.11978959292173386,
      "learning_rate": 5.209876543209876e-05,
      "loss": 1.0497,
      "step": 1139
    },
    {
      "epoch": 2.529118136439268,
      "grad_norm": 0.0916508138179779,
      "learning_rate": 5.185185185185184e-05,
      "loss": 0.7715,
      "step": 1140
    },
    {
      "epoch": 2.531336661120355,
      "grad_norm": 0.09982048720121384,
      "learning_rate": 5.160493827160494e-05,
      "loss": 0.7529,
      "step": 1141
    },
    {
      "epoch": 2.533555185801442,
      "grad_norm": 0.0816536694765091,
      "learning_rate": 5.135802469135802e-05,
      "loss": 0.767,
      "step": 1142
    },
    {
      "epoch": 2.535773710482529,
      "grad_norm": 0.11165288090705872,
      "learning_rate": 5.111111111111111e-05,
      "loss": 0.892,
      "step": 1143
    },
    {
      "epoch": 2.537992235163616,
      "grad_norm": 0.11610313504934311,
      "learning_rate": 5.086419753086419e-05,
      "loss": 1.1991,
      "step": 1144
    },
    {
      "epoch": 2.5402107598447032,
      "grad_norm": 0.07252341508865356,
      "learning_rate": 5.0617283950617285e-05,
      "loss": 0.6513,
      "step": 1145
    },
    {
      "epoch": 2.5424292845257903,
      "grad_norm": 0.07541034370660782,
      "learning_rate": 5.0370370370370366e-05,
      "loss": 0.7617,
      "step": 1146
    },
    {
      "epoch": 2.5446478092068774,
      "grad_norm": 0.096015565097332,
      "learning_rate": 5.012345679012345e-05,
      "loss": 1.0903,
      "step": 1147
    },
    {
      "epoch": 2.5468663338879645,
      "grad_norm": 0.12408241629600525,
      "learning_rate": 4.9876543209876536e-05,
      "loss": 0.8169,
      "step": 1148
    },
    {
      "epoch": 2.5490848585690515,
      "grad_norm": 0.08961262553930283,
      "learning_rate": 4.9629629629629625e-05,
      "loss": 0.7883,
      "step": 1149
    },
    {
      "epoch": 2.5513033832501386,
      "grad_norm": 0.09675523638725281,
      "learning_rate": 4.938271604938271e-05,
      "loss": 0.7849,
      "step": 1150
    },
    {
      "epoch": 2.5535219079312257,
      "grad_norm": 0.08893217891454697,
      "learning_rate": 4.9135802469135795e-05,
      "loss": 0.7328,
      "step": 1151
    },
    {
      "epoch": 2.5557404326123128,
      "grad_norm": 0.09901981055736542,
      "learning_rate": 4.888888888888889e-05,
      "loss": 0.7258,
      "step": 1152
    },
    {
      "epoch": 2.5579589572934,
      "grad_norm": 0.07090885192155838,
      "learning_rate": 4.864197530864197e-05,
      "loss": 0.6692,
      "step": 1153
    },
    {
      "epoch": 2.560177481974487,
      "grad_norm": 0.11335237324237823,
      "learning_rate": 4.839506172839506e-05,
      "loss": 0.7076,
      "step": 1154
    },
    {
      "epoch": 2.562396006655574,
      "grad_norm": 0.08642097562551498,
      "learning_rate": 4.814814814814814e-05,
      "loss": 0.8377,
      "step": 1155
    },
    {
      "epoch": 2.564614531336661,
      "grad_norm": 0.0857468694448471,
      "learning_rate": 4.7901234567901237e-05,
      "loss": 0.7658,
      "step": 1156
    },
    {
      "epoch": 2.566833056017748,
      "grad_norm": 0.08223691582679749,
      "learning_rate": 4.765432098765432e-05,
      "loss": 0.5822,
      "step": 1157
    },
    {
      "epoch": 2.569051580698835,
      "grad_norm": 0.06728120148181915,
      "learning_rate": 4.74074074074074e-05,
      "loss": 0.7681,
      "step": 1158
    },
    {
      "epoch": 2.5712701053799223,
      "grad_norm": 0.08319597691297531,
      "learning_rate": 4.716049382716049e-05,
      "loss": 0.7727,
      "step": 1159
    },
    {
      "epoch": 2.5734886300610094,
      "grad_norm": 0.10695908218622208,
      "learning_rate": 4.6913580246913576e-05,
      "loss": 0.7678,
      "step": 1160
    },
    {
      "epoch": 2.5757071547420964,
      "grad_norm": 0.07292738556861877,
      "learning_rate": 4.6666666666666665e-05,
      "loss": 0.7539,
      "step": 1161
    },
    {
      "epoch": 2.5779256794231835,
      "grad_norm": 0.06973250210285187,
      "learning_rate": 4.6419753086419746e-05,
      "loss": 0.6934,
      "step": 1162
    },
    {
      "epoch": 2.5801442041042706,
      "grad_norm": 0.07317639887332916,
      "learning_rate": 4.6172839506172835e-05,
      "loss": 0.8969,
      "step": 1163
    },
    {
      "epoch": 2.5823627287853577,
      "grad_norm": 0.09355457872152328,
      "learning_rate": 4.592592592592592e-05,
      "loss": 0.7796,
      "step": 1164
    },
    {
      "epoch": 2.5845812534664447,
      "grad_norm": 0.06710159778594971,
      "learning_rate": 4.567901234567901e-05,
      "loss": 0.6187,
      "step": 1165
    },
    {
      "epoch": 2.586799778147532,
      "grad_norm": 0.07069924473762512,
      "learning_rate": 4.543209876543209e-05,
      "loss": 0.5608,
      "step": 1166
    },
    {
      "epoch": 2.589018302828619,
      "grad_norm": 0.09057248383760452,
      "learning_rate": 4.518518518518519e-05,
      "loss": 0.9258,
      "step": 1167
    },
    {
      "epoch": 2.591236827509706,
      "grad_norm": 0.1104285717010498,
      "learning_rate": 4.493827160493827e-05,
      "loss": 0.7712,
      "step": 1168
    },
    {
      "epoch": 2.593455352190793,
      "grad_norm": 0.07302739471197128,
      "learning_rate": 4.469135802469135e-05,
      "loss": 0.6391,
      "step": 1169
    },
    {
      "epoch": 2.59567387687188,
      "grad_norm": 0.07616756111383438,
      "learning_rate": 4.444444444444444e-05,
      "loss": 0.7367,
      "step": 1170
    },
    {
      "epoch": 2.597892401552967,
      "grad_norm": 0.08171606063842773,
      "learning_rate": 4.419753086419753e-05,
      "loss": 0.7505,
      "step": 1171
    },
    {
      "epoch": 2.6001109262340543,
      "grad_norm": 0.07602343708276749,
      "learning_rate": 4.3950617283950617e-05,
      "loss": 0.8015,
      "step": 1172
    },
    {
      "epoch": 2.6023294509151413,
      "grad_norm": 0.09432733058929443,
      "learning_rate": 4.37037037037037e-05,
      "loss": 0.7976,
      "step": 1173
    },
    {
      "epoch": 2.6045479755962284,
      "grad_norm": 0.07081367075443268,
      "learning_rate": 4.3456790123456786e-05,
      "loss": 0.6454,
      "step": 1174
    },
    {
      "epoch": 2.6067665002773155,
      "grad_norm": 0.10675236582756042,
      "learning_rate": 4.3209876543209875e-05,
      "loss": 0.7544,
      "step": 1175
    },
    {
      "epoch": 2.6089850249584026,
      "grad_norm": 0.07426029443740845,
      "learning_rate": 4.296296296296296e-05,
      "loss": 0.6691,
      "step": 1176
    },
    {
      "epoch": 2.6112035496394896,
      "grad_norm": 0.08253014087677002,
      "learning_rate": 4.2716049382716045e-05,
      "loss": 0.8313,
      "step": 1177
    },
    {
      "epoch": 2.6134220743205767,
      "grad_norm": 0.07643367350101471,
      "learning_rate": 4.2469135802469126e-05,
      "loss": 0.8126,
      "step": 1178
    },
    {
      "epoch": 2.615640599001664,
      "grad_norm": 0.08502630144357681,
      "learning_rate": 4.222222222222222e-05,
      "loss": 0.8444,
      "step": 1179
    },
    {
      "epoch": 2.617859123682751,
      "grad_norm": 0.06839875876903534,
      "learning_rate": 4.19753086419753e-05,
      "loss": 0.7964,
      "step": 1180
    },
    {
      "epoch": 2.620077648363838,
      "grad_norm": 0.07805034518241882,
      "learning_rate": 4.172839506172839e-05,
      "loss": 0.6951,
      "step": 1181
    },
    {
      "epoch": 2.622296173044925,
      "grad_norm": 0.07529859244823456,
      "learning_rate": 4.148148148148147e-05,
      "loss": 0.836,
      "step": 1182
    },
    {
      "epoch": 2.624514697726012,
      "grad_norm": 0.11477915197610855,
      "learning_rate": 4.123456790123457e-05,
      "loss": 0.539,
      "step": 1183
    },
    {
      "epoch": 2.626733222407099,
      "grad_norm": 0.09865379333496094,
      "learning_rate": 4.098765432098765e-05,
      "loss": 0.7956,
      "step": 1184
    },
    {
      "epoch": 2.6289517470881862,
      "grad_norm": 0.10417494177818298,
      "learning_rate": 4.074074074074074e-05,
      "loss": 0.8614,
      "step": 1185
    },
    {
      "epoch": 2.6311702717692733,
      "grad_norm": 0.07929234951734543,
      "learning_rate": 4.0493827160493827e-05,
      "loss": 0.6292,
      "step": 1186
    },
    {
      "epoch": 2.6333887964503604,
      "grad_norm": 0.09960918128490448,
      "learning_rate": 4.0246913580246915e-05,
      "loss": 0.9595,
      "step": 1187
    },
    {
      "epoch": 2.6356073211314475,
      "grad_norm": 0.08050335198640823,
      "learning_rate": 3.9999999999999996e-05,
      "loss": 0.6097,
      "step": 1188
    },
    {
      "epoch": 2.6378258458125345,
      "grad_norm": 0.08106682449579239,
      "learning_rate": 3.975308641975308e-05,
      "loss": 0.7013,
      "step": 1189
    },
    {
      "epoch": 2.6400443704936216,
      "grad_norm": 0.0610768161714077,
      "learning_rate": 3.950617283950617e-05,
      "loss": 0.636,
      "step": 1190
    },
    {
      "epoch": 2.6422628951747087,
      "grad_norm": 0.08693917095661163,
      "learning_rate": 3.9259259259259255e-05,
      "loss": 0.6689,
      "step": 1191
    },
    {
      "epoch": 2.6444814198557958,
      "grad_norm": 0.06706726551055908,
      "learning_rate": 3.901234567901234e-05,
      "loss": 0.7939,
      "step": 1192
    },
    {
      "epoch": 2.646699944536883,
      "grad_norm": 0.08642245084047318,
      "learning_rate": 3.8765432098765425e-05,
      "loss": 0.9162,
      "step": 1193
    },
    {
      "epoch": 2.64891846921797,
      "grad_norm": 0.11163870245218277,
      "learning_rate": 3.851851851851852e-05,
      "loss": 0.7702,
      "step": 1194
    },
    {
      "epoch": 2.651136993899057,
      "grad_norm": 0.09669745713472366,
      "learning_rate": 3.82716049382716e-05,
      "loss": 0.6845,
      "step": 1195
    },
    {
      "epoch": 2.653355518580144,
      "grad_norm": 0.12039913982152939,
      "learning_rate": 3.802469135802468e-05,
      "loss": 0.6818,
      "step": 1196
    },
    {
      "epoch": 2.655574043261231,
      "grad_norm": 0.09419562667608261,
      "learning_rate": 3.777777777777777e-05,
      "loss": 0.6344,
      "step": 1197
    },
    {
      "epoch": 2.657792567942318,
      "grad_norm": 0.09924124926328659,
      "learning_rate": 3.7530864197530867e-05,
      "loss": 0.6709,
      "step": 1198
    },
    {
      "epoch": 2.6600110926234053,
      "grad_norm": 0.08572228997945786,
      "learning_rate": 3.728395061728395e-05,
      "loss": 0.783,
      "step": 1199
    },
    {
      "epoch": 2.6622296173044924,
      "grad_norm": 0.1327355057001114,
      "learning_rate": 3.703703703703703e-05,
      "loss": 0.8958,
      "step": 1200
    },
    {
      "epoch": 2.6644481419855794,
      "grad_norm": 0.09971456974744797,
      "learning_rate": 3.679012345679012e-05,
      "loss": 0.6587,
      "step": 1201
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 0.0805245190858841,
      "learning_rate": 3.6543209876543207e-05,
      "loss": 0.7695,
      "step": 1202
    },
    {
      "epoch": 2.6688851913477536,
      "grad_norm": 0.07940240949392319,
      "learning_rate": 3.6296296296296295e-05,
      "loss": 0.913,
      "step": 1203
    },
    {
      "epoch": 2.6711037160288407,
      "grad_norm": 0.0797734186053276,
      "learning_rate": 3.604938271604938e-05,
      "loss": 1.0077,
      "step": 1204
    },
    {
      "epoch": 2.6733222407099277,
      "grad_norm": 0.0713106095790863,
      "learning_rate": 3.5802469135802465e-05,
      "loss": 0.6747,
      "step": 1205
    },
    {
      "epoch": 2.675540765391015,
      "grad_norm": 0.09578016400337219,
      "learning_rate": 3.555555555555555e-05,
      "loss": 0.7653,
      "step": 1206
    },
    {
      "epoch": 2.677759290072102,
      "grad_norm": 0.07910098880529404,
      "learning_rate": 3.5308641975308635e-05,
      "loss": 0.746,
      "step": 1207
    },
    {
      "epoch": 2.679977814753189,
      "grad_norm": 0.09315955638885498,
      "learning_rate": 3.506172839506172e-05,
      "loss": 0.8559,
      "step": 1208
    },
    {
      "epoch": 2.682196339434276,
      "grad_norm": 0.07543308287858963,
      "learning_rate": 3.481481481481481e-05,
      "loss": 0.7036,
      "step": 1209
    },
    {
      "epoch": 2.684414864115363,
      "grad_norm": 0.11614494770765305,
      "learning_rate": 3.45679012345679e-05,
      "loss": 0.6888,
      "step": 1210
    },
    {
      "epoch": 2.68663338879645,
      "grad_norm": 0.08119738847017288,
      "learning_rate": 3.432098765432098e-05,
      "loss": 0.6997,
      "step": 1211
    },
    {
      "epoch": 2.6888519134775377,
      "grad_norm": 0.08399448543787003,
      "learning_rate": 3.407407407407407e-05,
      "loss": 0.8787,
      "step": 1212
    },
    {
      "epoch": 2.691070438158625,
      "grad_norm": 0.0808977484703064,
      "learning_rate": 3.382716049382716e-05,
      "loss": 0.7238,
      "step": 1213
    },
    {
      "epoch": 2.693288962839712,
      "grad_norm": 0.08603453636169434,
      "learning_rate": 3.3580246913580247e-05,
      "loss": 0.7871,
      "step": 1214
    },
    {
      "epoch": 2.695507487520799,
      "grad_norm": 0.0741838812828064,
      "learning_rate": 3.333333333333333e-05,
      "loss": 0.7083,
      "step": 1215
    },
    {
      "epoch": 2.697726012201886,
      "grad_norm": 0.10202973335981369,
      "learning_rate": 3.3086419753086417e-05,
      "loss": 0.8896,
      "step": 1216
    },
    {
      "epoch": 2.699944536882973,
      "grad_norm": 0.07235730439424515,
      "learning_rate": 3.2839506172839505e-05,
      "loss": 0.8439,
      "step": 1217
    },
    {
      "epoch": 2.70216306156406,
      "grad_norm": 0.10732976347208023,
      "learning_rate": 3.2592592592592586e-05,
      "loss": 0.7875,
      "step": 1218
    },
    {
      "epoch": 2.7043815862451472,
      "grad_norm": 0.0936044454574585,
      "learning_rate": 3.2345679012345675e-05,
      "loss": 0.8022,
      "step": 1219
    },
    {
      "epoch": 2.7066001109262343,
      "grad_norm": 0.0995996966958046,
      "learning_rate": 3.209876543209876e-05,
      "loss": 0.8157,
      "step": 1220
    },
    {
      "epoch": 2.7088186356073214,
      "grad_norm": 0.0844205766916275,
      "learning_rate": 3.185185185185185e-05,
      "loss": 0.7842,
      "step": 1221
    },
    {
      "epoch": 2.7110371602884085,
      "grad_norm": 0.07094638794660568,
      "learning_rate": 3.160493827160493e-05,
      "loss": 1.0153,
      "step": 1222
    },
    {
      "epoch": 2.7132556849694955,
      "grad_norm": 0.0819460079073906,
      "learning_rate": 3.135802469135802e-05,
      "loss": 0.861,
      "step": 1223
    },
    {
      "epoch": 2.7154742096505826,
      "grad_norm": 0.08398749679327011,
      "learning_rate": 3.111111111111111e-05,
      "loss": 0.8695,
      "step": 1224
    },
    {
      "epoch": 2.7176927343316697,
      "grad_norm": 0.15525184571743011,
      "learning_rate": 3.08641975308642e-05,
      "loss": 0.8451,
      "step": 1225
    },
    {
      "epoch": 2.7199112590127568,
      "grad_norm": 0.07175765931606293,
      "learning_rate": 3.061728395061728e-05,
      "loss": 0.6729,
      "step": 1226
    },
    {
      "epoch": 2.722129783693844,
      "grad_norm": 0.08030920475721359,
      "learning_rate": 3.0370370370370368e-05,
      "loss": 0.7713,
      "step": 1227
    },
    {
      "epoch": 2.724348308374931,
      "grad_norm": 0.07488817721605301,
      "learning_rate": 3.0123456790123453e-05,
      "loss": 0.7931,
      "step": 1228
    },
    {
      "epoch": 2.726566833056018,
      "grad_norm": 0.08629157394170761,
      "learning_rate": 2.987654320987654e-05,
      "loss": 0.785,
      "step": 1229
    },
    {
      "epoch": 2.728785357737105,
      "grad_norm": 0.07143526524305344,
      "learning_rate": 2.9629629629629627e-05,
      "loss": 0.698,
      "step": 1230
    },
    {
      "epoch": 2.731003882418192,
      "grad_norm": 0.08486952632665634,
      "learning_rate": 2.9382716049382715e-05,
      "loss": 0.8236,
      "step": 1231
    },
    {
      "epoch": 2.733222407099279,
      "grad_norm": 0.1564619243144989,
      "learning_rate": 2.9135802469135796e-05,
      "loss": 0.5746,
      "step": 1232
    },
    {
      "epoch": 2.7354409317803663,
      "grad_norm": 0.0844772681593895,
      "learning_rate": 2.8888888888888885e-05,
      "loss": 0.682,
      "step": 1233
    },
    {
      "epoch": 2.7376594564614534,
      "grad_norm": 0.0690489113330841,
      "learning_rate": 2.8641975308641973e-05,
      "loss": 0.6823,
      "step": 1234
    },
    {
      "epoch": 2.7398779811425404,
      "grad_norm": 0.08275934308767319,
      "learning_rate": 2.8395061728395058e-05,
      "loss": 0.741,
      "step": 1235
    },
    {
      "epoch": 2.7420965058236275,
      "grad_norm": 0.10229187458753586,
      "learning_rate": 2.8148148148148147e-05,
      "loss": 0.717,
      "step": 1236
    },
    {
      "epoch": 2.7443150305047146,
      "grad_norm": 0.06817882508039474,
      "learning_rate": 2.790123456790123e-05,
      "loss": 0.7984,
      "step": 1237
    },
    {
      "epoch": 2.7465335551858017,
      "grad_norm": 0.09821635484695435,
      "learning_rate": 2.765432098765432e-05,
      "loss": 0.8234,
      "step": 1238
    },
    {
      "epoch": 2.7487520798668887,
      "grad_norm": 0.08816348016262054,
      "learning_rate": 2.7407407407407405e-05,
      "loss": 0.6842,
      "step": 1239
    },
    {
      "epoch": 2.750970604547976,
      "grad_norm": 0.08404859155416489,
      "learning_rate": 2.7160493827160493e-05,
      "loss": 0.8151,
      "step": 1240
    },
    {
      "epoch": 2.753189129229063,
      "grad_norm": 0.08402937650680542,
      "learning_rate": 2.6913580246913578e-05,
      "loss": 0.8128,
      "step": 1241
    },
    {
      "epoch": 2.75540765391015,
      "grad_norm": 0.06476356089115143,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 0.7128,
      "step": 1242
    },
    {
      "epoch": 2.757626178591237,
      "grad_norm": 0.0770738422870636,
      "learning_rate": 2.6419753086419748e-05,
      "loss": 0.7066,
      "step": 1243
    },
    {
      "epoch": 2.759844703272324,
      "grad_norm": 0.10589257627725601,
      "learning_rate": 2.6172839506172837e-05,
      "loss": 0.8717,
      "step": 1244
    },
    {
      "epoch": 2.762063227953411,
      "grad_norm": 0.08934485167264938,
      "learning_rate": 2.592592592592592e-05,
      "loss": 0.9219,
      "step": 1245
    },
    {
      "epoch": 2.7642817526344983,
      "grad_norm": 0.06935290992259979,
      "learning_rate": 2.567901234567901e-05,
      "loss": 0.7757,
      "step": 1246
    },
    {
      "epoch": 2.7665002773155853,
      "grad_norm": 0.08091390132904053,
      "learning_rate": 2.5432098765432095e-05,
      "loss": 0.7724,
      "step": 1247
    },
    {
      "epoch": 2.7687188019966724,
      "grad_norm": 0.0819263756275177,
      "learning_rate": 2.5185185185185183e-05,
      "loss": 0.6966,
      "step": 1248
    },
    {
      "epoch": 2.7709373266777595,
      "grad_norm": 0.0771152526140213,
      "learning_rate": 2.4938271604938268e-05,
      "loss": 0.8774,
      "step": 1249
    },
    {
      "epoch": 2.7731558513588466,
      "grad_norm": 0.06884437054395676,
      "learning_rate": 2.4691358024691357e-05,
      "loss": 0.7288,
      "step": 1250
    },
    {
      "epoch": 2.7753743760399336,
      "grad_norm": 0.08990071713924408,
      "learning_rate": 2.4444444444444445e-05,
      "loss": 0.9148,
      "step": 1251
    },
    {
      "epoch": 2.7775929007210207,
      "grad_norm": 0.08048328757286072,
      "learning_rate": 2.419753086419753e-05,
      "loss": 0.6825,
      "step": 1252
    },
    {
      "epoch": 2.779811425402108,
      "grad_norm": 0.11374068260192871,
      "learning_rate": 2.3950617283950618e-05,
      "loss": 1.0171,
      "step": 1253
    },
    {
      "epoch": 2.782029950083195,
      "grad_norm": 0.08833570778369904,
      "learning_rate": 2.37037037037037e-05,
      "loss": 0.7012,
      "step": 1254
    },
    {
      "epoch": 2.784248474764282,
      "grad_norm": 0.05778530240058899,
      "learning_rate": 2.3456790123456788e-05,
      "loss": 0.7072,
      "step": 1255
    },
    {
      "epoch": 2.786466999445369,
      "grad_norm": 0.08792801201343536,
      "learning_rate": 2.3209876543209873e-05,
      "loss": 0.8068,
      "step": 1256
    },
    {
      "epoch": 2.788685524126456,
      "grad_norm": 0.08987090736627579,
      "learning_rate": 2.296296296296296e-05,
      "loss": 0.931,
      "step": 1257
    },
    {
      "epoch": 2.790904048807543,
      "grad_norm": 0.07250697910785675,
      "learning_rate": 2.2716049382716047e-05,
      "loss": 0.6927,
      "step": 1258
    },
    {
      "epoch": 2.7931225734886302,
      "grad_norm": 0.08322115242481232,
      "learning_rate": 2.2469135802469135e-05,
      "loss": 0.9791,
      "step": 1259
    },
    {
      "epoch": 2.7953410981697173,
      "grad_norm": 0.07254292070865631,
      "learning_rate": 2.222222222222222e-05,
      "loss": 0.7819,
      "step": 1260
    },
    {
      "epoch": 2.7975596228508044,
      "grad_norm": 0.14736898243427277,
      "learning_rate": 2.1975308641975308e-05,
      "loss": 0.6221,
      "step": 1261
    },
    {
      "epoch": 2.7997781475318915,
      "grad_norm": 0.07786618918180466,
      "learning_rate": 2.1728395061728393e-05,
      "loss": 0.7904,
      "step": 1262
    },
    {
      "epoch": 2.8019966722129785,
      "grad_norm": 0.08506270498037338,
      "learning_rate": 2.148148148148148e-05,
      "loss": 0.6685,
      "step": 1263
    },
    {
      "epoch": 2.8042151968940656,
      "grad_norm": 0.0936833992600441,
      "learning_rate": 2.1234567901234563e-05,
      "loss": 0.8163,
      "step": 1264
    },
    {
      "epoch": 2.8064337215751527,
      "grad_norm": 0.09610911458730698,
      "learning_rate": 2.098765432098765e-05,
      "loss": 0.9261,
      "step": 1265
    },
    {
      "epoch": 2.8086522462562398,
      "grad_norm": 0.09674936532974243,
      "learning_rate": 2.0740740740740737e-05,
      "loss": 0.7081,
      "step": 1266
    },
    {
      "epoch": 2.810870770937327,
      "grad_norm": 0.07615646719932556,
      "learning_rate": 2.0493827160493825e-05,
      "loss": 0.884,
      "step": 1267
    },
    {
      "epoch": 2.813089295618414,
      "grad_norm": 0.10574216395616531,
      "learning_rate": 2.0246913580246913e-05,
      "loss": 0.6754,
      "step": 1268
    },
    {
      "epoch": 2.815307820299501,
      "grad_norm": 0.08234966546297073,
      "learning_rate": 1.9999999999999998e-05,
      "loss": 0.6387,
      "step": 1269
    },
    {
      "epoch": 2.817526344980588,
      "grad_norm": 0.06741941720247269,
      "learning_rate": 1.9753086419753087e-05,
      "loss": 0.6786,
      "step": 1270
    },
    {
      "epoch": 2.819744869661675,
      "grad_norm": 0.08968403190374374,
      "learning_rate": 1.950617283950617e-05,
      "loss": 0.74,
      "step": 1271
    },
    {
      "epoch": 2.821963394342762,
      "grad_norm": 0.07589440047740936,
      "learning_rate": 1.925925925925926e-05,
      "loss": 0.7276,
      "step": 1272
    },
    {
      "epoch": 2.8241819190238493,
      "grad_norm": 0.07184835523366928,
      "learning_rate": 1.901234567901234e-05,
      "loss": 0.7186,
      "step": 1273
    },
    {
      "epoch": 2.8264004437049364,
      "grad_norm": 0.07523677498102188,
      "learning_rate": 1.8765432098765433e-05,
      "loss": 0.7736,
      "step": 1274
    },
    {
      "epoch": 2.8286189683860234,
      "grad_norm": 0.08530844002962112,
      "learning_rate": 1.8518518518518515e-05,
      "loss": 0.9237,
      "step": 1275
    },
    {
      "epoch": 2.8308374930671105,
      "grad_norm": 0.08034783601760864,
      "learning_rate": 1.8271604938271603e-05,
      "loss": 0.6767,
      "step": 1276
    },
    {
      "epoch": 2.8330560177481976,
      "grad_norm": 0.09826391935348511,
      "learning_rate": 1.802469135802469e-05,
      "loss": 0.8808,
      "step": 1277
    },
    {
      "epoch": 2.8352745424292847,
      "grad_norm": 0.11243614554405212,
      "learning_rate": 1.7777777777777777e-05,
      "loss": 0.6711,
      "step": 1278
    },
    {
      "epoch": 2.8374930671103717,
      "grad_norm": 0.08941835910081863,
      "learning_rate": 1.753086419753086e-05,
      "loss": 0.7288,
      "step": 1279
    },
    {
      "epoch": 2.839711591791459,
      "grad_norm": 0.09980852156877518,
      "learning_rate": 1.728395061728395e-05,
      "loss": 0.7536,
      "step": 1280
    },
    {
      "epoch": 2.841930116472546,
      "grad_norm": 0.07636159658432007,
      "learning_rate": 1.7037037037037035e-05,
      "loss": 0.6024,
      "step": 1281
    },
    {
      "epoch": 2.844148641153633,
      "grad_norm": 0.09484035521745682,
      "learning_rate": 1.6790123456790123e-05,
      "loss": 0.9191,
      "step": 1282
    },
    {
      "epoch": 2.84636716583472,
      "grad_norm": 0.07220458984375,
      "learning_rate": 1.6543209876543208e-05,
      "loss": 0.6475,
      "step": 1283
    },
    {
      "epoch": 2.848585690515807,
      "grad_norm": 0.07945550233125687,
      "learning_rate": 1.6296296296296293e-05,
      "loss": 0.5889,
      "step": 1284
    },
    {
      "epoch": 2.850804215196894,
      "grad_norm": 0.07872219383716583,
      "learning_rate": 1.604938271604938e-05,
      "loss": 0.7174,
      "step": 1285
    },
    {
      "epoch": 2.8530227398779813,
      "grad_norm": 0.06488676369190216,
      "learning_rate": 1.5802469135802467e-05,
      "loss": 0.642,
      "step": 1286
    },
    {
      "epoch": 2.8552412645590683,
      "grad_norm": 0.09177283942699432,
      "learning_rate": 1.5555555555555555e-05,
      "loss": 0.5129,
      "step": 1287
    },
    {
      "epoch": 2.8574597892401554,
      "grad_norm": 0.07603426277637482,
      "learning_rate": 1.530864197530864e-05,
      "loss": 0.6882,
      "step": 1288
    },
    {
      "epoch": 2.8596783139212425,
      "grad_norm": 0.14070254564285278,
      "learning_rate": 1.5061728395061727e-05,
      "loss": 0.7828,
      "step": 1289
    },
    {
      "epoch": 2.8618968386023296,
      "grad_norm": 0.07658109068870544,
      "learning_rate": 1.4814814814814813e-05,
      "loss": 0.6644,
      "step": 1290
    },
    {
      "epoch": 2.8641153632834166,
      "grad_norm": 0.07812785357236862,
      "learning_rate": 1.4567901234567898e-05,
      "loss": 0.7311,
      "step": 1291
    },
    {
      "epoch": 2.8663338879645037,
      "grad_norm": 0.07735467702150345,
      "learning_rate": 1.4320987654320987e-05,
      "loss": 0.6902,
      "step": 1292
    },
    {
      "epoch": 2.868552412645591,
      "grad_norm": 0.09400403499603271,
      "learning_rate": 1.4074074074074073e-05,
      "loss": 0.9644,
      "step": 1293
    },
    {
      "epoch": 2.870770937326678,
      "grad_norm": 0.07522409409284592,
      "learning_rate": 1.382716049382716e-05,
      "loss": 0.8609,
      "step": 1294
    },
    {
      "epoch": 2.872989462007765,
      "grad_norm": 0.09235885739326477,
      "learning_rate": 1.3580246913580247e-05,
      "loss": 0.8582,
      "step": 1295
    },
    {
      "epoch": 2.875207986688852,
      "grad_norm": 0.06240527331829071,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.6742,
      "step": 1296
    },
    {
      "epoch": 2.877426511369939,
      "grad_norm": 0.07458188384771347,
      "learning_rate": 1.3086419753086418e-05,
      "loss": 0.6912,
      "step": 1297
    },
    {
      "epoch": 2.879645036051026,
      "grad_norm": 0.08338135480880737,
      "learning_rate": 1.2839506172839505e-05,
      "loss": 0.7935,
      "step": 1298
    },
    {
      "epoch": 2.8818635607321132,
      "grad_norm": 0.08543558418750763,
      "learning_rate": 1.2592592592592592e-05,
      "loss": 0.9224,
      "step": 1299
    },
    {
      "epoch": 2.8840820854132003,
      "grad_norm": 0.09064888209104538,
      "learning_rate": 1.2345679012345678e-05,
      "loss": 0.6312,
      "step": 1300
    },
    {
      "epoch": 2.8863006100942874,
      "grad_norm": 0.080633744597435,
      "learning_rate": 1.2098765432098765e-05,
      "loss": 0.8092,
      "step": 1301
    },
    {
      "epoch": 2.8885191347753745,
      "grad_norm": 0.0881345346570015,
      "learning_rate": 1.185185185185185e-05,
      "loss": 0.632,
      "step": 1302
    },
    {
      "epoch": 2.8907376594564616,
      "grad_norm": 0.1319723278284073,
      "learning_rate": 1.1604938271604937e-05,
      "loss": 0.811,
      "step": 1303
    },
    {
      "epoch": 2.8929561841375486,
      "grad_norm": 0.08666954934597015,
      "learning_rate": 1.1358024691358023e-05,
      "loss": 0.7783,
      "step": 1304
    },
    {
      "epoch": 2.8951747088186357,
      "grad_norm": 0.07887360453605652,
      "learning_rate": 1.111111111111111e-05,
      "loss": 0.7889,
      "step": 1305
    },
    {
      "epoch": 2.8973932334997228,
      "grad_norm": 0.08616329729557037,
      "learning_rate": 1.0864197530864197e-05,
      "loss": 0.7569,
      "step": 1306
    },
    {
      "epoch": 2.89961175818081,
      "grad_norm": 0.11278702318668365,
      "learning_rate": 1.0617283950617282e-05,
      "loss": 0.7869,
      "step": 1307
    },
    {
      "epoch": 2.901830282861897,
      "grad_norm": 0.09346611052751541,
      "learning_rate": 1.0370370370370368e-05,
      "loss": 0.7858,
      "step": 1308
    },
    {
      "epoch": 2.904048807542984,
      "grad_norm": 0.08503006398677826,
      "learning_rate": 1.0123456790123457e-05,
      "loss": 0.7206,
      "step": 1309
    },
    {
      "epoch": 2.906267332224071,
      "grad_norm": 0.08722062408924103,
      "learning_rate": 9.876543209876543e-06,
      "loss": 0.4914,
      "step": 1310
    },
    {
      "epoch": 2.908485856905158,
      "grad_norm": 0.06704442948102951,
      "learning_rate": 9.62962962962963e-06,
      "loss": 0.7548,
      "step": 1311
    },
    {
      "epoch": 2.9107043815862452,
      "grad_norm": 0.09000910073518753,
      "learning_rate": 9.382716049382717e-06,
      "loss": 0.7719,
      "step": 1312
    },
    {
      "epoch": 2.9129229062673323,
      "grad_norm": 0.09796265512704849,
      "learning_rate": 9.135802469135802e-06,
      "loss": 0.6644,
      "step": 1313
    },
    {
      "epoch": 2.9151414309484194,
      "grad_norm": 0.07261394709348679,
      "learning_rate": 8.888888888888888e-06,
      "loss": 0.7425,
      "step": 1314
    },
    {
      "epoch": 2.9173599556295065,
      "grad_norm": 0.07742348313331604,
      "learning_rate": 8.641975308641975e-06,
      "loss": 0.762,
      "step": 1315
    },
    {
      "epoch": 2.9195784803105935,
      "grad_norm": 0.0739155039191246,
      "learning_rate": 8.395061728395062e-06,
      "loss": 0.997,
      "step": 1316
    },
    {
      "epoch": 2.9217970049916806,
      "grad_norm": 0.08261583745479584,
      "learning_rate": 8.148148148148147e-06,
      "loss": 0.8333,
      "step": 1317
    },
    {
      "epoch": 2.9240155296727677,
      "grad_norm": 0.08104324340820312,
      "learning_rate": 7.901234567901233e-06,
      "loss": 0.7875,
      "step": 1318
    },
    {
      "epoch": 2.9262340543538548,
      "grad_norm": 0.07566863298416138,
      "learning_rate": 7.65432098765432e-06,
      "loss": 0.7145,
      "step": 1319
    },
    {
      "epoch": 2.928452579034942,
      "grad_norm": 0.07343873381614685,
      "learning_rate": 7.407407407407407e-06,
      "loss": 0.8513,
      "step": 1320
    },
    {
      "epoch": 2.930671103716029,
      "grad_norm": 0.0896615982055664,
      "learning_rate": 7.160493827160493e-06,
      "loss": 0.8914,
      "step": 1321
    },
    {
      "epoch": 2.932889628397116,
      "grad_norm": 0.09784901142120361,
      "learning_rate": 6.91358024691358e-06,
      "loss": 0.7416,
      "step": 1322
    },
    {
      "epoch": 2.935108153078203,
      "grad_norm": 0.10139190405607224,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.7592,
      "step": 1323
    },
    {
      "epoch": 2.93732667775929,
      "grad_norm": 0.07628582417964935,
      "learning_rate": 6.4197530864197525e-06,
      "loss": 0.7299,
      "step": 1324
    },
    {
      "epoch": 2.939545202440377,
      "grad_norm": 0.10902193188667297,
      "learning_rate": 6.172839506172839e-06,
      "loss": 0.8774,
      "step": 1325
    },
    {
      "epoch": 2.9417637271214643,
      "grad_norm": 0.09532027691602707,
      "learning_rate": 5.925925925925925e-06,
      "loss": 0.7569,
      "step": 1326
    },
    {
      "epoch": 2.9439822518025514,
      "grad_norm": 0.08896414935588837,
      "learning_rate": 5.679012345679012e-06,
      "loss": 0.8022,
      "step": 1327
    },
    {
      "epoch": 2.9462007764836384,
      "grad_norm": 0.0703958123922348,
      "learning_rate": 5.432098765432098e-06,
      "loss": 0.8747,
      "step": 1328
    },
    {
      "epoch": 2.9484193011647255,
      "grad_norm": 0.1257905811071396,
      "learning_rate": 5.185185185185184e-06,
      "loss": 0.65,
      "step": 1329
    },
    {
      "epoch": 2.9506378258458126,
      "grad_norm": 0.09206582605838776,
      "learning_rate": 4.938271604938272e-06,
      "loss": 0.8152,
      "step": 1330
    },
    {
      "epoch": 2.9528563505268997,
      "grad_norm": 0.093979611992836,
      "learning_rate": 4.691358024691358e-06,
      "loss": 0.8865,
      "step": 1331
    },
    {
      "epoch": 2.9550748752079867,
      "grad_norm": 0.06463106721639633,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.7402,
      "step": 1332
    },
    {
      "epoch": 2.957293399889074,
      "grad_norm": 0.10955110192298889,
      "learning_rate": 4.197530864197531e-06,
      "loss": 0.682,
      "step": 1333
    },
    {
      "epoch": 2.959511924570161,
      "grad_norm": 0.0951935350894928,
      "learning_rate": 3.950617283950617e-06,
      "loss": 0.7996,
      "step": 1334
    },
    {
      "epoch": 2.961730449251248,
      "grad_norm": 0.078733429312706,
      "learning_rate": 3.7037037037037033e-06,
      "loss": 0.8665,
      "step": 1335
    },
    {
      "epoch": 2.963948973932335,
      "grad_norm": 0.10701534152030945,
      "learning_rate": 3.45679012345679e-06,
      "loss": 0.9935,
      "step": 1336
    },
    {
      "epoch": 2.966167498613422,
      "grad_norm": 0.08705060929059982,
      "learning_rate": 3.2098765432098762e-06,
      "loss": 0.7156,
      "step": 1337
    },
    {
      "epoch": 2.968386023294509,
      "grad_norm": 0.06746901571750641,
      "learning_rate": 2.9629629629629625e-06,
      "loss": 0.6993,
      "step": 1338
    },
    {
      "epoch": 2.9706045479755963,
      "grad_norm": 0.07382374256849289,
      "learning_rate": 2.716049382716049e-06,
      "loss": 0.6077,
      "step": 1339
    },
    {
      "epoch": 2.9728230726566833,
      "grad_norm": 0.06843526661396027,
      "learning_rate": 2.469135802469136e-06,
      "loss": 0.8196,
      "step": 1340
    },
    {
      "epoch": 2.9750415973377704,
      "grad_norm": 0.10533407330513,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.8364,
      "step": 1341
    },
    {
      "epoch": 2.9772601220188575,
      "grad_norm": 0.09831984341144562,
      "learning_rate": 1.9753086419753083e-06,
      "loss": 0.9394,
      "step": 1342
    },
    {
      "epoch": 2.9794786466999446,
      "grad_norm": 0.08849385380744934,
      "learning_rate": 1.728395061728395e-06,
      "loss": 0.7184,
      "step": 1343
    },
    {
      "epoch": 2.9816971713810316,
      "grad_norm": 0.09752389788627625,
      "learning_rate": 1.4814814814814812e-06,
      "loss": 0.8466,
      "step": 1344
    },
    {
      "epoch": 2.9839156960621187,
      "grad_norm": 0.06069264933466911,
      "learning_rate": 1.234567901234568e-06,
      "loss": 0.9308,
      "step": 1345
    },
    {
      "epoch": 2.986134220743206,
      "grad_norm": 0.0823795273900032,
      "learning_rate": 9.876543209876542e-07,
      "loss": 0.9502,
      "step": 1346
    },
    {
      "epoch": 2.988352745424293,
      "grad_norm": 0.07283660024404526,
      "learning_rate": 7.407407407407406e-07,
      "loss": 0.7044,
      "step": 1347
    },
    {
      "epoch": 2.99057127010538,
      "grad_norm": 0.09030569344758987,
      "learning_rate": 4.938271604938271e-07,
      "loss": 0.7503,
      "step": 1348
    },
    {
      "epoch": 2.992789794786467,
      "grad_norm": 0.08363115787506104,
      "learning_rate": 2.4691358024691354e-07,
      "loss": 0.7874,
      "step": 1349
    },
    {
      "epoch": 2.995008319467554,
      "grad_norm": 0.08536290377378464,
      "learning_rate": 0.0,
      "loss": 0.6587,
      "step": 1350
    },
    {
      "epoch": 2.995008319467554,
      "step": 1350,
      "total_flos": 2.325536422179635e+16,
      "train_loss": 0.7775403798068011,
      "train_runtime": 1848.9044,
      "train_samples_per_second": 11.7,
      "train_steps_per_second": 0.73
    }
  ],
  "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": 2.325536422179635e+16,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}