{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 5427,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 1.0148043235115958,
      "learning_rate": 1.226993865030675e-06,
      "loss": 2.2813,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3322791870446757,
      "learning_rate": 2.45398773006135e-06,
      "loss": 2.2296,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4663907879429692,
      "learning_rate": 3.680981595092025e-06,
      "loss": 2.2275,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.242837455798734,
      "learning_rate": 4.9079754601227e-06,
      "loss": 2.2625,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4319082986555842,
      "learning_rate": 6.134969325153374e-06,
      "loss": 2.1741,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5573921380949731,
      "learning_rate": 7.36196319018405e-06,
      "loss": 2.3279,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5303704386022488,
      "learning_rate": 8.588957055214725e-06,
      "loss": 2.0636,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7911837177329323,
      "learning_rate": 9.8159509202454e-06,
      "loss": 1.823,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5279544947343284,
      "learning_rate": 1.1042944785276074e-05,
      "loss": 2.1524,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8410701794916629,
      "learning_rate": 1.2269938650306748e-05,
      "loss": 1.9526,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8357212058293427,
      "learning_rate": 1.3496932515337424e-05,
      "loss": 1.9432,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8097079176087884,
      "learning_rate": 1.47239263803681e-05,
      "loss": 1.9705,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7429817009304395,
      "learning_rate": 1.5950920245398772e-05,
      "loss": 1.8853,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.5799419714518482,
      "learning_rate": 1.717791411042945e-05,
      "loss": 2.0684,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8699604225931582,
      "learning_rate": 1.8404907975460123e-05,
      "loss": 1.9457,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6004273329905194,
      "learning_rate": 1.96319018404908e-05,
      "loss": 1.7456,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.5668891934711069,
      "learning_rate": 2.085889570552147e-05,
      "loss": 2.1531,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.5301126416332612,
      "learning_rate": 2.208588957055215e-05,
      "loss": 1.6625,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.5249315093530881,
      "learning_rate": 2.3312883435582822e-05,
      "loss": 1.9496,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.5040378781872191,
      "learning_rate": 2.4539877300613496e-05,
      "loss": 2.0025,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.42883596055885237,
      "learning_rate": 2.5766871165644174e-05,
      "loss": 1.853,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.45918930299106364,
      "learning_rate": 2.6993865030674848e-05,
      "loss": 1.8424,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.40656981661235736,
      "learning_rate": 2.822085889570552e-05,
      "loss": 1.6939,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.4535865246318442,
      "learning_rate": 2.94478527607362e-05,
      "loss": 1.6126,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.324457046877525,
      "learning_rate": 3.067484662576687e-05,
      "loss": 1.4436,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.33176512036458355,
      "learning_rate": 3.1901840490797544e-05,
      "loss": 1.563,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.3698739216899826,
      "learning_rate": 3.312883435582822e-05,
      "loss": 1.6393,
      "step": 27
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.32902100429939113,
      "learning_rate": 3.43558282208589e-05,
      "loss": 1.6071,
      "step": 28
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.3465279831682197,
      "learning_rate": 3.558282208588957e-05,
      "loss": 1.6111,
      "step": 29
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.3429633625921972,
      "learning_rate": 3.6809815950920246e-05,
      "loss": 1.5906,
      "step": 30
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.3877743027230956,
      "learning_rate": 3.8036809815950924e-05,
      "loss": 1.6473,
      "step": 31
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.31265007161032915,
      "learning_rate": 3.92638036809816e-05,
      "loss": 1.5313,
      "step": 32
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.28755605509003546,
      "learning_rate": 4.049079754601227e-05,
      "loss": 1.553,
      "step": 33
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2824886979410489,
      "learning_rate": 4.171779141104294e-05,
      "loss": 1.4578,
      "step": 34
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.3722721962227805,
      "learning_rate": 4.2944785276073626e-05,
      "loss": 1.5176,
      "step": 35
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.29865108668611,
      "learning_rate": 4.41717791411043e-05,
      "loss": 1.6054,
      "step": 36
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2863226395344051,
      "learning_rate": 4.539877300613497e-05,
      "loss": 1.4599,
      "step": 37
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.25113715497338135,
      "learning_rate": 4.6625766871165645e-05,
      "loss": 1.6262,
      "step": 38
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2573170971439386,
      "learning_rate": 4.785276073619632e-05,
      "loss": 1.4571,
      "step": 39
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.3457432626059451,
      "learning_rate": 4.907975460122699e-05,
      "loss": 1.4403,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.27409178168640386,
      "learning_rate": 5.030674846625767e-05,
      "loss": 1.4078,
      "step": 41
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.27264156953481894,
      "learning_rate": 5.153374233128835e-05,
      "loss": 1.6047,
      "step": 42
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.21864380338605302,
      "learning_rate": 5.276073619631902e-05,
      "loss": 1.5205,
      "step": 43
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.3423667074472969,
      "learning_rate": 5.3987730061349695e-05,
      "loss": 1.4992,
      "step": 44
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.24737740692871324,
      "learning_rate": 5.521472392638037e-05,
      "loss": 1.6514,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2856091166045163,
      "learning_rate": 5.644171779141104e-05,
      "loss": 1.3787,
      "step": 46
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2386044601004921,
      "learning_rate": 5.766871165644172e-05,
      "loss": 1.3535,
      "step": 47
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2658117162222297,
      "learning_rate": 5.88957055214724e-05,
      "loss": 1.5327,
      "step": 48
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2272874554587197,
      "learning_rate": 6.012269938650307e-05,
      "loss": 1.4831,
      "step": 49
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.23693782032486257,
      "learning_rate": 6.134969325153375e-05,
      "loss": 1.4069,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2226600005886651,
      "learning_rate": 6.257668711656443e-05,
      "loss": 1.5937,
      "step": 51
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2090585051929551,
      "learning_rate": 6.380368098159509e-05,
      "loss": 1.4519,
      "step": 52
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2176030197809254,
      "learning_rate": 6.503067484662577e-05,
      "loss": 1.2628,
      "step": 53
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.22521859401346678,
      "learning_rate": 6.625766871165644e-05,
      "loss": 1.3848,
      "step": 54
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.22921759493023355,
      "learning_rate": 6.748466257668711e-05,
      "loss": 1.2734,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.33168472179157416,
      "learning_rate": 6.87116564417178e-05,
      "loss": 1.5422,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.18998448040115593,
      "learning_rate": 6.993865030674847e-05,
      "loss": 1.362,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2341563952231503,
      "learning_rate": 7.116564417177914e-05,
      "loss": 1.3436,
      "step": 58
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2285533575524817,
      "learning_rate": 7.239263803680982e-05,
      "loss": 1.532,
      "step": 59
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.25615324619980484,
      "learning_rate": 7.361963190184049e-05,
      "loss": 1.4599,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.26890944212289586,
      "learning_rate": 7.484662576687118e-05,
      "loss": 1.488,
      "step": 61
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2267182060380428,
      "learning_rate": 7.607361963190185e-05,
      "loss": 1.2055,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.20383331236076888,
      "learning_rate": 7.730061349693252e-05,
      "loss": 1.2319,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.23307079355793528,
      "learning_rate": 7.85276073619632e-05,
      "loss": 1.3735,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.21234795717987556,
      "learning_rate": 7.975460122699386e-05,
      "loss": 1.4103,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.22724053834150898,
      "learning_rate": 8.098159509202454e-05,
      "loss": 1.3588,
      "step": 66
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2596545056025517,
      "learning_rate": 8.220858895705523e-05,
      "loss": 1.2844,
      "step": 67
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2546836388980528,
      "learning_rate": 8.343558282208588e-05,
      "loss": 1.4699,
      "step": 68
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.30155161619533194,
      "learning_rate": 8.466257668711657e-05,
      "loss": 1.2968,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.21788606247517553,
      "learning_rate": 8.588957055214725e-05,
      "loss": 1.4752,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.27920100956350496,
      "learning_rate": 8.711656441717791e-05,
      "loss": 1.2399,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.3044298776719394,
      "learning_rate": 8.83435582822086e-05,
      "loss": 1.351,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.22185603130297002,
      "learning_rate": 8.957055214723928e-05,
      "loss": 1.303,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2302584002061702,
      "learning_rate": 9.079754601226993e-05,
      "loss": 1.2424,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.3023764430960956,
      "learning_rate": 9.202453987730062e-05,
      "loss": 1.242,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2320448295630494,
      "learning_rate": 9.325153374233129e-05,
      "loss": 1.4734,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.3714614857198299,
      "learning_rate": 9.447852760736196e-05,
      "loss": 1.1169,
      "step": 77
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2507610602516276,
      "learning_rate": 9.570552147239264e-05,
      "loss": 1.3574,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2875828871653168,
      "learning_rate": 9.693251533742331e-05,
      "loss": 1.4003,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.21468030112987907,
      "learning_rate": 9.815950920245399e-05,
      "loss": 1.1039,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.22578244452000046,
      "learning_rate": 9.938650306748467e-05,
      "loss": 1.3834,
      "step": 81
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.274360583836876,
      "learning_rate": 0.00010061349693251534,
      "loss": 1.0805,
      "step": 82
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.25696998767238255,
      "learning_rate": 0.00010184049079754601,
      "loss": 1.1788,
      "step": 83
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.22912564310672595,
      "learning_rate": 0.0001030674846625767,
      "loss": 1.2721,
      "step": 84
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.26174247518558263,
      "learning_rate": 0.00010429447852760737,
      "loss": 1.2469,
      "step": 85
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2331395630569126,
      "learning_rate": 0.00010552147239263804,
      "loss": 1.535,
      "step": 86
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.3460315246767122,
      "learning_rate": 0.00010674846625766872,
      "loss": 1.3999,
      "step": 87
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2795035774780289,
      "learning_rate": 0.00010797546012269939,
      "loss": 1.413,
      "step": 88
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2336430684297083,
      "learning_rate": 0.00010920245398773006,
      "loss": 1.1871,
      "step": 89
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.24696168476550445,
      "learning_rate": 0.00011042944785276075,
      "loss": 1.1908,
      "step": 90
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.21697786381609932,
      "learning_rate": 0.00011165644171779142,
      "loss": 1.3284,
      "step": 91
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.3157474556162222,
      "learning_rate": 0.00011288343558282209,
      "loss": 1.4175,
      "step": 92
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.35872199797464704,
      "learning_rate": 0.00011411042944785277,
      "loss": 1.2369,
      "step": 93
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2881188442078251,
      "learning_rate": 0.00011533742331288344,
      "loss": 1.4235,
      "step": 94
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.28559565323967145,
      "learning_rate": 0.00011656441717791411,
      "loss": 1.4024,
      "step": 95
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2385867103071378,
      "learning_rate": 0.0001177914110429448,
      "loss": 1.428,
      "step": 96
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.3868069707473173,
      "learning_rate": 0.00011901840490797547,
      "loss": 1.2835,
      "step": 97
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.22925209706637337,
      "learning_rate": 0.00012024539877300614,
      "loss": 1.4382,
      "step": 98
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.23157652808379847,
      "learning_rate": 0.00012147239263803682,
      "loss": 1.2111,
      "step": 99
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.28687527815197095,
      "learning_rate": 0.0001226993865030675,
      "loss": 1.4126,
      "step": 100
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.19671573737228432,
      "learning_rate": 0.00012392638036809816,
      "loss": 1.267,
      "step": 101
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.21436163056412721,
      "learning_rate": 0.00012515337423312886,
      "loss": 1.2862,
      "step": 102
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.27510334063607444,
      "learning_rate": 0.00012638036809815953,
      "loss": 1.1924,
      "step": 103
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.22587719181444318,
      "learning_rate": 0.00012760736196319017,
      "loss": 1.2288,
      "step": 104
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2138074762047092,
      "learning_rate": 0.00012883435582822084,
      "loss": 1.3467,
      "step": 105
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2166706682085098,
      "learning_rate": 0.00013006134969325154,
      "loss": 1.4491,
      "step": 106
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.25668014626041247,
      "learning_rate": 0.0001312883435582822,
      "loss": 1.1421,
      "step": 107
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2079608292715333,
      "learning_rate": 0.00013251533742331288,
      "loss": 1.2608,
      "step": 108
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2381553646129259,
      "learning_rate": 0.00013374233128834358,
      "loss": 1.2144,
      "step": 109
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2979577715894489,
      "learning_rate": 0.00013496932515337422,
      "loss": 1.3204,
      "step": 110
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.24547377222237482,
      "learning_rate": 0.0001361963190184049,
      "loss": 1.1792,
      "step": 111
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.26332639905476435,
      "learning_rate": 0.0001374233128834356,
      "loss": 1.2755,
      "step": 112
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.20806141946752668,
      "learning_rate": 0.00013865030674846626,
      "loss": 1.3175,
      "step": 113
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2232732150692507,
      "learning_rate": 0.00013987730061349693,
      "loss": 1.3261,
      "step": 114
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.25952716179133417,
      "learning_rate": 0.00014110429447852763,
      "loss": 1.0965,
      "step": 115
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.21927499950517107,
      "learning_rate": 0.00014233128834355828,
      "loss": 1.4258,
      "step": 116
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.20544339138513396,
      "learning_rate": 0.00014355828220858895,
      "loss": 1.3065,
      "step": 117
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2618613248843029,
      "learning_rate": 0.00014478527607361964,
      "loss": 1.2181,
      "step": 118
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.26131710948776965,
      "learning_rate": 0.00014601226993865031,
      "loss": 1.2479,
      "step": 119
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2664818179791093,
      "learning_rate": 0.00014723926380368098,
      "loss": 1.3084,
      "step": 120
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.25952338840372746,
      "learning_rate": 0.00014846625766871168,
      "loss": 1.294,
      "step": 121
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.23781738571319358,
      "learning_rate": 0.00014969325153374235,
      "loss": 1.2,
      "step": 122
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.20385804728873808,
      "learning_rate": 0.000150920245398773,
      "loss": 1.2929,
      "step": 123
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2434201621156122,
      "learning_rate": 0.0001521472392638037,
      "loss": 1.2379,
      "step": 124
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.24146213570729191,
      "learning_rate": 0.00015337423312883436,
      "loss": 1.252,
      "step": 125
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.38566913808660613,
      "learning_rate": 0.00015460122699386504,
      "loss": 1.0108,
      "step": 126
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.22452255326813914,
      "learning_rate": 0.00015582822085889573,
      "loss": 1.2809,
      "step": 127
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.23686375714097596,
      "learning_rate": 0.0001570552147239264,
      "loss": 1.2761,
      "step": 128
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2227310732644674,
      "learning_rate": 0.00015828220858895705,
      "loss": 1.2896,
      "step": 129
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.28088151004217965,
      "learning_rate": 0.00015950920245398772,
      "loss": 1.1347,
      "step": 130
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.24045339726688883,
      "learning_rate": 0.00016073619631901842,
      "loss": 1.1489,
      "step": 131
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2789184477721326,
      "learning_rate": 0.00016196319018404909,
      "loss": 1.2516,
      "step": 132
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.23167365099767578,
      "learning_rate": 0.00016319018404907976,
      "loss": 1.3565,
      "step": 133
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.24757419364515354,
      "learning_rate": 0.00016441717791411045,
      "loss": 1.1836,
      "step": 134
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2639068480251012,
      "learning_rate": 0.0001656441717791411,
      "loss": 1.1324,
      "step": 135
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.2010298162702872,
      "learning_rate": 0.00016687116564417177,
      "loss": 1.2528,
      "step": 136
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.24417247722445862,
      "learning_rate": 0.00016809815950920247,
      "loss": 1.0779,
      "step": 137
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.2529008024851062,
      "learning_rate": 0.00016932515337423314,
      "loss": 1.229,
      "step": 138
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.23217769824750578,
      "learning_rate": 0.0001705521472392638,
      "loss": 1.2311,
      "step": 139
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.24125216301590402,
      "learning_rate": 0.0001717791411042945,
      "loss": 1.4631,
      "step": 140
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.21003023331580384,
      "learning_rate": 0.00017300613496932518,
      "loss": 1.2867,
      "step": 141
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.23805317940909018,
      "learning_rate": 0.00017423312883435582,
      "loss": 1.119,
      "step": 142
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.21662651308463768,
      "learning_rate": 0.00017546012269938652,
      "loss": 1.2801,
      "step": 143
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.2153328541526383,
      "learning_rate": 0.0001766871165644172,
      "loss": 1.2005,
      "step": 144
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.2183497054798141,
      "learning_rate": 0.00017791411042944786,
      "loss": 1.1818,
      "step": 145
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.2241462979371002,
      "learning_rate": 0.00017914110429447856,
      "loss": 1.2328,
      "step": 146
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.31530294557529104,
      "learning_rate": 0.00018036809815950923,
      "loss": 1.2544,
      "step": 147
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.21732899005584305,
      "learning_rate": 0.00018159509202453987,
      "loss": 1.1391,
      "step": 148
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.22083223423861686,
      "learning_rate": 0.00018282208588957057,
      "loss": 1.1858,
      "step": 149
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.23778702086917966,
      "learning_rate": 0.00018404907975460124,
      "loss": 1.166,
      "step": 150
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1961629896590094,
      "learning_rate": 0.0001852760736196319,
      "loss": 1.1412,
      "step": 151
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.20907974554995556,
      "learning_rate": 0.00018650306748466258,
      "loss": 1.1865,
      "step": 152
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.243413987524455,
      "learning_rate": 0.00018773006134969328,
      "loss": 1.2774,
      "step": 153
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.21762351051140028,
      "learning_rate": 0.00018895705521472392,
      "loss": 1.2473,
      "step": 154
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.23270057923828139,
      "learning_rate": 0.0001901840490797546,
      "loss": 1.1827,
      "step": 155
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.20612258943535702,
      "learning_rate": 0.0001914110429447853,
      "loss": 1.0926,
      "step": 156
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.20528464353689116,
      "learning_rate": 0.00019263803680981596,
      "loss": 1.1694,
      "step": 157
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.23199069731562152,
      "learning_rate": 0.00019386503067484663,
      "loss": 1.1143,
      "step": 158
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.22790265202397628,
      "learning_rate": 0.00019509202453987733,
      "loss": 1.2251,
      "step": 159
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.21650277786853234,
      "learning_rate": 0.00019631901840490797,
      "loss": 1.2938,
      "step": 160
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.227552916262772,
      "learning_rate": 0.00019754601226993864,
      "loss": 1.2709,
      "step": 161
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.19981708000619516,
      "learning_rate": 0.00019877300613496934,
      "loss": 1.1182,
      "step": 162
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.2828421918419577,
      "learning_rate": 0.0002,
      "loss": 1.2686,
      "step": 163
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.2277551295109352,
      "learning_rate": 0.0001999999821910639,
      "loss": 1.2327,
      "step": 164
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.19990022096960525,
      "learning_rate": 0.00019999992876426194,
      "loss": 1.2076,
      "step": 165
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1889823307100826,
      "learning_rate": 0.00019999983971961317,
      "loss": 1.3262,
      "step": 166
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.2344775425848839,
      "learning_rate": 0.00019999971505714928,
      "loss": 1.2065,
      "step": 167
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.2252071265791597,
      "learning_rate": 0.00019999955477691467,
      "loss": 1.1581,
      "step": 168
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1840468367970235,
      "learning_rate": 0.00019999935887896647,
      "loss": 1.0759,
      "step": 169
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.22840673003147222,
      "learning_rate": 0.00019999912736337437,
      "loss": 1.2241,
      "step": 170
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.2207728205891687,
      "learning_rate": 0.0001999988602302209,
      "loss": 1.1205,
      "step": 171
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.20681822138263956,
      "learning_rate": 0.00019999855747960122,
      "loss": 1.1912,
      "step": 172
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.19690253103295044,
      "learning_rate": 0.00019999821911162316,
      "loss": 1.1742,
      "step": 173
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.18959522828368186,
      "learning_rate": 0.00019999784512640716,
      "loss": 1.2582,
      "step": 174
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.22383075091085267,
      "learning_rate": 0.00019999743552408648,
      "loss": 1.2754,
      "step": 175
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.27019735104197234,
      "learning_rate": 0.00019999699030480704,
      "loss": 1.487,
      "step": 176
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.22837799159872169,
      "learning_rate": 0.00019999650946872738,
      "loss": 1.1002,
      "step": 177
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.19939753279528766,
      "learning_rate": 0.0001999959930160188,
      "loss": 1.0589,
      "step": 178
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.23629675582097978,
      "learning_rate": 0.0001999954409468652,
      "loss": 1.3453,
      "step": 179
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.2884919748691849,
      "learning_rate": 0.00019999485326146324,
      "loss": 1.213,
      "step": 180
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.22332561197903636,
      "learning_rate": 0.0001999942299600222,
      "loss": 1.2863,
      "step": 181
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.23189634815483703,
      "learning_rate": 0.0001999935710427642,
      "loss": 1.2445,
      "step": 182
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.22575290876324208,
      "learning_rate": 0.0001999928765099238,
      "loss": 1.2639,
      "step": 183
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.21707254930658554,
      "learning_rate": 0.00019999214636174845,
      "loss": 1.4306,
      "step": 184
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.27765169299379383,
      "learning_rate": 0.00019999138059849822,
      "loss": 1.0523,
      "step": 185
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.2270015874944441,
      "learning_rate": 0.00019999057922044582,
      "loss": 1.2412,
      "step": 186
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.20135599707159021,
      "learning_rate": 0.0001999897422278767,
      "loss": 1.3082,
      "step": 187
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.2838581649183917,
      "learning_rate": 0.000199988869621089,
      "loss": 1.2549,
      "step": 188
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.265078114831888,
      "learning_rate": 0.00019998796140039348,
      "loss": 1.3298,
      "step": 189
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.20995698048547656,
      "learning_rate": 0.00019998701756611368,
      "loss": 1.1571,
      "step": 190
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.22768366122479866,
      "learning_rate": 0.00019998603811858571,
      "loss": 1.2892,
      "step": 191
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.19466597443548833,
      "learning_rate": 0.00019998502305815848,
      "loss": 1.2027,
      "step": 192
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2338358845712038,
      "learning_rate": 0.00019998397238519354,
      "loss": 1.1584,
      "step": 193
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.19531261079089393,
      "learning_rate": 0.0001999828861000651,
      "loss": 1.1453,
      "step": 194
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.21591014766253525,
      "learning_rate": 0.00019998176420316002,
      "loss": 1.0409,
      "step": 195
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1998586616050853,
      "learning_rate": 0.000199980606694878,
      "loss": 1.1743,
      "step": 196
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.20640462264265017,
      "learning_rate": 0.0001999794135756312,
      "loss": 1.1228,
      "step": 197
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.19844807748919077,
      "learning_rate": 0.0001999781848458447,
      "loss": 1.1942,
      "step": 198
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2421192353609468,
      "learning_rate": 0.00019997692050595606,
      "loss": 1.3072,
      "step": 199
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.21078192244794589,
      "learning_rate": 0.00019997562055641568,
      "loss": 1.3501,
      "step": 200
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.29199418333708144,
      "learning_rate": 0.0001999742849976865,
      "loss": 1.1568,
      "step": 201
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.220239894105908,
      "learning_rate": 0.00019997291383024428,
      "loss": 1.2389,
      "step": 202
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.3291145260499344,
      "learning_rate": 0.0001999715070545774,
      "loss": 1.3219,
      "step": 203
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.19373247927563342,
      "learning_rate": 0.00019997006467118686,
      "loss": 1.2989,
      "step": 204
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.28770581174429816,
      "learning_rate": 0.00019996858668058646,
      "loss": 0.9912,
      "step": 205
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.20706652879650247,
      "learning_rate": 0.00019996707308330263,
      "loss": 1.2243,
      "step": 206
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2940713899010905,
      "learning_rate": 0.00019996552387987446,
      "loss": 1.1988,
      "step": 207
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.21717636038659388,
      "learning_rate": 0.00019996393907085376,
      "loss": 1.2056,
      "step": 208
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.21350628783689565,
      "learning_rate": 0.00019996231865680497,
      "loss": 1.4948,
      "step": 209
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1816249495494853,
      "learning_rate": 0.00019996066263830531,
      "loss": 1.028,
      "step": 210
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.264664050753348,
      "learning_rate": 0.00019995897101594454,
      "loss": 1.1789,
      "step": 211
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2135652933074264,
      "learning_rate": 0.00019995724379032526,
      "loss": 1.1478,
      "step": 212
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2158779697981514,
      "learning_rate": 0.0001999554809620626,
      "loss": 1.2079,
      "step": 213
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2715865661072702,
      "learning_rate": 0.0001999536825317845,
      "loss": 0.9652,
      "step": 214
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4152184843105218,
      "learning_rate": 0.00019995184850013148,
      "loss": 1.276,
      "step": 215
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.26942156193946915,
      "learning_rate": 0.0001999499788677568,
      "loss": 1.3055,
      "step": 216
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2407088068794907,
      "learning_rate": 0.00019994807363532638,
      "loss": 1.2214,
      "step": 217
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.37694661327706896,
      "learning_rate": 0.00019994613280351886,
      "loss": 1.453,
      "step": 218
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.18294293702476752,
      "learning_rate": 0.00019994415637302547,
      "loss": 1.1737,
      "step": 219
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2215806528385957,
      "learning_rate": 0.0001999421443445502,
      "loss": 1.4757,
      "step": 220
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2056363037627169,
      "learning_rate": 0.00019994009671880968,
      "loss": 1.1896,
      "step": 221
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.20722571923959887,
      "learning_rate": 0.00019993801349653322,
      "loss": 1.2009,
      "step": 222
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2041976850701629,
      "learning_rate": 0.00019993589467846284,
      "loss": 1.2025,
      "step": 223
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.31075836606750323,
      "learning_rate": 0.0001999337402653532,
      "loss": 1.1181,
      "step": 224
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.20086746515721549,
      "learning_rate": 0.00019993155025797168,
      "loss": 1.1311,
      "step": 225
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.24317296430083665,
      "learning_rate": 0.0001999293246570983,
      "loss": 1.1872,
      "step": 226
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.20260828736110617,
      "learning_rate": 0.00019992706346352577,
      "loss": 1.0261,
      "step": 227
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2292236301177998,
      "learning_rate": 0.00019992476667805953,
      "loss": 1.183,
      "step": 228
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.23784123466710946,
      "learning_rate": 0.00019992243430151755,
      "loss": 1.1806,
      "step": 229
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.22480150882942793,
      "learning_rate": 0.00019992006633473062,
      "loss": 1.2433,
      "step": 230
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.23244779465806836,
      "learning_rate": 0.00019991766277854218,
      "loss": 1.2814,
      "step": 231
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.3558788730199244,
      "learning_rate": 0.0001999152236338083,
      "loss": 1.2259,
      "step": 232
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.24178159117364034,
      "learning_rate": 0.00019991274890139774,
      "loss": 1.199,
      "step": 233
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.19458758688453406,
      "learning_rate": 0.000199910238582192,
      "loss": 1.2952,
      "step": 234
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.22533759687261,
      "learning_rate": 0.00019990769267708516,
      "loss": 1.1831,
      "step": 235
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.19249925054680586,
      "learning_rate": 0.00019990511118698402,
      "loss": 1.3273,
      "step": 236
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2029017847905593,
      "learning_rate": 0.00019990249411280803,
      "loss": 1.2233,
      "step": 237
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2599912675581288,
      "learning_rate": 0.0001998998414554894,
      "loss": 1.2163,
      "step": 238
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.17928260514283526,
      "learning_rate": 0.00019989715321597288,
      "loss": 1.1483,
      "step": 239
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2498595940173184,
      "learning_rate": 0.00019989442939521602,
      "loss": 1.219,
      "step": 240
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2120788382582323,
      "learning_rate": 0.00019989166999418898,
      "loss": 1.201,
      "step": 241
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2515629582877367,
      "learning_rate": 0.00019988887501387454,
      "loss": 1.2016,
      "step": 242
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.23204840367282664,
      "learning_rate": 0.00019988604445526827,
      "loss": 1.2172,
      "step": 243
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.23959500508487447,
      "learning_rate": 0.00019988317831937837,
      "loss": 1.2166,
      "step": 244
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.20971059237513717,
      "learning_rate": 0.00019988027660722567,
      "loss": 1.3466,
      "step": 245
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.18269186637916812,
      "learning_rate": 0.00019987733931984365,
      "loss": 1.4088,
      "step": 246
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.21016812394454315,
      "learning_rate": 0.0001998743664582786,
      "loss": 1.2337,
      "step": 247
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2589458253609085,
      "learning_rate": 0.00019987135802358933,
      "loss": 1.2116,
      "step": 248
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1982242181291204,
      "learning_rate": 0.00019986831401684735,
      "loss": 1.2339,
      "step": 249
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2129932785185532,
      "learning_rate": 0.00019986523443913697,
      "loss": 1.0661,
      "step": 250
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2091237157530521,
      "learning_rate": 0.000199862119291555,
      "loss": 1.3012,
      "step": 251
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.22976076066136278,
      "learning_rate": 0.000199858968575211,
      "loss": 1.1762,
      "step": 252
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2185129919291678,
      "learning_rate": 0.00019985578229122722,
      "loss": 1.1947,
      "step": 253
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.19744849948929674,
      "learning_rate": 0.00019985256044073848,
      "loss": 1.185,
      "step": 254
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2538369027204267,
      "learning_rate": 0.0001998493030248924,
      "loss": 1.1968,
      "step": 255
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2212523966492473,
      "learning_rate": 0.00019984601004484915,
      "loss": 1.1921,
      "step": 256
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.18069243009914504,
      "learning_rate": 0.00019984268150178167,
      "loss": 1.1014,
      "step": 257
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.20743696368584674,
      "learning_rate": 0.00019983931739687546,
      "loss": 1.3353,
      "step": 258
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2488500491014322,
      "learning_rate": 0.00019983591773132882,
      "loss": 1.3725,
      "step": 259
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.23713849760278274,
      "learning_rate": 0.00019983248250635257,
      "loss": 1.2318,
      "step": 260
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.22445059111168716,
      "learning_rate": 0.00019982901172317031,
      "loss": 1.1894,
      "step": 261
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.20090886328036578,
      "learning_rate": 0.0001998255053830182,
      "loss": 1.2416,
      "step": 262
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1654891072033405,
      "learning_rate": 0.00019982196348714517,
      "loss": 1.0336,
      "step": 263
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.15479392620634833,
      "learning_rate": 0.00019981838603681278,
      "loss": 1.1825,
      "step": 264
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.26177626219973893,
      "learning_rate": 0.00019981477303329518,
      "loss": 1.3395,
      "step": 265
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.20440082940729465,
      "learning_rate": 0.00019981112447787933,
      "loss": 1.1369,
      "step": 266
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.21050721987310655,
      "learning_rate": 0.00019980744037186469,
      "loss": 1.1582,
      "step": 267
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2510991417795394,
      "learning_rate": 0.00019980372071656352,
      "loss": 1.2989,
      "step": 268
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.21640271085572407,
      "learning_rate": 0.00019979996551330065,
      "loss": 1.3807,
      "step": 269
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1862579873798564,
      "learning_rate": 0.0001997961747634136,
      "loss": 1.3383,
      "step": 270
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2160616007141558,
      "learning_rate": 0.00019979234846825253,
      "loss": 1.0041,
      "step": 271
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.22092740111970002,
      "learning_rate": 0.00019978848662918037,
      "loss": 1.222,
      "step": 272
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.23225213505053285,
      "learning_rate": 0.00019978458924757255,
      "loss": 1.228,
      "step": 273
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1774592531042566,
      "learning_rate": 0.00019978065632481726,
      "loss": 1.2026,
      "step": 274
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.27066957859958113,
      "learning_rate": 0.00019977668786231534,
      "loss": 1.131,
      "step": 275
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1990294406986911,
      "learning_rate": 0.00019977268386148022,
      "loss": 1.1461,
      "step": 276
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.203510820058379,
      "learning_rate": 0.0001997686443237381,
      "loss": 1.296,
      "step": 277
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.20189284289785964,
      "learning_rate": 0.00019976456925052772,
      "loss": 1.0728,
      "step": 278
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.26139156970979127,
      "learning_rate": 0.00019976045864330058,
      "loss": 1.2386,
      "step": 279
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.20555908808267298,
      "learning_rate": 0.00019975631250352077,
      "loss": 1.1525,
      "step": 280
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.24107025493497006,
      "learning_rate": 0.00019975213083266508,
      "loss": 1.0784,
      "step": 281
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.24570654303263298,
      "learning_rate": 0.0001997479136322229,
      "loss": 1.187,
      "step": 282
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.21276640072282454,
      "learning_rate": 0.0001997436609036963,
      "loss": 1.2926,
      "step": 283
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.21311145348893315,
      "learning_rate": 0.00019973937264860004,
      "loss": 1.0807,
      "step": 284
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2071560672761906,
      "learning_rate": 0.0001997350488684615,
      "loss": 1.126,
      "step": 285
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.21465887721792545,
      "learning_rate": 0.00019973068956482073,
      "loss": 1.2472,
      "step": 286
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.227144794689696,
      "learning_rate": 0.00019972629473923042,
      "loss": 1.3718,
      "step": 287
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2314806256500085,
      "learning_rate": 0.00019972186439325587,
      "loss": 1.2821,
      "step": 288
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2368742977348321,
      "learning_rate": 0.00019971739852847514,
      "loss": 1.2266,
      "step": 289
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.19792500073173364,
      "learning_rate": 0.0001997128971464788,
      "loss": 1.0951,
      "step": 290
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.18430325854205346,
      "learning_rate": 0.0001997083602488702,
      "loss": 1.1853,
      "step": 291
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.183875701104134,
      "learning_rate": 0.00019970378783726528,
      "loss": 1.3154,
      "step": 292
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.17924551441738473,
      "learning_rate": 0.00019969917991329267,
      "loss": 1.3428,
      "step": 293
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.35835835931929977,
      "learning_rate": 0.0001996945364785935,
      "loss": 1.3346,
      "step": 294
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.19233779842092588,
      "learning_rate": 0.0001996898575348218,
      "loss": 1.1201,
      "step": 295
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2535892237779721,
      "learning_rate": 0.00019968514308364398,
      "loss": 1.0216,
      "step": 296
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.21400121879871672,
      "learning_rate": 0.00019968039312673935,
      "loss": 1.2325,
      "step": 297
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.22268837889745946,
      "learning_rate": 0.0001996756076657997,
      "loss": 1.2038,
      "step": 298
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.19082769127647187,
      "learning_rate": 0.00019967078670252945,
      "loss": 1.1408,
      "step": 299
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.18585946568078116,
      "learning_rate": 0.0001996659302386458,
      "loss": 1.2645,
      "step": 300
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2127378991607637,
      "learning_rate": 0.00019966103827587846,
      "loss": 1.0756,
      "step": 301
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2133716183260187,
      "learning_rate": 0.00019965611081596994,
      "loss": 1.2047,
      "step": 302
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.20053294198604407,
      "learning_rate": 0.00019965114786067516,
      "loss": 1.2974,
      "step": 303
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.203688011889805,
      "learning_rate": 0.00019964614941176195,
      "loss": 1.2864,
      "step": 304
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.841382037366666,
      "learning_rate": 0.00019964111547101054,
      "loss": 1.1717,
      "step": 305
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.18406914935092553,
      "learning_rate": 0.00019963604604021398,
      "loss": 1.2628,
      "step": 306
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.20925585383313813,
      "learning_rate": 0.00019963094112117785,
      "loss": 1.2507,
      "step": 307
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.20867409386973482,
      "learning_rate": 0.0001996258007157205,
      "loss": 1.1925,
      "step": 308
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.19503949624461953,
      "learning_rate": 0.00019962062482567274,
      "loss": 1.2259,
      "step": 309
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1726673230872083,
      "learning_rate": 0.00019961541345287815,
      "loss": 1.2931,
      "step": 310
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.22201612499721576,
      "learning_rate": 0.0001996101665991929,
      "loss": 1.0959,
      "step": 311
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.19232149618599756,
      "learning_rate": 0.0001996048842664858,
      "loss": 1.1631,
      "step": 312
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.18960259250639283,
      "learning_rate": 0.00019959956645663836,
      "loss": 1.2274,
      "step": 313
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.22338704014045083,
      "learning_rate": 0.0001995942131715446,
      "loss": 1.2054,
      "step": 314
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.20397007527852085,
      "learning_rate": 0.00019958882441311126,
      "loss": 1.3453,
      "step": 315
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.21403462765523532,
      "learning_rate": 0.00019958340018325776,
      "loss": 1.2018,
      "step": 316
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.17736292820159197,
      "learning_rate": 0.00019957794048391602,
      "loss": 1.1153,
      "step": 317
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.23451211765548047,
      "learning_rate": 0.0001995724453170307,
      "loss": 1.1227,
      "step": 318
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2544695877253687,
      "learning_rate": 0.00019956691468455906,
      "loss": 1.2896,
      "step": 319
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1858279774319209,
      "learning_rate": 0.000199561348588471,
      "loss": 1.3418,
      "step": 320
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.193117405740899,
      "learning_rate": 0.00019955574703074905,
      "loss": 1.1315,
      "step": 321
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.19920697075979782,
      "learning_rate": 0.00019955011001338832,
      "loss": 1.1439,
      "step": 322
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2148449290163328,
      "learning_rate": 0.00019954443753839667,
      "loss": 1.2623,
      "step": 323
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.22697303930574209,
      "learning_rate": 0.00019953872960779445,
      "loss": 1.1144,
      "step": 324
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.19232613005581636,
      "learning_rate": 0.00019953298622361475,
      "loss": 1.1659,
      "step": 325
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.278795142973777,
      "learning_rate": 0.0001995272073879032,
      "loss": 1.0193,
      "step": 326
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.18603520900056855,
      "learning_rate": 0.00019952139310271813,
      "loss": 1.2718,
      "step": 327
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.19763462358857814,
      "learning_rate": 0.00019951554337013048,
      "loss": 1.1152,
      "step": 328
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.19632717060698035,
      "learning_rate": 0.00019950965819222374,
      "loss": 1.333,
      "step": 329
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.20164157437587263,
      "learning_rate": 0.00019950373757109412,
      "loss": 1.1877,
      "step": 330
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.15933217552012272,
      "learning_rate": 0.00019949778150885042,
      "loss": 1.1772,
      "step": 331
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3028652481995604,
      "learning_rate": 0.00019949179000761406,
      "loss": 1.2446,
      "step": 332
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2025041873884704,
      "learning_rate": 0.0001994857630695191,
      "loss": 1.0836,
      "step": 333
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.21654966394350378,
      "learning_rate": 0.00019947970069671218,
      "loss": 1.2884,
      "step": 334
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.22426980411734665,
      "learning_rate": 0.0001994736028913526,
      "loss": 1.1684,
      "step": 335
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.20643334883499503,
      "learning_rate": 0.0001994674696556123,
      "loss": 1.1372,
      "step": 336
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2608271470048531,
      "learning_rate": 0.00019946130099167574,
      "loss": 1.0714,
      "step": 337
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2236558610128359,
      "learning_rate": 0.00019945509690174012,
      "loss": 1.3485,
      "step": 338
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2062419018227342,
      "learning_rate": 0.00019944885738801518,
      "loss": 1.2446,
      "step": 339
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.23757538993255942,
      "learning_rate": 0.00019944258245272332,
      "loss": 1.0551,
      "step": 340
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1829793689054642,
      "learning_rate": 0.00019943627209809952,
      "loss": 1.1752,
      "step": 341
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.31500356280905295,
      "learning_rate": 0.00019942992632639142,
      "loss": 1.3005,
      "step": 342
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.21788958660805977,
      "learning_rate": 0.00019942354513985923,
      "loss": 1.1933,
      "step": 343
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.22337809173725642,
      "learning_rate": 0.0001994171285407758,
      "loss": 1.0594,
      "step": 344
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.23190638270002797,
      "learning_rate": 0.00019941067653142657,
      "loss": 1.223,
      "step": 345
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.23379576397438637,
      "learning_rate": 0.00019940418911410963,
      "loss": 1.2221,
      "step": 346
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.18486110820158827,
      "learning_rate": 0.00019939766629113566,
      "loss": 1.1398,
      "step": 347
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.20528390240989733,
      "learning_rate": 0.0001993911080648279,
      "loss": 1.1452,
      "step": 348
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.17585444170582631,
      "learning_rate": 0.00019938451443752233,
      "loss": 1.317,
      "step": 349
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2768020650338486,
      "learning_rate": 0.0001993778854115674,
      "loss": 1.367,
      "step": 350
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2719738720045931,
      "learning_rate": 0.00019937122098932428,
      "loss": 1.2077,
      "step": 351
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2968436470723722,
      "learning_rate": 0.00019936452117316663,
      "loss": 1.1542,
      "step": 352
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.21979912571855506,
      "learning_rate": 0.00019935778596548085,
      "loss": 1.1854,
      "step": 353
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.22220287152323806,
      "learning_rate": 0.00019935101536866582,
      "loss": 1.1919,
      "step": 354
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1955114893784741,
      "learning_rate": 0.00019934420938513313,
      "loss": 1.1834,
      "step": 355
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2734130190269747,
      "learning_rate": 0.00019933736801730687,
      "loss": 1.2351,
      "step": 356
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.27438364374203394,
      "learning_rate": 0.00019933049126762383,
      "loss": 1.0797,
      "step": 357
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.194204025811214,
      "learning_rate": 0.00019932357913853336,
      "loss": 1.1402,
      "step": 358
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.20966424850596913,
      "learning_rate": 0.00019931663163249742,
      "loss": 1.2229,
      "step": 359
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2464839511807341,
      "learning_rate": 0.00019930964875199055,
      "loss": 1.2033,
      "step": 360
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.3090094180197901,
      "learning_rate": 0.0001993026304994999,
      "loss": 1.0624,
      "step": 361
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.27751784966574766,
      "learning_rate": 0.00019929557687752524,
      "loss": 1.0754,
      "step": 362
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.22505041950854596,
      "learning_rate": 0.00019928848788857887,
      "loss": 1.2439,
      "step": 363
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.31117324881526326,
      "learning_rate": 0.00019928136353518583,
      "loss": 1.1993,
      "step": 364
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.20739562593693112,
      "learning_rate": 0.00019927420381988357,
      "loss": 1.0782,
      "step": 365
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2104774964962633,
      "learning_rate": 0.00019926700874522228,
      "loss": 1.3869,
      "step": 366
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.24464536236575016,
      "learning_rate": 0.00019925977831376467,
      "loss": 1.0542,
      "step": 367
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.18461227215883924,
      "learning_rate": 0.0001992525125280861,
      "loss": 1.0644,
      "step": 368
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.316085166312366,
      "learning_rate": 0.00019924521139077443,
      "loss": 1.0042,
      "step": 369
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.24110240021343549,
      "learning_rate": 0.0001992378749044302,
      "loss": 1.1567,
      "step": 370
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1872717868859075,
      "learning_rate": 0.00019923050307166655,
      "loss": 1.1689,
      "step": 371
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.22187598577627998,
      "learning_rate": 0.00019922309589510913,
      "loss": 1.1952,
      "step": 372
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.21584920719031392,
      "learning_rate": 0.0001992156533773962,
      "loss": 1.2742,
      "step": 373
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.22489765295686015,
      "learning_rate": 0.00019920817552117863,
      "loss": 1.1285,
      "step": 374
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.22108342749333607,
      "learning_rate": 0.00019920066232911992,
      "loss": 1.085,
      "step": 375
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.23960006903089073,
      "learning_rate": 0.0001991931138038961,
      "loss": 1.2287,
      "step": 376
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.20671379793314584,
      "learning_rate": 0.00019918552994819573,
      "loss": 1.0847,
      "step": 377
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.29977612996525366,
      "learning_rate": 0.00019917791076472008,
      "loss": 1.0665,
      "step": 378
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.22432757302231116,
      "learning_rate": 0.00019917025625618292,
      "loss": 1.1086,
      "step": 379
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.29843617152680396,
      "learning_rate": 0.00019916256642531064,
      "loss": 1.2458,
      "step": 380
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.19060588051265395,
      "learning_rate": 0.00019915484127484216,
      "loss": 1.2365,
      "step": 381
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2236413839195913,
      "learning_rate": 0.00019914708080752907,
      "loss": 1.2091,
      "step": 382
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.20389534544812019,
      "learning_rate": 0.0001991392850261354,
      "loss": 1.0977,
      "step": 383
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2430859804106905,
      "learning_rate": 0.00019913145393343795,
      "loss": 1.2079,
      "step": 384
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.21254331633033063,
      "learning_rate": 0.00019912358753222588,
      "loss": 1.1051,
      "step": 385
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.20067370348089433,
      "learning_rate": 0.00019911568582530113,
      "loss": 1.2277,
      "step": 386
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2201880457574964,
      "learning_rate": 0.000199107748815478,
      "loss": 1.1356,
      "step": 387
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2392600461904044,
      "learning_rate": 0.00019909977650558363,
      "loss": 1.1625,
      "step": 388
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.24283980783949602,
      "learning_rate": 0.0001990917688984575,
      "loss": 1.059,
      "step": 389
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.21643406336499474,
      "learning_rate": 0.0001990837259969517,
      "loss": 1.1075,
      "step": 390
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2119118307266263,
      "learning_rate": 0.00019907564780393108,
      "loss": 1.2178,
      "step": 391
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.18908220119155875,
      "learning_rate": 0.0001990675343222728,
      "loss": 0.9684,
      "step": 392
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2334600351914233,
      "learning_rate": 0.00019905938555486675,
      "loss": 1.0799,
      "step": 393
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.23144074343864235,
      "learning_rate": 0.0001990512015046154,
      "loss": 1.1614,
      "step": 394
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.18938380176215613,
      "learning_rate": 0.00019904298217443366,
      "loss": 1.2323,
      "step": 395
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.21036969328931157,
      "learning_rate": 0.0001990347275672491,
      "loss": 1.092,
      "step": 396
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.24290131562188558,
      "learning_rate": 0.00019902643768600185,
      "loss": 1.28,
      "step": 397
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2033990067192421,
      "learning_rate": 0.00019901811253364456,
      "loss": 1.1653,
      "step": 398
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.20330813330069622,
      "learning_rate": 0.0001990097521131425,
      "loss": 1.2955,
      "step": 399
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2702370063665531,
      "learning_rate": 0.00019900135642747347,
      "loss": 1.1457,
      "step": 400
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2053770177080164,
      "learning_rate": 0.00019899292547962788,
      "loss": 1.0957,
      "step": 401
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2866281906354625,
      "learning_rate": 0.00019898445927260853,
      "loss": 1.1287,
      "step": 402
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2331802357070062,
      "learning_rate": 0.00019897595780943102,
      "loss": 1.1997,
      "step": 403
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2379338212763173,
      "learning_rate": 0.00019896742109312335,
      "loss": 1.3517,
      "step": 404
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.19451704666011146,
      "learning_rate": 0.0001989588491267261,
      "loss": 1.1539,
      "step": 405
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.24730434059252204,
      "learning_rate": 0.00019895024191329247,
      "loss": 0.9807,
      "step": 406
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.21765378905389487,
      "learning_rate": 0.00019894159945588812,
      "loss": 1.1905,
      "step": 407
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.21725115215258817,
      "learning_rate": 0.00019893292175759131,
      "loss": 1.1764,
      "step": 408
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.25822813557064234,
      "learning_rate": 0.00019892420882149287,
      "loss": 1.0822,
      "step": 409
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.18848660122267671,
      "learning_rate": 0.00019891546065069614,
      "loss": 1.2668,
      "step": 410
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.19118488446538415,
      "learning_rate": 0.00019890667724831707,
      "loss": 1.0674,
      "step": 411
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.19183361966584409,
      "learning_rate": 0.00019889785861748412,
      "loss": 1.0796,
      "step": 412
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.22838858328717446,
      "learning_rate": 0.00019888900476133824,
      "loss": 1.2182,
      "step": 413
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.167585780504192,
      "learning_rate": 0.00019888011568303305,
      "loss": 1.0468,
      "step": 414
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.18130678252249474,
      "learning_rate": 0.0001988711913857346,
      "loss": 1.1894,
      "step": 415
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2084019241985263,
      "learning_rate": 0.00019886223187262162,
      "loss": 1.2717,
      "step": 416
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1908622674303808,
      "learning_rate": 0.00019885323714688517,
      "loss": 1.2637,
      "step": 417
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.23357313862285684,
      "learning_rate": 0.0001988442072117291,
      "loss": 1.1136,
      "step": 418
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2615655320804707,
      "learning_rate": 0.00019883514207036956,
      "loss": 1.2705,
      "step": 419
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.20516501674168414,
      "learning_rate": 0.0001988260417260355,
      "loss": 1.2194,
      "step": 420
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.18646576174616036,
      "learning_rate": 0.00019881690618196816,
      "loss": 1.1806,
      "step": 421
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.19532144033089757,
      "learning_rate": 0.00019880773544142148,
      "loss": 1.2292,
      "step": 422
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1759000561087604,
      "learning_rate": 0.00019879852950766187,
      "loss": 1.354,
      "step": 423
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.25499003937555126,
      "learning_rate": 0.0001987892883839683,
      "loss": 1.2197,
      "step": 424
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.22445228599328343,
      "learning_rate": 0.0001987800120736322,
      "loss": 1.1374,
      "step": 425
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.19265344748973182,
      "learning_rate": 0.00019877070057995768,
      "loss": 1.0707,
      "step": 426
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.17841392090718583,
      "learning_rate": 0.00019876135390626122,
      "loss": 1.2806,
      "step": 427
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.18027844963894255,
      "learning_rate": 0.00019875197205587196,
      "loss": 1.1032,
      "step": 428
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2201877729403843,
      "learning_rate": 0.00019874255503213152,
      "loss": 1.0093,
      "step": 429
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.18992653175685417,
      "learning_rate": 0.00019873310283839398,
      "loss": 1.2248,
      "step": 430
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2416505173957846,
      "learning_rate": 0.0001987236154780261,
      "loss": 1.2583,
      "step": 431
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.171179586933959,
      "learning_rate": 0.000198714092954407,
      "loss": 1.1778,
      "step": 432
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2151684780152867,
      "learning_rate": 0.00019870453527092845,
      "loss": 1.2491,
      "step": 433
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1870516550586188,
      "learning_rate": 0.00019869494243099465,
      "loss": 1.1905,
      "step": 434
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.21222091452198352,
      "learning_rate": 0.0001986853144380224,
      "loss": 1.2836,
      "step": 435
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.18304088986875705,
      "learning_rate": 0.00019867565129544096,
      "loss": 1.0375,
      "step": 436
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1796132780513688,
      "learning_rate": 0.00019866595300669217,
      "loss": 1.2099,
      "step": 437
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1815234811370433,
      "learning_rate": 0.00019865621957523033,
      "loss": 1.0752,
      "step": 438
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.19958200016996508,
      "learning_rate": 0.00019864645100452227,
      "loss": 1.2177,
      "step": 439
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.19230726361324876,
      "learning_rate": 0.00019863664729804738,
      "loss": 1.0088,
      "step": 440
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.20476000189817084,
      "learning_rate": 0.00019862680845929748,
      "loss": 1.1933,
      "step": 441
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.22778445521876994,
      "learning_rate": 0.00019861693449177702,
      "loss": 1.3138,
      "step": 442
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.18476329965171195,
      "learning_rate": 0.00019860702539900287,
      "loss": 1.026,
      "step": 443
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.22340456682335597,
      "learning_rate": 0.00019859708118450441,
      "loss": 1.1639,
      "step": 444
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1782276097949383,
      "learning_rate": 0.0001985871018518236,
      "loss": 1.3299,
      "step": 445
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.25153065203270947,
      "learning_rate": 0.00019857708740451483,
      "loss": 1.438,
      "step": 446
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2180265767229636,
      "learning_rate": 0.00019856703784614505,
      "loss": 1.1176,
      "step": 447
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.19756650646075252,
      "learning_rate": 0.00019855695318029372,
      "loss": 1.2234,
      "step": 448
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.22911668657785256,
      "learning_rate": 0.00019854683341055275,
      "loss": 1.1018,
      "step": 449
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.20032214301708415,
      "learning_rate": 0.00019853667854052663,
      "loss": 1.2047,
      "step": 450
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.20930398102049413,
      "learning_rate": 0.00019852648857383222,
      "loss": 1.1613,
      "step": 451
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.20342263849026296,
      "learning_rate": 0.00019851626351409906,
      "loss": 1.0138,
      "step": 452
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.23083113069580422,
      "learning_rate": 0.00019850600336496904,
      "loss": 1.2336,
      "step": 453
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.18947841067634055,
      "learning_rate": 0.0001984957081300967,
      "loss": 1.1252,
      "step": 454
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3242653567355775,
      "learning_rate": 0.00019848537781314883,
      "loss": 1.2019,
      "step": 455
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.18566382963786518,
      "learning_rate": 0.00019847501241780502,
      "loss": 1.1132,
      "step": 456
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.20651405048482732,
      "learning_rate": 0.00019846461194775712,
      "loss": 1.1669,
      "step": 457
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.21092505828627395,
      "learning_rate": 0.00019845417640670956,
      "loss": 1.1024,
      "step": 458
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1988118995854171,
      "learning_rate": 0.00019844370579837927,
      "loss": 0.9855,
      "step": 459
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.22557749448702027,
      "learning_rate": 0.0001984332001264957,
      "loss": 1.3387,
      "step": 460
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2798193662296762,
      "learning_rate": 0.00019842265939480064,
      "loss": 1.2776,
      "step": 461
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.219141492109134,
      "learning_rate": 0.00019841208360704858,
      "loss": 1.2126,
      "step": 462
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.22201091899869388,
      "learning_rate": 0.00019840147276700636,
      "loss": 1.2013,
      "step": 463
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.20311895327568805,
      "learning_rate": 0.00019839082687845335,
      "loss": 1.303,
      "step": 464
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.319413130844874,
      "learning_rate": 0.00019838014594518132,
      "loss": 1.0314,
      "step": 465
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.22611870158387662,
      "learning_rate": 0.00019836942997099465,
      "loss": 1.131,
      "step": 466
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2349241704407767,
      "learning_rate": 0.00019835867895971014,
      "loss": 1.2342,
      "step": 467
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1998276296158922,
      "learning_rate": 0.00019834789291515705,
      "loss": 1.1855,
      "step": 468
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.19698950114987138,
      "learning_rate": 0.00019833707184117717,
      "loss": 1.3863,
      "step": 469
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.19987590348431458,
      "learning_rate": 0.00019832621574162468,
      "loss": 1.101,
      "step": 470
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1795677508579765,
      "learning_rate": 0.00019831532462036636,
      "loss": 1.1542,
      "step": 471
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2203121504983519,
      "learning_rate": 0.00019830439848128134,
      "loss": 1.061,
      "step": 472
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.21110020996893614,
      "learning_rate": 0.0001982934373282613,
      "loss": 1.1383,
      "step": 473
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.20417618199935336,
      "learning_rate": 0.0001982824411652104,
      "loss": 1.2256,
      "step": 474
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1815925161615057,
      "learning_rate": 0.0001982714099960452,
      "loss": 1.1537,
      "step": 475
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.23397895201016838,
      "learning_rate": 0.00019826034382469478,
      "loss": 1.2344,
      "step": 476
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.20127229258679785,
      "learning_rate": 0.00019824924265510066,
      "loss": 1.1075,
      "step": 477
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.20747044443017496,
      "learning_rate": 0.00019823810649121688,
      "loss": 1.2464,
      "step": 478
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.20922803609747143,
      "learning_rate": 0.00019822693533700988,
      "loss": 1.2391,
      "step": 479
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.19577935713894568,
      "learning_rate": 0.0001982157291964586,
      "loss": 1.1326,
      "step": 480
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1963537100116952,
      "learning_rate": 0.00019820448807355437,
      "loss": 1.0876,
      "step": 481
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.23110432537681827,
      "learning_rate": 0.00019819321197230109,
      "loss": 1.1663,
      "step": 482
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.16127345468602303,
      "learning_rate": 0.00019818190089671508,
      "loss": 0.9878,
      "step": 483
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.20232546987320463,
      "learning_rate": 0.00019817055485082507,
      "loss": 1.2621,
      "step": 484
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2422699578874898,
      "learning_rate": 0.0001981591738386723,
      "loss": 1.0697,
      "step": 485
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.25497738577368445,
      "learning_rate": 0.00019814775786431045,
      "loss": 1.0537,
      "step": 486
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.24958325633402495,
      "learning_rate": 0.00019813630693180562,
      "loss": 1.1341,
      "step": 487
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1956727863376643,
      "learning_rate": 0.00019812482104523645,
      "loss": 1.1604,
      "step": 488
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1900400569702414,
      "learning_rate": 0.00019811330020869385,
      "loss": 1.1529,
      "step": 489
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.24297172100804898,
      "learning_rate": 0.00019810174442628142,
      "loss": 1.0738,
      "step": 490
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.17147825947919465,
      "learning_rate": 0.00019809015370211502,
      "loss": 1.1965,
      "step": 491
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.21016279002794075,
      "learning_rate": 0.00019807852804032305,
      "loss": 1.292,
      "step": 492
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2941945362020012,
      "learning_rate": 0.0001980668674450463,
      "loss": 1.2064,
      "step": 493
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2182571038301763,
      "learning_rate": 0.00019805517192043802,
      "loss": 1.134,
      "step": 494
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.17906710621934946,
      "learning_rate": 0.00019804344147066397,
      "loss": 1.1433,
      "step": 495
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1879387609422267,
      "learning_rate": 0.00019803167609990218,
      "loss": 1.2137,
      "step": 496
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2103033560293586,
      "learning_rate": 0.0001980198758123433,
      "loss": 1.2649,
      "step": 497
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.21364435788578495,
      "learning_rate": 0.00019800804061219032,
      "loss": 1.2699,
      "step": 498
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.18472334414653627,
      "learning_rate": 0.0001979961705036587,
      "loss": 1.1553,
      "step": 499
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.19230442487952146,
      "learning_rate": 0.0001979842654909763,
      "loss": 1.2503,
      "step": 500
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.18534880694959333,
      "learning_rate": 0.00019797232557838343,
      "loss": 1.1547,
      "step": 501
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2028946579505726,
      "learning_rate": 0.00019796035077013287,
      "loss": 1.3535,
      "step": 502
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.18488788573804846,
      "learning_rate": 0.00019794834107048973,
      "loss": 1.1292,
      "step": 503
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2082111004833595,
      "learning_rate": 0.0001979362964837317,
      "loss": 1.1109,
      "step": 504
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.21491892389920744,
      "learning_rate": 0.00019792421701414871,
      "loss": 1.2037,
      "step": 505
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2205354401800191,
      "learning_rate": 0.00019791210266604327,
      "loss": 1.1873,
      "step": 506
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.22179315347577744,
      "learning_rate": 0.00019789995344373024,
      "loss": 1.2086,
      "step": 507
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.18930355406630314,
      "learning_rate": 0.0001978877693515369,
      "loss": 1.188,
      "step": 508
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.19224181270595123,
      "learning_rate": 0.00019787555039380296,
      "loss": 1.3488,
      "step": 509
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.22223855240049203,
      "learning_rate": 0.0001978632965748806,
      "loss": 1.0814,
      "step": 510
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.23480982764791694,
      "learning_rate": 0.00019785100789913428,
      "loss": 1.1082,
      "step": 511
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.20365134658783918,
      "learning_rate": 0.0001978386843709411,
      "loss": 1.1671,
      "step": 512
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.18225770722017595,
      "learning_rate": 0.0001978263259946903,
      "loss": 1.1337,
      "step": 513
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.18199944680225413,
      "learning_rate": 0.00019781393277478373,
      "loss": 1.2614,
      "step": 514
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.19995283438889022,
      "learning_rate": 0.00019780150471563558,
      "loss": 1.1866,
      "step": 515
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.2931874462747061,
      "learning_rate": 0.00019778904182167254,
      "loss": 1.1045,
      "step": 516
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1960842794922477,
      "learning_rate": 0.0001977765440973335,
      "loss": 1.1089,
      "step": 517
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.21100581263762386,
      "learning_rate": 0.00019776401154706993,
      "loss": 1.0746,
      "step": 518
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.2002837507261299,
      "learning_rate": 0.00019775144417534573,
      "loss": 1.2559,
      "step": 519
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.20411060904049189,
      "learning_rate": 0.00019773884198663702,
      "loss": 1.0598,
      "step": 520
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.17410980171516469,
      "learning_rate": 0.00019772620498543247,
      "loss": 1.1492,
      "step": 521
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.15774469408992367,
      "learning_rate": 0.00019771353317623317,
      "loss": 1.0607,
      "step": 522
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.21482265667620218,
      "learning_rate": 0.0001977008265635525,
      "loss": 1.3472,
      "step": 523
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.21840407330035777,
      "learning_rate": 0.0001976880851519163,
      "loss": 1.1075,
      "step": 524
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.18065012164400016,
      "learning_rate": 0.00019767530894586273,
      "loss": 1.1134,
      "step": 525
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.198953847400624,
      "learning_rate": 0.00019766249794994246,
      "loss": 1.0756,
      "step": 526
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1991338269350163,
      "learning_rate": 0.00019764965216871846,
      "loss": 1.1753,
      "step": 527
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.2000476340362523,
      "learning_rate": 0.0001976367716067662,
      "loss": 1.2633,
      "step": 528
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1965128417896058,
      "learning_rate": 0.00019762385626867336,
      "loss": 1.0225,
      "step": 529
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1766429672055576,
      "learning_rate": 0.00019761090615904022,
      "loss": 1.0976,
      "step": 530
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.23576777467034396,
      "learning_rate": 0.00019759792128247922,
      "loss": 1.1303,
      "step": 531
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.22655687037702682,
      "learning_rate": 0.00019758490164361537,
      "loss": 1.1343,
      "step": 532
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.20372176958834837,
      "learning_rate": 0.00019757184724708595,
      "loss": 1.1119,
      "step": 533
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.23542740888329428,
      "learning_rate": 0.00019755875809754068,
      "loss": 1.2438,
      "step": 534
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.20575751899853492,
      "learning_rate": 0.00019754563419964165,
      "loss": 1.1717,
      "step": 535
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.17734581760621068,
      "learning_rate": 0.0001975324755580633,
      "loss": 0.9808,
      "step": 536
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.17007186356094306,
      "learning_rate": 0.0001975192821774924,
      "loss": 0.9031,
      "step": 537
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.23415987746724165,
      "learning_rate": 0.00019750605406262825,
      "loss": 1.1747,
      "step": 538
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1817074313781565,
      "learning_rate": 0.00019749279121818235,
      "loss": 1.2423,
      "step": 539
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.2382767499320742,
      "learning_rate": 0.00019747949364887872,
      "loss": 1.1767,
      "step": 540
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.26545183195698957,
      "learning_rate": 0.00019746616135945356,
      "loss": 1.076,
      "step": 541
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.19679679560777397,
      "learning_rate": 0.00019745279435465565,
      "loss": 1.1387,
      "step": 542
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.33157018005891276,
      "learning_rate": 0.00019743939263924598,
      "loss": 1.1328,
      "step": 543
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.19651623495044224,
      "learning_rate": 0.00019742595621799795,
      "loss": 1.1988,
      "step": 544
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.23169624830134442,
      "learning_rate": 0.00019741248509569736,
      "loss": 1.113,
      "step": 545
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.20924133129744726,
      "learning_rate": 0.00019739897927714234,
      "loss": 1.0479,
      "step": 546
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.21171921182731004,
      "learning_rate": 0.00019738543876714334,
      "loss": 1.0555,
      "step": 547
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.20037957725050964,
      "learning_rate": 0.00019737186357052323,
      "loss": 1.1871,
      "step": 548
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.17590720595752335,
      "learning_rate": 0.0001973582536921172,
      "loss": 1.2277,
      "step": 549
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.18712245227445545,
      "learning_rate": 0.00019734460913677277,
      "loss": 1.1086,
      "step": 550
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.2056221135103932,
      "learning_rate": 0.00019733092990934984,
      "loss": 1.2261,
      "step": 551
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.22078249048471216,
      "learning_rate": 0.00019731721601472075,
      "loss": 1.2247,
      "step": 552
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.21189286332100862,
      "learning_rate": 0.00019730346745776997,
      "loss": 1.0765,
      "step": 553
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.18345773320313882,
      "learning_rate": 0.00019728968424339454,
      "loss": 1.2813,
      "step": 554
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.2688202539446119,
      "learning_rate": 0.00019727586637650373,
      "loss": 1.0793,
      "step": 555
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.210866999830138,
      "learning_rate": 0.0001972620138620191,
      "loss": 1.0304,
      "step": 556
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1865675362553537,
      "learning_rate": 0.00019724812670487475,
      "loss": 1.1259,
      "step": 557
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.21219766364086948,
      "learning_rate": 0.00019723420491001684,
      "loss": 1.175,
      "step": 558
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.25072784652021046,
      "learning_rate": 0.00019722024848240413,
      "loss": 1.2141,
      "step": 559
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.20477442198794102,
      "learning_rate": 0.00019720625742700756,
      "loss": 1.3129,
      "step": 560
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.2597289213359491,
      "learning_rate": 0.00019719223174881044,
      "loss": 1.1747,
      "step": 561
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.19342255440290604,
      "learning_rate": 0.00019717817145280844,
      "loss": 1.0059,
      "step": 562
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.18943376505892176,
      "learning_rate": 0.00019716407654400952,
      "loss": 1.1002,
      "step": 563
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.17261502217583144,
      "learning_rate": 0.00019714994702743396,
      "loss": 1.2101,
      "step": 564
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.18846315747976963,
      "learning_rate": 0.00019713578290811448,
      "loss": 1.2292,
      "step": 565
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.21201855929909813,
      "learning_rate": 0.00019712158419109597,
      "loss": 1.2598,
      "step": 566
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.21945538842546583,
      "learning_rate": 0.0001971073508814357,
      "loss": 0.9913,
      "step": 567
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.19781339667670664,
      "learning_rate": 0.0001970930829842033,
      "loss": 1.2012,
      "step": 568
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.21589054383877998,
      "learning_rate": 0.0001970787805044807,
      "loss": 1.0999,
      "step": 569
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.18516970778747321,
      "learning_rate": 0.00019706444344736213,
      "loss": 1.0931,
      "step": 570
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.238062864878335,
      "learning_rate": 0.00019705007181795416,
      "loss": 1.3207,
      "step": 571
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.22243829803840592,
      "learning_rate": 0.00019703566562137562,
      "loss": 1.0618,
      "step": 572
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.22437191505302434,
      "learning_rate": 0.0001970212248627577,
      "loss": 1.0917,
      "step": 573
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1932244190372955,
      "learning_rate": 0.00019700674954724388,
      "loss": 1.1526,
      "step": 574
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.28414380131218553,
      "learning_rate": 0.00019699223967999005,
      "loss": 1.1582,
      "step": 575
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2891445471626401,
      "learning_rate": 0.0001969776952661642,
      "loss": 1.0203,
      "step": 576
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.25017894540508634,
      "learning_rate": 0.0001969631163109468,
      "loss": 1.2667,
      "step": 577
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.23618920284864958,
      "learning_rate": 0.00019694850281953055,
      "loss": 1.2219,
      "step": 578
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.18502157908923197,
      "learning_rate": 0.00019693385479712048,
      "loss": 1.1015,
      "step": 579
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.19916827997613862,
      "learning_rate": 0.00019691917224893386,
      "loss": 1.2291,
      "step": 580
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2085452151572698,
      "learning_rate": 0.00019690445518020036,
      "loss": 1.1418,
      "step": 581
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1905184300592896,
      "learning_rate": 0.00019688970359616183,
      "loss": 1.1545,
      "step": 582
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.19318952740883058,
      "learning_rate": 0.00019687491750207254,
      "loss": 1.1089,
      "step": 583
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.20302145601987806,
      "learning_rate": 0.0001968600969031989,
      "loss": 1.1141,
      "step": 584
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1811066516611338,
      "learning_rate": 0.00019684524180481976,
      "loss": 1.2128,
      "step": 585
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.21536591835504865,
      "learning_rate": 0.00019683035221222618,
      "loss": 1.0344,
      "step": 586
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.22377792668530985,
      "learning_rate": 0.00019681542813072145,
      "loss": 1.2354,
      "step": 587
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.17114628027817708,
      "learning_rate": 0.0001968004695656213,
      "loss": 1.0387,
      "step": 588
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2009714582768279,
      "learning_rate": 0.00019678547652225356,
      "loss": 1.2265,
      "step": 589
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2103966063527931,
      "learning_rate": 0.00019677044900595853,
      "loss": 1.1376,
      "step": 590
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2603944562171985,
      "learning_rate": 0.00019675538702208862,
      "loss": 1.058,
      "step": 591
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.18362917871539217,
      "learning_rate": 0.00019674029057600864,
      "loss": 1.1147,
      "step": 592
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.20923842780812535,
      "learning_rate": 0.00019672515967309557,
      "loss": 1.1203,
      "step": 593
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.18780982101749502,
      "learning_rate": 0.00019670999431873874,
      "loss": 1.13,
      "step": 594
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.21651304664245158,
      "learning_rate": 0.00019669479451833976,
      "loss": 1.1218,
      "step": 595
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2220128941880201,
      "learning_rate": 0.00019667956027731242,
      "loss": 1.2173,
      "step": 596
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.23793586021275281,
      "learning_rate": 0.0001966642916010829,
      "loss": 1.0877,
      "step": 597
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.18770396797233427,
      "learning_rate": 0.0001966489884950895,
      "loss": 1.1191,
      "step": 598
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.229362250037237,
      "learning_rate": 0.0001966336509647829,
      "loss": 1.2415,
      "step": 599
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.19254855040191735,
      "learning_rate": 0.000196618279015626,
      "loss": 1.2529,
      "step": 600
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.21769445517573777,
      "learning_rate": 0.00019660287265309395,
      "loss": 1.1482,
      "step": 601
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.233553555568854,
      "learning_rate": 0.0001965874318826742,
      "loss": 1.2262,
      "step": 602
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.26808253791848524,
      "learning_rate": 0.00019657195670986637,
      "loss": 1.2388,
      "step": 603
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2130430134270506,
      "learning_rate": 0.00019655644714018243,
      "loss": 1.0454,
      "step": 604
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2580129587206687,
      "learning_rate": 0.00019654090317914656,
      "loss": 1.1463,
      "step": 605
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2094404064397588,
      "learning_rate": 0.00019652532483229514,
      "loss": 1.2308,
      "step": 606
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.19962222436953725,
      "learning_rate": 0.00019650971210517692,
      "loss": 1.2162,
      "step": 607
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.20204973095843265,
      "learning_rate": 0.00019649406500335276,
      "loss": 0.9682,
      "step": 608
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.18343087148698847,
      "learning_rate": 0.00019647838353239588,
      "loss": 1.0321,
      "step": 609
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.22991383933567452,
      "learning_rate": 0.0001964626676978916,
      "loss": 1.0358,
      "step": 610
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1996129157411663,
      "learning_rate": 0.00019644691750543767,
      "loss": 1.1176,
      "step": 611
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.23359721189702037,
      "learning_rate": 0.0001964311329606439,
      "loss": 0.9577,
      "step": 612
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.22339145979352493,
      "learning_rate": 0.00019641531406913243,
      "loss": 1.2358,
      "step": 613
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.23267539903586845,
      "learning_rate": 0.00019639946083653762,
      "loss": 1.1693,
      "step": 614
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2194177176286009,
      "learning_rate": 0.00019638357326850602,
      "loss": 1.0192,
      "step": 615
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.20352281332699462,
      "learning_rate": 0.00019636765137069653,
      "loss": 1.1262,
      "step": 616
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.21192005345983062,
      "learning_rate": 0.00019635169514878008,
      "loss": 1.1622,
      "step": 617
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.24667657431852077,
      "learning_rate": 0.00019633570460844002,
      "loss": 1.1018,
      "step": 618
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.21190069769529363,
      "learning_rate": 0.0001963196797553718,
      "loss": 1.0267,
      "step": 619
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.20662492717239656,
      "learning_rate": 0.00019630362059528316,
      "loss": 1.0683,
      "step": 620
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1713284093259186,
      "learning_rate": 0.00019628752713389398,
      "loss": 1.132,
      "step": 621
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.21437935358774554,
      "learning_rate": 0.00019627139937693645,
      "loss": 1.2149,
      "step": 622
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1951278825075767,
      "learning_rate": 0.00019625523733015493,
      "loss": 1.2288,
      "step": 623
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1809888457443517,
      "learning_rate": 0.00019623904099930602,
      "loss": 1.2027,
      "step": 624
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.21027935263423034,
      "learning_rate": 0.00019622281039015843,
      "loss": 0.9381,
      "step": 625
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.25799617366532535,
      "learning_rate": 0.00019620654550849323,
      "loss": 1.1244,
      "step": 626
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.18779446925692506,
      "learning_rate": 0.00019619024636010363,
      "loss": 1.0928,
      "step": 627
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.20786112828172001,
      "learning_rate": 0.000196173912950795,
      "loss": 1.0439,
      "step": 628
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1904989557452055,
      "learning_rate": 0.00019615754528638495,
      "loss": 1.1753,
      "step": 629
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2331150517719497,
      "learning_rate": 0.0001961411433727033,
      "loss": 1.146,
      "step": 630
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.19810400388747818,
      "learning_rate": 0.00019612470721559207,
      "loss": 1.1106,
      "step": 631
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2250292446881762,
      "learning_rate": 0.0001961082368209055,
      "loss": 0.9509,
      "step": 632
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.23805346289598223,
      "learning_rate": 0.00019609173219450998,
      "loss": 1.0222,
      "step": 633
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.19723386026827014,
      "learning_rate": 0.00019607519334228406,
      "loss": 1.0178,
      "step": 634
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.23685847423988238,
      "learning_rate": 0.00019605862027011856,
      "loss": 1.1724,
      "step": 635
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.21527444243553687,
      "learning_rate": 0.00019604201298391646,
      "loss": 1.1243,
      "step": 636
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.34611403037076865,
      "learning_rate": 0.00019602537148959292,
      "loss": 0.9613,
      "step": 637
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.29890883935643453,
      "learning_rate": 0.0001960086957930753,
      "loss": 1.0383,
      "step": 638
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1741330162241599,
      "learning_rate": 0.0001959919859003031,
      "loss": 1.1818,
      "step": 639
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2078750685990968,
      "learning_rate": 0.00019597524181722803,
      "loss": 1.2091,
      "step": 640
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.196730361092366,
      "learning_rate": 0.000195958463549814,
      "loss": 1.158,
      "step": 641
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.22181382820312778,
      "learning_rate": 0.00019594165110403707,
      "loss": 1.1222,
      "step": 642
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.19263992473533859,
      "learning_rate": 0.00019592480448588542,
      "loss": 1.1203,
      "step": 643
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.191113043935449,
      "learning_rate": 0.00019590792370135953,
      "loss": 1.3292,
      "step": 644
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.19311351152657869,
      "learning_rate": 0.00019589100875647196,
      "loss": 1.1795,
      "step": 645
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1825986681885503,
      "learning_rate": 0.0001958740596572474,
      "loss": 1.2548,
      "step": 646
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.18959767249352485,
      "learning_rate": 0.0001958570764097228,
      "loss": 1.0723,
      "step": 647
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.17837004136870088,
      "learning_rate": 0.0001958400590199473,
      "loss": 1.1616,
      "step": 648
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.21861636144035046,
      "learning_rate": 0.000195823007493982,
      "loss": 1.1953,
      "step": 649
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.17957667112706155,
      "learning_rate": 0.00019580592183790035,
      "loss": 1.1569,
      "step": 650
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1989840496231291,
      "learning_rate": 0.00019578880205778793,
      "loss": 1.2021,
      "step": 651
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.22928257124437573,
      "learning_rate": 0.0001957716481597424,
      "loss": 1.1489,
      "step": 652
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.17429146996417114,
      "learning_rate": 0.00019575446014987363,
      "loss": 1.0042,
      "step": 653
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.22706853883115607,
      "learning_rate": 0.0001957372380343036,
      "loss": 1.0921,
      "step": 654
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.22231730498647867,
      "learning_rate": 0.0001957199818191665,
      "loss": 1.0559,
      "step": 655
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1931588426162047,
      "learning_rate": 0.0001957026915106086,
      "loss": 1.2408,
      "step": 656
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.18990189747981445,
      "learning_rate": 0.00019568536711478835,
      "loss": 1.0935,
      "step": 657
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.18885372110832524,
      "learning_rate": 0.00019566800863787633,
      "loss": 1.0736,
      "step": 658
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.23685103523834503,
      "learning_rate": 0.00019565061608605526,
      "loss": 1.0922,
      "step": 659
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1769058978610504,
      "learning_rate": 0.00019563318946551998,
      "loss": 1.2226,
      "step": 660
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2151104379999775,
      "learning_rate": 0.00019561572878247748,
      "loss": 1.1731,
      "step": 661
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.18721214221347515,
      "learning_rate": 0.00019559823404314696,
      "loss": 1.0206,
      "step": 662
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.21520794828260967,
      "learning_rate": 0.00019558070525375952,
      "loss": 1.0244,
      "step": 663
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.20768255139749559,
      "learning_rate": 0.00019556314242055872,
      "loss": 1.184,
      "step": 664
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2214062657832189,
      "learning_rate": 0.00019554554554979995,
      "loss": 1.0665,
      "step": 665
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.17947043518341496,
      "learning_rate": 0.00019552791464775088,
      "loss": 1.0413,
      "step": 666
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2618071551130766,
      "learning_rate": 0.00019551024972069126,
      "loss": 1.0794,
      "step": 667
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2834650141025881,
      "learning_rate": 0.00019549255077491292,
      "loss": 1.2702,
      "step": 668
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.19771826249673324,
      "learning_rate": 0.00019547481781671992,
      "loss": 1.0793,
      "step": 669
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.24150545904941914,
      "learning_rate": 0.00019545705085242834,
      "loss": 1.0933,
      "step": 670
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.21443845952885185,
      "learning_rate": 0.00019543924988836637,
      "loss": 1.1766,
      "step": 671
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.17901394537244114,
      "learning_rate": 0.00019542141493087434,
      "loss": 1.0678,
      "step": 672
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.17238028027467786,
      "learning_rate": 0.00019540354598630467,
      "loss": 1.1354,
      "step": 673
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.28426237740186566,
      "learning_rate": 0.00019538564306102195,
      "loss": 1.1177,
      "step": 674
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.21036845929911582,
      "learning_rate": 0.00019536770616140276,
      "loss": 1.061,
      "step": 675
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.19204523357232325,
      "learning_rate": 0.0001953497352938359,
      "loss": 1.1607,
      "step": 676
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.21621226572697952,
      "learning_rate": 0.00019533173046472216,
      "loss": 1.1655,
      "step": 677
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.21974829901137624,
      "learning_rate": 0.0001953136916804745,
      "loss": 1.2908,
      "step": 678
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.15761955778862552,
      "learning_rate": 0.00019529561894751795,
      "loss": 1.0133,
      "step": 679
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.19317611315028724,
      "learning_rate": 0.00019527751227228963,
      "loss": 1.1478,
      "step": 680
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.18765693022086538,
      "learning_rate": 0.00019525937166123877,
      "loss": 1.0848,
      "step": 681
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.21944147478427775,
      "learning_rate": 0.00019524119712082666,
      "loss": 1.1947,
      "step": 682
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.19164411974574297,
      "learning_rate": 0.0001952229886575266,
      "loss": 1.1191,
      "step": 683
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.21085523035781403,
      "learning_rate": 0.00019520474627782424,
      "loss": 1.0862,
      "step": 684
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.34652835631652407,
      "learning_rate": 0.00019518646998821696,
      "loss": 1.2697,
      "step": 685
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2004156958683983,
      "learning_rate": 0.00019516815979521445,
      "loss": 1.0495,
      "step": 686
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.19802846990117293,
      "learning_rate": 0.00019514981570533846,
      "loss": 1.1019,
      "step": 687
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.19115484336350716,
      "learning_rate": 0.00019513143772512267,
      "loss": 1.0367,
      "step": 688
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.19586770369858,
      "learning_rate": 0.00019511302586111296,
      "loss": 1.0957,
      "step": 689
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.25424589418492444,
      "learning_rate": 0.00019509458011986728,
      "loss": 1.0535,
      "step": 690
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.19510930589172296,
      "learning_rate": 0.00019507610050795558,
      "loss": 1.2027,
      "step": 691
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.18093255658983826,
      "learning_rate": 0.00019505758703195988,
      "loss": 1.2559,
      "step": 692
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.19287803305062456,
      "learning_rate": 0.00019503903969847437,
      "loss": 1.0619,
      "step": 693
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.322291585549914,
      "learning_rate": 0.00019502045851410514,
      "loss": 1.0163,
      "step": 694
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1767569884296946,
      "learning_rate": 0.00019500184348547042,
      "loss": 1.0983,
      "step": 695
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2757354423555372,
      "learning_rate": 0.0001949831946192005,
      "loss": 1.0448,
      "step": 696
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.19522305545719887,
      "learning_rate": 0.0001949645119219377,
      "loss": 1.1201,
      "step": 697
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1662332402438219,
      "learning_rate": 0.0001949457954003364,
      "loss": 1.1922,
      "step": 698
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.20898062788471258,
      "learning_rate": 0.0001949270450610631,
      "loss": 1.1878,
      "step": 699
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2470609735333508,
      "learning_rate": 0.00019490826091079615,
      "loss": 1.1374,
      "step": 700
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2995632789980061,
      "learning_rate": 0.00019488944295622615,
      "loss": 1.0358,
      "step": 701
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.18162875385819727,
      "learning_rate": 0.0001948705912040556,
      "loss": 1.1779,
      "step": 702
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1986724560374396,
      "learning_rate": 0.00019485170566099913,
      "loss": 1.1268,
      "step": 703
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.23793863949998018,
      "learning_rate": 0.00019483278633378336,
      "loss": 1.043,
      "step": 704
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.17533351441457976,
      "learning_rate": 0.00019481383322914698,
      "loss": 1.1015,
      "step": 705
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.20002451386592693,
      "learning_rate": 0.00019479484635384061,
      "loss": 1.1835,
      "step": 706
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.19500626711679075,
      "learning_rate": 0.00019477582571462705,
      "loss": 1.0435,
      "step": 707
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.17193230192805972,
      "learning_rate": 0.00019475677131828095,
      "loss": 1.0803,
      "step": 708
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.21534116539549927,
      "learning_rate": 0.0001947376831715892,
      "loss": 1.0854,
      "step": 709
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3916886023805627,
      "learning_rate": 0.0001947185612813505,
      "loss": 1.1722,
      "step": 710
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.17618660141530446,
      "learning_rate": 0.00019469940565437573,
      "loss": 1.2001,
      "step": 711
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.193478841778336,
      "learning_rate": 0.00019468021629748768,
      "loss": 1.0815,
      "step": 712
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.19221270886865552,
      "learning_rate": 0.00019466099321752115,
      "loss": 1.0939,
      "step": 713
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.21483033562369008,
      "learning_rate": 0.00019464173642132307,
      "loss": 1.2669,
      "step": 714
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.24597953723443294,
      "learning_rate": 0.00019462244591575222,
      "loss": 1.1729,
      "step": 715
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1898605530534347,
      "learning_rate": 0.00019460312170767956,
      "loss": 1.0759,
      "step": 716
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.22809473824611423,
      "learning_rate": 0.00019458376380398788,
      "loss": 1.1395,
      "step": 717
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.21264528334383048,
      "learning_rate": 0.0001945643722115721,
      "loss": 1.1306,
      "step": 718
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3040222269539394,
      "learning_rate": 0.0001945449469373391,
      "loss": 1.1612,
      "step": 719
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1699884635483011,
      "learning_rate": 0.00019452548798820768,
      "loss": 1.0719,
      "step": 720
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.18957117285960337,
      "learning_rate": 0.00019450599537110881,
      "loss": 1.22,
      "step": 721
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.22221111634622825,
      "learning_rate": 0.00019448646909298525,
      "loss": 1.1677,
      "step": 722
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.21747796161904262,
      "learning_rate": 0.0001944669091607919,
      "loss": 1.1021,
      "step": 723
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.18961323645756234,
      "learning_rate": 0.00019444731558149555,
      "loss": 1.0738,
      "step": 724
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.19367527993941547,
      "learning_rate": 0.00019442768836207508,
      "loss": 1.0545,
      "step": 725
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.22989394117630835,
      "learning_rate": 0.00019440802750952122,
      "loss": 1.1303,
      "step": 726
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.258070266511484,
      "learning_rate": 0.00019438833303083678,
      "loss": 1.2039,
      "step": 727
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.23155389189525274,
      "learning_rate": 0.0001943686049330365,
      "loss": 1.13,
      "step": 728
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.20838776571421158,
      "learning_rate": 0.00019434884322314713,
      "loss": 1.1767,
      "step": 729
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.22694723215071583,
      "learning_rate": 0.00019432904790820735,
      "loss": 1.1438,
      "step": 730
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.24563629960646507,
      "learning_rate": 0.00019430921899526787,
      "loss": 1.1025,
      "step": 731
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.27580688894717664,
      "learning_rate": 0.00019428935649139123,
      "loss": 1.1116,
      "step": 732
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2130976334853932,
      "learning_rate": 0.00019426946040365217,
      "loss": 1.2604,
      "step": 733
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.17266617855364141,
      "learning_rate": 0.00019424953073913714,
      "loss": 1.0357,
      "step": 734
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.21894636508069637,
      "learning_rate": 0.00019422956750494473,
      "loss": 0.9474,
      "step": 735
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1735354250027064,
      "learning_rate": 0.00019420957070818536,
      "loss": 1.0519,
      "step": 736
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2173428987846353,
      "learning_rate": 0.00019418954035598152,
      "loss": 1.2557,
      "step": 737
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.20224034463178714,
      "learning_rate": 0.00019416947645546756,
      "loss": 1.2728,
      "step": 738
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.19786437926013134,
      "learning_rate": 0.00019414937901378982,
      "loss": 1.1077,
      "step": 739
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.196829871815927,
      "learning_rate": 0.0001941292480381066,
      "loss": 1.1184,
      "step": 740
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.21807699621983648,
      "learning_rate": 0.0001941090835355881,
      "loss": 1.1223,
      "step": 741
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2459976096374817,
      "learning_rate": 0.0001940888855134165,
      "loss": 1.1868,
      "step": 742
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1837498540212578,
      "learning_rate": 0.00019406865397878588,
      "loss": 1.2567,
      "step": 743
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.17309500559289104,
      "learning_rate": 0.00019404838893890235,
      "loss": 1.1347,
      "step": 744
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.17869524145661156,
      "learning_rate": 0.00019402809040098382,
      "loss": 1.1932,
      "step": 745
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.20563802094343459,
      "learning_rate": 0.00019400775837226023,
      "loss": 1.3487,
      "step": 746
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.25540543758418355,
      "learning_rate": 0.0001939873928599734,
      "loss": 1.2055,
      "step": 747
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.241807351789914,
      "learning_rate": 0.00019396699387137708,
      "loss": 1.2011,
      "step": 748
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.19435674821726592,
      "learning_rate": 0.00019394656141373702,
      "loss": 1.0187,
      "step": 749
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2682987646147018,
      "learning_rate": 0.0001939260954943307,
      "loss": 0.8387,
      "step": 750
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.16932251092143208,
      "learning_rate": 0.0001939055961204478,
      "loss": 1.0765,
      "step": 751
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.22527547682503019,
      "learning_rate": 0.00019388506329938966,
      "loss": 1.1452,
      "step": 752
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.18931199882327715,
      "learning_rate": 0.00019386449703846968,
      "loss": 1.06,
      "step": 753
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2540834605643452,
      "learning_rate": 0.00019384389734501307,
      "loss": 1.084,
      "step": 754
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.23834752426515182,
      "learning_rate": 0.00019382326422635705,
      "loss": 1.0589,
      "step": 755
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.19196911434707684,
      "learning_rate": 0.00019380259768985068,
      "loss": 1.1789,
      "step": 756
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1925147484873092,
      "learning_rate": 0.00019378189774285496,
      "loss": 1.1549,
      "step": 757
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.23494699132667687,
      "learning_rate": 0.00019376116439274275,
      "loss": 1.1434,
      "step": 758
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.20764859768907795,
      "learning_rate": 0.00019374039764689883,
      "loss": 1.0657,
      "step": 759
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.18996993042613416,
      "learning_rate": 0.00019371959751271985,
      "loss": 1.2402,
      "step": 760
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.20177617823909216,
      "learning_rate": 0.00019369876399761443,
      "loss": 1.1542,
      "step": 761
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.268968488920488,
      "learning_rate": 0.000193677897109003,
      "loss": 1.1893,
      "step": 762
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.23092237419455375,
      "learning_rate": 0.0001936569968543179,
      "loss": 1.1496,
      "step": 763
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.19124315700382777,
      "learning_rate": 0.00019363606324100335,
      "loss": 1.0519,
      "step": 764
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2644700937184694,
      "learning_rate": 0.00019361509627651547,
      "loss": 1.17,
      "step": 765
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.22815232244938094,
      "learning_rate": 0.00019359409596832224,
      "loss": 1.1915,
      "step": 766
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.21139433918289433,
      "learning_rate": 0.0001935730623239035,
      "loss": 1.0168,
      "step": 767
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.18042308357865783,
      "learning_rate": 0.00019355199535075102,
      "loss": 1.164,
      "step": 768
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.21273571394330082,
      "learning_rate": 0.00019353089505636838,
      "loss": 1.1924,
      "step": 769
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.19083381152703166,
      "learning_rate": 0.00019350976144827107,
      "loss": 1.0372,
      "step": 770
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.16233729028659158,
      "learning_rate": 0.00019348859453398646,
      "loss": 1.1275,
      "step": 771
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.23788991907839113,
      "learning_rate": 0.00019346739432105374,
      "loss": 0.9711,
      "step": 772
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.19187112268178225,
      "learning_rate": 0.00019344616081702393,
      "loss": 1.2024,
      "step": 773
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.23199932968233944,
      "learning_rate": 0.00019342489402945998,
      "loss": 1.2033,
      "step": 774
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.19903571845433524,
      "learning_rate": 0.00019340359396593672,
      "loss": 1.3405,
      "step": 775
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.40100114711859214,
      "learning_rate": 0.0001933822606340407,
      "loss": 1.0081,
      "step": 776
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.19723374679669187,
      "learning_rate": 0.0001933608940413704,
      "loss": 1.1911,
      "step": 777
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.18085387209496784,
      "learning_rate": 0.00019333949419553622,
      "loss": 1.1147,
      "step": 778
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2610320117505477,
      "learning_rate": 0.00019331806110416027,
      "loss": 1.254,
      "step": 779
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.22874811992093352,
      "learning_rate": 0.0001932965947748766,
      "loss": 1.1903,
      "step": 780
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.19663449532083435,
      "learning_rate": 0.000193275095215331,
      "loss": 1.1733,
      "step": 781
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.17580809341948747,
      "learning_rate": 0.0001932535624331812,
      "loss": 0.8971,
      "step": 782
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.18019120618642187,
      "learning_rate": 0.00019323199643609673,
      "loss": 1.3178,
      "step": 783
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.20467492163879253,
      "learning_rate": 0.00019321039723175888,
      "loss": 1.2474,
      "step": 784
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.21063360951537816,
      "learning_rate": 0.00019318876482786094,
      "loss": 1.2061,
      "step": 785
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.26205160681941403,
      "learning_rate": 0.00019316709923210775,
      "loss": 1.2426,
      "step": 786
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1881183148161407,
      "learning_rate": 0.00019314540045221626,
      "loss": 1.1638,
      "step": 787
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.17768331740163343,
      "learning_rate": 0.0001931236684959151,
      "loss": 1.0157,
      "step": 788
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.20825977905618442,
      "learning_rate": 0.0001931019033709447,
      "loss": 1.131,
      "step": 789
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.22715277036658332,
      "learning_rate": 0.00019308010508505734,
      "loss": 1.0625,
      "step": 790
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.23157265160826004,
      "learning_rate": 0.0001930582736460171,
      "loss": 1.1585,
      "step": 791
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.25124208420220134,
      "learning_rate": 0.0001930364090615999,
      "loss": 1.1692,
      "step": 792
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2160678395422809,
      "learning_rate": 0.0001930145113395934,
      "loss": 1.1019,
      "step": 793
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1923108486975328,
      "learning_rate": 0.00019299258048779714,
      "loss": 1.1644,
      "step": 794
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.19822545876133052,
      "learning_rate": 0.00019297061651402236,
      "loss": 1.054,
      "step": 795
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1893723309604151,
      "learning_rate": 0.00019294861942609228,
      "loss": 1.0813,
      "step": 796
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.23631379073144693,
      "learning_rate": 0.00019292658923184165,
      "loss": 1.0548,
      "step": 797
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.22310953318788448,
      "learning_rate": 0.0001929045259391173,
      "loss": 1.2385,
      "step": 798
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.19547798767766228,
      "learning_rate": 0.00019288242955577757,
      "loss": 1.1791,
      "step": 799
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.21229462613875963,
      "learning_rate": 0.00019286030008969283,
      "loss": 1.2298,
      "step": 800
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.19733052732198522,
      "learning_rate": 0.00019283813754874504,
      "loss": 1.0721,
      "step": 801
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.18283130291634275,
      "learning_rate": 0.00019281594194082811,
      "loss": 1.1323,
      "step": 802
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1880863333374766,
      "learning_rate": 0.0001927937132738476,
      "loss": 1.2104,
      "step": 803
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2910072467622591,
      "learning_rate": 0.00019277145155572085,
      "loss": 1.1389,
      "step": 804
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1906555642113488,
      "learning_rate": 0.00019274915679437705,
      "loss": 1.0503,
      "step": 805
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1874441648312473,
      "learning_rate": 0.0001927268289977571,
      "loss": 1.0222,
      "step": 806
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.24661299391694375,
      "learning_rate": 0.00019270446817381377,
      "loss": 1.0086,
      "step": 807
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.17536274329755308,
      "learning_rate": 0.0001926820743305114,
      "loss": 1.1539,
      "step": 808
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1877998688580868,
      "learning_rate": 0.00019265964747582626,
      "loss": 1.1268,
      "step": 809
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.20158450611730283,
      "learning_rate": 0.00019263718761774627,
      "loss": 1.265,
      "step": 810
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2183456214316565,
      "learning_rate": 0.0001926146947642712,
      "loss": 1.2623,
      "step": 811
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1825864762457915,
      "learning_rate": 0.00019259216892341253,
      "loss": 1.2273,
      "step": 812
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2424516071838109,
      "learning_rate": 0.00019256961010319344,
      "loss": 1.0758,
      "step": 813
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1642785470497717,
      "learning_rate": 0.00019254701831164893,
      "loss": 1.137,
      "step": 814
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1764341993506038,
      "learning_rate": 0.00019252439355682573,
      "loss": 1.1499,
      "step": 815
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.18785165447998006,
      "learning_rate": 0.00019250173584678225,
      "loss": 1.2377,
      "step": 816
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.20948942144601246,
      "learning_rate": 0.00019247904518958876,
      "loss": 1.038,
      "step": 817
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.22634496102728796,
      "learning_rate": 0.0001924563215933271,
      "loss": 1.2,
      "step": 818
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.19574918016525766,
      "learning_rate": 0.000192433565066091,
      "loss": 0.9961,
      "step": 819
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1802906696766204,
      "learning_rate": 0.0001924107756159858,
      "loss": 1.1507,
      "step": 820
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2039722261820194,
      "learning_rate": 0.0001923879532511287,
      "loss": 1.063,
      "step": 821
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.26688070373852657,
      "learning_rate": 0.00019236509797964843,
      "loss": 1.0329,
      "step": 822
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.19902849596898198,
      "learning_rate": 0.0001923422098096856,
      "loss": 1.2284,
      "step": 823
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.19053285487457783,
      "learning_rate": 0.00019231928874939255,
      "loss": 1.1499,
      "step": 824
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2444621066916314,
      "learning_rate": 0.00019229633480693316,
      "loss": 1.171,
      "step": 825
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.18776003293871582,
      "learning_rate": 0.0001922733479904832,
      "loss": 1.0682,
      "step": 826
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.21729577066355513,
      "learning_rate": 0.00019225032830823011,
      "loss": 1.0995,
      "step": 827
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.17612105644734707,
      "learning_rate": 0.00019222727576837295,
      "loss": 1.0697,
      "step": 828
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.21701137675026277,
      "learning_rate": 0.00019220419037912261,
      "loss": 1.1262,
      "step": 829
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2184791338039768,
      "learning_rate": 0.00019218107214870155,
      "loss": 1.0756,
      "step": 830
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1921895245050513,
      "learning_rate": 0.000192157921085344,
      "loss": 1.0805,
      "step": 831
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.565713398915103,
      "learning_rate": 0.00019213473719729594,
      "loss": 1.2401,
      "step": 832
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.23469413762549707,
      "learning_rate": 0.0001921115204928149,
      "loss": 1.1555,
      "step": 833
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2093277117797833,
      "learning_rate": 0.00019208827098017022,
      "loss": 1.0571,
      "step": 834
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.21205887242166577,
      "learning_rate": 0.00019206498866764288,
      "loss": 1.0822,
      "step": 835
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.20815630745735855,
      "learning_rate": 0.00019204167356352554,
      "loss": 1.2809,
      "step": 836
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1922887924784784,
      "learning_rate": 0.00019201832567612253,
      "loss": 1.18,
      "step": 837
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.257543155515596,
      "learning_rate": 0.0001919949450137499,
      "loss": 1.0609,
      "step": 838
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.21934508957763327,
      "learning_rate": 0.00019197153158473528,
      "loss": 1.2232,
      "step": 839
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1841365390160391,
      "learning_rate": 0.00019194808539741814,
      "loss": 1.0713,
      "step": 840
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.17862761275240194,
      "learning_rate": 0.00019192460646014938,
      "loss": 1.1453,
      "step": 841
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.19131466885795945,
      "learning_rate": 0.0001919010947812918,
      "loss": 0.9882,
      "step": 842
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2476823059941616,
      "learning_rate": 0.00019187755036921978,
      "loss": 1.2539,
      "step": 843
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.17138726308726968,
      "learning_rate": 0.00019185397323231925,
      "loss": 1.1099,
      "step": 844
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.17632138116717783,
      "learning_rate": 0.0001918303633789879,
      "loss": 1.0751,
      "step": 845
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.18924548979877584,
      "learning_rate": 0.0001918067208176351,
      "loss": 1.1761,
      "step": 846
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.19988609597893606,
      "learning_rate": 0.00019178304555668186,
      "loss": 1.0336,
      "step": 847
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.17936398744515042,
      "learning_rate": 0.00019175933760456069,
      "loss": 1.0954,
      "step": 848
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1932092296515352,
      "learning_rate": 0.00019173559696971594,
      "loss": 1.1411,
      "step": 849
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.17332883926260856,
      "learning_rate": 0.0001917118236606035,
      "loss": 1.1303,
      "step": 850
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.18456640523402593,
      "learning_rate": 0.0001916880176856909,
      "loss": 1.1344,
      "step": 851
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.17555239951527604,
      "learning_rate": 0.00019166417905345736,
      "loss": 1.0842,
      "step": 852
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3378891166193439,
      "learning_rate": 0.00019164030777239367,
      "loss": 0.9772,
      "step": 853
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.20818427366230915,
      "learning_rate": 0.00019161640385100228,
      "loss": 1.2091,
      "step": 854
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2249168433367931,
      "learning_rate": 0.00019159246729779725,
      "loss": 1.1413,
      "step": 855
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2102743711763353,
      "learning_rate": 0.00019156849812130427,
      "loss": 1.2279,
      "step": 856
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.22301106090924536,
      "learning_rate": 0.00019154449633006064,
      "loss": 1.0462,
      "step": 857
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.19805780455459707,
      "learning_rate": 0.0001915204619326153,
      "loss": 1.1278,
      "step": 858
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.19922243643745485,
      "learning_rate": 0.0001914963949375288,
      "loss": 1.2572,
      "step": 859
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2543154270351458,
      "learning_rate": 0.0001914722953533733,
      "loss": 1.0555,
      "step": 860
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2189555071583148,
      "learning_rate": 0.00019144816318873253,
      "loss": 1.2858,
      "step": 861
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.17456621324254548,
      "learning_rate": 0.00019142399845220188,
      "loss": 1.1006,
      "step": 862
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.20408822104756547,
      "learning_rate": 0.00019139980115238827,
      "loss": 1.199,
      "step": 863
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.26846745805127376,
      "learning_rate": 0.00019137557129791032,
      "loss": 1.1437,
      "step": 864
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.18877009209153783,
      "learning_rate": 0.00019135130889739817,
      "loss": 1.2925,
      "step": 865
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1994139784089647,
      "learning_rate": 0.00019132701395949356,
      "loss": 1.2245,
      "step": 866
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.18735746164130782,
      "learning_rate": 0.0001913026864928498,
      "loss": 1.145,
      "step": 867
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.24986351086363562,
      "learning_rate": 0.00019127832650613189,
      "loss": 1.0134,
      "step": 868
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.16262764691460993,
      "learning_rate": 0.0001912539340080163,
      "loss": 1.1845,
      "step": 869
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.24555168286368192,
      "learning_rate": 0.0001912295090071911,
      "loss": 1.1026,
      "step": 870
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.19434559670452772,
      "learning_rate": 0.00019120505151235597,
      "loss": 0.9872,
      "step": 871
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.24291788985577473,
      "learning_rate": 0.0001911805615322222,
      "loss": 1.0413,
      "step": 872
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1809596208015643,
      "learning_rate": 0.0001911560390755125,
      "loss": 1.2179,
      "step": 873
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.16185201026996995,
      "learning_rate": 0.0001911314841509613,
      "loss": 1.0603,
      "step": 874
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.21803570104925452,
      "learning_rate": 0.00019110689676731454,
      "loss": 1.1065,
      "step": 875
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1843894181665819,
      "learning_rate": 0.00019108227693332975,
      "loss": 1.2312,
      "step": 876
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1932574985269317,
      "learning_rate": 0.00019105762465777598,
      "loss": 1.0059,
      "step": 877
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.24455132985411082,
      "learning_rate": 0.00019103293994943382,
      "loss": 1.1382,
      "step": 878
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.17422757793440272,
      "learning_rate": 0.00019100822281709542,
      "loss": 1.144,
      "step": 879
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.18525102646031505,
      "learning_rate": 0.00019098347326956453,
      "loss": 1.111,
      "step": 880
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.19389121513270044,
      "learning_rate": 0.00019095869131565644,
      "loss": 1.1196,
      "step": 881
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.20310829132422573,
      "learning_rate": 0.00019093387696419788,
      "loss": 1.1662,
      "step": 882
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.20576609844074456,
      "learning_rate": 0.00019090903022402729,
      "loss": 1.0549,
      "step": 883
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.19319396202346864,
      "learning_rate": 0.00019088415110399444,
      "loss": 1.0071,
      "step": 884
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.19013309688506239,
      "learning_rate": 0.00019085923961296083,
      "loss": 0.9603,
      "step": 885
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.21560440719762183,
      "learning_rate": 0.00019083429575979935,
      "loss": 1.046,
      "step": 886
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.18833456797487602,
      "learning_rate": 0.00019080931955339455,
      "loss": 1.1091,
      "step": 887
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.17681096039986982,
      "learning_rate": 0.00019078431100264234,
      "loss": 1.286,
      "step": 888
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.23969651105589987,
      "learning_rate": 0.00019075927011645024,
      "loss": 1.1717,
      "step": 889
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2513134181428817,
      "learning_rate": 0.00019073419690373731,
      "loss": 1.1607,
      "step": 890
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.19180062027323155,
      "learning_rate": 0.00019070909137343408,
      "loss": 1.0674,
      "step": 891
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.21197629420862527,
      "learning_rate": 0.0001906839535344826,
      "loss": 1.1193,
      "step": 892
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1975243568168105,
      "learning_rate": 0.00019065878339583647,
      "loss": 0.9513,
      "step": 893
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.24228534114878716,
      "learning_rate": 0.00019063358096646073,
      "loss": 0.9612,
      "step": 894
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.19536318215922346,
      "learning_rate": 0.00019060834625533196,
      "loss": 1.152,
      "step": 895
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2012977252994394,
      "learning_rate": 0.0001905830792714382,
      "loss": 1.0888,
      "step": 896
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.20150272013806544,
      "learning_rate": 0.00019055778002377905,
      "loss": 1.0181,
      "step": 897
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.18843418098355513,
      "learning_rate": 0.00019053244852136554,
      "loss": 1.0828,
      "step": 898
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2507435661878565,
      "learning_rate": 0.00019050708477322018,
      "loss": 1.069,
      "step": 899
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.21832961787402988,
      "learning_rate": 0.00019048168878837708,
      "loss": 1.194,
      "step": 900
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.17305903043914805,
      "learning_rate": 0.00019045626057588168,
      "loss": 1.0927,
      "step": 901
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.17732184011601893,
      "learning_rate": 0.00019043080014479098,
      "loss": 0.9873,
      "step": 902
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.21132508812541184,
      "learning_rate": 0.00019040530750417348,
      "loss": 1.2261,
      "step": 903
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1771282200996104,
      "learning_rate": 0.0001903797826631091,
      "loss": 1.1993,
      "step": 904
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.16740585593441254,
      "learning_rate": 0.0001903542256306892,
      "loss": 1.1088,
      "step": 905
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.18141734419435757,
      "learning_rate": 0.0001903286364160167,
      "loss": 1.1721,
      "step": 906
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.19143594417763296,
      "learning_rate": 0.00019030301502820596,
      "loss": 1.1138,
      "step": 907
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.17417102954058605,
      "learning_rate": 0.0001902773614763827,
      "loss": 1.162,
      "step": 908
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.20092969208606223,
      "learning_rate": 0.00019025167576968422,
      "loss": 1.1446,
      "step": 909
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.19567240928299587,
      "learning_rate": 0.00019022595791725916,
      "loss": 1.2393,
      "step": 910
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.16671076987398448,
      "learning_rate": 0.00019020020792826777,
      "loss": 1.1101,
      "step": 911
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1724346980739866,
      "learning_rate": 0.0001901744258118816,
      "loss": 1.2579,
      "step": 912
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.19683772755612944,
      "learning_rate": 0.00019014861157728364,
      "loss": 1.2163,
      "step": 913
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.20990109428699386,
      "learning_rate": 0.00019012276523366848,
      "loss": 1.1165,
      "step": 914
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1870243619718924,
      "learning_rate": 0.0001900968867902419,
      "loss": 1.2505,
      "step": 915
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.18451049909711914,
      "learning_rate": 0.0001900709762562214,
      "loss": 1.1787,
      "step": 916
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.19920887194344467,
      "learning_rate": 0.00019004503364083566,
      "loss": 1.1147,
      "step": 917
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.16583639431857317,
      "learning_rate": 0.00019001905895332493,
      "loss": 1.2291,
      "step": 918
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2093265271126185,
      "learning_rate": 0.0001899930522029408,
      "loss": 1.0674,
      "step": 919
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.18171960620590794,
      "learning_rate": 0.00018996701339894635,
      "loss": 1.0815,
      "step": 920
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.24709077102008434,
      "learning_rate": 0.00018994094255061605,
      "loss": 1.1614,
      "step": 921
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.23069949460580608,
      "learning_rate": 0.0001899148396672358,
      "loss": 1.0803,
      "step": 922
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.23558806348011596,
      "learning_rate": 0.00018988870475810282,
      "loss": 1.0327,
      "step": 923
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.22006322005570317,
      "learning_rate": 0.00018986253783252592,
      "loss": 1.0864,
      "step": 924
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2265531932470592,
      "learning_rate": 0.00018983633889982512,
      "loss": 1.2989,
      "step": 925
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.22264878019744272,
      "learning_rate": 0.0001898101079693319,
      "loss": 1.1688,
      "step": 926
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.20056552248633036,
      "learning_rate": 0.00018978384505038924,
      "loss": 1.177,
      "step": 927
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.23275889516798043,
      "learning_rate": 0.0001897575501523514,
      "loss": 1.0759,
      "step": 928
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.18737067629602402,
      "learning_rate": 0.00018973122328458405,
      "loss": 1.043,
      "step": 929
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.17924764205110572,
      "learning_rate": 0.00018970486445646424,
      "loss": 1.1685,
      "step": 930
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.21015059904549405,
      "learning_rate": 0.00018967847367738048,
      "loss": 1.1183,
      "step": 931
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.20630165975139797,
      "learning_rate": 0.00018965205095673255,
      "loss": 1.2519,
      "step": 932
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1747415244930879,
      "learning_rate": 0.00018962559630393173,
      "loss": 1.0546,
      "step": 933
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.21205312308931176,
      "learning_rate": 0.00018959910972840053,
      "loss": 1.1101,
      "step": 934
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.18230943711691003,
      "learning_rate": 0.00018957259123957295,
      "loss": 1.2225,
      "step": 935
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2008622531393162,
      "learning_rate": 0.00018954604084689427,
      "loss": 1.1827,
      "step": 936
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.23468746895854828,
      "learning_rate": 0.00018951945855982122,
      "loss": 1.0414,
      "step": 937
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2590849192463141,
      "learning_rate": 0.00018949284438782182,
      "loss": 1.0162,
      "step": 938
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.19887953277430936,
      "learning_rate": 0.00018946619834037546,
      "loss": 1.0724,
      "step": 939
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1893164746278322,
      "learning_rate": 0.00018943952042697296,
      "loss": 1.1466,
      "step": 940
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1969402944746232,
      "learning_rate": 0.00018941281065711635,
      "loss": 0.9862,
      "step": 941
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.19250281615310166,
      "learning_rate": 0.00018938606904031914,
      "loss": 1.0924,
      "step": 942
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.18594048173211006,
      "learning_rate": 0.00018935929558610605,
      "loss": 0.9794,
      "step": 943
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.19589864933840867,
      "learning_rate": 0.00018933249030401327,
      "loss": 1.1468,
      "step": 944
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.18919275706348268,
      "learning_rate": 0.00018930565320358827,
      "loss": 1.2059,
      "step": 945
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.22271707167436927,
      "learning_rate": 0.00018927878429438985,
      "loss": 1.1446,
      "step": 946
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.19872914347170756,
      "learning_rate": 0.00018925188358598813,
      "loss": 1.1087,
      "step": 947
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.17973892923410023,
      "learning_rate": 0.00018922495108796459,
      "loss": 0.9879,
      "step": 948
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.19022878486943073,
      "learning_rate": 0.000189197986809912,
      "loss": 1.0133,
      "step": 949
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1860372634456013,
      "learning_rate": 0.00018917099076143443,
      "loss": 1.1998,
      "step": 950
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.17014031818456024,
      "learning_rate": 0.00018914396295214737,
      "loss": 1.1141,
      "step": 951
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1965189785892883,
      "learning_rate": 0.0001891169033916775,
      "loss": 1.2042,
      "step": 952
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.21052456886691934,
      "learning_rate": 0.00018908981208966288,
      "loss": 1.1949,
      "step": 953
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1961319780735773,
      "learning_rate": 0.00018906268905575285,
      "loss": 1.0485,
      "step": 954
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.27308862950942364,
      "learning_rate": 0.00018903553429960802,
      "loss": 1.0097,
      "step": 955
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.18002384306063007,
      "learning_rate": 0.00018900834783090043,
      "loss": 1.1419,
      "step": 956
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.28092737064880435,
      "learning_rate": 0.0001889811296593132,
      "loss": 1.1084,
      "step": 957
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.22615301709820884,
      "learning_rate": 0.000188953879794541,
      "loss": 1.0057,
      "step": 958
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.20317245124469724,
      "learning_rate": 0.00018892659824628952,
      "loss": 1.1216,
      "step": 959
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.18585071617932364,
      "learning_rate": 0.0001888992850242759,
      "loss": 1.1398,
      "step": 960
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.20540970368251563,
      "learning_rate": 0.00018887194013822862,
      "loss": 1.2583,
      "step": 961
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1843684553131277,
      "learning_rate": 0.00018884456359788724,
      "loss": 1.1206,
      "step": 962
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2085987086671775,
      "learning_rate": 0.00018881715541300276,
      "loss": 1.1877,
      "step": 963
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2017652157449278,
      "learning_rate": 0.00018878971559333735,
      "loss": 1.1305,
      "step": 964
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.22464468412449745,
      "learning_rate": 0.00018876224414866454,
      "loss": 1.0692,
      "step": 965
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.17990717761459527,
      "learning_rate": 0.00018873474108876902,
      "loss": 1.0852,
      "step": 966
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.17124859673122353,
      "learning_rate": 0.0001887072064234469,
      "loss": 1.2157,
      "step": 967
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.30334533232863153,
      "learning_rate": 0.00018867964016250528,
      "loss": 1.0607,
      "step": 968
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3104705014974402,
      "learning_rate": 0.00018865204231576277,
      "loss": 1.054,
      "step": 969
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.33274030010281436,
      "learning_rate": 0.00018862441289304915,
      "loss": 1.0533,
      "step": 970
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.20350225357811227,
      "learning_rate": 0.00018859675190420537,
      "loss": 1.2056,
      "step": 971
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.22963097418664258,
      "learning_rate": 0.0001885690593590838,
      "loss": 1.1378,
      "step": 972
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.18124591354504965,
      "learning_rate": 0.0001885413352675478,
      "loss": 1.1096,
      "step": 973
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.19704445023765382,
      "learning_rate": 0.00018851357963947218,
      "loss": 1.1166,
      "step": 974
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1880473219761876,
      "learning_rate": 0.00018848579248474288,
      "loss": 1.0781,
      "step": 975
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.18686963344995972,
      "learning_rate": 0.00018845797381325708,
      "loss": 1.1286,
      "step": 976
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.25061365410965286,
      "learning_rate": 0.00018843012363492325,
      "loss": 1.0294,
      "step": 977
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2060561827185824,
      "learning_rate": 0.000188402241959661,
      "loss": 1.3003,
      "step": 978
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1982249667791955,
      "learning_rate": 0.00018837432879740114,
      "loss": 1.1768,
      "step": 979
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.19569640753472303,
      "learning_rate": 0.00018834638415808584,
      "loss": 1.1682,
      "step": 980
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2102426417888484,
      "learning_rate": 0.0001883184080516683,
      "loss": 1.2813,
      "step": 981
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.19718504071066958,
      "learning_rate": 0.00018829040048811305,
      "loss": 1.2461,
      "step": 982
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2217369695256459,
      "learning_rate": 0.0001882623614773958,
      "loss": 1.1038,
      "step": 983
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.16893715242213916,
      "learning_rate": 0.00018823429102950346,
      "loss": 1.2407,
      "step": 984
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.22816878574578828,
      "learning_rate": 0.00018820618915443412,
      "loss": 1.0719,
      "step": 985
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2137554111784116,
      "learning_rate": 0.000188178055862197,
      "loss": 1.1265,
      "step": 986
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.24143280265365982,
      "learning_rate": 0.0001881498911628127,
      "loss": 1.0682,
      "step": 987
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.19755364113824087,
      "learning_rate": 0.00018812169506631277,
      "loss": 1.1166,
      "step": 988
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1930006008881704,
      "learning_rate": 0.00018809346758274012,
      "loss": 0.9653,
      "step": 989
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.18285535605832368,
      "learning_rate": 0.00018806520872214879,
      "loss": 1.1199,
      "step": 990
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.19234418569001388,
      "learning_rate": 0.00018803691849460393,
      "loss": 1.0967,
      "step": 991
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.19421597741414354,
      "learning_rate": 0.000188008596910182,
      "loss": 1.1808,
      "step": 992
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2420023547359673,
      "learning_rate": 0.00018798024397897046,
      "loss": 1.1418,
      "step": 993
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1717991374918012,
      "learning_rate": 0.00018795185971106805,
      "loss": 1.1409,
      "step": 994
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1978317899349855,
      "learning_rate": 0.00018792344411658468,
      "loss": 1.1002,
      "step": 995
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.19825402184765545,
      "learning_rate": 0.00018789499720564136,
      "loss": 1.1452,
      "step": 996
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.19730870346200027,
      "learning_rate": 0.00018786651898837023,
      "loss": 1.1941,
      "step": 997
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.18755176939951207,
      "learning_rate": 0.00018783800947491468,
      "loss": 1.0839,
      "step": 998
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2902569670642694,
      "learning_rate": 0.00018780946867542919,
      "loss": 1.1784,
      "step": 999
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.20685528286074686,
      "learning_rate": 0.00018778089660007936,
      "loss": 1.1947,
      "step": 1000
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.19782244139228697,
      "learning_rate": 0.00018775229325904193,
      "loss": 1.1653,
      "step": 1001
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.20972093386139862,
      "learning_rate": 0.00018772365866250485,
      "loss": 1.2382,
      "step": 1002
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1950492459867921,
      "learning_rate": 0.00018769499282066717,
      "loss": 1.1549,
      "step": 1003
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.18712950292162633,
      "learning_rate": 0.000187666295743739,
      "loss": 0.957,
      "step": 1004
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.19668965428035592,
      "learning_rate": 0.00018763756744194163,
      "loss": 1.1355,
      "step": 1005
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.20096601326608016,
      "learning_rate": 0.00018760880792550753,
      "loss": 1.353,
      "step": 1006
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.21297091222087727,
      "learning_rate": 0.00018758001720468016,
      "loss": 1.1636,
      "step": 1007
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.16913066610751823,
      "learning_rate": 0.0001875511952897142,
      "loss": 1.29,
      "step": 1008
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.19631220641065875,
      "learning_rate": 0.00018752234219087538,
      "loss": 1.1282,
      "step": 1009
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2472593023150154,
      "learning_rate": 0.0001874934579184406,
      "loss": 1.1022,
      "step": 1010
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2228754116521443,
      "learning_rate": 0.00018746454248269777,
      "loss": 1.0838,
      "step": 1011
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.22059737786825884,
      "learning_rate": 0.00018743559589394596,
      "loss": 1.1439,
      "step": 1012
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2014216917174859,
      "learning_rate": 0.00018740661816249538,
      "loss": 1.1658,
      "step": 1013
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.22175931017953898,
      "learning_rate": 0.00018737760929866726,
      "loss": 1.0249,
      "step": 1014
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.23767789295037564,
      "learning_rate": 0.0001873485693127939,
      "loss": 1.1309,
      "step": 1015
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2093303421724811,
      "learning_rate": 0.00018731949821521878,
      "loss": 1.0272,
      "step": 1016
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.17835355142897164,
      "learning_rate": 0.00018729039601629634,
      "loss": 1.1256,
      "step": 1017
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.25385961219443537,
      "learning_rate": 0.00018726126272639225,
      "loss": 1.0992,
      "step": 1018
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2181206956246946,
      "learning_rate": 0.0001872320983558831,
      "loss": 1.1729,
      "step": 1019
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.19092289938022972,
      "learning_rate": 0.00018720290291515665,
      "loss": 1.1388,
      "step": 1020
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.21224471695457647,
      "learning_rate": 0.00018717367641461166,
      "loss": 1.184,
      "step": 1021
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1784789766678782,
      "learning_rate": 0.00018714441886465803,
      "loss": 1.0931,
      "step": 1022
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2097885456717679,
      "learning_rate": 0.0001871151302757167,
      "loss": 1.025,
      "step": 1023
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.17910922442260957,
      "learning_rate": 0.00018708581065821954,
      "loss": 1.0759,
      "step": 1024
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.19474470922110965,
      "learning_rate": 0.00018705646002260964,
      "loss": 1.0519,
      "step": 1025
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.18985925290518954,
      "learning_rate": 0.00018702707837934111,
      "loss": 0.9182,
      "step": 1026
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1918910644837151,
      "learning_rate": 0.000186997665738879,
      "loss": 1.0142,
      "step": 1027
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2676463652803269,
      "learning_rate": 0.00018696822211169948,
      "loss": 1.004,
      "step": 1028
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.18689403749909728,
      "learning_rate": 0.00018693874750828977,
      "loss": 1.2045,
      "step": 1029
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.24577822776999822,
      "learning_rate": 0.00018690924193914807,
      "loss": 1.2038,
      "step": 1030
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2466119150351976,
      "learning_rate": 0.00018687970541478364,
      "loss": 1.1461,
      "step": 1031
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1836226209933837,
      "learning_rate": 0.0001868501379457168,
      "loss": 1.1421,
      "step": 1032
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1975095398270261,
      "learning_rate": 0.0001868205395424788,
      "loss": 1.2178,
      "step": 1033
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3066295294869232,
      "learning_rate": 0.000186790910215612,
      "loss": 1.0323,
      "step": 1034
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.20715247716108018,
      "learning_rate": 0.0001867612499756697,
      "loss": 1.014,
      "step": 1035
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.21348288932855566,
      "learning_rate": 0.00018673155883321626,
      "loss": 1.1189,
      "step": 1036
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2116783106907934,
      "learning_rate": 0.00018670183679882704,
      "loss": 1.2004,
      "step": 1037
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.19465381754525393,
      "learning_rate": 0.00018667208388308841,
      "loss": 0.9843,
      "step": 1038
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.21370629541340644,
      "learning_rate": 0.00018664230009659768,
      "loss": 1.2301,
      "step": 1039
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2254011367227908,
      "learning_rate": 0.00018661248544996325,
      "loss": 1.1584,
      "step": 1040
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.23371980476571444,
      "learning_rate": 0.00018658263995380446,
      "loss": 1.0623,
      "step": 1041
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.18076440897058937,
      "learning_rate": 0.00018655276361875163,
      "loss": 1.1989,
      "step": 1042
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.17294171076375237,
      "learning_rate": 0.00018652285645544603,
      "loss": 1.0999,
      "step": 1043
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.22291415327315972,
      "learning_rate": 0.00018649291847454,
      "loss": 1.2062,
      "step": 1044
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1834475783829089,
      "learning_rate": 0.00018646294968669685,
      "loss": 1.1239,
      "step": 1045
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.17898080867935234,
      "learning_rate": 0.00018643295010259075,
      "loss": 1.1366,
      "step": 1046
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.20959263767986588,
      "learning_rate": 0.00018640291973290695,
      "loss": 1.0879,
      "step": 1047
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2409956241930389,
      "learning_rate": 0.00018637285858834162,
      "loss": 1.0832,
      "step": 1048
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.16837588339377638,
      "learning_rate": 0.0001863427666796019,
      "loss": 1.0826,
      "step": 1049
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.21223417362726515,
      "learning_rate": 0.0001863126440174059,
      "loss": 1.1398,
      "step": 1050
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.21585678416371817,
      "learning_rate": 0.00018628249061248262,
      "loss": 0.9497,
      "step": 1051
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.20348135215023525,
      "learning_rate": 0.00018625230647557217,
      "loss": 1.1363,
      "step": 1052
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.17809453708147063,
      "learning_rate": 0.00018622209161742536,
      "loss": 1.0475,
      "step": 1053
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.23229070202307564,
      "learning_rate": 0.00018619184604880422,
      "loss": 1.0783,
      "step": 1054
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.20618053725900856,
      "learning_rate": 0.00018616156978048146,
      "loss": 1.0218,
      "step": 1055
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.222637036588439,
      "learning_rate": 0.00018613126282324092,
      "loss": 1.1602,
      "step": 1056
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2722617299359072,
      "learning_rate": 0.00018610092518787723,
      "loss": 0.9537,
      "step": 1057
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.19417338095945538,
      "learning_rate": 0.00018607055688519606,
      "loss": 0.9812,
      "step": 1058
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.16544582841784186,
      "learning_rate": 0.00018604015792601396,
      "loss": 1.0398,
      "step": 1059
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.18644070884512448,
      "learning_rate": 0.00018600972832115832,
      "loss": 1.1418,
      "step": 1060
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2352894787695743,
      "learning_rate": 0.0001859792680814676,
      "loss": 1.0803,
      "step": 1061
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.16431569356609757,
      "learning_rate": 0.00018594877721779106,
      "loss": 1.206,
      "step": 1062
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.23246813245481043,
      "learning_rate": 0.0001859182557409889,
      "loss": 1.0795,
      "step": 1063
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2589427196927336,
      "learning_rate": 0.0001858877036619322,
      "loss": 1.0564,
      "step": 1064
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.20719636710516848,
      "learning_rate": 0.000185857120991503,
      "loss": 0.993,
      "step": 1065
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.22270968444277792,
      "learning_rate": 0.0001858265077405941,
      "loss": 1.1896,
      "step": 1066
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.18926641591751742,
      "learning_rate": 0.00018579586392010943,
      "loss": 1.0802,
      "step": 1067
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1725861071881398,
      "learning_rate": 0.00018576518954096358,
      "loss": 1.0259,
      "step": 1068
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.18832909320897356,
      "learning_rate": 0.0001857344846140821,
      "loss": 0.9742,
      "step": 1069
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1941751306530333,
      "learning_rate": 0.0001857037491504015,
      "loss": 1.0376,
      "step": 1070
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.20345945643060143,
      "learning_rate": 0.00018567298316086902,
      "loss": 1.1142,
      "step": 1071
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.207456146599232,
      "learning_rate": 0.00018564218665644293,
      "loss": 1.1403,
      "step": 1072
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.18814168425572136,
      "learning_rate": 0.00018561135964809223,
      "loss": 1.2273,
      "step": 1073
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.18686652974600235,
      "learning_rate": 0.00018558050214679685,
      "loss": 1.1713,
      "step": 1074
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.18451138479905227,
      "learning_rate": 0.0001855496141635476,
      "loss": 1.0476,
      "step": 1075
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.22368183799545302,
      "learning_rate": 0.00018551869570934608,
      "loss": 1.0024,
      "step": 1076
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.18495814056225904,
      "learning_rate": 0.00018548774679520485,
      "loss": 0.9883,
      "step": 1077
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.20339654028470228,
      "learning_rate": 0.0001854567674321472,
      "loss": 1.1356,
      "step": 1078
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1916671618977106,
      "learning_rate": 0.00018542575763120739,
      "loss": 1.102,
      "step": 1079
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.19889946203491554,
      "learning_rate": 0.00018539471740343033,
      "loss": 1.1144,
      "step": 1080
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2238013368347142,
      "learning_rate": 0.00018536364675987202,
      "loss": 1.0734,
      "step": 1081
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.19922825594479285,
      "learning_rate": 0.00018533254571159903,
      "loss": 1.1001,
      "step": 1082
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.20512338013962012,
      "learning_rate": 0.00018530141426968902,
      "loss": 1.0995,
      "step": 1083
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.19018885684436956,
      "learning_rate": 0.0001852702524452303,
      "loss": 0.9615,
      "step": 1084
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.18071866764418687,
      "learning_rate": 0.00018523906024932202,
      "loss": 1.263,
      "step": 1085
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1987625725660759,
      "learning_rate": 0.0001852078376930742,
      "loss": 1.2415,
      "step": 1086
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.17761873096829026,
      "learning_rate": 0.0001851765847876076,
      "loss": 1.1041,
      "step": 1087
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.18595114240489133,
      "learning_rate": 0.00018514530154405394,
      "loss": 1.0581,
      "step": 1088
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.18384915193439685,
      "learning_rate": 0.0001851139879735556,
      "loss": 1.0253,
      "step": 1089
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2714236864101341,
      "learning_rate": 0.00018508264408726577,
      "loss": 1.1379,
      "step": 1090
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.20447471221686292,
      "learning_rate": 0.0001850512698963485,
      "loss": 1.0581,
      "step": 1091
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.21200098593110972,
      "learning_rate": 0.00018501986541197865,
      "loss": 1.0509,
      "step": 1092
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.20139296759395944,
      "learning_rate": 0.00018498843064534176,
      "loss": 1.0589,
      "step": 1093
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.18673890807055787,
      "learning_rate": 0.00018495696560763428,
      "loss": 1.2223,
      "step": 1094
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.20498431583962276,
      "learning_rate": 0.00018492547031006336,
      "loss": 1.0476,
      "step": 1095
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2087288022834574,
      "learning_rate": 0.00018489394476384696,
      "loss": 1.1975,
      "step": 1096
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1987275079157112,
      "learning_rate": 0.0001848623889802138,
      "loss": 1.0675,
      "step": 1097
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.18981680932116696,
      "learning_rate": 0.0001848308029704034,
      "loss": 1.1602,
      "step": 1098
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.15751292025693925,
      "learning_rate": 0.000184799186745666,
      "loss": 1.2354,
      "step": 1099
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1811344174826825,
      "learning_rate": 0.00018476754031726265,
      "loss": 1.1868,
      "step": 1100
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.23555558938784238,
      "learning_rate": 0.00018473586369646512,
      "loss": 1.0398,
      "step": 1101
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.19430903498051455,
      "learning_rate": 0.00018470415689455597,
      "loss": 1.1714,
      "step": 1102
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1822479342838426,
      "learning_rate": 0.00018467241992282843,
      "loss": 1.1233,
      "step": 1103
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.19605184849725055,
      "learning_rate": 0.00018464065279258658,
      "loss": 1.137,
      "step": 1104
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.20678670552687875,
      "learning_rate": 0.00018460885551514522,
      "loss": 1.137,
      "step": 1105
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.18460137834858797,
      "learning_rate": 0.0001845770281018298,
      "loss": 1.1701,
      "step": 1106
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1840772454524905,
      "learning_rate": 0.00018454517056397661,
      "loss": 0.9591,
      "step": 1107
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.18896771524998301,
      "learning_rate": 0.00018451328291293264,
      "loss": 0.9848,
      "step": 1108
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.20756794846617668,
      "learning_rate": 0.00018448136516005555,
      "loss": 1.1251,
      "step": 1109
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.18785062264803348,
      "learning_rate": 0.00018444941731671376,
      "loss": 1.0788,
      "step": 1110
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.21250428411434674,
      "learning_rate": 0.00018441743939428648,
      "loss": 1.0893,
      "step": 1111
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2239719861190515,
      "learning_rate": 0.00018438543140416352,
      "loss": 1.0951,
      "step": 1112
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2274524115081882,
      "learning_rate": 0.00018435339335774544,
      "loss": 1.0004,
      "step": 1113
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1927145702867546,
      "learning_rate": 0.0001843213252664435,
      "loss": 1.1981,
      "step": 1114
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.18634773199055438,
      "learning_rate": 0.0001842892271416797,
      "loss": 1.2866,
      "step": 1115
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.17164016765801582,
      "learning_rate": 0.00018425709899488673,
      "loss": 1.1546,
      "step": 1116
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.21342469261917063,
      "learning_rate": 0.0001842249408375079,
      "loss": 1.1731,
      "step": 1117
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.163430669338529,
      "learning_rate": 0.00018419275268099727,
      "loss": 1.1906,
      "step": 1118
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.17419950652892494,
      "learning_rate": 0.0001841605345368196,
      "loss": 1.0601,
      "step": 1119
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.193400218731286,
      "learning_rate": 0.00018412828641645027,
      "loss": 1.0508,
      "step": 1120
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.18256292443767475,
      "learning_rate": 0.0001840960083313754,
      "loss": 1.1151,
      "step": 1121
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.21676881086390898,
      "learning_rate": 0.0001840637002930918,
      "loss": 1.2297,
      "step": 1122
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.20099818765485086,
      "learning_rate": 0.00018403136231310684,
      "loss": 1.1578,
      "step": 1123
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.21208376399972623,
      "learning_rate": 0.00018399899440293865,
      "loss": 1.144,
      "step": 1124
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.18262441199483323,
      "learning_rate": 0.00018396659657411596,
      "loss": 1.1163,
      "step": 1125
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.21239351872284015,
      "learning_rate": 0.00018393416883817824,
      "loss": 1.2263,
      "step": 1126
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.18819910723657926,
      "learning_rate": 0.00018390171120667554,
      "loss": 1.0208,
      "step": 1127
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.20383310371681015,
      "learning_rate": 0.00018386922369116856,
      "loss": 1.1501,
      "step": 1128
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1874222938567548,
      "learning_rate": 0.00018383670630322865,
      "loss": 0.9801,
      "step": 1129
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.23631594691497104,
      "learning_rate": 0.00018380415905443784,
      "loss": 1.1327,
      "step": 1130
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.19823026830725587,
      "learning_rate": 0.00018377158195638876,
      "loss": 1.0031,
      "step": 1131
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2043674982882651,
      "learning_rate": 0.0001837389750206847,
      "loss": 1.1921,
      "step": 1132
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2086619396685083,
      "learning_rate": 0.00018370633825893955,
      "loss": 1.1234,
      "step": 1133
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.19011488288005862,
      "learning_rate": 0.00018367367168277779,
      "loss": 1.1154,
      "step": 1134
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.19407555932908294,
      "learning_rate": 0.00018364097530383457,
      "loss": 1.0979,
      "step": 1135
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1893009304919917,
      "learning_rate": 0.00018360824913375568,
      "loss": 1.127,
      "step": 1136
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.18895165235717815,
      "learning_rate": 0.00018357549318419745,
      "loss": 1.0937,
      "step": 1137
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2084016755129995,
      "learning_rate": 0.00018354270746682687,
      "loss": 1.0051,
      "step": 1138
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.20247332136571436,
      "learning_rate": 0.00018350989199332154,
      "loss": 1.1287,
      "step": 1139
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.19644299451522365,
      "learning_rate": 0.00018347704677536955,
      "loss": 1.0978,
      "step": 1140
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2385606999583217,
      "learning_rate": 0.00018344417182466978,
      "loss": 0.9887,
      "step": 1141
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.17398663221984204,
      "learning_rate": 0.0001834112671529315,
      "loss": 1.2397,
      "step": 1142
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2756380020393565,
      "learning_rate": 0.00018337833277187472,
      "loss": 1.0534,
      "step": 1143
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.18800841857146702,
      "learning_rate": 0.00018334536869322988,
      "loss": 1.1943,
      "step": 1144
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.17763329910241496,
      "learning_rate": 0.00018331237492873818,
      "loss": 1.1881,
      "step": 1145
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2452289946276053,
      "learning_rate": 0.00018327935149015123,
      "loss": 1.0322,
      "step": 1146
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.18563881894990295,
      "learning_rate": 0.00018324629838923132,
      "loss": 0.963,
      "step": 1147
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.18656694429763715,
      "learning_rate": 0.0001832132156377512,
      "loss": 1.1469,
      "step": 1148
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.16852591054871488,
      "learning_rate": 0.0001831801032474943,
      "loss": 1.1499,
      "step": 1149
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.17254739108502007,
      "learning_rate": 0.00018314696123025454,
      "loss": 1.0437,
      "step": 1150
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.15836265841575975,
      "learning_rate": 0.00018311378959783638,
      "loss": 1.0293,
      "step": 1151
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.19790686384741726,
      "learning_rate": 0.00018308058836205487,
      "loss": 1.0682,
      "step": 1152
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.17142733507732177,
      "learning_rate": 0.00018304735753473555,
      "loss": 1.1977,
      "step": 1153
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2594190987835753,
      "learning_rate": 0.0001830140971277146,
      "loss": 1.1281,
      "step": 1154
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.17839731662746108,
      "learning_rate": 0.00018298080715283858,
      "loss": 0.9929,
      "step": 1155
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.26423821651002954,
      "learning_rate": 0.00018294748762196473,
      "loss": 1.1442,
      "step": 1156
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.20864867327071665,
      "learning_rate": 0.00018291413854696077,
      "loss": 1.1268,
      "step": 1157
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.26142307923077057,
      "learning_rate": 0.0001828807599397049,
      "loss": 1.0843,
      "step": 1158
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.25868628223679363,
      "learning_rate": 0.00018284735181208587,
      "loss": 1.0105,
      "step": 1159
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.20604655690292736,
      "learning_rate": 0.00018281391417600295,
      "loss": 1.1425,
      "step": 1160
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1584333371771988,
      "learning_rate": 0.00018278044704336593,
      "loss": 0.9995,
      "step": 1161
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.25311116623057756,
      "learning_rate": 0.00018274695042609507,
      "loss": 1.0208,
      "step": 1162
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1624732823695372,
      "learning_rate": 0.00018271342433612113,
      "loss": 1.0404,
      "step": 1163
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.21158608094059161,
      "learning_rate": 0.00018267986878538545,
      "loss": 1.0629,
      "step": 1164
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.19047532916056906,
      "learning_rate": 0.00018264628378583976,
      "loss": 1.1312,
      "step": 1165
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.24961897355352297,
      "learning_rate": 0.00018261266934944633,
      "loss": 1.1401,
      "step": 1166
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.18015414639574395,
      "learning_rate": 0.00018257902548817794,
      "loss": 1.006,
      "step": 1167
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.19178624807348033,
      "learning_rate": 0.00018254535221401774,
      "loss": 1.009,
      "step": 1168
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1893533246690579,
      "learning_rate": 0.0001825116495389595,
      "loss": 0.9794,
      "step": 1169
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.21197713290632827,
      "learning_rate": 0.0001824779174750074,
      "loss": 1.2474,
      "step": 1170
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2979195669949457,
      "learning_rate": 0.00018244415603417603,
      "loss": 0.9765,
      "step": 1171
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2150301911158246,
      "learning_rate": 0.00018241036522849055,
      "loss": 1.0846,
      "step": 1172
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2068002142851417,
      "learning_rate": 0.0001823765450699865,
      "loss": 1.2224,
      "step": 1173
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.18151246571044846,
      "learning_rate": 0.0001823426955707099,
      "loss": 1.1379,
      "step": 1174
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.18964438598607394,
      "learning_rate": 0.00018230881674271722,
      "loss": 1.103,
      "step": 1175
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.21473088572358698,
      "learning_rate": 0.00018227490859807542,
      "loss": 1.198,
      "step": 1176
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.17737206855760917,
      "learning_rate": 0.00018224097114886178,
      "loss": 0.9906,
      "step": 1177
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.17892671975669458,
      "learning_rate": 0.00018220700440716413,
      "loss": 1.0493,
      "step": 1178
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.24886714017633207,
      "learning_rate": 0.00018217300838508073,
      "loss": 1.0844,
      "step": 1179
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.19142839978871318,
      "learning_rate": 0.00018213898309472025,
      "loss": 1.1239,
      "step": 1180
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.17090140819041472,
      "learning_rate": 0.00018210492854820172,
      "loss": 1.2546,
      "step": 1181
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.24364560609379193,
      "learning_rate": 0.00018207084475765467,
      "loss": 1.049,
      "step": 1182
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.18827003067602996,
      "learning_rate": 0.00018203673173521902,
      "loss": 1.0424,
      "step": 1183
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2031060558104765,
      "learning_rate": 0.0001820025894930451,
      "loss": 1.1619,
      "step": 1184
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.17605459263149506,
      "learning_rate": 0.00018196841804329368,
      "loss": 1.1868,
      "step": 1185
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1720687061353425,
      "learning_rate": 0.00018193421739813587,
      "loss": 1.1271,
      "step": 1186
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.17989503580633245,
      "learning_rate": 0.00018189998756975318,
      "loss": 0.9016,
      "step": 1187
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1899035956077032,
      "learning_rate": 0.0001818657285703376,
      "loss": 1.1361,
      "step": 1188
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.23380800587694436,
      "learning_rate": 0.00018183144041209145,
      "loss": 1.1394,
      "step": 1189
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.22059053640449425,
      "learning_rate": 0.00018179712310722746,
      "loss": 1.2245,
      "step": 1190
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.20611113877653472,
      "learning_rate": 0.00018176277666796868,
      "loss": 1.0429,
      "step": 1191
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2695683851417219,
      "learning_rate": 0.00018172840110654862,
      "loss": 1.0906,
      "step": 1192
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.20444161270814198,
      "learning_rate": 0.00018169399643521109,
      "loss": 1.0995,
      "step": 1193
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.20135355177782246,
      "learning_rate": 0.00018165956266621029,
      "loss": 1.131,
      "step": 1194
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.25610107747073163,
      "learning_rate": 0.00018162509981181084,
      "loss": 1.043,
      "step": 1195
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.21755146292681563,
      "learning_rate": 0.00018159060788428767,
      "loss": 1.1847,
      "step": 1196
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.18399875107766955,
      "learning_rate": 0.00018155608689592604,
      "loss": 1.1823,
      "step": 1197
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.195714319522421,
      "learning_rate": 0.0001815215368590216,
      "loss": 1.0841,
      "step": 1198
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.18488984048753326,
      "learning_rate": 0.00018148695778588033,
      "loss": 1.2161,
      "step": 1199
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2109865709175713,
      "learning_rate": 0.00018145234968881864,
      "loss": 1.0243,
      "step": 1200
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.22489828396382053,
      "learning_rate": 0.0001814177125801631,
      "loss": 1.1902,
      "step": 1201
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.18274834276926402,
      "learning_rate": 0.00018138304647225074,
      "loss": 1.1678,
      "step": 1202
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1729803437070135,
      "learning_rate": 0.0001813483513774289,
      "loss": 1.1867,
      "step": 1203
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.19133303788777714,
      "learning_rate": 0.00018131362730805522,
      "loss": 1.0443,
      "step": 1204
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.202467589661501,
      "learning_rate": 0.0001812788742764977,
      "loss": 1.2598,
      "step": 1205
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.262497814787232,
      "learning_rate": 0.00018124409229513462,
      "loss": 1.1659,
      "step": 1206
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2133483132524922,
      "learning_rate": 0.00018120928137635455,
      "loss": 1.1819,
      "step": 1207
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.18073990756953448,
      "learning_rate": 0.00018117444153255646,
      "loss": 1.0498,
      "step": 1208
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.19687152599053795,
      "learning_rate": 0.00018113957277614948,
      "loss": 1.1321,
      "step": 1209
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.18830170549535924,
      "learning_rate": 0.0001811046751195532,
      "loss": 1.0192,
      "step": 1210
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1754996571884831,
      "learning_rate": 0.00018106974857519736,
      "loss": 1.1963,
      "step": 1211
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.18467856015696452,
      "learning_rate": 0.0001810347931555221,
      "loss": 1.0698,
      "step": 1212
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.16650917355709766,
      "learning_rate": 0.00018099980887297781,
      "loss": 1.1535,
      "step": 1213
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.18827847742005788,
      "learning_rate": 0.00018096479574002507,
      "loss": 1.066,
      "step": 1214
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2022940374465093,
      "learning_rate": 0.0001809297537691349,
      "loss": 1.0258,
      "step": 1215
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2068562429343576,
      "learning_rate": 0.0001808946829727884,
      "loss": 1.2207,
      "step": 1216
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1856491141285739,
      "learning_rate": 0.00018085958336347714,
      "loss": 1.1309,
      "step": 1217
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1917501231383203,
      "learning_rate": 0.0001808244549537028,
      "loss": 1.0418,
      "step": 1218
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1881686576343932,
      "learning_rate": 0.0001807892977559774,
      "loss": 1.2113,
      "step": 1219
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1843105422044677,
      "learning_rate": 0.0001807541117828232,
      "loss": 0.9638,
      "step": 1220
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2193194846057264,
      "learning_rate": 0.0001807188970467726,
      "loss": 1.1053,
      "step": 1221
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2228248629218564,
      "learning_rate": 0.00018068365356036848,
      "loss": 1.0473,
      "step": 1222
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.19325730422914378,
      "learning_rate": 0.00018064838133616368,
      "loss": 0.9923,
      "step": 1223
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.19101367515348872,
      "learning_rate": 0.00018061308038672154,
      "loss": 1.1382,
      "step": 1224
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2069945852802502,
      "learning_rate": 0.00018057775072461544,
      "loss": 1.152,
      "step": 1225
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.20150160671046718,
      "learning_rate": 0.00018054239236242907,
      "loss": 1.0214,
      "step": 1226
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.19894106300691772,
      "learning_rate": 0.0001805070053127563,
      "loss": 1.006,
      "step": 1227
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.21118255274088896,
      "learning_rate": 0.0001804715895882013,
      "loss": 1.0977,
      "step": 1228
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.22240629848974247,
      "learning_rate": 0.00018043614520137834,
      "loss": 1.1228,
      "step": 1229
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.21670786588640592,
      "learning_rate": 0.00018040067216491197,
      "loss": 1.2211,
      "step": 1230
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.20065567558874692,
      "learning_rate": 0.00018036517049143695,
      "loss": 1.2134,
      "step": 1231
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.18199733192942724,
      "learning_rate": 0.0001803296401935982,
      "loss": 1.0404,
      "step": 1232
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.22641006616219359,
      "learning_rate": 0.0001802940812840509,
      "loss": 1.1571,
      "step": 1233
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2052661203888911,
      "learning_rate": 0.00018025849377546033,
      "loss": 1.1652,
      "step": 1234
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.19730088124045037,
      "learning_rate": 0.00018022287768050202,
      "loss": 1.116,
      "step": 1235
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.315985329540019,
      "learning_rate": 0.00018018723301186165,
      "loss": 1.0005,
      "step": 1236
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.17067745803409204,
      "learning_rate": 0.0001801515597822351,
      "loss": 1.1247,
      "step": 1237
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.26985364130539524,
      "learning_rate": 0.00018011585800432845,
      "loss": 0.8288,
      "step": 1238
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.20518712088176877,
      "learning_rate": 0.00018008012769085786,
      "loss": 1.0928,
      "step": 1239
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2022420990915898,
      "learning_rate": 0.00018004436885454977,
      "loss": 1.0576,
      "step": 1240
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.18840184322608453,
      "learning_rate": 0.0001800085815081406,
      "loss": 1.2192,
      "step": 1241
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.24873361740101027,
      "learning_rate": 0.00017997276566437717,
      "loss": 1.054,
      "step": 1242
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.21401646462660862,
      "learning_rate": 0.0001799369213360163,
      "loss": 1.0831,
      "step": 1243
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.18993812312726974,
      "learning_rate": 0.00017990104853582493,
      "loss": 1.017,
      "step": 1244
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.18692246280312377,
      "learning_rate": 0.00017986514727658017,
      "loss": 1.0482,
      "step": 1245
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.23376259413714168,
      "learning_rate": 0.00017982921757106937,
      "loss": 1.2093,
      "step": 1246
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2883867061236747,
      "learning_rate": 0.00017979325943208983,
      "loss": 1.017,
      "step": 1247
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.19118792376232221,
      "learning_rate": 0.00017975727287244914,
      "loss": 1.027,
      "step": 1248
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.22763293980163182,
      "learning_rate": 0.00017972125790496497,
      "loss": 0.975,
      "step": 1249
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.19085354154643305,
      "learning_rate": 0.000179685214542465,
      "loss": 1.1536,
      "step": 1250
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1722446971801036,
      "learning_rate": 0.00017964914279778715,
      "loss": 1.1997,
      "step": 1251
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.18313872567192535,
      "learning_rate": 0.0001796130426837794,
      "loss": 1.1069,
      "step": 1252
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2032846271745325,
      "learning_rate": 0.00017957691421329987,
      "loss": 1.0662,
      "step": 1253
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2023942250947665,
      "learning_rate": 0.00017954075739921675,
      "loss": 1.0896,
      "step": 1254
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.20854627386331295,
      "learning_rate": 0.0001795045722544083,
      "loss": 0.9648,
      "step": 1255
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.19135982493080428,
      "learning_rate": 0.00017946835879176294,
      "loss": 1.0477,
      "step": 1256
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.21182854882679625,
      "learning_rate": 0.00017943211702417907,
      "loss": 1.1647,
      "step": 1257
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2229935618282827,
      "learning_rate": 0.00017939584696456527,
      "loss": 1.2171,
      "step": 1258
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2588873638903722,
      "learning_rate": 0.00017935954862584018,
      "loss": 1.1071,
      "step": 1259
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.25045635995676824,
      "learning_rate": 0.00017932322202093249,
      "loss": 1.0259,
      "step": 1260
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.17330737822988146,
      "learning_rate": 0.00017928686716278094,
      "loss": 0.9633,
      "step": 1261
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.22238346339821247,
      "learning_rate": 0.00017925048406433437,
      "loss": 1.1039,
      "step": 1262
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.19583723961673694,
      "learning_rate": 0.0001792140727385517,
      "loss": 1.0961,
      "step": 1263
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.23489827626562457,
      "learning_rate": 0.00017917763319840182,
      "loss": 1.2576,
      "step": 1264
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.22182692843857044,
      "learning_rate": 0.00017914116545686372,
      "loss": 1.0249,
      "step": 1265
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.23506147952850573,
      "learning_rate": 0.00017910466952692647,
      "loss": 1.0449,
      "step": 1266
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.19799616171283616,
      "learning_rate": 0.0001790681454215891,
      "loss": 1.2171,
      "step": 1267
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.199574915713241,
      "learning_rate": 0.0001790315931538608,
      "loss": 1.159,
      "step": 1268
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2083332510428531,
      "learning_rate": 0.0001789950127367606,
      "loss": 1.2897,
      "step": 1269
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.22910792026786556,
      "learning_rate": 0.00017895840418331773,
      "loss": 1.2347,
      "step": 1270
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2035080655607523,
      "learning_rate": 0.00017892176750657136,
      "loss": 1.1595,
      "step": 1271
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.23428052336899255,
      "learning_rate": 0.00017888510271957074,
      "loss": 1.0946,
      "step": 1272
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.19728320160026994,
      "learning_rate": 0.000178848409835375,
      "loss": 1.1843,
      "step": 1273
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1761264857400885,
      "learning_rate": 0.00017881168886705343,
      "loss": 1.0752,
      "step": 1274
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.21117883950746433,
      "learning_rate": 0.00017877493982768527,
      "loss": 1.1692,
      "step": 1275
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.24346352293839263,
      "learning_rate": 0.0001787381627303597,
      "loss": 1.0807,
      "step": 1276
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.18040576775195208,
      "learning_rate": 0.00017870135758817592,
      "loss": 1.3328,
      "step": 1277
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.22495555966355119,
      "learning_rate": 0.00017866452441424321,
      "loss": 0.9916,
      "step": 1278
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.21728970485717292,
      "learning_rate": 0.00017862766322168075,
      "loss": 0.9926,
      "step": 1279
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2414104185267528,
      "learning_rate": 0.00017859077402361764,
      "loss": 1.0865,
      "step": 1280
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.16850679403274574,
      "learning_rate": 0.0001785538568331931,
      "loss": 1.1157,
      "step": 1281
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.24602179068764166,
      "learning_rate": 0.00017851691166355623,
      "loss": 1.0659,
      "step": 1282
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.19737430538848116,
      "learning_rate": 0.0001784799385278661,
      "loss": 1.1478,
      "step": 1283
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.20662772316025982,
      "learning_rate": 0.00017844293743929176,
      "loss": 1.0759,
      "step": 1284
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.16163282551126518,
      "learning_rate": 0.00017840590841101224,
      "loss": 1.2077,
      "step": 1285
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2192608594611785,
      "learning_rate": 0.00017836885145621644,
      "loss": 1.1452,
      "step": 1286
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.19723669850935066,
      "learning_rate": 0.0001783317665881033,
      "loss": 1.1247,
      "step": 1287
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.16716078618889207,
      "learning_rate": 0.00017829465381988163,
      "loss": 0.9455,
      "step": 1288
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1973333142578793,
      "learning_rate": 0.00017825751316477022,
      "loss": 1.0644,
      "step": 1289
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.21011699896186398,
      "learning_rate": 0.00017822034463599778,
      "loss": 1.1338,
      "step": 1290
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.19844291908590203,
      "learning_rate": 0.000178183148246803,
      "loss": 1.2365,
      "step": 1291
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.20306478817498397,
      "learning_rate": 0.00017814592401043435,
      "loss": 1.1717,
      "step": 1292
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.16680184196235273,
      "learning_rate": 0.00017810867194015037,
      "loss": 1.1966,
      "step": 1293
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.20290031063186972,
      "learning_rate": 0.00017807139204921945,
      "loss": 1.2535,
      "step": 1294
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.21984924038196615,
      "learning_rate": 0.0001780340843509199,
      "loss": 1.0433,
      "step": 1295
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.23126520445337823,
      "learning_rate": 0.00017799674885853988,
      "loss": 1.2066,
      "step": 1296
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.18895876445298393,
      "learning_rate": 0.00017795938558537753,
      "loss": 1.0864,
      "step": 1297
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.20555605388224749,
      "learning_rate": 0.00017792199454474088,
      "loss": 1.1175,
      "step": 1298
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2529640778312501,
      "learning_rate": 0.00017788457574994778,
      "loss": 1.2342,
      "step": 1299
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.23300340059129743,
      "learning_rate": 0.00017784712921432603,
      "loss": 1.0536,
      "step": 1300
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.22937279604784025,
      "learning_rate": 0.0001778096549512133,
      "loss": 1.2636,
      "step": 1301
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1929635985226523,
      "learning_rate": 0.00017777215297395707,
      "loss": 1.0914,
      "step": 1302
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1901548172078195,
      "learning_rate": 0.0001777346232959148,
      "loss": 1.2037,
      "step": 1303
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1817706028578597,
      "learning_rate": 0.0001776970659304538,
      "loss": 1.0747,
      "step": 1304
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1945117531877038,
      "learning_rate": 0.0001776594808909511,
      "loss": 1.2138,
      "step": 1305
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.21583016930330956,
      "learning_rate": 0.0001776218681907938,
      "loss": 1.0954,
      "step": 1306
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2109376089718572,
      "learning_rate": 0.00017758422784337863,
      "loss": 1.2327,
      "step": 1307
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.17201621018190055,
      "learning_rate": 0.00017754655986211234,
      "loss": 1.2223,
      "step": 1308
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.17487475204433658,
      "learning_rate": 0.00017750886426041152,
      "loss": 1.0832,
      "step": 1309
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2585159326769062,
      "learning_rate": 0.00017747114105170245,
      "loss": 1.0459,
      "step": 1310
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.19886976059517006,
      "learning_rate": 0.00017743339024942135,
      "loss": 1.1646,
      "step": 1311
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.205433958349088,
      "learning_rate": 0.0001773956118670143,
      "loss": 1.0478,
      "step": 1312
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2131635227995025,
      "learning_rate": 0.00017735780591793712,
      "loss": 1.1077,
      "step": 1313
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.17246046743579357,
      "learning_rate": 0.00017731997241565549,
      "loss": 1.0252,
      "step": 1314
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.18122346198947345,
      "learning_rate": 0.00017728211137364489,
      "loss": 1.068,
      "step": 1315
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.17374155656535425,
      "learning_rate": 0.00017724422280539063,
      "loss": 1.0881,
      "step": 1316
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.18642384424114752,
      "learning_rate": 0.00017720630672438782,
      "loss": 1.1702,
      "step": 1317
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.19990259701343557,
      "learning_rate": 0.00017716836314414137,
      "loss": 1.0129,
      "step": 1318
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.18634183029506304,
      "learning_rate": 0.00017713039207816592,
      "loss": 1.0495,
      "step": 1319
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.22749933923612708,
      "learning_rate": 0.000177092393539986,
      "loss": 1.1136,
      "step": 1320
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.22829750946526756,
      "learning_rate": 0.0001770543675431359,
      "loss": 1.1577,
      "step": 1321
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.20199576611691744,
      "learning_rate": 0.0001770163141011596,
      "loss": 1.1529,
      "step": 1322
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.201204582954191,
      "learning_rate": 0.000176978233227611,
      "loss": 1.0017,
      "step": 1323
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1797489813776261,
      "learning_rate": 0.0001769401249360537,
      "loss": 1.1287,
      "step": 1324
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.18967444585195378,
      "learning_rate": 0.00017690198924006097,
      "loss": 1.1968,
      "step": 1325
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.17845824912902744,
      "learning_rate": 0.00017686382615321602,
      "loss": 1.323,
      "step": 1326
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.20852331055391404,
      "learning_rate": 0.0001768256356891117,
      "loss": 1.0939,
      "step": 1327
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.20973483789346298,
      "learning_rate": 0.00017678741786135066,
      "loss": 1.1356,
      "step": 1328
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2495797724461576,
      "learning_rate": 0.00017674917268354522,
      "loss": 1.225,
      "step": 1329
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2060009689672564,
      "learning_rate": 0.0001767109001693176,
      "loss": 1.1324,
      "step": 1330
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.22506683857663828,
      "learning_rate": 0.00017667260033229953,
      "loss": 1.1898,
      "step": 1331
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2150224730184994,
      "learning_rate": 0.0001766342731861327,
      "loss": 0.9397,
      "step": 1332
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.17261108578743964,
      "learning_rate": 0.00017659591874446837,
      "loss": 1.0589,
      "step": 1333
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.220383564388057,
      "learning_rate": 0.0001765575370209676,
      "loss": 1.1561,
      "step": 1334
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.17439059982991842,
      "learning_rate": 0.00017651912802930112,
      "loss": 1.0741,
      "step": 1335
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.17672276141180868,
      "learning_rate": 0.00017648069178314945,
      "loss": 1.0916,
      "step": 1336
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.18289184219340002,
      "learning_rate": 0.0001764422282962027,
      "loss": 1.1989,
      "step": 1337
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.19030287104601268,
      "learning_rate": 0.00017640373758216077,
      "loss": 1.1484,
      "step": 1338
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.16984504259541516,
      "learning_rate": 0.00017636521965473323,
      "loss": 1.1472,
      "step": 1339
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.26702167660216114,
      "learning_rate": 0.00017632667452763934,
      "loss": 0.8822,
      "step": 1340
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.17296691276809845,
      "learning_rate": 0.00017628810221460806,
      "loss": 1.1123,
      "step": 1341
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.18958457126751646,
      "learning_rate": 0.00017624950272937802,
      "loss": 1.0785,
      "step": 1342
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.17267508140585927,
      "learning_rate": 0.00017621087608569758,
      "loss": 1.2146,
      "step": 1343
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.20292517452005904,
      "learning_rate": 0.0001761722222973247,
      "loss": 1.0867,
      "step": 1344
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.198128095261918,
      "learning_rate": 0.000176133541378027,
      "loss": 1.046,
      "step": 1345
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2377378544651746,
      "learning_rate": 0.00017609483334158181,
      "loss": 1.0393,
      "step": 1346
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.19403436296369667,
      "learning_rate": 0.00017605609820177617,
      "loss": 1.0791,
      "step": 1347
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2050724336390839,
      "learning_rate": 0.00017601733597240665,
      "loss": 1.1654,
      "step": 1348
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2471415222672082,
      "learning_rate": 0.0001759785466672796,
      "loss": 1.0501,
      "step": 1349
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.19284600729250403,
      "learning_rate": 0.00017593973030021086,
      "loss": 1.027,
      "step": 1350
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.19752755106262704,
      "learning_rate": 0.00017590088688502605,
      "loss": 1.0481,
      "step": 1351
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.18205146287852,
      "learning_rate": 0.00017586201643556031,
      "loss": 0.9689,
      "step": 1352
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1784913570743702,
      "learning_rate": 0.00017582311896565856,
      "loss": 1.076,
      "step": 1353
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2160588661123538,
      "learning_rate": 0.00017578419448917515,
      "loss": 1.0964,
      "step": 1354
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.20318437101277184,
      "learning_rate": 0.00017574524301997423,
      "loss": 1.0539,
      "step": 1355
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.17792385705148372,
      "learning_rate": 0.00017570626457192944,
      "loss": 1.0061,
      "step": 1356
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.17559034587194813,
      "learning_rate": 0.00017566725915892412,
      "loss": 0.9777,
      "step": 1357
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1546379895621558,
      "learning_rate": 0.00017562822679485112,
      "loss": 1.1487,
      "step": 1358
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.22281369485357053,
      "learning_rate": 0.00017558916749361295,
      "loss": 1.0605,
      "step": 1359
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2532497340586483,
      "learning_rate": 0.00017555008126912168,
      "loss": 1.0822,
      "step": 1360
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.18426734789496027,
      "learning_rate": 0.00017551096813529907,
      "loss": 1.0238,
      "step": 1361
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.23279934828305462,
      "learning_rate": 0.00017547182810607627,
      "loss": 1.1488,
      "step": 1362
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2059643420539777,
      "learning_rate": 0.00017543266119539422,
      "loss": 1.033,
      "step": 1363
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1977759602467135,
      "learning_rate": 0.00017539346741720334,
      "loss": 1.1157,
      "step": 1364
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.20102943842182427,
      "learning_rate": 0.00017535424678546355,
      "loss": 0.9854,
      "step": 1365
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1839034249136764,
      "learning_rate": 0.00017531499931414444,
      "loss": 1.1709,
      "step": 1366
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.20543915298608512,
      "learning_rate": 0.00017527572501722512,
      "loss": 1.0071,
      "step": 1367
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.19843650392960305,
      "learning_rate": 0.0001752364239086943,
      "loss": 1.1694,
      "step": 1368
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.17106859748791745,
      "learning_rate": 0.00017519709600255016,
      "loss": 1.1176,
      "step": 1369
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.19821579026478658,
      "learning_rate": 0.00017515774131280046,
      "loss": 1.0893,
      "step": 1370
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2013815340165207,
      "learning_rate": 0.00017511835985346253,
      "loss": 1.0985,
      "step": 1371
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.23619636973476701,
      "learning_rate": 0.00017507895163856315,
      "loss": 1.1344,
      "step": 1372
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.23537339304058025,
      "learning_rate": 0.00017503951668213874,
      "loss": 1.0611,
      "step": 1373
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.26011796288261796,
      "learning_rate": 0.00017500005499823519,
      "loss": 0.8503,
      "step": 1374
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.18629671963701744,
      "learning_rate": 0.0001749605666009079,
      "loss": 1.0657,
      "step": 1375
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.17701091707953903,
      "learning_rate": 0.0001749210515042218,
      "loss": 1.0206,
      "step": 1376
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1957191055206299,
      "learning_rate": 0.0001748815097222513,
      "loss": 1.1598,
      "step": 1377
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2142309388434044,
      "learning_rate": 0.00017484194126908038,
      "loss": 1.1323,
      "step": 1378
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.19848259822120987,
      "learning_rate": 0.00017480234615880247,
      "loss": 1.272,
      "step": 1379
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.18851745655366253,
      "learning_rate": 0.0001747627244055205,
      "loss": 1.1177,
      "step": 1380
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2219538500332957,
      "learning_rate": 0.00017472307602334693,
      "loss": 1.2046,
      "step": 1381
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.19567383770527763,
      "learning_rate": 0.0001746834010264036,
      "loss": 1.2709,
      "step": 1382
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.18971940255651468,
      "learning_rate": 0.00017464369942882195,
      "loss": 1.0228,
      "step": 1383
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.19977575563346725,
      "learning_rate": 0.00017460397124474284,
      "loss": 1.0874,
      "step": 1384
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.17146733947691373,
      "learning_rate": 0.00017456421648831655,
      "loss": 1.1109,
      "step": 1385
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.20109622508748523,
      "learning_rate": 0.000174524435173703,
      "loss": 1.0381,
      "step": 1386
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.20332348974454828,
      "learning_rate": 0.0001744846273150713,
      "loss": 1.0831,
      "step": 1387
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.25574345950923444,
      "learning_rate": 0.0001744447929266003,
      "loss": 1.0149,
      "step": 1388
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.17600565592454354,
      "learning_rate": 0.00017440493202247804,
      "loss": 1.0378,
      "step": 1389
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2480049566333743,
      "learning_rate": 0.00017436504461690222,
      "loss": 1.1651,
      "step": 1390
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.18493788180959905,
      "learning_rate": 0.00017432513072407983,
      "loss": 1.0778,
      "step": 1391
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.202121338581316,
      "learning_rate": 0.00017428519035822738,
      "loss": 1.0002,
      "step": 1392
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1665096412694077,
      "learning_rate": 0.00017424522353357076,
      "loss": 1.024,
      "step": 1393
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.17616614256290758,
      "learning_rate": 0.0001742052302643453,
      "loss": 1.2669,
      "step": 1394
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.20390121591580174,
      "learning_rate": 0.00017416521056479577,
      "loss": 1.073,
      "step": 1395
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1840107739423383,
      "learning_rate": 0.0001741251644491763,
      "loss": 1.1472,
      "step": 1396
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2536613907876674,
      "learning_rate": 0.00017408509193175053,
      "loss": 1.2525,
      "step": 1397
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2151706789947928,
      "learning_rate": 0.00017404499302679137,
      "loss": 0.9603,
      "step": 1398
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1798146029518046,
      "learning_rate": 0.00017400486774858122,
      "loss": 1.0151,
      "step": 1399
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.19680967419426856,
      "learning_rate": 0.00017396471611141187,
      "loss": 1.0945,
      "step": 1400
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.26977071174512585,
      "learning_rate": 0.00017392453812958446,
      "loss": 1.0922,
      "step": 1401
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2113339831014254,
      "learning_rate": 0.00017388433381740952,
      "loss": 1.1359,
      "step": 1402
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.20762723163004596,
      "learning_rate": 0.00017384410318920697,
      "loss": 1.2246,
      "step": 1403
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1993430144507201,
      "learning_rate": 0.00017380384625930616,
      "loss": 1.1605,
      "step": 1404
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.20970206485374757,
      "learning_rate": 0.0001737635630420457,
      "loss": 1.2146,
      "step": 1405
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.19487766367587894,
      "learning_rate": 0.0001737232535517736,
      "loss": 0.9914,
      "step": 1406
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1955959536283365,
      "learning_rate": 0.00017368291780284732,
      "loss": 1.1326,
      "step": 1407
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.20762711759300673,
      "learning_rate": 0.0001736425558096335,
      "loss": 1.1716,
      "step": 1408
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1604314530233634,
      "learning_rate": 0.00017360216758650828,
      "loss": 1.0562,
      "step": 1409
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.20114184232109858,
      "learning_rate": 0.00017356175314785707,
      "loss": 1.1196,
      "step": 1410
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.16961256816761291,
      "learning_rate": 0.00017352131250807467,
      "loss": 1.2885,
      "step": 1411
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.19944288789120676,
      "learning_rate": 0.00017348084568156509,
      "loss": 1.0465,
      "step": 1412
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.205860212850987,
      "learning_rate": 0.0001734403526827418,
      "loss": 0.9349,
      "step": 1413
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2244494650683235,
      "learning_rate": 0.0001733998335260276,
      "loss": 1.0955,
      "step": 1414
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.20900184178683429,
      "learning_rate": 0.00017335928822585447,
      "loss": 0.9056,
      "step": 1415
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.17342000854567585,
      "learning_rate": 0.0001733187167966638,
      "loss": 0.9929,
      "step": 1416
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.240553107625911,
      "learning_rate": 0.00017327811925290631,
      "loss": 1.0663,
      "step": 1417
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.18712081656625426,
      "learning_rate": 0.00017323749560904195,
      "loss": 1.0289,
      "step": 1418
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.25447563716877974,
      "learning_rate": 0.00017319684587954002,
      "loss": 1.2,
      "step": 1419
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.18195703720436374,
      "learning_rate": 0.00017315617007887903,
      "loss": 1.1074,
      "step": 1420
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1607411385718242,
      "learning_rate": 0.00017311546822154687,
      "loss": 0.9781,
      "step": 1421
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1552287106290681,
      "learning_rate": 0.0001730747403220407,
      "loss": 1.1322,
      "step": 1422
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2017836210601039,
      "learning_rate": 0.00017303398639486695,
      "loss": 1.143,
      "step": 1423
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1921409851282997,
      "learning_rate": 0.00017299320645454124,
      "loss": 1.0925,
      "step": 1424
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1829064470082964,
      "learning_rate": 0.00017295240051558853,
      "loss": 1.1818,
      "step": 1425
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.19235244742192903,
      "learning_rate": 0.00017291156859254302,
      "loss": 1.2848,
      "step": 1426
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.19169166117077643,
      "learning_rate": 0.0001728707106999482,
      "loss": 1.2028,
      "step": 1427
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.21881085781048698,
      "learning_rate": 0.0001728298268523568,
      "loss": 1.1743,
      "step": 1428
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1877551879381556,
      "learning_rate": 0.00017278891706433072,
      "loss": 1.0584,
      "step": 1429
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.22494487750367367,
      "learning_rate": 0.00017274798135044118,
      "loss": 1.1244,
      "step": 1430
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.18861961042353653,
      "learning_rate": 0.00017270701972526862,
      "loss": 1.0078,
      "step": 1431
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.18407682078094487,
      "learning_rate": 0.0001726660322034027,
      "loss": 1.0752,
      "step": 1432
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.18591757163126862,
      "learning_rate": 0.00017262501879944232,
      "loss": 1.1029,
      "step": 1433
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.20622343747777308,
      "learning_rate": 0.00017258397952799555,
      "loss": 1.0509,
      "step": 1434
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.18077283678324976,
      "learning_rate": 0.00017254291440367968,
      "loss": 1.1127,
      "step": 1435
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2267517177985081,
      "learning_rate": 0.0001725018234411213,
      "loss": 0.9374,
      "step": 1436
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1896511687988133,
      "learning_rate": 0.00017246070665495608,
      "loss": 1.0706,
      "step": 1437
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.21003591179007672,
      "learning_rate": 0.000172419564059829,
      "loss": 1.1481,
      "step": 1438
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.18490696165391635,
      "learning_rate": 0.00017237839567039412,
      "loss": 1.087,
      "step": 1439
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.18617903077986905,
      "learning_rate": 0.0001723372015013148,
      "loss": 1.1051,
      "step": 1440
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2185877162071503,
      "learning_rate": 0.00017229598156726347,
      "loss": 1.1507,
      "step": 1441
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1888047248274932,
      "learning_rate": 0.0001722547358829218,
      "loss": 1.0914,
      "step": 1442
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1874869672584748,
      "learning_rate": 0.0001722134644629807,
      "loss": 0.9193,
      "step": 1443
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.17252404913535743,
      "learning_rate": 0.0001721721673221401,
      "loss": 1.1879,
      "step": 1444
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.19859023316474889,
      "learning_rate": 0.00017213084447510916,
      "loss": 1.0676,
      "step": 1445
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1803478844557159,
      "learning_rate": 0.0001720894959366062,
      "loss": 1.0782,
      "step": 1446
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1927601804866452,
      "learning_rate": 0.00017204812172135874,
      "loss": 0.8615,
      "step": 1447
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1824885578753956,
      "learning_rate": 0.00017200672184410336,
      "loss": 1.0204,
      "step": 1448
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.18269157569970249,
      "learning_rate": 0.0001719652963195858,
      "loss": 1.2365,
      "step": 1449
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.21301090932749303,
      "learning_rate": 0.00017192384516256102,
      "loss": 1.0308,
      "step": 1450
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.22128645249040277,
      "learning_rate": 0.00017188236838779295,
      "loss": 1.1486,
      "step": 1451
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2192351509035511,
      "learning_rate": 0.00017184086601005477,
      "loss": 1.0552,
      "step": 1452
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.17760855252011354,
      "learning_rate": 0.00017179933804412875,
      "loss": 1.0648,
      "step": 1453
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.20467551982889898,
      "learning_rate": 0.00017175778450480625,
      "loss": 1.0358,
      "step": 1454
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2584404502929534,
      "learning_rate": 0.00017171620540688782,
      "loss": 1.0546,
      "step": 1455
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.18797546793189354,
      "learning_rate": 0.00017167460076518299,
      "loss": 1.0487,
      "step": 1456
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.23637393020319272,
      "learning_rate": 0.00017163297059451043,
      "loss": 1.1412,
      "step": 1457
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.17987270669170616,
      "learning_rate": 0.00017159131490969797,
      "loss": 1.0747,
      "step": 1458
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.20993108966749982,
      "learning_rate": 0.00017154963372558246,
      "loss": 1.0914,
      "step": 1459
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1984366687325768,
      "learning_rate": 0.00017150792705700982,
      "loss": 1.0502,
      "step": 1460
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1643809971930811,
      "learning_rate": 0.00017146619491883514,
      "loss": 1.0452,
      "step": 1461
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1786999938587481,
      "learning_rate": 0.00017142443732592248,
      "loss": 0.9935,
      "step": 1462
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1837287326242118,
      "learning_rate": 0.000171382654293145,
      "loss": 0.8825,
      "step": 1463
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.18813348481280526,
      "learning_rate": 0.000171340845835385,
      "loss": 0.9621,
      "step": 1464
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.24278831987840852,
      "learning_rate": 0.00017129901196753363,
      "loss": 1.0329,
      "step": 1465
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1770410361479277,
      "learning_rate": 0.00017125715270449138,
      "loss": 1.0872,
      "step": 1466
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.19600797827445007,
      "learning_rate": 0.00017121526806116748,
      "loss": 1.1718,
      "step": 1467
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.15303098213466199,
      "learning_rate": 0.0001711733580524804,
      "loss": 1.0328,
      "step": 1468
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.19322962465910712,
      "learning_rate": 0.00017113142269335768,
      "loss": 1.1533,
      "step": 1469
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.22943385989425258,
      "learning_rate": 0.00017108946199873568,
      "loss": 1.1189,
      "step": 1470
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.17548100568200764,
      "learning_rate": 0.00017104747598355996,
      "loss": 1.1242,
      "step": 1471
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2012976032782674,
      "learning_rate": 0.00017100546466278504,
      "loss": 0.7114,
      "step": 1472
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.21325104944066284,
      "learning_rate": 0.00017096342805137448,
      "loss": 1.0513,
      "step": 1473
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.20447568592127294,
      "learning_rate": 0.00017092136616430082,
      "loss": 1.2142,
      "step": 1474
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.21696177763650926,
      "learning_rate": 0.00017087927901654557,
      "loss": 1.0677,
      "step": 1475
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.20911601782094433,
      "learning_rate": 0.00017083716662309931,
      "loss": 1.0145,
      "step": 1476
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.20422347561113147,
      "learning_rate": 0.00017079502899896158,
      "loss": 1.1706,
      "step": 1477
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.22204665280977814,
      "learning_rate": 0.00017075286615914087,
      "loss": 1.1057,
      "step": 1478
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.18189740604272864,
      "learning_rate": 0.00017071067811865476,
      "loss": 0.9241,
      "step": 1479
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.19511430755177303,
      "learning_rate": 0.00017066846489252968,
      "loss": 1.2317,
      "step": 1480
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.23408313598538294,
      "learning_rate": 0.00017062622649580104,
      "loss": 1.0503,
      "step": 1481
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1668733260134105,
      "learning_rate": 0.00017058396294351334,
      "loss": 1.2693,
      "step": 1482
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.19720659983657401,
      "learning_rate": 0.00017054167425071995,
      "loss": 1.2197,
      "step": 1483
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2180925962551316,
      "learning_rate": 0.00017049936043248316,
      "loss": 1.0841,
      "step": 1484
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.18378341928085476,
      "learning_rate": 0.00017045702150387428,
      "loss": 0.9912,
      "step": 1485
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.17874376013421517,
      "learning_rate": 0.00017041465747997349,
      "loss": 1.1904,
      "step": 1486
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.24292377895430678,
      "learning_rate": 0.00017037226837587002,
      "loss": 1.215,
      "step": 1487
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.21479703380917398,
      "learning_rate": 0.0001703298542066619,
      "loss": 1.1107,
      "step": 1488
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.21874232310860603,
      "learning_rate": 0.00017028741498745621,
      "loss": 1.0522,
      "step": 1489
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.22018170033040682,
      "learning_rate": 0.0001702449507333689,
      "loss": 0.9989,
      "step": 1490
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.19176709321680305,
      "learning_rate": 0.00017020246145952477,
      "loss": 1.178,
      "step": 1491
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.19562551055040567,
      "learning_rate": 0.00017015994718105768,
      "loss": 0.9499,
      "step": 1492
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.19764743099485402,
      "learning_rate": 0.00017011740791311026,
      "loss": 0.9287,
      "step": 1493
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.17955934789548353,
      "learning_rate": 0.00017007484367083408,
      "loss": 1.0599,
      "step": 1494
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.19732776703649876,
      "learning_rate": 0.00017003225446938965,
      "loss": 1.1236,
      "step": 1495
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.17488631497380652,
      "learning_rate": 0.00016998964032394632,
      "loss": 1.2152,
      "step": 1496
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.18214077552847988,
      "learning_rate": 0.00016994700124968234,
      "loss": 1.2255,
      "step": 1497
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.20065680545725417,
      "learning_rate": 0.00016990433726178487,
      "loss": 1.0436,
      "step": 1498
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.23259764815254957,
      "learning_rate": 0.00016986164837544987,
      "loss": 1.1287,
      "step": 1499
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.21788475026139542,
      "learning_rate": 0.00016981893460588223,
      "loss": 1.0841,
      "step": 1500
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2392697191051156,
      "learning_rate": 0.0001697761959682957,
      "loss": 0.958,
      "step": 1501
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.21349588489284793,
      "learning_rate": 0.00016973343247791287,
      "loss": 1.2637,
      "step": 1502
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.18547288794643638,
      "learning_rate": 0.00016969064414996518,
      "loss": 1.1899,
      "step": 1503
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.23438588331579382,
      "learning_rate": 0.0001696478309996929,
      "loss": 1.1632,
      "step": 1504
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2180008704003826,
      "learning_rate": 0.00016960499304234522,
      "loss": 1.1859,
      "step": 1505
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.185950021166035,
      "learning_rate": 0.00016956213029318005,
      "loss": 1.212,
      "step": 1506
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.23082196993902193,
      "learning_rate": 0.00016951924276746425,
      "loss": 1.0485,
      "step": 1507
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2577667548186816,
      "learning_rate": 0.00016947633048047337,
      "loss": 0.9819,
      "step": 1508
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1945804283833549,
      "learning_rate": 0.0001694333934474919,
      "loss": 1.1307,
      "step": 1509
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.239088697785574,
      "learning_rate": 0.00016939043168381307,
      "loss": 0.9738,
      "step": 1510
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.18766859880749145,
      "learning_rate": 0.000169347445204739,
      "loss": 1.0899,
      "step": 1511
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.18955019977604168,
      "learning_rate": 0.00016930443402558048,
      "loss": 1.0965,
      "step": 1512
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.18513158591561235,
      "learning_rate": 0.00016926139816165727,
      "loss": 1.2336,
      "step": 1513
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2093118822313966,
      "learning_rate": 0.00016921833762829772,
      "loss": 1.1038,
      "step": 1514
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.19882818807513464,
      "learning_rate": 0.00016917525244083918,
      "loss": 1.0643,
      "step": 1515
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.20394291041791446,
      "learning_rate": 0.00016913214261462763,
      "loss": 1.1229,
      "step": 1516
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.19649532094218422,
      "learning_rate": 0.00016908900816501788,
      "loss": 1.2441,
      "step": 1517
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.19137866123010386,
      "learning_rate": 0.00016904584910737346,
      "loss": 1.0843,
      "step": 1518
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.16427267214417818,
      "learning_rate": 0.00016900266545706678,
      "loss": 1.0515,
      "step": 1519
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2158571953087659,
      "learning_rate": 0.00016895945722947888,
      "loss": 1.1202,
      "step": 1520
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.20102864582851243,
      "learning_rate": 0.00016891622443999965,
      "loss": 1.1404,
      "step": 1521
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.21663082617251397,
      "learning_rate": 0.00016887296710402767,
      "loss": 1.1389,
      "step": 1522
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1841352260092895,
      "learning_rate": 0.00016882968523697028,
      "loss": 1.0273,
      "step": 1523
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.19098039349869936,
      "learning_rate": 0.00016878637885424358,
      "loss": 1.2024,
      "step": 1524
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2109416062715069,
      "learning_rate": 0.00016874304797127236,
      "loss": 1.1465,
      "step": 1525
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2102048768537668,
      "learning_rate": 0.00016869969260349018,
      "loss": 1.0448,
      "step": 1526
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.24564933419548438,
      "learning_rate": 0.0001686563127663393,
      "loss": 1.0518,
      "step": 1527
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.17439148821414585,
      "learning_rate": 0.00016861290847527066,
      "loss": 1.0786,
      "step": 1528
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.19175501165235917,
      "learning_rate": 0.00016856947974574395,
      "loss": 1.1266,
      "step": 1529
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.21414342758442,
      "learning_rate": 0.00016852602659322757,
      "loss": 1.2101,
      "step": 1530
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.18419625656599814,
      "learning_rate": 0.00016848254903319867,
      "loss": 1.2041,
      "step": 1531
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.20834894626410402,
      "learning_rate": 0.0001684390470811429,
      "loss": 1.18,
      "step": 1532
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.22149203064517162,
      "learning_rate": 0.0001683955207525549,
      "loss": 1.1371,
      "step": 1533
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.19192150034612085,
      "learning_rate": 0.00016835197006293763,
      "loss": 1.1449,
      "step": 1534
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.19152701765139854,
      "learning_rate": 0.0001683083950278031,
      "loss": 1.0763,
      "step": 1535
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.16896200841767367,
      "learning_rate": 0.00016826479566267168,
      "loss": 1.1147,
      "step": 1536
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1775809827374719,
      "learning_rate": 0.0001682211719830726,
      "loss": 0.9867,
      "step": 1537
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.17301380471981861,
      "learning_rate": 0.00016817752400454368,
      "loss": 1.1248,
      "step": 1538
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.17284721468982342,
      "learning_rate": 0.00016813385174263137,
      "loss": 1.2102,
      "step": 1539
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.18084303181280037,
      "learning_rate": 0.00016809015521289085,
      "loss": 1.2305,
      "step": 1540
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.18616281402046464,
      "learning_rate": 0.0001680464344308858,
      "loss": 1.109,
      "step": 1541
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.16749114781420424,
      "learning_rate": 0.00016800268941218876,
      "loss": 1.1251,
      "step": 1542
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1858857794740941,
      "learning_rate": 0.0001679589201723807,
      "loss": 1.1692,
      "step": 1543
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.18260676433592044,
      "learning_rate": 0.0001679151267270513,
      "loss": 1.1329,
      "step": 1544
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.21654021799864293,
      "learning_rate": 0.00016787130909179883,
      "loss": 1.0661,
      "step": 1545
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.21386025055337154,
      "learning_rate": 0.00016782746728223028,
      "loss": 1.1001,
      "step": 1546
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1758480096900126,
      "learning_rate": 0.0001677836013139611,
      "loss": 1.142,
      "step": 1547
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.20258785606353077,
      "learning_rate": 0.00016773971120261544,
      "loss": 1.133,
      "step": 1548
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.21016369593585674,
      "learning_rate": 0.00016769579696382597,
      "loss": 0.9542,
      "step": 1549
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.15763530479448093,
      "learning_rate": 0.0001676518586132341,
      "loss": 1.134,
      "step": 1550
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.18800574782544638,
      "learning_rate": 0.00016760789616648966,
      "loss": 1.0824,
      "step": 1551
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.21511289266032793,
      "learning_rate": 0.0001675639096392512,
      "loss": 1.0546,
      "step": 1552
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.22238728324753626,
      "learning_rate": 0.00016751989904718572,
      "loss": 1.072,
      "step": 1553
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.187278524049031,
      "learning_rate": 0.0001674758644059689,
      "loss": 1.0242,
      "step": 1554
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.15361731801823234,
      "learning_rate": 0.00016743180573128495,
      "loss": 1.1272,
      "step": 1555
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.18258202679751664,
      "learning_rate": 0.00016738772303882658,
      "loss": 1.2289,
      "step": 1556
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1634986742454566,
      "learning_rate": 0.00016734361634429515,
      "loss": 1.0942,
      "step": 1557
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.24226132051990726,
      "learning_rate": 0.0001672994856634005,
      "loss": 1.1374,
      "step": 1558
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.21666501569813781,
      "learning_rate": 0.0001672553310118611,
      "loss": 1.0894,
      "step": 1559
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.20622988126137204,
      "learning_rate": 0.00016721115240540384,
      "loss": 1.053,
      "step": 1560
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2452294120460336,
      "learning_rate": 0.0001671669498597642,
      "loss": 1.0298,
      "step": 1561
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3562153699138546,
      "learning_rate": 0.0001671227233906862,
      "loss": 1.1215,
      "step": 1562
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.18575396053797735,
      "learning_rate": 0.00016707847301392236,
      "loss": 1.0673,
      "step": 1563
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.17137457500825062,
      "learning_rate": 0.00016703419874523374,
      "loss": 1.0213,
      "step": 1564
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.17310229629899773,
      "learning_rate": 0.00016698990060038985,
      "loss": 1.026,
      "step": 1565
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.20057575297480884,
      "learning_rate": 0.00016694557859516881,
      "loss": 0.897,
      "step": 1566
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.19346200671539074,
      "learning_rate": 0.00016690123274535715,
      "loss": 1.2329,
      "step": 1567
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2143704844536373,
      "learning_rate": 0.00016685686306674988,
      "loss": 1.1245,
      "step": 1568
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.18291346928747126,
      "learning_rate": 0.0001668124695751506,
      "loss": 1.2155,
      "step": 1569
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.17581429020091283,
      "learning_rate": 0.00016676805228637128,
      "loss": 1.035,
      "step": 1570
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2066673645607881,
      "learning_rate": 0.00016672361121623238,
      "loss": 1.0291,
      "step": 1571
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2745396182185379,
      "learning_rate": 0.00016667914638056298,
      "loss": 1.0732,
      "step": 1572
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.18368004234969912,
      "learning_rate": 0.0001666346577952004,
      "loss": 1.1442,
      "step": 1573
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2029438147510384,
      "learning_rate": 0.0001665901454759906,
      "loss": 0.9413,
      "step": 1574
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.18918770932288398,
      "learning_rate": 0.00016654560943878783,
      "loss": 1.1468,
      "step": 1575
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.19726268379434211,
      "learning_rate": 0.00016650104969945497,
      "loss": 1.0626,
      "step": 1576
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.19735615664376224,
      "learning_rate": 0.0001664564662738632,
      "loss": 0.9685,
      "step": 1577
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.23070912523926293,
      "learning_rate": 0.00016641185917789226,
      "loss": 1.0377,
      "step": 1578
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.23349246324327133,
      "learning_rate": 0.00016636722842743013,
      "loss": 1.061,
      "step": 1579
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.19614366783856524,
      "learning_rate": 0.00016632257403837345,
      "loss": 1.0909,
      "step": 1580
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.20837016996714738,
      "learning_rate": 0.00016627789602662708,
      "loss": 1.0613,
      "step": 1581
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1705934368906592,
      "learning_rate": 0.0001662331944081044,
      "loss": 0.9408,
      "step": 1582
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.21152612960317232,
      "learning_rate": 0.00016618846919872718,
      "loss": 1.0922,
      "step": 1583
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.24478840606437027,
      "learning_rate": 0.0001661437204144256,
      "loss": 1.1413,
      "step": 1584
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.17740359017320742,
      "learning_rate": 0.00016609894807113822,
      "loss": 1.162,
      "step": 1585
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.20328836303839629,
      "learning_rate": 0.00016605415218481199,
      "loss": 1.0876,
      "step": 1586
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.15698367907397126,
      "learning_rate": 0.0001660093327714022,
      "loss": 0.9946,
      "step": 1587
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.23596894198734558,
      "learning_rate": 0.00016596448984687267,
      "loss": 1.0431,
      "step": 1588
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2038352919681052,
      "learning_rate": 0.00016591962342719544,
      "loss": 1.0503,
      "step": 1589
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1680030722637562,
      "learning_rate": 0.000165874733528351,
      "loss": 1.2673,
      "step": 1590
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.22463934390805584,
      "learning_rate": 0.00016582982016632818,
      "loss": 0.934,
      "step": 1591
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.24761549951580572,
      "learning_rate": 0.00016578488335712409,
      "loss": 1.1276,
      "step": 1592
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.19653684808260985,
      "learning_rate": 0.00016573992311674437,
      "loss": 1.1345,
      "step": 1593
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.17612150716422817,
      "learning_rate": 0.00016569493946120282,
      "loss": 1.0125,
      "step": 1594
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1814334363129602,
      "learning_rate": 0.0001656499324065217,
      "loss": 1.1374,
      "step": 1595
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.20799549391098196,
      "learning_rate": 0.00016560490196873155,
      "loss": 0.9831,
      "step": 1596
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2045546281268181,
      "learning_rate": 0.00016555984816387124,
      "loss": 1.0707,
      "step": 1597
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1977454427656333,
      "learning_rate": 0.00016551477100798805,
      "loss": 1.1885,
      "step": 1598
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.17193744341409142,
      "learning_rate": 0.00016546967051713743,
      "loss": 1.0364,
      "step": 1599
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.17760983558453627,
      "learning_rate": 0.0001654245467073832,
      "loss": 1.2138,
      "step": 1600
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.16849161863897164,
      "learning_rate": 0.00016537939959479755,
      "loss": 1.1289,
      "step": 1601
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.29022448542449686,
      "learning_rate": 0.00016533422919546092,
      "loss": 1.0722,
      "step": 1602
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1731439546300329,
      "learning_rate": 0.00016528903552546207,
      "loss": 1.0762,
      "step": 1603
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.22018356233771263,
      "learning_rate": 0.00016524381860089795,
      "loss": 0.9259,
      "step": 1604
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.18744673927958075,
      "learning_rate": 0.00016519857843787388,
      "loss": 1.2033,
      "step": 1605
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.22278070685188125,
      "learning_rate": 0.00016515331505250348,
      "loss": 1.1452,
      "step": 1606
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.19369642435790632,
      "learning_rate": 0.00016510802846090862,
      "loss": 1.203,
      "step": 1607
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.18348837963165082,
      "learning_rate": 0.00016506271867921935,
      "loss": 1.0682,
      "step": 1608
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.21047713017584754,
      "learning_rate": 0.00016501738572357412,
      "loss": 1.0936,
      "step": 1609
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1766998262077258,
      "learning_rate": 0.00016497202961011952,
      "loss": 1.0832,
      "step": 1610
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2088666178023322,
      "learning_rate": 0.00016492665035501046,
      "loss": 1.0225,
      "step": 1611
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.14410420364127122,
      "learning_rate": 0.00016488124797441004,
      "loss": 1.1019,
      "step": 1612
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.222127684041611,
      "learning_rate": 0.00016483582248448964,
      "loss": 1.0446,
      "step": 1613
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2057970068117427,
      "learning_rate": 0.0001647903739014289,
      "loss": 1.1419,
      "step": 1614
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.23106469208876815,
      "learning_rate": 0.0001647449022414155,
      "loss": 1.1271,
      "step": 1615
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.19081499292694315,
      "learning_rate": 0.00016469940752064562,
      "loss": 1.0741,
      "step": 1616
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.17497489495003402,
      "learning_rate": 0.00016465388975532344,
      "loss": 1.1581,
      "step": 1617
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2181856270423759,
      "learning_rate": 0.0001646083489616614,
      "loss": 1.1143,
      "step": 1618
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.18437776332021352,
      "learning_rate": 0.00016456278515588024,
      "loss": 1.0595,
      "step": 1619
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.19767987774849632,
      "learning_rate": 0.00016451719835420877,
      "loss": 1.1102,
      "step": 1620
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.17644272388464513,
      "learning_rate": 0.00016447158857288403,
      "loss": 1.1351,
      "step": 1621
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.17442664088595564,
      "learning_rate": 0.00016442595582815127,
      "loss": 1.084,
      "step": 1622
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.19681294836663418,
      "learning_rate": 0.00016438030013626393,
      "loss": 1.0437,
      "step": 1623
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.20555982685797422,
      "learning_rate": 0.00016433462151348353,
      "loss": 1.1078,
      "step": 1624
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.20171771842535946,
      "learning_rate": 0.00016428891997607986,
      "loss": 1.0917,
      "step": 1625
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1915072582838596,
      "learning_rate": 0.00016424319554033084,
      "loss": 0.9974,
      "step": 1626
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.19238180484047998,
      "learning_rate": 0.00016419744822252253,
      "loss": 1.0752,
      "step": 1627
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.21298953102484977,
      "learning_rate": 0.0001641516780389492,
      "loss": 1.1721,
      "step": 1628
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.24199564936950413,
      "learning_rate": 0.00016410588500591312,
      "loss": 1.1287,
      "step": 1629
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.18261007682666922,
      "learning_rate": 0.00016406006913972484,
      "loss": 1.0294,
      "step": 1630
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.22642671762182778,
      "learning_rate": 0.00016401423045670304,
      "loss": 1.0694,
      "step": 1631
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.24190012230412575,
      "learning_rate": 0.00016396836897317446,
      "loss": 1.1028,
      "step": 1632
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.24632449500009815,
      "learning_rate": 0.00016392248470547394,
      "loss": 1.0764,
      "step": 1633
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1997759115115665,
      "learning_rate": 0.00016387657766994451,
      "loss": 1.0334,
      "step": 1634
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.20299246302100488,
      "learning_rate": 0.0001638306478829373,
      "loss": 1.0715,
      "step": 1635
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.18415347608167387,
      "learning_rate": 0.00016378469536081148,
      "loss": 1.0378,
      "step": 1636
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.18801457583891187,
      "learning_rate": 0.00016373872011993438,
      "loss": 1.0431,
      "step": 1637
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.22667590428037881,
      "learning_rate": 0.00016369272217668144,
      "loss": 1.0257,
      "step": 1638
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.18405933023714238,
      "learning_rate": 0.00016364670154743607,
      "loss": 1.0897,
      "step": 1639
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.204220972801132,
      "learning_rate": 0.00016360065824858993,
      "loss": 1.1884,
      "step": 1640
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1688239479890036,
      "learning_rate": 0.00016355459229654257,
      "loss": 0.9212,
      "step": 1641
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.17650191850693778,
      "learning_rate": 0.00016350850370770178,
      "loss": 1.1026,
      "step": 1642
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2272620370938987,
      "learning_rate": 0.0001634623924984833,
      "loss": 1.0629,
      "step": 1643
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.19279532138505412,
      "learning_rate": 0.00016341625868531096,
      "loss": 1.0562,
      "step": 1644
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.24147408997279846,
      "learning_rate": 0.00016337010228461668,
      "loss": 1.2092,
      "step": 1645
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.19481819546294576,
      "learning_rate": 0.0001633239233128403,
      "loss": 1.3011,
      "step": 1646
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1943266386011017,
      "learning_rate": 0.00016327772178642986,
      "loss": 1.0289,
      "step": 1647
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2058601537630965,
      "learning_rate": 0.00016323149772184134,
      "loss": 1.0327,
      "step": 1648
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1757551683171292,
      "learning_rate": 0.0001631852511355388,
      "loss": 1.1788,
      "step": 1649
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.19155107700361582,
      "learning_rate": 0.00016313898204399423,
      "loss": 1.2208,
      "step": 1650
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.19539150329855856,
      "learning_rate": 0.00016309269046368776,
      "loss": 1.0003,
      "step": 1651
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.19827089254253186,
      "learning_rate": 0.0001630463764111074,
      "loss": 1.0289,
      "step": 1652
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.17329718615820294,
      "learning_rate": 0.00016300003990274927,
      "loss": 1.1131,
      "step": 1653
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.22419158495020386,
      "learning_rate": 0.00016295368095511746,
      "loss": 1.1165,
      "step": 1654
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.20836494633852226,
      "learning_rate": 0.000162907299584724,
      "loss": 1.1592,
      "step": 1655
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.503677340111375,
      "learning_rate": 0.000162860895808089,
      "loss": 1.1974,
      "step": 1656
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2748874222767157,
      "learning_rate": 0.00016281446964174046,
      "loss": 1.0332,
      "step": 1657
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.18422062973684725,
      "learning_rate": 0.0001627680211022144,
      "loss": 1.1177,
      "step": 1658
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1818596828646734,
      "learning_rate": 0.0001627215502060548,
      "loss": 0.9859,
      "step": 1659
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.24631886638303516,
      "learning_rate": 0.0001626750569698136,
      "loss": 1.0792,
      "step": 1660
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1839192066523034,
      "learning_rate": 0.0001626285414100507,
      "loss": 1.0229,
      "step": 1661
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2781527869407188,
      "learning_rate": 0.00016258200354333396,
      "loss": 1.0798,
      "step": 1662
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.178037767222137,
      "learning_rate": 0.00016253544338623916,
      "loss": 1.1761,
      "step": 1663
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.20504495822634144,
      "learning_rate": 0.0001624888609553501,
      "loss": 1.237,
      "step": 1664
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.22010396493217707,
      "learning_rate": 0.00016244225626725836,
      "loss": 1.0898,
      "step": 1665
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2095879843695362,
      "learning_rate": 0.00016239562933856365,
      "loss": 0.9977,
      "step": 1666
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1815593142139667,
      "learning_rate": 0.00016234898018587337,
      "loss": 1.0974,
      "step": 1667
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.23442825248859386,
      "learning_rate": 0.00016230230882580303,
      "loss": 1.1322,
      "step": 1668
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.19379851479178067,
      "learning_rate": 0.00016225561527497594,
      "loss": 0.9703,
      "step": 1669
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1936605434224726,
      "learning_rate": 0.00016220889955002337,
      "loss": 1.0364,
      "step": 1670
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.17623798588743833,
      "learning_rate": 0.00016216216166758447,
      "loss": 1.2825,
      "step": 1671
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.19053317900050862,
      "learning_rate": 0.00016211540164430625,
      "loss": 1.0478,
      "step": 1672
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1504633944522574,
      "learning_rate": 0.00016206861949684366,
      "loss": 1.0967,
      "step": 1673
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.17865103744910196,
      "learning_rate": 0.00016202181524185952,
      "loss": 1.1588,
      "step": 1674
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.24069067313629003,
      "learning_rate": 0.00016197498889602448,
      "loss": 1.0517,
      "step": 1675
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.17811337691961912,
      "learning_rate": 0.00016192814047601712,
      "loss": 1.1453,
      "step": 1676
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.22872350490533025,
      "learning_rate": 0.00016188126999852381,
      "loss": 1.0732,
      "step": 1677
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.17791544811482804,
      "learning_rate": 0.00016183437748023885,
      "loss": 1.0599,
      "step": 1678
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1755715426943051,
      "learning_rate": 0.00016178746293786433,
      "loss": 0.9545,
      "step": 1679
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2101147551384495,
      "learning_rate": 0.00016174052638811024,
      "loss": 1.2556,
      "step": 1680
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.20275995600977176,
      "learning_rate": 0.00016169356784769435,
      "loss": 0.9592,
      "step": 1681
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2182803402136901,
      "learning_rate": 0.0001616465873333423,
      "loss": 1.0389,
      "step": 1682
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.19932378662688313,
      "learning_rate": 0.0001615995848617876,
      "loss": 1.3032,
      "step": 1683
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.22833739182005938,
      "learning_rate": 0.00016155256044977145,
      "loss": 1.0693,
      "step": 1684
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.20528664508977765,
      "learning_rate": 0.000161505514114043,
      "loss": 0.9935,
      "step": 1685
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.165584827713049,
      "learning_rate": 0.00016145844587135913,
      "loss": 1.0356,
      "step": 1686
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.18571389612749292,
      "learning_rate": 0.00016141135573848457,
      "loss": 1.0891,
      "step": 1687
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.20473585341890022,
      "learning_rate": 0.0001613642437321918,
      "loss": 1.1939,
      "step": 1688
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.18493784372762254,
      "learning_rate": 0.00016131710986926108,
      "loss": 1.0977,
      "step": 1689
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.18538653640446107,
      "learning_rate": 0.00016126995416648057,
      "loss": 1.1579,
      "step": 1690
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2205535828819987,
      "learning_rate": 0.0001612227766406461,
      "loss": 1.1104,
      "step": 1691
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.23456116756434406,
      "learning_rate": 0.00016117557730856125,
      "loss": 1.0937,
      "step": 1692
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.20397159869044793,
      "learning_rate": 0.00016112835618703746,
      "loss": 1.2707,
      "step": 1693
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.17547430493883961,
      "learning_rate": 0.0001610811132928939,
      "loss": 1.1116,
      "step": 1694
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1809968195833916,
      "learning_rate": 0.0001610338486429575,
      "loss": 1.0071,
      "step": 1695
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1721881478595582,
      "learning_rate": 0.00016098656225406287,
      "loss": 1.0196,
      "step": 1696
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.17747910810622755,
      "learning_rate": 0.0001609392541430524,
      "loss": 1.2694,
      "step": 1697
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.18367938695454783,
      "learning_rate": 0.00016089192432677634,
      "loss": 1.1308,
      "step": 1698
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.21335139920301197,
      "learning_rate": 0.00016084457282209243,
      "loss": 1.0612,
      "step": 1699
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.18653516661111677,
      "learning_rate": 0.00016079719964586637,
      "loss": 1.0655,
      "step": 1700
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.17906602933456575,
      "learning_rate": 0.0001607498048149714,
      "loss": 1.1568,
      "step": 1701
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.23679569529730154,
      "learning_rate": 0.00016070238834628865,
      "loss": 0.9413,
      "step": 1702
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1948322488247879,
      "learning_rate": 0.00016065495025670675,
      "loss": 1.0051,
      "step": 1703
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.18338463935212934,
      "learning_rate": 0.00016060749056312222,
      "loss": 1.1352,
      "step": 1704
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.19611889523853765,
      "learning_rate": 0.00016056000928243912,
      "loss": 1.0661,
      "step": 1705
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2456472325264068,
      "learning_rate": 0.0001605125064315693,
      "loss": 1.0977,
      "step": 1706
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.18522869910683362,
      "learning_rate": 0.00016046498202743233,
      "loss": 0.9136,
      "step": 1707
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2291313308471223,
      "learning_rate": 0.00016041743608695533,
      "loss": 1.0156,
      "step": 1708
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.18273837857355651,
      "learning_rate": 0.0001603698686270731,
      "loss": 1.0348,
      "step": 1709
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.16912918689092987,
      "learning_rate": 0.00016032227966472827,
      "loss": 1.0357,
      "step": 1710
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.20346694227121587,
      "learning_rate": 0.00016027466921687093,
      "loss": 1.1316,
      "step": 1711
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1661910229622768,
      "learning_rate": 0.00016022703730045895,
      "loss": 1.0595,
      "step": 1712
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.18318415134698318,
      "learning_rate": 0.0001601793839324578,
      "loss": 1.2349,
      "step": 1713
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.17429054705698152,
      "learning_rate": 0.00016013170912984058,
      "loss": 1.0943,
      "step": 1714
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.17302274788445282,
      "learning_rate": 0.00016008401290958807,
      "loss": 1.1394,
      "step": 1715
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.17332142377171414,
      "learning_rate": 0.00016003629528868859,
      "loss": 1.1031,
      "step": 1716
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.20239042863375165,
      "learning_rate": 0.00015998855628413823,
      "loss": 1.3007,
      "step": 1717
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.17719067100919503,
      "learning_rate": 0.00015994079591294055,
      "loss": 1.146,
      "step": 1718
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1883041431347506,
      "learning_rate": 0.00015989301419210675,
      "loss": 0.9498,
      "step": 1719
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.20538936966411905,
      "learning_rate": 0.00015984521113865572,
      "loss": 1.1351,
      "step": 1720
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.16131900456190876,
      "learning_rate": 0.00015979738676961388,
      "loss": 1.0148,
      "step": 1721
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.22299994080000135,
      "learning_rate": 0.00015974954110201526,
      "loss": 1.1456,
      "step": 1722
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1883322234504808,
      "learning_rate": 0.0001597016741529014,
      "loss": 1.1532,
      "step": 1723
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.16750987836533707,
      "learning_rate": 0.00015965378593932157,
      "loss": 0.9879,
      "step": 1724
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.21743970740297244,
      "learning_rate": 0.00015960587647833249,
      "loss": 1.1652,
      "step": 1725
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.24908371416694755,
      "learning_rate": 0.0001595579457869985,
      "loss": 1.1631,
      "step": 1726
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.24624736846427572,
      "learning_rate": 0.0001595099938823915,
      "loss": 1.0685,
      "step": 1727
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.19955121701514023,
      "learning_rate": 0.00015946202078159094,
      "loss": 1.0621,
      "step": 1728
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.18399754045875774,
      "learning_rate": 0.00015941402650168382,
      "loss": 1.0919,
      "step": 1729
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.22524790978286993,
      "learning_rate": 0.00015936601105976465,
      "loss": 1.1753,
      "step": 1730
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.22945349461254647,
      "learning_rate": 0.00015931797447293552,
      "loss": 1.2038,
      "step": 1731
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.17557686941632697,
      "learning_rate": 0.0001592699167583061,
      "loss": 1.1414,
      "step": 1732
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.19482743229178756,
      "learning_rate": 0.00015922183793299342,
      "loss": 1.1418,
      "step": 1733
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.21963220236877576,
      "learning_rate": 0.00015917373801412223,
      "loss": 1.0701,
      "step": 1734
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.193425867433454,
      "learning_rate": 0.00015912561701882463,
      "loss": 1.0509,
      "step": 1735
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.172505030570464,
      "learning_rate": 0.00015907747496424034,
      "loss": 1.0472,
      "step": 1736
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.196370687834003,
      "learning_rate": 0.00015902931186751652,
      "loss": 1.0905,
      "step": 1737
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1784143613819339,
      "learning_rate": 0.00015898112774580784,
      "loss": 1.1772,
      "step": 1738
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.20499271259621663,
      "learning_rate": 0.00015893292261627643,
      "loss": 1.0979,
      "step": 1739
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.20641329370953354,
      "learning_rate": 0.00015888469649609198,
      "loss": 1.1594,
      "step": 1740
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.21839388747058797,
      "learning_rate": 0.00015883644940243162,
      "loss": 1.1733,
      "step": 1741
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2199034196752957,
      "learning_rate": 0.00015878818135247987,
      "loss": 0.952,
      "step": 1742
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.18458549099525792,
      "learning_rate": 0.0001587398923634288,
      "loss": 1.2036,
      "step": 1743
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.18634061976112606,
      "learning_rate": 0.00015869158245247796,
      "loss": 1.2543,
      "step": 1744
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.22581964595821977,
      "learning_rate": 0.00015864325163683431,
      "loss": 1.0237,
      "step": 1745
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.20151092683599253,
      "learning_rate": 0.0001585948999337122,
      "loss": 0.9642,
      "step": 1746
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.19535111247195275,
      "learning_rate": 0.00015854652736033354,
      "loss": 0.9206,
      "step": 1747
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.18933028891490647,
      "learning_rate": 0.0001584981339339276,
      "loss": 1.045,
      "step": 1748
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.16877162120522693,
      "learning_rate": 0.00015844971967173102,
      "loss": 1.1121,
      "step": 1749
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.19774025845996226,
      "learning_rate": 0.00015840128459098805,
      "loss": 1.0232,
      "step": 1750
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.20129906029928094,
      "learning_rate": 0.00015835282870895015,
      "loss": 1.1057,
      "step": 1751
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.17418518836927713,
      "learning_rate": 0.00015830435204287628,
      "loss": 1.1062,
      "step": 1752
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2251636682153992,
      "learning_rate": 0.00015825585461003283,
      "loss": 1.1664,
      "step": 1753
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.19767006367671242,
      "learning_rate": 0.0001582073364276935,
      "loss": 0.9666,
      "step": 1754
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.21334068703139364,
      "learning_rate": 0.00015815879751313955,
      "loss": 1.2704,
      "step": 1755
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2980475362326206,
      "learning_rate": 0.00015811023788365937,
      "loss": 1.0899,
      "step": 1756
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.19919030172251037,
      "learning_rate": 0.00015806165755654893,
      "loss": 1.0216,
      "step": 1757
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.240233247165891,
      "learning_rate": 0.0001580130565491115,
      "loss": 1.0327,
      "step": 1758
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.21605708372901608,
      "learning_rate": 0.00015796443487865776,
      "loss": 1.0102,
      "step": 1759
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.19774426295995912,
      "learning_rate": 0.00015791579256250565,
      "loss": 1.0978,
      "step": 1760
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.23271235929680936,
      "learning_rate": 0.0001578671296179806,
      "loss": 1.1194,
      "step": 1761
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.18839629084882972,
      "learning_rate": 0.00015781844606241525,
      "loss": 0.9939,
      "step": 1762
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.21863765677344318,
      "learning_rate": 0.0001577697419131497,
      "loss": 1.0748,
      "step": 1763
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.18445495431477837,
      "learning_rate": 0.00015772101718753125,
      "loss": 1.1842,
      "step": 1764
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.23191250051668882,
      "learning_rate": 0.00015767227190291468,
      "loss": 1.013,
      "step": 1765
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.19267933582234376,
      "learning_rate": 0.00015762350607666204,
      "loss": 1.1233,
      "step": 1766
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1970607721606187,
      "learning_rate": 0.00015757471972614265,
      "loss": 1.2892,
      "step": 1767
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.22421777820478211,
      "learning_rate": 0.00015752591286873313,
      "loss": 1.0431,
      "step": 1768
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.17052513669691896,
      "learning_rate": 0.0001574770855218175,
      "loss": 1.0459,
      "step": 1769
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.17999640127828637,
      "learning_rate": 0.00015742823770278702,
      "loss": 0.98,
      "step": 1770
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.19653623488147562,
      "learning_rate": 0.00015737936942904023,
      "loss": 1.1477,
      "step": 1771
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1953803515669757,
      "learning_rate": 0.00015733048071798296,
      "loss": 1.1702,
      "step": 1772
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.22566864722897303,
      "learning_rate": 0.00015728157158702833,
      "loss": 0.9676,
      "step": 1773
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.19262061690754448,
      "learning_rate": 0.00015723264205359675,
      "loss": 1.1645,
      "step": 1774
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1879159370909707,
      "learning_rate": 0.00015718369213511585,
      "loss": 1.0412,
      "step": 1775
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1855183206366637,
      "learning_rate": 0.00015713472184902057,
      "loss": 1.2157,
      "step": 1776
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.19040489961321952,
      "learning_rate": 0.0001570857312127531,
      "loss": 1.0994,
      "step": 1777
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.17877673041189715,
      "learning_rate": 0.00015703672024376282,
      "loss": 1.128,
      "step": 1778
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.18491234640978016,
      "learning_rate": 0.00015698768895950642,
      "loss": 1.0841,
      "step": 1779
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.24848014635670912,
      "learning_rate": 0.0001569386373774478,
      "loss": 1.1481,
      "step": 1780
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.20214229362238073,
      "learning_rate": 0.0001568895655150581,
      "loss": 1.096,
      "step": 1781
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.17716286781475293,
      "learning_rate": 0.00015684047338981568,
      "loss": 1.1209,
      "step": 1782
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.21025413638116489,
      "learning_rate": 0.00015679136101920605,
      "loss": 1.0888,
      "step": 1783
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.20214127161800435,
      "learning_rate": 0.00015674222842072203,
      "loss": 1.223,
      "step": 1784
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.19754034306046095,
      "learning_rate": 0.00015669307561186363,
      "loss": 1.1421,
      "step": 1785
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1877022632021359,
      "learning_rate": 0.00015664390261013798,
      "loss": 0.9455,
      "step": 1786
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.20580749716472654,
      "learning_rate": 0.00015659470943305955,
      "loss": 1.2442,
      "step": 1787
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.20869911710578434,
      "learning_rate": 0.0001565454960981498,
      "loss": 1.0135,
      "step": 1788
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.19208330714105418,
      "learning_rate": 0.0001564962626229375,
      "loss": 1.2567,
      "step": 1789
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.18422468666912195,
      "learning_rate": 0.0001564470090249586,
      "loss": 1.1372,
      "step": 1790
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.17842166291630876,
      "learning_rate": 0.00015639773532175616,
      "loss": 1.0992,
      "step": 1791
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.20137385158143206,
      "learning_rate": 0.00015634844153088043,
      "loss": 1.0369,
      "step": 1792
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.23699024935897403,
      "learning_rate": 0.00015629912766988884,
      "loss": 1.1804,
      "step": 1793
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.17621632359491682,
      "learning_rate": 0.00015624979375634588,
      "loss": 1.1924,
      "step": 1794
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2250199465926083,
      "learning_rate": 0.00015620043980782327,
      "loss": 1.0684,
      "step": 1795
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.16436394999052403,
      "learning_rate": 0.00015615106584189982,
      "loss": 1.0579,
      "step": 1796
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.18613280806774415,
      "learning_rate": 0.00015610167187616152,
      "loss": 1.0282,
      "step": 1797
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1820037012510398,
      "learning_rate": 0.00015605225792820142,
      "loss": 0.9534,
      "step": 1798
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2039899488628761,
      "learning_rate": 0.0001560028240156197,
      "loss": 1.035,
      "step": 1799
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1962581948283822,
      "learning_rate": 0.00015595337015602377,
      "loss": 1.099,
      "step": 1800
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1807500238734824,
      "learning_rate": 0.0001559038963670279,
      "loss": 1.1518,
      "step": 1801
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.21148326261139552,
      "learning_rate": 0.0001558544026662537,
      "loss": 0.9015,
      "step": 1802
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2553837986125019,
      "learning_rate": 0.00015580488907132974,
      "loss": 0.9852,
      "step": 1803
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.15879227264045775,
      "learning_rate": 0.00015575535559989167,
      "loss": 1.1639,
      "step": 1804
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.18312876562851918,
      "learning_rate": 0.00015570580226958232,
      "loss": 1.1956,
      "step": 1805
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.187572357646709,
      "learning_rate": 0.00015565622909805154,
      "loss": 1.0222,
      "step": 1806
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.22128784881437177,
      "learning_rate": 0.00015560663610295617,
      "loss": 0.9933,
      "step": 1807
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1746590676113906,
      "learning_rate": 0.00015555702330196023,
      "loss": 0.9572,
      "step": 1808
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.180430890220108,
      "learning_rate": 0.00015550739071273475,
      "loss": 1.0476,
      "step": 1809
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2023873559565717,
      "learning_rate": 0.00015545773835295775,
      "loss": 1.087,
      "step": 1810
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2008630794770439,
      "learning_rate": 0.00015540806624031442,
      "loss": 1.2042,
      "step": 1811
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2727121662565032,
      "learning_rate": 0.00015535837439249682,
      "loss": 1.0897,
      "step": 1812
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.17713221058150166,
      "learning_rate": 0.0001553086628272042,
      "loss": 1.091,
      "step": 1813
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.20150013866856825,
      "learning_rate": 0.00015525893156214274,
      "loss": 1.0843,
      "step": 1814
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.17843932420352088,
      "learning_rate": 0.00015520918061502569,
      "loss": 1.1216,
      "step": 1815
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.18238466681616203,
      "learning_rate": 0.0001551594100035732,
      "loss": 1.0658,
      "step": 1816
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.18559908814197262,
      "learning_rate": 0.00015510961974551257,
      "loss": 0.9051,
      "step": 1817
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.184184158481438,
      "learning_rate": 0.000155059809858578,
      "loss": 1.1837,
      "step": 1818
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.20179077620577537,
      "learning_rate": 0.00015500998036051074,
      "loss": 1.0694,
      "step": 1819
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.17535690251255495,
      "learning_rate": 0.00015496013126905896,
      "loss": 1.094,
      "step": 1820
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.16352264841786254,
      "learning_rate": 0.00015491026260197786,
      "loss": 1.046,
      "step": 1821
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.22598240855362675,
      "learning_rate": 0.0001548603743770296,
      "loss": 1.0258,
      "step": 1822
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.20835605257169884,
      "learning_rate": 0.00015481046661198334,
      "loss": 1.0639,
      "step": 1823
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2018575785721617,
      "learning_rate": 0.00015476053932461512,
      "loss": 1.3218,
      "step": 1824
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.20182702564222121,
      "learning_rate": 0.00015471059253270797,
      "loss": 1.1266,
      "step": 1825
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2519980128367271,
      "learning_rate": 0.00015466062625405189,
      "loss": 1.0273,
      "step": 1826
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.16173301132366036,
      "learning_rate": 0.0001546106405064438,
      "loss": 1.0092,
      "step": 1827
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.23456087503760617,
      "learning_rate": 0.00015456063530768757,
      "loss": 0.955,
      "step": 1828
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1820889286042369,
      "learning_rate": 0.000154510610675594,
      "loss": 1.1868,
      "step": 1829
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.29847104583499834,
      "learning_rate": 0.0001544605666279808,
      "loss": 1.072,
      "step": 1830
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2022745079542504,
      "learning_rate": 0.00015441050318267254,
      "loss": 1.0391,
      "step": 1831
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.17455507507478818,
      "learning_rate": 0.00015436042035750082,
      "loss": 1.1116,
      "step": 1832
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.20502535605204855,
      "learning_rate": 0.00015431031817030402,
      "loss": 0.9904,
      "step": 1833
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.17527253347814958,
      "learning_rate": 0.00015426019663892753,
      "loss": 1.0055,
      "step": 1834
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1993217195159309,
      "learning_rate": 0.00015421005578122356,
      "loss": 0.924,
      "step": 1835
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1670591826602783,
      "learning_rate": 0.00015415989561505118,
      "loss": 1.1915,
      "step": 1836
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1944975958499578,
      "learning_rate": 0.0001541097161582764,
      "loss": 1.0465,
      "step": 1837
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.21433864271280834,
      "learning_rate": 0.0001540595174287721,
      "loss": 1.017,
      "step": 1838
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.19749030968319412,
      "learning_rate": 0.00015400929944441792,
      "loss": 1.3254,
      "step": 1839
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.19122074582355164,
      "learning_rate": 0.0001539590622231005,
      "loss": 1.1679,
      "step": 1840
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.19124423273770028,
      "learning_rate": 0.00015390880578271328,
      "loss": 1.2174,
      "step": 1841
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2398918623130293,
      "learning_rate": 0.0001538585301411565,
      "loss": 1.1284,
      "step": 1842
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.20259763187326343,
      "learning_rate": 0.00015380823531633729,
      "loss": 1.1989,
      "step": 1843
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.23205322916349952,
      "learning_rate": 0.00015375792132616957,
      "loss": 1.046,
      "step": 1844
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.23241853496894194,
      "learning_rate": 0.00015370758818857415,
      "loss": 1.0903,
      "step": 1845
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.21953076251194936,
      "learning_rate": 0.0001536572359214786,
      "loss": 1.0754,
      "step": 1846
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2106611127360362,
      "learning_rate": 0.0001536068645428174,
      "loss": 1.1662,
      "step": 1847
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.23653106252884293,
      "learning_rate": 0.00015355647407053162,
      "loss": 1.1365,
      "step": 1848
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.21963966312548885,
      "learning_rate": 0.00015350606452256934,
      "loss": 1.027,
      "step": 1849
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.20189338508714186,
      "learning_rate": 0.0001534556359168854,
      "loss": 1.0507,
      "step": 1850
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.162155777740831,
      "learning_rate": 0.00015340518827144145,
      "loss": 0.9376,
      "step": 1851
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1930128817754185,
      "learning_rate": 0.0001533547216042057,
      "loss": 1.2659,
      "step": 1852
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.23403416268709792,
      "learning_rate": 0.0001533042359331534,
      "loss": 1.0545,
      "step": 1853
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1825792921020711,
      "learning_rate": 0.00015325373127626652,
      "loss": 1.0215,
      "step": 1854
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1671353491596232,
      "learning_rate": 0.00015320320765153367,
      "loss": 1.0959,
      "step": 1855
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.16726730633976483,
      "learning_rate": 0.0001531526650769503,
      "loss": 1.2187,
      "step": 1856
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2604817226340623,
      "learning_rate": 0.00015310210357051863,
      "loss": 1.139,
      "step": 1857
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.23052492774533673,
      "learning_rate": 0.00015305152315024756,
      "loss": 1.0361,
      "step": 1858
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.19006465305097944,
      "learning_rate": 0.00015300092383415282,
      "loss": 0.9769,
      "step": 1859
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.19822440888272427,
      "learning_rate": 0.0001529503056402567,
      "loss": 1.1806,
      "step": 1860
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.22107069536397608,
      "learning_rate": 0.00015289966858658842,
      "loss": 1.0263,
      "step": 1861
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.20123758221821236,
      "learning_rate": 0.00015284901269118376,
      "loss": 1.0974,
      "step": 1862
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.17964169350061768,
      "learning_rate": 0.00015279833797208535,
      "loss": 0.9077,
      "step": 1863
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2479720717154668,
      "learning_rate": 0.00015274764444734234,
      "loss": 0.9646,
      "step": 1864
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.242478110009814,
      "learning_rate": 0.0001526969321350108,
      "loss": 1.1998,
      "step": 1865
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.18940820000801187,
      "learning_rate": 0.00015264620105315328,
      "loss": 1.1156,
      "step": 1866
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.18769481736477547,
      "learning_rate": 0.0001525954512198392,
      "loss": 1.1336,
      "step": 1867
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1839460847726564,
      "learning_rate": 0.00015254468265314447,
      "loss": 1.0253,
      "step": 1868
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1585715578075997,
      "learning_rate": 0.00015249389537115183,
      "loss": 1.0486,
      "step": 1869
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2041860536478953,
      "learning_rate": 0.00015244308939195065,
      "loss": 1.1588,
      "step": 1870
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.19915413541786364,
      "learning_rate": 0.00015239226473363687,
      "loss": 1.0591,
      "step": 1871
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.20750196808889496,
      "learning_rate": 0.00015234142141431323,
      "loss": 1.1441,
      "step": 1872
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2566590628002439,
      "learning_rate": 0.000152290559452089,
      "loss": 1.1264,
      "step": 1873
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.17446069313507923,
      "learning_rate": 0.00015223967886508012,
      "loss": 1.0712,
      "step": 1874
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.21473481057807287,
      "learning_rate": 0.0001521887796714092,
      "loss": 0.973,
      "step": 1875
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.16858417531577213,
      "learning_rate": 0.00015213786188920546,
      "loss": 1.0674,
      "step": 1876
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.23173193317980295,
      "learning_rate": 0.00015208692553660467,
      "loss": 1.2003,
      "step": 1877
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1642919304151443,
      "learning_rate": 0.0001520359706317493,
      "loss": 0.9787,
      "step": 1878
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2582899812755441,
      "learning_rate": 0.0001519849971927884,
      "loss": 1.075,
      "step": 1879
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1839968572288907,
      "learning_rate": 0.00015193400523787768,
      "loss": 1.074,
      "step": 1880
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.24041224791567173,
      "learning_rate": 0.00015188299478517933,
      "loss": 1.0568,
      "step": 1881
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1992482596790568,
      "learning_rate": 0.00015183196585286218,
      "loss": 1.1817,
      "step": 1882
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3008609523496942,
      "learning_rate": 0.0001517809184591017,
      "loss": 0.9848,
      "step": 1883
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2069617847039776,
      "learning_rate": 0.00015172985262207987,
      "loss": 1.1477,
      "step": 1884
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.20301162122901817,
      "learning_rate": 0.00015167876835998524,
      "loss": 1.0638,
      "step": 1885
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.17381755032838045,
      "learning_rate": 0.0001516276656910129,
      "loss": 1.238,
      "step": 1886
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.175125514772444,
      "learning_rate": 0.0001515765446333646,
      "loss": 1.204,
      "step": 1887
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.23691725434028377,
      "learning_rate": 0.0001515254052052486,
      "loss": 0.9579,
      "step": 1888
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.23276103056918637,
      "learning_rate": 0.00015147424742487957,
      "loss": 1.0928,
      "step": 1889
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2099915109108064,
      "learning_rate": 0.00015142307131047885,
      "loss": 1.0954,
      "step": 1890
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.27147674881279643,
      "learning_rate": 0.00015137187688027436,
      "loss": 1.039,
      "step": 1891
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.21543024558195986,
      "learning_rate": 0.00015132066415250042,
      "loss": 1.1211,
      "step": 1892
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1726205542275041,
      "learning_rate": 0.00015126943314539788,
      "loss": 1.2154,
      "step": 1893
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.20685023973602987,
      "learning_rate": 0.00015121818387721416,
      "loss": 1.0175,
      "step": 1894
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.19604909662538428,
      "learning_rate": 0.00015116691636620318,
      "loss": 1.0052,
      "step": 1895
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1760404285302947,
      "learning_rate": 0.00015111563063062534,
      "loss": 0.9463,
      "step": 1896
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.21751636970080548,
      "learning_rate": 0.00015106432668874747,
      "loss": 1.15,
      "step": 1897
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.21225488088490976,
      "learning_rate": 0.00015101300455884297,
      "loss": 1.0558,
      "step": 1898
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.22202616026687788,
      "learning_rate": 0.00015096166425919175,
      "loss": 1.1187,
      "step": 1899
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2256460238871411,
      "learning_rate": 0.0001509103058080801,
      "loss": 1.1075,
      "step": 1900
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.21138318569980136,
      "learning_rate": 0.00015085892922380074,
      "loss": 1.0825,
      "step": 1901
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1913527122260372,
      "learning_rate": 0.00015080753452465296,
      "loss": 1.0666,
      "step": 1902
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.19268020029275873,
      "learning_rate": 0.00015075612172894246,
      "loss": 1.0972,
      "step": 1903
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2187917463057192,
      "learning_rate": 0.00015070469085498144,
      "loss": 1.1008,
      "step": 1904
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.18770683419985607,
      "learning_rate": 0.00015065324192108838,
      "loss": 1.0314,
      "step": 1905
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.18667532212889215,
      "learning_rate": 0.0001506017749455884,
      "loss": 1.1554,
      "step": 1906
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.16146415106356812,
      "learning_rate": 0.00015055028994681284,
      "loss": 1.0313,
      "step": 1907
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.18015079272224066,
      "learning_rate": 0.00015049878694309965,
      "loss": 1.0502,
      "step": 1908
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1609265502299731,
      "learning_rate": 0.00015044726595279305,
      "loss": 1.0565,
      "step": 1909
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.19233652939653306,
      "learning_rate": 0.00015039572699424375,
      "loss": 1.1005,
      "step": 1910
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2094880170090412,
      "learning_rate": 0.00015034417008580875,
      "loss": 1.1038,
      "step": 1911
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.21093422854269164,
      "learning_rate": 0.0001502925952458517,
      "loss": 1.234,
      "step": 1912
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.22822316648015117,
      "learning_rate": 0.00015024100249274227,
      "loss": 1.0267,
      "step": 1913
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2057646776352659,
      "learning_rate": 0.00015018939184485678,
      "loss": 1.1572,
      "step": 1914
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1963745268836647,
      "learning_rate": 0.00015013776332057786,
      "loss": 1.1645,
      "step": 1915
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2609128982851452,
      "learning_rate": 0.00015008611693829446,
      "loss": 1.1801,
      "step": 1916
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.19429961656011766,
      "learning_rate": 0.00015003445271640197,
      "loss": 1.1136,
      "step": 1917
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2143670520405398,
      "learning_rate": 0.00014998277067330203,
      "loss": 1.0922,
      "step": 1918
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.17259470751245615,
      "learning_rate": 0.0001499310708274027,
      "loss": 1.0072,
      "step": 1919
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.31512961535082185,
      "learning_rate": 0.00014987935319711842,
      "loss": 1.1264,
      "step": 1920
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.22719569736646492,
      "learning_rate": 0.00014982761780086987,
      "loss": 1.121,
      "step": 1921
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.21871040862525673,
      "learning_rate": 0.00014977586465708407,
      "loss": 1.0871,
      "step": 1922
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.17878712649489692,
      "learning_rate": 0.0001497240937841944,
      "loss": 1.0563,
      "step": 1923
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.18783847027398,
      "learning_rate": 0.00014967230520064056,
      "loss": 1.1465,
      "step": 1924
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1965251785607461,
      "learning_rate": 0.0001496204989248685,
      "loss": 1.1766,
      "step": 1925
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2097145912966756,
      "learning_rate": 0.00014956867497533058,
      "loss": 1.0638,
      "step": 1926
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.20861454758344267,
      "learning_rate": 0.00014951683337048537,
      "loss": 1.1536,
      "step": 1927
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.20404262641002857,
      "learning_rate": 0.00014946497412879773,
      "loss": 1.1512,
      "step": 1928
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.18258421320704654,
      "learning_rate": 0.00014941309726873879,
      "loss": 1.0231,
      "step": 1929
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2187002915431527,
      "learning_rate": 0.00014936120280878605,
      "loss": 1.0562,
      "step": 1930
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2871163636575551,
      "learning_rate": 0.00014930929076742316,
      "loss": 1.0443,
      "step": 1931
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.19132004187285692,
      "learning_rate": 0.0001492573611631401,
      "loss": 1.1009,
      "step": 1932
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.16641590711650714,
      "learning_rate": 0.00014920541401443313,
      "loss": 1.1562,
      "step": 1933
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1977778618520939,
      "learning_rate": 0.0001491534493398046,
      "loss": 1.0737,
      "step": 1934
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.20640794958685663,
      "learning_rate": 0.00014910146715776338,
      "loss": 1.1412,
      "step": 1935
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1760487989462219,
      "learning_rate": 0.00014904946748682432,
      "loss": 1.2531,
      "step": 1936
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.178375352392081,
      "learning_rate": 0.0001489974503455086,
      "loss": 1.1187,
      "step": 1937
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.17578958712865683,
      "learning_rate": 0.00014894541575234365,
      "loss": 1.1953,
      "step": 1938
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1835893783669967,
      "learning_rate": 0.00014889336372586305,
      "loss": 1.1877,
      "step": 1939
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.20007618292324983,
      "learning_rate": 0.00014884129428460665,
      "loss": 1.0451,
      "step": 1940
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.24422244285557773,
      "learning_rate": 0.0001487892074471205,
      "loss": 1.1053,
      "step": 1941
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.20911609240288073,
      "learning_rate": 0.00014873710323195675,
      "loss": 1.1404,
      "step": 1942
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.24522281970028792,
      "learning_rate": 0.00014868498165767388,
      "loss": 1.0082,
      "step": 1943
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.19547903076296047,
      "learning_rate": 0.00014863284274283648,
      "loss": 1.1256,
      "step": 1944
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.325007695594078,
      "learning_rate": 0.00014858068650601527,
      "loss": 1.0572,
      "step": 1945
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2153075197231093,
      "learning_rate": 0.00014852851296578724,
      "loss": 0.9631,
      "step": 1946
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.19554437402053784,
      "learning_rate": 0.00014847632214073548,
      "loss": 1.18,
      "step": 1947
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.17994105888338371,
      "learning_rate": 0.00014842411404944927,
      "loss": 1.1167,
      "step": 1948
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.19721160339820237,
      "learning_rate": 0.000148371888710524,
      "loss": 1.0629,
      "step": 1949
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.16655867145851,
      "learning_rate": 0.0001483196461425612,
      "loss": 1.0457,
      "step": 1950
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.19789533769502857,
      "learning_rate": 0.00014826738636416863,
      "loss": 1.022,
      "step": 1951
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.17833834019736458,
      "learning_rate": 0.00014821510939396003,
      "loss": 0.9049,
      "step": 1952
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2001350143558408,
      "learning_rate": 0.00014816281525055543,
      "loss": 1.0525,
      "step": 1953
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.25082501621499154,
      "learning_rate": 0.00014811050395258083,
      "loss": 0.9559,
      "step": 1954
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2091748323061642,
      "learning_rate": 0.00014805817551866838,
      "loss": 1.0472,
      "step": 1955
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1949120178188088,
      "learning_rate": 0.00014800582996745643,
      "loss": 0.9957,
      "step": 1956
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.29129000709153885,
      "learning_rate": 0.0001479534673175893,
      "loss": 0.9974,
      "step": 1957
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2042930200430595,
      "learning_rate": 0.00014790108758771747,
      "loss": 1.131,
      "step": 1958
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2098224914748471,
      "learning_rate": 0.0001478486907964975,
      "loss": 0.9471,
      "step": 1959
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.17298840985650468,
      "learning_rate": 0.00014779627696259195,
      "loss": 1.0126,
      "step": 1960
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.20259027004282035,
      "learning_rate": 0.0001477438461046696,
      "loss": 1.1805,
      "step": 1961
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.17135334219717704,
      "learning_rate": 0.00014769139824140516,
      "loss": 1.0839,
      "step": 1962
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.19840874401662886,
      "learning_rate": 0.0001476389333914794,
      "loss": 1.0091,
      "step": 1963
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.23769655321450323,
      "learning_rate": 0.00014758645157357926,
      "loss": 1.2119,
      "step": 1964
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.17654847254517145,
      "learning_rate": 0.00014753395280639762,
      "loss": 1.2281,
      "step": 1965
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.16726656941204698,
      "learning_rate": 0.0001474814371086334,
      "loss": 1.0759,
      "step": 1966
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.18081207251785691,
      "learning_rate": 0.00014742890449899161,
      "loss": 0.915,
      "step": 1967
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3020563395251193,
      "learning_rate": 0.0001473763549961832,
      "loss": 0.9311,
      "step": 1968
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.17710194874988416,
      "learning_rate": 0.00014732378861892525,
      "loss": 1.0872,
      "step": 1969
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.29336232725648526,
      "learning_rate": 0.00014727120538594074,
      "loss": 1.0132,
      "step": 1970
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.20827698029989114,
      "learning_rate": 0.0001472186053159587,
      "loss": 1.0184,
      "step": 1971
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.23896303015423442,
      "learning_rate": 0.00014716598842771413,
      "loss": 1.1613,
      "step": 1972
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.20857515561527204,
      "learning_rate": 0.00014711335473994815,
      "loss": 1.0246,
      "step": 1973
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.18661385054989016,
      "learning_rate": 0.00014706070427140765,
      "loss": 1.1067,
      "step": 1974
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.18130545908455187,
      "learning_rate": 0.00014700803704084564,
      "loss": 1.1987,
      "step": 1975
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2027941594162983,
      "learning_rate": 0.00014695535306702109,
      "loss": 1.1558,
      "step": 1976
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.16655979984874045,
      "learning_rate": 0.00014690265236869888,
      "loss": 1.1194,
      "step": 1977
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.20524787256723082,
      "learning_rate": 0.0001468499349646499,
      "loss": 1.0176,
      "step": 1978
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2186088981843399,
      "learning_rate": 0.00014679720087365096,
      "loss": 0.972,
      "step": 1979
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.21059219319191452,
      "learning_rate": 0.0001467444501144848,
      "loss": 0.9794,
      "step": 1980
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.21091564809194627,
      "learning_rate": 0.00014669168270594018,
      "loss": 0.9346,
      "step": 1981
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.21048099660454125,
      "learning_rate": 0.00014663889866681164,
      "loss": 1.0871,
      "step": 1982
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.17512334243338962,
      "learning_rate": 0.00014658609801589982,
      "loss": 1.0985,
      "step": 1983
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.26948924177056993,
      "learning_rate": 0.00014653328077201115,
      "loss": 1.1372,
      "step": 1984
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.23402220575012145,
      "learning_rate": 0.00014648044695395797,
      "loss": 0.9702,
      "step": 1985
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.18101358815255322,
      "learning_rate": 0.0001464275965805586,
      "loss": 1.052,
      "step": 1986
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1780303984587088,
      "learning_rate": 0.0001463747296706372,
      "loss": 0.9809,
      "step": 1987
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.17563879677394037,
      "learning_rate": 0.0001463218462430239,
      "loss": 1.2471,
      "step": 1988
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.186757875970591,
      "learning_rate": 0.00014626894631655457,
      "loss": 1.252,
      "step": 1989
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.20927441696544438,
      "learning_rate": 0.0001462160299100711,
      "loss": 1.1344,
      "step": 1990
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.20463847013241465,
      "learning_rate": 0.00014616309704242113,
      "loss": 1.0729,
      "step": 1991
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.17512707246557943,
      "learning_rate": 0.00014611014773245823,
      "loss": 1.0925,
      "step": 1992
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2337355434874305,
      "learning_rate": 0.0001460571819990419,
      "loss": 1.1416,
      "step": 1993
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.19396490652250026,
      "learning_rate": 0.00014600419986103734,
      "loss": 1.2132,
      "step": 1994
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1897341015374581,
      "learning_rate": 0.00014595120133731565,
      "loss": 1.0853,
      "step": 1995
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2007524818271093,
      "learning_rate": 0.00014589818644675378,
      "loss": 1.0488,
      "step": 1996
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.19473377885762871,
      "learning_rate": 0.00014584515520823453,
      "loss": 1.0399,
      "step": 1997
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.18821379253376577,
      "learning_rate": 0.0001457921076406465,
      "loss": 1.0501,
      "step": 1998
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.20090437352763146,
      "learning_rate": 0.00014573904376288409,
      "loss": 1.1779,
      "step": 1999
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.21397204818030546,
      "learning_rate": 0.0001456859635938475,
      "loss": 1.0322,
      "step": 2000
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.22468478359963506,
      "learning_rate": 0.00014563286715244282,
      "loss": 1.0754,
      "step": 2001
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.21194500846950778,
      "learning_rate": 0.00014557975445758181,
      "loss": 0.9793,
      "step": 2002
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.19946644576184505,
      "learning_rate": 0.0001455266255281821,
      "loss": 1.0825,
      "step": 2003
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2115287040389139,
      "learning_rate": 0.00014547348038316713,
      "loss": 1.2049,
      "step": 2004
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.18768148219783592,
      "learning_rate": 0.00014542031904146602,
      "loss": 1.1694,
      "step": 2005
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.22764194361777007,
      "learning_rate": 0.00014536714152201373,
      "loss": 1.2441,
      "step": 2006
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.18575752016528083,
      "learning_rate": 0.00014531394784375094,
      "loss": 1.1221,
      "step": 2007
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.23651162064004674,
      "learning_rate": 0.0001452607380256241,
      "loss": 1.0204,
      "step": 2008
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.21506904139847516,
      "learning_rate": 0.00014520751208658546,
      "loss": 1.1022,
      "step": 2009
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.19580368537627516,
      "learning_rate": 0.00014515427004559292,
      "loss": 1.121,
      "step": 2010
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.17826067697165213,
      "learning_rate": 0.00014510101192161018,
      "loss": 1.0701,
      "step": 2011
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1919536183653056,
      "learning_rate": 0.00014504773773360666,
      "loss": 1.0328,
      "step": 2012
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.23209513580983893,
      "learning_rate": 0.00014499444750055754,
      "loss": 1.1371,
      "step": 2013
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.17744479026462626,
      "learning_rate": 0.00014494114124144353,
      "loss": 1.0939,
      "step": 2014
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.21279844160931016,
      "learning_rate": 0.00014488781897525131,
      "loss": 1.0516,
      "step": 2015
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.18611245953695352,
      "learning_rate": 0.00014483448072097308,
      "loss": 1.0441,
      "step": 2016
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2170670077505332,
      "learning_rate": 0.0001447811264976068,
      "loss": 1.1458,
      "step": 2017
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.21375108480586144,
      "learning_rate": 0.00014472775632415615,
      "loss": 1.1828,
      "step": 2018
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.18457549471771328,
      "learning_rate": 0.0001446743702196304,
      "loss": 1.1163,
      "step": 2019
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.17846542544368152,
      "learning_rate": 0.00014462096820304455,
      "loss": 1.0005,
      "step": 2020
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.16996477801525023,
      "learning_rate": 0.00014456755029341924,
      "loss": 1.2065,
      "step": 2021
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.16921960939247174,
      "learning_rate": 0.00014451411650978087,
      "loss": 1.0473,
      "step": 2022
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1836363911014366,
      "learning_rate": 0.00014446066687116136,
      "loss": 0.9552,
      "step": 2023
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.18026569084256497,
      "learning_rate": 0.00014440720139659833,
      "loss": 0.958,
      "step": 2024
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2010292970390006,
      "learning_rate": 0.00014435372010513509,
      "loss": 0.9423,
      "step": 2025
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.21437696703974454,
      "learning_rate": 0.0001443002230158205,
      "loss": 1.0153,
      "step": 2026
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1881178849366857,
      "learning_rate": 0.00014424671014770906,
      "loss": 1.0544,
      "step": 2027
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2409703672542879,
      "learning_rate": 0.00014419318151986097,
      "loss": 1.3103,
      "step": 2028
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.20989125273123083,
      "learning_rate": 0.00014413963715134196,
      "loss": 1.1175,
      "step": 2029
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.229457021077695,
      "learning_rate": 0.0001440860770612234,
      "loss": 1.1004,
      "step": 2030
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.18052707822321432,
      "learning_rate": 0.00014403250126858224,
      "loss": 1.1528,
      "step": 2031
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2059598784725111,
      "learning_rate": 0.00014397890979250108,
      "loss": 0.9908,
      "step": 2032
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.21051354552155443,
      "learning_rate": 0.000143925302652068,
      "loss": 1.0944,
      "step": 2033
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2216816487950454,
      "learning_rate": 0.0001438716798663768,
      "loss": 1.1377,
      "step": 2034
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.22404359249845998,
      "learning_rate": 0.0001438180414545267,
      "loss": 1.0043,
      "step": 2035
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2446108814565959,
      "learning_rate": 0.00014376438743562262,
      "loss": 1.0252,
      "step": 2036
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.18989733934739758,
      "learning_rate": 0.00014371071782877495,
      "loss": 0.9674,
      "step": 2037
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.186870057624777,
      "learning_rate": 0.00014365703265309968,
      "loss": 1.1308,
      "step": 2038
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.19401089264620788,
      "learning_rate": 0.0001436033319277183,
      "loss": 1.1486,
      "step": 2039
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2002010190420035,
      "learning_rate": 0.00014354961567175786,
      "loss": 1.0494,
      "step": 2040
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1883358824669382,
      "learning_rate": 0.000143495883904351,
      "loss": 1.1456,
      "step": 2041
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.22686715899736773,
      "learning_rate": 0.00014344213664463576,
      "loss": 1.0798,
      "step": 2042
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.21682964722732803,
      "learning_rate": 0.00014338837391175582,
      "loss": 1.0696,
      "step": 2043
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.19504687296262635,
      "learning_rate": 0.00014333459572486032,
      "loss": 1.0435,
      "step": 2044
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.20380193402774466,
      "learning_rate": 0.00014328080210310388,
      "loss": 1.2295,
      "step": 2045
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2112966714942451,
      "learning_rate": 0.00014322699306564668,
      "loss": 1.0906,
      "step": 2046
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.27097482436554415,
      "learning_rate": 0.00014317316863165434,
      "loss": 1.1716,
      "step": 2047
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.20167293769645273,
      "learning_rate": 0.00014311932882029796,
      "loss": 1.0091,
      "step": 2048
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.17735786442686727,
      "learning_rate": 0.00014306547365075414,
      "loss": 1.0512,
      "step": 2049
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1982192755572155,
      "learning_rate": 0.00014301160314220498,
      "loss": 0.9924,
      "step": 2050
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.162860071233834,
      "learning_rate": 0.00014295771731383797,
      "loss": 1.0675,
      "step": 2051
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.18744715972740086,
      "learning_rate": 0.0001429038161848461,
      "loss": 1.1271,
      "step": 2052
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.20996602675600423,
      "learning_rate": 0.0001428498997744278,
      "loss": 1.3314,
      "step": 2053
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.19186783348281655,
      "learning_rate": 0.000142795968101787,
      "loss": 1.1437,
      "step": 2054
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.20930240360959307,
      "learning_rate": 0.00014274202118613294,
      "loss": 0.9109,
      "step": 2055
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1821118322247781,
      "learning_rate": 0.0001426880590466804,
      "loss": 1.0196,
      "step": 2056
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.24224885489392226,
      "learning_rate": 0.00014263408170264955,
      "loss": 0.9411,
      "step": 2057
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.17569449906918605,
      "learning_rate": 0.00014258008917326597,
      "loss": 1.0039,
      "step": 2058
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.20544512492474515,
      "learning_rate": 0.00014252608147776065,
      "loss": 1.1334,
      "step": 2059
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.21696332483463968,
      "learning_rate": 0.00014247205863536996,
      "loss": 1.2875,
      "step": 2060
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.22065992992121114,
      "learning_rate": 0.0001424180206653357,
      "loss": 1.0453,
      "step": 2061
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.21105998798380235,
      "learning_rate": 0.0001423639675869051,
      "loss": 1.0833,
      "step": 2062
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.25202796505353825,
      "learning_rate": 0.00014230989941933065,
      "loss": 1.0137,
      "step": 2063
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.24802330228189284,
      "learning_rate": 0.00014225581618187026,
      "loss": 1.2399,
      "step": 2064
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.18500290916031847,
      "learning_rate": 0.00014220171789378728,
      "loss": 0.9875,
      "step": 2065
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.18044015331799776,
      "learning_rate": 0.00014214760457435037,
      "loss": 0.9593,
      "step": 2066
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.27051524091168505,
      "learning_rate": 0.0001420934762428335,
      "loss": 0.9435,
      "step": 2067
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.26181309650066514,
      "learning_rate": 0.00014203933291851608,
      "loss": 1.0552,
      "step": 2068
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.23320864867966978,
      "learning_rate": 0.00014198517462068279,
      "loss": 1.1784,
      "step": 2069
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.20773777920706296,
      "learning_rate": 0.00014193100136862365,
      "loss": 1.0463,
      "step": 2070
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1850906160901623,
      "learning_rate": 0.00014187681318163404,
      "loss": 1.0875,
      "step": 2071
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.173622603063349,
      "learning_rate": 0.00014182261007901462,
      "loss": 1.0924,
      "step": 2072
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.19754868049351884,
      "learning_rate": 0.0001417683920800714,
      "loss": 1.0375,
      "step": 2073
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2186728881046753,
      "learning_rate": 0.00014171415920411566,
      "loss": 1.0071,
      "step": 2074
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.22098816687902,
      "learning_rate": 0.00014165991147046403,
      "loss": 0.9717,
      "step": 2075
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.22919335768230656,
      "learning_rate": 0.00014160564889843832,
      "loss": 1.1438,
      "step": 2076
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.19266429320511852,
      "learning_rate": 0.0001415513715073658,
      "loss": 1.1538,
      "step": 2077
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.18025495887269685,
      "learning_rate": 0.00014149707931657886,
      "loss": 1.2723,
      "step": 2078
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1953518952274364,
      "learning_rate": 0.0001414427723454153,
      "loss": 1.0314,
      "step": 2079
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.18758242527455968,
      "learning_rate": 0.00014138845061321796,
      "loss": 1.0811,
      "step": 2080
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1765296618174252,
      "learning_rate": 0.00014133411413933523,
      "loss": 1.109,
      "step": 2081
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.17523149183412876,
      "learning_rate": 0.00014127976294312053,
      "loss": 1.121,
      "step": 2082
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.16255846267089863,
      "learning_rate": 0.00014122539704393265,
      "loss": 0.9846,
      "step": 2083
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.20534591091825236,
      "learning_rate": 0.0001411710164611355,
      "loss": 1.0321,
      "step": 2084
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.17893708266973318,
      "learning_rate": 0.00014111662121409833,
      "loss": 1.164,
      "step": 2085
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.278847281571289,
      "learning_rate": 0.0001410622113221956,
      "loss": 1.0247,
      "step": 2086
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2168217568979397,
      "learning_rate": 0.0001410077868048069,
      "loss": 1.1336,
      "step": 2087
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5435171545295869,
      "learning_rate": 0.0001409533476813171,
      "loss": 1.2003,
      "step": 2088
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.21304326292056644,
      "learning_rate": 0.00014089889397111624,
      "loss": 1.0777,
      "step": 2089
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.17117744755906694,
      "learning_rate": 0.00014084442569359964,
      "loss": 0.9832,
      "step": 2090
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.20341825273019515,
      "learning_rate": 0.00014078994286816768,
      "loss": 1.1366,
      "step": 2091
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.19309693838449915,
      "learning_rate": 0.00014073544551422597,
      "loss": 1.2368,
      "step": 2092
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.20131754499176147,
      "learning_rate": 0.00014068093365118535,
      "loss": 1.042,
      "step": 2093
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2047221208765798,
      "learning_rate": 0.00014062640729846178,
      "loss": 1.1304,
      "step": 2094
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.19152896506645553,
      "learning_rate": 0.0001405718664754764,
      "loss": 1.106,
      "step": 2095
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.29882232939121045,
      "learning_rate": 0.00014051731120165542,
      "loss": 1.1398,
      "step": 2096
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.23612426688257424,
      "learning_rate": 0.00014046274149643036,
      "loss": 1.3082,
      "step": 2097
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.17578379199115043,
      "learning_rate": 0.0001404081573792377,
      "loss": 1.1979,
      "step": 2098
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1843421893380292,
      "learning_rate": 0.00014035355886951923,
      "loss": 1.1535,
      "step": 2099
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.20823668762520953,
      "learning_rate": 0.00014029894598672173,
      "loss": 1.1638,
      "step": 2100
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2599152938427948,
      "learning_rate": 0.00014024431875029713,
      "loss": 1.0453,
      "step": 2101
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2212648724956023,
      "learning_rate": 0.00014018967717970254,
      "loss": 1.1626,
      "step": 2102
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2548747725892823,
      "learning_rate": 0.00014013502129440005,
      "loss": 1.0045,
      "step": 2103
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.229798932906175,
      "learning_rate": 0.000140080351113857,
      "loss": 1.1286,
      "step": 2104
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.16071014196180528,
      "learning_rate": 0.0001400256666575457,
      "loss": 1.0231,
      "step": 2105
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.19317409629757948,
      "learning_rate": 0.0001399709679449436,
      "loss": 1.1288,
      "step": 2106
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1701803610988948,
      "learning_rate": 0.00013991625499553325,
      "loss": 1.0305,
      "step": 2107
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.17293789108454383,
      "learning_rate": 0.0001398615278288022,
      "loss": 1.032,
      "step": 2108
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.18453842134472426,
      "learning_rate": 0.0001398067864642431,
      "loss": 1.0995,
      "step": 2109
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.16651844867866739,
      "learning_rate": 0.00013975203092135365,
      "loss": 0.948,
      "step": 2110
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2035881314909315,
      "learning_rate": 0.00013969726121963667,
      "loss": 1.1177,
      "step": 2111
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1900982353133268,
      "learning_rate": 0.0001396424773785999,
      "loss": 1.1586,
      "step": 2112
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.20269073106358154,
      "learning_rate": 0.00013958767941775618,
      "loss": 0.8901,
      "step": 2113
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.21595488215733866,
      "learning_rate": 0.00013953286735662342,
      "loss": 1.2443,
      "step": 2114
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.22510277270234366,
      "learning_rate": 0.0001394780412147245,
      "loss": 1.2375,
      "step": 2115
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.16392582416549148,
      "learning_rate": 0.00013942320101158732,
      "loss": 0.9406,
      "step": 2116
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.19000127572407713,
      "learning_rate": 0.00013936834676674478,
      "loss": 1.1576,
      "step": 2117
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.24419431826675453,
      "learning_rate": 0.00013931347849973475,
      "loss": 1.0182,
      "step": 2118
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.195977681247544,
      "learning_rate": 0.00013925859623010024,
      "loss": 1.0875,
      "step": 2119
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1756572937397054,
      "learning_rate": 0.00013920369997738908,
      "loss": 0.9699,
      "step": 2120
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.166885696747153,
      "learning_rate": 0.0001391487897611542,
      "loss": 1.1321,
      "step": 2121
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.22080559354573223,
      "learning_rate": 0.0001390938656009534,
      "loss": 1.1888,
      "step": 2122
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.20190776833303498,
      "learning_rate": 0.00013903892751634947,
      "loss": 1.1325,
      "step": 2123
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1911024079055726,
      "learning_rate": 0.0001389839755269103,
      "loss": 1.1581,
      "step": 2124
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.23439945859655895,
      "learning_rate": 0.0001389290096522085,
      "loss": 1.1514,
      "step": 2125
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.17621646430174948,
      "learning_rate": 0.00013887402991182178,
      "loss": 0.9712,
      "step": 2126
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.19016863758088529,
      "learning_rate": 0.0001388190363253328,
      "loss": 1.1614,
      "step": 2127
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.19858511281333116,
      "learning_rate": 0.00013876402891232906,
      "loss": 1.2483,
      "step": 2128
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.20080437054036906,
      "learning_rate": 0.00013870900769240305,
      "loss": 0.9882,
      "step": 2129
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.19860261456546038,
      "learning_rate": 0.00013865397268515215,
      "loss": 1.0862,
      "step": 2130
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1713587785750049,
      "learning_rate": 0.00013859892391017865,
      "loss": 1.0627,
      "step": 2131
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.20332879800339887,
      "learning_rate": 0.00013854386138708978,
      "loss": 0.9718,
      "step": 2132
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.22275556964820434,
      "learning_rate": 0.0001384887851354976,
      "loss": 1.0985,
      "step": 2133
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.17821890459325135,
      "learning_rate": 0.00013843369517501912,
      "loss": 0.9905,
      "step": 2134
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.17312076201336857,
      "learning_rate": 0.00013837859152527623,
      "loss": 0.9232,
      "step": 2135
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.21191961923488897,
      "learning_rate": 0.00013832347420589564,
      "loss": 1.0287,
      "step": 2136
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.17825752630797378,
      "learning_rate": 0.000138268343236509,
      "loss": 1.1751,
      "step": 2137
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.18738147082091805,
      "learning_rate": 0.00013821319863675276,
      "loss": 0.9993,
      "step": 2138
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.20839419218553998,
      "learning_rate": 0.00013815804042626828,
      "loss": 0.9946,
      "step": 2139
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1830360901127199,
      "learning_rate": 0.0001381028686247017,
      "loss": 1.1565,
      "step": 2140
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.15972571450421819,
      "learning_rate": 0.00013804768325170408,
      "loss": 1.189,
      "step": 2141
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.16239911209014696,
      "learning_rate": 0.00013799248432693127,
      "loss": 1.0115,
      "step": 2142
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.18761757020222103,
      "learning_rate": 0.00013793727187004395,
      "loss": 0.8983,
      "step": 2143
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.17542426250198898,
      "learning_rate": 0.00013788204590070764,
      "loss": 0.998,
      "step": 2144
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.19228422415314644,
      "learning_rate": 0.0001378268064385926,
      "loss": 1.1616,
      "step": 2145
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.18739055686533304,
      "learning_rate": 0.000137771553503374,
      "loss": 1.1317,
      "step": 2146
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.20065468044771667,
      "learning_rate": 0.00013771628711473172,
      "loss": 0.9614,
      "step": 2147
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2363656506436528,
      "learning_rate": 0.00013766100729235053,
      "loss": 1.128,
      "step": 2148
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1890250796742403,
      "learning_rate": 0.0001376057140559199,
      "loss": 0.87,
      "step": 2149
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.19222113242147343,
      "learning_rate": 0.00013755040742513406,
      "loss": 1.1817,
      "step": 2150
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.18749948765119065,
      "learning_rate": 0.00013749508741969213,
      "loss": 1.1658,
      "step": 2151
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.20477125381475192,
      "learning_rate": 0.00013743975405929786,
      "loss": 0.9127,
      "step": 2152
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.20583200510517746,
      "learning_rate": 0.00013738440736365984,
      "loss": 1.1929,
      "step": 2153
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1958596639405671,
      "learning_rate": 0.0001373290473524914,
      "loss": 0.9653,
      "step": 2154
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.17365971844504646,
      "learning_rate": 0.00013727367404551055,
      "loss": 0.9141,
      "step": 2155
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.21850952461114478,
      "learning_rate": 0.00013721828746244015,
      "loss": 0.9966,
      "step": 2156
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1664632172684075,
      "learning_rate": 0.00013716288762300765,
      "loss": 1.1212,
      "step": 2157
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.21345948340647453,
      "learning_rate": 0.00013710747454694536,
      "loss": 0.9823,
      "step": 2158
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.20490262808976345,
      "learning_rate": 0.0001370520482539902,
      "loss": 1.1086,
      "step": 2159
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.25778493295659494,
      "learning_rate": 0.00013699660876388383,
      "loss": 1.0289,
      "step": 2160
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.16419857118292244,
      "learning_rate": 0.00013694115609637263,
      "loss": 1.0024,
      "step": 2161
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.18920923966000153,
      "learning_rate": 0.00013688569027120768,
      "loss": 1.0792,
      "step": 2162
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2982687221501605,
      "learning_rate": 0.0001368302113081447,
      "loss": 1.1286,
      "step": 2163
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.19315182176445336,
      "learning_rate": 0.00013677471922694418,
      "loss": 0.9451,
      "step": 2164
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.19989569806756832,
      "learning_rate": 0.0001367192140473711,
      "loss": 0.8118,
      "step": 2165
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.18195614595481557,
      "learning_rate": 0.00013666369578919528,
      "loss": 0.995,
      "step": 2166
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.19527396420879406,
      "learning_rate": 0.0001366081644721912,
      "loss": 1.1257,
      "step": 2167
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.16981502433737258,
      "learning_rate": 0.00013655262011613785,
      "loss": 1.1087,
      "step": 2168
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.20983633182006592,
      "learning_rate": 0.000136497062740819,
      "loss": 1.1148,
      "step": 2169
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.19526956506187504,
      "learning_rate": 0.00013644149236602298,
      "loss": 1.0606,
      "step": 2170
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.20217602895911196,
      "learning_rate": 0.00013638590901154276,
      "loss": 1.2047,
      "step": 2171
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.186303151286183,
      "learning_rate": 0.000136330312697176,
      "loss": 1.2717,
      "step": 2172
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.17644194984838524,
      "learning_rate": 0.0001362747034427249,
      "loss": 0.9117,
      "step": 2173
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.19770676890798142,
      "learning_rate": 0.00013621908126799622,
      "loss": 0.977,
      "step": 2174
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.20868361880149655,
      "learning_rate": 0.00013616344619280156,
      "loss": 1.1023,
      "step": 2175
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.17260775664765837,
      "learning_rate": 0.0001361077982369568,
      "loss": 1.0485,
      "step": 2176
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.24691208299216535,
      "learning_rate": 0.0001360521374202826,
      "loss": 1.0707,
      "step": 2177
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.24048586260940027,
      "learning_rate": 0.00013599646376260418,
      "loss": 1.1721,
      "step": 2178
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.20801279752064367,
      "learning_rate": 0.00013594077728375128,
      "loss": 1.2655,
      "step": 2179
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.18923306669688733,
      "learning_rate": 0.0001358850780035583,
      "loss": 1.0034,
      "step": 2180
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1749897115033667,
      "learning_rate": 0.0001358293659418641,
      "loss": 1.0287,
      "step": 2181
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.23045038446997249,
      "learning_rate": 0.00013577364111851207,
      "loss": 0.9841,
      "step": 2182
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2468441930273697,
      "learning_rate": 0.0001357179035533503,
      "loss": 1.0753,
      "step": 2183
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.18049226961905787,
      "learning_rate": 0.0001356621532662313,
      "loss": 1.0408,
      "step": 2184
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1908316366657128,
      "learning_rate": 0.00013560639027701215,
      "loss": 1.0635,
      "step": 2185
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.17769870994008707,
      "learning_rate": 0.00013555061460555437,
      "loss": 1.0679,
      "step": 2186
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.19621683297217238,
      "learning_rate": 0.0001354948262717241,
      "loss": 1.0858,
      "step": 2187
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.16692645519481344,
      "learning_rate": 0.00013543902529539206,
      "loss": 1.1463,
      "step": 2188
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.21477628095000786,
      "learning_rate": 0.0001353832116964332,
      "loss": 1.0697,
      "step": 2189
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.16563984639111648,
      "learning_rate": 0.00013532738549472723,
      "loss": 1.1104,
      "step": 2190
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.21253675627494586,
      "learning_rate": 0.00013527154671015825,
      "loss": 0.9797,
      "step": 2191
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2603013112428631,
      "learning_rate": 0.00013521569536261487,
      "loss": 0.9518,
      "step": 2192
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.21822178758629465,
      "learning_rate": 0.00013515983147199007,
      "loss": 1.016,
      "step": 2193
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.20791791025353287,
      "learning_rate": 0.00013510395505818146,
      "loss": 1.0892,
      "step": 2194
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.22311025749537874,
      "learning_rate": 0.00013504806614109098,
      "loss": 1.1206,
      "step": 2195
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.18796912082760162,
      "learning_rate": 0.00013499216474062512,
      "loss": 1.0879,
      "step": 2196
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2197513974138735,
      "learning_rate": 0.00013493625087669474,
      "loss": 1.072,
      "step": 2197
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.21404190480487714,
      "learning_rate": 0.00013488032456921518,
      "loss": 1.1601,
      "step": 2198
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2064516000086343,
      "learning_rate": 0.00013482438583810614,
      "loss": 1.1787,
      "step": 2199
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.16759266655167315,
      "learning_rate": 0.0001347684347032919,
      "loss": 1.0398,
      "step": 2200
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.20424908759811086,
      "learning_rate": 0.00013471247118470103,
      "loss": 1.0393,
      "step": 2201
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.18593993446824725,
      "learning_rate": 0.00013465649530226652,
      "loss": 1.1468,
      "step": 2202
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.17384113901517934,
      "learning_rate": 0.0001346005070759258,
      "loss": 1.0381,
      "step": 2203
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.18228125098834405,
      "learning_rate": 0.00013454450652562071,
      "loss": 1.2263,
      "step": 2204
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.17397370679176657,
      "learning_rate": 0.0001344884936712974,
      "loss": 1.0407,
      "step": 2205
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.17339066237423584,
      "learning_rate": 0.00013443246853290654,
      "loss": 1.0357,
      "step": 2206
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1909401386662949,
      "learning_rate": 0.00013437643113040301,
      "loss": 1.141,
      "step": 2207
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.17154611203630948,
      "learning_rate": 0.00013432038148374618,
      "loss": 1.1105,
      "step": 2208
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1674556838815995,
      "learning_rate": 0.00013426431961289975,
      "loss": 1.3235,
      "step": 2209
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.17680448134840074,
      "learning_rate": 0.0001342082455378317,
      "loss": 1.1872,
      "step": 2210
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.23922128635945303,
      "learning_rate": 0.0001341521592785145,
      "loss": 1.0145,
      "step": 2211
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1840089421141917,
      "learning_rate": 0.00013409606085492486,
      "loss": 1.1841,
      "step": 2212
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.18564767552491648,
      "learning_rate": 0.00013403995028704385,
      "loss": 1.1327,
      "step": 2213
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2420259244275171,
      "learning_rate": 0.00013398382759485683,
      "loss": 1.1808,
      "step": 2214
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2840336867707975,
      "learning_rate": 0.00013392769279835354,
      "loss": 1.1358,
      "step": 2215
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.19653180143965554,
      "learning_rate": 0.00013387154591752799,
      "loss": 1.2368,
      "step": 2216
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.17662402366046578,
      "learning_rate": 0.00013381538697237853,
      "loss": 0.9985,
      "step": 2217
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1897423085787927,
      "learning_rate": 0.0001337592159829077,
      "loss": 1.1223,
      "step": 2218
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2097299201689303,
      "learning_rate": 0.00013370303296912249,
      "loss": 1.2238,
      "step": 2219
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.16968032003048725,
      "learning_rate": 0.00013364683795103404,
      "loss": 1.2042,
      "step": 2220
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2700431330912942,
      "learning_rate": 0.0001335906309486579,
      "loss": 1.0762,
      "step": 2221
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1959296249269031,
      "learning_rate": 0.00013353441198201373,
      "loss": 0.9994,
      "step": 2222
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.18523139501169936,
      "learning_rate": 0.00013347818107112554,
      "loss": 1.04,
      "step": 2223
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.21886543604262804,
      "learning_rate": 0.00013342193823602164,
      "loss": 1.0557,
      "step": 2224
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.19966068331252604,
      "learning_rate": 0.00013336568349673448,
      "loss": 1.0135,
      "step": 2225
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.17234307299526447,
      "learning_rate": 0.00013330941687330082,
      "loss": 1.132,
      "step": 2226
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.17447746117894217,
      "learning_rate": 0.0001332531383857616,
      "loss": 1.3064,
      "step": 2227
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.20136850592209132,
      "learning_rate": 0.00013319684805416208,
      "loss": 0.9464,
      "step": 2228
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.22727836387348777,
      "learning_rate": 0.00013314054589855164,
      "loss": 1.115,
      "step": 2229
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.18427820664302694,
      "learning_rate": 0.00013308423193898392,
      "loss": 1.1588,
      "step": 2230
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.18102158251118092,
      "learning_rate": 0.00013302790619551674,
      "loss": 1.0488,
      "step": 2231
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.25063507107473487,
      "learning_rate": 0.00013297156868821212,
      "loss": 1.1264,
      "step": 2232
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.19651515779136397,
      "learning_rate": 0.00013291521943713632,
      "loss": 1.1655,
      "step": 2233
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.17532218874917838,
      "learning_rate": 0.00013285885846235973,
      "loss": 1.0681,
      "step": 2234
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.20571796481407176,
      "learning_rate": 0.0001328024857839569,
      "loss": 1.01,
      "step": 2235
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.16790192443576837,
      "learning_rate": 0.00013274610142200664,
      "loss": 0.9334,
      "step": 2236
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.19060099153903107,
      "learning_rate": 0.00013268970539659182,
      "loss": 1.2665,
      "step": 2237
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.19154296822343622,
      "learning_rate": 0.00013263329772779953,
      "loss": 1.1386,
      "step": 2238
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2108912559555725,
      "learning_rate": 0.00013257687843572092,
      "loss": 1.0192,
      "step": 2239
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.20440581904768845,
      "learning_rate": 0.00013252044754045138,
      "loss": 1.0795,
      "step": 2240
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.18465851559484042,
      "learning_rate": 0.00013246400506209045,
      "loss": 0.9879,
      "step": 2241
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.18710781117138967,
      "learning_rate": 0.00013240755102074162,
      "loss": 1.0031,
      "step": 2242
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1839918039809768,
      "learning_rate": 0.00013235108543651272,
      "loss": 1.074,
      "step": 2243
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.18615647877261282,
      "learning_rate": 0.00013229460832951554,
      "loss": 1.1524,
      "step": 2244
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.20611384513248135,
      "learning_rate": 0.00013223811971986606,
      "loss": 1.1301,
      "step": 2245
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.19173063015527259,
      "learning_rate": 0.00013218161962768427,
      "loss": 1.1208,
      "step": 2246
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1954800988814553,
      "learning_rate": 0.00013212510807309434,
      "loss": 0.998,
      "step": 2247
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.21078937047901422,
      "learning_rate": 0.00013206858507622447,
      "loss": 1.1379,
      "step": 2248
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.19346315023618096,
      "learning_rate": 0.00013201205065720698,
      "loss": 1.1037,
      "step": 2249
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1925326092293825,
      "learning_rate": 0.00013195550483617818,
      "loss": 1.1315,
      "step": 2250
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.17706065935150112,
      "learning_rate": 0.0001318989476332785,
      "loss": 1.137,
      "step": 2251
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.20192177241316533,
      "learning_rate": 0.00013184237906865243,
      "loss": 0.8633,
      "step": 2252
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.20777970007724375,
      "learning_rate": 0.00013178579916244847,
      "loss": 0.9084,
      "step": 2253
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1846838205225053,
      "learning_rate": 0.00013172920793481918,
      "loss": 1.1645,
      "step": 2254
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.18156717062508007,
      "learning_rate": 0.00013167260540592114,
      "loss": 1.2298,
      "step": 2255
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.20846205888612265,
      "learning_rate": 0.00013161599159591502,
      "loss": 1.2661,
      "step": 2256
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.20921480193409225,
      "learning_rate": 0.00013155936652496544,
      "loss": 1.0412,
      "step": 2257
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.19167550528822047,
      "learning_rate": 0.00013150273021324095,
      "loss": 1.0796,
      "step": 2258
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2055800122082977,
      "learning_rate": 0.00013144608268091435,
      "loss": 1.1822,
      "step": 2259
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.19312607934200535,
      "learning_rate": 0.00013138942394816215,
      "loss": 1.2399,
      "step": 2260
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1684476034227272,
      "learning_rate": 0.00013133275403516507,
      "loss": 1.1509,
      "step": 2261
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.18550963954241426,
      "learning_rate": 0.00013127607296210773,
      "loss": 0.8515,
      "step": 2262
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2360475894069583,
      "learning_rate": 0.00013121938074917865,
      "loss": 1.0808,
      "step": 2263
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.19309144200079048,
      "learning_rate": 0.00013116267741657045,
      "loss": 1.1255,
      "step": 2264
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.22789578239088526,
      "learning_rate": 0.00013110596298447963,
      "loss": 1.0672,
      "step": 2265
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1851490151059784,
      "learning_rate": 0.00013104923747310666,
      "loss": 0.9917,
      "step": 2266
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.17944739996772266,
      "learning_rate": 0.000130992500902656,
      "loss": 1.1504,
      "step": 2267
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2086173594786595,
      "learning_rate": 0.0001309357532933359,
      "loss": 1.078,
      "step": 2268
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1818436886744165,
      "learning_rate": 0.0001308789946653588,
      "loss": 1.1571,
      "step": 2269
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.23425685784015585,
      "learning_rate": 0.00013082222503894085,
      "loss": 0.9925,
      "step": 2270
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2171127231993143,
      "learning_rate": 0.00013076544443430213,
      "loss": 1.0811,
      "step": 2271
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.21532215687931688,
      "learning_rate": 0.00013070865287166677,
      "loss": 1.0476,
      "step": 2272
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.18828664425556557,
      "learning_rate": 0.00013065185037126263,
      "loss": 0.951,
      "step": 2273
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.18863986724262008,
      "learning_rate": 0.00013059503695332165,
      "loss": 1.0584,
      "step": 2274
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.20091454807855338,
      "learning_rate": 0.00013053821263807946,
      "loss": 1.1206,
      "step": 2275
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1947598702368196,
      "learning_rate": 0.00013048137744577572,
      "loss": 1.1352,
      "step": 2276
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2333773248472431,
      "learning_rate": 0.00013042453139665397,
      "loss": 1.1653,
      "step": 2277
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1998308875602797,
      "learning_rate": 0.00013036767451096148,
      "loss": 1.1969,
      "step": 2278
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1835577853360749,
      "learning_rate": 0.0001303108068089495,
      "loss": 1.0974,
      "step": 2279
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.18461231926404076,
      "learning_rate": 0.00013025392831087307,
      "loss": 1.0304,
      "step": 2280
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.18372197494835232,
      "learning_rate": 0.00013019703903699114,
      "loss": 1.0685,
      "step": 2281
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.18770259063757627,
      "learning_rate": 0.00013014013900756644,
      "loss": 1.0619,
      "step": 2282
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2384097294809393,
      "learning_rate": 0.00013008322824286555,
      "loss": 1.0358,
      "step": 2283
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.20092626520746382,
      "learning_rate": 0.00013002630676315884,
      "loss": 1.1867,
      "step": 2284
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.19661407470105077,
      "learning_rate": 0.00012996937458872057,
      "loss": 1.0814,
      "step": 2285
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2151259445321274,
      "learning_rate": 0.00012991243173982877,
      "loss": 1.003,
      "step": 2286
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.19051682877444615,
      "learning_rate": 0.00012985547823676525,
      "loss": 1.0623,
      "step": 2287
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2698364906896789,
      "learning_rate": 0.00012979851409981565,
      "loss": 0.9544,
      "step": 2288
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2265557241386887,
      "learning_rate": 0.00012974153934926938,
      "loss": 1.0335,
      "step": 2289
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.18583706466032493,
      "learning_rate": 0.00012968455400541966,
      "loss": 1.0357,
      "step": 2290
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.19928637416393463,
      "learning_rate": 0.00012962755808856342,
      "loss": 1.0997,
      "step": 2291
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.16943226864753638,
      "learning_rate": 0.00012957055161900137,
      "loss": 1.0883,
      "step": 2292
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.18291058356986467,
      "learning_rate": 0.00012951353461703808,
      "loss": 1.1382,
      "step": 2293
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.19096648164877966,
      "learning_rate": 0.00012945650710298175,
      "loss": 1.1564,
      "step": 2294
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.18294120161615185,
      "learning_rate": 0.00012939946909714433,
      "loss": 1.1643,
      "step": 2295
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.22596424031396065,
      "learning_rate": 0.0001293424206198416,
      "loss": 0.962,
      "step": 2296
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.18233423699635073,
      "learning_rate": 0.000129285361691393,
      "loss": 1.1607,
      "step": 2297
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.20124580164187206,
      "learning_rate": 0.00012922829233212165,
      "loss": 1.1971,
      "step": 2298
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1874642713261692,
      "learning_rate": 0.00012917121256235455,
      "loss": 1.0947,
      "step": 2299
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.20323197611796556,
      "learning_rate": 0.0001291141224024222,
      "loss": 1.1557,
      "step": 2300
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.20034620050654248,
      "learning_rate": 0.00012905702187265895,
      "loss": 0.9281,
      "step": 2301
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1836515770148713,
      "learning_rate": 0.00012899991099340284,
      "loss": 1.1104,
      "step": 2302
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.18128883573977342,
      "learning_rate": 0.00012894278978499543,
      "loss": 1.1527,
      "step": 2303
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.21198722785448984,
      "learning_rate": 0.00012888565826778216,
      "loss": 1.1489,
      "step": 2304
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.17236934761040135,
      "learning_rate": 0.00012882851646211206,
      "loss": 1.1845,
      "step": 2305
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.200889990161615,
      "learning_rate": 0.0001287713643883378,
      "loss": 1.0594,
      "step": 2306
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1898233586666023,
      "learning_rate": 0.00012871420206681571,
      "loss": 1.1232,
      "step": 2307
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.21800408522893627,
      "learning_rate": 0.00012865702951790587,
      "loss": 0.8957,
      "step": 2308
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.16268245172636736,
      "learning_rate": 0.00012859984676197187,
      "loss": 1.0916,
      "step": 2309
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.19365529011236857,
      "learning_rate": 0.000128542653819381,
      "loss": 1.1519,
      "step": 2310
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1889076720064626,
      "learning_rate": 0.00012848545071050414,
      "loss": 1.0236,
      "step": 2311
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.20111278250192866,
      "learning_rate": 0.00012842823745571588,
      "loss": 0.9467,
      "step": 2312
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.16621286724082296,
      "learning_rate": 0.00012837101407539434,
      "loss": 1.0474,
      "step": 2313
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.24354076629143798,
      "learning_rate": 0.00012831378058992128,
      "loss": 0.8674,
      "step": 2314
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.18426785383283917,
      "learning_rate": 0.000128256537019682,
      "loss": 1.0882,
      "step": 2315
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.18373890149396838,
      "learning_rate": 0.00012819928338506548,
      "loss": 1.146,
      "step": 2316
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.173211105565574,
      "learning_rate": 0.00012814201970646425,
      "loss": 1.0378,
      "step": 2317
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.20846459492204225,
      "learning_rate": 0.0001280847460042744,
      "loss": 1.0985,
      "step": 2318
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.17717794363471392,
      "learning_rate": 0.00012802746229889563,
      "loss": 1.0004,
      "step": 2319
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.16199498584542904,
      "learning_rate": 0.00012797016861073113,
      "loss": 1.2404,
      "step": 2320
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2059934372407809,
      "learning_rate": 0.00012791286496018772,
      "loss": 1.0468,
      "step": 2321
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.21879664858756365,
      "learning_rate": 0.00012785555136767576,
      "loss": 1.0484,
      "step": 2322
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.20504644431712793,
      "learning_rate": 0.00012779822785360912,
      "loss": 1.2377,
      "step": 2323
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1823661887722081,
      "learning_rate": 0.0001277408944384052,
      "loss": 1.0922,
      "step": 2324
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1559755958708782,
      "learning_rate": 0.00012768355114248494,
      "loss": 1.0415,
      "step": 2325
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.18356730631006365,
      "learning_rate": 0.00012762619798627282,
      "loss": 1.0832,
      "step": 2326
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.16964117647770094,
      "learning_rate": 0.00012756883499019683,
      "loss": 1.117,
      "step": 2327
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.20286989731674332,
      "learning_rate": 0.0001275114621746884,
      "loss": 1.1692,
      "step": 2328
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2408439318071222,
      "learning_rate": 0.00012745407956018254,
      "loss": 1.1368,
      "step": 2329
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1938777737555802,
      "learning_rate": 0.00012739668716711772,
      "loss": 1.0295,
      "step": 2330
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.17180347503576646,
      "learning_rate": 0.00012733928501593587,
      "loss": 1.0987,
      "step": 2331
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1707612629897477,
      "learning_rate": 0.00012728187312708238,
      "loss": 1.052,
      "step": 2332
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2271940596328298,
      "learning_rate": 0.00012722445152100624,
      "loss": 1.0114,
      "step": 2333
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.19963140156371992,
      "learning_rate": 0.00012716702021815977,
      "loss": 1.0782,
      "step": 2334
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.19741322225185523,
      "learning_rate": 0.0001271095792389987,
      "loss": 0.9718,
      "step": 2335
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.18717722900485845,
      "learning_rate": 0.0001270521286039824,
      "loss": 1.0873,
      "step": 2336
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.21050927862491356,
      "learning_rate": 0.00012699466833357342,
      "loss": 1.1356,
      "step": 2337
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2062590888571451,
      "learning_rate": 0.00012693719844823805,
      "loss": 1.0463,
      "step": 2338
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.18965851808744097,
      "learning_rate": 0.00012687971896844575,
      "loss": 1.0775,
      "step": 2339
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.18850291046928616,
      "learning_rate": 0.00012682222991466948,
      "loss": 1.1007,
      "step": 2340
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.18598038976489803,
      "learning_rate": 0.00012676473130738566,
      "loss": 1.3753,
      "step": 2341
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1924532555686447,
      "learning_rate": 0.00012670722316707403,
      "loss": 1.0853,
      "step": 2342
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.18703367623448114,
      "learning_rate": 0.00012664970551421777,
      "loss": 0.9487,
      "step": 2343
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1860359825175446,
      "learning_rate": 0.0001265921783693035,
      "loss": 1.0198,
      "step": 2344
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.22018554867363754,
      "learning_rate": 0.00012653464175282108,
      "loss": 1.043,
      "step": 2345
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2009948331029508,
      "learning_rate": 0.0001264770956852639,
      "loss": 1.1466,
      "step": 2346
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.18821022870920012,
      "learning_rate": 0.00012641954018712863,
      "loss": 1.0328,
      "step": 2347
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1652267488522754,
      "learning_rate": 0.00012636197527891526,
      "loss": 1.1141,
      "step": 2348
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1958135434665089,
      "learning_rate": 0.00012630440098112725,
      "loss": 1.0763,
      "step": 2349
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.19812607662593842,
      "learning_rate": 0.0001262468173142713,
      "loss": 1.1097,
      "step": 2350
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.18446026336145654,
      "learning_rate": 0.00012618922429885753,
      "loss": 1.139,
      "step": 2351
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.18243708383157273,
      "learning_rate": 0.0001261316219553993,
      "loss": 1.0299,
      "step": 2352
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.256560170497605,
      "learning_rate": 0.00012607401030441336,
      "loss": 1.0797,
      "step": 2353
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.22476888929157265,
      "learning_rate": 0.00012601638936641976,
      "loss": 1.2486,
      "step": 2354
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.18647068784172244,
      "learning_rate": 0.00012595875916194188,
      "loss": 1.1125,
      "step": 2355
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.17827566894732516,
      "learning_rate": 0.00012590111971150626,
      "loss": 0.9674,
      "step": 2356
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.19407810523726898,
      "learning_rate": 0.00012584347103564297,
      "loss": 0.9884,
      "step": 2357
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1936345032699986,
      "learning_rate": 0.00012578581315488514,
      "loss": 1.1703,
      "step": 2358
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.23977502249270416,
      "learning_rate": 0.0001257281460897694,
      "loss": 1.0862,
      "step": 2359
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.20569425680199618,
      "learning_rate": 0.00012567046986083544,
      "loss": 1.1384,
      "step": 2360
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1983480312696826,
      "learning_rate": 0.00012561278448862634,
      "loss": 1.0451,
      "step": 2361
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.20114725041539808,
      "learning_rate": 0.00012555508999368838,
      "loss": 1.054,
      "step": 2362
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.18138358187580494,
      "learning_rate": 0.00012549738639657115,
      "loss": 1.0609,
      "step": 2363
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.17113703861461238,
      "learning_rate": 0.00012543967371782738,
      "loss": 1.179,
      "step": 2364
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2227025097580783,
      "learning_rate": 0.00012538195197801317,
      "loss": 1.2359,
      "step": 2365
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1849721978129481,
      "learning_rate": 0.00012532422119768775,
      "loss": 1.1193,
      "step": 2366
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1886472224333886,
      "learning_rate": 0.0001252664813974136,
      "loss": 1.082,
      "step": 2367
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.20166150734162874,
      "learning_rate": 0.00012520873259775636,
      "loss": 1.1664,
      "step": 2368
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1901340596227541,
      "learning_rate": 0.00012515097481928495,
      "loss": 1.1152,
      "step": 2369
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1709847169578024,
      "learning_rate": 0.0001250932080825715,
      "loss": 1.1426,
      "step": 2370
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.18984416216500788,
      "learning_rate": 0.00012503543240819127,
      "loss": 1.0707,
      "step": 2371
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.18961685276326973,
      "learning_rate": 0.0001249776478167227,
      "loss": 1.0467,
      "step": 2372
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2009665704426878,
      "learning_rate": 0.00012491985432874745,
      "loss": 1.1171,
      "step": 2373
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2014365989793528,
      "learning_rate": 0.00012486205196485034,
      "loss": 1.2107,
      "step": 2374
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.18098949618585697,
      "learning_rate": 0.00012480424074561933,
      "loss": 0.904,
      "step": 2375
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.17085747910940538,
      "learning_rate": 0.0001247464206916456,
      "loss": 1.0462,
      "step": 2376
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.18932586829101192,
      "learning_rate": 0.00012468859182352332,
      "loss": 1.2413,
      "step": 2377
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2574273191986423,
      "learning_rate": 0.00012463075416184994,
      "loss": 1.1175,
      "step": 2378
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2037278989468581,
      "learning_rate": 0.00012457290772722608,
      "loss": 1.2421,
      "step": 2379
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.18456957581564662,
      "learning_rate": 0.0001245150525402553,
      "loss": 1.0423,
      "step": 2380
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.16877441904399051,
      "learning_rate": 0.00012445718862154445,
      "loss": 1.0992,
      "step": 2381
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.20428757163135539,
      "learning_rate": 0.00012439931599170342,
      "loss": 1.0668,
      "step": 2382
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.20500398411711587,
      "learning_rate": 0.00012434143467134523,
      "loss": 0.9925,
      "step": 2383
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.19049311312608058,
      "learning_rate": 0.0001242835446810859,
      "loss": 1.0883,
      "step": 2384
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1870787091757859,
      "learning_rate": 0.00012422564604154467,
      "loss": 1.1532,
      "step": 2385
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.17637336320256547,
      "learning_rate": 0.0001241677387733438,
      "loss": 1.0969,
      "step": 2386
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1898665399346918,
      "learning_rate": 0.00012410982289710865,
      "loss": 1.0771,
      "step": 2387
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.19515006959445794,
      "learning_rate": 0.00012405189843346756,
      "loss": 1.0358,
      "step": 2388
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2076290661365544,
      "learning_rate": 0.00012399396540305205,
      "loss": 1.1038,
      "step": 2389
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1761987474852816,
      "learning_rate": 0.00012393602382649655,
      "loss": 1.0615,
      "step": 2390
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.16196788780032115,
      "learning_rate": 0.00012387807372443873,
      "loss": 1.1384,
      "step": 2391
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.18512960694549632,
      "learning_rate": 0.00012382011511751913,
      "loss": 1.0957,
      "step": 2392
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.164437031249974,
      "learning_rate": 0.0001237621480263813,
      "loss": 1.0053,
      "step": 2393
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.15819630106567842,
      "learning_rate": 0.00012370417247167202,
      "loss": 1.0879,
      "step": 2394
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.19393333500727927,
      "learning_rate": 0.0001236461884740409,
      "loss": 1.1424,
      "step": 2395
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.22853625873533967,
      "learning_rate": 0.0001235881960541405,
      "loss": 1.0415,
      "step": 2396
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2205787828541167,
      "learning_rate": 0.00012353019523262665,
      "loss": 1.1401,
      "step": 2397
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.18830404059409414,
      "learning_rate": 0.00012347218603015793,
      "loss": 0.9408,
      "step": 2398
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.28150642336805853,
      "learning_rate": 0.00012341416846739595,
      "loss": 1.1999,
      "step": 2399
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1960187258780947,
      "learning_rate": 0.0001233561425650054,
      "loss": 0.9744,
      "step": 2400
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.18479761128499236,
      "learning_rate": 0.00012329810834365383,
      "loss": 1.0812,
      "step": 2401
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1778455057969572,
      "learning_rate": 0.0001232400658240118,
      "loss": 1.0222,
      "step": 2402
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.21100005634530483,
      "learning_rate": 0.00012318201502675285,
      "loss": 1.1689,
      "step": 2403
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.23770763283581958,
      "learning_rate": 0.0001231239559725534,
      "loss": 1.0402,
      "step": 2404
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.21251748091895942,
      "learning_rate": 0.00012306588868209284,
      "loss": 1.1487,
      "step": 2405
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.20336179027592358,
      "learning_rate": 0.00012300781317605355,
      "loss": 1.0335,
      "step": 2406
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.20603555917860883,
      "learning_rate": 0.0001229497294751208,
      "loss": 1.0934,
      "step": 2407
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.16934384734816849,
      "learning_rate": 0.00012289163759998269,
      "loss": 1.0142,
      "step": 2408
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2077261517424253,
      "learning_rate": 0.00012283353757133037,
      "loss": 1.1664,
      "step": 2409
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2671351618946086,
      "learning_rate": 0.00012277542940985779,
      "loss": 1.0849,
      "step": 2410
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.20903655085305675,
      "learning_rate": 0.0001227173131362619,
      "loss": 1.0508,
      "step": 2411
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.18576636414517894,
      "learning_rate": 0.00012265918877124246,
      "loss": 1.1124,
      "step": 2412
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.19877996458273717,
      "learning_rate": 0.0001226010563355021,
      "loss": 1.1543,
      "step": 2413
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1788512796259946,
      "learning_rate": 0.00012254291584974636,
      "loss": 1.2145,
      "step": 2414
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2244198889965178,
      "learning_rate": 0.00012248476733468368,
      "loss": 1.0512,
      "step": 2415
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1885506688812071,
      "learning_rate": 0.00012242661081102534,
      "loss": 1.0434,
      "step": 2416
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2026084482644867,
      "learning_rate": 0.00012236844629948538,
      "loss": 1.1521,
      "step": 2417
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1929948323506084,
      "learning_rate": 0.00012231027382078081,
      "loss": 1.065,
      "step": 2418
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.17520180571352503,
      "learning_rate": 0.00012225209339563145,
      "loss": 1.0371,
      "step": 2419
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.22218055448724192,
      "learning_rate": 0.00012219390504475988,
      "loss": 0.9381,
      "step": 2420
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.20674891663671355,
      "learning_rate": 0.0001221357087888916,
      "loss": 1.0314,
      "step": 2421
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.16696420158581307,
      "learning_rate": 0.00012207750464875484,
      "loss": 1.0377,
      "step": 2422
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.18494180760399917,
      "learning_rate": 0.00012201929264508065,
      "loss": 1.091,
      "step": 2423
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.20746048735482595,
      "learning_rate": 0.00012196107279860301,
      "loss": 1.0022,
      "step": 2424
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.17561998812108842,
      "learning_rate": 0.00012190284513005849,
      "loss": 1.0324,
      "step": 2425
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.23321339632102997,
      "learning_rate": 0.00012184460966018659,
      "loss": 1.1179,
      "step": 2426
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.22552858564516925,
      "learning_rate": 0.00012178636640972953,
      "loss": 1.1315,
      "step": 2427
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.17177698011062492,
      "learning_rate": 0.00012172811539943234,
      "loss": 1.0564,
      "step": 2428
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.23115386396931248,
      "learning_rate": 0.00012166985665004278,
      "loss": 0.9748,
      "step": 2429
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.17732013607899272,
      "learning_rate": 0.00012161159018231136,
      "loss": 1.1216,
      "step": 2430
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.15996782459734518,
      "learning_rate": 0.00012155331601699136,
      "loss": 0.9433,
      "step": 2431
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.179689156608436,
      "learning_rate": 0.00012149503417483885,
      "loss": 1.2598,
      "step": 2432
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2103006433187239,
      "learning_rate": 0.00012143674467661248,
      "loss": 1.1288,
      "step": 2433
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.21232021072178756,
      "learning_rate": 0.00012137844754307381,
      "loss": 0.8793,
      "step": 2434
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.20792487522670972,
      "learning_rate": 0.00012132014279498703,
      "loss": 0.987,
      "step": 2435
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.183385530487637,
      "learning_rate": 0.00012126183045311904,
      "loss": 1.1892,
      "step": 2436
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.19685731577012183,
      "learning_rate": 0.00012120351053823942,
      "loss": 1.0421,
      "step": 2437
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1886332991313991,
      "learning_rate": 0.00012114518307112053,
      "loss": 1.1159,
      "step": 2438
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.21810705080901205,
      "learning_rate": 0.00012108684807253734,
      "loss": 1.0067,
      "step": 2439
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2368926369544662,
      "learning_rate": 0.00012102850556326761,
      "loss": 1.1543,
      "step": 2440
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.21170498769586066,
      "learning_rate": 0.00012097015556409161,
      "loss": 1.2124,
      "step": 2441
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1989708625086486,
      "learning_rate": 0.00012091179809579238,
      "loss": 1.04,
      "step": 2442
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.22458378707752397,
      "learning_rate": 0.00012085343317915565,
      "loss": 1.0428,
      "step": 2443
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.20074332906510634,
      "learning_rate": 0.00012079506083496974,
      "loss": 1.1306,
      "step": 2444
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.22341749125544538,
      "learning_rate": 0.00012073668108402565,
      "loss": 1.0022,
      "step": 2445
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.16734327109591557,
      "learning_rate": 0.00012067829394711696,
      "loss": 1.1929,
      "step": 2446
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.21246402143477452,
      "learning_rate": 0.00012061989944503998,
      "loss": 0.9733,
      "step": 2447
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.18647855203471406,
      "learning_rate": 0.00012056149759859357,
      "loss": 1.1285,
      "step": 2448
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.17765778319426676,
      "learning_rate": 0.0001205030884285792,
      "loss": 1.0417,
      "step": 2449
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.18501474976520035,
      "learning_rate": 0.00012044467195580102,
      "loss": 1.0586,
      "step": 2450
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.20904210632134607,
      "learning_rate": 0.00012038624820106572,
      "loss": 1.1352,
      "step": 2451
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.19515519995788574,
      "learning_rate": 0.00012032781718518257,
      "loss": 1.0765,
      "step": 2452
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.17739475443845326,
      "learning_rate": 0.00012026937892896351,
      "loss": 1.0163,
      "step": 2453
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.20515352367872966,
      "learning_rate": 0.00012021093345322292,
      "loss": 1.23,
      "step": 2454
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.19071147662762897,
      "learning_rate": 0.0001201524807787779,
      "loss": 1.027,
      "step": 2455
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.20889510969996555,
      "learning_rate": 0.00012009402092644804,
      "loss": 1.1896,
      "step": 2456
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.22939781915468926,
      "learning_rate": 0.00012003555391705545,
      "loss": 1.0476,
      "step": 2457
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.20395108739617987,
      "learning_rate": 0.00011997707977142492,
      "loss": 1.0598,
      "step": 2458
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2020262134327061,
      "learning_rate": 0.0001199185985103836,
      "loss": 1.0641,
      "step": 2459
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1877484085161883,
      "learning_rate": 0.00011986011015476138,
      "loss": 1.0247,
      "step": 2460
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.21596647293995272,
      "learning_rate": 0.00011980161472539046,
      "loss": 1.0561,
      "step": 2461
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.18466600083246432,
      "learning_rate": 0.00011974311224310573,
      "loss": 1.0043,
      "step": 2462
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.23641206554626507,
      "learning_rate": 0.0001196846027287445,
      "loss": 1.0547,
      "step": 2463
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1810059177063893,
      "learning_rate": 0.00011962608620314664,
      "loss": 1.1033,
      "step": 2464
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.19172146338379523,
      "learning_rate": 0.00011956756268715448,
      "loss": 1.1041,
      "step": 2465
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.17387805972081838,
      "learning_rate": 0.00011950903220161285,
      "loss": 0.9612,
      "step": 2466
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.18242323617308587,
      "learning_rate": 0.00011945049476736905,
      "loss": 1.0284,
      "step": 2467
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.21264561547892752,
      "learning_rate": 0.0001193919504052729,
      "loss": 1.0762,
      "step": 2468
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.21600642741609521,
      "learning_rate": 0.0001193333991361766,
      "loss": 1.0284,
      "step": 2469
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2387689450722008,
      "learning_rate": 0.00011927484098093493,
      "loss": 1.1625,
      "step": 2470
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.19053787800614674,
      "learning_rate": 0.00011921627596040503,
      "loss": 0.9716,
      "step": 2471
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.203220607335337,
      "learning_rate": 0.00011915770409544654,
      "loss": 1.1141,
      "step": 2472
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2331371397467109,
      "learning_rate": 0.00011909912540692148,
      "loss": 1.0552,
      "step": 2473
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.19145223426756325,
      "learning_rate": 0.00011904053991569433,
      "loss": 1.0612,
      "step": 2474
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.16177685764598634,
      "learning_rate": 0.00011898194764263197,
      "loss": 0.9685,
      "step": 2475
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.21335489160209764,
      "learning_rate": 0.00011892334860860382,
      "loss": 1.056,
      "step": 2476
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.20589524234723192,
      "learning_rate": 0.00011886474283448153,
      "loss": 1.1138,
      "step": 2477
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.17320229976895202,
      "learning_rate": 0.00011880613034113921,
      "loss": 1.1115,
      "step": 2478
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.16419910489360068,
      "learning_rate": 0.00011874751114945346,
      "loss": 1.0618,
      "step": 2479
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.18441870973564703,
      "learning_rate": 0.00011868888528030312,
      "loss": 1.0836,
      "step": 2480
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.19679385933225496,
      "learning_rate": 0.00011863025275456953,
      "loss": 1.0493,
      "step": 2481
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1913787184232436,
      "learning_rate": 0.00011857161359313633,
      "loss": 1.0449,
      "step": 2482
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2225215743019526,
      "learning_rate": 0.00011851296781688952,
      "loss": 1.0462,
      "step": 2483
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2463262204730667,
      "learning_rate": 0.00011845431544671748,
      "loss": 0.9099,
      "step": 2484
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.19908179302636622,
      "learning_rate": 0.00011839565650351097,
      "loss": 0.8799,
      "step": 2485
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.22062533165407022,
      "learning_rate": 0.00011833699100816302,
      "loss": 1.1355,
      "step": 2486
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.18493552187357645,
      "learning_rate": 0.00011827831898156905,
      "loss": 1.0578,
      "step": 2487
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.18715069298881631,
      "learning_rate": 0.00011821964044462679,
      "loss": 1.0889,
      "step": 2488
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.25888943923975954,
      "learning_rate": 0.00011816095541823629,
      "loss": 1.0158,
      "step": 2489
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.21654154312788013,
      "learning_rate": 0.00011810226392329988,
      "loss": 1.0928,
      "step": 2490
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1776832634879973,
      "learning_rate": 0.00011804356598072223,
      "loss": 1.0973,
      "step": 2491
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1956010877908206,
      "learning_rate": 0.00011798486161141034,
      "loss": 0.8495,
      "step": 2492
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2069349056965757,
      "learning_rate": 0.00011792615083627345,
      "loss": 0.9948,
      "step": 2493
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.21040610881833502,
      "learning_rate": 0.00011786743367622301,
      "loss": 1.055,
      "step": 2494
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1944801419564681,
      "learning_rate": 0.0001178087101521729,
      "loss": 1.0977,
      "step": 2495
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.19494097299496352,
      "learning_rate": 0.00011774998028503916,
      "loss": 1.0296,
      "step": 2496
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.20957148405687587,
      "learning_rate": 0.00011769124409574014,
      "loss": 1.0103,
      "step": 2497
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.17354131366929665,
      "learning_rate": 0.00011763250160519639,
      "loss": 0.9452,
      "step": 2498
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.19675792449383755,
      "learning_rate": 0.00011757375283433076,
      "loss": 1.0164,
      "step": 2499
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2209491075143487,
      "learning_rate": 0.00011751499780406829,
      "loss": 0.9473,
      "step": 2500
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.20789959389599819,
      "learning_rate": 0.00011745623653533633,
      "loss": 1.0866,
      "step": 2501
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.20632164520406215,
      "learning_rate": 0.0001173974690490643,
      "loss": 0.9617,
      "step": 2502
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.25996568404276266,
      "learning_rate": 0.000117338695366184,
      "loss": 1.0658,
      "step": 2503
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.17520076107176052,
      "learning_rate": 0.00011727991550762936,
      "loss": 0.8955,
      "step": 2504
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.17414879431520974,
      "learning_rate": 0.0001172211294943365,
      "loss": 1.1154,
      "step": 2505
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1841717801547216,
      "learning_rate": 0.00011716233734724371,
      "loss": 1.0385,
      "step": 2506
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.22069321324333974,
      "learning_rate": 0.00011710353908729156,
      "loss": 1.2158,
      "step": 2507
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.18601502459167965,
      "learning_rate": 0.0001170447347354227,
      "loss": 1.1743,
      "step": 2508
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.20022500300964435,
      "learning_rate": 0.00011698592431258205,
      "loss": 1.1577,
      "step": 2509
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.17957556615587322,
      "learning_rate": 0.00011692710783971654,
      "loss": 1.2182,
      "step": 2510
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.18083336710742354,
      "learning_rate": 0.00011686828533777539,
      "loss": 1.0589,
      "step": 2511
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.23237447695620753,
      "learning_rate": 0.00011680945682770994,
      "loss": 1.1927,
      "step": 2512
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.22088909672720067,
      "learning_rate": 0.00011675062233047364,
      "loss": 1.0521,
      "step": 2513
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.23686606820039993,
      "learning_rate": 0.00011669178186702212,
      "loss": 0.9396,
      "step": 2514
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.16892925148100965,
      "learning_rate": 0.00011663293545831302,
      "loss": 0.8753,
      "step": 2515
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.19759926670295877,
      "learning_rate": 0.00011657408312530624,
      "loss": 1.1786,
      "step": 2516
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2746885699104194,
      "learning_rate": 0.00011651522488896371,
      "loss": 1.035,
      "step": 2517
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.18142982664437315,
      "learning_rate": 0.00011645636077024949,
      "loss": 1.2743,
      "step": 2518
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.20664526373149608,
      "learning_rate": 0.00011639749079012972,
      "loss": 1.2304,
      "step": 2519
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.20621349230016534,
      "learning_rate": 0.00011633861496957263,
      "loss": 1.0248,
      "step": 2520
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1755479538730198,
      "learning_rate": 0.00011627973332954855,
      "loss": 0.9525,
      "step": 2521
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.19351218340429444,
      "learning_rate": 0.00011622084589102986,
      "loss": 1.0884,
      "step": 2522
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1798851853885994,
      "learning_rate": 0.00011616195267499102,
      "loss": 1.0595,
      "step": 2523
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.20977180729412018,
      "learning_rate": 0.00011610305370240851,
      "loss": 0.9447,
      "step": 2524
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2148144760477042,
      "learning_rate": 0.00011604414899426093,
      "loss": 1.0117,
      "step": 2525
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.19821071580476557,
      "learning_rate": 0.00011598523857152887,
      "loss": 1.0738,
      "step": 2526
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.19670855695254008,
      "learning_rate": 0.00011592632245519495,
      "loss": 1.1271,
      "step": 2527
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.15789516489891559,
      "learning_rate": 0.00011586740066624384,
      "loss": 0.9327,
      "step": 2528
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2027592224893308,
      "learning_rate": 0.00011580847322566224,
      "loss": 0.9838,
      "step": 2529
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.23751021467344643,
      "learning_rate": 0.00011574954015443888,
      "loss": 1.028,
      "step": 2530
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.22005376929878165,
      "learning_rate": 0.00011569060147356441,
      "loss": 1.0341,
      "step": 2531
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.19747985519637337,
      "learning_rate": 0.00011563165720403157,
      "loss": 0.9529,
      "step": 2532
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.23997001578418492,
      "learning_rate": 0.00011557270736683505,
      "loss": 0.9856,
      "step": 2533
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.18412681172017611,
      "learning_rate": 0.00011551375198297151,
      "loss": 1.1299,
      "step": 2534
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2530550112565592,
      "learning_rate": 0.00011545479107343963,
      "loss": 1.098,
      "step": 2535
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.16364028997953145,
      "learning_rate": 0.00011539582465924001,
      "loss": 1.0171,
      "step": 2536
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.19241810292819858,
      "learning_rate": 0.00011533685276137522,
      "loss": 1.0442,
      "step": 2537
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1919873556898766,
      "learning_rate": 0.00011527787540084985,
      "loss": 1.1949,
      "step": 2538
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1869256955400382,
      "learning_rate": 0.00011521889259867032,
      "loss": 1.0125,
      "step": 2539
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.19135641716898094,
      "learning_rate": 0.00011515990437584507,
      "loss": 1.0685,
      "step": 2540
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1625729371361019,
      "learning_rate": 0.00011510091075338444,
      "loss": 0.755,
      "step": 2541
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.23414550134371803,
      "learning_rate": 0.00011504191175230077,
      "loss": 1.1633,
      "step": 2542
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1908172384956217,
      "learning_rate": 0.00011498290739360815,
      "loss": 1.1263,
      "step": 2543
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1542392497714385,
      "learning_rate": 0.00011492389769832275,
      "loss": 0.8954,
      "step": 2544
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2026297702212014,
      "learning_rate": 0.00011486488268746254,
      "loss": 1.0038,
      "step": 2545
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2374682857923721,
      "learning_rate": 0.0001148058623820474,
      "loss": 1.0358,
      "step": 2546
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.17546831572378607,
      "learning_rate": 0.00011474683680309912,
      "loss": 1.1828,
      "step": 2547
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.24875508456030598,
      "learning_rate": 0.00011468780597164135,
      "loss": 1.0358,
      "step": 2548
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1657946368477515,
      "learning_rate": 0.00011462876990869962,
      "loss": 0.8428,
      "step": 2549
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.17953652373126122,
      "learning_rate": 0.00011456972863530133,
      "loss": 0.993,
      "step": 2550
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.20038654186432478,
      "learning_rate": 0.0001145106821724757,
      "loss": 1.1036,
      "step": 2551
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.17586189902712532,
      "learning_rate": 0.00011445163054125385,
      "loss": 1.1391,
      "step": 2552
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.16535246492819297,
      "learning_rate": 0.00011439257376266868,
      "loss": 1.097,
      "step": 2553
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.20696641987692221,
      "learning_rate": 0.00011433351185775502,
      "loss": 1.2086,
      "step": 2554
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.18853961802495534,
      "learning_rate": 0.0001142744448475494,
      "loss": 1.1296,
      "step": 2555
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.17324804415984213,
      "learning_rate": 0.00011421537275309027,
      "loss": 0.8319,
      "step": 2556
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.18525039295529483,
      "learning_rate": 0.00011415629559541782,
      "loss": 1.0695,
      "step": 2557
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.17568073689448901,
      "learning_rate": 0.00011409721339557414,
      "loss": 1.0499,
      "step": 2558
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.20614148598179852,
      "learning_rate": 0.00011403812617460296,
      "loss": 1.0415,
      "step": 2559
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1891497361903277,
      "learning_rate": 0.00011397903395354996,
      "loss": 1.1526,
      "step": 2560
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2005058032586489,
      "learning_rate": 0.00011391993675346251,
      "loss": 0.9973,
      "step": 2561
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.199357453550534,
      "learning_rate": 0.00011386083459538978,
      "loss": 1.0585,
      "step": 2562
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2202353064758207,
      "learning_rate": 0.00011380172750038269,
      "loss": 1.0933,
      "step": 2563
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2070080199699779,
      "learning_rate": 0.00011374261548949395,
      "loss": 1.0325,
      "step": 2564
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.19525786784574142,
      "learning_rate": 0.00011368349858377796,
      "loss": 0.9947,
      "step": 2565
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1771579860883639,
      "learning_rate": 0.00011362437680429097,
      "loss": 1.0995,
      "step": 2566
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.17726504894589842,
      "learning_rate": 0.00011356525017209085,
      "loss": 1.0168,
      "step": 2567
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.19697167035422858,
      "learning_rate": 0.00011350611870823726,
      "loss": 1.2302,
      "step": 2568
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2061349839328106,
      "learning_rate": 0.00011344698243379154,
      "loss": 1.1953,
      "step": 2569
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.17302666167560768,
      "learning_rate": 0.00011338784136981687,
      "loss": 1.1971,
      "step": 2570
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.23112676928139197,
      "learning_rate": 0.0001133286955373779,
      "loss": 0.9848,
      "step": 2571
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.19881004184003945,
      "learning_rate": 0.00011326954495754121,
      "loss": 1.1637,
      "step": 2572
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1921911010790494,
      "learning_rate": 0.00011321038965137496,
      "loss": 1.2325,
      "step": 2573
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2044260219289887,
      "learning_rate": 0.00011315122963994903,
      "loss": 1.0669,
      "step": 2574
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.22530738111886262,
      "learning_rate": 0.00011309206494433488,
      "loss": 1.1624,
      "step": 2575
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.20983035139909229,
      "learning_rate": 0.0001130328955856058,
      "loss": 1.0714,
      "step": 2576
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2520119811451323,
      "learning_rate": 0.0001129737215848366,
      "loss": 1.1268,
      "step": 2577
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.18017325814932458,
      "learning_rate": 0.00011291454296310389,
      "loss": 1.1242,
      "step": 2578
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.16970227292922735,
      "learning_rate": 0.00011285535974148576,
      "loss": 1.1228,
      "step": 2579
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1968429725601562,
      "learning_rate": 0.00011279617194106198,
      "loss": 1.0639,
      "step": 2580
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.20286388764260943,
      "learning_rate": 0.00011273697958291405,
      "loss": 0.9842,
      "step": 2581
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1835958662127655,
      "learning_rate": 0.00011267778268812504,
      "loss": 1.2114,
      "step": 2582
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1943802489815404,
      "learning_rate": 0.00011261858127777956,
      "loss": 1.0487,
      "step": 2583
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.19892084630346576,
      "learning_rate": 0.00011255937537296393,
      "loss": 0.9825,
      "step": 2584
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1981246202738597,
      "learning_rate": 0.00011250016499476602,
      "loss": 0.9858,
      "step": 2585
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.20759305374394801,
      "learning_rate": 0.00011244095016427534,
      "loss": 1.1147,
      "step": 2586
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2584361446960796,
      "learning_rate": 0.00011238173090258293,
      "loss": 1.0998,
      "step": 2587
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.17522340119224702,
      "learning_rate": 0.00011232250723078141,
      "loss": 1.2119,
      "step": 2588
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1920280264903842,
      "learning_rate": 0.00011226327916996504,
      "loss": 1.1335,
      "step": 2589
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1816925254172026,
      "learning_rate": 0.00011220404674122955,
      "loss": 0.9997,
      "step": 2590
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.18111798646152152,
      "learning_rate": 0.00011214480996567232,
      "loss": 1.025,
      "step": 2591
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2220303992626808,
      "learning_rate": 0.00011208556886439216,
      "loss": 0.9599,
      "step": 2592
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3134843501269834,
      "learning_rate": 0.00011202632345848954,
      "loss": 1.0803,
      "step": 2593
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1798040598992936,
      "learning_rate": 0.00011196707376906641,
      "loss": 1.1139,
      "step": 2594
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.19456252256868758,
      "learning_rate": 0.00011190781981722623,
      "loss": 0.9849,
      "step": 2595
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.19353220817383965,
      "learning_rate": 0.00011184856162407398,
      "loss": 0.9121,
      "step": 2596
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.21637960529054945,
      "learning_rate": 0.00011178929921071622,
      "loss": 1.115,
      "step": 2597
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2463703631746263,
      "learning_rate": 0.00011173003259826094,
      "loss": 1.081,
      "step": 2598
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1790509939917149,
      "learning_rate": 0.00011167076180781764,
      "loss": 1.0025,
      "step": 2599
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2089294824064154,
      "learning_rate": 0.0001116114868604973,
      "loss": 0.9253,
      "step": 2600
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.23915547727881153,
      "learning_rate": 0.0001115522077774124,
      "loss": 1.0168,
      "step": 2601
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.17978867579365215,
      "learning_rate": 0.00011149292457967691,
      "loss": 1.1175,
      "step": 2602
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.20132339584929815,
      "learning_rate": 0.00011143363728840625,
      "loss": 1.1105,
      "step": 2603
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.24312399233388904,
      "learning_rate": 0.00011137434592471725,
      "loss": 1.0047,
      "step": 2604
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.20050426383658096,
      "learning_rate": 0.00011131505050972826,
      "loss": 1.0581,
      "step": 2605
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1927635198274872,
      "learning_rate": 0.00011125575106455906,
      "loss": 1.0972,
      "step": 2606
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2775378344842026,
      "learning_rate": 0.00011119644761033078,
      "loss": 0.8965,
      "step": 2607
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2140134887358899,
      "learning_rate": 0.00011113714016816614,
      "loss": 1.124,
      "step": 2608
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.21133723873737634,
      "learning_rate": 0.00011107782875918913,
      "loss": 1.0076,
      "step": 2609
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.22275906920265293,
      "learning_rate": 0.00011101851340452527,
      "loss": 1.0899,
      "step": 2610
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.18160836878479047,
      "learning_rate": 0.00011095919412530136,
      "loss": 1.0982,
      "step": 2611
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.20862197186464054,
      "learning_rate": 0.00011089987094264569,
      "loss": 1.1262,
      "step": 2612
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.20214483851568044,
      "learning_rate": 0.0001108405438776879,
      "loss": 0.9613,
      "step": 2613
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1851481843903897,
      "learning_rate": 0.00011078121295155905,
      "loss": 1.122,
      "step": 2614
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1895048785408802,
      "learning_rate": 0.00011072187818539159,
      "loss": 1.0482,
      "step": 2615
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.23308125684698586,
      "learning_rate": 0.00011066253960031923,
      "loss": 0.9999,
      "step": 2616
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.17953028591267897,
      "learning_rate": 0.00011060319721747714,
      "loss": 1.0049,
      "step": 2617
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.24356523459540677,
      "learning_rate": 0.0001105438510580018,
      "loss": 0.9983,
      "step": 2618
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1915242658747499,
      "learning_rate": 0.0001104845011430311,
      "loss": 1.0695,
      "step": 2619
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.15783681100967906,
      "learning_rate": 0.00011042514749370418,
      "loss": 1.0569,
      "step": 2620
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.19626811972125924,
      "learning_rate": 0.00011036579013116155,
      "loss": 1.0344,
      "step": 2621
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1756870202065642,
      "learning_rate": 0.00011030642907654503,
      "loss": 1.1199,
      "step": 2622
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.21512039302424926,
      "learning_rate": 0.00011024706435099776,
      "loss": 0.9867,
      "step": 2623
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.18817483179655217,
      "learning_rate": 0.0001101876959756642,
      "loss": 1.1621,
      "step": 2624
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.18125399520544302,
      "learning_rate": 0.00011012832397169013,
      "loss": 1.1853,
      "step": 2625
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2063173359262591,
      "learning_rate": 0.00011006894836022256,
      "loss": 1.1769,
      "step": 2626
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.20270537606239716,
      "learning_rate": 0.00011000956916240985,
      "loss": 0.9722,
      "step": 2627
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2153382777344495,
      "learning_rate": 0.00010995018639940156,
      "loss": 0.9753,
      "step": 2628
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.19769131565532025,
      "learning_rate": 0.0001098908000923486,
      "loss": 1.0418,
      "step": 2629
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.23938266436777858,
      "learning_rate": 0.00010983141026240311,
      "loss": 1.2375,
      "step": 2630
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.17476770053518653,
      "learning_rate": 0.00010977201693071852,
      "loss": 1.1719,
      "step": 2631
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.20960163900744294,
      "learning_rate": 0.00010971262011844938,
      "loss": 1.0322,
      "step": 2632
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1828507779802881,
      "learning_rate": 0.0001096532198467516,
      "loss": 1.1302,
      "step": 2633
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.19844901119861663,
      "learning_rate": 0.00010959381613678235,
      "loss": 1.0121,
      "step": 2634
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.15442788914470837,
      "learning_rate": 0.00010953440900969994,
      "loss": 0.9812,
      "step": 2635
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.20443666209620695,
      "learning_rate": 0.00010947499848666388,
      "loss": 1.0615,
      "step": 2636
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1661005285554268,
      "learning_rate": 0.00010941558458883496,
      "loss": 1.1497,
      "step": 2637
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1638689169145988,
      "learning_rate": 0.00010935616733737517,
      "loss": 0.9505,
      "step": 2638
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1695426221936716,
      "learning_rate": 0.00010929674675344767,
      "loss": 0.9835,
      "step": 2639
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.20195464569414284,
      "learning_rate": 0.00010923732285821678,
      "loss": 0.9276,
      "step": 2640
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.22081167204787305,
      "learning_rate": 0.00010917789567284804,
      "loss": 1.0366,
      "step": 2641
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.22263315126320857,
      "learning_rate": 0.00010911846521850817,
      "loss": 1.019,
      "step": 2642
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.18198940971909977,
      "learning_rate": 0.00010905903151636501,
      "loss": 1.1785,
      "step": 2643
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2012146927885766,
      "learning_rate": 0.00010899959458758754,
      "loss": 1.0046,
      "step": 2644
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.24503165397380208,
      "learning_rate": 0.000108940154453346,
      "loss": 1.0509,
      "step": 2645
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.224704711973627,
      "learning_rate": 0.00010888071113481164,
      "loss": 1.0439,
      "step": 2646
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1875308354920906,
      "learning_rate": 0.00010882126465315698,
      "loss": 1.1903,
      "step": 2647
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.19398580148660632,
      "learning_rate": 0.00010876181502955553,
      "loss": 0.9849,
      "step": 2648
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2277695186226238,
      "learning_rate": 0.000108702362285182,
      "loss": 1.1764,
      "step": 2649
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.20407688658487236,
      "learning_rate": 0.0001086429064412122,
      "loss": 1.3296,
      "step": 2650
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.21747425478795693,
      "learning_rate": 0.00010858344751882304,
      "loss": 1.1957,
      "step": 2651
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1816427240257697,
      "learning_rate": 0.00010852398553919252,
      "loss": 0.985,
      "step": 2652
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1725225593560006,
      "learning_rate": 0.0001084645205234997,
      "loss": 1.1295,
      "step": 2653
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.18804691250691408,
      "learning_rate": 0.00010840505249292476,
      "loss": 1.1672,
      "step": 2654
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1804557647162797,
      "learning_rate": 0.000108345581468649,
      "loss": 1.0246,
      "step": 2655
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.17147607881334712,
      "learning_rate": 0.00010828610747185468,
      "loss": 1.0374,
      "step": 2656
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.22032188135455896,
      "learning_rate": 0.0001082266305237252,
      "loss": 0.9544,
      "step": 2657
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.20196435009784383,
      "learning_rate": 0.00010816715064544495,
      "loss": 1.049,
      "step": 2658
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.23669666566552275,
      "learning_rate": 0.00010810766785819946,
      "loss": 1.2736,
      "step": 2659
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.34978380219252975,
      "learning_rate": 0.00010804818218317517,
      "loss": 1.092,
      "step": 2660
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.21188976667180495,
      "learning_rate": 0.00010798869364155963,
      "loss": 1.0775,
      "step": 2661
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1855424809880326,
      "learning_rate": 0.00010792920225454139,
      "loss": 0.9922,
      "step": 2662
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.44484859206457833,
      "learning_rate": 0.00010786970804331007,
      "loss": 1.1483,
      "step": 2663
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1737692616557497,
      "learning_rate": 0.00010781021102905617,
      "loss": 1.2105,
      "step": 2664
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1883109218261582,
      "learning_rate": 0.00010775071123297128,
      "loss": 1.0638,
      "step": 2665
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.22421579194108304,
      "learning_rate": 0.00010769120867624798,
      "loss": 1.1382,
      "step": 2666
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.19055470552336365,
      "learning_rate": 0.00010763170338007978,
      "loss": 1.1766,
      "step": 2667
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.18373773535010526,
      "learning_rate": 0.00010757219536566125,
      "loss": 0.9669,
      "step": 2668
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.18217032418747184,
      "learning_rate": 0.00010751268465418785,
      "loss": 1.086,
      "step": 2669
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1739931520499227,
      "learning_rate": 0.000107453171266856,
      "loss": 1.1767,
      "step": 2670
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.19130464318440055,
      "learning_rate": 0.00010739365522486318,
      "loss": 1.0356,
      "step": 2671
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.194648932102174,
      "learning_rate": 0.00010733413654940767,
      "loss": 1.1165,
      "step": 2672
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.20884787543862837,
      "learning_rate": 0.00010727461526168878,
      "loss": 1.1055,
      "step": 2673
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2472526455103853,
      "learning_rate": 0.0001072150913829067,
      "loss": 1.0352,
      "step": 2674
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1815867977921345,
      "learning_rate": 0.00010715556493426262,
      "loss": 0.9355,
      "step": 2675
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.19867127590281614,
      "learning_rate": 0.00010709603593695856,
      "loss": 1.197,
      "step": 2676
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1845458771233655,
      "learning_rate": 0.00010703650441219749,
      "loss": 1.0803,
      "step": 2677
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.18589953787422733,
      "learning_rate": 0.00010697697038118326,
      "loss": 1.0433,
      "step": 2678
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.24820716344545388,
      "learning_rate": 0.00010691743386512064,
      "loss": 1.1262,
      "step": 2679
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.22682893398691342,
      "learning_rate": 0.00010685789488521527,
      "loss": 1.098,
      "step": 2680
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.30152258021240347,
      "learning_rate": 0.00010679835346267366,
      "loss": 1.0759,
      "step": 2681
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.19785601186355595,
      "learning_rate": 0.0001067388096187032,
      "loss": 1.14,
      "step": 2682
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.18208884639381998,
      "learning_rate": 0.00010667926337451217,
      "loss": 1.0227,
      "step": 2683
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.20991379525044426,
      "learning_rate": 0.00010661971475130963,
      "loss": 0.9756,
      "step": 2684
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.18314873715783483,
      "learning_rate": 0.00010656016377030555,
      "loss": 1.0372,
      "step": 2685
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.22820986633428797,
      "learning_rate": 0.00010650061045271072,
      "loss": 1.0167,
      "step": 2686
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2302380822585077,
      "learning_rate": 0.00010644105481973674,
      "loss": 1.1123,
      "step": 2687
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.20090165737213253,
      "learning_rate": 0.00010638149689259613,
      "loss": 1.0459,
      "step": 2688
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.20831276271771057,
      "learning_rate": 0.00010632193669250207,
      "loss": 1.0432,
      "step": 2689
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.19102553654606144,
      "learning_rate": 0.00010626237424066872,
      "loss": 1.0568,
      "step": 2690
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1854669862192371,
      "learning_rate": 0.00010620280955831087,
      "loss": 0.98,
      "step": 2691
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.234459253185243,
      "learning_rate": 0.00010614324266664428,
      "loss": 0.9888,
      "step": 2692
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.20769762089705296,
      "learning_rate": 0.00010608367358688536,
      "loss": 1.1249,
      "step": 2693
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.17157421107495952,
      "learning_rate": 0.00010602410234025136,
      "loss": 1.1307,
      "step": 2694
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.20063961019637958,
      "learning_rate": 0.00010596452894796028,
      "loss": 1.0931,
      "step": 2695
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.22214127735139458,
      "learning_rate": 0.0001059049534312309,
      "loss": 1.1084,
      "step": 2696
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.16745569611184993,
      "learning_rate": 0.00010584537581128276,
      "loss": 1.076,
      "step": 2697
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.18138847553259604,
      "learning_rate": 0.00010578579610933612,
      "loss": 0.9899,
      "step": 2698
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.19159020272175403,
      "learning_rate": 0.000105726214346612,
      "loss": 0.9575,
      "step": 2699
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.21477105689904769,
      "learning_rate": 0.00010566663054433223,
      "loss": 0.9792,
      "step": 2700
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.20045135392881525,
      "learning_rate": 0.00010560704472371919,
      "loss": 1.143,
      "step": 2701
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.20065085097821775,
      "learning_rate": 0.00010554745690599612,
      "loss": 1.0601,
      "step": 2702
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1745029308636996,
      "learning_rate": 0.00010548786711238697,
      "loss": 1.209,
      "step": 2703
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2193306172201266,
      "learning_rate": 0.00010542827536411632,
      "loss": 1.1522,
      "step": 2704
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.20762466819141692,
      "learning_rate": 0.00010536868168240949,
      "loss": 0.9475,
      "step": 2705
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1959644826486763,
      "learning_rate": 0.00010530908608849246,
      "loss": 0.9438,
      "step": 2706
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2921231523493975,
      "learning_rate": 0.00010524948860359193,
      "loss": 1.0406,
      "step": 2707
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.20805068061545207,
      "learning_rate": 0.0001051898892489353,
      "loss": 1.1527,
      "step": 2708
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2232965978334706,
      "learning_rate": 0.00010513028804575049,
      "loss": 1.0428,
      "step": 2709
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.19275470155119148,
      "learning_rate": 0.00010507068501526627,
      "loss": 0.9299,
      "step": 2710
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1974935745830824,
      "learning_rate": 0.00010501108017871192,
      "loss": 1.1142,
      "step": 2711
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.17264792087382685,
      "learning_rate": 0.0001049514735573175,
      "loss": 1.042,
      "step": 2712
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.18156390435409503,
      "learning_rate": 0.00010489186517231348,
      "loss": 1.0878,
      "step": 2713
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.20399296529418304,
      "learning_rate": 0.00010483225504493122,
      "loss": 1.0816,
      "step": 2714
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.22140648762154747,
      "learning_rate": 0.00010477264319640252,
      "loss": 0.9818,
      "step": 2715
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.20407406659807878,
      "learning_rate": 0.00010471302964795988,
      "loss": 0.948,
      "step": 2716
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2100693051105504,
      "learning_rate": 0.00010465341442083635,
      "loss": 1.0612,
      "step": 2717
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1984266489381907,
      "learning_rate": 0.00010459379753626562,
      "loss": 0.9216,
      "step": 2718
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.16275433594004612,
      "learning_rate": 0.00010453417901548195,
      "loss": 1.0311,
      "step": 2719
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.18284071621657,
      "learning_rate": 0.00010447455887972022,
      "loss": 1.2197,
      "step": 2720
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.19539452240767985,
      "learning_rate": 0.0001044149371502158,
      "loss": 1.1046,
      "step": 2721
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2575648110080922,
      "learning_rate": 0.0001043553138482047,
      "loss": 1.1987,
      "step": 2722
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1966838978372754,
      "learning_rate": 0.00010429568899492348,
      "loss": 1.0152,
      "step": 2723
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2642331319374317,
      "learning_rate": 0.00010423606261160928,
      "loss": 1.1011,
      "step": 2724
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1957340158490089,
      "learning_rate": 0.00010417643471949967,
      "loss": 1.0613,
      "step": 2725
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2873641975654169,
      "learning_rate": 0.00010411680533983288,
      "loss": 1.0705,
      "step": 2726
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.22894734747604945,
      "learning_rate": 0.00010405717449384766,
      "loss": 0.9465,
      "step": 2727
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2267891542874127,
      "learning_rate": 0.00010399754220278319,
      "loss": 1.1452,
      "step": 2728
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2079884803455379,
      "learning_rate": 0.00010393790848787928,
      "loss": 1.1325,
      "step": 2729
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2052365783566343,
      "learning_rate": 0.00010387827337037612,
      "loss": 1.0912,
      "step": 2730
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.21042496456117693,
      "learning_rate": 0.0001038186368715145,
      "loss": 1.1265,
      "step": 2731
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.21249415793372892,
      "learning_rate": 0.00010375899901253572,
      "loss": 0.9923,
      "step": 2732
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.21186517843835356,
      "learning_rate": 0.00010369935981468144,
      "loss": 1.1316,
      "step": 2733
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.19659272626028515,
      "learning_rate": 0.00010363971929919391,
      "loss": 1.1692,
      "step": 2734
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.20528280130988638,
      "learning_rate": 0.00010358007748731582,
      "loss": 1.1258,
      "step": 2735
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.20285230304931443,
      "learning_rate": 0.00010352043440029033,
      "loss": 1.0926,
      "step": 2736
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.23827829213110044,
      "learning_rate": 0.00010346079005936102,
      "loss": 1.1181,
      "step": 2737
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.17572400760331186,
      "learning_rate": 0.00010340114448577189,
      "loss": 1.0623,
      "step": 2738
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.20860993465851002,
      "learning_rate": 0.00010334149770076747,
      "loss": 1.1096,
      "step": 2739
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.22513532210768972,
      "learning_rate": 0.00010328184972559267,
      "loss": 0.9908,
      "step": 2740
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.15729352197294483,
      "learning_rate": 0.00010322220058149286,
      "loss": 0.9656,
      "step": 2741
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.19808189180492897,
      "learning_rate": 0.00010316255028971374,
      "loss": 0.9463,
      "step": 2742
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.21900262434406542,
      "learning_rate": 0.00010310289887150149,
      "loss": 1.0806,
      "step": 2743
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1818181706306018,
      "learning_rate": 0.00010304324634810268,
      "loss": 0.9896,
      "step": 2744
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2172121619124102,
      "learning_rate": 0.0001029835927407643,
      "loss": 0.9601,
      "step": 2745
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.21702616447103854,
      "learning_rate": 0.00010292393807073366,
      "loss": 0.9647,
      "step": 2746
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.18873526375912825,
      "learning_rate": 0.00010286428235925849,
      "loss": 1.1053,
      "step": 2747
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.197019794187196,
      "learning_rate": 0.0001028046256275869,
      "loss": 1.0151,
      "step": 2748
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.19990366913838006,
      "learning_rate": 0.00010274496789696734,
      "loss": 0.9413,
      "step": 2749
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.18424315408471312,
      "learning_rate": 0.00010268530918864863,
      "loss": 1.1184,
      "step": 2750
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.24345291911234115,
      "learning_rate": 0.0001026256495238799,
      "loss": 0.9768,
      "step": 2751
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.22330656270520383,
      "learning_rate": 0.00010256598892391069,
      "loss": 0.9545,
      "step": 2752
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.20484581403911353,
      "learning_rate": 0.00010250632740999084,
      "loss": 0.9019,
      "step": 2753
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.17631878253415142,
      "learning_rate": 0.00010244666500337046,
      "loss": 0.988,
      "step": 2754
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.19230568742091655,
      "learning_rate": 0.00010238700172530009,
      "loss": 1.0489,
      "step": 2755
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2662543442863366,
      "learning_rate": 0.00010232733759703047,
      "loss": 0.9885,
      "step": 2756
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.22745098603042369,
      "learning_rate": 0.00010226767263981276,
      "loss": 1.2417,
      "step": 2757
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.17426197463446907,
      "learning_rate": 0.00010220800687489831,
      "loss": 1.1303,
      "step": 2758
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.20249362085673808,
      "learning_rate": 0.00010214834032353877,
      "loss": 1.0221,
      "step": 2759
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1928600425242116,
      "learning_rate": 0.00010208867300698612,
      "loss": 1.1376,
      "step": 2760
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.20125209076374262,
      "learning_rate": 0.00010202900494649259,
      "loss": 1.0773,
      "step": 2761
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.18017401996590188,
      "learning_rate": 0.00010196933616331067,
      "loss": 1.1008,
      "step": 2762
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.19104070629120856,
      "learning_rate": 0.0001019096666786931,
      "loss": 1.1396,
      "step": 2763
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.19764082449790749,
      "learning_rate": 0.00010184999651389291,
      "loss": 1.2003,
      "step": 2764
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.19984999067432654,
      "learning_rate": 0.00010179032569016334,
      "loss": 1.1262,
      "step": 2765
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2737041839016041,
      "learning_rate": 0.00010173065422875783,
      "loss": 0.9041,
      "step": 2766
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1875377182089994,
      "learning_rate": 0.00010167098215093009,
      "loss": 1.2032,
      "step": 2767
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1947180316736362,
      "learning_rate": 0.0001016113094779341,
      "loss": 1.1227,
      "step": 2768
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1802339512596935,
      "learning_rate": 0.00010155163623102398,
      "loss": 1.0604,
      "step": 2769
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2841144950870126,
      "learning_rate": 0.000101491962431454,
      "loss": 1.1511,
      "step": 2770
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.20283016656499706,
      "learning_rate": 0.00010143228810047875,
      "loss": 1.1417,
      "step": 2771
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.19111467235178303,
      "learning_rate": 0.00010137261325935297,
      "loss": 1.039,
      "step": 2772
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.19037214603972924,
      "learning_rate": 0.00010131293792933157,
      "loss": 1.1751,
      "step": 2773
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1734978096438846,
      "learning_rate": 0.00010125326213166959,
      "loss": 1.2604,
      "step": 2774
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.21453641852455696,
      "learning_rate": 0.00010119358588762232,
      "loss": 1.1733,
      "step": 2775
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.23329872308621932,
      "learning_rate": 0.00010113390921844512,
      "loss": 1.0633,
      "step": 2776
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1928916685709993,
      "learning_rate": 0.00010107423214539365,
      "loss": 1.3138,
      "step": 2777
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.21418475931252418,
      "learning_rate": 0.0001010145546897235,
      "loss": 1.0044,
      "step": 2778
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2273650392425167,
      "learning_rate": 0.00010095487687269054,
      "loss": 1.0996,
      "step": 2779
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.19035049627759415,
      "learning_rate": 0.0001008951987155508,
      "loss": 1.2178,
      "step": 2780
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.205854443528465,
      "learning_rate": 0.00010083552023956031,
      "loss": 1.0204,
      "step": 2781
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.18907367489851476,
      "learning_rate": 0.00010077584146597529,
      "loss": 1.2302,
      "step": 2782
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2034367977619389,
      "learning_rate": 0.00010071616241605205,
      "loss": 1.0236,
      "step": 2783
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1664326718742048,
      "learning_rate": 0.00010065648311104698,
      "loss": 1.1072,
      "step": 2784
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.20642696491715834,
      "learning_rate": 0.00010059680357221662,
      "loss": 1.0807,
      "step": 2785
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.19085987896715703,
      "learning_rate": 0.00010053712382081751,
      "loss": 1.1852,
      "step": 2786
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1975609035274681,
      "learning_rate": 0.00010047744387810632,
      "loss": 1.1096,
      "step": 2787
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.195160015472382,
      "learning_rate": 0.00010041776376533978,
      "loss": 1.0604,
      "step": 2788
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.16671537337537917,
      "learning_rate": 0.00010035808350377469,
      "loss": 1.2377,
      "step": 2789
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.17300781846482002,
      "learning_rate": 0.00010029840311466787,
      "loss": 1.0826,
      "step": 2790
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1994121243314807,
      "learning_rate": 0.00010023872261927621,
      "loss": 0.9525,
      "step": 2791
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.17298632826301122,
      "learning_rate": 0.00010017904203885661,
      "loss": 1.0569,
      "step": 2792
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.24057210349885075,
      "learning_rate": 0.00010011936139466607,
      "loss": 0.9519,
      "step": 2793
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2272676137510552,
      "learning_rate": 0.00010005968070796154,
      "loss": 0.8874,
      "step": 2794
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.200884145521064,
      "learning_rate": 0.0001,
      "loss": 1.1609,
      "step": 2795
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2151949928164997,
      "learning_rate": 9.99403192920385e-05,
      "loss": 1.0245,
      "step": 2796
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.18205232563452253,
      "learning_rate": 9.988063860533395e-05,
      "loss": 1.1563,
      "step": 2797
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1819689089668428,
      "learning_rate": 9.982095796114342e-05,
      "loss": 0.9887,
      "step": 2798
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.34368247751147,
      "learning_rate": 9.976127738072381e-05,
      "loss": 1.2438,
      "step": 2799
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.21566704200577866,
      "learning_rate": 9.970159688533216e-05,
      "loss": 1.0991,
      "step": 2800
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.18820601492420183,
      "learning_rate": 9.964191649622532e-05,
      "loss": 1.0172,
      "step": 2801
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.19785927561146463,
      "learning_rate": 9.958223623466025e-05,
      "loss": 0.9243,
      "step": 2802
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.17947162362205832,
      "learning_rate": 9.952255612189368e-05,
      "loss": 1.119,
      "step": 2803
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.18090366492218876,
      "learning_rate": 9.946287617918252e-05,
      "loss": 1.0366,
      "step": 2804
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.16416980962645508,
      "learning_rate": 9.94031964277834e-05,
      "loss": 0.9865,
      "step": 2805
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2172435881997548,
      "learning_rate": 9.934351688895305e-05,
      "loss": 1.1089,
      "step": 2806
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.22312400390831474,
      "learning_rate": 9.9283837583948e-05,
      "loss": 1.0796,
      "step": 2807
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.19574343788251736,
      "learning_rate": 9.922415853402472e-05,
      "loss": 1.2238,
      "step": 2808
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.19028747764746312,
      "learning_rate": 9.916447976043972e-05,
      "loss": 1.038,
      "step": 2809
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.19485629172378108,
      "learning_rate": 9.910480128444922e-05,
      "loss": 0.9695,
      "step": 2810
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.18996485684137396,
      "learning_rate": 9.904512312730948e-05,
      "loss": 1.2031,
      "step": 2811
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.20652302651221763,
      "learning_rate": 9.898544531027652e-05,
      "loss": 1.0833,
      "step": 2812
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.19771766538452806,
      "learning_rate": 9.892576785460639e-05,
      "loss": 1.2523,
      "step": 2813
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.19578597587663413,
      "learning_rate": 9.886609078155488e-05,
      "loss": 1.0853,
      "step": 2814
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.22259503918395723,
      "learning_rate": 9.880641411237772e-05,
      "loss": 1.1202,
      "step": 2815
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1804630010394837,
      "learning_rate": 9.874673786833041e-05,
      "loss": 1.1789,
      "step": 2816
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.18932051474808423,
      "learning_rate": 9.868706207066846e-05,
      "loss": 1.1085,
      "step": 2817
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.216444775526756,
      "learning_rate": 9.862738674064705e-05,
      "loss": 0.978,
      "step": 2818
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2188260129654793,
      "learning_rate": 9.856771189952126e-05,
      "loss": 1.0772,
      "step": 2819
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1829221807660715,
      "learning_rate": 9.850803756854604e-05,
      "loss": 1.1817,
      "step": 2820
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4584363784357222,
      "learning_rate": 9.844836376897605e-05,
      "loss": 1.0785,
      "step": 2821
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.21131962375955182,
      "learning_rate": 9.838869052206594e-05,
      "loss": 1.173,
      "step": 2822
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.20471640766586638,
      "learning_rate": 9.83290178490699e-05,
      "loss": 0.9613,
      "step": 2823
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1775626136778261,
      "learning_rate": 9.826934577124221e-05,
      "loss": 1.1328,
      "step": 2824
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.16396191586782977,
      "learning_rate": 9.820967430983668e-05,
      "loss": 1.0596,
      "step": 2825
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.24448733804488368,
      "learning_rate": 9.815000348610713e-05,
      "loss": 1.106,
      "step": 2826
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1876356143509577,
      "learning_rate": 9.809033332130693e-05,
      "loss": 0.9115,
      "step": 2827
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1932877243197304,
      "learning_rate": 9.803066383668935e-05,
      "loss": 1.204,
      "step": 2828
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.24683312717380138,
      "learning_rate": 9.797099505350744e-05,
      "loss": 1.0291,
      "step": 2829
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.216785058582211,
      "learning_rate": 9.79113269930139e-05,
      "loss": 1.0908,
      "step": 2830
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.20386654108115032,
      "learning_rate": 9.785165967646127e-05,
      "loss": 1.0332,
      "step": 2831
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.19586067380176614,
      "learning_rate": 9.779199312510171e-05,
      "loss": 0.9949,
      "step": 2832
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1939668410487833,
      "learning_rate": 9.773232736018725e-05,
      "loss": 1.0301,
      "step": 2833
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.23121639674762248,
      "learning_rate": 9.767266240296952e-05,
      "loss": 0.8553,
      "step": 2834
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.30404705645684205,
      "learning_rate": 9.761299827469992e-05,
      "loss": 1.0472,
      "step": 2835
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.20549184063246262,
      "learning_rate": 9.755333499662954e-05,
      "loss": 0.9688,
      "step": 2836
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.18966001974590488,
      "learning_rate": 9.749367259000918e-05,
      "loss": 1.1453,
      "step": 2837
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.24702650084202368,
      "learning_rate": 9.743401107608933e-05,
      "loss": 1.1944,
      "step": 2838
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.288119525659629,
      "learning_rate": 9.737435047612012e-05,
      "loss": 1.0715,
      "step": 2839
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.21701418737309489,
      "learning_rate": 9.731469081135141e-05,
      "loss": 1.0472,
      "step": 2840
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.20977877847012363,
      "learning_rate": 9.725503210303266e-05,
      "loss": 1.0985,
      "step": 2841
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.20223018634267687,
      "learning_rate": 9.719537437241312e-05,
      "loss": 0.9704,
      "step": 2842
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1802813931141754,
      "learning_rate": 9.713571764074152e-05,
      "loss": 1.1535,
      "step": 2843
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.19673584596888147,
      "learning_rate": 9.707606192926637e-05,
      "loss": 1.0775,
      "step": 2844
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.19711122561512684,
      "learning_rate": 9.70164072592357e-05,
      "loss": 1.1467,
      "step": 2845
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2392777485636965,
      "learning_rate": 9.695675365189734e-05,
      "loss": 1.0721,
      "step": 2846
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.22681264583674346,
      "learning_rate": 9.689710112849852e-05,
      "loss": 1.0864,
      "step": 2847
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.23980114983768294,
      "learning_rate": 9.68374497102863e-05,
      "loss": 1.0327,
      "step": 2848
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.18931204055600187,
      "learning_rate": 9.677779941850719e-05,
      "loss": 0.8208,
      "step": 2849
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.20370104324281071,
      "learning_rate": 9.671815027440734e-05,
      "loss": 1.0897,
      "step": 2850
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.20206934561648535,
      "learning_rate": 9.665850229923258e-05,
      "loss": 1.0773,
      "step": 2851
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.19127640805339505,
      "learning_rate": 9.659885551422814e-05,
      "loss": 1.2131,
      "step": 2852
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.19583033223052787,
      "learning_rate": 9.653920994063903e-05,
      "loss": 1.0776,
      "step": 2853
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.18203750959290071,
      "learning_rate": 9.647956559970968e-05,
      "loss": 0.9674,
      "step": 2854
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.18205507734830892,
      "learning_rate": 9.641992251268419e-05,
      "loss": 1.0551,
      "step": 2855
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.20930252585749787,
      "learning_rate": 9.636028070080608e-05,
      "loss": 1.0186,
      "step": 2856
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.18688372524153132,
      "learning_rate": 9.630064018531858e-05,
      "loss": 0.8997,
      "step": 2857
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.15689818989846466,
      "learning_rate": 9.62410009874643e-05,
      "loss": 0.8372,
      "step": 2858
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.17529766473023467,
      "learning_rate": 9.618136312848551e-05,
      "loss": 1.1093,
      "step": 2859
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.19316861064782942,
      "learning_rate": 9.612172662962393e-05,
      "loss": 0.9816,
      "step": 2860
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1845530760838809,
      "learning_rate": 9.606209151212076e-05,
      "loss": 1.0164,
      "step": 2861
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2089411689712405,
      "learning_rate": 9.600245779721682e-05,
      "loss": 1.0652,
      "step": 2862
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.16898513238795246,
      "learning_rate": 9.594282550615235e-05,
      "loss": 1.2056,
      "step": 2863
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.18299865787972283,
      "learning_rate": 9.588319466016713e-05,
      "loss": 1.2044,
      "step": 2864
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.19755083058682066,
      "learning_rate": 9.582356528050034e-05,
      "loss": 1.0683,
      "step": 2865
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1821422520374318,
      "learning_rate": 9.576393738839076e-05,
      "loss": 1.0969,
      "step": 2866
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.20952771025560235,
      "learning_rate": 9.570431100507651e-05,
      "loss": 1.1699,
      "step": 2867
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1817709057571404,
      "learning_rate": 9.564468615179533e-05,
      "loss": 1.0624,
      "step": 2868
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2143923509382737,
      "learning_rate": 9.558506284978422e-05,
      "loss": 1.1947,
      "step": 2869
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2197241684595144,
      "learning_rate": 9.552544112027981e-05,
      "loss": 1.2282,
      "step": 2870
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2044265139292913,
      "learning_rate": 9.546582098451809e-05,
      "loss": 1.1061,
      "step": 2871
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.22239847163024717,
      "learning_rate": 9.540620246373439e-05,
      "loss": 1.1786,
      "step": 2872
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.22181258837358592,
      "learning_rate": 9.534658557916368e-05,
      "loss": 1.1067,
      "step": 2873
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2156728652317449,
      "learning_rate": 9.528697035204016e-05,
      "loss": 1.0797,
      "step": 2874
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2832238192995393,
      "learning_rate": 9.52273568035975e-05,
      "loss": 1.068,
      "step": 2875
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.18100537426533767,
      "learning_rate": 9.516774495506879e-05,
      "loss": 1.0849,
      "step": 2876
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.17928060792106384,
      "learning_rate": 9.510813482768653e-05,
      "loss": 1.0942,
      "step": 2877
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1780380010754709,
      "learning_rate": 9.504852644268254e-05,
      "loss": 1.098,
      "step": 2878
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.15609043521546168,
      "learning_rate": 9.498891982128809e-05,
      "loss": 1.1741,
      "step": 2879
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.17280311404979734,
      "learning_rate": 9.492931498473377e-05,
      "loss": 1.0195,
      "step": 2880
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2513135985518441,
      "learning_rate": 9.486971195424952e-05,
      "loss": 0.9864,
      "step": 2881
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.22831653901006727,
      "learning_rate": 9.481011075106475e-05,
      "loss": 0.9932,
      "step": 2882
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.20564120602148334,
      "learning_rate": 9.475051139640809e-05,
      "loss": 1.1601,
      "step": 2883
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1834523693159964,
      "learning_rate": 9.469091391150758e-05,
      "loss": 0.9655,
      "step": 2884
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.22607440031835216,
      "learning_rate": 9.463131831759053e-05,
      "loss": 1.0682,
      "step": 2885
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.20222807540647483,
      "learning_rate": 9.457172463588371e-05,
      "loss": 1.0525,
      "step": 2886
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.17770709004303026,
      "learning_rate": 9.451213288761304e-05,
      "loss": 1.0607,
      "step": 2887
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.180499714599528,
      "learning_rate": 9.445254309400389e-05,
      "loss": 1.0932,
      "step": 2888
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.20351763058024883,
      "learning_rate": 9.439295527628081e-05,
      "loss": 1.1301,
      "step": 2889
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2239235370233788,
      "learning_rate": 9.43333694556678e-05,
      "loss": 1.0639,
      "step": 2890
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.18347120443814666,
      "learning_rate": 9.4273785653388e-05,
      "loss": 1.2024,
      "step": 2891
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1862831702291195,
      "learning_rate": 9.42142038906639e-05,
      "loss": 1.0133,
      "step": 2892
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2109986899089801,
      "learning_rate": 9.41546241887173e-05,
      "loss": 1.0835,
      "step": 2893
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.16622239773504324,
      "learning_rate": 9.40950465687691e-05,
      "loss": 1.0304,
      "step": 2894
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.22620212627962955,
      "learning_rate": 9.403547105203974e-05,
      "loss": 1.0776,
      "step": 2895
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.17248421808129785,
      "learning_rate": 9.397589765974865e-05,
      "loss": 1.1352,
      "step": 2896
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.18144705395492025,
      "learning_rate": 9.391632641311466e-05,
      "loss": 1.2199,
      "step": 2897
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.18242166767770368,
      "learning_rate": 9.385675733335573e-05,
      "loss": 1.0456,
      "step": 2898
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.18531214868236665,
      "learning_rate": 9.379719044168915e-05,
      "loss": 1.1436,
      "step": 2899
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.18896322552439498,
      "learning_rate": 9.373762575933129e-05,
      "loss": 0.9723,
      "step": 2900
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.16644024251924996,
      "learning_rate": 9.367806330749794e-05,
      "loss": 1.048,
      "step": 2901
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.17853274812771852,
      "learning_rate": 9.36185031074039e-05,
      "loss": 1.1908,
      "step": 2902
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.21965070571225714,
      "learning_rate": 9.355894518026328e-05,
      "loss": 1.1216,
      "step": 2903
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.21926727683538771,
      "learning_rate": 9.349938954728934e-05,
      "loss": 0.9294,
      "step": 2904
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.17986546917936466,
      "learning_rate": 9.343983622969448e-05,
      "loss": 1.0989,
      "step": 2905
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.22363566122476594,
      "learning_rate": 9.338028524869041e-05,
      "loss": 1.0195,
      "step": 2906
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1921302222565514,
      "learning_rate": 9.332073662548784e-05,
      "loss": 1.1712,
      "step": 2907
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.20129114460517036,
      "learning_rate": 9.326119038129681e-05,
      "loss": 1.0616,
      "step": 2908
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.22319610809943072,
      "learning_rate": 9.320164653732634e-05,
      "loss": 0.9788,
      "step": 2909
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.18917646116322026,
      "learning_rate": 9.314210511478474e-05,
      "loss": 1.1099,
      "step": 2910
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.24247052408997857,
      "learning_rate": 9.308256613487937e-05,
      "loss": 1.0895,
      "step": 2911
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.21255848409669084,
      "learning_rate": 9.302302961881676e-05,
      "loss": 1.0948,
      "step": 2912
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2102302545729101,
      "learning_rate": 9.296349558780255e-05,
      "loss": 1.1942,
      "step": 2913
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.18781460249009943,
      "learning_rate": 9.290396406304145e-05,
      "loss": 0.9653,
      "step": 2914
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.18649485179417344,
      "learning_rate": 9.28444350657374e-05,
      "loss": 1.1127,
      "step": 2915
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1852500805492639,
      "learning_rate": 9.27849086170933e-05,
      "loss": 1.1454,
      "step": 2916
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.16853450770474324,
      "learning_rate": 9.272538473831126e-05,
      "loss": 1.184,
      "step": 2917
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.18984408080132015,
      "learning_rate": 9.266586345059234e-05,
      "loss": 1.1036,
      "step": 2918
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2103776986515545,
      "learning_rate": 9.260634477513683e-05,
      "loss": 1.111,
      "step": 2919
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.18379331993016756,
      "learning_rate": 9.254682873314399e-05,
      "loss": 1.1245,
      "step": 2920
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.24000740829928557,
      "learning_rate": 9.248731534581218e-05,
      "loss": 1.0162,
      "step": 2921
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.24689286121069315,
      "learning_rate": 9.242780463433875e-05,
      "loss": 0.9618,
      "step": 2922
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.26249925585335143,
      "learning_rate": 9.236829661992023e-05,
      "loss": 1.1511,
      "step": 2923
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.19109383543849473,
      "learning_rate": 9.230879132375207e-05,
      "loss": 1.077,
      "step": 2924
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.17418881476472808,
      "learning_rate": 9.224928876702874e-05,
      "loss": 1.132,
      "step": 2925
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1841157490482925,
      "learning_rate": 9.218978897094387e-05,
      "loss": 0.997,
      "step": 2926
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.18547234646986913,
      "learning_rate": 9.213029195668996e-05,
      "loss": 1.0452,
      "step": 2927
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.19003220916576127,
      "learning_rate": 9.207079774545863e-05,
      "loss": 1.0648,
      "step": 2928
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.17428301775352123,
      "learning_rate": 9.201130635844038e-05,
      "loss": 1.1269,
      "step": 2929
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.18719169829424273,
      "learning_rate": 9.195181781682485e-05,
      "loss": 1.0726,
      "step": 2930
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.20920984466244147,
      "learning_rate": 9.189233214180056e-05,
      "loss": 1.0908,
      "step": 2931
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1928800873708427,
      "learning_rate": 9.183284935455507e-05,
      "loss": 1.1213,
      "step": 2932
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.19092204274847227,
      "learning_rate": 9.177336947627486e-05,
      "loss": 1.261,
      "step": 2933
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.22912035484204948,
      "learning_rate": 9.171389252814534e-05,
      "loss": 1.02,
      "step": 2934
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1793574283421613,
      "learning_rate": 9.165441853135104e-05,
      "loss": 1.0667,
      "step": 2935
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.20406473836086209,
      "learning_rate": 9.159494750707526e-05,
      "loss": 1.0462,
      "step": 2936
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.19129417822372696,
      "learning_rate": 9.153547947650036e-05,
      "loss": 0.9273,
      "step": 2937
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.18102222237549914,
      "learning_rate": 9.147601446080751e-05,
      "loss": 0.7723,
      "step": 2938
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1987869959666654,
      "learning_rate": 9.141655248117698e-05,
      "loss": 1.0996,
      "step": 2939
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.18776369043273905,
      "learning_rate": 9.135709355878779e-05,
      "loss": 0.9316,
      "step": 2940
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.21813425190138513,
      "learning_rate": 9.129763771481802e-05,
      "loss": 1.0988,
      "step": 2941
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2215580471177792,
      "learning_rate": 9.123818497044446e-05,
      "loss": 1.0804,
      "step": 2942
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.23385350355043397,
      "learning_rate": 9.117873534684303e-05,
      "loss": 1.0127,
      "step": 2943
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.19585980849981952,
      "learning_rate": 9.111928886518837e-05,
      "loss": 1.0533,
      "step": 2944
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1782097274047375,
      "learning_rate": 9.105984554665404e-05,
      "loss": 1.0241,
      "step": 2945
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.20111246042406195,
      "learning_rate": 9.10004054124125e-05,
      "loss": 1.151,
      "step": 2946
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.20539052845278613,
      "learning_rate": 9.094096848363502e-05,
      "loss": 1.0918,
      "step": 2947
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1746518449748338,
      "learning_rate": 9.088153478149189e-05,
      "loss": 0.9948,
      "step": 2948
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.17723375794516266,
      "learning_rate": 9.082210432715197e-05,
      "loss": 0.9018,
      "step": 2949
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1876474063615073,
      "learning_rate": 9.076267714178323e-05,
      "loss": 1.0981,
      "step": 2950
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.19233753186959424,
      "learning_rate": 9.070325324655234e-05,
      "loss": 1.0876,
      "step": 2951
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.17010960526671073,
      "learning_rate": 9.064383266262485e-05,
      "loss": 1.2556,
      "step": 2952
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.199134667698522,
      "learning_rate": 9.058441541116503e-05,
      "loss": 1.0871,
      "step": 2953
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1888532963023887,
      "learning_rate": 9.052500151333615e-05,
      "loss": 0.9505,
      "step": 2954
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.23270093137775028,
      "learning_rate": 9.046559099030012e-05,
      "loss": 0.9868,
      "step": 2955
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.19437711186723106,
      "learning_rate": 9.040618386321768e-05,
      "loss": 1.0877,
      "step": 2956
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.18363313814416826,
      "learning_rate": 9.034678015324842e-05,
      "loss": 1.1768,
      "step": 2957
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1935609679621308,
      "learning_rate": 9.028737988155065e-05,
      "loss": 0.9171,
      "step": 2958
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.20239675833812237,
      "learning_rate": 9.022798306928153e-05,
      "loss": 0.884,
      "step": 2959
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.20512249221981887,
      "learning_rate": 9.016858973759689e-05,
      "loss": 1.0233,
      "step": 2960
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2029691113507339,
      "learning_rate": 9.010919990765141e-05,
      "loss": 1.0988,
      "step": 2961
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2124321106679958,
      "learning_rate": 9.004981360059844e-05,
      "loss": 0.9535,
      "step": 2962
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.316738212461984,
      "learning_rate": 8.999043083759017e-05,
      "loss": 1.0878,
      "step": 2963
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.17715833502748238,
      "learning_rate": 8.993105163977745e-05,
      "loss": 1.0743,
      "step": 2964
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.19019478222282057,
      "learning_rate": 8.987167602830989e-05,
      "loss": 0.9804,
      "step": 2965
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.21174583799833296,
      "learning_rate": 8.981230402433583e-05,
      "loss": 1.1567,
      "step": 2966
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.19635586595419696,
      "learning_rate": 8.975293564900226e-05,
      "loss": 1.0446,
      "step": 2967
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.19688279068557918,
      "learning_rate": 8.9693570923455e-05,
      "loss": 1.0646,
      "step": 2968
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.19838620974111726,
      "learning_rate": 8.963420986883848e-05,
      "loss": 0.8439,
      "step": 2969
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.18691964611962955,
      "learning_rate": 8.957485250629584e-05,
      "loss": 0.9429,
      "step": 2970
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.16742275073195392,
      "learning_rate": 8.951549885696889e-05,
      "loss": 1.2675,
      "step": 2971
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.24356142174914447,
      "learning_rate": 8.945614894199821e-05,
      "loss": 0.9248,
      "step": 2972
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.20794798324309585,
      "learning_rate": 8.939680278252287e-05,
      "loss": 1.0118,
      "step": 2973
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.20609056232282413,
      "learning_rate": 8.93374603996808e-05,
      "loss": 1.0189,
      "step": 2974
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.18019615939697692,
      "learning_rate": 8.927812181460843e-05,
      "loss": 1.0663,
      "step": 2975
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2665975369918319,
      "learning_rate": 8.921878704844096e-05,
      "loss": 0.8615,
      "step": 2976
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.21524390640827087,
      "learning_rate": 8.915945612231214e-05,
      "loss": 1.1293,
      "step": 2977
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1822284989186338,
      "learning_rate": 8.910012905735435e-05,
      "loss": 1.0701,
      "step": 2978
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1958358236598516,
      "learning_rate": 8.904080587469868e-05,
      "loss": 1.1779,
      "step": 2979
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.21507122658854455,
      "learning_rate": 8.898148659547476e-05,
      "loss": 0.9588,
      "step": 2980
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.19451685702065458,
      "learning_rate": 8.892217124081088e-05,
      "loss": 1.092,
      "step": 2981
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.19278004485297376,
      "learning_rate": 8.886285983183386e-05,
      "loss": 1.1317,
      "step": 2982
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.23077298679366348,
      "learning_rate": 8.880355238966923e-05,
      "loss": 1.1177,
      "step": 2983
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1978853327408698,
      "learning_rate": 8.874424893544098e-05,
      "loss": 1.08,
      "step": 2984
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1931415957979613,
      "learning_rate": 8.868494949027176e-05,
      "loss": 1.0792,
      "step": 2985
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.20372749579186816,
      "learning_rate": 8.862565407528275e-05,
      "loss": 1.1385,
      "step": 2986
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.18810902975512445,
      "learning_rate": 8.856636271159377e-05,
      "loss": 1.0814,
      "step": 2987
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1759081994998572,
      "learning_rate": 8.85070754203231e-05,
      "loss": 1.0875,
      "step": 2988
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2128698435393099,
      "learning_rate": 8.844779222258761e-05,
      "loss": 1.1175,
      "step": 2989
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.21157256379479356,
      "learning_rate": 8.838851313950275e-05,
      "loss": 1.0297,
      "step": 2990
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.25462440787480745,
      "learning_rate": 8.832923819218238e-05,
      "loss": 0.889,
      "step": 2991
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.18988780719338955,
      "learning_rate": 8.826996740173908e-05,
      "loss": 1.1732,
      "step": 2992
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.20998582750317546,
      "learning_rate": 8.821070078928379e-05,
      "loss": 0.9908,
      "step": 2993
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.20318634638433347,
      "learning_rate": 8.815143837592604e-05,
      "loss": 1.1797,
      "step": 2994
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1799590142710945,
      "learning_rate": 8.809218018277378e-05,
      "loss": 1.0796,
      "step": 2995
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.19909424612424043,
      "learning_rate": 8.803292623093363e-05,
      "loss": 1.101,
      "step": 2996
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2866538148515401,
      "learning_rate": 8.797367654151052e-05,
      "loss": 0.9387,
      "step": 2997
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.19902654769389883,
      "learning_rate": 8.791443113560787e-05,
      "loss": 1.0645,
      "step": 2998
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1811081408977768,
      "learning_rate": 8.785519003432774e-05,
      "loss": 0.9192,
      "step": 2999
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.20391366306585829,
      "learning_rate": 8.779595325877047e-05,
      "loss": 0.9478,
      "step": 3000
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1935972910433075,
      "learning_rate": 8.7736720830035e-05,
      "loss": 1.2335,
      "step": 3001
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.18897346166250942,
      "learning_rate": 8.767749276921859e-05,
      "loss": 1.1779,
      "step": 3002
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.21323984764522544,
      "learning_rate": 8.761826909741709e-05,
      "loss": 1.1421,
      "step": 3003
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.17606592055000816,
      "learning_rate": 8.755904983572466e-05,
      "loss": 1.0442,
      "step": 3004
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.16781464008210642,
      "learning_rate": 8.7499835005234e-05,
      "loss": 1.22,
      "step": 3005
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.19199983002721158,
      "learning_rate": 8.744062462703608e-05,
      "loss": 1.0204,
      "step": 3006
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.17692094915480808,
      "learning_rate": 8.738141872222048e-05,
      "loss": 0.9654,
      "step": 3007
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.19194903869948116,
      "learning_rate": 8.732221731187501e-05,
      "loss": 1.0166,
      "step": 3008
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.18393460003177403,
      "learning_rate": 8.726302041708597e-05,
      "loss": 0.8403,
      "step": 3009
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.17219878707696212,
      "learning_rate": 8.720382805893805e-05,
      "loss": 0.9598,
      "step": 3010
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2590363647011609,
      "learning_rate": 8.714464025851427e-05,
      "loss": 1.0194,
      "step": 3011
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.18976663486448317,
      "learning_rate": 8.708545703689613e-05,
      "loss": 1.0833,
      "step": 3012
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.20636019827920782,
      "learning_rate": 8.702627841516338e-05,
      "loss": 1.0673,
      "step": 3013
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2149869579973603,
      "learning_rate": 8.696710441439421e-05,
      "loss": 1.049,
      "step": 3014
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.19411545941828792,
      "learning_rate": 8.690793505566511e-05,
      "loss": 1.2201,
      "step": 3015
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1980732539386561,
      "learning_rate": 8.684877036005101e-05,
      "loss": 1.0827,
      "step": 3016
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.19657462682272656,
      "learning_rate": 8.678961034862505e-05,
      "loss": 1.0105,
      "step": 3017
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2128371407870446,
      "learning_rate": 8.673045504245881e-05,
      "loss": 1.0551,
      "step": 3018
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.19106981041589785,
      "learning_rate": 8.667130446262214e-05,
      "loss": 1.1299,
      "step": 3019
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.20591288011677983,
      "learning_rate": 8.661215863018317e-05,
      "loss": 1.111,
      "step": 3020
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.20174978131792476,
      "learning_rate": 8.655301756620847e-05,
      "loss": 1.0112,
      "step": 3021
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.20938197804240982,
      "learning_rate": 8.649388129176277e-05,
      "loss": 1.0486,
      "step": 3022
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1797651671361572,
      "learning_rate": 8.643474982790919e-05,
      "loss": 1.213,
      "step": 3023
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2300623051403798,
      "learning_rate": 8.637562319570905e-05,
      "loss": 1.0294,
      "step": 3024
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.18604212442286752,
      "learning_rate": 8.631650141622207e-05,
      "loss": 1.2258,
      "step": 3025
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.20551789959702285,
      "learning_rate": 8.625738451050606e-05,
      "loss": 1.1835,
      "step": 3026
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.181934149643112,
      "learning_rate": 8.619827249961733e-05,
      "loss": 1.0305,
      "step": 3027
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.18719139261753834,
      "learning_rate": 8.613916540461024e-05,
      "loss": 1.1938,
      "step": 3028
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.20080530206216313,
      "learning_rate": 8.608006324653753e-05,
      "loss": 1.0813,
      "step": 3029
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1833573982057168,
      "learning_rate": 8.602096604645009e-05,
      "loss": 1.1619,
      "step": 3030
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1988909136884465,
      "learning_rate": 8.596187382539705e-05,
      "loss": 1.0485,
      "step": 3031
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.20310359153216373,
      "learning_rate": 8.590278660442591e-05,
      "loss": 1.078,
      "step": 3032
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.17574145524758072,
      "learning_rate": 8.584370440458219e-05,
      "loss": 1.0835,
      "step": 3033
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.18740041636824784,
      "learning_rate": 8.578462724690977e-05,
      "loss": 1.0832,
      "step": 3034
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.18556802256745109,
      "learning_rate": 8.57255551524506e-05,
      "loss": 1.0314,
      "step": 3035
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.19637219727472416,
      "learning_rate": 8.5666488142245e-05,
      "loss": 0.9929,
      "step": 3036
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.20975365211069547,
      "learning_rate": 8.560742623733131e-05,
      "loss": 1.0343,
      "step": 3037
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.22519191033834057,
      "learning_rate": 8.554836945874618e-05,
      "loss": 0.9343,
      "step": 3038
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.24345051044959493,
      "learning_rate": 8.548931782752429e-05,
      "loss": 1.0477,
      "step": 3039
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.30775645154560605,
      "learning_rate": 8.543027136469868e-05,
      "loss": 1.0429,
      "step": 3040
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1748680720179969,
      "learning_rate": 8.53712300913004e-05,
      "loss": 1.017,
      "step": 3041
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.21027837193544408,
      "learning_rate": 8.531219402835867e-05,
      "loss": 1.0439,
      "step": 3042
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.20320036973511216,
      "learning_rate": 8.525316319690092e-05,
      "loss": 0.9662,
      "step": 3043
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.165594677495696,
      "learning_rate": 8.519413761795261e-05,
      "loss": 1.0656,
      "step": 3044
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.21583774534335445,
      "learning_rate": 8.51351173125375e-05,
      "loss": 0.841,
      "step": 3045
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.18476906656488795,
      "learning_rate": 8.507610230167726e-05,
      "loss": 1.1086,
      "step": 3046
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.19798860316175337,
      "learning_rate": 8.501709260639186e-05,
      "loss": 1.2362,
      "step": 3047
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.17204699943068816,
      "learning_rate": 8.495808824769924e-05,
      "loss": 1.0004,
      "step": 3048
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.20896989537676336,
      "learning_rate": 8.489908924661557e-05,
      "loss": 1.1202,
      "step": 3049
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.20845774514052512,
      "learning_rate": 8.484009562415496e-05,
      "loss": 1.0349,
      "step": 3050
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.18426949520931685,
      "learning_rate": 8.47811074013297e-05,
      "loss": 0.9946,
      "step": 3051
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.17278606728138868,
      "learning_rate": 8.472212459915019e-05,
      "loss": 1.24,
      "step": 3052
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.24031872232394852,
      "learning_rate": 8.46631472386248e-05,
      "loss": 0.9329,
      "step": 3053
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.21926515814465614,
      "learning_rate": 8.460417534076004e-05,
      "loss": 1.1933,
      "step": 3054
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.18739761302238064,
      "learning_rate": 8.454520892656038e-05,
      "loss": 1.0928,
      "step": 3055
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.22587704957180832,
      "learning_rate": 8.448624801702851e-05,
      "loss": 0.9551,
      "step": 3056
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1849894612767889,
      "learning_rate": 8.442729263316497e-05,
      "loss": 1.2719,
      "step": 3057
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1821290320555773,
      "learning_rate": 8.436834279596845e-05,
      "loss": 1.1031,
      "step": 3058
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.21954802704774534,
      "learning_rate": 8.430939852643558e-05,
      "loss": 1.0429,
      "step": 3059
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.20218113211475724,
      "learning_rate": 8.425045984556114e-05,
      "loss": 1.0753,
      "step": 3060
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.20322819700428618,
      "learning_rate": 8.419152677433778e-05,
      "loss": 1.0969,
      "step": 3061
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.24087827359482827,
      "learning_rate": 8.413259933375618e-05,
      "loss": 1.0555,
      "step": 3062
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1873185891276744,
      "learning_rate": 8.407367754480511e-05,
      "loss": 1.0551,
      "step": 3063
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.23581585333396624,
      "learning_rate": 8.401476142847115e-05,
      "loss": 1.0312,
      "step": 3064
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.23830452164460708,
      "learning_rate": 8.39558510057391e-05,
      "loss": 0.9543,
      "step": 3065
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.22057298421861352,
      "learning_rate": 8.38969462975915e-05,
      "loss": 1.0016,
      "step": 3066
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2818590611464797,
      "learning_rate": 8.383804732500902e-05,
      "loss": 1.1883,
      "step": 3067
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2058977386248464,
      "learning_rate": 8.377915410897013e-05,
      "loss": 1.109,
      "step": 3068
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.17929641644793323,
      "learning_rate": 8.372026667045146e-05,
      "loss": 0.9235,
      "step": 3069
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.16341315388370675,
      "learning_rate": 8.366138503042738e-05,
      "loss": 0.9626,
      "step": 3070
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.23344156259674562,
      "learning_rate": 8.36025092098703e-05,
      "loss": 1.1746,
      "step": 3071
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.18345309463242304,
      "learning_rate": 8.354363922975055e-05,
      "loss": 0.9964,
      "step": 3072
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2048576131129775,
      "learning_rate": 8.34847751110363e-05,
      "loss": 1.0134,
      "step": 3073
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.20905297054637934,
      "learning_rate": 8.342591687469381e-05,
      "loss": 1.0762,
      "step": 3074
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1651696812008982,
      "learning_rate": 8.336706454168701e-05,
      "loss": 1.0061,
      "step": 3075
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.19740682996098113,
      "learning_rate": 8.330821813297793e-05,
      "loss": 1.0341,
      "step": 3076
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.24094571419064367,
      "learning_rate": 8.324937766952638e-05,
      "loss": 1.073,
      "step": 3077
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.19447403053009246,
      "learning_rate": 8.319054317229009e-05,
      "loss": 1.0728,
      "step": 3078
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1820690504026213,
      "learning_rate": 8.313171466222461e-05,
      "loss": 1.0778,
      "step": 3079
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.18357620974882827,
      "learning_rate": 8.307289216028348e-05,
      "loss": 1.1979,
      "step": 3080
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.21507600493332288,
      "learning_rate": 8.301407568741799e-05,
      "loss": 0.9069,
      "step": 3081
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.18341997780336353,
      "learning_rate": 8.295526526457732e-05,
      "loss": 1.077,
      "step": 3082
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.18673333132352016,
      "learning_rate": 8.289646091270849e-05,
      "loss": 1.116,
      "step": 3083
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.20577219168477567,
      "learning_rate": 8.28376626527563e-05,
      "loss": 1.1018,
      "step": 3084
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.19574634072191222,
      "learning_rate": 8.277887050566354e-05,
      "loss": 1.1334,
      "step": 3085
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.24252080194522013,
      "learning_rate": 8.272008449237065e-05,
      "loss": 1.0564,
      "step": 3086
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.18409411581376753,
      "learning_rate": 8.266130463381602e-05,
      "loss": 0.9952,
      "step": 3087
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.21813199267477842,
      "learning_rate": 8.260253095093569e-05,
      "loss": 0.9745,
      "step": 3088
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1908857768216001,
      "learning_rate": 8.254376346466371e-05,
      "loss": 1.1244,
      "step": 3089
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.22171479960910712,
      "learning_rate": 8.24850021959317e-05,
      "loss": 1.1684,
      "step": 3090
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1972671122001174,
      "learning_rate": 8.242624716566927e-05,
      "loss": 0.8157,
      "step": 3091
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.20323341188307553,
      "learning_rate": 8.236749839480361e-05,
      "loss": 1.1049,
      "step": 3092
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2614760672784847,
      "learning_rate": 8.230875590425988e-05,
      "loss": 0.8874,
      "step": 3093
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.16870292977504087,
      "learning_rate": 8.225001971496087e-05,
      "loss": 1.0198,
      "step": 3094
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.20244287445204606,
      "learning_rate": 8.219128984782712e-05,
      "loss": 1.206,
      "step": 3095
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.21120844271988556,
      "learning_rate": 8.213256632377702e-05,
      "loss": 1.0533,
      "step": 3096
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2678492276511325,
      "learning_rate": 8.207384916372658e-05,
      "loss": 0.9928,
      "step": 3097
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.17709424309718697,
      "learning_rate": 8.20151383885897e-05,
      "loss": 1.0861,
      "step": 3098
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1920947331229081,
      "learning_rate": 8.195643401927777e-05,
      "loss": 1.1053,
      "step": 3099
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.183228916424188,
      "learning_rate": 8.189773607670015e-05,
      "loss": 1.0577,
      "step": 3100
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.18080716823012272,
      "learning_rate": 8.183904458176374e-05,
      "loss": 1.0974,
      "step": 3101
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.20224698857006607,
      "learning_rate": 8.178035955537325e-05,
      "loss": 0.8841,
      "step": 3102
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1952115054335303,
      "learning_rate": 8.172168101843099e-05,
      "loss": 1.0582,
      "step": 3103
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.20995723819471296,
      "learning_rate": 8.1663008991837e-05,
      "loss": 1.0323,
      "step": 3104
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.20544290928594466,
      "learning_rate": 8.160434349648907e-05,
      "loss": 1.0515,
      "step": 3105
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.17785938116144173,
      "learning_rate": 8.154568455328254e-05,
      "loss": 1.0098,
      "step": 3106
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.19852824205999814,
      "learning_rate": 8.148703218311053e-05,
      "loss": 1.0197,
      "step": 3107
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.16828132704490556,
      "learning_rate": 8.14283864068637e-05,
      "loss": 1.0609,
      "step": 3108
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.17247572531603522,
      "learning_rate": 8.136974724543049e-05,
      "loss": 1.2136,
      "step": 3109
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.17529705889524996,
      "learning_rate": 8.131111471969688e-05,
      "loss": 1.0989,
      "step": 3110
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.20406732006680495,
      "learning_rate": 8.125248885054656e-05,
      "loss": 1.1639,
      "step": 3111
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.18103005791592697,
      "learning_rate": 8.119386965886077e-05,
      "loss": 1.154,
      "step": 3112
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.257287531028331,
      "learning_rate": 8.113525716551851e-05,
      "loss": 1.1002,
      "step": 3113
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.21249934803480122,
      "learning_rate": 8.107665139139622e-05,
      "loss": 1.0203,
      "step": 3114
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1923951420012075,
      "learning_rate": 8.101805235736804e-05,
      "loss": 1.0552,
      "step": 3115
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.21478937777487814,
      "learning_rate": 8.095946008430573e-05,
      "loss": 1.0899,
      "step": 3116
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1954632950530588,
      "learning_rate": 8.090087459307854e-05,
      "loss": 1.1693,
      "step": 3117
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.21708942921693564,
      "learning_rate": 8.084229590455349e-05,
      "loss": 0.94,
      "step": 3118
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1799735011966122,
      "learning_rate": 8.078372403959498e-05,
      "loss": 1.1951,
      "step": 3119
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1659921711563436,
      "learning_rate": 8.072515901906508e-05,
      "loss": 1.1562,
      "step": 3120
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.19612312053072925,
      "learning_rate": 8.06666008638234e-05,
      "loss": 1.0697,
      "step": 3121
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.19383688937437293,
      "learning_rate": 8.060804959472713e-05,
      "loss": 1.1149,
      "step": 3122
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.19467302335504663,
      "learning_rate": 8.054950523263096e-05,
      "loss": 1.1226,
      "step": 3123
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.20445925519583832,
      "learning_rate": 8.049096779838719e-05,
      "loss": 1.134,
      "step": 3124
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2446088019098038,
      "learning_rate": 8.043243731284557e-05,
      "loss": 1.036,
      "step": 3125
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.20249797275644082,
      "learning_rate": 8.03739137968534e-05,
      "loss": 1.0469,
      "step": 3126
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.19449003668949366,
      "learning_rate": 8.031539727125554e-05,
      "loss": 1.0689,
      "step": 3127
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.15804030530021196,
      "learning_rate": 8.025688775689428e-05,
      "loss": 1.0948,
      "step": 3128
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1901406444095396,
      "learning_rate": 8.019838527460957e-05,
      "loss": 1.0513,
      "step": 3129
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1855416533845084,
      "learning_rate": 8.013988984523866e-05,
      "loss": 1.0099,
      "step": 3130
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.19475472318175693,
      "learning_rate": 8.008140148961641e-05,
      "loss": 1.0881,
      "step": 3131
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.19123514111073542,
      "learning_rate": 8.002292022857509e-05,
      "loss": 1.116,
      "step": 3132
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1888697964461829,
      "learning_rate": 7.996444608294456e-05,
      "loss": 1.092,
      "step": 3133
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.23180511689995112,
      "learning_rate": 7.990597907355198e-05,
      "loss": 1.1129,
      "step": 3134
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.17741161156811425,
      "learning_rate": 7.984751922122214e-05,
      "loss": 1.0132,
      "step": 3135
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1908117653746611,
      "learning_rate": 7.978906654677713e-05,
      "loss": 1.1232,
      "step": 3136
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.16362338980620167,
      "learning_rate": 7.973062107103653e-05,
      "loss": 1.0088,
      "step": 3137
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.20113861475102282,
      "learning_rate": 7.967218281481745e-05,
      "loss": 1.0358,
      "step": 3138
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.21202283174893222,
      "learning_rate": 7.96137517989343e-05,
      "loss": 1.0431,
      "step": 3139
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.18348897522567362,
      "learning_rate": 7.955532804419899e-05,
      "loss": 1.1856,
      "step": 3140
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.22528051476110325,
      "learning_rate": 7.949691157142079e-05,
      "loss": 0.9707,
      "step": 3141
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1857507339521687,
      "learning_rate": 7.943850240140644e-05,
      "loss": 0.8772,
      "step": 3142
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1978603215762609,
      "learning_rate": 7.938010055496002e-05,
      "loss": 1.1841,
      "step": 3143
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.21025083679860812,
      "learning_rate": 7.932170605288307e-05,
      "loss": 0.9151,
      "step": 3144
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.22853169490516062,
      "learning_rate": 7.926331891597436e-05,
      "loss": 1.0708,
      "step": 3145
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2025350082239774,
      "learning_rate": 7.920493916503027e-05,
      "loss": 1.1327,
      "step": 3146
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.18224600150122286,
      "learning_rate": 7.914656682084437e-05,
      "loss": 1.0023,
      "step": 3147
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2045808701146211,
      "learning_rate": 7.908820190420763e-05,
      "loss": 1.0043,
      "step": 3148
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.18764735383838269,
      "learning_rate": 7.902984443590844e-05,
      "loss": 1.1809,
      "step": 3149
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.17269598070332332,
      "learning_rate": 7.89714944367324e-05,
      "loss": 1.2413,
      "step": 3150
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.20610064146999188,
      "learning_rate": 7.891315192746267e-05,
      "loss": 1.058,
      "step": 3151
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.21490076099126124,
      "learning_rate": 7.885481692887947e-05,
      "loss": 1.1499,
      "step": 3152
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.19956891902544868,
      "learning_rate": 7.87964894617606e-05,
      "loss": 0.9767,
      "step": 3153
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.19598582172406706,
      "learning_rate": 7.873816954688099e-05,
      "loss": 0.9316,
      "step": 3154
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.20355379635825713,
      "learning_rate": 7.867985720501301e-05,
      "loss": 0.9028,
      "step": 3155
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.18599630168033973,
      "learning_rate": 7.862155245692623e-05,
      "loss": 1.07,
      "step": 3156
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1782894351720859,
      "learning_rate": 7.856325532338753e-05,
      "loss": 1.0225,
      "step": 3157
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1973274115135599,
      "learning_rate": 7.85049658251612e-05,
      "loss": 1.137,
      "step": 3158
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.22194782651751374,
      "learning_rate": 7.844668398300865e-05,
      "loss": 1.0748,
      "step": 3159
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.16919250065941038,
      "learning_rate": 7.838840981768868e-05,
      "loss": 1.128,
      "step": 3160
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2310797890517031,
      "learning_rate": 7.833014334995723e-05,
      "loss": 0.9231,
      "step": 3161
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.30841041921581885,
      "learning_rate": 7.827188460056768e-05,
      "loss": 1.0533,
      "step": 3162
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2699314223991941,
      "learning_rate": 7.821363359027048e-05,
      "loss": 1.135,
      "step": 3163
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.19818675295985788,
      "learning_rate": 7.815539033981344e-05,
      "loss": 1.0387,
      "step": 3164
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.18093120309525512,
      "learning_rate": 7.809715486994152e-05,
      "loss": 1.1657,
      "step": 3165
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.20006325692518415,
      "learning_rate": 7.803892720139701e-05,
      "loss": 1.0564,
      "step": 3166
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.17740226621786787,
      "learning_rate": 7.798070735491936e-05,
      "loss": 1.0525,
      "step": 3167
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2020003171009865,
      "learning_rate": 7.79224953512452e-05,
      "loss": 0.9565,
      "step": 3168
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.17936028935822507,
      "learning_rate": 7.786429121110845e-05,
      "loss": 1.0223,
      "step": 3169
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2587935017908667,
      "learning_rate": 7.780609495524012e-05,
      "loss": 1.1996,
      "step": 3170
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.22012990129108082,
      "learning_rate": 7.774790660436858e-05,
      "loss": 0.9865,
      "step": 3171
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.17639616689449425,
      "learning_rate": 7.76897261792192e-05,
      "loss": 0.9994,
      "step": 3172
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.19152683163766548,
      "learning_rate": 7.763155370051465e-05,
      "loss": 1.0994,
      "step": 3173
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.20650490345539033,
      "learning_rate": 7.757338918897468e-05,
      "loss": 0.943,
      "step": 3174
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.22930963501789214,
      "learning_rate": 7.751523266531634e-05,
      "loss": 1.1259,
      "step": 3175
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.18316496791337866,
      "learning_rate": 7.745708415025363e-05,
      "loss": 1.0121,
      "step": 3176
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.27338467603995975,
      "learning_rate": 7.739894366449794e-05,
      "loss": 0.9672,
      "step": 3177
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2270992093718511,
      "learning_rate": 7.73408112287576e-05,
      "loss": 0.9192,
      "step": 3178
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.18208970448051154,
      "learning_rate": 7.728268686373814e-05,
      "loss": 1.0019,
      "step": 3179
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.21836797374537023,
      "learning_rate": 7.722457059014225e-05,
      "loss": 0.9463,
      "step": 3180
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.17650485951106706,
      "learning_rate": 7.716646242866967e-05,
      "loss": 1.1104,
      "step": 3181
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2045164818017761,
      "learning_rate": 7.710836240001737e-05,
      "loss": 1.1299,
      "step": 3182
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.18312631644556016,
      "learning_rate": 7.705027052487925e-05,
      "loss": 1.2097,
      "step": 3183
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.17572403103046905,
      "learning_rate": 7.699218682394649e-05,
      "loss": 1.1024,
      "step": 3184
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1958414322609213,
      "learning_rate": 7.693411131790717e-05,
      "loss": 0.9981,
      "step": 3185
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.18172840520120823,
      "learning_rate": 7.687604402744663e-05,
      "loss": 1.2195,
      "step": 3186
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.20024879918062796,
      "learning_rate": 7.681798497324716e-05,
      "loss": 1.1234,
      "step": 3187
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.19222727609061907,
      "learning_rate": 7.675993417598821e-05,
      "loss": 1.0208,
      "step": 3188
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.27394289299845764,
      "learning_rate": 7.670189165634621e-05,
      "loss": 0.9992,
      "step": 3189
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.20497597920749955,
      "learning_rate": 7.664385743499462e-05,
      "loss": 1.0125,
      "step": 3190
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2202236412189105,
      "learning_rate": 7.658583153260407e-05,
      "loss": 1.1025,
      "step": 3191
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2584628846196161,
      "learning_rate": 7.65278139698421e-05,
      "loss": 0.995,
      "step": 3192
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.18756974055578993,
      "learning_rate": 7.646980476737338e-05,
      "loss": 1.0579,
      "step": 3193
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1961134818053853,
      "learning_rate": 7.641180394585947e-05,
      "loss": 1.1875,
      "step": 3194
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1674146897094984,
      "learning_rate": 7.635381152595915e-05,
      "loss": 1.0559,
      "step": 3195
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.20433523268787213,
      "learning_rate": 7.629582752832798e-05,
      "loss": 1.0825,
      "step": 3196
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1915146442230961,
      "learning_rate": 7.62378519736187e-05,
      "loss": 1.0215,
      "step": 3197
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.22252160870191795,
      "learning_rate": 7.617988488248089e-05,
      "loss": 1.092,
      "step": 3198
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.22648644446034696,
      "learning_rate": 7.612192627556128e-05,
      "loss": 1.1192,
      "step": 3199
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2113354127875412,
      "learning_rate": 7.606397617350349e-05,
      "loss": 1.1155,
      "step": 3200
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2008986090981962,
      "learning_rate": 7.600603459694799e-05,
      "loss": 1.0599,
      "step": 3201
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2217527628886672,
      "learning_rate": 7.594810156653247e-05,
      "loss": 1.1136,
      "step": 3202
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.20614915471212744,
      "learning_rate": 7.589017710289139e-05,
      "loss": 1.1017,
      "step": 3203
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1947013297782604,
      "learning_rate": 7.583226122665622e-05,
      "loss": 0.9905,
      "step": 3204
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.23502844718861457,
      "learning_rate": 7.577435395845533e-05,
      "loss": 0.9939,
      "step": 3205
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.18262924055927643,
      "learning_rate": 7.571645531891411e-05,
      "loss": 1.0279,
      "step": 3206
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.18342817677267917,
      "learning_rate": 7.56585653286548e-05,
      "loss": 0.9736,
      "step": 3207
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1855882423235474,
      "learning_rate": 7.56006840082966e-05,
      "loss": 1.0271,
      "step": 3208
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1892810009004192,
      "learning_rate": 7.554281137845558e-05,
      "loss": 1.1138,
      "step": 3209
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.19947508325236796,
      "learning_rate": 7.54849474597447e-05,
      "loss": 1.0413,
      "step": 3210
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.20235719031682278,
      "learning_rate": 7.542709227277396e-05,
      "loss": 1.0877,
      "step": 3211
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.19578056794826595,
      "learning_rate": 7.536924583815007e-05,
      "loss": 1.1097,
      "step": 3212
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.18490725140934813,
      "learning_rate": 7.531140817647673e-05,
      "loss": 1.0244,
      "step": 3213
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.18933828811726106,
      "learning_rate": 7.525357930835444e-05,
      "loss": 0.9989,
      "step": 3214
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.19695162819858686,
      "learning_rate": 7.519575925438067e-05,
      "loss": 1.0023,
      "step": 3215
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2000558006060464,
      "learning_rate": 7.513794803514966e-05,
      "loss": 1.2765,
      "step": 3216
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.18887804304760383,
      "learning_rate": 7.508014567125256e-05,
      "loss": 1.0017,
      "step": 3217
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.18379441304266628,
      "learning_rate": 7.502235218327731e-05,
      "loss": 0.949,
      "step": 3218
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.21573056622357753,
      "learning_rate": 7.496456759180875e-05,
      "loss": 1.1293,
      "step": 3219
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.20029260506891308,
      "learning_rate": 7.490679191742853e-05,
      "loss": 1.1133,
      "step": 3220
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.16237254547831764,
      "learning_rate": 7.484902518071506e-05,
      "loss": 0.9821,
      "step": 3221
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1998001932163252,
      "learning_rate": 7.47912674022437e-05,
      "loss": 1.0178,
      "step": 3222
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2023043285380274,
      "learning_rate": 7.473351860258643e-05,
      "loss": 0.9952,
      "step": 3223
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.165818840486047,
      "learning_rate": 7.467577880231227e-05,
      "loss": 0.9145,
      "step": 3224
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.22849320448609742,
      "learning_rate": 7.461804802198681e-05,
      "loss": 1.0815,
      "step": 3225
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.21574885627930962,
      "learning_rate": 7.456032628217262e-05,
      "loss": 1.0132,
      "step": 3226
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2504707403784515,
      "learning_rate": 7.450261360342888e-05,
      "loss": 1.1337,
      "step": 3227
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.24856308288331994,
      "learning_rate": 7.444491000631165e-05,
      "loss": 1.0064,
      "step": 3228
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.20923471569110125,
      "learning_rate": 7.438721551137367e-05,
      "loss": 0.9231,
      "step": 3229
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.19310322477563444,
      "learning_rate": 7.432953013916457e-05,
      "loss": 0.9673,
      "step": 3230
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.19463950688393716,
      "learning_rate": 7.427185391023063e-05,
      "loss": 0.9844,
      "step": 3231
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.21974931540822742,
      "learning_rate": 7.421418684511486e-05,
      "loss": 1.1421,
      "step": 3232
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.20230267081365785,
      "learning_rate": 7.41565289643571e-05,
      "loss": 0.918,
      "step": 3233
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.19494677928118065,
      "learning_rate": 7.409888028849377e-05,
      "loss": 1.2306,
      "step": 3234
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.17040082165066758,
      "learning_rate": 7.404124083805819e-05,
      "loss": 1.0256,
      "step": 3235
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.21748737344417812,
      "learning_rate": 7.398361063358025e-05,
      "loss": 1.1364,
      "step": 3236
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1902656172343539,
      "learning_rate": 7.392598969558667e-05,
      "loss": 1.0866,
      "step": 3237
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.21419274992622206,
      "learning_rate": 7.38683780446007e-05,
      "loss": 1.1135,
      "step": 3238
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3005458851997846,
      "learning_rate": 7.381077570114249e-05,
      "loss": 0.8785,
      "step": 3239
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2197920641094139,
      "learning_rate": 7.375318268572871e-05,
      "loss": 1.0081,
      "step": 3240
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.22663136061209535,
      "learning_rate": 7.369559901887277e-05,
      "loss": 1.0684,
      "step": 3241
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.18196454299173223,
      "learning_rate": 7.363802472108477e-05,
      "loss": 1.0174,
      "step": 3242
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2275948537993288,
      "learning_rate": 7.358045981287141e-05,
      "loss": 1.1407,
      "step": 3243
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.17688542545646613,
      "learning_rate": 7.352290431473612e-05,
      "loss": 1.108,
      "step": 3244
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.17799472912719175,
      "learning_rate": 7.346535824717893e-05,
      "loss": 1.0488,
      "step": 3245
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.21753723506555891,
      "learning_rate": 7.340782163069655e-05,
      "loss": 0.9883,
      "step": 3246
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.18623202628409802,
      "learning_rate": 7.335029448578222e-05,
      "loss": 1.2639,
      "step": 3247
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2753099827685581,
      "learning_rate": 7.329277683292599e-05,
      "loss": 0.9348,
      "step": 3248
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.20814975557028031,
      "learning_rate": 7.323526869261437e-05,
      "loss": 0.8368,
      "step": 3249
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1952841628659988,
      "learning_rate": 7.317777008533054e-05,
      "loss": 1.0869,
      "step": 3250
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.20832345330919613,
      "learning_rate": 7.312028103155426e-05,
      "loss": 0.873,
      "step": 3251
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2110721034843065,
      "learning_rate": 7.306280155176198e-05,
      "loss": 1.1238,
      "step": 3252
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.16258779128359685,
      "learning_rate": 7.30053316664266e-05,
      "loss": 1.0223,
      "step": 3253
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.15904390760611614,
      "learning_rate": 7.294787139601764e-05,
      "loss": 1.0308,
      "step": 3254
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.20758817072117586,
      "learning_rate": 7.289042076100132e-05,
      "loss": 1.1846,
      "step": 3255
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.21410459740326365,
      "learning_rate": 7.283297978184027e-05,
      "loss": 1.1097,
      "step": 3256
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.18250530122785222,
      "learning_rate": 7.277554847899378e-05,
      "loss": 1.064,
      "step": 3257
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.28206451243853375,
      "learning_rate": 7.271812687291759e-05,
      "loss": 1.0628,
      "step": 3258
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2177994578921701,
      "learning_rate": 7.266071498406417e-05,
      "loss": 0.9605,
      "step": 3259
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.24577571201642057,
      "learning_rate": 7.26033128328823e-05,
      "loss": 1.0945,
      "step": 3260
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.19604973840068812,
      "learning_rate": 7.254592043981748e-05,
      "loss": 1.0283,
      "step": 3261
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.20495806244890802,
      "learning_rate": 7.248853782531164e-05,
      "loss": 1.1257,
      "step": 3262
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2013258906632663,
      "learning_rate": 7.243116500980318e-05,
      "loss": 0.9344,
      "step": 3263
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1975575097982678,
      "learning_rate": 7.237380201372719e-05,
      "loss": 0.9797,
      "step": 3264
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.19801826865732608,
      "learning_rate": 7.231644885751507e-05,
      "loss": 0.9857,
      "step": 3265
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.19910095290281501,
      "learning_rate": 7.225910556159483e-05,
      "loss": 1.1289,
      "step": 3266
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.18282833210290106,
      "learning_rate": 7.220177214639088e-05,
      "loss": 1.0625,
      "step": 3267
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2020700210760485,
      "learning_rate": 7.214444863232425e-05,
      "loss": 1.1036,
      "step": 3268
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2650726109527545,
      "learning_rate": 7.208713503981228e-05,
      "loss": 0.9274,
      "step": 3269
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.20715680000161252,
      "learning_rate": 7.202983138926888e-05,
      "loss": 1.0418,
      "step": 3270
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1709874351056194,
      "learning_rate": 7.197253770110438e-05,
      "loss": 1.0141,
      "step": 3271
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1865046400711849,
      "learning_rate": 7.191525399572561e-05,
      "loss": 1.0695,
      "step": 3272
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.19587746686574967,
      "learning_rate": 7.185798029353579e-05,
      "loss": 1.1665,
      "step": 3273
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.16698083840710118,
      "learning_rate": 7.180071661493454e-05,
      "loss": 0.9758,
      "step": 3274
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1941362385916029,
      "learning_rate": 7.174346298031804e-05,
      "loss": 0.9433,
      "step": 3275
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.19143068935741323,
      "learning_rate": 7.168621941007873e-05,
      "loss": 1.1528,
      "step": 3276
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.22435422476358777,
      "learning_rate": 7.16289859246057e-05,
      "loss": 1.1541,
      "step": 3277
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.21073930137533484,
      "learning_rate": 7.157176254428411e-05,
      "loss": 1.0425,
      "step": 3278
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2088794102224457,
      "learning_rate": 7.151454928949585e-05,
      "loss": 1.1867,
      "step": 3279
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1929302061681802,
      "learning_rate": 7.145734618061902e-05,
      "loss": 1.1228,
      "step": 3280
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1923662243427163,
      "learning_rate": 7.140015323802816e-05,
      "loss": 1.1361,
      "step": 3281
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.17686592472157925,
      "learning_rate": 7.134297048209412e-05,
      "loss": 1.053,
      "step": 3282
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.18351708182129933,
      "learning_rate": 7.128579793318428e-05,
      "loss": 0.979,
      "step": 3283
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.19992831460295668,
      "learning_rate": 7.122863561166224e-05,
      "loss": 1.12,
      "step": 3284
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.18618366468162711,
      "learning_rate": 7.117148353788797e-05,
      "loss": 0.9452,
      "step": 3285
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2181663427592499,
      "learning_rate": 7.111434173221786e-05,
      "loss": 1.0592,
      "step": 3286
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1698386220752922,
      "learning_rate": 7.105721021500458e-05,
      "loss": 0.9816,
      "step": 3287
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1936492383884415,
      "learning_rate": 7.100008900659721e-05,
      "loss": 0.8501,
      "step": 3288
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.20968622381438437,
      "learning_rate": 7.094297812734103e-05,
      "loss": 1.0339,
      "step": 3289
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.20132648596470729,
      "learning_rate": 7.088587759757782e-05,
      "loss": 1.0786,
      "step": 3290
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.16952083005169036,
      "learning_rate": 7.082878743764545e-05,
      "loss": 0.9941,
      "step": 3291
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1862042923835141,
      "learning_rate": 7.077170766787835e-05,
      "loss": 1.2529,
      "step": 3292
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.18551514137616784,
      "learning_rate": 7.071463830860703e-05,
      "loss": 1.2455,
      "step": 3293
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.18265966820181012,
      "learning_rate": 7.065757938015843e-05,
      "loss": 1.0544,
      "step": 3294
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.18533093252063693,
      "learning_rate": 7.060053090285572e-05,
      "loss": 1.1056,
      "step": 3295
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.20279303768922927,
      "learning_rate": 7.054349289701828e-05,
      "loss": 0.8,
      "step": 3296
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2079889839405287,
      "learning_rate": 7.048646538296195e-05,
      "loss": 1.1088,
      "step": 3297
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.20073370233362697,
      "learning_rate": 7.042944838099863e-05,
      "loss": 1.0832,
      "step": 3298
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.21922248671727684,
      "learning_rate": 7.037244191143661e-05,
      "loss": 1.0639,
      "step": 3299
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2630248977221537,
      "learning_rate": 7.031544599458036e-05,
      "loss": 1.0995,
      "step": 3300
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2409056255750436,
      "learning_rate": 7.025846065073064e-05,
      "loss": 1.0527,
      "step": 3301
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.22538053508676537,
      "learning_rate": 7.020148590018435e-05,
      "loss": 1.0427,
      "step": 3302
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.20895497012874503,
      "learning_rate": 7.014452176323477e-05,
      "loss": 1.1823,
      "step": 3303
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.20727590363632573,
      "learning_rate": 7.008756826017124e-05,
      "loss": 1.0405,
      "step": 3304
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.20636177250475232,
      "learning_rate": 7.003062541127946e-05,
      "loss": 1.0071,
      "step": 3305
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.19149476409179805,
      "learning_rate": 6.997369323684123e-05,
      "loss": 1.1031,
      "step": 3306
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2196143754834626,
      "learning_rate": 6.991677175713449e-05,
      "loss": 1.0541,
      "step": 3307
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.18562685155004174,
      "learning_rate": 6.98598609924336e-05,
      "loss": 1.063,
      "step": 3308
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2028491392020017,
      "learning_rate": 6.98029609630089e-05,
      "loss": 0.8397,
      "step": 3309
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.18609008484543277,
      "learning_rate": 6.974607168912696e-05,
      "loss": 0.9361,
      "step": 3310
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.18553551228451837,
      "learning_rate": 6.968919319105052e-05,
      "loss": 0.9662,
      "step": 3311
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.19252385495499796,
      "learning_rate": 6.963232548903853e-05,
      "loss": 1.0532,
      "step": 3312
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.22213687437678997,
      "learning_rate": 6.957546860334604e-05,
      "loss": 1.0848,
      "step": 3313
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1784459241508724,
      "learning_rate": 6.951862255422427e-05,
      "loss": 1.1023,
      "step": 3314
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2460049804437254,
      "learning_rate": 6.946178736192053e-05,
      "loss": 0.9828,
      "step": 3315
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2873237914784205,
      "learning_rate": 6.940496304667839e-05,
      "loss": 1.0295,
      "step": 3316
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.16099057476986792,
      "learning_rate": 6.934814962873739e-05,
      "loss": 1.008,
      "step": 3317
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1811737095994218,
      "learning_rate": 6.929134712833328e-05,
      "loss": 1.0207,
      "step": 3318
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.19766946932563978,
      "learning_rate": 6.92345555656979e-05,
      "loss": 1.2122,
      "step": 3319
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.18218051826676745,
      "learning_rate": 6.917777496105917e-05,
      "loss": 1.0701,
      "step": 3320
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.20687090961661608,
      "learning_rate": 6.912100533464121e-05,
      "loss": 0.9753,
      "step": 3321
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.22866506960559055,
      "learning_rate": 6.906424670666409e-05,
      "loss": 1.0109,
      "step": 3322
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.21440381295484132,
      "learning_rate": 6.900749909734406e-05,
      "loss": 1.0371,
      "step": 3323
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.20122458168316484,
      "learning_rate": 6.895076252689335e-05,
      "loss": 1.0425,
      "step": 3324
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.24195910795782494,
      "learning_rate": 6.889403701552039e-05,
      "loss": 0.9608,
      "step": 3325
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.17975134817805583,
      "learning_rate": 6.88373225834296e-05,
      "loss": 1.1089,
      "step": 3326
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.19371573352450508,
      "learning_rate": 6.878061925082137e-05,
      "loss": 1.0329,
      "step": 3327
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.21282849223728253,
      "learning_rate": 6.872392703789232e-05,
      "loss": 1.0089,
      "step": 3328
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.19465101379111616,
      "learning_rate": 6.866724596483495e-05,
      "loss": 0.9509,
      "step": 3329
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.22762489814300446,
      "learning_rate": 6.861057605183787e-05,
      "loss": 1.0925,
      "step": 3330
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.19519677994265686,
      "learning_rate": 6.855391731908567e-05,
      "loss": 1.1376,
      "step": 3331
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.17382587915636893,
      "learning_rate": 6.849726978675906e-05,
      "loss": 1.0675,
      "step": 3332
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.22978940524348307,
      "learning_rate": 6.84406334750346e-05,
      "loss": 0.9633,
      "step": 3333
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.18544454802442306,
      "learning_rate": 6.838400840408499e-05,
      "loss": 1.0881,
      "step": 3334
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.20850339955040942,
      "learning_rate": 6.832739459407885e-05,
      "loss": 0.9377,
      "step": 3335
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.18495516341650467,
      "learning_rate": 6.827079206518083e-05,
      "loss": 1.0372,
      "step": 3336
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.18987099852937678,
      "learning_rate": 6.821420083755157e-05,
      "loss": 1.1377,
      "step": 3337
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1840027200701242,
      "learning_rate": 6.81576209313476e-05,
      "loss": 1.0019,
      "step": 3338
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.19439590528444184,
      "learning_rate": 6.810105236672155e-05,
      "loss": 1.0973,
      "step": 3339
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.19642850162811473,
      "learning_rate": 6.804449516382183e-05,
      "loss": 1.0352,
      "step": 3340
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.17561533212672342,
      "learning_rate": 6.798794934279304e-05,
      "loss": 1.0329,
      "step": 3341
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2084563483819395,
      "learning_rate": 6.793141492377552e-05,
      "loss": 1.0162,
      "step": 3342
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.19676996760058554,
      "learning_rate": 6.787489192690569e-05,
      "loss": 0.9459,
      "step": 3343
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.19794571421789428,
      "learning_rate": 6.781838037231571e-05,
      "loss": 1.0488,
      "step": 3344
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.20216473581700573,
      "learning_rate": 6.776188028013396e-05,
      "loss": 0.9912,
      "step": 3345
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2090737828007055,
      "learning_rate": 6.770539167048446e-05,
      "loss": 1.0021,
      "step": 3346
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.28529821635344205,
      "learning_rate": 6.764891456348729e-05,
      "loss": 1.0172,
      "step": 3347
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.18334767937100319,
      "learning_rate": 6.759244897925841e-05,
      "loss": 1.0905,
      "step": 3348
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1743848947571326,
      "learning_rate": 6.753599493790959e-05,
      "loss": 1.011,
      "step": 3349
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.19513320935198947,
      "learning_rate": 6.747955245954862e-05,
      "loss": 1.1405,
      "step": 3350
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.20758665892146175,
      "learning_rate": 6.74231215642791e-05,
      "loss": 1.0485,
      "step": 3351
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.20867158743460493,
      "learning_rate": 6.736670227220052e-05,
      "loss": 0.9386,
      "step": 3352
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1611325574702465,
      "learning_rate": 6.73102946034082e-05,
      "loss": 1.0438,
      "step": 3353
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.17622391469931212,
      "learning_rate": 6.725389857799337e-05,
      "loss": 0.9753,
      "step": 3354
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.16480449780881365,
      "learning_rate": 6.719751421604309e-05,
      "loss": 1.0936,
      "step": 3355
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.18516139977071222,
      "learning_rate": 6.714114153764029e-05,
      "loss": 1.1523,
      "step": 3356
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.20792558530012517,
      "learning_rate": 6.708478056286369e-05,
      "loss": 1.1442,
      "step": 3357
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.17860179165253476,
      "learning_rate": 6.70284313117879e-05,
      "loss": 1.0824,
      "step": 3358
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1810738671286442,
      "learning_rate": 6.697209380448333e-05,
      "loss": 1.0461,
      "step": 3359
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.20635226019155425,
      "learning_rate": 6.691576806101612e-05,
      "loss": 0.9909,
      "step": 3360
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.21235768520616466,
      "learning_rate": 6.68594541014484e-05,
      "loss": 1.1198,
      "step": 3361
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.19636400900365444,
      "learning_rate": 6.680315194583794e-05,
      "loss": 0.9739,
      "step": 3362
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.22163255970070675,
      "learning_rate": 6.674686161423843e-05,
      "loss": 1.1354,
      "step": 3363
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2095143601917929,
      "learning_rate": 6.669058312669919e-05,
      "loss": 1.0272,
      "step": 3364
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1786925572999645,
      "learning_rate": 6.663431650326553e-05,
      "loss": 0.9764,
      "step": 3365
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1808559721019589,
      "learning_rate": 6.657806176397836e-05,
      "loss": 0.9117,
      "step": 3366
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.19382699214543384,
      "learning_rate": 6.652181892887446e-05,
      "loss": 1.0854,
      "step": 3367
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.24748522691692962,
      "learning_rate": 6.646558801798627e-05,
      "loss": 1.0513,
      "step": 3368
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1619195184387168,
      "learning_rate": 6.640936905134212e-05,
      "loss": 1.0777,
      "step": 3369
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.18978704374434496,
      "learning_rate": 6.635316204896598e-05,
      "loss": 1.0985,
      "step": 3370
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.22158297492705478,
      "learning_rate": 6.629696703087754e-05,
      "loss": 0.9559,
      "step": 3371
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.19559349485972194,
      "learning_rate": 6.624078401709235e-05,
      "loss": 1.2444,
      "step": 3372
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.18698786008843693,
      "learning_rate": 6.61846130276215e-05,
      "loss": 1.0541,
      "step": 3373
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.21101556731605306,
      "learning_rate": 6.612845408247204e-05,
      "loss": 1.0098,
      "step": 3374
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.18978753967068968,
      "learning_rate": 6.607230720164647e-05,
      "loss": 1.0645,
      "step": 3375
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.23731114274823087,
      "learning_rate": 6.60161724051432e-05,
      "loss": 1.2091,
      "step": 3376
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1831249008719852,
      "learning_rate": 6.596004971295615e-05,
      "loss": 0.9532,
      "step": 3377
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.22062085742244852,
      "learning_rate": 6.590393914507517e-05,
      "loss": 1.0202,
      "step": 3378
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.22428330750885728,
      "learning_rate": 6.584784072148555e-05,
      "loss": 1.0918,
      "step": 3379
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.19051818074096244,
      "learning_rate": 6.579175446216833e-05,
      "loss": 1.2785,
      "step": 3380
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.17891107503398995,
      "learning_rate": 6.573568038710032e-05,
      "loss": 1.1025,
      "step": 3381
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2020520180791638,
      "learning_rate": 6.567961851625386e-05,
      "loss": 1.1382,
      "step": 3382
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.18079241074039856,
      "learning_rate": 6.562356886959704e-05,
      "loss": 1.0272,
      "step": 3383
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1784704870684247,
      "learning_rate": 6.556753146709348e-05,
      "loss": 1.2168,
      "step": 3384
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.19869489872578147,
      "learning_rate": 6.55115063287026e-05,
      "loss": 1.1138,
      "step": 3385
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.208538961099462,
      "learning_rate": 6.545549347437931e-05,
      "loss": 1.1452,
      "step": 3386
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.19587299279356035,
      "learning_rate": 6.539949292407421e-05,
      "loss": 1.1676,
      "step": 3387
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1728141409914368,
      "learning_rate": 6.534350469773348e-05,
      "loss": 1.0678,
      "step": 3388
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.19420531851574058,
      "learning_rate": 6.528752881529898e-05,
      "loss": 1.0237,
      "step": 3389
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.19081138488822547,
      "learning_rate": 6.523156529670811e-05,
      "loss": 0.9184,
      "step": 3390
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2186771855056742,
      "learning_rate": 6.517561416189387e-05,
      "loss": 0.9946,
      "step": 3391
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.21915857655385224,
      "learning_rate": 6.511967543078488e-05,
      "loss": 0.967,
      "step": 3392
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.20977556553491472,
      "learning_rate": 6.506374912330527e-05,
      "loss": 1.012,
      "step": 3393
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.21776767299111116,
      "learning_rate": 6.50078352593749e-05,
      "loss": 1.0892,
      "step": 3394
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.17228017338908855,
      "learning_rate": 6.495193385890901e-05,
      "loss": 1.0422,
      "step": 3395
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2897363669131825,
      "learning_rate": 6.489604494181857e-05,
      "loss": 0.9939,
      "step": 3396
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.21521089752105207,
      "learning_rate": 6.484016852800993e-05,
      "loss": 0.9785,
      "step": 3397
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2080939076671122,
      "learning_rate": 6.478430463738516e-05,
      "loss": 0.9202,
      "step": 3398
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.19970859631865667,
      "learning_rate": 6.472845328984175e-05,
      "loss": 1.0753,
      "step": 3399
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.18456848154294184,
      "learning_rate": 6.467261450527279e-05,
      "loss": 0.9934,
      "step": 3400
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.17387835562632703,
      "learning_rate": 6.461678830356684e-05,
      "loss": 1.0129,
      "step": 3401
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.18810450753588262,
      "learning_rate": 6.456097470460798e-05,
      "loss": 1.0839,
      "step": 3402
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.19821099605234482,
      "learning_rate": 6.450517372827591e-05,
      "loss": 1.0472,
      "step": 3403
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.19770220833298002,
      "learning_rate": 6.444938539444565e-05,
      "loss": 0.972,
      "step": 3404
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.18472295431643318,
      "learning_rate": 6.43936097229879e-05,
      "loss": 1.0858,
      "step": 3405
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1984065783646406,
      "learning_rate": 6.43378467337687e-05,
      "loss": 1.0224,
      "step": 3406
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.20798316524627058,
      "learning_rate": 6.42820964466497e-05,
      "loss": 1.1429,
      "step": 3407
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.18371393621111912,
      "learning_rate": 6.422635888148791e-05,
      "loss": 1.0911,
      "step": 3408
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.20363986744366203,
      "learning_rate": 6.417063405813594e-05,
      "loss": 1.028,
      "step": 3409
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.23086491380039037,
      "learning_rate": 6.411492199644171e-05,
      "loss": 1.1293,
      "step": 3410
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.20004769768465172,
      "learning_rate": 6.405922271624874e-05,
      "loss": 1.0765,
      "step": 3411
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2159214870199943,
      "learning_rate": 6.400353623739587e-05,
      "loss": 0.9625,
      "step": 3412
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.17839898938786367,
      "learning_rate": 6.394786257971743e-05,
      "loss": 1.1141,
      "step": 3413
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.23458676405598414,
      "learning_rate": 6.389220176304325e-05,
      "loss": 0.9072,
      "step": 3414
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.20792741611702323,
      "learning_rate": 6.383655380719848e-05,
      "loss": 1.0418,
      "step": 3415
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2119192667395123,
      "learning_rate": 6.378091873200379e-05,
      "loss": 1.0969,
      "step": 3416
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.18466818320877992,
      "learning_rate": 6.372529655727513e-05,
      "loss": 0.963,
      "step": 3417
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.20551682389882203,
      "learning_rate": 6.366968730282403e-05,
      "loss": 0.9684,
      "step": 3418
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2131006816214632,
      "learning_rate": 6.361409098845725e-05,
      "loss": 0.999,
      "step": 3419
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.18724739701342796,
      "learning_rate": 6.355850763397707e-05,
      "loss": 1.0564,
      "step": 3420
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1774961693064699,
      "learning_rate": 6.350293725918102e-05,
      "loss": 1.0687,
      "step": 3421
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.23512344629797038,
      "learning_rate": 6.344737988386215e-05,
      "loss": 1.0477,
      "step": 3422
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2067065383150378,
      "learning_rate": 6.339183552780884e-05,
      "loss": 1.0383,
      "step": 3423
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.19886397289352456,
      "learning_rate": 6.333630421080472e-05,
      "loss": 1.0905,
      "step": 3424
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.19736812191691858,
      "learning_rate": 6.328078595262894e-05,
      "loss": 1.0803,
      "step": 3425
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2717562455981034,
      "learning_rate": 6.322528077305585e-05,
      "loss": 0.9899,
      "step": 3426
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.24379998436521913,
      "learning_rate": 6.316978869185532e-05,
      "loss": 0.9913,
      "step": 3427
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.19851689559074606,
      "learning_rate": 6.311430972879232e-05,
      "loss": 1.0013,
      "step": 3428
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1915356609565645,
      "learning_rate": 6.305884390362738e-05,
      "loss": 1.1568,
      "step": 3429
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.19351488494943098,
      "learning_rate": 6.300339123611618e-05,
      "loss": 1.0104,
      "step": 3430
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.19099227661735,
      "learning_rate": 6.294795174600985e-05,
      "loss": 0.9184,
      "step": 3431
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.22198153983721444,
      "learning_rate": 6.28925254530547e-05,
      "loss": 1.011,
      "step": 3432
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.24382943466936263,
      "learning_rate": 6.283711237699236e-05,
      "loss": 0.899,
      "step": 3433
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.23831380872489238,
      "learning_rate": 6.27817125375599e-05,
      "loss": 0.9261,
      "step": 3434
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2099936321839458,
      "learning_rate": 6.272632595448947e-05,
      "loss": 1.1658,
      "step": 3435
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1877473194660829,
      "learning_rate": 6.267095264750866e-05,
      "loss": 0.903,
      "step": 3436
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2117029129632022,
      "learning_rate": 6.261559263634017e-05,
      "loss": 0.9104,
      "step": 3437
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.18289964584565532,
      "learning_rate": 6.256024594070217e-05,
      "loss": 1.0717,
      "step": 3438
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1920854690413216,
      "learning_rate": 6.250491258030791e-05,
      "loss": 1.0526,
      "step": 3439
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.20042728799000945,
      "learning_rate": 6.244959257486596e-05,
      "loss": 1.0667,
      "step": 3440
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.18408809131801665,
      "learning_rate": 6.239428594408013e-05,
      "loss": 0.9952,
      "step": 3441
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.18558182263221737,
      "learning_rate": 6.23389927076495e-05,
      "loss": 1.0261,
      "step": 3442
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.20911347690614526,
      "learning_rate": 6.22837128852683e-05,
      "loss": 0.9908,
      "step": 3443
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.23232248292696223,
      "learning_rate": 6.222844649662604e-05,
      "loss": 1.0148,
      "step": 3444
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.18597985193230898,
      "learning_rate": 6.217319356140745e-05,
      "loss": 1.0293,
      "step": 3445
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.23301636401379716,
      "learning_rate": 6.211795409929238e-05,
      "loss": 0.9706,
      "step": 3446
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1803829411546901,
      "learning_rate": 6.206272812995607e-05,
      "loss": 0.9708,
      "step": 3447
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2394071082130075,
      "learning_rate": 6.200751567306874e-05,
      "loss": 1.0272,
      "step": 3448
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1812504670127948,
      "learning_rate": 6.195231674829593e-05,
      "loss": 1.0883,
      "step": 3449
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2791748147804822,
      "learning_rate": 6.18971313752983e-05,
      "loss": 1.0916,
      "step": 3450
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1584271304791279,
      "learning_rate": 6.184195957373176e-05,
      "loss": 0.987,
      "step": 3451
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2572448435682796,
      "learning_rate": 6.178680136324727e-05,
      "loss": 1.0257,
      "step": 3452
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.16661494556376183,
      "learning_rate": 6.173165676349103e-05,
      "loss": 1.1056,
      "step": 3453
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.18169825022106126,
      "learning_rate": 6.167652579410439e-05,
      "loss": 1.0725,
      "step": 3454
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.22269581454704857,
      "learning_rate": 6.162140847472381e-05,
      "loss": 1.0773,
      "step": 3455
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.17339937303832376,
      "learning_rate": 6.156630482498092e-05,
      "loss": 0.9238,
      "step": 3456
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.21076235757461823,
      "learning_rate": 6.151121486450243e-05,
      "loss": 1.1383,
      "step": 3457
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.17571282911521466,
      "learning_rate": 6.145613861291026e-05,
      "loss": 1.116,
      "step": 3458
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.19937378533069824,
      "learning_rate": 6.140107608982136e-05,
      "loss": 0.9633,
      "step": 3459
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.18900485516190585,
      "learning_rate": 6.134602731484788e-05,
      "loss": 1.0755,
      "step": 3460
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.18587354989416494,
      "learning_rate": 6.129099230759695e-05,
      "loss": 1.0243,
      "step": 3461
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.21656486616083692,
      "learning_rate": 6.123597108767095e-05,
      "loss": 1.0652,
      "step": 3462
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.25035738868042845,
      "learning_rate": 6.118096367466721e-05,
      "loss": 1.2397,
      "step": 3463
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.18036912337263944,
      "learning_rate": 6.112597008817823e-05,
      "loss": 1.0048,
      "step": 3464
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.20061163490564762,
      "learning_rate": 6.107099034779155e-05,
      "loss": 1.0914,
      "step": 3465
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.19375733850520657,
      "learning_rate": 6.1016024473089726e-05,
      "loss": 0.9203,
      "step": 3466
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.17995873207847815,
      "learning_rate": 6.0961072483650526e-05,
      "loss": 0.8893,
      "step": 3467
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2072574202087396,
      "learning_rate": 6.090613439904662e-05,
      "loss": 0.8973,
      "step": 3468
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2491183078835181,
      "learning_rate": 6.0851210238845834e-05,
      "loss": 0.9906,
      "step": 3469
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.18819402763315068,
      "learning_rate": 6.07963000226109e-05,
      "loss": 1.0463,
      "step": 3470
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.23202796476938567,
      "learning_rate": 6.0741403769899765e-05,
      "loss": 1.1098,
      "step": 3471
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.20895092132092727,
      "learning_rate": 6.0686521500265234e-05,
      "loss": 1.0053,
      "step": 3472
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.19652189860426791,
      "learning_rate": 6.063165323325526e-05,
      "loss": 0.9424,
      "step": 3473
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.20214128334903114,
      "learning_rate": 6.057679898841269e-05,
      "loss": 0.8755,
      "step": 3474
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.16906055595255357,
      "learning_rate": 6.05219587852755e-05,
      "loss": 1.0508,
      "step": 3475
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2110692382202643,
      "learning_rate": 6.046713264337658e-05,
      "loss": 1.0662,
      "step": 3476
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1739022561735214,
      "learning_rate": 6.041232058224382e-05,
      "loss": 1.0731,
      "step": 3477
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.22731012520296937,
      "learning_rate": 6.0357522621400134e-05,
      "loss": 1.1145,
      "step": 3478
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.21131169590825305,
      "learning_rate": 6.030273878036334e-05,
      "loss": 1.1545,
      "step": 3479
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.19141584802655975,
      "learning_rate": 6.0247969078646384e-05,
      "loss": 1.0608,
      "step": 3480
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.18330703188381464,
      "learning_rate": 6.019321353575692e-05,
      "loss": 1.0658,
      "step": 3481
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.22220111045923613,
      "learning_rate": 6.013847217119784e-05,
      "loss": 0.9482,
      "step": 3482
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.19323881165978105,
      "learning_rate": 6.008374500446676e-05,
      "loss": 0.9311,
      "step": 3483
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.21818216014849218,
      "learning_rate": 6.002903205505641e-05,
      "loss": 0.9242,
      "step": 3484
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.17266784141029717,
      "learning_rate": 5.9974333342454326e-05,
      "loss": 1.189,
      "step": 3485
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.22973784414770837,
      "learning_rate": 5.991964888614301e-05,
      "loss": 1.0207,
      "step": 3486
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.21968898551216196,
      "learning_rate": 5.986497870559997e-05,
      "loss": 1.0308,
      "step": 3487
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1886055962489234,
      "learning_rate": 5.98103228202975e-05,
      "loss": 1.1582,
      "step": 3488
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.20444718462337644,
      "learning_rate": 5.9755681249702896e-05,
      "loss": 1.0151,
      "step": 3489
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.190485544326146,
      "learning_rate": 5.970105401327828e-05,
      "loss": 0.9437,
      "step": 3490
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.21585040060877406,
      "learning_rate": 5.964644113048079e-05,
      "loss": 1.0855,
      "step": 3491
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.21967616599438888,
      "learning_rate": 5.9591842620762294e-05,
      "loss": 0.9522,
      "step": 3492
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.20298012465764492,
      "learning_rate": 5.953725850356968e-05,
      "loss": 1.0134,
      "step": 3493
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2081610622071685,
      "learning_rate": 5.948268879834458e-05,
      "loss": 1.0097,
      "step": 3494
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.19787254969546084,
      "learning_rate": 5.9428133524523646e-05,
      "loss": 1.1084,
      "step": 3495
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2285576835598241,
      "learning_rate": 5.937359270153825e-05,
      "loss": 0.9504,
      "step": 3496
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.17443240962318404,
      "learning_rate": 5.931906634881468e-05,
      "loss": 0.8848,
      "step": 3497
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.19870248338397906,
      "learning_rate": 5.926455448577407e-05,
      "loss": 1.0956,
      "step": 3498
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.21833565865099705,
      "learning_rate": 5.921005713183235e-05,
      "loss": 1.0696,
      "step": 3499
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2588215891811804,
      "learning_rate": 5.9155574306400395e-05,
      "loss": 0.967,
      "step": 3500
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.20578751159579423,
      "learning_rate": 5.910110602888377e-05,
      "loss": 1.1066,
      "step": 3501
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1836973004385898,
      "learning_rate": 5.904665231868294e-05,
      "loss": 1.1017,
      "step": 3502
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1788755275662708,
      "learning_rate": 5.89922131951931e-05,
      "loss": 0.8107,
      "step": 3503
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.23841196580159374,
      "learning_rate": 5.893778867780444e-05,
      "loss": 1.1264,
      "step": 3504
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1944819894285701,
      "learning_rate": 5.8883378785901644e-05,
      "loss": 1.108,
      "step": 3505
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1893718992886241,
      "learning_rate": 5.88289835388645e-05,
      "loss": 1.0112,
      "step": 3506
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.21155222203137203,
      "learning_rate": 5.877460295606738e-05,
      "loss": 1.1437,
      "step": 3507
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2056420316674882,
      "learning_rate": 5.872023705687948e-05,
      "loss": 0.9524,
      "step": 3508
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2410736539756741,
      "learning_rate": 5.866588586066481e-05,
      "loss": 0.9087,
      "step": 3509
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.20299373153543984,
      "learning_rate": 5.861154938678204e-05,
      "loss": 1.0751,
      "step": 3510
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.20971852366871654,
      "learning_rate": 5.8557227654584754e-05,
      "loss": 0.9556,
      "step": 3511
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.22285080471141686,
      "learning_rate": 5.8502920683421134e-05,
      "loss": 1.1218,
      "step": 3512
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.20141844800938288,
      "learning_rate": 5.8448628492634225e-05,
      "loss": 1.1053,
      "step": 3513
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.20152781428243444,
      "learning_rate": 5.839435110156167e-05,
      "loss": 0.958,
      "step": 3514
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.20228910846420714,
      "learning_rate": 5.834008852953603e-05,
      "loss": 1.1015,
      "step": 3515
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2124836333876322,
      "learning_rate": 5.828584079588435e-05,
      "loss": 1.1207,
      "step": 3516
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1754069248165951,
      "learning_rate": 5.8231607919928636e-05,
      "loss": 0.948,
      "step": 3517
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.165153277421249,
      "learning_rate": 5.817738992098543e-05,
      "loss": 0.9029,
      "step": 3518
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1966669916349029,
      "learning_rate": 5.812318681836597e-05,
      "loss": 1.1127,
      "step": 3519
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.20086412418159574,
      "learning_rate": 5.8068998631376384e-05,
      "loss": 1.0937,
      "step": 3520
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.21454231438985813,
      "learning_rate": 5.801482537931722e-05,
      "loss": 1.0455,
      "step": 3521
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.18767912573332962,
      "learning_rate": 5.796066708148393e-05,
      "loss": 1.0677,
      "step": 3522
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.20918299521831563,
      "learning_rate": 5.790652375716652e-05,
      "loss": 1.1486,
      "step": 3523
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.19691618002568542,
      "learning_rate": 5.785239542564968e-05,
      "loss": 1.1281,
      "step": 3524
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2010967191754908,
      "learning_rate": 5.779828210621273e-05,
      "loss": 0.982,
      "step": 3525
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.22863346051244285,
      "learning_rate": 5.774418381812977e-05,
      "loss": 0.8849,
      "step": 3526
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1813708004042739,
      "learning_rate": 5.769010058066937e-05,
      "loss": 1.0535,
      "step": 3527
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.19492657883499687,
      "learning_rate": 5.763603241309492e-05,
      "loss": 1.0631,
      "step": 3528
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2086020562676018,
      "learning_rate": 5.758197933466429e-05,
      "loss": 1.0884,
      "step": 3529
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.17433913749610308,
      "learning_rate": 5.7527941364630066e-05,
      "loss": 0.902,
      "step": 3530
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.21312448531487954,
      "learning_rate": 5.74739185222394e-05,
      "loss": 1.1295,
      "step": 3531
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.18806986150263288,
      "learning_rate": 5.7419910826734056e-05,
      "loss": 1.1727,
      "step": 3532
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.21496669151709513,
      "learning_rate": 5.7365918297350484e-05,
      "loss": 1.1012,
      "step": 3533
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.20461321216690584,
      "learning_rate": 5.731194095331961e-05,
      "loss": 1.0875,
      "step": 3534
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2076350033984962,
      "learning_rate": 5.7257978813867094e-05,
      "loss": 1.0967,
      "step": 3535
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.199486033835761,
      "learning_rate": 5.720403189821302e-05,
      "loss": 1.2484,
      "step": 3536
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.18644380524731174,
      "learning_rate": 5.71501002255722e-05,
      "loss": 0.9498,
      "step": 3537
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2163703333287637,
      "learning_rate": 5.709618381515394e-05,
      "loss": 1.0206,
      "step": 3538
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1791802667409674,
      "learning_rate": 5.704228268616208e-05,
      "loss": 1.0393,
      "step": 3539
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.19423875617778955,
      "learning_rate": 5.698839685779507e-05,
      "loss": 0.9888,
      "step": 3540
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.20893033129839395,
      "learning_rate": 5.693452634924587e-05,
      "loss": 1.0409,
      "step": 3541
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.17442455124369532,
      "learning_rate": 5.688067117970207e-05,
      "loss": 1.1199,
      "step": 3542
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.18128359037050878,
      "learning_rate": 5.682683136834567e-05,
      "loss": 1.0706,
      "step": 3543
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1682435074667977,
      "learning_rate": 5.6773006934353326e-05,
      "loss": 1.1587,
      "step": 3544
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2120852324103909,
      "learning_rate": 5.6719197896896106e-05,
      "loss": 1.2523,
      "step": 3545
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.16894850836028402,
      "learning_rate": 5.666540427513969e-05,
      "loss": 1.0681,
      "step": 3546
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.19709321380887518,
      "learning_rate": 5.6611626088244194e-05,
      "loss": 0.9412,
      "step": 3547
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2024608722718471,
      "learning_rate": 5.6557863355364286e-05,
      "loss": 0.9479,
      "step": 3548
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.17734843194964034,
      "learning_rate": 5.650411609564907e-05,
      "loss": 1.1302,
      "step": 3549
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.24042977981413133,
      "learning_rate": 5.645038432824217e-05,
      "loss": 1.0384,
      "step": 3550
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.18116685672977775,
      "learning_rate": 5.639666807228175e-05,
      "loss": 1.0538,
      "step": 3551
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.17546852839913454,
      "learning_rate": 5.634296734690035e-05,
      "loss": 0.9891,
      "step": 3552
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.18288162808199646,
      "learning_rate": 5.628928217122507e-05,
      "loss": 1.0825,
      "step": 3553
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.182350170418,
      "learning_rate": 5.623561256437736e-05,
      "loss": 1.0557,
      "step": 3554
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.19742977910004209,
      "learning_rate": 5.6181958545473325e-05,
      "loss": 1.1228,
      "step": 3555
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1877646637917707,
      "learning_rate": 5.612832013362321e-05,
      "loss": 1.1113,
      "step": 3556
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.23465878872801002,
      "learning_rate": 5.6074697347932e-05,
      "loss": 1.1095,
      "step": 3557
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.21850598192515708,
      "learning_rate": 5.6021090207498926e-05,
      "loss": 1.083,
      "step": 3558
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.22329754987418116,
      "learning_rate": 5.5967498731417765e-05,
      "loss": 1.1127,
      "step": 3559
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2274919362417169,
      "learning_rate": 5.5913922938776645e-05,
      "loss": 0.9994,
      "step": 3560
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2373741678363117,
      "learning_rate": 5.586036284865805e-05,
      "loss": 1.04,
      "step": 3561
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.20549140303887567,
      "learning_rate": 5.580681848013906e-05,
      "loss": 1.0674,
      "step": 3562
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.16249552596567618,
      "learning_rate": 5.575328985229098e-05,
      "loss": 1.2072,
      "step": 3563
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.19102904300990428,
      "learning_rate": 5.569977698417956e-05,
      "loss": 0.9458,
      "step": 3564
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.24994115075346954,
      "learning_rate": 5.564627989486494e-05,
      "loss": 1.0408,
      "step": 3565
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.19981279416937792,
      "learning_rate": 5.5592798603401696e-05,
      "loss": 1.2005,
      "step": 3566
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.20650931508376838,
      "learning_rate": 5.5539333128838653e-05,
      "loss": 0.9864,
      "step": 3567
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1660613460185073,
      "learning_rate": 5.548588349021915e-05,
      "loss": 1.0319,
      "step": 3568
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.19206682414202586,
      "learning_rate": 5.543244970658074e-05,
      "loss": 0.874,
      "step": 3569
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.18755677787372974,
      "learning_rate": 5.537903179695547e-05,
      "loss": 1.0187,
      "step": 3570
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.21900846766024207,
      "learning_rate": 5.5325629780369635e-05,
      "loss": 0.8988,
      "step": 3571
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1869592004076636,
      "learning_rate": 5.5272243675843896e-05,
      "loss": 1.1329,
      "step": 3572
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.17144927972019577,
      "learning_rate": 5.521887350239323e-05,
      "loss": 1.0179,
      "step": 3573
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.19145282941044867,
      "learning_rate": 5.5165519279026935e-05,
      "loss": 1.0476,
      "step": 3574
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3140075644475535,
      "learning_rate": 5.511218102474872e-05,
      "loss": 1.1652,
      "step": 3575
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.190610975994275,
      "learning_rate": 5.505885875855647e-05,
      "loss": 1.0834,
      "step": 3576
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.19458300958957683,
      "learning_rate": 5.5005552499442504e-05,
      "loss": 1.1332,
      "step": 3577
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1762852442010008,
      "learning_rate": 5.495226226639331e-05,
      "loss": 1.0109,
      "step": 3578
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.18402773592040547,
      "learning_rate": 5.48989880783898e-05,
      "loss": 1.1675,
      "step": 3579
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.19094308270408314,
      "learning_rate": 5.484572995440709e-05,
      "loss": 1.0068,
      "step": 3580
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1875577856656065,
      "learning_rate": 5.479248791341458e-05,
      "loss": 1.1394,
      "step": 3581
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.17865219627026352,
      "learning_rate": 5.473926197437594e-05,
      "loss": 0.9371,
      "step": 3582
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.16909628208747524,
      "learning_rate": 5.468605215624909e-05,
      "loss": 0.8987,
      "step": 3583
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.23032718235574717,
      "learning_rate": 5.463285847798632e-05,
      "loss": 1.0291,
      "step": 3584
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2039671864648694,
      "learning_rate": 5.4579680958533986e-05,
      "loss": 1.0486,
      "step": 3585
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.18918421335729174,
      "learning_rate": 5.4526519616832885e-05,
      "loss": 1.0821,
      "step": 3586
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.19595904363105265,
      "learning_rate": 5.4473374471817906e-05,
      "loss": 1.0139,
      "step": 3587
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.19726552537368675,
      "learning_rate": 5.442024554241822e-05,
      "loss": 0.8569,
      "step": 3588
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.20297987822071661,
      "learning_rate": 5.436713284755721e-05,
      "loss": 1.1164,
      "step": 3589
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.17705212564362574,
      "learning_rate": 5.431403640615252e-05,
      "loss": 1.0517,
      "step": 3590
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.17076853957735164,
      "learning_rate": 5.4260956237115965e-05,
      "loss": 0.9569,
      "step": 3591
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.16915213326956086,
      "learning_rate": 5.420789235935353e-05,
      "loss": 0.9955,
      "step": 3592
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.20296218575414318,
      "learning_rate": 5.415484479176549e-05,
      "loss": 0.9951,
      "step": 3593
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.18536617232940225,
      "learning_rate": 5.410181355324622e-05,
      "loss": 1.0934,
      "step": 3594
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.25144904451358674,
      "learning_rate": 5.4048798662684376e-05,
      "loss": 1.1148,
      "step": 3595
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1791410200811393,
      "learning_rate": 5.39958001389627e-05,
      "loss": 1.0018,
      "step": 3596
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.17724610683739123,
      "learning_rate": 5.394281800095813e-05,
      "loss": 1.0521,
      "step": 3597
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1938627788319221,
      "learning_rate": 5.388985226754175e-05,
      "loss": 1.0088,
      "step": 3598
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.17918798375352982,
      "learning_rate": 5.3836902957578904e-05,
      "loss": 1.1079,
      "step": 3599
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.19136905944231636,
      "learning_rate": 5.378397008992893e-05,
      "loss": 1.1115,
      "step": 3600
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.19484428227864797,
      "learning_rate": 5.373105368344544e-05,
      "loss": 1.1029,
      "step": 3601
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1802429376646844,
      "learning_rate": 5.3678153756976135e-05,
      "loss": 1.1249,
      "step": 3602
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2087992304541337,
      "learning_rate": 5.362527032936277e-05,
      "loss": 1.1758,
      "step": 3603
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.20292166231277167,
      "learning_rate": 5.357240341944145e-05,
      "loss": 1.1248,
      "step": 3604
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.19082033276568736,
      "learning_rate": 5.351955304604207e-05,
      "loss": 0.9199,
      "step": 3605
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1972232445996486,
      "learning_rate": 5.346671922798891e-05,
      "loss": 0.9955,
      "step": 3606
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1953016899850395,
      "learning_rate": 5.341390198410019e-05,
      "loss": 1.1084,
      "step": 3607
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.20746472750179454,
      "learning_rate": 5.336110133318837e-05,
      "loss": 0.9665,
      "step": 3608
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.19191241027063916,
      "learning_rate": 5.3308317294059826e-05,
      "loss": 0.9331,
      "step": 3609
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.18136826493857525,
      "learning_rate": 5.32555498855152e-05,
      "loss": 1.229,
      "step": 3610
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2018782460551806,
      "learning_rate": 5.320279912634907e-05,
      "loss": 1.1581,
      "step": 3611
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.18428430382388547,
      "learning_rate": 5.3150065035350136e-05,
      "loss": 1.1616,
      "step": 3612
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.17590016835934832,
      "learning_rate": 5.309734763130116e-05,
      "loss": 1.1027,
      "step": 3613
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.21460632860529047,
      "learning_rate": 5.304464693297895e-05,
      "loss": 0.9999,
      "step": 3614
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.17765277685801162,
      "learning_rate": 5.299196295915441e-05,
      "loss": 0.8694,
      "step": 3615
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.17539561936702688,
      "learning_rate": 5.2939295728592374e-05,
      "loss": 0.8886,
      "step": 3616
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.22459152314551425,
      "learning_rate": 5.2886645260051894e-05,
      "loss": 1.0865,
      "step": 3617
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.18799967728821002,
      "learning_rate": 5.283401157228584e-05,
      "loss": 0.9813,
      "step": 3618
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.19072843985842736,
      "learning_rate": 5.278139468404133e-05,
      "loss": 1.1958,
      "step": 3619
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1817375232659677,
      "learning_rate": 5.2728794614059306e-05,
      "loss": 1.1736,
      "step": 3620
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.21919495010294876,
      "learning_rate": 5.26762113810748e-05,
      "loss": 1.1499,
      "step": 3621
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.18989598945248132,
      "learning_rate": 5.2623645003816804e-05,
      "loss": 0.9676,
      "step": 3622
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.18204196253782895,
      "learning_rate": 5.257109550100843e-05,
      "loss": 1.1908,
      "step": 3623
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.20465810549036248,
      "learning_rate": 5.251856289136664e-05,
      "loss": 1.0381,
      "step": 3624
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.24435076699237118,
      "learning_rate": 5.2466047193602395e-05,
      "loss": 1.0467,
      "step": 3625
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.21263150649344162,
      "learning_rate": 5.241354842642077e-05,
      "loss": 1.0981,
      "step": 3626
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.20743848647535798,
      "learning_rate": 5.236106660852058e-05,
      "loss": 1.0784,
      "step": 3627
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.19113831598603281,
      "learning_rate": 5.2308601758594866e-05,
      "loss": 1.063,
      "step": 3628
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.21305084984643552,
      "learning_rate": 5.225615389533043e-05,
      "loss": 0.9224,
      "step": 3629
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.16908721325070006,
      "learning_rate": 5.220372303740807e-05,
      "loss": 0.9291,
      "step": 3630
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.17215153920893264,
      "learning_rate": 5.215130920350252e-05,
      "loss": 1.0296,
      "step": 3631
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.19313929895790988,
      "learning_rate": 5.209891241228254e-05,
      "loss": 1.0597,
      "step": 3632
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2408871760564238,
      "learning_rate": 5.204653268241069e-05,
      "loss": 1.0367,
      "step": 3633
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.24785057875700806,
      "learning_rate": 5.199417003254358e-05,
      "loss": 1.1314,
      "step": 3634
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.18208680693794946,
      "learning_rate": 5.1941824481331626e-05,
      "loss": 0.9761,
      "step": 3635
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2036983746375491,
      "learning_rate": 5.1889496047419216e-05,
      "loss": 1.0554,
      "step": 3636
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1964140071578769,
      "learning_rate": 5.183718474944464e-05,
      "loss": 1.1047,
      "step": 3637
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.19388114489165265,
      "learning_rate": 5.1784890606039984e-05,
      "loss": 1.0475,
      "step": 3638
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2080388339512544,
      "learning_rate": 5.173261363583142e-05,
      "loss": 0.9338,
      "step": 3639
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1855361670474136,
      "learning_rate": 5.168035385743879e-05,
      "loss": 1.0354,
      "step": 3640
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.24731337193502964,
      "learning_rate": 5.162811128947602e-05,
      "loss": 1.055,
      "step": 3641
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.19137662774298073,
      "learning_rate": 5.157588595055073e-05,
      "loss": 1.0918,
      "step": 3642
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.19099229492589737,
      "learning_rate": 5.1523677859264516e-05,
      "loss": 1.1003,
      "step": 3643
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.20765536720077868,
      "learning_rate": 5.147148703421277e-05,
      "loss": 1.0379,
      "step": 3644
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.21925955823356597,
      "learning_rate": 5.141931349398477e-05,
      "loss": 1.123,
      "step": 3645
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.20654082448481326,
      "learning_rate": 5.136715725716358e-05,
      "loss": 1.0345,
      "step": 3646
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2163432793082301,
      "learning_rate": 5.1315018342326124e-05,
      "loss": 1.0873,
      "step": 3647
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.20942514298171064,
      "learning_rate": 5.126289676804328e-05,
      "loss": 1.0379,
      "step": 3648
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.22256202513814938,
      "learning_rate": 5.1210792552879525e-05,
      "loss": 1.152,
      "step": 3649
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1908673935623469,
      "learning_rate": 5.1158705715393364e-05,
      "loss": 0.9017,
      "step": 3650
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.18131079355064286,
      "learning_rate": 5.110663627413694e-05,
      "loss": 1.0157,
      "step": 3651
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2088719233153322,
      "learning_rate": 5.105458424765637e-05,
      "loss": 0.9347,
      "step": 3652
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.21536456269043822,
      "learning_rate": 5.100254965449143e-05,
      "loss": 1.1603,
      "step": 3653
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.20818312175435674,
      "learning_rate": 5.095053251317572e-05,
      "loss": 1.1198,
      "step": 3654
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1602723202918872,
      "learning_rate": 5.089853284223667e-05,
      "loss": 1.1022,
      "step": 3655
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.19966451609777996,
      "learning_rate": 5.0846550660195394e-05,
      "loss": 1.0088,
      "step": 3656
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2089231939790632,
      "learning_rate": 5.079458598556692e-05,
      "loss": 1.0021,
      "step": 3657
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2208191055414772,
      "learning_rate": 5.07426388368599e-05,
      "loss": 1.0348,
      "step": 3658
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.20098038748257707,
      "learning_rate": 5.069070923257685e-05,
      "loss": 0.988,
      "step": 3659
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.19280608758916798,
      "learning_rate": 5.063879719121398e-05,
      "loss": 1.057,
      "step": 3660
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.21271730645540293,
      "learning_rate": 5.058690273126123e-05,
      "loss": 0.9503,
      "step": 3661
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.19018187957959942,
      "learning_rate": 5.053502587120229e-05,
      "loss": 1.1181,
      "step": 3662
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.20300699834688196,
      "learning_rate": 5.0483166629514654e-05,
      "loss": 0.9693,
      "step": 3663
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.21219285231004878,
      "learning_rate": 5.043132502466941e-05,
      "loss": 0.9532,
      "step": 3664
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1768337633938046,
      "learning_rate": 5.0379501075131494e-05,
      "loss": 0.8854,
      "step": 3665
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.21389225242055693,
      "learning_rate": 5.032769479935948e-05,
      "loss": 1.0099,
      "step": 3666
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.22543604727030686,
      "learning_rate": 5.0275906215805625e-05,
      "loss": 0.9659,
      "step": 3667
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.212300407437688,
      "learning_rate": 5.022413534291597e-05,
      "loss": 1.001,
      "step": 3668
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.24778553174466852,
      "learning_rate": 5.0172382199130167e-05,
      "loss": 1.0164,
      "step": 3669
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.18056897292188998,
      "learning_rate": 5.01206468028816e-05,
      "loss": 1.0613,
      "step": 3670
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1886802128946995,
      "learning_rate": 5.006892917259728e-05,
      "loss": 1.0057,
      "step": 3671
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.17263658634208556,
      "learning_rate": 5.001722932669799e-05,
      "loss": 1.0344,
      "step": 3672
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.20021725161783574,
      "learning_rate": 4.996554728359804e-05,
      "loss": 1.1263,
      "step": 3673
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.21027627811913677,
      "learning_rate": 4.9913883061705534e-05,
      "loss": 1.0169,
      "step": 3674
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.21231331570666115,
      "learning_rate": 4.986223667942214e-05,
      "loss": 1.1573,
      "step": 3675
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2396464903424923,
      "learning_rate": 4.981060815514322e-05,
      "loss": 0.9921,
      "step": 3676
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.19482291415581407,
      "learning_rate": 4.9758997507257763e-05,
      "loss": 0.9014,
      "step": 3677
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1991095634324599,
      "learning_rate": 4.970740475414837e-05,
      "loss": 1.2293,
      "step": 3678
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.18062372013867783,
      "learning_rate": 4.965582991419127e-05,
      "loss": 1.1882,
      "step": 3679
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.18814870047680152,
      "learning_rate": 4.960427300575629e-05,
      "loss": 1.0646,
      "step": 3680
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2057543640680821,
      "learning_rate": 4.9552734047206986e-05,
      "loss": 0.9651,
      "step": 3681
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.19672686331611536,
      "learning_rate": 4.9501213056900355e-05,
      "loss": 1.0634,
      "step": 3682
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1852455194510284,
      "learning_rate": 4.944971005318716e-05,
      "loss": 1.0763,
      "step": 3683
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1864117556737071,
      "learning_rate": 4.939822505441163e-05,
      "loss": 1.1493,
      "step": 3684
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2017517268683026,
      "learning_rate": 4.9346758078911634e-05,
      "loss": 0.8787,
      "step": 3685
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.16081001515023188,
      "learning_rate": 4.929530914501857e-05,
      "loss": 0.9266,
      "step": 3686
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.20780037258374115,
      "learning_rate": 4.924387827105754e-05,
      "loss": 1.2286,
      "step": 3687
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.17823953275696702,
      "learning_rate": 4.919246547534708e-05,
      "loss": 1.0422,
      "step": 3688
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.18916001857904094,
      "learning_rate": 4.9141070776199296e-05,
      "loss": 1.1896,
      "step": 3689
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.22696106431200386,
      "learning_rate": 4.908969419191997e-05,
      "loss": 1.1571,
      "step": 3690
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.19914727451703348,
      "learning_rate": 4.903833574080825e-05,
      "loss": 1.0,
      "step": 3691
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.19441469335782055,
      "learning_rate": 4.8986995441157016e-05,
      "loss": 0.9919,
      "step": 3692
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.22796441184597116,
      "learning_rate": 4.893567331125256e-05,
      "loss": 1.0381,
      "step": 3693
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1941344804723652,
      "learning_rate": 4.8884369369374715e-05,
      "loss": 1.1646,
      "step": 3694
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.27828544472708044,
      "learning_rate": 4.8833083633796826e-05,
      "loss": 0.9378,
      "step": 3695
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.18703288183059807,
      "learning_rate": 4.878181612278586e-05,
      "loss": 0.9746,
      "step": 3696
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.23155549654632934,
      "learning_rate": 4.873056685460213e-05,
      "loss": 1.0518,
      "step": 3697
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2550628813249342,
      "learning_rate": 4.867933584749961e-05,
      "loss": 0.8962,
      "step": 3698
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1962057093405795,
      "learning_rate": 4.862812311972567e-05,
      "loss": 1.0105,
      "step": 3699
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.16414306227732933,
      "learning_rate": 4.857692868952113e-05,
      "loss": 1.153,
      "step": 3700
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.18967834208354664,
      "learning_rate": 4.8525752575120454e-05,
      "loss": 1.1719,
      "step": 3701
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.22593224955101798,
      "learning_rate": 4.8474594794751447e-05,
      "loss": 0.9273,
      "step": 3702
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2022068811369253,
      "learning_rate": 4.8423455366635404e-05,
      "loss": 1.0023,
      "step": 3703
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2154581034882868,
      "learning_rate": 4.8372334308987086e-05,
      "loss": 1.0489,
      "step": 3704
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2073812919380501,
      "learning_rate": 4.8321231640014794e-05,
      "loss": 1.0532,
      "step": 3705
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.22651458558693097,
      "learning_rate": 4.827014737792013e-05,
      "loss": 0.8826,
      "step": 3706
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2194373471986128,
      "learning_rate": 4.8219081540898295e-05,
      "loss": 1.1054,
      "step": 3707
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.181372646727082,
      "learning_rate": 4.816803414713782e-05,
      "loss": 1.1404,
      "step": 3708
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.19492842672950092,
      "learning_rate": 4.811700521482071e-05,
      "loss": 1.0302,
      "step": 3709
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2622198896691045,
      "learning_rate": 4.806599476212237e-05,
      "loss": 1.0559,
      "step": 3710
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.21606559512659923,
      "learning_rate": 4.80150028072116e-05,
      "loss": 1.0186,
      "step": 3711
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.17762295938654565,
      "learning_rate": 4.796402936825074e-05,
      "loss": 1.2251,
      "step": 3712
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.17835674804399082,
      "learning_rate": 4.791307446339536e-05,
      "loss": 1.0327,
      "step": 3713
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.22610868096976697,
      "learning_rate": 4.786213811079459e-05,
      "loss": 1.1647,
      "step": 3714
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.24225893143628863,
      "learning_rate": 4.781122032859079e-05,
      "loss": 1.0669,
      "step": 3715
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.21248920203161656,
      "learning_rate": 4.7760321134919875e-05,
      "loss": 1.0014,
      "step": 3716
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.182782654036042,
      "learning_rate": 4.770944054791101e-05,
      "loss": 1.0468,
      "step": 3717
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2144843952430665,
      "learning_rate": 4.7658578585686784e-05,
      "loss": 1.0063,
      "step": 3718
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.24927574677915101,
      "learning_rate": 4.760773526636315e-05,
      "loss": 0.9352,
      "step": 3719
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.18611731423760272,
      "learning_rate": 4.755691060804938e-05,
      "loss": 1.169,
      "step": 3720
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1802773060098242,
      "learning_rate": 4.7506104628848204e-05,
      "loss": 0.977,
      "step": 3721
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.18053864966521616,
      "learning_rate": 4.7455317346855546e-05,
      "loss": 1.0342,
      "step": 3722
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.20039718658916522,
      "learning_rate": 4.740454878016084e-05,
      "loss": 1.1209,
      "step": 3723
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.18348512604496406,
      "learning_rate": 4.735379894684671e-05,
      "loss": 1.1569,
      "step": 3724
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2186572838856738,
      "learning_rate": 4.730306786498921e-05,
      "loss": 0.9395,
      "step": 3725
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.24864428469940208,
      "learning_rate": 4.7252355552657657e-05,
      "loss": 0.9821,
      "step": 3726
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1591789916498842,
      "learning_rate": 4.72016620279147e-05,
      "loss": 0.9446,
      "step": 3727
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.20969218848706528,
      "learning_rate": 4.715098730881623e-05,
      "loss": 1.1807,
      "step": 3728
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.19912093995738595,
      "learning_rate": 4.71003314134116e-05,
      "loss": 0.9708,
      "step": 3729
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.22553539864380623,
      "learning_rate": 4.704969435974334e-05,
      "loss": 1.2266,
      "step": 3730
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1696780478260381,
      "learning_rate": 4.699907616584721e-05,
      "loss": 1.0116,
      "step": 3731
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2559204796815543,
      "learning_rate": 4.694847684975245e-05,
      "loss": 1.0883,
      "step": 3732
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.22086032426981042,
      "learning_rate": 4.68978964294814e-05,
      "loss": 1.0834,
      "step": 3733
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.16762565896540008,
      "learning_rate": 4.684733492304974e-05,
      "loss": 1.104,
      "step": 3734
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1782121169757056,
      "learning_rate": 4.6796792348466356e-05,
      "loss": 1.0054,
      "step": 3735
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.19033742806580706,
      "learning_rate": 4.674626872373351e-05,
      "loss": 1.0815,
      "step": 3736
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2102492173114342,
      "learning_rate": 4.669576406684658e-05,
      "loss": 1.0815,
      "step": 3737
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1898374942590929,
      "learning_rate": 4.6645278395794324e-05,
      "loss": 0.9718,
      "step": 3738
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.19784626324730373,
      "learning_rate": 4.6594811728558584e-05,
      "loss": 1.1052,
      "step": 3739
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.22310061375770035,
      "learning_rate": 4.6544364083114575e-05,
      "loss": 1.09,
      "step": 3740
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1791247852455991,
      "learning_rate": 4.6493935477430664e-05,
      "loss": 1.0485,
      "step": 3741
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.18952234431991302,
      "learning_rate": 4.644352592946842e-05,
      "loss": 1.0973,
      "step": 3742
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.20513984686591907,
      "learning_rate": 4.6393135457182676e-05,
      "loss": 0.9158,
      "step": 3743
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.19031753835314083,
      "learning_rate": 4.634276407852139e-05,
      "loss": 1.185,
      "step": 3744
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1667073549886076,
      "learning_rate": 4.629241181142587e-05,
      "loss": 0.9876,
      "step": 3745
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.30440993498351154,
      "learning_rate": 4.6242078673830426e-05,
      "loss": 1.0311,
      "step": 3746
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.20502359353408572,
      "learning_rate": 4.6191764683662744e-05,
      "loss": 0.921,
      "step": 3747
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.21197483713777004,
      "learning_rate": 4.61414698588435e-05,
      "loss": 1.1552,
      "step": 3748
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.21501718601143452,
      "learning_rate": 4.609119421728674e-05,
      "loss": 1.2088,
      "step": 3749
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.22467784042525285,
      "learning_rate": 4.6040937776899517e-05,
      "loss": 1.0654,
      "step": 3750
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.20012101654781547,
      "learning_rate": 4.599070055558211e-05,
      "loss": 1.117,
      "step": 3751
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.19929575510118974,
      "learning_rate": 4.594048257122797e-05,
      "loss": 1.0009,
      "step": 3752
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.19897900222329626,
      "learning_rate": 4.589028384172361e-05,
      "loss": 1.0021,
      "step": 3753
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.20616439903875752,
      "learning_rate": 4.584010438494885e-05,
      "loss": 0.9866,
      "step": 3754
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1790943589488259,
      "learning_rate": 4.578994421877645e-05,
      "loss": 1.0134,
      "step": 3755
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1885269666083371,
      "learning_rate": 4.573980336107247e-05,
      "loss": 1.0436,
      "step": 3756
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.20107178812281074,
      "learning_rate": 4.568968182969595e-05,
      "loss": 0.9019,
      "step": 3757
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.18324784132144223,
      "learning_rate": 4.563957964249923e-05,
      "loss": 1.0649,
      "step": 3758
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.22044808551689943,
      "learning_rate": 4.558949681732747e-05,
      "loss": 0.9102,
      "step": 3759
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1854168492517875,
      "learning_rate": 4.5539433372019246e-05,
      "loss": 1.1894,
      "step": 3760
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.18666200903387634,
      "learning_rate": 4.548938932440604e-05,
      "loss": 1.0968,
      "step": 3761
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.21266928016638564,
      "learning_rate": 4.543936469231244e-05,
      "loss": 1.0521,
      "step": 3762
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.20805321712115002,
      "learning_rate": 4.538935949355623e-05,
      "loss": 1.0682,
      "step": 3763
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.27360020051688766,
      "learning_rate": 4.533937374594813e-05,
      "loss": 0.8665,
      "step": 3764
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1882145414897454,
      "learning_rate": 4.5289407467292064e-05,
      "loss": 1.0938,
      "step": 3765
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2002764441843406,
      "learning_rate": 4.5239460675384936e-05,
      "loss": 1.0381,
      "step": 3766
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.19444811415298918,
      "learning_rate": 4.5189533388016715e-05,
      "loss": 0.9586,
      "step": 3767
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1902679776128321,
      "learning_rate": 4.5139625622970404e-05,
      "loss": 1.0638,
      "step": 3768
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.19128291879412088,
      "learning_rate": 4.508973739802217e-05,
      "loss": 1.0626,
      "step": 3769
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.21724895586616852,
      "learning_rate": 4.5039868730941046e-05,
      "loss": 1.0239,
      "step": 3770
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2099032529544164,
      "learning_rate": 4.499001963948929e-05,
      "loss": 1.056,
      "step": 3771
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.20146658829615616,
      "learning_rate": 4.494019014142202e-05,
      "loss": 0.9806,
      "step": 3772
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.19412847245334391,
      "learning_rate": 4.489038025448743e-05,
      "loss": 0.9798,
      "step": 3773
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.19742075005994597,
      "learning_rate": 4.484058999642681e-05,
      "loss": 1.1281,
      "step": 3774
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.22063927940913472,
      "learning_rate": 4.479081938497435e-05,
      "loss": 1.0666,
      "step": 3775
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2619118696637535,
      "learning_rate": 4.474106843785728e-05,
      "loss": 1.0298,
      "step": 3776
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.19952296252510207,
      "learning_rate": 4.46913371727958e-05,
      "loss": 1.0851,
      "step": 3777
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.18659686221262264,
      "learning_rate": 4.464162560750319e-05,
      "loss": 1.0578,
      "step": 3778
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1606694529637802,
      "learning_rate": 4.45919337596856e-05,
      "loss": 1.0104,
      "step": 3779
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.20452658201037796,
      "learning_rate": 4.4542261647042263e-05,
      "loss": 1.1181,
      "step": 3780
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.19554162540532125,
      "learning_rate": 4.4492609287265254e-05,
      "loss": 0.881,
      "step": 3781
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.24158862866032849,
      "learning_rate": 4.444297669803981e-05,
      "loss": 1.1687,
      "step": 3782
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.18334138540849246,
      "learning_rate": 4.439336389704388e-05,
      "loss": 1.1242,
      "step": 3783
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.21471109347428063,
      "learning_rate": 4.43437709019485e-05,
      "loss": 0.9737,
      "step": 3784
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.19267417868884834,
      "learning_rate": 4.4294197730417705e-05,
      "loss": 1.0025,
      "step": 3785
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1824530691145748,
      "learning_rate": 4.424464440010834e-05,
      "loss": 0.9791,
      "step": 3786
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1985484698550346,
      "learning_rate": 4.41951109286703e-05,
      "loss": 1.1443,
      "step": 3787
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.20205634611047146,
      "learning_rate": 4.4145597333746314e-05,
      "loss": 1.1249,
      "step": 3788
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.17429078659267913,
      "learning_rate": 4.409610363297211e-05,
      "loss": 1.0322,
      "step": 3789
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.20252079800715697,
      "learning_rate": 4.4046629843976275e-05,
      "loss": 1.0515,
      "step": 3790
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.22839544199253026,
      "learning_rate": 4.39971759843803e-05,
      "loss": 1.0126,
      "step": 3791
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2155737198551163,
      "learning_rate": 4.394774207179859e-05,
      "loss": 1.0149,
      "step": 3792
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.21609170349355492,
      "learning_rate": 4.38983281238385e-05,
      "loss": 0.9111,
      "step": 3793
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.21011317117154585,
      "learning_rate": 4.384893415810021e-05,
      "loss": 1.0752,
      "step": 3794
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.21254804639679203,
      "learning_rate": 4.379956019217675e-05,
      "loss": 0.9022,
      "step": 3795
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.18544049623489417,
      "learning_rate": 4.3750206243654146e-05,
      "loss": 1.0951,
      "step": 3796
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1977040286406834,
      "learning_rate": 4.3700872330111166e-05,
      "loss": 1.0608,
      "step": 3797
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.20417576650503128,
      "learning_rate": 4.365155846911957e-05,
      "loss": 1.0926,
      "step": 3798
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.20259469811341535,
      "learning_rate": 4.3602264678243854e-05,
      "loss": 1.1192,
      "step": 3799
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.20936965292888043,
      "learning_rate": 4.355299097504143e-05,
      "loss": 0.9605,
      "step": 3800
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.21737126359180992,
      "learning_rate": 4.350373737706251e-05,
      "loss": 1.1242,
      "step": 3801
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.32207546833866973,
      "learning_rate": 4.345450390185024e-05,
      "loss": 1.0423,
      "step": 3802
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1769417114933679,
      "learning_rate": 4.340529056694047e-05,
      "loss": 1.1835,
      "step": 3803
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.22147735848144884,
      "learning_rate": 4.335609738986201e-05,
      "loss": 1.1794,
      "step": 3804
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.17882668249406414,
      "learning_rate": 4.3306924388136405e-05,
      "loss": 1.0897,
      "step": 3805
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.22424478415719942,
      "learning_rate": 4.3257771579277964e-05,
      "loss": 1.0534,
      "step": 3806
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.22450796033683057,
      "learning_rate": 4.3208638980794e-05,
      "loss": 1.143,
      "step": 3807
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.187740617320915,
      "learning_rate": 4.3159526610184365e-05,
      "loss": 0.9006,
      "step": 3808
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.24569883996843828,
      "learning_rate": 4.3110434484941934e-05,
      "loss": 1.1081,
      "step": 3809
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.20024604999587173,
      "learning_rate": 4.306136262255219e-05,
      "loss": 1.0183,
      "step": 3810
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2186326782614719,
      "learning_rate": 4.3012311040493594e-05,
      "loss": 1.0587,
      "step": 3811
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.19329695109343067,
      "learning_rate": 4.296327975623716e-05,
      "loss": 1.1265,
      "step": 3812
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.19771518625023163,
      "learning_rate": 4.291426878724691e-05,
      "loss": 1.0933,
      "step": 3813
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.18451397444815648,
      "learning_rate": 4.286527815097944e-05,
      "loss": 1.0215,
      "step": 3814
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.19980706189150496,
      "learning_rate": 4.281630786488418e-05,
      "loss": 1.0381,
      "step": 3815
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2097524627439575,
      "learning_rate": 4.27673579464033e-05,
      "loss": 1.2137,
      "step": 3816
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.30741050206178067,
      "learning_rate": 4.271842841297169e-05,
      "loss": 0.965,
      "step": 3817
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.15836382650563222,
      "learning_rate": 4.266951928201708e-05,
      "loss": 0.9834,
      "step": 3818
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.25228541201491805,
      "learning_rate": 4.2620630570959775e-05,
      "loss": 0.9886,
      "step": 3819
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.20970874698490244,
      "learning_rate": 4.257176229721299e-05,
      "loss": 1.0813,
      "step": 3820
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.20345839222543805,
      "learning_rate": 4.252291447818247e-05,
      "loss": 1.0465,
      "step": 3821
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.206743067488478,
      "learning_rate": 4.247408713126687e-05,
      "loss": 1.2117,
      "step": 3822
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2102053285262028,
      "learning_rate": 4.242528027385738e-05,
      "loss": 0.9534,
      "step": 3823
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.21401788100499336,
      "learning_rate": 4.237649392333799e-05,
      "loss": 1.0255,
      "step": 3824
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.20572637362008164,
      "learning_rate": 4.232772809708535e-05,
      "loss": 1.1783,
      "step": 3825
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.21363989513757015,
      "learning_rate": 4.227898281246877e-05,
      "loss": 1.1044,
      "step": 3826
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.21928897691107407,
      "learning_rate": 4.2230258086850374e-05,
      "loss": 1.0638,
      "step": 3827
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.20593722607128845,
      "learning_rate": 4.2181553937584775e-05,
      "loss": 1.0259,
      "step": 3828
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2517637397699032,
      "learning_rate": 4.213287038201943e-05,
      "loss": 0.9628,
      "step": 3829
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2012111722274609,
      "learning_rate": 4.208420743749434e-05,
      "loss": 1.0812,
      "step": 3830
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.185523968158013,
      "learning_rate": 4.2035565121342246e-05,
      "loss": 1.0608,
      "step": 3831
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2050714984975699,
      "learning_rate": 4.1986943450888506e-05,
      "loss": 1.047,
      "step": 3832
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.22487181061614459,
      "learning_rate": 4.1938342443451095e-05,
      "loss": 1.1069,
      "step": 3833
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2171675157926399,
      "learning_rate": 4.1889762116340635e-05,
      "loss": 1.2897,
      "step": 3834
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.17613902654963323,
      "learning_rate": 4.184120248686048e-05,
      "loss": 1.0322,
      "step": 3835
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.21016912497179774,
      "learning_rate": 4.179266357230649e-05,
      "loss": 0.9735,
      "step": 3836
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3777461208526261,
      "learning_rate": 4.174414538996717e-05,
      "loss": 1.0634,
      "step": 3837
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1948775017025412,
      "learning_rate": 4.1695647957123727e-05,
      "loss": 0.9456,
      "step": 3838
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1533559981468097,
      "learning_rate": 4.164717129104988e-05,
      "loss": 1.0069,
      "step": 3839
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.20934949366359598,
      "learning_rate": 4.159871540901199e-05,
      "loss": 1.2327,
      "step": 3840
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1756657030828761,
      "learning_rate": 4.155028032826897e-05,
      "loss": 1.0486,
      "step": 3841
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.20709901305433062,
      "learning_rate": 4.1501866066072436e-05,
      "loss": 0.9911,
      "step": 3842
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.18522833416730516,
      "learning_rate": 4.1453472639666457e-05,
      "loss": 1.0039,
      "step": 3843
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.23682858200309936,
      "learning_rate": 4.1405100066287803e-05,
      "loss": 0.9103,
      "step": 3844
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.20122457039327202,
      "learning_rate": 4.135674836316569e-05,
      "loss": 0.972,
      "step": 3845
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.24696739741073864,
      "learning_rate": 4.130841754752203e-05,
      "loss": 0.9453,
      "step": 3846
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.21226204622501577,
      "learning_rate": 4.12601076365712e-05,
      "loss": 0.9296,
      "step": 3847
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.19619252635739934,
      "learning_rate": 4.1211818647520164e-05,
      "loss": 0.9715,
      "step": 3848
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1835533342705209,
      "learning_rate": 4.116355059756843e-05,
      "loss": 1.046,
      "step": 3849
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.17886274732943244,
      "learning_rate": 4.111530350390801e-05,
      "loss": 0.9962,
      "step": 3850
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.22247027942972858,
      "learning_rate": 4.106707738372357e-05,
      "loss": 1.019,
      "step": 3851
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1862398652707597,
      "learning_rate": 4.1018872254192166e-05,
      "loss": 1.0398,
      "step": 3852
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2055921056395116,
      "learning_rate": 4.097068813248349e-05,
      "loss": 1.2226,
      "step": 3853
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.31038791217476047,
      "learning_rate": 4.092252503575965e-05,
      "loss": 1.0425,
      "step": 3854
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.16438811451444924,
      "learning_rate": 4.087438298117536e-05,
      "loss": 1.0221,
      "step": 3855
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2089382641416534,
      "learning_rate": 4.082626198587779e-05,
      "loss": 0.9056,
      "step": 3856
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.235986440187898,
      "learning_rate": 4.07781620670066e-05,
      "loss": 1.1151,
      "step": 3857
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.18751043462173056,
      "learning_rate": 4.073008324169396e-05,
      "loss": 0.9414,
      "step": 3858
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.17581222223344511,
      "learning_rate": 4.0682025527064486e-05,
      "loss": 0.9691,
      "step": 3859
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.18485423658517303,
      "learning_rate": 4.063398894023538e-05,
      "loss": 0.9483,
      "step": 3860
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.17903839054587398,
      "learning_rate": 4.05859734983162e-05,
      "loss": 1.0246,
      "step": 3861
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.21382034316605242,
      "learning_rate": 4.053797921840907e-05,
      "loss": 1.0039,
      "step": 3862
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.18853214378637662,
      "learning_rate": 4.049000611760852e-05,
      "loss": 1.2096,
      "step": 3863
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2569952890338686,
      "learning_rate": 4.044205421300154e-05,
      "loss": 1.1289,
      "step": 3864
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.224093232698004,
      "learning_rate": 4.0394123521667524e-05,
      "loss": 0.947,
      "step": 3865
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.20602595278505953,
      "learning_rate": 4.034621406067847e-05,
      "loss": 1.0311,
      "step": 3866
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.18577911597101454,
      "learning_rate": 4.029832584709864e-05,
      "loss": 1.0318,
      "step": 3867
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.22354265152279706,
      "learning_rate": 4.025045889798478e-05,
      "loss": 1.0287,
      "step": 3868
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.19137250146366883,
      "learning_rate": 4.020261323038614e-05,
      "loss": 1.0535,
      "step": 3869
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.18964191758359708,
      "learning_rate": 4.015478886134427e-05,
      "loss": 1.0929,
      "step": 3870
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1887063864828889,
      "learning_rate": 4.010698580789325e-05,
      "loss": 0.984,
      "step": 3871
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1866527660863658,
      "learning_rate": 4.0059204087059485e-05,
      "loss": 1.0354,
      "step": 3872
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.16946090272551093,
      "learning_rate": 4.001144371586181e-05,
      "loss": 0.9663,
      "step": 3873
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2261190083961037,
      "learning_rate": 3.9963704711311404e-05,
      "loss": 0.8204,
      "step": 3874
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.19486186972898262,
      "learning_rate": 3.991598709041196e-05,
      "loss": 1.1177,
      "step": 3875
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1988593744275189,
      "learning_rate": 3.9868290870159405e-05,
      "loss": 1.0254,
      "step": 3876
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.16411345484422074,
      "learning_rate": 3.9820616067542204e-05,
      "loss": 0.9671,
      "step": 3877
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1982698620863389,
      "learning_rate": 3.9772962699541075e-05,
      "loss": 1.1264,
      "step": 3878
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.26877080310538726,
      "learning_rate": 3.972533078312907e-05,
      "loss": 0.9066,
      "step": 3879
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.18655242353340432,
      "learning_rate": 3.967772033527175e-05,
      "loss": 1.0256,
      "step": 3880
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1836659946101919,
      "learning_rate": 3.963013137292691e-05,
      "loss": 1.042,
      "step": 3881
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2082492209549578,
      "learning_rate": 3.958256391304472e-05,
      "loss": 0.9938,
      "step": 3882
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.24288138428202877,
      "learning_rate": 3.9535017972567675e-05,
      "loss": 0.9392,
      "step": 3883
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.19251503584971608,
      "learning_rate": 3.948749356843069e-05,
      "loss": 1.1942,
      "step": 3884
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2127490980360366,
      "learning_rate": 3.943999071756088e-05,
      "loss": 1.043,
      "step": 3885
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.24536558310945994,
      "learning_rate": 3.93925094368778e-05,
      "loss": 0.9177,
      "step": 3886
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2424297858968054,
      "learning_rate": 3.934504974329326e-05,
      "loss": 0.9864,
      "step": 3887
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.21894889845312424,
      "learning_rate": 3.9297611653711383e-05,
      "loss": 1.0254,
      "step": 3888
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.19987187740005344,
      "learning_rate": 3.925019518502861e-05,
      "loss": 0.9182,
      "step": 3889
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.17949790531193038,
      "learning_rate": 3.920280035413365e-05,
      "loss": 1.079,
      "step": 3890
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2179937525213683,
      "learning_rate": 3.915542717790759e-05,
      "loss": 1.1032,
      "step": 3891
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2816902235223324,
      "learning_rate": 3.910807567322369e-05,
      "loss": 0.9855,
      "step": 3892
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.23789791910132488,
      "learning_rate": 3.9060745856947603e-05,
      "loss": 1.0641,
      "step": 3893
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.18824952561228495,
      "learning_rate": 3.901343774593714e-05,
      "loss": 1.1125,
      "step": 3894
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.21237357905510645,
      "learning_rate": 3.896615135704251e-05,
      "loss": 1.0511,
      "step": 3895
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.19489855696936187,
      "learning_rate": 3.8918886707106106e-05,
      "loss": 1.0608,
      "step": 3896
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.21069122920277003,
      "learning_rate": 3.887164381296256e-05,
      "loss": 1.098,
      "step": 3897
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2591149331569423,
      "learning_rate": 3.882442269143877e-05,
      "loss": 0.9819,
      "step": 3898
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.19900310245831138,
      "learning_rate": 3.877722335935394e-05,
      "loss": 1.1597,
      "step": 3899
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.17590679365742687,
      "learning_rate": 3.8730045833519465e-05,
      "loss": 1.0006,
      "step": 3900
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.20870270953090012,
      "learning_rate": 3.868289013073893e-05,
      "loss": 0.9206,
      "step": 3901
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.19241413964392723,
      "learning_rate": 3.863575626780826e-05,
      "loss": 0.8724,
      "step": 3902
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.19816729077817533,
      "learning_rate": 3.858864426151545e-05,
      "loss": 1.0017,
      "step": 3903
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.20229307600132815,
      "learning_rate": 3.854155412864089e-05,
      "loss": 1.0162,
      "step": 3904
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1989886848348331,
      "learning_rate": 3.849448588595703e-05,
      "loss": 0.9675,
      "step": 3905
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.20439553304646402,
      "learning_rate": 3.844743955022859e-05,
      "loss": 0.9536,
      "step": 3906
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.20519802370354287,
      "learning_rate": 3.840041513821243e-05,
      "loss": 1.0878,
      "step": 3907
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.21578486173294684,
      "learning_rate": 3.8353412666657705e-05,
      "loss": 1.043,
      "step": 3908
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2054329121268554,
      "learning_rate": 3.830643215230565e-05,
      "loss": 1.0422,
      "step": 3909
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.18063355030582293,
      "learning_rate": 3.825947361188977e-05,
      "loss": 1.1562,
      "step": 3910
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1681027294348282,
      "learning_rate": 3.821253706213569e-05,
      "loss": 1.0771,
      "step": 3911
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.18360638076321092,
      "learning_rate": 3.816562251976119e-05,
      "loss": 1.0332,
      "step": 3912
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.21994633537794014,
      "learning_rate": 3.811873000147622e-05,
      "loss": 1.0244,
      "step": 3913
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.21402187954608554,
      "learning_rate": 3.80718595239829e-05,
      "loss": 1.0654,
      "step": 3914
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.22442094542193547,
      "learning_rate": 3.802501110397553e-05,
      "loss": 1.2019,
      "step": 3915
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.19256490589961628,
      "learning_rate": 3.7978184758140476e-05,
      "loss": 1.022,
      "step": 3916
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.19333605495423525,
      "learning_rate": 3.793138050315635e-05,
      "loss": 1.121,
      "step": 3917
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.189201830532633,
      "learning_rate": 3.788459835569373e-05,
      "loss": 0.9695,
      "step": 3918
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1883560915763127,
      "learning_rate": 3.783783833241554e-05,
      "loss": 1.1195,
      "step": 3919
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2191899105290029,
      "learning_rate": 3.779110044997665e-05,
      "loss": 0.9932,
      "step": 3920
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2859424022607958,
      "learning_rate": 3.774438472502409e-05,
      "loss": 0.911,
      "step": 3921
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.249901571339984,
      "learning_rate": 3.769769117419703e-05,
      "loss": 1.0791,
      "step": 3922
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.20149514445093483,
      "learning_rate": 3.7651019814126654e-05,
      "loss": 1.0155,
      "step": 3923
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2193992417997238,
      "learning_rate": 3.760437066143641e-05,
      "loss": 0.9274,
      "step": 3924
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.17209124009021323,
      "learning_rate": 3.7557743732741625e-05,
      "loss": 0.9865,
      "step": 3925
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.17949752171882058,
      "learning_rate": 3.7511139044649926e-05,
      "loss": 1.0597,
      "step": 3926
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1940182758692254,
      "learning_rate": 3.7464556613760826e-05,
      "loss": 0.9521,
      "step": 3927
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.21271446214023632,
      "learning_rate": 3.741799645666606e-05,
      "loss": 1.0267,
      "step": 3928
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.17312669776748982,
      "learning_rate": 3.737145858994934e-05,
      "loss": 1.1007,
      "step": 3929
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.21070641341897198,
      "learning_rate": 3.7324943030186455e-05,
      "loss": 1.069,
      "step": 3930
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.20277248885960938,
      "learning_rate": 3.727844979394526e-05,
      "loss": 0.9343,
      "step": 3931
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.20339354198178988,
      "learning_rate": 3.7231978897785635e-05,
      "loss": 1.1421,
      "step": 3932
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.19302470399600025,
      "learning_rate": 3.7185530358259565e-05,
      "loss": 1.0445,
      "step": 3933
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.25480009592359093,
      "learning_rate": 3.7139104191911004e-05,
      "loss": 1.0881,
      "step": 3934
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1818183757763989,
      "learning_rate": 3.709270041527599e-05,
      "loss": 0.9683,
      "step": 3935
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2268590156590052,
      "learning_rate": 3.704631904488256e-05,
      "loss": 1.005,
      "step": 3936
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.22996737056089173,
      "learning_rate": 3.6999960097250755e-05,
      "loss": 1.0446,
      "step": 3937
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2205222013724746,
      "learning_rate": 3.6953623588892603e-05,
      "loss": 1.1651,
      "step": 3938
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.25683722272044296,
      "learning_rate": 3.6907309536312276e-05,
      "loss": 1.0256,
      "step": 3939
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.17874177731129884,
      "learning_rate": 3.686101795600576e-05,
      "loss": 1.1775,
      "step": 3940
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.17582287917374526,
      "learning_rate": 3.6814748864461216e-05,
      "loss": 1.1529,
      "step": 3941
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.17083255991661236,
      "learning_rate": 3.6768502278158665e-05,
      "loss": 1.0947,
      "step": 3942
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.22113702856575182,
      "learning_rate": 3.672227821357014e-05,
      "loss": 0.9891,
      "step": 3943
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.23889378180638288,
      "learning_rate": 3.667607668715972e-05,
      "loss": 1.0186,
      "step": 3944
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.17625682411922436,
      "learning_rate": 3.662989771538338e-05,
      "loss": 0.9765,
      "step": 3945
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.20183753715144873,
      "learning_rate": 3.6583741314689076e-05,
      "loss": 1.0036,
      "step": 3946
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1984769884810475,
      "learning_rate": 3.6537607501516715e-05,
      "loss": 1.2281,
      "step": 3947
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2048338765095224,
      "learning_rate": 3.6491496292298246e-05,
      "loss": 1.0577,
      "step": 3948
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.19035052871181724,
      "learning_rate": 3.644540770345742e-05,
      "loss": 1.0202,
      "step": 3949
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.24729797473440931,
      "learning_rate": 3.63993417514101e-05,
      "loss": 1.0302,
      "step": 3950
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.19221234036370713,
      "learning_rate": 3.635329845256391e-05,
      "loss": 1.0922,
      "step": 3951
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.17309507861884282,
      "learning_rate": 3.6307277823318574e-05,
      "loss": 1.1552,
      "step": 3952
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1824057895095207,
      "learning_rate": 3.626127988006563e-05,
      "loss": 0.8883,
      "step": 3953
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2035648365777874,
      "learning_rate": 3.6215304639188555e-05,
      "loss": 1.1091,
      "step": 3954
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.19346100864156596,
      "learning_rate": 3.616935211706275e-05,
      "loss": 1.0519,
      "step": 3955
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1874566249351973,
      "learning_rate": 3.61234223300555e-05,
      "loss": 0.9218,
      "step": 3956
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.18436651116250527,
      "learning_rate": 3.607751529452609e-05,
      "loss": 1.1057,
      "step": 3957
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.22461363577213095,
      "learning_rate": 3.6031631026825554e-05,
      "loss": 1.1968,
      "step": 3958
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.16331886240206722,
      "learning_rate": 3.598576954329695e-05,
      "loss": 0.9641,
      "step": 3959
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2501771716878642,
      "learning_rate": 3.5939930860275115e-05,
      "loss": 1.0431,
      "step": 3960
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2062122584482938,
      "learning_rate": 3.589411499408691e-05,
      "loss": 1.0233,
      "step": 3961
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.19099814599417136,
      "learning_rate": 3.5848321961050823e-05,
      "loss": 0.8929,
      "step": 3962
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2086497145867355,
      "learning_rate": 3.5802551777477476e-05,
      "loss": 1.0646,
      "step": 3963
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.21883704454308409,
      "learning_rate": 3.575680445966918e-05,
      "loss": 1.0608,
      "step": 3964
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.20225348986792074,
      "learning_rate": 3.571108002392014e-05,
      "loss": 1.1406,
      "step": 3965
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1824243433842943,
      "learning_rate": 3.56653784865165e-05,
      "loss": 1.1179,
      "step": 3966
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.15261154397017446,
      "learning_rate": 3.5619699863736076e-05,
      "loss": 1.1616,
      "step": 3967
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.18065259417579044,
      "learning_rate": 3.557404417184873e-05,
      "loss": 0.9765,
      "step": 3968
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.17076774285910976,
      "learning_rate": 3.5528411427115984e-05,
      "loss": 1.0709,
      "step": 3969
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.20575131592310605,
      "learning_rate": 3.548280164579126e-05,
      "loss": 1.0631,
      "step": 3970
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2090243601855649,
      "learning_rate": 3.543721484411976e-05,
      "loss": 1.0311,
      "step": 3971
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.23507103463508522,
      "learning_rate": 3.53916510383386e-05,
      "loss": 0.9733,
      "step": 3972
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.19040508012156943,
      "learning_rate": 3.5346110244676564e-05,
      "loss": 0.9564,
      "step": 3973
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.17430147486609576,
      "learning_rate": 3.53005924793544e-05,
      "loss": 1.1007,
      "step": 3974
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.20657459554042928,
      "learning_rate": 3.525509775858451e-05,
      "loss": 1.147,
      "step": 3975
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1930403635465425,
      "learning_rate": 3.520962609857112e-05,
      "loss": 1.0936,
      "step": 3976
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.21166703060423703,
      "learning_rate": 3.516417751551035e-05,
      "loss": 1.1826,
      "step": 3977
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.19508929779574263,
      "learning_rate": 3.5118752025589965e-05,
      "loss": 0.8471,
      "step": 3978
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.20465450244668645,
      "learning_rate": 3.5073349644989564e-05,
      "loss": 1.1826,
      "step": 3979
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1766111905620841,
      "learning_rate": 3.5027970389880486e-05,
      "loss": 1.1388,
      "step": 3980
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.23210210128050685,
      "learning_rate": 3.498261427642589e-05,
      "loss": 1.038,
      "step": 3981
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.22411499022030282,
      "learning_rate": 3.493728132078064e-05,
      "loss": 1.0853,
      "step": 3982
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1971033219145773,
      "learning_rate": 3.489197153909139e-05,
      "loss": 1.1351,
      "step": 3983
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.22295631230755705,
      "learning_rate": 3.484668494749653e-05,
      "loss": 1.0527,
      "step": 3984
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.26147870885609825,
      "learning_rate": 3.4801421562126144e-05,
      "loss": 1.0803,
      "step": 3985
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.20542600494611601,
      "learning_rate": 3.47561813991021e-05,
      "loss": 1.009,
      "step": 3986
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.17856576955447553,
      "learning_rate": 3.4710964474537966e-05,
      "loss": 1.1168,
      "step": 3987
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.22540227653514214,
      "learning_rate": 3.466577080453909e-05,
      "loss": 1.0033,
      "step": 3988
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.24288635840030962,
      "learning_rate": 3.462060040520244e-05,
      "loss": 0.969,
      "step": 3989
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.20189732675924968,
      "learning_rate": 3.457545329261681e-05,
      "loss": 0.9974,
      "step": 3990
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1783759442449083,
      "learning_rate": 3.4530329482862587e-05,
      "loss": 1.1631,
      "step": 3991
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2061914699080313,
      "learning_rate": 3.448522899201198e-05,
      "loss": 1.065,
      "step": 3992
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.18772001319112772,
      "learning_rate": 3.444015183612876e-05,
      "loss": 1.1556,
      "step": 3993
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.20077988948800252,
      "learning_rate": 3.439509803126848e-05,
      "loss": 1.1075,
      "step": 3994
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.18976880265653243,
      "learning_rate": 3.4350067593478356e-05,
      "loss": 1.1259,
      "step": 3995
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.22118836945584747,
      "learning_rate": 3.4305060538797206e-05,
      "loss": 1.1071,
      "step": 3996
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.20220804753865204,
      "learning_rate": 3.426007688325566e-05,
      "loss": 1.057,
      "step": 3997
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.21284496097226835,
      "learning_rate": 3.42151166428759e-05,
      "loss": 1.287,
      "step": 3998
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.20458660397534,
      "learning_rate": 3.4170179833671846e-05,
      "loss": 1.1163,
      "step": 3999
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1926278327903263,
      "learning_rate": 3.4125266471648976e-05,
      "loss": 1.0933,
      "step": 4000
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.18763631988994692,
      "learning_rate": 3.408037657280454e-05,
      "loss": 1.052,
      "step": 4001
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.18389597018058387,
      "learning_rate": 3.403551015312734e-05,
      "loss": 1.0234,
      "step": 4002
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2058746202245206,
      "learning_rate": 3.399066722859782e-05,
      "loss": 1.1368,
      "step": 4003
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.17973763087465797,
      "learning_rate": 3.394584781518805e-05,
      "loss": 0.9542,
      "step": 4004
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.20052509467456414,
      "learning_rate": 3.390105192886182e-05,
      "loss": 1.0304,
      "step": 4005
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.20665072191705278,
      "learning_rate": 3.385627958557445e-05,
      "loss": 1.0191,
      "step": 4006
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1910896972369258,
      "learning_rate": 3.381153080127284e-05,
      "loss": 1.0235,
      "step": 4007
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.21737510528810514,
      "learning_rate": 3.3766805591895636e-05,
      "loss": 0.9598,
      "step": 4008
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.19997104222451334,
      "learning_rate": 3.372210397337294e-05,
      "loss": 1.0487,
      "step": 4009
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.166430105027456,
      "learning_rate": 3.367742596162661e-05,
      "loss": 1.0433,
      "step": 4010
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.20339187621568186,
      "learning_rate": 3.363277157256988e-05,
      "loss": 1.0509,
      "step": 4011
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.16747266657021392,
      "learning_rate": 3.3588140822107784e-05,
      "loss": 1.1187,
      "step": 4012
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.18801338250935398,
      "learning_rate": 3.354353372613678e-05,
      "loss": 1.1,
      "step": 4013
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1946515816819778,
      "learning_rate": 3.349895030054504e-05,
      "loss": 1.0032,
      "step": 4014
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.18716153372913655,
      "learning_rate": 3.345439056121216e-05,
      "loss": 0.9618,
      "step": 4015
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.18245095054762195,
      "learning_rate": 3.340985452400943e-05,
      "loss": 1.0955,
      "step": 4016
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1997179748116679,
      "learning_rate": 3.336534220479961e-05,
      "loss": 1.0648,
      "step": 4017
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1815044587603543,
      "learning_rate": 3.332085361943705e-05,
      "loss": 1.0117,
      "step": 4018
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2369608873917558,
      "learning_rate": 3.327638878376764e-05,
      "loss": 1.0498,
      "step": 4019
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.17781602542370079,
      "learning_rate": 3.323194771362875e-05,
      "loss": 1.0744,
      "step": 4020
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.17088915663431237,
      "learning_rate": 3.318753042484944e-05,
      "loss": 1.0735,
      "step": 4021
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.23125739097634337,
      "learning_rate": 3.3143136933250117e-05,
      "loss": 0.9827,
      "step": 4022
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.18015080616616755,
      "learning_rate": 3.3098767254642883e-05,
      "loss": 1.0201,
      "step": 4023
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3005272845891236,
      "learning_rate": 3.305442140483118e-05,
      "loss": 0.9651,
      "step": 4024
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.20672819761388792,
      "learning_rate": 3.301009939961014e-05,
      "loss": 1.0048,
      "step": 4025
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.18575448907572573,
      "learning_rate": 3.296580125476629e-05,
      "loss": 1.0446,
      "step": 4026
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.19692904189692287,
      "learning_rate": 3.292152698607768e-05,
      "loss": 1.0984,
      "step": 4027
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.20020444927931505,
      "learning_rate": 3.287727660931385e-05,
      "loss": 0.9902,
      "step": 4028
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1644454250198622,
      "learning_rate": 3.2833050140235824e-05,
      "loss": 1.0923,
      "step": 4029
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.24004188023915177,
      "learning_rate": 3.278884759459619e-05,
      "loss": 1.0006,
      "step": 4030
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.20430839728059985,
      "learning_rate": 3.27446689881389e-05,
      "loss": 1.1117,
      "step": 4031
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.17990729689923787,
      "learning_rate": 3.2700514336599483e-05,
      "loss": 1.0961,
      "step": 4032
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.21776824329657418,
      "learning_rate": 3.2656383655704835e-05,
      "loss": 1.1259,
      "step": 4033
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.16186470711718717,
      "learning_rate": 3.2612276961173416e-05,
      "loss": 1.0214,
      "step": 4034
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2001478168767885,
      "learning_rate": 3.256819426871507e-05,
      "loss": 1.0209,
      "step": 4035
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.20819511125117746,
      "learning_rate": 3.2524135594031116e-05,
      "loss": 1.0408,
      "step": 4036
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.202735732785506,
      "learning_rate": 3.2480100952814305e-05,
      "loss": 0.9796,
      "step": 4037
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.199799335809838,
      "learning_rate": 3.243609036074882e-05,
      "loss": 1.0987,
      "step": 4038
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.21686318392201315,
      "learning_rate": 3.239210383351034e-05,
      "loss": 1.0879,
      "step": 4039
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1825063491118461,
      "learning_rate": 3.23481413867659e-05,
      "loss": 1.0148,
      "step": 4040
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.16292149987652255,
      "learning_rate": 3.2304203036174016e-05,
      "loss": 1.045,
      "step": 4041
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.18747891498821959,
      "learning_rate": 3.22602887973846e-05,
      "loss": 1.1488,
      "step": 4042
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.18742123712957576,
      "learning_rate": 3.2216398686038926e-05,
      "loss": 1.0949,
      "step": 4043
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.18149635749489604,
      "learning_rate": 3.2172532717769734e-05,
      "loss": 1.0103,
      "step": 4044
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2000695860987263,
      "learning_rate": 3.212869090820116e-05,
      "loss": 0.9737,
      "step": 4045
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.17983991307788508,
      "learning_rate": 3.2084873272948705e-05,
      "loss": 0.8792,
      "step": 4046
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1882555514961866,
      "learning_rate": 3.2041079827619314e-05,
      "loss": 1.0692,
      "step": 4047
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.19633796579397522,
      "learning_rate": 3.199731058781127e-05,
      "loss": 1.0376,
      "step": 4048
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2064074341761319,
      "learning_rate": 3.195356556911418e-05,
      "loss": 1.1379,
      "step": 4049
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.27830622217350665,
      "learning_rate": 3.1909844787109186e-05,
      "loss": 1.1326,
      "step": 4050
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.18380829815142613,
      "learning_rate": 3.1866148257368665e-05,
      "loss": 1.064,
      "step": 4051
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.20738765471865964,
      "learning_rate": 3.182247599545637e-05,
      "loss": 1.0488,
      "step": 4052
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.19803626067592744,
      "learning_rate": 3.177882801692742e-05,
      "loss": 1.0077,
      "step": 4053
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.17996830928688567,
      "learning_rate": 3.173520433732834e-05,
      "loss": 1.1522,
      "step": 4054
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.23592938100652738,
      "learning_rate": 3.169160497219692e-05,
      "loss": 1.0118,
      "step": 4055
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.17592491596326862,
      "learning_rate": 3.164802993706236e-05,
      "loss": 1.0173,
      "step": 4056
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1863568761520776,
      "learning_rate": 3.160447924744512e-05,
      "loss": 0.9045,
      "step": 4057
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.21176043604123665,
      "learning_rate": 3.156095291885708e-05,
      "loss": 1.1064,
      "step": 4058
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.20526135693621614,
      "learning_rate": 3.15174509668014e-05,
      "loss": 0.8742,
      "step": 4059
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2163193627108563,
      "learning_rate": 3.147397340677244e-05,
      "loss": 1.1559,
      "step": 4060
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.25034839721355384,
      "learning_rate": 3.14305202542561e-05,
      "loss": 1.1806,
      "step": 4061
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1877799190139149,
      "learning_rate": 3.138709152472938e-05,
      "loss": 0.9067,
      "step": 4062
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1690656909250283,
      "learning_rate": 3.1343687233660754e-05,
      "loss": 1.1159,
      "step": 4063
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.20575956652524136,
      "learning_rate": 3.130030739650983e-05,
      "loss": 1.0649,
      "step": 4064
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.19876463154456844,
      "learning_rate": 3.125695202872765e-05,
      "loss": 0.9742,
      "step": 4065
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.18370536145109798,
      "learning_rate": 3.1213621145756447e-05,
      "loss": 1.0549,
      "step": 4066
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.18442889416049407,
      "learning_rate": 3.117031476302975e-05,
      "loss": 1.23,
      "step": 4067
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.21169569313796605,
      "learning_rate": 3.112703289597234e-05,
      "loss": 1.026,
      "step": 4068
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.18568495010942423,
      "learning_rate": 3.108377556000037e-05,
      "loss": 1.1699,
      "step": 4069
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.19423479634904492,
      "learning_rate": 3.1040542770521156e-05,
      "loss": 0.8972,
      "step": 4070
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.19549733898785324,
      "learning_rate": 3.099733454293324e-05,
      "loss": 1.0952,
      "step": 4071
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.21955102376338026,
      "learning_rate": 3.095415089262657e-05,
      "loss": 1.0081,
      "step": 4072
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.21928294905076673,
      "learning_rate": 3.091099183498214e-05,
      "loss": 0.9338,
      "step": 4073
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.23758611231444393,
      "learning_rate": 3.086785738537239e-05,
      "loss": 1.2163,
      "step": 4074
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2030667234536358,
      "learning_rate": 3.0824747559160836e-05,
      "loss": 0.9896,
      "step": 4075
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.18338065712913965,
      "learning_rate": 3.078166237170229e-05,
      "loss": 0.9898,
      "step": 4076
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.24950983734900278,
      "learning_rate": 3.073860183834276e-05,
      "loss": 1.0693,
      "step": 4077
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.21391298538286735,
      "learning_rate": 3.069556597441953e-05,
      "loss": 1.1417,
      "step": 4078
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.20943814353543816,
      "learning_rate": 3.065255479526101e-05,
      "loss": 0.9186,
      "step": 4079
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.20205923796446545,
      "learning_rate": 3.0609568316186935e-05,
      "loss": 0.9005,
      "step": 4080
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.18954214363003463,
      "learning_rate": 3.056660655250813e-05,
      "loss": 1.1136,
      "step": 4081
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.18143490611824156,
      "learning_rate": 3.052366951952663e-05,
      "loss": 0.8761,
      "step": 4082
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1892778878317215,
      "learning_rate": 3.0480757232535772e-05,
      "loss": 0.9959,
      "step": 4083
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.21257138794560096,
      "learning_rate": 3.0437869706819955e-05,
      "loss": 1.0654,
      "step": 4084
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.19267356465958452,
      "learning_rate": 3.0395006957654802e-05,
      "loss": 0.969,
      "step": 4085
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2063789570268476,
      "learning_rate": 3.0352169000307086e-05,
      "loss": 0.9996,
      "step": 4086
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.17643236123687564,
      "learning_rate": 3.0309355850034837e-05,
      "loss": 1.1559,
      "step": 4087
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.20038764807721016,
      "learning_rate": 3.026656752208712e-05,
      "loss": 1.1918,
      "step": 4088
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.21691989703808665,
      "learning_rate": 3.0223804031704295e-05,
      "loss": 1.0603,
      "step": 4089
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.20327958089394152,
      "learning_rate": 3.018106539411778e-05,
      "loss": 1.1024,
      "step": 4090
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.19548051443936726,
      "learning_rate": 3.0138351624550164e-05,
      "loss": 0.9677,
      "step": 4091
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.22527336196970185,
      "learning_rate": 3.009566273821518e-05,
      "loss": 1.0121,
      "step": 4092
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.18431236453547648,
      "learning_rate": 3.0052998750317675e-05,
      "loss": 0.9478,
      "step": 4093
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.22506737728293613,
      "learning_rate": 3.0010359676053713e-05,
      "loss": 0.9889,
      "step": 4094
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.16504178651643,
      "learning_rate": 2.9967745530610357e-05,
      "loss": 1.1707,
      "step": 4095
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2083858381908209,
      "learning_rate": 2.9925156329165937e-05,
      "loss": 1.0744,
      "step": 4096
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.18925087915904087,
      "learning_rate": 2.9882592086889736e-05,
      "loss": 1.0557,
      "step": 4097
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.177657776872756,
      "learning_rate": 2.984005281894232e-05,
      "loss": 1.0051,
      "step": 4098
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.19278717062697434,
      "learning_rate": 2.979753854047522e-05,
      "loss": 1.191,
      "step": 4099
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.21280691273707616,
      "learning_rate": 2.975504926663113e-05,
      "loss": 1.134,
      "step": 4100
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.18658813223806306,
      "learning_rate": 2.9712585012543816e-05,
      "loss": 1.0517,
      "step": 4101
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1974254525435282,
      "learning_rate": 2.967014579333811e-05,
      "loss": 1.0424,
      "step": 4102
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.22656827274788938,
      "learning_rate": 2.9627731624130016e-05,
      "loss": 0.9968,
      "step": 4103
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.20261716206004812,
      "learning_rate": 2.9585342520026514e-05,
      "loss": 0.8867,
      "step": 4104
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.18611497814695707,
      "learning_rate": 2.9542978496125762e-05,
      "loss": 1.0578,
      "step": 4105
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.22406824474177828,
      "learning_rate": 2.9500639567516842e-05,
      "loss": 1.1476,
      "step": 4106
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2192577505978438,
      "learning_rate": 2.9458325749280057e-05,
      "loss": 1.0432,
      "step": 4107
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1953254448144994,
      "learning_rate": 2.9416037056486658e-05,
      "loss": 0.9753,
      "step": 4108
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.16780149429949093,
      "learning_rate": 2.9373773504198963e-05,
      "loss": 1.1845,
      "step": 4109
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.26911482674495796,
      "learning_rate": 2.9331535107470342e-05,
      "loss": 1.1265,
      "step": 4110
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.17421430855153827,
      "learning_rate": 2.9289321881345254e-05,
      "loss": 1.0944,
      "step": 4111
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.19849248221811144,
      "learning_rate": 2.924713384085913e-05,
      "loss": 0.945,
      "step": 4112
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2075839156998247,
      "learning_rate": 2.9204971001038428e-05,
      "loss": 0.8247,
      "step": 4113
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1728118455298111,
      "learning_rate": 2.91628333769007e-05,
      "loss": 0.9255,
      "step": 4114
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1964163092723989,
      "learning_rate": 2.9120720983454463e-05,
      "loss": 0.9555,
      "step": 4115
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.19387108718838664,
      "learning_rate": 2.9078633835699233e-05,
      "loss": 0.9521,
      "step": 4116
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.19609848961764786,
      "learning_rate": 2.9036571948625537e-05,
      "loss": 1.0531,
      "step": 4117
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.17825639982059685,
      "learning_rate": 2.8994535337214978e-05,
      "loss": 1.0929,
      "step": 4118
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.23464736347577805,
      "learning_rate": 2.8952524016440052e-05,
      "loss": 0.9492,
      "step": 4119
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.20663456965855653,
      "learning_rate": 2.8910538001264354e-05,
      "loss": 1.0968,
      "step": 4120
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.19408559276722126,
      "learning_rate": 2.8868577306642343e-05,
      "loss": 1.0548,
      "step": 4121
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.20483264440604756,
      "learning_rate": 2.8826641947519595e-05,
      "loss": 1.0463,
      "step": 4122
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.28323926299345764,
      "learning_rate": 2.8784731938832556e-05,
      "loss": 1.0223,
      "step": 4123
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.18930135815287832,
      "learning_rate": 2.8742847295508678e-05,
      "loss": 1.0848,
      "step": 4124
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.24678254777355443,
      "learning_rate": 2.8700988032466392e-05,
      "loss": 1.1168,
      "step": 4125
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.18949050228290526,
      "learning_rate": 2.865915416461503e-05,
      "loss": 1.0992,
      "step": 4126
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.17820264079368653,
      "learning_rate": 2.8617345706855003e-05,
      "loss": 1.0209,
      "step": 4127
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.19876237786440876,
      "learning_rate": 2.8575562674077517e-05,
      "loss": 1.0398,
      "step": 4128
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1880162389938706,
      "learning_rate": 2.8533805081164867e-05,
      "loss": 1.089,
      "step": 4129
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2508561468653909,
      "learning_rate": 2.8492072942990157e-05,
      "loss": 1.2083,
      "step": 4130
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.22516773485612188,
      "learning_rate": 2.845036627441755e-05,
      "loss": 1.0658,
      "step": 4131
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.20292356619389745,
      "learning_rate": 2.8408685090302044e-05,
      "loss": 1.0784,
      "step": 4132
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.21318642077567182,
      "learning_rate": 2.8367029405489597e-05,
      "loss": 0.9487,
      "step": 4133
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.22101062044712352,
      "learning_rate": 2.8325399234817064e-05,
      "loss": 0.9498,
      "step": 4134
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.19047061795284043,
      "learning_rate": 2.828379459311219e-05,
      "loss": 1.0127,
      "step": 4135
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.17863938580184502,
      "learning_rate": 2.8242215495193746e-05,
      "loss": 0.9993,
      "step": 4136
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2076418646132173,
      "learning_rate": 2.8200661955871242e-05,
      "loss": 1.0148,
      "step": 4137
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2416029354521076,
      "learning_rate": 2.815913398994524e-05,
      "loss": 1.0222,
      "step": 4138
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.19098023579494416,
      "learning_rate": 2.8117631612207084e-05,
      "loss": 1.1867,
      "step": 4139
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.17352643026914347,
      "learning_rate": 2.8076154837439027e-05,
      "loss": 1.0581,
      "step": 4140
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.17673579026784728,
      "learning_rate": 2.8034703680414186e-05,
      "loss": 1.3195,
      "step": 4141
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.22112855206402635,
      "learning_rate": 2.7993278155896653e-05,
      "loss": 1.0781,
      "step": 4142
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.17899530968152055,
      "learning_rate": 2.7951878278641285e-05,
      "loss": 0.9867,
      "step": 4143
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2231840150257677,
      "learning_rate": 2.7910504063393793e-05,
      "loss": 1.0281,
      "step": 4144
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.20467947935445452,
      "learning_rate": 2.786915552489088e-05,
      "loss": 0.9792,
      "step": 4145
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.18830198073136004,
      "learning_rate": 2.782783267785992e-05,
      "loss": 1.0599,
      "step": 4146
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1985258887415792,
      "learning_rate": 2.778653553701932e-05,
      "loss": 1.1232,
      "step": 4147
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.22558669803841508,
      "learning_rate": 2.7745264117078207e-05,
      "loss": 1.0396,
      "step": 4148
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.20307522915340193,
      "learning_rate": 2.7704018432736566e-05,
      "loss": 0.9484,
      "step": 4149
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.20670632096450664,
      "learning_rate": 2.7662798498685227e-05,
      "loss": 0.7949,
      "step": 4150
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.16923090829159382,
      "learning_rate": 2.7621604329605898e-05,
      "loss": 1.0197,
      "step": 4151
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.19038734987623507,
      "learning_rate": 2.7580435940171e-05,
      "loss": 0.9988,
      "step": 4152
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.21642487582668304,
      "learning_rate": 2.7539293345043926e-05,
      "loss": 1.1246,
      "step": 4153
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.21912517704766055,
      "learning_rate": 2.749817655887873e-05,
      "loss": 1.0594,
      "step": 4154
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.19727488606218696,
      "learning_rate": 2.745708559632032e-05,
      "loss": 0.9973,
      "step": 4155
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.18270320222418704,
      "learning_rate": 2.741602047200448e-05,
      "loss": 0.9462,
      "step": 4156
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.21958721079673718,
      "learning_rate": 2.7374981200557703e-05,
      "loss": 1.0145,
      "step": 4157
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3062557528843728,
      "learning_rate": 2.7333967796597315e-05,
      "loss": 1.0677,
      "step": 4158
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.16583958815849312,
      "learning_rate": 2.7292980274731382e-05,
      "loss": 1.1535,
      "step": 4159
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.19269848539074258,
      "learning_rate": 2.7252018649558842e-05,
      "loss": 1.0047,
      "step": 4160
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2219033174474081,
      "learning_rate": 2.7211082935669285e-05,
      "loss": 1.025,
      "step": 4161
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1796080202321257,
      "learning_rate": 2.717017314764323e-05,
      "loss": 0.9119,
      "step": 4162
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2050429806921932,
      "learning_rate": 2.7129289300051787e-05,
      "loss": 0.92,
      "step": 4163
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.17268925146467373,
      "learning_rate": 2.7088431407457015e-05,
      "loss": 1.042,
      "step": 4164
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.20464115735105406,
      "learning_rate": 2.7047599484411535e-05,
      "loss": 1.1206,
      "step": 4165
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1929740897514812,
      "learning_rate": 2.7006793545458807e-05,
      "loss": 1.2083,
      "step": 4166
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.19750760685744204,
      "learning_rate": 2.6966013605133088e-05,
      "loss": 1.1546,
      "step": 4167
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.18522950852889136,
      "learning_rate": 2.692525967795929e-05,
      "loss": 1.1122,
      "step": 4168
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.19844846616944214,
      "learning_rate": 2.6884531778453126e-05,
      "loss": 1.1068,
      "step": 4169
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.19213032885077366,
      "learning_rate": 2.684382992112098e-05,
      "loss": 1.1014,
      "step": 4170
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1990775909839337,
      "learning_rate": 2.6803154120460007e-05,
      "loss": 1.0569,
      "step": 4171
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.21338966670107457,
      "learning_rate": 2.6762504390958066e-05,
      "loss": 1.1042,
      "step": 4172
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.19701972213148514,
      "learning_rate": 2.672188074709371e-05,
      "loss": 1.1708,
      "step": 4173
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.18591445049631802,
      "learning_rate": 2.6681283203336193e-05,
      "loss": 1.0761,
      "step": 4174
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.16273084287923367,
      "learning_rate": 2.6640711774145543e-05,
      "loss": 1.0203,
      "step": 4175
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.20794026728870946,
      "learning_rate": 2.660016647397243e-05,
      "loss": 0.9769,
      "step": 4176
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.22084370194835007,
      "learning_rate": 2.6559647317258183e-05,
      "loss": 0.8847,
      "step": 4177
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.19232075340085963,
      "learning_rate": 2.6519154318434934e-05,
      "loss": 0.9635,
      "step": 4178
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2071180218558523,
      "learning_rate": 2.6478687491925357e-05,
      "loss": 1.0429,
      "step": 4179
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.18091741706858672,
      "learning_rate": 2.6438246852142934e-05,
      "loss": 1.128,
      "step": 4180
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2257562726877083,
      "learning_rate": 2.639783241349174e-05,
      "loss": 1.0412,
      "step": 4181
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1755614206766948,
      "learning_rate": 2.635744419036653e-05,
      "loss": 0.9327,
      "step": 4182
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.22425399064347515,
      "learning_rate": 2.63170821971527e-05,
      "loss": 0.9394,
      "step": 4183
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2058161256131396,
      "learning_rate": 2.6276746448226398e-05,
      "loss": 0.9062,
      "step": 4184
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.20904507627129976,
      "learning_rate": 2.6236436957954304e-05,
      "loss": 1.069,
      "step": 4185
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.19013712508296277,
      "learning_rate": 2.6196153740693842e-05,
      "loss": 1.1333,
      "step": 4186
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.17136423885770088,
      "learning_rate": 2.6155896810793036e-05,
      "loss": 0.8611,
      "step": 4187
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.238489464430118,
      "learning_rate": 2.6115666182590513e-05,
      "loss": 1.0065,
      "step": 4188
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.20740962316257922,
      "learning_rate": 2.6075461870415595e-05,
      "loss": 0.9357,
      "step": 4189
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.20916506496335519,
      "learning_rate": 2.6035283888588158e-05,
      "loss": 0.8919,
      "step": 4190
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.21461382060444747,
      "learning_rate": 2.599513225141881e-05,
      "loss": 0.9587,
      "step": 4191
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.21192430526379694,
      "learning_rate": 2.595500697320865e-05,
      "loss": 1.0311,
      "step": 4192
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.21700129734655024,
      "learning_rate": 2.59149080682495e-05,
      "loss": 1.1456,
      "step": 4193
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2131817786740426,
      "learning_rate": 2.5874835550823683e-05,
      "loss": 1.1244,
      "step": 4194
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.18840585421325987,
      "learning_rate": 2.5834789435204243e-05,
      "loss": 1.129,
      "step": 4195
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.20881764545868065,
      "learning_rate": 2.5794769735654723e-05,
      "loss": 1.1299,
      "step": 4196
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.20126684478082238,
      "learning_rate": 2.5754776466429274e-05,
      "loss": 1.1166,
      "step": 4197
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.17743614734571128,
      "learning_rate": 2.5714809641772664e-05,
      "loss": 0.9933,
      "step": 4198
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.19709235772681585,
      "learning_rate": 2.567486927592018e-05,
      "loss": 1.0575,
      "step": 4199
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.19243027342720215,
      "learning_rate": 2.5634955383097804e-05,
      "loss": 0.8241,
      "step": 4200
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.19186001482651507,
      "learning_rate": 2.559506797752196e-05,
      "loss": 1.1015,
      "step": 4201
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1826160921082101,
      "learning_rate": 2.5555207073399733e-05,
      "loss": 1.0497,
      "step": 4202
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.18367440964177165,
      "learning_rate": 2.5515372684928683e-05,
      "loss": 0.9147,
      "step": 4203
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.19426695132654717,
      "learning_rate": 2.5475564826297027e-05,
      "loss": 1.1329,
      "step": 4204
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2033793469772638,
      "learning_rate": 2.5435783511683443e-05,
      "loss": 1.0358,
      "step": 4205
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.17206390895274287,
      "learning_rate": 2.5396028755257207e-05,
      "loss": 1.1727,
      "step": 4206
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.17991575230362933,
      "learning_rate": 2.5356300571178094e-05,
      "loss": 0.9104,
      "step": 4207
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.22107735056978275,
      "learning_rate": 2.531659897359643e-05,
      "loss": 0.9897,
      "step": 4208
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1857103258763657,
      "learning_rate": 2.527692397665311e-05,
      "loss": 1.1056,
      "step": 4209
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.21770689295541767,
      "learning_rate": 2.523727559447949e-05,
      "loss": 0.9022,
      "step": 4210
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.20991541310676784,
      "learning_rate": 2.5197653841197543e-05,
      "loss": 1.1101,
      "step": 4211
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1743947808894751,
      "learning_rate": 2.5158058730919608e-05,
      "loss": 0.9028,
      "step": 4212
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.32666957993225226,
      "learning_rate": 2.511849027774874e-05,
      "loss": 1.0936,
      "step": 4213
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2013922121516079,
      "learning_rate": 2.5078948495778233e-05,
      "loss": 1.1456,
      "step": 4214
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.17874642515737166,
      "learning_rate": 2.503943339909214e-05,
      "loss": 1.0355,
      "step": 4215
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2191374446950704,
      "learning_rate": 2.4999945001764824e-05,
      "loss": 1.0289,
      "step": 4216
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2198686157002135,
      "learning_rate": 2.4960483317861273e-05,
      "loss": 1.0535,
      "step": 4217
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.20718161602655352,
      "learning_rate": 2.4921048361436882e-05,
      "loss": 1.1337,
      "step": 4218
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.19415391033203425,
      "learning_rate": 2.4881640146537498e-05,
      "loss": 0.9529,
      "step": 4219
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.21902025502866518,
      "learning_rate": 2.484225868719956e-05,
      "loss": 1.1763,
      "step": 4220
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.224481850058425,
      "learning_rate": 2.4802903997449865e-05,
      "loss": 1.1016,
      "step": 4221
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.20454591455033733,
      "learning_rate": 2.476357609130573e-05,
      "loss": 1.05,
      "step": 4222
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.17913543814286015,
      "learning_rate": 2.4724274982774865e-05,
      "loss": 1.1706,
      "step": 4223
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1987086495536307,
      "learning_rate": 2.4685000685855576e-05,
      "loss": 0.9087,
      "step": 4224
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.20583687812413756,
      "learning_rate": 2.464575321453646e-05,
      "loss": 1.0976,
      "step": 4225
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.22044111968337904,
      "learning_rate": 2.4606532582796693e-05,
      "loss": 1.1048,
      "step": 4226
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.19133240156303916,
      "learning_rate": 2.4567338804605756e-05,
      "loss": 0.94,
      "step": 4227
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.18628693199735688,
      "learning_rate": 2.4528171893923723e-05,
      "loss": 0.8602,
      "step": 4228
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.21203874400437656,
      "learning_rate": 2.448903186470096e-05,
      "loss": 0.9488,
      "step": 4229
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.24941677634731893,
      "learning_rate": 2.444991873087833e-05,
      "loss": 1.0134,
      "step": 4230
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2424902791672178,
      "learning_rate": 2.4410832506387095e-05,
      "loss": 1.0041,
      "step": 4231
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2762771197066873,
      "learning_rate": 2.4371773205148895e-05,
      "loss": 1.0287,
      "step": 4232
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1933067981078012,
      "learning_rate": 2.4332740841075908e-05,
      "loss": 1.1573,
      "step": 4233
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.17325284468506136,
      "learning_rate": 2.429373542807054e-05,
      "loss": 1.0318,
      "step": 4234
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.21607696707316157,
      "learning_rate": 2.4254756980025773e-05,
      "loss": 1.1052,
      "step": 4235
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.20557961507158493,
      "learning_rate": 2.4215805510824828e-05,
      "loss": 0.9245,
      "step": 4236
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.22153477843548117,
      "learning_rate": 2.417688103434146e-05,
      "loss": 0.9715,
      "step": 4237
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1895454440324327,
      "learning_rate": 2.413798356443969e-05,
      "loss": 0.9809,
      "step": 4238
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1905053122936568,
      "learning_rate": 2.4099113114973992e-05,
      "loss": 1.0685,
      "step": 4239
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1993238771644452,
      "learning_rate": 2.4060269699789184e-05,
      "loss": 0.9807,
      "step": 4240
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1949514248402561,
      "learning_rate": 2.4021453332720422e-05,
      "loss": 1.0401,
      "step": 4241
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2665469101274107,
      "learning_rate": 2.398266402759335e-05,
      "loss": 1.0602,
      "step": 4242
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.21383342017770296,
      "learning_rate": 2.394390179822382e-05,
      "loss": 0.9656,
      "step": 4243
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2128293859050976,
      "learning_rate": 2.3905166658418188e-05,
      "loss": 1.071,
      "step": 4244
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.20128142513834654,
      "learning_rate": 2.386645862197303e-05,
      "loss": 1.206,
      "step": 4245
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.21104499765820103,
      "learning_rate": 2.3827777702675348e-05,
      "loss": 1.0937,
      "step": 4246
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.17815671989508056,
      "learning_rate": 2.3789123914302436e-05,
      "loss": 1.071,
      "step": 4247
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.19269052474550116,
      "learning_rate": 2.3750497270621986e-05,
      "loss": 1.2335,
      "step": 4248
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1894150953807836,
      "learning_rate": 2.371189778539197e-05,
      "loss": 1.1987,
      "step": 4249
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.19966207380225667,
      "learning_rate": 2.3673325472360687e-05,
      "loss": 1.0714,
      "step": 4250
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.20755689335157657,
      "learning_rate": 2.3634780345266806e-05,
      "loss": 0.9733,
      "step": 4251
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.21605533369430627,
      "learning_rate": 2.3596262417839255e-05,
      "loss": 0.9587,
      "step": 4252
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.18048393793276,
      "learning_rate": 2.3557771703797325e-05,
      "loss": 1.0678,
      "step": 4253
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.19223507658994216,
      "learning_rate": 2.3519308216850577e-05,
      "loss": 1.1123,
      "step": 4254
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.17850586274124747,
      "learning_rate": 2.348087197069889e-05,
      "loss": 1.045,
      "step": 4255
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2190263366205762,
      "learning_rate": 2.3442462979032408e-05,
      "loss": 1.1638,
      "step": 4256
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.23554812058501032,
      "learning_rate": 2.3404081255531652e-05,
      "loss": 0.9725,
      "step": 4257
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2403049957509684,
      "learning_rate": 2.3365726813867304e-05,
      "loss": 1.0326,
      "step": 4258
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2493885873422945,
      "learning_rate": 2.3327399667700477e-05,
      "loss": 0.9642,
      "step": 4259
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.22927856178668599,
      "learning_rate": 2.3289099830682448e-05,
      "loss": 1.1253,
      "step": 4260
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2676002335469014,
      "learning_rate": 2.325082731645477e-05,
      "loss": 1.0198,
      "step": 4261
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.20285891741546352,
      "learning_rate": 2.3212582138649396e-05,
      "loss": 0.9396,
      "step": 4262
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.18946583161217642,
      "learning_rate": 2.317436431088832e-05,
      "loss": 1.0478,
      "step": 4263
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2112258979828523,
      "learning_rate": 2.3136173846784014e-05,
      "loss": 1.062,
      "step": 4264
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.23648531147358848,
      "learning_rate": 2.3098010759939036e-05,
      "loss": 1.0379,
      "step": 4265
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.24767395062236636,
      "learning_rate": 2.3059875063946347e-05,
      "loss": 1.2026,
      "step": 4266
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.18409283118730185,
      "learning_rate": 2.3021766772388986e-05,
      "loss": 1.1555,
      "step": 4267
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3122394976011344,
      "learning_rate": 2.298368589884039e-05,
      "loss": 0.9359,
      "step": 4268
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.21640120161593898,
      "learning_rate": 2.294563245686413e-05,
      "loss": 0.9972,
      "step": 4269
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.21473602050731214,
      "learning_rate": 2.290760646001402e-05,
      "loss": 1.0866,
      "step": 4270
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2065889170025819,
      "learning_rate": 2.2869607921834113e-05,
      "loss": 0.9989,
      "step": 4271
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.20854440897091575,
      "learning_rate": 2.2831636855858663e-05,
      "loss": 1.111,
      "step": 4272
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.19290560054897055,
      "learning_rate": 2.2793693275612204e-05,
      "loss": 1.0607,
      "step": 4273
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2541648280369107,
      "learning_rate": 2.2755777194609374e-05,
      "loss": 0.9502,
      "step": 4274
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.18249108432377634,
      "learning_rate": 2.2717888626355134e-05,
      "loss": 1.0944,
      "step": 4275
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2038194875465556,
      "learning_rate": 2.268002758434452e-05,
      "loss": 1.1194,
      "step": 4276
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1832372870044055,
      "learning_rate": 2.2642194082062895e-05,
      "loss": 1.0663,
      "step": 4277
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.18210209553615547,
      "learning_rate": 2.260438813298572e-05,
      "loss": 0.9049,
      "step": 4278
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.19401791300270543,
      "learning_rate": 2.2566609750578673e-05,
      "loss": 1.0471,
      "step": 4279
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.21205005876053737,
      "learning_rate": 2.252885894829757e-05,
      "loss": 0.9977,
      "step": 4280
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.179611163703741,
      "learning_rate": 2.249113573958852e-05,
      "loss": 1.1306,
      "step": 4281
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.18317210271200793,
      "learning_rate": 2.245344013788767e-05,
      "loss": 0.9568,
      "step": 4282
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.20112080147206532,
      "learning_rate": 2.2415772156621382e-05,
      "loss": 1.0135,
      "step": 4283
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.19529438189252463,
      "learning_rate": 2.2378131809206248e-05,
      "loss": 1.0963,
      "step": 4284
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.21564414603273016,
      "learning_rate": 2.2340519109048898e-05,
      "loss": 1.1793,
      "step": 4285
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.19903261507180564,
      "learning_rate": 2.2302934069546222e-05,
      "loss": 1.0428,
      "step": 4286
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.19168520929586574,
      "learning_rate": 2.2265376704085193e-05,
      "loss": 1.026,
      "step": 4287
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.20577161582176,
      "learning_rate": 2.2227847026042946e-05,
      "loss": 0.9704,
      "step": 4288
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.18588980839040004,
      "learning_rate": 2.2190345048786733e-05,
      "loss": 1.1257,
      "step": 4289
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.20790002027013366,
      "learning_rate": 2.2152870785673996e-05,
      "loss": 0.9833,
      "step": 4290
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2364760194769093,
      "learning_rate": 2.211542425005223e-05,
      "loss": 1.0701,
      "step": 4291
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.20098814901253384,
      "learning_rate": 2.2078005455259143e-05,
      "loss": 0.9114,
      "step": 4292
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.24185834338941223,
      "learning_rate": 2.2040614414622484e-05,
      "loss": 1.0405,
      "step": 4293
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.17014877547867152,
      "learning_rate": 2.2003251141460156e-05,
      "loss": 1.1158,
      "step": 4294
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.18207066230585253,
      "learning_rate": 2.196591564908016e-05,
      "loss": 1.0647,
      "step": 4295
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.27541398102744147,
      "learning_rate": 2.1928607950780566e-05,
      "loss": 0.9785,
      "step": 4296
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.21685295109855088,
      "learning_rate": 2.189132805984965e-05,
      "loss": 0.9079,
      "step": 4297
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.19941405299538667,
      "learning_rate": 2.1854075989565658e-05,
      "loss": 1.0982,
      "step": 4298
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.19544964593454425,
      "learning_rate": 2.181685175319702e-05,
      "loss": 0.915,
      "step": 4299
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.20401907355100118,
      "learning_rate": 2.1779655364002195e-05,
      "loss": 0.9443,
      "step": 4300
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2207543773203643,
      "learning_rate": 2.174248683522978e-05,
      "loss": 0.9824,
      "step": 4301
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2176474222761429,
      "learning_rate": 2.170534618011839e-05,
      "loss": 1.0515,
      "step": 4302
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.21060524427274363,
      "learning_rate": 2.1668233411896734e-05,
      "loss": 1.1985,
      "step": 4303
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1979979780620034,
      "learning_rate": 2.1631148543783597e-05,
      "loss": 0.9462,
      "step": 4304
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.22159349016649646,
      "learning_rate": 2.1594091588987774e-05,
      "loss": 1.0942,
      "step": 4305
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1695754335137129,
      "learning_rate": 2.1557062560708252e-05,
      "loss": 1.1186,
      "step": 4306
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.201722990779671,
      "learning_rate": 2.1520061472133902e-05,
      "loss": 1.0447,
      "step": 4307
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2500623869640117,
      "learning_rate": 2.148308833644378e-05,
      "loss": 1.0893,
      "step": 4308
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1833695306571161,
      "learning_rate": 2.14461431668069e-05,
      "loss": 1.013,
      "step": 4309
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1958731102989515,
      "learning_rate": 2.1409225976382363e-05,
      "loss": 0.9689,
      "step": 4310
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.24935413983821753,
      "learning_rate": 2.137233677831929e-05,
      "loss": 1.1488,
      "step": 4311
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.19826877567538084,
      "learning_rate": 2.1335475585756803e-05,
      "loss": 1.026,
      "step": 4312
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.19478957140550213,
      "learning_rate": 2.12986424118241e-05,
      "loss": 1.0529,
      "step": 4313
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.207817280145275,
      "learning_rate": 2.126183726964034e-05,
      "loss": 1.1093,
      "step": 4314
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.21500101871245123,
      "learning_rate": 2.122506017231477e-05,
      "loss": 0.9559,
      "step": 4315
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.21896197960819438,
      "learning_rate": 2.118831113294657e-05,
      "loss": 0.9619,
      "step": 4316
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.20269305196036733,
      "learning_rate": 2.1151590164625012e-05,
      "loss": 1.0343,
      "step": 4317
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.26939710863825805,
      "learning_rate": 2.1114897280429304e-05,
      "loss": 0.9536,
      "step": 4318
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1924625231923352,
      "learning_rate": 2.1078232493428664e-05,
      "loss": 1.1765,
      "step": 4319
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.19152799154785527,
      "learning_rate": 2.104159581668229e-05,
      "loss": 1.0877,
      "step": 4320
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2178209134611367,
      "learning_rate": 2.100498726323943e-05,
      "loss": 1.0227,
      "step": 4321
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.232290321303052,
      "learning_rate": 2.0968406846139233e-05,
      "loss": 1.2414,
      "step": 4322
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.20612475870282684,
      "learning_rate": 2.0931854578410905e-05,
      "loss": 0.8809,
      "step": 4323
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.20368900019907454,
      "learning_rate": 2.089533047307356e-05,
      "loss": 1.1152,
      "step": 4324
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1919567923613192,
      "learning_rate": 2.0858834543136286e-05,
      "loss": 1.0151,
      "step": 4325
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.223464821063959,
      "learning_rate": 2.0822366801598204e-05,
      "loss": 1.0651,
      "step": 4326
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.193277606384106,
      "learning_rate": 2.0785927261448335e-05,
      "loss": 1.1022,
      "step": 4327
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.21640884442047897,
      "learning_rate": 2.0749515935665654e-05,
      "loss": 1.0801,
      "step": 4328
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1953999190555029,
      "learning_rate": 2.0713132837219075e-05,
      "loss": 0.9579,
      "step": 4329
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.21083923972927654,
      "learning_rate": 2.067677797906754e-05,
      "loss": 1.0124,
      "step": 4330
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.18094445314922852,
      "learning_rate": 2.064045137415982e-05,
      "loss": 1.1651,
      "step": 4331
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.18536818851758405,
      "learning_rate": 2.0604153035434735e-05,
      "loss": 1.1122,
      "step": 4332
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.23448027216952344,
      "learning_rate": 2.056788297582093e-05,
      "loss": 1.0083,
      "step": 4333
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2031910890212284,
      "learning_rate": 2.0531641208237084e-05,
      "loss": 1.0456,
      "step": 4334
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.23836778683740534,
      "learning_rate": 2.04954277455917e-05,
      "loss": 1.2086,
      "step": 4335
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.20552007817932133,
      "learning_rate": 2.045924260078327e-05,
      "loss": 0.8719,
      "step": 4336
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.22336817267745246,
      "learning_rate": 2.0423085786700135e-05,
      "loss": 0.9201,
      "step": 4337
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.20807537719391986,
      "learning_rate": 2.03869573162206e-05,
      "loss": 1.0602,
      "step": 4338
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.20977542750731,
      "learning_rate": 2.035085720221288e-05,
      "loss": 0.9333,
      "step": 4339
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2098080075015176,
      "learning_rate": 2.0314785457535014e-05,
      "loss": 1.062,
      "step": 4340
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.16006884058095353,
      "learning_rate": 2.0278742095035065e-05,
      "loss": 1.076,
      "step": 4341
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1789668810415249,
      "learning_rate": 2.0242727127550865e-05,
      "loss": 0.9483,
      "step": 4342
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.15645915142268996,
      "learning_rate": 2.0206740567910187e-05,
      "loss": 0.9577,
      "step": 4343
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.19570617029278378,
      "learning_rate": 2.0170782428930658e-05,
      "loss": 1.0253,
      "step": 4344
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.22330640515846553,
      "learning_rate": 2.013485272341984e-05,
      "loss": 0.9634,
      "step": 4345
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2488898857504409,
      "learning_rate": 2.009895146417512e-05,
      "loss": 0.9877,
      "step": 4346
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.21410131383612735,
      "learning_rate": 2.0063078663983714e-05,
      "loss": 0.9829,
      "step": 4347
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.20017213760824318,
      "learning_rate": 2.0027234335622826e-05,
      "loss": 0.8912,
      "step": 4348
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2445889105695649,
      "learning_rate": 1.999141849185938e-05,
      "loss": 1.0693,
      "step": 4349
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.20541460251234966,
      "learning_rate": 1.9955631145450272e-05,
      "loss": 1.1618,
      "step": 4350
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.18252210997854876,
      "learning_rate": 1.9919872309142152e-05,
      "loss": 0.9892,
      "step": 4351
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.18066312838141874,
      "learning_rate": 1.988414199567158e-05,
      "loss": 1.0426,
      "step": 4352
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.23850564038466818,
      "learning_rate": 1.9848440217764897e-05,
      "loss": 0.9631,
      "step": 4353
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.18510206622193176,
      "learning_rate": 1.9812766988138375e-05,
      "loss": 1.0228,
      "step": 4354
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.18918727372360683,
      "learning_rate": 1.9777122319497986e-05,
      "loss": 1.0117,
      "step": 4355
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.20641573670204022,
      "learning_rate": 1.9741506224539687e-05,
      "loss": 0.9426,
      "step": 4356
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2062977255101869,
      "learning_rate": 1.9705918715949123e-05,
      "loss": 1.006,
      "step": 4357
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.21013338090662564,
      "learning_rate": 1.9670359806401785e-05,
      "loss": 1.0225,
      "step": 4358
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2308068394820628,
      "learning_rate": 1.9634829508563057e-05,
      "loss": 0.8979,
      "step": 4359
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1699858758421207,
      "learning_rate": 1.9599327835088056e-05,
      "loss": 1.1663,
      "step": 4360
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2213920899277503,
      "learning_rate": 1.9563854798621706e-05,
      "loss": 0.9849,
      "step": 4361
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2193200849663855,
      "learning_rate": 1.9528410411798727e-05,
      "loss": 1.0925,
      "step": 4362
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1934613414124001,
      "learning_rate": 1.9492994687243714e-05,
      "loss": 1.0119,
      "step": 4363
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.20432638550892301,
      "learning_rate": 1.945760763757094e-05,
      "loss": 1.0153,
      "step": 4364
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2053226489743393,
      "learning_rate": 1.9422249275384573e-05,
      "loss": 1.0557,
      "step": 4365
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.21270587288134354,
      "learning_rate": 1.9386919613278475e-05,
      "loss": 1.0429,
      "step": 4366
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.208889652330382,
      "learning_rate": 1.9351618663836325e-05,
      "loss": 1.2259,
      "step": 4367
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1522447137443068,
      "learning_rate": 1.9316346439631573e-05,
      "loss": 0.9942,
      "step": 4368
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.19250446997019258,
      "learning_rate": 1.9281102953227416e-05,
      "loss": 0.9497,
      "step": 4369
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.19865928437378025,
      "learning_rate": 1.9245888217176854e-05,
      "loss": 1.015,
      "step": 4370
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1933853497393748,
      "learning_rate": 1.9210702244022617e-05,
      "loss": 0.9822,
      "step": 4371
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2208471278501074,
      "learning_rate": 1.9175545046297218e-05,
      "loss": 1.0171,
      "step": 4372
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1961246954905173,
      "learning_rate": 1.9140416636522872e-05,
      "loss": 1.1121,
      "step": 4373
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.19004616292316215,
      "learning_rate": 1.9105317027211612e-05,
      "loss": 1.1019,
      "step": 4374
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1921718538922879,
      "learning_rate": 1.907024623086515e-05,
      "loss": 1.0475,
      "step": 4375
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.18497811079167045,
      "learning_rate": 1.903520425997496e-05,
      "loss": 0.9724,
      "step": 4376
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.20880201738471355,
      "learning_rate": 1.9000191127022237e-05,
      "loss": 1.0303,
      "step": 4377
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.16677990132810072,
      "learning_rate": 1.896520684447789e-05,
      "loss": 0.9778,
      "step": 4378
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2114389111024812,
      "learning_rate": 1.893025142480265e-05,
      "loss": 0.8753,
      "step": 4379
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2139633970773668,
      "learning_rate": 1.889532488044681e-05,
      "loss": 0.9434,
      "step": 4380
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.23788652313069653,
      "learning_rate": 1.886042722385053e-05,
      "loss": 1.0332,
      "step": 4381
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.16615210496185392,
      "learning_rate": 1.882555846744356e-05,
      "loss": 0.9758,
      "step": 4382
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1922499972211335,
      "learning_rate": 1.8790718623645455e-05,
      "loss": 1.0181,
      "step": 4383
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.18860918400528887,
      "learning_rate": 1.8755907704865416e-05,
      "loss": 1.036,
      "step": 4384
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.20392843860723173,
      "learning_rate": 1.872112572350233e-05,
      "loss": 0.9828,
      "step": 4385
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.17106196186776146,
      "learning_rate": 1.8686372691944787e-05,
      "loss": 0.9015,
      "step": 4386
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.19012798784173215,
      "learning_rate": 1.8651648622571128e-05,
      "loss": 0.8195,
      "step": 4387
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.21140134225697618,
      "learning_rate": 1.861695352774929e-05,
      "loss": 1.1543,
      "step": 4388
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.19931972755881375,
      "learning_rate": 1.858228741983692e-05,
      "loss": 1.0065,
      "step": 4389
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.21966988763115838,
      "learning_rate": 1.8547650311181387e-05,
      "loss": 1.0478,
      "step": 4390
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.18794316222398857,
      "learning_rate": 1.851304221411967e-05,
      "loss": 1.1724,
      "step": 4391
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.21787466263132077,
      "learning_rate": 1.847846314097843e-05,
      "loss": 0.951,
      "step": 4392
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.20229332541417913,
      "learning_rate": 1.8443913104073983e-05,
      "loss": 1.0378,
      "step": 4393
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.16949707837654146,
      "learning_rate": 1.8409392115712366e-05,
      "loss": 1.0906,
      "step": 4394
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1909764372344428,
      "learning_rate": 1.837490018818917e-05,
      "loss": 1.0557,
      "step": 4395
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.22087211187730868,
      "learning_rate": 1.8340437333789727e-05,
      "loss": 1.0742,
      "step": 4396
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.21485095675250612,
      "learning_rate": 1.830600356478893e-05,
      "loss": 1.0629,
      "step": 4397
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.213041510198585,
      "learning_rate": 1.8271598893451404e-05,
      "loss": 1.1875,
      "step": 4398
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1966642531827405,
      "learning_rate": 1.8237223332031327e-05,
      "loss": 1.1179,
      "step": 4399
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.16902452325800393,
      "learning_rate": 1.820287689277256e-05,
      "loss": 1.083,
      "step": 4400
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.20019562139646316,
      "learning_rate": 1.8168559587908552e-05,
      "loss": 0.9733,
      "step": 4401
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.19723333526828102,
      "learning_rate": 1.8134271429662396e-05,
      "loss": 1.1242,
      "step": 4402
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2330926451904508,
      "learning_rate": 1.8100012430246837e-05,
      "loss": 0.9288,
      "step": 4403
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.22942761385089058,
      "learning_rate": 1.8065782601864156e-05,
      "loss": 1.1442,
      "step": 4404
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.16657033223856366,
      "learning_rate": 1.8031581956706346e-05,
      "loss": 1.0901,
      "step": 4405
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1840243623755023,
      "learning_rate": 1.7997410506954893e-05,
      "loss": 1.1194,
      "step": 4406
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.16456063931152926,
      "learning_rate": 1.7963268264780985e-05,
      "loss": 1.0623,
      "step": 4407
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.22071691209291,
      "learning_rate": 1.792915524234534e-05,
      "loss": 1.0147,
      "step": 4408
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.19563770679785955,
      "learning_rate": 1.7895071451798306e-05,
      "loss": 1.065,
      "step": 4409
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1792272677596631,
      "learning_rate": 1.7861016905279782e-05,
      "loss": 0.9465,
      "step": 4410
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.23260398163172108,
      "learning_rate": 1.7826991614919265e-05,
      "loss": 1.0067,
      "step": 4411
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.19401416607894956,
      "learning_rate": 1.779299559283588e-05,
      "loss": 1.0375,
      "step": 4412
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.20341226994752903,
      "learning_rate": 1.775902885113824e-05,
      "loss": 1.0067,
      "step": 4413
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.17546590650264712,
      "learning_rate": 1.772509140192462e-05,
      "loss": 0.9967,
      "step": 4414
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.19635700334132436,
      "learning_rate": 1.7691183257282772e-05,
      "loss": 1.1124,
      "step": 4415
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.21163569724324374,
      "learning_rate": 1.7657304429290134e-05,
      "loss": 0.9381,
      "step": 4416
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.21814897574241834,
      "learning_rate": 1.7623454930013515e-05,
      "loss": 1.0715,
      "step": 4417
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.20992778380583227,
      "learning_rate": 1.7589634771509465e-05,
      "loss": 0.8807,
      "step": 4418
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.19843897094106694,
      "learning_rate": 1.7555843965823992e-05,
      "loss": 1.0869,
      "step": 4419
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.18908915898209905,
      "learning_rate": 1.752208252499261e-05,
      "loss": 0.8879,
      "step": 4420
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.19035750360223036,
      "learning_rate": 1.7488350461040504e-05,
      "loss": 1.151,
      "step": 4421
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.17859235438194063,
      "learning_rate": 1.7454647785982257e-05,
      "loss": 0.9726,
      "step": 4422
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.17798450033489494,
      "learning_rate": 1.7420974511822085e-05,
      "loss": 0.9383,
      "step": 4423
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.18589306947611697,
      "learning_rate": 1.7387330650553678e-05,
      "loss": 1.0845,
      "step": 4424
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.21745609130650487,
      "learning_rate": 1.7353716214160266e-05,
      "loss": 1.0755,
      "step": 4425
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.19283233994165858,
      "learning_rate": 1.7320131214614566e-05,
      "loss": 1.0128,
      "step": 4426
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1893424390691021,
      "learning_rate": 1.7286575663878877e-05,
      "loss": 1.0755,
      "step": 4427
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1769156972411495,
      "learning_rate": 1.7253049573904944e-05,
      "loss": 1.0335,
      "step": 4428
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.20961991124672785,
      "learning_rate": 1.721955295663409e-05,
      "loss": 1.0538,
      "step": 4429
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.16555983269772792,
      "learning_rate": 1.7186085823997067e-05,
      "loss": 1.0663,
      "step": 4430
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.18209259213263765,
      "learning_rate": 1.7152648187914134e-05,
      "loss": 1.0704,
      "step": 4431
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1828231062543661,
      "learning_rate": 1.7119240060295105e-05,
      "loss": 1.0465,
      "step": 4432
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.24946454563448164,
      "learning_rate": 1.7085861453039243e-05,
      "loss": 1.0162,
      "step": 4433
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.33115932485286936,
      "learning_rate": 1.705251237803528e-05,
      "loss": 1.0409,
      "step": 4434
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.20320314596569372,
      "learning_rate": 1.7019192847161425e-05,
      "loss": 0.9591,
      "step": 4435
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1828829231984977,
      "learning_rate": 1.698590287228544e-05,
      "loss": 1.0967,
      "step": 4436
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.19258561450670794,
      "learning_rate": 1.6952642465264447e-05,
      "loss": 1.035,
      "step": 4437
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1810102718297075,
      "learning_rate": 1.6919411637945147e-05,
      "loss": 0.9279,
      "step": 4438
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.20682868232402932,
      "learning_rate": 1.6886210402163615e-05,
      "loss": 0.9166,
      "step": 4439
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.192830988722829,
      "learning_rate": 1.6853038769745467e-05,
      "loss": 0.9737,
      "step": 4440
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2446561083685997,
      "learning_rate": 1.6819896752505738e-05,
      "loss": 0.8722,
      "step": 4441
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2555247328447526,
      "learning_rate": 1.6786784362248818e-05,
      "loss": 0.9948,
      "step": 4442
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.19697134376888473,
      "learning_rate": 1.6753701610768724e-05,
      "loss": 0.9752,
      "step": 4443
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.19194035396284317,
      "learning_rate": 1.672064850984878e-05,
      "loss": 1.0578,
      "step": 4444
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.17189500043485823,
      "learning_rate": 1.668762507126185e-05,
      "loss": 1.0185,
      "step": 4445
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.18244243301454105,
      "learning_rate": 1.665463130677011e-05,
      "loss": 0.9775,
      "step": 4446
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2384386367279086,
      "learning_rate": 1.6621667228125302e-05,
      "loss": 0.9072,
      "step": 4447
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.19876547854655702,
      "learning_rate": 1.6588732847068513e-05,
      "loss": 1.0552,
      "step": 4448
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2081695521725839,
      "learning_rate": 1.6555828175330245e-05,
      "loss": 0.9747,
      "step": 4449
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1846526535556492,
      "learning_rate": 1.6522953224630443e-05,
      "loss": 0.8497,
      "step": 4450
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.28156495734141895,
      "learning_rate": 1.6490108006678494e-05,
      "loss": 0.9968,
      "step": 4451
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2115772094455779,
      "learning_rate": 1.6457292533173153e-05,
      "loss": 0.915,
      "step": 4452
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.21691373972807554,
      "learning_rate": 1.6424506815802566e-05,
      "loss": 1.047,
      "step": 4453
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.19106275440823758,
      "learning_rate": 1.639175086624435e-05,
      "loss": 1.1063,
      "step": 4454
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.21290481314320808,
      "learning_rate": 1.635902469616544e-05,
      "loss": 0.8331,
      "step": 4455
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.24306032243560208,
      "learning_rate": 1.6326328317222238e-05,
      "loss": 1.0084,
      "step": 4456
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.18229719353908158,
      "learning_rate": 1.6293661741060485e-05,
      "loss": 1.0485,
      "step": 4457
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.17445592002264862,
      "learning_rate": 1.626102497931532e-05,
      "loss": 1.1052,
      "step": 4458
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.22250073973375298,
      "learning_rate": 1.6228418043611227e-05,
      "loss": 0.9887,
      "step": 4459
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1691117678783188,
      "learning_rate": 1.619584094556217e-05,
      "loss": 0.9106,
      "step": 4460
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.19416488499144566,
      "learning_rate": 1.616329369677134e-05,
      "loss": 1.1624,
      "step": 4461
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1789264658569218,
      "learning_rate": 1.613077630883145e-05,
      "loss": 1.0015,
      "step": 4462
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2010971162351002,
      "learning_rate": 1.6098288793324477e-05,
      "loss": 1.1546,
      "step": 4463
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.18860508936317635,
      "learning_rate": 1.6065831161821744e-05,
      "loss": 1.0918,
      "step": 4464
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.196312693652578,
      "learning_rate": 1.6033403425884052e-05,
      "loss": 1.0635,
      "step": 4465
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.23800978200027065,
      "learning_rate": 1.600100559706137e-05,
      "loss": 1.0661,
      "step": 4466
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.18181323283001327,
      "learning_rate": 1.5968637686893186e-05,
      "loss": 1.114,
      "step": 4467
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.20273628131491533,
      "learning_rate": 1.5936299706908208e-05,
      "loss": 0.8812,
      "step": 4468
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.19914286968802322,
      "learning_rate": 1.5903991668624597e-05,
      "loss": 0.9112,
      "step": 4469
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.22941237973476705,
      "learning_rate": 1.587171358354973e-05,
      "loss": 1.0287,
      "step": 4470
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.21176686860376176,
      "learning_rate": 1.5839465463180425e-05,
      "loss": 0.9191,
      "step": 4471
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.20424048914185078,
      "learning_rate": 1.580724731900276e-05,
      "loss": 1.0957,
      "step": 4472
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.19551901399703756,
      "learning_rate": 1.5775059162492146e-05,
      "loss": 1.0755,
      "step": 4473
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.19351828213697864,
      "learning_rate": 1.574290100511332e-05,
      "loss": 1.0397,
      "step": 4474
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1730454527390594,
      "learning_rate": 1.57107728583203e-05,
      "loss": 1.0771,
      "step": 4475
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.24016815173258627,
      "learning_rate": 1.5678674733556507e-05,
      "loss": 1.0065,
      "step": 4476
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.20138463864058676,
      "learning_rate": 1.564660664225457e-05,
      "loss": 1.1211,
      "step": 4477
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.20412396950802023,
      "learning_rate": 1.5614568595836498e-05,
      "loss": 0.9894,
      "step": 4478
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.18405916145855133,
      "learning_rate": 1.5582560605713504e-05,
      "loss": 1.1177,
      "step": 4479
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.19538077983355817,
      "learning_rate": 1.555058268328622e-05,
      "loss": 0.8167,
      "step": 4480
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.20122198183933576,
      "learning_rate": 1.5518634839944468e-05,
      "loss": 1.0251,
      "step": 4481
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.22591996659955957,
      "learning_rate": 1.5486717087067393e-05,
      "loss": 0.8777,
      "step": 4482
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.20221855294379643,
      "learning_rate": 1.545482943602341e-05,
      "loss": 1.0578,
      "step": 4483
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.24113629900976427,
      "learning_rate": 1.5422971898170212e-05,
      "loss": 1.202,
      "step": 4484
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.18952224935763348,
      "learning_rate": 1.5391144484854815e-05,
      "loss": 0.9362,
      "step": 4485
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.20483710689430007,
      "learning_rate": 1.535934720741342e-05,
      "loss": 1.0146,
      "step": 4486
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.19668374940050554,
      "learning_rate": 1.5327580077171587e-05,
      "loss": 1.2185,
      "step": 4487
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.19031618784200907,
      "learning_rate": 1.5295843105444053e-05,
      "loss": 0.9581,
      "step": 4488
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.20753650413971397,
      "learning_rate": 1.5264136303534894e-05,
      "loss": 0.9338,
      "step": 4489
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.18258058277530664,
      "learning_rate": 1.5232459682737366e-05,
      "loss": 0.9764,
      "step": 4490
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.17779995652878064,
      "learning_rate": 1.5200813254334012e-05,
      "loss": 0.9977,
      "step": 4491
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.18711836563557213,
      "learning_rate": 1.5169197029596604e-05,
      "loss": 1.0579,
      "step": 4492
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.19553123035651512,
      "learning_rate": 1.5137611019786213e-05,
      "loss": 1.1458,
      "step": 4493
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.19295902428310754,
      "learning_rate": 1.5106055236153061e-05,
      "loss": 1.049,
      "step": 4494
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.19193948644983824,
      "learning_rate": 1.5074529689936645e-05,
      "loss": 1.0233,
      "step": 4495
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.19383415749019875,
      "learning_rate": 1.5043034392365729e-05,
      "loss": 0.9229,
      "step": 4496
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.19061575574415265,
      "learning_rate": 1.501156935465825e-05,
      "loss": 1.1066,
      "step": 4497
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.17190193545373308,
      "learning_rate": 1.4980134588021377e-05,
      "loss": 0.9107,
      "step": 4498
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.19610499048359292,
      "learning_rate": 1.4948730103651498e-05,
      "loss": 0.9381,
      "step": 4499
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1724180531809725,
      "learning_rate": 1.4917355912734243e-05,
      "loss": 1.1551,
      "step": 4500
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.21500380284717716,
      "learning_rate": 1.4886012026444406e-05,
      "loss": 0.9888,
      "step": 4501
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2333278343339504,
      "learning_rate": 1.4854698455946058e-05,
      "loss": 1.1326,
      "step": 4502
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.20870763363111441,
      "learning_rate": 1.4823415212392377e-05,
      "loss": 1.0307,
      "step": 4503
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.278973067360034,
      "learning_rate": 1.4792162306925827e-05,
      "loss": 0.9521,
      "step": 4504
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.18398077859129675,
      "learning_rate": 1.4760939750678004e-05,
      "loss": 1.1069,
      "step": 4505
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.20541818811865722,
      "learning_rate": 1.4729747554769735e-05,
      "loss": 1.0176,
      "step": 4506
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.21200914975620946,
      "learning_rate": 1.4698585730310998e-05,
      "loss": 1.0949,
      "step": 4507
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1957643120149438,
      "learning_rate": 1.4667454288400962e-05,
      "loss": 1.1158,
      "step": 4508
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.18967291494726776,
      "learning_rate": 1.463635324012802e-05,
      "loss": 0.9959,
      "step": 4509
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1688502743019516,
      "learning_rate": 1.460528259656967e-05,
      "loss": 1.1948,
      "step": 4510
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.22880989530563797,
      "learning_rate": 1.4574242368792657e-05,
      "loss": 1.1529,
      "step": 4511
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.15728767046824335,
      "learning_rate": 1.4543232567852783e-05,
      "loss": 1.0329,
      "step": 4512
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.22148057192218462,
      "learning_rate": 1.4512253204795156e-05,
      "loss": 1.0912,
      "step": 4513
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.20402518044104298,
      "learning_rate": 1.4481304290653919e-05,
      "loss": 0.9708,
      "step": 4514
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2021994745843028,
      "learning_rate": 1.4450385836452429e-05,
      "loss": 1.2641,
      "step": 4515
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.23859749857607757,
      "learning_rate": 1.4419497853203189e-05,
      "loss": 0.8093,
      "step": 4516
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.20176929226311408,
      "learning_rate": 1.4388640351907801e-05,
      "loss": 1.0924,
      "step": 4517
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.196506498163887,
      "learning_rate": 1.4357813343557103e-05,
      "loss": 1.1211,
      "step": 4518
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1814250397544929,
      "learning_rate": 1.4327016839130969e-05,
      "loss": 1.0378,
      "step": 4519
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.19281165856206242,
      "learning_rate": 1.4296250849598514e-05,
      "loss": 0.9318,
      "step": 4520
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.17828194252707885,
      "learning_rate": 1.4265515385917904e-05,
      "loss": 1.0185,
      "step": 4521
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.18519988014585723,
      "learning_rate": 1.4234810459036452e-05,
      "loss": 1.1325,
      "step": 4522
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2141517229500967,
      "learning_rate": 1.4204136079890584e-05,
      "loss": 1.0653,
      "step": 4523
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.19826041652665083,
      "learning_rate": 1.4173492259405896e-05,
      "loss": 1.0661,
      "step": 4524
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1937072647959065,
      "learning_rate": 1.4142879008497056e-05,
      "loss": 0.9147,
      "step": 4525
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.198613815062234,
      "learning_rate": 1.411229633806781e-05,
      "loss": 0.9204,
      "step": 4526
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.20400283963055602,
      "learning_rate": 1.408174425901112e-05,
      "loss": 1.112,
      "step": 4527
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.17250814733035513,
      "learning_rate": 1.4051222782208939e-05,
      "loss": 0.8923,
      "step": 4528
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.21238044353142657,
      "learning_rate": 1.4020731918532393e-05,
      "loss": 0.9304,
      "step": 4529
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.19811684466360382,
      "learning_rate": 1.3990271678841682e-05,
      "loss": 0.9926,
      "step": 4530
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.21776148242375018,
      "learning_rate": 1.3959842073986085e-05,
      "loss": 0.9874,
      "step": 4531
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.21491000030990146,
      "learning_rate": 1.3929443114803942e-05,
      "loss": 1.0798,
      "step": 4532
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.18091756545768098,
      "learning_rate": 1.389907481212278e-05,
      "loss": 0.9853,
      "step": 4533
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.20729628437135794,
      "learning_rate": 1.3868737176759106e-05,
      "loss": 1.2331,
      "step": 4534
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1773211832311069,
      "learning_rate": 1.383843021951855e-05,
      "loss": 1.2262,
      "step": 4535
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.19311766111524017,
      "learning_rate": 1.3808153951195812e-05,
      "loss": 0.9222,
      "step": 4536
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.19564116283545835,
      "learning_rate": 1.3777908382574622e-05,
      "loss": 0.9104,
      "step": 4537
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2051355239579708,
      "learning_rate": 1.3747693524427862e-05,
      "loss": 1.0195,
      "step": 4538
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.18855583440520485,
      "learning_rate": 1.371750938751739e-05,
      "loss": 1.1688,
      "step": 4539
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1881129237426398,
      "learning_rate": 1.368735598259414e-05,
      "loss": 1.1406,
      "step": 4540
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.18663155635038964,
      "learning_rate": 1.3657233320398122e-05,
      "loss": 1.2301,
      "step": 4541
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.21500834635219435,
      "learning_rate": 1.3627141411658406e-05,
      "loss": 1.0285,
      "step": 4542
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.18634414626676912,
      "learning_rate": 1.3597080267093066e-05,
      "loss": 0.9986,
      "step": 4543
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.18279739386265403,
      "learning_rate": 1.3567049897409268e-05,
      "loss": 1.0412,
      "step": 4544
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2202517732755465,
      "learning_rate": 1.3537050313303179e-05,
      "loss": 0.983,
      "step": 4545
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.21108535287750393,
      "learning_rate": 1.3507081525459997e-05,
      "loss": 0.9857,
      "step": 4546
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.18281513076122738,
      "learning_rate": 1.3477143544553995e-05,
      "loss": 1.082,
      "step": 4547
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.18568700884094913,
      "learning_rate": 1.3447236381248395e-05,
      "loss": 0.9519,
      "step": 4548
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1947797273613745,
      "learning_rate": 1.341736004619556e-05,
      "loss": 0.8594,
      "step": 4549
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.22647867640081798,
      "learning_rate": 1.338751455003674e-05,
      "loss": 1.0285,
      "step": 4550
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2137013831523536,
      "learning_rate": 1.335769990340232e-05,
      "loss": 0.9378,
      "step": 4551
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.18405770445481198,
      "learning_rate": 1.33279161169116e-05,
      "loss": 1.1244,
      "step": 4552
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.18448919319244708,
      "learning_rate": 1.3298163201172964e-05,
      "loss": 1.0762,
      "step": 4553
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.19514148543040022,
      "learning_rate": 1.3268441166783752e-05,
      "loss": 1.1032,
      "step": 4554
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.16173985162304366,
      "learning_rate": 1.3238750024330338e-05,
      "loss": 0.9419,
      "step": 4555
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.24707332394396836,
      "learning_rate": 1.3209089784388019e-05,
      "loss": 1.0957,
      "step": 4556
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.21265043985826276,
      "learning_rate": 1.317946045752122e-05,
      "loss": 0.9155,
      "step": 4557
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.19821343755361046,
      "learning_rate": 1.3149862054283225e-05,
      "loss": 1.0429,
      "step": 4558
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.19455364141765658,
      "learning_rate": 1.3120294585216353e-05,
      "loss": 0.9005,
      "step": 4559
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.26170747388552495,
      "learning_rate": 1.3090758060851938e-05,
      "loss": 0.8939,
      "step": 4560
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.18888515962165778,
      "learning_rate": 1.3061252491710229e-05,
      "loss": 1.1547,
      "step": 4561
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.21352598840856876,
      "learning_rate": 1.3031777888300523e-05,
      "loss": 0.9187,
      "step": 4562
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2033367647014099,
      "learning_rate": 1.300233426112103e-05,
      "loss": 1.0472,
      "step": 4563
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2678157950189738,
      "learning_rate": 1.2972921620658928e-05,
      "loss": 0.9198,
      "step": 4564
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.19029002922744792,
      "learning_rate": 1.2943539977390361e-05,
      "loss": 1.0054,
      "step": 4565
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1957112888207703,
      "learning_rate": 1.291418934178048e-05,
      "loss": 1.0405,
      "step": 4566
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.21347998707604784,
      "learning_rate": 1.2884869724283332e-05,
      "loss": 0.9263,
      "step": 4567
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.16290377707857684,
      "learning_rate": 1.2855581135341976e-05,
      "loss": 1.1004,
      "step": 4568
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.22492673794851614,
      "learning_rate": 1.2826323585388355e-05,
      "loss": 1.0934,
      "step": 4569
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.19294583042484828,
      "learning_rate": 1.279709708484339e-05,
      "loss": 1.0961,
      "step": 4570
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.18560159652960698,
      "learning_rate": 1.2767901644116941e-05,
      "loss": 1.2006,
      "step": 4571
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.17667632087927737,
      "learning_rate": 1.273873727360777e-05,
      "loss": 1.1138,
      "step": 4572
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.19202629647774064,
      "learning_rate": 1.2709603983703678e-05,
      "loss": 0.867,
      "step": 4573
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.17756089375172424,
      "learning_rate": 1.2680501784781241e-05,
      "loss": 0.9696,
      "step": 4574
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1762956394317298,
      "learning_rate": 1.2651430687206112e-05,
      "loss": 1.07,
      "step": 4575
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.20468658432948988,
      "learning_rate": 1.2622390701332754e-05,
      "loss": 0.9198,
      "step": 4576
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.17245763838386213,
      "learning_rate": 1.2593381837504625e-05,
      "loss": 1.0215,
      "step": 4577
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.19225091492224558,
      "learning_rate": 1.256440410605404e-05,
      "loss": 1.0859,
      "step": 4578
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.18781952302740842,
      "learning_rate": 1.2535457517302263e-05,
      "loss": 1.1227,
      "step": 4579
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.17287561685078012,
      "learning_rate": 1.2506542081559447e-05,
      "loss": 1.2041,
      "step": 4580
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2017870060670738,
      "learning_rate": 1.2477657809124631e-05,
      "loss": 0.9916,
      "step": 4581
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.19454542028709979,
      "learning_rate": 1.2448804710285833e-05,
      "loss": 1.0166,
      "step": 4582
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2010595065168683,
      "learning_rate": 1.2419982795319852e-05,
      "loss": 0.9668,
      "step": 4583
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.21317377773539575,
      "learning_rate": 1.2391192074492496e-05,
      "loss": 1.0154,
      "step": 4584
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.24090271569910082,
      "learning_rate": 1.2362432558058356e-05,
      "loss": 1.1268,
      "step": 4585
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2353401370017217,
      "learning_rate": 1.2333704256261014e-05,
      "loss": 1.0764,
      "step": 4586
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.20430237270553744,
      "learning_rate": 1.230500717933285e-05,
      "loss": 1.0446,
      "step": 4587
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.18290919675940356,
      "learning_rate": 1.2276341337495157e-05,
      "loss": 1.001,
      "step": 4588
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1943057306465002,
      "learning_rate": 1.2247706740958087e-05,
      "loss": 1.179,
      "step": 4589
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.19058264787866694,
      "learning_rate": 1.2219103399920673e-05,
      "loss": 0.993,
      "step": 4590
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2080775732978633,
      "learning_rate": 1.2190531324570843e-05,
      "loss": 0.834,
      "step": 4591
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1757850313280347,
      "learning_rate": 1.216199052508532e-05,
      "loss": 1.0424,
      "step": 4592
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.22577746403364155,
      "learning_rate": 1.2133481011629777e-05,
      "loss": 1.069,
      "step": 4593
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.18914223882604503,
      "learning_rate": 1.2105002794358666e-05,
      "loss": 1.1366,
      "step": 4594
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.18973733542623558,
      "learning_rate": 1.207655588341534e-05,
      "loss": 1.022,
      "step": 4595
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1863503098854822,
      "learning_rate": 1.2048140288931942e-05,
      "loss": 0.9503,
      "step": 4596
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.17157038904948033,
      "learning_rate": 1.201975602102956e-05,
      "loss": 0.9141,
      "step": 4597
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.186524146273874,
      "learning_rate": 1.199140308981802e-05,
      "loss": 1.0685,
      "step": 4598
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2565943825712755,
      "learning_rate": 1.1963081505396068e-05,
      "loss": 1.02,
      "step": 4599
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.24230877761091127,
      "learning_rate": 1.1934791277851242e-05,
      "loss": 1.0481,
      "step": 4600
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1805975368373766,
      "learning_rate": 1.1906532417259886e-05,
      "loss": 1.0494,
      "step": 4601
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.19179908996256126,
      "learning_rate": 1.1878304933687246e-05,
      "loss": 0.9993,
      "step": 4602
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2050155215374758,
      "learning_rate": 1.1850108837187335e-05,
      "loss": 0.9274,
      "step": 4603
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2345240577969505,
      "learning_rate": 1.1821944137803009e-05,
      "loss": 0.8967,
      "step": 4604
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.20576311692073965,
      "learning_rate": 1.1793810845565901e-05,
      "loss": 1.0927,
      "step": 4605
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.22386115027863043,
      "learning_rate": 1.1765708970496548e-05,
      "loss": 1.0238,
      "step": 4606
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.16247705075914073,
      "learning_rate": 1.1737638522604177e-05,
      "loss": 0.979,
      "step": 4607
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1852319097174306,
      "learning_rate": 1.1709599511886948e-05,
      "loss": 1.1576,
      "step": 4608
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2018566223443251,
      "learning_rate": 1.1681591948331705e-05,
      "loss": 0.9788,
      "step": 4609
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.20709608655523626,
      "learning_rate": 1.1653615841914189e-05,
      "loss": 0.9091,
      "step": 4610
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.17891691619012584,
      "learning_rate": 1.1625671202598875e-05,
      "loss": 1.0607,
      "step": 4611
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.20303189002330077,
      "learning_rate": 1.1597758040339046e-05,
      "loss": 1.0196,
      "step": 4612
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.23234559815342265,
      "learning_rate": 1.1569876365076793e-05,
      "loss": 1.0911,
      "step": 4613
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.18932675742474514,
      "learning_rate": 1.1542026186742926e-05,
      "loss": 1.004,
      "step": 4614
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2008564830443081,
      "learning_rate": 1.1514207515257147e-05,
      "loss": 1.1055,
      "step": 4615
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2008226387643322,
      "learning_rate": 1.1486420360527838e-05,
      "loss": 0.9498,
      "step": 4616
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.21430177199963182,
      "learning_rate": 1.1458664732452218e-05,
      "loss": 1.0583,
      "step": 4617
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.26386754851556515,
      "learning_rate": 1.1430940640916221e-05,
      "loss": 0.8809,
      "step": 4618
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.17797714476926343,
      "learning_rate": 1.1403248095794628e-05,
      "loss": 1.1101,
      "step": 4619
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.19928145442485098,
      "learning_rate": 1.1375587106950858e-05,
      "loss": 1.0866,
      "step": 4620
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2564393315064652,
      "learning_rate": 1.1347957684237243e-05,
      "loss": 0.9254,
      "step": 4621
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.18221348051756212,
      "learning_rate": 1.1320359837494754e-05,
      "loss": 1.1116,
      "step": 4622
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2198725990317425,
      "learning_rate": 1.129279357655315e-05,
      "loss": 1.0046,
      "step": 4623
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.20486822799845453,
      "learning_rate": 1.1265258911230981e-05,
      "loss": 0.9429,
      "step": 4624
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.21031925172777133,
      "learning_rate": 1.1237755851335463e-05,
      "loss": 0.9562,
      "step": 4625
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2151697587700019,
      "learning_rate": 1.1210284406662652e-05,
      "loss": 1.0735,
      "step": 4626
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.23955181661835956,
      "learning_rate": 1.1182844586997266e-05,
      "loss": 0.9646,
      "step": 4627
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1701914017777097,
      "learning_rate": 1.1155436402112785e-05,
      "loss": 1.0959,
      "step": 4628
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.20735641202153682,
      "learning_rate": 1.11280598617714e-05,
      "loss": 1.0491,
      "step": 4629
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.17933365697158649,
      "learning_rate": 1.1100714975724103e-05,
      "loss": 1.0403,
      "step": 4630
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2154242417912937,
      "learning_rate": 1.107340175371051e-05,
      "loss": 0.996,
      "step": 4631
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.20575401735656593,
      "learning_rate": 1.1046120205459043e-05,
      "loss": 0.9848,
      "step": 4632
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.19031085434394182,
      "learning_rate": 1.1018870340686804e-05,
      "loss": 1.0667,
      "step": 4633
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.17258577265673358,
      "learning_rate": 1.0991652169099598e-05,
      "loss": 1.017,
      "step": 4634
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.24242476817722028,
      "learning_rate": 1.096446570039198e-05,
      "loss": 0.8777,
      "step": 4635
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2507440251368827,
      "learning_rate": 1.0937310944247181e-05,
      "loss": 1.1364,
      "step": 4636
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1863123274427878,
      "learning_rate": 1.0910187910337144e-05,
      "loss": 0.9939,
      "step": 4637
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1911428009031162,
      "learning_rate": 1.0883096608322507e-05,
      "loss": 1.2648,
      "step": 4638
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.18814061813462998,
      "learning_rate": 1.085603704785264e-05,
      "loss": 1.012,
      "step": 4639
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.19502856498161603,
      "learning_rate": 1.0829009238565558e-05,
      "loss": 0.9819,
      "step": 4640
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2212494219371657,
      "learning_rate": 1.0802013190088024e-05,
      "loss": 1.0252,
      "step": 4641
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2109991257274615,
      "learning_rate": 1.0775048912035435e-05,
      "loss": 1.0513,
      "step": 4642
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.21939974332374756,
      "learning_rate": 1.0748116414011888e-05,
      "loss": 1.1417,
      "step": 4643
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19157455489822162,
      "learning_rate": 1.0721215705610178e-05,
      "loss": 1.0335,
      "step": 4644
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1870768969469803,
      "learning_rate": 1.0694346796411747e-05,
      "loss": 1.0167,
      "step": 4645
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2294101698746466,
      "learning_rate": 1.0667509695986754e-05,
      "loss": 1.0658,
      "step": 4646
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1999187327631815,
      "learning_rate": 1.0640704413893965e-05,
      "loss": 0.9328,
      "step": 4647
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19191131354128899,
      "learning_rate": 1.0613930959680906e-05,
      "loss": 1.0509,
      "step": 4648
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.20115458636176456,
      "learning_rate": 1.0587189342883652e-05,
      "loss": 1.0335,
      "step": 4649
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.18396106024800618,
      "learning_rate": 1.056047957302705e-05,
      "loss": 1.025,
      "step": 4650
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.18557504448117232,
      "learning_rate": 1.0533801659624531e-05,
      "loss": 1.0929,
      "step": 4651
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.20473428497627438,
      "learning_rate": 1.0507155612178199e-05,
      "loss": 0.9498,
      "step": 4652
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.20475165964369754,
      "learning_rate": 1.0480541440178816e-05,
      "loss": 0.9681,
      "step": 4653
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.18927147256472437,
      "learning_rate": 1.0453959153105752e-05,
      "loss": 1.1355,
      "step": 4654
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.25569614562513104,
      "learning_rate": 1.0427408760427093e-05,
      "loss": 1.0885,
      "step": 4655
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2225385619117924,
      "learning_rate": 1.0400890271599483e-05,
      "loss": 0.9789,
      "step": 4656
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.17926076190133738,
      "learning_rate": 1.0374403696068291e-05,
      "loss": 1.2494,
      "step": 4657
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.21674357725839605,
      "learning_rate": 1.0347949043267436e-05,
      "loss": 1.0452,
      "step": 4658
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.17974779777810795,
      "learning_rate": 1.0321526322619534e-05,
      "loss": 1.0495,
      "step": 4659
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.21068076673582678,
      "learning_rate": 1.0295135543535772e-05,
      "loss": 1.0601,
      "step": 4660
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.20511902720852862,
      "learning_rate": 1.0268776715415984e-05,
      "loss": 0.9823,
      "step": 4661
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.18309800618084324,
      "learning_rate": 1.0242449847648616e-05,
      "loss": 1.0691,
      "step": 4662
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2307540598597255,
      "learning_rate": 1.0216154949610768e-05,
      "loss": 1.0777,
      "step": 4663
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.18826105054265568,
      "learning_rate": 1.0189892030668114e-05,
      "loss": 1.024,
      "step": 4664
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.18355971337374824,
      "learning_rate": 1.0163661100174904e-05,
      "loss": 1.0229,
      "step": 4665
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.20994927887867557,
      "learning_rate": 1.013746216747411e-05,
      "loss": 1.0593,
      "step": 4666
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1747531756468684,
      "learning_rate": 1.0111295241897157e-05,
      "loss": 1.0851,
      "step": 4667
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.21166366635135023,
      "learning_rate": 1.0085160332764242e-05,
      "loss": 1.0629,
      "step": 4668
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.23197907380850738,
      "learning_rate": 1.0059057449383958e-05,
      "loss": 0.8312,
      "step": 4669
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.20599370135379125,
      "learning_rate": 1.0032986601053673e-05,
      "loss": 1.0008,
      "step": 4670
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2147706698050782,
      "learning_rate": 1.0006947797059219e-05,
      "loss": 1.0505,
      "step": 4671
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.20798215173116194,
      "learning_rate": 9.980941046675107e-06,
      "loss": 0.992,
      "step": 4672
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19341030570153875,
      "learning_rate": 9.954966359164352e-06,
      "loss": 0.9777,
      "step": 4673
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2002380188486891,
      "learning_rate": 9.92902374377862e-06,
      "loss": 0.9721,
      "step": 4674
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19266938943679365,
      "learning_rate": 9.903113209758096e-06,
      "loss": 0.9087,
      "step": 4675
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.17915950040191417,
      "learning_rate": 9.877234766331567e-06,
      "loss": 0.9461,
      "step": 4676
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.18741804272657225,
      "learning_rate": 9.851388422716379e-06,
      "loss": 1.0533,
      "step": 4677
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.18460450510340481,
      "learning_rate": 9.825574188118437e-06,
      "loss": 1.0655,
      "step": 4678
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19090569921221331,
      "learning_rate": 9.79979207173225e-06,
      "loss": 1.0482,
      "step": 4679
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.21353295508525616,
      "learning_rate": 9.774042082740831e-06,
      "loss": 1.134,
      "step": 4680
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.18312545780377093,
      "learning_rate": 9.748324230315809e-06,
      "loss": 0.9659,
      "step": 4681
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1993986455812144,
      "learning_rate": 9.72263852361731e-06,
      "loss": 1.1015,
      "step": 4682
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19653540959337218,
      "learning_rate": 9.696984971794065e-06,
      "loss": 1.0058,
      "step": 4683
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.17828144693209436,
      "learning_rate": 9.671363583983295e-06,
      "loss": 1.0994,
      "step": 4684
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.24356594087697042,
      "learning_rate": 9.64577436931081e-06,
      "loss": 1.1925,
      "step": 4685
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2022387004029611,
      "learning_rate": 9.620217336890936e-06,
      "loss": 1.2042,
      "step": 4686
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19084097559408142,
      "learning_rate": 9.594692495826529e-06,
      "loss": 1.0621,
      "step": 4687
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19835265565941065,
      "learning_rate": 9.569199855209033e-06,
      "loss": 0.9603,
      "step": 4688
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.17969739384159092,
      "learning_rate": 9.543739424118336e-06,
      "loss": 1.0487,
      "step": 4689
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.17658125004480646,
      "learning_rate": 9.518311211622944e-06,
      "loss": 1.083,
      "step": 4690
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.24036016527926282,
      "learning_rate": 9.492915226779808e-06,
      "loss": 0.9578,
      "step": 4691
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19914965532626744,
      "learning_rate": 9.467551478634485e-06,
      "loss": 1.032,
      "step": 4692
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.16569682132530647,
      "learning_rate": 9.442219976220979e-06,
      "loss": 1.0278,
      "step": 4693
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.16913982114137002,
      "learning_rate": 9.416920728561818e-06,
      "loss": 0.9608,
      "step": 4694
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.19951771343006852,
      "learning_rate": 9.391653744668072e-06,
      "loss": 1.0966,
      "step": 4695
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1893030645450548,
      "learning_rate": 9.366419033539275e-06,
      "loss": 1.2282,
      "step": 4696
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.20494865455895422,
      "learning_rate": 9.34121660416354e-06,
      "loss": 0.9445,
      "step": 4697
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1955164683822614,
      "learning_rate": 9.316046465517381e-06,
      "loss": 1.0838,
      "step": 4698
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1944052779410043,
      "learning_rate": 9.29090862656593e-06,
      "loss": 0.9572,
      "step": 4699
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.18838946800819184,
      "learning_rate": 9.265803096262715e-06,
      "loss": 0.9262,
      "step": 4700
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1740635585136717,
      "learning_rate": 9.240729883549792e-06,
      "loss": 1.2435,
      "step": 4701
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1962679511225165,
      "learning_rate": 9.215688997357686e-06,
      "loss": 0.9501,
      "step": 4702
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2156704822195121,
      "learning_rate": 9.19068044660547e-06,
      "loss": 1.1608,
      "step": 4703
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.19127005218295573,
      "learning_rate": 9.165704240200623e-06,
      "loss": 0.9705,
      "step": 4704
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.19001720881930412,
      "learning_rate": 9.14076038703917e-06,
      "loss": 1.0289,
      "step": 4705
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.18343318418771795,
      "learning_rate": 9.115848896005575e-06,
      "loss": 0.8375,
      "step": 4706
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.181761788660198,
      "learning_rate": 9.090969775972736e-06,
      "loss": 1.05,
      "step": 4707
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.21110229066877423,
      "learning_rate": 9.066123035802122e-06,
      "loss": 1.0057,
      "step": 4708
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2096048232473462,
      "learning_rate": 9.041308684343596e-06,
      "loss": 1.1278,
      "step": 4709
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.22850708490448654,
      "learning_rate": 9.016526730435493e-06,
      "loss": 1.0044,
      "step": 4710
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.18333510091273772,
      "learning_rate": 8.991777182904605e-06,
      "loss": 1.2417,
      "step": 4711
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.20242485413355363,
      "learning_rate": 8.967060050566224e-06,
      "loss": 1.1153,
      "step": 4712
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.17149656211987038,
      "learning_rate": 8.942375342224041e-06,
      "loss": 0.9211,
      "step": 4713
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1960914845690546,
      "learning_rate": 8.917723066670247e-06,
      "loss": 1.1691,
      "step": 4714
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.21731692949308512,
      "learning_rate": 8.89310323268544e-06,
      "loss": 1.009,
      "step": 4715
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.17832201556271157,
      "learning_rate": 8.868515849038705e-06,
      "loss": 1.0854,
      "step": 4716
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.20696837413040092,
      "learning_rate": 8.843960924487538e-06,
      "loss": 1.0308,
      "step": 4717
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.21521649293328357,
      "learning_rate": 8.819438467777841e-06,
      "loss": 1.0588,
      "step": 4718
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.25138175651245653,
      "learning_rate": 8.79494848764404e-06,
      "loss": 1.1134,
      "step": 4719
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.26926081887222736,
      "learning_rate": 8.770490992808911e-06,
      "loss": 1.0345,
      "step": 4720
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.27964361285463624,
      "learning_rate": 8.74606599198372e-06,
      "loss": 0.9701,
      "step": 4721
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.18598708547017712,
      "learning_rate": 8.72167349386811e-06,
      "loss": 1.0787,
      "step": 4722
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.18284803581103118,
      "learning_rate": 8.697313507150184e-06,
      "loss": 1.1176,
      "step": 4723
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.239641809877285,
      "learning_rate": 8.672986040506459e-06,
      "loss": 1.0029,
      "step": 4724
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.17798238563451493,
      "learning_rate": 8.648691102601858e-06,
      "loss": 1.0983,
      "step": 4725
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.18170354924579818,
      "learning_rate": 8.624428702089682e-06,
      "loss": 0.8803,
      "step": 4726
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.20597144112042012,
      "learning_rate": 8.600198847611729e-06,
      "loss": 1.0478,
      "step": 4727
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.18567914441206892,
      "learning_rate": 8.576001547798152e-06,
      "loss": 1.1374,
      "step": 4728
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.21271150038146822,
      "learning_rate": 8.551836811267478e-06,
      "loss": 1.0315,
      "step": 4729
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2091518777021944,
      "learning_rate": 8.527704646626711e-06,
      "loss": 0.8971,
      "step": 4730
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.23733422696520065,
      "learning_rate": 8.503605062471187e-06,
      "loss": 0.9933,
      "step": 4731
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.23526847238585744,
      "learning_rate": 8.479538067384706e-06,
      "loss": 0.9762,
      "step": 4732
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2048115373280049,
      "learning_rate": 8.45550366993938e-06,
      "loss": 1.0068,
      "step": 4733
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.21357225688028086,
      "learning_rate": 8.431501878695769e-06,
      "loss": 1.0198,
      "step": 4734
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2339904983678498,
      "learning_rate": 8.40753270220277e-06,
      "loss": 1.1147,
      "step": 4735
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.18361600523238716,
      "learning_rate": 8.383596148997742e-06,
      "loss": 0.9818,
      "step": 4736
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1984096662711891,
      "learning_rate": 8.35969222760633e-06,
      "loss": 1.0212,
      "step": 4737
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.28395736878591016,
      "learning_rate": 8.33582094654265e-06,
      "loss": 0.9659,
      "step": 4738
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.18563460488428127,
      "learning_rate": 8.311982314309109e-06,
      "loss": 1.0351,
      "step": 4739
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.23705212254821958,
      "learning_rate": 8.288176339396514e-06,
      "loss": 1.1522,
      "step": 4740
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.18020768292977657,
      "learning_rate": 8.264403030284074e-06,
      "loss": 1.0087,
      "step": 4741
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.25450774711358326,
      "learning_rate": 8.240662395439325e-06,
      "loss": 1.0577,
      "step": 4742
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.21180887808771656,
      "learning_rate": 8.216954443318181e-06,
      "loss": 0.9716,
      "step": 4743
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.19069127920932522,
      "learning_rate": 8.193279182364888e-06,
      "loss": 1.0477,
      "step": 4744
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.17476830395685616,
      "learning_rate": 8.1696366210121e-06,
      "loss": 1.1031,
      "step": 4745
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.20202124123320828,
      "learning_rate": 8.146026767680758e-06,
      "loss": 0.8984,
      "step": 4746
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.18375966649043643,
      "learning_rate": 8.122449630780238e-06,
      "loss": 0.982,
      "step": 4747
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.19874001576477673,
      "learning_rate": 8.098905218708186e-06,
      "loss": 1.0125,
      "step": 4748
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.19649194723397612,
      "learning_rate": 8.075393539850628e-06,
      "loss": 1.1532,
      "step": 4749
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.18770899461096507,
      "learning_rate": 8.051914602581912e-06,
      "loss": 0.9618,
      "step": 4750
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.208189809817218,
      "learning_rate": 8.028468415264723e-06,
      "loss": 0.938,
      "step": 4751
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.21670894034491905,
      "learning_rate": 8.005054986250137e-06,
      "loss": 1.0331,
      "step": 4752
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.191077170459351,
      "learning_rate": 7.981674323877487e-06,
      "loss": 1.0903,
      "step": 4753
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.19123706893124195,
      "learning_rate": 7.958326436474473e-06,
      "loss": 1.1098,
      "step": 4754
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.19407765557298348,
      "learning_rate": 7.935011332357112e-06,
      "loss": 0.9447,
      "step": 4755
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.23264983112254053,
      "learning_rate": 7.911729019829783e-06,
      "loss": 0.9702,
      "step": 4756
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2034562337239418,
      "learning_rate": 7.888479507185109e-06,
      "loss": 1.0729,
      "step": 4757
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.20204010694628477,
      "learning_rate": 7.865262802704088e-06,
      "loss": 0.9726,
      "step": 4758
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.20240367227536954,
      "learning_rate": 7.842078914656015e-06,
      "loss": 1.0849,
      "step": 4759
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.24989228314874695,
      "learning_rate": 7.818927851298474e-06,
      "loss": 0.9524,
      "step": 4760
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1981952790187592,
      "learning_rate": 7.795809620877426e-06,
      "loss": 1.2266,
      "step": 4761
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.21378780111469964,
      "learning_rate": 7.772724231627048e-06,
      "loss": 0.9465,
      "step": 4762
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.20859187700966164,
      "learning_rate": 7.749671691769911e-06,
      "loss": 1.1193,
      "step": 4763
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2212007787773554,
      "learning_rate": 7.726652009516788e-06,
      "loss": 0.9441,
      "step": 4764
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.25711452186203065,
      "learning_rate": 7.703665193066845e-06,
      "loss": 1.0176,
      "step": 4765
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.17514155394203493,
      "learning_rate": 7.680711250607487e-06,
      "loss": 1.0057,
      "step": 4766
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.16936016049918046,
      "learning_rate": 7.657790190314406e-06,
      "loss": 0.947,
      "step": 4767
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2275486799340905,
      "learning_rate": 7.634902020351575e-06,
      "loss": 0.9554,
      "step": 4768
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.19849145662561302,
      "learning_rate": 7.612046748871327e-06,
      "loss": 1.1409,
      "step": 4769
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1733024983611077,
      "learning_rate": 7.589224384014193e-06,
      "loss": 1.0303,
      "step": 4770
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2212265747173984,
      "learning_rate": 7.566434933909006e-06,
      "loss": 1.0971,
      "step": 4771
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.20273438403171395,
      "learning_rate": 7.543678406672905e-06,
      "loss": 1.0597,
      "step": 4772
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.17806200906819178,
      "learning_rate": 7.520954810411262e-06,
      "loss": 1.0315,
      "step": 4773
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.20924588164715524,
      "learning_rate": 7.49826415321776e-06,
      "loss": 0.8668,
      "step": 4774
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2595256050458666,
      "learning_rate": 7.475606443174288e-06,
      "loss": 1.1696,
      "step": 4775
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.19261813443808595,
      "learning_rate": 7.452981688351079e-06,
      "loss": 1.1152,
      "step": 4776
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2149209536787687,
      "learning_rate": 7.430389896806567e-06,
      "loss": 0.9723,
      "step": 4777
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1694438016433491,
      "learning_rate": 7.40783107658749e-06,
      "loss": 1.0564,
      "step": 4778
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.22744889002576596,
      "learning_rate": 7.385305235728801e-06,
      "loss": 1.0539,
      "step": 4779
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.22685286517554265,
      "learning_rate": 7.362812382253737e-06,
      "loss": 0.9806,
      "step": 4780
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.21325648877987005,
      "learning_rate": 7.340352524173777e-06,
      "loss": 0.8951,
      "step": 4781
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1831611044489155,
      "learning_rate": 7.317925669488623e-06,
      "loss": 0.9221,
      "step": 4782
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.22934420150085807,
      "learning_rate": 7.295531826186264e-06,
      "loss": 1.0963,
      "step": 4783
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1843911186562692,
      "learning_rate": 7.273171002242885e-06,
      "loss": 1.1348,
      "step": 4784
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.22262885757936313,
      "learning_rate": 7.25084320562297e-06,
      "loss": 1.0967,
      "step": 4785
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2056250536467575,
      "learning_rate": 7.2285484442791705e-06,
      "loss": 1.0723,
      "step": 4786
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.22264255451337525,
      "learning_rate": 7.206286726152434e-06,
      "loss": 1.1813,
      "step": 4787
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2861727729589411,
      "learning_rate": 7.184058059171894e-06,
      "loss": 1.0488,
      "step": 4788
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.16706246824670887,
      "learning_rate": 7.161862451254941e-06,
      "loss": 1.0091,
      "step": 4789
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.18140893758799403,
      "learning_rate": 7.139699910307185e-06,
      "loss": 0.8839,
      "step": 4790
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3006805911335413,
      "learning_rate": 7.117570444222432e-06,
      "loss": 1.0249,
      "step": 4791
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.22345357060160803,
      "learning_rate": 7.095474060882734e-06,
      "loss": 0.9637,
      "step": 4792
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.19318596389850923,
      "learning_rate": 7.073410768158339e-06,
      "loss": 0.8542,
      "step": 4793
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.18954511103991747,
      "learning_rate": 7.051380573907751e-06,
      "loss": 1.1033,
      "step": 4794
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.22517308218391796,
      "learning_rate": 7.029383485977625e-06,
      "loss": 1.1152,
      "step": 4795
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.20817817833792918,
      "learning_rate": 7.007419512202884e-06,
      "loss": 1.0623,
      "step": 4796
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.22881779635121344,
      "learning_rate": 6.985488660406625e-06,
      "loss": 1.0701,
      "step": 4797
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1872240326574512,
      "learning_rate": 6.963590938400133e-06,
      "loss": 0.9564,
      "step": 4798
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.17968437740597196,
      "learning_rate": 6.941726353982913e-06,
      "loss": 0.9389,
      "step": 4799
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1973491689134908,
      "learning_rate": 6.919894914942682e-06,
      "loss": 1.0059,
      "step": 4800
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2018273387238054,
      "learning_rate": 6.898096629055318e-06,
      "loss": 1.173,
      "step": 4801
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.17505772248542398,
      "learning_rate": 6.876331504084899e-06,
      "loss": 1.1053,
      "step": 4802
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1843977213260931,
      "learning_rate": 6.854599547783736e-06,
      "loss": 1.033,
      "step": 4803
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.20948486195731403,
      "learning_rate": 6.832900767892237e-06,
      "loss": 0.9635,
      "step": 4804
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.21328794403200704,
      "learning_rate": 6.811235172139097e-06,
      "loss": 1.146,
      "step": 4805
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.20900942437674017,
      "learning_rate": 6.7896027682411215e-06,
      "loss": 0.9465,
      "step": 4806
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2039999253536364,
      "learning_rate": 6.768003563903303e-06,
      "loss": 1.0527,
      "step": 4807
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.20574393408749328,
      "learning_rate": 6.7464375668188085e-06,
      "loss": 1.0425,
      "step": 4808
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1883558889335681,
      "learning_rate": 6.724904784669028e-06,
      "loss": 1.1408,
      "step": 4809
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.23365340445107158,
      "learning_rate": 6.703405225123427e-06,
      "loss": 1.1274,
      "step": 4810
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.23205171407364342,
      "learning_rate": 6.681938895839746e-06,
      "loss": 1.1534,
      "step": 4811
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.18611001652293402,
      "learning_rate": 6.660505804463802e-06,
      "loss": 1.1189,
      "step": 4812
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.21367527858058574,
      "learning_rate": 6.639105958629599e-06,
      "loss": 0.8637,
      "step": 4813
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.19954827806412773,
      "learning_rate": 6.61773936595933e-06,
      "loss": 1.0606,
      "step": 4814
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.19587463595712734,
      "learning_rate": 6.596406034063318e-06,
      "loss": 1.1699,
      "step": 4815
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1965410684524785,
      "learning_rate": 6.5751059705400295e-06,
      "loss": 0.9731,
      "step": 4816
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2018956058492079,
      "learning_rate": 6.553839182976085e-06,
      "loss": 0.9522,
      "step": 4817
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.21251562914704886,
      "learning_rate": 6.532605678946291e-06,
      "loss": 1.0475,
      "step": 4818
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.20414804530416664,
      "learning_rate": 6.5114054660135315e-06,
      "loss": 1.0804,
      "step": 4819
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1902269720350212,
      "learning_rate": 6.490238551728922e-06,
      "loss": 1.012,
      "step": 4820
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.18613347004927205,
      "learning_rate": 6.469104943631621e-06,
      "loss": 0.9558,
      "step": 4821
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.23157691681651815,
      "learning_rate": 6.448004649249018e-06,
      "loss": 1.0275,
      "step": 4822
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2073121458969831,
      "learning_rate": 6.426937676096534e-06,
      "loss": 1.0526,
      "step": 4823
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.19637004291783344,
      "learning_rate": 6.405904031677801e-06,
      "loss": 0.9245,
      "step": 4824
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.19558623729650135,
      "learning_rate": 6.3849037234845584e-06,
      "loss": 1.0957,
      "step": 4825
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.21759401848786616,
      "learning_rate": 6.363936758996658e-06,
      "loss": 1.1051,
      "step": 4826
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.21192397946508712,
      "learning_rate": 6.343003145682114e-06,
      "loss": 0.843,
      "step": 4827
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.20768361626970047,
      "learning_rate": 6.322102890996995e-06,
      "loss": 1.0602,
      "step": 4828
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.20200240693493574,
      "learning_rate": 6.301236002385568e-06,
      "loss": 1.0824,
      "step": 4829
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.19908889667912008,
      "learning_rate": 6.280402487280157e-06,
      "loss": 1.0929,
      "step": 4830
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2195803972376425,
      "learning_rate": 6.2596023531012015e-06,
      "loss": 0.9119,
      "step": 4831
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.22095177938545327,
      "learning_rate": 6.2388356072572765e-06,
      "loss": 0.8359,
      "step": 4832
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.17966727850820632,
      "learning_rate": 6.218102257145064e-06,
      "loss": 0.9377,
      "step": 4833
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.18514643947798057,
      "learning_rate": 6.1974023101493364e-06,
      "loss": 1.2079,
      "step": 4834
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.20931847108841412,
      "learning_rate": 6.176735773642961e-06,
      "loss": 1.0737,
      "step": 4835
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.18097270119901235,
      "learning_rate": 6.156102654986951e-06,
      "loss": 0.9858,
      "step": 4836
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2376431774139092,
      "learning_rate": 6.1355029615303375e-06,
      "loss": 1.1329,
      "step": 4837
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1804187534488962,
      "learning_rate": 6.114936700610341e-06,
      "loss": 0.9255,
      "step": 4838
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.20284721723085794,
      "learning_rate": 6.094403879552213e-06,
      "loss": 1.0058,
      "step": 4839
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.20572659785496206,
      "learning_rate": 6.0739045056692925e-06,
      "loss": 1.0751,
      "step": 4840
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.21946885975120842,
      "learning_rate": 6.053438586263005e-06,
      "loss": 0.9946,
      "step": 4841
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.21736507440498073,
      "learning_rate": 6.03300612862292e-06,
      "loss": 1.1336,
      "step": 4842
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.19844874856489017,
      "learning_rate": 6.012607140026605e-06,
      "loss": 0.8474,
      "step": 4843
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1796158523851667,
      "learning_rate": 5.992241627739781e-06,
      "loss": 1.103,
      "step": 4844
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.19311578361619003,
      "learning_rate": 5.97190959901619e-06,
      "loss": 1.1922,
      "step": 4845
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.23530539325221933,
      "learning_rate": 5.951611061097672e-06,
      "loss": 0.9534,
      "step": 4846
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.20702223645064327,
      "learning_rate": 5.931346021214123e-06,
      "loss": 1.2219,
      "step": 4847
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1835568418254046,
      "learning_rate": 5.911114486583525e-06,
      "loss": 0.9488,
      "step": 4848
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.19305519580447836,
      "learning_rate": 5.890916464411933e-06,
      "loss": 1.1524,
      "step": 4849
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1827290828520329,
      "learning_rate": 5.870751961893417e-06,
      "loss": 0.866,
      "step": 4850
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1852099886069203,
      "learning_rate": 5.850620986210198e-06,
      "loss": 0.8698,
      "step": 4851
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1982854604240957,
      "learning_rate": 5.830523544532451e-06,
      "loss": 0.9891,
      "step": 4852
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.20307618373080394,
      "learning_rate": 5.810459644018506e-06,
      "loss": 0.919,
      "step": 4853
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.19806531948756775,
      "learning_rate": 5.790429291814659e-06,
      "loss": 1.1381,
      "step": 4854
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.18436990171396503,
      "learning_rate": 5.770432495055311e-06,
      "loss": 0.9319,
      "step": 4855
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1768330107157463,
      "learning_rate": 5.7504692608628785e-06,
      "loss": 0.9863,
      "step": 4856
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.18661147531179817,
      "learning_rate": 5.730539596347851e-06,
      "loss": 1.1562,
      "step": 4857
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19469538317190957,
      "learning_rate": 5.710643508608771e-06,
      "loss": 1.0721,
      "step": 4858
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.17462201430377317,
      "learning_rate": 5.69078100473216e-06,
      "loss": 1.0998,
      "step": 4859
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2211206250330354,
      "learning_rate": 5.670952091792658e-06,
      "loss": 0.952,
      "step": 4860
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19307403937868828,
      "learning_rate": 5.6511567768528835e-06,
      "loss": 1.0149,
      "step": 4861
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.18007857218078355,
      "learning_rate": 5.631395066963507e-06,
      "loss": 1.1073,
      "step": 4862
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2058735630002251,
      "learning_rate": 5.611666969163243e-06,
      "loss": 1.1574,
      "step": 4863
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1792505011172691,
      "learning_rate": 5.591972490478814e-06,
      "loss": 0.9729,
      "step": 4864
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.17722741018619667,
      "learning_rate": 5.572311637924965e-06,
      "loss": 1.1676,
      "step": 4865
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2056972875222307,
      "learning_rate": 5.552684418504462e-06,
      "loss": 0.9038,
      "step": 4866
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19412242764974683,
      "learning_rate": 5.533090839208133e-06,
      "loss": 1.0452,
      "step": 4867
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1904363064724335,
      "learning_rate": 5.513530907014763e-06,
      "loss": 0.9768,
      "step": 4868
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.21052416458960227,
      "learning_rate": 5.494004628891214e-06,
      "loss": 1.1611,
      "step": 4869
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2148969293657254,
      "learning_rate": 5.47451201179231e-06,
      "loss": 1.0716,
      "step": 4870
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.217129882323183,
      "learning_rate": 5.455053062660931e-06,
      "loss": 0.9793,
      "step": 4871
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.18633536874962414,
      "learning_rate": 5.4356277884278996e-06,
      "loss": 0.9538,
      "step": 4872
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19884529322041994,
      "learning_rate": 5.416236196012126e-06,
      "loss": 1.1464,
      "step": 4873
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1826118332642266,
      "learning_rate": 5.396878292320451e-06,
      "loss": 1.0876,
      "step": 4874
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19361850197647382,
      "learning_rate": 5.377554084247771e-06,
      "loss": 1.1883,
      "step": 4875
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.23693174080663656,
      "learning_rate": 5.358263578676959e-06,
      "loss": 1.0216,
      "step": 4876
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.21054721339110974,
      "learning_rate": 5.33900678247885e-06,
      "loss": 1.051,
      "step": 4877
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.20317979262249547,
      "learning_rate": 5.319783702512349e-06,
      "loss": 1.0997,
      "step": 4878
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19314331196574355,
      "learning_rate": 5.300594345624288e-06,
      "loss": 1.028,
      "step": 4879
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.20876357080827743,
      "learning_rate": 5.281438718649501e-06,
      "loss": 0.8794,
      "step": 4880
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.18855396847577346,
      "learning_rate": 5.262316828410807e-06,
      "loss": 1.0494,
      "step": 4881
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1755390127096264,
      "learning_rate": 5.243228681719048e-06,
      "loss": 0.9654,
      "step": 4882
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.21087515842689172,
      "learning_rate": 5.224174285372974e-06,
      "loss": 1.0584,
      "step": 4883
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2085378721520731,
      "learning_rate": 5.2051536461594e-06,
      "loss": 0.9563,
      "step": 4884
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19033884464734316,
      "learning_rate": 5.18616677085304e-06,
      "loss": 1.1305,
      "step": 4885
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2153933930509571,
      "learning_rate": 5.16721366621663e-06,
      "loss": 1.0118,
      "step": 4886
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1951899907517869,
      "learning_rate": 5.14829433900087e-06,
      "loss": 1.1349,
      "step": 4887
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.17701263755522414,
      "learning_rate": 5.129408795944412e-06,
      "loss": 1.155,
      "step": 4888
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.21223491152450344,
      "learning_rate": 5.110557043773878e-06,
      "loss": 0.9794,
      "step": 4889
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.17164660453372446,
      "learning_rate": 5.091739089203862e-06,
      "loss": 1.0754,
      "step": 4890
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.20450061924934781,
      "learning_rate": 5.0729549389369245e-06,
      "loss": 0.9209,
      "step": 4891
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2513414208190392,
      "learning_rate": 5.054204599663581e-06,
      "loss": 1.0902,
      "step": 4892
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.16175081298619864,
      "learning_rate": 5.035488078062301e-06,
      "loss": 1.0895,
      "step": 4893
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.21303794968215273,
      "learning_rate": 5.0168053807995055e-06,
      "loss": 0.9999,
      "step": 4894
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.21090722475096968,
      "learning_rate": 4.998156514529595e-06,
      "loss": 1.0275,
      "step": 4895
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19834101518382322,
      "learning_rate": 4.979541485894889e-06,
      "loss": 1.0239,
      "step": 4896
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.18832263038106345,
      "learning_rate": 4.960960301525652e-06,
      "loss": 1.0372,
      "step": 4897
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.21722042146821433,
      "learning_rate": 4.942412968040111e-06,
      "loss": 1.07,
      "step": 4898
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.24383733354310771,
      "learning_rate": 4.923899492044437e-06,
      "loss": 1.0338,
      "step": 4899
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.18099852911522135,
      "learning_rate": 4.905419880132733e-06,
      "loss": 1.0508,
      "step": 4900
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19164846471023791,
      "learning_rate": 4.886974138887035e-06,
      "loss": 1.0246,
      "step": 4901
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19014649943611053,
      "learning_rate": 4.868562274877353e-06,
      "loss": 1.0691,
      "step": 4902
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.207182296752186,
      "learning_rate": 4.850184294661575e-06,
      "loss": 0.936,
      "step": 4903
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2031510832157943,
      "learning_rate": 4.831840204785553e-06,
      "loss": 1.0478,
      "step": 4904
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2211021936055546,
      "learning_rate": 4.813530011783052e-06,
      "loss": 1.0667,
      "step": 4905
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19062205488785586,
      "learning_rate": 4.795253722175786e-06,
      "loss": 1.0903,
      "step": 4906
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2021750199103566,
      "learning_rate": 4.777011342473392e-06,
      "loss": 0.8962,
      "step": 4907
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.17417924436946824,
      "learning_rate": 4.758802879173374e-06,
      "loss": 1.0018,
      "step": 4908
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2272181456969855,
      "learning_rate": 4.740628338761255e-06,
      "loss": 1.1546,
      "step": 4909
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1834833216326334,
      "learning_rate": 4.722487727710368e-06,
      "loss": 0.9314,
      "step": 4910
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.216453901208962,
      "learning_rate": 4.704381052482055e-06,
      "loss": 1.0771,
      "step": 4911
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.19496355930473747,
      "learning_rate": 4.686308319525512e-06,
      "loss": 1.0187,
      "step": 4912
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2148183212972903,
      "learning_rate": 4.668269535277858e-06,
      "loss": 1.1373,
      "step": 4913
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.19733972106384917,
      "learning_rate": 4.650264706164109e-06,
      "loss": 1.1809,
      "step": 4914
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.16166308902635265,
      "learning_rate": 4.632293838597246e-06,
      "loss": 1.1787,
      "step": 4915
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.19502817449159052,
      "learning_rate": 4.614356938978059e-06,
      "loss": 1.1187,
      "step": 4916
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.221461292291132,
      "learning_rate": 4.596454013695328e-06,
      "loss": 0.997,
      "step": 4917
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.20949687957385577,
      "learning_rate": 4.578585069125685e-06,
      "loss": 1.2415,
      "step": 4918
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.18430876268891327,
      "learning_rate": 4.560750111633638e-06,
      "loss": 1.0557,
      "step": 4919
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.21765532380515995,
      "learning_rate": 4.542949147571684e-06,
      "loss": 1.0777,
      "step": 4920
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1939867220207641,
      "learning_rate": 4.5251821832800746e-06,
      "loss": 1.0785,
      "step": 4921
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.20634171703064896,
      "learning_rate": 4.50744922508708e-06,
      "loss": 0.9599,
      "step": 4922
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.19946341143691773,
      "learning_rate": 4.489750279308757e-06,
      "loss": 1.0354,
      "step": 4923
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.18628228459332077,
      "learning_rate": 4.472085352249134e-06,
      "loss": 1.0438,
      "step": 4924
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.17319450910541323,
      "learning_rate": 4.454454450200051e-06,
      "loss": 0.9887,
      "step": 4925
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.18656609387810696,
      "learning_rate": 4.436857579441289e-06,
      "loss": 0.9895,
      "step": 4926
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2151041222068984,
      "learning_rate": 4.4192947462404654e-06,
      "loss": 1.2273,
      "step": 4927
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2028009209017946,
      "learning_rate": 4.401765956853088e-06,
      "loss": 0.9401,
      "step": 4928
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1727560292786269,
      "learning_rate": 4.3842712175225285e-06,
      "loss": 1.0991,
      "step": 4929
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2223315152738978,
      "learning_rate": 4.366810534480037e-06,
      "loss": 1.0575,
      "step": 4930
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.21031773659487732,
      "learning_rate": 4.349383913944771e-06,
      "loss": 0.8656,
      "step": 4931
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.21960837903064157,
      "learning_rate": 4.331991362123689e-06,
      "loss": 1.0144,
      "step": 4932
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.20880746935407032,
      "learning_rate": 4.314632885211667e-06,
      "loss": 0.9549,
      "step": 4933
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.17951438173442866,
      "learning_rate": 4.297308489391405e-06,
      "loss": 1.0395,
      "step": 4934
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.32583713463651187,
      "learning_rate": 4.280018180833501e-06,
      "loss": 1.0225,
      "step": 4935
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.18683734105232122,
      "learning_rate": 4.262761965696404e-06,
      "loss": 1.1472,
      "step": 4936
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.20428237252889003,
      "learning_rate": 4.24553985012639e-06,
      "loss": 0.9529,
      "step": 4937
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.21952176906159648,
      "learning_rate": 4.228351840257605e-06,
      "loss": 0.9935,
      "step": 4938
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.21211208062044323,
      "learning_rate": 4.211197942212086e-06,
      "loss": 1.0965,
      "step": 4939
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2222916404983346,
      "learning_rate": 4.194078162099657e-06,
      "loss": 0.9833,
      "step": 4940
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.19178181429486238,
      "learning_rate": 4.176992506018018e-06,
      "loss": 1.0948,
      "step": 4941
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.16834315884058665,
      "learning_rate": 4.159940980052734e-06,
      "loss": 0.9467,
      "step": 4942
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2237345003077268,
      "learning_rate": 4.142923590277181e-06,
      "loss": 1.0519,
      "step": 4943
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.17667487385476854,
      "learning_rate": 4.12594034275261e-06,
      "loss": 1.1255,
      "step": 4944
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2004683114225347,
      "learning_rate": 4.1089912435280705e-06,
      "loss": 0.9303,
      "step": 4945
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.19351457364205815,
      "learning_rate": 4.09207629864049e-06,
      "loss": 1.0454,
      "step": 4946
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.23347322581146873,
      "learning_rate": 4.075195514114593e-06,
      "loss": 1.0144,
      "step": 4947
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.20280677718382026,
      "learning_rate": 4.058348895962971e-06,
      "loss": 0.977,
      "step": 4948
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.19881585206340177,
      "learning_rate": 4.041536450186012e-06,
      "loss": 0.8928,
      "step": 4949
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.19122622519264193,
      "learning_rate": 4.024758182771982e-06,
      "loss": 1.0141,
      "step": 4950
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.18629309436387545,
      "learning_rate": 4.008014099696922e-06,
      "loss": 1.0432,
      "step": 4951
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.22098178231665147,
      "learning_rate": 3.991304206924728e-06,
      "loss": 1.1383,
      "step": 4952
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.24598833261250974,
      "learning_rate": 3.974628510407097e-06,
      "loss": 1.0577,
      "step": 4953
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.19794929567832054,
      "learning_rate": 3.957987016083553e-06,
      "loss": 0.9431,
      "step": 4954
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.208544606978024,
      "learning_rate": 3.941379729881456e-06,
      "loss": 0.8522,
      "step": 4955
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2055569413296833,
      "learning_rate": 3.924806657715951e-06,
      "loss": 0.953,
      "step": 4956
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.22565775957798362,
      "learning_rate": 3.908267805490051e-06,
      "loss": 1.0408,
      "step": 4957
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.23574125559040074,
      "learning_rate": 3.8917631790944985e-06,
      "loss": 1.0238,
      "step": 4958
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1881903875777562,
      "learning_rate": 3.875292784407924e-06,
      "loss": 1.1119,
      "step": 4959
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2521841124103999,
      "learning_rate": 3.858856627296714e-06,
      "loss": 1.0176,
      "step": 4960
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.17773952694154213,
      "learning_rate": 3.842454713615084e-06,
      "loss": 1.0965,
      "step": 4961
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1862783960531154,
      "learning_rate": 3.826087049205041e-06,
      "loss": 1.0109,
      "step": 4962
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.23404647804347584,
      "learning_rate": 3.8097536398963963e-06,
      "loss": 0.977,
      "step": 4963
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.17307447855762736,
      "learning_rate": 3.793454491506776e-06,
      "loss": 1.0416,
      "step": 4964
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2202341831938504,
      "learning_rate": 3.7771896098415695e-06,
      "loss": 1.057,
      "step": 4965
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.18717735098373028,
      "learning_rate": 3.760959000694009e-06,
      "loss": 1.0858,
      "step": 4966
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.18076198702255433,
      "learning_rate": 3.74476266984507e-06,
      "loss": 0.911,
      "step": 4967
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.21942051846966046,
      "learning_rate": 3.7286006230635495e-06,
      "loss": 1.0313,
      "step": 4968
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.16357406532300095,
      "learning_rate": 3.7124728661060317e-06,
      "loss": 1.0246,
      "step": 4969
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.19158064973631245,
      "learning_rate": 3.6963794047168763e-06,
      "loss": 1.094,
      "step": 4970
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1963240673520008,
      "learning_rate": 3.6803202446282214e-06,
      "loss": 0.9148,
      "step": 4971
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.19080894164225345,
      "learning_rate": 3.664295391559991e-06,
      "loss": 1.0089,
      "step": 4972
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.18470721133970353,
      "learning_rate": 3.648304851219919e-06,
      "loss": 1.0759,
      "step": 4973
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.198249572564054,
      "learning_rate": 3.6323486293034835e-06,
      "loss": 1.0799,
      "step": 4974
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2146748113565556,
      "learning_rate": 3.6164267314939713e-06,
      "loss": 1.0045,
      "step": 4975
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.22233665514702658,
      "learning_rate": 3.600539163462402e-06,
      "loss": 1.0368,
      "step": 4976
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.19812662247758228,
      "learning_rate": 3.5846859308675927e-06,
      "loss": 0.9841,
      "step": 4977
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.22504831589165658,
      "learning_rate": 3.568867039356116e-06,
      "loss": 0.9943,
      "step": 4978
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.20943091349932083,
      "learning_rate": 3.5530824945623542e-06,
      "loss": 0.937,
      "step": 4979
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2066868231233379,
      "learning_rate": 3.5373323021083873e-06,
      "loss": 1.0383,
      "step": 4980
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.22074462775517467,
      "learning_rate": 3.5216164676041386e-06,
      "loss": 0.9919,
      "step": 4981
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.19722030578543762,
      "learning_rate": 3.505934996647242e-06,
      "loss": 0.9156,
      "step": 4982
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.19164419462890206,
      "learning_rate": 3.4902878948230854e-06,
      "loss": 1.0497,
      "step": 4983
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1632238040214539,
      "learning_rate": 3.474675167704855e-06,
      "loss": 1.1138,
      "step": 4984
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.25232633786556635,
      "learning_rate": 3.459096820853469e-06,
      "loss": 1.0158,
      "step": 4985
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.18440106255579147,
      "learning_rate": 3.4435528598175894e-06,
      "loss": 1.0606,
      "step": 4986
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.16032153112940564,
      "learning_rate": 3.4280432901336425e-06,
      "loss": 0.9844,
      "step": 4987
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.19069400812770543,
      "learning_rate": 3.4125681173258316e-06,
      "loss": 1.0875,
      "step": 4988
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.23849863262505291,
      "learning_rate": 3.3971273469060592e-06,
      "loss": 0.9293,
      "step": 4989
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.189663249390863,
      "learning_rate": 3.381720984374015e-06,
      "loss": 0.9515,
      "step": 4990
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.20028580559832942,
      "learning_rate": 3.366349035217109e-06,
      "loss": 1.119,
      "step": 4991
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.20848292548806688,
      "learning_rate": 3.351011504910517e-06,
      "loss": 1.0176,
      "step": 4992
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.19035079758641635,
      "learning_rate": 3.3357083989171256e-06,
      "loss": 1.1209,
      "step": 4993
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.19357639443382815,
      "learning_rate": 3.320439722687574e-06,
      "loss": 1.0506,
      "step": 4994
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.19153309625175077,
      "learning_rate": 3.3052054816602452e-06,
      "loss": 1.0777,
      "step": 4995
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.16839202955419777,
      "learning_rate": 3.2900056812612436e-06,
      "loss": 1.0261,
      "step": 4996
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.22003372042160663,
      "learning_rate": 3.274840326904438e-06,
      "loss": 0.9038,
      "step": 4997
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2773489565849313,
      "learning_rate": 3.259709423991375e-06,
      "loss": 1.0193,
      "step": 4998
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.20564489868029642,
      "learning_rate": 3.2446129779113877e-06,
      "loss": 0.9314,
      "step": 4999
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.21326997091492367,
      "learning_rate": 3.229550994041497e-06,
      "loss": 0.9579,
      "step": 5000
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2461489115480062,
      "learning_rate": 3.214523477746456e-06,
      "loss": 1.1307,
      "step": 5001
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2249169196065553,
      "learning_rate": 3.199530434378739e-06,
      "loss": 0.8585,
      "step": 5002
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1814606381245753,
      "learning_rate": 3.184571869278574e-06,
      "loss": 1.023,
      "step": 5003
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.18388114108876297,
      "learning_rate": 3.169647787773866e-06,
      "loss": 1.1634,
      "step": 5004
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2700208982487346,
      "learning_rate": 3.154758195180252e-06,
      "loss": 1.0184,
      "step": 5005
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.19587656007858775,
      "learning_rate": 3.1399030968011e-06,
      "loss": 1.1042,
      "step": 5006
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.19277202892382284,
      "learning_rate": 3.1250824979274675e-06,
      "loss": 1.0822,
      "step": 5007
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.18110406026588483,
      "learning_rate": 3.1102964038381644e-06,
      "loss": 1.0021,
      "step": 5008
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.17448179589330118,
      "learning_rate": 3.095544819799656e-06,
      "loss": 1.05,
      "step": 5009
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.20012646829632294,
      "learning_rate": 3.080827751066151e-06,
      "loss": 0.9935,
      "step": 5010
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.18099750141253615,
      "learning_rate": 3.0661452028795336e-06,
      "loss": 1.0529,
      "step": 5011
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1870712490413967,
      "learning_rate": 3.0514971804694535e-06,
      "loss": 0.9477,
      "step": 5012
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.19261464155001654,
      "learning_rate": 3.036883689053205e-06,
      "loss": 1.0759,
      "step": 5013
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.25302128941288404,
      "learning_rate": 3.0223047338358013e-06,
      "loss": 0.9959,
      "step": 5014
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1771223590945275,
      "learning_rate": 3.007760320009967e-06,
      "loss": 1.0103,
      "step": 5015
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.21530293533554098,
      "learning_rate": 2.9932504527560913e-06,
      "loss": 1.2167,
      "step": 5016
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.20527376980055387,
      "learning_rate": 2.9787751372423066e-06,
      "loss": 1.1347,
      "step": 5017
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2237065093455414,
      "learning_rate": 2.9643343786244006e-06,
      "loss": 1.0494,
      "step": 5018
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.23705323774241746,
      "learning_rate": 2.9499281820458692e-06,
      "loss": 1.0157,
      "step": 5019
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1719174934514086,
      "learning_rate": 2.9355565526378635e-06,
      "loss": 0.7988,
      "step": 5020
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18561565675713085,
      "learning_rate": 2.921219495519301e-06,
      "loss": 1.1061,
      "step": 5021
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.21278569802729988,
      "learning_rate": 2.906917015796695e-06,
      "loss": 0.9568,
      "step": 5022
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2176561095633566,
      "learning_rate": 2.8926491185643058e-06,
      "loss": 1.0559,
      "step": 5023
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.23263694184170347,
      "learning_rate": 2.8784158089040556e-06,
      "loss": 1.1689,
      "step": 5024
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2508972618109188,
      "learning_rate": 2.8642170918855347e-06,
      "loss": 1.0417,
      "step": 5025
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1743876417560049,
      "learning_rate": 2.8500529725660417e-06,
      "loss": 1.0786,
      "step": 5026
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18697297468343893,
      "learning_rate": 2.835923455990508e-06,
      "loss": 1.1147,
      "step": 5027
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.19546609519440794,
      "learning_rate": 2.821828547191585e-06,
      "loss": 1.1205,
      "step": 5028
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1908060088736038,
      "learning_rate": 2.8077682511895687e-06,
      "loss": 0.9956,
      "step": 5029
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18634624031054864,
      "learning_rate": 2.7937425729924527e-06,
      "loss": 0.9708,
      "step": 5030
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1962395305070063,
      "learning_rate": 2.7797515175958744e-06,
      "loss": 1.1026,
      "step": 5031
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1973095085326209,
      "learning_rate": 2.7657950899831585e-06,
      "loss": 1.0479,
      "step": 5032
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2242930447362119,
      "learning_rate": 2.7518732951252847e-06,
      "loss": 1.104,
      "step": 5033
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1811067410671908,
      "learning_rate": 2.7379861379808856e-06,
      "loss": 1.0137,
      "step": 5034
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.16964296268043494,
      "learning_rate": 2.7241336234962944e-06,
      "loss": 0.9864,
      "step": 5035
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.23860238225033253,
      "learning_rate": 2.7103157566054525e-06,
      "loss": 0.9671,
      "step": 5036
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.20480075661972763,
      "learning_rate": 2.6965325422300237e-06,
      "loss": 1.0258,
      "step": 5037
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.233640505466791,
      "learning_rate": 2.682783985279269e-06,
      "loss": 0.9644,
      "step": 5038
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.19750509364321234,
      "learning_rate": 2.6690700906501496e-06,
      "loss": 0.9374,
      "step": 5039
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.22045780809306198,
      "learning_rate": 2.655390863227247e-06,
      "loss": 0.9419,
      "step": 5040
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.20337337174213402,
      "learning_rate": 2.6417463078828307e-06,
      "loss": 0.9543,
      "step": 5041
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1963875358355897,
      "learning_rate": 2.628136429476791e-06,
      "loss": 0.948,
      "step": 5042
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2433402467855819,
      "learning_rate": 2.6145612328566717e-06,
      "loss": 0.8326,
      "step": 5043
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.21695862645739125,
      "learning_rate": 2.6010207228576723e-06,
      "loss": 1.0654,
      "step": 5044
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.19886580585257402,
      "learning_rate": 2.5875149043026348e-06,
      "loss": 1.0488,
      "step": 5045
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.20219165282560367,
      "learning_rate": 2.574043782002056e-06,
      "loss": 1.1297,
      "step": 5046
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1870271924210793,
      "learning_rate": 2.5606073607540305e-06,
      "loss": 0.8607,
      "step": 5047
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18352039812326199,
      "learning_rate": 2.5472056453443638e-06,
      "loss": 1.0053,
      "step": 5048
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.21071746392592877,
      "learning_rate": 2.5338386405464377e-06,
      "loss": 1.0269,
      "step": 5049
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1520102434897998,
      "learning_rate": 2.52050635112131e-06,
      "loss": 0.9191,
      "step": 5050
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18083060410003773,
      "learning_rate": 2.5072087818176382e-06,
      "loss": 1.0658,
      "step": 5051
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2358680616165229,
      "learning_rate": 2.4939459373717556e-06,
      "loss": 1.2598,
      "step": 5052
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.20433068324093703,
      "learning_rate": 2.4807178225075945e-06,
      "loss": 1.0311,
      "step": 5053
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.20697078982423914,
      "learning_rate": 2.4675244419367305e-06,
      "loss": 1.0673,
      "step": 5054
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18658369495572083,
      "learning_rate": 2.4543658003583604e-06,
      "loss": 0.9461,
      "step": 5055
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18664305543512472,
      "learning_rate": 2.4412419024593127e-06,
      "loss": 1.0241,
      "step": 5056
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.21192946408506907,
      "learning_rate": 2.428152752914059e-06,
      "loss": 0.9962,
      "step": 5057
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1870561573656927,
      "learning_rate": 2.4150983563846486e-06,
      "loss": 1.0752,
      "step": 5058
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.22877446820385985,
      "learning_rate": 2.402078717520795e-06,
      "loss": 0.986,
      "step": 5059
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1883206899814044,
      "learning_rate": 2.3890938409597995e-06,
      "loss": 1.124,
      "step": 5060
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18984012935734396,
      "learning_rate": 2.37614373132663e-06,
      "loss": 0.9243,
      "step": 5061
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.23417653951535256,
      "learning_rate": 2.3632283932338074e-06,
      "loss": 1.0964,
      "step": 5062
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.22795039577795284,
      "learning_rate": 2.3503478312815298e-06,
      "loss": 0.8504,
      "step": 5063
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.21881091301102432,
      "learning_rate": 2.3375020500575606e-06,
      "loss": 1.0639,
      "step": 5064
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1757163101411366,
      "learning_rate": 2.324691054137296e-06,
      "loss": 1.27,
      "step": 5065
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2141767591289641,
      "learning_rate": 2.3119148480837406e-06,
      "loss": 1.2416,
      "step": 5066
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1765522439877872,
      "learning_rate": 2.2991734364475215e-06,
      "loss": 1.0603,
      "step": 5067
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.21493875251440292,
      "learning_rate": 2.286466823766831e-06,
      "loss": 0.9838,
      "step": 5068
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.22143932577240671,
      "learning_rate": 2.2737950145675147e-06,
      "loss": 0.9663,
      "step": 5069
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.20064129230923017,
      "learning_rate": 2.261158013362996e-06,
      "loss": 1.2167,
      "step": 5070
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.20966143177387328,
      "learning_rate": 2.248555824654297e-06,
      "loss": 1.0605,
      "step": 5071
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.19614869498930826,
      "learning_rate": 2.2359884529300603e-06,
      "loss": 1.135,
      "step": 5072
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.20448465638279956,
      "learning_rate": 2.2234559026665157e-06,
      "loss": 1.1211,
      "step": 5073
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18024961450791685,
      "learning_rate": 2.2109581783275047e-06,
      "loss": 0.957,
      "step": 5074
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1836776406802868,
      "learning_rate": 2.19849528436441e-06,
      "loss": 1.1809,
      "step": 5075
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2201061852428156,
      "learning_rate": 2.186067225216293e-06,
      "loss": 1.0434,
      "step": 5076
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.17738875018679964,
      "learning_rate": 2.1736740053097338e-06,
      "loss": 1.0439,
      "step": 5077
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.20954184347992122,
      "learning_rate": 2.1613156290589465e-06,
      "loss": 1.1121,
      "step": 5078
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.20029193386121988,
      "learning_rate": 2.148992100865721e-06,
      "loss": 0.9368,
      "step": 5079
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1856022787062861,
      "learning_rate": 2.136703425119424e-06,
      "loss": 0.9449,
      "step": 5080
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.18246134017839494,
      "learning_rate": 2.1244496061970433e-06,
      "loss": 0.9927,
      "step": 5081
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.17162854966140922,
      "learning_rate": 2.1122306484631204e-06,
      "loss": 0.9769,
      "step": 5082
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.21036216913991768,
      "learning_rate": 2.1000465562697856e-06,
      "loss": 0.9907,
      "step": 5083
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.24464812653474752,
      "learning_rate": 2.087897333956734e-06,
      "loss": 1.0263,
      "step": 5084
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.20284114950507412,
      "learning_rate": 2.075782985851293e-06,
      "loss": 0.8931,
      "step": 5085
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.19303491973022277,
      "learning_rate": 2.0637035162683115e-06,
      "loss": 0.9181,
      "step": 5086
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1768365013800959,
      "learning_rate": 2.0516589295102583e-06,
      "loss": 1.141,
      "step": 5087
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1803154938801285,
      "learning_rate": 2.0396492298671467e-06,
      "loss": 0.991,
      "step": 5088
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1976639469791424,
      "learning_rate": 2.0276744216165654e-06,
      "loss": 1.0269,
      "step": 5089
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.23230390027522213,
      "learning_rate": 2.0157345090237145e-06,
      "loss": 1.0282,
      "step": 5090
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.21505619124810696,
      "learning_rate": 2.003829496341325e-06,
      "loss": 1.0185,
      "step": 5091
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.18553925451353645,
      "learning_rate": 1.9919593878096944e-06,
      "loss": 1.0357,
      "step": 5092
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1831186617981922,
      "learning_rate": 1.980124187656718e-06,
      "loss": 1.1365,
      "step": 5093
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1763016010999441,
      "learning_rate": 1.968323900097835e-06,
      "loss": 1.0656,
      "step": 5094
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.19384378252376147,
      "learning_rate": 1.956558529336061e-06,
      "loss": 1.0246,
      "step": 5095
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.20184043678170815,
      "learning_rate": 1.944828079561978e-06,
      "loss": 1.1648,
      "step": 5096
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.20341194393647904,
      "learning_rate": 1.9331325549536984e-06,
      "loss": 1.0661,
      "step": 5097
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.19781262797645494,
      "learning_rate": 1.921471959676957e-06,
      "loss": 1.1819,
      "step": 5098
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.19244552934404416,
      "learning_rate": 1.9098462978849873e-06,
      "loss": 1.1579,
      "step": 5099
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.18038277772294206,
      "learning_rate": 1.898255573718588e-06,
      "loss": 1.1592,
      "step": 5100
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2312649510250303,
      "learning_rate": 1.8866997913061457e-06,
      "loss": 0.9483,
      "step": 5101
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.18305828311905328,
      "learning_rate": 1.87517895476359e-06,
      "loss": 1.0497,
      "step": 5102
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2253126705324755,
      "learning_rate": 1.8636930681943831e-06,
      "loss": 1.1793,
      "step": 5103
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2109049132889334,
      "learning_rate": 1.8522421356895637e-06,
      "loss": 1.0107,
      "step": 5104
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1824294779139822,
      "learning_rate": 1.8408261613277022e-06,
      "loss": 1.0182,
      "step": 5105
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2007516552419863,
      "learning_rate": 1.8294451491749464e-06,
      "loss": 1.0137,
      "step": 5106
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2034330052663484,
      "learning_rate": 1.8180991032849426e-06,
      "loss": 0.9579,
      "step": 5107
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.17626968891840109,
      "learning_rate": 1.8067880276989134e-06,
      "loss": 1.0326,
      "step": 5108
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.17065068474908424,
      "learning_rate": 1.7955119264456476e-06,
      "loss": 1.1522,
      "step": 5109
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.21360527461916753,
      "learning_rate": 1.7842708035414434e-06,
      "loss": 1.0928,
      "step": 5110
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.21032670208163565,
      "learning_rate": 1.7730646629901315e-06,
      "loss": 1.093,
      "step": 5111
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1977941055654304,
      "learning_rate": 1.761893508783119e-06,
      "loss": 1.2832,
      "step": 5112
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.195883124850744,
      "learning_rate": 1.7507573448993231e-06,
      "loss": 1.0665,
      "step": 5113
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.19271215336763758,
      "learning_rate": 1.7396561753052154e-06,
      "loss": 1.076,
      "step": 5114
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.17696972510132578,
      "learning_rate": 1.7285900039547998e-06,
      "loss": 1.1334,
      "step": 5115
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2282466807477438,
      "learning_rate": 1.717558834789601e-06,
      "loss": 0.9713,
      "step": 5116
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.19009654686659266,
      "learning_rate": 1.7065626717386874e-06,
      "loss": 1.1004,
      "step": 5117
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.20267601550053135,
      "learning_rate": 1.695601518718659e-06,
      "loss": 1.1715,
      "step": 5118
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.16983942986101175,
      "learning_rate": 1.684675379633649e-06,
      "loss": 0.9997,
      "step": 5119
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.21837947413586434,
      "learning_rate": 1.6737842583753216e-06,
      "loss": 1.032,
      "step": 5120
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.20821629568492228,
      "learning_rate": 1.662928158822863e-06,
      "loss": 0.9948,
      "step": 5121
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.21248657472169477,
      "learning_rate": 1.6521070848429576e-06,
      "loss": 1.0939,
      "step": 5122
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.19339562181313208,
      "learning_rate": 1.6413210402898893e-06,
      "loss": 0.9533,
      "step": 5123
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.19916152814429994,
      "learning_rate": 1.6305700290053627e-06,
      "loss": 1.0323,
      "step": 5124
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.20615636997604941,
      "learning_rate": 1.6198540548187036e-06,
      "loss": 1.0257,
      "step": 5125
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.21591003493195407,
      "learning_rate": 1.609173121546681e-06,
      "loss": 0.9616,
      "step": 5126
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.18117760708190556,
      "learning_rate": 1.5985272329936407e-06,
      "loss": 1.0387,
      "step": 5127
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.20671837409164412,
      "learning_rate": 1.5879163929514052e-06,
      "loss": 1.0568,
      "step": 5128
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.18484535519271972,
      "learning_rate": 1.577340605199351e-06,
      "loss": 0.7749,
      "step": 5129
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.25185090990757314,
      "learning_rate": 1.5667998735043321e-06,
      "loss": 0.9978,
      "step": 5130
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2498535997252898,
      "learning_rate": 1.5562942016207338e-06,
      "loss": 1.0466,
      "step": 5131
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.20319604756449913,
      "learning_rate": 1.545823593290463e-06,
      "loss": 0.9608,
      "step": 5132
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.21545565505997535,
      "learning_rate": 1.5353880522429032e-06,
      "loss": 1.1388,
      "step": 5133
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.20218243358456175,
      "learning_rate": 1.5249875821950034e-06,
      "loss": 1.0857,
      "step": 5134
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.20259714189189662,
      "learning_rate": 1.5146221868511668e-06,
      "loss": 1.0833,
      "step": 5135
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.21234201676866954,
      "learning_rate": 1.5042918699033403e-06,
      "loss": 0.8823,
      "step": 5136
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.17409399408200385,
      "learning_rate": 1.4939966350309476e-06,
      "loss": 1.0562,
      "step": 5137
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.20936090746461414,
      "learning_rate": 1.483736485900944e-06,
      "loss": 1.1446,
      "step": 5138
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.18580158504714436,
      "learning_rate": 1.473511426167784e-06,
      "loss": 0.8934,
      "step": 5139
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.19611364925428776,
      "learning_rate": 1.4633214594734102e-06,
      "loss": 1.0437,
      "step": 5140
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.20555682049203797,
      "learning_rate": 1.4531665894472634e-06,
      "loss": 0.8831,
      "step": 5141
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.19569272578495348,
      "learning_rate": 1.4430468197062841e-06,
      "loss": 1.0984,
      "step": 5142
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2174537919450885,
      "learning_rate": 1.4329621538549444e-06,
      "loss": 0.9746,
      "step": 5143
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3504908738243488,
      "learning_rate": 1.4229125954851707e-06,
      "loss": 1.1785,
      "step": 5144
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1725059311379513,
      "learning_rate": 1.4128981481764115e-06,
      "loss": 1.0502,
      "step": 5145
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.21128908103490948,
      "learning_rate": 1.4029188154955908e-06,
      "loss": 0.9665,
      "step": 5146
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.18180482061949782,
      "learning_rate": 1.3929746009971433e-06,
      "loss": 1.0824,
      "step": 5147
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.26379563670469425,
      "learning_rate": 1.3830655082229803e-06,
      "loss": 0.9444,
      "step": 5148
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.21209189253889094,
      "learning_rate": 1.373191540702523e-06,
      "loss": 1.0135,
      "step": 5149
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.22990337121524237,
      "learning_rate": 1.363352701952636e-06,
      "loss": 0.9565,
      "step": 5150
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2363629694413125,
      "learning_rate": 1.353548995477738e-06,
      "loss": 1.0337,
      "step": 5151
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.18984716319936487,
      "learning_rate": 1.3437804247696917e-06,
      "loss": 1.0425,
      "step": 5152
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.27512961439408856,
      "learning_rate": 1.3340469933078359e-06,
      "loss": 1.1547,
      "step": 5153
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2050579361551144,
      "learning_rate": 1.324348704559042e-06,
      "loss": 0.8764,
      "step": 5154
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.20828252792469817,
      "learning_rate": 1.3146855619776134e-06,
      "loss": 0.9665,
      "step": 5155
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.19728745314899043,
      "learning_rate": 1.3050575690053636e-06,
      "loss": 1.2193,
      "step": 5156
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.17773598644041083,
      "learning_rate": 1.2954647290715605e-06,
      "loss": 1.1771,
      "step": 5157
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2080599078437032,
      "learning_rate": 1.2859070455930045e-06,
      "loss": 1.0618,
      "step": 5158
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.19686558792017217,
      "learning_rate": 1.2763845219739056e-06,
      "loss": 0.9634,
      "step": 5159
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.24490850106425285,
      "learning_rate": 1.2668971616060066e-06,
      "loss": 1.0578,
      "step": 5160
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.23818196856518561,
      "learning_rate": 1.257444967868482e-06,
      "loss": 1.0199,
      "step": 5161
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.19451189576727954,
      "learning_rate": 1.2480279441280274e-06,
      "loss": 0.9506,
      "step": 5162
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.20209143205524965,
      "learning_rate": 1.2386460937387822e-06,
      "loss": 1.1086,
      "step": 5163
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.23823655901093946,
      "learning_rate": 1.2292994200423403e-06,
      "loss": 0.8449,
      "step": 5164
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1791662175091192,
      "learning_rate": 1.219987926367816e-06,
      "loss": 1.0014,
      "step": 5165
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1966889375511569,
      "learning_rate": 1.2107116160317344e-06,
      "loss": 0.9366,
      "step": 5166
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.18557807254993552,
      "learning_rate": 1.2014704923381414e-06,
      "loss": 1.1426,
      "step": 5167
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2319200833737467,
      "learning_rate": 1.1922645585785264e-06,
      "loss": 1.0604,
      "step": 5168
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.17905371164987322,
      "learning_rate": 1.183093818031844e-06,
      "loss": 1.1061,
      "step": 5169
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.18217400633338615,
      "learning_rate": 1.1739582739645038e-06,
      "loss": 0.9882,
      "step": 5170
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.20594663770835267,
      "learning_rate": 1.1648579296304253e-06,
      "loss": 1.0628,
      "step": 5171
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.20476541639631612,
      "learning_rate": 1.1557927882709264e-06,
      "loss": 0.9856,
      "step": 5172
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1992428255573377,
      "learning_rate": 1.1467628531148357e-06,
      "loss": 0.9695,
      "step": 5173
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.20783811061297414,
      "learning_rate": 1.1377681273784135e-06,
      "loss": 1.2135,
      "step": 5174
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.17521814993836363,
      "learning_rate": 1.1288086142653864e-06,
      "loss": 0.9539,
      "step": 5175
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1960435156546422,
      "learning_rate": 1.119884316966957e-06,
      "loss": 1.0013,
      "step": 5176
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.21206043931593802,
      "learning_rate": 1.1109952386617605e-06,
      "loss": 1.05,
      "step": 5177
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2304153813500629,
      "learning_rate": 1.1021413825158977e-06,
      "loss": 1.1332,
      "step": 5178
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2084187343521844,
      "learning_rate": 1.0933227516829347e-06,
      "loss": 0.8861,
      "step": 5179
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.21332548654535924,
      "learning_rate": 1.0845393493038592e-06,
      "loss": 1.0758,
      "step": 5180
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.20675189336677666,
      "learning_rate": 1.0757911785071463e-06,
      "loss": 1.0433,
      "step": 5181
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.18320404211025865,
      "learning_rate": 1.0670782424087033e-06,
      "loss": 1.0696,
      "step": 5182
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.18911822510959142,
      "learning_rate": 1.0584005441119038e-06,
      "loss": 1.0711,
      "step": 5183
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2085184971816588,
      "learning_rate": 1.0497580867075418e-06,
      "loss": 1.0083,
      "step": 5184
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.22945124918831655,
      "learning_rate": 1.0411508732738884e-06,
      "loss": 1.015,
      "step": 5185
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.23627896945534907,
      "learning_rate": 1.0325789068766467e-06,
      "loss": 1.1819,
      "step": 5186
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.20879783935929233,
      "learning_rate": 1.0240421905689745e-06,
      "loss": 0.8604,
      "step": 5187
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.20049056033945656,
      "learning_rate": 1.0155407273914619e-06,
      "loss": 1.108,
      "step": 5188
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.24979051581359,
      "learning_rate": 1.0070745203721532e-06,
      "loss": 0.9677,
      "step": 5189
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2070744812463853,
      "learning_rate": 9.986435725265141e-07,
      "loss": 1.1124,
      "step": 5190
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.16463729082511916,
      "learning_rate": 9.902478868574982e-07,
      "loss": 1.0081,
      "step": 5191
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.27941554308168887,
      "learning_rate": 9.818874663554357e-07,
      "loss": 0.9455,
      "step": 5192
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.19880765096553626,
      "learning_rate": 9.735623139981664e-07,
      "loss": 1.0517,
      "step": 5193
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.21994888908004773,
      "learning_rate": 9.65272432750919e-07,
      "loss": 1.1835,
      "step": 5194
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.17645386538080027,
      "learning_rate": 9.570178255663532e-07,
      "loss": 0.924,
      "step": 5195
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.21261102860732464,
      "learning_rate": 9.487984953846174e-07,
      "loss": 1.1245,
      "step": 5196
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.259733562188254,
      "learning_rate": 9.406144451332366e-07,
      "loss": 1.1136,
      "step": 5197
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.18288223955872332,
      "learning_rate": 9.324656777272123e-07,
      "loss": 1.039,
      "step": 5198
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2264964292912454,
      "learning_rate": 9.24352196068945e-07,
      "loss": 1.0793,
      "step": 5199
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1998841263239624,
      "learning_rate": 9.162740030482897e-07,
      "loss": 0.9321,
      "step": 5200
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.18050621258217128,
      "learning_rate": 9.082311015425338e-07,
      "loss": 1.007,
      "step": 5201
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.17820427086284776,
      "learning_rate": 9.002234944163967e-07,
      "loss": 1.0607,
      "step": 5202
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2032053524057955,
      "learning_rate": 8.922511845219971e-07,
      "loss": 1.1313,
      "step": 5203
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.24693207940506684,
      "learning_rate": 8.843141746989081e-07,
      "loss": 1.1578,
      "step": 5204
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.24598401803828573,
      "learning_rate": 8.764124677741236e-07,
      "loss": 0.9217,
      "step": 5205
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.21471198766859625,
      "learning_rate": 8.685460665620704e-07,
      "loss": 1.0152,
      "step": 5206
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.29549203213725056,
      "learning_rate": 8.607149738645848e-07,
      "loss": 0.9464,
      "step": 5207
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.17270422482233638,
      "learning_rate": 8.529191924709468e-07,
      "loss": 0.973,
      "step": 5208
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2150072975072605,
      "learning_rate": 8.451587251578352e-07,
      "loss": 1.0337,
      "step": 5209
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.21129611795256525,
      "learning_rate": 8.374335746893613e-07,
      "loss": 1.0846,
      "step": 5210
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2494845142084077,
      "learning_rate": 8.297437438170797e-07,
      "loss": 1.1601,
      "step": 5211
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.20377413708789216,
      "learning_rate": 8.220892352799215e-07,
      "loss": 1.1254,
      "step": 5212
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.18489661721902806,
      "learning_rate": 8.144700518042836e-07,
      "loss": 1.0493,
      "step": 5213
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.20445392723000855,
      "learning_rate": 8.068861961039287e-07,
      "loss": 1.1612,
      "step": 5214
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2227222884446238,
      "learning_rate": 7.993376708800848e-07,
      "loss": 1.1765,
      "step": 5215
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.19920787769572984,
      "learning_rate": 7.91824478821368e-07,
      "loss": 1.1422,
      "step": 5216
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.20194782900518968,
      "learning_rate": 7.843466226038155e-07,
      "loss": 1.0598,
      "step": 5217
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.21267990518156057,
      "learning_rate": 7.769041048908965e-07,
      "loss": 1.104,
      "step": 5218
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.24405958903911207,
      "learning_rate": 7.694969283334575e-07,
      "loss": 0.885,
      "step": 5219
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1905073994288828,
      "learning_rate": 7.621250955697878e-07,
      "loss": 0.984,
      "step": 5220
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.20423767486465055,
      "learning_rate": 7.547886092255762e-07,
      "loss": 0.9927,
      "step": 5221
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.19974201235615374,
      "learning_rate": 7.474874719139214e-07,
      "loss": 1.0348,
      "step": 5222
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.17705510084179746,
      "learning_rate": 7.402216862353317e-07,
      "loss": 1.1311,
      "step": 5223
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.20321379984682958,
      "learning_rate": 7.329912547777262e-07,
      "loss": 0.9349,
      "step": 5224
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1789475744829409,
      "learning_rate": 7.257961801164337e-07,
      "loss": 1.0585,
      "step": 5225
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.21007061342733133,
      "learning_rate": 7.186364648141819e-07,
      "loss": 1.034,
      "step": 5226
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.22939842867020369,
      "learning_rate": 7.115121114211199e-07,
      "loss": 1.1323,
      "step": 5227
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.18987079829904252,
      "learning_rate": 7.044231224747844e-07,
      "loss": 1.168,
      "step": 5228
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.20669416949282465,
      "learning_rate": 6.973695005001113e-07,
      "loss": 1.0589,
      "step": 5229
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.18169869755615406,
      "learning_rate": 6.903512480094576e-07,
      "loss": 0.9812,
      "step": 5230
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.20236391354829414,
      "learning_rate": 6.833683675025904e-07,
      "loss": 1.1629,
      "step": 5231
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2181664189357876,
      "learning_rate": 6.764208614666423e-07,
      "loss": 0.9455,
      "step": 5232
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.21027811049480444,
      "learning_rate": 6.695087323761784e-07,
      "loss": 1.0343,
      "step": 5233
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.23035317939717523,
      "learning_rate": 6.626319826931404e-07,
      "loss": 0.9138,
      "step": 5234
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2144906106343999,
      "learning_rate": 6.557906148669024e-07,
      "loss": 1.0822,
      "step": 5235
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.18453295361269745,
      "learning_rate": 6.48984631334204e-07,
      "loss": 1.0392,
      "step": 5236
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.213570309916313,
      "learning_rate": 6.422140345191729e-07,
      "loss": 0.9795,
      "step": 5237
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.19139655843550224,
      "learning_rate": 6.3547882683338e-07,
      "loss": 1.1172,
      "step": 5238
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.18133739441858718,
      "learning_rate": 6.287790106757396e-07,
      "loss": 1.0255,
      "step": 5239
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.21638754301574079,
      "learning_rate": 6.221145884326096e-07,
      "loss": 1.0756,
      "step": 5240
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.20352968534102112,
      "learning_rate": 6.154855624776801e-07,
      "loss": 1.1045,
      "step": 5241
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.17023380899092486,
      "learning_rate": 6.088919351721067e-07,
      "loss": 1.0245,
      "step": 5242
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.20219260394937863,
      "learning_rate": 6.023337088643665e-07,
      "loss": 0.9546,
      "step": 5243
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.18886292207550137,
      "learning_rate": 5.958108858903799e-07,
      "loss": 0.972,
      "step": 5244
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.20209469127338156,
      "learning_rate": 5.893234685734439e-07,
      "loss": 1.0527,
      "step": 5245
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.17726596119356436,
      "learning_rate": 5.828714592242212e-07,
      "loss": 0.9963,
      "step": 5246
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.21629653846039087,
      "learning_rate": 5.764548601407849e-07,
      "loss": 1.0674,
      "step": 5247
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.19174879406292233,
      "learning_rate": 5.700736736085843e-07,
      "loss": 1.064,
      "step": 5248
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1734032405061531,
      "learning_rate": 5.637279019004793e-07,
      "loss": 1.058,
      "step": 5249
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.25729346883780924,
      "learning_rate": 5.57417547276684e-07,
      "loss": 1.1847,
      "step": 5250
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.20932718187424823,
      "learning_rate": 5.51142611984834e-07,
      "loss": 0.9691,
      "step": 5251
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.23723534583516057,
      "learning_rate": 5.449030982598968e-07,
      "loss": 1.044,
      "step": 5252
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.18035810199098898,
      "learning_rate": 5.386990083242837e-07,
      "loss": 1.0547,
      "step": 5253
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.21666934037201277,
      "learning_rate": 5.325303443877272e-07,
      "loss": 1.0256,
      "step": 5254
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1820057982939214,
      "learning_rate": 5.26397108647414e-07,
      "loss": 0.9804,
      "step": 5255
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2696070862209236,
      "learning_rate": 5.202993032878301e-07,
      "loss": 1.0481,
      "step": 5256
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.20315153057923846,
      "learning_rate": 5.142369304809158e-07,
      "loss": 0.9977,
      "step": 5257
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1827062238792182,
      "learning_rate": 5.082099923859551e-07,
      "loss": 1.0531,
      "step": 5258
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.17801525217857772,
      "learning_rate": 5.022184911495864e-07,
      "loss": 1.0341,
      "step": 5259
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1751624973016687,
      "learning_rate": 4.962624289059026e-07,
      "loss": 0.9883,
      "step": 5260
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.20572284702742738,
      "learning_rate": 4.903418077762844e-07,
      "loss": 1.2216,
      "step": 5261
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2352067221076941,
      "learning_rate": 4.844566298695563e-07,
      "loss": 1.0276,
      "step": 5262
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.25587554510750377,
      "learning_rate": 4.786068972818747e-07,
      "loss": 1.0421,
      "step": 5263
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2764012874385684,
      "learning_rate": 4.7279261209680627e-07,
      "loss": 0.9778,
      "step": 5264
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.19437073798835294,
      "learning_rate": 4.670137763852611e-07,
      "loss": 0.9406,
      "step": 5265
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2069999987749179,
      "learning_rate": 4.6127039220555943e-07,
      "loss": 0.9377,
      "step": 5266
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1874919400185965,
      "learning_rate": 4.555624616033427e-07,
      "loss": 1.1094,
      "step": 5267
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.18854023712432008,
      "learning_rate": 4.498899866116735e-07,
      "loss": 1.0971,
      "step": 5268
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.27826028634725314,
      "learning_rate": 4.442529692509689e-07,
      "loss": 1.0409,
      "step": 5269
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.20333514043523374,
      "learning_rate": 4.386514115290119e-07,
      "loss": 1.0392,
      "step": 5270
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.23391483988823045,
      "learning_rate": 4.3308531544095087e-07,
      "loss": 1.0333,
      "step": 5271
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.18652869600123717,
      "learning_rate": 4.2755468296931113e-07,
      "loss": 1.043,
      "step": 5272
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.18787014960625395,
      "learning_rate": 4.220595160840057e-07,
      "loss": 0.9945,
      "step": 5273
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1770727821622203,
      "learning_rate": 4.16599816742258e-07,
      "loss": 1.0377,
      "step": 5274
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.24476268475518242,
      "learning_rate": 4.111755868887346e-07,
      "loss": 1.0616,
      "step": 5275
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2033324990740237,
      "learning_rate": 4.0578682845541226e-07,
      "loss": 0.9091,
      "step": 5276
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2012633128721873,
      "learning_rate": 4.0043354336165573e-07,
      "loss": 0.967,
      "step": 5277
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.25742281037823284,
      "learning_rate": 3.951157335141953e-07,
      "loss": 1.0529,
      "step": 5278
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.20379250312210442,
      "learning_rate": 3.8983340080710476e-07,
      "loss": 0.8732,
      "step": 5279
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.24709769508210133,
      "learning_rate": 3.845865471218679e-07,
      "loss": 1.0104,
      "step": 5280
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.208456016357541,
      "learning_rate": 3.7937517432726777e-07,
      "loss": 1.0445,
      "step": 5281
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.26248857151314325,
      "learning_rate": 3.7419928427951944e-07,
      "loss": 0.9831,
      "step": 5282
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.17836507024340054,
      "learning_rate": 3.6905887882213717e-07,
      "loss": 1.1654,
      "step": 5283
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1914726756697256,
      "learning_rate": 3.639539597860342e-07,
      "loss": 1.0253,
      "step": 5284
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.17972732181122045,
      "learning_rate": 3.588845289894782e-07,
      "loss": 1.0013,
      "step": 5285
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.23305989317901013,
      "learning_rate": 3.5385058823809156e-07,
      "loss": 0.9084,
      "step": 5286
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.17903796664768,
      "learning_rate": 3.488521393248401e-07,
      "loss": 1.0525,
      "step": 5287
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.21529135638767294,
      "learning_rate": 3.438891840300995e-07,
      "loss": 1.0651,
      "step": 5288
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.16688820954718936,
      "learning_rate": 3.389617241215337e-07,
      "loss": 0.9933,
      "step": 5289
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.19716048768984537,
      "learning_rate": 3.3406976135421653e-07,
      "loss": 0.929,
      "step": 5290
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2242183856259502,
      "learning_rate": 3.292132974705653e-07,
      "loss": 1.1035,
      "step": 5291
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19759461887269858,
      "learning_rate": 3.2439233420032965e-07,
      "loss": 0.9792,
      "step": 5292
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.18579671096431907,
      "learning_rate": 3.196068732606583e-07,
      "loss": 1.151,
      "step": 5293
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.18275406933449032,
      "learning_rate": 3.1485691635601e-07,
      "loss": 0.9503,
      "step": 5294
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.21940703547917434,
      "learning_rate": 3.1014246517823145e-07,
      "loss": 0.9517,
      "step": 5295
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2027370208306989,
      "learning_rate": 3.054635214065016e-07,
      "loss": 1.0966,
      "step": 5296
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1974822763826935,
      "learning_rate": 3.0082008670736517e-07,
      "loss": 0.9784,
      "step": 5297
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.20836513035988374,
      "learning_rate": 2.962121627347214e-07,
      "loss": 0.9669,
      "step": 5298
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19729310751885298,
      "learning_rate": 2.916397511298019e-07,
      "loss": 1.1684,
      "step": 5299
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2577557127256229,
      "learning_rate": 2.8710285352120393e-07,
      "loss": 1.0029,
      "step": 5300
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.18588044082192423,
      "learning_rate": 2.826014715248904e-07,
      "loss": 1.0847,
      "step": 5301
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2321889084992382,
      "learning_rate": 2.781356067441343e-07,
      "loss": 0.911,
      "step": 5302
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2052315070772051,
      "learning_rate": 2.7370526076959666e-07,
      "loss": 1.179,
      "step": 5303
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19390060811589638,
      "learning_rate": 2.693104351792708e-07,
      "loss": 0.9248,
      "step": 5304
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.18283022146630531,
      "learning_rate": 2.649511315384934e-07,
      "loss": 0.9594,
      "step": 5305
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2236989455156508,
      "learning_rate": 2.6062735139996684e-07,
      "loss": 1.0244,
      "step": 5306
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1906556817227987,
      "learning_rate": 2.5633909630371487e-07,
      "loss": 1.0366,
      "step": 5307
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.17033156283534787,
      "learning_rate": 2.520863677771268e-07,
      "loss": 0.8434,
      "step": 5308
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19068049048517122,
      "learning_rate": 2.478691673349465e-07,
      "loss": 1.1267,
      "step": 5309
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19117385421439617,
      "learning_rate": 2.4368749647925013e-07,
      "loss": 1.2162,
      "step": 5310
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.20412778455163205,
      "learning_rate": 2.395413566994353e-07,
      "loss": 1.0489,
      "step": 5311
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.17904392899296162,
      "learning_rate": 2.3543074947229848e-07,
      "loss": 0.9667,
      "step": 5312
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19626551646728516,
      "learning_rate": 2.313556762619351e-07,
      "loss": 0.9385,
      "step": 5313
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1976750034985333,
      "learning_rate": 2.273161385197953e-07,
      "loss": 1.1432,
      "step": 5314
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.20388158876383816,
      "learning_rate": 2.2331213768468363e-07,
      "loss": 1.0303,
      "step": 5315
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.20126715947988313,
      "learning_rate": 2.1934367518274823e-07,
      "loss": 1.0126,
      "step": 5316
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19730881848424858,
      "learning_rate": 2.1541075242745846e-07,
      "loss": 1.1663,
      "step": 5317
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19564961746819995,
      "learning_rate": 2.1151337081963817e-07,
      "loss": 1.0002,
      "step": 5318
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19856094919036632,
      "learning_rate": 2.0765153174745478e-07,
      "loss": 1.1812,
      "step": 5319
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19122016107258286,
      "learning_rate": 2.0382523658641906e-07,
      "loss": 0.8479,
      "step": 5320
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.18328329146524822,
      "learning_rate": 2.000344866993631e-07,
      "loss": 0.9685,
      "step": 5321
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1836792906884857,
      "learning_rate": 1.962792834364846e-07,
      "loss": 1.0034,
      "step": 5322
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.189639098257973,
      "learning_rate": 1.925596281353026e-07,
      "loss": 1.0596,
      "step": 5323
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.22701728508374155,
      "learning_rate": 1.888755221206795e-07,
      "loss": 0.9038,
      "step": 5324
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19135774152828255,
      "learning_rate": 1.852269667048101e-07,
      "loss": 1.0188,
      "step": 5325
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19130368766045497,
      "learning_rate": 1.8161396318723266e-07,
      "loss": 0.9203,
      "step": 5326
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19293963509468587,
      "learning_rate": 1.7803651285482893e-07,
      "loss": 1.0571,
      "step": 5327
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.16706914556041969,
      "learning_rate": 1.744946169818018e-07,
      "loss": 1.0199,
      "step": 5328
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.180069723534374,
      "learning_rate": 1.7098827682970886e-07,
      "loss": 1.0049,
      "step": 5329
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.20453307299190576,
      "learning_rate": 1.6751749364743995e-07,
      "loss": 1.1368,
      "step": 5330
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2132891937228494,
      "learning_rate": 1.6408226867118403e-07,
      "loss": 0.9258,
      "step": 5331
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.18219389398507924,
      "learning_rate": 1.6068260312452898e-07,
      "loss": 1.0041,
      "step": 5332
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1847398823169709,
      "learning_rate": 1.5731849821833954e-07,
      "loss": 1.2405,
      "step": 5333
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19318056818930043,
      "learning_rate": 1.5398995515085724e-07,
      "loss": 0.9438,
      "step": 5334
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.20747769947457811,
      "learning_rate": 1.506969751076226e-07,
      "loss": 1.166,
      "step": 5335
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2312045230412564,
      "learning_rate": 1.4743955926154185e-07,
      "loss": 1.0032,
      "step": 5336
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19716701763584707,
      "learning_rate": 1.4421770877280915e-07,
      "loss": 0.8745,
      "step": 5337
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.18039135159128455,
      "learning_rate": 1.4103142478900655e-07,
      "loss": 1.0481,
      "step": 5338
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19738048206556102,
      "learning_rate": 1.378807084450151e-07,
      "loss": 1.0252,
      "step": 5339
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2865145025856236,
      "learning_rate": 1.3476556086303715e-07,
      "loss": 0.9149,
      "step": 5340
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.21540404518590653,
      "learning_rate": 1.3168598315264068e-07,
      "loss": 1.0452,
      "step": 5341
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.20396649787786525,
      "learning_rate": 1.2864197641070387e-07,
      "loss": 0.9719,
      "step": 5342
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.20197287494190122,
      "learning_rate": 1.2563354172142606e-07,
      "loss": 0.9541,
      "step": 5343
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.23177163205378654,
      "learning_rate": 1.2266068015635013e-07,
      "loss": 1.0625,
      "step": 5344
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19798285283384662,
      "learning_rate": 1.1972339277435129e-07,
      "loss": 0.9978,
      "step": 5345
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.19081982785461826,
      "learning_rate": 1.1682168062163713e-07,
      "loss": 1.2753,
      "step": 5346
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.20206921435021655,
      "learning_rate": 1.1395554473171422e-07,
      "loss": 1.0178,
      "step": 5347
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.27245281702613705,
      "learning_rate": 1.1112498612545486e-07,
      "loss": 1.0388,
      "step": 5348
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.19093129604911852,
      "learning_rate": 1.0833000581103036e-07,
      "loss": 0.9957,
      "step": 5349
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.21980310971000788,
      "learning_rate": 1.0557060478396663e-07,
      "loss": 1.1269,
      "step": 5350
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.23962536772876233,
      "learning_rate": 1.0284678402711079e-07,
      "loss": 0.916,
      "step": 5351
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.20526838743796094,
      "learning_rate": 1.0015854451060901e-07,
      "loss": 0.9994,
      "step": 5352
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.18223443069326922,
      "learning_rate": 9.750588719196207e-08,
      "loss": 1.0887,
      "step": 5353
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.19160346035056464,
      "learning_rate": 9.488881301599195e-08,
      "loss": 1.0999,
      "step": 5354
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.21524362311356224,
      "learning_rate": 9.230732291485301e-08,
      "loss": 1.0165,
      "step": 5355
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1964306031699503,
      "learning_rate": 8.976141780799862e-08,
      "loss": 0.9627,
      "step": 5356
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.19386971868076142,
      "learning_rate": 8.725109860224789e-08,
      "loss": 0.9363,
      "step": 5357
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.24271072440475833,
      "learning_rate": 8.477636619170781e-08,
      "loss": 0.9118,
      "step": 5358
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2352706595193733,
      "learning_rate": 8.233722145782885e-08,
      "loss": 0.9661,
      "step": 5359
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.18890855196966905,
      "learning_rate": 7.993366526938273e-08,
      "loss": 1.0258,
      "step": 5360
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.21639385929002078,
      "learning_rate": 7.756569848246242e-08,
      "loss": 1.1137,
      "step": 5361
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.22833359676394602,
      "learning_rate": 7.523332194049326e-08,
      "loss": 1.1786,
      "step": 5362
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.22167472114823275,
      "learning_rate": 7.293653647421073e-08,
      "loss": 0.9469,
      "step": 5363
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2047368163020116,
      "learning_rate": 7.067534290169376e-08,
      "loss": 1.2272,
      "step": 5364
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.24106074713351883,
      "learning_rate": 6.844974202832033e-08,
      "loss": 1.1718,
      "step": 5365
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1831197552140803,
      "learning_rate": 6.625973464680079e-08,
      "loss": 1.008,
      "step": 5366
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.23384887098095805,
      "learning_rate": 6.410532153717786e-08,
      "loss": 0.9069,
      "step": 5367
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.22182722407381672,
      "learning_rate": 6.198650346680435e-08,
      "loss": 1.0417,
      "step": 5368
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1869414703206219,
      "learning_rate": 5.990328119034328e-08,
      "loss": 1.006,
      "step": 5369
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2172547622403695,
      "learning_rate": 5.7855655449823297e-08,
      "loss": 1.083,
      "step": 5370
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.220030305894611,
      "learning_rate": 5.584362697453882e-08,
      "loss": 1.2049,
      "step": 5371
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2010404458939176,
      "learning_rate": 5.3867196481149904e-08,
      "loss": 0.9382,
      "step": 5372
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1978095800388953,
      "learning_rate": 5.192636467360457e-08,
      "loss": 0.9617,
      "step": 5373
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.19237364391106637,
      "learning_rate": 5.002113224319427e-08,
      "loss": 0.967,
      "step": 5374
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1814900563026185,
      "learning_rate": 4.8151499868520634e-08,
      "loss": 0.9959,
      "step": 5375
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.24453854477370904,
      "learning_rate": 4.631746821550653e-08,
      "loss": 1.1157,
      "step": 5376
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.19549223396290644,
      "learning_rate": 4.451903793740719e-08,
      "loss": 1.2614,
      "step": 5377
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1945724844164026,
      "learning_rate": 4.275620967475469e-08,
      "loss": 1.068,
      "step": 5378
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1956548570855024,
      "learning_rate": 4.102898405545785e-08,
      "loss": 1.0163,
      "step": 5379
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.19443257813645984,
      "learning_rate": 3.933736169471347e-08,
      "loss": 1.054,
      "step": 5380
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1966671085235745,
      "learning_rate": 3.7681343195028474e-08,
      "loss": 1.027,
      "step": 5381
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.22690528299728405,
      "learning_rate": 3.606092914625325e-08,
      "loss": 1.102,
      "step": 5382
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.18044527411038044,
      "learning_rate": 3.447612012554835e-08,
      "loss": 1.0306,
      "step": 5383
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.16041434980864355,
      "learning_rate": 3.2926916697373357e-08,
      "loss": 0.997,
      "step": 5384
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.18509021748597368,
      "learning_rate": 3.141331941354242e-08,
      "loss": 1.1331,
      "step": 5385
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.18970505459218193,
      "learning_rate": 2.9935328813146536e-08,
      "loss": 1.1642,
      "step": 5386
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.19958857628463506,
      "learning_rate": 2.8492945422620155e-08,
      "loss": 0.9752,
      "step": 5387
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.20339627247952374,
      "learning_rate": 2.7086169755718983e-08,
      "loss": 1.226,
      "step": 5388
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2115265835991377,
      "learning_rate": 2.5715002313497772e-08,
      "loss": 1.1262,
      "step": 5389
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.25867098889405316,
      "learning_rate": 2.437944358433253e-08,
      "loss": 1.0598,
      "step": 5390
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.17771126778436572,
      "learning_rate": 2.307949404394272e-08,
      "loss": 1.1049,
      "step": 5391
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1917454406402132,
      "learning_rate": 2.1815154155313543e-08,
      "loss": 1.169,
      "step": 5392
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.16586432403443507,
      "learning_rate": 2.0586424368795875e-08,
      "loss": 1.182,
      "step": 5393
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.21953463756527536,
      "learning_rate": 1.9393305122028528e-08,
      "loss": 1.0073,
      "step": 5394
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1964093628291568,
      "learning_rate": 1.8235796839982665e-08,
      "loss": 0.99,
      "step": 5395
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1883182517687202,
      "learning_rate": 1.7113899934939616e-08,
      "loss": 1.0305,
      "step": 5396
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2014239396275436,
      "learning_rate": 1.6027614806479742e-08,
      "loss": 1.0649,
      "step": 5397
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1810901207539633,
      "learning_rate": 1.497694184151577e-08,
      "loss": 1.084,
      "step": 5398
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.215878080900582,
      "learning_rate": 1.3961881414292778e-08,
      "loss": 0.9439,
      "step": 5399
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.16581059310132049,
      "learning_rate": 1.2982433886354894e-08,
      "loss": 0.9964,
      "step": 5400
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.18963993553272432,
      "learning_rate": 1.2038599606534195e-08,
      "loss": 1.056,
      "step": 5401
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.18358889831605774,
      "learning_rate": 1.1130378911028416e-08,
      "loss": 1.0794,
      "step": 5402
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.20195937182750637,
      "learning_rate": 1.0257772123312137e-08,
      "loss": 1.1297,
      "step": 5403
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.20145224636107967,
      "learning_rate": 9.420779554192294e-09,
      "loss": 1.1045,
      "step": 5404
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2147098427851325,
      "learning_rate": 8.619401501797075e-09,
      "loss": 0.9602,
      "step": 5405
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.20417724104824472,
      "learning_rate": 7.853638251553718e-09,
      "loss": 0.8955,
      "step": 5406
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.24027319442280018,
      "learning_rate": 7.123490076210715e-09,
      "loss": 1.183,
      "step": 5407
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.16861398262254687,
      "learning_rate": 6.428957235826705e-09,
      "loss": 0.8369,
      "step": 5408
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.17684642499881187,
      "learning_rate": 5.770039977792685e-09,
      "loss": 1.054,
      "step": 5409
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2174923453334809,
      "learning_rate": 5.146738536787599e-09,
      "loss": 0.998,
      "step": 5410
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2124685801725982,
      "learning_rate": 4.559053134822744e-09,
      "loss": 0.9469,
      "step": 5411
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2036391556430673,
      "learning_rate": 4.0069839812195696e-09,
      "loss": 1.0013,
      "step": 5412
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.21333060813538304,
      "learning_rate": 3.4905312726207783e-09,
      "loss": 1.0092,
      "step": 5413
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.19188791215283835,
      "learning_rate": 3.0096951929570184e-09,
      "loss": 1.0479,
      "step": 5414
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.22407406013119954,
      "learning_rate": 2.5644759135134976e-09,
      "loss": 0.9628,
      "step": 5415
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.18053981084128395,
      "learning_rate": 2.154873592852269e-09,
      "loss": 1.0499,
      "step": 5416
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2451452125008284,
      "learning_rate": 1.7808883768677398e-09,
      "loss": 1.1062,
      "step": 5417
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.19834103143593856,
      "learning_rate": 1.4425203987755709e-09,
      "loss": 1.0032,
      "step": 5418
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.18906386649672507,
      "learning_rate": 1.1397697790793694e-09,
      "loss": 0.9748,
      "step": 5419
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2233100958692608,
      "learning_rate": 8.726366256261998e-10,
      "loss": 1.1204,
      "step": 5420
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.19073473816203593,
      "learning_rate": 6.411210335621753e-10,
      "loss": 0.9806,
      "step": 5421
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2077794600512692,
      "learning_rate": 4.4522308533245793e-10,
      "loss": 1.0751,
      "step": 5422
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.22167559228844644,
      "learning_rate": 2.849428507367691e-10,
      "loss": 0.9851,
      "step": 5423
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.19377062546021537,
      "learning_rate": 1.6028038684057222e-10,
      "loss": 0.8334,
      "step": 5424
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.22845523052535235,
      "learning_rate": 7.123573805278838e-11,
      "loss": 1.0112,
      "step": 5425
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.23344047004316082,
      "learning_rate": 1.7808936103591577e-11,
      "loss": 0.896,
      "step": 5426
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.18863184516707757,
      "learning_rate": 0.0,
      "loss": 1.0993,
      "step": 5427
    },
    {
      "epoch": 1.0,
      "step": 5427,
      "total_flos": 4808138970562560.0,
      "train_loss": 1.0867616527313815,
      "train_runtime": 28984.4948,
      "train_samples_per_second": 23.965,
      "train_steps_per_second": 0.187
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 5427,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "total_flos": 4808138970562560.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}