{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 7177,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00013934369121438027,
      "grad_norm": 68.35508728027344,
      "learning_rate": 0.0,
      "loss": 0.741943359375,
      "step": 1
    },
    {
      "epoch": 0.00027868738242876054,
      "grad_norm": 84.05374908447266,
      "learning_rate": 7.8125e-08,
      "loss": 0.7530517578125,
      "step": 2
    },
    {
      "epoch": 0.0004180310736431408,
      "grad_norm": 80.69912719726562,
      "learning_rate": 1.5625e-07,
      "loss": 0.74884033203125,
      "step": 3
    },
    {
      "epoch": 0.0005573747648575211,
      "grad_norm": 82.28425598144531,
      "learning_rate": 2.3437500000000003e-07,
      "loss": 0.74957275390625,
      "step": 4
    },
    {
      "epoch": 0.0006967184560719013,
      "grad_norm": 86.44834899902344,
      "learning_rate": 3.125e-07,
      "loss": 0.7557373046875,
      "step": 5
    },
    {
      "epoch": 0.0008360621472862816,
      "grad_norm": 80.1391830444336,
      "learning_rate": 3.90625e-07,
      "loss": 0.74884033203125,
      "step": 6
    },
    {
      "epoch": 0.0009754058385006619,
      "grad_norm": 86.425537109375,
      "learning_rate": 4.6875000000000006e-07,
      "loss": 0.7528076171875,
      "step": 7
    },
    {
      "epoch": 0.0011147495297150422,
      "grad_norm": 86.5018310546875,
      "learning_rate": 5.468750000000001e-07,
      "loss": 0.754638671875,
      "step": 8
    },
    {
      "epoch": 0.0012540932209294225,
      "grad_norm": 84.22899627685547,
      "learning_rate": 6.25e-07,
      "loss": 0.7506103515625,
      "step": 9
    },
    {
      "epoch": 0.0013934369121438026,
      "grad_norm": 84.8850326538086,
      "learning_rate": 7.03125e-07,
      "loss": 0.75335693359375,
      "step": 10
    },
    {
      "epoch": 0.001532780603358183,
      "grad_norm": 83.78080749511719,
      "learning_rate": 7.8125e-07,
      "loss": 0.75177001953125,
      "step": 11
    },
    {
      "epoch": 0.0016721242945725633,
      "grad_norm": 81.85120391845703,
      "learning_rate": 8.59375e-07,
      "loss": 0.74407958984375,
      "step": 12
    },
    {
      "epoch": 0.0018114679857869436,
      "grad_norm": 82.96898651123047,
      "learning_rate": 9.375000000000001e-07,
      "loss": 0.74395751953125,
      "step": 13
    },
    {
      "epoch": 0.0019508116770013237,
      "grad_norm": 85.06578063964844,
      "learning_rate": 1.0156250000000001e-06,
      "loss": 0.7449951171875,
      "step": 14
    },
    {
      "epoch": 0.0020901553682157042,
      "grad_norm": 80.7475357055664,
      "learning_rate": 1.0937500000000001e-06,
      "loss": 0.740966796875,
      "step": 15
    },
    {
      "epoch": 0.0022294990594300844,
      "grad_norm": 80.70791625976562,
      "learning_rate": 1.1718750000000001e-06,
      "loss": 0.726318359375,
      "step": 16
    },
    {
      "epoch": 0.0023688427506444645,
      "grad_norm": 78.53620147705078,
      "learning_rate": 1.25e-06,
      "loss": 0.7271728515625,
      "step": 17
    },
    {
      "epoch": 0.002508186441858845,
      "grad_norm": 81.7895278930664,
      "learning_rate": 1.328125e-06,
      "loss": 0.72772216796875,
      "step": 18
    },
    {
      "epoch": 0.002647530133073225,
      "grad_norm": 84.47947692871094,
      "learning_rate": 1.40625e-06,
      "loss": 0.725341796875,
      "step": 19
    },
    {
      "epoch": 0.0027868738242876052,
      "grad_norm": 73.71009826660156,
      "learning_rate": 1.484375e-06,
      "loss": 0.7186279296875,
      "step": 20
    },
    {
      "epoch": 0.0029262175155019858,
      "grad_norm": 78.1681900024414,
      "learning_rate": 1.5625e-06,
      "loss": 0.71649169921875,
      "step": 21
    },
    {
      "epoch": 0.003065561206716366,
      "grad_norm": 75.11559295654297,
      "learning_rate": 1.640625e-06,
      "loss": 0.66864013671875,
      "step": 22
    },
    {
      "epoch": 0.003204904897930746,
      "grad_norm": 76.30701446533203,
      "learning_rate": 1.71875e-06,
      "loss": 0.66497802734375,
      "step": 23
    },
    {
      "epoch": 0.0033442485891451265,
      "grad_norm": 71.80799102783203,
      "learning_rate": 1.796875e-06,
      "loss": 0.6649169921875,
      "step": 24
    },
    {
      "epoch": 0.0034835922803595066,
      "grad_norm": 70.86400604248047,
      "learning_rate": 1.8750000000000003e-06,
      "loss": 0.66485595703125,
      "step": 25
    },
    {
      "epoch": 0.003622935971573887,
      "grad_norm": 69.09696197509766,
      "learning_rate": 1.953125e-06,
      "loss": 0.6510009765625,
      "step": 26
    },
    {
      "epoch": 0.0037622796627882673,
      "grad_norm": 69.65038299560547,
      "learning_rate": 2.0312500000000002e-06,
      "loss": 0.64556884765625,
      "step": 27
    },
    {
      "epoch": 0.0039016233540026474,
      "grad_norm": 74.79225158691406,
      "learning_rate": 2.109375e-06,
      "loss": 0.6395263671875,
      "step": 28
    },
    {
      "epoch": 0.0040409670452170275,
      "grad_norm": 69.77416229248047,
      "learning_rate": 2.1875000000000002e-06,
      "loss": 0.63531494140625,
      "step": 29
    },
    {
      "epoch": 0.0041803107364314085,
      "grad_norm": 61.20261001586914,
      "learning_rate": 2.265625e-06,
      "loss": 0.54632568359375,
      "step": 30
    },
    {
      "epoch": 0.004319654427645789,
      "grad_norm": 59.32276916503906,
      "learning_rate": 2.3437500000000002e-06,
      "loss": 0.53826904296875,
      "step": 31
    },
    {
      "epoch": 0.004458998118860169,
      "grad_norm": 61.9091911315918,
      "learning_rate": 2.421875e-06,
      "loss": 0.53729248046875,
      "step": 32
    },
    {
      "epoch": 0.004598341810074549,
      "grad_norm": 59.664695739746094,
      "learning_rate": 2.5e-06,
      "loss": 0.52093505859375,
      "step": 33
    },
    {
      "epoch": 0.004737685501288929,
      "grad_norm": 51.16754150390625,
      "learning_rate": 2.5781250000000004e-06,
      "loss": 0.53924560546875,
      "step": 34
    },
    {
      "epoch": 0.004877029192503309,
      "grad_norm": 53.8609619140625,
      "learning_rate": 2.65625e-06,
      "loss": 0.521240234375,
      "step": 35
    },
    {
      "epoch": 0.00501637288371769,
      "grad_norm": 53.472801208496094,
      "learning_rate": 2.7343750000000004e-06,
      "loss": 0.496063232421875,
      "step": 36
    },
    {
      "epoch": 0.00515571657493207,
      "grad_norm": 47.95134353637695,
      "learning_rate": 2.8125e-06,
      "loss": 0.498260498046875,
      "step": 37
    },
    {
      "epoch": 0.00529506026614645,
      "grad_norm": 47.91554260253906,
      "learning_rate": 2.8906250000000004e-06,
      "loss": 0.484832763671875,
      "step": 38
    },
    {
      "epoch": 0.00543440395736083,
      "grad_norm": 51.750431060791016,
      "learning_rate": 2.96875e-06,
      "loss": 0.448883056640625,
      "step": 39
    },
    {
      "epoch": 0.0055737476485752105,
      "grad_norm": 46.16836166381836,
      "learning_rate": 3.0468750000000004e-06,
      "loss": 0.428558349609375,
      "step": 40
    },
    {
      "epoch": 0.005713091339789591,
      "grad_norm": 31.973407745361328,
      "learning_rate": 3.125e-06,
      "loss": 0.42529296875,
      "step": 41
    },
    {
      "epoch": 0.0058524350310039715,
      "grad_norm": 30.66975975036621,
      "learning_rate": 3.2031250000000004e-06,
      "loss": 0.397003173828125,
      "step": 42
    },
    {
      "epoch": 0.005991778722218352,
      "grad_norm": 24.170297622680664,
      "learning_rate": 3.28125e-06,
      "loss": 0.4156494140625,
      "step": 43
    },
    {
      "epoch": 0.006131122413432732,
      "grad_norm": 29.654438018798828,
      "learning_rate": 3.3593750000000003e-06,
      "loss": 0.369110107421875,
      "step": 44
    },
    {
      "epoch": 0.006270466104647112,
      "grad_norm": 19.66484260559082,
      "learning_rate": 3.4375e-06,
      "loss": 0.42010498046875,
      "step": 45
    },
    {
      "epoch": 0.006409809795861492,
      "grad_norm": 26.667516708374023,
      "learning_rate": 3.5156250000000003e-06,
      "loss": 0.364288330078125,
      "step": 46
    },
    {
      "epoch": 0.006549153487075873,
      "grad_norm": 15.48472785949707,
      "learning_rate": 3.59375e-06,
      "loss": 0.424407958984375,
      "step": 47
    },
    {
      "epoch": 0.006688497178290253,
      "grad_norm": 19.41010856628418,
      "learning_rate": 3.6718750000000003e-06,
      "loss": 0.380035400390625,
      "step": 48
    },
    {
      "epoch": 0.006827840869504633,
      "grad_norm": 24.582704544067383,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.332916259765625,
      "step": 49
    },
    {
      "epoch": 0.006967184560719013,
      "grad_norm": 14.221268653869629,
      "learning_rate": 3.828125000000001e-06,
      "loss": 0.38800048828125,
      "step": 50
    },
    {
      "epoch": 0.007106528251933393,
      "grad_norm": 17.17940330505371,
      "learning_rate": 3.90625e-06,
      "loss": 0.3555755615234375,
      "step": 51
    },
    {
      "epoch": 0.007245871943147774,
      "grad_norm": 12.101567268371582,
      "learning_rate": 3.984375e-06,
      "loss": 0.3737945556640625,
      "step": 52
    },
    {
      "epoch": 0.0073852156343621545,
      "grad_norm": 16.21722412109375,
      "learning_rate": 4.0625000000000005e-06,
      "loss": 0.318939208984375,
      "step": 53
    },
    {
      "epoch": 0.007524559325576535,
      "grad_norm": 9.64992904663086,
      "learning_rate": 4.140625000000001e-06,
      "loss": 0.364654541015625,
      "step": 54
    },
    {
      "epoch": 0.007663903016790915,
      "grad_norm": 11.83438491821289,
      "learning_rate": 4.21875e-06,
      "loss": 0.32879638671875,
      "step": 55
    },
    {
      "epoch": 0.007803246708005295,
      "grad_norm": 2.6852619647979736,
      "learning_rate": 4.296875e-06,
      "loss": 0.403411865234375,
      "step": 56
    },
    {
      "epoch": 0.007942590399219676,
      "grad_norm": 4.491562843322754,
      "learning_rate": 4.3750000000000005e-06,
      "loss": 0.375762939453125,
      "step": 57
    },
    {
      "epoch": 0.008081934090434055,
      "grad_norm": 10.005256652832031,
      "learning_rate": 4.453125000000001e-06,
      "loss": 0.3044281005859375,
      "step": 58
    },
    {
      "epoch": 0.008221277781648436,
      "grad_norm": 8.927207946777344,
      "learning_rate": 4.53125e-06,
      "loss": 0.302581787109375,
      "step": 59
    },
    {
      "epoch": 0.008360621472862817,
      "grad_norm": 4.227772235870361,
      "learning_rate": 4.609375e-06,
      "loss": 0.339599609375,
      "step": 60
    },
    {
      "epoch": 0.008499965164077196,
      "grad_norm": 2.812326192855835,
      "learning_rate": 4.6875000000000004e-06,
      "loss": 0.3450164794921875,
      "step": 61
    },
    {
      "epoch": 0.008639308855291577,
      "grad_norm": 3.775153875350952,
      "learning_rate": 4.765625000000001e-06,
      "loss": 0.3248138427734375,
      "step": 62
    },
    {
      "epoch": 0.008778652546505956,
      "grad_norm": 2.7568106651306152,
      "learning_rate": 4.84375e-06,
      "loss": 0.379791259765625,
      "step": 63
    },
    {
      "epoch": 0.008917996237720337,
      "grad_norm": 5.9498748779296875,
      "learning_rate": 4.921875e-06,
      "loss": 0.3110198974609375,
      "step": 64
    },
    {
      "epoch": 0.009057339928934717,
      "grad_norm": 4.063409805297852,
      "learning_rate": 5e-06,
      "loss": 0.3192901611328125,
      "step": 65
    },
    {
      "epoch": 0.009196683620149098,
      "grad_norm": 7.123007297515869,
      "learning_rate": 5.078125000000001e-06,
      "loss": 0.258636474609375,
      "step": 66
    },
    {
      "epoch": 0.009336027311363479,
      "grad_norm": 8.098702430725098,
      "learning_rate": 5.156250000000001e-06,
      "loss": 0.225128173828125,
      "step": 67
    },
    {
      "epoch": 0.009475371002577858,
      "grad_norm": 7.1579155921936035,
      "learning_rate": 5.234375e-06,
      "loss": 0.216796875,
      "step": 68
    },
    {
      "epoch": 0.009614714693792239,
      "grad_norm": 7.160254955291748,
      "learning_rate": 5.3125e-06,
      "loss": 0.32779693603515625,
      "step": 69
    },
    {
      "epoch": 0.009754058385006618,
      "grad_norm": 6.66569185256958,
      "learning_rate": 5.390625000000001e-06,
      "loss": 0.231903076171875,
      "step": 70
    },
    {
      "epoch": 0.009893402076220999,
      "grad_norm": 5.218270778656006,
      "learning_rate": 5.468750000000001e-06,
      "loss": 0.3233489990234375,
      "step": 71
    },
    {
      "epoch": 0.01003274576743538,
      "grad_norm": 7.696992874145508,
      "learning_rate": 5.546875e-06,
      "loss": 0.22046661376953125,
      "step": 72
    },
    {
      "epoch": 0.01017208945864976,
      "grad_norm": 10.82709789276123,
      "learning_rate": 5.625e-06,
      "loss": 0.3422698974609375,
      "step": 73
    },
    {
      "epoch": 0.01031143314986414,
      "grad_norm": 8.295283317565918,
      "learning_rate": 5.7031250000000006e-06,
      "loss": 0.22650909423828125,
      "step": 74
    },
    {
      "epoch": 0.01045077684107852,
      "grad_norm": 20.828475952148438,
      "learning_rate": 5.781250000000001e-06,
      "loss": 0.196014404296875,
      "step": 75
    },
    {
      "epoch": 0.0105901205322929,
      "grad_norm": 15.837902069091797,
      "learning_rate": 5.859375e-06,
      "loss": 0.3138885498046875,
      "step": 76
    },
    {
      "epoch": 0.010729464223507281,
      "grad_norm": 8.306597709655762,
      "learning_rate": 5.9375e-06,
      "loss": 0.25958251953125,
      "step": 77
    },
    {
      "epoch": 0.01086880791472166,
      "grad_norm": 13.491727828979492,
      "learning_rate": 6.0156250000000005e-06,
      "loss": 0.22585296630859375,
      "step": 78
    },
    {
      "epoch": 0.011008151605936042,
      "grad_norm": 11.723892211914062,
      "learning_rate": 6.093750000000001e-06,
      "loss": 0.2397003173828125,
      "step": 79
    },
    {
      "epoch": 0.011147495297150421,
      "grad_norm": 20.432104110717773,
      "learning_rate": 6.171875e-06,
      "loss": 0.2823944091796875,
      "step": 80
    },
    {
      "epoch": 0.011286838988364802,
      "grad_norm": 18.704376220703125,
      "learning_rate": 6.25e-06,
      "loss": 0.3331146240234375,
      "step": 81
    },
    {
      "epoch": 0.011426182679579183,
      "grad_norm": 16.293437957763672,
      "learning_rate": 6.3281250000000005e-06,
      "loss": 0.3228607177734375,
      "step": 82
    },
    {
      "epoch": 0.011565526370793562,
      "grad_norm": 7.765333652496338,
      "learning_rate": 6.406250000000001e-06,
      "loss": 0.3239898681640625,
      "step": 83
    },
    {
      "epoch": 0.011704870062007943,
      "grad_norm": 9.332883834838867,
      "learning_rate": 6.484375000000001e-06,
      "loss": 0.24146270751953125,
      "step": 84
    },
    {
      "epoch": 0.011844213753222322,
      "grad_norm": 5.795038223266602,
      "learning_rate": 6.5625e-06,
      "loss": 0.23305511474609375,
      "step": 85
    },
    {
      "epoch": 0.011983557444436703,
      "grad_norm": 13.246281623840332,
      "learning_rate": 6.6406250000000005e-06,
      "loss": 0.2350921630859375,
      "step": 86
    },
    {
      "epoch": 0.012122901135651083,
      "grad_norm": 12.693568229675293,
      "learning_rate": 6.718750000000001e-06,
      "loss": 0.20955657958984375,
      "step": 87
    },
    {
      "epoch": 0.012262244826865464,
      "grad_norm": 11.37897777557373,
      "learning_rate": 6.796875000000001e-06,
      "loss": 0.2831573486328125,
      "step": 88
    },
    {
      "epoch": 0.012401588518079844,
      "grad_norm": 13.740562438964844,
      "learning_rate": 6.875e-06,
      "loss": 0.25287628173828125,
      "step": 89
    },
    {
      "epoch": 0.012540932209294224,
      "grad_norm": 5.004226207733154,
      "learning_rate": 6.9531250000000004e-06,
      "loss": 0.23177337646484375,
      "step": 90
    },
    {
      "epoch": 0.012680275900508605,
      "grad_norm": 16.822385787963867,
      "learning_rate": 7.031250000000001e-06,
      "loss": 0.261810302734375,
      "step": 91
    },
    {
      "epoch": 0.012819619591722984,
      "grad_norm": 25.2241153717041,
      "learning_rate": 7.109375000000001e-06,
      "loss": 0.2317962646484375,
      "step": 92
    },
    {
      "epoch": 0.012958963282937365,
      "grad_norm": 12.065925598144531,
      "learning_rate": 7.1875e-06,
      "loss": 0.323699951171875,
      "step": 93
    },
    {
      "epoch": 0.013098306974151746,
      "grad_norm": 3.920778512954712,
      "learning_rate": 7.265625e-06,
      "loss": 0.22322845458984375,
      "step": 94
    },
    {
      "epoch": 0.013237650665366125,
      "grad_norm": 8.746639251708984,
      "learning_rate": 7.343750000000001e-06,
      "loss": 0.208465576171875,
      "step": 95
    },
    {
      "epoch": 0.013376994356580506,
      "grad_norm": 4.796881198883057,
      "learning_rate": 7.421875000000001e-06,
      "loss": 0.18829345703125,
      "step": 96
    },
    {
      "epoch": 0.013516338047794885,
      "grad_norm": 6.854090213775635,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.21773910522460938,
      "step": 97
    },
    {
      "epoch": 0.013655681739009266,
      "grad_norm": 6.848060607910156,
      "learning_rate": 7.578125e-06,
      "loss": 0.235931396484375,
      "step": 98
    },
    {
      "epoch": 0.013795025430223647,
      "grad_norm": 5.420156002044678,
      "learning_rate": 7.656250000000001e-06,
      "loss": 0.25177001953125,
      "step": 99
    },
    {
      "epoch": 0.013934369121438027,
      "grad_norm": 8.819807052612305,
      "learning_rate": 7.734375e-06,
      "loss": 0.214385986328125,
      "step": 100
    },
    {
      "epoch": 0.014073712812652408,
      "grad_norm": 4.034743309020996,
      "learning_rate": 7.8125e-06,
      "loss": 0.1653900146484375,
      "step": 101
    },
    {
      "epoch": 0.014213056503866787,
      "grad_norm": 20.99355697631836,
      "learning_rate": 7.890625e-06,
      "loss": 0.28530120849609375,
      "step": 102
    },
    {
      "epoch": 0.014352400195081168,
      "grad_norm": 15.552170753479004,
      "learning_rate": 7.96875e-06,
      "loss": 0.23798370361328125,
      "step": 103
    },
    {
      "epoch": 0.014491743886295549,
      "grad_norm": 7.223971366882324,
      "learning_rate": 8.046875e-06,
      "loss": 0.21608734130859375,
      "step": 104
    },
    {
      "epoch": 0.014631087577509928,
      "grad_norm": 10.722821235656738,
      "learning_rate": 8.125000000000001e-06,
      "loss": 0.17705535888671875,
      "step": 105
    },
    {
      "epoch": 0.014770431268724309,
      "grad_norm": 11.949408531188965,
      "learning_rate": 8.203125000000001e-06,
      "loss": 0.23856353759765625,
      "step": 106
    },
    {
      "epoch": 0.014909774959938688,
      "grad_norm": 9.366409301757812,
      "learning_rate": 8.281250000000001e-06,
      "loss": 0.203033447265625,
      "step": 107
    },
    {
      "epoch": 0.01504911865115307,
      "grad_norm": 15.073811531066895,
      "learning_rate": 8.359375e-06,
      "loss": 0.14673614501953125,
      "step": 108
    },
    {
      "epoch": 0.01518846234236745,
      "grad_norm": 14.343490600585938,
      "learning_rate": 8.4375e-06,
      "loss": 0.2232666015625,
      "step": 109
    },
    {
      "epoch": 0.01532780603358183,
      "grad_norm": 3.198679208755493,
      "learning_rate": 8.515625e-06,
      "loss": 0.2050323486328125,
      "step": 110
    },
    {
      "epoch": 0.01546714972479621,
      "grad_norm": 16.795978546142578,
      "learning_rate": 8.59375e-06,
      "loss": 0.19673538208007812,
      "step": 111
    },
    {
      "epoch": 0.01560649341601059,
      "grad_norm": 15.135960578918457,
      "learning_rate": 8.671875e-06,
      "loss": 0.238311767578125,
      "step": 112
    },
    {
      "epoch": 0.01574583710722497,
      "grad_norm": 10.938837051391602,
      "learning_rate": 8.750000000000001e-06,
      "loss": 0.29259490966796875,
      "step": 113
    },
    {
      "epoch": 0.01588518079843935,
      "grad_norm": 15.526944160461426,
      "learning_rate": 8.828125000000001e-06,
      "loss": 0.235565185546875,
      "step": 114
    },
    {
      "epoch": 0.016024524489653733,
      "grad_norm": 20.863676071166992,
      "learning_rate": 8.906250000000001e-06,
      "loss": 0.2219696044921875,
      "step": 115
    },
    {
      "epoch": 0.01616386818086811,
      "grad_norm": 15.12514877319336,
      "learning_rate": 8.984375000000002e-06,
      "loss": 0.2725677490234375,
      "step": 116
    },
    {
      "epoch": 0.01630321187208249,
      "grad_norm": 8.126269340515137,
      "learning_rate": 9.0625e-06,
      "loss": 0.17389678955078125,
      "step": 117
    },
    {
      "epoch": 0.016442555563296872,
      "grad_norm": 19.26151466369629,
      "learning_rate": 9.140625e-06,
      "loss": 0.1754608154296875,
      "step": 118
    },
    {
      "epoch": 0.016581899254511253,
      "grad_norm": 25.02276039123535,
      "learning_rate": 9.21875e-06,
      "loss": 0.2781982421875,
      "step": 119
    },
    {
      "epoch": 0.016721242945725634,
      "grad_norm": 18.551816940307617,
      "learning_rate": 9.296875e-06,
      "loss": 0.27923583984375,
      "step": 120
    },
    {
      "epoch": 0.01686058663694001,
      "grad_norm": 9.031113624572754,
      "learning_rate": 9.375000000000001e-06,
      "loss": 0.2947731018066406,
      "step": 121
    },
    {
      "epoch": 0.016999930328154392,
      "grad_norm": 6.5948028564453125,
      "learning_rate": 9.453125000000001e-06,
      "loss": 0.19278717041015625,
      "step": 122
    },
    {
      "epoch": 0.017139274019368773,
      "grad_norm": 8.066917419433594,
      "learning_rate": 9.531250000000001e-06,
      "loss": 0.19611740112304688,
      "step": 123
    },
    {
      "epoch": 0.017278617710583154,
      "grad_norm": 4.516763210296631,
      "learning_rate": 9.609375000000001e-06,
      "loss": 0.24005889892578125,
      "step": 124
    },
    {
      "epoch": 0.017417961401797532,
      "grad_norm": 2.6600706577301025,
      "learning_rate": 9.6875e-06,
      "loss": 0.1641998291015625,
      "step": 125
    },
    {
      "epoch": 0.017557305093011913,
      "grad_norm": 10.423636436462402,
      "learning_rate": 9.765625e-06,
      "loss": 0.18814468383789062,
      "step": 126
    },
    {
      "epoch": 0.017696648784226294,
      "grad_norm": 6.515997409820557,
      "learning_rate": 9.84375e-06,
      "loss": 0.2266082763671875,
      "step": 127
    },
    {
      "epoch": 0.017835992475440675,
      "grad_norm": 8.942510604858398,
      "learning_rate": 9.921875e-06,
      "loss": 0.232635498046875,
      "step": 128
    },
    {
      "epoch": 0.017975336166655056,
      "grad_norm": 11.724843978881836,
      "learning_rate": 1e-05,
      "loss": 0.1920928955078125,
      "step": 129
    },
    {
      "epoch": 0.018114679857869433,
      "grad_norm": 2.1824944019317627,
      "learning_rate": 1.0078125000000001e-05,
      "loss": 0.1387939453125,
      "step": 130
    },
    {
      "epoch": 0.018254023549083814,
      "grad_norm": 16.392610549926758,
      "learning_rate": 1.0156250000000001e-05,
      "loss": 0.22784423828125,
      "step": 131
    },
    {
      "epoch": 0.018393367240298195,
      "grad_norm": 5.487381935119629,
      "learning_rate": 1.0234375000000001e-05,
      "loss": 0.2693672180175781,
      "step": 132
    },
    {
      "epoch": 0.018532710931512576,
      "grad_norm": 7.823380470275879,
      "learning_rate": 1.0312500000000002e-05,
      "loss": 0.142913818359375,
      "step": 133
    },
    {
      "epoch": 0.018672054622726957,
      "grad_norm": 4.9395270347595215,
      "learning_rate": 1.0390625e-05,
      "loss": 0.14363861083984375,
      "step": 134
    },
    {
      "epoch": 0.018811398313941335,
      "grad_norm": 6.06682014465332,
      "learning_rate": 1.046875e-05,
      "loss": 0.15842819213867188,
      "step": 135
    },
    {
      "epoch": 0.018950742005155716,
      "grad_norm": 6.153921604156494,
      "learning_rate": 1.0546875e-05,
      "loss": 0.2611846923828125,
      "step": 136
    },
    {
      "epoch": 0.019090085696370097,
      "grad_norm": 4.895573616027832,
      "learning_rate": 1.0625e-05,
      "loss": 0.1944427490234375,
      "step": 137
    },
    {
      "epoch": 0.019229429387584478,
      "grad_norm": 3.2464022636413574,
      "learning_rate": 1.0703125000000001e-05,
      "loss": 0.19963836669921875,
      "step": 138
    },
    {
      "epoch": 0.01936877307879886,
      "grad_norm": 12.111172676086426,
      "learning_rate": 1.0781250000000001e-05,
      "loss": 0.23938751220703125,
      "step": 139
    },
    {
      "epoch": 0.019508116770013236,
      "grad_norm": 13.649538040161133,
      "learning_rate": 1.0859375000000001e-05,
      "loss": 0.179962158203125,
      "step": 140
    },
    {
      "epoch": 0.019647460461227617,
      "grad_norm": 4.919685363769531,
      "learning_rate": 1.0937500000000002e-05,
      "loss": 0.24884796142578125,
      "step": 141
    },
    {
      "epoch": 0.019786804152441998,
      "grad_norm": 5.723228931427002,
      "learning_rate": 1.1015625e-05,
      "loss": 0.19905853271484375,
      "step": 142
    },
    {
      "epoch": 0.01992614784365638,
      "grad_norm": 2.247729778289795,
      "learning_rate": 1.109375e-05,
      "loss": 0.23711776733398438,
      "step": 143
    },
    {
      "epoch": 0.02006549153487076,
      "grad_norm": 10.623004913330078,
      "learning_rate": 1.1171875e-05,
      "loss": 0.21348953247070312,
      "step": 144
    },
    {
      "epoch": 0.020204835226085138,
      "grad_norm": 6.32315731048584,
      "learning_rate": 1.125e-05,
      "loss": 0.18719482421875,
      "step": 145
    },
    {
      "epoch": 0.02034417891729952,
      "grad_norm": 6.062091827392578,
      "learning_rate": 1.1328125000000001e-05,
      "loss": 0.17003250122070312,
      "step": 146
    },
    {
      "epoch": 0.0204835226085139,
      "grad_norm": 10.080375671386719,
      "learning_rate": 1.1406250000000001e-05,
      "loss": 0.21138763427734375,
      "step": 147
    },
    {
      "epoch": 0.02062286629972828,
      "grad_norm": 7.955326557159424,
      "learning_rate": 1.1484375000000001e-05,
      "loss": 0.17786407470703125,
      "step": 148
    },
    {
      "epoch": 0.02076220999094266,
      "grad_norm": 2.162188768386841,
      "learning_rate": 1.1562500000000002e-05,
      "loss": 0.14757537841796875,
      "step": 149
    },
    {
      "epoch": 0.02090155368215704,
      "grad_norm": 6.504528045654297,
      "learning_rate": 1.1640625000000002e-05,
      "loss": 0.17267608642578125,
      "step": 150
    },
    {
      "epoch": 0.02104089737337142,
      "grad_norm": 4.625658988952637,
      "learning_rate": 1.171875e-05,
      "loss": 0.229766845703125,
      "step": 151
    },
    {
      "epoch": 0.0211802410645858,
      "grad_norm": 5.853835105895996,
      "learning_rate": 1.1796875e-05,
      "loss": 0.21680450439453125,
      "step": 152
    },
    {
      "epoch": 0.021319584755800182,
      "grad_norm": 8.290628433227539,
      "learning_rate": 1.1875e-05,
      "loss": 0.12215805053710938,
      "step": 153
    },
    {
      "epoch": 0.021458928447014563,
      "grad_norm": 6.803259372711182,
      "learning_rate": 1.1953125000000001e-05,
      "loss": 0.1794586181640625,
      "step": 154
    },
    {
      "epoch": 0.02159827213822894,
      "grad_norm": 5.6353440284729,
      "learning_rate": 1.2031250000000001e-05,
      "loss": 0.19565582275390625,
      "step": 155
    },
    {
      "epoch": 0.02173761582944332,
      "grad_norm": 8.689154624938965,
      "learning_rate": 1.2109375000000001e-05,
      "loss": 0.1830902099609375,
      "step": 156
    },
    {
      "epoch": 0.021876959520657702,
      "grad_norm": 5.814373970031738,
      "learning_rate": 1.2187500000000001e-05,
      "loss": 0.1688690185546875,
      "step": 157
    },
    {
      "epoch": 0.022016303211872083,
      "grad_norm": 15.974237442016602,
      "learning_rate": 1.2265625000000002e-05,
      "loss": 0.2555427551269531,
      "step": 158
    },
    {
      "epoch": 0.022155646903086464,
      "grad_norm": 5.6902618408203125,
      "learning_rate": 1.234375e-05,
      "loss": 0.15194320678710938,
      "step": 159
    },
    {
      "epoch": 0.022294990594300842,
      "grad_norm": 9.141617774963379,
      "learning_rate": 1.2421875e-05,
      "loss": 0.22792816162109375,
      "step": 160
    },
    {
      "epoch": 0.022434334285515223,
      "grad_norm": 4.283246994018555,
      "learning_rate": 1.25e-05,
      "loss": 0.2251434326171875,
      "step": 161
    },
    {
      "epoch": 0.022573677976729604,
      "grad_norm": 7.396460056304932,
      "learning_rate": 1.2578125e-05,
      "loss": 0.1446075439453125,
      "step": 162
    },
    {
      "epoch": 0.022713021667943985,
      "grad_norm": 5.757813930511475,
      "learning_rate": 1.2656250000000001e-05,
      "loss": 0.18130874633789062,
      "step": 163
    },
    {
      "epoch": 0.022852365359158366,
      "grad_norm": 3.1287357807159424,
      "learning_rate": 1.2734375000000001e-05,
      "loss": 0.118560791015625,
      "step": 164
    },
    {
      "epoch": 0.022991709050372743,
      "grad_norm": 1.7641234397888184,
      "learning_rate": 1.2812500000000001e-05,
      "loss": 0.16223526000976562,
      "step": 165
    },
    {
      "epoch": 0.023131052741587124,
      "grad_norm": 3.406259536743164,
      "learning_rate": 1.2890625000000002e-05,
      "loss": 0.180267333984375,
      "step": 166
    },
    {
      "epoch": 0.023270396432801505,
      "grad_norm": 11.442927360534668,
      "learning_rate": 1.2968750000000002e-05,
      "loss": 0.179901123046875,
      "step": 167
    },
    {
      "epoch": 0.023409740124015886,
      "grad_norm": 4.9085493087768555,
      "learning_rate": 1.3046875e-05,
      "loss": 0.11784744262695312,
      "step": 168
    },
    {
      "epoch": 0.023549083815230267,
      "grad_norm": 5.811831474304199,
      "learning_rate": 1.3125e-05,
      "loss": 0.12638092041015625,
      "step": 169
    },
    {
      "epoch": 0.023688427506444645,
      "grad_norm": 3.0889089107513428,
      "learning_rate": 1.3203125e-05,
      "loss": 0.10998153686523438,
      "step": 170
    },
    {
      "epoch": 0.023827771197659026,
      "grad_norm": 2.942074775695801,
      "learning_rate": 1.3281250000000001e-05,
      "loss": 0.18724441528320312,
      "step": 171
    },
    {
      "epoch": 0.023967114888873407,
      "grad_norm": 4.158938407897949,
      "learning_rate": 1.3359375000000001e-05,
      "loss": 0.1234893798828125,
      "step": 172
    },
    {
      "epoch": 0.024106458580087788,
      "grad_norm": 4.3631367683410645,
      "learning_rate": 1.3437500000000001e-05,
      "loss": 0.1478424072265625,
      "step": 173
    },
    {
      "epoch": 0.024245802271302165,
      "grad_norm": 5.135581016540527,
      "learning_rate": 1.3515625000000002e-05,
      "loss": 0.17469406127929688,
      "step": 174
    },
    {
      "epoch": 0.024385145962516546,
      "grad_norm": 7.9102301597595215,
      "learning_rate": 1.3593750000000002e-05,
      "loss": 0.20298004150390625,
      "step": 175
    },
    {
      "epoch": 0.024524489653730927,
      "grad_norm": 2.550529956817627,
      "learning_rate": 1.3671875e-05,
      "loss": 0.15830612182617188,
      "step": 176
    },
    {
      "epoch": 0.024663833344945308,
      "grad_norm": 10.557239532470703,
      "learning_rate": 1.375e-05,
      "loss": 0.14163589477539062,
      "step": 177
    },
    {
      "epoch": 0.02480317703615969,
      "grad_norm": 7.7039899826049805,
      "learning_rate": 1.3828125e-05,
      "loss": 0.1522216796875,
      "step": 178
    },
    {
      "epoch": 0.024942520727374067,
      "grad_norm": 4.780228614807129,
      "learning_rate": 1.3906250000000001e-05,
      "loss": 0.12770843505859375,
      "step": 179
    },
    {
      "epoch": 0.025081864418588447,
      "grad_norm": 7.108807563781738,
      "learning_rate": 1.3984375000000001e-05,
      "loss": 0.106536865234375,
      "step": 180
    },
    {
      "epoch": 0.02522120810980283,
      "grad_norm": 10.260586738586426,
      "learning_rate": 1.4062500000000001e-05,
      "loss": 0.17214584350585938,
      "step": 181
    },
    {
      "epoch": 0.02536055180101721,
      "grad_norm": 15.482001304626465,
      "learning_rate": 1.4140625000000002e-05,
      "loss": 0.23189544677734375,
      "step": 182
    },
    {
      "epoch": 0.02549989549223159,
      "grad_norm": 3.990216016769409,
      "learning_rate": 1.4218750000000002e-05,
      "loss": 0.2648162841796875,
      "step": 183
    },
    {
      "epoch": 0.025639239183445968,
      "grad_norm": 11.725672721862793,
      "learning_rate": 1.4296875000000002e-05,
      "loss": 0.22397232055664062,
      "step": 184
    },
    {
      "epoch": 0.02577858287466035,
      "grad_norm": 4.92960786819458,
      "learning_rate": 1.4375e-05,
      "loss": 0.15979385375976562,
      "step": 185
    },
    {
      "epoch": 0.02591792656587473,
      "grad_norm": 4.962447643280029,
      "learning_rate": 1.4453125e-05,
      "loss": 0.19137191772460938,
      "step": 186
    },
    {
      "epoch": 0.02605727025708911,
      "grad_norm": 4.462624549865723,
      "learning_rate": 1.453125e-05,
      "loss": 0.1897125244140625,
      "step": 187
    },
    {
      "epoch": 0.026196613948303492,
      "grad_norm": 4.031778335571289,
      "learning_rate": 1.4609375000000001e-05,
      "loss": 0.17697906494140625,
      "step": 188
    },
    {
      "epoch": 0.02633595763951787,
      "grad_norm": 6.0412139892578125,
      "learning_rate": 1.4687500000000001e-05,
      "loss": 0.13458251953125,
      "step": 189
    },
    {
      "epoch": 0.02647530133073225,
      "grad_norm": 6.443297863006592,
      "learning_rate": 1.4765625000000001e-05,
      "loss": 0.19022369384765625,
      "step": 190
    },
    {
      "epoch": 0.02661464502194663,
      "grad_norm": 1.3980852365493774,
      "learning_rate": 1.4843750000000002e-05,
      "loss": 0.1270732879638672,
      "step": 191
    },
    {
      "epoch": 0.026753988713161012,
      "grad_norm": 8.832712173461914,
      "learning_rate": 1.4921875000000002e-05,
      "loss": 0.22046661376953125,
      "step": 192
    },
    {
      "epoch": 0.026893332404375393,
      "grad_norm": 6.981844425201416,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.2545623779296875,
      "step": 193
    },
    {
      "epoch": 0.02703267609558977,
      "grad_norm": 3.730447292327881,
      "learning_rate": 1.5078125e-05,
      "loss": 0.13394546508789062,
      "step": 194
    },
    {
      "epoch": 0.02717201978680415,
      "grad_norm": 3.3465733528137207,
      "learning_rate": 1.515625e-05,
      "loss": 0.17260360717773438,
      "step": 195
    },
    {
      "epoch": 0.027311363478018533,
      "grad_norm": 5.559998989105225,
      "learning_rate": 1.5234375000000001e-05,
      "loss": 0.12349700927734375,
      "step": 196
    },
    {
      "epoch": 0.027450707169232914,
      "grad_norm": 6.5464701652526855,
      "learning_rate": 1.5312500000000003e-05,
      "loss": 0.2064361572265625,
      "step": 197
    },
    {
      "epoch": 0.027590050860447295,
      "grad_norm": 7.421088695526123,
      "learning_rate": 1.5390625e-05,
      "loss": 0.167144775390625,
      "step": 198
    },
    {
      "epoch": 0.027729394551661672,
      "grad_norm": 5.992583274841309,
      "learning_rate": 1.546875e-05,
      "loss": 0.15998458862304688,
      "step": 199
    },
    {
      "epoch": 0.027868738242876053,
      "grad_norm": 2.1265034675598145,
      "learning_rate": 1.5546875e-05,
      "loss": 0.17952728271484375,
      "step": 200
    },
    {
      "epoch": 0.028008081934090434,
      "grad_norm": 4.205785274505615,
      "learning_rate": 1.5625e-05,
      "loss": 0.13156509399414062,
      "step": 201
    },
    {
      "epoch": 0.028147425625304815,
      "grad_norm": 1.694059133529663,
      "learning_rate": 1.5703125e-05,
      "loss": 0.0891876220703125,
      "step": 202
    },
    {
      "epoch": 0.028286769316519196,
      "grad_norm": 2.5963356494903564,
      "learning_rate": 1.578125e-05,
      "loss": 0.10303878784179688,
      "step": 203
    },
    {
      "epoch": 0.028426113007733574,
      "grad_norm": 2.343632698059082,
      "learning_rate": 1.5859375e-05,
      "loss": 0.12206268310546875,
      "step": 204
    },
    {
      "epoch": 0.028565456698947955,
      "grad_norm": 2.1490259170532227,
      "learning_rate": 1.59375e-05,
      "loss": 0.13367843627929688,
      "step": 205
    },
    {
      "epoch": 0.028704800390162336,
      "grad_norm": 3.200058698654175,
      "learning_rate": 1.6015625e-05,
      "loss": 0.11779022216796875,
      "step": 206
    },
    {
      "epoch": 0.028844144081376717,
      "grad_norm": 9.46463680267334,
      "learning_rate": 1.609375e-05,
      "loss": 0.15291595458984375,
      "step": 207
    },
    {
      "epoch": 0.028983487772591097,
      "grad_norm": 5.547858715057373,
      "learning_rate": 1.6171875000000002e-05,
      "loss": 0.14309310913085938,
      "step": 208
    },
    {
      "epoch": 0.029122831463805475,
      "grad_norm": 4.851221561431885,
      "learning_rate": 1.6250000000000002e-05,
      "loss": 0.0955352783203125,
      "step": 209
    },
    {
      "epoch": 0.029262175155019856,
      "grad_norm": 7.888828277587891,
      "learning_rate": 1.6328125000000002e-05,
      "loss": 0.20464324951171875,
      "step": 210
    },
    {
      "epoch": 0.029401518846234237,
      "grad_norm": 3.4378020763397217,
      "learning_rate": 1.6406250000000002e-05,
      "loss": 0.12322235107421875,
      "step": 211
    },
    {
      "epoch": 0.029540862537448618,
      "grad_norm": 4.337862968444824,
      "learning_rate": 1.6484375000000003e-05,
      "loss": 0.14989852905273438,
      "step": 212
    },
    {
      "epoch": 0.029680206228663,
      "grad_norm": 3.9836552143096924,
      "learning_rate": 1.6562500000000003e-05,
      "loss": 0.18027496337890625,
      "step": 213
    },
    {
      "epoch": 0.029819549919877376,
      "grad_norm": 6.980125427246094,
      "learning_rate": 1.6640625000000003e-05,
      "loss": 0.196807861328125,
      "step": 214
    },
    {
      "epoch": 0.029958893611091757,
      "grad_norm": 8.426046371459961,
      "learning_rate": 1.671875e-05,
      "loss": 0.21418380737304688,
      "step": 215
    },
    {
      "epoch": 0.03009823730230614,
      "grad_norm": 7.015588283538818,
      "learning_rate": 1.6796875e-05,
      "loss": 0.22662734985351562,
      "step": 216
    },
    {
      "epoch": 0.03023758099352052,
      "grad_norm": 4.552795886993408,
      "learning_rate": 1.6875e-05,
      "loss": 0.15364456176757812,
      "step": 217
    },
    {
      "epoch": 0.0303769246847349,
      "grad_norm": 1.7086564302444458,
      "learning_rate": 1.6953125e-05,
      "loss": 0.15375518798828125,
      "step": 218
    },
    {
      "epoch": 0.030516268375949278,
      "grad_norm": 0.8052099943161011,
      "learning_rate": 1.703125e-05,
      "loss": 0.11156082153320312,
      "step": 219
    },
    {
      "epoch": 0.03065561206716366,
      "grad_norm": 4.236763954162598,
      "learning_rate": 1.7109375e-05,
      "loss": 0.13831329345703125,
      "step": 220
    },
    {
      "epoch": 0.03079495575837804,
      "grad_norm": 2.7073967456817627,
      "learning_rate": 1.71875e-05,
      "loss": 0.1667022705078125,
      "step": 221
    },
    {
      "epoch": 0.03093429944959242,
      "grad_norm": 4.411389350891113,
      "learning_rate": 1.7265625e-05,
      "loss": 0.23344802856445312,
      "step": 222
    },
    {
      "epoch": 0.031073643140806798,
      "grad_norm": 19.97957992553711,
      "learning_rate": 1.734375e-05,
      "loss": 0.21508026123046875,
      "step": 223
    },
    {
      "epoch": 0.03121298683202118,
      "grad_norm": 10.919645309448242,
      "learning_rate": 1.7421875e-05,
      "loss": 0.1551361083984375,
      "step": 224
    },
    {
      "epoch": 0.031352330523235564,
      "grad_norm": 1.9346929788589478,
      "learning_rate": 1.7500000000000002e-05,
      "loss": 0.14713668823242188,
      "step": 225
    },
    {
      "epoch": 0.03149167421444994,
      "grad_norm": 3.8985822200775146,
      "learning_rate": 1.7578125000000002e-05,
      "loss": 0.13894271850585938,
      "step": 226
    },
    {
      "epoch": 0.03163101790566432,
      "grad_norm": 2.8252902030944824,
      "learning_rate": 1.7656250000000002e-05,
      "loss": 0.133819580078125,
      "step": 227
    },
    {
      "epoch": 0.0317703615968787,
      "grad_norm": 4.966582298278809,
      "learning_rate": 1.7734375000000002e-05,
      "loss": 0.17972183227539062,
      "step": 228
    },
    {
      "epoch": 0.03190970528809308,
      "grad_norm": 11.945621490478516,
      "learning_rate": 1.7812500000000003e-05,
      "loss": 0.21394729614257812,
      "step": 229
    },
    {
      "epoch": 0.032049048979307465,
      "grad_norm": 15.203856468200684,
      "learning_rate": 1.7890625000000003e-05,
      "loss": 0.2523040771484375,
      "step": 230
    },
    {
      "epoch": 0.03218839267052184,
      "grad_norm": 4.826852798461914,
      "learning_rate": 1.7968750000000003e-05,
      "loss": 0.14131927490234375,
      "step": 231
    },
    {
      "epoch": 0.03232773636173622,
      "grad_norm": 9.49569034576416,
      "learning_rate": 1.8046875e-05,
      "loss": 0.22652435302734375,
      "step": 232
    },
    {
      "epoch": 0.032467080052950605,
      "grad_norm": 4.826241970062256,
      "learning_rate": 1.8125e-05,
      "loss": 0.13573455810546875,
      "step": 233
    },
    {
      "epoch": 0.03260642374416498,
      "grad_norm": 9.664764404296875,
      "learning_rate": 1.8203125e-05,
      "loss": 0.2448883056640625,
      "step": 234
    },
    {
      "epoch": 0.032745767435379367,
      "grad_norm": 2.496797561645508,
      "learning_rate": 1.828125e-05,
      "loss": 0.1580657958984375,
      "step": 235
    },
    {
      "epoch": 0.032885111126593744,
      "grad_norm": 9.663092613220215,
      "learning_rate": 1.8359375e-05,
      "loss": 0.13799285888671875,
      "step": 236
    },
    {
      "epoch": 0.03302445481780812,
      "grad_norm": 7.511887550354004,
      "learning_rate": 1.84375e-05,
      "loss": 0.14389991760253906,
      "step": 237
    },
    {
      "epoch": 0.033163798509022506,
      "grad_norm": 6.70507287979126,
      "learning_rate": 1.8515625e-05,
      "loss": 0.13750076293945312,
      "step": 238
    },
    {
      "epoch": 0.033303142200236883,
      "grad_norm": 1.5982306003570557,
      "learning_rate": 1.859375e-05,
      "loss": 0.09104156494140625,
      "step": 239
    },
    {
      "epoch": 0.03344248589145127,
      "grad_norm": 4.109487533569336,
      "learning_rate": 1.8671875e-05,
      "loss": 0.17372512817382812,
      "step": 240
    },
    {
      "epoch": 0.033581829582665645,
      "grad_norm": 5.061545372009277,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 0.12432098388671875,
      "step": 241
    },
    {
      "epoch": 0.03372117327388002,
      "grad_norm": 3.4716570377349854,
      "learning_rate": 1.8828125000000002e-05,
      "loss": 0.11724090576171875,
      "step": 242
    },
    {
      "epoch": 0.03386051696509441,
      "grad_norm": 5.689507007598877,
      "learning_rate": 1.8906250000000002e-05,
      "loss": 0.19164276123046875,
      "step": 243
    },
    {
      "epoch": 0.033999860656308785,
      "grad_norm": 16.495922088623047,
      "learning_rate": 1.8984375000000002e-05,
      "loss": 0.20988845825195312,
      "step": 244
    },
    {
      "epoch": 0.03413920434752317,
      "grad_norm": 18.91434097290039,
      "learning_rate": 1.9062500000000003e-05,
      "loss": 0.18117904663085938,
      "step": 245
    },
    {
      "epoch": 0.03427854803873755,
      "grad_norm": 5.467302322387695,
      "learning_rate": 1.9140625000000003e-05,
      "loss": 0.0908203125,
      "step": 246
    },
    {
      "epoch": 0.034417891729951924,
      "grad_norm": 2.3467586040496826,
      "learning_rate": 1.9218750000000003e-05,
      "loss": 0.13916015625,
      "step": 247
    },
    {
      "epoch": 0.03455723542116631,
      "grad_norm": 3.496011734008789,
      "learning_rate": 1.9296875000000003e-05,
      "loss": 0.15574264526367188,
      "step": 248
    },
    {
      "epoch": 0.034696579112380686,
      "grad_norm": 4.236489295959473,
      "learning_rate": 1.9375e-05,
      "loss": 0.18981170654296875,
      "step": 249
    },
    {
      "epoch": 0.034835922803595064,
      "grad_norm": 3.7420201301574707,
      "learning_rate": 1.9453125e-05,
      "loss": 0.11133575439453125,
      "step": 250
    },
    {
      "epoch": 0.03497526649480945,
      "grad_norm": 4.348264217376709,
      "learning_rate": 1.953125e-05,
      "loss": 0.14104843139648438,
      "step": 251
    },
    {
      "epoch": 0.035114610186023826,
      "grad_norm": 13.566645622253418,
      "learning_rate": 1.9609375e-05,
      "loss": 0.23079681396484375,
      "step": 252
    },
    {
      "epoch": 0.03525395387723821,
      "grad_norm": 10.602228164672852,
      "learning_rate": 1.96875e-05,
      "loss": 0.18766403198242188,
      "step": 253
    },
    {
      "epoch": 0.03539329756845259,
      "grad_norm": 6.669156074523926,
      "learning_rate": 1.9765625e-05,
      "loss": 0.134124755859375,
      "step": 254
    },
    {
      "epoch": 0.035532641259666965,
      "grad_norm": 11.8833646774292,
      "learning_rate": 1.984375e-05,
      "loss": 0.19338607788085938,
      "step": 255
    },
    {
      "epoch": 0.03567198495088135,
      "grad_norm": 10.189719200134277,
      "learning_rate": 1.9921875e-05,
      "loss": 0.19855880737304688,
      "step": 256
    },
    {
      "epoch": 0.03581132864209573,
      "grad_norm": 3.4376537799835205,
      "learning_rate": 2e-05,
      "loss": 0.16809844970703125,
      "step": 257
    },
    {
      "epoch": 0.03595067233331011,
      "grad_norm": 2.0516419410705566,
      "learning_rate": 2.0078125000000002e-05,
      "loss": 0.11817741394042969,
      "step": 258
    },
    {
      "epoch": 0.03609001602452449,
      "grad_norm": 2.8684804439544678,
      "learning_rate": 2.0156250000000002e-05,
      "loss": 0.16216659545898438,
      "step": 259
    },
    {
      "epoch": 0.03622935971573887,
      "grad_norm": 3.486982822418213,
      "learning_rate": 2.0234375000000002e-05,
      "loss": 0.1451244354248047,
      "step": 260
    },
    {
      "epoch": 0.03636870340695325,
      "grad_norm": 12.981520652770996,
      "learning_rate": 2.0312500000000002e-05,
      "loss": 0.1826019287109375,
      "step": 261
    },
    {
      "epoch": 0.03650804709816763,
      "grad_norm": 10.362968444824219,
      "learning_rate": 2.0390625000000003e-05,
      "loss": 0.17943191528320312,
      "step": 262
    },
    {
      "epoch": 0.03664739078938201,
      "grad_norm": 8.4119291305542,
      "learning_rate": 2.0468750000000003e-05,
      "loss": 0.19427490234375,
      "step": 263
    },
    {
      "epoch": 0.03678673448059639,
      "grad_norm": 8.52756404876709,
      "learning_rate": 2.0546875000000003e-05,
      "loss": 0.18929672241210938,
      "step": 264
    },
    {
      "epoch": 0.03692607817181077,
      "grad_norm": 8.45046329498291,
      "learning_rate": 2.0625000000000003e-05,
      "loss": 0.20080184936523438,
      "step": 265
    },
    {
      "epoch": 0.03706542186302515,
      "grad_norm": 4.45372200012207,
      "learning_rate": 2.0703125e-05,
      "loss": 0.11498641967773438,
      "step": 266
    },
    {
      "epoch": 0.03720476555423953,
      "grad_norm": 2.6231820583343506,
      "learning_rate": 2.078125e-05,
      "loss": 0.11147689819335938,
      "step": 267
    },
    {
      "epoch": 0.037344109245453914,
      "grad_norm": 6.669605731964111,
      "learning_rate": 2.0859375e-05,
      "loss": 0.17826080322265625,
      "step": 268
    },
    {
      "epoch": 0.03748345293666829,
      "grad_norm": 3.3699822425842285,
      "learning_rate": 2.09375e-05,
      "loss": 0.1303253173828125,
      "step": 269
    },
    {
      "epoch": 0.03762279662788267,
      "grad_norm": 4.310925006866455,
      "learning_rate": 2.1015625e-05,
      "loss": 0.13889694213867188,
      "step": 270
    },
    {
      "epoch": 0.037762140319097054,
      "grad_norm": 2.983015537261963,
      "learning_rate": 2.109375e-05,
      "loss": 0.13807296752929688,
      "step": 271
    },
    {
      "epoch": 0.03790148401031143,
      "grad_norm": 1.8394473791122437,
      "learning_rate": 2.1171875e-05,
      "loss": 0.14680099487304688,
      "step": 272
    },
    {
      "epoch": 0.038040827701525816,
      "grad_norm": 3.6735901832580566,
      "learning_rate": 2.125e-05,
      "loss": 0.13889694213867188,
      "step": 273
    },
    {
      "epoch": 0.03818017139274019,
      "grad_norm": 1.6112126111984253,
      "learning_rate": 2.1328125000000002e-05,
      "loss": 0.09775924682617188,
      "step": 274
    },
    {
      "epoch": 0.03831951508395457,
      "grad_norm": 7.074946880340576,
      "learning_rate": 2.1406250000000002e-05,
      "loss": 0.13819503784179688,
      "step": 275
    },
    {
      "epoch": 0.038458858775168955,
      "grad_norm": 1.4056625366210938,
      "learning_rate": 2.1484375000000002e-05,
      "loss": 0.08742141723632812,
      "step": 276
    },
    {
      "epoch": 0.03859820246638333,
      "grad_norm": 4.244787216186523,
      "learning_rate": 2.1562500000000002e-05,
      "loss": 0.17275238037109375,
      "step": 277
    },
    {
      "epoch": 0.03873754615759772,
      "grad_norm": 4.395383834838867,
      "learning_rate": 2.1640625000000003e-05,
      "loss": 0.15907669067382812,
      "step": 278
    },
    {
      "epoch": 0.038876889848812095,
      "grad_norm": 2.8337953090667725,
      "learning_rate": 2.1718750000000003e-05,
      "loss": 0.13778114318847656,
      "step": 279
    },
    {
      "epoch": 0.03901623354002647,
      "grad_norm": 3.2872002124786377,
      "learning_rate": 2.1796875000000003e-05,
      "loss": 0.11849594116210938,
      "step": 280
    },
    {
      "epoch": 0.03915557723124086,
      "grad_norm": 5.852701663970947,
      "learning_rate": 2.1875000000000003e-05,
      "loss": 0.1680908203125,
      "step": 281
    },
    {
      "epoch": 0.039294920922455234,
      "grad_norm": 4.413242816925049,
      "learning_rate": 2.1953125000000003e-05,
      "loss": 0.11581039428710938,
      "step": 282
    },
    {
      "epoch": 0.03943426461366962,
      "grad_norm": 4.618696689605713,
      "learning_rate": 2.203125e-05,
      "loss": 0.16373443603515625,
      "step": 283
    },
    {
      "epoch": 0.039573608304883996,
      "grad_norm": 11.284041404724121,
      "learning_rate": 2.2109375e-05,
      "loss": 0.2421875,
      "step": 284
    },
    {
      "epoch": 0.039712951996098374,
      "grad_norm": 5.369922637939453,
      "learning_rate": 2.21875e-05,
      "loss": 0.1548919677734375,
      "step": 285
    },
    {
      "epoch": 0.03985229568731276,
      "grad_norm": 7.530871868133545,
      "learning_rate": 2.2265625e-05,
      "loss": 0.15246963500976562,
      "step": 286
    },
    {
      "epoch": 0.039991639378527136,
      "grad_norm": 3.955171585083008,
      "learning_rate": 2.234375e-05,
      "loss": 0.14166259765625,
      "step": 287
    },
    {
      "epoch": 0.04013098306974152,
      "grad_norm": 5.170267581939697,
      "learning_rate": 2.2421875e-05,
      "loss": 0.11017227172851562,
      "step": 288
    },
    {
      "epoch": 0.0402703267609559,
      "grad_norm": 11.165712356567383,
      "learning_rate": 2.25e-05,
      "loss": 0.15656661987304688,
      "step": 289
    },
    {
      "epoch": 0.040409670452170275,
      "grad_norm": 6.673206329345703,
      "learning_rate": 2.2578125e-05,
      "loss": 0.15043258666992188,
      "step": 290
    },
    {
      "epoch": 0.04054901414338466,
      "grad_norm": 2.142934799194336,
      "learning_rate": 2.2656250000000002e-05,
      "loss": 0.144195556640625,
      "step": 291
    },
    {
      "epoch": 0.04068835783459904,
      "grad_norm": 4.6472954750061035,
      "learning_rate": 2.2734375000000002e-05,
      "loss": 0.17383956909179688,
      "step": 292
    },
    {
      "epoch": 0.04082770152581342,
      "grad_norm": 8.35328483581543,
      "learning_rate": 2.2812500000000002e-05,
      "loss": 0.21802902221679688,
      "step": 293
    },
    {
      "epoch": 0.0409670452170278,
      "grad_norm": 4.26718282699585,
      "learning_rate": 2.2890625000000002e-05,
      "loss": 0.14316177368164062,
      "step": 294
    },
    {
      "epoch": 0.04110638890824218,
      "grad_norm": 2.589521884918213,
      "learning_rate": 2.2968750000000003e-05,
      "loss": 0.12925338745117188,
      "step": 295
    },
    {
      "epoch": 0.04124573259945656,
      "grad_norm": 4.902878761291504,
      "learning_rate": 2.3046875000000003e-05,
      "loss": 0.17634201049804688,
      "step": 296
    },
    {
      "epoch": 0.04138507629067094,
      "grad_norm": 3.3919310569763184,
      "learning_rate": 2.3125000000000003e-05,
      "loss": 0.11818695068359375,
      "step": 297
    },
    {
      "epoch": 0.04152441998188532,
      "grad_norm": 0.8714953660964966,
      "learning_rate": 2.3203125000000003e-05,
      "loss": 0.09313583374023438,
      "step": 298
    },
    {
      "epoch": 0.0416637636730997,
      "grad_norm": 6.001412391662598,
      "learning_rate": 2.3281250000000003e-05,
      "loss": 0.09900283813476562,
      "step": 299
    },
    {
      "epoch": 0.04180310736431408,
      "grad_norm": 7.798113822937012,
      "learning_rate": 2.3359375e-05,
      "loss": 0.17302322387695312,
      "step": 300
    },
    {
      "epoch": 0.04194245105552846,
      "grad_norm": 6.590585231781006,
      "learning_rate": 2.34375e-05,
      "loss": 0.18585968017578125,
      "step": 301
    },
    {
      "epoch": 0.04208179474674284,
      "grad_norm": 1.1921802759170532,
      "learning_rate": 2.3515625e-05,
      "loss": 0.1658782958984375,
      "step": 302
    },
    {
      "epoch": 0.042221138437957224,
      "grad_norm": 4.3022942543029785,
      "learning_rate": 2.359375e-05,
      "loss": 0.11975669860839844,
      "step": 303
    },
    {
      "epoch": 0.0423604821291716,
      "grad_norm": 13.332063674926758,
      "learning_rate": 2.3671875e-05,
      "loss": 0.2271881103515625,
      "step": 304
    },
    {
      "epoch": 0.04249982582038598,
      "grad_norm": 2.675811290740967,
      "learning_rate": 2.375e-05,
      "loss": 0.12166213989257812,
      "step": 305
    },
    {
      "epoch": 0.042639169511600364,
      "grad_norm": 9.621895790100098,
      "learning_rate": 2.3828125e-05,
      "loss": 0.21872711181640625,
      "step": 306
    },
    {
      "epoch": 0.04277851320281474,
      "grad_norm": 16.944185256958008,
      "learning_rate": 2.3906250000000002e-05,
      "loss": 0.19298171997070312,
      "step": 307
    },
    {
      "epoch": 0.042917856894029126,
      "grad_norm": 7.998727321624756,
      "learning_rate": 2.3984375000000002e-05,
      "loss": 0.14169692993164062,
      "step": 308
    },
    {
      "epoch": 0.0430572005852435,
      "grad_norm": 4.063196182250977,
      "learning_rate": 2.4062500000000002e-05,
      "loss": 0.07884597778320312,
      "step": 309
    },
    {
      "epoch": 0.04319654427645788,
      "grad_norm": 3.240393877029419,
      "learning_rate": 2.4140625000000002e-05,
      "loss": 0.1863555908203125,
      "step": 310
    },
    {
      "epoch": 0.043335887967672265,
      "grad_norm": 1.6797086000442505,
      "learning_rate": 2.4218750000000003e-05,
      "loss": 0.14199447631835938,
      "step": 311
    },
    {
      "epoch": 0.04347523165888664,
      "grad_norm": 13.882161140441895,
      "learning_rate": 2.4296875000000003e-05,
      "loss": 0.22545623779296875,
      "step": 312
    },
    {
      "epoch": 0.04361457535010103,
      "grad_norm": 6.29648494720459,
      "learning_rate": 2.4375000000000003e-05,
      "loss": 0.14227676391601562,
      "step": 313
    },
    {
      "epoch": 0.043753919041315405,
      "grad_norm": 5.53711462020874,
      "learning_rate": 2.4453125000000003e-05,
      "loss": 0.13063430786132812,
      "step": 314
    },
    {
      "epoch": 0.04389326273252978,
      "grad_norm": 9.892417907714844,
      "learning_rate": 2.4531250000000003e-05,
      "loss": 0.15546798706054688,
      "step": 315
    },
    {
      "epoch": 0.04403260642374417,
      "grad_norm": 1.3137129545211792,
      "learning_rate": 2.4609375000000004e-05,
      "loss": 0.12632369995117188,
      "step": 316
    },
    {
      "epoch": 0.044171950114958544,
      "grad_norm": 13.980039596557617,
      "learning_rate": 2.46875e-05,
      "loss": 0.1938323974609375,
      "step": 317
    },
    {
      "epoch": 0.04431129380617293,
      "grad_norm": 13.280157089233398,
      "learning_rate": 2.4765625e-05,
      "loss": 0.16693878173828125,
      "step": 318
    },
    {
      "epoch": 0.044450637497387306,
      "grad_norm": 3.1769421100616455,
      "learning_rate": 2.484375e-05,
      "loss": 0.12438201904296875,
      "step": 319
    },
    {
      "epoch": 0.044589981188601684,
      "grad_norm": 6.686270236968994,
      "learning_rate": 2.4921875e-05,
      "loss": 0.12722015380859375,
      "step": 320
    },
    {
      "epoch": 0.04472932487981607,
      "grad_norm": 6.62825870513916,
      "learning_rate": 2.5e-05,
      "loss": 0.1279144287109375,
      "step": 321
    },
    {
      "epoch": 0.044868668571030446,
      "grad_norm": 4.808536529541016,
      "learning_rate": 2.5078125e-05,
      "loss": 0.12734603881835938,
      "step": 322
    },
    {
      "epoch": 0.04500801226224483,
      "grad_norm": 6.01970100402832,
      "learning_rate": 2.515625e-05,
      "loss": 0.160400390625,
      "step": 323
    },
    {
      "epoch": 0.04514735595345921,
      "grad_norm": 5.805986404418945,
      "learning_rate": 2.5234375000000002e-05,
      "loss": 0.106903076171875,
      "step": 324
    },
    {
      "epoch": 0.045286699644673585,
      "grad_norm": 7.291723728179932,
      "learning_rate": 2.5312500000000002e-05,
      "loss": 0.16937255859375,
      "step": 325
    },
    {
      "epoch": 0.04542604333588797,
      "grad_norm": 4.367210388183594,
      "learning_rate": 2.5390625000000002e-05,
      "loss": 0.16176605224609375,
      "step": 326
    },
    {
      "epoch": 0.04556538702710235,
      "grad_norm": 5.535088539123535,
      "learning_rate": 2.5468750000000002e-05,
      "loss": 0.10674285888671875,
      "step": 327
    },
    {
      "epoch": 0.04570473071831673,
      "grad_norm": 6.321102619171143,
      "learning_rate": 2.5546875000000003e-05,
      "loss": 0.17055892944335938,
      "step": 328
    },
    {
      "epoch": 0.04584407440953111,
      "grad_norm": 1.4896820783615112,
      "learning_rate": 2.5625000000000003e-05,
      "loss": 0.1331329345703125,
      "step": 329
    },
    {
      "epoch": 0.045983418100745486,
      "grad_norm": 4.083991527557373,
      "learning_rate": 2.5703125000000003e-05,
      "loss": 0.15977096557617188,
      "step": 330
    },
    {
      "epoch": 0.04612276179195987,
      "grad_norm": 3.0442798137664795,
      "learning_rate": 2.5781250000000003e-05,
      "loss": 0.1445465087890625,
      "step": 331
    },
    {
      "epoch": 0.04626210548317425,
      "grad_norm": 1.5406112670898438,
      "learning_rate": 2.5859375000000003e-05,
      "loss": 0.11664199829101562,
      "step": 332
    },
    {
      "epoch": 0.04640144917438863,
      "grad_norm": 3.74257493019104,
      "learning_rate": 2.5937500000000004e-05,
      "loss": 0.11249160766601562,
      "step": 333
    },
    {
      "epoch": 0.04654079286560301,
      "grad_norm": 2.8007924556732178,
      "learning_rate": 2.6015625e-05,
      "loss": 0.13175392150878906,
      "step": 334
    },
    {
      "epoch": 0.04668013655681739,
      "grad_norm": 2.975522041320801,
      "learning_rate": 2.609375e-05,
      "loss": 0.142822265625,
      "step": 335
    },
    {
      "epoch": 0.04681948024803177,
      "grad_norm": 2.471824884414673,
      "learning_rate": 2.6171875e-05,
      "loss": 0.1080474853515625,
      "step": 336
    },
    {
      "epoch": 0.04695882393924615,
      "grad_norm": 0.7004625797271729,
      "learning_rate": 2.625e-05,
      "loss": 0.09199905395507812,
      "step": 337
    },
    {
      "epoch": 0.047098167630460534,
      "grad_norm": 2.7238762378692627,
      "learning_rate": 2.6328125e-05,
      "loss": 0.1332111358642578,
      "step": 338
    },
    {
      "epoch": 0.04723751132167491,
      "grad_norm": 5.013106346130371,
      "learning_rate": 2.640625e-05,
      "loss": 0.11324310302734375,
      "step": 339
    },
    {
      "epoch": 0.04737685501288929,
      "grad_norm": 3.931610584259033,
      "learning_rate": 2.6484375000000002e-05,
      "loss": 0.10091400146484375,
      "step": 340
    },
    {
      "epoch": 0.047516198704103674,
      "grad_norm": 2.0242345333099365,
      "learning_rate": 2.6562500000000002e-05,
      "loss": 0.13353347778320312,
      "step": 341
    },
    {
      "epoch": 0.04765554239531805,
      "grad_norm": 2.6253623962402344,
      "learning_rate": 2.6640625000000002e-05,
      "loss": 0.15896224975585938,
      "step": 342
    },
    {
      "epoch": 0.047794886086532436,
      "grad_norm": 1.82095468044281,
      "learning_rate": 2.6718750000000002e-05,
      "loss": 0.11186981201171875,
      "step": 343
    },
    {
      "epoch": 0.04793422977774681,
      "grad_norm": 1.6914509534835815,
      "learning_rate": 2.6796875000000003e-05,
      "loss": 0.14142608642578125,
      "step": 344
    },
    {
      "epoch": 0.04807357346896119,
      "grad_norm": 2.262786626815796,
      "learning_rate": 2.6875000000000003e-05,
      "loss": 0.120819091796875,
      "step": 345
    },
    {
      "epoch": 0.048212917160175575,
      "grad_norm": 2.7549540996551514,
      "learning_rate": 2.6953125000000003e-05,
      "loss": 0.10667991638183594,
      "step": 346
    },
    {
      "epoch": 0.04835226085138995,
      "grad_norm": 3.977987051010132,
      "learning_rate": 2.7031250000000003e-05,
      "loss": 0.196563720703125,
      "step": 347
    },
    {
      "epoch": 0.04849160454260433,
      "grad_norm": 1.7890863418579102,
      "learning_rate": 2.7109375000000003e-05,
      "loss": 0.12773895263671875,
      "step": 348
    },
    {
      "epoch": 0.048630948233818715,
      "grad_norm": 10.143086433410645,
      "learning_rate": 2.7187500000000004e-05,
      "loss": 0.18898773193359375,
      "step": 349
    },
    {
      "epoch": 0.04877029192503309,
      "grad_norm": 7.500576972961426,
      "learning_rate": 2.7265625000000004e-05,
      "loss": 0.13497543334960938,
      "step": 350
    },
    {
      "epoch": 0.04890963561624748,
      "grad_norm": 4.198714256286621,
      "learning_rate": 2.734375e-05,
      "loss": 0.1424713134765625,
      "step": 351
    },
    {
      "epoch": 0.049048979307461854,
      "grad_norm": 1.4235715866088867,
      "learning_rate": 2.7421875e-05,
      "loss": 0.10097122192382812,
      "step": 352
    },
    {
      "epoch": 0.04918832299867623,
      "grad_norm": 1.9646923542022705,
      "learning_rate": 2.75e-05,
      "loss": 0.0894927978515625,
      "step": 353
    },
    {
      "epoch": 0.049327666689890616,
      "grad_norm": 3.6632654666900635,
      "learning_rate": 2.7578125e-05,
      "loss": 0.1464996337890625,
      "step": 354
    },
    {
      "epoch": 0.049467010381104994,
      "grad_norm": 5.117610931396484,
      "learning_rate": 2.765625e-05,
      "loss": 0.115447998046875,
      "step": 355
    },
    {
      "epoch": 0.04960635407231938,
      "grad_norm": 3.7435388565063477,
      "learning_rate": 2.7734375e-05,
      "loss": 0.18556594848632812,
      "step": 356
    },
    {
      "epoch": 0.049745697763533755,
      "grad_norm": 2.6390440464019775,
      "learning_rate": 2.7812500000000002e-05,
      "loss": 0.14746856689453125,
      "step": 357
    },
    {
      "epoch": 0.04988504145474813,
      "grad_norm": 1.234492301940918,
      "learning_rate": 2.7890625000000002e-05,
      "loss": 0.11200141906738281,
      "step": 358
    },
    {
      "epoch": 0.05002438514596252,
      "grad_norm": 4.065972328186035,
      "learning_rate": 2.7968750000000002e-05,
      "loss": 0.10887908935546875,
      "step": 359
    },
    {
      "epoch": 0.050163728837176895,
      "grad_norm": 9.417527198791504,
      "learning_rate": 2.8046875000000002e-05,
      "loss": 0.21554946899414062,
      "step": 360
    },
    {
      "epoch": 0.05030307252839128,
      "grad_norm": 2.6062357425689697,
      "learning_rate": 2.8125000000000003e-05,
      "loss": 0.12706375122070312,
      "step": 361
    },
    {
      "epoch": 0.05044241621960566,
      "grad_norm": 4.077699661254883,
      "learning_rate": 2.8203125000000003e-05,
      "loss": 0.1599864959716797,
      "step": 362
    },
    {
      "epoch": 0.050581759910820034,
      "grad_norm": 1.8388473987579346,
      "learning_rate": 2.8281250000000003e-05,
      "loss": 0.13700485229492188,
      "step": 363
    },
    {
      "epoch": 0.05072110360203442,
      "grad_norm": 1.6937845945358276,
      "learning_rate": 2.8359375000000003e-05,
      "loss": 0.122955322265625,
      "step": 364
    },
    {
      "epoch": 0.050860447293248796,
      "grad_norm": 0.7536792159080505,
      "learning_rate": 2.8437500000000003e-05,
      "loss": 0.08925247192382812,
      "step": 365
    },
    {
      "epoch": 0.05099979098446318,
      "grad_norm": 7.48211145401001,
      "learning_rate": 2.8515625000000004e-05,
      "loss": 0.15163040161132812,
      "step": 366
    },
    {
      "epoch": 0.05113913467567756,
      "grad_norm": 3.3823533058166504,
      "learning_rate": 2.8593750000000004e-05,
      "loss": 0.12523651123046875,
      "step": 367
    },
    {
      "epoch": 0.051278478366891936,
      "grad_norm": 3.116825819015503,
      "learning_rate": 2.8671875e-05,
      "loss": 0.13550949096679688,
      "step": 368
    },
    {
      "epoch": 0.05141782205810632,
      "grad_norm": 5.276050090789795,
      "learning_rate": 2.875e-05,
      "loss": 0.1670074462890625,
      "step": 369
    },
    {
      "epoch": 0.0515571657493207,
      "grad_norm": 2.1061322689056396,
      "learning_rate": 2.8828125e-05,
      "loss": 0.16495513916015625,
      "step": 370
    },
    {
      "epoch": 0.05169650944053508,
      "grad_norm": 1.6421290636062622,
      "learning_rate": 2.890625e-05,
      "loss": 0.13971328735351562,
      "step": 371
    },
    {
      "epoch": 0.05183585313174946,
      "grad_norm": 0.5310415029525757,
      "learning_rate": 2.8984375e-05,
      "loss": 0.07225799560546875,
      "step": 372
    },
    {
      "epoch": 0.05197519682296384,
      "grad_norm": 5.098465919494629,
      "learning_rate": 2.90625e-05,
      "loss": 0.1422252655029297,
      "step": 373
    },
    {
      "epoch": 0.05211454051417822,
      "grad_norm": 3.1248228549957275,
      "learning_rate": 2.9140625000000002e-05,
      "loss": 0.14033889770507812,
      "step": 374
    },
    {
      "epoch": 0.0522538842053926,
      "grad_norm": 2.333101987838745,
      "learning_rate": 2.9218750000000002e-05,
      "loss": 0.14675140380859375,
      "step": 375
    },
    {
      "epoch": 0.052393227896606984,
      "grad_norm": 4.245767593383789,
      "learning_rate": 2.9296875000000002e-05,
      "loss": 0.17270660400390625,
      "step": 376
    },
    {
      "epoch": 0.05253257158782136,
      "grad_norm": 1.8721506595611572,
      "learning_rate": 2.9375000000000003e-05,
      "loss": 0.13094711303710938,
      "step": 377
    },
    {
      "epoch": 0.05267191527903574,
      "grad_norm": 4.1797637939453125,
      "learning_rate": 2.9453125000000003e-05,
      "loss": 0.13178253173828125,
      "step": 378
    },
    {
      "epoch": 0.05281125897025012,
      "grad_norm": 2.8279387950897217,
      "learning_rate": 2.9531250000000003e-05,
      "loss": 0.20691680908203125,
      "step": 379
    },
    {
      "epoch": 0.0529506026614645,
      "grad_norm": 1.7068419456481934,
      "learning_rate": 2.9609375000000003e-05,
      "loss": 0.1299285888671875,
      "step": 380
    },
    {
      "epoch": 0.053089946352678885,
      "grad_norm": 2.8568222522735596,
      "learning_rate": 2.9687500000000003e-05,
      "loss": 0.11894607543945312,
      "step": 381
    },
    {
      "epoch": 0.05322929004389326,
      "grad_norm": 2.934680223464966,
      "learning_rate": 2.9765625000000004e-05,
      "loss": 0.13636016845703125,
      "step": 382
    },
    {
      "epoch": 0.05336863373510764,
      "grad_norm": 6.454765796661377,
      "learning_rate": 2.9843750000000004e-05,
      "loss": 0.13864517211914062,
      "step": 383
    },
    {
      "epoch": 0.053507977426322025,
      "grad_norm": 4.41960334777832,
      "learning_rate": 2.9921875000000004e-05,
      "loss": 0.11544609069824219,
      "step": 384
    },
    {
      "epoch": 0.0536473211175364,
      "grad_norm": 2.246563196182251,
      "learning_rate": 3.0000000000000004e-05,
      "loss": 0.10828399658203125,
      "step": 385
    },
    {
      "epoch": 0.053786664808750786,
      "grad_norm": 5.776893138885498,
      "learning_rate": 3.0078125e-05,
      "loss": 0.15362930297851562,
      "step": 386
    },
    {
      "epoch": 0.053926008499965164,
      "grad_norm": 3.4696993827819824,
      "learning_rate": 3.015625e-05,
      "loss": 0.12880706787109375,
      "step": 387
    },
    {
      "epoch": 0.05406535219117954,
      "grad_norm": 6.263038158416748,
      "learning_rate": 3.0234375e-05,
      "loss": 0.21252822875976562,
      "step": 388
    },
    {
      "epoch": 0.054204695882393926,
      "grad_norm": 4.023487091064453,
      "learning_rate": 3.03125e-05,
      "loss": 0.10901260375976562,
      "step": 389
    },
    {
      "epoch": 0.0543440395736083,
      "grad_norm": 1.549216628074646,
      "learning_rate": 3.0390625000000002e-05,
      "loss": 0.1363372802734375,
      "step": 390
    },
    {
      "epoch": 0.05448338326482269,
      "grad_norm": 3.0464742183685303,
      "learning_rate": 3.0468750000000002e-05,
      "loss": 0.11837959289550781,
      "step": 391
    },
    {
      "epoch": 0.054622726956037065,
      "grad_norm": 3.3431432247161865,
      "learning_rate": 3.0546875e-05,
      "loss": 0.13915252685546875,
      "step": 392
    },
    {
      "epoch": 0.05476207064725144,
      "grad_norm": 1.6485289335250854,
      "learning_rate": 3.0625000000000006e-05,
      "loss": 0.13364219665527344,
      "step": 393
    },
    {
      "epoch": 0.05490141433846583,
      "grad_norm": 1.2598912715911865,
      "learning_rate": 3.0703125e-05,
      "loss": 0.12411880493164062,
      "step": 394
    },
    {
      "epoch": 0.055040758029680205,
      "grad_norm": 12.061439514160156,
      "learning_rate": 3.078125e-05,
      "loss": 0.16320037841796875,
      "step": 395
    },
    {
      "epoch": 0.05518010172089459,
      "grad_norm": 6.43983793258667,
      "learning_rate": 3.0859375e-05,
      "loss": 0.11378097534179688,
      "step": 396
    },
    {
      "epoch": 0.05531944541210897,
      "grad_norm": 4.606500148773193,
      "learning_rate": 3.09375e-05,
      "loss": 0.14632034301757812,
      "step": 397
    },
    {
      "epoch": 0.055458789103323344,
      "grad_norm": 6.551544666290283,
      "learning_rate": 3.1015625000000003e-05,
      "loss": 0.15517234802246094,
      "step": 398
    },
    {
      "epoch": 0.05559813279453773,
      "grad_norm": 2.759211778640747,
      "learning_rate": 3.109375e-05,
      "loss": 0.11321258544921875,
      "step": 399
    },
    {
      "epoch": 0.055737476485752106,
      "grad_norm": 1.0862674713134766,
      "learning_rate": 3.1171875000000004e-05,
      "loss": 0.07699394226074219,
      "step": 400
    },
    {
      "epoch": 0.05587682017696649,
      "grad_norm": 5.297611236572266,
      "learning_rate": 3.125e-05,
      "loss": 0.179534912109375,
      "step": 401
    },
    {
      "epoch": 0.05601616386818087,
      "grad_norm": 1.2492098808288574,
      "learning_rate": 3.1328125000000004e-05,
      "loss": 0.08292007446289062,
      "step": 402
    },
    {
      "epoch": 0.056155507559395246,
      "grad_norm": 7.252022743225098,
      "learning_rate": 3.140625e-05,
      "loss": 0.18150711059570312,
      "step": 403
    },
    {
      "epoch": 0.05629485125060963,
      "grad_norm": 1.6567667722702026,
      "learning_rate": 3.1484375000000005e-05,
      "loss": 0.09032630920410156,
      "step": 404
    },
    {
      "epoch": 0.05643419494182401,
      "grad_norm": 1.7720317840576172,
      "learning_rate": 3.15625e-05,
      "loss": 0.1399078369140625,
      "step": 405
    },
    {
      "epoch": 0.05657353863303839,
      "grad_norm": 4.111324787139893,
      "learning_rate": 3.1640625000000005e-05,
      "loss": 0.11572647094726562,
      "step": 406
    },
    {
      "epoch": 0.05671288232425277,
      "grad_norm": 4.905533790588379,
      "learning_rate": 3.171875e-05,
      "loss": 0.162567138671875,
      "step": 407
    },
    {
      "epoch": 0.05685222601546715,
      "grad_norm": 1.1801806688308716,
      "learning_rate": 3.1796875000000005e-05,
      "loss": 0.11251449584960938,
      "step": 408
    },
    {
      "epoch": 0.05699156970668153,
      "grad_norm": 3.205890417098999,
      "learning_rate": 3.1875e-05,
      "loss": 0.11421585083007812,
      "step": 409
    },
    {
      "epoch": 0.05713091339789591,
      "grad_norm": 3.4559805393218994,
      "learning_rate": 3.1953125000000006e-05,
      "loss": 0.10849380493164062,
      "step": 410
    },
    {
      "epoch": 0.057270257089110294,
      "grad_norm": 2.342633008956909,
      "learning_rate": 3.203125e-05,
      "loss": 0.1396617889404297,
      "step": 411
    },
    {
      "epoch": 0.05740960078032467,
      "grad_norm": 2.9076180458068848,
      "learning_rate": 3.2109375e-05,
      "loss": 0.111907958984375,
      "step": 412
    },
    {
      "epoch": 0.05754894447153905,
      "grad_norm": 5.0726237297058105,
      "learning_rate": 3.21875e-05,
      "loss": 0.13541793823242188,
      "step": 413
    },
    {
      "epoch": 0.05768828816275343,
      "grad_norm": 3.734821319580078,
      "learning_rate": 3.2265625e-05,
      "loss": 0.14260101318359375,
      "step": 414
    },
    {
      "epoch": 0.05782763185396781,
      "grad_norm": 1.1307181119918823,
      "learning_rate": 3.2343750000000004e-05,
      "loss": 0.11335182189941406,
      "step": 415
    },
    {
      "epoch": 0.057966975545182195,
      "grad_norm": 2.9182047843933105,
      "learning_rate": 3.2421875e-05,
      "loss": 0.10137557983398438,
      "step": 416
    },
    {
      "epoch": 0.05810631923639657,
      "grad_norm": 2.3167152404785156,
      "learning_rate": 3.2500000000000004e-05,
      "loss": 0.13050460815429688,
      "step": 417
    },
    {
      "epoch": 0.05824566292761095,
      "grad_norm": 2.6518657207489014,
      "learning_rate": 3.2578125e-05,
      "loss": 0.16229629516601562,
      "step": 418
    },
    {
      "epoch": 0.058385006618825334,
      "grad_norm": 3.3731040954589844,
      "learning_rate": 3.2656250000000004e-05,
      "loss": 0.1067962646484375,
      "step": 419
    },
    {
      "epoch": 0.05852435031003971,
      "grad_norm": 5.558904647827148,
      "learning_rate": 3.2734375e-05,
      "loss": 0.13186264038085938,
      "step": 420
    },
    {
      "epoch": 0.058663694001254096,
      "grad_norm": 3.7551989555358887,
      "learning_rate": 3.2812500000000005e-05,
      "loss": 0.11958885192871094,
      "step": 421
    },
    {
      "epoch": 0.058803037692468474,
      "grad_norm": 1.2587206363677979,
      "learning_rate": 3.2890625e-05,
      "loss": 0.08900070190429688,
      "step": 422
    },
    {
      "epoch": 0.05894238138368285,
      "grad_norm": 3.7682955265045166,
      "learning_rate": 3.2968750000000005e-05,
      "loss": 0.12882614135742188,
      "step": 423
    },
    {
      "epoch": 0.059081725074897236,
      "grad_norm": 3.5053179264068604,
      "learning_rate": 3.3046875e-05,
      "loss": 0.15697860717773438,
      "step": 424
    },
    {
      "epoch": 0.05922106876611161,
      "grad_norm": 5.757637977600098,
      "learning_rate": 3.3125000000000006e-05,
      "loss": 0.20546913146972656,
      "step": 425
    },
    {
      "epoch": 0.059360412457326,
      "grad_norm": 4.957822322845459,
      "learning_rate": 3.3203125e-05,
      "loss": 0.21409988403320312,
      "step": 426
    },
    {
      "epoch": 0.059499756148540375,
      "grad_norm": 4.916723251342773,
      "learning_rate": 3.3281250000000006e-05,
      "loss": 0.13035202026367188,
      "step": 427
    },
    {
      "epoch": 0.05963909983975475,
      "grad_norm": 6.045250415802002,
      "learning_rate": 3.3359375e-05,
      "loss": 0.15678024291992188,
      "step": 428
    },
    {
      "epoch": 0.05977844353096914,
      "grad_norm": 2.259713649749756,
      "learning_rate": 3.34375e-05,
      "loss": 0.16791152954101562,
      "step": 429
    },
    {
      "epoch": 0.059917787222183515,
      "grad_norm": 5.927862167358398,
      "learning_rate": 3.3515625e-05,
      "loss": 0.10385513305664062,
      "step": 430
    },
    {
      "epoch": 0.0600571309133979,
      "grad_norm": 6.808668613433838,
      "learning_rate": 3.359375e-05,
      "loss": 0.13460254669189453,
      "step": 431
    },
    {
      "epoch": 0.06019647460461228,
      "grad_norm": 2.4134442806243896,
      "learning_rate": 3.3671875000000004e-05,
      "loss": 0.1080322265625,
      "step": 432
    },
    {
      "epoch": 0.060335818295826654,
      "grad_norm": 2.871001958847046,
      "learning_rate": 3.375e-05,
      "loss": 0.10614013671875,
      "step": 433
    },
    {
      "epoch": 0.06047516198704104,
      "grad_norm": 4.3150434494018555,
      "learning_rate": 3.3828125000000004e-05,
      "loss": 0.18203353881835938,
      "step": 434
    },
    {
      "epoch": 0.060614505678255416,
      "grad_norm": 2.9419655799865723,
      "learning_rate": 3.390625e-05,
      "loss": 0.10440826416015625,
      "step": 435
    },
    {
      "epoch": 0.0607538493694698,
      "grad_norm": 1.0252058506011963,
      "learning_rate": 3.3984375000000004e-05,
      "loss": 0.10599517822265625,
      "step": 436
    },
    {
      "epoch": 0.06089319306068418,
      "grad_norm": 5.004328727722168,
      "learning_rate": 3.40625e-05,
      "loss": 0.13195419311523438,
      "step": 437
    },
    {
      "epoch": 0.061032536751898556,
      "grad_norm": 2.9312875270843506,
      "learning_rate": 3.4140625000000005e-05,
      "loss": 0.10009765625,
      "step": 438
    },
    {
      "epoch": 0.06117188044311294,
      "grad_norm": 1.5946803092956543,
      "learning_rate": 3.421875e-05,
      "loss": 0.08932876586914062,
      "step": 439
    },
    {
      "epoch": 0.06131122413432732,
      "grad_norm": 2.1674325466156006,
      "learning_rate": 3.4296875000000005e-05,
      "loss": 0.1359996795654297,
      "step": 440
    },
    {
      "epoch": 0.0614505678255417,
      "grad_norm": 1.8363351821899414,
      "learning_rate": 3.4375e-05,
      "loss": 0.12226295471191406,
      "step": 441
    },
    {
      "epoch": 0.06158991151675608,
      "grad_norm": 3.685462474822998,
      "learning_rate": 3.4453125000000006e-05,
      "loss": 0.1614227294921875,
      "step": 442
    },
    {
      "epoch": 0.06172925520797046,
      "grad_norm": 1.6046937704086304,
      "learning_rate": 3.453125e-05,
      "loss": 0.149322509765625,
      "step": 443
    },
    {
      "epoch": 0.06186859889918484,
      "grad_norm": 2.0443170070648193,
      "learning_rate": 3.4609375000000006e-05,
      "loss": 0.14379310607910156,
      "step": 444
    },
    {
      "epoch": 0.06200794259039922,
      "grad_norm": 2.646277904510498,
      "learning_rate": 3.46875e-05,
      "loss": 0.17947006225585938,
      "step": 445
    },
    {
      "epoch": 0.062147286281613597,
      "grad_norm": 1.5388216972351074,
      "learning_rate": 3.4765625e-05,
      "loss": 0.10263633728027344,
      "step": 446
    },
    {
      "epoch": 0.06228662997282798,
      "grad_norm": 1.6689749956130981,
      "learning_rate": 3.484375e-05,
      "loss": 0.11783790588378906,
      "step": 447
    },
    {
      "epoch": 0.06242597366404236,
      "grad_norm": 1.6699832677841187,
      "learning_rate": 3.4921875e-05,
      "loss": 0.11031818389892578,
      "step": 448
    },
    {
      "epoch": 0.06256531735525674,
      "grad_norm": 1.9755361080169678,
      "learning_rate": 3.5000000000000004e-05,
      "loss": 0.10696601867675781,
      "step": 449
    },
    {
      "epoch": 0.06270466104647113,
      "grad_norm": 3.1808133125305176,
      "learning_rate": 3.5078125e-05,
      "loss": 0.20627593994140625,
      "step": 450
    },
    {
      "epoch": 0.0628440047376855,
      "grad_norm": 4.196004390716553,
      "learning_rate": 3.5156250000000004e-05,
      "loss": 0.1648101806640625,
      "step": 451
    },
    {
      "epoch": 0.06298334842889988,
      "grad_norm": 5.337859153747559,
      "learning_rate": 3.5234375e-05,
      "loss": 0.16832733154296875,
      "step": 452
    },
    {
      "epoch": 0.06312269212011426,
      "grad_norm": 1.4184443950653076,
      "learning_rate": 3.5312500000000005e-05,
      "loss": 0.11067008972167969,
      "step": 453
    },
    {
      "epoch": 0.06326203581132864,
      "grad_norm": 1.2673379182815552,
      "learning_rate": 3.5390625e-05,
      "loss": 0.10303115844726562,
      "step": 454
    },
    {
      "epoch": 0.06340137950254303,
      "grad_norm": 1.0493322610855103,
      "learning_rate": 3.5468750000000005e-05,
      "loss": 0.08939552307128906,
      "step": 455
    },
    {
      "epoch": 0.0635407231937574,
      "grad_norm": 0.7348464131355286,
      "learning_rate": 3.5546875e-05,
      "loss": 0.08004379272460938,
      "step": 456
    },
    {
      "epoch": 0.06368006688497178,
      "grad_norm": 1.8887932300567627,
      "learning_rate": 3.5625000000000005e-05,
      "loss": 0.11274528503417969,
      "step": 457
    },
    {
      "epoch": 0.06381941057618616,
      "grad_norm": 4.843327522277832,
      "learning_rate": 3.5703125e-05,
      "loss": 0.1579742431640625,
      "step": 458
    },
    {
      "epoch": 0.06395875426740054,
      "grad_norm": 4.328675746917725,
      "learning_rate": 3.5781250000000006e-05,
      "loss": 0.1220855712890625,
      "step": 459
    },
    {
      "epoch": 0.06409809795861493,
      "grad_norm": 3.246380567550659,
      "learning_rate": 3.5859375e-05,
      "loss": 0.157257080078125,
      "step": 460
    },
    {
      "epoch": 0.06423744164982931,
      "grad_norm": 1.911177635192871,
      "learning_rate": 3.5937500000000006e-05,
      "loss": 0.14912033081054688,
      "step": 461
    },
    {
      "epoch": 0.06437678534104369,
      "grad_norm": 6.6284942626953125,
      "learning_rate": 3.6015625e-05,
      "loss": 0.12305450439453125,
      "step": 462
    },
    {
      "epoch": 0.06451612903225806,
      "grad_norm": 3.0399420261383057,
      "learning_rate": 3.609375e-05,
      "loss": 0.09240055084228516,
      "step": 463
    },
    {
      "epoch": 0.06465547272347244,
      "grad_norm": 2.766639471054077,
      "learning_rate": 3.6171875000000003e-05,
      "loss": 0.09424209594726562,
      "step": 464
    },
    {
      "epoch": 0.06479481641468683,
      "grad_norm": 1.9291465282440186,
      "learning_rate": 3.625e-05,
      "loss": 0.09503936767578125,
      "step": 465
    },
    {
      "epoch": 0.06493416010590121,
      "grad_norm": 3.721626043319702,
      "learning_rate": 3.6328125000000004e-05,
      "loss": 0.13573837280273438,
      "step": 466
    },
    {
      "epoch": 0.06507350379711559,
      "grad_norm": 4.3513617515563965,
      "learning_rate": 3.640625e-05,
      "loss": 0.189666748046875,
      "step": 467
    },
    {
      "epoch": 0.06521284748832996,
      "grad_norm": 1.758995771408081,
      "learning_rate": 3.6484375000000004e-05,
      "loss": 0.09131813049316406,
      "step": 468
    },
    {
      "epoch": 0.06535219117954434,
      "grad_norm": 1.7660777568817139,
      "learning_rate": 3.65625e-05,
      "loss": 0.10843467712402344,
      "step": 469
    },
    {
      "epoch": 0.06549153487075873,
      "grad_norm": 1.3174102306365967,
      "learning_rate": 3.6640625000000005e-05,
      "loss": 0.08036994934082031,
      "step": 470
    },
    {
      "epoch": 0.06563087856197311,
      "grad_norm": 2.3893778324127197,
      "learning_rate": 3.671875e-05,
      "loss": 0.13868331909179688,
      "step": 471
    },
    {
      "epoch": 0.06577022225318749,
      "grad_norm": 3.097043514251709,
      "learning_rate": 3.6796875000000005e-05,
      "loss": 0.1457672119140625,
      "step": 472
    },
    {
      "epoch": 0.06590956594440187,
      "grad_norm": 7.362919807434082,
      "learning_rate": 3.6875e-05,
      "loss": 0.15611648559570312,
      "step": 473
    },
    {
      "epoch": 0.06604890963561624,
      "grad_norm": 4.099730014801025,
      "learning_rate": 3.6953125000000005e-05,
      "loss": 0.10845565795898438,
      "step": 474
    },
    {
      "epoch": 0.06618825332683063,
      "grad_norm": 3.938004493713379,
      "learning_rate": 3.703125e-05,
      "loss": 0.13916778564453125,
      "step": 475
    },
    {
      "epoch": 0.06632759701804501,
      "grad_norm": 1.572511076927185,
      "learning_rate": 3.7109375000000006e-05,
      "loss": 0.0987396240234375,
      "step": 476
    },
    {
      "epoch": 0.06646694070925939,
      "grad_norm": 1.9598009586334229,
      "learning_rate": 3.71875e-05,
      "loss": 0.1291370391845703,
      "step": 477
    },
    {
      "epoch": 0.06660628440047377,
      "grad_norm": 0.9886524677276611,
      "learning_rate": 3.7265625000000006e-05,
      "loss": 0.08448982238769531,
      "step": 478
    },
    {
      "epoch": 0.06674562809168814,
      "grad_norm": 3.3393375873565674,
      "learning_rate": 3.734375e-05,
      "loss": 0.163421630859375,
      "step": 479
    },
    {
      "epoch": 0.06688497178290254,
      "grad_norm": 3.6082115173339844,
      "learning_rate": 3.7421875e-05,
      "loss": 0.15908050537109375,
      "step": 480
    },
    {
      "epoch": 0.06702431547411691,
      "grad_norm": 4.145823955535889,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.138275146484375,
      "step": 481
    },
    {
      "epoch": 0.06716365916533129,
      "grad_norm": 2.339674234390259,
      "learning_rate": 3.7578125e-05,
      "loss": 0.115234375,
      "step": 482
    },
    {
      "epoch": 0.06730300285654567,
      "grad_norm": 4.810821056365967,
      "learning_rate": 3.7656250000000004e-05,
      "loss": 0.17440032958984375,
      "step": 483
    },
    {
      "epoch": 0.06744234654776005,
      "grad_norm": 3.7228634357452393,
      "learning_rate": 3.7734375e-05,
      "loss": 0.13367462158203125,
      "step": 484
    },
    {
      "epoch": 0.06758169023897444,
      "grad_norm": 1.312095284461975,
      "learning_rate": 3.7812500000000004e-05,
      "loss": 0.14334487915039062,
      "step": 485
    },
    {
      "epoch": 0.06772103393018881,
      "grad_norm": 0.9424468874931335,
      "learning_rate": 3.7890625e-05,
      "loss": 0.09403419494628906,
      "step": 486
    },
    {
      "epoch": 0.06786037762140319,
      "grad_norm": 2.94515061378479,
      "learning_rate": 3.7968750000000005e-05,
      "loss": 0.10296058654785156,
      "step": 487
    },
    {
      "epoch": 0.06799972131261757,
      "grad_norm": 0.9378728866577148,
      "learning_rate": 3.8046875e-05,
      "loss": 0.10338401794433594,
      "step": 488
    },
    {
      "epoch": 0.06813906500383195,
      "grad_norm": 1.9752261638641357,
      "learning_rate": 3.8125000000000005e-05,
      "loss": 0.09709548950195312,
      "step": 489
    },
    {
      "epoch": 0.06827840869504634,
      "grad_norm": 2.080390453338623,
      "learning_rate": 3.8203125e-05,
      "loss": 0.12378692626953125,
      "step": 490
    },
    {
      "epoch": 0.06841775238626072,
      "grad_norm": 2.281527519226074,
      "learning_rate": 3.8281250000000006e-05,
      "loss": 0.11740684509277344,
      "step": 491
    },
    {
      "epoch": 0.0685570960774751,
      "grad_norm": 3.26632022857666,
      "learning_rate": 3.8359375e-05,
      "loss": 0.11805152893066406,
      "step": 492
    },
    {
      "epoch": 0.06869643976868947,
      "grad_norm": 1.0444692373275757,
      "learning_rate": 3.8437500000000006e-05,
      "loss": 0.08564186096191406,
      "step": 493
    },
    {
      "epoch": 0.06883578345990385,
      "grad_norm": 5.619449615478516,
      "learning_rate": 3.8515625e-05,
      "loss": 0.20371627807617188,
      "step": 494
    },
    {
      "epoch": 0.06897512715111823,
      "grad_norm": 2.2947535514831543,
      "learning_rate": 3.8593750000000006e-05,
      "loss": 0.11037445068359375,
      "step": 495
    },
    {
      "epoch": 0.06911447084233262,
      "grad_norm": 3.0819923877716064,
      "learning_rate": 3.8671875e-05,
      "loss": 0.095550537109375,
      "step": 496
    },
    {
      "epoch": 0.069253814533547,
      "grad_norm": 1.4950711727142334,
      "learning_rate": 3.875e-05,
      "loss": 0.07056236267089844,
      "step": 497
    },
    {
      "epoch": 0.06939315822476137,
      "grad_norm": 3.981626033782959,
      "learning_rate": 3.8828125000000004e-05,
      "loss": 0.12194633483886719,
      "step": 498
    },
    {
      "epoch": 0.06953250191597575,
      "grad_norm": 2.9882898330688477,
      "learning_rate": 3.890625e-05,
      "loss": 0.11988258361816406,
      "step": 499
    },
    {
      "epoch": 0.06967184560719013,
      "grad_norm": 4.005314350128174,
      "learning_rate": 3.8984375000000004e-05,
      "loss": 0.14714622497558594,
      "step": 500
    },
    {
      "epoch": 0.06981118929840452,
      "grad_norm": 4.22213077545166,
      "learning_rate": 3.90625e-05,
      "loss": 0.12076568603515625,
      "step": 501
    },
    {
      "epoch": 0.0699505329896189,
      "grad_norm": 2.8216464519500732,
      "learning_rate": 3.9140625000000004e-05,
      "loss": 0.1625518798828125,
      "step": 502
    },
    {
      "epoch": 0.07008987668083327,
      "grad_norm": 1.3072019815444946,
      "learning_rate": 3.921875e-05,
      "loss": 0.11846733093261719,
      "step": 503
    },
    {
      "epoch": 0.07022922037204765,
      "grad_norm": 5.674465656280518,
      "learning_rate": 3.9296875000000005e-05,
      "loss": 0.1237335205078125,
      "step": 504
    },
    {
      "epoch": 0.07036856406326203,
      "grad_norm": 7.615915298461914,
      "learning_rate": 3.9375e-05,
      "loss": 0.1512165069580078,
      "step": 505
    },
    {
      "epoch": 0.07050790775447642,
      "grad_norm": 3.9900150299072266,
      "learning_rate": 3.9453125000000005e-05,
      "loss": 0.13902664184570312,
      "step": 506
    },
    {
      "epoch": 0.0706472514456908,
      "grad_norm": 3.043977737426758,
      "learning_rate": 3.953125e-05,
      "loss": 0.12251091003417969,
      "step": 507
    },
    {
      "epoch": 0.07078659513690518,
      "grad_norm": 2.323931932449341,
      "learning_rate": 3.9609375000000006e-05,
      "loss": 0.11706161499023438,
      "step": 508
    },
    {
      "epoch": 0.07092593882811955,
      "grad_norm": 2.765200614929199,
      "learning_rate": 3.96875e-05,
      "loss": 0.10280609130859375,
      "step": 509
    },
    {
      "epoch": 0.07106528251933393,
      "grad_norm": 3.848862648010254,
      "learning_rate": 3.9765625000000006e-05,
      "loss": 0.13304519653320312,
      "step": 510
    },
    {
      "epoch": 0.07120462621054832,
      "grad_norm": 3.36061429977417,
      "learning_rate": 3.984375e-05,
      "loss": 0.17521286010742188,
      "step": 511
    },
    {
      "epoch": 0.0713439699017627,
      "grad_norm": 2.0747926235198975,
      "learning_rate": 3.9921875000000006e-05,
      "loss": 0.13555145263671875,
      "step": 512
    },
    {
      "epoch": 0.07148331359297708,
      "grad_norm": 3.8085427284240723,
      "learning_rate": 4e-05,
      "loss": 0.14349365234375,
      "step": 513
    },
    {
      "epoch": 0.07162265728419145,
      "grad_norm": 3.5413095951080322,
      "learning_rate": 3.999999777822831e-05,
      "loss": 0.125152587890625,
      "step": 514
    },
    {
      "epoch": 0.07176200097540583,
      "grad_norm": 1.5531997680664062,
      "learning_rate": 3.999999111291371e-05,
      "loss": 0.11183738708496094,
      "step": 515
    },
    {
      "epoch": 0.07190134466662022,
      "grad_norm": 4.626058578491211,
      "learning_rate": 3.9999980004057696e-05,
      "loss": 0.13446426391601562,
      "step": 516
    },
    {
      "epoch": 0.0720406883578346,
      "grad_norm": 8.27481746673584,
      "learning_rate": 3.999996445166274e-05,
      "loss": 0.17162132263183594,
      "step": 517
    },
    {
      "epoch": 0.07218003204904898,
      "grad_norm": 6.6619954109191895,
      "learning_rate": 3.9999944455732284e-05,
      "loss": 0.21932220458984375,
      "step": 518
    },
    {
      "epoch": 0.07231937574026336,
      "grad_norm": 4.600326061248779,
      "learning_rate": 3.9999920016270776e-05,
      "loss": 0.173004150390625,
      "step": 519
    },
    {
      "epoch": 0.07245871943147773,
      "grad_norm": 4.08194637298584,
      "learning_rate": 3.999989113328364e-05,
      "loss": 0.134185791015625,
      "step": 520
    },
    {
      "epoch": 0.07259806312269212,
      "grad_norm": 4.183300971984863,
      "learning_rate": 3.999985780677731e-05,
      "loss": 0.15105056762695312,
      "step": 521
    },
    {
      "epoch": 0.0727374068139065,
      "grad_norm": 1.868992567062378,
      "learning_rate": 3.999982003675918e-05,
      "loss": 0.09186744689941406,
      "step": 522
    },
    {
      "epoch": 0.07287675050512088,
      "grad_norm": 2.105400562286377,
      "learning_rate": 3.999977782323763e-05,
      "loss": 0.1583728790283203,
      "step": 523
    },
    {
      "epoch": 0.07301609419633526,
      "grad_norm": 3.8349478244781494,
      "learning_rate": 3.9999731166222065e-05,
      "loss": 0.14019012451171875,
      "step": 524
    },
    {
      "epoch": 0.07315543788754963,
      "grad_norm": 5.373597621917725,
      "learning_rate": 3.9999680065722826e-05,
      "loss": 0.17078781127929688,
      "step": 525
    },
    {
      "epoch": 0.07329478157876403,
      "grad_norm": 4.404664993286133,
      "learning_rate": 3.999962452175128e-05,
      "loss": 0.15452957153320312,
      "step": 526
    },
    {
      "epoch": 0.0734341252699784,
      "grad_norm": 2.5311670303344727,
      "learning_rate": 3.9999564534319764e-05,
      "loss": 0.10028266906738281,
      "step": 527
    },
    {
      "epoch": 0.07357346896119278,
      "grad_norm": 1.473449945449829,
      "learning_rate": 3.9999500103441604e-05,
      "loss": 0.11912918090820312,
      "step": 528
    },
    {
      "epoch": 0.07371281265240716,
      "grad_norm": 3.460256338119507,
      "learning_rate": 3.999943122913112e-05,
      "loss": 0.2085418701171875,
      "step": 529
    },
    {
      "epoch": 0.07385215634362154,
      "grad_norm": 1.859515905380249,
      "learning_rate": 3.9999357911403613e-05,
      "loss": 0.10864639282226562,
      "step": 530
    },
    {
      "epoch": 0.07399150003483593,
      "grad_norm": 0.9655384421348572,
      "learning_rate": 3.9999280150275375e-05,
      "loss": 0.10744380950927734,
      "step": 531
    },
    {
      "epoch": 0.0741308437260503,
      "grad_norm": 0.957709550857544,
      "learning_rate": 3.999919794576367e-05,
      "loss": 0.13984298706054688,
      "step": 532
    },
    {
      "epoch": 0.07427018741726468,
      "grad_norm": 2.0696256160736084,
      "learning_rate": 3.9999111297886774e-05,
      "loss": 0.08688831329345703,
      "step": 533
    },
    {
      "epoch": 0.07440953110847906,
      "grad_norm": 0.8702329993247986,
      "learning_rate": 3.999902020666394e-05,
      "loss": 0.08815574645996094,
      "step": 534
    },
    {
      "epoch": 0.07454887479969344,
      "grad_norm": 1.270948052406311,
      "learning_rate": 3.99989246721154e-05,
      "loss": 0.11433601379394531,
      "step": 535
    },
    {
      "epoch": 0.07468821849090783,
      "grad_norm": 2.3735063076019287,
      "learning_rate": 3.9998824694262376e-05,
      "loss": 0.10658454895019531,
      "step": 536
    },
    {
      "epoch": 0.0748275621821222,
      "grad_norm": 3.470846176147461,
      "learning_rate": 3.999872027312709e-05,
      "loss": 0.11987113952636719,
      "step": 537
    },
    {
      "epoch": 0.07496690587333658,
      "grad_norm": 2.8136346340179443,
      "learning_rate": 3.999861140873274e-05,
      "loss": 0.14839935302734375,
      "step": 538
    },
    {
      "epoch": 0.07510624956455096,
      "grad_norm": 4.242218494415283,
      "learning_rate": 3.99984981011035e-05,
      "loss": 0.15454864501953125,
      "step": 539
    },
    {
      "epoch": 0.07524559325576534,
      "grad_norm": 2.2143540382385254,
      "learning_rate": 3.999838035026456e-05,
      "loss": 0.1316814422607422,
      "step": 540
    },
    {
      "epoch": 0.07538493694697973,
      "grad_norm": 1.0337727069854736,
      "learning_rate": 3.999825815624208e-05,
      "loss": 0.08818244934082031,
      "step": 541
    },
    {
      "epoch": 0.07552428063819411,
      "grad_norm": 2.531571626663208,
      "learning_rate": 3.9998131519063204e-05,
      "loss": 0.12324905395507812,
      "step": 542
    },
    {
      "epoch": 0.07566362432940849,
      "grad_norm": 1.8717105388641357,
      "learning_rate": 3.999800043875607e-05,
      "loss": 0.14538955688476562,
      "step": 543
    },
    {
      "epoch": 0.07580296802062286,
      "grad_norm": 0.8989245891571045,
      "learning_rate": 3.999786491534981e-05,
      "loss": 0.08049392700195312,
      "step": 544
    },
    {
      "epoch": 0.07594231171183724,
      "grad_norm": 0.8702663779258728,
      "learning_rate": 3.9997724948874514e-05,
      "loss": 0.11128807067871094,
      "step": 545
    },
    {
      "epoch": 0.07608165540305163,
      "grad_norm": 1.7686147689819336,
      "learning_rate": 3.999758053936129e-05,
      "loss": 0.1466217041015625,
      "step": 546
    },
    {
      "epoch": 0.07622099909426601,
      "grad_norm": 0.9515817761421204,
      "learning_rate": 3.999743168684223e-05,
      "loss": 0.08151435852050781,
      "step": 547
    },
    {
      "epoch": 0.07636034278548039,
      "grad_norm": 5.572052001953125,
      "learning_rate": 3.9997278391350395e-05,
      "loss": 0.15542984008789062,
      "step": 548
    },
    {
      "epoch": 0.07649968647669476,
      "grad_norm": 2.3883323669433594,
      "learning_rate": 3.999712065291984e-05,
      "loss": 0.12106704711914062,
      "step": 549
    },
    {
      "epoch": 0.07663903016790914,
      "grad_norm": 1.4061657190322876,
      "learning_rate": 3.999695847158562e-05,
      "loss": 0.15160751342773438,
      "step": 550
    },
    {
      "epoch": 0.07677837385912353,
      "grad_norm": 2.090426206588745,
      "learning_rate": 3.999679184738377e-05,
      "loss": 0.11965560913085938,
      "step": 551
    },
    {
      "epoch": 0.07691771755033791,
      "grad_norm": 4.149261474609375,
      "learning_rate": 3.9996620780351306e-05,
      "loss": 0.12900733947753906,
      "step": 552
    },
    {
      "epoch": 0.07705706124155229,
      "grad_norm": 2.549879789352417,
      "learning_rate": 3.9996445270526235e-05,
      "loss": 0.12584304809570312,
      "step": 553
    },
    {
      "epoch": 0.07719640493276667,
      "grad_norm": 5.93257999420166,
      "learning_rate": 3.999626531794755e-05,
      "loss": 0.16759109497070312,
      "step": 554
    },
    {
      "epoch": 0.07733574862398104,
      "grad_norm": 0.6048027873039246,
      "learning_rate": 3.9996080922655236e-05,
      "loss": 0.0752716064453125,
      "step": 555
    },
    {
      "epoch": 0.07747509231519543,
      "grad_norm": 2.0776069164276123,
      "learning_rate": 3.9995892084690256e-05,
      "loss": 0.12067604064941406,
      "step": 556
    },
    {
      "epoch": 0.07761443600640981,
      "grad_norm": 0.923984944820404,
      "learning_rate": 3.999569880409458e-05,
      "loss": 0.13747787475585938,
      "step": 557
    },
    {
      "epoch": 0.07775377969762419,
      "grad_norm": 0.7828582525253296,
      "learning_rate": 3.9995501080911124e-05,
      "loss": 0.12700462341308594,
      "step": 558
    },
    {
      "epoch": 0.07789312338883857,
      "grad_norm": 1.2990708351135254,
      "learning_rate": 3.999529891518384e-05,
      "loss": 0.10113906860351562,
      "step": 559
    },
    {
      "epoch": 0.07803246708005294,
      "grad_norm": 1.3646270036697388,
      "learning_rate": 3.9995092306957636e-05,
      "loss": 0.12221717834472656,
      "step": 560
    },
    {
      "epoch": 0.07817181077126734,
      "grad_norm": 1.6404664516448975,
      "learning_rate": 3.9994881256278424e-05,
      "loss": 0.10071563720703125,
      "step": 561
    },
    {
      "epoch": 0.07831115446248171,
      "grad_norm": 2.1226539611816406,
      "learning_rate": 3.999466576319308e-05,
      "loss": 0.16773414611816406,
      "step": 562
    },
    {
      "epoch": 0.07845049815369609,
      "grad_norm": 1.2778706550598145,
      "learning_rate": 3.999444582774949e-05,
      "loss": 0.14202308654785156,
      "step": 563
    },
    {
      "epoch": 0.07858984184491047,
      "grad_norm": 3.3511712551116943,
      "learning_rate": 3.999422144999652e-05,
      "loss": 0.18107986450195312,
      "step": 564
    },
    {
      "epoch": 0.07872918553612485,
      "grad_norm": 2.2309207916259766,
      "learning_rate": 3.9993992629984016e-05,
      "loss": 0.1256866455078125,
      "step": 565
    },
    {
      "epoch": 0.07886852922733924,
      "grad_norm": 1.0485955476760864,
      "learning_rate": 3.9993759367762825e-05,
      "loss": 0.07908248901367188,
      "step": 566
    },
    {
      "epoch": 0.07900787291855361,
      "grad_norm": 2.0486409664154053,
      "learning_rate": 3.9993521663384766e-05,
      "loss": 0.12654876708984375,
      "step": 567
    },
    {
      "epoch": 0.07914721660976799,
      "grad_norm": 3.582606792449951,
      "learning_rate": 3.999327951690265e-05,
      "loss": 0.17953109741210938,
      "step": 568
    },
    {
      "epoch": 0.07928656030098237,
      "grad_norm": 2.542111396789551,
      "learning_rate": 3.9993032928370284e-05,
      "loss": 0.14783859252929688,
      "step": 569
    },
    {
      "epoch": 0.07942590399219675,
      "grad_norm": 2.0648233890533447,
      "learning_rate": 3.999278189784245e-05,
      "loss": 0.10119247436523438,
      "step": 570
    },
    {
      "epoch": 0.07956524768341114,
      "grad_norm": 2.798393487930298,
      "learning_rate": 3.9992526425374916e-05,
      "loss": 0.12763214111328125,
      "step": 571
    },
    {
      "epoch": 0.07970459137462552,
      "grad_norm": 2.1240041255950928,
      "learning_rate": 3.999226651102445e-05,
      "loss": 0.1259765625,
      "step": 572
    },
    {
      "epoch": 0.0798439350658399,
      "grad_norm": 1.9640324115753174,
      "learning_rate": 3.99920021548488e-05,
      "loss": 0.13722610473632812,
      "step": 573
    },
    {
      "epoch": 0.07998327875705427,
      "grad_norm": 2.151902675628662,
      "learning_rate": 3.99917333569067e-05,
      "loss": 0.13800811767578125,
      "step": 574
    },
    {
      "epoch": 0.08012262244826865,
      "grad_norm": 2.2750742435455322,
      "learning_rate": 3.9991460117257856e-05,
      "loss": 0.15391159057617188,
      "step": 575
    },
    {
      "epoch": 0.08026196613948304,
      "grad_norm": 1.5828267335891724,
      "learning_rate": 3.9991182435962994e-05,
      "loss": 0.1822643280029297,
      "step": 576
    },
    {
      "epoch": 0.08040130983069742,
      "grad_norm": 2.76873517036438,
      "learning_rate": 3.99909003130838e-05,
      "loss": 0.11618995666503906,
      "step": 577
    },
    {
      "epoch": 0.0805406535219118,
      "grad_norm": 3.414393424987793,
      "learning_rate": 3.9990613748682954e-05,
      "loss": 0.1348114013671875,
      "step": 578
    },
    {
      "epoch": 0.08067999721312617,
      "grad_norm": 1.2800782918930054,
      "learning_rate": 3.9990322742824126e-05,
      "loss": 0.10827255249023438,
      "step": 579
    },
    {
      "epoch": 0.08081934090434055,
      "grad_norm": 3.1290533542633057,
      "learning_rate": 3.999002729557197e-05,
      "loss": 0.10898971557617188,
      "step": 580
    },
    {
      "epoch": 0.08095868459555494,
      "grad_norm": 3.069255828857422,
      "learning_rate": 3.9989727406992136e-05,
      "loss": 0.1311187744140625,
      "step": 581
    },
    {
      "epoch": 0.08109802828676932,
      "grad_norm": 1.0050139427185059,
      "learning_rate": 3.998942307715124e-05,
      "loss": 0.11010932922363281,
      "step": 582
    },
    {
      "epoch": 0.0812373719779837,
      "grad_norm": 2.393117666244507,
      "learning_rate": 3.998911430611691e-05,
      "loss": 0.14769744873046875,
      "step": 583
    },
    {
      "epoch": 0.08137671566919807,
      "grad_norm": 1.805238127708435,
      "learning_rate": 3.9988801093957735e-05,
      "loss": 0.10110664367675781,
      "step": 584
    },
    {
      "epoch": 0.08151605936041245,
      "grad_norm": 0.9078276753425598,
      "learning_rate": 3.9988483440743306e-05,
      "loss": 0.08968353271484375,
      "step": 585
    },
    {
      "epoch": 0.08165540305162684,
      "grad_norm": 2.747990131378174,
      "learning_rate": 3.998816134654421e-05,
      "loss": 0.1086568832397461,
      "step": 586
    },
    {
      "epoch": 0.08179474674284122,
      "grad_norm": 4.451938152313232,
      "learning_rate": 3.9987834811431986e-05,
      "loss": 0.1411724090576172,
      "step": 587
    },
    {
      "epoch": 0.0819340904340556,
      "grad_norm": 1.203033685684204,
      "learning_rate": 3.998750383547921e-05,
      "loss": 0.09840202331542969,
      "step": 588
    },
    {
      "epoch": 0.08207343412526998,
      "grad_norm": 1.3895381689071655,
      "learning_rate": 3.9987168418759396e-05,
      "loss": 0.11197090148925781,
      "step": 589
    },
    {
      "epoch": 0.08221277781648435,
      "grad_norm": 2.0719799995422363,
      "learning_rate": 3.998682856134708e-05,
      "loss": 0.10160064697265625,
      "step": 590
    },
    {
      "epoch": 0.08235212150769874,
      "grad_norm": 3.161625385284424,
      "learning_rate": 3.9986484263317766e-05,
      "loss": 0.115814208984375,
      "step": 591
    },
    {
      "epoch": 0.08249146519891312,
      "grad_norm": 1.7461684942245483,
      "learning_rate": 3.9986135524747945e-05,
      "loss": 0.10396003723144531,
      "step": 592
    },
    {
      "epoch": 0.0826308088901275,
      "grad_norm": 0.9677025675773621,
      "learning_rate": 3.9985782345715105e-05,
      "loss": 0.11051368713378906,
      "step": 593
    },
    {
      "epoch": 0.08277015258134188,
      "grad_norm": 1.5929374694824219,
      "learning_rate": 3.99854247262977e-05,
      "loss": 0.09242630004882812,
      "step": 594
    },
    {
      "epoch": 0.08290949627255625,
      "grad_norm": 1.2520278692245483,
      "learning_rate": 3.99850626665752e-05,
      "loss": 0.1183319091796875,
      "step": 595
    },
    {
      "epoch": 0.08304883996377065,
      "grad_norm": 1.7947602272033691,
      "learning_rate": 3.998469616662805e-05,
      "loss": 0.10550308227539062,
      "step": 596
    },
    {
      "epoch": 0.08318818365498502,
      "grad_norm": 1.553332805633545,
      "learning_rate": 3.9984325226537665e-05,
      "loss": 0.09752845764160156,
      "step": 597
    },
    {
      "epoch": 0.0833275273461994,
      "grad_norm": 0.8475018739700317,
      "learning_rate": 3.998394984638647e-05,
      "loss": 0.08748245239257812,
      "step": 598
    },
    {
      "epoch": 0.08346687103741378,
      "grad_norm": 1.5756542682647705,
      "learning_rate": 3.9983570026257844e-05,
      "loss": 0.1014251708984375,
      "step": 599
    },
    {
      "epoch": 0.08360621472862816,
      "grad_norm": 0.6557784676551819,
      "learning_rate": 3.998318576623621e-05,
      "loss": 0.10439300537109375,
      "step": 600
    },
    {
      "epoch": 0.08374555841984255,
      "grad_norm": 1.7600188255310059,
      "learning_rate": 3.998279706640691e-05,
      "loss": 0.09660530090332031,
      "step": 601
    },
    {
      "epoch": 0.08388490211105692,
      "grad_norm": 1.3026453256607056,
      "learning_rate": 3.998240392685633e-05,
      "loss": 0.10809516906738281,
      "step": 602
    },
    {
      "epoch": 0.0840242458022713,
      "grad_norm": 1.9691437482833862,
      "learning_rate": 3.9982006347671796e-05,
      "loss": 0.11550712585449219,
      "step": 603
    },
    {
      "epoch": 0.08416358949348568,
      "grad_norm": 1.722626805305481,
      "learning_rate": 3.998160432894164e-05,
      "loss": 0.09701156616210938,
      "step": 604
    },
    {
      "epoch": 0.08430293318470006,
      "grad_norm": 0.47721943259239197,
      "learning_rate": 3.99811978707552e-05,
      "loss": 0.06611824035644531,
      "step": 605
    },
    {
      "epoch": 0.08444227687591445,
      "grad_norm": 1.7810955047607422,
      "learning_rate": 3.998078697320277e-05,
      "loss": 0.14276695251464844,
      "step": 606
    },
    {
      "epoch": 0.08458162056712883,
      "grad_norm": 1.7148725986480713,
      "learning_rate": 3.9980371636375645e-05,
      "loss": 0.10901641845703125,
      "step": 607
    },
    {
      "epoch": 0.0847209642583432,
      "grad_norm": 1.0007349252700806,
      "learning_rate": 3.99799518603661e-05,
      "loss": 0.09817886352539062,
      "step": 608
    },
    {
      "epoch": 0.08486030794955758,
      "grad_norm": 0.8453435897827148,
      "learning_rate": 3.9979527645267395e-05,
      "loss": 0.1143951416015625,
      "step": 609
    },
    {
      "epoch": 0.08499965164077196,
      "grad_norm": 1.9799214601516724,
      "learning_rate": 3.99790989911738e-05,
      "loss": 0.09308624267578125,
      "step": 610
    },
    {
      "epoch": 0.08513899533198635,
      "grad_norm": 1.374338150024414,
      "learning_rate": 3.997866589818053e-05,
      "loss": 0.11553955078125,
      "step": 611
    },
    {
      "epoch": 0.08527833902320073,
      "grad_norm": 1.1998134851455688,
      "learning_rate": 3.997822836638382e-05,
      "loss": 0.11280250549316406,
      "step": 612
    },
    {
      "epoch": 0.0854176827144151,
      "grad_norm": 1.1562647819519043,
      "learning_rate": 3.997778639588088e-05,
      "loss": 0.09814262390136719,
      "step": 613
    },
    {
      "epoch": 0.08555702640562948,
      "grad_norm": 2.2079992294311523,
      "learning_rate": 3.9977339986769905e-05,
      "loss": 0.11584091186523438,
      "step": 614
    },
    {
      "epoch": 0.08569637009684386,
      "grad_norm": 2.0636987686157227,
      "learning_rate": 3.9976889139150074e-05,
      "loss": 0.1290607452392578,
      "step": 615
    },
    {
      "epoch": 0.08583571378805825,
      "grad_norm": 1.7909809350967407,
      "learning_rate": 3.997643385312156e-05,
      "loss": 0.11124229431152344,
      "step": 616
    },
    {
      "epoch": 0.08597505747927263,
      "grad_norm": 2.6300296783447266,
      "learning_rate": 3.9975974128785505e-05,
      "loss": 0.11059188842773438,
      "step": 617
    },
    {
      "epoch": 0.086114401170487,
      "grad_norm": 3.255582809448242,
      "learning_rate": 3.997550996624406e-05,
      "loss": 0.12091636657714844,
      "step": 618
    },
    {
      "epoch": 0.08625374486170138,
      "grad_norm": 3.166318655014038,
      "learning_rate": 3.997504136560036e-05,
      "loss": 0.10851001739501953,
      "step": 619
    },
    {
      "epoch": 0.08639308855291576,
      "grad_norm": 1.74764883518219,
      "learning_rate": 3.997456832695849e-05,
      "loss": 0.0956878662109375,
      "step": 620
    },
    {
      "epoch": 0.08653243224413015,
      "grad_norm": 0.7127701044082642,
      "learning_rate": 3.997409085042358e-05,
      "loss": 0.1027069091796875,
      "step": 621
    },
    {
      "epoch": 0.08667177593534453,
      "grad_norm": 2.7335541248321533,
      "learning_rate": 3.997360893610169e-05,
      "loss": 0.11098098754882812,
      "step": 622
    },
    {
      "epoch": 0.08681111962655891,
      "grad_norm": 5.717250347137451,
      "learning_rate": 3.99731225840999e-05,
      "loss": 0.17713165283203125,
      "step": 623
    },
    {
      "epoch": 0.08695046331777329,
      "grad_norm": 2.7387754917144775,
      "learning_rate": 3.9972631794526265e-05,
      "loss": 0.13355255126953125,
      "step": 624
    },
    {
      "epoch": 0.08708980700898766,
      "grad_norm": 1.081800937652588,
      "learning_rate": 3.9972136567489836e-05,
      "loss": 0.172821044921875,
      "step": 625
    },
    {
      "epoch": 0.08722915070020205,
      "grad_norm": 1.5363692045211792,
      "learning_rate": 3.997163690310063e-05,
      "loss": 0.11891937255859375,
      "step": 626
    },
    {
      "epoch": 0.08736849439141643,
      "grad_norm": 1.4293453693389893,
      "learning_rate": 3.997113280146966e-05,
      "loss": 0.15371322631835938,
      "step": 627
    },
    {
      "epoch": 0.08750783808263081,
      "grad_norm": 1.8178093433380127,
      "learning_rate": 3.9970624262708934e-05,
      "loss": 0.1451873779296875,
      "step": 628
    },
    {
      "epoch": 0.08764718177384519,
      "grad_norm": 0.8165032863616943,
      "learning_rate": 3.997011128693143e-05,
      "loss": 0.07885360717773438,
      "step": 629
    },
    {
      "epoch": 0.08778652546505956,
      "grad_norm": 0.7978515028953552,
      "learning_rate": 3.996959387425113e-05,
      "loss": 0.08555412292480469,
      "step": 630
    },
    {
      "epoch": 0.08792586915627396,
      "grad_norm": 1.210871696472168,
      "learning_rate": 3.996907202478298e-05,
      "loss": 0.1194610595703125,
      "step": 631
    },
    {
      "epoch": 0.08806521284748833,
      "grad_norm": 1.2033125162124634,
      "learning_rate": 3.996854573864293e-05,
      "loss": 0.11191177368164062,
      "step": 632
    },
    {
      "epoch": 0.08820455653870271,
      "grad_norm": 0.672520101070404,
      "learning_rate": 3.9968015015947904e-05,
      "loss": 0.10470771789550781,
      "step": 633
    },
    {
      "epoch": 0.08834390022991709,
      "grad_norm": 0.7171827554702759,
      "learning_rate": 3.996747985681582e-05,
      "loss": 0.10395431518554688,
      "step": 634
    },
    {
      "epoch": 0.08848324392113147,
      "grad_norm": 1.9401113986968994,
      "learning_rate": 3.9966940261365576e-05,
      "loss": 0.12075424194335938,
      "step": 635
    },
    {
      "epoch": 0.08862258761234586,
      "grad_norm": 1.069329023361206,
      "learning_rate": 3.996639622971706e-05,
      "loss": 0.08725929260253906,
      "step": 636
    },
    {
      "epoch": 0.08876193130356023,
      "grad_norm": 1.750138521194458,
      "learning_rate": 3.996584776199114e-05,
      "loss": 0.1767597198486328,
      "step": 637
    },
    {
      "epoch": 0.08890127499477461,
      "grad_norm": 1.0362471342086792,
      "learning_rate": 3.9965294858309685e-05,
      "loss": 0.10419464111328125,
      "step": 638
    },
    {
      "epoch": 0.08904061868598899,
      "grad_norm": 1.1169592142105103,
      "learning_rate": 3.996473751879552e-05,
      "loss": 0.10972785949707031,
      "step": 639
    },
    {
      "epoch": 0.08917996237720337,
      "grad_norm": 0.9762294292449951,
      "learning_rate": 3.996417574357248e-05,
      "loss": 0.12237358093261719,
      "step": 640
    },
    {
      "epoch": 0.08931930606841776,
      "grad_norm": 2.073392868041992,
      "learning_rate": 3.996360953276538e-05,
      "loss": 0.15527725219726562,
      "step": 641
    },
    {
      "epoch": 0.08945864975963214,
      "grad_norm": 0.6941425204277039,
      "learning_rate": 3.996303888650002e-05,
      "loss": 0.10612106323242188,
      "step": 642
    },
    {
      "epoch": 0.08959799345084651,
      "grad_norm": 0.5668501257896423,
      "learning_rate": 3.996246380490319e-05,
      "loss": 0.10363006591796875,
      "step": 643
    },
    {
      "epoch": 0.08973733714206089,
      "grad_norm": 2.4187774658203125,
      "learning_rate": 3.996188428810264e-05,
      "loss": 0.15231704711914062,
      "step": 644
    },
    {
      "epoch": 0.08987668083327527,
      "grad_norm": 1.8461713790893555,
      "learning_rate": 3.9961300336227146e-05,
      "loss": 0.12401580810546875,
      "step": 645
    },
    {
      "epoch": 0.09001602452448966,
      "grad_norm": 1.252118468284607,
      "learning_rate": 3.996071194940644e-05,
      "loss": 0.10144615173339844,
      "step": 646
    },
    {
      "epoch": 0.09015536821570404,
      "grad_norm": 0.6942083835601807,
      "learning_rate": 3.996011912777126e-05,
      "loss": 0.09859275817871094,
      "step": 647
    },
    {
      "epoch": 0.09029471190691842,
      "grad_norm": 2.204408884048462,
      "learning_rate": 3.995952187145329e-05,
      "loss": 0.1428050994873047,
      "step": 648
    },
    {
      "epoch": 0.09043405559813279,
      "grad_norm": 0.7580353617668152,
      "learning_rate": 3.995892018058525e-05,
      "loss": 0.09755802154541016,
      "step": 649
    },
    {
      "epoch": 0.09057339928934717,
      "grad_norm": 2.8576738834381104,
      "learning_rate": 3.995831405530082e-05,
      "loss": 0.1951770782470703,
      "step": 650
    },
    {
      "epoch": 0.09071274298056156,
      "grad_norm": 2.6492650508880615,
      "learning_rate": 3.9957703495734666e-05,
      "loss": 0.1256256103515625,
      "step": 651
    },
    {
      "epoch": 0.09085208667177594,
      "grad_norm": 1.157537817955017,
      "learning_rate": 3.9957088502022426e-05,
      "loss": 0.09515380859375,
      "step": 652
    },
    {
      "epoch": 0.09099143036299032,
      "grad_norm": 1.6019444465637207,
      "learning_rate": 3.995646907430076e-05,
      "loss": 0.09901046752929688,
      "step": 653
    },
    {
      "epoch": 0.0911307740542047,
      "grad_norm": 4.919183254241943,
      "learning_rate": 3.995584521270727e-05,
      "loss": 0.14474105834960938,
      "step": 654
    },
    {
      "epoch": 0.09127011774541907,
      "grad_norm": 0.8669459819793701,
      "learning_rate": 3.995521691738058e-05,
      "loss": 0.11705398559570312,
      "step": 655
    },
    {
      "epoch": 0.09140946143663346,
      "grad_norm": 1.8246808052062988,
      "learning_rate": 3.995458418846028e-05,
      "loss": 0.10042953491210938,
      "step": 656
    },
    {
      "epoch": 0.09154880512784784,
      "grad_norm": 3.4547104835510254,
      "learning_rate": 3.995394702608693e-05,
      "loss": 0.15300750732421875,
      "step": 657
    },
    {
      "epoch": 0.09168814881906222,
      "grad_norm": 1.3384592533111572,
      "learning_rate": 3.995330543040212e-05,
      "loss": 0.08542442321777344,
      "step": 658
    },
    {
      "epoch": 0.0918274925102766,
      "grad_norm": 1.164960503578186,
      "learning_rate": 3.995265940154838e-05,
      "loss": 0.1305999755859375,
      "step": 659
    },
    {
      "epoch": 0.09196683620149097,
      "grad_norm": 1.5734573602676392,
      "learning_rate": 3.995200893966925e-05,
      "loss": 0.19118690490722656,
      "step": 660
    },
    {
      "epoch": 0.09210617989270536,
      "grad_norm": 2.191829204559326,
      "learning_rate": 3.9951354044909246e-05,
      "loss": 0.11335182189941406,
      "step": 661
    },
    {
      "epoch": 0.09224552358391974,
      "grad_norm": 1.332816243171692,
      "learning_rate": 3.995069471741387e-05,
      "loss": 0.10511970520019531,
      "step": 662
    },
    {
      "epoch": 0.09238486727513412,
      "grad_norm": 1.7221275568008423,
      "learning_rate": 3.9950030957329604e-05,
      "loss": 0.13498878479003906,
      "step": 663
    },
    {
      "epoch": 0.0925242109663485,
      "grad_norm": 1.7761082649230957,
      "learning_rate": 3.9949362764803934e-05,
      "loss": 0.0808258056640625,
      "step": 664
    },
    {
      "epoch": 0.09266355465756287,
      "grad_norm": 1.271019697189331,
      "learning_rate": 3.9948690139985305e-05,
      "loss": 0.10305976867675781,
      "step": 665
    },
    {
      "epoch": 0.09280289834877727,
      "grad_norm": 1.014792799949646,
      "learning_rate": 3.994801308302316e-05,
      "loss": 0.07171058654785156,
      "step": 666
    },
    {
      "epoch": 0.09294224203999164,
      "grad_norm": 2.9300577640533447,
      "learning_rate": 3.994733159406794e-05,
      "loss": 0.12253570556640625,
      "step": 667
    },
    {
      "epoch": 0.09308158573120602,
      "grad_norm": 2.817939043045044,
      "learning_rate": 3.9946645673271034e-05,
      "loss": 0.13839149475097656,
      "step": 668
    },
    {
      "epoch": 0.0932209294224204,
      "grad_norm": 1.1543421745300293,
      "learning_rate": 3.994595532078486e-05,
      "loss": 0.11301040649414062,
      "step": 669
    },
    {
      "epoch": 0.09336027311363478,
      "grad_norm": 1.3884141445159912,
      "learning_rate": 3.9945260536762775e-05,
      "loss": 0.06984329223632812,
      "step": 670
    },
    {
      "epoch": 0.09349961680484917,
      "grad_norm": 3.0776305198669434,
      "learning_rate": 3.994456132135916e-05,
      "loss": 0.15468692779541016,
      "step": 671
    },
    {
      "epoch": 0.09363896049606354,
      "grad_norm": 2.6249451637268066,
      "learning_rate": 3.994385767472937e-05,
      "loss": 0.14349746704101562,
      "step": 672
    },
    {
      "epoch": 0.09377830418727792,
      "grad_norm": 0.8873428106307983,
      "learning_rate": 3.9943149597029724e-05,
      "loss": 0.08478355407714844,
      "step": 673
    },
    {
      "epoch": 0.0939176478784923,
      "grad_norm": 1.360435962677002,
      "learning_rate": 3.994243708841755e-05,
      "loss": 0.1453990936279297,
      "step": 674
    },
    {
      "epoch": 0.09405699156970668,
      "grad_norm": 2.7673938274383545,
      "learning_rate": 3.9941720149051146e-05,
      "loss": 0.16139602661132812,
      "step": 675
    },
    {
      "epoch": 0.09419633526092107,
      "grad_norm": 1.5676738023757935,
      "learning_rate": 3.99409987790898e-05,
      "loss": 0.11298370361328125,
      "step": 676
    },
    {
      "epoch": 0.09433567895213545,
      "grad_norm": 0.7559939622879028,
      "learning_rate": 3.9940272978693795e-05,
      "loss": 0.08307456970214844,
      "step": 677
    },
    {
      "epoch": 0.09447502264334982,
      "grad_norm": 3.1241300106048584,
      "learning_rate": 3.993954274802437e-05,
      "loss": 0.17406082153320312,
      "step": 678
    },
    {
      "epoch": 0.0946143663345642,
      "grad_norm": 2.981058359146118,
      "learning_rate": 3.993880808724378e-05,
      "loss": 0.14940643310546875,
      "step": 679
    },
    {
      "epoch": 0.09475371002577858,
      "grad_norm": 1.125257134437561,
      "learning_rate": 3.993806899651524e-05,
      "loss": 0.10942649841308594,
      "step": 680
    },
    {
      "epoch": 0.09489305371699297,
      "grad_norm": 1.3269895315170288,
      "learning_rate": 3.9937325476002955e-05,
      "loss": 0.103607177734375,
      "step": 681
    },
    {
      "epoch": 0.09503239740820735,
      "grad_norm": 0.6538723111152649,
      "learning_rate": 3.993657752587214e-05,
      "loss": 0.09619331359863281,
      "step": 682
    },
    {
      "epoch": 0.09517174109942172,
      "grad_norm": 1.5624948740005493,
      "learning_rate": 3.993582514628895e-05,
      "loss": 0.14385604858398438,
      "step": 683
    },
    {
      "epoch": 0.0953110847906361,
      "grad_norm": 1.3760669231414795,
      "learning_rate": 3.9935068337420564e-05,
      "loss": 0.09527206420898438,
      "step": 684
    },
    {
      "epoch": 0.09545042848185048,
      "grad_norm": 0.34413033723831177,
      "learning_rate": 3.99343070994351e-05,
      "loss": 0.07621192932128906,
      "step": 685
    },
    {
      "epoch": 0.09558977217306487,
      "grad_norm": 1.0510213375091553,
      "learning_rate": 3.993354143250173e-05,
      "loss": 0.09285545349121094,
      "step": 686
    },
    {
      "epoch": 0.09572911586427925,
      "grad_norm": 1.4344948530197144,
      "learning_rate": 3.993277133679053e-05,
      "loss": 0.12052345275878906,
      "step": 687
    },
    {
      "epoch": 0.09586845955549363,
      "grad_norm": 1.9553276300430298,
      "learning_rate": 3.993199681247261e-05,
      "loss": 0.08988189697265625,
      "step": 688
    },
    {
      "epoch": 0.096007803246708,
      "grad_norm": 2.0521926879882812,
      "learning_rate": 3.9931217859720066e-05,
      "loss": 0.14418792724609375,
      "step": 689
    },
    {
      "epoch": 0.09614714693792238,
      "grad_norm": 0.8239341378211975,
      "learning_rate": 3.993043447870594e-05,
      "loss": 0.12273788452148438,
      "step": 690
    },
    {
      "epoch": 0.09628649062913676,
      "grad_norm": 1.9166098833084106,
      "learning_rate": 3.99296466696043e-05,
      "loss": 0.11394309997558594,
      "step": 691
    },
    {
      "epoch": 0.09642583432035115,
      "grad_norm": 3.601027011871338,
      "learning_rate": 3.9928854432590166e-05,
      "loss": 0.1578388214111328,
      "step": 692
    },
    {
      "epoch": 0.09656517801156553,
      "grad_norm": 2.117563009262085,
      "learning_rate": 3.9928057767839565e-05,
      "loss": 0.11327934265136719,
      "step": 693
    },
    {
      "epoch": 0.0967045217027799,
      "grad_norm": 0.7808533906936646,
      "learning_rate": 3.992725667552948e-05,
      "loss": 0.101318359375,
      "step": 694
    },
    {
      "epoch": 0.09684386539399428,
      "grad_norm": 1.9542772769927979,
      "learning_rate": 3.9926451155837925e-05,
      "loss": 0.11246109008789062,
      "step": 695
    },
    {
      "epoch": 0.09698320908520866,
      "grad_norm": 2.3494012355804443,
      "learning_rate": 3.9925641208943846e-05,
      "loss": 0.11851882934570312,
      "step": 696
    },
    {
      "epoch": 0.09712255277642305,
      "grad_norm": 1.7434897422790527,
      "learning_rate": 3.99248268350272e-05,
      "loss": 0.1325836181640625,
      "step": 697
    },
    {
      "epoch": 0.09726189646763743,
      "grad_norm": 1.3294047117233276,
      "learning_rate": 3.992400803426892e-05,
      "loss": 0.11371994018554688,
      "step": 698
    },
    {
      "epoch": 0.0974012401588518,
      "grad_norm": 1.560742974281311,
      "learning_rate": 3.992318480685094e-05,
      "loss": 0.11021995544433594,
      "step": 699
    },
    {
      "epoch": 0.09754058385006618,
      "grad_norm": 1.5074039697647095,
      "learning_rate": 3.992235715295614e-05,
      "loss": 0.11312103271484375,
      "step": 700
    },
    {
      "epoch": 0.09767992754128056,
      "grad_norm": 0.7238860726356506,
      "learning_rate": 3.992152507276841e-05,
      "loss": 0.09823417663574219,
      "step": 701
    },
    {
      "epoch": 0.09781927123249495,
      "grad_norm": 1.6033929586410522,
      "learning_rate": 3.9920688566472636e-05,
      "loss": 0.13399314880371094,
      "step": 702
    },
    {
      "epoch": 0.09795861492370933,
      "grad_norm": 0.9758917093276978,
      "learning_rate": 3.991984763425465e-05,
      "loss": 0.0968780517578125,
      "step": 703
    },
    {
      "epoch": 0.09809795861492371,
      "grad_norm": 0.8494994640350342,
      "learning_rate": 3.99190022763013e-05,
      "loss": 0.105987548828125,
      "step": 704
    },
    {
      "epoch": 0.09823730230613809,
      "grad_norm": 1.23043692111969,
      "learning_rate": 3.991815249280041e-05,
      "loss": 0.1386699676513672,
      "step": 705
    },
    {
      "epoch": 0.09837664599735246,
      "grad_norm": 0.9754325747489929,
      "learning_rate": 3.991729828394078e-05,
      "loss": 0.09955787658691406,
      "step": 706
    },
    {
      "epoch": 0.09851598968856685,
      "grad_norm": 0.7522068023681641,
      "learning_rate": 3.9916439649912175e-05,
      "loss": 0.10589981079101562,
      "step": 707
    },
    {
      "epoch": 0.09865533337978123,
      "grad_norm": 0.9560259580612183,
      "learning_rate": 3.9915576590905385e-05,
      "loss": 0.17127609252929688,
      "step": 708
    },
    {
      "epoch": 0.09879467707099561,
      "grad_norm": 1.1445032358169556,
      "learning_rate": 3.991470910711216e-05,
      "loss": 0.08207130432128906,
      "step": 709
    },
    {
      "epoch": 0.09893402076220999,
      "grad_norm": 1.9059768915176392,
      "learning_rate": 3.9913837198725224e-05,
      "loss": 0.13786697387695312,
      "step": 710
    },
    {
      "epoch": 0.09907336445342436,
      "grad_norm": 0.48259973526000977,
      "learning_rate": 3.9912960865938316e-05,
      "loss": 0.09104347229003906,
      "step": 711
    },
    {
      "epoch": 0.09921270814463876,
      "grad_norm": 0.6930782198905945,
      "learning_rate": 3.9912080108946115e-05,
      "loss": 0.09864234924316406,
      "step": 712
    },
    {
      "epoch": 0.09935205183585313,
      "grad_norm": 0.9861940741539001,
      "learning_rate": 3.9911194927944315e-05,
      "loss": 0.07714271545410156,
      "step": 713
    },
    {
      "epoch": 0.09949139552706751,
      "grad_norm": 0.9397109150886536,
      "learning_rate": 3.991030532312959e-05,
      "loss": 0.14679718017578125,
      "step": 714
    },
    {
      "epoch": 0.09963073921828189,
      "grad_norm": 0.7344309687614441,
      "learning_rate": 3.990941129469957e-05,
      "loss": 0.0991067886352539,
      "step": 715
    },
    {
      "epoch": 0.09977008290949627,
      "grad_norm": 1.9518712759017944,
      "learning_rate": 3.9908512842852906e-05,
      "loss": 0.1284027099609375,
      "step": 716
    },
    {
      "epoch": 0.09990942660071066,
      "grad_norm": 1.9064795970916748,
      "learning_rate": 3.990760996778921e-05,
      "loss": 0.118255615234375,
      "step": 717
    },
    {
      "epoch": 0.10004877029192503,
      "grad_norm": 1.0310511589050293,
      "learning_rate": 3.9906702669709074e-05,
      "loss": 0.10207366943359375,
      "step": 718
    },
    {
      "epoch": 0.10018811398313941,
      "grad_norm": 2.126049757003784,
      "learning_rate": 3.9905790948814086e-05,
      "loss": 0.16246414184570312,
      "step": 719
    },
    {
      "epoch": 0.10032745767435379,
      "grad_norm": 1.5851454734802246,
      "learning_rate": 3.9904874805306804e-05,
      "loss": 0.1069793701171875,
      "step": 720
    },
    {
      "epoch": 0.10046680136556817,
      "grad_norm": 1.3377631902694702,
      "learning_rate": 3.990395423939077e-05,
      "loss": 0.1281890869140625,
      "step": 721
    },
    {
      "epoch": 0.10060614505678256,
      "grad_norm": 0.7208927273750305,
      "learning_rate": 3.9903029251270535e-05,
      "loss": 0.07527732849121094,
      "step": 722
    },
    {
      "epoch": 0.10074548874799694,
      "grad_norm": 0.5075308680534363,
      "learning_rate": 3.990209984115158e-05,
      "loss": 0.07642745971679688,
      "step": 723
    },
    {
      "epoch": 0.10088483243921131,
      "grad_norm": 1.675383448600769,
      "learning_rate": 3.990116600924042e-05,
      "loss": 0.09997749328613281,
      "step": 724
    },
    {
      "epoch": 0.10102417613042569,
      "grad_norm": 1.5877801179885864,
      "learning_rate": 3.9900227755744515e-05,
      "loss": 0.07759857177734375,
      "step": 725
    },
    {
      "epoch": 0.10116351982164007,
      "grad_norm": 2.3962759971618652,
      "learning_rate": 3.9899285080872346e-05,
      "loss": 0.19809341430664062,
      "step": 726
    },
    {
      "epoch": 0.10130286351285446,
      "grad_norm": 0.696289598941803,
      "learning_rate": 3.9898337984833334e-05,
      "loss": 0.08775138854980469,
      "step": 727
    },
    {
      "epoch": 0.10144220720406884,
      "grad_norm": 3.3508143424987793,
      "learning_rate": 3.98973864678379e-05,
      "loss": 0.15690231323242188,
      "step": 728
    },
    {
      "epoch": 0.10158155089528322,
      "grad_norm": 1.2874993085861206,
      "learning_rate": 3.989643053009747e-05,
      "loss": 0.12796401977539062,
      "step": 729
    },
    {
      "epoch": 0.10172089458649759,
      "grad_norm": 0.5695470571517944,
      "learning_rate": 3.989547017182442e-05,
      "loss": 0.10289192199707031,
      "step": 730
    },
    {
      "epoch": 0.10186023827771197,
      "grad_norm": 1.7321703433990479,
      "learning_rate": 3.989450539323211e-05,
      "loss": 0.10570526123046875,
      "step": 731
    },
    {
      "epoch": 0.10199958196892636,
      "grad_norm": 1.1347671747207642,
      "learning_rate": 3.989353619453491e-05,
      "loss": 0.11036109924316406,
      "step": 732
    },
    {
      "epoch": 0.10213892566014074,
      "grad_norm": 0.5553204417228699,
      "learning_rate": 3.989256257594814e-05,
      "loss": 0.07574272155761719,
      "step": 733
    },
    {
      "epoch": 0.10227826935135512,
      "grad_norm": 1.4166901111602783,
      "learning_rate": 3.989158453768812e-05,
      "loss": 0.1366710662841797,
      "step": 734
    },
    {
      "epoch": 0.1024176130425695,
      "grad_norm": 0.8847742676734924,
      "learning_rate": 3.989060207997215e-05,
      "loss": 0.11824798583984375,
      "step": 735
    },
    {
      "epoch": 0.10255695673378387,
      "grad_norm": 1.0696070194244385,
      "learning_rate": 3.98896152030185e-05,
      "loss": 0.11084556579589844,
      "step": 736
    },
    {
      "epoch": 0.10269630042499826,
      "grad_norm": 0.5786455869674683,
      "learning_rate": 3.988862390704645e-05,
      "loss": 0.10807418823242188,
      "step": 737
    },
    {
      "epoch": 0.10283564411621264,
      "grad_norm": 1.4585732221603394,
      "learning_rate": 3.988762819227623e-05,
      "loss": 0.10635185241699219,
      "step": 738
    },
    {
      "epoch": 0.10297498780742702,
      "grad_norm": 1.1260654926300049,
      "learning_rate": 3.988662805892907e-05,
      "loss": 0.12812042236328125,
      "step": 739
    },
    {
      "epoch": 0.1031143314986414,
      "grad_norm": 0.5539922118186951,
      "learning_rate": 3.988562350722717e-05,
      "loss": 0.09383201599121094,
      "step": 740
    },
    {
      "epoch": 0.10325367518985577,
      "grad_norm": 0.560459315776825,
      "learning_rate": 3.9884614537393724e-05,
      "loss": 0.0829010009765625,
      "step": 741
    },
    {
      "epoch": 0.10339301888107016,
      "grad_norm": 1.5291521549224854,
      "learning_rate": 3.98836011496529e-05,
      "loss": 0.12047004699707031,
      "step": 742
    },
    {
      "epoch": 0.10353236257228454,
      "grad_norm": 1.3131424188613892,
      "learning_rate": 3.9882583344229856e-05,
      "loss": 0.11011505126953125,
      "step": 743
    },
    {
      "epoch": 0.10367170626349892,
      "grad_norm": 1.412013292312622,
      "learning_rate": 3.9881561121350725e-05,
      "loss": 0.143463134765625,
      "step": 744
    },
    {
      "epoch": 0.1038110499547133,
      "grad_norm": 1.620991826057434,
      "learning_rate": 3.988053448124261e-05,
      "loss": 0.1324920654296875,
      "step": 745
    },
    {
      "epoch": 0.10395039364592767,
      "grad_norm": 1.2912758588790894,
      "learning_rate": 3.9879503424133606e-05,
      "loss": 0.1139373779296875,
      "step": 746
    },
    {
      "epoch": 0.10408973733714207,
      "grad_norm": 1.3240872621536255,
      "learning_rate": 3.9878467950252807e-05,
      "loss": 0.11318206787109375,
      "step": 747
    },
    {
      "epoch": 0.10422908102835644,
      "grad_norm": 1.0506455898284912,
      "learning_rate": 3.9877428059830256e-05,
      "loss": 0.0853271484375,
      "step": 748
    },
    {
      "epoch": 0.10436842471957082,
      "grad_norm": 0.7446185946464539,
      "learning_rate": 3.9876383753097004e-05,
      "loss": 0.11227798461914062,
      "step": 749
    },
    {
      "epoch": 0.1045077684107852,
      "grad_norm": 0.5639259219169617,
      "learning_rate": 3.987533503028507e-05,
      "loss": 0.09223556518554688,
      "step": 750
    },
    {
      "epoch": 0.10464711210199958,
      "grad_norm": 0.9279486536979675,
      "learning_rate": 3.987428189162745e-05,
      "loss": 0.09468841552734375,
      "step": 751
    },
    {
      "epoch": 0.10478645579321397,
      "grad_norm": 1.568182110786438,
      "learning_rate": 3.9873224337358134e-05,
      "loss": 0.12052536010742188,
      "step": 752
    },
    {
      "epoch": 0.10492579948442834,
      "grad_norm": 1.6366362571716309,
      "learning_rate": 3.987216236771208e-05,
      "loss": 0.13929367065429688,
      "step": 753
    },
    {
      "epoch": 0.10506514317564272,
      "grad_norm": 2.1157543659210205,
      "learning_rate": 3.987109598292524e-05,
      "loss": 0.13171768188476562,
      "step": 754
    },
    {
      "epoch": 0.1052044868668571,
      "grad_norm": 1.521085262298584,
      "learning_rate": 3.9870025183234536e-05,
      "loss": 0.10310173034667969,
      "step": 755
    },
    {
      "epoch": 0.10534383055807148,
      "grad_norm": 1.1753852367401123,
      "learning_rate": 3.986894996887788e-05,
      "loss": 0.11014366149902344,
      "step": 756
    },
    {
      "epoch": 0.10548317424928587,
      "grad_norm": 0.9424460530281067,
      "learning_rate": 3.986787034009416e-05,
      "loss": 0.10788917541503906,
      "step": 757
    },
    {
      "epoch": 0.10562251794050025,
      "grad_norm": 3.1568048000335693,
      "learning_rate": 3.986678629712323e-05,
      "loss": 0.13332176208496094,
      "step": 758
    },
    {
      "epoch": 0.10576186163171462,
      "grad_norm": 1.9697014093399048,
      "learning_rate": 3.9865697840205955e-05,
      "loss": 0.1415119171142578,
      "step": 759
    },
    {
      "epoch": 0.105901205322929,
      "grad_norm": 1.201587200164795,
      "learning_rate": 3.986460496958416e-05,
      "loss": 0.12663841247558594,
      "step": 760
    },
    {
      "epoch": 0.10604054901414338,
      "grad_norm": 1.1773841381072998,
      "learning_rate": 3.986350768550066e-05,
      "loss": 0.10579490661621094,
      "step": 761
    },
    {
      "epoch": 0.10617989270535777,
      "grad_norm": 1.499826431274414,
      "learning_rate": 3.986240598819925e-05,
      "loss": 0.13878631591796875,
      "step": 762
    },
    {
      "epoch": 0.10631923639657215,
      "grad_norm": 0.4687274992465973,
      "learning_rate": 3.986129987792469e-05,
      "loss": 0.08551979064941406,
      "step": 763
    },
    {
      "epoch": 0.10645858008778653,
      "grad_norm": 1.034368634223938,
      "learning_rate": 3.986018935492274e-05,
      "loss": 0.1614990234375,
      "step": 764
    },
    {
      "epoch": 0.1065979237790009,
      "grad_norm": 1.261549472808838,
      "learning_rate": 3.985907441944013e-05,
      "loss": 0.0895071029663086,
      "step": 765
    },
    {
      "epoch": 0.10673726747021528,
      "grad_norm": 1.2000449895858765,
      "learning_rate": 3.9857955071724575e-05,
      "loss": 0.11925697326660156,
      "step": 766
    },
    {
      "epoch": 0.10687661116142967,
      "grad_norm": 2.438431978225708,
      "learning_rate": 3.9856831312024765e-05,
      "loss": 0.1375598907470703,
      "step": 767
    },
    {
      "epoch": 0.10701595485264405,
      "grad_norm": 1.2121024131774902,
      "learning_rate": 3.985570314059038e-05,
      "loss": 0.10234451293945312,
      "step": 768
    },
    {
      "epoch": 0.10715529854385843,
      "grad_norm": 1.5067170858383179,
      "learning_rate": 3.9854570557672073e-05,
      "loss": 0.10780715942382812,
      "step": 769
    },
    {
      "epoch": 0.1072946422350728,
      "grad_norm": 0.8413577079772949,
      "learning_rate": 3.985343356352147e-05,
      "loss": 0.13018798828125,
      "step": 770
    },
    {
      "epoch": 0.10743398592628718,
      "grad_norm": 3.81839656829834,
      "learning_rate": 3.985229215839119e-05,
      "loss": 0.15117835998535156,
      "step": 771
    },
    {
      "epoch": 0.10757332961750157,
      "grad_norm": 1.1108512878417969,
      "learning_rate": 3.985114634253483e-05,
      "loss": 0.08644676208496094,
      "step": 772
    },
    {
      "epoch": 0.10771267330871595,
      "grad_norm": 0.8536274433135986,
      "learning_rate": 3.9849996116206966e-05,
      "loss": 0.08495903015136719,
      "step": 773
    },
    {
      "epoch": 0.10785201699993033,
      "grad_norm": 0.5108976364135742,
      "learning_rate": 3.9848841479663146e-05,
      "loss": 0.10399341583251953,
      "step": 774
    },
    {
      "epoch": 0.1079913606911447,
      "grad_norm": 0.9988969564437866,
      "learning_rate": 3.984768243315991e-05,
      "loss": 0.10687065124511719,
      "step": 775
    },
    {
      "epoch": 0.10813070438235908,
      "grad_norm": 0.6871193647384644,
      "learning_rate": 3.984651897695476e-05,
      "loss": 0.09221076965332031,
      "step": 776
    },
    {
      "epoch": 0.10827004807357347,
      "grad_norm": 2.1453189849853516,
      "learning_rate": 3.9845351111306196e-05,
      "loss": 0.118377685546875,
      "step": 777
    },
    {
      "epoch": 0.10840939176478785,
      "grad_norm": 2.358560800552368,
      "learning_rate": 3.98441788364737e-05,
      "loss": 0.1477518081665039,
      "step": 778
    },
    {
      "epoch": 0.10854873545600223,
      "grad_norm": 0.4648269712924957,
      "learning_rate": 3.984300215271771e-05,
      "loss": 0.0981597900390625,
      "step": 779
    },
    {
      "epoch": 0.1086880791472166,
      "grad_norm": 1.3363220691680908,
      "learning_rate": 3.984182106029967e-05,
      "loss": 0.08282470703125,
      "step": 780
    },
    {
      "epoch": 0.10882742283843098,
      "grad_norm": 1.3709101676940918,
      "learning_rate": 3.984063555948198e-05,
      "loss": 0.11137962341308594,
      "step": 781
    },
    {
      "epoch": 0.10896676652964538,
      "grad_norm": 1.0121681690216064,
      "learning_rate": 3.9839445650528046e-05,
      "loss": 0.1295604705810547,
      "step": 782
    },
    {
      "epoch": 0.10910611022085975,
      "grad_norm": 2.054105520248413,
      "learning_rate": 3.983825133370223e-05,
      "loss": 0.12508392333984375,
      "step": 783
    },
    {
      "epoch": 0.10924545391207413,
      "grad_norm": 0.7920443415641785,
      "learning_rate": 3.983705260926988e-05,
      "loss": 0.14387130737304688,
      "step": 784
    },
    {
      "epoch": 0.10938479760328851,
      "grad_norm": 2.895064353942871,
      "learning_rate": 3.983584947749733e-05,
      "loss": 0.14847755432128906,
      "step": 785
    },
    {
      "epoch": 0.10952414129450289,
      "grad_norm": 0.8737805485725403,
      "learning_rate": 3.983464193865188e-05,
      "loss": 0.11993789672851562,
      "step": 786
    },
    {
      "epoch": 0.10966348498571728,
      "grad_norm": 0.842776358127594,
      "learning_rate": 3.9833429993001827e-05,
      "loss": 0.08897781372070312,
      "step": 787
    },
    {
      "epoch": 0.10980282867693165,
      "grad_norm": 2.2271933555603027,
      "learning_rate": 3.983221364081644e-05,
      "loss": 0.15197372436523438,
      "step": 788
    },
    {
      "epoch": 0.10994217236814603,
      "grad_norm": 0.9457564353942871,
      "learning_rate": 3.983099288236595e-05,
      "loss": 0.09394645690917969,
      "step": 789
    },
    {
      "epoch": 0.11008151605936041,
      "grad_norm": 0.9889166355133057,
      "learning_rate": 3.98297677179216e-05,
      "loss": 0.1539154052734375,
      "step": 790
    },
    {
      "epoch": 0.11022085975057479,
      "grad_norm": 2.6877601146698,
      "learning_rate": 3.982853814775558e-05,
      "loss": 0.15602493286132812,
      "step": 791
    },
    {
      "epoch": 0.11036020344178918,
      "grad_norm": 2.448507308959961,
      "learning_rate": 3.982730417214107e-05,
      "loss": 0.12117385864257812,
      "step": 792
    },
    {
      "epoch": 0.11049954713300356,
      "grad_norm": 2.2778351306915283,
      "learning_rate": 3.982606579135225e-05,
      "loss": 0.1409473419189453,
      "step": 793
    },
    {
      "epoch": 0.11063889082421793,
      "grad_norm": 0.857330858707428,
      "learning_rate": 3.982482300566424e-05,
      "loss": 0.09689712524414062,
      "step": 794
    },
    {
      "epoch": 0.11077823451543231,
      "grad_norm": 1.0533100366592407,
      "learning_rate": 3.982357581535317e-05,
      "loss": 0.10053062438964844,
      "step": 795
    },
    {
      "epoch": 0.11091757820664669,
      "grad_norm": 1.054516077041626,
      "learning_rate": 3.9822324220696134e-05,
      "loss": 0.11700630187988281,
      "step": 796
    },
    {
      "epoch": 0.11105692189786108,
      "grad_norm": 1.549259901046753,
      "learning_rate": 3.98210682219712e-05,
      "loss": 0.12303924560546875,
      "step": 797
    },
    {
      "epoch": 0.11119626558907546,
      "grad_norm": 1.1947675943374634,
      "learning_rate": 3.9819807819457444e-05,
      "loss": 0.12286567687988281,
      "step": 798
    },
    {
      "epoch": 0.11133560928028984,
      "grad_norm": 1.2768586874008179,
      "learning_rate": 3.9818543013434874e-05,
      "loss": 0.13268661499023438,
      "step": 799
    },
    {
      "epoch": 0.11147495297150421,
      "grad_norm": 1.2711249589920044,
      "learning_rate": 3.9817273804184514e-05,
      "loss": 0.10602188110351562,
      "step": 800
    },
    {
      "epoch": 0.11161429666271859,
      "grad_norm": 2.547293186187744,
      "learning_rate": 3.981600019198835e-05,
      "loss": 0.14692306518554688,
      "step": 801
    },
    {
      "epoch": 0.11175364035393298,
      "grad_norm": 1.1622370481491089,
      "learning_rate": 3.981472217712935e-05,
      "loss": 0.10278034210205078,
      "step": 802
    },
    {
      "epoch": 0.11189298404514736,
      "grad_norm": 0.5008900761604309,
      "learning_rate": 3.9813439759891466e-05,
      "loss": 0.09461212158203125,
      "step": 803
    },
    {
      "epoch": 0.11203232773636174,
      "grad_norm": 1.5992443561553955,
      "learning_rate": 3.981215294055961e-05,
      "loss": 0.11413002014160156,
      "step": 804
    },
    {
      "epoch": 0.11217167142757611,
      "grad_norm": 1.0794358253479004,
      "learning_rate": 3.981086171941969e-05,
      "loss": 0.105010986328125,
      "step": 805
    },
    {
      "epoch": 0.11231101511879049,
      "grad_norm": 1.069921612739563,
      "learning_rate": 3.9809566096758586e-05,
      "loss": 0.09340667724609375,
      "step": 806
    },
    {
      "epoch": 0.11245035881000488,
      "grad_norm": 1.005802035331726,
      "learning_rate": 3.9808266072864156e-05,
      "loss": 0.11477279663085938,
      "step": 807
    },
    {
      "epoch": 0.11258970250121926,
      "grad_norm": 0.7565615177154541,
      "learning_rate": 3.980696164802523e-05,
      "loss": 0.1026611328125,
      "step": 808
    },
    {
      "epoch": 0.11272904619243364,
      "grad_norm": 1.95918607711792,
      "learning_rate": 3.980565282253164e-05,
      "loss": 0.13885116577148438,
      "step": 809
    },
    {
      "epoch": 0.11286838988364802,
      "grad_norm": 2.142925262451172,
      "learning_rate": 3.9804339596674146e-05,
      "loss": 0.1059713363647461,
      "step": 810
    },
    {
      "epoch": 0.11300773357486239,
      "grad_norm": 1.4271775484085083,
      "learning_rate": 3.980302197074455e-05,
      "loss": 0.08303260803222656,
      "step": 811
    },
    {
      "epoch": 0.11314707726607678,
      "grad_norm": 1.2232320308685303,
      "learning_rate": 3.9801699945035573e-05,
      "loss": 0.10942459106445312,
      "step": 812
    },
    {
      "epoch": 0.11328642095729116,
      "grad_norm": 1.390616774559021,
      "learning_rate": 3.980037351984095e-05,
      "loss": 0.10650444030761719,
      "step": 813
    },
    {
      "epoch": 0.11342576464850554,
      "grad_norm": 2.2584481239318848,
      "learning_rate": 3.979904269545538e-05,
      "loss": 0.12761497497558594,
      "step": 814
    },
    {
      "epoch": 0.11356510833971992,
      "grad_norm": 1.9571324586868286,
      "learning_rate": 3.979770747217455e-05,
      "loss": 0.10061168670654297,
      "step": 815
    },
    {
      "epoch": 0.1137044520309343,
      "grad_norm": 1.5997616052627563,
      "learning_rate": 3.97963678502951e-05,
      "loss": 0.09047698974609375,
      "step": 816
    },
    {
      "epoch": 0.11384379572214869,
      "grad_norm": 0.9450207948684692,
      "learning_rate": 3.979502383011468e-05,
      "loss": 0.07411384582519531,
      "step": 817
    },
    {
      "epoch": 0.11398313941336306,
      "grad_norm": 1.723956823348999,
      "learning_rate": 3.979367541193189e-05,
      "loss": 0.14814186096191406,
      "step": 818
    },
    {
      "epoch": 0.11412248310457744,
      "grad_norm": 1.6721744537353516,
      "learning_rate": 3.9792322596046326e-05,
      "loss": 0.12276840209960938,
      "step": 819
    },
    {
      "epoch": 0.11426182679579182,
      "grad_norm": 1.1367372274398804,
      "learning_rate": 3.979096538275854e-05,
      "loss": 0.11266899108886719,
      "step": 820
    },
    {
      "epoch": 0.1144011704870062,
      "grad_norm": 1.3513717651367188,
      "learning_rate": 3.978960377237009e-05,
      "loss": 0.16111183166503906,
      "step": 821
    },
    {
      "epoch": 0.11454051417822059,
      "grad_norm": 1.443853735923767,
      "learning_rate": 3.978823776518348e-05,
      "loss": 0.15060043334960938,
      "step": 822
    },
    {
      "epoch": 0.11467985786943496,
      "grad_norm": 2.0307443141937256,
      "learning_rate": 3.978686736150221e-05,
      "loss": 0.1636981964111328,
      "step": 823
    },
    {
      "epoch": 0.11481920156064934,
      "grad_norm": 1.1807634830474854,
      "learning_rate": 3.978549256163075e-05,
      "loss": 0.12828445434570312,
      "step": 824
    },
    {
      "epoch": 0.11495854525186372,
      "grad_norm": 0.577113687992096,
      "learning_rate": 3.978411336587457e-05,
      "loss": 0.10004234313964844,
      "step": 825
    },
    {
      "epoch": 0.1150978889430781,
      "grad_norm": 1.1561959981918335,
      "learning_rate": 3.978272977454006e-05,
      "loss": 0.08631706237792969,
      "step": 826
    },
    {
      "epoch": 0.11523723263429249,
      "grad_norm": 0.4761141240596771,
      "learning_rate": 3.978134178793465e-05,
      "loss": 0.09083938598632812,
      "step": 827
    },
    {
      "epoch": 0.11537657632550687,
      "grad_norm": 0.4336426258087158,
      "learning_rate": 3.977994940636671e-05,
      "loss": 0.10993766784667969,
      "step": 828
    },
    {
      "epoch": 0.11551592001672124,
      "grad_norm": 0.4969298839569092,
      "learning_rate": 3.9778552630145595e-05,
      "loss": 0.08218574523925781,
      "step": 829
    },
    {
      "epoch": 0.11565526370793562,
      "grad_norm": 2.006443738937378,
      "learning_rate": 3.977715145958164e-05,
      "loss": 0.2059001922607422,
      "step": 830
    },
    {
      "epoch": 0.11579460739915,
      "grad_norm": 0.9672330021858215,
      "learning_rate": 3.9775745894986155e-05,
      "loss": 0.12047958374023438,
      "step": 831
    },
    {
      "epoch": 0.11593395109036439,
      "grad_norm": 1.297072410583496,
      "learning_rate": 3.9774335936671414e-05,
      "loss": 0.11907958984375,
      "step": 832
    },
    {
      "epoch": 0.11607329478157877,
      "grad_norm": 0.895574688911438,
      "learning_rate": 3.977292158495068e-05,
      "loss": 0.09975624084472656,
      "step": 833
    },
    {
      "epoch": 0.11621263847279314,
      "grad_norm": 0.6433578133583069,
      "learning_rate": 3.9771502840138196e-05,
      "loss": 0.08802413940429688,
      "step": 834
    },
    {
      "epoch": 0.11635198216400752,
      "grad_norm": 0.6772577166557312,
      "learning_rate": 3.9770079702549174e-05,
      "loss": 0.09848785400390625,
      "step": 835
    },
    {
      "epoch": 0.1164913258552219,
      "grad_norm": 0.8166855573654175,
      "learning_rate": 3.9768652172499804e-05,
      "loss": 0.09238052368164062,
      "step": 836
    },
    {
      "epoch": 0.11663066954643629,
      "grad_norm": 1.3394217491149902,
      "learning_rate": 3.9767220250307244e-05,
      "loss": 0.09633636474609375,
      "step": 837
    },
    {
      "epoch": 0.11677001323765067,
      "grad_norm": 0.967022716999054,
      "learning_rate": 3.976578393628963e-05,
      "loss": 0.10444450378417969,
      "step": 838
    },
    {
      "epoch": 0.11690935692886505,
      "grad_norm": 0.5620585680007935,
      "learning_rate": 3.9764343230766096e-05,
      "loss": 0.10424232482910156,
      "step": 839
    },
    {
      "epoch": 0.11704870062007942,
      "grad_norm": 0.8165538311004639,
      "learning_rate": 3.976289813405672e-05,
      "loss": 0.09117507934570312,
      "step": 840
    },
    {
      "epoch": 0.1171880443112938,
      "grad_norm": 0.811055064201355,
      "learning_rate": 3.9761448646482576e-05,
      "loss": 0.09662818908691406,
      "step": 841
    },
    {
      "epoch": 0.11732738800250819,
      "grad_norm": 1.4214133024215698,
      "learning_rate": 3.975999476836571e-05,
      "loss": 0.10876274108886719,
      "step": 842
    },
    {
      "epoch": 0.11746673169372257,
      "grad_norm": 0.5501617789268494,
      "learning_rate": 3.9758536500029116e-05,
      "loss": 0.08095359802246094,
      "step": 843
    },
    {
      "epoch": 0.11760607538493695,
      "grad_norm": 0.5954941511154175,
      "learning_rate": 3.975707384179682e-05,
      "loss": 0.09542465209960938,
      "step": 844
    },
    {
      "epoch": 0.11774541907615133,
      "grad_norm": 1.2494981288909912,
      "learning_rate": 3.9755606793993776e-05,
      "loss": 0.10699462890625,
      "step": 845
    },
    {
      "epoch": 0.1178847627673657,
      "grad_norm": 1.5655019283294678,
      "learning_rate": 3.9754135356945934e-05,
      "loss": 0.12392997741699219,
      "step": 846
    },
    {
      "epoch": 0.1180241064585801,
      "grad_norm": 0.6734741926193237,
      "learning_rate": 3.9752659530980205e-05,
      "loss": 0.10831451416015625,
      "step": 847
    },
    {
      "epoch": 0.11816345014979447,
      "grad_norm": 2.138023853302002,
      "learning_rate": 3.975117931642449e-05,
      "loss": 0.17435073852539062,
      "step": 848
    },
    {
      "epoch": 0.11830279384100885,
      "grad_norm": 0.9267823696136475,
      "learning_rate": 3.9749694713607654e-05,
      "loss": 0.09423065185546875,
      "step": 849
    },
    {
      "epoch": 0.11844213753222323,
      "grad_norm": 0.5772899985313416,
      "learning_rate": 3.974820572285955e-05,
      "loss": 0.07808494567871094,
      "step": 850
    },
    {
      "epoch": 0.1185814812234376,
      "grad_norm": 0.5302408933639526,
      "learning_rate": 3.9746712344510996e-05,
      "loss": 0.094329833984375,
      "step": 851
    },
    {
      "epoch": 0.118720824914652,
      "grad_norm": 0.8932099342346191,
      "learning_rate": 3.9745214578893784e-05,
      "loss": 0.09273433685302734,
      "step": 852
    },
    {
      "epoch": 0.11886016860586637,
      "grad_norm": 1.260711669921875,
      "learning_rate": 3.974371242634068e-05,
      "loss": 0.16221237182617188,
      "step": 853
    },
    {
      "epoch": 0.11899951229708075,
      "grad_norm": 0.7477459907531738,
      "learning_rate": 3.9742205887185434e-05,
      "loss": 0.09038543701171875,
      "step": 854
    },
    {
      "epoch": 0.11913885598829513,
      "grad_norm": 1.0658917427062988,
      "learning_rate": 3.974069496176277e-05,
      "loss": 0.11784934997558594,
      "step": 855
    },
    {
      "epoch": 0.1192781996795095,
      "grad_norm": 0.6801803708076477,
      "learning_rate": 3.973917965040836e-05,
      "loss": 0.11389923095703125,
      "step": 856
    },
    {
      "epoch": 0.1194175433707239,
      "grad_norm": 0.9303820133209229,
      "learning_rate": 3.973765995345889e-05,
      "loss": 0.1399993896484375,
      "step": 857
    },
    {
      "epoch": 0.11955688706193827,
      "grad_norm": 1.3556439876556396,
      "learning_rate": 3.9736135871251994e-05,
      "loss": 0.1107940673828125,
      "step": 858
    },
    {
      "epoch": 0.11969623075315265,
      "grad_norm": 0.8433564901351929,
      "learning_rate": 3.9734607404126293e-05,
      "loss": 0.11928367614746094,
      "step": 859
    },
    {
      "epoch": 0.11983557444436703,
      "grad_norm": 0.6797468066215515,
      "learning_rate": 3.973307455242138e-05,
      "loss": 0.10315418243408203,
      "step": 860
    },
    {
      "epoch": 0.11997491813558141,
      "grad_norm": 1.3954963684082031,
      "learning_rate": 3.9731537316477806e-05,
      "loss": 0.12972259521484375,
      "step": 861
    },
    {
      "epoch": 0.1201142618267958,
      "grad_norm": 0.9100663661956787,
      "learning_rate": 3.9729995696637125e-05,
      "loss": 0.1049652099609375,
      "step": 862
    },
    {
      "epoch": 0.12025360551801018,
      "grad_norm": 0.7759761810302734,
      "learning_rate": 3.972844969324184e-05,
      "loss": 0.08396720886230469,
      "step": 863
    },
    {
      "epoch": 0.12039294920922455,
      "grad_norm": 4.135852336883545,
      "learning_rate": 3.9726899306635446e-05,
      "loss": 0.21060943603515625,
      "step": 864
    },
    {
      "epoch": 0.12053229290043893,
      "grad_norm": 1.6119786500930786,
      "learning_rate": 3.9725344537162394e-05,
      "loss": 0.13420677185058594,
      "step": 865
    },
    {
      "epoch": 0.12067163659165331,
      "grad_norm": 0.5810844898223877,
      "learning_rate": 3.972378538516813e-05,
      "loss": 0.11736488342285156,
      "step": 866
    },
    {
      "epoch": 0.1208109802828677,
      "grad_norm": 1.3256585597991943,
      "learning_rate": 3.972222185099905e-05,
      "loss": 0.13250732421875,
      "step": 867
    },
    {
      "epoch": 0.12095032397408208,
      "grad_norm": 0.759692370891571,
      "learning_rate": 3.972065393500254e-05,
      "loss": 0.07603263854980469,
      "step": 868
    },
    {
      "epoch": 0.12108966766529645,
      "grad_norm": 1.2146332263946533,
      "learning_rate": 3.971908163752696e-05,
      "loss": 0.14098024368286133,
      "step": 869
    },
    {
      "epoch": 0.12122901135651083,
      "grad_norm": 0.5106416940689087,
      "learning_rate": 3.9717504958921634e-05,
      "loss": 0.10230636596679688,
      "step": 870
    },
    {
      "epoch": 0.12136835504772521,
      "grad_norm": 0.8769246935844421,
      "learning_rate": 3.971592389953686e-05,
      "loss": 0.11970901489257812,
      "step": 871
    },
    {
      "epoch": 0.1215076987389396,
      "grad_norm": 0.863617479801178,
      "learning_rate": 3.9714338459723924e-05,
      "loss": 0.12817001342773438,
      "step": 872
    },
    {
      "epoch": 0.12164704243015398,
      "grad_norm": 0.7919303774833679,
      "learning_rate": 3.9712748639835056e-05,
      "loss": 0.10923385620117188,
      "step": 873
    },
    {
      "epoch": 0.12178638612136836,
      "grad_norm": 0.9119038581848145,
      "learning_rate": 3.97111544402235e-05,
      "loss": 0.1362323760986328,
      "step": 874
    },
    {
      "epoch": 0.12192572981258273,
      "grad_norm": 0.9829415082931519,
      "learning_rate": 3.970955586124344e-05,
      "loss": 0.10671043395996094,
      "step": 875
    },
    {
      "epoch": 0.12206507350379711,
      "grad_norm": 0.785045862197876,
      "learning_rate": 3.9707952903250045e-05,
      "loss": 0.14325714111328125,
      "step": 876
    },
    {
      "epoch": 0.1222044171950115,
      "grad_norm": 0.38225650787353516,
      "learning_rate": 3.9706345566599454e-05,
      "loss": 0.06811904907226562,
      "step": 877
    },
    {
      "epoch": 0.12234376088622588,
      "grad_norm": 1.3383413553237915,
      "learning_rate": 3.9704733851648785e-05,
      "loss": 0.1023101806640625,
      "step": 878
    },
    {
      "epoch": 0.12248310457744026,
      "grad_norm": 1.366274118423462,
      "learning_rate": 3.970311775875611e-05,
      "loss": 0.12511253356933594,
      "step": 879
    },
    {
      "epoch": 0.12262244826865464,
      "grad_norm": 0.8313373923301697,
      "learning_rate": 3.9701497288280506e-05,
      "loss": 0.1096954345703125,
      "step": 880
    },
    {
      "epoch": 0.12276179195986901,
      "grad_norm": 0.739745557308197,
      "learning_rate": 3.9699872440582e-05,
      "loss": 0.11933517456054688,
      "step": 881
    },
    {
      "epoch": 0.1229011356510834,
      "grad_norm": 0.6557244658470154,
      "learning_rate": 3.969824321602159e-05,
      "loss": 0.09633255004882812,
      "step": 882
    },
    {
      "epoch": 0.12304047934229778,
      "grad_norm": 1.0902858972549438,
      "learning_rate": 3.969660961496126e-05,
      "loss": 0.11241912841796875,
      "step": 883
    },
    {
      "epoch": 0.12317982303351216,
      "grad_norm": 1.4936882257461548,
      "learning_rate": 3.969497163776395e-05,
      "loss": 0.13085269927978516,
      "step": 884
    },
    {
      "epoch": 0.12331916672472654,
      "grad_norm": 2.4463281631469727,
      "learning_rate": 3.9693329284793586e-05,
      "loss": 0.16193389892578125,
      "step": 885
    },
    {
      "epoch": 0.12345851041594091,
      "grad_norm": 0.9028124213218689,
      "learning_rate": 3.9691682556415064e-05,
      "loss": 0.14135169982910156,
      "step": 886
    },
    {
      "epoch": 0.12359785410715529,
      "grad_norm": 1.8216041326522827,
      "learning_rate": 3.969003145299424e-05,
      "loss": 0.1624908447265625,
      "step": 887
    },
    {
      "epoch": 0.12373719779836968,
      "grad_norm": 0.8420243263244629,
      "learning_rate": 3.968837597489797e-05,
      "loss": 0.1012115478515625,
      "step": 888
    },
    {
      "epoch": 0.12387654148958406,
      "grad_norm": 0.6212747097015381,
      "learning_rate": 3.968671612249404e-05,
      "loss": 0.10637664794921875,
      "step": 889
    },
    {
      "epoch": 0.12401588518079844,
      "grad_norm": 0.8405910730361938,
      "learning_rate": 3.968505189615125e-05,
      "loss": 0.10589408874511719,
      "step": 890
    },
    {
      "epoch": 0.12415522887201282,
      "grad_norm": 1.072261095046997,
      "learning_rate": 3.9683383296239345e-05,
      "loss": 0.09920883178710938,
      "step": 891
    },
    {
      "epoch": 0.12429457256322719,
      "grad_norm": 2.0888612270355225,
      "learning_rate": 3.968171032312905e-05,
      "loss": 0.1413726806640625,
      "step": 892
    },
    {
      "epoch": 0.12443391625444158,
      "grad_norm": 0.7281012535095215,
      "learning_rate": 3.968003297719206e-05,
      "loss": 0.09989166259765625,
      "step": 893
    },
    {
      "epoch": 0.12457325994565596,
      "grad_norm": 0.6744201183319092,
      "learning_rate": 3.9678351258801046e-05,
      "loss": 0.10813140869140625,
      "step": 894
    },
    {
      "epoch": 0.12471260363687034,
      "grad_norm": 1.953149676322937,
      "learning_rate": 3.9676665168329655e-05,
      "loss": 0.1544818878173828,
      "step": 895
    },
    {
      "epoch": 0.12485194732808472,
      "grad_norm": 1.028326153755188,
      "learning_rate": 3.967497470615248e-05,
      "loss": 0.13013648986816406,
      "step": 896
    },
    {
      "epoch": 0.1249912910192991,
      "grad_norm": 0.8883559703826904,
      "learning_rate": 3.967327987264512e-05,
      "loss": 0.12459373474121094,
      "step": 897
    },
    {
      "epoch": 0.12513063471051347,
      "grad_norm": 0.9527366757392883,
      "learning_rate": 3.967158066818411e-05,
      "loss": 0.13219070434570312,
      "step": 898
    },
    {
      "epoch": 0.12526997840172785,
      "grad_norm": 0.9646626114845276,
      "learning_rate": 3.9669877093146995e-05,
      "loss": 0.11056900024414062,
      "step": 899
    },
    {
      "epoch": 0.12540932209294225,
      "grad_norm": 0.8447823524475098,
      "learning_rate": 3.966816914791226e-05,
      "loss": 0.10135078430175781,
      "step": 900
    },
    {
      "epoch": 0.12554866578415663,
      "grad_norm": 1.1658153533935547,
      "learning_rate": 3.9666456832859365e-05,
      "loss": 0.1222381591796875,
      "step": 901
    },
    {
      "epoch": 0.125688009475371,
      "grad_norm": 1.094548225402832,
      "learning_rate": 3.966474014836876e-05,
      "loss": 0.09234237670898438,
      "step": 902
    },
    {
      "epoch": 0.1258273531665854,
      "grad_norm": 1.0840222835540771,
      "learning_rate": 3.9663019094821843e-05,
      "loss": 0.1737346649169922,
      "step": 903
    },
    {
      "epoch": 0.12596669685779976,
      "grad_norm": 0.8250463604927063,
      "learning_rate": 3.9661293672601006e-05,
      "loss": 0.10967826843261719,
      "step": 904
    },
    {
      "epoch": 0.12610604054901414,
      "grad_norm": 0.7635146379470825,
      "learning_rate": 3.965956388208959e-05,
      "loss": 0.10135078430175781,
      "step": 905
    },
    {
      "epoch": 0.12624538424022852,
      "grad_norm": 2.660240411758423,
      "learning_rate": 3.965782972367191e-05,
      "loss": 0.1481761932373047,
      "step": 906
    },
    {
      "epoch": 0.1263847279314429,
      "grad_norm": 1.2389339208602905,
      "learning_rate": 3.965609119773326e-05,
      "loss": 0.13171768188476562,
      "step": 907
    },
    {
      "epoch": 0.12652407162265727,
      "grad_norm": 0.7233284115791321,
      "learning_rate": 3.9654348304659905e-05,
      "loss": 0.10956573486328125,
      "step": 908
    },
    {
      "epoch": 0.12666341531387165,
      "grad_norm": 0.9505659341812134,
      "learning_rate": 3.965260104483907e-05,
      "loss": 0.12093353271484375,
      "step": 909
    },
    {
      "epoch": 0.12680275900508606,
      "grad_norm": 0.5517993569374084,
      "learning_rate": 3.965084941865896e-05,
      "loss": 0.10152244567871094,
      "step": 910
    },
    {
      "epoch": 0.12694210269630044,
      "grad_norm": 0.6311435699462891,
      "learning_rate": 3.964909342650875e-05,
      "loss": 0.10203361511230469,
      "step": 911
    },
    {
      "epoch": 0.1270814463875148,
      "grad_norm": 0.8900637030601501,
      "learning_rate": 3.964733306877857e-05,
      "loss": 0.15645980834960938,
      "step": 912
    },
    {
      "epoch": 0.1272207900787292,
      "grad_norm": 0.28614675998687744,
      "learning_rate": 3.964556834585954e-05,
      "loss": 0.06302642822265625,
      "step": 913
    },
    {
      "epoch": 0.12736013376994357,
      "grad_norm": 0.8549137711524963,
      "learning_rate": 3.9643799258143745e-05,
      "loss": 0.12442493438720703,
      "step": 914
    },
    {
      "epoch": 0.12749947746115795,
      "grad_norm": 0.5446961522102356,
      "learning_rate": 3.9642025806024226e-05,
      "loss": 0.09725761413574219,
      "step": 915
    },
    {
      "epoch": 0.12763882115237232,
      "grad_norm": 0.457271933555603,
      "learning_rate": 3.964024798989501e-05,
      "loss": 0.06842231750488281,
      "step": 916
    },
    {
      "epoch": 0.1277781648435867,
      "grad_norm": 0.5276815295219421,
      "learning_rate": 3.963846581015109e-05,
      "loss": 0.09306907653808594,
      "step": 917
    },
    {
      "epoch": 0.12791750853480108,
      "grad_norm": 0.6969532370567322,
      "learning_rate": 3.963667926718841e-05,
      "loss": 0.08910179138183594,
      "step": 918
    },
    {
      "epoch": 0.12805685222601546,
      "grad_norm": 1.6845422983169556,
      "learning_rate": 3.9634888361403916e-05,
      "loss": 0.11983680725097656,
      "step": 919
    },
    {
      "epoch": 0.12819619591722986,
      "grad_norm": 0.48111677169799805,
      "learning_rate": 3.96330930931955e-05,
      "loss": 0.08892822265625,
      "step": 920
    },
    {
      "epoch": 0.12833553960844424,
      "grad_norm": 0.7504842877388,
      "learning_rate": 3.963129346296203e-05,
      "loss": 0.09876632690429688,
      "step": 921
    },
    {
      "epoch": 0.12847488329965862,
      "grad_norm": 0.8554426431655884,
      "learning_rate": 3.9629489471103334e-05,
      "loss": 0.09718894958496094,
      "step": 922
    },
    {
      "epoch": 0.128614226990873,
      "grad_norm": 0.825312614440918,
      "learning_rate": 3.962768111802023e-05,
      "loss": 0.10096549987792969,
      "step": 923
    },
    {
      "epoch": 0.12875357068208737,
      "grad_norm": 1.7146801948547363,
      "learning_rate": 3.96258684041145e-05,
      "loss": 0.21262741088867188,
      "step": 924
    },
    {
      "epoch": 0.12889291437330175,
      "grad_norm": 0.929318368434906,
      "learning_rate": 3.9624051329788875e-05,
      "loss": 0.14669322967529297,
      "step": 925
    },
    {
      "epoch": 0.12903225806451613,
      "grad_norm": 1.0911771059036255,
      "learning_rate": 3.9622229895447054e-05,
      "loss": 0.10153388977050781,
      "step": 926
    },
    {
      "epoch": 0.1291716017557305,
      "grad_norm": 1.5078743696212769,
      "learning_rate": 3.962040410149375e-05,
      "loss": 0.14362144470214844,
      "step": 927
    },
    {
      "epoch": 0.12931094544694488,
      "grad_norm": 0.8618125915527344,
      "learning_rate": 3.961857394833459e-05,
      "loss": 0.10207939147949219,
      "step": 928
    },
    {
      "epoch": 0.12945028913815926,
      "grad_norm": 0.47343820333480835,
      "learning_rate": 3.96167394363762e-05,
      "loss": 0.08660888671875,
      "step": 929
    },
    {
      "epoch": 0.12958963282937366,
      "grad_norm": 0.825143039226532,
      "learning_rate": 3.9614900566026154e-05,
      "loss": 0.08667755126953125,
      "step": 930
    },
    {
      "epoch": 0.12972897652058804,
      "grad_norm": 0.7102736234664917,
      "learning_rate": 3.961305733769303e-05,
      "loss": 0.1053924560546875,
      "step": 931
    },
    {
      "epoch": 0.12986832021180242,
      "grad_norm": 0.8108076453208923,
      "learning_rate": 3.961120975178634e-05,
      "loss": 0.0892791748046875,
      "step": 932
    },
    {
      "epoch": 0.1300076639030168,
      "grad_norm": 1.125351071357727,
      "learning_rate": 3.960935780871657e-05,
      "loss": 0.13264846801757812,
      "step": 933
    },
    {
      "epoch": 0.13014700759423117,
      "grad_norm": 1.4581644535064697,
      "learning_rate": 3.9607501508895185e-05,
      "loss": 0.16429901123046875,
      "step": 934
    },
    {
      "epoch": 0.13028635128544555,
      "grad_norm": 0.5428231954574585,
      "learning_rate": 3.960564085273461e-05,
      "loss": 0.1180422306060791,
      "step": 935
    },
    {
      "epoch": 0.13042569497665993,
      "grad_norm": 1.9891698360443115,
      "learning_rate": 3.9603775840648243e-05,
      "loss": 0.19652366638183594,
      "step": 936
    },
    {
      "epoch": 0.1305650386678743,
      "grad_norm": 0.7734001874923706,
      "learning_rate": 3.9601906473050446e-05,
      "loss": 0.10776519775390625,
      "step": 937
    },
    {
      "epoch": 0.13070438235908868,
      "grad_norm": 0.48310205340385437,
      "learning_rate": 3.960003275035655e-05,
      "loss": 0.08874893188476562,
      "step": 938
    },
    {
      "epoch": 0.13084372605030306,
      "grad_norm": 0.6201640963554382,
      "learning_rate": 3.959815467298285e-05,
      "loss": 0.08957290649414062,
      "step": 939
    },
    {
      "epoch": 0.13098306974151747,
      "grad_norm": 1.7165266275405884,
      "learning_rate": 3.9596272241346625e-05,
      "loss": 0.11921882629394531,
      "step": 940
    },
    {
      "epoch": 0.13112241343273184,
      "grad_norm": 0.4996984004974365,
      "learning_rate": 3.959438545586609e-05,
      "loss": 0.10185050964355469,
      "step": 941
    },
    {
      "epoch": 0.13126175712394622,
      "grad_norm": 0.6177219152450562,
      "learning_rate": 3.959249431696046e-05,
      "loss": 0.10330581665039062,
      "step": 942
    },
    {
      "epoch": 0.1314011008151606,
      "grad_norm": 0.3020438551902771,
      "learning_rate": 3.9590598825049896e-05,
      "loss": 0.06919097900390625,
      "step": 943
    },
    {
      "epoch": 0.13154044450637498,
      "grad_norm": 0.6501237750053406,
      "learning_rate": 3.958869898055553e-05,
      "loss": 0.10583877563476562,
      "step": 944
    },
    {
      "epoch": 0.13167978819758935,
      "grad_norm": 0.5967585444450378,
      "learning_rate": 3.9586794783899464e-05,
      "loss": 0.10979461669921875,
      "step": 945
    },
    {
      "epoch": 0.13181913188880373,
      "grad_norm": 0.8810577392578125,
      "learning_rate": 3.958488623550478e-05,
      "loss": 0.09739875793457031,
      "step": 946
    },
    {
      "epoch": 0.1319584755800181,
      "grad_norm": 1.4574545621871948,
      "learning_rate": 3.95829733357955e-05,
      "loss": 0.12108612060546875,
      "step": 947
    },
    {
      "epoch": 0.1320978192712325,
      "grad_norm": 0.6736417412757874,
      "learning_rate": 3.958105608519663e-05,
      "loss": 0.07841110229492188,
      "step": 948
    },
    {
      "epoch": 0.13223716296244686,
      "grad_norm": 1.2602319717407227,
      "learning_rate": 3.957913448413415e-05,
      "loss": 0.11762428283691406,
      "step": 949
    },
    {
      "epoch": 0.13237650665366127,
      "grad_norm": 0.9865601658821106,
      "learning_rate": 3.957720853303499e-05,
      "loss": 0.182098388671875,
      "step": 950
    },
    {
      "epoch": 0.13251585034487565,
      "grad_norm": 1.799268126487732,
      "learning_rate": 3.9575278232327036e-05,
      "loss": 0.09648895263671875,
      "step": 951
    },
    {
      "epoch": 0.13265519403609002,
      "grad_norm": 3.0479214191436768,
      "learning_rate": 3.957334358243917e-05,
      "loss": 0.16586875915527344,
      "step": 952
    },
    {
      "epoch": 0.1327945377273044,
      "grad_norm": 1.4267535209655762,
      "learning_rate": 3.957140458380123e-05,
      "loss": 0.13303184509277344,
      "step": 953
    },
    {
      "epoch": 0.13293388141851878,
      "grad_norm": 0.8347949385643005,
      "learning_rate": 3.956946123684402e-05,
      "loss": 0.10215568542480469,
      "step": 954
    },
    {
      "epoch": 0.13307322510973316,
      "grad_norm": 0.5874918103218079,
      "learning_rate": 3.95675135419993e-05,
      "loss": 0.10650253295898438,
      "step": 955
    },
    {
      "epoch": 0.13321256880094753,
      "grad_norm": 1.711199164390564,
      "learning_rate": 3.9565561499699795e-05,
      "loss": 0.0952911376953125,
      "step": 956
    },
    {
      "epoch": 0.1333519124921619,
      "grad_norm": 2.169102191925049,
      "learning_rate": 3.9563605110379224e-05,
      "loss": 0.1320209503173828,
      "step": 957
    },
    {
      "epoch": 0.1334912561833763,
      "grad_norm": 0.47926661372184753,
      "learning_rate": 3.956164437447224e-05,
      "loss": 0.08258724212646484,
      "step": 958
    },
    {
      "epoch": 0.13363059987459067,
      "grad_norm": 0.9093363881111145,
      "learning_rate": 3.955967929241447e-05,
      "loss": 0.09659385681152344,
      "step": 959
    },
    {
      "epoch": 0.13376994356580507,
      "grad_norm": 1.6823540925979614,
      "learning_rate": 3.955770986464253e-05,
      "loss": 0.159820556640625,
      "step": 960
    },
    {
      "epoch": 0.13390928725701945,
      "grad_norm": 0.6723421216011047,
      "learning_rate": 3.955573609159395e-05,
      "loss": 0.08875465393066406,
      "step": 961
    },
    {
      "epoch": 0.13404863094823383,
      "grad_norm": 0.49799638986587524,
      "learning_rate": 3.95537579737073e-05,
      "loss": 0.11854362487792969,
      "step": 962
    },
    {
      "epoch": 0.1341879746394482,
      "grad_norm": 0.9390732049942017,
      "learning_rate": 3.955177551142202e-05,
      "loss": 0.10270500183105469,
      "step": 963
    },
    {
      "epoch": 0.13432731833066258,
      "grad_norm": 0.7228181958198547,
      "learning_rate": 3.954978870517861e-05,
      "loss": 0.10239601135253906,
      "step": 964
    },
    {
      "epoch": 0.13446666202187696,
      "grad_norm": 0.4605611264705658,
      "learning_rate": 3.954779755541848e-05,
      "loss": 0.08382606506347656,
      "step": 965
    },
    {
      "epoch": 0.13460600571309134,
      "grad_norm": 0.44877952337265015,
      "learning_rate": 3.954580206258402e-05,
      "loss": 0.11230278015136719,
      "step": 966
    },
    {
      "epoch": 0.13474534940430571,
      "grad_norm": 0.6742639541625977,
      "learning_rate": 3.9543802227118574e-05,
      "loss": 0.10134696960449219,
      "step": 967
    },
    {
      "epoch": 0.1348846930955201,
      "grad_norm": 0.8711440563201904,
      "learning_rate": 3.954179804946647e-05,
      "loss": 0.12663841247558594,
      "step": 968
    },
    {
      "epoch": 0.13502403678673447,
      "grad_norm": 0.756864607334137,
      "learning_rate": 3.953978953007299e-05,
      "loss": 0.0988168716430664,
      "step": 969
    },
    {
      "epoch": 0.13516338047794887,
      "grad_norm": 1.2521171569824219,
      "learning_rate": 3.953777666938436e-05,
      "loss": 0.10729122161865234,
      "step": 970
    },
    {
      "epoch": 0.13530272416916325,
      "grad_norm": 0.9330423474311829,
      "learning_rate": 3.953575946784782e-05,
      "loss": 0.1218109130859375,
      "step": 971
    },
    {
      "epoch": 0.13544206786037763,
      "grad_norm": 1.0559850931167603,
      "learning_rate": 3.953373792591154e-05,
      "loss": 0.12021064758300781,
      "step": 972
    },
    {
      "epoch": 0.135581411551592,
      "grad_norm": 0.5243715047836304,
      "learning_rate": 3.953171204402465e-05,
      "loss": 0.1219024658203125,
      "step": 973
    },
    {
      "epoch": 0.13572075524280638,
      "grad_norm": 0.8299070596694946,
      "learning_rate": 3.952968182263726e-05,
      "loss": 0.11920547485351562,
      "step": 974
    },
    {
      "epoch": 0.13586009893402076,
      "grad_norm": 0.9031861424446106,
      "learning_rate": 3.9527647262200444e-05,
      "loss": 0.10433006286621094,
      "step": 975
    },
    {
      "epoch": 0.13599944262523514,
      "grad_norm": 0.9984228014945984,
      "learning_rate": 3.9525608363166225e-05,
      "loss": 0.11217880249023438,
      "step": 976
    },
    {
      "epoch": 0.13613878631644952,
      "grad_norm": 1.5464277267456055,
      "learning_rate": 3.9523565125987606e-05,
      "loss": 0.1273517608642578,
      "step": 977
    },
    {
      "epoch": 0.1362781300076639,
      "grad_norm": 0.6148502230644226,
      "learning_rate": 3.952151755111855e-05,
      "loss": 0.07797527313232422,
      "step": 978
    },
    {
      "epoch": 0.13641747369887827,
      "grad_norm": 1.7016847133636475,
      "learning_rate": 3.951946563901397e-05,
      "loss": 0.1307659149169922,
      "step": 979
    },
    {
      "epoch": 0.13655681739009268,
      "grad_norm": 0.8098174333572388,
      "learning_rate": 3.951740939012977e-05,
      "loss": 0.1078033447265625,
      "step": 980
    },
    {
      "epoch": 0.13669616108130705,
      "grad_norm": 0.7261725664138794,
      "learning_rate": 3.951534880492279e-05,
      "loss": 0.10004234313964844,
      "step": 981
    },
    {
      "epoch": 0.13683550477252143,
      "grad_norm": 0.9728497862815857,
      "learning_rate": 3.951328388385085e-05,
      "loss": 0.14611434936523438,
      "step": 982
    },
    {
      "epoch": 0.1369748484637358,
      "grad_norm": 0.500342845916748,
      "learning_rate": 3.951121462737273e-05,
      "loss": 0.08334541320800781,
      "step": 983
    },
    {
      "epoch": 0.1371141921549502,
      "grad_norm": 1.0606285333633423,
      "learning_rate": 3.9509141035948156e-05,
      "loss": 0.14150047302246094,
      "step": 984
    },
    {
      "epoch": 0.13725353584616456,
      "grad_norm": 1.2684240341186523,
      "learning_rate": 3.950706311003785e-05,
      "loss": 0.09601974487304688,
      "step": 985
    },
    {
      "epoch": 0.13739287953737894,
      "grad_norm": 1.4431241750717163,
      "learning_rate": 3.950498085010348e-05,
      "loss": 0.10696983337402344,
      "step": 986
    },
    {
      "epoch": 0.13753222322859332,
      "grad_norm": 1.647621512413025,
      "learning_rate": 3.950289425660767e-05,
      "loss": 0.1531829833984375,
      "step": 987
    },
    {
      "epoch": 0.1376715669198077,
      "grad_norm": 0.8080485463142395,
      "learning_rate": 3.950080333001402e-05,
      "loss": 0.11075592041015625,
      "step": 988
    },
    {
      "epoch": 0.13781091061102207,
      "grad_norm": 0.9858983755111694,
      "learning_rate": 3.9498708070787076e-05,
      "loss": 0.11331939697265625,
      "step": 989
    },
    {
      "epoch": 0.13795025430223645,
      "grad_norm": 0.6059591174125671,
      "learning_rate": 3.949660847939236e-05,
      "loss": 0.10280036926269531,
      "step": 990
    },
    {
      "epoch": 0.13808959799345086,
      "grad_norm": 0.8609289526939392,
      "learning_rate": 3.949450455629635e-05,
      "loss": 0.131927490234375,
      "step": 991
    },
    {
      "epoch": 0.13822894168466524,
      "grad_norm": 1.354343295097351,
      "learning_rate": 3.9492396301966504e-05,
      "loss": 0.09990119934082031,
      "step": 992
    },
    {
      "epoch": 0.1383682853758796,
      "grad_norm": 0.821421205997467,
      "learning_rate": 3.9490283716871214e-05,
      "loss": 0.08404922485351562,
      "step": 993
    },
    {
      "epoch": 0.138507629067094,
      "grad_norm": 0.9992891550064087,
      "learning_rate": 3.948816680147986e-05,
      "loss": 0.12139129638671875,
      "step": 994
    },
    {
      "epoch": 0.13864697275830837,
      "grad_norm": 0.6031719446182251,
      "learning_rate": 3.9486045556262756e-05,
      "loss": 0.11076164245605469,
      "step": 995
    },
    {
      "epoch": 0.13878631644952275,
      "grad_norm": 1.0184024572372437,
      "learning_rate": 3.948391998169121e-05,
      "loss": 0.12169456481933594,
      "step": 996
    },
    {
      "epoch": 0.13892566014073712,
      "grad_norm": 0.4988100230693817,
      "learning_rate": 3.948179007823746e-05,
      "loss": 0.08377456665039062,
      "step": 997
    },
    {
      "epoch": 0.1390650038319515,
      "grad_norm": 0.44008979201316833,
      "learning_rate": 3.947965584637474e-05,
      "loss": 0.07970237731933594,
      "step": 998
    },
    {
      "epoch": 0.13920434752316588,
      "grad_norm": 0.5558716654777527,
      "learning_rate": 3.947751728657722e-05,
      "loss": 0.110076904296875,
      "step": 999
    },
    {
      "epoch": 0.13934369121438026,
      "grad_norm": 0.829971194267273,
      "learning_rate": 3.9475374399320036e-05,
      "loss": 0.14620590209960938,
      "step": 1000
    },
    {
      "epoch": 0.13948303490559466,
      "grad_norm": 0.3911738395690918,
      "learning_rate": 3.947322718507929e-05,
      "loss": 0.10377883911132812,
      "step": 1001
    },
    {
      "epoch": 0.13962237859680904,
      "grad_norm": 0.6939563751220703,
      "learning_rate": 3.947107564433204e-05,
      "loss": 0.09389305114746094,
      "step": 1002
    },
    {
      "epoch": 0.13976172228802342,
      "grad_norm": 2.19252872467041,
      "learning_rate": 3.946891977755632e-05,
      "loss": 0.1535186767578125,
      "step": 1003
    },
    {
      "epoch": 0.1399010659792378,
      "grad_norm": 0.3593987822532654,
      "learning_rate": 3.946675958523111e-05,
      "loss": 0.09176063537597656,
      "step": 1004
    },
    {
      "epoch": 0.14004040967045217,
      "grad_norm": 0.787429690361023,
      "learning_rate": 3.946459506783635e-05,
      "loss": 0.08251571655273438,
      "step": 1005
    },
    {
      "epoch": 0.14017975336166655,
      "grad_norm": 0.5198335647583008,
      "learning_rate": 3.9462426225852954e-05,
      "loss": 0.08567237854003906,
      "step": 1006
    },
    {
      "epoch": 0.14031909705288093,
      "grad_norm": 0.5118117928504944,
      "learning_rate": 3.946025305976278e-05,
      "loss": 0.07783889770507812,
      "step": 1007
    },
    {
      "epoch": 0.1404584407440953,
      "grad_norm": 0.898271381855011,
      "learning_rate": 3.9458075570048666e-05,
      "loss": 0.09580421447753906,
      "step": 1008
    },
    {
      "epoch": 0.14059778443530968,
      "grad_norm": 1.2402944564819336,
      "learning_rate": 3.945589375719439e-05,
      "loss": 0.1432056427001953,
      "step": 1009
    },
    {
      "epoch": 0.14073712812652406,
      "grad_norm": 0.44301941990852356,
      "learning_rate": 3.9453707621684714e-05,
      "loss": 0.09371280670166016,
      "step": 1010
    },
    {
      "epoch": 0.14087647181773846,
      "grad_norm": 0.6207379698753357,
      "learning_rate": 3.945151716400534e-05,
      "loss": 0.07216835021972656,
      "step": 1011
    },
    {
      "epoch": 0.14101581550895284,
      "grad_norm": 1.2328846454620361,
      "learning_rate": 3.944932238464293e-05,
      "loss": 0.17153167724609375,
      "step": 1012
    },
    {
      "epoch": 0.14115515920016722,
      "grad_norm": 0.9434775710105896,
      "learning_rate": 3.944712328408513e-05,
      "loss": 0.09087181091308594,
      "step": 1013
    },
    {
      "epoch": 0.1412945028913816,
      "grad_norm": 0.9901787638664246,
      "learning_rate": 3.9444919862820514e-05,
      "loss": 0.08065032958984375,
      "step": 1014
    },
    {
      "epoch": 0.14143384658259597,
      "grad_norm": 1.2479723691940308,
      "learning_rate": 3.944271212133864e-05,
      "loss": 0.1379070281982422,
      "step": 1015
    },
    {
      "epoch": 0.14157319027381035,
      "grad_norm": 0.5679019093513489,
      "learning_rate": 3.9440500060130025e-05,
      "loss": 0.08939170837402344,
      "step": 1016
    },
    {
      "epoch": 0.14171253396502473,
      "grad_norm": 0.9982486963272095,
      "learning_rate": 3.943828367968613e-05,
      "loss": 0.10690498352050781,
      "step": 1017
    },
    {
      "epoch": 0.1418518776562391,
      "grad_norm": 0.4452458322048187,
      "learning_rate": 3.9436062980499376e-05,
      "loss": 0.079620361328125,
      "step": 1018
    },
    {
      "epoch": 0.14199122134745348,
      "grad_norm": 0.6532194018363953,
      "learning_rate": 3.943383796306317e-05,
      "loss": 0.10463333129882812,
      "step": 1019
    },
    {
      "epoch": 0.14213056503866786,
      "grad_norm": 0.9193634986877441,
      "learning_rate": 3.9431608627871845e-05,
      "loss": 0.10005378723144531,
      "step": 1020
    },
    {
      "epoch": 0.14226990872988227,
      "grad_norm": 0.35800793766975403,
      "learning_rate": 3.9429374975420714e-05,
      "loss": 0.08039093017578125,
      "step": 1021
    },
    {
      "epoch": 0.14240925242109664,
      "grad_norm": 0.5265731811523438,
      "learning_rate": 3.942713700620605e-05,
      "loss": 0.10545730590820312,
      "step": 1022
    },
    {
      "epoch": 0.14254859611231102,
      "grad_norm": 0.4224061369895935,
      "learning_rate": 3.942489472072507e-05,
      "loss": 0.09466171264648438,
      "step": 1023
    },
    {
      "epoch": 0.1426879398035254,
      "grad_norm": 0.9892920255661011,
      "learning_rate": 3.942264811947596e-05,
      "loss": 0.16904067993164062,
      "step": 1024
    },
    {
      "epoch": 0.14282728349473978,
      "grad_norm": 0.5567622184753418,
      "learning_rate": 3.9420397202957854e-05,
      "loss": 0.11565399169921875,
      "step": 1025
    },
    {
      "epoch": 0.14296662718595415,
      "grad_norm": 0.6593400835990906,
      "learning_rate": 3.941814197167087e-05,
      "loss": 0.09443092346191406,
      "step": 1026
    },
    {
      "epoch": 0.14310597087716853,
      "grad_norm": 0.9876372218132019,
      "learning_rate": 3.941588242611607e-05,
      "loss": 0.06507492065429688,
      "step": 1027
    },
    {
      "epoch": 0.1432453145683829,
      "grad_norm": 1.0169603824615479,
      "learning_rate": 3.9413618566795465e-05,
      "loss": 0.103790283203125,
      "step": 1028
    },
    {
      "epoch": 0.1433846582595973,
      "grad_norm": 0.7091437578201294,
      "learning_rate": 3.941135039421204e-05,
      "loss": 0.11845016479492188,
      "step": 1029
    },
    {
      "epoch": 0.14352400195081166,
      "grad_norm": 0.9706575274467468,
      "learning_rate": 3.940907790886971e-05,
      "loss": 0.10301971435546875,
      "step": 1030
    },
    {
      "epoch": 0.14366334564202607,
      "grad_norm": 1.2931971549987793,
      "learning_rate": 3.94068011112734e-05,
      "loss": 0.132659912109375,
      "step": 1031
    },
    {
      "epoch": 0.14380268933324045,
      "grad_norm": 1.0244883298873901,
      "learning_rate": 3.9404520001928945e-05,
      "loss": 0.09354782104492188,
      "step": 1032
    },
    {
      "epoch": 0.14394203302445482,
      "grad_norm": 1.0689970254898071,
      "learning_rate": 3.940223458134316e-05,
      "loss": 0.104095458984375,
      "step": 1033
    },
    {
      "epoch": 0.1440813767156692,
      "grad_norm": 0.8139352798461914,
      "learning_rate": 3.939994485002381e-05,
      "loss": 0.11745643615722656,
      "step": 1034
    },
    {
      "epoch": 0.14422072040688358,
      "grad_norm": 0.3693269193172455,
      "learning_rate": 3.939765080847962e-05,
      "loss": 0.0854644775390625,
      "step": 1035
    },
    {
      "epoch": 0.14436006409809796,
      "grad_norm": 1.1354705095291138,
      "learning_rate": 3.9395352457220275e-05,
      "loss": 0.10104179382324219,
      "step": 1036
    },
    {
      "epoch": 0.14449940778931233,
      "grad_norm": 0.8560324907302856,
      "learning_rate": 3.939304979675642e-05,
      "loss": 0.12490081787109375,
      "step": 1037
    },
    {
      "epoch": 0.1446387514805267,
      "grad_norm": 1.307073950767517,
      "learning_rate": 3.939074282759965e-05,
      "loss": 0.12902259826660156,
      "step": 1038
    },
    {
      "epoch": 0.1447780951717411,
      "grad_norm": 1.1590675115585327,
      "learning_rate": 3.938843155026252e-05,
      "loss": 0.09788894653320312,
      "step": 1039
    },
    {
      "epoch": 0.14491743886295547,
      "grad_norm": 0.3905137777328491,
      "learning_rate": 3.938611596525855e-05,
      "loss": 0.08150291442871094,
      "step": 1040
    },
    {
      "epoch": 0.14505678255416987,
      "grad_norm": 1.2704097032546997,
      "learning_rate": 3.9383796073102206e-05,
      "loss": 0.11468887329101562,
      "step": 1041
    },
    {
      "epoch": 0.14519612624538425,
      "grad_norm": 1.2073370218276978,
      "learning_rate": 3.9381471874308916e-05,
      "loss": 0.11696434020996094,
      "step": 1042
    },
    {
      "epoch": 0.14533546993659863,
      "grad_norm": 0.9552823305130005,
      "learning_rate": 3.9379143369395054e-05,
      "loss": 0.10514163970947266,
      "step": 1043
    },
    {
      "epoch": 0.145474813627813,
      "grad_norm": 0.6865363121032715,
      "learning_rate": 3.937681055887797e-05,
      "loss": 0.1565380096435547,
      "step": 1044
    },
    {
      "epoch": 0.14561415731902738,
      "grad_norm": 0.7349458932876587,
      "learning_rate": 3.937447344327596e-05,
      "loss": 0.11250877380371094,
      "step": 1045
    },
    {
      "epoch": 0.14575350101024176,
      "grad_norm": 1.6060528755187988,
      "learning_rate": 3.937213202310828e-05,
      "loss": 0.12238121032714844,
      "step": 1046
    },
    {
      "epoch": 0.14589284470145614,
      "grad_norm": 0.7500627636909485,
      "learning_rate": 3.9369786298895144e-05,
      "loss": 0.08775520324707031,
      "step": 1047
    },
    {
      "epoch": 0.14603218839267051,
      "grad_norm": 0.4673171937465668,
      "learning_rate": 3.93674362711577e-05,
      "loss": 0.11436271667480469,
      "step": 1048
    },
    {
      "epoch": 0.1461715320838849,
      "grad_norm": 1.5575499534606934,
      "learning_rate": 3.936508194041809e-05,
      "loss": 0.12291145324707031,
      "step": 1049
    },
    {
      "epoch": 0.14631087577509927,
      "grad_norm": 0.48581135272979736,
      "learning_rate": 3.936272330719938e-05,
      "loss": 0.07053375244140625,
      "step": 1050
    },
    {
      "epoch": 0.14645021946631367,
      "grad_norm": 0.7019978761672974,
      "learning_rate": 3.936036037202561e-05,
      "loss": 0.10374259948730469,
      "step": 1051
    },
    {
      "epoch": 0.14658956315752805,
      "grad_norm": 0.9891214370727539,
      "learning_rate": 3.935799313542178e-05,
      "loss": 0.11409378051757812,
      "step": 1052
    },
    {
      "epoch": 0.14672890684874243,
      "grad_norm": 1.101868987083435,
      "learning_rate": 3.935562159791381e-05,
      "loss": 0.09305763244628906,
      "step": 1053
    },
    {
      "epoch": 0.1468682505399568,
      "grad_norm": 0.5298671722412109,
      "learning_rate": 3.9353245760028634e-05,
      "loss": 0.08376121520996094,
      "step": 1054
    },
    {
      "epoch": 0.14700759423117118,
      "grad_norm": 0.5000380873680115,
      "learning_rate": 3.935086562229408e-05,
      "loss": 0.09716224670410156,
      "step": 1055
    },
    {
      "epoch": 0.14714693792238556,
      "grad_norm": 1.1321380138397217,
      "learning_rate": 3.9348481185238976e-05,
      "loss": 0.21497344970703125,
      "step": 1056
    },
    {
      "epoch": 0.14728628161359994,
      "grad_norm": 0.9351024627685547,
      "learning_rate": 3.9346092449393084e-05,
      "loss": 0.11600494384765625,
      "step": 1057
    },
    {
      "epoch": 0.14742562530481432,
      "grad_norm": 1.4278959035873413,
      "learning_rate": 3.934369941528713e-05,
      "loss": 0.10428047180175781,
      "step": 1058
    },
    {
      "epoch": 0.1475649689960287,
      "grad_norm": 1.376546859741211,
      "learning_rate": 3.93413020834528e-05,
      "loss": 0.11654090881347656,
      "step": 1059
    },
    {
      "epoch": 0.14770431268724307,
      "grad_norm": 0.989070475101471,
      "learning_rate": 3.9338900454422704e-05,
      "loss": 0.11585235595703125,
      "step": 1060
    },
    {
      "epoch": 0.14784365637845748,
      "grad_norm": 0.3304140865802765,
      "learning_rate": 3.933649452873044e-05,
      "loss": 0.07465457916259766,
      "step": 1061
    },
    {
      "epoch": 0.14798300006967186,
      "grad_norm": 0.8556541800498962,
      "learning_rate": 3.933408430691055e-05,
      "loss": 0.09260177612304688,
      "step": 1062
    },
    {
      "epoch": 0.14812234376088623,
      "grad_norm": 0.8454442620277405,
      "learning_rate": 3.933166978949855e-05,
      "loss": 0.09960174560546875,
      "step": 1063
    },
    {
      "epoch": 0.1482616874521006,
      "grad_norm": 0.848351776599884,
      "learning_rate": 3.932925097703086e-05,
      "loss": 0.09482097625732422,
      "step": 1064
    },
    {
      "epoch": 0.148401031143315,
      "grad_norm": 1.447342038154602,
      "learning_rate": 3.932682787004489e-05,
      "loss": 0.12227439880371094,
      "step": 1065
    },
    {
      "epoch": 0.14854037483452937,
      "grad_norm": 0.9167925715446472,
      "learning_rate": 3.932440046907902e-05,
      "loss": 0.12813568115234375,
      "step": 1066
    },
    {
      "epoch": 0.14867971852574374,
      "grad_norm": 0.6223666071891785,
      "learning_rate": 3.932196877467254e-05,
      "loss": 0.08332061767578125,
      "step": 1067
    },
    {
      "epoch": 0.14881906221695812,
      "grad_norm": 1.1545028686523438,
      "learning_rate": 3.9319532787365733e-05,
      "loss": 0.11682510375976562,
      "step": 1068
    },
    {
      "epoch": 0.1489584059081725,
      "grad_norm": 0.7685822248458862,
      "learning_rate": 3.931709250769981e-05,
      "loss": 0.0981903076171875,
      "step": 1069
    },
    {
      "epoch": 0.14909774959938688,
      "grad_norm": 0.5964155197143555,
      "learning_rate": 3.931464793621695e-05,
      "loss": 0.10311508178710938,
      "step": 1070
    },
    {
      "epoch": 0.14923709329060128,
      "grad_norm": 0.506064772605896,
      "learning_rate": 3.931219907346028e-05,
      "loss": 0.08450508117675781,
      "step": 1071
    },
    {
      "epoch": 0.14937643698181566,
      "grad_norm": 0.288532555103302,
      "learning_rate": 3.930974591997387e-05,
      "loss": 0.07798957824707031,
      "step": 1072
    },
    {
      "epoch": 0.14951578067303004,
      "grad_norm": 0.3222642242908478,
      "learning_rate": 3.930728847630278e-05,
      "loss": 0.07794952392578125,
      "step": 1073
    },
    {
      "epoch": 0.1496551243642444,
      "grad_norm": 0.8542212843894958,
      "learning_rate": 3.930482674299297e-05,
      "loss": 0.11896514892578125,
      "step": 1074
    },
    {
      "epoch": 0.1497944680554588,
      "grad_norm": 0.4300534725189209,
      "learning_rate": 3.930236072059141e-05,
      "loss": 0.07259368896484375,
      "step": 1075
    },
    {
      "epoch": 0.14993381174667317,
      "grad_norm": 0.7836169600486755,
      "learning_rate": 3.929989040964596e-05,
      "loss": 0.09284210205078125,
      "step": 1076
    },
    {
      "epoch": 0.15007315543788755,
      "grad_norm": 0.5698338150978088,
      "learning_rate": 3.92974158107055e-05,
      "loss": 0.11429595947265625,
      "step": 1077
    },
    {
      "epoch": 0.15021249912910192,
      "grad_norm": 0.4571463167667389,
      "learning_rate": 3.929493692431981e-05,
      "loss": 0.08469867706298828,
      "step": 1078
    },
    {
      "epoch": 0.1503518428203163,
      "grad_norm": 0.9946839809417725,
      "learning_rate": 3.929245375103965e-05,
      "loss": 0.12943077087402344,
      "step": 1079
    },
    {
      "epoch": 0.15049118651153068,
      "grad_norm": 0.6127795577049255,
      "learning_rate": 3.928996629141671e-05,
      "loss": 0.1240234375,
      "step": 1080
    },
    {
      "epoch": 0.15063053020274508,
      "grad_norm": 0.6540313363075256,
      "learning_rate": 3.928747454600367e-05,
      "loss": 0.11668777465820312,
      "step": 1081
    },
    {
      "epoch": 0.15076987389395946,
      "grad_norm": 0.44004571437835693,
      "learning_rate": 3.928497851535412e-05,
      "loss": 0.08819961547851562,
      "step": 1082
    },
    {
      "epoch": 0.15090921758517384,
      "grad_norm": 0.7766216993331909,
      "learning_rate": 3.9282478200022624e-05,
      "loss": 0.1381053924560547,
      "step": 1083
    },
    {
      "epoch": 0.15104856127638822,
      "grad_norm": 0.49621301889419556,
      "learning_rate": 3.9279973600564706e-05,
      "loss": 0.0928659439086914,
      "step": 1084
    },
    {
      "epoch": 0.1511879049676026,
      "grad_norm": 0.658751904964447,
      "learning_rate": 3.9277464717536815e-05,
      "loss": 0.10831832885742188,
      "step": 1085
    },
    {
      "epoch": 0.15132724865881697,
      "grad_norm": 0.6189517974853516,
      "learning_rate": 3.927495155149639e-05,
      "loss": 0.09692955017089844,
      "step": 1086
    },
    {
      "epoch": 0.15146659235003135,
      "grad_norm": 0.7042515277862549,
      "learning_rate": 3.927243410300177e-05,
      "loss": 0.09633827209472656,
      "step": 1087
    },
    {
      "epoch": 0.15160593604124573,
      "grad_norm": 0.7039734125137329,
      "learning_rate": 3.9269912372612295e-05,
      "loss": 0.0843353271484375,
      "step": 1088
    },
    {
      "epoch": 0.1517452797324601,
      "grad_norm": 1.2230112552642822,
      "learning_rate": 3.926738636088823e-05,
      "loss": 0.11907577514648438,
      "step": 1089
    },
    {
      "epoch": 0.15188462342367448,
      "grad_norm": 0.6470591425895691,
      "learning_rate": 3.92648560683908e-05,
      "loss": 0.08072471618652344,
      "step": 1090
    },
    {
      "epoch": 0.15202396711488889,
      "grad_norm": 0.7570467591285706,
      "learning_rate": 3.926232149568217e-05,
      "loss": 0.09984970092773438,
      "step": 1091
    },
    {
      "epoch": 0.15216331080610326,
      "grad_norm": 0.9719909429550171,
      "learning_rate": 3.925978264332548e-05,
      "loss": 0.09764480590820312,
      "step": 1092
    },
    {
      "epoch": 0.15230265449731764,
      "grad_norm": 1.181147575378418,
      "learning_rate": 3.925723951188478e-05,
      "loss": 0.09545707702636719,
      "step": 1093
    },
    {
      "epoch": 0.15244199818853202,
      "grad_norm": 0.8334307074546814,
      "learning_rate": 3.925469210192512e-05,
      "loss": 0.0977621078491211,
      "step": 1094
    },
    {
      "epoch": 0.1525813418797464,
      "grad_norm": 0.49978500604629517,
      "learning_rate": 3.9252140414012465e-05,
      "loss": 0.09285736083984375,
      "step": 1095
    },
    {
      "epoch": 0.15272068557096077,
      "grad_norm": 1.2212903499603271,
      "learning_rate": 3.9249584448713746e-05,
      "loss": 0.11314582824707031,
      "step": 1096
    },
    {
      "epoch": 0.15286002926217515,
      "grad_norm": 1.1671745777130127,
      "learning_rate": 3.9247024206596836e-05,
      "loss": 0.125457763671875,
      "step": 1097
    },
    {
      "epoch": 0.15299937295338953,
      "grad_norm": 1.296875238418579,
      "learning_rate": 3.924445968823057e-05,
      "loss": 0.11758708953857422,
      "step": 1098
    },
    {
      "epoch": 0.1531387166446039,
      "grad_norm": 0.8075811862945557,
      "learning_rate": 3.924189089418471e-05,
      "loss": 0.10266685485839844,
      "step": 1099
    },
    {
      "epoch": 0.15327806033581828,
      "grad_norm": 0.6042525768280029,
      "learning_rate": 3.923931782503e-05,
      "loss": 0.09482955932617188,
      "step": 1100
    },
    {
      "epoch": 0.1534174040270327,
      "grad_norm": 0.48374706506729126,
      "learning_rate": 3.923674048133811e-05,
      "loss": 0.11186790466308594,
      "step": 1101
    },
    {
      "epoch": 0.15355674771824707,
      "grad_norm": 1.056330680847168,
      "learning_rate": 3.923415886368166e-05,
      "loss": 0.1139373779296875,
      "step": 1102
    },
    {
      "epoch": 0.15369609140946144,
      "grad_norm": 0.6797927618026733,
      "learning_rate": 3.923157297263425e-05,
      "loss": 0.09133529663085938,
      "step": 1103
    },
    {
      "epoch": 0.15383543510067582,
      "grad_norm": 0.6420765519142151,
      "learning_rate": 3.922898280877037e-05,
      "loss": 0.10420417785644531,
      "step": 1104
    },
    {
      "epoch": 0.1539747787918902,
      "grad_norm": 0.38202783465385437,
      "learning_rate": 3.922638837266552e-05,
      "loss": 0.09093666076660156,
      "step": 1105
    },
    {
      "epoch": 0.15411412248310458,
      "grad_norm": 0.39757218956947327,
      "learning_rate": 3.9223789664896136e-05,
      "loss": 0.09690666198730469,
      "step": 1106
    },
    {
      "epoch": 0.15425346617431895,
      "grad_norm": 0.8744697570800781,
      "learning_rate": 3.922118668603956e-05,
      "loss": 0.12566852569580078,
      "step": 1107
    },
    {
      "epoch": 0.15439280986553333,
      "grad_norm": 0.34353718161582947,
      "learning_rate": 3.9218579436674134e-05,
      "loss": 0.07489585876464844,
      "step": 1108
    },
    {
      "epoch": 0.1545321535567477,
      "grad_norm": 0.5745982527732849,
      "learning_rate": 3.921596791737912e-05,
      "loss": 0.11295318603515625,
      "step": 1109
    },
    {
      "epoch": 0.1546714972479621,
      "grad_norm": 0.5885826349258423,
      "learning_rate": 3.9213352128734746e-05,
      "loss": 0.08493804931640625,
      "step": 1110
    },
    {
      "epoch": 0.1548108409391765,
      "grad_norm": 1.3089457750320435,
      "learning_rate": 3.9210732071322175e-05,
      "loss": 0.10105705261230469,
      "step": 1111
    },
    {
      "epoch": 0.15495018463039087,
      "grad_norm": 1.3515362739562988,
      "learning_rate": 3.920810774572353e-05,
      "loss": 0.10039710998535156,
      "step": 1112
    },
    {
      "epoch": 0.15508952832160525,
      "grad_norm": 0.6509199142456055,
      "learning_rate": 3.9205479152521874e-05,
      "loss": 0.10841178894042969,
      "step": 1113
    },
    {
      "epoch": 0.15522887201281962,
      "grad_norm": 0.9489114880561829,
      "learning_rate": 3.920284629230121e-05,
      "loss": 0.10499000549316406,
      "step": 1114
    },
    {
      "epoch": 0.155368215704034,
      "grad_norm": 0.9219138622283936,
      "learning_rate": 3.920020916564652e-05,
      "loss": 0.1313457489013672,
      "step": 1115
    },
    {
      "epoch": 0.15550755939524838,
      "grad_norm": 0.8144559264183044,
      "learning_rate": 3.919756777314369e-05,
      "loss": 0.10344886779785156,
      "step": 1116
    },
    {
      "epoch": 0.15564690308646276,
      "grad_norm": 1.4209098815917969,
      "learning_rate": 3.9194922115379596e-05,
      "loss": 0.12770843505859375,
      "step": 1117
    },
    {
      "epoch": 0.15578624677767713,
      "grad_norm": 0.4071964919567108,
      "learning_rate": 3.919227219294204e-05,
      "loss": 0.10225105285644531,
      "step": 1118
    },
    {
      "epoch": 0.1559255904688915,
      "grad_norm": 0.663932204246521,
      "learning_rate": 3.918961800641976e-05,
      "loss": 0.11481094360351562,
      "step": 1119
    },
    {
      "epoch": 0.1560649341601059,
      "grad_norm": 0.38957080245018005,
      "learning_rate": 3.918695955640247e-05,
      "loss": 0.0826263427734375,
      "step": 1120
    },
    {
      "epoch": 0.1562042778513203,
      "grad_norm": 1.3244646787643433,
      "learning_rate": 3.9184296843480816e-05,
      "loss": 0.15844154357910156,
      "step": 1121
    },
    {
      "epoch": 0.15634362154253467,
      "grad_norm": 0.7474139332771301,
      "learning_rate": 3.918162986824638e-05,
      "loss": 0.12818527221679688,
      "step": 1122
    },
    {
      "epoch": 0.15648296523374905,
      "grad_norm": 1.1751627922058105,
      "learning_rate": 3.9178958631291715e-05,
      "loss": 0.1099395751953125,
      "step": 1123
    },
    {
      "epoch": 0.15662230892496343,
      "grad_norm": 0.6349981427192688,
      "learning_rate": 3.91762831332103e-05,
      "loss": 0.10024833679199219,
      "step": 1124
    },
    {
      "epoch": 0.1567616526161778,
      "grad_norm": 0.6428050398826599,
      "learning_rate": 3.917360337459658e-05,
      "loss": 0.08183479309082031,
      "step": 1125
    },
    {
      "epoch": 0.15690099630739218,
      "grad_norm": 0.7403628826141357,
      "learning_rate": 3.9170919356045935e-05,
      "loss": 0.12365913391113281,
      "step": 1126
    },
    {
      "epoch": 0.15704033999860656,
      "grad_norm": 0.5638999938964844,
      "learning_rate": 3.916823107815469e-05,
      "loss": 0.14251708984375,
      "step": 1127
    },
    {
      "epoch": 0.15717968368982094,
      "grad_norm": 0.8675004839897156,
      "learning_rate": 3.916553854152011e-05,
      "loss": 0.09070968627929688,
      "step": 1128
    },
    {
      "epoch": 0.15731902738103531,
      "grad_norm": 0.9647676348686218,
      "learning_rate": 3.916284174674042e-05,
      "loss": 0.10032272338867188,
      "step": 1129
    },
    {
      "epoch": 0.1574583710722497,
      "grad_norm": 0.8367294073104858,
      "learning_rate": 3.9160140694414796e-05,
      "loss": 0.10583114624023438,
      "step": 1130
    },
    {
      "epoch": 0.1575977147634641,
      "grad_norm": 0.7427570819854736,
      "learning_rate": 3.915743538514334e-05,
      "loss": 0.10660457611083984,
      "step": 1131
    },
    {
      "epoch": 0.15773705845467847,
      "grad_norm": 1.2409693002700806,
      "learning_rate": 3.915472581952711e-05,
      "loss": 0.15674591064453125,
      "step": 1132
    },
    {
      "epoch": 0.15787640214589285,
      "grad_norm": 1.4524359703063965,
      "learning_rate": 3.915201199816812e-05,
      "loss": 0.1296405792236328,
      "step": 1133
    },
    {
      "epoch": 0.15801574583710723,
      "grad_norm": 0.6520192623138428,
      "learning_rate": 3.914929392166931e-05,
      "loss": 0.07789421081542969,
      "step": 1134
    },
    {
      "epoch": 0.1581550895283216,
      "grad_norm": 0.8502861261367798,
      "learning_rate": 3.914657159063458e-05,
      "loss": 0.09886455535888672,
      "step": 1135
    },
    {
      "epoch": 0.15829443321953598,
      "grad_norm": 0.5840429067611694,
      "learning_rate": 3.914384500566876e-05,
      "loss": 0.10323143005371094,
      "step": 1136
    },
    {
      "epoch": 0.15843377691075036,
      "grad_norm": 1.6788294315338135,
      "learning_rate": 3.9141114167377636e-05,
      "loss": 0.13954925537109375,
      "step": 1137
    },
    {
      "epoch": 0.15857312060196474,
      "grad_norm": 0.4864406883716583,
      "learning_rate": 3.9138379076367956e-05,
      "loss": 0.08394241333007812,
      "step": 1138
    },
    {
      "epoch": 0.15871246429317912,
      "grad_norm": 0.914016842842102,
      "learning_rate": 3.913563973324738e-05,
      "loss": 0.10073471069335938,
      "step": 1139
    },
    {
      "epoch": 0.1588518079843935,
      "grad_norm": 1.6128093004226685,
      "learning_rate": 3.913289613862452e-05,
      "loss": 0.12928390502929688,
      "step": 1140
    },
    {
      "epoch": 0.1589911516756079,
      "grad_norm": 0.9807238578796387,
      "learning_rate": 3.913014829310895e-05,
      "loss": 0.12134933471679688,
      "step": 1141
    },
    {
      "epoch": 0.15913049536682228,
      "grad_norm": 0.9340755939483643,
      "learning_rate": 3.9127396197311185e-05,
      "loss": 0.10314369201660156,
      "step": 1142
    },
    {
      "epoch": 0.15926983905803666,
      "grad_norm": 0.6124973297119141,
      "learning_rate": 3.9124639851842666e-05,
      "loss": 0.10646629333496094,
      "step": 1143
    },
    {
      "epoch": 0.15940918274925103,
      "grad_norm": 1.4228442907333374,
      "learning_rate": 3.91218792573158e-05,
      "loss": 0.12665557861328125,
      "step": 1144
    },
    {
      "epoch": 0.1595485264404654,
      "grad_norm": 1.0365920066833496,
      "learning_rate": 3.911911441434392e-05,
      "loss": 0.13811492919921875,
      "step": 1145
    },
    {
      "epoch": 0.1596878701316798,
      "grad_norm": 0.6849114298820496,
      "learning_rate": 3.911634532354131e-05,
      "loss": 0.11650276184082031,
      "step": 1146
    },
    {
      "epoch": 0.15982721382289417,
      "grad_norm": 1.1287925243377686,
      "learning_rate": 3.911357198552321e-05,
      "loss": 0.10275745391845703,
      "step": 1147
    },
    {
      "epoch": 0.15996655751410854,
      "grad_norm": 0.7633118033409119,
      "learning_rate": 3.9110794400905785e-05,
      "loss": 0.09051322937011719,
      "step": 1148
    },
    {
      "epoch": 0.16010590120532292,
      "grad_norm": 0.43828871846199036,
      "learning_rate": 3.9108012570306143e-05,
      "loss": 0.10183525085449219,
      "step": 1149
    },
    {
      "epoch": 0.1602452448965373,
      "grad_norm": 0.5595589280128479,
      "learning_rate": 3.910522649434236e-05,
      "loss": 0.12334251403808594,
      "step": 1150
    },
    {
      "epoch": 0.1603845885877517,
      "grad_norm": 0.7032245993614197,
      "learning_rate": 3.9102436173633425e-05,
      "loss": 0.10942268371582031,
      "step": 1151
    },
    {
      "epoch": 0.16052393227896608,
      "grad_norm": 0.4530486464500427,
      "learning_rate": 3.9099641608799286e-05,
      "loss": 0.07540130615234375,
      "step": 1152
    },
    {
      "epoch": 0.16066327597018046,
      "grad_norm": 0.7467103004455566,
      "learning_rate": 3.9096842800460836e-05,
      "loss": 0.10842704772949219,
      "step": 1153
    },
    {
      "epoch": 0.16080261966139484,
      "grad_norm": 0.7465243935585022,
      "learning_rate": 3.909403974923991e-05,
      "loss": 0.12427902221679688,
      "step": 1154
    },
    {
      "epoch": 0.1609419633526092,
      "grad_norm": 0.8461906909942627,
      "learning_rate": 3.9091232455759274e-05,
      "loss": 0.13983154296875,
      "step": 1155
    },
    {
      "epoch": 0.1610813070438236,
      "grad_norm": 0.33672839403152466,
      "learning_rate": 3.908842092064264e-05,
      "loss": 0.08561325073242188,
      "step": 1156
    },
    {
      "epoch": 0.16122065073503797,
      "grad_norm": 0.6950038075447083,
      "learning_rate": 3.9085605144514674e-05,
      "loss": 0.09231948852539062,
      "step": 1157
    },
    {
      "epoch": 0.16135999442625235,
      "grad_norm": 1.0353983640670776,
      "learning_rate": 3.908278512800098e-05,
      "loss": 0.1248016357421875,
      "step": 1158
    },
    {
      "epoch": 0.16149933811746672,
      "grad_norm": 0.8223158121109009,
      "learning_rate": 3.9079960871728094e-05,
      "loss": 0.09426498413085938,
      "step": 1159
    },
    {
      "epoch": 0.1616386818086811,
      "grad_norm": 0.9021525382995605,
      "learning_rate": 3.907713237632351e-05,
      "loss": 0.13553619384765625,
      "step": 1160
    },
    {
      "epoch": 0.1617780254998955,
      "grad_norm": 0.7501322031021118,
      "learning_rate": 3.907429964241565e-05,
      "loss": 0.1172332763671875,
      "step": 1161
    },
    {
      "epoch": 0.16191736919110988,
      "grad_norm": 0.27710118889808655,
      "learning_rate": 3.907146267063389e-05,
      "loss": 0.06795883178710938,
      "step": 1162
    },
    {
      "epoch": 0.16205671288232426,
      "grad_norm": 0.34633028507232666,
      "learning_rate": 3.906862146160852e-05,
      "loss": 0.09498214721679688,
      "step": 1163
    },
    {
      "epoch": 0.16219605657353864,
      "grad_norm": 0.6766300201416016,
      "learning_rate": 3.9065776015970815e-05,
      "loss": 0.09692192077636719,
      "step": 1164
    },
    {
      "epoch": 0.16233540026475302,
      "grad_norm": 1.2373223304748535,
      "learning_rate": 3.906292633435295e-05,
      "loss": 0.11261558532714844,
      "step": 1165
    },
    {
      "epoch": 0.1624747439559674,
      "grad_norm": 1.3194924592971802,
      "learning_rate": 3.906007241738807e-05,
      "loss": 0.1446857452392578,
      "step": 1166
    },
    {
      "epoch": 0.16261408764718177,
      "grad_norm": 1.782264232635498,
      "learning_rate": 3.9057214265710245e-05,
      "loss": 0.1426067352294922,
      "step": 1167
    },
    {
      "epoch": 0.16275343133839615,
      "grad_norm": 0.9703130722045898,
      "learning_rate": 3.9054351879954505e-05,
      "loss": 0.08833503723144531,
      "step": 1168
    },
    {
      "epoch": 0.16289277502961053,
      "grad_norm": 2.350311756134033,
      "learning_rate": 3.905148526075679e-05,
      "loss": 0.15925025939941406,
      "step": 1169
    },
    {
      "epoch": 0.1630321187208249,
      "grad_norm": 1.0752640962600708,
      "learning_rate": 3.9048614408754e-05,
      "loss": 0.1352405548095703,
      "step": 1170
    },
    {
      "epoch": 0.1631714624120393,
      "grad_norm": 0.7407616972923279,
      "learning_rate": 3.904573932458398e-05,
      "loss": 0.08530235290527344,
      "step": 1171
    },
    {
      "epoch": 0.1633108061032537,
      "grad_norm": 0.7590469717979431,
      "learning_rate": 3.90428600088855e-05,
      "loss": 0.09467887878417969,
      "step": 1172
    },
    {
      "epoch": 0.16345014979446806,
      "grad_norm": 0.5962256193161011,
      "learning_rate": 3.9039976462298284e-05,
      "loss": 0.09307670593261719,
      "step": 1173
    },
    {
      "epoch": 0.16358949348568244,
      "grad_norm": 0.6715301871299744,
      "learning_rate": 3.9037088685462985e-05,
      "loss": 0.11414718627929688,
      "step": 1174
    },
    {
      "epoch": 0.16372883717689682,
      "grad_norm": 0.5469124913215637,
      "learning_rate": 3.9034196679021206e-05,
      "loss": 0.09520244598388672,
      "step": 1175
    },
    {
      "epoch": 0.1638681808681112,
      "grad_norm": 0.6244786381721497,
      "learning_rate": 3.903130044361549e-05,
      "loss": 0.07747459411621094,
      "step": 1176
    },
    {
      "epoch": 0.16400752455932557,
      "grad_norm": 1.0193482637405396,
      "learning_rate": 3.902839997988929e-05,
      "loss": 0.11363029479980469,
      "step": 1177
    },
    {
      "epoch": 0.16414686825053995,
      "grad_norm": 0.6269405484199524,
      "learning_rate": 3.902549528848705e-05,
      "loss": 0.08823013305664062,
      "step": 1178
    },
    {
      "epoch": 0.16428621194175433,
      "grad_norm": 0.9852508306503296,
      "learning_rate": 3.902258637005412e-05,
      "loss": 0.13587188720703125,
      "step": 1179
    },
    {
      "epoch": 0.1644255556329687,
      "grad_norm": 0.9998050332069397,
      "learning_rate": 3.901967322523679e-05,
      "loss": 0.09872817993164062,
      "step": 1180
    },
    {
      "epoch": 0.16456489932418308,
      "grad_norm": 1.4214932918548584,
      "learning_rate": 3.901675585468229e-05,
      "loss": 0.07799911499023438,
      "step": 1181
    },
    {
      "epoch": 0.1647042430153975,
      "grad_norm": 1.1722124814987183,
      "learning_rate": 3.9013834259038805e-05,
      "loss": 0.14375686645507812,
      "step": 1182
    },
    {
      "epoch": 0.16484358670661187,
      "grad_norm": 0.5825862288475037,
      "learning_rate": 3.9010908438955436e-05,
      "loss": 0.11171340942382812,
      "step": 1183
    },
    {
      "epoch": 0.16498293039782624,
      "grad_norm": 0.3763311207294464,
      "learning_rate": 3.900797839508225e-05,
      "loss": 0.08905601501464844,
      "step": 1184
    },
    {
      "epoch": 0.16512227408904062,
      "grad_norm": 1.557778000831604,
      "learning_rate": 3.900504412807021e-05,
      "loss": 0.13108444213867188,
      "step": 1185
    },
    {
      "epoch": 0.165261617780255,
      "grad_norm": 0.5098754167556763,
      "learning_rate": 3.900210563857127e-05,
      "loss": 0.09978866577148438,
      "step": 1186
    },
    {
      "epoch": 0.16540096147146938,
      "grad_norm": 0.5490496754646301,
      "learning_rate": 3.8999162927238274e-05,
      "loss": 0.09748268127441406,
      "step": 1187
    },
    {
      "epoch": 0.16554030516268375,
      "grad_norm": 0.7629265785217285,
      "learning_rate": 3.899621599472504e-05,
      "loss": 0.1368427276611328,
      "step": 1188
    },
    {
      "epoch": 0.16567964885389813,
      "grad_norm": 0.4778222143650055,
      "learning_rate": 3.899326484168629e-05,
      "loss": 0.09628868103027344,
      "step": 1189
    },
    {
      "epoch": 0.1658189925451125,
      "grad_norm": 0.4263807535171509,
      "learning_rate": 3.899030946877773e-05,
      "loss": 0.07402992248535156,
      "step": 1190
    },
    {
      "epoch": 0.1659583362363269,
      "grad_norm": 0.2838360667228699,
      "learning_rate": 3.898734987665596e-05,
      "loss": 0.07217597961425781,
      "step": 1191
    },
    {
      "epoch": 0.1660976799275413,
      "grad_norm": 0.4744914472103119,
      "learning_rate": 3.898438606597853e-05,
      "loss": 0.09826469421386719,
      "step": 1192
    },
    {
      "epoch": 0.16623702361875567,
      "grad_norm": 0.7216706275939941,
      "learning_rate": 3.898141803740393e-05,
      "loss": 0.10869979858398438,
      "step": 1193
    },
    {
      "epoch": 0.16637636730997005,
      "grad_norm": 0.6144678592681885,
      "learning_rate": 3.897844579159161e-05,
      "loss": 0.09122848510742188,
      "step": 1194
    },
    {
      "epoch": 0.16651571100118442,
      "grad_norm": 0.5022600293159485,
      "learning_rate": 3.897546932920191e-05,
      "loss": 0.08252906799316406,
      "step": 1195
    },
    {
      "epoch": 0.1666550546923988,
      "grad_norm": 0.6524208784103394,
      "learning_rate": 3.897248865089615e-05,
      "loss": 0.09522819519042969,
      "step": 1196
    },
    {
      "epoch": 0.16679439838361318,
      "grad_norm": 1.0800602436065674,
      "learning_rate": 3.8969503757336564e-05,
      "loss": 0.11676979064941406,
      "step": 1197
    },
    {
      "epoch": 0.16693374207482756,
      "grad_norm": 0.49601659178733826,
      "learning_rate": 3.896651464918632e-05,
      "loss": 0.07990837097167969,
      "step": 1198
    },
    {
      "epoch": 0.16707308576604193,
      "grad_norm": 0.4810534417629242,
      "learning_rate": 3.896352132710953e-05,
      "loss": 0.0899658203125,
      "step": 1199
    },
    {
      "epoch": 0.1672124294572563,
      "grad_norm": 1.1268420219421387,
      "learning_rate": 3.896052379177125e-05,
      "loss": 0.11725997924804688,
      "step": 1200
    },
    {
      "epoch": 0.1673517731484707,
      "grad_norm": 0.6495039463043213,
      "learning_rate": 3.895752204383746e-05,
      "loss": 0.09661197662353516,
      "step": 1201
    },
    {
      "epoch": 0.1674911168396851,
      "grad_norm": 0.819868803024292,
      "learning_rate": 3.8954516083975075e-05,
      "loss": 0.08772659301757812,
      "step": 1202
    },
    {
      "epoch": 0.16763046053089947,
      "grad_norm": 0.6955235004425049,
      "learning_rate": 3.8951505912851956e-05,
      "loss": 0.11422920227050781,
      "step": 1203
    },
    {
      "epoch": 0.16776980422211385,
      "grad_norm": 0.8201819658279419,
      "learning_rate": 3.89484915311369e-05,
      "loss": 0.09656143188476562,
      "step": 1204
    },
    {
      "epoch": 0.16790914791332823,
      "grad_norm": 0.7101025581359863,
      "learning_rate": 3.8945472939499616e-05,
      "loss": 0.11252784729003906,
      "step": 1205
    },
    {
      "epoch": 0.1680484916045426,
      "grad_norm": 0.3870401680469513,
      "learning_rate": 3.894245013861079e-05,
      "loss": 0.09090042114257812,
      "step": 1206
    },
    {
      "epoch": 0.16818783529575698,
      "grad_norm": 0.5592317581176758,
      "learning_rate": 3.8939423129141996e-05,
      "loss": 0.07826042175292969,
      "step": 1207
    },
    {
      "epoch": 0.16832717898697136,
      "grad_norm": 1.2223643064498901,
      "learning_rate": 3.8936391911765784e-05,
      "loss": 0.1281757354736328,
      "step": 1208
    },
    {
      "epoch": 0.16846652267818574,
      "grad_norm": 1.2784321308135986,
      "learning_rate": 3.893335648715561e-05,
      "loss": 0.14121246337890625,
      "step": 1209
    },
    {
      "epoch": 0.16860586636940011,
      "grad_norm": 0.6280313730239868,
      "learning_rate": 3.893031685598588e-05,
      "loss": 0.12414360046386719,
      "step": 1210
    },
    {
      "epoch": 0.1687452100606145,
      "grad_norm": 0.5173035264015198,
      "learning_rate": 3.8927273018931934e-05,
      "loss": 0.09871864318847656,
      "step": 1211
    },
    {
      "epoch": 0.1688845537518289,
      "grad_norm": 0.5425534248352051,
      "learning_rate": 3.892422497667004e-05,
      "loss": 0.07556819915771484,
      "step": 1212
    },
    {
      "epoch": 0.16902389744304328,
      "grad_norm": 0.7019771337509155,
      "learning_rate": 3.89211727298774e-05,
      "loss": 0.09800529479980469,
      "step": 1213
    },
    {
      "epoch": 0.16916324113425765,
      "grad_norm": 0.6072948575019836,
      "learning_rate": 3.891811627923216e-05,
      "loss": 0.09136390686035156,
      "step": 1214
    },
    {
      "epoch": 0.16930258482547203,
      "grad_norm": 1.2019731998443604,
      "learning_rate": 3.89150556254134e-05,
      "loss": 0.1353759765625,
      "step": 1215
    },
    {
      "epoch": 0.1694419285166864,
      "grad_norm": 0.5014553070068359,
      "learning_rate": 3.89119907691011e-05,
      "loss": 0.09474945068359375,
      "step": 1216
    },
    {
      "epoch": 0.16958127220790079,
      "grad_norm": 0.509661853313446,
      "learning_rate": 3.8908921710976234e-05,
      "loss": 0.13306427001953125,
      "step": 1217
    },
    {
      "epoch": 0.16972061589911516,
      "grad_norm": 0.9827958941459656,
      "learning_rate": 3.890584845172066e-05,
      "loss": 0.1262645721435547,
      "step": 1218
    },
    {
      "epoch": 0.16985995959032954,
      "grad_norm": 0.6204047203063965,
      "learning_rate": 3.890277099201718e-05,
      "loss": 0.09029579162597656,
      "step": 1219
    },
    {
      "epoch": 0.16999930328154392,
      "grad_norm": 0.5655670166015625,
      "learning_rate": 3.889968933254954e-05,
      "loss": 0.11165618896484375,
      "step": 1220
    },
    {
      "epoch": 0.1701386469727583,
      "grad_norm": 0.8767064809799194,
      "learning_rate": 3.889660347400243e-05,
      "loss": 0.10636234283447266,
      "step": 1221
    },
    {
      "epoch": 0.1702779906639727,
      "grad_norm": 1.2394474744796753,
      "learning_rate": 3.889351341706144e-05,
      "loss": 0.11235618591308594,
      "step": 1222
    },
    {
      "epoch": 0.17041733435518708,
      "grad_norm": 1.1138149499893188,
      "learning_rate": 3.8890419162413114e-05,
      "loss": 0.1050882339477539,
      "step": 1223
    },
    {
      "epoch": 0.17055667804640146,
      "grad_norm": 0.616955578327179,
      "learning_rate": 3.8887320710744923e-05,
      "loss": 0.07587432861328125,
      "step": 1224
    },
    {
      "epoch": 0.17069602173761583,
      "grad_norm": 0.2585721015930176,
      "learning_rate": 3.888421806274528e-05,
      "loss": 0.06199359893798828,
      "step": 1225
    },
    {
      "epoch": 0.1708353654288302,
      "grad_norm": 1.174164891242981,
      "learning_rate": 3.8881111219103516e-05,
      "loss": 0.08356475830078125,
      "step": 1226
    },
    {
      "epoch": 0.1709747091200446,
      "grad_norm": 1.522684097290039,
      "learning_rate": 3.88780001805099e-05,
      "loss": 0.1101837158203125,
      "step": 1227
    },
    {
      "epoch": 0.17111405281125897,
      "grad_norm": 1.34418523311615,
      "learning_rate": 3.8874884947655636e-05,
      "loss": 0.10564422607421875,
      "step": 1228
    },
    {
      "epoch": 0.17125339650247334,
      "grad_norm": 1.5710673332214355,
      "learning_rate": 3.8871765521232865e-05,
      "loss": 0.11625289916992188,
      "step": 1229
    },
    {
      "epoch": 0.17139274019368772,
      "grad_norm": 0.6479690670967102,
      "learning_rate": 3.8868641901934636e-05,
      "loss": 0.11217880249023438,
      "step": 1230
    },
    {
      "epoch": 0.1715320838849021,
      "grad_norm": 0.7428600788116455,
      "learning_rate": 3.886551409045496e-05,
      "loss": 0.1382732391357422,
      "step": 1231
    },
    {
      "epoch": 0.1716714275761165,
      "grad_norm": 1.1228066682815552,
      "learning_rate": 3.886238208748876e-05,
      "loss": 0.08400344848632812,
      "step": 1232
    },
    {
      "epoch": 0.17181077126733088,
      "grad_norm": 1.0170844793319702,
      "learning_rate": 3.885924589373189e-05,
      "loss": 0.10818862915039062,
      "step": 1233
    },
    {
      "epoch": 0.17195011495854526,
      "grad_norm": 1.093823790550232,
      "learning_rate": 3.885610550988115e-05,
      "loss": 0.08206653594970703,
      "step": 1234
    },
    {
      "epoch": 0.17208945864975964,
      "grad_norm": 1.119759440422058,
      "learning_rate": 3.885296093663426e-05,
      "loss": 0.11375808715820312,
      "step": 1235
    },
    {
      "epoch": 0.172228802340974,
      "grad_norm": 0.4808886647224426,
      "learning_rate": 3.884981217468987e-05,
      "loss": 0.08153915405273438,
      "step": 1236
    },
    {
      "epoch": 0.1723681460321884,
      "grad_norm": 0.4569849967956543,
      "learning_rate": 3.884665922474756e-05,
      "loss": 0.08654022216796875,
      "step": 1237
    },
    {
      "epoch": 0.17250748972340277,
      "grad_norm": 1.3318341970443726,
      "learning_rate": 3.884350208750784e-05,
      "loss": 0.12014198303222656,
      "step": 1238
    },
    {
      "epoch": 0.17264683341461715,
      "grad_norm": 0.34421777725219727,
      "learning_rate": 3.884034076367218e-05,
      "loss": 0.08241653442382812,
      "step": 1239
    },
    {
      "epoch": 0.17278617710583152,
      "grad_norm": 1.0937418937683105,
      "learning_rate": 3.883717525394292e-05,
      "loss": 0.120574951171875,
      "step": 1240
    },
    {
      "epoch": 0.1729255207970459,
      "grad_norm": 0.6259034872055054,
      "learning_rate": 3.883400555902338e-05,
      "loss": 0.09210014343261719,
      "step": 1241
    },
    {
      "epoch": 0.1730648644882603,
      "grad_norm": 0.722667396068573,
      "learning_rate": 3.88308316796178e-05,
      "loss": 0.089874267578125,
      "step": 1242
    },
    {
      "epoch": 0.17320420817947468,
      "grad_norm": 0.677371084690094,
      "learning_rate": 3.882765361643133e-05,
      "loss": 0.12020492553710938,
      "step": 1243
    },
    {
      "epoch": 0.17334355187068906,
      "grad_norm": 0.6265414357185364,
      "learning_rate": 3.882447137017007e-05,
      "loss": 0.10531425476074219,
      "step": 1244
    },
    {
      "epoch": 0.17348289556190344,
      "grad_norm": 0.8198543787002563,
      "learning_rate": 3.882128494154104e-05,
      "loss": 0.11435508728027344,
      "step": 1245
    },
    {
      "epoch": 0.17362223925311782,
      "grad_norm": 0.7933017611503601,
      "learning_rate": 3.8818094331252194e-05,
      "loss": 0.10502433776855469,
      "step": 1246
    },
    {
      "epoch": 0.1737615829443322,
      "grad_norm": 0.855887770652771,
      "learning_rate": 3.881489954001241e-05,
      "loss": 0.11224746704101562,
      "step": 1247
    },
    {
      "epoch": 0.17390092663554657,
      "grad_norm": 0.5570697784423828,
      "learning_rate": 3.88117005685315e-05,
      "loss": 0.08007240295410156,
      "step": 1248
    },
    {
      "epoch": 0.17404027032676095,
      "grad_norm": 0.36361071467399597,
      "learning_rate": 3.88084974175202e-05,
      "loss": 0.07812881469726562,
      "step": 1249
    },
    {
      "epoch": 0.17417961401797533,
      "grad_norm": 0.6572778224945068,
      "learning_rate": 3.8805290087690196e-05,
      "loss": 0.09654045104980469,
      "step": 1250
    },
    {
      "epoch": 0.1743189577091897,
      "grad_norm": 1.4070894718170166,
      "learning_rate": 3.880207857975405e-05,
      "loss": 0.16957855224609375,
      "step": 1251
    },
    {
      "epoch": 0.1744583014004041,
      "grad_norm": 0.8828078508377075,
      "learning_rate": 3.879886289442531e-05,
      "loss": 0.11992454528808594,
      "step": 1252
    },
    {
      "epoch": 0.1745976450916185,
      "grad_norm": 0.34186825156211853,
      "learning_rate": 3.879564303241841e-05,
      "loss": 0.07204246520996094,
      "step": 1253
    },
    {
      "epoch": 0.17473698878283286,
      "grad_norm": 2.0179011821746826,
      "learning_rate": 3.8792418994448746e-05,
      "loss": 0.17632484436035156,
      "step": 1254
    },
    {
      "epoch": 0.17487633247404724,
      "grad_norm": 0.9922803044319153,
      "learning_rate": 3.8789190781232626e-05,
      "loss": 0.1372966766357422,
      "step": 1255
    },
    {
      "epoch": 0.17501567616526162,
      "grad_norm": 0.3026036024093628,
      "learning_rate": 3.878595839348727e-05,
      "loss": 0.0642852783203125,
      "step": 1256
    },
    {
      "epoch": 0.175155019856476,
      "grad_norm": 0.7335386872291565,
      "learning_rate": 3.878272183193085e-05,
      "loss": 0.10975360870361328,
      "step": 1257
    },
    {
      "epoch": 0.17529436354769037,
      "grad_norm": 0.6881363391876221,
      "learning_rate": 3.8779481097282464e-05,
      "loss": 0.09327507019042969,
      "step": 1258
    },
    {
      "epoch": 0.17543370723890475,
      "grad_norm": 1.0665518045425415,
      "learning_rate": 3.8776236190262114e-05,
      "loss": 0.11595344543457031,
      "step": 1259
    },
    {
      "epoch": 0.17557305093011913,
      "grad_norm": 0.3905077278614044,
      "learning_rate": 3.877298711159076e-05,
      "loss": 0.08664131164550781,
      "step": 1260
    },
    {
      "epoch": 0.1757123946213335,
      "grad_norm": 0.4281877875328064,
      "learning_rate": 3.876973386199025e-05,
      "loss": 0.059622764587402344,
      "step": 1261
    },
    {
      "epoch": 0.1758517383125479,
      "grad_norm": 0.8304551839828491,
      "learning_rate": 3.87664764421834e-05,
      "loss": 0.10391426086425781,
      "step": 1262
    },
    {
      "epoch": 0.1759910820037623,
      "grad_norm": 0.7611992359161377,
      "learning_rate": 3.876321485289394e-05,
      "loss": 0.10839557647705078,
      "step": 1263
    },
    {
      "epoch": 0.17613042569497667,
      "grad_norm": 0.44577690958976746,
      "learning_rate": 3.87599490948465e-05,
      "loss": 0.092315673828125,
      "step": 1264
    },
    {
      "epoch": 0.17626976938619104,
      "grad_norm": 0.5830724239349365,
      "learning_rate": 3.875667916876668e-05,
      "loss": 0.09503555297851562,
      "step": 1265
    },
    {
      "epoch": 0.17640911307740542,
      "grad_norm": 0.49108240008354187,
      "learning_rate": 3.875340507538096e-05,
      "loss": 0.07281875610351562,
      "step": 1266
    },
    {
      "epoch": 0.1765484567686198,
      "grad_norm": 0.4723402261734009,
      "learning_rate": 3.875012681541678e-05,
      "loss": 0.11981391906738281,
      "step": 1267
    },
    {
      "epoch": 0.17668780045983418,
      "grad_norm": 0.7398895621299744,
      "learning_rate": 3.87468443896025e-05,
      "loss": 0.098968505859375,
      "step": 1268
    },
    {
      "epoch": 0.17682714415104855,
      "grad_norm": 0.6669520139694214,
      "learning_rate": 3.8743557798667395e-05,
      "loss": 0.10626983642578125,
      "step": 1269
    },
    {
      "epoch": 0.17696648784226293,
      "grad_norm": 1.3382683992385864,
      "learning_rate": 3.874026704334167e-05,
      "loss": 0.14954757690429688,
      "step": 1270
    },
    {
      "epoch": 0.1771058315334773,
      "grad_norm": 0.5279415845870972,
      "learning_rate": 3.873697212435645e-05,
      "loss": 0.13609886169433594,
      "step": 1271
    },
    {
      "epoch": 0.17724517522469171,
      "grad_norm": 0.9191502928733826,
      "learning_rate": 3.87336730424438e-05,
      "loss": 0.11865234375,
      "step": 1272
    },
    {
      "epoch": 0.1773845189159061,
      "grad_norm": 0.9079429507255554,
      "learning_rate": 3.87303697983367e-05,
      "loss": 0.11186027526855469,
      "step": 1273
    },
    {
      "epoch": 0.17752386260712047,
      "grad_norm": 1.1797435283660889,
      "learning_rate": 3.872706239276904e-05,
      "loss": 0.11644363403320312,
      "step": 1274
    },
    {
      "epoch": 0.17766320629833485,
      "grad_norm": 0.5337759256362915,
      "learning_rate": 3.8723750826475674e-05,
      "loss": 0.12830543518066406,
      "step": 1275
    },
    {
      "epoch": 0.17780254998954922,
      "grad_norm": 0.5210356712341309,
      "learning_rate": 3.872043510019235e-05,
      "loss": 0.09801864624023438,
      "step": 1276
    },
    {
      "epoch": 0.1779418936807636,
      "grad_norm": 0.7031455039978027,
      "learning_rate": 3.871711521465573e-05,
      "loss": 0.09349822998046875,
      "step": 1277
    },
    {
      "epoch": 0.17808123737197798,
      "grad_norm": 0.5507082939147949,
      "learning_rate": 3.871379117060343e-05,
      "loss": 0.09299468994140625,
      "step": 1278
    },
    {
      "epoch": 0.17822058106319236,
      "grad_norm": 0.5519847869873047,
      "learning_rate": 3.871046296877398e-05,
      "loss": 0.10372352600097656,
      "step": 1279
    },
    {
      "epoch": 0.17835992475440673,
      "grad_norm": 0.41661226749420166,
      "learning_rate": 3.870713060990682e-05,
      "loss": 0.08462715148925781,
      "step": 1280
    },
    {
      "epoch": 0.1784992684456211,
      "grad_norm": 0.5839899778366089,
      "learning_rate": 3.870379409474233e-05,
      "loss": 0.11013317108154297,
      "step": 1281
    },
    {
      "epoch": 0.17863861213683552,
      "grad_norm": 0.6457597017288208,
      "learning_rate": 3.870045342402181e-05,
      "loss": 0.08987808227539062,
      "step": 1282
    },
    {
      "epoch": 0.1787779558280499,
      "grad_norm": 0.3778478801250458,
      "learning_rate": 3.8697108598487474e-05,
      "loss": 0.0996694564819336,
      "step": 1283
    },
    {
      "epoch": 0.17891729951926427,
      "grad_norm": 0.8293259143829346,
      "learning_rate": 3.8693759618882475e-05,
      "loss": 0.11539459228515625,
      "step": 1284
    },
    {
      "epoch": 0.17905664321047865,
      "grad_norm": 0.920018196105957,
      "learning_rate": 3.8690406485950874e-05,
      "loss": 0.09552955627441406,
      "step": 1285
    },
    {
      "epoch": 0.17919598690169303,
      "grad_norm": 0.4179550111293793,
      "learning_rate": 3.868704920043766e-05,
      "loss": 0.08619308471679688,
      "step": 1286
    },
    {
      "epoch": 0.1793353305929074,
      "grad_norm": 0.8999516367912292,
      "learning_rate": 3.8683687763088745e-05,
      "loss": 0.16223907470703125,
      "step": 1287
    },
    {
      "epoch": 0.17947467428412178,
      "grad_norm": 1.5918910503387451,
      "learning_rate": 3.868032217465097e-05,
      "loss": 0.15572357177734375,
      "step": 1288
    },
    {
      "epoch": 0.17961401797533616,
      "grad_norm": 0.9245583415031433,
      "learning_rate": 3.867695243587207e-05,
      "loss": 0.12650299072265625,
      "step": 1289
    },
    {
      "epoch": 0.17975336166655054,
      "grad_norm": 0.6573454737663269,
      "learning_rate": 3.8673578547500754e-05,
      "loss": 0.10108566284179688,
      "step": 1290
    },
    {
      "epoch": 0.17989270535776491,
      "grad_norm": 0.7187705039978027,
      "learning_rate": 3.867020051028661e-05,
      "loss": 0.10828971862792969,
      "step": 1291
    },
    {
      "epoch": 0.18003204904897932,
      "grad_norm": 0.686320960521698,
      "learning_rate": 3.8666818324980165e-05,
      "loss": 0.14690113067626953,
      "step": 1292
    },
    {
      "epoch": 0.1801713927401937,
      "grad_norm": 1.088727355003357,
      "learning_rate": 3.866343199233285e-05,
      "loss": 0.10896492004394531,
      "step": 1293
    },
    {
      "epoch": 0.18031073643140808,
      "grad_norm": 0.9809849858283997,
      "learning_rate": 3.866004151309704e-05,
      "loss": 0.1352214813232422,
      "step": 1294
    },
    {
      "epoch": 0.18045008012262245,
      "grad_norm": 0.3993098735809326,
      "learning_rate": 3.8656646888026026e-05,
      "loss": 0.08288383483886719,
      "step": 1295
    },
    {
      "epoch": 0.18058942381383683,
      "grad_norm": 0.683144748210907,
      "learning_rate": 3.8653248117874015e-05,
      "loss": 0.11868476867675781,
      "step": 1296
    },
    {
      "epoch": 0.1807287675050512,
      "grad_norm": 0.6973060965538025,
      "learning_rate": 3.8649845203396125e-05,
      "loss": 0.07170772552490234,
      "step": 1297
    },
    {
      "epoch": 0.18086811119626559,
      "grad_norm": 0.8602525591850281,
      "learning_rate": 3.8646438145348415e-05,
      "loss": 0.14191436767578125,
      "step": 1298
    },
    {
      "epoch": 0.18100745488747996,
      "grad_norm": 0.7572188377380371,
      "learning_rate": 3.8643026944487856e-05,
      "loss": 0.12969970703125,
      "step": 1299
    },
    {
      "epoch": 0.18114679857869434,
      "grad_norm": 0.35731810331344604,
      "learning_rate": 3.8639611601572345e-05,
      "loss": 0.09122085571289062,
      "step": 1300
    },
    {
      "epoch": 0.18128614226990872,
      "grad_norm": 0.39366886019706726,
      "learning_rate": 3.8636192117360676e-05,
      "loss": 0.1086273193359375,
      "step": 1301
    },
    {
      "epoch": 0.18142548596112312,
      "grad_norm": 0.46176785230636597,
      "learning_rate": 3.8632768492612596e-05,
      "loss": 0.08187675476074219,
      "step": 1302
    },
    {
      "epoch": 0.1815648296523375,
      "grad_norm": 1.469296932220459,
      "learning_rate": 3.862934072808875e-05,
      "loss": 0.11510562896728516,
      "step": 1303
    },
    {
      "epoch": 0.18170417334355188,
      "grad_norm": 1.019187331199646,
      "learning_rate": 3.86259088245507e-05,
      "loss": 0.12382888793945312,
      "step": 1304
    },
    {
      "epoch": 0.18184351703476626,
      "grad_norm": 0.5867264866828918,
      "learning_rate": 3.8622472782760956e-05,
      "loss": 0.15715408325195312,
      "step": 1305
    },
    {
      "epoch": 0.18198286072598063,
      "grad_norm": 0.5373423099517822,
      "learning_rate": 3.861903260348291e-05,
      "loss": 0.11078834533691406,
      "step": 1306
    },
    {
      "epoch": 0.182122204417195,
      "grad_norm": 0.8571909666061401,
      "learning_rate": 3.8615588287480906e-05,
      "loss": 0.13559341430664062,
      "step": 1307
    },
    {
      "epoch": 0.1822615481084094,
      "grad_norm": 1.1384767293930054,
      "learning_rate": 3.861213983552018e-05,
      "loss": 0.10149955749511719,
      "step": 1308
    },
    {
      "epoch": 0.18240089179962377,
      "grad_norm": 0.8413253426551819,
      "learning_rate": 3.860868724836691e-05,
      "loss": 0.1054534912109375,
      "step": 1309
    },
    {
      "epoch": 0.18254023549083814,
      "grad_norm": 0.5824040770530701,
      "learning_rate": 3.860523052678818e-05,
      "loss": 0.10660362243652344,
      "step": 1310
    },
    {
      "epoch": 0.18267957918205252,
      "grad_norm": 0.576574981212616,
      "learning_rate": 3.860176967155198e-05,
      "loss": 0.08987808227539062,
      "step": 1311
    },
    {
      "epoch": 0.18281892287326693,
      "grad_norm": 0.6395757794380188,
      "learning_rate": 3.8598304683427257e-05,
      "loss": 0.08406829833984375,
      "step": 1312
    },
    {
      "epoch": 0.1829582665644813,
      "grad_norm": 0.6899009943008423,
      "learning_rate": 3.859483556318384e-05,
      "loss": 0.09299850463867188,
      "step": 1313
    },
    {
      "epoch": 0.18309761025569568,
      "grad_norm": 0.7416677474975586,
      "learning_rate": 3.859136231159248e-05,
      "loss": 0.08332252502441406,
      "step": 1314
    },
    {
      "epoch": 0.18323695394691006,
      "grad_norm": 0.5145108103752136,
      "learning_rate": 3.858788492942486e-05,
      "loss": 0.12900733947753906,
      "step": 1315
    },
    {
      "epoch": 0.18337629763812444,
      "grad_norm": 0.480636864900589,
      "learning_rate": 3.8584403417453586e-05,
      "loss": 0.08121299743652344,
      "step": 1316
    },
    {
      "epoch": 0.1835156413293388,
      "grad_norm": 1.873380184173584,
      "learning_rate": 3.858091777645216e-05,
      "loss": 0.14214706420898438,
      "step": 1317
    },
    {
      "epoch": 0.1836549850205532,
      "grad_norm": 1.2113927602767944,
      "learning_rate": 3.857742800719501e-05,
      "loss": 0.08494377136230469,
      "step": 1318
    },
    {
      "epoch": 0.18379432871176757,
      "grad_norm": 0.44741958379745483,
      "learning_rate": 3.857393411045749e-05,
      "loss": 0.085693359375,
      "step": 1319
    },
    {
      "epoch": 0.18393367240298195,
      "grad_norm": 0.5662305355072021,
      "learning_rate": 3.8570436087015855e-05,
      "loss": 0.10150623321533203,
      "step": 1320
    },
    {
      "epoch": 0.18407301609419632,
      "grad_norm": 0.5008440613746643,
      "learning_rate": 3.8566933937647294e-05,
      "loss": 0.08993721008300781,
      "step": 1321
    },
    {
      "epoch": 0.18421235978541073,
      "grad_norm": 0.8278573751449585,
      "learning_rate": 3.856342766312991e-05,
      "loss": 0.1232757568359375,
      "step": 1322
    },
    {
      "epoch": 0.1843517034766251,
      "grad_norm": 0.5242437720298767,
      "learning_rate": 3.85599172642427e-05,
      "loss": 0.11133146286010742,
      "step": 1323
    },
    {
      "epoch": 0.18449104716783948,
      "grad_norm": 0.5966528058052063,
      "learning_rate": 3.855640274176561e-05,
      "loss": 0.08792304992675781,
      "step": 1324
    },
    {
      "epoch": 0.18463039085905386,
      "grad_norm": 0.3977065682411194,
      "learning_rate": 3.8552884096479476e-05,
      "loss": 0.09332084655761719,
      "step": 1325
    },
    {
      "epoch": 0.18476973455026824,
      "grad_norm": 0.689285933971405,
      "learning_rate": 3.854936132916607e-05,
      "loss": 0.11291122436523438,
      "step": 1326
    },
    {
      "epoch": 0.18490907824148262,
      "grad_norm": 1.0005016326904297,
      "learning_rate": 3.854583444060806e-05,
      "loss": 0.12997817993164062,
      "step": 1327
    },
    {
      "epoch": 0.185048421932697,
      "grad_norm": 0.4531738758087158,
      "learning_rate": 3.854230343158906e-05,
      "loss": 0.08896446228027344,
      "step": 1328
    },
    {
      "epoch": 0.18518776562391137,
      "grad_norm": 0.4677441418170929,
      "learning_rate": 3.8538768302893544e-05,
      "loss": 0.08194732666015625,
      "step": 1329
    },
    {
      "epoch": 0.18532710931512575,
      "grad_norm": 1.2755392789840698,
      "learning_rate": 3.853522905530698e-05,
      "loss": 0.11968421936035156,
      "step": 1330
    },
    {
      "epoch": 0.18546645300634013,
      "grad_norm": 0.7076494693756104,
      "learning_rate": 3.853168568961567e-05,
      "loss": 0.10310173034667969,
      "step": 1331
    },
    {
      "epoch": 0.18560579669755453,
      "grad_norm": 0.7455466389656067,
      "learning_rate": 3.852813820660689e-05,
      "loss": 0.13688278198242188,
      "step": 1332
    },
    {
      "epoch": 0.1857451403887689,
      "grad_norm": 0.9865037202835083,
      "learning_rate": 3.852458660706881e-05,
      "loss": 0.12118148803710938,
      "step": 1333
    },
    {
      "epoch": 0.1858844840799833,
      "grad_norm": 0.9995445013046265,
      "learning_rate": 3.85210308917905e-05,
      "loss": 0.0994405746459961,
      "step": 1334
    },
    {
      "epoch": 0.18602382777119766,
      "grad_norm": 0.48521867394447327,
      "learning_rate": 3.8517471061561974e-05,
      "loss": 0.08416080474853516,
      "step": 1335
    },
    {
      "epoch": 0.18616317146241204,
      "grad_norm": 0.42504364252090454,
      "learning_rate": 3.851390711717414e-05,
      "loss": 0.08894538879394531,
      "step": 1336
    },
    {
      "epoch": 0.18630251515362642,
      "grad_norm": 0.24130935966968536,
      "learning_rate": 3.851033905941882e-05,
      "loss": 0.06106853485107422,
      "step": 1337
    },
    {
      "epoch": 0.1864418588448408,
      "grad_norm": 0.8676820397377014,
      "learning_rate": 3.850676688908877e-05,
      "loss": 0.10792732238769531,
      "step": 1338
    },
    {
      "epoch": 0.18658120253605517,
      "grad_norm": 0.7004497051239014,
      "learning_rate": 3.8503190606977624e-05,
      "loss": 0.07880973815917969,
      "step": 1339
    },
    {
      "epoch": 0.18672054622726955,
      "grad_norm": 0.48882248997688293,
      "learning_rate": 3.849961021387996e-05,
      "loss": 0.09171295166015625,
      "step": 1340
    },
    {
      "epoch": 0.18685988991848393,
      "grad_norm": 0.9736695289611816,
      "learning_rate": 3.849602571059127e-05,
      "loss": 0.1238861083984375,
      "step": 1341
    },
    {
      "epoch": 0.18699923360969833,
      "grad_norm": 0.3489600419998169,
      "learning_rate": 3.849243709790793e-05,
      "loss": 0.07259559631347656,
      "step": 1342
    },
    {
      "epoch": 0.1871385773009127,
      "grad_norm": 0.7062949538230896,
      "learning_rate": 3.848884437662725e-05,
      "loss": 0.10225868225097656,
      "step": 1343
    },
    {
      "epoch": 0.1872779209921271,
      "grad_norm": 0.7970095276832581,
      "learning_rate": 3.8485247547547465e-05,
      "loss": 0.1495208740234375,
      "step": 1344
    },
    {
      "epoch": 0.18741726468334147,
      "grad_norm": 0.3726244568824768,
      "learning_rate": 3.8481646611467704e-05,
      "loss": 0.09688758850097656,
      "step": 1345
    },
    {
      "epoch": 0.18755660837455584,
      "grad_norm": 0.7556194067001343,
      "learning_rate": 3.8478041569188e-05,
      "loss": 0.12447166442871094,
      "step": 1346
    },
    {
      "epoch": 0.18769595206577022,
      "grad_norm": 1.2556729316711426,
      "learning_rate": 3.8474432421509324e-05,
      "loss": 0.1305103302001953,
      "step": 1347
    },
    {
      "epoch": 0.1878352957569846,
      "grad_norm": 0.9894823431968689,
      "learning_rate": 3.847081916923355e-05,
      "loss": 0.12862586975097656,
      "step": 1348
    },
    {
      "epoch": 0.18797463944819898,
      "grad_norm": 0.37070226669311523,
      "learning_rate": 3.846720181316344e-05,
      "loss": 0.08415603637695312,
      "step": 1349
    },
    {
      "epoch": 0.18811398313941335,
      "grad_norm": 1.2234729528427124,
      "learning_rate": 3.846358035410271e-05,
      "loss": 0.12853240966796875,
      "step": 1350
    },
    {
      "epoch": 0.18825332683062773,
      "grad_norm": 0.48727330565452576,
      "learning_rate": 3.845995479285595e-05,
      "loss": 0.08751487731933594,
      "step": 1351
    },
    {
      "epoch": 0.18839267052184214,
      "grad_norm": 0.5283886194229126,
      "learning_rate": 3.845632513022869e-05,
      "loss": 0.1024932861328125,
      "step": 1352
    },
    {
      "epoch": 0.18853201421305651,
      "grad_norm": 1.0089560747146606,
      "learning_rate": 3.845269136702734e-05,
      "loss": 0.11927986145019531,
      "step": 1353
    },
    {
      "epoch": 0.1886713579042709,
      "grad_norm": 1.1472110748291016,
      "learning_rate": 3.844905350405926e-05,
      "loss": 0.14356613159179688,
      "step": 1354
    },
    {
      "epoch": 0.18881070159548527,
      "grad_norm": 0.8192573189735413,
      "learning_rate": 3.8445411542132684e-05,
      "loss": 0.156646728515625,
      "step": 1355
    },
    {
      "epoch": 0.18895004528669965,
      "grad_norm": 0.3285615146160126,
      "learning_rate": 3.8441765482056783e-05,
      "loss": 0.0782928466796875,
      "step": 1356
    },
    {
      "epoch": 0.18908938897791402,
      "grad_norm": 1.471805453300476,
      "learning_rate": 3.843811532464163e-05,
      "loss": 0.16009140014648438,
      "step": 1357
    },
    {
      "epoch": 0.1892287326691284,
      "grad_norm": 0.6613411903381348,
      "learning_rate": 3.8434461070698194e-05,
      "loss": 0.14098548889160156,
      "step": 1358
    },
    {
      "epoch": 0.18936807636034278,
      "grad_norm": 0.34900209307670593,
      "learning_rate": 3.843080272103837e-05,
      "loss": 0.10255050659179688,
      "step": 1359
    },
    {
      "epoch": 0.18950742005155716,
      "grad_norm": 0.5859901905059814,
      "learning_rate": 3.842714027647497e-05,
      "loss": 0.11478805541992188,
      "step": 1360
    },
    {
      "epoch": 0.18964676374277153,
      "grad_norm": 0.4439310133457184,
      "learning_rate": 3.8423473737821705e-05,
      "loss": 0.09545707702636719,
      "step": 1361
    },
    {
      "epoch": 0.18978610743398594,
      "grad_norm": 0.5977876782417297,
      "learning_rate": 3.8419803105893175e-05,
      "loss": 0.103790283203125,
      "step": 1362
    },
    {
      "epoch": 0.18992545112520032,
      "grad_norm": 0.6895452737808228,
      "learning_rate": 3.841612838150494e-05,
      "loss": 0.11831474304199219,
      "step": 1363
    },
    {
      "epoch": 0.1900647948164147,
      "grad_norm": 0.6594695448875427,
      "learning_rate": 3.8412449565473414e-05,
      "loss": 0.11533546447753906,
      "step": 1364
    },
    {
      "epoch": 0.19020413850762907,
      "grad_norm": 0.6502701640129089,
      "learning_rate": 3.840876665861597e-05,
      "loss": 0.11409378051757812,
      "step": 1365
    },
    {
      "epoch": 0.19034348219884345,
      "grad_norm": 0.7259324789047241,
      "learning_rate": 3.840507966175085e-05,
      "loss": 0.125457763671875,
      "step": 1366
    },
    {
      "epoch": 0.19048282589005783,
      "grad_norm": 0.34324994683265686,
      "learning_rate": 3.840138857569722e-05,
      "loss": 0.08785438537597656,
      "step": 1367
    },
    {
      "epoch": 0.1906221695812722,
      "grad_norm": 0.6383782029151917,
      "learning_rate": 3.8397693401275165e-05,
      "loss": 0.10250473022460938,
      "step": 1368
    },
    {
      "epoch": 0.19076151327248658,
      "grad_norm": 0.4695620834827423,
      "learning_rate": 3.8393994139305656e-05,
      "loss": 0.08772659301757812,
      "step": 1369
    },
    {
      "epoch": 0.19090085696370096,
      "grad_norm": 0.5953870415687561,
      "learning_rate": 3.8390290790610595e-05,
      "loss": 0.09298896789550781,
      "step": 1370
    },
    {
      "epoch": 0.19104020065491534,
      "grad_norm": 0.20506471395492554,
      "learning_rate": 3.838658335601278e-05,
      "loss": 0.06280708312988281,
      "step": 1371
    },
    {
      "epoch": 0.19117954434612974,
      "grad_norm": 0.7083378434181213,
      "learning_rate": 3.838287183633591e-05,
      "loss": 0.12908172607421875,
      "step": 1372
    },
    {
      "epoch": 0.19131888803734412,
      "grad_norm": 0.6947250962257385,
      "learning_rate": 3.837915623240462e-05,
      "loss": 0.14439964294433594,
      "step": 1373
    },
    {
      "epoch": 0.1914582317285585,
      "grad_norm": 0.7821818590164185,
      "learning_rate": 3.837543654504441e-05,
      "loss": 0.10806083679199219,
      "step": 1374
    },
    {
      "epoch": 0.19159757541977288,
      "grad_norm": 1.120538592338562,
      "learning_rate": 3.837171277508171e-05,
      "loss": 0.11051559448242188,
      "step": 1375
    },
    {
      "epoch": 0.19173691911098725,
      "grad_norm": 0.6469658017158508,
      "learning_rate": 3.836798492334387e-05,
      "loss": 0.12801742553710938,
      "step": 1376
    },
    {
      "epoch": 0.19187626280220163,
      "grad_norm": 0.38692253828048706,
      "learning_rate": 3.836425299065913e-05,
      "loss": 0.09635066986083984,
      "step": 1377
    },
    {
      "epoch": 0.192015606493416,
      "grad_norm": 0.3441038131713867,
      "learning_rate": 3.836051697785664e-05,
      "loss": 0.08381271362304688,
      "step": 1378
    },
    {
      "epoch": 0.19215495018463039,
      "grad_norm": 0.7907981276512146,
      "learning_rate": 3.8356776885766456e-05,
      "loss": 0.10181045532226562,
      "step": 1379
    },
    {
      "epoch": 0.19229429387584476,
      "grad_norm": 0.6198570132255554,
      "learning_rate": 3.8353032715219534e-05,
      "loss": 0.09758758544921875,
      "step": 1380
    },
    {
      "epoch": 0.19243363756705914,
      "grad_norm": 0.4087033271789551,
      "learning_rate": 3.834928446704775e-05,
      "loss": 0.09213066101074219,
      "step": 1381
    },
    {
      "epoch": 0.19257298125827352,
      "grad_norm": 0.8019723296165466,
      "learning_rate": 3.834553214208389e-05,
      "loss": 0.09819984436035156,
      "step": 1382
    },
    {
      "epoch": 0.19271232494948792,
      "grad_norm": 0.8303682804107666,
      "learning_rate": 3.834177574116161e-05,
      "loss": 0.08117103576660156,
      "step": 1383
    },
    {
      "epoch": 0.1928516686407023,
      "grad_norm": 0.6511688828468323,
      "learning_rate": 3.833801526511552e-05,
      "loss": 0.0792388916015625,
      "step": 1384
    },
    {
      "epoch": 0.19299101233191668,
      "grad_norm": 0.850060760974884,
      "learning_rate": 3.83342507147811e-05,
      "loss": 0.10252189636230469,
      "step": 1385
    },
    {
      "epoch": 0.19313035602313106,
      "grad_norm": 0.3934380114078522,
      "learning_rate": 3.833048209099474e-05,
      "loss": 0.07392120361328125,
      "step": 1386
    },
    {
      "epoch": 0.19326969971434543,
      "grad_norm": 0.7533586025238037,
      "learning_rate": 3.832670939459376e-05,
      "loss": 0.09762191772460938,
      "step": 1387
    },
    {
      "epoch": 0.1934090434055598,
      "grad_norm": 0.4304204285144806,
      "learning_rate": 3.832293262641636e-05,
      "loss": 0.07915878295898438,
      "step": 1388
    },
    {
      "epoch": 0.1935483870967742,
      "grad_norm": 1.1906611919403076,
      "learning_rate": 3.8319151787301644e-05,
      "loss": 0.12559127807617188,
      "step": 1389
    },
    {
      "epoch": 0.19368773078798857,
      "grad_norm": 0.693774402141571,
      "learning_rate": 3.831536687808964e-05,
      "loss": 0.09670066833496094,
      "step": 1390
    },
    {
      "epoch": 0.19382707447920294,
      "grad_norm": 0.7913341522216797,
      "learning_rate": 3.831157789962126e-05,
      "loss": 0.10721015930175781,
      "step": 1391
    },
    {
      "epoch": 0.19396641817041732,
      "grad_norm": 0.48556143045425415,
      "learning_rate": 3.830778485273833e-05,
      "loss": 0.09945297241210938,
      "step": 1392
    },
    {
      "epoch": 0.19410576186163173,
      "grad_norm": 0.5833231210708618,
      "learning_rate": 3.830398773828358e-05,
      "loss": 0.1147308349609375,
      "step": 1393
    },
    {
      "epoch": 0.1942451055528461,
      "grad_norm": 0.9020781517028809,
      "learning_rate": 3.830018655710064e-05,
      "loss": 0.11287307739257812,
      "step": 1394
    },
    {
      "epoch": 0.19438444924406048,
      "grad_norm": 0.34590286016464233,
      "learning_rate": 3.829638131003405e-05,
      "loss": 0.0658721923828125,
      "step": 1395
    },
    {
      "epoch": 0.19452379293527486,
      "grad_norm": 1.3885587453842163,
      "learning_rate": 3.829257199792925e-05,
      "loss": 0.11473846435546875,
      "step": 1396
    },
    {
      "epoch": 0.19466313662648924,
      "grad_norm": 0.45876994729042053,
      "learning_rate": 3.828875862163258e-05,
      "loss": 0.09295082092285156,
      "step": 1397
    },
    {
      "epoch": 0.1948024803177036,
      "grad_norm": 0.8232770562171936,
      "learning_rate": 3.828494118199127e-05,
      "loss": 0.11212730407714844,
      "step": 1398
    },
    {
      "epoch": 0.194941824008918,
      "grad_norm": 0.7585610747337341,
      "learning_rate": 3.828111967985349e-05,
      "loss": 0.10973358154296875,
      "step": 1399
    },
    {
      "epoch": 0.19508116770013237,
      "grad_norm": 0.4467112123966217,
      "learning_rate": 3.8277294116068285e-05,
      "loss": 0.09208488464355469,
      "step": 1400
    },
    {
      "epoch": 0.19522051139134675,
      "grad_norm": 2.012354850769043,
      "learning_rate": 3.8273464491485596e-05,
      "loss": 0.15551376342773438,
      "step": 1401
    },
    {
      "epoch": 0.19535985508256112,
      "grad_norm": 1.0755252838134766,
      "learning_rate": 3.82696308069563e-05,
      "loss": 0.10248565673828125,
      "step": 1402
    },
    {
      "epoch": 0.19549919877377553,
      "grad_norm": 0.7562090158462524,
      "learning_rate": 3.8265793063332135e-05,
      "loss": 0.09842872619628906,
      "step": 1403
    },
    {
      "epoch": 0.1956385424649899,
      "grad_norm": 0.823768138885498,
      "learning_rate": 3.826195126146576e-05,
      "loss": 0.1348285675048828,
      "step": 1404
    },
    {
      "epoch": 0.19577788615620428,
      "grad_norm": 0.3633580207824707,
      "learning_rate": 3.8258105402210755e-05,
      "loss": 0.09007930755615234,
      "step": 1405
    },
    {
      "epoch": 0.19591722984741866,
      "grad_norm": 0.4143328368663788,
      "learning_rate": 3.825425548642156e-05,
      "loss": 0.07435894012451172,
      "step": 1406
    },
    {
      "epoch": 0.19605657353863304,
      "grad_norm": 0.6795492172241211,
      "learning_rate": 3.8250401514953557e-05,
      "loss": 0.1013193130493164,
      "step": 1407
    },
    {
      "epoch": 0.19619591722984742,
      "grad_norm": 0.8466312885284424,
      "learning_rate": 3.824654348866299e-05,
      "loss": 0.08840751647949219,
      "step": 1408
    },
    {
      "epoch": 0.1963352609210618,
      "grad_norm": 0.8269075155258179,
      "learning_rate": 3.824268140840704e-05,
      "loss": 0.11049461364746094,
      "step": 1409
    },
    {
      "epoch": 0.19647460461227617,
      "grad_norm": 1.1659709215164185,
      "learning_rate": 3.823881527504377e-05,
      "loss": 0.14566993713378906,
      "step": 1410
    },
    {
      "epoch": 0.19661394830349055,
      "grad_norm": 0.4802590608596802,
      "learning_rate": 3.823494508943214e-05,
      "loss": 0.10031318664550781,
      "step": 1411
    },
    {
      "epoch": 0.19675329199470493,
      "grad_norm": 1.2405728101730347,
      "learning_rate": 3.8231070852432035e-05,
      "loss": 0.17720794677734375,
      "step": 1412
    },
    {
      "epoch": 0.19689263568591933,
      "grad_norm": 0.33376267552375793,
      "learning_rate": 3.82271925649042e-05,
      "loss": 0.08036231994628906,
      "step": 1413
    },
    {
      "epoch": 0.1970319793771337,
      "grad_norm": 0.7798534035682678,
      "learning_rate": 3.822331022771031e-05,
      "loss": 0.09315681457519531,
      "step": 1414
    },
    {
      "epoch": 0.1971713230683481,
      "grad_norm": 0.26651304960250854,
      "learning_rate": 3.8219423841712935e-05,
      "loss": 0.06585693359375,
      "step": 1415
    },
    {
      "epoch": 0.19731066675956246,
      "grad_norm": 0.33895596861839294,
      "learning_rate": 3.821553340777553e-05,
      "loss": 0.07099342346191406,
      "step": 1416
    },
    {
      "epoch": 0.19745001045077684,
      "grad_norm": 0.7491000890731812,
      "learning_rate": 3.821163892676248e-05,
      "loss": 0.1431446075439453,
      "step": 1417
    },
    {
      "epoch": 0.19758935414199122,
      "grad_norm": 0.48909980058670044,
      "learning_rate": 3.820774039953904e-05,
      "loss": 0.10382652282714844,
      "step": 1418
    },
    {
      "epoch": 0.1977286978332056,
      "grad_norm": 0.596738874912262,
      "learning_rate": 3.820383782697136e-05,
      "loss": 0.09536552429199219,
      "step": 1419
    },
    {
      "epoch": 0.19786804152441997,
      "grad_norm": 0.5971279144287109,
      "learning_rate": 3.819993120992653e-05,
      "loss": 0.10203933715820312,
      "step": 1420
    },
    {
      "epoch": 0.19800738521563435,
      "grad_norm": 0.37086930871009827,
      "learning_rate": 3.819602054927249e-05,
      "loss": 0.09574699401855469,
      "step": 1421
    },
    {
      "epoch": 0.19814672890684873,
      "grad_norm": 0.8277623057365417,
      "learning_rate": 3.8192105845878106e-05,
      "loss": 0.11203765869140625,
      "step": 1422
    },
    {
      "epoch": 0.19828607259806313,
      "grad_norm": 0.38132745027542114,
      "learning_rate": 3.818818710061314e-05,
      "loss": 0.08694267272949219,
      "step": 1423
    },
    {
      "epoch": 0.1984254162892775,
      "grad_norm": 0.7666764259338379,
      "learning_rate": 3.818426431434824e-05,
      "loss": 0.13161468505859375,
      "step": 1424
    },
    {
      "epoch": 0.1985647599804919,
      "grad_norm": 0.892369270324707,
      "learning_rate": 3.818033748795497e-05,
      "loss": 0.1602935791015625,
      "step": 1425
    },
    {
      "epoch": 0.19870410367170627,
      "grad_norm": 0.7345067262649536,
      "learning_rate": 3.817640662230576e-05,
      "loss": 0.11780166625976562,
      "step": 1426
    },
    {
      "epoch": 0.19884344736292064,
      "grad_norm": 0.22475594282150269,
      "learning_rate": 3.8172471718273986e-05,
      "loss": 0.06590080261230469,
      "step": 1427
    },
    {
      "epoch": 0.19898279105413502,
      "grad_norm": 0.31588122248649597,
      "learning_rate": 3.8168532776733874e-05,
      "loss": 0.08957290649414062,
      "step": 1428
    },
    {
      "epoch": 0.1991221347453494,
      "grad_norm": 0.40985652804374695,
      "learning_rate": 3.816458979856058e-05,
      "loss": 0.09764671325683594,
      "step": 1429
    },
    {
      "epoch": 0.19926147843656378,
      "grad_norm": 0.5436139702796936,
      "learning_rate": 3.816064278463013e-05,
      "loss": 0.12009620666503906,
      "step": 1430
    },
    {
      "epoch": 0.19940082212777815,
      "grad_norm": 0.6112926602363586,
      "learning_rate": 3.815669173581947e-05,
      "loss": 0.08493614196777344,
      "step": 1431
    },
    {
      "epoch": 0.19954016581899253,
      "grad_norm": 0.8178542852401733,
      "learning_rate": 3.8152736653006434e-05,
      "loss": 0.10950279235839844,
      "step": 1432
    },
    {
      "epoch": 0.19967950951020694,
      "grad_norm": 0.958256721496582,
      "learning_rate": 3.8148777537069745e-05,
      "loss": 0.13604164123535156,
      "step": 1433
    },
    {
      "epoch": 0.19981885320142131,
      "grad_norm": 0.5863037705421448,
      "learning_rate": 3.8144814388889034e-05,
      "loss": 0.10473060607910156,
      "step": 1434
    },
    {
      "epoch": 0.1999581968926357,
      "grad_norm": 0.38497716188430786,
      "learning_rate": 3.814084720934482e-05,
      "loss": 0.08390045166015625,
      "step": 1435
    },
    {
      "epoch": 0.20009754058385007,
      "grad_norm": 1.1983671188354492,
      "learning_rate": 3.813687599931851e-05,
      "loss": 0.10915184020996094,
      "step": 1436
    },
    {
      "epoch": 0.20023688427506445,
      "grad_norm": 0.5354716181755066,
      "learning_rate": 3.813290075969243e-05,
      "loss": 0.08707809448242188,
      "step": 1437
    },
    {
      "epoch": 0.20037622796627882,
      "grad_norm": 0.7907063364982605,
      "learning_rate": 3.812892149134978e-05,
      "loss": 0.09860992431640625,
      "step": 1438
    },
    {
      "epoch": 0.2005155716574932,
      "grad_norm": 1.1451308727264404,
      "learning_rate": 3.812493819517467e-05,
      "loss": 0.1474018096923828,
      "step": 1439
    },
    {
      "epoch": 0.20065491534870758,
      "grad_norm": 0.5149797797203064,
      "learning_rate": 3.812095087205209e-05,
      "loss": 0.09700965881347656,
      "step": 1440
    },
    {
      "epoch": 0.20079425903992196,
      "grad_norm": 0.588095486164093,
      "learning_rate": 3.811695952286793e-05,
      "loss": 0.09161949157714844,
      "step": 1441
    },
    {
      "epoch": 0.20093360273113633,
      "grad_norm": 1.0182271003723145,
      "learning_rate": 3.8112964148508986e-05,
      "loss": 0.1202545166015625,
      "step": 1442
    },
    {
      "epoch": 0.20107294642235074,
      "grad_norm": 0.3629237115383148,
      "learning_rate": 3.810896474986294e-05,
      "loss": 0.07866287231445312,
      "step": 1443
    },
    {
      "epoch": 0.20121229011356512,
      "grad_norm": 0.6830618381500244,
      "learning_rate": 3.8104961327818354e-05,
      "loss": 0.1313323974609375,
      "step": 1444
    },
    {
      "epoch": 0.2013516338047795,
      "grad_norm": 1.473646640777588,
      "learning_rate": 3.8100953883264705e-05,
      "loss": 0.11252784729003906,
      "step": 1445
    },
    {
      "epoch": 0.20149097749599387,
      "grad_norm": 0.8773378133773804,
      "learning_rate": 3.809694241709235e-05,
      "loss": 0.11846733093261719,
      "step": 1446
    },
    {
      "epoch": 0.20163032118720825,
      "grad_norm": 0.29767510294914246,
      "learning_rate": 3.8092926930192555e-05,
      "loss": 0.07369613647460938,
      "step": 1447
    },
    {
      "epoch": 0.20176966487842263,
      "grad_norm": 0.3374917805194855,
      "learning_rate": 3.8088907423457466e-05,
      "loss": 0.06881523132324219,
      "step": 1448
    },
    {
      "epoch": 0.201909008569637,
      "grad_norm": 0.33682623505592346,
      "learning_rate": 3.8084883897780126e-05,
      "loss": 0.06980705261230469,
      "step": 1449
    },
    {
      "epoch": 0.20204835226085138,
      "grad_norm": 0.9696002006530762,
      "learning_rate": 3.808085635405446e-05,
      "loss": 0.09891414642333984,
      "step": 1450
    },
    {
      "epoch": 0.20218769595206576,
      "grad_norm": 0.8802271485328674,
      "learning_rate": 3.807682479317531e-05,
      "loss": 0.10804176330566406,
      "step": 1451
    },
    {
      "epoch": 0.20232703964328014,
      "grad_norm": 0.3295086920261383,
      "learning_rate": 3.80727892160384e-05,
      "loss": 0.09547042846679688,
      "step": 1452
    },
    {
      "epoch": 0.20246638333449454,
      "grad_norm": 0.7114360332489014,
      "learning_rate": 3.806874962354033e-05,
      "loss": 0.13481712341308594,
      "step": 1453
    },
    {
      "epoch": 0.20260572702570892,
      "grad_norm": 0.49520426988601685,
      "learning_rate": 3.806470601657861e-05,
      "loss": 0.09775924682617188,
      "step": 1454
    },
    {
      "epoch": 0.2027450707169233,
      "grad_norm": 0.4204261302947998,
      "learning_rate": 3.806065839605163e-05,
      "loss": 0.08002281188964844,
      "step": 1455
    },
    {
      "epoch": 0.20288441440813768,
      "grad_norm": 0.5384228825569153,
      "learning_rate": 3.805660676285869e-05,
      "loss": 0.09677314758300781,
      "step": 1456
    },
    {
      "epoch": 0.20302375809935205,
      "grad_norm": 0.31039959192276,
      "learning_rate": 3.805255111789997e-05,
      "loss": 0.059444427490234375,
      "step": 1457
    },
    {
      "epoch": 0.20316310179056643,
      "grad_norm": 0.6278024315834045,
      "learning_rate": 3.804849146207654e-05,
      "loss": 0.1160116195678711,
      "step": 1458
    },
    {
      "epoch": 0.2033024454817808,
      "grad_norm": 0.37210944294929504,
      "learning_rate": 3.804442779629035e-05,
      "loss": 0.08463001251220703,
      "step": 1459
    },
    {
      "epoch": 0.20344178917299519,
      "grad_norm": 0.5949891805648804,
      "learning_rate": 3.804036012144428e-05,
      "loss": 0.06990814208984375,
      "step": 1460
    },
    {
      "epoch": 0.20358113286420956,
      "grad_norm": 0.4826754033565521,
      "learning_rate": 3.8036288438442056e-05,
      "loss": 0.09526634216308594,
      "step": 1461
    },
    {
      "epoch": 0.20372047655542394,
      "grad_norm": 0.30970972776412964,
      "learning_rate": 3.8032212748188306e-05,
      "loss": 0.068878173828125,
      "step": 1462
    },
    {
      "epoch": 0.20385982024663835,
      "grad_norm": 0.4353504776954651,
      "learning_rate": 3.802813305158857e-05,
      "loss": 0.08207130432128906,
      "step": 1463
    },
    {
      "epoch": 0.20399916393785272,
      "grad_norm": 0.31890976428985596,
      "learning_rate": 3.802404934954926e-05,
      "loss": 0.06662368774414062,
      "step": 1464
    },
    {
      "epoch": 0.2041385076290671,
      "grad_norm": 0.4211995005607605,
      "learning_rate": 3.801996164297769e-05,
      "loss": 0.11152839660644531,
      "step": 1465
    },
    {
      "epoch": 0.20427785132028148,
      "grad_norm": 1.0828603506088257,
      "learning_rate": 3.8015869932782034e-05,
      "loss": 0.14220237731933594,
      "step": 1466
    },
    {
      "epoch": 0.20441719501149586,
      "grad_norm": 0.4392499625682831,
      "learning_rate": 3.801177421987139e-05,
      "loss": 0.08716583251953125,
      "step": 1467
    },
    {
      "epoch": 0.20455653870271023,
      "grad_norm": 0.5736841559410095,
      "learning_rate": 3.800767450515574e-05,
      "loss": 0.10194778442382812,
      "step": 1468
    },
    {
      "epoch": 0.2046958823939246,
      "grad_norm": 0.7148897647857666,
      "learning_rate": 3.800357078954593e-05,
      "loss": 0.11293697357177734,
      "step": 1469
    },
    {
      "epoch": 0.204835226085139,
      "grad_norm": 0.5559396147727966,
      "learning_rate": 3.7999463073953715e-05,
      "loss": 0.11779403686523438,
      "step": 1470
    },
    {
      "epoch": 0.20497456977635337,
      "grad_norm": 0.5491883754730225,
      "learning_rate": 3.7995351359291743e-05,
      "loss": 0.10757064819335938,
      "step": 1471
    },
    {
      "epoch": 0.20511391346756774,
      "grad_norm": 0.7194819450378418,
      "learning_rate": 3.799123564647354e-05,
      "loss": 0.12995338439941406,
      "step": 1472
    },
    {
      "epoch": 0.20525325715878215,
      "grad_norm": 0.5404316782951355,
      "learning_rate": 3.7987115936413526e-05,
      "loss": 0.11382675170898438,
      "step": 1473
    },
    {
      "epoch": 0.20539260084999653,
      "grad_norm": 0.8655185103416443,
      "learning_rate": 3.7982992230027e-05,
      "loss": 0.11959648132324219,
      "step": 1474
    },
    {
      "epoch": 0.2055319445412109,
      "grad_norm": 0.678997814655304,
      "learning_rate": 3.797886452823016e-05,
      "loss": 0.10365104675292969,
      "step": 1475
    },
    {
      "epoch": 0.20567128823242528,
      "grad_norm": 0.8390398621559143,
      "learning_rate": 3.797473283194009e-05,
      "loss": 0.12501144409179688,
      "step": 1476
    },
    {
      "epoch": 0.20581063192363966,
      "grad_norm": 1.0266169309616089,
      "learning_rate": 3.797059714207475e-05,
      "loss": 0.12553787231445312,
      "step": 1477
    },
    {
      "epoch": 0.20594997561485404,
      "grad_norm": 0.30015701055526733,
      "learning_rate": 3.7966457459553e-05,
      "loss": 0.07081985473632812,
      "step": 1478
    },
    {
      "epoch": 0.2060893193060684,
      "grad_norm": 0.538057267665863,
      "learning_rate": 3.796231378529458e-05,
      "loss": 0.10770225524902344,
      "step": 1479
    },
    {
      "epoch": 0.2062286629972828,
      "grad_norm": 0.4557957053184509,
      "learning_rate": 3.795816612022014e-05,
      "loss": 0.10361862182617188,
      "step": 1480
    },
    {
      "epoch": 0.20636800668849717,
      "grad_norm": 0.5970976948738098,
      "learning_rate": 3.795401446525117e-05,
      "loss": 0.13608360290527344,
      "step": 1481
    },
    {
      "epoch": 0.20650735037971155,
      "grad_norm": 1.1295959949493408,
      "learning_rate": 3.794985882131008e-05,
      "loss": 0.11529922485351562,
      "step": 1482
    },
    {
      "epoch": 0.20664669407092595,
      "grad_norm": 0.8950627446174622,
      "learning_rate": 3.794569918932016e-05,
      "loss": 0.08821678161621094,
      "step": 1483
    },
    {
      "epoch": 0.20678603776214033,
      "grad_norm": 0.955672562122345,
      "learning_rate": 3.79415355702056e-05,
      "loss": 0.10879707336425781,
      "step": 1484
    },
    {
      "epoch": 0.2069253814533547,
      "grad_norm": 0.6705617308616638,
      "learning_rate": 3.793736796489143e-05,
      "loss": 0.10563850402832031,
      "step": 1485
    },
    {
      "epoch": 0.20706472514456908,
      "grad_norm": 0.437788188457489,
      "learning_rate": 3.7933196374303636e-05,
      "loss": 0.08569145202636719,
      "step": 1486
    },
    {
      "epoch": 0.20720406883578346,
      "grad_norm": 0.7959402799606323,
      "learning_rate": 3.792902079936902e-05,
      "loss": 0.08927726745605469,
      "step": 1487
    },
    {
      "epoch": 0.20734341252699784,
      "grad_norm": 0.936450183391571,
      "learning_rate": 3.79248412410153e-05,
      "loss": 0.10674571990966797,
      "step": 1488
    },
    {
      "epoch": 0.20748275621821222,
      "grad_norm": 0.7912521958351135,
      "learning_rate": 3.79206577001711e-05,
      "loss": 0.09006309509277344,
      "step": 1489
    },
    {
      "epoch": 0.2076220999094266,
      "grad_norm": 0.491150438785553,
      "learning_rate": 3.791647017776589e-05,
      "loss": 0.06692314147949219,
      "step": 1490
    },
    {
      "epoch": 0.20776144360064097,
      "grad_norm": 0.40566304326057434,
      "learning_rate": 3.791227867473004e-05,
      "loss": 0.09125900268554688,
      "step": 1491
    },
    {
      "epoch": 0.20790078729185535,
      "grad_norm": 0.3640974462032318,
      "learning_rate": 3.790808319199483e-05,
      "loss": 0.0757293701171875,
      "step": 1492
    },
    {
      "epoch": 0.20804013098306975,
      "grad_norm": 0.3945387899875641,
      "learning_rate": 3.790388373049236e-05,
      "loss": 0.06107330322265625,
      "step": 1493
    },
    {
      "epoch": 0.20817947467428413,
      "grad_norm": 1.162039875984192,
      "learning_rate": 3.78996802911557e-05,
      "loss": 0.12734031677246094,
      "step": 1494
    },
    {
      "epoch": 0.2083188183654985,
      "grad_norm": 0.4436519742012024,
      "learning_rate": 3.789547287491872e-05,
      "loss": 0.06573104858398438,
      "step": 1495
    },
    {
      "epoch": 0.2084581620567129,
      "grad_norm": 0.4242134392261505,
      "learning_rate": 3.789126148271624e-05,
      "loss": 0.088714599609375,
      "step": 1496
    },
    {
      "epoch": 0.20859750574792726,
      "grad_norm": 0.4224456548690796,
      "learning_rate": 3.7887046115483914e-05,
      "loss": 0.0977325439453125,
      "step": 1497
    },
    {
      "epoch": 0.20873684943914164,
      "grad_norm": 0.5047667026519775,
      "learning_rate": 3.788282677415831e-05,
      "loss": 0.07758140563964844,
      "step": 1498
    },
    {
      "epoch": 0.20887619313035602,
      "grad_norm": 0.6682465076446533,
      "learning_rate": 3.787860345967687e-05,
      "loss": 0.09996986389160156,
      "step": 1499
    },
    {
      "epoch": 0.2090155368215704,
      "grad_norm": 0.4682313799858093,
      "learning_rate": 3.787437617297792e-05,
      "loss": 0.113739013671875,
      "step": 1500
    },
    {
      "epoch": 0.20915488051278477,
      "grad_norm": 0.6314793229103088,
      "learning_rate": 3.787014491500066e-05,
      "loss": 0.0919036865234375,
      "step": 1501
    },
    {
      "epoch": 0.20929422420399915,
      "grad_norm": 0.3438890874385834,
      "learning_rate": 3.786590968668518e-05,
      "loss": 0.09236717224121094,
      "step": 1502
    },
    {
      "epoch": 0.20943356789521356,
      "grad_norm": 0.4249441921710968,
      "learning_rate": 3.7861670488972464e-05,
      "loss": 0.08953285217285156,
      "step": 1503
    },
    {
      "epoch": 0.20957291158642793,
      "grad_norm": 0.47448834776878357,
      "learning_rate": 3.7857427322804346e-05,
      "loss": 0.08986759185791016,
      "step": 1504
    },
    {
      "epoch": 0.2097122552776423,
      "grad_norm": 0.5752113461494446,
      "learning_rate": 3.785318018912357e-05,
      "loss": 0.11834144592285156,
      "step": 1505
    },
    {
      "epoch": 0.2098515989688567,
      "grad_norm": 0.8963272571563721,
      "learning_rate": 3.784892908887375e-05,
      "loss": 0.10187339782714844,
      "step": 1506
    },
    {
      "epoch": 0.20999094266007107,
      "grad_norm": 0.9021040797233582,
      "learning_rate": 3.7844674022999387e-05,
      "loss": 0.13229751586914062,
      "step": 1507
    },
    {
      "epoch": 0.21013028635128544,
      "grad_norm": 0.4937811493873596,
      "learning_rate": 3.784041499244585e-05,
      "loss": 0.09146881103515625,
      "step": 1508
    },
    {
      "epoch": 0.21026963004249982,
      "grad_norm": 0.9222509264945984,
      "learning_rate": 3.783615199815941e-05,
      "loss": 0.09781455993652344,
      "step": 1509
    },
    {
      "epoch": 0.2104089737337142,
      "grad_norm": 0.7012776732444763,
      "learning_rate": 3.78318850410872e-05,
      "loss": 0.1532745361328125,
      "step": 1510
    },
    {
      "epoch": 0.21054831742492858,
      "grad_norm": 0.325534850358963,
      "learning_rate": 3.782761412217725e-05,
      "loss": 0.08605194091796875,
      "step": 1511
    },
    {
      "epoch": 0.21068766111614295,
      "grad_norm": 0.6870124936103821,
      "learning_rate": 3.7823339242378445e-05,
      "loss": 0.12457275390625,
      "step": 1512
    },
    {
      "epoch": 0.21082700480735736,
      "grad_norm": 0.43713685870170593,
      "learning_rate": 3.7819060402640577e-05,
      "loss": 0.08042716979980469,
      "step": 1513
    },
    {
      "epoch": 0.21096634849857174,
      "grad_norm": 0.29261136054992676,
      "learning_rate": 3.7814777603914305e-05,
      "loss": 0.07395172119140625,
      "step": 1514
    },
    {
      "epoch": 0.21110569218978611,
      "grad_norm": 0.8282667994499207,
      "learning_rate": 3.781049084715117e-05,
      "loss": 0.09408378601074219,
      "step": 1515
    },
    {
      "epoch": 0.2112450358810005,
      "grad_norm": 0.6816376447677612,
      "learning_rate": 3.780620013330358e-05,
      "loss": 0.14255142211914062,
      "step": 1516
    },
    {
      "epoch": 0.21138437957221487,
      "grad_norm": 0.8154549598693848,
      "learning_rate": 3.7801905463324855e-05,
      "loss": 0.17228317260742188,
      "step": 1517
    },
    {
      "epoch": 0.21152372326342925,
      "grad_norm": 0.551977813243866,
      "learning_rate": 3.7797606838169156e-05,
      "loss": 0.0966033935546875,
      "step": 1518
    },
    {
      "epoch": 0.21166306695464362,
      "grad_norm": 0.600788414478302,
      "learning_rate": 3.7793304258791544e-05,
      "loss": 0.08553504943847656,
      "step": 1519
    },
    {
      "epoch": 0.211802410645858,
      "grad_norm": 0.3054784834384918,
      "learning_rate": 3.778899772614795e-05,
      "loss": 0.06509208679199219,
      "step": 1520
    },
    {
      "epoch": 0.21194175433707238,
      "grad_norm": 0.7253639698028564,
      "learning_rate": 3.7784687241195195e-05,
      "loss": 0.12289810180664062,
      "step": 1521
    },
    {
      "epoch": 0.21208109802828676,
      "grad_norm": 0.7096953988075256,
      "learning_rate": 3.778037280489096e-05,
      "loss": 0.09976005554199219,
      "step": 1522
    },
    {
      "epoch": 0.21222044171950116,
      "grad_norm": 0.348112553358078,
      "learning_rate": 3.777605441819383e-05,
      "loss": 0.09137344360351562,
      "step": 1523
    },
    {
      "epoch": 0.21235978541071554,
      "grad_norm": 0.5054752826690674,
      "learning_rate": 3.777173208206323e-05,
      "loss": 0.06853675842285156,
      "step": 1524
    },
    {
      "epoch": 0.21249912910192992,
      "grad_norm": 0.5703735947608948,
      "learning_rate": 3.776740579745951e-05,
      "loss": 0.12443733215332031,
      "step": 1525
    },
    {
      "epoch": 0.2126384727931443,
      "grad_norm": 0.7497169971466064,
      "learning_rate": 3.776307556534385e-05,
      "loss": 0.09358978271484375,
      "step": 1526
    },
    {
      "epoch": 0.21277781648435867,
      "grad_norm": 0.8233155608177185,
      "learning_rate": 3.775874138667834e-05,
      "loss": 0.1333599090576172,
      "step": 1527
    },
    {
      "epoch": 0.21291716017557305,
      "grad_norm": 0.395746111869812,
      "learning_rate": 3.775440326242593e-05,
      "loss": 0.08567619323730469,
      "step": 1528
    },
    {
      "epoch": 0.21305650386678743,
      "grad_norm": 0.5177441239356995,
      "learning_rate": 3.775006119355047e-05,
      "loss": 0.08508682250976562,
      "step": 1529
    },
    {
      "epoch": 0.2131958475580018,
      "grad_norm": 0.7314834594726562,
      "learning_rate": 3.7745715181016634e-05,
      "loss": 0.12312889099121094,
      "step": 1530
    },
    {
      "epoch": 0.21333519124921618,
      "grad_norm": 0.4594397246837616,
      "learning_rate": 3.774136522579004e-05,
      "loss": 0.08123970031738281,
      "step": 1531
    },
    {
      "epoch": 0.21347453494043056,
      "grad_norm": 0.6907541155815125,
      "learning_rate": 3.773701132883712e-05,
      "loss": 0.1280040740966797,
      "step": 1532
    },
    {
      "epoch": 0.21361387863164497,
      "grad_norm": 0.7328150868415833,
      "learning_rate": 3.773265349112524e-05,
      "loss": 0.10618972778320312,
      "step": 1533
    },
    {
      "epoch": 0.21375322232285934,
      "grad_norm": 0.3529180586338043,
      "learning_rate": 3.772829171362259e-05,
      "loss": 0.07922935485839844,
      "step": 1534
    },
    {
      "epoch": 0.21389256601407372,
      "grad_norm": 0.3982096314430237,
      "learning_rate": 3.772392599729827e-05,
      "loss": 0.09147262573242188,
      "step": 1535
    },
    {
      "epoch": 0.2140319097052881,
      "grad_norm": 0.38216084241867065,
      "learning_rate": 3.7719556343122236e-05,
      "loss": 0.08507156372070312,
      "step": 1536
    },
    {
      "epoch": 0.21417125339650248,
      "grad_norm": 0.5215528607368469,
      "learning_rate": 3.771518275206532e-05,
      "loss": 0.0823822021484375,
      "step": 1537
    },
    {
      "epoch": 0.21431059708771685,
      "grad_norm": 0.38749194145202637,
      "learning_rate": 3.771080522509925e-05,
      "loss": 0.10201835632324219,
      "step": 1538
    },
    {
      "epoch": 0.21444994077893123,
      "grad_norm": 0.46555837988853455,
      "learning_rate": 3.77064237631966e-05,
      "loss": 0.09074020385742188,
      "step": 1539
    },
    {
      "epoch": 0.2145892844701456,
      "grad_norm": 0.47891974449157715,
      "learning_rate": 3.770203836733084e-05,
      "loss": 0.08310127258300781,
      "step": 1540
    },
    {
      "epoch": 0.21472862816135999,
      "grad_norm": 0.43048909306526184,
      "learning_rate": 3.769764903847629e-05,
      "loss": 0.09546852111816406,
      "step": 1541
    },
    {
      "epoch": 0.21486797185257436,
      "grad_norm": 0.7208696603775024,
      "learning_rate": 3.769325577760817e-05,
      "loss": 0.09722328186035156,
      "step": 1542
    },
    {
      "epoch": 0.21500731554378877,
      "grad_norm": 0.4891384243965149,
      "learning_rate": 3.7688858585702564e-05,
      "loss": 0.10654830932617188,
      "step": 1543
    },
    {
      "epoch": 0.21514665923500315,
      "grad_norm": 0.5755232572555542,
      "learning_rate": 3.768445746373642e-05,
      "loss": 0.12471199035644531,
      "step": 1544
    },
    {
      "epoch": 0.21528600292621752,
      "grad_norm": 0.2976595163345337,
      "learning_rate": 3.768005241268757e-05,
      "loss": 0.07563018798828125,
      "step": 1545
    },
    {
      "epoch": 0.2154253466174319,
      "grad_norm": 0.8567584753036499,
      "learning_rate": 3.7675643433534725e-05,
      "loss": 0.12109756469726562,
      "step": 1546
    },
    {
      "epoch": 0.21556469030864628,
      "grad_norm": 0.3276172876358032,
      "learning_rate": 3.767123052725744e-05,
      "loss": 0.07333564758300781,
      "step": 1547
    },
    {
      "epoch": 0.21570403399986066,
      "grad_norm": 0.4382215142250061,
      "learning_rate": 3.7666813694836176e-05,
      "loss": 0.10529899597167969,
      "step": 1548
    },
    {
      "epoch": 0.21584337769107503,
      "grad_norm": 0.369022011756897,
      "learning_rate": 3.7662392937252255e-05,
      "loss": 0.07841300964355469,
      "step": 1549
    },
    {
      "epoch": 0.2159827213822894,
      "grad_norm": 0.8088367581367493,
      "learning_rate": 3.7657968255487854e-05,
      "loss": 0.11362838745117188,
      "step": 1550
    },
    {
      "epoch": 0.2161220650735038,
      "grad_norm": 0.8999560475349426,
      "learning_rate": 3.765353965052605e-05,
      "loss": 0.11891365051269531,
      "step": 1551
    },
    {
      "epoch": 0.21626140876471817,
      "grad_norm": 0.6453534364700317,
      "learning_rate": 3.764910712335077e-05,
      "loss": 0.1423654556274414,
      "step": 1552
    },
    {
      "epoch": 0.21640075245593257,
      "grad_norm": 0.2893756628036499,
      "learning_rate": 3.764467067494683e-05,
      "loss": 0.08387374877929688,
      "step": 1553
    },
    {
      "epoch": 0.21654009614714695,
      "grad_norm": 0.5407136082649231,
      "learning_rate": 3.7640230306299895e-05,
      "loss": 0.09052658081054688,
      "step": 1554
    },
    {
      "epoch": 0.21667943983836133,
      "grad_norm": 0.7949237823486328,
      "learning_rate": 3.7635786018396524e-05,
      "loss": 0.11100196838378906,
      "step": 1555
    },
    {
      "epoch": 0.2168187835295757,
      "grad_norm": 1.0256104469299316,
      "learning_rate": 3.763133781222412e-05,
      "loss": 0.11709213256835938,
      "step": 1556
    },
    {
      "epoch": 0.21695812722079008,
      "grad_norm": 0.6628548502922058,
      "learning_rate": 3.762688568877099e-05,
      "loss": 0.13437652587890625,
      "step": 1557
    },
    {
      "epoch": 0.21709747091200446,
      "grad_norm": 0.3250269889831543,
      "learning_rate": 3.762242964902629e-05,
      "loss": 0.08572006225585938,
      "step": 1558
    },
    {
      "epoch": 0.21723681460321884,
      "grad_norm": 0.7066490650177002,
      "learning_rate": 3.761796969398005e-05,
      "loss": 0.10484790802001953,
      "step": 1559
    },
    {
      "epoch": 0.2173761582944332,
      "grad_norm": 0.4039357602596283,
      "learning_rate": 3.761350582462317e-05,
      "loss": 0.09900856018066406,
      "step": 1560
    },
    {
      "epoch": 0.2175155019856476,
      "grad_norm": 0.7094775438308716,
      "learning_rate": 3.760903804194742e-05,
      "loss": 0.1310443878173828,
      "step": 1561
    },
    {
      "epoch": 0.21765484567686197,
      "grad_norm": 0.798835039138794,
      "learning_rate": 3.7604566346945437e-05,
      "loss": 0.10375595092773438,
      "step": 1562
    },
    {
      "epoch": 0.21779418936807637,
      "grad_norm": 0.8003969192504883,
      "learning_rate": 3.760009074061073e-05,
      "loss": 0.17800521850585938,
      "step": 1563
    },
    {
      "epoch": 0.21793353305929075,
      "grad_norm": 0.5331871509552002,
      "learning_rate": 3.759561122393767e-05,
      "loss": 0.12304496765136719,
      "step": 1564
    },
    {
      "epoch": 0.21807287675050513,
      "grad_norm": 0.45761337876319885,
      "learning_rate": 3.7591127797921523e-05,
      "loss": 0.09998798370361328,
      "step": 1565
    },
    {
      "epoch": 0.2182122204417195,
      "grad_norm": 0.36916062235832214,
      "learning_rate": 3.7586640463558384e-05,
      "loss": 0.10434627532958984,
      "step": 1566
    },
    {
      "epoch": 0.21835156413293388,
      "grad_norm": 0.9214900732040405,
      "learning_rate": 3.758214922184525e-05,
      "loss": 0.1218109130859375,
      "step": 1567
    },
    {
      "epoch": 0.21849090782414826,
      "grad_norm": 0.27142274379730225,
      "learning_rate": 3.7577654073779956e-05,
      "loss": 0.08076286315917969,
      "step": 1568
    },
    {
      "epoch": 0.21863025151536264,
      "grad_norm": 0.41765594482421875,
      "learning_rate": 3.757315502036124e-05,
      "loss": 0.10165786743164062,
      "step": 1569
    },
    {
      "epoch": 0.21876959520657702,
      "grad_norm": 0.4707895517349243,
      "learning_rate": 3.756865206258868e-05,
      "loss": 0.1075286865234375,
      "step": 1570
    },
    {
      "epoch": 0.2189089388977914,
      "grad_norm": 0.6539376974105835,
      "learning_rate": 3.7564145201462726e-05,
      "loss": 0.09635591506958008,
      "step": 1571
    },
    {
      "epoch": 0.21904828258900577,
      "grad_norm": 0.6401981711387634,
      "learning_rate": 3.755963443798471e-05,
      "loss": 0.11151885986328125,
      "step": 1572
    },
    {
      "epoch": 0.21918762628022015,
      "grad_norm": 0.3274432420730591,
      "learning_rate": 3.7555119773156815e-05,
      "loss": 0.07125282287597656,
      "step": 1573
    },
    {
      "epoch": 0.21932696997143455,
      "grad_norm": 0.3339978754520416,
      "learning_rate": 3.755060120798209e-05,
      "loss": 0.09644699096679688,
      "step": 1574
    },
    {
      "epoch": 0.21946631366264893,
      "grad_norm": 0.6201578974723816,
      "learning_rate": 3.754607874346447e-05,
      "loss": 0.09425735473632812,
      "step": 1575
    },
    {
      "epoch": 0.2196056573538633,
      "grad_norm": 0.518813967704773,
      "learning_rate": 3.7541552380608734e-05,
      "loss": 0.10203742980957031,
      "step": 1576
    },
    {
      "epoch": 0.2197450010450777,
      "grad_norm": 0.6740968823432922,
      "learning_rate": 3.753702212042054e-05,
      "loss": 0.14271163940429688,
      "step": 1577
    },
    {
      "epoch": 0.21988434473629206,
      "grad_norm": 0.5926182270050049,
      "learning_rate": 3.753248796390641e-05,
      "loss": 0.1084604263305664,
      "step": 1578
    },
    {
      "epoch": 0.22002368842750644,
      "grad_norm": 0.3018651604652405,
      "learning_rate": 3.7527949912073725e-05,
      "loss": 0.08038520812988281,
      "step": 1579
    },
    {
      "epoch": 0.22016303211872082,
      "grad_norm": 0.5300496220588684,
      "learning_rate": 3.752340796593074e-05,
      "loss": 0.12004852294921875,
      "step": 1580
    },
    {
      "epoch": 0.2203023758099352,
      "grad_norm": 0.5574085116386414,
      "learning_rate": 3.751886212648657e-05,
      "loss": 0.12952613830566406,
      "step": 1581
    },
    {
      "epoch": 0.22044171950114957,
      "grad_norm": 0.9039328694343567,
      "learning_rate": 3.75143123947512e-05,
      "loss": 0.1310100555419922,
      "step": 1582
    },
    {
      "epoch": 0.22058106319236395,
      "grad_norm": 0.317617267370224,
      "learning_rate": 3.7509758771735475e-05,
      "loss": 0.07017707824707031,
      "step": 1583
    },
    {
      "epoch": 0.22072040688357836,
      "grad_norm": 0.21542635560035706,
      "learning_rate": 3.75052012584511e-05,
      "loss": 0.06874847412109375,
      "step": 1584
    },
    {
      "epoch": 0.22085975057479273,
      "grad_norm": 0.5125584602355957,
      "learning_rate": 3.750063985591067e-05,
      "loss": 0.12099838256835938,
      "step": 1585
    },
    {
      "epoch": 0.2209990942660071,
      "grad_norm": 0.49988046288490295,
      "learning_rate": 3.749607456512759e-05,
      "loss": 0.10702323913574219,
      "step": 1586
    },
    {
      "epoch": 0.2211384379572215,
      "grad_norm": 0.5726056694984436,
      "learning_rate": 3.74915053871162e-05,
      "loss": 0.07961177825927734,
      "step": 1587
    },
    {
      "epoch": 0.22127778164843587,
      "grad_norm": 0.34733280539512634,
      "learning_rate": 3.7486932322891646e-05,
      "loss": 0.07981109619140625,
      "step": 1588
    },
    {
      "epoch": 0.22141712533965024,
      "grad_norm": 0.5153899192810059,
      "learning_rate": 3.748235537346996e-05,
      "loss": 0.09248733520507812,
      "step": 1589
    },
    {
      "epoch": 0.22155646903086462,
      "grad_norm": 0.30907776951789856,
      "learning_rate": 3.747777453986804e-05,
      "loss": 0.08660602569580078,
      "step": 1590
    },
    {
      "epoch": 0.221695812722079,
      "grad_norm": 0.6186462640762329,
      "learning_rate": 3.7473189823103645e-05,
      "loss": 0.15837669372558594,
      "step": 1591
    },
    {
      "epoch": 0.22183515641329338,
      "grad_norm": 0.2438896894454956,
      "learning_rate": 3.746860122419539e-05,
      "loss": 0.0655527114868164,
      "step": 1592
    },
    {
      "epoch": 0.22197450010450775,
      "grad_norm": 0.34336549043655396,
      "learning_rate": 3.746400874416276e-05,
      "loss": 0.0745391845703125,
      "step": 1593
    },
    {
      "epoch": 0.22211384379572216,
      "grad_norm": 0.4868924617767334,
      "learning_rate": 3.745941238402609e-05,
      "loss": 0.11593246459960938,
      "step": 1594
    },
    {
      "epoch": 0.22225318748693654,
      "grad_norm": 0.6077494025230408,
      "learning_rate": 3.74548121448066e-05,
      "loss": 0.11920928955078125,
      "step": 1595
    },
    {
      "epoch": 0.22239253117815092,
      "grad_norm": 0.36852696537971497,
      "learning_rate": 3.745020802752635e-05,
      "loss": 0.07178401947021484,
      "step": 1596
    },
    {
      "epoch": 0.2225318748693653,
      "grad_norm": 0.6349460482597351,
      "learning_rate": 3.744560003320827e-05,
      "loss": 0.121734619140625,
      "step": 1597
    },
    {
      "epoch": 0.22267121856057967,
      "grad_norm": 0.27636298537254333,
      "learning_rate": 3.744098816287616e-05,
      "loss": 0.07383441925048828,
      "step": 1598
    },
    {
      "epoch": 0.22281056225179405,
      "grad_norm": 0.7512742877006531,
      "learning_rate": 3.743637241755465e-05,
      "loss": 0.11266708374023438,
      "step": 1599
    },
    {
      "epoch": 0.22294990594300843,
      "grad_norm": 0.47874268889427185,
      "learning_rate": 3.743175279826928e-05,
      "loss": 0.10490226745605469,
      "step": 1600
    },
    {
      "epoch": 0.2230892496342228,
      "grad_norm": 1.0714977979660034,
      "learning_rate": 3.7427129306046406e-05,
      "loss": 0.16431427001953125,
      "step": 1601
    },
    {
      "epoch": 0.22322859332543718,
      "grad_norm": 0.9433776140213013,
      "learning_rate": 3.7422501941913274e-05,
      "loss": 0.15381336212158203,
      "step": 1602
    },
    {
      "epoch": 0.22336793701665156,
      "grad_norm": 0.5465538501739502,
      "learning_rate": 3.7417870706897964e-05,
      "loss": 0.11712074279785156,
      "step": 1603
    },
    {
      "epoch": 0.22350728070786596,
      "grad_norm": 0.6326457262039185,
      "learning_rate": 3.7413235602029445e-05,
      "loss": 0.10058212280273438,
      "step": 1604
    },
    {
      "epoch": 0.22364662439908034,
      "grad_norm": 0.6444109678268433,
      "learning_rate": 3.740859662833753e-05,
      "loss": 0.13872146606445312,
      "step": 1605
    },
    {
      "epoch": 0.22378596809029472,
      "grad_norm": 0.5282261371612549,
      "learning_rate": 3.7403953786852884e-05,
      "loss": 0.1089630126953125,
      "step": 1606
    },
    {
      "epoch": 0.2239253117815091,
      "grad_norm": 0.5039969086647034,
      "learning_rate": 3.739930707860705e-05,
      "loss": 0.09267616271972656,
      "step": 1607
    },
    {
      "epoch": 0.22406465547272347,
      "grad_norm": 0.7581688165664673,
      "learning_rate": 3.739465650463241e-05,
      "loss": 0.13608360290527344,
      "step": 1608
    },
    {
      "epoch": 0.22420399916393785,
      "grad_norm": 0.40179604291915894,
      "learning_rate": 3.739000206596222e-05,
      "loss": 0.06851577758789062,
      "step": 1609
    },
    {
      "epoch": 0.22434334285515223,
      "grad_norm": 0.5212522149085999,
      "learning_rate": 3.7385343763630594e-05,
      "loss": 0.10344123840332031,
      "step": 1610
    },
    {
      "epoch": 0.2244826865463666,
      "grad_norm": 0.6378124356269836,
      "learning_rate": 3.738068159867251e-05,
      "loss": 0.09065818786621094,
      "step": 1611
    },
    {
      "epoch": 0.22462203023758098,
      "grad_norm": 0.27570831775665283,
      "learning_rate": 3.7376015572123766e-05,
      "loss": 0.058109283447265625,
      "step": 1612
    },
    {
      "epoch": 0.22476137392879536,
      "grad_norm": 0.40337181091308594,
      "learning_rate": 3.737134568502107e-05,
      "loss": 0.07692718505859375,
      "step": 1613
    },
    {
      "epoch": 0.22490071762000977,
      "grad_norm": 0.8806300163269043,
      "learning_rate": 3.7366671938401954e-05,
      "loss": 0.12884902954101562,
      "step": 1614
    },
    {
      "epoch": 0.22504006131122414,
      "grad_norm": 0.9037048816680908,
      "learning_rate": 3.736199433330483e-05,
      "loss": 0.10411643981933594,
      "step": 1615
    },
    {
      "epoch": 0.22517940500243852,
      "grad_norm": 0.19295407831668854,
      "learning_rate": 3.735731287076893e-05,
      "loss": 0.05302619934082031,
      "step": 1616
    },
    {
      "epoch": 0.2253187486936529,
      "grad_norm": 0.7655208110809326,
      "learning_rate": 3.73526275518344e-05,
      "loss": 0.11087894439697266,
      "step": 1617
    },
    {
      "epoch": 0.22545809238486728,
      "grad_norm": 0.4710557758808136,
      "learning_rate": 3.734793837754219e-05,
      "loss": 0.09324455261230469,
      "step": 1618
    },
    {
      "epoch": 0.22559743607608165,
      "grad_norm": 0.3750317096710205,
      "learning_rate": 3.734324534893413e-05,
      "loss": 0.08462333679199219,
      "step": 1619
    },
    {
      "epoch": 0.22573677976729603,
      "grad_norm": 0.5708099007606506,
      "learning_rate": 3.733854846705291e-05,
      "loss": 0.08513259887695312,
      "step": 1620
    },
    {
      "epoch": 0.2258761234585104,
      "grad_norm": 0.3139928877353668,
      "learning_rate": 3.733384773294207e-05,
      "loss": 0.08656501770019531,
      "step": 1621
    },
    {
      "epoch": 0.22601546714972479,
      "grad_norm": 0.8058632612228394,
      "learning_rate": 3.7329143147645994e-05,
      "loss": 0.11630439758300781,
      "step": 1622
    },
    {
      "epoch": 0.22615481084093916,
      "grad_norm": 0.5171427726745605,
      "learning_rate": 3.732443471220994e-05,
      "loss": 0.09388160705566406,
      "step": 1623
    },
    {
      "epoch": 0.22629415453215357,
      "grad_norm": 0.8346560597419739,
      "learning_rate": 3.731972242768002e-05,
      "loss": 0.1201171875,
      "step": 1624
    },
    {
      "epoch": 0.22643349822336795,
      "grad_norm": 0.4235908091068268,
      "learning_rate": 3.73150062951032e-05,
      "loss": 0.09433937072753906,
      "step": 1625
    },
    {
      "epoch": 0.22657284191458232,
      "grad_norm": 0.4598454236984253,
      "learning_rate": 3.731028631552728e-05,
      "loss": 0.06681632995605469,
      "step": 1626
    },
    {
      "epoch": 0.2267121856057967,
      "grad_norm": 0.7440026998519897,
      "learning_rate": 3.7305562490000944e-05,
      "loss": 0.10416984558105469,
      "step": 1627
    },
    {
      "epoch": 0.22685152929701108,
      "grad_norm": 0.49010762572288513,
      "learning_rate": 3.730083481957372e-05,
      "loss": 0.07461071014404297,
      "step": 1628
    },
    {
      "epoch": 0.22699087298822546,
      "grad_norm": 1.408625602722168,
      "learning_rate": 3.729610330529598e-05,
      "loss": 0.12647628784179688,
      "step": 1629
    },
    {
      "epoch": 0.22713021667943983,
      "grad_norm": 0.26831820607185364,
      "learning_rate": 3.7291367948218964e-05,
      "loss": 0.06605100631713867,
      "step": 1630
    },
    {
      "epoch": 0.2272695603706542,
      "grad_norm": 0.3455255627632141,
      "learning_rate": 3.7286628749394754e-05,
      "loss": 0.06838417053222656,
      "step": 1631
    },
    {
      "epoch": 0.2274089040618686,
      "grad_norm": 0.8549287915229797,
      "learning_rate": 3.72818857098763e-05,
      "loss": 0.10059928894042969,
      "step": 1632
    },
    {
      "epoch": 0.22754824775308297,
      "grad_norm": 1.1519582271575928,
      "learning_rate": 3.727713883071739e-05,
      "loss": 0.106719970703125,
      "step": 1633
    },
    {
      "epoch": 0.22768759144429737,
      "grad_norm": 1.2846801280975342,
      "learning_rate": 3.727238811297268e-05,
      "loss": 0.16149425506591797,
      "step": 1634
    },
    {
      "epoch": 0.22782693513551175,
      "grad_norm": 0.37317559123039246,
      "learning_rate": 3.7267633557697666e-05,
      "loss": 0.08448028564453125,
      "step": 1635
    },
    {
      "epoch": 0.22796627882672613,
      "grad_norm": 0.994754433631897,
      "learning_rate": 3.72628751659487e-05,
      "loss": 0.1497478485107422,
      "step": 1636
    },
    {
      "epoch": 0.2281056225179405,
      "grad_norm": 0.520328164100647,
      "learning_rate": 3.725811293878299e-05,
      "loss": 0.09092330932617188,
      "step": 1637
    },
    {
      "epoch": 0.22824496620915488,
      "grad_norm": 0.4498412311077118,
      "learning_rate": 3.72533468772586e-05,
      "loss": 0.11457443237304688,
      "step": 1638
    },
    {
      "epoch": 0.22838430990036926,
      "grad_norm": 1.1230679750442505,
      "learning_rate": 3.724857698243443e-05,
      "loss": 0.1331329345703125,
      "step": 1639
    },
    {
      "epoch": 0.22852365359158364,
      "grad_norm": 0.7834765315055847,
      "learning_rate": 3.724380325537024e-05,
      "loss": 0.12019729614257812,
      "step": 1640
    },
    {
      "epoch": 0.22866299728279801,
      "grad_norm": 0.4696687161922455,
      "learning_rate": 3.723902569712666e-05,
      "loss": 0.09833908081054688,
      "step": 1641
    },
    {
      "epoch": 0.2288023409740124,
      "grad_norm": 0.4047260880470276,
      "learning_rate": 3.7234244308765136e-05,
      "loss": 0.10163688659667969,
      "step": 1642
    },
    {
      "epoch": 0.22894168466522677,
      "grad_norm": 0.31232279539108276,
      "learning_rate": 3.7229459091348e-05,
      "loss": 0.06556320190429688,
      "step": 1643
    },
    {
      "epoch": 0.22908102835644117,
      "grad_norm": 0.43527713418006897,
      "learning_rate": 3.7224670045938406e-05,
      "loss": 0.08668136596679688,
      "step": 1644
    },
    {
      "epoch": 0.22922037204765555,
      "grad_norm": 0.5505969524383545,
      "learning_rate": 3.721987717360037e-05,
      "loss": 0.08183765411376953,
      "step": 1645
    },
    {
      "epoch": 0.22935971573886993,
      "grad_norm": 1.3112683296203613,
      "learning_rate": 3.721508047539877e-05,
      "loss": 0.15229225158691406,
      "step": 1646
    },
    {
      "epoch": 0.2294990594300843,
      "grad_norm": 0.725774884223938,
      "learning_rate": 3.72102799523993e-05,
      "loss": 0.10330009460449219,
      "step": 1647
    },
    {
      "epoch": 0.22963840312129868,
      "grad_norm": 1.221149206161499,
      "learning_rate": 3.720547560566855e-05,
      "loss": 0.13100051879882812,
      "step": 1648
    },
    {
      "epoch": 0.22977774681251306,
      "grad_norm": 0.651091456413269,
      "learning_rate": 3.720066743627393e-05,
      "loss": 0.09765815734863281,
      "step": 1649
    },
    {
      "epoch": 0.22991709050372744,
      "grad_norm": 0.5582399368286133,
      "learning_rate": 3.719585544528371e-05,
      "loss": 0.10725593566894531,
      "step": 1650
    },
    {
      "epoch": 0.23005643419494182,
      "grad_norm": 0.3642353415489197,
      "learning_rate": 3.719103963376699e-05,
      "loss": 0.07566070556640625,
      "step": 1651
    },
    {
      "epoch": 0.2301957778861562,
      "grad_norm": 0.9978375434875488,
      "learning_rate": 3.718622000279374e-05,
      "loss": 0.0920867919921875,
      "step": 1652
    },
    {
      "epoch": 0.23033512157737057,
      "grad_norm": 0.8710653781890869,
      "learning_rate": 3.718139655343477e-05,
      "loss": 0.09995841979980469,
      "step": 1653
    },
    {
      "epoch": 0.23047446526858498,
      "grad_norm": 0.6143290400505066,
      "learning_rate": 3.717656928676175e-05,
      "loss": 0.08996772766113281,
      "step": 1654
    },
    {
      "epoch": 0.23061380895979935,
      "grad_norm": 0.30669403076171875,
      "learning_rate": 3.7171738203847185e-05,
      "loss": 0.07484626770019531,
      "step": 1655
    },
    {
      "epoch": 0.23075315265101373,
      "grad_norm": 1.0691511631011963,
      "learning_rate": 3.7166903305764426e-05,
      "loss": 0.15988540649414062,
      "step": 1656
    },
    {
      "epoch": 0.2308924963422281,
      "grad_norm": 1.2117648124694824,
      "learning_rate": 3.716206459358768e-05,
      "loss": 0.09419822692871094,
      "step": 1657
    },
    {
      "epoch": 0.2310318400334425,
      "grad_norm": 0.63372802734375,
      "learning_rate": 3.7157222068392e-05,
      "loss": 0.09680366516113281,
      "step": 1658
    },
    {
      "epoch": 0.23117118372465686,
      "grad_norm": 0.8234993815422058,
      "learning_rate": 3.715237573125328e-05,
      "loss": 0.07779693603515625,
      "step": 1659
    },
    {
      "epoch": 0.23131052741587124,
      "grad_norm": 0.9243489503860474,
      "learning_rate": 3.7147525583248264e-05,
      "loss": 0.14867782592773438,
      "step": 1660
    },
    {
      "epoch": 0.23144987110708562,
      "grad_norm": 0.2812195122241974,
      "learning_rate": 3.714267162545455e-05,
      "loss": 0.0814666748046875,
      "step": 1661
    },
    {
      "epoch": 0.2315892147983,
      "grad_norm": 0.42775958776474,
      "learning_rate": 3.7137813858950576e-05,
      "loss": 0.08056831359863281,
      "step": 1662
    },
    {
      "epoch": 0.23172855848951437,
      "grad_norm": 0.6298354864120483,
      "learning_rate": 3.713295228481563e-05,
      "loss": 0.10430717468261719,
      "step": 1663
    },
    {
      "epoch": 0.23186790218072878,
      "grad_norm": 1.0439720153808594,
      "learning_rate": 3.712808690412983e-05,
      "loss": 0.09417533874511719,
      "step": 1664
    },
    {
      "epoch": 0.23200724587194316,
      "grad_norm": 0.9177694320678711,
      "learning_rate": 3.7123217717974166e-05,
      "loss": 0.11417770385742188,
      "step": 1665
    },
    {
      "epoch": 0.23214658956315753,
      "grad_norm": 0.6626611351966858,
      "learning_rate": 3.711834472743045e-05,
      "loss": 0.10962867736816406,
      "step": 1666
    },
    {
      "epoch": 0.2322859332543719,
      "grad_norm": 0.5949239730834961,
      "learning_rate": 3.7113467933581364e-05,
      "loss": 0.08469581604003906,
      "step": 1667
    },
    {
      "epoch": 0.2324252769455863,
      "grad_norm": 0.9350483417510986,
      "learning_rate": 3.7108587337510405e-05,
      "loss": 0.13500404357910156,
      "step": 1668
    },
    {
      "epoch": 0.23256462063680067,
      "grad_norm": 0.43504223227500916,
      "learning_rate": 3.7103702940301934e-05,
      "loss": 0.08293819427490234,
      "step": 1669
    },
    {
      "epoch": 0.23270396432801504,
      "grad_norm": 0.43150565028190613,
      "learning_rate": 3.709881474304115e-05,
      "loss": 0.05774497985839844,
      "step": 1670
    },
    {
      "epoch": 0.23284330801922942,
      "grad_norm": 0.5862683653831482,
      "learning_rate": 3.7093922746814104e-05,
      "loss": 0.09478378295898438,
      "step": 1671
    },
    {
      "epoch": 0.2329826517104438,
      "grad_norm": 0.41796886920928955,
      "learning_rate": 3.7089026952707695e-05,
      "loss": 0.06665515899658203,
      "step": 1672
    },
    {
      "epoch": 0.23312199540165818,
      "grad_norm": 0.7624088525772095,
      "learning_rate": 3.7084127361809636e-05,
      "loss": 0.18622589111328125,
      "step": 1673
    },
    {
      "epoch": 0.23326133909287258,
      "grad_norm": 0.45114749670028687,
      "learning_rate": 3.707922397520852e-05,
      "loss": 0.07932472229003906,
      "step": 1674
    },
    {
      "epoch": 0.23340068278408696,
      "grad_norm": 0.6560578346252441,
      "learning_rate": 3.707431679399375e-05,
      "loss": 0.11545562744140625,
      "step": 1675
    },
    {
      "epoch": 0.23354002647530134,
      "grad_norm": 0.5277032852172852,
      "learning_rate": 3.7069405819255615e-05,
      "loss": 0.1023702621459961,
      "step": 1676
    },
    {
      "epoch": 0.23367937016651572,
      "grad_norm": 0.47121530771255493,
      "learning_rate": 3.706449105208521e-05,
      "loss": 0.09769821166992188,
      "step": 1677
    },
    {
      "epoch": 0.2338187138577301,
      "grad_norm": 0.6647270917892456,
      "learning_rate": 3.705957249357447e-05,
      "loss": 0.10782814025878906,
      "step": 1678
    },
    {
      "epoch": 0.23395805754894447,
      "grad_norm": 0.5387539863586426,
      "learning_rate": 3.70546501448162e-05,
      "loss": 0.11089038848876953,
      "step": 1679
    },
    {
      "epoch": 0.23409740124015885,
      "grad_norm": 0.6609600782394409,
      "learning_rate": 3.704972400690404e-05,
      "loss": 0.11353492736816406,
      "step": 1680
    },
    {
      "epoch": 0.23423674493137323,
      "grad_norm": 0.5125299096107483,
      "learning_rate": 3.704479408093245e-05,
      "loss": 0.12015151977539062,
      "step": 1681
    },
    {
      "epoch": 0.2343760886225876,
      "grad_norm": 0.3488992154598236,
      "learning_rate": 3.703986036799676e-05,
      "loss": 0.08428764343261719,
      "step": 1682
    },
    {
      "epoch": 0.23451543231380198,
      "grad_norm": 0.33674168586730957,
      "learning_rate": 3.7034922869193125e-05,
      "loss": 0.0851583480834961,
      "step": 1683
    },
    {
      "epoch": 0.23465477600501639,
      "grad_norm": 0.9553260207176208,
      "learning_rate": 3.702998158561854e-05,
      "loss": 0.11856651306152344,
      "step": 1684
    },
    {
      "epoch": 0.23479411969623076,
      "grad_norm": 0.35411953926086426,
      "learning_rate": 3.7025036518370846e-05,
      "loss": 0.09574508666992188,
      "step": 1685
    },
    {
      "epoch": 0.23493346338744514,
      "grad_norm": 0.6003347039222717,
      "learning_rate": 3.7020087668548725e-05,
      "loss": 0.10250091552734375,
      "step": 1686
    },
    {
      "epoch": 0.23507280707865952,
      "grad_norm": 0.44941961765289307,
      "learning_rate": 3.7015135037251696e-05,
      "loss": 0.11246109008789062,
      "step": 1687
    },
    {
      "epoch": 0.2352121507698739,
      "grad_norm": 0.7339297533035278,
      "learning_rate": 3.7010178625580134e-05,
      "loss": 0.09427452087402344,
      "step": 1688
    },
    {
      "epoch": 0.23535149446108827,
      "grad_norm": 0.7725930213928223,
      "learning_rate": 3.700521843463522e-05,
      "loss": 0.10149765014648438,
      "step": 1689
    },
    {
      "epoch": 0.23549083815230265,
      "grad_norm": 0.6277204155921936,
      "learning_rate": 3.7000254465519014e-05,
      "loss": 0.12342453002929688,
      "step": 1690
    },
    {
      "epoch": 0.23563018184351703,
      "grad_norm": 0.8732476830482483,
      "learning_rate": 3.6995286719334385e-05,
      "loss": 0.12891197204589844,
      "step": 1691
    },
    {
      "epoch": 0.2357695255347314,
      "grad_norm": 0.5717354416847229,
      "learning_rate": 3.6990315197185054e-05,
      "loss": 0.09356117248535156,
      "step": 1692
    },
    {
      "epoch": 0.23590886922594578,
      "grad_norm": 0.6999872922897339,
      "learning_rate": 3.6985339900175584e-05,
      "loss": 0.1067495346069336,
      "step": 1693
    },
    {
      "epoch": 0.2360482129171602,
      "grad_norm": 0.8686468005180359,
      "learning_rate": 3.698036082941137e-05,
      "loss": 0.11345100402832031,
      "step": 1694
    },
    {
      "epoch": 0.23618755660837457,
      "grad_norm": 0.7100614905357361,
      "learning_rate": 3.6975377985998645e-05,
      "loss": 0.10672760009765625,
      "step": 1695
    },
    {
      "epoch": 0.23632690029958894,
      "grad_norm": 0.9861154556274414,
      "learning_rate": 3.697039137104449e-05,
      "loss": 0.10477638244628906,
      "step": 1696
    },
    {
      "epoch": 0.23646624399080332,
      "grad_norm": 1.8330574035644531,
      "learning_rate": 3.696540098565681e-05,
      "loss": 0.14670944213867188,
      "step": 1697
    },
    {
      "epoch": 0.2366055876820177,
      "grad_norm": 0.48330506682395935,
      "learning_rate": 3.696040683094436e-05,
      "loss": 0.1071624755859375,
      "step": 1698
    },
    {
      "epoch": 0.23674493137323208,
      "grad_norm": 0.93628990650177,
      "learning_rate": 3.695540890801672e-05,
      "loss": 0.1279582977294922,
      "step": 1699
    },
    {
      "epoch": 0.23688427506444645,
      "grad_norm": 0.6754939556121826,
      "learning_rate": 3.6950407217984326e-05,
      "loss": 0.07445335388183594,
      "step": 1700
    },
    {
      "epoch": 0.23702361875566083,
      "grad_norm": 1.127946376800537,
      "learning_rate": 3.694540176195843e-05,
      "loss": 0.11805343627929688,
      "step": 1701
    },
    {
      "epoch": 0.2371629624468752,
      "grad_norm": 0.5287918448448181,
      "learning_rate": 3.694039254105113e-05,
      "loss": 0.0856943130493164,
      "step": 1702
    },
    {
      "epoch": 0.23730230613808959,
      "grad_norm": 0.4176771640777588,
      "learning_rate": 3.693537955637537e-05,
      "loss": 0.09712791442871094,
      "step": 1703
    },
    {
      "epoch": 0.237441649829304,
      "grad_norm": 0.418195903301239,
      "learning_rate": 3.6930362809044906e-05,
      "loss": 0.10551071166992188,
      "step": 1704
    },
    {
      "epoch": 0.23758099352051837,
      "grad_norm": 0.3771388828754425,
      "learning_rate": 3.692534230017436e-05,
      "loss": 0.1008758544921875,
      "step": 1705
    },
    {
      "epoch": 0.23772033721173275,
      "grad_norm": 1.5284042358398438,
      "learning_rate": 3.692031803087916e-05,
      "loss": 0.11960411071777344,
      "step": 1706
    },
    {
      "epoch": 0.23785968090294712,
      "grad_norm": 0.515853762626648,
      "learning_rate": 3.691529000227559e-05,
      "loss": 0.11124038696289062,
      "step": 1707
    },
    {
      "epoch": 0.2379990245941615,
      "grad_norm": 0.6337981224060059,
      "learning_rate": 3.691025821548077e-05,
      "loss": 0.10515499114990234,
      "step": 1708
    },
    {
      "epoch": 0.23813836828537588,
      "grad_norm": 0.6571255922317505,
      "learning_rate": 3.6905222671612634e-05,
      "loss": 0.11510848999023438,
      "step": 1709
    },
    {
      "epoch": 0.23827771197659026,
      "grad_norm": 0.4082661271095276,
      "learning_rate": 3.6900183371789984e-05,
      "loss": 0.08769607543945312,
      "step": 1710
    },
    {
      "epoch": 0.23841705566780463,
      "grad_norm": 0.4568444490432739,
      "learning_rate": 3.689514031713242e-05,
      "loss": 0.07398223876953125,
      "step": 1711
    },
    {
      "epoch": 0.238556399359019,
      "grad_norm": 0.41724255681037903,
      "learning_rate": 3.68900935087604e-05,
      "loss": 0.08727788925170898,
      "step": 1712
    },
    {
      "epoch": 0.2386957430502334,
      "grad_norm": 0.29547029733657837,
      "learning_rate": 3.688504294779521e-05,
      "loss": 0.08379745483398438,
      "step": 1713
    },
    {
      "epoch": 0.2388350867414478,
      "grad_norm": 0.36507725715637207,
      "learning_rate": 3.687998863535897e-05,
      "loss": 0.093017578125,
      "step": 1714
    },
    {
      "epoch": 0.23897443043266217,
      "grad_norm": 0.9347231388092041,
      "learning_rate": 3.687493057257464e-05,
      "loss": 0.10626411437988281,
      "step": 1715
    },
    {
      "epoch": 0.23911377412387655,
      "grad_norm": 0.30378079414367676,
      "learning_rate": 3.686986876056599e-05,
      "loss": 0.07088088989257812,
      "step": 1716
    },
    {
      "epoch": 0.23925311781509093,
      "grad_norm": 0.40904057025909424,
      "learning_rate": 3.6864803200457646e-05,
      "loss": 0.08650398254394531,
      "step": 1717
    },
    {
      "epoch": 0.2393924615063053,
      "grad_norm": 0.5277330279350281,
      "learning_rate": 3.685973389337506e-05,
      "loss": 0.10448455810546875,
      "step": 1718
    },
    {
      "epoch": 0.23953180519751968,
      "grad_norm": 0.7666447162628174,
      "learning_rate": 3.6854660840444524e-05,
      "loss": 0.11308670043945312,
      "step": 1719
    },
    {
      "epoch": 0.23967114888873406,
      "grad_norm": 0.959380030632019,
      "learning_rate": 3.6849584042793145e-05,
      "loss": 0.10850334167480469,
      "step": 1720
    },
    {
      "epoch": 0.23981049257994844,
      "grad_norm": 0.3588741719722748,
      "learning_rate": 3.6844503501548866e-05,
      "loss": 0.06542205810546875,
      "step": 1721
    },
    {
      "epoch": 0.23994983627116281,
      "grad_norm": 0.32009726762771606,
      "learning_rate": 3.6839419217840486e-05,
      "loss": 0.06527137756347656,
      "step": 1722
    },
    {
      "epoch": 0.2400891799623772,
      "grad_norm": 1.0041871070861816,
      "learning_rate": 3.6834331192797606e-05,
      "loss": 0.1396961212158203,
      "step": 1723
    },
    {
      "epoch": 0.2402285236535916,
      "grad_norm": 0.7159838080406189,
      "learning_rate": 3.682923942755066e-05,
      "loss": 0.10334014892578125,
      "step": 1724
    },
    {
      "epoch": 0.24036786734480597,
      "grad_norm": 0.5270859003067017,
      "learning_rate": 3.6824143923230936e-05,
      "loss": 0.11982345581054688,
      "step": 1725
    },
    {
      "epoch": 0.24050721103602035,
      "grad_norm": 0.3235551118850708,
      "learning_rate": 3.681904468097054e-05,
      "loss": 0.08557510375976562,
      "step": 1726
    },
    {
      "epoch": 0.24064655472723473,
      "grad_norm": 0.3953123688697815,
      "learning_rate": 3.681394170190239e-05,
      "loss": 0.08114433288574219,
      "step": 1727
    },
    {
      "epoch": 0.2407858984184491,
      "grad_norm": 0.7354132533073425,
      "learning_rate": 3.6808834987160276e-05,
      "loss": 0.09092140197753906,
      "step": 1728
    },
    {
      "epoch": 0.24092524210966348,
      "grad_norm": 0.40588700771331787,
      "learning_rate": 3.680372453787877e-05,
      "loss": 0.09728336334228516,
      "step": 1729
    },
    {
      "epoch": 0.24106458580087786,
      "grad_norm": 0.314870148897171,
      "learning_rate": 3.679861035519331e-05,
      "loss": 0.08397102355957031,
      "step": 1730
    },
    {
      "epoch": 0.24120392949209224,
      "grad_norm": 0.7342545986175537,
      "learning_rate": 3.679349244024015e-05,
      "loss": 0.09243202209472656,
      "step": 1731
    },
    {
      "epoch": 0.24134327318330662,
      "grad_norm": 1.2203713655471802,
      "learning_rate": 3.6788370794156366e-05,
      "loss": 0.14073753356933594,
      "step": 1732
    },
    {
      "epoch": 0.241482616874521,
      "grad_norm": 0.954064667224884,
      "learning_rate": 3.678324541807988e-05,
      "loss": 0.140106201171875,
      "step": 1733
    },
    {
      "epoch": 0.2416219605657354,
      "grad_norm": 0.49176448583602905,
      "learning_rate": 3.677811631314943e-05,
      "loss": 0.10079383850097656,
      "step": 1734
    },
    {
      "epoch": 0.24176130425694978,
      "grad_norm": 0.4475259482860565,
      "learning_rate": 3.677298348050459e-05,
      "loss": 0.09651851654052734,
      "step": 1735
    },
    {
      "epoch": 0.24190064794816415,
      "grad_norm": 0.616462767124176,
      "learning_rate": 3.676784692128575e-05,
      "loss": 0.09023666381835938,
      "step": 1736
    },
    {
      "epoch": 0.24203999163937853,
      "grad_norm": 0.7639544606208801,
      "learning_rate": 3.676270663663414e-05,
      "loss": 0.09742546081542969,
      "step": 1737
    },
    {
      "epoch": 0.2421793353305929,
      "grad_norm": 0.5675891041755676,
      "learning_rate": 3.675756262769182e-05,
      "loss": 0.07899665832519531,
      "step": 1738
    },
    {
      "epoch": 0.2423186790218073,
      "grad_norm": 0.9190603494644165,
      "learning_rate": 3.6752414895601656e-05,
      "loss": 0.11309814453125,
      "step": 1739
    },
    {
      "epoch": 0.24245802271302166,
      "grad_norm": 0.39199456572532654,
      "learning_rate": 3.674726344150737e-05,
      "loss": 0.0918426513671875,
      "step": 1740
    },
    {
      "epoch": 0.24259736640423604,
      "grad_norm": 0.7018279433250427,
      "learning_rate": 3.67421082665535e-05,
      "loss": 0.12424945831298828,
      "step": 1741
    },
    {
      "epoch": 0.24273671009545042,
      "grad_norm": 0.6182534694671631,
      "learning_rate": 3.6736949371885395e-05,
      "loss": 0.08600616455078125,
      "step": 1742
    },
    {
      "epoch": 0.2428760537866648,
      "grad_norm": 0.6947794556617737,
      "learning_rate": 3.6731786758649255e-05,
      "loss": 0.11729240417480469,
      "step": 1743
    },
    {
      "epoch": 0.2430153974778792,
      "grad_norm": 1.1020642518997192,
      "learning_rate": 3.672662042799209e-05,
      "loss": 0.15318679809570312,
      "step": 1744
    },
    {
      "epoch": 0.24315474116909358,
      "grad_norm": 0.41621577739715576,
      "learning_rate": 3.672145038106174e-05,
      "loss": 0.06282329559326172,
      "step": 1745
    },
    {
      "epoch": 0.24329408486030796,
      "grad_norm": 0.9790002107620239,
      "learning_rate": 3.6716276619006874e-05,
      "loss": 0.10218238830566406,
      "step": 1746
    },
    {
      "epoch": 0.24343342855152234,
      "grad_norm": 0.5793771147727966,
      "learning_rate": 3.671109914297698e-05,
      "loss": 0.08719062805175781,
      "step": 1747
    },
    {
      "epoch": 0.2435727722427367,
      "grad_norm": 0.39084383845329285,
      "learning_rate": 3.670591795412238e-05,
      "loss": 0.08290958404541016,
      "step": 1748
    },
    {
      "epoch": 0.2437121159339511,
      "grad_norm": 0.6570208072662354,
      "learning_rate": 3.670073305359421e-05,
      "loss": 0.09575653076171875,
      "step": 1749
    },
    {
      "epoch": 0.24385145962516547,
      "grad_norm": 0.9504016637802124,
      "learning_rate": 3.669554444254444e-05,
      "loss": 0.08401107788085938,
      "step": 1750
    },
    {
      "epoch": 0.24399080331637985,
      "grad_norm": 1.0595790147781372,
      "learning_rate": 3.6690352122125867e-05,
      "loss": 0.1108551025390625,
      "step": 1751
    },
    {
      "epoch": 0.24413014700759422,
      "grad_norm": 0.9910376071929932,
      "learning_rate": 3.668515609349209e-05,
      "loss": 0.13036346435546875,
      "step": 1752
    },
    {
      "epoch": 0.2442694906988086,
      "grad_norm": 0.5814268589019775,
      "learning_rate": 3.667995635779756e-05,
      "loss": 0.11112785339355469,
      "step": 1753
    },
    {
      "epoch": 0.244408834390023,
      "grad_norm": 0.5021295547485352,
      "learning_rate": 3.667475291619754e-05,
      "loss": 0.0863189697265625,
      "step": 1754
    },
    {
      "epoch": 0.24454817808123738,
      "grad_norm": 0.46807432174682617,
      "learning_rate": 3.6669545769848115e-05,
      "loss": 0.10615730285644531,
      "step": 1755
    },
    {
      "epoch": 0.24468752177245176,
      "grad_norm": 0.5841719508171082,
      "learning_rate": 3.666433491990619e-05,
      "loss": 0.08770942687988281,
      "step": 1756
    },
    {
      "epoch": 0.24482686546366614,
      "grad_norm": 0.7530962228775024,
      "learning_rate": 3.66591203675295e-05,
      "loss": 0.0893707275390625,
      "step": 1757
    },
    {
      "epoch": 0.24496620915488052,
      "grad_norm": 0.8663804531097412,
      "learning_rate": 3.665390211387659e-05,
      "loss": 0.09127616882324219,
      "step": 1758
    },
    {
      "epoch": 0.2451055528460949,
      "grad_norm": 0.6627533435821533,
      "learning_rate": 3.664868016010686e-05,
      "loss": 0.07779502868652344,
      "step": 1759
    },
    {
      "epoch": 0.24524489653730927,
      "grad_norm": 0.9758980870246887,
      "learning_rate": 3.664345450738048e-05,
      "loss": 0.1236114501953125,
      "step": 1760
    },
    {
      "epoch": 0.24538424022852365,
      "grad_norm": 0.49459898471832275,
      "learning_rate": 3.6638225156858494e-05,
      "loss": 0.11063766479492188,
      "step": 1761
    },
    {
      "epoch": 0.24552358391973803,
      "grad_norm": 0.4179801046848297,
      "learning_rate": 3.663299210970273e-05,
      "loss": 0.10082244873046875,
      "step": 1762
    },
    {
      "epoch": 0.2456629276109524,
      "grad_norm": 0.9350152611732483,
      "learning_rate": 3.662775536707586e-05,
      "loss": 0.1270885467529297,
      "step": 1763
    },
    {
      "epoch": 0.2458022713021668,
      "grad_norm": 0.6197609305381775,
      "learning_rate": 3.662251493014137e-05,
      "loss": 0.07829666137695312,
      "step": 1764
    },
    {
      "epoch": 0.24594161499338119,
      "grad_norm": 0.8590683341026306,
      "learning_rate": 3.661727080006356e-05,
      "loss": 0.09772300720214844,
      "step": 1765
    },
    {
      "epoch": 0.24608095868459556,
      "grad_norm": 0.7772808074951172,
      "learning_rate": 3.6612022978007546e-05,
      "loss": 0.11162948608398438,
      "step": 1766
    },
    {
      "epoch": 0.24622030237580994,
      "grad_norm": 0.6488801836967468,
      "learning_rate": 3.66067714651393e-05,
      "loss": 0.1325855255126953,
      "step": 1767
    },
    {
      "epoch": 0.24635964606702432,
      "grad_norm": 0.35372066497802734,
      "learning_rate": 3.6601516262625556e-05,
      "loss": 0.07141876220703125,
      "step": 1768
    },
    {
      "epoch": 0.2464989897582387,
      "grad_norm": 0.9762297868728638,
      "learning_rate": 3.6596257371633926e-05,
      "loss": 0.10553169250488281,
      "step": 1769
    },
    {
      "epoch": 0.24663833344945307,
      "grad_norm": 0.7614570260047913,
      "learning_rate": 3.659099479333281e-05,
      "loss": 0.11890792846679688,
      "step": 1770
    },
    {
      "epoch": 0.24677767714066745,
      "grad_norm": 0.699006199836731,
      "learning_rate": 3.658572852889143e-05,
      "loss": 0.10899066925048828,
      "step": 1771
    },
    {
      "epoch": 0.24691702083188183,
      "grad_norm": 0.6220634579658508,
      "learning_rate": 3.658045857947983e-05,
      "loss": 0.10459518432617188,
      "step": 1772
    },
    {
      "epoch": 0.2470563645230962,
      "grad_norm": 1.01050865650177,
      "learning_rate": 3.657518494626887e-05,
      "loss": 0.13403701782226562,
      "step": 1773
    },
    {
      "epoch": 0.24719570821431058,
      "grad_norm": 0.4430351257324219,
      "learning_rate": 3.6569907630430234e-05,
      "loss": 0.102386474609375,
      "step": 1774
    },
    {
      "epoch": 0.247335051905525,
      "grad_norm": 0.29094040393829346,
      "learning_rate": 3.6564626633136416e-05,
      "loss": 0.0772705078125,
      "step": 1775
    },
    {
      "epoch": 0.24747439559673937,
      "grad_norm": 0.4308490455150604,
      "learning_rate": 3.6559341955560744e-05,
      "loss": 0.08751487731933594,
      "step": 1776
    },
    {
      "epoch": 0.24761373928795374,
      "grad_norm": 0.9212446212768555,
      "learning_rate": 3.655405359887734e-05,
      "loss": 0.126312255859375,
      "step": 1777
    },
    {
      "epoch": 0.24775308297916812,
      "grad_norm": 0.8355321288108826,
      "learning_rate": 3.654876156426116e-05,
      "loss": 0.11662101745605469,
      "step": 1778
    },
    {
      "epoch": 0.2478924266703825,
      "grad_norm": 0.4076789617538452,
      "learning_rate": 3.6543465852887975e-05,
      "loss": 0.11370086669921875,
      "step": 1779
    },
    {
      "epoch": 0.24803177036159688,
      "grad_norm": 0.6194537281990051,
      "learning_rate": 3.6538166465934375e-05,
      "loss": 0.1217193603515625,
      "step": 1780
    },
    {
      "epoch": 0.24817111405281125,
      "grad_norm": 0.678620457649231,
      "learning_rate": 3.653286340457776e-05,
      "loss": 0.11448097229003906,
      "step": 1781
    },
    {
      "epoch": 0.24831045774402563,
      "grad_norm": 0.3153510093688965,
      "learning_rate": 3.6527556669996345e-05,
      "loss": 0.07822227478027344,
      "step": 1782
    },
    {
      "epoch": 0.24844980143524,
      "grad_norm": 0.8410484790802002,
      "learning_rate": 3.6522246263369174e-05,
      "loss": 0.11531257629394531,
      "step": 1783
    },
    {
      "epoch": 0.24858914512645439,
      "grad_norm": 0.5270363092422485,
      "learning_rate": 3.6516932185876085e-05,
      "loss": 0.08694648742675781,
      "step": 1784
    },
    {
      "epoch": 0.2487284888176688,
      "grad_norm": 0.7318198084831238,
      "learning_rate": 3.651161443869776e-05,
      "loss": 0.10180282592773438,
      "step": 1785
    },
    {
      "epoch": 0.24886783250888317,
      "grad_norm": 0.5747535228729248,
      "learning_rate": 3.650629302301567e-05,
      "loss": 0.11678314208984375,
      "step": 1786
    },
    {
      "epoch": 0.24900717620009755,
      "grad_norm": 0.7830781936645508,
      "learning_rate": 3.650096794001211e-05,
      "loss": 0.14970016479492188,
      "step": 1787
    },
    {
      "epoch": 0.24914651989131192,
      "grad_norm": 1.100631594657898,
      "learning_rate": 3.6495639190870204e-05,
      "loss": 0.15678787231445312,
      "step": 1788
    },
    {
      "epoch": 0.2492858635825263,
      "grad_norm": 0.4166242182254791,
      "learning_rate": 3.649030677677387e-05,
      "loss": 0.08962821960449219,
      "step": 1789
    },
    {
      "epoch": 0.24942520727374068,
      "grad_norm": 0.4716280400753021,
      "learning_rate": 3.648497069890785e-05,
      "loss": 0.09626197814941406,
      "step": 1790
    },
    {
      "epoch": 0.24956455096495506,
      "grad_norm": 0.5687309503555298,
      "learning_rate": 3.6479630958457696e-05,
      "loss": 0.08848762512207031,
      "step": 1791
    },
    {
      "epoch": 0.24970389465616943,
      "grad_norm": 0.787609338760376,
      "learning_rate": 3.647428755660978e-05,
      "loss": 0.11913871765136719,
      "step": 1792
    },
    {
      "epoch": 0.2498432383473838,
      "grad_norm": 0.6540242433547974,
      "learning_rate": 3.646894049455129e-05,
      "loss": 0.10382843017578125,
      "step": 1793
    },
    {
      "epoch": 0.2499825820385982,
      "grad_norm": 0.22238531708717346,
      "learning_rate": 3.646358977347021e-05,
      "loss": 0.06094551086425781,
      "step": 1794
    },
    {
      "epoch": 0.25012192572981257,
      "grad_norm": 0.48430097103118896,
      "learning_rate": 3.6458235394555345e-05,
      "loss": 0.07753944396972656,
      "step": 1795
    },
    {
      "epoch": 0.25026126942102694,
      "grad_norm": 0.497452974319458,
      "learning_rate": 3.6452877358996336e-05,
      "loss": 0.09486007690429688,
      "step": 1796
    },
    {
      "epoch": 0.2504006131122413,
      "grad_norm": 0.6213709115982056,
      "learning_rate": 3.64475156679836e-05,
      "loss": 0.09573936462402344,
      "step": 1797
    },
    {
      "epoch": 0.2505399568034557,
      "grad_norm": 1.040291428565979,
      "learning_rate": 3.644215032270838e-05,
      "loss": 0.1268463134765625,
      "step": 1798
    },
    {
      "epoch": 0.2506793004946701,
      "grad_norm": 0.47205644845962524,
      "learning_rate": 3.643678132436274e-05,
      "loss": 0.09928512573242188,
      "step": 1799
    },
    {
      "epoch": 0.2508186441858845,
      "grad_norm": 0.32311198115348816,
      "learning_rate": 3.643140867413956e-05,
      "loss": 0.08672714233398438,
      "step": 1800
    },
    {
      "epoch": 0.2509579878770989,
      "grad_norm": 0.2704096734523773,
      "learning_rate": 3.642603237323249e-05,
      "loss": 0.08147811889648438,
      "step": 1801
    },
    {
      "epoch": 0.25109733156831326,
      "grad_norm": 0.5769910216331482,
      "learning_rate": 3.6420652422836046e-05,
      "loss": 0.09059906005859375,
      "step": 1802
    },
    {
      "epoch": 0.25123667525952764,
      "grad_norm": 0.7045993208885193,
      "learning_rate": 3.641526882414553e-05,
      "loss": 0.1567707061767578,
      "step": 1803
    },
    {
      "epoch": 0.251376018950742,
      "grad_norm": 0.27969956398010254,
      "learning_rate": 3.640988157835704e-05,
      "loss": 0.06602287292480469,
      "step": 1804
    },
    {
      "epoch": 0.2515153626419564,
      "grad_norm": 0.4284462034702301,
      "learning_rate": 3.640449068666751e-05,
      "loss": 0.09215736389160156,
      "step": 1805
    },
    {
      "epoch": 0.2516547063331708,
      "grad_norm": 0.5937240123748779,
      "learning_rate": 3.639909615027468e-05,
      "loss": 0.09843826293945312,
      "step": 1806
    },
    {
      "epoch": 0.25179405002438515,
      "grad_norm": 0.28131163120269775,
      "learning_rate": 3.6393697970377074e-05,
      "loss": 0.07300758361816406,
      "step": 1807
    },
    {
      "epoch": 0.25193339371559953,
      "grad_norm": 0.37684229016304016,
      "learning_rate": 3.638829614817405e-05,
      "loss": 0.08911418914794922,
      "step": 1808
    },
    {
      "epoch": 0.2520727374068139,
      "grad_norm": 0.4463822841644287,
      "learning_rate": 3.638289068486577e-05,
      "loss": 0.11150932312011719,
      "step": 1809
    },
    {
      "epoch": 0.2522120810980283,
      "grad_norm": 1.3823834657669067,
      "learning_rate": 3.6377481581653225e-05,
      "loss": 0.1859283447265625,
      "step": 1810
    },
    {
      "epoch": 0.25235142478924266,
      "grad_norm": 0.5966286659240723,
      "learning_rate": 3.637206883973816e-05,
      "loss": 0.163330078125,
      "step": 1811
    },
    {
      "epoch": 0.25249076848045704,
      "grad_norm": 0.4158175587654114,
      "learning_rate": 3.6366652460323186e-05,
      "loss": 0.09689521789550781,
      "step": 1812
    },
    {
      "epoch": 0.2526301121716714,
      "grad_norm": 0.4966145157814026,
      "learning_rate": 3.6361232444611695e-05,
      "loss": 0.08305549621582031,
      "step": 1813
    },
    {
      "epoch": 0.2527694558628858,
      "grad_norm": 0.4486822187900543,
      "learning_rate": 3.635580879380788e-05,
      "loss": 0.11853408813476562,
      "step": 1814
    },
    {
      "epoch": 0.25290879955410017,
      "grad_norm": 0.36528071761131287,
      "learning_rate": 3.635038150911677e-05,
      "loss": 0.09031486511230469,
      "step": 1815
    },
    {
      "epoch": 0.25304814324531455,
      "grad_norm": 0.7639957666397095,
      "learning_rate": 3.634495059174417e-05,
      "loss": 0.14404869079589844,
      "step": 1816
    },
    {
      "epoch": 0.2531874869365289,
      "grad_norm": 0.8938929438591003,
      "learning_rate": 3.633951604289671e-05,
      "loss": 0.14026641845703125,
      "step": 1817
    },
    {
      "epoch": 0.2533268306277433,
      "grad_norm": 0.5533015131950378,
      "learning_rate": 3.633407786378182e-05,
      "loss": 0.08431243896484375,
      "step": 1818
    },
    {
      "epoch": 0.2534661743189577,
      "grad_norm": 0.9017908573150635,
      "learning_rate": 3.632863605560775e-05,
      "loss": 0.10787582397460938,
      "step": 1819
    },
    {
      "epoch": 0.2536055180101721,
      "grad_norm": 0.8543798923492432,
      "learning_rate": 3.632319061958353e-05,
      "loss": 0.14708709716796875,
      "step": 1820
    },
    {
      "epoch": 0.2537448617013865,
      "grad_norm": 0.5239341259002686,
      "learning_rate": 3.631774155691902e-05,
      "loss": 0.1273975372314453,
      "step": 1821
    },
    {
      "epoch": 0.25388420539260087,
      "grad_norm": 0.6726720929145813,
      "learning_rate": 3.631228886882488e-05,
      "loss": 0.10326766967773438,
      "step": 1822
    },
    {
      "epoch": 0.25402354908381525,
      "grad_norm": 0.6107078790664673,
      "learning_rate": 3.630683255651256e-05,
      "loss": 0.06628799438476562,
      "step": 1823
    },
    {
      "epoch": 0.2541628927750296,
      "grad_norm": 0.7050951719284058,
      "learning_rate": 3.6301372621194354e-05,
      "loss": 0.09975624084472656,
      "step": 1824
    },
    {
      "epoch": 0.254302236466244,
      "grad_norm": 0.8883379101753235,
      "learning_rate": 3.6295909064083305e-05,
      "loss": 0.08823585510253906,
      "step": 1825
    },
    {
      "epoch": 0.2544415801574584,
      "grad_norm": 1.1860942840576172,
      "learning_rate": 3.629044188639331e-05,
      "loss": 0.12037277221679688,
      "step": 1826
    },
    {
      "epoch": 0.25458092384867276,
      "grad_norm": 0.3693036735057831,
      "learning_rate": 3.628497108933904e-05,
      "loss": 0.09196090698242188,
      "step": 1827
    },
    {
      "epoch": 0.25472026753988714,
      "grad_norm": 0.6446684002876282,
      "learning_rate": 3.6279496674135985e-05,
      "loss": 0.1194305419921875,
      "step": 1828
    },
    {
      "epoch": 0.2548596112311015,
      "grad_norm": 0.7448377013206482,
      "learning_rate": 3.6274018642000445e-05,
      "loss": 0.0910491943359375,
      "step": 1829
    },
    {
      "epoch": 0.2549989549223159,
      "grad_norm": 0.8398482799530029,
      "learning_rate": 3.62685369941495e-05,
      "loss": 0.0888671875,
      "step": 1830
    },
    {
      "epoch": 0.25513829861353027,
      "grad_norm": 0.4471357464790344,
      "learning_rate": 3.626305173180105e-05,
      "loss": 0.1293811798095703,
      "step": 1831
    },
    {
      "epoch": 0.25527764230474465,
      "grad_norm": 0.5265861749649048,
      "learning_rate": 3.62575628561738e-05,
      "loss": 0.09464073181152344,
      "step": 1832
    },
    {
      "epoch": 0.255416985995959,
      "grad_norm": 0.3279283940792084,
      "learning_rate": 3.6252070368487246e-05,
      "loss": 0.0825953483581543,
      "step": 1833
    },
    {
      "epoch": 0.2555563296871734,
      "grad_norm": 0.40153518319129944,
      "learning_rate": 3.62465742699617e-05,
      "loss": 0.09640884399414062,
      "step": 1834
    },
    {
      "epoch": 0.2556956733783878,
      "grad_norm": 0.6762891411781311,
      "learning_rate": 3.624107456181826e-05,
      "loss": 0.13918304443359375,
      "step": 1835
    },
    {
      "epoch": 0.25583501706960216,
      "grad_norm": 0.3646334707736969,
      "learning_rate": 3.623557124527885e-05,
      "loss": 0.09185600280761719,
      "step": 1836
    },
    {
      "epoch": 0.25597436076081653,
      "grad_norm": 0.38965800404548645,
      "learning_rate": 3.623006432156618e-05,
      "loss": 0.09678077697753906,
      "step": 1837
    },
    {
      "epoch": 0.2561137044520309,
      "grad_norm": 0.4134117364883423,
      "learning_rate": 3.6224553791903744e-05,
      "loss": 0.10531234741210938,
      "step": 1838
    },
    {
      "epoch": 0.2562530481432453,
      "grad_norm": 0.5846555233001709,
      "learning_rate": 3.621903965751588e-05,
      "loss": 0.0870981216430664,
      "step": 1839
    },
    {
      "epoch": 0.2563923918344597,
      "grad_norm": 0.8281081318855286,
      "learning_rate": 3.6213521919627675e-05,
      "loss": 0.1316070556640625,
      "step": 1840
    },
    {
      "epoch": 0.2565317355256741,
      "grad_norm": 0.49357369542121887,
      "learning_rate": 3.6208000579465075e-05,
      "loss": 0.08634757995605469,
      "step": 1841
    },
    {
      "epoch": 0.2566710792168885,
      "grad_norm": 0.4211686849594116,
      "learning_rate": 3.620247563825477e-05,
      "loss": 0.09738731384277344,
      "step": 1842
    },
    {
      "epoch": 0.25681042290810285,
      "grad_norm": 0.4803521931171417,
      "learning_rate": 3.619694709722429e-05,
      "loss": 0.06947898864746094,
      "step": 1843
    },
    {
      "epoch": 0.25694976659931723,
      "grad_norm": 0.5710007548332214,
      "learning_rate": 3.619141495760196e-05,
      "loss": 0.08675956726074219,
      "step": 1844
    },
    {
      "epoch": 0.2570891102905316,
      "grad_norm": 1.355593204498291,
      "learning_rate": 3.618587922061687e-05,
      "loss": 0.11847305297851562,
      "step": 1845
    },
    {
      "epoch": 0.257228453981746,
      "grad_norm": 0.44880548119544983,
      "learning_rate": 3.6180339887498953e-05,
      "loss": 0.09994697570800781,
      "step": 1846
    },
    {
      "epoch": 0.25736779767296036,
      "grad_norm": 0.7390772700309753,
      "learning_rate": 3.617479695947891e-05,
      "loss": 0.09039878845214844,
      "step": 1847
    },
    {
      "epoch": 0.25750714136417474,
      "grad_norm": 0.5073873996734619,
      "learning_rate": 3.616925043778826e-05,
      "loss": 0.09923648834228516,
      "step": 1848
    },
    {
      "epoch": 0.2576464850553891,
      "grad_norm": 0.5397263169288635,
      "learning_rate": 3.6163700323659327e-05,
      "loss": 0.10794830322265625,
      "step": 1849
    },
    {
      "epoch": 0.2577858287466035,
      "grad_norm": 0.4900830388069153,
      "learning_rate": 3.615814661832519e-05,
      "loss": 0.1214599609375,
      "step": 1850
    },
    {
      "epoch": 0.2579251724378179,
      "grad_norm": 0.5768001675605774,
      "learning_rate": 3.6152589323019775e-05,
      "loss": 0.0909423828125,
      "step": 1851
    },
    {
      "epoch": 0.25806451612903225,
      "grad_norm": 0.5526692867279053,
      "learning_rate": 3.614702843897779e-05,
      "loss": 0.08661842346191406,
      "step": 1852
    },
    {
      "epoch": 0.25820385982024663,
      "grad_norm": 0.38397684693336487,
      "learning_rate": 3.6141463967434715e-05,
      "loss": 0.08755874633789062,
      "step": 1853
    },
    {
      "epoch": 0.258343203511461,
      "grad_norm": 0.6625437140464783,
      "learning_rate": 3.613589590962687e-05,
      "loss": 0.13588714599609375,
      "step": 1854
    },
    {
      "epoch": 0.2584825472026754,
      "grad_norm": 0.3794143795967102,
      "learning_rate": 3.6130324266791344e-05,
      "loss": 0.07888603210449219,
      "step": 1855
    },
    {
      "epoch": 0.25862189089388976,
      "grad_norm": 0.544284999370575,
      "learning_rate": 3.612474904016602e-05,
      "loss": 0.11255073547363281,
      "step": 1856
    },
    {
      "epoch": 0.25876123458510414,
      "grad_norm": 0.5632256269454956,
      "learning_rate": 3.61191702309896e-05,
      "loss": 0.09134483337402344,
      "step": 1857
    },
    {
      "epoch": 0.2589005782763185,
      "grad_norm": 0.35321030020713806,
      "learning_rate": 3.611358784050157e-05,
      "loss": 0.08018302917480469,
      "step": 1858
    },
    {
      "epoch": 0.2590399219675329,
      "grad_norm": 0.5794420838356018,
      "learning_rate": 3.610800186994219e-05,
      "loss": 0.10342025756835938,
      "step": 1859
    },
    {
      "epoch": 0.2591792656587473,
      "grad_norm": 0.5234116911888123,
      "learning_rate": 3.6102412320552546e-05,
      "loss": 0.11651039123535156,
      "step": 1860
    },
    {
      "epoch": 0.2593186093499617,
      "grad_norm": 0.6101605892181396,
      "learning_rate": 3.609681919357451e-05,
      "loss": 0.09016799926757812,
      "step": 1861
    },
    {
      "epoch": 0.2594579530411761,
      "grad_norm": 0.46730512380599976,
      "learning_rate": 3.609122249025075e-05,
      "loss": 0.09749794006347656,
      "step": 1862
    },
    {
      "epoch": 0.25959729673239046,
      "grad_norm": 0.2993644177913666,
      "learning_rate": 3.608562221182472e-05,
      "loss": 0.08300399780273438,
      "step": 1863
    },
    {
      "epoch": 0.25973664042360484,
      "grad_norm": 0.32802021503448486,
      "learning_rate": 3.608001835954067e-05,
      "loss": 0.09124946594238281,
      "step": 1864
    },
    {
      "epoch": 0.2598759841148192,
      "grad_norm": 0.4572458565235138,
      "learning_rate": 3.607441093464366e-05,
      "loss": 0.08585739135742188,
      "step": 1865
    },
    {
      "epoch": 0.2600153278060336,
      "grad_norm": 0.411652147769928,
      "learning_rate": 3.606879993837952e-05,
      "loss": 0.08230209350585938,
      "step": 1866
    },
    {
      "epoch": 0.26015467149724797,
      "grad_norm": 0.397482305765152,
      "learning_rate": 3.60631853719949e-05,
      "loss": 0.06582069396972656,
      "step": 1867
    },
    {
      "epoch": 0.26029401518846235,
      "grad_norm": 0.7697508335113525,
      "learning_rate": 3.6057567236737206e-05,
      "loss": 0.09256935119628906,
      "step": 1868
    },
    {
      "epoch": 0.2604333588796767,
      "grad_norm": 0.34800007939338684,
      "learning_rate": 3.605194553385468e-05,
      "loss": 0.08139228820800781,
      "step": 1869
    },
    {
      "epoch": 0.2605727025708911,
      "grad_norm": 0.4215603172779083,
      "learning_rate": 3.6046320264596324e-05,
      "loss": 0.08046340942382812,
      "step": 1870
    },
    {
      "epoch": 0.2607120462621055,
      "grad_norm": 1.389743447303772,
      "learning_rate": 3.6040691430211955e-05,
      "loss": 0.1567974090576172,
      "step": 1871
    },
    {
      "epoch": 0.26085138995331986,
      "grad_norm": 0.303472638130188,
      "learning_rate": 3.603505903195217e-05,
      "loss": 0.07074832916259766,
      "step": 1872
    },
    {
      "epoch": 0.26099073364453423,
      "grad_norm": 0.5202268958091736,
      "learning_rate": 3.602942307106834e-05,
      "loss": 0.09238624572753906,
      "step": 1873
    },
    {
      "epoch": 0.2611300773357486,
      "grad_norm": 0.9400713443756104,
      "learning_rate": 3.602378354881267e-05,
      "loss": 0.1450824737548828,
      "step": 1874
    },
    {
      "epoch": 0.261269421026963,
      "grad_norm": 0.4340181350708008,
      "learning_rate": 3.601814046643813e-05,
      "loss": 0.09440040588378906,
      "step": 1875
    },
    {
      "epoch": 0.26140876471817737,
      "grad_norm": 0.5446973443031311,
      "learning_rate": 3.6012493825198466e-05,
      "loss": 0.14310836791992188,
      "step": 1876
    },
    {
      "epoch": 0.26154810840939174,
      "grad_norm": 0.8607577085494995,
      "learning_rate": 3.600684362634826e-05,
      "loss": 0.134735107421875,
      "step": 1877
    },
    {
      "epoch": 0.2616874521006061,
      "grad_norm": 0.28477218747138977,
      "learning_rate": 3.600118987114283e-05,
      "loss": 0.08120393753051758,
      "step": 1878
    },
    {
      "epoch": 0.2618267957918205,
      "grad_norm": 1.1631417274475098,
      "learning_rate": 3.599553256083833e-05,
      "loss": 0.1341075897216797,
      "step": 1879
    },
    {
      "epoch": 0.26196613948303493,
      "grad_norm": 0.4370209574699402,
      "learning_rate": 3.598987169669168e-05,
      "loss": 0.08785724639892578,
      "step": 1880
    },
    {
      "epoch": 0.2621054831742493,
      "grad_norm": 0.5547659993171692,
      "learning_rate": 3.598420727996059e-05,
      "loss": 0.1258544921875,
      "step": 1881
    },
    {
      "epoch": 0.2622448268654637,
      "grad_norm": 0.395129919052124,
      "learning_rate": 3.597853931190357e-05,
      "loss": 0.07674217224121094,
      "step": 1882
    },
    {
      "epoch": 0.26238417055667806,
      "grad_norm": 0.5151911973953247,
      "learning_rate": 3.597286779377991e-05,
      "loss": 0.09795475006103516,
      "step": 1883
    },
    {
      "epoch": 0.26252351424789244,
      "grad_norm": 0.3010560870170593,
      "learning_rate": 3.5967192726849694e-05,
      "loss": 0.07511711120605469,
      "step": 1884
    },
    {
      "epoch": 0.2626628579391068,
      "grad_norm": 0.26032501459121704,
      "learning_rate": 3.59615141123738e-05,
      "loss": 0.06875133514404297,
      "step": 1885
    },
    {
      "epoch": 0.2628022016303212,
      "grad_norm": 0.3008735179901123,
      "learning_rate": 3.5955831951613866e-05,
      "loss": 0.07623672485351562,
      "step": 1886
    },
    {
      "epoch": 0.2629415453215356,
      "grad_norm": 0.5896360874176025,
      "learning_rate": 3.595014624583235e-05,
      "loss": 0.096923828125,
      "step": 1887
    },
    {
      "epoch": 0.26308088901274995,
      "grad_norm": 0.5309996008872986,
      "learning_rate": 3.5944456996292486e-05,
      "loss": 0.08529853820800781,
      "step": 1888
    },
    {
      "epoch": 0.26322023270396433,
      "grad_norm": 0.2979205548763275,
      "learning_rate": 3.5938764204258306e-05,
      "loss": 0.07126903533935547,
      "step": 1889
    },
    {
      "epoch": 0.2633595763951787,
      "grad_norm": 0.3421195149421692,
      "learning_rate": 3.59330678709946e-05,
      "loss": 0.08875370025634766,
      "step": 1890
    },
    {
      "epoch": 0.2634989200863931,
      "grad_norm": 0.3483293354511261,
      "learning_rate": 3.5927367997766974e-05,
      "loss": 0.07441329956054688,
      "step": 1891
    },
    {
      "epoch": 0.26363826377760746,
      "grad_norm": 0.38319557905197144,
      "learning_rate": 3.592166458584181e-05,
      "loss": 0.06830978393554688,
      "step": 1892
    },
    {
      "epoch": 0.26377760746882184,
      "grad_norm": 0.2799524962902069,
      "learning_rate": 3.591595763648626e-05,
      "loss": 0.08103752136230469,
      "step": 1893
    },
    {
      "epoch": 0.2639169511600362,
      "grad_norm": 0.3626091480255127,
      "learning_rate": 3.59102471509683e-05,
      "loss": 0.06842613220214844,
      "step": 1894
    },
    {
      "epoch": 0.2640562948512506,
      "grad_norm": 0.3763192296028137,
      "learning_rate": 3.590453313055666e-05,
      "loss": 0.09151077270507812,
      "step": 1895
    },
    {
      "epoch": 0.264195638542465,
      "grad_norm": 0.4718570113182068,
      "learning_rate": 3.589881557652087e-05,
      "loss": 0.11451530456542969,
      "step": 1896
    },
    {
      "epoch": 0.26433498223367935,
      "grad_norm": 0.7787143588066101,
      "learning_rate": 3.5893094490131224e-05,
      "loss": 0.14797306060791016,
      "step": 1897
    },
    {
      "epoch": 0.2644743259248937,
      "grad_norm": 0.5604667067527771,
      "learning_rate": 3.588736987265884e-05,
      "loss": 0.1374530792236328,
      "step": 1898
    },
    {
      "epoch": 0.2646136696161081,
      "grad_norm": 0.912446141242981,
      "learning_rate": 3.588164172537557e-05,
      "loss": 0.16248512268066406,
      "step": 1899
    },
    {
      "epoch": 0.26475301330732254,
      "grad_norm": 0.32521966099739075,
      "learning_rate": 3.58759100495541e-05,
      "loss": 0.06964683532714844,
      "step": 1900
    },
    {
      "epoch": 0.2648923569985369,
      "grad_norm": 0.6443588137626648,
      "learning_rate": 3.587017484646787e-05,
      "loss": 0.1448040008544922,
      "step": 1901
    },
    {
      "epoch": 0.2650317006897513,
      "grad_norm": 0.49545955657958984,
      "learning_rate": 3.586443611739111e-05,
      "loss": 0.1255340576171875,
      "step": 1902
    },
    {
      "epoch": 0.26517104438096567,
      "grad_norm": 0.5963937640190125,
      "learning_rate": 3.585869386359884e-05,
      "loss": 0.09311866760253906,
      "step": 1903
    },
    {
      "epoch": 0.26531038807218005,
      "grad_norm": 0.5033314824104309,
      "learning_rate": 3.5852948086366855e-05,
      "loss": 0.09543323516845703,
      "step": 1904
    },
    {
      "epoch": 0.2654497317633944,
      "grad_norm": 0.3215762674808502,
      "learning_rate": 3.584719878697173e-05,
      "loss": 0.07210159301757812,
      "step": 1905
    },
    {
      "epoch": 0.2655890754546088,
      "grad_norm": 0.5398496389389038,
      "learning_rate": 3.5841445966690834e-05,
      "loss": 0.10020923614501953,
      "step": 1906
    },
    {
      "epoch": 0.2657284191458232,
      "grad_norm": 0.4440378248691559,
      "learning_rate": 3.583568962680231e-05,
      "loss": 0.10395622253417969,
      "step": 1907
    },
    {
      "epoch": 0.26586776283703756,
      "grad_norm": 0.36795106530189514,
      "learning_rate": 3.5829929768585086e-05,
      "loss": 0.09747314453125,
      "step": 1908
    },
    {
      "epoch": 0.26600710652825194,
      "grad_norm": 0.3598285913467407,
      "learning_rate": 3.582416639331886e-05,
      "loss": 0.09177112579345703,
      "step": 1909
    },
    {
      "epoch": 0.2661464502194663,
      "grad_norm": 0.639314591884613,
      "learning_rate": 3.5818399502284154e-05,
      "loss": 0.14349937438964844,
      "step": 1910
    },
    {
      "epoch": 0.2662857939106807,
      "grad_norm": 0.6710983514785767,
      "learning_rate": 3.581262909676221e-05,
      "loss": 0.12166118621826172,
      "step": 1911
    },
    {
      "epoch": 0.26642513760189507,
      "grad_norm": 0.3551844656467438,
      "learning_rate": 3.5806855178035085e-05,
      "loss": 0.07610893249511719,
      "step": 1912
    },
    {
      "epoch": 0.26656448129310945,
      "grad_norm": 0.3430061936378479,
      "learning_rate": 3.580107774738562e-05,
      "loss": 0.06243133544921875,
      "step": 1913
    },
    {
      "epoch": 0.2667038249843238,
      "grad_norm": 0.37956351041793823,
      "learning_rate": 3.579529680609742e-05,
      "loss": 0.0810842514038086,
      "step": 1914
    },
    {
      "epoch": 0.2668431686755382,
      "grad_norm": 0.6682817339897156,
      "learning_rate": 3.578951235545489e-05,
      "loss": 0.1505908966064453,
      "step": 1915
    },
    {
      "epoch": 0.2669825123667526,
      "grad_norm": 0.7081629633903503,
      "learning_rate": 3.578372439674319e-05,
      "loss": 0.13405227661132812,
      "step": 1916
    },
    {
      "epoch": 0.26712185605796696,
      "grad_norm": 0.9852317571640015,
      "learning_rate": 3.577793293124828e-05,
      "loss": 0.14369964599609375,
      "step": 1917
    },
    {
      "epoch": 0.26726119974918133,
      "grad_norm": 0.4601384997367859,
      "learning_rate": 3.577213796025689e-05,
      "loss": 0.080413818359375,
      "step": 1918
    },
    {
      "epoch": 0.2674005434403957,
      "grad_norm": 0.3979266285896301,
      "learning_rate": 3.5766339485056524e-05,
      "loss": 0.07751178741455078,
      "step": 1919
    },
    {
      "epoch": 0.26753988713161014,
      "grad_norm": 0.4357164800167084,
      "learning_rate": 3.5760537506935475e-05,
      "loss": 0.11382484436035156,
      "step": 1920
    },
    {
      "epoch": 0.2676792308228245,
      "grad_norm": 0.4433880150318146,
      "learning_rate": 3.575473202718282e-05,
      "loss": 0.09043121337890625,
      "step": 1921
    },
    {
      "epoch": 0.2678185745140389,
      "grad_norm": 0.32676956057548523,
      "learning_rate": 3.574892304708839e-05,
      "loss": 0.09177398681640625,
      "step": 1922
    },
    {
      "epoch": 0.2679579182052533,
      "grad_norm": 0.6375782489776611,
      "learning_rate": 3.5743110567942815e-05,
      "loss": 0.10851478576660156,
      "step": 1923
    },
    {
      "epoch": 0.26809726189646765,
      "grad_norm": 0.8322335481643677,
      "learning_rate": 3.573729459103749e-05,
      "loss": 0.12631607055664062,
      "step": 1924
    },
    {
      "epoch": 0.26823660558768203,
      "grad_norm": 0.286443293094635,
      "learning_rate": 3.573147511766459e-05,
      "loss": 0.08636665344238281,
      "step": 1925
    },
    {
      "epoch": 0.2683759492788964,
      "grad_norm": 0.3363337218761444,
      "learning_rate": 3.5725652149117085e-05,
      "loss": 0.08887481689453125,
      "step": 1926
    },
    {
      "epoch": 0.2685152929701108,
      "grad_norm": 0.41697755455970764,
      "learning_rate": 3.571982568668869e-05,
      "loss": 0.08847975730895996,
      "step": 1927
    },
    {
      "epoch": 0.26865463666132516,
      "grad_norm": 0.47883492708206177,
      "learning_rate": 3.571399573167392e-05,
      "loss": 0.09887313842773438,
      "step": 1928
    },
    {
      "epoch": 0.26879398035253954,
      "grad_norm": 0.33878427743911743,
      "learning_rate": 3.570816228536806e-05,
      "loss": 0.07528114318847656,
      "step": 1929
    },
    {
      "epoch": 0.2689333240437539,
      "grad_norm": 0.5932261943817139,
      "learning_rate": 3.570232534906716e-05,
      "loss": 0.0956277847290039,
      "step": 1930
    },
    {
      "epoch": 0.2690726677349683,
      "grad_norm": 1.261786699295044,
      "learning_rate": 3.569648492406805e-05,
      "loss": 0.14262104034423828,
      "step": 1931
    },
    {
      "epoch": 0.2692120114261827,
      "grad_norm": 0.8415667414665222,
      "learning_rate": 3.569064101166835e-05,
      "loss": 0.12322330474853516,
      "step": 1932
    },
    {
      "epoch": 0.26935135511739705,
      "grad_norm": 0.6066407561302185,
      "learning_rate": 3.568479361316644e-05,
      "loss": 0.10774803161621094,
      "step": 1933
    },
    {
      "epoch": 0.26949069880861143,
      "grad_norm": 0.44882458448410034,
      "learning_rate": 3.567894272986149e-05,
      "loss": 0.08070564270019531,
      "step": 1934
    },
    {
      "epoch": 0.2696300424998258,
      "grad_norm": 0.3789900243282318,
      "learning_rate": 3.567308836305341e-05,
      "loss": 0.09707069396972656,
      "step": 1935
    },
    {
      "epoch": 0.2697693861910402,
      "grad_norm": 0.3694375455379486,
      "learning_rate": 3.566723051404292e-05,
      "loss": 0.09214401245117188,
      "step": 1936
    },
    {
      "epoch": 0.26990872988225456,
      "grad_norm": 0.5905842781066895,
      "learning_rate": 3.56613691841315e-05,
      "loss": 0.11587715148925781,
      "step": 1937
    },
    {
      "epoch": 0.27004807357346894,
      "grad_norm": 0.5479986071586609,
      "learning_rate": 3.5655504374621404e-05,
      "loss": 0.11419105529785156,
      "step": 1938
    },
    {
      "epoch": 0.2701874172646833,
      "grad_norm": 0.28680363297462463,
      "learning_rate": 3.5649636086815656e-05,
      "loss": 0.07530784606933594,
      "step": 1939
    },
    {
      "epoch": 0.27032676095589775,
      "grad_norm": 0.5132629871368408,
      "learning_rate": 3.5643764322018054e-05,
      "loss": 0.1015777587890625,
      "step": 1940
    },
    {
      "epoch": 0.2704661046471121,
      "grad_norm": 0.5591539740562439,
      "learning_rate": 3.563788908153317e-05,
      "loss": 0.14066696166992188,
      "step": 1941
    },
    {
      "epoch": 0.2706054483383265,
      "grad_norm": 0.18093343079090118,
      "learning_rate": 3.563201036666636e-05,
      "loss": 0.06077384948730469,
      "step": 1942
    },
    {
      "epoch": 0.2707447920295409,
      "grad_norm": 0.36041295528411865,
      "learning_rate": 3.562612817872373e-05,
      "loss": 0.10089111328125,
      "step": 1943
    },
    {
      "epoch": 0.27088413572075526,
      "grad_norm": 0.5118692517280579,
      "learning_rate": 3.5620242519012164e-05,
      "loss": 0.10852622985839844,
      "step": 1944
    },
    {
      "epoch": 0.27102347941196964,
      "grad_norm": 0.4373393654823303,
      "learning_rate": 3.561435338883933e-05,
      "loss": 0.09810447692871094,
      "step": 1945
    },
    {
      "epoch": 0.271162823103184,
      "grad_norm": 0.37534528970718384,
      "learning_rate": 3.560846078951366e-05,
      "loss": 0.10475921630859375,
      "step": 1946
    },
    {
      "epoch": 0.2713021667943984,
      "grad_norm": 1.006746530532837,
      "learning_rate": 3.560256472234434e-05,
      "loss": 0.12412261962890625,
      "step": 1947
    },
    {
      "epoch": 0.27144151048561277,
      "grad_norm": 0.4274523854255676,
      "learning_rate": 3.559666518864136e-05,
      "loss": 0.101806640625,
      "step": 1948
    },
    {
      "epoch": 0.27158085417682715,
      "grad_norm": 0.595859944820404,
      "learning_rate": 3.5590762189715445e-05,
      "loss": 0.1027994155883789,
      "step": 1949
    },
    {
      "epoch": 0.2717201978680415,
      "grad_norm": 0.8337340354919434,
      "learning_rate": 3.558485572687812e-05,
      "loss": 0.11211013793945312,
      "step": 1950
    },
    {
      "epoch": 0.2718595415592559,
      "grad_norm": 0.5119706392288208,
      "learning_rate": 3.557894580144166e-05,
      "loss": 0.07381820678710938,
      "step": 1951
    },
    {
      "epoch": 0.2719988852504703,
      "grad_norm": 0.8109081387519836,
      "learning_rate": 3.5573032414719116e-05,
      "loss": 0.1616191864013672,
      "step": 1952
    },
    {
      "epoch": 0.27213822894168466,
      "grad_norm": 0.7485285401344299,
      "learning_rate": 3.556711556802431e-05,
      "loss": 0.11441993713378906,
      "step": 1953
    },
    {
      "epoch": 0.27227757263289903,
      "grad_norm": 0.40780994296073914,
      "learning_rate": 3.556119526267182e-05,
      "loss": 0.08874130249023438,
      "step": 1954
    },
    {
      "epoch": 0.2724169163241134,
      "grad_norm": 0.3430754542350769,
      "learning_rate": 3.5555271499977015e-05,
      "loss": 0.07614707946777344,
      "step": 1955
    },
    {
      "epoch": 0.2725562600153278,
      "grad_norm": 0.28607431054115295,
      "learning_rate": 3.554934428125602e-05,
      "loss": 0.08019256591796875,
      "step": 1956
    },
    {
      "epoch": 0.27269560370654217,
      "grad_norm": 0.4054071307182312,
      "learning_rate": 3.554341360782572e-05,
      "loss": 0.10265922546386719,
      "step": 1957
    },
    {
      "epoch": 0.27283494739775654,
      "grad_norm": 0.5661808848381042,
      "learning_rate": 3.553747948100378e-05,
      "loss": 0.1094202995300293,
      "step": 1958
    },
    {
      "epoch": 0.2729742910889709,
      "grad_norm": 0.40176090598106384,
      "learning_rate": 3.5531541902108624e-05,
      "loss": 0.06313610076904297,
      "step": 1959
    },
    {
      "epoch": 0.27311363478018535,
      "grad_norm": 0.9042211771011353,
      "learning_rate": 3.5525600872459444e-05,
      "loss": 0.140472412109375,
      "step": 1960
    },
    {
      "epoch": 0.27325297847139973,
      "grad_norm": 0.30686596035957336,
      "learning_rate": 3.551965639337621e-05,
      "loss": 0.0782928466796875,
      "step": 1961
    },
    {
      "epoch": 0.2733923221626141,
      "grad_norm": 0.7576156854629517,
      "learning_rate": 3.5513708466179647e-05,
      "loss": 0.10603523254394531,
      "step": 1962
    },
    {
      "epoch": 0.2735316658538285,
      "grad_norm": 0.43598252534866333,
      "learning_rate": 3.550775709219125e-05,
      "loss": 0.11345672607421875,
      "step": 1963
    },
    {
      "epoch": 0.27367100954504286,
      "grad_norm": 0.4188361167907715,
      "learning_rate": 3.550180227273327e-05,
      "loss": 0.09899330139160156,
      "step": 1964
    },
    {
      "epoch": 0.27381035323625724,
      "grad_norm": 1.1931653022766113,
      "learning_rate": 3.549584400912874e-05,
      "loss": 0.12101173400878906,
      "step": 1965
    },
    {
      "epoch": 0.2739496969274716,
      "grad_norm": 0.6831144094467163,
      "learning_rate": 3.5489882302701445e-05,
      "loss": 0.10581588745117188,
      "step": 1966
    },
    {
      "epoch": 0.274089040618686,
      "grad_norm": 0.46209463477134705,
      "learning_rate": 3.548391715477594e-05,
      "loss": 0.08888626098632812,
      "step": 1967
    },
    {
      "epoch": 0.2742283843099004,
      "grad_norm": 0.3034350275993347,
      "learning_rate": 3.547794856667756e-05,
      "loss": 0.07871818542480469,
      "step": 1968
    },
    {
      "epoch": 0.27436772800111475,
      "grad_norm": 0.33257126808166504,
      "learning_rate": 3.547197653973236e-05,
      "loss": 0.08785438537597656,
      "step": 1969
    },
    {
      "epoch": 0.27450707169232913,
      "grad_norm": 0.44992557168006897,
      "learning_rate": 3.546600107526721e-05,
      "loss": 0.09186553955078125,
      "step": 1970
    },
    {
      "epoch": 0.2746464153835435,
      "grad_norm": 0.3259158432483673,
      "learning_rate": 3.546002217460971e-05,
      "loss": 0.07746505737304688,
      "step": 1971
    },
    {
      "epoch": 0.2747857590747579,
      "grad_norm": 0.23460717499256134,
      "learning_rate": 3.5454039839088256e-05,
      "loss": 0.07041549682617188,
      "step": 1972
    },
    {
      "epoch": 0.27492510276597226,
      "grad_norm": 0.6748077869415283,
      "learning_rate": 3.544805407003196e-05,
      "loss": 0.1273965835571289,
      "step": 1973
    },
    {
      "epoch": 0.27506444645718664,
      "grad_norm": 0.49056944251060486,
      "learning_rate": 3.544206486877073e-05,
      "loss": 0.07704544067382812,
      "step": 1974
    },
    {
      "epoch": 0.275203790148401,
      "grad_norm": 0.34405627846717834,
      "learning_rate": 3.543607223663524e-05,
      "loss": 0.08992862701416016,
      "step": 1975
    },
    {
      "epoch": 0.2753431338396154,
      "grad_norm": 0.8867620825767517,
      "learning_rate": 3.543007617495692e-05,
      "loss": 0.10330009460449219,
      "step": 1976
    },
    {
      "epoch": 0.2754824775308298,
      "grad_norm": 0.2595212459564209,
      "learning_rate": 3.5424076685067935e-05,
      "loss": 0.07495498657226562,
      "step": 1977
    },
    {
      "epoch": 0.27562182122204415,
      "grad_norm": 0.5839775204658508,
      "learning_rate": 3.5418073768301254e-05,
      "loss": 0.12533950805664062,
      "step": 1978
    },
    {
      "epoch": 0.2757611649132585,
      "grad_norm": 0.47888192534446716,
      "learning_rate": 3.5412067425990585e-05,
      "loss": 0.10244560241699219,
      "step": 1979
    },
    {
      "epoch": 0.2759005086044729,
      "grad_norm": 0.47099775075912476,
      "learning_rate": 3.54060576594704e-05,
      "loss": 0.11858177185058594,
      "step": 1980
    },
    {
      "epoch": 0.27603985229568734,
      "grad_norm": 1.0311931371688843,
      "learning_rate": 3.540004447007592e-05,
      "loss": 0.1495380401611328,
      "step": 1981
    },
    {
      "epoch": 0.2761791959869017,
      "grad_norm": 0.2896905541419983,
      "learning_rate": 3.5394027859143154e-05,
      "loss": 0.07521533966064453,
      "step": 1982
    },
    {
      "epoch": 0.2763185396781161,
      "grad_norm": 0.7052919268608093,
      "learning_rate": 3.5388007828008845e-05,
      "loss": 0.11617851257324219,
      "step": 1983
    },
    {
      "epoch": 0.27645788336933047,
      "grad_norm": 0.5670589208602905,
      "learning_rate": 3.5381984378010513e-05,
      "loss": 0.10597896575927734,
      "step": 1984
    },
    {
      "epoch": 0.27659722706054485,
      "grad_norm": 0.2592502236366272,
      "learning_rate": 3.5375957510486426e-05,
      "loss": 0.07413768768310547,
      "step": 1985
    },
    {
      "epoch": 0.2767365707517592,
      "grad_norm": 0.3200737535953522,
      "learning_rate": 3.5369927226775625e-05,
      "loss": 0.07195186614990234,
      "step": 1986
    },
    {
      "epoch": 0.2768759144429736,
      "grad_norm": 0.34834355115890503,
      "learning_rate": 3.536389352821789e-05,
      "loss": 0.09104728698730469,
      "step": 1987
    },
    {
      "epoch": 0.277015258134188,
      "grad_norm": 0.4452013671398163,
      "learning_rate": 3.535785641615378e-05,
      "loss": 0.10104560852050781,
      "step": 1988
    },
    {
      "epoch": 0.27715460182540236,
      "grad_norm": 0.33302363753318787,
      "learning_rate": 3.53518158919246e-05,
      "loss": 0.09110260009765625,
      "step": 1989
    },
    {
      "epoch": 0.27729394551661674,
      "grad_norm": 0.19890393316745758,
      "learning_rate": 3.5345771956872416e-05,
      "loss": 0.06351852416992188,
      "step": 1990
    },
    {
      "epoch": 0.2774332892078311,
      "grad_norm": 0.30907997488975525,
      "learning_rate": 3.5339724612340055e-05,
      "loss": 0.08650779724121094,
      "step": 1991
    },
    {
      "epoch": 0.2775726328990455,
      "grad_norm": 0.3389478325843811,
      "learning_rate": 3.5333673859671095e-05,
      "loss": 0.08023452758789062,
      "step": 1992
    },
    {
      "epoch": 0.27771197659025987,
      "grad_norm": 0.5167315006256104,
      "learning_rate": 3.532761970020987e-05,
      "loss": 0.1363372802734375,
      "step": 1993
    },
    {
      "epoch": 0.27785132028147425,
      "grad_norm": 0.27678409218788147,
      "learning_rate": 3.532156213530149e-05,
      "loss": 0.07525062561035156,
      "step": 1994
    },
    {
      "epoch": 0.2779906639726886,
      "grad_norm": 0.9599758982658386,
      "learning_rate": 3.5315501166291806e-05,
      "loss": 0.12007713317871094,
      "step": 1995
    },
    {
      "epoch": 0.278130007663903,
      "grad_norm": 0.33756813406944275,
      "learning_rate": 3.530943679452742e-05,
      "loss": 0.07428741455078125,
      "step": 1996
    },
    {
      "epoch": 0.2782693513551174,
      "grad_norm": 0.38301870226860046,
      "learning_rate": 3.530336902135569e-05,
      "loss": 0.09238243103027344,
      "step": 1997
    },
    {
      "epoch": 0.27840869504633176,
      "grad_norm": 0.36167415976524353,
      "learning_rate": 3.5297297848124756e-05,
      "loss": 0.08146858215332031,
      "step": 1998
    },
    {
      "epoch": 0.27854803873754613,
      "grad_norm": 1.2600332498550415,
      "learning_rate": 3.5291223276183476e-05,
      "loss": 0.1834716796875,
      "step": 1999
    },
    {
      "epoch": 0.2786873824287605,
      "grad_norm": 0.39971303939819336,
      "learning_rate": 3.528514530688149e-05,
      "loss": 0.09444046020507812,
      "step": 2000
    },
    {
      "epoch": 0.27882672611997494,
      "grad_norm": 0.41816550493240356,
      "learning_rate": 3.527906394156919e-05,
      "loss": 0.09694671630859375,
      "step": 2001
    },
    {
      "epoch": 0.2789660698111893,
      "grad_norm": 0.5608137845993042,
      "learning_rate": 3.52729791815977e-05,
      "loss": 0.10853004455566406,
      "step": 2002
    },
    {
      "epoch": 0.2791054135024037,
      "grad_norm": 0.5796910524368286,
      "learning_rate": 3.526689102831892e-05,
      "loss": 0.13431644439697266,
      "step": 2003
    },
    {
      "epoch": 0.2792447571936181,
      "grad_norm": 0.35114586353302,
      "learning_rate": 3.526079948308551e-05,
      "loss": 0.1079401969909668,
      "step": 2004
    },
    {
      "epoch": 0.27938410088483245,
      "grad_norm": 0.39952677488327026,
      "learning_rate": 3.525470454725087e-05,
      "loss": 0.06465911865234375,
      "step": 2005
    },
    {
      "epoch": 0.27952344457604683,
      "grad_norm": 0.36104124784469604,
      "learning_rate": 3.524860622216914e-05,
      "loss": 0.09728240966796875,
      "step": 2006
    },
    {
      "epoch": 0.2796627882672612,
      "grad_norm": 0.7839861512184143,
      "learning_rate": 3.524250450919524e-05,
      "loss": 0.1350231170654297,
      "step": 2007
    },
    {
      "epoch": 0.2798021319584756,
      "grad_norm": 0.3416142165660858,
      "learning_rate": 3.523639940968484e-05,
      "loss": 0.07802391052246094,
      "step": 2008
    },
    {
      "epoch": 0.27994147564968996,
      "grad_norm": 0.2836419641971588,
      "learning_rate": 3.5230290924994334e-05,
      "loss": 0.06949806213378906,
      "step": 2009
    },
    {
      "epoch": 0.28008081934090434,
      "grad_norm": 0.41860708594322205,
      "learning_rate": 3.5224179056480906e-05,
      "loss": 0.08610343933105469,
      "step": 2010
    },
    {
      "epoch": 0.2802201630321187,
      "grad_norm": 0.6027640700340271,
      "learning_rate": 3.521806380550246e-05,
      "loss": 0.14846229553222656,
      "step": 2011
    },
    {
      "epoch": 0.2803595067233331,
      "grad_norm": 0.45950642228126526,
      "learning_rate": 3.5211945173417674e-05,
      "loss": 0.10294914245605469,
      "step": 2012
    },
    {
      "epoch": 0.2804988504145475,
      "grad_norm": 0.34128785133361816,
      "learning_rate": 3.520582316158596e-05,
      "loss": 0.08978843688964844,
      "step": 2013
    },
    {
      "epoch": 0.28063819410576185,
      "grad_norm": 0.29848596453666687,
      "learning_rate": 3.5199697771367494e-05,
      "loss": 0.06291389465332031,
      "step": 2014
    },
    {
      "epoch": 0.28077753779697623,
      "grad_norm": 1.010061264038086,
      "learning_rate": 3.5193569004123204e-05,
      "loss": 0.102081298828125,
      "step": 2015
    },
    {
      "epoch": 0.2809168814881906,
      "grad_norm": 0.3186851441860199,
      "learning_rate": 3.518743686121475e-05,
      "loss": 0.07619190216064453,
      "step": 2016
    },
    {
      "epoch": 0.281056225179405,
      "grad_norm": 0.3428669571876526,
      "learning_rate": 3.5181301344004574e-05,
      "loss": 0.07553482055664062,
      "step": 2017
    },
    {
      "epoch": 0.28119556887061936,
      "grad_norm": 0.35573920607566833,
      "learning_rate": 3.517516245385582e-05,
      "loss": 0.09304332733154297,
      "step": 2018
    },
    {
      "epoch": 0.28133491256183374,
      "grad_norm": 0.576701283454895,
      "learning_rate": 3.5169020192132425e-05,
      "loss": 0.1267108917236328,
      "step": 2019
    },
    {
      "epoch": 0.2814742562530481,
      "grad_norm": 0.4636387825012207,
      "learning_rate": 3.516287456019907e-05,
      "loss": 0.09097862243652344,
      "step": 2020
    },
    {
      "epoch": 0.28161359994426255,
      "grad_norm": 0.5109367966651917,
      "learning_rate": 3.515672555942115e-05,
      "loss": 0.12767410278320312,
      "step": 2021
    },
    {
      "epoch": 0.2817529436354769,
      "grad_norm": 0.8496819138526917,
      "learning_rate": 3.5150573191164855e-05,
      "loss": 0.15622329711914062,
      "step": 2022
    },
    {
      "epoch": 0.2818922873266913,
      "grad_norm": 0.5370823740959167,
      "learning_rate": 3.514441745679708e-05,
      "loss": 0.09035301208496094,
      "step": 2023
    },
    {
      "epoch": 0.2820316310179057,
      "grad_norm": 0.6654703617095947,
      "learning_rate": 3.5138258357685494e-05,
      "loss": 0.1130685806274414,
      "step": 2024
    },
    {
      "epoch": 0.28217097470912006,
      "grad_norm": 0.5739553570747375,
      "learning_rate": 3.513209589519853e-05,
      "loss": 0.11667251586914062,
      "step": 2025
    },
    {
      "epoch": 0.28231031840033444,
      "grad_norm": 0.3623608350753784,
      "learning_rate": 3.512593007070532e-05,
      "loss": 0.07126045227050781,
      "step": 2026
    },
    {
      "epoch": 0.2824496620915488,
      "grad_norm": 0.5074231028556824,
      "learning_rate": 3.5119760885575785e-05,
      "loss": 0.10278511047363281,
      "step": 2027
    },
    {
      "epoch": 0.2825890057827632,
      "grad_norm": 0.5079374313354492,
      "learning_rate": 3.5113588341180564e-05,
      "loss": 0.11852073669433594,
      "step": 2028
    },
    {
      "epoch": 0.28272834947397757,
      "grad_norm": 0.38028639554977417,
      "learning_rate": 3.510741243889106e-05,
      "loss": 0.08341026306152344,
      "step": 2029
    },
    {
      "epoch": 0.28286769316519195,
      "grad_norm": 0.3194384276866913,
      "learning_rate": 3.510123318007943e-05,
      "loss": 0.06251907348632812,
      "step": 2030
    },
    {
      "epoch": 0.2830070368564063,
      "grad_norm": 1.5679850578308105,
      "learning_rate": 3.509505056611855e-05,
      "loss": 0.15052032470703125,
      "step": 2031
    },
    {
      "epoch": 0.2831463805476207,
      "grad_norm": 0.7880901098251343,
      "learning_rate": 3.508886459838206e-05,
      "loss": 0.11232566833496094,
      "step": 2032
    },
    {
      "epoch": 0.2832857242388351,
      "grad_norm": 0.3304464817047119,
      "learning_rate": 3.508267527824434e-05,
      "loss": 0.07713890075683594,
      "step": 2033
    },
    {
      "epoch": 0.28342506793004946,
      "grad_norm": 0.5241314172744751,
      "learning_rate": 3.5076482607080513e-05,
      "loss": 0.09955596923828125,
      "step": 2034
    },
    {
      "epoch": 0.28356441162126383,
      "grad_norm": 0.7181317806243896,
      "learning_rate": 3.507028658626646e-05,
      "loss": 0.10213279724121094,
      "step": 2035
    },
    {
      "epoch": 0.2837037553124782,
      "grad_norm": 0.5204509496688843,
      "learning_rate": 3.5064087217178787e-05,
      "loss": 0.08506011962890625,
      "step": 2036
    },
    {
      "epoch": 0.2838430990036926,
      "grad_norm": 0.30570340156555176,
      "learning_rate": 3.505788450119485e-05,
      "loss": 0.07583999633789062,
      "step": 2037
    },
    {
      "epoch": 0.28398244269490697,
      "grad_norm": 0.4159783124923706,
      "learning_rate": 3.505167843969276e-05,
      "loss": 0.08308029174804688,
      "step": 2038
    },
    {
      "epoch": 0.28412178638612134,
      "grad_norm": 0.4155747592449188,
      "learning_rate": 3.504546903405135e-05,
      "loss": 0.09405517578125,
      "step": 2039
    },
    {
      "epoch": 0.2842611300773357,
      "grad_norm": 0.5757187008857727,
      "learning_rate": 3.5039256285650214e-05,
      "loss": 0.14109039306640625,
      "step": 2040
    },
    {
      "epoch": 0.28440047376855015,
      "grad_norm": 0.3665321469306946,
      "learning_rate": 3.5033040195869685e-05,
      "loss": 0.09366226196289062,
      "step": 2041
    },
    {
      "epoch": 0.28453981745976453,
      "grad_norm": 0.2809184491634369,
      "learning_rate": 3.502682076609084e-05,
      "loss": 0.06767559051513672,
      "step": 2042
    },
    {
      "epoch": 0.2846791611509789,
      "grad_norm": 0.5288949608802795,
      "learning_rate": 3.5020597997695484e-05,
      "loss": 0.08463287353515625,
      "step": 2043
    },
    {
      "epoch": 0.2848185048421933,
      "grad_norm": 0.41560277342796326,
      "learning_rate": 3.501437189206618e-05,
      "loss": 0.08261299133300781,
      "step": 2044
    },
    {
      "epoch": 0.28495784853340766,
      "grad_norm": 1.0148755311965942,
      "learning_rate": 3.5008142450586226e-05,
      "loss": 0.17163658142089844,
      "step": 2045
    },
    {
      "epoch": 0.28509719222462204,
      "grad_norm": 0.539237916469574,
      "learning_rate": 3.500190967463966e-05,
      "loss": 0.10205078125,
      "step": 2046
    },
    {
      "epoch": 0.2852365359158364,
      "grad_norm": 0.5211705565452576,
      "learning_rate": 3.4995673565611265e-05,
      "loss": 0.09119796752929688,
      "step": 2047
    },
    {
      "epoch": 0.2853758796070508,
      "grad_norm": 0.3513891100883484,
      "learning_rate": 3.498943412488656e-05,
      "loss": 0.06157684326171875,
      "step": 2048
    },
    {
      "epoch": 0.2855152232982652,
      "grad_norm": 0.5245388746261597,
      "learning_rate": 3.4983191353851804e-05,
      "loss": 0.1049041748046875,
      "step": 2049
    },
    {
      "epoch": 0.28565456698947955,
      "grad_norm": 0.4318659007549286,
      "learning_rate": 3.4976945253894e-05,
      "loss": 0.12069511413574219,
      "step": 2050
    },
    {
      "epoch": 0.28579391068069393,
      "grad_norm": 0.45237475633621216,
      "learning_rate": 3.49706958264009e-05,
      "loss": 0.11353111267089844,
      "step": 2051
    },
    {
      "epoch": 0.2859332543719083,
      "grad_norm": 0.6031201481819153,
      "learning_rate": 3.496444307276097e-05,
      "loss": 0.09246444702148438,
      "step": 2052
    },
    {
      "epoch": 0.2860725980631227,
      "grad_norm": 0.3734297752380371,
      "learning_rate": 3.495818699436343e-05,
      "loss": 0.08231544494628906,
      "step": 2053
    },
    {
      "epoch": 0.28621194175433706,
      "grad_norm": 0.37651169300079346,
      "learning_rate": 3.495192759259824e-05,
      "loss": 0.07795906066894531,
      "step": 2054
    },
    {
      "epoch": 0.28635128544555144,
      "grad_norm": 0.42590630054473877,
      "learning_rate": 3.49456648688561e-05,
      "loss": 0.10091209411621094,
      "step": 2055
    },
    {
      "epoch": 0.2864906291367658,
      "grad_norm": 0.2602212727069855,
      "learning_rate": 3.493939882452845e-05,
      "loss": 0.08231163024902344,
      "step": 2056
    },
    {
      "epoch": 0.2866299728279802,
      "grad_norm": 0.502697765827179,
      "learning_rate": 3.493312946100743e-05,
      "loss": 0.10119915008544922,
      "step": 2057
    },
    {
      "epoch": 0.2867693165191946,
      "grad_norm": 0.44224539399147034,
      "learning_rate": 3.4926856779685993e-05,
      "loss": 0.07082748413085938,
      "step": 2058
    },
    {
      "epoch": 0.28690866021040895,
      "grad_norm": 1.285374402999878,
      "learning_rate": 3.492058078195776e-05,
      "loss": 0.12043190002441406,
      "step": 2059
    },
    {
      "epoch": 0.2870480039016233,
      "grad_norm": 0.5392195582389832,
      "learning_rate": 3.491430146921712e-05,
      "loss": 0.11692142486572266,
      "step": 2060
    },
    {
      "epoch": 0.28718734759283776,
      "grad_norm": 0.5769364237785339,
      "learning_rate": 3.49080188428592e-05,
      "loss": 0.0934600830078125,
      "step": 2061
    },
    {
      "epoch": 0.28732669128405214,
      "grad_norm": 0.4602513313293457,
      "learning_rate": 3.490173290427984e-05,
      "loss": 0.10551166534423828,
      "step": 2062
    },
    {
      "epoch": 0.2874660349752665,
      "grad_norm": 0.44367870688438416,
      "learning_rate": 3.489544365487564e-05,
      "loss": 0.1076812744140625,
      "step": 2063
    },
    {
      "epoch": 0.2876053786664809,
      "grad_norm": 0.47916075587272644,
      "learning_rate": 3.488915109604393e-05,
      "loss": 0.07186508178710938,
      "step": 2064
    },
    {
      "epoch": 0.28774472235769527,
      "grad_norm": 0.6950048208236694,
      "learning_rate": 3.488285522918277e-05,
      "loss": 0.11661529541015625,
      "step": 2065
    },
    {
      "epoch": 0.28788406604890965,
      "grad_norm": 0.5778775811195374,
      "learning_rate": 3.487655605569096e-05,
      "loss": 0.10140800476074219,
      "step": 2066
    },
    {
      "epoch": 0.288023409740124,
      "grad_norm": 0.7643184065818787,
      "learning_rate": 3.487025357696804e-05,
      "loss": 0.11916732788085938,
      "step": 2067
    },
    {
      "epoch": 0.2881627534313384,
      "grad_norm": 0.5924773216247559,
      "learning_rate": 3.486394779441426e-05,
      "loss": 0.0875244140625,
      "step": 2068
    },
    {
      "epoch": 0.2883020971225528,
      "grad_norm": 0.5425692200660706,
      "learning_rate": 3.485763870943064e-05,
      "loss": 0.11858940124511719,
      "step": 2069
    },
    {
      "epoch": 0.28844144081376716,
      "grad_norm": 0.3038046061992645,
      "learning_rate": 3.48513263234189e-05,
      "loss": 0.07624244689941406,
      "step": 2070
    },
    {
      "epoch": 0.28858078450498154,
      "grad_norm": 0.9170233607292175,
      "learning_rate": 3.484501063778151e-05,
      "loss": 0.11957359313964844,
      "step": 2071
    },
    {
      "epoch": 0.2887201281961959,
      "grad_norm": 0.5894279479980469,
      "learning_rate": 3.483869165392167e-05,
      "loss": 0.10427093505859375,
      "step": 2072
    },
    {
      "epoch": 0.2888594718874103,
      "grad_norm": 0.5294084548950195,
      "learning_rate": 3.483236937324332e-05,
      "loss": 0.11343002319335938,
      "step": 2073
    },
    {
      "epoch": 0.28899881557862467,
      "grad_norm": 0.5838432908058167,
      "learning_rate": 3.482604379715113e-05,
      "loss": 0.132110595703125,
      "step": 2074
    },
    {
      "epoch": 0.28913815926983905,
      "grad_norm": 0.5252535939216614,
      "learning_rate": 3.481971492705048e-05,
      "loss": 0.08907890319824219,
      "step": 2075
    },
    {
      "epoch": 0.2892775029610534,
      "grad_norm": 0.32209885120391846,
      "learning_rate": 3.481338276434753e-05,
      "loss": 0.07363319396972656,
      "step": 2076
    },
    {
      "epoch": 0.2894168466522678,
      "grad_norm": 0.4487036466598511,
      "learning_rate": 3.480704731044911e-05,
      "loss": 0.09509468078613281,
      "step": 2077
    },
    {
      "epoch": 0.2895561903434822,
      "grad_norm": 0.2984406054019928,
      "learning_rate": 3.480070856676283e-05,
      "loss": 0.08282279968261719,
      "step": 2078
    },
    {
      "epoch": 0.28969553403469656,
      "grad_norm": 0.4058975279331207,
      "learning_rate": 3.479436653469702e-05,
      "loss": 0.08962631225585938,
      "step": 2079
    },
    {
      "epoch": 0.28983487772591093,
      "grad_norm": 0.46609628200531006,
      "learning_rate": 3.478802121566073e-05,
      "loss": 0.08637428283691406,
      "step": 2080
    },
    {
      "epoch": 0.28997422141712537,
      "grad_norm": 0.6410079598426819,
      "learning_rate": 3.478167261106373e-05,
      "loss": 0.10193061828613281,
      "step": 2081
    },
    {
      "epoch": 0.29011356510833974,
      "grad_norm": 0.7031494379043579,
      "learning_rate": 3.4775320722316555e-05,
      "loss": 0.1073760986328125,
      "step": 2082
    },
    {
      "epoch": 0.2902529087995541,
      "grad_norm": 0.42779994010925293,
      "learning_rate": 3.476896555083044e-05,
      "loss": 0.08971118927001953,
      "step": 2083
    },
    {
      "epoch": 0.2903922524907685,
      "grad_norm": 0.41394221782684326,
      "learning_rate": 3.476260709801736e-05,
      "loss": 0.11905288696289062,
      "step": 2084
    },
    {
      "epoch": 0.2905315961819829,
      "grad_norm": 0.3321639895439148,
      "learning_rate": 3.475624536529002e-05,
      "loss": 0.0707244873046875,
      "step": 2085
    },
    {
      "epoch": 0.29067093987319725,
      "grad_norm": 0.5003089308738708,
      "learning_rate": 3.4749880354061855e-05,
      "loss": 0.08369827270507812,
      "step": 2086
    },
    {
      "epoch": 0.29081028356441163,
      "grad_norm": 0.3917432129383087,
      "learning_rate": 3.474351206574701e-05,
      "loss": 0.10224151611328125,
      "step": 2087
    },
    {
      "epoch": 0.290949627255626,
      "grad_norm": 0.5467779040336609,
      "learning_rate": 3.4737140501760396e-05,
      "loss": 0.10201454162597656,
      "step": 2088
    },
    {
      "epoch": 0.2910889709468404,
      "grad_norm": 0.1576787531375885,
      "learning_rate": 3.473076566351761e-05,
      "loss": 0.056171417236328125,
      "step": 2089
    },
    {
      "epoch": 0.29122831463805476,
      "grad_norm": 0.31077930331230164,
      "learning_rate": 3.4724387552435004e-05,
      "loss": 0.08428478240966797,
      "step": 2090
    },
    {
      "epoch": 0.29136765832926914,
      "grad_norm": 0.3415067791938782,
      "learning_rate": 3.471800616992965e-05,
      "loss": 0.08790397644042969,
      "step": 2091
    },
    {
      "epoch": 0.2915070020204835,
      "grad_norm": 0.4265667498111725,
      "learning_rate": 3.471162151741934e-05,
      "loss": 0.09807395935058594,
      "step": 2092
    },
    {
      "epoch": 0.2916463457116979,
      "grad_norm": 1.0415289402008057,
      "learning_rate": 3.47052335963226e-05,
      "loss": 0.15485382080078125,
      "step": 2093
    },
    {
      "epoch": 0.2917856894029123,
      "grad_norm": 0.642889678478241,
      "learning_rate": 3.469884240805869e-05,
      "loss": 0.13623619079589844,
      "step": 2094
    },
    {
      "epoch": 0.29192503309412665,
      "grad_norm": 0.4386214315891266,
      "learning_rate": 3.4692447954047566e-05,
      "loss": 0.08283615112304688,
      "step": 2095
    },
    {
      "epoch": 0.29206437678534103,
      "grad_norm": 0.33510154485702515,
      "learning_rate": 3.468605023570993e-05,
      "loss": 0.0673208236694336,
      "step": 2096
    },
    {
      "epoch": 0.2922037204765554,
      "grad_norm": 1.1273136138916016,
      "learning_rate": 3.4679649254467244e-05,
      "loss": 0.14962387084960938,
      "step": 2097
    },
    {
      "epoch": 0.2923430641677698,
      "grad_norm": 0.35053569078445435,
      "learning_rate": 3.467324501174163e-05,
      "loss": 0.07761764526367188,
      "step": 2098
    },
    {
      "epoch": 0.29248240785898416,
      "grad_norm": 0.23390597105026245,
      "learning_rate": 3.466683750895596e-05,
      "loss": 0.0808868408203125,
      "step": 2099
    },
    {
      "epoch": 0.29262175155019854,
      "grad_norm": 0.38983651995658875,
      "learning_rate": 3.4660426747533846e-05,
      "loss": 0.0955963134765625,
      "step": 2100
    },
    {
      "epoch": 0.29276109524141297,
      "grad_norm": 0.4707218408584595,
      "learning_rate": 3.4654012728899624e-05,
      "loss": 0.09681129455566406,
      "step": 2101
    },
    {
      "epoch": 0.29290043893262735,
      "grad_norm": 0.38831308484077454,
      "learning_rate": 3.464759545447832e-05,
      "loss": 0.0960540771484375,
      "step": 2102
    },
    {
      "epoch": 0.2930397826238417,
      "grad_norm": 0.7193843126296997,
      "learning_rate": 3.4641174925695716e-05,
      "loss": 0.15797805786132812,
      "step": 2103
    },
    {
      "epoch": 0.2931791263150561,
      "grad_norm": 0.464728444814682,
      "learning_rate": 3.4634751143978317e-05,
      "loss": 0.08670616149902344,
      "step": 2104
    },
    {
      "epoch": 0.2933184700062705,
      "grad_norm": 0.5166411399841309,
      "learning_rate": 3.4628324110753326e-05,
      "loss": 0.10496711730957031,
      "step": 2105
    },
    {
      "epoch": 0.29345781369748486,
      "grad_norm": 0.28511741757392883,
      "learning_rate": 3.462189382744869e-05,
      "loss": 0.078155517578125,
      "step": 2106
    },
    {
      "epoch": 0.29359715738869924,
      "grad_norm": 0.37454676628112793,
      "learning_rate": 3.461546029549306e-05,
      "loss": 0.08242607116699219,
      "step": 2107
    },
    {
      "epoch": 0.2937365010799136,
      "grad_norm": 0.3371250629425049,
      "learning_rate": 3.4609023516315834e-05,
      "loss": 0.08198261260986328,
      "step": 2108
    },
    {
      "epoch": 0.293875844771128,
      "grad_norm": 0.30086231231689453,
      "learning_rate": 3.4602583491347116e-05,
      "loss": 0.08827877044677734,
      "step": 2109
    },
    {
      "epoch": 0.29401518846234237,
      "grad_norm": 0.37922486662864685,
      "learning_rate": 3.4596140222017725e-05,
      "loss": 0.08135032653808594,
      "step": 2110
    },
    {
      "epoch": 0.29415453215355675,
      "grad_norm": 0.31655359268188477,
      "learning_rate": 3.4589693709759216e-05,
      "loss": 0.07038688659667969,
      "step": 2111
    },
    {
      "epoch": 0.2942938758447711,
      "grad_norm": 0.5995901823043823,
      "learning_rate": 3.4583243956003847e-05,
      "loss": 0.08998775482177734,
      "step": 2112
    },
    {
      "epoch": 0.2944332195359855,
      "grad_norm": 0.35880595445632935,
      "learning_rate": 3.457679096218461e-05,
      "loss": 0.09398841857910156,
      "step": 2113
    },
    {
      "epoch": 0.2945725632271999,
      "grad_norm": 0.32209083437919617,
      "learning_rate": 3.457033472973523e-05,
      "loss": 0.0819540023803711,
      "step": 2114
    },
    {
      "epoch": 0.29471190691841426,
      "grad_norm": 0.47115588188171387,
      "learning_rate": 3.4563875260090114e-05,
      "loss": 0.10219383239746094,
      "step": 2115
    },
    {
      "epoch": 0.29485125060962863,
      "grad_norm": 0.40643271803855896,
      "learning_rate": 3.45574125546844e-05,
      "loss": 0.09634780883789062,
      "step": 2116
    },
    {
      "epoch": 0.294990594300843,
      "grad_norm": 0.7056506872177124,
      "learning_rate": 3.4550946614953984e-05,
      "loss": 0.1078338623046875,
      "step": 2117
    },
    {
      "epoch": 0.2951299379920574,
      "grad_norm": 0.4892171025276184,
      "learning_rate": 3.454447744233543e-05,
      "loss": 0.1060495376586914,
      "step": 2118
    },
    {
      "epoch": 0.29526928168327177,
      "grad_norm": 0.4710906147956848,
      "learning_rate": 3.453800503826604e-05,
      "loss": 0.09679031372070312,
      "step": 2119
    },
    {
      "epoch": 0.29540862537448614,
      "grad_norm": 0.4885799288749695,
      "learning_rate": 3.453152940418384e-05,
      "loss": 0.094482421875,
      "step": 2120
    },
    {
      "epoch": 0.2955479690657006,
      "grad_norm": 0.37788811326026917,
      "learning_rate": 3.4525050541527566e-05,
      "loss": 0.0719156265258789,
      "step": 2121
    },
    {
      "epoch": 0.29568731275691496,
      "grad_norm": 0.2686011493206024,
      "learning_rate": 3.4518568451736675e-05,
      "loss": 0.06802749633789062,
      "step": 2122
    },
    {
      "epoch": 0.29582665644812933,
      "grad_norm": 0.2841027081012726,
      "learning_rate": 3.4512083136251346e-05,
      "loss": 0.05787467956542969,
      "step": 2123
    },
    {
      "epoch": 0.2959660001393437,
      "grad_norm": 0.40339699387550354,
      "learning_rate": 3.450559459651245e-05,
      "loss": 0.0903482437133789,
      "step": 2124
    },
    {
      "epoch": 0.2961053438305581,
      "grad_norm": 0.22828826308250427,
      "learning_rate": 3.449910283396161e-05,
      "loss": 0.07097625732421875,
      "step": 2125
    },
    {
      "epoch": 0.29624468752177247,
      "grad_norm": 0.4794375002384186,
      "learning_rate": 3.4492607850041136e-05,
      "loss": 0.1063995361328125,
      "step": 2126
    },
    {
      "epoch": 0.29638403121298684,
      "grad_norm": 0.37380251288414,
      "learning_rate": 3.448610964619407e-05,
      "loss": 0.09469985961914062,
      "step": 2127
    },
    {
      "epoch": 0.2965233749042012,
      "grad_norm": 0.630882740020752,
      "learning_rate": 3.447960822386417e-05,
      "loss": 0.1557788848876953,
      "step": 2128
    },
    {
      "epoch": 0.2966627185954156,
      "grad_norm": 1.3417961597442627,
      "learning_rate": 3.4473103584495894e-05,
      "loss": 0.13213729858398438,
      "step": 2129
    },
    {
      "epoch": 0.29680206228663,
      "grad_norm": 0.4532499611377716,
      "learning_rate": 3.446659572953443e-05,
      "loss": 0.10228919982910156,
      "step": 2130
    },
    {
      "epoch": 0.29694140597784435,
      "grad_norm": 0.31804201006889343,
      "learning_rate": 3.446008466042566e-05,
      "loss": 0.08793830871582031,
      "step": 2131
    },
    {
      "epoch": 0.29708074966905873,
      "grad_norm": 0.4352450966835022,
      "learning_rate": 3.445357037861622e-05,
      "loss": 0.09606552124023438,
      "step": 2132
    },
    {
      "epoch": 0.2972200933602731,
      "grad_norm": 0.35623499751091003,
      "learning_rate": 3.4447052885553424e-05,
      "loss": 0.07125282287597656,
      "step": 2133
    },
    {
      "epoch": 0.2973594370514875,
      "grad_norm": 0.4456421732902527,
      "learning_rate": 3.44405321826853e-05,
      "loss": 0.09111595153808594,
      "step": 2134
    },
    {
      "epoch": 0.29749878074270186,
      "grad_norm": 0.4499965310096741,
      "learning_rate": 3.443400827146062e-05,
      "loss": 0.0946044921875,
      "step": 2135
    },
    {
      "epoch": 0.29763812443391624,
      "grad_norm": 0.40589338541030884,
      "learning_rate": 3.442748115332882e-05,
      "loss": 0.09238433837890625,
      "step": 2136
    },
    {
      "epoch": 0.2977774681251306,
      "grad_norm": 0.3958226144313812,
      "learning_rate": 3.442095082974011e-05,
      "loss": 0.10013580322265625,
      "step": 2137
    },
    {
      "epoch": 0.297916811816345,
      "grad_norm": 0.5077462792396545,
      "learning_rate": 3.441441730214535e-05,
      "loss": 0.08050918579101562,
      "step": 2138
    },
    {
      "epoch": 0.2980561555075594,
      "grad_norm": 0.3427429795265198,
      "learning_rate": 3.440788057199616e-05,
      "loss": 0.08791351318359375,
      "step": 2139
    },
    {
      "epoch": 0.29819549919877375,
      "grad_norm": 0.63393634557724,
      "learning_rate": 3.440134064074483e-05,
      "loss": 0.13384628295898438,
      "step": 2140
    },
    {
      "epoch": 0.2983348428899882,
      "grad_norm": 0.4252096712589264,
      "learning_rate": 3.4394797509844415e-05,
      "loss": 0.08881092071533203,
      "step": 2141
    },
    {
      "epoch": 0.29847418658120256,
      "grad_norm": 0.30493465065956116,
      "learning_rate": 3.438825118074863e-05,
      "loss": 0.06873321533203125,
      "step": 2142
    },
    {
      "epoch": 0.29861353027241694,
      "grad_norm": 0.608375608921051,
      "learning_rate": 3.4381701654911915e-05,
      "loss": 0.1201934814453125,
      "step": 2143
    },
    {
      "epoch": 0.2987528739636313,
      "grad_norm": 0.49948617815971375,
      "learning_rate": 3.437514893378943e-05,
      "loss": 0.09692001342773438,
      "step": 2144
    },
    {
      "epoch": 0.2988922176548457,
      "grad_norm": 0.30702710151672363,
      "learning_rate": 3.4368593018837046e-05,
      "loss": 0.07010269165039062,
      "step": 2145
    },
    {
      "epoch": 0.29903156134606007,
      "grad_norm": 0.5603824853897095,
      "learning_rate": 3.4362033911511336e-05,
      "loss": 0.10648155212402344,
      "step": 2146
    },
    {
      "epoch": 0.29917090503727445,
      "grad_norm": 0.5419380068778992,
      "learning_rate": 3.435547161326958e-05,
      "loss": 0.133544921875,
      "step": 2147
    },
    {
      "epoch": 0.2993102487284888,
      "grad_norm": 0.9297410249710083,
      "learning_rate": 3.434890612556977e-05,
      "loss": 0.13912200927734375,
      "step": 2148
    },
    {
      "epoch": 0.2994495924197032,
      "grad_norm": 0.6058835983276367,
      "learning_rate": 3.434233744987061e-05,
      "loss": 0.13639259338378906,
      "step": 2149
    },
    {
      "epoch": 0.2995889361109176,
      "grad_norm": 0.32640838623046875,
      "learning_rate": 3.433576558763151e-05,
      "loss": 0.07975959777832031,
      "step": 2150
    },
    {
      "epoch": 0.29972827980213196,
      "grad_norm": 0.3459543287754059,
      "learning_rate": 3.4329190540312596e-05,
      "loss": 0.08760643005371094,
      "step": 2151
    },
    {
      "epoch": 0.29986762349334634,
      "grad_norm": 0.33536621928215027,
      "learning_rate": 3.432261230937468e-05,
      "loss": 0.08847427368164062,
      "step": 2152
    },
    {
      "epoch": 0.3000069671845607,
      "grad_norm": 0.7921039462089539,
      "learning_rate": 3.431603089627929e-05,
      "loss": 0.1289386749267578,
      "step": 2153
    },
    {
      "epoch": 0.3001463108757751,
      "grad_norm": 0.220619335770607,
      "learning_rate": 3.4309446302488686e-05,
      "loss": 0.05788612365722656,
      "step": 2154
    },
    {
      "epoch": 0.30028565456698947,
      "grad_norm": 0.6694440841674805,
      "learning_rate": 3.4302858529465806e-05,
      "loss": 0.10042381286621094,
      "step": 2155
    },
    {
      "epoch": 0.30042499825820385,
      "grad_norm": 0.41608506441116333,
      "learning_rate": 3.429626757867429e-05,
      "loss": 0.10307121276855469,
      "step": 2156
    },
    {
      "epoch": 0.3005643419494182,
      "grad_norm": 0.6776466369628906,
      "learning_rate": 3.428967345157852e-05,
      "loss": 0.13116455078125,
      "step": 2157
    },
    {
      "epoch": 0.3007036856406326,
      "grad_norm": 0.6127967834472656,
      "learning_rate": 3.428307614964354e-05,
      "loss": 0.11480522155761719,
      "step": 2158
    },
    {
      "epoch": 0.300843029331847,
      "grad_norm": 1.661865234375,
      "learning_rate": 3.427647567433512e-05,
      "loss": 0.14780902862548828,
      "step": 2159
    },
    {
      "epoch": 0.30098237302306136,
      "grad_norm": 0.7376072406768799,
      "learning_rate": 3.426987202711976e-05,
      "loss": 0.12442779541015625,
      "step": 2160
    },
    {
      "epoch": 0.3011217167142758,
      "grad_norm": 0.5364863872528076,
      "learning_rate": 3.4263265209464606e-05,
      "loss": 0.06836128234863281,
      "step": 2161
    },
    {
      "epoch": 0.30126106040549017,
      "grad_norm": 0.3374653458595276,
      "learning_rate": 3.4256655222837574e-05,
      "loss": 0.08504009246826172,
      "step": 2162
    },
    {
      "epoch": 0.30140040409670454,
      "grad_norm": 0.3822915256023407,
      "learning_rate": 3.425004206870723e-05,
      "loss": 0.07573127746582031,
      "step": 2163
    },
    {
      "epoch": 0.3015397477879189,
      "grad_norm": 0.9280564188957214,
      "learning_rate": 3.424342574854286e-05,
      "loss": 0.124542236328125,
      "step": 2164
    },
    {
      "epoch": 0.3016790914791333,
      "grad_norm": 0.5022726058959961,
      "learning_rate": 3.423680626381449e-05,
      "loss": 0.09351921081542969,
      "step": 2165
    },
    {
      "epoch": 0.3018184351703477,
      "grad_norm": 0.455067902803421,
      "learning_rate": 3.423018361599279e-05,
      "loss": 0.1001739501953125,
      "step": 2166
    },
    {
      "epoch": 0.30195777886156205,
      "grad_norm": 0.5218072533607483,
      "learning_rate": 3.4223557806549175e-05,
      "loss": 0.12045764923095703,
      "step": 2167
    },
    {
      "epoch": 0.30209712255277643,
      "grad_norm": 0.4877760410308838,
      "learning_rate": 3.421692883695574e-05,
      "loss": 0.10284614562988281,
      "step": 2168
    },
    {
      "epoch": 0.3022364662439908,
      "grad_norm": 0.4854704439640045,
      "learning_rate": 3.4210296708685303e-05,
      "loss": 0.0814208984375,
      "step": 2169
    },
    {
      "epoch": 0.3023758099352052,
      "grad_norm": 0.3876161277294159,
      "learning_rate": 3.420366142321136e-05,
      "loss": 0.0862274169921875,
      "step": 2170
    },
    {
      "epoch": 0.30251515362641956,
      "grad_norm": 0.519382655620575,
      "learning_rate": 3.419702298200812e-05,
      "loss": 0.11932945251464844,
      "step": 2171
    },
    {
      "epoch": 0.30265449731763394,
      "grad_norm": 0.3671526610851288,
      "learning_rate": 3.41903813865505e-05,
      "loss": 0.09107780456542969,
      "step": 2172
    },
    {
      "epoch": 0.3027938410088483,
      "grad_norm": 0.5745139122009277,
      "learning_rate": 3.418373663831411e-05,
      "loss": 0.10152053833007812,
      "step": 2173
    },
    {
      "epoch": 0.3029331847000627,
      "grad_norm": 0.46935635805130005,
      "learning_rate": 3.4177088738775254e-05,
      "loss": 0.09948158264160156,
      "step": 2174
    },
    {
      "epoch": 0.3030725283912771,
      "grad_norm": 1.0329387187957764,
      "learning_rate": 3.417043768941095e-05,
      "loss": 0.1117544174194336,
      "step": 2175
    },
    {
      "epoch": 0.30321187208249145,
      "grad_norm": 1.037937045097351,
      "learning_rate": 3.416378349169891e-05,
      "loss": 0.117767333984375,
      "step": 2176
    },
    {
      "epoch": 0.30335121577370583,
      "grad_norm": 0.49376338720321655,
      "learning_rate": 3.415712614711755e-05,
      "loss": 0.10068893432617188,
      "step": 2177
    },
    {
      "epoch": 0.3034905594649202,
      "grad_norm": 0.3958212733268738,
      "learning_rate": 3.4150465657145964e-05,
      "loss": 0.07669639587402344,
      "step": 2178
    },
    {
      "epoch": 0.3036299031561346,
      "grad_norm": 0.2620695233345032,
      "learning_rate": 3.414380202326397e-05,
      "loss": 0.0654592514038086,
      "step": 2179
    },
    {
      "epoch": 0.30376924684734896,
      "grad_norm": 0.5514135360717773,
      "learning_rate": 3.413713524695208e-05,
      "loss": 0.09516525268554688,
      "step": 2180
    },
    {
      "epoch": 0.30390859053856334,
      "grad_norm": 0.4060947000980377,
      "learning_rate": 3.413046532969149e-05,
      "loss": 0.07757759094238281,
      "step": 2181
    },
    {
      "epoch": 0.30404793422977777,
      "grad_norm": 0.4602896571159363,
      "learning_rate": 3.412379227296411e-05,
      "loss": 0.07181739807128906,
      "step": 2182
    },
    {
      "epoch": 0.30418727792099215,
      "grad_norm": 0.43439313769340515,
      "learning_rate": 3.411711607825253e-05,
      "loss": 0.07863235473632812,
      "step": 2183
    },
    {
      "epoch": 0.3043266216122065,
      "grad_norm": 0.3391636312007904,
      "learning_rate": 3.411043674704007e-05,
      "loss": 0.08497810363769531,
      "step": 2184
    },
    {
      "epoch": 0.3044659653034209,
      "grad_norm": 0.5677062273025513,
      "learning_rate": 3.4103754280810705e-05,
      "loss": 0.07610607147216797,
      "step": 2185
    },
    {
      "epoch": 0.3046053089946353,
      "grad_norm": 0.5425165891647339,
      "learning_rate": 3.409706868104913e-05,
      "loss": 0.12258052825927734,
      "step": 2186
    },
    {
      "epoch": 0.30474465268584966,
      "grad_norm": 0.26035138964653015,
      "learning_rate": 3.409037994924074e-05,
      "loss": 0.06361007690429688,
      "step": 2187
    },
    {
      "epoch": 0.30488399637706404,
      "grad_norm": 0.5197556614875793,
      "learning_rate": 3.408368808687161e-05,
      "loss": 0.1010751724243164,
      "step": 2188
    },
    {
      "epoch": 0.3050233400682784,
      "grad_norm": 0.6934305429458618,
      "learning_rate": 3.407699309542853e-05,
      "loss": 0.10816001892089844,
      "step": 2189
    },
    {
      "epoch": 0.3051626837594928,
      "grad_norm": 0.3629651963710785,
      "learning_rate": 3.407029497639896e-05,
      "loss": 0.10110282897949219,
      "step": 2190
    },
    {
      "epoch": 0.30530202745070717,
      "grad_norm": 0.9198886752128601,
      "learning_rate": 3.406359373127108e-05,
      "loss": 0.1447925567626953,
      "step": 2191
    },
    {
      "epoch": 0.30544137114192155,
      "grad_norm": 0.5217763185501099,
      "learning_rate": 3.405688936153375e-05,
      "loss": 0.09319305419921875,
      "step": 2192
    },
    {
      "epoch": 0.3055807148331359,
      "grad_norm": 0.81162428855896,
      "learning_rate": 3.405018186867653e-05,
      "loss": 0.10169029235839844,
      "step": 2193
    },
    {
      "epoch": 0.3057200585243503,
      "grad_norm": 0.4772144556045532,
      "learning_rate": 3.404347125418967e-05,
      "loss": 0.08695220947265625,
      "step": 2194
    },
    {
      "epoch": 0.3058594022155647,
      "grad_norm": 0.4903084933757782,
      "learning_rate": 3.4036757519564116e-05,
      "loss": 0.09914004802703857,
      "step": 2195
    },
    {
      "epoch": 0.30599874590677906,
      "grad_norm": 0.6063910722732544,
      "learning_rate": 3.40300406662915e-05,
      "loss": 0.1074676513671875,
      "step": 2196
    },
    {
      "epoch": 0.30613808959799343,
      "grad_norm": 1.0479623079299927,
      "learning_rate": 3.402332069586416e-05,
      "loss": 0.11800861358642578,
      "step": 2197
    },
    {
      "epoch": 0.3062774332892078,
      "grad_norm": 0.8403177261352539,
      "learning_rate": 3.401659760977513e-05,
      "loss": 0.09145736694335938,
      "step": 2198
    },
    {
      "epoch": 0.3064167769804222,
      "grad_norm": 0.44769829511642456,
      "learning_rate": 3.4009871409518104e-05,
      "loss": 0.08980751037597656,
      "step": 2199
    },
    {
      "epoch": 0.30655612067163657,
      "grad_norm": 0.21064536273479462,
      "learning_rate": 3.40031420965875e-05,
      "loss": 0.06361770629882812,
      "step": 2200
    },
    {
      "epoch": 0.30669546436285094,
      "grad_norm": 0.5954223275184631,
      "learning_rate": 3.399640967247843e-05,
      "loss": 0.08922767639160156,
      "step": 2201
    },
    {
      "epoch": 0.3068348080540654,
      "grad_norm": 0.3422307074069977,
      "learning_rate": 3.398967413868666e-05,
      "loss": 0.08225822448730469,
      "step": 2202
    },
    {
      "epoch": 0.30697415174527976,
      "grad_norm": 0.5269918441772461,
      "learning_rate": 3.3982935496708704e-05,
      "loss": 0.08499908447265625,
      "step": 2203
    },
    {
      "epoch": 0.30711349543649413,
      "grad_norm": 0.5362560153007507,
      "learning_rate": 3.397619374804171e-05,
      "loss": 0.10434722900390625,
      "step": 2204
    },
    {
      "epoch": 0.3072528391277085,
      "grad_norm": 0.560541570186615,
      "learning_rate": 3.3969448894183536e-05,
      "loss": 0.090484619140625,
      "step": 2205
    },
    {
      "epoch": 0.3073921828189229,
      "grad_norm": 0.6833621263504028,
      "learning_rate": 3.396270093663276e-05,
      "loss": 0.08775997161865234,
      "step": 2206
    },
    {
      "epoch": 0.30753152651013727,
      "grad_norm": 0.6220759153366089,
      "learning_rate": 3.39559498768886e-05,
      "loss": 0.10339927673339844,
      "step": 2207
    },
    {
      "epoch": 0.30767087020135164,
      "grad_norm": 0.5056524276733398,
      "learning_rate": 3.3949195716451004e-05,
      "loss": 0.10686302185058594,
      "step": 2208
    },
    {
      "epoch": 0.307810213892566,
      "grad_norm": 0.5647354125976562,
      "learning_rate": 3.394243845682058e-05,
      "loss": 0.11247062683105469,
      "step": 2209
    },
    {
      "epoch": 0.3079495575837804,
      "grad_norm": 0.32976895570755005,
      "learning_rate": 3.3935678099498644e-05,
      "loss": 0.07585620880126953,
      "step": 2210
    },
    {
      "epoch": 0.3080889012749948,
      "grad_norm": 0.4210070073604584,
      "learning_rate": 3.392891464598719e-05,
      "loss": 0.09619712829589844,
      "step": 2211
    },
    {
      "epoch": 0.30822824496620915,
      "grad_norm": 0.9212906956672668,
      "learning_rate": 3.3922148097788906e-05,
      "loss": 0.1073150634765625,
      "step": 2212
    },
    {
      "epoch": 0.30836758865742353,
      "grad_norm": 0.39670461416244507,
      "learning_rate": 3.3915378456407167e-05,
      "loss": 0.09037399291992188,
      "step": 2213
    },
    {
      "epoch": 0.3085069323486379,
      "grad_norm": 0.5660991668701172,
      "learning_rate": 3.390860572334602e-05,
      "loss": 0.07079696655273438,
      "step": 2214
    },
    {
      "epoch": 0.3086462760398523,
      "grad_norm": 0.406087189912796,
      "learning_rate": 3.390182990011022e-05,
      "loss": 0.06858158111572266,
      "step": 2215
    },
    {
      "epoch": 0.30878561973106666,
      "grad_norm": 0.35389095544815063,
      "learning_rate": 3.389505098820521e-05,
      "loss": 0.09491920471191406,
      "step": 2216
    },
    {
      "epoch": 0.30892496342228104,
      "grad_norm": 0.27242687344551086,
      "learning_rate": 3.388826898913709e-05,
      "loss": 0.07175445556640625,
      "step": 2217
    },
    {
      "epoch": 0.3090643071134954,
      "grad_norm": 0.2788008749485016,
      "learning_rate": 3.3881483904412685e-05,
      "loss": 0.08760643005371094,
      "step": 2218
    },
    {
      "epoch": 0.3092036508047098,
      "grad_norm": 0.5513746738433838,
      "learning_rate": 3.3874695735539467e-05,
      "loss": 0.11725044250488281,
      "step": 2219
    },
    {
      "epoch": 0.3093429944959242,
      "grad_norm": 0.4611223340034485,
      "learning_rate": 3.3867904484025626e-05,
      "loss": 0.09033870697021484,
      "step": 2220
    },
    {
      "epoch": 0.30948233818713855,
      "grad_norm": 0.4473017156124115,
      "learning_rate": 3.3861110151380015e-05,
      "loss": 0.10693931579589844,
      "step": 2221
    },
    {
      "epoch": 0.309621681878353,
      "grad_norm": 0.3503588140010834,
      "learning_rate": 3.3854312739112186e-05,
      "loss": 0.07238960266113281,
      "step": 2222
    },
    {
      "epoch": 0.30976102556956736,
      "grad_norm": 0.31143879890441895,
      "learning_rate": 3.384751224873237e-05,
      "loss": 0.07255363464355469,
      "step": 2223
    },
    {
      "epoch": 0.30990036926078174,
      "grad_norm": 0.5713512897491455,
      "learning_rate": 3.384070868175146e-05,
      "loss": 0.10464668273925781,
      "step": 2224
    },
    {
      "epoch": 0.3100397129519961,
      "grad_norm": 0.411577045917511,
      "learning_rate": 3.383390203968109e-05,
      "loss": 0.09222984313964844,
      "step": 2225
    },
    {
      "epoch": 0.3101790566432105,
      "grad_norm": 0.48561152815818787,
      "learning_rate": 3.38270923240335e-05,
      "loss": 0.07882118225097656,
      "step": 2226
    },
    {
      "epoch": 0.31031840033442487,
      "grad_norm": 0.39045268297195435,
      "learning_rate": 3.382027953632169e-05,
      "loss": 0.09436225891113281,
      "step": 2227
    },
    {
      "epoch": 0.31045774402563925,
      "grad_norm": 0.5065590143203735,
      "learning_rate": 3.381346367805928e-05,
      "loss": 0.116546630859375,
      "step": 2228
    },
    {
      "epoch": 0.3105970877168536,
      "grad_norm": 0.2066865712404251,
      "learning_rate": 3.3806644750760615e-05,
      "loss": 0.06245994567871094,
      "step": 2229
    },
    {
      "epoch": 0.310736431408068,
      "grad_norm": 0.765433132648468,
      "learning_rate": 3.3799822755940694e-05,
      "loss": 0.09021663665771484,
      "step": 2230
    },
    {
      "epoch": 0.3108757750992824,
      "grad_norm": 0.4410562515258789,
      "learning_rate": 3.379299769511521e-05,
      "loss": 0.09302711486816406,
      "step": 2231
    },
    {
      "epoch": 0.31101511879049676,
      "grad_norm": 0.790136456489563,
      "learning_rate": 3.3786169569800534e-05,
      "loss": 0.16009521484375,
      "step": 2232
    },
    {
      "epoch": 0.31115446248171114,
      "grad_norm": 0.6399275660514832,
      "learning_rate": 3.377933838151374e-05,
      "loss": 0.10659217834472656,
      "step": 2233
    },
    {
      "epoch": 0.3112938061729255,
      "grad_norm": 0.451762855052948,
      "learning_rate": 3.377250413177253e-05,
      "loss": 0.10004997253417969,
      "step": 2234
    },
    {
      "epoch": 0.3114331498641399,
      "grad_norm": 0.4026309847831726,
      "learning_rate": 3.3765666822095336e-05,
      "loss": 0.09274101257324219,
      "step": 2235
    },
    {
      "epoch": 0.31157249355535427,
      "grad_norm": 0.48816579580307007,
      "learning_rate": 3.375882645400125e-05,
      "loss": 0.10879135131835938,
      "step": 2236
    },
    {
      "epoch": 0.31171183724656865,
      "grad_norm": 0.8027631640434265,
      "learning_rate": 3.375198302901004e-05,
      "loss": 0.1460552215576172,
      "step": 2237
    },
    {
      "epoch": 0.311851180937783,
      "grad_norm": 0.3675945997238159,
      "learning_rate": 3.3745136548642175e-05,
      "loss": 0.11374282836914062,
      "step": 2238
    },
    {
      "epoch": 0.3119905246289974,
      "grad_norm": 0.4578910171985626,
      "learning_rate": 3.373828701441877e-05,
      "loss": 0.11362457275390625,
      "step": 2239
    },
    {
      "epoch": 0.3121298683202118,
      "grad_norm": 0.5062185525894165,
      "learning_rate": 3.3731434427861644e-05,
      "loss": 0.0947265625,
      "step": 2240
    },
    {
      "epoch": 0.31226921201142616,
      "grad_norm": 0.3537193238735199,
      "learning_rate": 3.372457879049328e-05,
      "loss": 0.0774688720703125,
      "step": 2241
    },
    {
      "epoch": 0.3124085557026406,
      "grad_norm": 0.3560682237148285,
      "learning_rate": 3.3717720103836846e-05,
      "loss": 0.0877695083618164,
      "step": 2242
    },
    {
      "epoch": 0.31254789939385497,
      "grad_norm": 0.628003716468811,
      "learning_rate": 3.371085836941618e-05,
      "loss": 0.12603378295898438,
      "step": 2243
    },
    {
      "epoch": 0.31268724308506934,
      "grad_norm": 0.3049165904521942,
      "learning_rate": 3.370399358875582e-05,
      "loss": 0.07451343536376953,
      "step": 2244
    },
    {
      "epoch": 0.3128265867762837,
      "grad_norm": 0.7593726515769958,
      "learning_rate": 3.3697125763380944e-05,
      "loss": 0.12245655059814453,
      "step": 2245
    },
    {
      "epoch": 0.3129659304674981,
      "grad_norm": 0.5958145260810852,
      "learning_rate": 3.369025489481744e-05,
      "loss": 0.11583518981933594,
      "step": 2246
    },
    {
      "epoch": 0.3131052741587125,
      "grad_norm": 0.4666860103607178,
      "learning_rate": 3.3683380984591845e-05,
      "loss": 0.09042549133300781,
      "step": 2247
    },
    {
      "epoch": 0.31324461784992685,
      "grad_norm": 0.7894067168235779,
      "learning_rate": 3.36765040342314e-05,
      "loss": 0.14007186889648438,
      "step": 2248
    },
    {
      "epoch": 0.31338396154114123,
      "grad_norm": 0.42105093598365784,
      "learning_rate": 3.3669624045264e-05,
      "loss": 0.1000518798828125,
      "step": 2249
    },
    {
      "epoch": 0.3135233052323556,
      "grad_norm": 0.45504456758499146,
      "learning_rate": 3.3662741019218206e-05,
      "loss": 0.08559226989746094,
      "step": 2250
    },
    {
      "epoch": 0.31366264892357,
      "grad_norm": 0.3422819674015045,
      "learning_rate": 3.3655854957623295e-05,
      "loss": 0.07419967651367188,
      "step": 2251
    },
    {
      "epoch": 0.31380199261478436,
      "grad_norm": 0.47475188970565796,
      "learning_rate": 3.3648965862009174e-05,
      "loss": 0.0955047607421875,
      "step": 2252
    },
    {
      "epoch": 0.31394133630599874,
      "grad_norm": 0.5955170392990112,
      "learning_rate": 3.364207373390645e-05,
      "loss": 0.13274765014648438,
      "step": 2253
    },
    {
      "epoch": 0.3140806799972131,
      "grad_norm": 0.37457892298698425,
      "learning_rate": 3.3635178574846403e-05,
      "loss": 0.10373687744140625,
      "step": 2254
    },
    {
      "epoch": 0.3142200236884275,
      "grad_norm": 0.2952471077442169,
      "learning_rate": 3.362828038636097e-05,
      "loss": 0.07410907745361328,
      "step": 2255
    },
    {
      "epoch": 0.3143593673796419,
      "grad_norm": 0.45747804641723633,
      "learning_rate": 3.3621379169982774e-05,
      "loss": 0.09686279296875,
      "step": 2256
    },
    {
      "epoch": 0.31449871107085625,
      "grad_norm": 0.29270270466804504,
      "learning_rate": 3.361447492724511e-05,
      "loss": 0.06411361694335938,
      "step": 2257
    },
    {
      "epoch": 0.31463805476207063,
      "grad_norm": 0.8030925393104553,
      "learning_rate": 3.3607567659681934e-05,
      "loss": 0.10518074035644531,
      "step": 2258
    },
    {
      "epoch": 0.314777398453285,
      "grad_norm": 0.45017898082733154,
      "learning_rate": 3.3600657368827894e-05,
      "loss": 0.1027374267578125,
      "step": 2259
    },
    {
      "epoch": 0.3149167421444994,
      "grad_norm": 0.34578078985214233,
      "learning_rate": 3.35937440562183e-05,
      "loss": 0.07234573364257812,
      "step": 2260
    },
    {
      "epoch": 0.31505608583571376,
      "grad_norm": 0.4659200608730316,
      "learning_rate": 3.358682772338912e-05,
      "loss": 0.11041736602783203,
      "step": 2261
    },
    {
      "epoch": 0.3151954295269282,
      "grad_norm": 0.21286527812480927,
      "learning_rate": 3.357990837187701e-05,
      "loss": 0.06775569915771484,
      "step": 2262
    },
    {
      "epoch": 0.31533477321814257,
      "grad_norm": 0.7375885248184204,
      "learning_rate": 3.35729860032193e-05,
      "loss": 0.11047935485839844,
      "step": 2263
    },
    {
      "epoch": 0.31547411690935695,
      "grad_norm": 0.5221675634384155,
      "learning_rate": 3.356606061895398e-05,
      "loss": 0.1099081039428711,
      "step": 2264
    },
    {
      "epoch": 0.3156134606005713,
      "grad_norm": 0.3933596611022949,
      "learning_rate": 3.35591322206197e-05,
      "loss": 0.09245967864990234,
      "step": 2265
    },
    {
      "epoch": 0.3157528042917857,
      "grad_norm": 0.5408973693847656,
      "learning_rate": 3.355220080975581e-05,
      "loss": 0.08696937561035156,
      "step": 2266
    },
    {
      "epoch": 0.3158921479830001,
      "grad_norm": 0.2773128151893616,
      "learning_rate": 3.3545266387902295e-05,
      "loss": 0.06807804107666016,
      "step": 2267
    },
    {
      "epoch": 0.31603149167421446,
      "grad_norm": 0.30009403824806213,
      "learning_rate": 3.353832895659984e-05,
      "loss": 0.08386993408203125,
      "step": 2268
    },
    {
      "epoch": 0.31617083536542884,
      "grad_norm": 0.47882503271102905,
      "learning_rate": 3.353138851738976e-05,
      "loss": 0.11203384399414062,
      "step": 2269
    },
    {
      "epoch": 0.3163101790566432,
      "grad_norm": 0.4055362045764923,
      "learning_rate": 3.352444507181409e-05,
      "loss": 0.09613800048828125,
      "step": 2270
    },
    {
      "epoch": 0.3164495227478576,
      "grad_norm": 0.4450908899307251,
      "learning_rate": 3.3517498621415496e-05,
      "loss": 0.09634971618652344,
      "step": 2271
    },
    {
      "epoch": 0.31658886643907197,
      "grad_norm": 0.6092128753662109,
      "learning_rate": 3.3510549167737316e-05,
      "loss": 0.11800003051757812,
      "step": 2272
    },
    {
      "epoch": 0.31672821013028635,
      "grad_norm": 0.4656425714492798,
      "learning_rate": 3.350359671232356e-05,
      "loss": 0.09119224548339844,
      "step": 2273
    },
    {
      "epoch": 0.3168675538215007,
      "grad_norm": 0.6673257350921631,
      "learning_rate": 3.349664125671891e-05,
      "loss": 0.12740707397460938,
      "step": 2274
    },
    {
      "epoch": 0.3170068975127151,
      "grad_norm": 0.33994215726852417,
      "learning_rate": 3.3489682802468704e-05,
      "loss": 0.07729053497314453,
      "step": 2275
    },
    {
      "epoch": 0.3171462412039295,
      "grad_norm": 0.34801408648490906,
      "learning_rate": 3.348272135111895e-05,
      "loss": 0.09315109252929688,
      "step": 2276
    },
    {
      "epoch": 0.31728558489514386,
      "grad_norm": 0.41522085666656494,
      "learning_rate": 3.347575690421633e-05,
      "loss": 0.08078384399414062,
      "step": 2277
    },
    {
      "epoch": 0.31742492858635823,
      "grad_norm": 0.608802080154419,
      "learning_rate": 3.346878946330819e-05,
      "loss": 0.11621761322021484,
      "step": 2278
    },
    {
      "epoch": 0.3175642722775726,
      "grad_norm": 0.7342398762702942,
      "learning_rate": 3.346181902994252e-05,
      "loss": 0.14013290405273438,
      "step": 2279
    },
    {
      "epoch": 0.317703615968787,
      "grad_norm": 0.6116031408309937,
      "learning_rate": 3.3454845605668e-05,
      "loss": 0.11138534545898438,
      "step": 2280
    },
    {
      "epoch": 0.31784295966000137,
      "grad_norm": 0.3514629900455475,
      "learning_rate": 3.3447869192033974e-05,
      "loss": 0.094879150390625,
      "step": 2281
    },
    {
      "epoch": 0.3179823033512158,
      "grad_norm": 0.4417176842689514,
      "learning_rate": 3.344088979059042e-05,
      "loss": 0.08700275421142578,
      "step": 2282
    },
    {
      "epoch": 0.3181216470424302,
      "grad_norm": 0.33068954944610596,
      "learning_rate": 3.343390740288803e-05,
      "loss": 0.07244300842285156,
      "step": 2283
    },
    {
      "epoch": 0.31826099073364456,
      "grad_norm": 0.5799704790115356,
      "learning_rate": 3.3426922030478106e-05,
      "loss": 0.09161376953125,
      "step": 2284
    },
    {
      "epoch": 0.31840033442485893,
      "grad_norm": 0.3916057348251343,
      "learning_rate": 3.341993367491266e-05,
      "loss": 0.07014656066894531,
      "step": 2285
    },
    {
      "epoch": 0.3185396781160733,
      "grad_norm": 0.9308236241340637,
      "learning_rate": 3.3412942337744326e-05,
      "loss": 0.11835098266601562,
      "step": 2286
    },
    {
      "epoch": 0.3186790218072877,
      "grad_norm": 0.32238301634788513,
      "learning_rate": 3.340594802052642e-05,
      "loss": 0.06805610656738281,
      "step": 2287
    },
    {
      "epoch": 0.31881836549850207,
      "grad_norm": 0.344722181558609,
      "learning_rate": 3.339895072481294e-05,
      "loss": 0.07584762573242188,
      "step": 2288
    },
    {
      "epoch": 0.31895770918971644,
      "grad_norm": 0.5328919291496277,
      "learning_rate": 3.3391950452158504e-05,
      "loss": 0.11554527282714844,
      "step": 2289
    },
    {
      "epoch": 0.3190970528809308,
      "grad_norm": 0.5167103409767151,
      "learning_rate": 3.338494720411842e-05,
      "loss": 0.09307098388671875,
      "step": 2290
    },
    {
      "epoch": 0.3192363965721452,
      "grad_norm": 0.659233033657074,
      "learning_rate": 3.337794098224866e-05,
      "loss": 0.12290191650390625,
      "step": 2291
    },
    {
      "epoch": 0.3193757402633596,
      "grad_norm": 0.4836517870426178,
      "learning_rate": 3.337093178810583e-05,
      "loss": 0.08838462829589844,
      "step": 2292
    },
    {
      "epoch": 0.31951508395457395,
      "grad_norm": 0.4207834303379059,
      "learning_rate": 3.336391962324722e-05,
      "loss": 0.08433914184570312,
      "step": 2293
    },
    {
      "epoch": 0.31965442764578833,
      "grad_norm": 0.38034912943840027,
      "learning_rate": 3.3356904489230784e-05,
      "loss": 0.0902099609375,
      "step": 2294
    },
    {
      "epoch": 0.3197937713370027,
      "grad_norm": 0.2723650336265564,
      "learning_rate": 3.3349886387615096e-05,
      "loss": 0.07352066040039062,
      "step": 2295
    },
    {
      "epoch": 0.3199331150282171,
      "grad_norm": 0.23615872859954834,
      "learning_rate": 3.334286531995945e-05,
      "loss": 0.07040023803710938,
      "step": 2296
    },
    {
      "epoch": 0.32007245871943146,
      "grad_norm": 0.48382166028022766,
      "learning_rate": 3.3335841287823746e-05,
      "loss": 0.10594940185546875,
      "step": 2297
    },
    {
      "epoch": 0.32021180241064584,
      "grad_norm": 0.2502215504646301,
      "learning_rate": 3.332881429276857e-05,
      "loss": 0.08007049560546875,
      "step": 2298
    },
    {
      "epoch": 0.3203511461018602,
      "grad_norm": 0.2912226915359497,
      "learning_rate": 3.3321784336355163e-05,
      "loss": 0.06398391723632812,
      "step": 2299
    },
    {
      "epoch": 0.3204904897930746,
      "grad_norm": 0.5374675989151001,
      "learning_rate": 3.331475142014542e-05,
      "loss": 0.10400581359863281,
      "step": 2300
    },
    {
      "epoch": 0.320629833484289,
      "grad_norm": 0.3450746238231659,
      "learning_rate": 3.3307715545701885e-05,
      "loss": 0.09013557434082031,
      "step": 2301
    },
    {
      "epoch": 0.3207691771755034,
      "grad_norm": 0.4561043381690979,
      "learning_rate": 3.3300676714587784e-05,
      "loss": 0.10864639282226562,
      "step": 2302
    },
    {
      "epoch": 0.3209085208667178,
      "grad_norm": 0.5234081149101257,
      "learning_rate": 3.329363492836697e-05,
      "loss": 0.09229087829589844,
      "step": 2303
    },
    {
      "epoch": 0.32104786455793216,
      "grad_norm": 0.6799538731575012,
      "learning_rate": 3.328659018860398e-05,
      "loss": 0.11582565307617188,
      "step": 2304
    },
    {
      "epoch": 0.32118720824914654,
      "grad_norm": 0.6993386745452881,
      "learning_rate": 3.3279542496863984e-05,
      "loss": 0.1275920867919922,
      "step": 2305
    },
    {
      "epoch": 0.3213265519403609,
      "grad_norm": 0.32343941926956177,
      "learning_rate": 3.3272491854712825e-05,
      "loss": 0.09890556335449219,
      "step": 2306
    },
    {
      "epoch": 0.3214658956315753,
      "grad_norm": 0.28908371925354004,
      "learning_rate": 3.326543826371699e-05,
      "loss": 0.06862640380859375,
      "step": 2307
    },
    {
      "epoch": 0.32160523932278967,
      "grad_norm": 0.3716498017311096,
      "learning_rate": 3.3258381725443625e-05,
      "loss": 0.08921146392822266,
      "step": 2308
    },
    {
      "epoch": 0.32174458301400405,
      "grad_norm": 0.26163074374198914,
      "learning_rate": 3.325132224146054e-05,
      "loss": 0.06203651428222656,
      "step": 2309
    },
    {
      "epoch": 0.3218839267052184,
      "grad_norm": 0.5224486589431763,
      "learning_rate": 3.3244259813336185e-05,
      "loss": 0.11212730407714844,
      "step": 2310
    },
    {
      "epoch": 0.3220232703964328,
      "grad_norm": 0.5506842136383057,
      "learning_rate": 3.323719444263967e-05,
      "loss": 0.1135406494140625,
      "step": 2311
    },
    {
      "epoch": 0.3221626140876472,
      "grad_norm": 0.5278815627098083,
      "learning_rate": 3.323012613094075e-05,
      "loss": 0.10018062591552734,
      "step": 2312
    },
    {
      "epoch": 0.32230195777886156,
      "grad_norm": 0.5578588247299194,
      "learning_rate": 3.322305487980987e-05,
      "loss": 0.10527992248535156,
      "step": 2313
    },
    {
      "epoch": 0.32244130147007594,
      "grad_norm": 0.34109964966773987,
      "learning_rate": 3.3215980690818076e-05,
      "loss": 0.07400131225585938,
      "step": 2314
    },
    {
      "epoch": 0.3225806451612903,
      "grad_norm": 0.5088338851928711,
      "learning_rate": 3.32089035655371e-05,
      "loss": 0.1118011474609375,
      "step": 2315
    },
    {
      "epoch": 0.3227199888525047,
      "grad_norm": 0.32191866636276245,
      "learning_rate": 3.320182350553931e-05,
      "loss": 0.0729684829711914,
      "step": 2316
    },
    {
      "epoch": 0.32285933254371907,
      "grad_norm": 0.5021439790725708,
      "learning_rate": 3.319474051239775e-05,
      "loss": 0.11319923400878906,
      "step": 2317
    },
    {
      "epoch": 0.32299867623493345,
      "grad_norm": 0.4694070518016815,
      "learning_rate": 3.318765458768608e-05,
      "loss": 0.10040473937988281,
      "step": 2318
    },
    {
      "epoch": 0.3231380199261478,
      "grad_norm": 0.435435950756073,
      "learning_rate": 3.318056573297864e-05,
      "loss": 0.09393119812011719,
      "step": 2319
    },
    {
      "epoch": 0.3232773636173622,
      "grad_norm": 0.5262898206710815,
      "learning_rate": 3.317347394985042e-05,
      "loss": 0.1358470916748047,
      "step": 2320
    },
    {
      "epoch": 0.3234167073085766,
      "grad_norm": 0.4200402796268463,
      "learning_rate": 3.316637923987704e-05,
      "loss": 0.08157157897949219,
      "step": 2321
    },
    {
      "epoch": 0.323556050999791,
      "grad_norm": 0.30181458592414856,
      "learning_rate": 3.315928160463478e-05,
      "loss": 0.07543563842773438,
      "step": 2322
    },
    {
      "epoch": 0.3236953946910054,
      "grad_norm": 0.3008681833744049,
      "learning_rate": 3.3152181045700584e-05,
      "loss": 0.08365631103515625,
      "step": 2323
    },
    {
      "epoch": 0.32383473838221977,
      "grad_norm": 0.30835631489753723,
      "learning_rate": 3.314507756465202e-05,
      "loss": 0.09855461120605469,
      "step": 2324
    },
    {
      "epoch": 0.32397408207343414,
      "grad_norm": 0.5550879240036011,
      "learning_rate": 3.313797116306734e-05,
      "loss": 0.14342689514160156,
      "step": 2325
    },
    {
      "epoch": 0.3241134257646485,
      "grad_norm": 0.35623231530189514,
      "learning_rate": 3.3130861842525416e-05,
      "loss": 0.09258079528808594,
      "step": 2326
    },
    {
      "epoch": 0.3242527694558629,
      "grad_norm": 0.6147512793540955,
      "learning_rate": 3.3123749604605765e-05,
      "loss": 0.134246826171875,
      "step": 2327
    },
    {
      "epoch": 0.3243921131470773,
      "grad_norm": 0.9250198006629944,
      "learning_rate": 3.311663445088858e-05,
      "loss": 0.16204833984375,
      "step": 2328
    },
    {
      "epoch": 0.32453145683829165,
      "grad_norm": 0.4717129170894623,
      "learning_rate": 3.310951638295467e-05,
      "loss": 0.09883499145507812,
      "step": 2329
    },
    {
      "epoch": 0.32467080052950603,
      "grad_norm": 0.24493888020515442,
      "learning_rate": 3.310239540238552e-05,
      "loss": 0.07372856140136719,
      "step": 2330
    },
    {
      "epoch": 0.3248101442207204,
      "grad_norm": 0.624098002910614,
      "learning_rate": 3.3095271510763234e-05,
      "loss": 0.13164329528808594,
      "step": 2331
    },
    {
      "epoch": 0.3249494879119348,
      "grad_norm": 0.22563324868679047,
      "learning_rate": 3.3088144709670596e-05,
      "loss": 0.06113624572753906,
      "step": 2332
    },
    {
      "epoch": 0.32508883160314916,
      "grad_norm": 0.45894870162010193,
      "learning_rate": 3.3081015000691014e-05,
      "loss": 0.08589839935302734,
      "step": 2333
    },
    {
      "epoch": 0.32522817529436354,
      "grad_norm": 0.2636943459510803,
      "learning_rate": 3.3073882385408535e-05,
      "loss": 0.0672459602355957,
      "step": 2334
    },
    {
      "epoch": 0.3253675189855779,
      "grad_norm": 0.6906598806381226,
      "learning_rate": 3.306674686540788e-05,
      "loss": 0.10051250457763672,
      "step": 2335
    },
    {
      "epoch": 0.3255068626767923,
      "grad_norm": 0.9698188900947571,
      "learning_rate": 3.305960844227439e-05,
      "loss": 0.15515518188476562,
      "step": 2336
    },
    {
      "epoch": 0.3256462063680067,
      "grad_norm": 0.2422633022069931,
      "learning_rate": 3.305246711759406e-05,
      "loss": 0.0769805908203125,
      "step": 2337
    },
    {
      "epoch": 0.32578555005922105,
      "grad_norm": 0.493241548538208,
      "learning_rate": 3.3045322892953524e-05,
      "loss": 0.1036825180053711,
      "step": 2338
    },
    {
      "epoch": 0.32592489375043543,
      "grad_norm": 0.36997658014297485,
      "learning_rate": 3.303817576994008e-05,
      "loss": 0.0882425308227539,
      "step": 2339
    },
    {
      "epoch": 0.3260642374416498,
      "grad_norm": 0.5829008221626282,
      "learning_rate": 3.303102575014164e-05,
      "loss": 0.09795761108398438,
      "step": 2340
    },
    {
      "epoch": 0.3262035811328642,
      "grad_norm": 0.7862222194671631,
      "learning_rate": 3.3023872835146775e-05,
      "loss": 0.10712337493896484,
      "step": 2341
    },
    {
      "epoch": 0.3263429248240786,
      "grad_norm": 0.32332322001457214,
      "learning_rate": 3.301671702654472e-05,
      "loss": 0.06812667846679688,
      "step": 2342
    },
    {
      "epoch": 0.326482268515293,
      "grad_norm": 0.2966870665550232,
      "learning_rate": 3.300955832592531e-05,
      "loss": 0.0689544677734375,
      "step": 2343
    },
    {
      "epoch": 0.3266216122065074,
      "grad_norm": 0.3951704204082489,
      "learning_rate": 3.300239673487905e-05,
      "loss": 0.10073089599609375,
      "step": 2344
    },
    {
      "epoch": 0.32676095589772175,
      "grad_norm": 0.573266327381134,
      "learning_rate": 3.299523225499709e-05,
      "loss": 0.10152053833007812,
      "step": 2345
    },
    {
      "epoch": 0.3269002995889361,
      "grad_norm": 0.34922972321510315,
      "learning_rate": 3.298806488787121e-05,
      "loss": 0.07849311828613281,
      "step": 2346
    },
    {
      "epoch": 0.3270396432801505,
      "grad_norm": 0.40301749110221863,
      "learning_rate": 3.2980894635093837e-05,
      "loss": 0.09319496154785156,
      "step": 2347
    },
    {
      "epoch": 0.3271789869713649,
      "grad_norm": 0.7585023045539856,
      "learning_rate": 3.297372149825803e-05,
      "loss": 0.1248016357421875,
      "step": 2348
    },
    {
      "epoch": 0.32731833066257926,
      "grad_norm": 0.3214467465877533,
      "learning_rate": 3.2966545478957504e-05,
      "loss": 0.08795547485351562,
      "step": 2349
    },
    {
      "epoch": 0.32745767435379364,
      "grad_norm": 0.6638066172599792,
      "learning_rate": 3.29593665787866e-05,
      "loss": 0.1151885986328125,
      "step": 2350
    },
    {
      "epoch": 0.327597018045008,
      "grad_norm": 0.4727615714073181,
      "learning_rate": 3.295218479934032e-05,
      "loss": 0.105743408203125,
      "step": 2351
    },
    {
      "epoch": 0.3277363617362224,
      "grad_norm": 0.3728771209716797,
      "learning_rate": 3.2945000142214274e-05,
      "loss": 0.0948953628540039,
      "step": 2352
    },
    {
      "epoch": 0.32787570542743677,
      "grad_norm": 0.2520064413547516,
      "learning_rate": 3.293781260900473e-05,
      "loss": 0.07794475555419922,
      "step": 2353
    },
    {
      "epoch": 0.32801504911865115,
      "grad_norm": 0.3920459449291229,
      "learning_rate": 3.29306222013086e-05,
      "loss": 0.084991455078125,
      "step": 2354
    },
    {
      "epoch": 0.3281543928098655,
      "grad_norm": 0.6644890904426575,
      "learning_rate": 3.292342892072344e-05,
      "loss": 0.10717296600341797,
      "step": 2355
    },
    {
      "epoch": 0.3282937365010799,
      "grad_norm": 0.323684424161911,
      "learning_rate": 3.2916232768847404e-05,
      "loss": 0.07300376892089844,
      "step": 2356
    },
    {
      "epoch": 0.3284330801922943,
      "grad_norm": 0.5622155070304871,
      "learning_rate": 3.2909033747279344e-05,
      "loss": 0.11129951477050781,
      "step": 2357
    },
    {
      "epoch": 0.32857242388350866,
      "grad_norm": 0.4102945625782013,
      "learning_rate": 3.29018318576187e-05,
      "loss": 0.09053707122802734,
      "step": 2358
    },
    {
      "epoch": 0.32871176757472303,
      "grad_norm": 0.29124870896339417,
      "learning_rate": 3.289462710146557e-05,
      "loss": 0.06834983825683594,
      "step": 2359
    },
    {
      "epoch": 0.3288511112659374,
      "grad_norm": 0.35967689752578735,
      "learning_rate": 3.288741948042069e-05,
      "loss": 0.09321403503417969,
      "step": 2360
    },
    {
      "epoch": 0.3289904549571518,
      "grad_norm": 0.39062491059303284,
      "learning_rate": 3.288020899608542e-05,
      "loss": 0.08416557312011719,
      "step": 2361
    },
    {
      "epoch": 0.32912979864836617,
      "grad_norm": 0.40589770674705505,
      "learning_rate": 3.287299565006177e-05,
      "loss": 0.07540512084960938,
      "step": 2362
    },
    {
      "epoch": 0.3292691423395806,
      "grad_norm": 0.45984339714050293,
      "learning_rate": 3.286577944395239e-05,
      "loss": 0.12878990173339844,
      "step": 2363
    },
    {
      "epoch": 0.329408486030795,
      "grad_norm": 0.3910803496837616,
      "learning_rate": 3.2858560379360546e-05,
      "loss": 0.09573841094970703,
      "step": 2364
    },
    {
      "epoch": 0.32954782972200936,
      "grad_norm": 0.34030821919441223,
      "learning_rate": 3.2851338457890154e-05,
      "loss": 0.09856033325195312,
      "step": 2365
    },
    {
      "epoch": 0.32968717341322373,
      "grad_norm": 0.2837493419647217,
      "learning_rate": 3.284411368114575e-05,
      "loss": 0.09913063049316406,
      "step": 2366
    },
    {
      "epoch": 0.3298265171044381,
      "grad_norm": 0.4155622720718384,
      "learning_rate": 3.283688605073253e-05,
      "loss": 0.0820608139038086,
      "step": 2367
    },
    {
      "epoch": 0.3299658607956525,
      "grad_norm": 0.5526661276817322,
      "learning_rate": 3.282965556825629e-05,
      "loss": 0.11838626861572266,
      "step": 2368
    },
    {
      "epoch": 0.33010520448686687,
      "grad_norm": 0.3923546075820923,
      "learning_rate": 3.282242223532349e-05,
      "loss": 0.0969076156616211,
      "step": 2369
    },
    {
      "epoch": 0.33024454817808124,
      "grad_norm": 0.4976341426372528,
      "learning_rate": 3.281518605354123e-05,
      "loss": 0.1115884780883789,
      "step": 2370
    },
    {
      "epoch": 0.3303838918692956,
      "grad_norm": 0.35734495520591736,
      "learning_rate": 3.280794702451719e-05,
      "loss": 0.07043647766113281,
      "step": 2371
    },
    {
      "epoch": 0.33052323556051,
      "grad_norm": 0.3963504731655121,
      "learning_rate": 3.2800705149859725e-05,
      "loss": 0.09220218658447266,
      "step": 2372
    },
    {
      "epoch": 0.3306625792517244,
      "grad_norm": 0.5049615502357483,
      "learning_rate": 3.2793460431177827e-05,
      "loss": 0.09704780578613281,
      "step": 2373
    },
    {
      "epoch": 0.33080192294293875,
      "grad_norm": 0.3017151355743408,
      "learning_rate": 3.27862128700811e-05,
      "loss": 0.08580589294433594,
      "step": 2374
    },
    {
      "epoch": 0.33094126663415313,
      "grad_norm": 0.35772690176963806,
      "learning_rate": 3.277896246817979e-05,
      "loss": 0.08547019958496094,
      "step": 2375
    },
    {
      "epoch": 0.3310806103253675,
      "grad_norm": 0.19267503917217255,
      "learning_rate": 3.277170922708477e-05,
      "loss": 0.06628799438476562,
      "step": 2376
    },
    {
      "epoch": 0.3312199540165819,
      "grad_norm": 0.4267140328884125,
      "learning_rate": 3.276445314840754e-05,
      "loss": 0.09536933898925781,
      "step": 2377
    },
    {
      "epoch": 0.33135929770779626,
      "grad_norm": 0.5670433640480042,
      "learning_rate": 3.275719423376024e-05,
      "loss": 0.10722541809082031,
      "step": 2378
    },
    {
      "epoch": 0.33149864139901064,
      "grad_norm": 0.33942684531211853,
      "learning_rate": 3.274993248475563e-05,
      "loss": 0.09595108032226562,
      "step": 2379
    },
    {
      "epoch": 0.331637985090225,
      "grad_norm": 0.4974816143512726,
      "learning_rate": 3.274266790300711e-05,
      "loss": 0.09083080291748047,
      "step": 2380
    },
    {
      "epoch": 0.3317773287814394,
      "grad_norm": 0.18702788650989532,
      "learning_rate": 3.2735400490128695e-05,
      "loss": 0.06667423248291016,
      "step": 2381
    },
    {
      "epoch": 0.3319166724726538,
      "grad_norm": 0.4869811534881592,
      "learning_rate": 3.272813024773506e-05,
      "loss": 0.10177230834960938,
      "step": 2382
    },
    {
      "epoch": 0.3320560161638682,
      "grad_norm": 0.39756399393081665,
      "learning_rate": 3.272085717744146e-05,
      "loss": 0.09592247009277344,
      "step": 2383
    },
    {
      "epoch": 0.3321953598550826,
      "grad_norm": 0.393979012966156,
      "learning_rate": 3.271358128086381e-05,
      "loss": 0.07917022705078125,
      "step": 2384
    },
    {
      "epoch": 0.33233470354629696,
      "grad_norm": 0.5237410664558411,
      "learning_rate": 3.270630255961867e-05,
      "loss": 0.13058853149414062,
      "step": 2385
    },
    {
      "epoch": 0.33247404723751134,
      "grad_norm": 0.5242125988006592,
      "learning_rate": 3.269902101532319e-05,
      "loss": 0.11997604370117188,
      "step": 2386
    },
    {
      "epoch": 0.3326133909287257,
      "grad_norm": 0.3407009541988373,
      "learning_rate": 3.269173664959516e-05,
      "loss": 0.10419654846191406,
      "step": 2387
    },
    {
      "epoch": 0.3327527346199401,
      "grad_norm": 0.5610882639884949,
      "learning_rate": 3.2684449464053006e-05,
      "loss": 0.10400962829589844,
      "step": 2388
    },
    {
      "epoch": 0.33289207831115447,
      "grad_norm": 0.4991978108882904,
      "learning_rate": 3.2677159460315766e-05,
      "loss": 0.1324005126953125,
      "step": 2389
    },
    {
      "epoch": 0.33303142200236885,
      "grad_norm": 0.7368584871292114,
      "learning_rate": 3.2669866640003124e-05,
      "loss": 0.1494121551513672,
      "step": 2390
    },
    {
      "epoch": 0.3331707656935832,
      "grad_norm": 0.5434150099754333,
      "learning_rate": 3.266257100473538e-05,
      "loss": 0.0918588638305664,
      "step": 2391
    },
    {
      "epoch": 0.3333101093847976,
      "grad_norm": 0.30917656421661377,
      "learning_rate": 3.2655272556133436e-05,
      "loss": 0.07354927062988281,
      "step": 2392
    },
    {
      "epoch": 0.333449453076012,
      "grad_norm": 0.3430536389350891,
      "learning_rate": 3.264797129581886e-05,
      "loss": 0.07726478576660156,
      "step": 2393
    },
    {
      "epoch": 0.33358879676722636,
      "grad_norm": 0.3795054256916046,
      "learning_rate": 3.264066722541382e-05,
      "loss": 0.10697364807128906,
      "step": 2394
    },
    {
      "epoch": 0.33372814045844074,
      "grad_norm": 0.40601861476898193,
      "learning_rate": 3.263336034654112e-05,
      "loss": 0.09152030944824219,
      "step": 2395
    },
    {
      "epoch": 0.3338674841496551,
      "grad_norm": 0.44102704524993896,
      "learning_rate": 3.262605066082417e-05,
      "loss": 0.08270454406738281,
      "step": 2396
    },
    {
      "epoch": 0.3340068278408695,
      "grad_norm": 0.3930211067199707,
      "learning_rate": 3.261873816988702e-05,
      "loss": 0.07275009155273438,
      "step": 2397
    },
    {
      "epoch": 0.33414617153208387,
      "grad_norm": 0.36082959175109863,
      "learning_rate": 3.261142287535433e-05,
      "loss": 0.0689554214477539,
      "step": 2398
    },
    {
      "epoch": 0.33428551522329825,
      "grad_norm": 0.4438636898994446,
      "learning_rate": 3.2604104778851416e-05,
      "loss": 0.10985755920410156,
      "step": 2399
    },
    {
      "epoch": 0.3344248589145126,
      "grad_norm": 0.31746554374694824,
      "learning_rate": 3.259678388200417e-05,
      "loss": 0.08141613006591797,
      "step": 2400
    },
    {
      "epoch": 0.334564202605727,
      "grad_norm": 0.4546755254268646,
      "learning_rate": 3.258946018643914e-05,
      "loss": 0.07799720764160156,
      "step": 2401
    },
    {
      "epoch": 0.3347035462969414,
      "grad_norm": 0.44316938519477844,
      "learning_rate": 3.2582133693783475e-05,
      "loss": 0.08976078033447266,
      "step": 2402
    },
    {
      "epoch": 0.3348428899881558,
      "grad_norm": 0.42000386118888855,
      "learning_rate": 3.257480440566496e-05,
      "loss": 0.08169174194335938,
      "step": 2403
    },
    {
      "epoch": 0.3349822336793702,
      "grad_norm": 0.2285415232181549,
      "learning_rate": 3.256747232371199e-05,
      "loss": 0.07017135620117188,
      "step": 2404
    },
    {
      "epoch": 0.33512157737058457,
      "grad_norm": 0.46919485926628113,
      "learning_rate": 3.256013744955359e-05,
      "loss": 0.12347221374511719,
      "step": 2405
    },
    {
      "epoch": 0.33526092106179894,
      "grad_norm": 0.6146566867828369,
      "learning_rate": 3.25527997848194e-05,
      "loss": 0.10839462280273438,
      "step": 2406
    },
    {
      "epoch": 0.3354002647530133,
      "grad_norm": 0.35578176379203796,
      "learning_rate": 3.2545459331139694e-05,
      "loss": 0.09717178344726562,
      "step": 2407
    },
    {
      "epoch": 0.3355396084442277,
      "grad_norm": 0.5027663111686707,
      "learning_rate": 3.253811609014533e-05,
      "loss": 0.11473417282104492,
      "step": 2408
    },
    {
      "epoch": 0.3356789521354421,
      "grad_norm": 0.2492402046918869,
      "learning_rate": 3.2530770063467835e-05,
      "loss": 0.07414054870605469,
      "step": 2409
    },
    {
      "epoch": 0.33581829582665645,
      "grad_norm": 0.6268411874771118,
      "learning_rate": 3.2523421252739295e-05,
      "loss": 0.10320281982421875,
      "step": 2410
    },
    {
      "epoch": 0.33595763951787083,
      "grad_norm": 0.266360878944397,
      "learning_rate": 3.2516069659592485e-05,
      "loss": 0.07423591613769531,
      "step": 2411
    },
    {
      "epoch": 0.3360969832090852,
      "grad_norm": 0.33042073249816895,
      "learning_rate": 3.2508715285660734e-05,
      "loss": 0.09781837463378906,
      "step": 2412
    },
    {
      "epoch": 0.3362363269002996,
      "grad_norm": 0.39457595348358154,
      "learning_rate": 3.250135813257803e-05,
      "loss": 0.11181449890136719,
      "step": 2413
    },
    {
      "epoch": 0.33637567059151396,
      "grad_norm": 0.46706685423851013,
      "learning_rate": 3.249399820197895e-05,
      "loss": 0.06671714782714844,
      "step": 2414
    },
    {
      "epoch": 0.33651501428272834,
      "grad_norm": 0.5006923675537109,
      "learning_rate": 3.248663549549872e-05,
      "loss": 0.09073925018310547,
      "step": 2415
    },
    {
      "epoch": 0.3366543579739427,
      "grad_norm": 0.2768676280975342,
      "learning_rate": 3.247927001477316e-05,
      "loss": 0.07122039794921875,
      "step": 2416
    },
    {
      "epoch": 0.3367937016651571,
      "grad_norm": 0.4138382375240326,
      "learning_rate": 3.247190176143871e-05,
      "loss": 0.08867454528808594,
      "step": 2417
    },
    {
      "epoch": 0.3369330453563715,
      "grad_norm": 0.4021446406841278,
      "learning_rate": 3.246453073713242e-05,
      "loss": 0.09118080139160156,
      "step": 2418
    },
    {
      "epoch": 0.33707238904758585,
      "grad_norm": 0.4186002314090729,
      "learning_rate": 3.245715694349197e-05,
      "loss": 0.08011531829833984,
      "step": 2419
    },
    {
      "epoch": 0.33721173273880023,
      "grad_norm": 0.44859442114830017,
      "learning_rate": 3.244978038215566e-05,
      "loss": 0.09143638610839844,
      "step": 2420
    },
    {
      "epoch": 0.3373510764300146,
      "grad_norm": 0.33427193760871887,
      "learning_rate": 3.244240105476237e-05,
      "loss": 0.08266639709472656,
      "step": 2421
    },
    {
      "epoch": 0.337490420121229,
      "grad_norm": 0.45313364267349243,
      "learning_rate": 3.243501896295164e-05,
      "loss": 0.09927177429199219,
      "step": 2422
    },
    {
      "epoch": 0.3376297638124434,
      "grad_norm": 0.7248185873031616,
      "learning_rate": 3.242763410836358e-05,
      "loss": 0.12139129638671875,
      "step": 2423
    },
    {
      "epoch": 0.3377691075036578,
      "grad_norm": 0.5738295912742615,
      "learning_rate": 3.242024649263896e-05,
      "loss": 0.08926010131835938,
      "step": 2424
    },
    {
      "epoch": 0.3379084511948722,
      "grad_norm": 0.7763272523880005,
      "learning_rate": 3.241285611741913e-05,
      "loss": 0.12429046630859375,
      "step": 2425
    },
    {
      "epoch": 0.33804779488608655,
      "grad_norm": 0.4029110074043274,
      "learning_rate": 3.240546298434606e-05,
      "loss": 0.08815956115722656,
      "step": 2426
    },
    {
      "epoch": 0.3381871385773009,
      "grad_norm": 0.3251311480998993,
      "learning_rate": 3.2398067095062325e-05,
      "loss": 0.09242057800292969,
      "step": 2427
    },
    {
      "epoch": 0.3383264822685153,
      "grad_norm": 0.41928163170814514,
      "learning_rate": 3.239066845121114e-05,
      "loss": 0.07892990112304688,
      "step": 2428
    },
    {
      "epoch": 0.3384658259597297,
      "grad_norm": 0.675408124923706,
      "learning_rate": 3.238326705443631e-05,
      "loss": 0.09620285034179688,
      "step": 2429
    },
    {
      "epoch": 0.33860516965094406,
      "grad_norm": 0.34462517499923706,
      "learning_rate": 3.237586290638226e-05,
      "loss": 0.09120559692382812,
      "step": 2430
    },
    {
      "epoch": 0.33874451334215844,
      "grad_norm": 0.3990703523159027,
      "learning_rate": 3.2368456008694014e-05,
      "loss": 0.09833526611328125,
      "step": 2431
    },
    {
      "epoch": 0.3388838570333728,
      "grad_norm": 0.7682449221611023,
      "learning_rate": 3.2361046363017216e-05,
      "loss": 0.14214134216308594,
      "step": 2432
    },
    {
      "epoch": 0.3390232007245872,
      "grad_norm": 0.4660167098045349,
      "learning_rate": 3.2353633970998135e-05,
      "loss": 0.09596443176269531,
      "step": 2433
    },
    {
      "epoch": 0.33916254441580157,
      "grad_norm": 0.314060240983963,
      "learning_rate": 3.2346218834283605e-05,
      "loss": 0.082183837890625,
      "step": 2434
    },
    {
      "epoch": 0.33930188810701595,
      "grad_norm": 0.5374533534049988,
      "learning_rate": 3.233880095452113e-05,
      "loss": 0.12430191040039062,
      "step": 2435
    },
    {
      "epoch": 0.3394412317982303,
      "grad_norm": 0.5047711133956909,
      "learning_rate": 3.2331380333358794e-05,
      "loss": 0.12836074829101562,
      "step": 2436
    },
    {
      "epoch": 0.3395805754894447,
      "grad_norm": 0.44736090302467346,
      "learning_rate": 3.232395697244526e-05,
      "loss": 0.10450935363769531,
      "step": 2437
    },
    {
      "epoch": 0.3397199191806591,
      "grad_norm": 0.31245410442352295,
      "learning_rate": 3.231653087342986e-05,
      "loss": 0.08687973022460938,
      "step": 2438
    },
    {
      "epoch": 0.33985926287187346,
      "grad_norm": 0.3404836058616638,
      "learning_rate": 3.230910203796248e-05,
      "loss": 0.08744049072265625,
      "step": 2439
    },
    {
      "epoch": 0.33999860656308784,
      "grad_norm": 0.21726205945014954,
      "learning_rate": 3.2301670467693654e-05,
      "loss": 0.06679916381835938,
      "step": 2440
    },
    {
      "epoch": 0.3401379502543022,
      "grad_norm": 0.5237830877304077,
      "learning_rate": 3.22942361642745e-05,
      "loss": 0.08211040496826172,
      "step": 2441
    },
    {
      "epoch": 0.3402772939455166,
      "grad_norm": 0.2824972867965698,
      "learning_rate": 3.228679912935675e-05,
      "loss": 0.08375740051269531,
      "step": 2442
    },
    {
      "epoch": 0.340416637636731,
      "grad_norm": 0.3641866147518158,
      "learning_rate": 3.227935936459276e-05,
      "loss": 0.11381912231445312,
      "step": 2443
    },
    {
      "epoch": 0.3405559813279454,
      "grad_norm": 0.6694666743278503,
      "learning_rate": 3.2271916871635455e-05,
      "loss": 0.14427757263183594,
      "step": 2444
    },
    {
      "epoch": 0.3406953250191598,
      "grad_norm": 0.541516125202179,
      "learning_rate": 3.226447165213839e-05,
      "loss": 0.11024284362792969,
      "step": 2445
    },
    {
      "epoch": 0.34083466871037416,
      "grad_norm": 0.3573431372642517,
      "learning_rate": 3.225702370775572e-05,
      "loss": 0.10333061218261719,
      "step": 2446
    },
    {
      "epoch": 0.34097401240158853,
      "grad_norm": 0.48090577125549316,
      "learning_rate": 3.224957304014223e-05,
      "loss": 0.1178741455078125,
      "step": 2447
    },
    {
      "epoch": 0.3411133560928029,
      "grad_norm": 0.4571234881877899,
      "learning_rate": 3.224211965095326e-05,
      "loss": 0.10483551025390625,
      "step": 2448
    },
    {
      "epoch": 0.3412526997840173,
      "grad_norm": 0.5651015043258667,
      "learning_rate": 3.2234663541844805e-05,
      "loss": 0.12268638610839844,
      "step": 2449
    },
    {
      "epoch": 0.34139204347523167,
      "grad_norm": 0.8033178448677063,
      "learning_rate": 3.222720471447343e-05,
      "loss": 0.10239124298095703,
      "step": 2450
    },
    {
      "epoch": 0.34153138716644604,
      "grad_norm": 0.5018488168716431,
      "learning_rate": 3.221974317049632e-05,
      "loss": 0.08542060852050781,
      "step": 2451
    },
    {
      "epoch": 0.3416707308576604,
      "grad_norm": 0.8319666981697083,
      "learning_rate": 3.221227891157125e-05,
      "loss": 0.1290435791015625,
      "step": 2452
    },
    {
      "epoch": 0.3418100745488748,
      "grad_norm": 0.5570886731147766,
      "learning_rate": 3.220481193935663e-05,
      "loss": 0.09189414978027344,
      "step": 2453
    },
    {
      "epoch": 0.3419494182400892,
      "grad_norm": 0.4300248324871063,
      "learning_rate": 3.219734225551143e-05,
      "loss": 0.1008768081665039,
      "step": 2454
    },
    {
      "epoch": 0.34208876193130355,
      "grad_norm": 0.4105485677719116,
      "learning_rate": 3.2189869861695254e-05,
      "loss": 0.08240699768066406,
      "step": 2455
    },
    {
      "epoch": 0.34222810562251793,
      "grad_norm": 0.44879260659217834,
      "learning_rate": 3.218239475956829e-05,
      "loss": 0.10956954956054688,
      "step": 2456
    },
    {
      "epoch": 0.3423674493137323,
      "grad_norm": 0.3781580626964569,
      "learning_rate": 3.217491695079134e-05,
      "loss": 0.09499740600585938,
      "step": 2457
    },
    {
      "epoch": 0.3425067930049467,
      "grad_norm": 0.20924383401870728,
      "learning_rate": 3.216743643702581e-05,
      "loss": 0.05468177795410156,
      "step": 2458
    },
    {
      "epoch": 0.34264613669616106,
      "grad_norm": 0.5202205777168274,
      "learning_rate": 3.215995321993368e-05,
      "loss": 0.10908985137939453,
      "step": 2459
    },
    {
      "epoch": 0.34278548038737544,
      "grad_norm": 0.7814775705337524,
      "learning_rate": 3.215246730117757e-05,
      "loss": 0.12062263488769531,
      "step": 2460
    },
    {
      "epoch": 0.3429248240785898,
      "grad_norm": 0.291515052318573,
      "learning_rate": 3.2144978682420664e-05,
      "loss": 0.09375190734863281,
      "step": 2461
    },
    {
      "epoch": 0.3430641677698042,
      "grad_norm": 0.5268447995185852,
      "learning_rate": 3.2137487365326773e-05,
      "loss": 0.1308135986328125,
      "step": 2462
    },
    {
      "epoch": 0.34320351146101863,
      "grad_norm": 0.5495997071266174,
      "learning_rate": 3.212999335156029e-05,
      "loss": 0.10853767395019531,
      "step": 2463
    },
    {
      "epoch": 0.343342855152233,
      "grad_norm": 0.4850027561187744,
      "learning_rate": 3.212249664278622e-05,
      "loss": 0.13413524627685547,
      "step": 2464
    },
    {
      "epoch": 0.3434821988434474,
      "grad_norm": 0.36481475830078125,
      "learning_rate": 3.211499724067016e-05,
      "loss": 0.08443450927734375,
      "step": 2465
    },
    {
      "epoch": 0.34362154253466176,
      "grad_norm": 0.2388271987438202,
      "learning_rate": 3.2107495146878295e-05,
      "loss": 0.07675743103027344,
      "step": 2466
    },
    {
      "epoch": 0.34376088622587614,
      "grad_norm": 0.2634911835193634,
      "learning_rate": 3.2099990363077434e-05,
      "loss": 0.08017158508300781,
      "step": 2467
    },
    {
      "epoch": 0.3439002299170905,
      "grad_norm": 1.1354248523712158,
      "learning_rate": 3.209248289093496e-05,
      "loss": 0.1427021026611328,
      "step": 2468
    },
    {
      "epoch": 0.3440395736083049,
      "grad_norm": 0.40939679741859436,
      "learning_rate": 3.208497273211886e-05,
      "loss": 0.0929250717163086,
      "step": 2469
    },
    {
      "epoch": 0.34417891729951927,
      "grad_norm": 0.24878805875778198,
      "learning_rate": 3.207745988829773e-05,
      "loss": 0.07772636413574219,
      "step": 2470
    },
    {
      "epoch": 0.34431826099073365,
      "grad_norm": 0.4284425675868988,
      "learning_rate": 3.206994436114074e-05,
      "loss": 0.07416725158691406,
      "step": 2471
    },
    {
      "epoch": 0.344457604681948,
      "grad_norm": 0.46328625082969666,
      "learning_rate": 3.206242615231768e-05,
      "loss": 0.07968711853027344,
      "step": 2472
    },
    {
      "epoch": 0.3445969483731624,
      "grad_norm": 0.5142204761505127,
      "learning_rate": 3.2054905263498916e-05,
      "loss": 0.08838224411010742,
      "step": 2473
    },
    {
      "epoch": 0.3447362920643768,
      "grad_norm": 0.40713605284690857,
      "learning_rate": 3.2047381696355424e-05,
      "loss": 0.09913825988769531,
      "step": 2474
    },
    {
      "epoch": 0.34487563575559116,
      "grad_norm": 0.512003481388092,
      "learning_rate": 3.2039855452558755e-05,
      "loss": 0.08614349365234375,
      "step": 2475
    },
    {
      "epoch": 0.34501497944680554,
      "grad_norm": 0.6679466366767883,
      "learning_rate": 3.203232653378109e-05,
      "loss": 0.12686920166015625,
      "step": 2476
    },
    {
      "epoch": 0.3451543231380199,
      "grad_norm": 0.25272807478904724,
      "learning_rate": 3.202479494169516e-05,
      "loss": 0.06745147705078125,
      "step": 2477
    },
    {
      "epoch": 0.3452936668292343,
      "grad_norm": 0.47477859258651733,
      "learning_rate": 3.2017260677974346e-05,
      "loss": 0.0842275619506836,
      "step": 2478
    },
    {
      "epoch": 0.34543301052044867,
      "grad_norm": 0.40169757604599,
      "learning_rate": 3.200972374429255e-05,
      "loss": 0.07072257995605469,
      "step": 2479
    },
    {
      "epoch": 0.34557235421166305,
      "grad_norm": 0.5866200923919678,
      "learning_rate": 3.200218414232433e-05,
      "loss": 0.1068878173828125,
      "step": 2480
    },
    {
      "epoch": 0.3457116979028774,
      "grad_norm": 0.3350193202495575,
      "learning_rate": 3.199464187374481e-05,
      "loss": 0.08732795715332031,
      "step": 2481
    },
    {
      "epoch": 0.3458510415940918,
      "grad_norm": 0.639800488948822,
      "learning_rate": 3.19870969402297e-05,
      "loss": 0.11146354675292969,
      "step": 2482
    },
    {
      "epoch": 0.34599038528530623,
      "grad_norm": 0.47244054079055786,
      "learning_rate": 3.197954934345533e-05,
      "loss": 0.07835960388183594,
      "step": 2483
    },
    {
      "epoch": 0.3461297289765206,
      "grad_norm": 0.3806304633617401,
      "learning_rate": 3.1971999085098583e-05,
      "loss": 0.08087348937988281,
      "step": 2484
    },
    {
      "epoch": 0.346269072667735,
      "grad_norm": 0.3921177387237549,
      "learning_rate": 3.196444616683698e-05,
      "loss": 0.08104801177978516,
      "step": 2485
    },
    {
      "epoch": 0.34640841635894937,
      "grad_norm": 0.7224194407463074,
      "learning_rate": 3.195689059034858e-05,
      "loss": 0.09578323364257812,
      "step": 2486
    },
    {
      "epoch": 0.34654776005016374,
      "grad_norm": 0.7315101623535156,
      "learning_rate": 3.194933235731207e-05,
      "loss": 0.10746383666992188,
      "step": 2487
    },
    {
      "epoch": 0.3466871037413781,
      "grad_norm": 0.3492971658706665,
      "learning_rate": 3.194177146940673e-05,
      "loss": 0.07104015350341797,
      "step": 2488
    },
    {
      "epoch": 0.3468264474325925,
      "grad_norm": 0.3480431139469147,
      "learning_rate": 3.193420792831239e-05,
      "loss": 0.08890533447265625,
      "step": 2489
    },
    {
      "epoch": 0.3469657911238069,
      "grad_norm": 0.37840065360069275,
      "learning_rate": 3.192664173570952e-05,
      "loss": 0.087982177734375,
      "step": 2490
    },
    {
      "epoch": 0.34710513481502125,
      "grad_norm": 0.44156667590141296,
      "learning_rate": 3.1919072893279144e-05,
      "loss": 0.10117149353027344,
      "step": 2491
    },
    {
      "epoch": 0.34724447850623563,
      "grad_norm": 0.3404620587825775,
      "learning_rate": 3.1911501402702886e-05,
      "loss": 0.07656288146972656,
      "step": 2492
    },
    {
      "epoch": 0.34738382219745,
      "grad_norm": 0.6152765154838562,
      "learning_rate": 3.1903927265662965e-05,
      "loss": 0.10150146484375,
      "step": 2493
    },
    {
      "epoch": 0.3475231658886644,
      "grad_norm": 0.6122232675552368,
      "learning_rate": 3.189635048384217e-05,
      "loss": 0.11285972595214844,
      "step": 2494
    },
    {
      "epoch": 0.34766250957987876,
      "grad_norm": 1.2048684358596802,
      "learning_rate": 3.18887710589239e-05,
      "loss": 0.1577892303466797,
      "step": 2495
    },
    {
      "epoch": 0.34780185327109314,
      "grad_norm": 0.5289834141731262,
      "learning_rate": 3.188118899259213e-05,
      "loss": 0.12551307678222656,
      "step": 2496
    },
    {
      "epoch": 0.3479411969623075,
      "grad_norm": 0.3810412287712097,
      "learning_rate": 3.1873604286531415e-05,
      "loss": 0.07194805145263672,
      "step": 2497
    },
    {
      "epoch": 0.3480805406535219,
      "grad_norm": 0.384438157081604,
      "learning_rate": 3.18660169424269e-05,
      "loss": 0.10820770263671875,
      "step": 2498
    },
    {
      "epoch": 0.3482198843447363,
      "grad_norm": 0.31853631138801575,
      "learning_rate": 3.185842696196434e-05,
      "loss": 0.0894460678100586,
      "step": 2499
    },
    {
      "epoch": 0.34835922803595065,
      "grad_norm": 0.30965960025787354,
      "learning_rate": 3.185083434683003e-05,
      "loss": 0.07777786254882812,
      "step": 2500
    },
    {
      "epoch": 0.34849857172716503,
      "grad_norm": 0.45966655015945435,
      "learning_rate": 3.184323909871089e-05,
      "loss": 0.10226249694824219,
      "step": 2501
    },
    {
      "epoch": 0.3486379154183794,
      "grad_norm": 0.49824199080467224,
      "learning_rate": 3.1835641219294414e-05,
      "loss": 0.10141277313232422,
      "step": 2502
    },
    {
      "epoch": 0.34877725910959384,
      "grad_norm": 0.4209989905357361,
      "learning_rate": 3.182804071026867e-05,
      "loss": 0.10384941101074219,
      "step": 2503
    },
    {
      "epoch": 0.3489166028008082,
      "grad_norm": 0.4403812885284424,
      "learning_rate": 3.1820437573322306e-05,
      "loss": 0.09401512145996094,
      "step": 2504
    },
    {
      "epoch": 0.3490559464920226,
      "grad_norm": 0.4102729260921478,
      "learning_rate": 3.181283181014459e-05,
      "loss": 0.09067916870117188,
      "step": 2505
    },
    {
      "epoch": 0.349195290183237,
      "grad_norm": 0.2255939096212387,
      "learning_rate": 3.1805223422425334e-05,
      "loss": 0.06485366821289062,
      "step": 2506
    },
    {
      "epoch": 0.34933463387445135,
      "grad_norm": 0.4002390503883362,
      "learning_rate": 3.179761241185495e-05,
      "loss": 0.11719512939453125,
      "step": 2507
    },
    {
      "epoch": 0.3494739775656657,
      "grad_norm": 0.2519642114639282,
      "learning_rate": 3.178999878012443e-05,
      "loss": 0.07923126220703125,
      "step": 2508
    },
    {
      "epoch": 0.3496133212568801,
      "grad_norm": 0.3779487609863281,
      "learning_rate": 3.178238252892536e-05,
      "loss": 0.08405208587646484,
      "step": 2509
    },
    {
      "epoch": 0.3497526649480945,
      "grad_norm": 0.4608590304851532,
      "learning_rate": 3.177476365994989e-05,
      "loss": 0.08600425720214844,
      "step": 2510
    },
    {
      "epoch": 0.34989200863930886,
      "grad_norm": 0.513778805732727,
      "learning_rate": 3.1767142174890746e-05,
      "loss": 0.11716270446777344,
      "step": 2511
    },
    {
      "epoch": 0.35003135233052324,
      "grad_norm": 0.4361875057220459,
      "learning_rate": 3.175951807544126e-05,
      "loss": 0.0967559814453125,
      "step": 2512
    },
    {
      "epoch": 0.3501706960217376,
      "grad_norm": 0.5316446423530579,
      "learning_rate": 3.1751891363295344e-05,
      "loss": 0.10214042663574219,
      "step": 2513
    },
    {
      "epoch": 0.350310039712952,
      "grad_norm": 0.6804463267326355,
      "learning_rate": 3.1744262040147454e-05,
      "loss": 0.1665191650390625,
      "step": 2514
    },
    {
      "epoch": 0.35044938340416637,
      "grad_norm": 0.40050479769706726,
      "learning_rate": 3.173663010769267e-05,
      "loss": 0.08290290832519531,
      "step": 2515
    },
    {
      "epoch": 0.35058872709538075,
      "grad_norm": 0.22831140458583832,
      "learning_rate": 3.172899556762663e-05,
      "loss": 0.0614471435546875,
      "step": 2516
    },
    {
      "epoch": 0.3507280707865951,
      "grad_norm": 0.5256701707839966,
      "learning_rate": 3.172135842164555e-05,
      "loss": 0.1012868881225586,
      "step": 2517
    },
    {
      "epoch": 0.3508674144778095,
      "grad_norm": 0.87656170129776,
      "learning_rate": 3.171371867144624e-05,
      "loss": 0.11284303665161133,
      "step": 2518
    },
    {
      "epoch": 0.3510067581690239,
      "grad_norm": 0.5709128379821777,
      "learning_rate": 3.1706076318726056e-05,
      "loss": 0.08726882934570312,
      "step": 2519
    },
    {
      "epoch": 0.35114610186023826,
      "grad_norm": 0.908566415309906,
      "learning_rate": 3.1698431365182974e-05,
      "loss": 0.14369583129882812,
      "step": 2520
    },
    {
      "epoch": 0.35128544555145264,
      "grad_norm": 0.35034847259521484,
      "learning_rate": 3.169078381251552e-05,
      "loss": 0.06421661376953125,
      "step": 2521
    },
    {
      "epoch": 0.351424789242667,
      "grad_norm": 0.6176842451095581,
      "learning_rate": 3.168313366242281e-05,
      "loss": 0.09510040283203125,
      "step": 2522
    },
    {
      "epoch": 0.35156413293388145,
      "grad_norm": 0.4559093713760376,
      "learning_rate": 3.167548091660454e-05,
      "loss": 0.09984970092773438,
      "step": 2523
    },
    {
      "epoch": 0.3517034766250958,
      "grad_norm": 0.3195243179798126,
      "learning_rate": 3.166782557676095e-05,
      "loss": 0.08394241333007812,
      "step": 2524
    },
    {
      "epoch": 0.3518428203163102,
      "grad_norm": 0.2382056564092636,
      "learning_rate": 3.1660167644592915e-05,
      "loss": 0.07370567321777344,
      "step": 2525
    },
    {
      "epoch": 0.3519821640075246,
      "grad_norm": 0.3201528787612915,
      "learning_rate": 3.165250712180182e-05,
      "loss": 0.08929443359375,
      "step": 2526
    },
    {
      "epoch": 0.35212150769873896,
      "grad_norm": 0.6946732997894287,
      "learning_rate": 3.1644844010089686e-05,
      "loss": 0.09323501586914062,
      "step": 2527
    },
    {
      "epoch": 0.35226085138995333,
      "grad_norm": 0.3719169795513153,
      "learning_rate": 3.163717831115906e-05,
      "loss": 0.07489013671875,
      "step": 2528
    },
    {
      "epoch": 0.3524001950811677,
      "grad_norm": 0.157009094953537,
      "learning_rate": 3.1629510026713095e-05,
      "loss": 0.05513572692871094,
      "step": 2529
    },
    {
      "epoch": 0.3525395387723821,
      "grad_norm": 0.5100787281990051,
      "learning_rate": 3.162183915845551e-05,
      "loss": 0.09361648559570312,
      "step": 2530
    },
    {
      "epoch": 0.35267888246359647,
      "grad_norm": 0.6863251328468323,
      "learning_rate": 3.16141657080906e-05,
      "loss": 0.09431695938110352,
      "step": 2531
    },
    {
      "epoch": 0.35281822615481084,
      "grad_norm": 0.4269714057445526,
      "learning_rate": 3.160648967732322e-05,
      "loss": 0.08947372436523438,
      "step": 2532
    },
    {
      "epoch": 0.3529575698460252,
      "grad_norm": 0.2729628384113312,
      "learning_rate": 3.159881106785882e-05,
      "loss": 0.07842445373535156,
      "step": 2533
    },
    {
      "epoch": 0.3530969135372396,
      "grad_norm": 0.4875258505344391,
      "learning_rate": 3.15911298814034e-05,
      "loss": 0.09406280517578125,
      "step": 2534
    },
    {
      "epoch": 0.353236257228454,
      "grad_norm": 0.445156067609787,
      "learning_rate": 3.1583446119663555e-05,
      "loss": 0.09652328491210938,
      "step": 2535
    },
    {
      "epoch": 0.35337560091966835,
      "grad_norm": 0.2816588878631592,
      "learning_rate": 3.1575759784346436e-05,
      "loss": 0.0845489501953125,
      "step": 2536
    },
    {
      "epoch": 0.35351494461088273,
      "grad_norm": 0.3294815123081207,
      "learning_rate": 3.1568070877159766e-05,
      "loss": 0.07814645767211914,
      "step": 2537
    },
    {
      "epoch": 0.3536542883020971,
      "grad_norm": 0.2936975955963135,
      "learning_rate": 3.1560379399811856e-05,
      "loss": 0.08336257934570312,
      "step": 2538
    },
    {
      "epoch": 0.3537936319933115,
      "grad_norm": 0.4594877362251282,
      "learning_rate": 3.155268535401157e-05,
      "loss": 0.10181427001953125,
      "step": 2539
    },
    {
      "epoch": 0.35393297568452586,
      "grad_norm": 0.4611753523349762,
      "learning_rate": 3.1544988741468353e-05,
      "loss": 0.11417961120605469,
      "step": 2540
    },
    {
      "epoch": 0.35407231937574024,
      "grad_norm": 0.2853682339191437,
      "learning_rate": 3.153728956389221e-05,
      "loss": 0.07401466369628906,
      "step": 2541
    },
    {
      "epoch": 0.3542116630669546,
      "grad_norm": 0.4520014524459839,
      "learning_rate": 3.152958782299373e-05,
      "loss": 0.08025074005126953,
      "step": 2542
    },
    {
      "epoch": 0.35435100675816905,
      "grad_norm": 0.330563485622406,
      "learning_rate": 3.152188352048406e-05,
      "loss": 0.07129669189453125,
      "step": 2543
    },
    {
      "epoch": 0.35449035044938343,
      "grad_norm": 0.4054713547229767,
      "learning_rate": 3.1514176658074925e-05,
      "loss": 0.09200286865234375,
      "step": 2544
    },
    {
      "epoch": 0.3546296941405978,
      "grad_norm": 0.33665603399276733,
      "learning_rate": 3.15064672374786e-05,
      "loss": 0.0760507583618164,
      "step": 2545
    },
    {
      "epoch": 0.3547690378318122,
      "grad_norm": 0.49538248777389526,
      "learning_rate": 3.149875526040796e-05,
      "loss": 0.10960197448730469,
      "step": 2546
    },
    {
      "epoch": 0.35490838152302656,
      "grad_norm": 0.7118152976036072,
      "learning_rate": 3.1491040728576416e-05,
      "loss": 0.10572242736816406,
      "step": 2547
    },
    {
      "epoch": 0.35504772521424094,
      "grad_norm": 0.3852025866508484,
      "learning_rate": 3.1483323643697965e-05,
      "loss": 0.07859420776367188,
      "step": 2548
    },
    {
      "epoch": 0.3551870689054553,
      "grad_norm": 0.4282594323158264,
      "learning_rate": 3.147560400748719e-05,
      "loss": 0.09489822387695312,
      "step": 2549
    },
    {
      "epoch": 0.3553264125966697,
      "grad_norm": 0.36423978209495544,
      "learning_rate": 3.146788182165917e-05,
      "loss": 0.10332298278808594,
      "step": 2550
    },
    {
      "epoch": 0.35546575628788407,
      "grad_norm": 0.20002001523971558,
      "learning_rate": 3.146015708792964e-05,
      "loss": 0.06299591064453125,
      "step": 2551
    },
    {
      "epoch": 0.35560509997909845,
      "grad_norm": 0.3775453269481659,
      "learning_rate": 3.1452429808014845e-05,
      "loss": 0.08882904052734375,
      "step": 2552
    },
    {
      "epoch": 0.3557444436703128,
      "grad_norm": 0.5813329219818115,
      "learning_rate": 3.1444699983631604e-05,
      "loss": 0.10185813903808594,
      "step": 2553
    },
    {
      "epoch": 0.3558837873615272,
      "grad_norm": 0.6250063180923462,
      "learning_rate": 3.143696761649732e-05,
      "loss": 0.10751152038574219,
      "step": 2554
    },
    {
      "epoch": 0.3560231310527416,
      "grad_norm": 1.0059397220611572,
      "learning_rate": 3.1429232708329935e-05,
      "loss": 0.1645793914794922,
      "step": 2555
    },
    {
      "epoch": 0.35616247474395596,
      "grad_norm": 0.2661990821361542,
      "learning_rate": 3.142149526084798e-05,
      "loss": 0.06608724594116211,
      "step": 2556
    },
    {
      "epoch": 0.35630181843517034,
      "grad_norm": 0.32851940393447876,
      "learning_rate": 3.1413755275770533e-05,
      "loss": 0.06689929962158203,
      "step": 2557
    },
    {
      "epoch": 0.3564411621263847,
      "grad_norm": 0.3427017033100128,
      "learning_rate": 3.1406012754817246e-05,
      "loss": 0.11832952499389648,
      "step": 2558
    },
    {
      "epoch": 0.3565805058175991,
      "grad_norm": 0.45496290922164917,
      "learning_rate": 3.139826769970833e-05,
      "loss": 0.11744880676269531,
      "step": 2559
    },
    {
      "epoch": 0.35671984950881347,
      "grad_norm": 0.38608700037002563,
      "learning_rate": 3.139052011216456e-05,
      "loss": 0.08851432800292969,
      "step": 2560
    },
    {
      "epoch": 0.35685919320002785,
      "grad_norm": 0.6037548184394836,
      "learning_rate": 3.138276999390726e-05,
      "loss": 0.10453224182128906,
      "step": 2561
    },
    {
      "epoch": 0.3569985368912422,
      "grad_norm": 0.3369615972042084,
      "learning_rate": 3.1375017346658354e-05,
      "loss": 0.09618949890136719,
      "step": 2562
    },
    {
      "epoch": 0.3571378805824566,
      "grad_norm": 0.386778861284256,
      "learning_rate": 3.136726217214028e-05,
      "loss": 0.0967249870300293,
      "step": 2563
    },
    {
      "epoch": 0.35727722427367103,
      "grad_norm": 0.6766453981399536,
      "learning_rate": 3.1359504472076074e-05,
      "loss": 0.11552047729492188,
      "step": 2564
    },
    {
      "epoch": 0.3574165679648854,
      "grad_norm": 0.21613693237304688,
      "learning_rate": 3.1351744248189314e-05,
      "loss": 0.07839202880859375,
      "step": 2565
    },
    {
      "epoch": 0.3575559116560998,
      "grad_norm": 0.4077081084251404,
      "learning_rate": 3.134398150220415e-05,
      "loss": 0.1036834716796875,
      "step": 2566
    },
    {
      "epoch": 0.35769525534731417,
      "grad_norm": 0.4320823550224304,
      "learning_rate": 3.133621623584528e-05,
      "loss": 0.08541488647460938,
      "step": 2567
    },
    {
      "epoch": 0.35783459903852854,
      "grad_norm": 0.23426011204719543,
      "learning_rate": 3.132844845083798e-05,
      "loss": 0.07692337036132812,
      "step": 2568
    },
    {
      "epoch": 0.3579739427297429,
      "grad_norm": 0.19030626118183136,
      "learning_rate": 3.132067814890806e-05,
      "loss": 0.05902862548828125,
      "step": 2569
    },
    {
      "epoch": 0.3581132864209573,
      "grad_norm": 0.40281978249549866,
      "learning_rate": 3.1312905331781914e-05,
      "loss": 0.10067176818847656,
      "step": 2570
    },
    {
      "epoch": 0.3582526301121717,
      "grad_norm": 0.529802143573761,
      "learning_rate": 3.130513000118648e-05,
      "loss": 0.12243175506591797,
      "step": 2571
    },
    {
      "epoch": 0.35839197380338605,
      "grad_norm": 0.47500455379486084,
      "learning_rate": 3.1297352158849264e-05,
      "loss": 0.1415557861328125,
      "step": 2572
    },
    {
      "epoch": 0.35853131749460043,
      "grad_norm": 0.392120897769928,
      "learning_rate": 3.128957180649832e-05,
      "loss": 0.08834648132324219,
      "step": 2573
    },
    {
      "epoch": 0.3586706611858148,
      "grad_norm": 0.75846266746521,
      "learning_rate": 3.128178894586226e-05,
      "loss": 0.13341522216796875,
      "step": 2574
    },
    {
      "epoch": 0.3588100048770292,
      "grad_norm": 0.6151486039161682,
      "learning_rate": 3.1274003578670264e-05,
      "loss": 0.13898277282714844,
      "step": 2575
    },
    {
      "epoch": 0.35894934856824356,
      "grad_norm": 0.3831477761268616,
      "learning_rate": 3.126621570665207e-05,
      "loss": 0.09893417358398438,
      "step": 2576
    },
    {
      "epoch": 0.35908869225945794,
      "grad_norm": 0.4112849235534668,
      "learning_rate": 3.125842533153796e-05,
      "loss": 0.08327484130859375,
      "step": 2577
    },
    {
      "epoch": 0.3592280359506723,
      "grad_norm": 0.6142802238464355,
      "learning_rate": 3.1250632455058764e-05,
      "loss": 0.10808181762695312,
      "step": 2578
    },
    {
      "epoch": 0.3593673796418867,
      "grad_norm": 0.5233549475669861,
      "learning_rate": 3.12428370789459e-05,
      "loss": 0.0998220443725586,
      "step": 2579
    },
    {
      "epoch": 0.3595067233331011,
      "grad_norm": 0.7534226775169373,
      "learning_rate": 3.1235039204931316e-05,
      "loss": 0.10317039489746094,
      "step": 2580
    },
    {
      "epoch": 0.35964606702431545,
      "grad_norm": 0.3940846621990204,
      "learning_rate": 3.122723883474752e-05,
      "loss": 0.08102226257324219,
      "step": 2581
    },
    {
      "epoch": 0.35978541071552983,
      "grad_norm": 1.2365977764129639,
      "learning_rate": 3.1219435970127574e-05,
      "loss": 0.1277332305908203,
      "step": 2582
    },
    {
      "epoch": 0.3599247544067442,
      "grad_norm": 0.5809354782104492,
      "learning_rate": 3.12116306128051e-05,
      "loss": 0.0838613510131836,
      "step": 2583
    },
    {
      "epoch": 0.36006409809795864,
      "grad_norm": 0.3610229194164276,
      "learning_rate": 3.1203822764514274e-05,
      "loss": 0.08929634094238281,
      "step": 2584
    },
    {
      "epoch": 0.360203441789173,
      "grad_norm": 0.7103058695793152,
      "learning_rate": 3.1196012426989814e-05,
      "loss": 0.1682300567626953,
      "step": 2585
    },
    {
      "epoch": 0.3603427854803874,
      "grad_norm": 0.43314695358276367,
      "learning_rate": 3.1188199601967e-05,
      "loss": 0.10078811645507812,
      "step": 2586
    },
    {
      "epoch": 0.3604821291716018,
      "grad_norm": 0.32209959626197815,
      "learning_rate": 3.118038429118167e-05,
      "loss": 0.07481765747070312,
      "step": 2587
    },
    {
      "epoch": 0.36062147286281615,
      "grad_norm": 0.3271544277667999,
      "learning_rate": 3.1172566496370205e-05,
      "loss": 0.0765218734741211,
      "step": 2588
    },
    {
      "epoch": 0.36076081655403053,
      "grad_norm": 0.3193660080432892,
      "learning_rate": 3.116474621926953e-05,
      "loss": 0.09214591979980469,
      "step": 2589
    },
    {
      "epoch": 0.3609001602452449,
      "grad_norm": 0.3044019937515259,
      "learning_rate": 3.115692346161715e-05,
      "loss": 0.07052040100097656,
      "step": 2590
    },
    {
      "epoch": 0.3610395039364593,
      "grad_norm": 0.2673049867153168,
      "learning_rate": 3.1149098225151086e-05,
      "loss": 0.07297134399414062,
      "step": 2591
    },
    {
      "epoch": 0.36117884762767366,
      "grad_norm": 0.46576112508773804,
      "learning_rate": 3.114127051160994e-05,
      "loss": 0.10613155364990234,
      "step": 2592
    },
    {
      "epoch": 0.36131819131888804,
      "grad_norm": 0.2673509120941162,
      "learning_rate": 3.1133440322732846e-05,
      "loss": 0.06754302978515625,
      "step": 2593
    },
    {
      "epoch": 0.3614575350101024,
      "grad_norm": 0.44114530086517334,
      "learning_rate": 3.112560766025949e-05,
      "loss": 0.08633804321289062,
      "step": 2594
    },
    {
      "epoch": 0.3615968787013168,
      "grad_norm": 0.3208615481853485,
      "learning_rate": 3.1117772525930115e-05,
      "loss": 0.09144401550292969,
      "step": 2595
    },
    {
      "epoch": 0.36173622239253117,
      "grad_norm": 0.4067465364933014,
      "learning_rate": 3.11099349214855e-05,
      "loss": 0.09580802917480469,
      "step": 2596
    },
    {
      "epoch": 0.36187556608374555,
      "grad_norm": 0.4733952581882477,
      "learning_rate": 3.1102094848667e-05,
      "loss": 0.09858512878417969,
      "step": 2597
    },
    {
      "epoch": 0.3620149097749599,
      "grad_norm": 0.473481148481369,
      "learning_rate": 3.109425230921649e-05,
      "loss": 0.07828235626220703,
      "step": 2598
    },
    {
      "epoch": 0.3621542534661743,
      "grad_norm": 0.21305640041828156,
      "learning_rate": 3.10864073048764e-05,
      "loss": 0.05056571960449219,
      "step": 2599
    },
    {
      "epoch": 0.3622935971573887,
      "grad_norm": 0.8415384888648987,
      "learning_rate": 3.107855983738971e-05,
      "loss": 0.12586402893066406,
      "step": 2600
    },
    {
      "epoch": 0.36243294084860306,
      "grad_norm": 0.3478691875934601,
      "learning_rate": 3.107070990849995e-05,
      "loss": 0.08661651611328125,
      "step": 2601
    },
    {
      "epoch": 0.36257228453981744,
      "grad_norm": 0.6279574036598206,
      "learning_rate": 3.1062857519951215e-05,
      "loss": 0.11527538299560547,
      "step": 2602
    },
    {
      "epoch": 0.3627116282310318,
      "grad_norm": 0.7989532947540283,
      "learning_rate": 3.105500267348809e-05,
      "loss": 0.12757110595703125,
      "step": 2603
    },
    {
      "epoch": 0.36285097192224625,
      "grad_norm": 0.39449018239974976,
      "learning_rate": 3.1047145370855764e-05,
      "loss": 0.08160972595214844,
      "step": 2604
    },
    {
      "epoch": 0.3629903156134606,
      "grad_norm": 0.5415140986442566,
      "learning_rate": 3.103928561379996e-05,
      "loss": 0.1280841827392578,
      "step": 2605
    },
    {
      "epoch": 0.363129659304675,
      "grad_norm": 0.35626840591430664,
      "learning_rate": 3.103142340406691e-05,
      "loss": 0.07424545288085938,
      "step": 2606
    },
    {
      "epoch": 0.3632690029958894,
      "grad_norm": 0.6165229678153992,
      "learning_rate": 3.102355874340343e-05,
      "loss": 0.07974624633789062,
      "step": 2607
    },
    {
      "epoch": 0.36340834668710376,
      "grad_norm": 0.5115894079208374,
      "learning_rate": 3.101569163355688e-05,
      "loss": 0.11807632446289062,
      "step": 2608
    },
    {
      "epoch": 0.36354769037831813,
      "grad_norm": 0.48784202337265015,
      "learning_rate": 3.100782207627513e-05,
      "loss": 0.09202194213867188,
      "step": 2609
    },
    {
      "epoch": 0.3636870340695325,
      "grad_norm": 0.5993459820747375,
      "learning_rate": 3.099995007330664e-05,
      "loss": 0.09099578857421875,
      "step": 2610
    },
    {
      "epoch": 0.3638263777607469,
      "grad_norm": 0.5753583908081055,
      "learning_rate": 3.099207562640037e-05,
      "loss": 0.10274505615234375,
      "step": 2611
    },
    {
      "epoch": 0.36396572145196127,
      "grad_norm": 0.7421638369560242,
      "learning_rate": 3.098419873730585e-05,
      "loss": 0.10706329345703125,
      "step": 2612
    },
    {
      "epoch": 0.36410506514317564,
      "grad_norm": 0.3109113872051239,
      "learning_rate": 3.097631940777314e-05,
      "loss": 0.08759498596191406,
      "step": 2613
    },
    {
      "epoch": 0.36424440883439,
      "grad_norm": 0.42095738649368286,
      "learning_rate": 3.096843763955285e-05,
      "loss": 0.08941078186035156,
      "step": 2614
    },
    {
      "epoch": 0.3643837525256044,
      "grad_norm": 0.23949962854385376,
      "learning_rate": 3.096055343439614e-05,
      "loss": 0.07265281677246094,
      "step": 2615
    },
    {
      "epoch": 0.3645230962168188,
      "grad_norm": 0.4515247941017151,
      "learning_rate": 3.095266679405468e-05,
      "loss": 0.09041976928710938,
      "step": 2616
    },
    {
      "epoch": 0.36466243990803315,
      "grad_norm": 0.5558367371559143,
      "learning_rate": 3.094477772028072e-05,
      "loss": 0.09982490539550781,
      "step": 2617
    },
    {
      "epoch": 0.36480178359924753,
      "grad_norm": 0.338630348443985,
      "learning_rate": 3.0936886214827024e-05,
      "loss": 0.08219337463378906,
      "step": 2618
    },
    {
      "epoch": 0.3649411272904619,
      "grad_norm": 0.3557131886482239,
      "learning_rate": 3.09289922794469e-05,
      "loss": 0.07383155822753906,
      "step": 2619
    },
    {
      "epoch": 0.3650804709816763,
      "grad_norm": 0.43100714683532715,
      "learning_rate": 3.092109591589421e-05,
      "loss": 0.09056472778320312,
      "step": 2620
    },
    {
      "epoch": 0.36521981467289066,
      "grad_norm": 0.709992527961731,
      "learning_rate": 3.091319712592333e-05,
      "loss": 0.10838508605957031,
      "step": 2621
    },
    {
      "epoch": 0.36535915836410504,
      "grad_norm": 0.7200655341148376,
      "learning_rate": 3.0905295911289216e-05,
      "loss": 0.11471366882324219,
      "step": 2622
    },
    {
      "epoch": 0.3654985020553194,
      "grad_norm": 0.27600622177124023,
      "learning_rate": 3.089739227374732e-05,
      "loss": 0.06329917907714844,
      "step": 2623
    },
    {
      "epoch": 0.36563784574653385,
      "grad_norm": 0.35377442836761475,
      "learning_rate": 3.088948621505364e-05,
      "loss": 0.0729217529296875,
      "step": 2624
    },
    {
      "epoch": 0.36577718943774823,
      "grad_norm": 0.6577832698822021,
      "learning_rate": 3.088157773696474e-05,
      "loss": 0.10921859741210938,
      "step": 2625
    },
    {
      "epoch": 0.3659165331289626,
      "grad_norm": 0.40780776739120483,
      "learning_rate": 3.08736668412377e-05,
      "loss": 0.09307670593261719,
      "step": 2626
    },
    {
      "epoch": 0.366055876820177,
      "grad_norm": 0.6203763484954834,
      "learning_rate": 3.0865753529630135e-05,
      "loss": 0.12179088592529297,
      "step": 2627
    },
    {
      "epoch": 0.36619522051139136,
      "grad_norm": 0.8463844656944275,
      "learning_rate": 3.085783780390021e-05,
      "loss": 0.14350128173828125,
      "step": 2628
    },
    {
      "epoch": 0.36633456420260574,
      "grad_norm": 0.4993295669555664,
      "learning_rate": 3.0849919665806605e-05,
      "loss": 0.08677291870117188,
      "step": 2629
    },
    {
      "epoch": 0.3664739078938201,
      "grad_norm": 0.6247550249099731,
      "learning_rate": 3.0841999117108564e-05,
      "loss": 0.09033966064453125,
      "step": 2630
    },
    {
      "epoch": 0.3666132515850345,
      "grad_norm": 0.3499354422092438,
      "learning_rate": 3.0834076159565847e-05,
      "loss": 0.082977294921875,
      "step": 2631
    },
    {
      "epoch": 0.36675259527624887,
      "grad_norm": 0.3901529312133789,
      "learning_rate": 3.082615079493876e-05,
      "loss": 0.09166908264160156,
      "step": 2632
    },
    {
      "epoch": 0.36689193896746325,
      "grad_norm": 0.5176382660865784,
      "learning_rate": 3.081822302498812e-05,
      "loss": 0.10700225830078125,
      "step": 2633
    },
    {
      "epoch": 0.3670312826586776,
      "grad_norm": 0.37949952483177185,
      "learning_rate": 3.081029285147531e-05,
      "loss": 0.12396240234375,
      "step": 2634
    },
    {
      "epoch": 0.367170626349892,
      "grad_norm": 0.16940274834632874,
      "learning_rate": 3.080236027616224e-05,
      "loss": 0.05377006530761719,
      "step": 2635
    },
    {
      "epoch": 0.3673099700411064,
      "grad_norm": 1.1633319854736328,
      "learning_rate": 3.079442530081133e-05,
      "loss": 0.13844871520996094,
      "step": 2636
    },
    {
      "epoch": 0.36744931373232076,
      "grad_norm": 0.5803998112678528,
      "learning_rate": 3.078648792718556e-05,
      "loss": 0.10044097900390625,
      "step": 2637
    },
    {
      "epoch": 0.36758865742353514,
      "grad_norm": 0.5642547011375427,
      "learning_rate": 3.0778548157048434e-05,
      "loss": 0.11301994323730469,
      "step": 2638
    },
    {
      "epoch": 0.3677280011147495,
      "grad_norm": 0.32669633626937866,
      "learning_rate": 3.0770605992163986e-05,
      "loss": 0.07831001281738281,
      "step": 2639
    },
    {
      "epoch": 0.3678673448059639,
      "grad_norm": 0.35218387842178345,
      "learning_rate": 3.076266143429679e-05,
      "loss": 0.093170166015625,
      "step": 2640
    },
    {
      "epoch": 0.36800668849717827,
      "grad_norm": 0.4158170819282532,
      "learning_rate": 3.0754714485211925e-05,
      "loss": 0.0768594741821289,
      "step": 2641
    },
    {
      "epoch": 0.36814603218839265,
      "grad_norm": 0.45826539397239685,
      "learning_rate": 3.0746765146675043e-05,
      "loss": 0.08993148803710938,
      "step": 2642
    },
    {
      "epoch": 0.368285375879607,
      "grad_norm": 0.5935287475585938,
      "learning_rate": 3.0738813420452295e-05,
      "loss": 0.1072225570678711,
      "step": 2643
    },
    {
      "epoch": 0.36842471957082146,
      "grad_norm": 0.22847943007946014,
      "learning_rate": 3.073085930831038e-05,
      "loss": 0.06132316589355469,
      "step": 2644
    },
    {
      "epoch": 0.36856406326203583,
      "grad_norm": 0.31108322739601135,
      "learning_rate": 3.072290281201652e-05,
      "loss": 0.08017349243164062,
      "step": 2645
    },
    {
      "epoch": 0.3687034069532502,
      "grad_norm": 0.4615246653556824,
      "learning_rate": 3.071494393333846e-05,
      "loss": 0.0951690673828125,
      "step": 2646
    },
    {
      "epoch": 0.3688427506444646,
      "grad_norm": 0.32167354226112366,
      "learning_rate": 3.0706982674044486e-05,
      "loss": 0.087005615234375,
      "step": 2647
    },
    {
      "epoch": 0.36898209433567897,
      "grad_norm": 0.4453098475933075,
      "learning_rate": 3.06990190359034e-05,
      "loss": 0.09072494506835938,
      "step": 2648
    },
    {
      "epoch": 0.36912143802689334,
      "grad_norm": 0.41377750039100647,
      "learning_rate": 3.069105302068455e-05,
      "loss": 0.09400367736816406,
      "step": 2649
    },
    {
      "epoch": 0.3692607817181077,
      "grad_norm": 0.42259132862091064,
      "learning_rate": 3.06830846301578e-05,
      "loss": 0.07926177978515625,
      "step": 2650
    },
    {
      "epoch": 0.3694001254093221,
      "grad_norm": 0.4374361038208008,
      "learning_rate": 3.067511386609354e-05,
      "loss": 0.08720970153808594,
      "step": 2651
    },
    {
      "epoch": 0.3695394691005365,
      "grad_norm": 0.353947252035141,
      "learning_rate": 3.0667140730262706e-05,
      "loss": 0.0829019546508789,
      "step": 2652
    },
    {
      "epoch": 0.36967881279175085,
      "grad_norm": 0.43192946910858154,
      "learning_rate": 3.065916522443673e-05,
      "loss": 0.09763145446777344,
      "step": 2653
    },
    {
      "epoch": 0.36981815648296523,
      "grad_norm": 0.4580514430999756,
      "learning_rate": 3.06511873503876e-05,
      "loss": 0.10834121704101562,
      "step": 2654
    },
    {
      "epoch": 0.3699575001741796,
      "grad_norm": 0.7476041316986084,
      "learning_rate": 3.0643207109887804e-05,
      "loss": 0.1402416229248047,
      "step": 2655
    },
    {
      "epoch": 0.370096843865394,
      "grad_norm": 0.31842339038848877,
      "learning_rate": 3.063522450471038e-05,
      "loss": 0.07881927490234375,
      "step": 2656
    },
    {
      "epoch": 0.37023618755660836,
      "grad_norm": 0.41811904311180115,
      "learning_rate": 3.062723953662888e-05,
      "loss": 0.10394477844238281,
      "step": 2657
    },
    {
      "epoch": 0.37037553124782274,
      "grad_norm": 0.48975223302841187,
      "learning_rate": 3.061925220741738e-05,
      "loss": 0.06064033508300781,
      "step": 2658
    },
    {
      "epoch": 0.3705148749390371,
      "grad_norm": 1.1146239042282104,
      "learning_rate": 3.0611262518850464e-05,
      "loss": 0.11457061767578125,
      "step": 2659
    },
    {
      "epoch": 0.3706542186302515,
      "grad_norm": 0.5867258906364441,
      "learning_rate": 3.0603270472703294e-05,
      "loss": 0.11674690246582031,
      "step": 2660
    },
    {
      "epoch": 0.3707935623214659,
      "grad_norm": 0.2562991976737976,
      "learning_rate": 3.05952760707515e-05,
      "loss": 0.07279300689697266,
      "step": 2661
    },
    {
      "epoch": 0.37093290601268025,
      "grad_norm": 0.674834132194519,
      "learning_rate": 3.0587279314771253e-05,
      "loss": 0.08330154418945312,
      "step": 2662
    },
    {
      "epoch": 0.37107224970389463,
      "grad_norm": 0.49679428339004517,
      "learning_rate": 3.057928020653925e-05,
      "loss": 0.08529090881347656,
      "step": 2663
    },
    {
      "epoch": 0.37121159339510906,
      "grad_norm": 0.7925440669059753,
      "learning_rate": 3.057127874783272e-05,
      "loss": 0.11136245727539062,
      "step": 2664
    },
    {
      "epoch": 0.37135093708632344,
      "grad_norm": 0.24290738999843597,
      "learning_rate": 3.0563274940429404e-05,
      "loss": 0.07623291015625,
      "step": 2665
    },
    {
      "epoch": 0.3714902807775378,
      "grad_norm": 0.7929752469062805,
      "learning_rate": 3.055526878610755e-05,
      "loss": 0.13171768188476562,
      "step": 2666
    },
    {
      "epoch": 0.3716296244687522,
      "grad_norm": 0.9059690237045288,
      "learning_rate": 3.054726028664595e-05,
      "loss": 0.1026153564453125,
      "step": 2667
    },
    {
      "epoch": 0.3717689681599666,
      "grad_norm": 0.32353541254997253,
      "learning_rate": 3.053924944382393e-05,
      "loss": 0.08931732177734375,
      "step": 2668
    },
    {
      "epoch": 0.37190831185118095,
      "grad_norm": 0.6103008985519409,
      "learning_rate": 3.053123625942128e-05,
      "loss": 0.10427093505859375,
      "step": 2669
    },
    {
      "epoch": 0.37204765554239533,
      "grad_norm": 0.599092423915863,
      "learning_rate": 3.052322073521837e-05,
      "loss": 0.08529090881347656,
      "step": 2670
    },
    {
      "epoch": 0.3721869992336097,
      "grad_norm": 0.6153716444969177,
      "learning_rate": 3.0515202872996067e-05,
      "loss": 0.09033584594726562,
      "step": 2671
    },
    {
      "epoch": 0.3723263429248241,
      "grad_norm": 0.24148611724376678,
      "learning_rate": 3.050718267453575e-05,
      "loss": 0.061034202575683594,
      "step": 2672
    },
    {
      "epoch": 0.37246568661603846,
      "grad_norm": 1.1671512126922607,
      "learning_rate": 3.0499160141619323e-05,
      "loss": 0.10204887390136719,
      "step": 2673
    },
    {
      "epoch": 0.37260503030725284,
      "grad_norm": 0.2623699903488159,
      "learning_rate": 3.049113527602922e-05,
      "loss": 0.07540130615234375,
      "step": 2674
    },
    {
      "epoch": 0.3727443739984672,
      "grad_norm": 0.2784218192100525,
      "learning_rate": 3.0483108079548366e-05,
      "loss": 0.09188270568847656,
      "step": 2675
    },
    {
      "epoch": 0.3728837176896816,
      "grad_norm": 0.3810289204120636,
      "learning_rate": 3.0475078553960234e-05,
      "loss": 0.09445762634277344,
      "step": 2676
    },
    {
      "epoch": 0.37302306138089597,
      "grad_norm": 0.7802804708480835,
      "learning_rate": 3.0467046701048795e-05,
      "loss": 0.13504409790039062,
      "step": 2677
    },
    {
      "epoch": 0.37316240507211035,
      "grad_norm": 0.5049695372581482,
      "learning_rate": 3.0459012522598553e-05,
      "loss": 0.1236572265625,
      "step": 2678
    },
    {
      "epoch": 0.3733017487633247,
      "grad_norm": 0.208942249417305,
      "learning_rate": 3.04509760203945e-05,
      "loss": 0.07639312744140625,
      "step": 2679
    },
    {
      "epoch": 0.3734410924545391,
      "grad_norm": 0.36450180411338806,
      "learning_rate": 3.0442937196222186e-05,
      "loss": 0.06351470947265625,
      "step": 2680
    },
    {
      "epoch": 0.3735804361457535,
      "grad_norm": 0.5124438405036926,
      "learning_rate": 3.043489605186764e-05,
      "loss": 0.06998443603515625,
      "step": 2681
    },
    {
      "epoch": 0.37371977983696786,
      "grad_norm": 0.7344912886619568,
      "learning_rate": 3.0426852589117422e-05,
      "loss": 0.11066246032714844,
      "step": 2682
    },
    {
      "epoch": 0.37385912352818224,
      "grad_norm": 0.4974328279495239,
      "learning_rate": 3.041880680975861e-05,
      "loss": 0.12248420715332031,
      "step": 2683
    },
    {
      "epoch": 0.37399846721939667,
      "grad_norm": 0.3984052836894989,
      "learning_rate": 3.041075871557879e-05,
      "loss": 0.08704471588134766,
      "step": 2684
    },
    {
      "epoch": 0.37413781091061105,
      "grad_norm": 0.29928821325302124,
      "learning_rate": 3.0402708308366066e-05,
      "loss": 0.07342529296875,
      "step": 2685
    },
    {
      "epoch": 0.3742771546018254,
      "grad_norm": 0.3556903898715973,
      "learning_rate": 3.039465558990905e-05,
      "loss": 0.08797454833984375,
      "step": 2686
    },
    {
      "epoch": 0.3744164982930398,
      "grad_norm": 0.5514016151428223,
      "learning_rate": 3.038660056199688e-05,
      "loss": 0.11268043518066406,
      "step": 2687
    },
    {
      "epoch": 0.3745558419842542,
      "grad_norm": 0.7045251131057739,
      "learning_rate": 3.037854322641919e-05,
      "loss": 0.11430931091308594,
      "step": 2688
    },
    {
      "epoch": 0.37469518567546856,
      "grad_norm": 0.35879555344581604,
      "learning_rate": 3.0370483584966144e-05,
      "loss": 0.07147026062011719,
      "step": 2689
    },
    {
      "epoch": 0.37483452936668293,
      "grad_norm": 0.6388649344444275,
      "learning_rate": 3.036242163942841e-05,
      "loss": 0.10208892822265625,
      "step": 2690
    },
    {
      "epoch": 0.3749738730578973,
      "grad_norm": 0.3103763163089752,
      "learning_rate": 3.035435739159716e-05,
      "loss": 0.09404563903808594,
      "step": 2691
    },
    {
      "epoch": 0.3751132167491117,
      "grad_norm": 0.6099969148635864,
      "learning_rate": 3.03462908432641e-05,
      "loss": 0.11745262145996094,
      "step": 2692
    },
    {
      "epoch": 0.37525256044032607,
      "grad_norm": 0.5519262552261353,
      "learning_rate": 3.033822199622142e-05,
      "loss": 0.10070419311523438,
      "step": 2693
    },
    {
      "epoch": 0.37539190413154044,
      "grad_norm": 0.3458658456802368,
      "learning_rate": 3.033015085226184e-05,
      "loss": 0.09826374053955078,
      "step": 2694
    },
    {
      "epoch": 0.3755312478227548,
      "grad_norm": 0.345923513174057,
      "learning_rate": 3.0322077413178578e-05,
      "loss": 0.07012939453125,
      "step": 2695
    },
    {
      "epoch": 0.3756705915139692,
      "grad_norm": 0.8151650428771973,
      "learning_rate": 3.0314001680765375e-05,
      "loss": 0.1175537109375,
      "step": 2696
    },
    {
      "epoch": 0.3758099352051836,
      "grad_norm": 0.33849337697029114,
      "learning_rate": 3.0305923656816473e-05,
      "loss": 0.07731914520263672,
      "step": 2697
    },
    {
      "epoch": 0.37594927889639795,
      "grad_norm": 0.714036226272583,
      "learning_rate": 3.0297843343126617e-05,
      "loss": 0.09450340270996094,
      "step": 2698
    },
    {
      "epoch": 0.37608862258761233,
      "grad_norm": 0.3599410653114319,
      "learning_rate": 3.0289760741491077e-05,
      "loss": 0.080169677734375,
      "step": 2699
    },
    {
      "epoch": 0.3762279662788267,
      "grad_norm": 0.2828015685081482,
      "learning_rate": 3.028167585370562e-05,
      "loss": 0.08306121826171875,
      "step": 2700
    },
    {
      "epoch": 0.3763673099700411,
      "grad_norm": 0.8141335248947144,
      "learning_rate": 3.0273588681566523e-05,
      "loss": 0.16182994842529297,
      "step": 2701
    },
    {
      "epoch": 0.37650665366125546,
      "grad_norm": 0.2689749598503113,
      "learning_rate": 3.026549922687057e-05,
      "loss": 0.08111095428466797,
      "step": 2702
    },
    {
      "epoch": 0.37664599735246984,
      "grad_norm": 0.9530807137489319,
      "learning_rate": 3.0257407491415053e-05,
      "loss": 0.10832405090332031,
      "step": 2703
    },
    {
      "epoch": 0.3767853410436843,
      "grad_norm": 0.4666097164154053,
      "learning_rate": 3.0249313476997772e-05,
      "loss": 0.08621978759765625,
      "step": 2704
    },
    {
      "epoch": 0.37692468473489865,
      "grad_norm": 0.5367114543914795,
      "learning_rate": 3.0241217185417034e-05,
      "loss": 0.10029792785644531,
      "step": 2705
    },
    {
      "epoch": 0.37706402842611303,
      "grad_norm": 0.3195343017578125,
      "learning_rate": 3.023311861847165e-05,
      "loss": 0.09476089477539062,
      "step": 2706
    },
    {
      "epoch": 0.3772033721173274,
      "grad_norm": 0.4547435939311981,
      "learning_rate": 3.0225017777960927e-05,
      "loss": 0.08100509643554688,
      "step": 2707
    },
    {
      "epoch": 0.3773427158085418,
      "grad_norm": 0.45441633462905884,
      "learning_rate": 3.0216914665684705e-05,
      "loss": 0.09466934204101562,
      "step": 2708
    },
    {
      "epoch": 0.37748205949975616,
      "grad_norm": 0.3753572702407837,
      "learning_rate": 3.020880928344329e-05,
      "loss": 0.07953548431396484,
      "step": 2709
    },
    {
      "epoch": 0.37762140319097054,
      "grad_norm": 0.8422170877456665,
      "learning_rate": 3.0200701633037534e-05,
      "loss": 0.1198434829711914,
      "step": 2710
    },
    {
      "epoch": 0.3777607468821849,
      "grad_norm": 0.4495704174041748,
      "learning_rate": 3.0192591716268755e-05,
      "loss": 0.09206199645996094,
      "step": 2711
    },
    {
      "epoch": 0.3779000905733993,
      "grad_norm": 0.38908272981643677,
      "learning_rate": 3.0184479534938797e-05,
      "loss": 0.10435104370117188,
      "step": 2712
    },
    {
      "epoch": 0.37803943426461367,
      "grad_norm": 1.0148074626922607,
      "learning_rate": 3.0176365090850005e-05,
      "loss": 0.15790557861328125,
      "step": 2713
    },
    {
      "epoch": 0.37817877795582805,
      "grad_norm": 0.3954848051071167,
      "learning_rate": 3.0168248385805223e-05,
      "loss": 0.11053276062011719,
      "step": 2714
    },
    {
      "epoch": 0.3783181216470424,
      "grad_norm": 0.2946760654449463,
      "learning_rate": 3.0160129421607792e-05,
      "loss": 0.07213020324707031,
      "step": 2715
    },
    {
      "epoch": 0.3784574653382568,
      "grad_norm": 0.5702301263809204,
      "learning_rate": 3.015200820006156e-05,
      "loss": 0.09881401062011719,
      "step": 2716
    },
    {
      "epoch": 0.3785968090294712,
      "grad_norm": 0.6966204643249512,
      "learning_rate": 3.014388472297088e-05,
      "loss": 0.09740638732910156,
      "step": 2717
    },
    {
      "epoch": 0.37873615272068556,
      "grad_norm": 0.45229408144950867,
      "learning_rate": 3.013575899214061e-05,
      "loss": 0.10313606262207031,
      "step": 2718
    },
    {
      "epoch": 0.37887549641189994,
      "grad_norm": 0.5059126019477844,
      "learning_rate": 3.0127631009376093e-05,
      "loss": 0.11496162414550781,
      "step": 2719
    },
    {
      "epoch": 0.3790148401031143,
      "grad_norm": 0.665940523147583,
      "learning_rate": 3.011950077648318e-05,
      "loss": 0.13916778564453125,
      "step": 2720
    },
    {
      "epoch": 0.3791541837943287,
      "grad_norm": 0.506413996219635,
      "learning_rate": 3.0111368295268225e-05,
      "loss": 0.08782577514648438,
      "step": 2721
    },
    {
      "epoch": 0.37929352748554307,
      "grad_norm": 0.2189047634601593,
      "learning_rate": 3.0103233567538086e-05,
      "loss": 0.0667104721069336,
      "step": 2722
    },
    {
      "epoch": 0.37943287117675745,
      "grad_norm": 0.3517168462276459,
      "learning_rate": 3.009509659510011e-05,
      "loss": 0.08195877075195312,
      "step": 2723
    },
    {
      "epoch": 0.3795722148679719,
      "grad_norm": 0.23586399853229523,
      "learning_rate": 3.008695737976214e-05,
      "loss": 0.06647872924804688,
      "step": 2724
    },
    {
      "epoch": 0.37971155855918626,
      "grad_norm": 0.6002417802810669,
      "learning_rate": 3.0078815923332532e-05,
      "loss": 0.09085273742675781,
      "step": 2725
    },
    {
      "epoch": 0.37985090225040063,
      "grad_norm": 0.5633906126022339,
      "learning_rate": 3.007067222762013e-05,
      "loss": 0.10212516784667969,
      "step": 2726
    },
    {
      "epoch": 0.379990245941615,
      "grad_norm": 0.5210935473442078,
      "learning_rate": 3.0062526294434273e-05,
      "loss": 0.10518836975097656,
      "step": 2727
    },
    {
      "epoch": 0.3801295896328294,
      "grad_norm": 0.37298083305358887,
      "learning_rate": 3.005437812558481e-05,
      "loss": 0.10016632080078125,
      "step": 2728
    },
    {
      "epoch": 0.38026893332404377,
      "grad_norm": 0.6972285509109497,
      "learning_rate": 3.004622772288207e-05,
      "loss": 0.10857200622558594,
      "step": 2729
    },
    {
      "epoch": 0.38040827701525814,
      "grad_norm": 0.4334034323692322,
      "learning_rate": 3.003807508813689e-05,
      "loss": 0.07750511169433594,
      "step": 2730
    },
    {
      "epoch": 0.3805476207064725,
      "grad_norm": 0.6945555806159973,
      "learning_rate": 3.00299202231606e-05,
      "loss": 0.09650707244873047,
      "step": 2731
    },
    {
      "epoch": 0.3806869643976869,
      "grad_norm": 0.41155219078063965,
      "learning_rate": 3.0021763129765024e-05,
      "loss": 0.08482933044433594,
      "step": 2732
    },
    {
      "epoch": 0.3808263080889013,
      "grad_norm": 0.5360231995582581,
      "learning_rate": 3.0013603809762473e-05,
      "loss": 0.10396766662597656,
      "step": 2733
    },
    {
      "epoch": 0.38096565178011566,
      "grad_norm": 0.49952879548072815,
      "learning_rate": 3.0005442264965778e-05,
      "loss": 0.10853004455566406,
      "step": 2734
    },
    {
      "epoch": 0.38110499547133003,
      "grad_norm": 0.31619298458099365,
      "learning_rate": 2.9997278497188236e-05,
      "loss": 0.09268379211425781,
      "step": 2735
    },
    {
      "epoch": 0.3812443391625444,
      "grad_norm": 0.3898692727088928,
      "learning_rate": 2.9989112508243655e-05,
      "loss": 0.08002281188964844,
      "step": 2736
    },
    {
      "epoch": 0.3813836828537588,
      "grad_norm": 0.5013592839241028,
      "learning_rate": 2.998094429994633e-05,
      "loss": 0.12183380126953125,
      "step": 2737
    },
    {
      "epoch": 0.38152302654497317,
      "grad_norm": 0.6989168524742126,
      "learning_rate": 2.9972773874111057e-05,
      "loss": 0.12021255493164062,
      "step": 2738
    },
    {
      "epoch": 0.38166237023618754,
      "grad_norm": 0.568914532661438,
      "learning_rate": 2.99646012325531e-05,
      "loss": 0.12770462036132812,
      "step": 2739
    },
    {
      "epoch": 0.3818017139274019,
      "grad_norm": 0.612546443939209,
      "learning_rate": 2.995642637708825e-05,
      "loss": 0.12451553344726562,
      "step": 2740
    },
    {
      "epoch": 0.3819410576186163,
      "grad_norm": 0.5529236793518066,
      "learning_rate": 2.9948249309532768e-05,
      "loss": 0.07091617584228516,
      "step": 2741
    },
    {
      "epoch": 0.3820804013098307,
      "grad_norm": 0.4931420087814331,
      "learning_rate": 2.9940070031703413e-05,
      "loss": 0.09982109069824219,
      "step": 2742
    },
    {
      "epoch": 0.38221974500104505,
      "grad_norm": 0.45883163809776306,
      "learning_rate": 2.9931888545417435e-05,
      "loss": 0.10249900817871094,
      "step": 2743
    },
    {
      "epoch": 0.3823590886922595,
      "grad_norm": 0.41469842195510864,
      "learning_rate": 2.9923704852492566e-05,
      "loss": 0.09606361389160156,
      "step": 2744
    },
    {
      "epoch": 0.38249843238347386,
      "grad_norm": 0.21494698524475098,
      "learning_rate": 2.9915518954747038e-05,
      "loss": 0.065521240234375,
      "step": 2745
    },
    {
      "epoch": 0.38263777607468824,
      "grad_norm": 0.35185182094573975,
      "learning_rate": 2.9907330853999583e-05,
      "loss": 0.10042285919189453,
      "step": 2746
    },
    {
      "epoch": 0.3827771197659026,
      "grad_norm": 0.39386314153671265,
      "learning_rate": 2.9899140552069396e-05,
      "loss": 0.09450912475585938,
      "step": 2747
    },
    {
      "epoch": 0.382916463457117,
      "grad_norm": 0.5368359088897705,
      "learning_rate": 2.989094805077618e-05,
      "loss": 0.1170501708984375,
      "step": 2748
    },
    {
      "epoch": 0.3830558071483314,
      "grad_norm": 0.5673030018806458,
      "learning_rate": 2.9882753351940115e-05,
      "loss": 0.09240961074829102,
      "step": 2749
    },
    {
      "epoch": 0.38319515083954575,
      "grad_norm": 0.4494551122188568,
      "learning_rate": 2.987455645738189e-05,
      "loss": 0.10066032409667969,
      "step": 2750
    },
    {
      "epoch": 0.38333449453076013,
      "grad_norm": 0.41963085532188416,
      "learning_rate": 2.9866357368922657e-05,
      "loss": 0.10503387451171875,
      "step": 2751
    },
    {
      "epoch": 0.3834738382219745,
      "grad_norm": 0.20522433519363403,
      "learning_rate": 2.985815608838407e-05,
      "loss": 0.060642242431640625,
      "step": 2752
    },
    {
      "epoch": 0.3836131819131889,
      "grad_norm": 0.7208806872367859,
      "learning_rate": 2.984995261758827e-05,
      "loss": 0.10602951049804688,
      "step": 2753
    },
    {
      "epoch": 0.38375252560440326,
      "grad_norm": 0.5711843371391296,
      "learning_rate": 2.984174695835787e-05,
      "loss": 0.12057304382324219,
      "step": 2754
    },
    {
      "epoch": 0.38389186929561764,
      "grad_norm": 0.278260201215744,
      "learning_rate": 2.983353911251599e-05,
      "loss": 0.069183349609375,
      "step": 2755
    },
    {
      "epoch": 0.384031212986832,
      "grad_norm": 0.6637197136878967,
      "learning_rate": 2.9825329081886222e-05,
      "loss": 0.11963844299316406,
      "step": 2756
    },
    {
      "epoch": 0.3841705566780464,
      "grad_norm": 0.25681716203689575,
      "learning_rate": 2.981711686829264e-05,
      "loss": 0.07241439819335938,
      "step": 2757
    },
    {
      "epoch": 0.38430990036926077,
      "grad_norm": 0.7481124997138977,
      "learning_rate": 2.9808902473559835e-05,
      "loss": 0.12314891815185547,
      "step": 2758
    },
    {
      "epoch": 0.38444924406047515,
      "grad_norm": 0.24926702678203583,
      "learning_rate": 2.9800685899512828e-05,
      "loss": 0.07191658020019531,
      "step": 2759
    },
    {
      "epoch": 0.3845885877516895,
      "grad_norm": 0.45428547263145447,
      "learning_rate": 2.9792467147977174e-05,
      "loss": 0.11787223815917969,
      "step": 2760
    },
    {
      "epoch": 0.3847279314429039,
      "grad_norm": 0.32125124335289,
      "learning_rate": 2.9784246220778885e-05,
      "loss": 0.08127880096435547,
      "step": 2761
    },
    {
      "epoch": 0.3848672751341183,
      "grad_norm": 0.2578955590724945,
      "learning_rate": 2.9776023119744462e-05,
      "loss": 0.07527732849121094,
      "step": 2762
    },
    {
      "epoch": 0.38500661882533266,
      "grad_norm": 0.4422241747379303,
      "learning_rate": 2.976779784670089e-05,
      "loss": 0.09022903442382812,
      "step": 2763
    },
    {
      "epoch": 0.38514596251654704,
      "grad_norm": 0.5517653822898865,
      "learning_rate": 2.9759570403475644e-05,
      "loss": 0.12624740600585938,
      "step": 2764
    },
    {
      "epoch": 0.38528530620776147,
      "grad_norm": 0.23249341547489166,
      "learning_rate": 2.975134079189667e-05,
      "loss": 0.07310295104980469,
      "step": 2765
    },
    {
      "epoch": 0.38542464989897585,
      "grad_norm": 0.6443906426429749,
      "learning_rate": 2.9743109013792395e-05,
      "loss": 0.11773872375488281,
      "step": 2766
    },
    {
      "epoch": 0.3855639935901902,
      "grad_norm": 0.2953103184700012,
      "learning_rate": 2.9734875070991736e-05,
      "loss": 0.08864402770996094,
      "step": 2767
    },
    {
      "epoch": 0.3857033372814046,
      "grad_norm": 0.32889044284820557,
      "learning_rate": 2.9726638965324088e-05,
      "loss": 0.07188940048217773,
      "step": 2768
    },
    {
      "epoch": 0.385842680972619,
      "grad_norm": 0.522597074508667,
      "learning_rate": 2.9718400698619327e-05,
      "loss": 0.1144256591796875,
      "step": 2769
    },
    {
      "epoch": 0.38598202466383336,
      "grad_norm": 0.421981543302536,
      "learning_rate": 2.9710160272707803e-05,
      "loss": 0.0855865478515625,
      "step": 2770
    },
    {
      "epoch": 0.38612136835504773,
      "grad_norm": 0.5474709868431091,
      "learning_rate": 2.9701917689420354e-05,
      "loss": 0.0909576416015625,
      "step": 2771
    },
    {
      "epoch": 0.3862607120462621,
      "grad_norm": 0.5345871448516846,
      "learning_rate": 2.9693672950588292e-05,
      "loss": 0.11491012573242188,
      "step": 2772
    },
    {
      "epoch": 0.3864000557374765,
      "grad_norm": 0.2521311342716217,
      "learning_rate": 2.9685426058043414e-05,
      "loss": 0.07110023498535156,
      "step": 2773
    },
    {
      "epoch": 0.38653939942869087,
      "grad_norm": 0.3329651355743408,
      "learning_rate": 2.9677177013617987e-05,
      "loss": 0.0724954605102539,
      "step": 2774
    },
    {
      "epoch": 0.38667874311990524,
      "grad_norm": 0.4630734324455261,
      "learning_rate": 2.9668925819144755e-05,
      "loss": 0.09348487854003906,
      "step": 2775
    },
    {
      "epoch": 0.3868180868111196,
      "grad_norm": 0.39918097853660583,
      "learning_rate": 2.966067247645696e-05,
      "loss": 0.1049041748046875,
      "step": 2776
    },
    {
      "epoch": 0.386957430502334,
      "grad_norm": 0.5957585573196411,
      "learning_rate": 2.965241698738829e-05,
      "loss": 0.11333656311035156,
      "step": 2777
    },
    {
      "epoch": 0.3870967741935484,
      "grad_norm": 0.4458919167518616,
      "learning_rate": 2.9644159353772937e-05,
      "loss": 0.09021186828613281,
      "step": 2778
    },
    {
      "epoch": 0.38723611788476275,
      "grad_norm": 0.3416711091995239,
      "learning_rate": 2.9635899577445558e-05,
      "loss": 0.0747528076171875,
      "step": 2779
    },
    {
      "epoch": 0.38737546157597713,
      "grad_norm": 0.4731020927429199,
      "learning_rate": 2.9627637660241283e-05,
      "loss": 0.114105224609375,
      "step": 2780
    },
    {
      "epoch": 0.3875148052671915,
      "grad_norm": 0.5346294045448303,
      "learning_rate": 2.9619373603995724e-05,
      "loss": 0.10131072998046875,
      "step": 2781
    },
    {
      "epoch": 0.3876541489584059,
      "grad_norm": 0.5147583484649658,
      "learning_rate": 2.9611107410544958e-05,
      "loss": 0.10749149322509766,
      "step": 2782
    },
    {
      "epoch": 0.38779349264962026,
      "grad_norm": 0.3791317343711853,
      "learning_rate": 2.9602839081725558e-05,
      "loss": 0.10463905334472656,
      "step": 2783
    },
    {
      "epoch": 0.38793283634083464,
      "grad_norm": 0.5833604335784912,
      "learning_rate": 2.959456861937455e-05,
      "loss": 0.0988922119140625,
      "step": 2784
    },
    {
      "epoch": 0.3880721800320491,
      "grad_norm": 0.3031501770019531,
      "learning_rate": 2.958629602532944e-05,
      "loss": 0.0731353759765625,
      "step": 2785
    },
    {
      "epoch": 0.38821152372326345,
      "grad_norm": 0.27346330881118774,
      "learning_rate": 2.9578021301428212e-05,
      "loss": 0.08133506774902344,
      "step": 2786
    },
    {
      "epoch": 0.38835086741447783,
      "grad_norm": 0.3339281678199768,
      "learning_rate": 2.9569744449509322e-05,
      "loss": 0.08997344970703125,
      "step": 2787
    },
    {
      "epoch": 0.3884902111056922,
      "grad_norm": 0.7204552292823792,
      "learning_rate": 2.9561465471411693e-05,
      "loss": 0.09783935546875,
      "step": 2788
    },
    {
      "epoch": 0.3886295547969066,
      "grad_norm": 0.2614377439022064,
      "learning_rate": 2.955318436897473e-05,
      "loss": 0.06814956665039062,
      "step": 2789
    },
    {
      "epoch": 0.38876889848812096,
      "grad_norm": 0.20632733404636383,
      "learning_rate": 2.9544901144038303e-05,
      "loss": 0.063232421875,
      "step": 2790
    },
    {
      "epoch": 0.38890824217933534,
      "grad_norm": 0.3239797353744507,
      "learning_rate": 2.9536615798442755e-05,
      "loss": 0.07658576965332031,
      "step": 2791
    },
    {
      "epoch": 0.3890475858705497,
      "grad_norm": 0.46518293023109436,
      "learning_rate": 2.9528328334028903e-05,
      "loss": 0.07776355743408203,
      "step": 2792
    },
    {
      "epoch": 0.3891869295617641,
      "grad_norm": 0.4190099835395813,
      "learning_rate": 2.952003875263803e-05,
      "loss": 0.09036064147949219,
      "step": 2793
    },
    {
      "epoch": 0.38932627325297847,
      "grad_norm": 0.5311266779899597,
      "learning_rate": 2.9511747056111893e-05,
      "loss": 0.11085891723632812,
      "step": 2794
    },
    {
      "epoch": 0.38946561694419285,
      "grad_norm": 0.37859615683555603,
      "learning_rate": 2.9503453246292716e-05,
      "loss": 0.09800529479980469,
      "step": 2795
    },
    {
      "epoch": 0.3896049606354072,
      "grad_norm": 0.6319118142127991,
      "learning_rate": 2.9495157325023195e-05,
      "loss": 0.11688613891601562,
      "step": 2796
    },
    {
      "epoch": 0.3897443043266216,
      "grad_norm": 0.9914302229881287,
      "learning_rate": 2.9486859294146497e-05,
      "loss": 0.1352405548095703,
      "step": 2797
    },
    {
      "epoch": 0.389883648017836,
      "grad_norm": 0.7224761843681335,
      "learning_rate": 2.9478559155506244e-05,
      "loss": 0.1156768798828125,
      "step": 2798
    },
    {
      "epoch": 0.39002299170905036,
      "grad_norm": 0.41691628098487854,
      "learning_rate": 2.9470256910946555e-05,
      "loss": 0.11707878112792969,
      "step": 2799
    },
    {
      "epoch": 0.39016233540026474,
      "grad_norm": 0.2999376058578491,
      "learning_rate": 2.946195256231199e-05,
      "loss": 0.0853729248046875,
      "step": 2800
    },
    {
      "epoch": 0.3903016790914791,
      "grad_norm": 0.9876853227615356,
      "learning_rate": 2.9453646111447582e-05,
      "loss": 0.17195415496826172,
      "step": 2801
    },
    {
      "epoch": 0.3904410227826935,
      "grad_norm": 0.5433264374732971,
      "learning_rate": 2.944533756019884e-05,
      "loss": 0.1074676513671875,
      "step": 2802
    },
    {
      "epoch": 0.39058036647390787,
      "grad_norm": 0.6145113110542297,
      "learning_rate": 2.9437026910411734e-05,
      "loss": 0.11833953857421875,
      "step": 2803
    },
    {
      "epoch": 0.39071971016512225,
      "grad_norm": 0.3740767538547516,
      "learning_rate": 2.9428714163932697e-05,
      "loss": 0.07789039611816406,
      "step": 2804
    },
    {
      "epoch": 0.3908590538563367,
      "grad_norm": 0.4510919153690338,
      "learning_rate": 2.9420399322608637e-05,
      "loss": 0.10042953491210938,
      "step": 2805
    },
    {
      "epoch": 0.39099839754755106,
      "grad_norm": 0.49457767605781555,
      "learning_rate": 2.9412082388286916e-05,
      "loss": 0.08388710021972656,
      "step": 2806
    },
    {
      "epoch": 0.39113774123876544,
      "grad_norm": 0.6645947694778442,
      "learning_rate": 2.940376336281537e-05,
      "loss": 0.11479473114013672,
      "step": 2807
    },
    {
      "epoch": 0.3912770849299798,
      "grad_norm": 0.3037877082824707,
      "learning_rate": 2.9395442248042297e-05,
      "loss": 0.08470916748046875,
      "step": 2808
    },
    {
      "epoch": 0.3914164286211942,
      "grad_norm": 0.632085382938385,
      "learning_rate": 2.9387119045816453e-05,
      "loss": 0.1334667205810547,
      "step": 2809
    },
    {
      "epoch": 0.39155577231240857,
      "grad_norm": 0.7185274958610535,
      "learning_rate": 2.9378793757987082e-05,
      "loss": 0.13877487182617188,
      "step": 2810
    },
    {
      "epoch": 0.39169511600362295,
      "grad_norm": 0.46159228682518005,
      "learning_rate": 2.9370466386403843e-05,
      "loss": 0.1102447509765625,
      "step": 2811
    },
    {
      "epoch": 0.3918344596948373,
      "grad_norm": 0.31972458958625793,
      "learning_rate": 2.9362136932916914e-05,
      "loss": 0.08866310119628906,
      "step": 2812
    },
    {
      "epoch": 0.3919738033860517,
      "grad_norm": 0.3936508595943451,
      "learning_rate": 2.93538053993769e-05,
      "loss": 0.09870052337646484,
      "step": 2813
    },
    {
      "epoch": 0.3921131470772661,
      "grad_norm": 0.4467742145061493,
      "learning_rate": 2.9345471787634873e-05,
      "loss": 0.10474395751953125,
      "step": 2814
    },
    {
      "epoch": 0.39225249076848046,
      "grad_norm": 0.43785348534584045,
      "learning_rate": 2.933713609954238e-05,
      "loss": 0.09344482421875,
      "step": 2815
    },
    {
      "epoch": 0.39239183445969483,
      "grad_norm": 0.6319786310195923,
      "learning_rate": 2.9328798336951415e-05,
      "loss": 0.09176921844482422,
      "step": 2816
    },
    {
      "epoch": 0.3925311781509092,
      "grad_norm": 0.3951207995414734,
      "learning_rate": 2.9320458501714437e-05,
      "loss": 0.08819580078125,
      "step": 2817
    },
    {
      "epoch": 0.3926705218421236,
      "grad_norm": 0.2947540283203125,
      "learning_rate": 2.931211659568437e-05,
      "loss": 0.0671548843383789,
      "step": 2818
    },
    {
      "epoch": 0.39280986553333797,
      "grad_norm": 0.4707702696323395,
      "learning_rate": 2.93037726207146e-05,
      "loss": 0.08539962768554688,
      "step": 2819
    },
    {
      "epoch": 0.39294920922455234,
      "grad_norm": 0.4581524431705475,
      "learning_rate": 2.929542657865896e-05,
      "loss": 0.09482574462890625,
      "step": 2820
    },
    {
      "epoch": 0.3930885529157667,
      "grad_norm": 0.31296730041503906,
      "learning_rate": 2.9287078471371747e-05,
      "loss": 0.07076263427734375,
      "step": 2821
    },
    {
      "epoch": 0.3932278966069811,
      "grad_norm": 0.5241219401359558,
      "learning_rate": 2.927872830070773e-05,
      "loss": 0.12018966674804688,
      "step": 2822
    },
    {
      "epoch": 0.3933672402981955,
      "grad_norm": 0.5782728791236877,
      "learning_rate": 2.9270376068522117e-05,
      "loss": 0.10172748565673828,
      "step": 2823
    },
    {
      "epoch": 0.39350658398940985,
      "grad_norm": 0.4213505685329437,
      "learning_rate": 2.9262021776670594e-05,
      "loss": 0.08958244323730469,
      "step": 2824
    },
    {
      "epoch": 0.3936459276806243,
      "grad_norm": 0.4802245795726776,
      "learning_rate": 2.9253665427009283e-05,
      "loss": 0.08833694458007812,
      "step": 2825
    },
    {
      "epoch": 0.39378527137183866,
      "grad_norm": 0.6023840308189392,
      "learning_rate": 2.9245307021394787e-05,
      "loss": 0.16649627685546875,
      "step": 2826
    },
    {
      "epoch": 0.39392461506305304,
      "grad_norm": 0.8773688673973083,
      "learning_rate": 2.9236946561684133e-05,
      "loss": 0.142608642578125,
      "step": 2827
    },
    {
      "epoch": 0.3940639587542674,
      "grad_norm": 0.45546507835388184,
      "learning_rate": 2.922858404973484e-05,
      "loss": 0.11491584777832031,
      "step": 2828
    },
    {
      "epoch": 0.3942033024454818,
      "grad_norm": 0.7287598252296448,
      "learning_rate": 2.922021948740487e-05,
      "loss": 0.12036895751953125,
      "step": 2829
    },
    {
      "epoch": 0.3943426461366962,
      "grad_norm": 0.4068584740161896,
      "learning_rate": 2.9211852876552624e-05,
      "loss": 0.10361480712890625,
      "step": 2830
    },
    {
      "epoch": 0.39448198982791055,
      "grad_norm": 0.8591437935829163,
      "learning_rate": 2.9203484219036986e-05,
      "loss": 0.1234292984008789,
      "step": 2831
    },
    {
      "epoch": 0.39462133351912493,
      "grad_norm": 0.41472795605659485,
      "learning_rate": 2.9195113516717267e-05,
      "loss": 0.08765792846679688,
      "step": 2832
    },
    {
      "epoch": 0.3947606772103393,
      "grad_norm": 0.33983391523361206,
      "learning_rate": 2.9186740771453253e-05,
      "loss": 0.08751392364501953,
      "step": 2833
    },
    {
      "epoch": 0.3949000209015537,
      "grad_norm": 0.4313623905181885,
      "learning_rate": 2.9178365985105182e-05,
      "loss": 0.08644580841064453,
      "step": 2834
    },
    {
      "epoch": 0.39503936459276806,
      "grad_norm": 0.4976264238357544,
      "learning_rate": 2.916998915953373e-05,
      "loss": 0.11478614807128906,
      "step": 2835
    },
    {
      "epoch": 0.39517870828398244,
      "grad_norm": 0.4120500981807709,
      "learning_rate": 2.916161029660004e-05,
      "loss": 0.09641838073730469,
      "step": 2836
    },
    {
      "epoch": 0.3953180519751968,
      "grad_norm": 0.5672963857650757,
      "learning_rate": 2.915322939816571e-05,
      "loss": 0.08379411697387695,
      "step": 2837
    },
    {
      "epoch": 0.3954573956664112,
      "grad_norm": 0.334119975566864,
      "learning_rate": 2.9144846466092773e-05,
      "loss": 0.06439781188964844,
      "step": 2838
    },
    {
      "epoch": 0.39559673935762557,
      "grad_norm": 0.6263145208358765,
      "learning_rate": 2.9136461502243735e-05,
      "loss": 0.11399173736572266,
      "step": 2839
    },
    {
      "epoch": 0.39573608304883995,
      "grad_norm": 0.3615053594112396,
      "learning_rate": 2.9128074508481544e-05,
      "loss": 0.0775289535522461,
      "step": 2840
    },
    {
      "epoch": 0.3958754267400543,
      "grad_norm": 0.4296785593032837,
      "learning_rate": 2.9119685486669587e-05,
      "loss": 0.08209228515625,
      "step": 2841
    },
    {
      "epoch": 0.3960147704312687,
      "grad_norm": 0.3805120587348938,
      "learning_rate": 2.911129443867173e-05,
      "loss": 0.09244251251220703,
      "step": 2842
    },
    {
      "epoch": 0.3961541141224831,
      "grad_norm": 0.6638845801353455,
      "learning_rate": 2.9102901366352254e-05,
      "loss": 0.10248088836669922,
      "step": 2843
    },
    {
      "epoch": 0.39629345781369746,
      "grad_norm": 0.23928716778755188,
      "learning_rate": 2.909450627157592e-05,
      "loss": 0.06439399719238281,
      "step": 2844
    },
    {
      "epoch": 0.3964328015049119,
      "grad_norm": 0.5579345226287842,
      "learning_rate": 2.9086109156207926e-05,
      "loss": 0.09394168853759766,
      "step": 2845
    },
    {
      "epoch": 0.39657214519612627,
      "grad_norm": 0.6000582575798035,
      "learning_rate": 2.9077710022113918e-05,
      "loss": 0.12530136108398438,
      "step": 2846
    },
    {
      "epoch": 0.39671148888734065,
      "grad_norm": 0.6729063987731934,
      "learning_rate": 2.906930887115999e-05,
      "loss": 0.11494636535644531,
      "step": 2847
    },
    {
      "epoch": 0.396850832578555,
      "grad_norm": 0.4611850678920746,
      "learning_rate": 2.906090570521268e-05,
      "loss": 0.09486770629882812,
      "step": 2848
    },
    {
      "epoch": 0.3969901762697694,
      "grad_norm": 0.3406074345111847,
      "learning_rate": 2.9052500526138994e-05,
      "loss": 0.07277584075927734,
      "step": 2849
    },
    {
      "epoch": 0.3971295199609838,
      "grad_norm": 0.4217374324798584,
      "learning_rate": 2.904409333580636e-05,
      "loss": 0.08868026733398438,
      "step": 2850
    },
    {
      "epoch": 0.39726886365219816,
      "grad_norm": 0.28599727153778076,
      "learning_rate": 2.903568413608267e-05,
      "loss": 0.07407951354980469,
      "step": 2851
    },
    {
      "epoch": 0.39740820734341253,
      "grad_norm": 0.31757208704948425,
      "learning_rate": 2.9027272928836248e-05,
      "loss": 0.09232902526855469,
      "step": 2852
    },
    {
      "epoch": 0.3975475510346269,
      "grad_norm": 0.4805624186992645,
      "learning_rate": 2.901885971593588e-05,
      "loss": 0.08654594421386719,
      "step": 2853
    },
    {
      "epoch": 0.3976868947258413,
      "grad_norm": 0.594313383102417,
      "learning_rate": 2.901044449925079e-05,
      "loss": 0.13433456420898438,
      "step": 2854
    },
    {
      "epoch": 0.39782623841705567,
      "grad_norm": 0.4714030623435974,
      "learning_rate": 2.9002027280650643e-05,
      "loss": 0.11092185974121094,
      "step": 2855
    },
    {
      "epoch": 0.39796558210827004,
      "grad_norm": 0.4146770238876343,
      "learning_rate": 2.899360806200555e-05,
      "loss": 0.09079456329345703,
      "step": 2856
    },
    {
      "epoch": 0.3981049257994844,
      "grad_norm": 0.3807482123374939,
      "learning_rate": 2.8985186845186077e-05,
      "loss": 0.09280967712402344,
      "step": 2857
    },
    {
      "epoch": 0.3982442694906988,
      "grad_norm": 0.4483656585216522,
      "learning_rate": 2.897676363206322e-05,
      "loss": 0.08497238159179688,
      "step": 2858
    },
    {
      "epoch": 0.3983836131819132,
      "grad_norm": 0.6151071190834045,
      "learning_rate": 2.8968338424508426e-05,
      "loss": 0.1251373291015625,
      "step": 2859
    },
    {
      "epoch": 0.39852295687312755,
      "grad_norm": 0.38683465123176575,
      "learning_rate": 2.895991122439359e-05,
      "loss": 0.06981563568115234,
      "step": 2860
    },
    {
      "epoch": 0.39866230056434193,
      "grad_norm": 0.539776086807251,
      "learning_rate": 2.895148203359103e-05,
      "loss": 0.10857963562011719,
      "step": 2861
    },
    {
      "epoch": 0.3988016442555563,
      "grad_norm": 0.7045628428459167,
      "learning_rate": 2.8943050853973536e-05,
      "loss": 0.0862579345703125,
      "step": 2862
    },
    {
      "epoch": 0.3989409879467707,
      "grad_norm": 0.5766880512237549,
      "learning_rate": 2.893461768741431e-05,
      "loss": 0.09726905822753906,
      "step": 2863
    },
    {
      "epoch": 0.39908033163798506,
      "grad_norm": 0.4177318811416626,
      "learning_rate": 2.892618253578702e-05,
      "loss": 0.10309982299804688,
      "step": 2864
    },
    {
      "epoch": 0.3992196753291995,
      "grad_norm": 1.0094282627105713,
      "learning_rate": 2.8917745400965755e-05,
      "loss": 0.11057281494140625,
      "step": 2865
    },
    {
      "epoch": 0.3993590190204139,
      "grad_norm": 0.39825648069381714,
      "learning_rate": 2.8909306284825058e-05,
      "loss": 0.07055854797363281,
      "step": 2866
    },
    {
      "epoch": 0.39949836271162825,
      "grad_norm": 0.3099052906036377,
      "learning_rate": 2.8900865189239907e-05,
      "loss": 0.07972526550292969,
      "step": 2867
    },
    {
      "epoch": 0.39963770640284263,
      "grad_norm": 0.3715667426586151,
      "learning_rate": 2.889242211608572e-05,
      "loss": 0.08117103576660156,
      "step": 2868
    },
    {
      "epoch": 0.399777050094057,
      "grad_norm": 0.3709195554256439,
      "learning_rate": 2.8883977067238363e-05,
      "loss": 0.07629013061523438,
      "step": 2869
    },
    {
      "epoch": 0.3999163937852714,
      "grad_norm": 0.451784610748291,
      "learning_rate": 2.887553004457412e-05,
      "loss": 0.07976913452148438,
      "step": 2870
    },
    {
      "epoch": 0.40005573747648576,
      "grad_norm": 0.4687831699848175,
      "learning_rate": 2.8867081049969738e-05,
      "loss": 0.10159111022949219,
      "step": 2871
    },
    {
      "epoch": 0.40019508116770014,
      "grad_norm": 0.5947216153144836,
      "learning_rate": 2.8858630085302378e-05,
      "loss": 0.1293649673461914,
      "step": 2872
    },
    {
      "epoch": 0.4003344248589145,
      "grad_norm": 0.2705707848072052,
      "learning_rate": 2.885017715244966e-05,
      "loss": 0.07787513732910156,
      "step": 2873
    },
    {
      "epoch": 0.4004737685501289,
      "grad_norm": 0.26179739832878113,
      "learning_rate": 2.884172225328964e-05,
      "loss": 0.06071281433105469,
      "step": 2874
    },
    {
      "epoch": 0.40061311224134327,
      "grad_norm": 0.7979472279548645,
      "learning_rate": 2.883326538970079e-05,
      "loss": 0.13092422485351562,
      "step": 2875
    },
    {
      "epoch": 0.40075245593255765,
      "grad_norm": 0.6518993973731995,
      "learning_rate": 2.8824806563562037e-05,
      "loss": 0.11884307861328125,
      "step": 2876
    },
    {
      "epoch": 0.400891799623772,
      "grad_norm": 0.3480386435985565,
      "learning_rate": 2.8816345776752737e-05,
      "loss": 0.08157920837402344,
      "step": 2877
    },
    {
      "epoch": 0.4010311433149864,
      "grad_norm": 0.2972913384437561,
      "learning_rate": 2.880788303115269e-05,
      "loss": 0.08409690856933594,
      "step": 2878
    },
    {
      "epoch": 0.4011704870062008,
      "grad_norm": 0.5967562794685364,
      "learning_rate": 2.8799418328642116e-05,
      "loss": 0.11619377136230469,
      "step": 2879
    },
    {
      "epoch": 0.40130983069741516,
      "grad_norm": 0.5714161992073059,
      "learning_rate": 2.879095167110169e-05,
      "loss": 0.09496879577636719,
      "step": 2880
    },
    {
      "epoch": 0.40144917438862954,
      "grad_norm": 0.634818971157074,
      "learning_rate": 2.8782483060412502e-05,
      "loss": 0.10003471374511719,
      "step": 2881
    },
    {
      "epoch": 0.4015885180798439,
      "grad_norm": 0.36395058035850525,
      "learning_rate": 2.8774012498456083e-05,
      "loss": 0.09754180908203125,
      "step": 2882
    },
    {
      "epoch": 0.4017278617710583,
      "grad_norm": 0.8031824827194214,
      "learning_rate": 2.8765539987114403e-05,
      "loss": 0.11302757263183594,
      "step": 2883
    },
    {
      "epoch": 0.40186720546227267,
      "grad_norm": 0.39306899905204773,
      "learning_rate": 2.8757065528269855e-05,
      "loss": 0.10762405395507812,
      "step": 2884
    },
    {
      "epoch": 0.4020065491534871,
      "grad_norm": 0.2913625240325928,
      "learning_rate": 2.8748589123805274e-05,
      "loss": 0.07906723022460938,
      "step": 2885
    },
    {
      "epoch": 0.4021458928447015,
      "grad_norm": 0.47080209851264954,
      "learning_rate": 2.874011077560393e-05,
      "loss": 0.11349868774414062,
      "step": 2886
    },
    {
      "epoch": 0.40228523653591586,
      "grad_norm": 0.5210425853729248,
      "learning_rate": 2.8731630485549504e-05,
      "loss": 0.11547279357910156,
      "step": 2887
    },
    {
      "epoch": 0.40242458022713024,
      "grad_norm": 0.22618699073791504,
      "learning_rate": 2.8723148255526138e-05,
      "loss": 0.06984329223632812,
      "step": 2888
    },
    {
      "epoch": 0.4025639239183446,
      "grad_norm": 0.44981807470321655,
      "learning_rate": 2.8714664087418374e-05,
      "loss": 0.09727668762207031,
      "step": 2889
    },
    {
      "epoch": 0.402703267609559,
      "grad_norm": 0.3494621813297272,
      "learning_rate": 2.8706177983111216e-05,
      "loss": 0.08778190612792969,
      "step": 2890
    },
    {
      "epoch": 0.40284261130077337,
      "grad_norm": 0.5396576523780823,
      "learning_rate": 2.869768994449007e-05,
      "loss": 0.09456253051757812,
      "step": 2891
    },
    {
      "epoch": 0.40298195499198775,
      "grad_norm": 0.2663780748844147,
      "learning_rate": 2.86891999734408e-05,
      "loss": 0.07070732116699219,
      "step": 2892
    },
    {
      "epoch": 0.4031212986832021,
      "grad_norm": 0.758431077003479,
      "learning_rate": 2.868070807184966e-05,
      "loss": 0.11065864562988281,
      "step": 2893
    },
    {
      "epoch": 0.4032606423744165,
      "grad_norm": 0.3624112904071808,
      "learning_rate": 2.867221424160338e-05,
      "loss": 0.08304405212402344,
      "step": 2894
    },
    {
      "epoch": 0.4033999860656309,
      "grad_norm": 0.8406776189804077,
      "learning_rate": 2.866371848458908e-05,
      "loss": 0.15859413146972656,
      "step": 2895
    },
    {
      "epoch": 0.40353932975684526,
      "grad_norm": 0.3445568084716797,
      "learning_rate": 2.8655220802694334e-05,
      "loss": 0.09676933288574219,
      "step": 2896
    },
    {
      "epoch": 0.40367867344805963,
      "grad_norm": 0.37975454330444336,
      "learning_rate": 2.864672119780713e-05,
      "loss": 0.10293006896972656,
      "step": 2897
    },
    {
      "epoch": 0.403818017139274,
      "grad_norm": 0.5747414827346802,
      "learning_rate": 2.8638219671815873e-05,
      "loss": 0.07542800903320312,
      "step": 2898
    },
    {
      "epoch": 0.4039573608304884,
      "grad_norm": 0.7423135042190552,
      "learning_rate": 2.8629716226609427e-05,
      "loss": 0.11152076721191406,
      "step": 2899
    },
    {
      "epoch": 0.40409670452170277,
      "grad_norm": 0.18259790539741516,
      "learning_rate": 2.8621210864077053e-05,
      "loss": 0.061981201171875,
      "step": 2900
    },
    {
      "epoch": 0.40423604821291714,
      "grad_norm": 0.42934221029281616,
      "learning_rate": 2.861270358610845e-05,
      "loss": 0.0935516357421875,
      "step": 2901
    },
    {
      "epoch": 0.4043753919041315,
      "grad_norm": 0.2880236506462097,
      "learning_rate": 2.8604194394593744e-05,
      "loss": 0.07050514221191406,
      "step": 2902
    },
    {
      "epoch": 0.4045147355953459,
      "grad_norm": 0.49966609477996826,
      "learning_rate": 2.8595683291423476e-05,
      "loss": 0.10454559326171875,
      "step": 2903
    },
    {
      "epoch": 0.4046540792865603,
      "grad_norm": 0.4908045530319214,
      "learning_rate": 2.858717027848863e-05,
      "loss": 0.09884488582611084,
      "step": 2904
    },
    {
      "epoch": 0.4047934229777747,
      "grad_norm": 0.5836314558982849,
      "learning_rate": 2.857865535768059e-05,
      "loss": 0.11692428588867188,
      "step": 2905
    },
    {
      "epoch": 0.4049327666689891,
      "grad_norm": 0.37020042538642883,
      "learning_rate": 2.8570138530891188e-05,
      "loss": 0.09495162963867188,
      "step": 2906
    },
    {
      "epoch": 0.40507211036020346,
      "grad_norm": 0.47730156779289246,
      "learning_rate": 2.8561619800012657e-05,
      "loss": 0.09744071960449219,
      "step": 2907
    },
    {
      "epoch": 0.40521145405141784,
      "grad_norm": 0.831049919128418,
      "learning_rate": 2.8553099166937685e-05,
      "loss": 0.1297168731689453,
      "step": 2908
    },
    {
      "epoch": 0.4053507977426322,
      "grad_norm": 0.45376503467559814,
      "learning_rate": 2.8544576633559335e-05,
      "loss": 0.08504104614257812,
      "step": 2909
    },
    {
      "epoch": 0.4054901414338466,
      "grad_norm": 0.28864362835884094,
      "learning_rate": 2.853605220177114e-05,
      "loss": 0.07505607604980469,
      "step": 2910
    },
    {
      "epoch": 0.405629485125061,
      "grad_norm": 0.5174407362937927,
      "learning_rate": 2.8527525873467022e-05,
      "loss": 0.1446208953857422,
      "step": 2911
    },
    {
      "epoch": 0.40576882881627535,
      "grad_norm": 0.31805428862571716,
      "learning_rate": 2.851899765054135e-05,
      "loss": 0.07036590576171875,
      "step": 2912
    },
    {
      "epoch": 0.40590817250748973,
      "grad_norm": 0.29066160321235657,
      "learning_rate": 2.8510467534888886e-05,
      "loss": 0.06470203399658203,
      "step": 2913
    },
    {
      "epoch": 0.4060475161987041,
      "grad_norm": 0.4190223515033722,
      "learning_rate": 2.8501935528404833e-05,
      "loss": 0.08709335327148438,
      "step": 2914
    },
    {
      "epoch": 0.4061868598899185,
      "grad_norm": 0.3024798035621643,
      "learning_rate": 2.849340163298481e-05,
      "loss": 0.09853553771972656,
      "step": 2915
    },
    {
      "epoch": 0.40632620358113286,
      "grad_norm": 0.30848273634910583,
      "learning_rate": 2.848486585052485e-05,
      "loss": 0.08180713653564453,
      "step": 2916
    },
    {
      "epoch": 0.40646554727234724,
      "grad_norm": 0.19469375908374786,
      "learning_rate": 2.8476328182921414e-05,
      "loss": 0.059627532958984375,
      "step": 2917
    },
    {
      "epoch": 0.4066048909635616,
      "grad_norm": 0.47342467308044434,
      "learning_rate": 2.8467788632071367e-05,
      "loss": 0.11958694458007812,
      "step": 2918
    },
    {
      "epoch": 0.406744234654776,
      "grad_norm": 0.5535681247711182,
      "learning_rate": 2.845924719987202e-05,
      "loss": 0.11155509948730469,
      "step": 2919
    },
    {
      "epoch": 0.40688357834599037,
      "grad_norm": 0.6099625825881958,
      "learning_rate": 2.8450703888221066e-05,
      "loss": 0.08853816986083984,
      "step": 2920
    },
    {
      "epoch": 0.40702292203720475,
      "grad_norm": 0.30032074451446533,
      "learning_rate": 2.844215869901664e-05,
      "loss": 0.0713491439819336,
      "step": 2921
    },
    {
      "epoch": 0.4071622657284191,
      "grad_norm": 0.6499759554862976,
      "learning_rate": 2.8433611634157293e-05,
      "loss": 0.10539436340332031,
      "step": 2922
    },
    {
      "epoch": 0.4073016094196335,
      "grad_norm": 0.35065481066703796,
      "learning_rate": 2.8425062695541975e-05,
      "loss": 0.08057212829589844,
      "step": 2923
    },
    {
      "epoch": 0.4074409531108479,
      "grad_norm": 0.38863053917884827,
      "learning_rate": 2.8416511885070085e-05,
      "loss": 0.07405853271484375,
      "step": 2924
    },
    {
      "epoch": 0.4075802968020623,
      "grad_norm": 0.3987928032875061,
      "learning_rate": 2.84079592046414e-05,
      "loss": 0.08418464660644531,
      "step": 2925
    },
    {
      "epoch": 0.4077196404932767,
      "grad_norm": 0.3448917865753174,
      "learning_rate": 2.839940465615614e-05,
      "loss": 0.08751296997070312,
      "step": 2926
    },
    {
      "epoch": 0.40785898418449107,
      "grad_norm": 0.31232085824012756,
      "learning_rate": 2.8390848241514918e-05,
      "loss": 0.08133506774902344,
      "step": 2927
    },
    {
      "epoch": 0.40799832787570545,
      "grad_norm": 0.8299886584281921,
      "learning_rate": 2.8382289962618793e-05,
      "loss": 0.13628768920898438,
      "step": 2928
    },
    {
      "epoch": 0.4081376715669198,
      "grad_norm": 0.5269566774368286,
      "learning_rate": 2.8373729821369206e-05,
      "loss": 0.093505859375,
      "step": 2929
    },
    {
      "epoch": 0.4082770152581342,
      "grad_norm": 0.5619516968727112,
      "learning_rate": 2.8365167819668027e-05,
      "loss": 0.11731243133544922,
      "step": 2930
    },
    {
      "epoch": 0.4084163589493486,
      "grad_norm": 0.7887362837791443,
      "learning_rate": 2.835660395941754e-05,
      "loss": 0.13588333129882812,
      "step": 2931
    },
    {
      "epoch": 0.40855570264056296,
      "grad_norm": 0.2972719967365265,
      "learning_rate": 2.8348038242520438e-05,
      "loss": 0.07999229431152344,
      "step": 2932
    },
    {
      "epoch": 0.40869504633177733,
      "grad_norm": 0.5917282700538635,
      "learning_rate": 2.833947067087983e-05,
      "loss": 0.10748863220214844,
      "step": 2933
    },
    {
      "epoch": 0.4088343900229917,
      "grad_norm": 0.5834023356437683,
      "learning_rate": 2.833090124639923e-05,
      "loss": 0.11511421203613281,
      "step": 2934
    },
    {
      "epoch": 0.4089737337142061,
      "grad_norm": 0.2627822458744049,
      "learning_rate": 2.832232997098257e-05,
      "loss": 0.08635711669921875,
      "step": 2935
    },
    {
      "epoch": 0.40911307740542047,
      "grad_norm": 0.4111623764038086,
      "learning_rate": 2.831375684653419e-05,
      "loss": 0.1255779266357422,
      "step": 2936
    },
    {
      "epoch": 0.40925242109663484,
      "grad_norm": 0.5223656892776489,
      "learning_rate": 2.8305181874958844e-05,
      "loss": 0.115966796875,
      "step": 2937
    },
    {
      "epoch": 0.4093917647878492,
      "grad_norm": 0.6312443614006042,
      "learning_rate": 2.82966050581617e-05,
      "loss": 0.12738037109375,
      "step": 2938
    },
    {
      "epoch": 0.4095311084790636,
      "grad_norm": 0.39587321877479553,
      "learning_rate": 2.8288026398048326e-05,
      "loss": 0.09307670593261719,
      "step": 2939
    },
    {
      "epoch": 0.409670452170278,
      "grad_norm": 0.5742647647857666,
      "learning_rate": 2.8279445896524705e-05,
      "loss": 0.09099006652832031,
      "step": 2940
    },
    {
      "epoch": 0.40980979586149235,
      "grad_norm": 0.4007696211338043,
      "learning_rate": 2.8270863555497227e-05,
      "loss": 0.08545875549316406,
      "step": 2941
    },
    {
      "epoch": 0.40994913955270673,
      "grad_norm": 0.47694236040115356,
      "learning_rate": 2.82622793768727e-05,
      "loss": 0.0868072509765625,
      "step": 2942
    },
    {
      "epoch": 0.4100884832439211,
      "grad_norm": 0.31636810302734375,
      "learning_rate": 2.8253693362558322e-05,
      "loss": 0.07649517059326172,
      "step": 2943
    },
    {
      "epoch": 0.4102278269351355,
      "grad_norm": 0.49565309286117554,
      "learning_rate": 2.8245105514461712e-05,
      "loss": 0.10487842559814453,
      "step": 2944
    },
    {
      "epoch": 0.4103671706263499,
      "grad_norm": 0.3699641227722168,
      "learning_rate": 2.82365158344909e-05,
      "loss": 0.10815048217773438,
      "step": 2945
    },
    {
      "epoch": 0.4105065143175643,
      "grad_norm": 0.5241638422012329,
      "learning_rate": 2.822792432455431e-05,
      "loss": 0.11094474792480469,
      "step": 2946
    },
    {
      "epoch": 0.4106458580087787,
      "grad_norm": 0.8141152262687683,
      "learning_rate": 2.8219330986560783e-05,
      "loss": 0.12200164794921875,
      "step": 2947
    },
    {
      "epoch": 0.41078520169999305,
      "grad_norm": 0.426407128572464,
      "learning_rate": 2.821073582241956e-05,
      "loss": 0.10487747192382812,
      "step": 2948
    },
    {
      "epoch": 0.41092454539120743,
      "grad_norm": 0.45884084701538086,
      "learning_rate": 2.820213883404029e-05,
      "loss": 0.0910797119140625,
      "step": 2949
    },
    {
      "epoch": 0.4110638890824218,
      "grad_norm": 0.4297236204147339,
      "learning_rate": 2.8193540023333033e-05,
      "loss": 0.07059669494628906,
      "step": 2950
    },
    {
      "epoch": 0.4112032327736362,
      "grad_norm": 0.5169134736061096,
      "learning_rate": 2.818493939220824e-05,
      "loss": 0.08985328674316406,
      "step": 2951
    },
    {
      "epoch": 0.41134257646485056,
      "grad_norm": 0.3720305562019348,
      "learning_rate": 2.8176336942576785e-05,
      "loss": 0.07719039916992188,
      "step": 2952
    },
    {
      "epoch": 0.41148192015606494,
      "grad_norm": 0.41053661704063416,
      "learning_rate": 2.816773267634993e-05,
      "loss": 0.08061790466308594,
      "step": 2953
    },
    {
      "epoch": 0.4116212638472793,
      "grad_norm": 0.5452696681022644,
      "learning_rate": 2.8159126595439344e-05,
      "loss": 0.126678466796875,
      "step": 2954
    },
    {
      "epoch": 0.4117606075384937,
      "grad_norm": 0.2549055218696594,
      "learning_rate": 2.8150518701757104e-05,
      "loss": 0.0647134780883789,
      "step": 2955
    },
    {
      "epoch": 0.4118999512297081,
      "grad_norm": 0.4326753318309784,
      "learning_rate": 2.814190899721569e-05,
      "loss": 0.09583473205566406,
      "step": 2956
    },
    {
      "epoch": 0.41203929492092245,
      "grad_norm": 0.6350720524787903,
      "learning_rate": 2.8133297483727972e-05,
      "loss": 0.08076858520507812,
      "step": 2957
    },
    {
      "epoch": 0.4121786386121368,
      "grad_norm": 0.5242518186569214,
      "learning_rate": 2.8124684163207252e-05,
      "loss": 0.08086252212524414,
      "step": 2958
    },
    {
      "epoch": 0.4123179823033512,
      "grad_norm": 0.4909093379974365,
      "learning_rate": 2.8116069037567187e-05,
      "loss": 0.09307193756103516,
      "step": 2959
    },
    {
      "epoch": 0.4124573259945656,
      "grad_norm": 0.7606944441795349,
      "learning_rate": 2.8107452108721887e-05,
      "loss": 0.12036895751953125,
      "step": 2960
    },
    {
      "epoch": 0.41259666968577996,
      "grad_norm": 0.479745477437973,
      "learning_rate": 2.809883337858582e-05,
      "loss": 0.10840225219726562,
      "step": 2961
    },
    {
      "epoch": 0.41273601337699434,
      "grad_norm": 0.4662981331348419,
      "learning_rate": 2.8090212849073877e-05,
      "loss": 0.10615825653076172,
      "step": 2962
    },
    {
      "epoch": 0.4128753570682087,
      "grad_norm": 0.5484395027160645,
      "learning_rate": 2.8081590522101342e-05,
      "loss": 0.1029977798461914,
      "step": 2963
    },
    {
      "epoch": 0.4130147007594231,
      "grad_norm": 0.38715115189552307,
      "learning_rate": 2.8072966399583897e-05,
      "loss": 0.08767318725585938,
      "step": 2964
    },
    {
      "epoch": 0.41315404445063747,
      "grad_norm": 0.4839401841163635,
      "learning_rate": 2.8064340483437625e-05,
      "loss": 0.10988616943359375,
      "step": 2965
    },
    {
      "epoch": 0.4132933881418519,
      "grad_norm": 0.3486270010471344,
      "learning_rate": 2.8055712775579012e-05,
      "loss": 0.07895278930664062,
      "step": 2966
    },
    {
      "epoch": 0.4134327318330663,
      "grad_norm": 0.3146027624607086,
      "learning_rate": 2.8047083277924935e-05,
      "loss": 0.08570671081542969,
      "step": 2967
    },
    {
      "epoch": 0.41357207552428066,
      "grad_norm": 0.41636112332344055,
      "learning_rate": 2.803845199239267e-05,
      "loss": 0.0882568359375,
      "step": 2968
    },
    {
      "epoch": 0.41371141921549504,
      "grad_norm": 0.4933621287345886,
      "learning_rate": 2.8029818920899902e-05,
      "loss": 0.10334396362304688,
      "step": 2969
    },
    {
      "epoch": 0.4138507629067094,
      "grad_norm": 0.18043652176856995,
      "learning_rate": 2.8021184065364684e-05,
      "loss": 0.06083488464355469,
      "step": 2970
    },
    {
      "epoch": 0.4139901065979238,
      "grad_norm": 0.2700120508670807,
      "learning_rate": 2.8012547427705497e-05,
      "loss": 0.08107757568359375,
      "step": 2971
    },
    {
      "epoch": 0.41412945028913817,
      "grad_norm": 0.3799881041049957,
      "learning_rate": 2.80039090098412e-05,
      "loss": 0.08291244506835938,
      "step": 2972
    },
    {
      "epoch": 0.41426879398035255,
      "grad_norm": 0.4240100681781769,
      "learning_rate": 2.7995268813691052e-05,
      "loss": 0.08331871032714844,
      "step": 2973
    },
    {
      "epoch": 0.4144081376715669,
      "grad_norm": 0.4528537392616272,
      "learning_rate": 2.7986626841174717e-05,
      "loss": 0.09042549133300781,
      "step": 2974
    },
    {
      "epoch": 0.4145474813627813,
      "grad_norm": 0.4231869876384735,
      "learning_rate": 2.7977983094212224e-05,
      "loss": 0.09463691711425781,
      "step": 2975
    },
    {
      "epoch": 0.4146868250539957,
      "grad_norm": 0.42183446884155273,
      "learning_rate": 2.7969337574724033e-05,
      "loss": 0.08738327026367188,
      "step": 2976
    },
    {
      "epoch": 0.41482616874521006,
      "grad_norm": 0.3546229898929596,
      "learning_rate": 2.7960690284630976e-05,
      "loss": 0.07401657104492188,
      "step": 2977
    },
    {
      "epoch": 0.41496551243642443,
      "grad_norm": 0.6682326197624207,
      "learning_rate": 2.7952041225854283e-05,
      "loss": 0.12703514099121094,
      "step": 2978
    },
    {
      "epoch": 0.4151048561276388,
      "grad_norm": 0.6465019583702087,
      "learning_rate": 2.7943390400315577e-05,
      "loss": 0.11440467834472656,
      "step": 2979
    },
    {
      "epoch": 0.4152441998188532,
      "grad_norm": 0.5679382681846619,
      "learning_rate": 2.793473780993688e-05,
      "loss": 0.1015777587890625,
      "step": 2980
    },
    {
      "epoch": 0.41538354351006757,
      "grad_norm": 0.3499954044818878,
      "learning_rate": 2.792608345664059e-05,
      "loss": 0.07579612731933594,
      "step": 2981
    },
    {
      "epoch": 0.41552288720128194,
      "grad_norm": 0.4513980746269226,
      "learning_rate": 2.791742734234951e-05,
      "loss": 0.09237957000732422,
      "step": 2982
    },
    {
      "epoch": 0.4156622308924963,
      "grad_norm": 0.43855834007263184,
      "learning_rate": 2.7908769468986837e-05,
      "loss": 0.08110666275024414,
      "step": 2983
    },
    {
      "epoch": 0.4158015745837107,
      "grad_norm": 0.29996269941329956,
      "learning_rate": 2.7900109838476138e-05,
      "loss": 0.0807342529296875,
      "step": 2984
    },
    {
      "epoch": 0.4159409182749251,
      "grad_norm": 0.4602603316307068,
      "learning_rate": 2.789144845274141e-05,
      "loss": 0.09967613220214844,
      "step": 2985
    },
    {
      "epoch": 0.4160802619661395,
      "grad_norm": 0.4237877130508423,
      "learning_rate": 2.7882785313706996e-05,
      "loss": 0.09588241577148438,
      "step": 2986
    },
    {
      "epoch": 0.4162196056573539,
      "grad_norm": 0.42571425437927246,
      "learning_rate": 2.787412042329765e-05,
      "loss": 0.08127212524414062,
      "step": 2987
    },
    {
      "epoch": 0.41635894934856826,
      "grad_norm": 0.248041570186615,
      "learning_rate": 2.7865453783438517e-05,
      "loss": 0.06004047393798828,
      "step": 2988
    },
    {
      "epoch": 0.41649829303978264,
      "grad_norm": 0.32979148626327515,
      "learning_rate": 2.785678539605512e-05,
      "loss": 0.07060050964355469,
      "step": 2989
    },
    {
      "epoch": 0.416637636730997,
      "grad_norm": 0.5158145427703857,
      "learning_rate": 2.7848115263073386e-05,
      "loss": 0.09709358215332031,
      "step": 2990
    },
    {
      "epoch": 0.4167769804222114,
      "grad_norm": 0.361479789018631,
      "learning_rate": 2.7839443386419613e-05,
      "loss": 0.09724617004394531,
      "step": 2991
    },
    {
      "epoch": 0.4169163241134258,
      "grad_norm": 0.5252524614334106,
      "learning_rate": 2.7830769768020504e-05,
      "loss": 0.10721015930175781,
      "step": 2992
    },
    {
      "epoch": 0.41705566780464015,
      "grad_norm": 0.4230610728263855,
      "learning_rate": 2.782209440980312e-05,
      "loss": 0.10601234436035156,
      "step": 2993
    },
    {
      "epoch": 0.41719501149585453,
      "grad_norm": 0.1995987594127655,
      "learning_rate": 2.781341731369495e-05,
      "loss": 0.05877208709716797,
      "step": 2994
    },
    {
      "epoch": 0.4173343551870689,
      "grad_norm": 0.469099760055542,
      "learning_rate": 2.780473848162383e-05,
      "loss": 0.09636878967285156,
      "step": 2995
    },
    {
      "epoch": 0.4174736988782833,
      "grad_norm": 0.41076505184173584,
      "learning_rate": 2.779605791551801e-05,
      "loss": 0.0984344482421875,
      "step": 2996
    },
    {
      "epoch": 0.41761304256949766,
      "grad_norm": 0.691967248916626,
      "learning_rate": 2.778737561730611e-05,
      "loss": 0.11032295227050781,
      "step": 2997
    },
    {
      "epoch": 0.41775238626071204,
      "grad_norm": 0.4118465781211853,
      "learning_rate": 2.7778691588917127e-05,
      "loss": 0.08633232116699219,
      "step": 2998
    },
    {
      "epoch": 0.4178917299519264,
      "grad_norm": 0.44758427143096924,
      "learning_rate": 2.777000583228047e-05,
      "loss": 0.09740543365478516,
      "step": 2999
    },
    {
      "epoch": 0.4180310736431408,
      "grad_norm": 0.3800640404224396,
      "learning_rate": 2.776131834932591e-05,
      "loss": 0.07234764099121094,
      "step": 3000
    },
    {
      "epoch": 0.41817041733435517,
      "grad_norm": 0.2727917432785034,
      "learning_rate": 2.7752629141983605e-05,
      "loss": 0.07259941101074219,
      "step": 3001
    },
    {
      "epoch": 0.41830976102556955,
      "grad_norm": 0.17896799743175507,
      "learning_rate": 2.77439382121841e-05,
      "loss": 0.05730438232421875,
      "step": 3002
    },
    {
      "epoch": 0.4184491047167839,
      "grad_norm": 0.3604126274585724,
      "learning_rate": 2.773524556185832e-05,
      "loss": 0.08320236206054688,
      "step": 3003
    },
    {
      "epoch": 0.4185884484079983,
      "grad_norm": 0.15601606667041779,
      "learning_rate": 2.7726551192937577e-05,
      "loss": 0.057816505432128906,
      "step": 3004
    },
    {
      "epoch": 0.4187277920992127,
      "grad_norm": 0.564013659954071,
      "learning_rate": 2.7717855107353557e-05,
      "loss": 0.09427070617675781,
      "step": 3005
    },
    {
      "epoch": 0.4188671357904271,
      "grad_norm": 0.2955789566040039,
      "learning_rate": 2.770915730703834e-05,
      "loss": 0.09270095825195312,
      "step": 3006
    },
    {
      "epoch": 0.4190064794816415,
      "grad_norm": 0.3839729428291321,
      "learning_rate": 2.7700457793924357e-05,
      "loss": 0.09511184692382812,
      "step": 3007
    },
    {
      "epoch": 0.41914582317285587,
      "grad_norm": 0.5673214793205261,
      "learning_rate": 2.7691756569944473e-05,
      "loss": 0.10733795166015625,
      "step": 3008
    },
    {
      "epoch": 0.41928516686407025,
      "grad_norm": 0.29565975069999695,
      "learning_rate": 2.7683053637031874e-05,
      "loss": 0.09211921691894531,
      "step": 3009
    },
    {
      "epoch": 0.4194245105552846,
      "grad_norm": 0.43969300389289856,
      "learning_rate": 2.7674348997120174e-05,
      "loss": 0.08995914459228516,
      "step": 3010
    },
    {
      "epoch": 0.419563854246499,
      "grad_norm": 0.38074493408203125,
      "learning_rate": 2.7665642652143327e-05,
      "loss": 0.10424518585205078,
      "step": 3011
    },
    {
      "epoch": 0.4197031979377134,
      "grad_norm": 0.3844943940639496,
      "learning_rate": 2.7656934604035694e-05,
      "loss": 0.09274864196777344,
      "step": 3012
    },
    {
      "epoch": 0.41984254162892776,
      "grad_norm": 0.4631916582584381,
      "learning_rate": 2.7648224854732005e-05,
      "loss": 0.09597206115722656,
      "step": 3013
    },
    {
      "epoch": 0.41998188532014213,
      "grad_norm": 0.2502036690711975,
      "learning_rate": 2.7639513406167363e-05,
      "loss": 0.0676126480102539,
      "step": 3014
    },
    {
      "epoch": 0.4201212290113565,
      "grad_norm": 0.2362333983182907,
      "learning_rate": 2.763080026027726e-05,
      "loss": 0.07489967346191406,
      "step": 3015
    },
    {
      "epoch": 0.4202605727025709,
      "grad_norm": 0.3751072287559509,
      "learning_rate": 2.762208541899755e-05,
      "loss": 0.093963623046875,
      "step": 3016
    },
    {
      "epoch": 0.42039991639378527,
      "grad_norm": 0.3404616415500641,
      "learning_rate": 2.761336888426448e-05,
      "loss": 0.06876182556152344,
      "step": 3017
    },
    {
      "epoch": 0.42053926008499964,
      "grad_norm": 0.45011523365974426,
      "learning_rate": 2.7604650658014648e-05,
      "loss": 0.09289169311523438,
      "step": 3018
    },
    {
      "epoch": 0.420678603776214,
      "grad_norm": 0.26344069838523865,
      "learning_rate": 2.7595930742185068e-05,
      "loss": 0.07769203186035156,
      "step": 3019
    },
    {
      "epoch": 0.4208179474674284,
      "grad_norm": 0.9897775650024414,
      "learning_rate": 2.758720913871309e-05,
      "loss": 0.12956809997558594,
      "step": 3020
    },
    {
      "epoch": 0.4209572911586428,
      "grad_norm": 0.9581022262573242,
      "learning_rate": 2.7578485849536464e-05,
      "loss": 0.11178398132324219,
      "step": 3021
    },
    {
      "epoch": 0.42109663484985715,
      "grad_norm": 0.31706711649894714,
      "learning_rate": 2.7569760876593298e-05,
      "loss": 0.0904083251953125,
      "step": 3022
    },
    {
      "epoch": 0.42123597854107153,
      "grad_norm": 0.41639766097068787,
      "learning_rate": 2.7561034221822085e-05,
      "loss": 0.09426307678222656,
      "step": 3023
    },
    {
      "epoch": 0.4213753222322859,
      "grad_norm": 0.6080905795097351,
      "learning_rate": 2.7552305887161693e-05,
      "loss": 0.16414642333984375,
      "step": 3024
    },
    {
      "epoch": 0.4215146659235003,
      "grad_norm": 0.8979246020317078,
      "learning_rate": 2.754357587455135e-05,
      "loss": 0.11898994445800781,
      "step": 3025
    },
    {
      "epoch": 0.4216540096147147,
      "grad_norm": 0.6686037182807922,
      "learning_rate": 2.7534844185930674e-05,
      "loss": 0.106475830078125,
      "step": 3026
    },
    {
      "epoch": 0.4217933533059291,
      "grad_norm": 0.4539884924888611,
      "learning_rate": 2.7526110823239647e-05,
      "loss": 0.10874176025390625,
      "step": 3027
    },
    {
      "epoch": 0.4219326969971435,
      "grad_norm": 0.4164336323738098,
      "learning_rate": 2.7517375788418613e-05,
      "loss": 0.11487579345703125,
      "step": 3028
    },
    {
      "epoch": 0.42207204068835785,
      "grad_norm": 0.3385312259197235,
      "learning_rate": 2.7508639083408306e-05,
      "loss": 0.08850669860839844,
      "step": 3029
    },
    {
      "epoch": 0.42221138437957223,
      "grad_norm": 0.3954416513442993,
      "learning_rate": 2.749990071014982e-05,
      "loss": 0.09598731994628906,
      "step": 3030
    },
    {
      "epoch": 0.4223507280707866,
      "grad_norm": 0.6479269862174988,
      "learning_rate": 2.749116067058462e-05,
      "loss": 0.1256732940673828,
      "step": 3031
    },
    {
      "epoch": 0.422490071762001,
      "grad_norm": 0.7110391855239868,
      "learning_rate": 2.748241896665455e-05,
      "loss": 0.12057876586914062,
      "step": 3032
    },
    {
      "epoch": 0.42262941545321536,
      "grad_norm": 0.5215420722961426,
      "learning_rate": 2.7473675600301807e-05,
      "loss": 0.08901786804199219,
      "step": 3033
    },
    {
      "epoch": 0.42276875914442974,
      "grad_norm": 0.5117837190628052,
      "learning_rate": 2.7464930573468973e-05,
      "loss": 0.07907485961914062,
      "step": 3034
    },
    {
      "epoch": 0.4229081028356441,
      "grad_norm": 0.29275521636009216,
      "learning_rate": 2.7456183888098995e-05,
      "loss": 0.06429100036621094,
      "step": 3035
    },
    {
      "epoch": 0.4230474465268585,
      "grad_norm": 0.5966756343841553,
      "learning_rate": 2.7447435546135186e-05,
      "loss": 0.08964920043945312,
      "step": 3036
    },
    {
      "epoch": 0.4231867902180729,
      "grad_norm": 0.7810686826705933,
      "learning_rate": 2.7438685549521228e-05,
      "loss": 0.12935447692871094,
      "step": 3037
    },
    {
      "epoch": 0.42332613390928725,
      "grad_norm": 0.30725204944610596,
      "learning_rate": 2.742993390020116e-05,
      "loss": 0.07145309448242188,
      "step": 3038
    },
    {
      "epoch": 0.4234654776005016,
      "grad_norm": 0.5059618949890137,
      "learning_rate": 2.742118060011941e-05,
      "loss": 0.0894937515258789,
      "step": 3039
    },
    {
      "epoch": 0.423604821291716,
      "grad_norm": 0.2828565835952759,
      "learning_rate": 2.7412425651220767e-05,
      "loss": 0.07688045501708984,
      "step": 3040
    },
    {
      "epoch": 0.4237441649829304,
      "grad_norm": 0.6171338558197021,
      "learning_rate": 2.7403669055450363e-05,
      "loss": 0.11657905578613281,
      "step": 3041
    },
    {
      "epoch": 0.42388350867414476,
      "grad_norm": 0.305751234292984,
      "learning_rate": 2.739491081475373e-05,
      "loss": 0.0838155746459961,
      "step": 3042
    },
    {
      "epoch": 0.42402285236535914,
      "grad_norm": 0.577755868434906,
      "learning_rate": 2.738615093107674e-05,
      "loss": 0.10146141052246094,
      "step": 3043
    },
    {
      "epoch": 0.4241621960565735,
      "grad_norm": 0.6004830598831177,
      "learning_rate": 2.7377389406365642e-05,
      "loss": 0.11208343505859375,
      "step": 3044
    },
    {
      "epoch": 0.4243015397477879,
      "grad_norm": 0.32550421357154846,
      "learning_rate": 2.7368626242567046e-05,
      "loss": 0.08430862426757812,
      "step": 3045
    },
    {
      "epoch": 0.4244408834390023,
      "grad_norm": 0.2881087064743042,
      "learning_rate": 2.735986144162793e-05,
      "loss": 0.0760040283203125,
      "step": 3046
    },
    {
      "epoch": 0.4245802271302167,
      "grad_norm": 0.4742453694343567,
      "learning_rate": 2.735109500549563e-05,
      "loss": 0.08654022216796875,
      "step": 3047
    },
    {
      "epoch": 0.4247195708214311,
      "grad_norm": 0.435769259929657,
      "learning_rate": 2.7342326936117847e-05,
      "loss": 0.10020637512207031,
      "step": 3048
    },
    {
      "epoch": 0.42485891451264546,
      "grad_norm": 0.34583523869514465,
      "learning_rate": 2.7333557235442648e-05,
      "loss": 0.08666181564331055,
      "step": 3049
    },
    {
      "epoch": 0.42499825820385984,
      "grad_norm": 0.4487597346305847,
      "learning_rate": 2.732478590541846e-05,
      "loss": 0.07819366455078125,
      "step": 3050
    },
    {
      "epoch": 0.4251376018950742,
      "grad_norm": 0.5147772431373596,
      "learning_rate": 2.7316012947994067e-05,
      "loss": 0.12531471252441406,
      "step": 3051
    },
    {
      "epoch": 0.4252769455862886,
      "grad_norm": 0.2830198407173157,
      "learning_rate": 2.730723836511863e-05,
      "loss": 0.06543350219726562,
      "step": 3052
    },
    {
      "epoch": 0.42541628927750297,
      "grad_norm": 0.2761100232601166,
      "learning_rate": 2.729846215874165e-05,
      "loss": 0.07118797302246094,
      "step": 3053
    },
    {
      "epoch": 0.42555563296871735,
      "grad_norm": 0.46774086356163025,
      "learning_rate": 2.728968433081301e-05,
      "loss": 0.11004447937011719,
      "step": 3054
    },
    {
      "epoch": 0.4256949766599317,
      "grad_norm": 0.40086862444877625,
      "learning_rate": 2.728090488328293e-05,
      "loss": 0.0838165283203125,
      "step": 3055
    },
    {
      "epoch": 0.4258343203511461,
      "grad_norm": 0.5478135347366333,
      "learning_rate": 2.727212381810202e-05,
      "loss": 0.09226226806640625,
      "step": 3056
    },
    {
      "epoch": 0.4259736640423605,
      "grad_norm": 0.44372865557670593,
      "learning_rate": 2.7263341137221217e-05,
      "loss": 0.09789466857910156,
      "step": 3057
    },
    {
      "epoch": 0.42611300773357486,
      "grad_norm": 0.6014997363090515,
      "learning_rate": 2.725455684259185e-05,
      "loss": 0.10030174255371094,
      "step": 3058
    },
    {
      "epoch": 0.42625235142478923,
      "grad_norm": 0.25951048731803894,
      "learning_rate": 2.724577093616556e-05,
      "loss": 0.07194232940673828,
      "step": 3059
    },
    {
      "epoch": 0.4263916951160036,
      "grad_norm": 0.3270566761493683,
      "learning_rate": 2.72369834198944e-05,
      "loss": 0.078399658203125,
      "step": 3060
    },
    {
      "epoch": 0.426531038807218,
      "grad_norm": 0.31611230969429016,
      "learning_rate": 2.7228194295730747e-05,
      "loss": 0.08364486694335938,
      "step": 3061
    },
    {
      "epoch": 0.42667038249843237,
      "grad_norm": 0.5268136262893677,
      "learning_rate": 2.7219403565627342e-05,
      "loss": 0.11836051940917969,
      "step": 3062
    },
    {
      "epoch": 0.42680972618964674,
      "grad_norm": 0.4463173449039459,
      "learning_rate": 2.721061123153729e-05,
      "loss": 0.11009025573730469,
      "step": 3063
    },
    {
      "epoch": 0.4269490698808611,
      "grad_norm": 0.4011792242527008,
      "learning_rate": 2.720181729541404e-05,
      "loss": 0.08667755126953125,
      "step": 3064
    },
    {
      "epoch": 0.4270884135720755,
      "grad_norm": 0.4120916426181793,
      "learning_rate": 2.719302175921141e-05,
      "loss": 0.08592605590820312,
      "step": 3065
    },
    {
      "epoch": 0.42722775726328993,
      "grad_norm": 0.35560086369514465,
      "learning_rate": 2.7184224624883566e-05,
      "loss": 0.08373451232910156,
      "step": 3066
    },
    {
      "epoch": 0.4273671009545043,
      "grad_norm": 0.37799519300460815,
      "learning_rate": 2.7175425894385026e-05,
      "loss": 0.08482170104980469,
      "step": 3067
    },
    {
      "epoch": 0.4275064446457187,
      "grad_norm": 0.4046212136745453,
      "learning_rate": 2.7166625569670664e-05,
      "loss": 0.08334159851074219,
      "step": 3068
    },
    {
      "epoch": 0.42764578833693306,
      "grad_norm": 0.23191151022911072,
      "learning_rate": 2.715782365269573e-05,
      "loss": 0.05487632751464844,
      "step": 3069
    },
    {
      "epoch": 0.42778513202814744,
      "grad_norm": 0.40622830390930176,
      "learning_rate": 2.714902014541579e-05,
      "loss": 0.09069061279296875,
      "step": 3070
    },
    {
      "epoch": 0.4279244757193618,
      "grad_norm": 0.7446821331977844,
      "learning_rate": 2.7140215049786783e-05,
      "loss": 0.105987548828125,
      "step": 3071
    },
    {
      "epoch": 0.4280638194105762,
      "grad_norm": 0.3094369173049927,
      "learning_rate": 2.7131408367765017e-05,
      "loss": 0.08099365234375,
      "step": 3072
    },
    {
      "epoch": 0.4282031631017906,
      "grad_norm": 0.538990318775177,
      "learning_rate": 2.7122600101307113e-05,
      "loss": 0.10638427734375,
      "step": 3073
    },
    {
      "epoch": 0.42834250679300495,
      "grad_norm": 0.35945823788642883,
      "learning_rate": 2.7113790252370093e-05,
      "loss": 0.0973052978515625,
      "step": 3074
    },
    {
      "epoch": 0.42848185048421933,
      "grad_norm": 0.5654674768447876,
      "learning_rate": 2.710497882291127e-05,
      "loss": 0.11312675476074219,
      "step": 3075
    },
    {
      "epoch": 0.4286211941754337,
      "grad_norm": 0.6413313150405884,
      "learning_rate": 2.7096165814888373e-05,
      "loss": 0.1388416290283203,
      "step": 3076
    },
    {
      "epoch": 0.4287605378666481,
      "grad_norm": 0.3120116591453552,
      "learning_rate": 2.7087351230259442e-05,
      "loss": 0.09258651733398438,
      "step": 3077
    },
    {
      "epoch": 0.42889988155786246,
      "grad_norm": 0.4430004358291626,
      "learning_rate": 2.7078535070982873e-05,
      "loss": 0.08355331420898438,
      "step": 3078
    },
    {
      "epoch": 0.42903922524907684,
      "grad_norm": 0.4878310561180115,
      "learning_rate": 2.7069717339017415e-05,
      "loss": 0.10353469848632812,
      "step": 3079
    },
    {
      "epoch": 0.4291785689402912,
      "grad_norm": 0.6105650663375854,
      "learning_rate": 2.706089803632217e-05,
      "loss": 0.10670280456542969,
      "step": 3080
    },
    {
      "epoch": 0.4293179126315056,
      "grad_norm": 0.6068555116653442,
      "learning_rate": 2.7052077164856584e-05,
      "loss": 0.0952911376953125,
      "step": 3081
    },
    {
      "epoch": 0.42945725632271997,
      "grad_norm": 0.4432108700275421,
      "learning_rate": 2.7043254726580457e-05,
      "loss": 0.10673332214355469,
      "step": 3082
    },
    {
      "epoch": 0.42959660001393435,
      "grad_norm": 0.45219582319259644,
      "learning_rate": 2.7034430723453925e-05,
      "loss": 0.09998226165771484,
      "step": 3083
    },
    {
      "epoch": 0.4297359437051487,
      "grad_norm": 0.7268244624137878,
      "learning_rate": 2.7025605157437483e-05,
      "loss": 0.10045242309570312,
      "step": 3084
    },
    {
      "epoch": 0.4298752873963631,
      "grad_norm": 0.5471758842468262,
      "learning_rate": 2.701677803049198e-05,
      "loss": 0.09918689727783203,
      "step": 3085
    },
    {
      "epoch": 0.43001463108757754,
      "grad_norm": 0.4069291651248932,
      "learning_rate": 2.700794934457859e-05,
      "loss": 0.0905914306640625,
      "step": 3086
    },
    {
      "epoch": 0.4301539747787919,
      "grad_norm": 0.43674585223197937,
      "learning_rate": 2.6999119101658854e-05,
      "loss": 0.08358478546142578,
      "step": 3087
    },
    {
      "epoch": 0.4302933184700063,
      "grad_norm": 0.41119179129600525,
      "learning_rate": 2.699028730369464e-05,
      "loss": 0.07599544525146484,
      "step": 3088
    },
    {
      "epoch": 0.43043266216122067,
      "grad_norm": 0.5414245128631592,
      "learning_rate": 2.6981453952648178e-05,
      "loss": 0.10011863708496094,
      "step": 3089
    },
    {
      "epoch": 0.43057200585243505,
      "grad_norm": 0.29360273480415344,
      "learning_rate": 2.6972619050482044e-05,
      "loss": 0.08612823486328125,
      "step": 3090
    },
    {
      "epoch": 0.4307113495436494,
      "grad_norm": 0.6129921078681946,
      "learning_rate": 2.6963782599159135e-05,
      "loss": 0.08502769470214844,
      "step": 3091
    },
    {
      "epoch": 0.4308506932348638,
      "grad_norm": 0.49789607524871826,
      "learning_rate": 2.6954944600642724e-05,
      "loss": 0.11270904541015625,
      "step": 3092
    },
    {
      "epoch": 0.4309900369260782,
      "grad_norm": 0.4156376123428345,
      "learning_rate": 2.6946105056896406e-05,
      "loss": 0.0885624885559082,
      "step": 3093
    },
    {
      "epoch": 0.43112938061729256,
      "grad_norm": 0.28965625166893005,
      "learning_rate": 2.693726396988413e-05,
      "loss": 0.06447792053222656,
      "step": 3094
    },
    {
      "epoch": 0.43126872430850693,
      "grad_norm": 0.40528684854507446,
      "learning_rate": 2.6928421341570178e-05,
      "loss": 0.09171295166015625,
      "step": 3095
    },
    {
      "epoch": 0.4314080679997213,
      "grad_norm": 0.6176245212554932,
      "learning_rate": 2.691957717391918e-05,
      "loss": 0.09287834167480469,
      "step": 3096
    },
    {
      "epoch": 0.4315474116909357,
      "grad_norm": 0.3871535658836365,
      "learning_rate": 2.6910731468896112e-05,
      "loss": 0.07746601104736328,
      "step": 3097
    },
    {
      "epoch": 0.43168675538215007,
      "grad_norm": 0.502102792263031,
      "learning_rate": 2.690188422846629e-05,
      "loss": 0.11438941955566406,
      "step": 3098
    },
    {
      "epoch": 0.43182609907336444,
      "grad_norm": 0.5268672108650208,
      "learning_rate": 2.6893035454595363e-05,
      "loss": 0.12081146240234375,
      "step": 3099
    },
    {
      "epoch": 0.4319654427645788,
      "grad_norm": 0.8232094049453735,
      "learning_rate": 2.688418514924932e-05,
      "loss": 0.11641883850097656,
      "step": 3100
    },
    {
      "epoch": 0.4321047864557932,
      "grad_norm": 0.41472020745277405,
      "learning_rate": 2.6875333314394517e-05,
      "loss": 0.09449195861816406,
      "step": 3101
    },
    {
      "epoch": 0.4322441301470076,
      "grad_norm": 0.5742473006248474,
      "learning_rate": 2.6866479951997616e-05,
      "loss": 0.08580589294433594,
      "step": 3102
    },
    {
      "epoch": 0.43238347383822195,
      "grad_norm": 0.3602655529975891,
      "learning_rate": 2.685762506402563e-05,
      "loss": 0.08642578125,
      "step": 3103
    },
    {
      "epoch": 0.43252281752943633,
      "grad_norm": 0.8047970533370972,
      "learning_rate": 2.6848768652445924e-05,
      "loss": 0.14673709869384766,
      "step": 3104
    },
    {
      "epoch": 0.4326621612206507,
      "grad_norm": 0.42502206563949585,
      "learning_rate": 2.6839910719226173e-05,
      "loss": 0.09339523315429688,
      "step": 3105
    },
    {
      "epoch": 0.43280150491186514,
      "grad_norm": 0.4490775167942047,
      "learning_rate": 2.683105126633443e-05,
      "loss": 0.11094474792480469,
      "step": 3106
    },
    {
      "epoch": 0.4329408486030795,
      "grad_norm": 0.4816376566886902,
      "learning_rate": 2.6822190295739038e-05,
      "loss": 0.09279060363769531,
      "step": 3107
    },
    {
      "epoch": 0.4330801922942939,
      "grad_norm": 0.6091939806938171,
      "learning_rate": 2.6813327809408723e-05,
      "loss": 0.10823249816894531,
      "step": 3108
    },
    {
      "epoch": 0.4332195359855083,
      "grad_norm": 0.5936306715011597,
      "learning_rate": 2.680446380931252e-05,
      "loss": 0.10489463806152344,
      "step": 3109
    },
    {
      "epoch": 0.43335887967672265,
      "grad_norm": 0.5704661011695862,
      "learning_rate": 2.6795598297419806e-05,
      "loss": 0.10129547119140625,
      "step": 3110
    },
    {
      "epoch": 0.43349822336793703,
      "grad_norm": 0.28526198863983154,
      "learning_rate": 2.6786731275700297e-05,
      "loss": 0.07546138763427734,
      "step": 3111
    },
    {
      "epoch": 0.4336375670591514,
      "grad_norm": 0.40714725852012634,
      "learning_rate": 2.6777862746124045e-05,
      "loss": 0.08675289154052734,
      "step": 3112
    },
    {
      "epoch": 0.4337769107503658,
      "grad_norm": 0.6405243873596191,
      "learning_rate": 2.6768992710661428e-05,
      "loss": 0.11686038970947266,
      "step": 3113
    },
    {
      "epoch": 0.43391625444158016,
      "grad_norm": 0.249418243765831,
      "learning_rate": 2.676012117128317e-05,
      "loss": 0.06254196166992188,
      "step": 3114
    },
    {
      "epoch": 0.43405559813279454,
      "grad_norm": 0.5908002853393555,
      "learning_rate": 2.6751248129960323e-05,
      "loss": 0.10704803466796875,
      "step": 3115
    },
    {
      "epoch": 0.4341949418240089,
      "grad_norm": 0.3314914405345917,
      "learning_rate": 2.6742373588664276e-05,
      "loss": 0.09360408782958984,
      "step": 3116
    },
    {
      "epoch": 0.4343342855152233,
      "grad_norm": 0.24369199573993683,
      "learning_rate": 2.673349754936675e-05,
      "loss": 0.06453514099121094,
      "step": 3117
    },
    {
      "epoch": 0.4344736292064377,
      "grad_norm": 0.3088749945163727,
      "learning_rate": 2.6724620014039794e-05,
      "loss": 0.0769195556640625,
      "step": 3118
    },
    {
      "epoch": 0.43461297289765205,
      "grad_norm": 0.607825517654419,
      "learning_rate": 2.67157409846558e-05,
      "loss": 0.13092994689941406,
      "step": 3119
    },
    {
      "epoch": 0.4347523165888664,
      "grad_norm": 0.5324974060058594,
      "learning_rate": 2.670686046318748e-05,
      "loss": 0.09222602844238281,
      "step": 3120
    },
    {
      "epoch": 0.4348916602800808,
      "grad_norm": 0.48396164178848267,
      "learning_rate": 2.669797845160788e-05,
      "loss": 0.10568428039550781,
      "step": 3121
    },
    {
      "epoch": 0.4350310039712952,
      "grad_norm": 0.2298593819141388,
      "learning_rate": 2.66890949518904e-05,
      "loss": 0.06252098083496094,
      "step": 3122
    },
    {
      "epoch": 0.43517034766250956,
      "grad_norm": 0.4656989574432373,
      "learning_rate": 2.6680209966008727e-05,
      "loss": 0.10078048706054688,
      "step": 3123
    },
    {
      "epoch": 0.43530969135372394,
      "grad_norm": 0.7908186316490173,
      "learning_rate": 2.6671323495936913e-05,
      "loss": 0.11881446838378906,
      "step": 3124
    },
    {
      "epoch": 0.4354490350449383,
      "grad_norm": 0.6851418018341064,
      "learning_rate": 2.666243554364932e-05,
      "loss": 0.09810638427734375,
      "step": 3125
    },
    {
      "epoch": 0.43558837873615275,
      "grad_norm": 0.5078226327896118,
      "learning_rate": 2.6653546111120664e-05,
      "loss": 0.1016073226928711,
      "step": 3126
    },
    {
      "epoch": 0.4357277224273671,
      "grad_norm": 0.38181978464126587,
      "learning_rate": 2.664465520032596e-05,
      "loss": 0.08620929718017578,
      "step": 3127
    },
    {
      "epoch": 0.4358670661185815,
      "grad_norm": 0.8450789451599121,
      "learning_rate": 2.6635762813240574e-05,
      "loss": 0.13358211517333984,
      "step": 3128
    },
    {
      "epoch": 0.4360064098097959,
      "grad_norm": 0.36622416973114014,
      "learning_rate": 2.662686895184019e-05,
      "loss": 0.06759262084960938,
      "step": 3129
    },
    {
      "epoch": 0.43614575350101026,
      "grad_norm": 0.389064222574234,
      "learning_rate": 2.6617973618100817e-05,
      "loss": 0.08028411865234375,
      "step": 3130
    },
    {
      "epoch": 0.43628509719222464,
      "grad_norm": 0.6169992685317993,
      "learning_rate": 2.6609076813998795e-05,
      "loss": 0.0865325927734375,
      "step": 3131
    },
    {
      "epoch": 0.436424440883439,
      "grad_norm": 0.5708268880844116,
      "learning_rate": 2.6600178541510792e-05,
      "loss": 0.08453083038330078,
      "step": 3132
    },
    {
      "epoch": 0.4365637845746534,
      "grad_norm": 0.5644346475601196,
      "learning_rate": 2.65912788026138e-05,
      "loss": 0.08947372436523438,
      "step": 3133
    },
    {
      "epoch": 0.43670312826586777,
      "grad_norm": 0.18803328275680542,
      "learning_rate": 2.6582377599285143e-05,
      "loss": 0.0657968521118164,
      "step": 3134
    },
    {
      "epoch": 0.43684247195708215,
      "grad_norm": 0.30600976943969727,
      "learning_rate": 2.6573474933502466e-05,
      "loss": 0.06857681274414062,
      "step": 3135
    },
    {
      "epoch": 0.4369818156482965,
      "grad_norm": 0.45744991302490234,
      "learning_rate": 2.6564570807243728e-05,
      "loss": 0.06744575500488281,
      "step": 3136
    },
    {
      "epoch": 0.4371211593395109,
      "grad_norm": 0.5566093325614929,
      "learning_rate": 2.655566522248723e-05,
      "loss": 0.09020137786865234,
      "step": 3137
    },
    {
      "epoch": 0.4372605030307253,
      "grad_norm": 0.5168794989585876,
      "learning_rate": 2.6546758181211593e-05,
      "loss": 0.10427665710449219,
      "step": 3138
    },
    {
      "epoch": 0.43739984672193966,
      "grad_norm": 0.6565899848937988,
      "learning_rate": 2.653784968539574e-05,
      "loss": 0.10200309753417969,
      "step": 3139
    },
    {
      "epoch": 0.43753919041315403,
      "grad_norm": 0.5680868625640869,
      "learning_rate": 2.652893973701896e-05,
      "loss": 0.1434001922607422,
      "step": 3140
    },
    {
      "epoch": 0.4376785341043684,
      "grad_norm": 0.4077380299568176,
      "learning_rate": 2.652002833806082e-05,
      "loss": 0.08700942993164062,
      "step": 3141
    },
    {
      "epoch": 0.4378178777955828,
      "grad_norm": 0.27713924646377563,
      "learning_rate": 2.6511115490501244e-05,
      "loss": 0.07441902160644531,
      "step": 3142
    },
    {
      "epoch": 0.43795722148679717,
      "grad_norm": 0.5776276588439941,
      "learning_rate": 2.650220119632046e-05,
      "loss": 0.1021881103515625,
      "step": 3143
    },
    {
      "epoch": 0.43809656517801154,
      "grad_norm": 0.2088954895734787,
      "learning_rate": 2.649328545749901e-05,
      "loss": 0.05626487731933594,
      "step": 3144
    },
    {
      "epoch": 0.4382359088692259,
      "grad_norm": 0.4736892580986023,
      "learning_rate": 2.648436827601778e-05,
      "loss": 0.09764480590820312,
      "step": 3145
    },
    {
      "epoch": 0.4383752525604403,
      "grad_norm": 0.275655597448349,
      "learning_rate": 2.6475449653857964e-05,
      "loss": 0.07464981079101562,
      "step": 3146
    },
    {
      "epoch": 0.43851459625165473,
      "grad_norm": 0.5600424408912659,
      "learning_rate": 2.6466529593001065e-05,
      "loss": 0.11087799072265625,
      "step": 3147
    },
    {
      "epoch": 0.4386539399428691,
      "grad_norm": 0.44646841287612915,
      "learning_rate": 2.6457608095428925e-05,
      "loss": 0.08103561401367188,
      "step": 3148
    },
    {
      "epoch": 0.4387932836340835,
      "grad_norm": 0.37932389974594116,
      "learning_rate": 2.64486851631237e-05,
      "loss": 0.07647895812988281,
      "step": 3149
    },
    {
      "epoch": 0.43893262732529786,
      "grad_norm": 0.5100890398025513,
      "learning_rate": 2.6439760798067854e-05,
      "loss": 0.09775733947753906,
      "step": 3150
    },
    {
      "epoch": 0.43907197101651224,
      "grad_norm": 0.6459035277366638,
      "learning_rate": 2.6430835002244183e-05,
      "loss": 0.10758018493652344,
      "step": 3151
    },
    {
      "epoch": 0.4392113147077266,
      "grad_norm": 0.3352036774158478,
      "learning_rate": 2.6421907777635793e-05,
      "loss": 0.08616065979003906,
      "step": 3152
    },
    {
      "epoch": 0.439350658398941,
      "grad_norm": 0.6230624914169312,
      "learning_rate": 2.641297912622611e-05,
      "loss": 0.11764907836914062,
      "step": 3153
    },
    {
      "epoch": 0.4394900020901554,
      "grad_norm": 0.4858156442642212,
      "learning_rate": 2.640404904999887e-05,
      "loss": 0.11335945129394531,
      "step": 3154
    },
    {
      "epoch": 0.43962934578136975,
      "grad_norm": 0.5678015947341919,
      "learning_rate": 2.639511755093814e-05,
      "loss": 0.1326580047607422,
      "step": 3155
    },
    {
      "epoch": 0.43976868947258413,
      "grad_norm": 0.3392687439918518,
      "learning_rate": 2.63861846310283e-05,
      "loss": 0.09939765930175781,
      "step": 3156
    },
    {
      "epoch": 0.4399080331637985,
      "grad_norm": 0.4600037634372711,
      "learning_rate": 2.6377250292254023e-05,
      "loss": 0.09518909454345703,
      "step": 3157
    },
    {
      "epoch": 0.4400473768550129,
      "grad_norm": 0.4338976740837097,
      "learning_rate": 2.6368314536600337e-05,
      "loss": 0.10296249389648438,
      "step": 3158
    },
    {
      "epoch": 0.44018672054622726,
      "grad_norm": 0.3142938017845154,
      "learning_rate": 2.6359377366052546e-05,
      "loss": 0.07747268676757812,
      "step": 3159
    },
    {
      "epoch": 0.44032606423744164,
      "grad_norm": 0.35865795612335205,
      "learning_rate": 2.6350438782596293e-05,
      "loss": 0.06481075286865234,
      "step": 3160
    },
    {
      "epoch": 0.440465407928656,
      "grad_norm": 0.7369528412818909,
      "learning_rate": 2.6341498788217527e-05,
      "loss": 0.13164520263671875,
      "step": 3161
    },
    {
      "epoch": 0.4406047516198704,
      "grad_norm": 0.2851244807243347,
      "learning_rate": 2.6332557384902506e-05,
      "loss": 0.07604408264160156,
      "step": 3162
    },
    {
      "epoch": 0.44074409531108477,
      "grad_norm": 0.2298494130373001,
      "learning_rate": 2.6323614574637812e-05,
      "loss": 0.07416200637817383,
      "step": 3163
    },
    {
      "epoch": 0.44088343900229915,
      "grad_norm": 0.4025443494319916,
      "learning_rate": 2.6314670359410332e-05,
      "loss": 0.08670997619628906,
      "step": 3164
    },
    {
      "epoch": 0.4410227826935135,
      "grad_norm": 0.515621542930603,
      "learning_rate": 2.630572474120726e-05,
      "loss": 0.10008668899536133,
      "step": 3165
    },
    {
      "epoch": 0.4411621263847279,
      "grad_norm": 0.28302887082099915,
      "learning_rate": 2.6296777722016108e-05,
      "loss": 0.07881355285644531,
      "step": 3166
    },
    {
      "epoch": 0.44130147007594234,
      "grad_norm": 0.509435772895813,
      "learning_rate": 2.6287829303824713e-05,
      "loss": 0.13525009155273438,
      "step": 3167
    },
    {
      "epoch": 0.4414408137671567,
      "grad_norm": 0.6929137110710144,
      "learning_rate": 2.6278879488621197e-05,
      "loss": 0.126373291015625,
      "step": 3168
    },
    {
      "epoch": 0.4415801574583711,
      "grad_norm": 0.42747148871421814,
      "learning_rate": 2.626992827839401e-05,
      "loss": 0.09135627746582031,
      "step": 3169
    },
    {
      "epoch": 0.44171950114958547,
      "grad_norm": 0.470749169588089,
      "learning_rate": 2.62609756751319e-05,
      "loss": 0.11327552795410156,
      "step": 3170
    },
    {
      "epoch": 0.44185884484079985,
      "grad_norm": 0.1951466202735901,
      "learning_rate": 2.6252021680823937e-05,
      "loss": 0.06481456756591797,
      "step": 3171
    },
    {
      "epoch": 0.4419981885320142,
      "grad_norm": 0.2759275436401367,
      "learning_rate": 2.6243066297459495e-05,
      "loss": 0.08458518981933594,
      "step": 3172
    },
    {
      "epoch": 0.4421375322232286,
      "grad_norm": 0.4868941903114319,
      "learning_rate": 2.623410952702825e-05,
      "loss": 0.11844444274902344,
      "step": 3173
    },
    {
      "epoch": 0.442276875914443,
      "grad_norm": 0.3588199019432068,
      "learning_rate": 2.62251513715202e-05,
      "loss": 0.10197257995605469,
      "step": 3174
    },
    {
      "epoch": 0.44241621960565736,
      "grad_norm": 0.4048108160495758,
      "learning_rate": 2.6216191832925634e-05,
      "loss": 0.09224319458007812,
      "step": 3175
    },
    {
      "epoch": 0.44255556329687173,
      "grad_norm": 0.21726645529270172,
      "learning_rate": 2.620723091323516e-05,
      "loss": 0.061919212341308594,
      "step": 3176
    },
    {
      "epoch": 0.4426949069880861,
      "grad_norm": 0.2929801344871521,
      "learning_rate": 2.6198268614439694e-05,
      "loss": 0.07763862609863281,
      "step": 3177
    },
    {
      "epoch": 0.4428342506793005,
      "grad_norm": 0.35297369956970215,
      "learning_rate": 2.618930493853045e-05,
      "loss": 0.094024658203125,
      "step": 3178
    },
    {
      "epoch": 0.44297359437051487,
      "grad_norm": 0.41006776690483093,
      "learning_rate": 2.618033988749895e-05,
      "loss": 0.10381317138671875,
      "step": 3179
    },
    {
      "epoch": 0.44311293806172924,
      "grad_norm": 0.5403784513473511,
      "learning_rate": 2.6171373463337028e-05,
      "loss": 0.1233210563659668,
      "step": 3180
    },
    {
      "epoch": 0.4432522817529436,
      "grad_norm": 0.2832668423652649,
      "learning_rate": 2.616240566803682e-05,
      "loss": 0.06217765808105469,
      "step": 3181
    },
    {
      "epoch": 0.443391625444158,
      "grad_norm": 0.39890509843826294,
      "learning_rate": 2.6153436503590765e-05,
      "loss": 0.11121845245361328,
      "step": 3182
    },
    {
      "epoch": 0.4435309691353724,
      "grad_norm": 0.4453290104866028,
      "learning_rate": 2.6144465971991596e-05,
      "loss": 0.1016693115234375,
      "step": 3183
    },
    {
      "epoch": 0.44367031282658675,
      "grad_norm": 0.3747658133506775,
      "learning_rate": 2.6135494075232366e-05,
      "loss": 0.11945152282714844,
      "step": 3184
    },
    {
      "epoch": 0.44380965651780113,
      "grad_norm": 0.3644007444381714,
      "learning_rate": 2.612652081530644e-05,
      "loss": 0.09680938720703125,
      "step": 3185
    },
    {
      "epoch": 0.4439490002090155,
      "grad_norm": 0.3593016564846039,
      "learning_rate": 2.6117546194207454e-05,
      "loss": 0.091400146484375,
      "step": 3186
    },
    {
      "epoch": 0.44408834390022994,
      "grad_norm": 0.37312301993370056,
      "learning_rate": 2.6108570213929366e-05,
      "loss": 0.07148265838623047,
      "step": 3187
    },
    {
      "epoch": 0.4442276875914443,
      "grad_norm": 0.18219980597496033,
      "learning_rate": 2.609959287646645e-05,
      "loss": 0.065521240234375,
      "step": 3188
    },
    {
      "epoch": 0.4443670312826587,
      "grad_norm": 0.4620741307735443,
      "learning_rate": 2.6090614183813236e-05,
      "loss": 0.08932113647460938,
      "step": 3189
    },
    {
      "epoch": 0.4445063749738731,
      "grad_norm": 0.27734750509262085,
      "learning_rate": 2.6081634137964615e-05,
      "loss": 0.07478904724121094,
      "step": 3190
    },
    {
      "epoch": 0.44464571866508745,
      "grad_norm": 0.47322383522987366,
      "learning_rate": 2.607265274091573e-05,
      "loss": 0.10181236267089844,
      "step": 3191
    },
    {
      "epoch": 0.44478506235630183,
      "grad_norm": 0.8202535510063171,
      "learning_rate": 2.6063669994662043e-05,
      "loss": 0.1680927276611328,
      "step": 3192
    },
    {
      "epoch": 0.4449244060475162,
      "grad_norm": 0.35152265429496765,
      "learning_rate": 2.605468590119932e-05,
      "loss": 0.0787191390991211,
      "step": 3193
    },
    {
      "epoch": 0.4450637497387306,
      "grad_norm": 0.2865127623081207,
      "learning_rate": 2.6045700462523625e-05,
      "loss": 0.08188438415527344,
      "step": 3194
    },
    {
      "epoch": 0.44520309342994496,
      "grad_norm": 0.3493317663669586,
      "learning_rate": 2.6036713680631312e-05,
      "loss": 0.07248115539550781,
      "step": 3195
    },
    {
      "epoch": 0.44534243712115934,
      "grad_norm": 0.34856924414634705,
      "learning_rate": 2.6027725557519037e-05,
      "loss": 0.08799171447753906,
      "step": 3196
    },
    {
      "epoch": 0.4454817808123737,
      "grad_norm": 0.3275778293609619,
      "learning_rate": 2.601873609518376e-05,
      "loss": 0.09891700744628906,
      "step": 3197
    },
    {
      "epoch": 0.4456211245035881,
      "grad_norm": 0.6363255381584167,
      "learning_rate": 2.600974529562273e-05,
      "loss": 0.11921310424804688,
      "step": 3198
    },
    {
      "epoch": 0.4457604681948025,
      "grad_norm": 0.2859232723712921,
      "learning_rate": 2.6000753160833506e-05,
      "loss": 0.07242870330810547,
      "step": 3199
    },
    {
      "epoch": 0.44589981188601685,
      "grad_norm": 0.30385327339172363,
      "learning_rate": 2.599175969281392e-05,
      "loss": 0.08919525146484375,
      "step": 3200
    },
    {
      "epoch": 0.44603915557723123,
      "grad_norm": 0.4143965542316437,
      "learning_rate": 2.5982764893562137e-05,
      "loss": 0.10561561584472656,
      "step": 3201
    },
    {
      "epoch": 0.4461784992684456,
      "grad_norm": 0.35189366340637207,
      "learning_rate": 2.5973768765076578e-05,
      "loss": 0.08381462097167969,
      "step": 3202
    },
    {
      "epoch": 0.44631784295966,
      "grad_norm": 0.4816943109035492,
      "learning_rate": 2.5964771309355978e-05,
      "loss": 0.13361740112304688,
      "step": 3203
    },
    {
      "epoch": 0.44645718665087436,
      "grad_norm": 0.23214365541934967,
      "learning_rate": 2.595577252839938e-05,
      "loss": 0.07532501220703125,
      "step": 3204
    },
    {
      "epoch": 0.44659653034208874,
      "grad_norm": 0.40004539489746094,
      "learning_rate": 2.594677242420609e-05,
      "loss": 0.11186981201171875,
      "step": 3205
    },
    {
      "epoch": 0.4467358740333031,
      "grad_norm": 0.36163726449012756,
      "learning_rate": 2.593777099877574e-05,
      "loss": 0.08881187438964844,
      "step": 3206
    },
    {
      "epoch": 0.44687521772451755,
      "grad_norm": 0.5212134718894958,
      "learning_rate": 2.592876825410823e-05,
      "loss": 0.10515213012695312,
      "step": 3207
    },
    {
      "epoch": 0.4470145614157319,
      "grad_norm": 0.511530876159668,
      "learning_rate": 2.5919764192203777e-05,
      "loss": 0.08733177185058594,
      "step": 3208
    },
    {
      "epoch": 0.4471539051069463,
      "grad_norm": 0.3775254786014557,
      "learning_rate": 2.591075881506287e-05,
      "loss": 0.09482669830322266,
      "step": 3209
    },
    {
      "epoch": 0.4472932487981607,
      "grad_norm": 0.2528819441795349,
      "learning_rate": 2.5901752124686294e-05,
      "loss": 0.06436920166015625,
      "step": 3210
    },
    {
      "epoch": 0.44743259248937506,
      "grad_norm": 0.38951048254966736,
      "learning_rate": 2.5892744123075138e-05,
      "loss": 0.090911865234375,
      "step": 3211
    },
    {
      "epoch": 0.44757193618058944,
      "grad_norm": 0.4155295491218567,
      "learning_rate": 2.5883734812230773e-05,
      "loss": 0.09823989868164062,
      "step": 3212
    },
    {
      "epoch": 0.4477112798718038,
      "grad_norm": 0.27602076530456543,
      "learning_rate": 2.587472419415486e-05,
      "loss": 0.08160400390625,
      "step": 3213
    },
    {
      "epoch": 0.4478506235630182,
      "grad_norm": 0.5807915329933167,
      "learning_rate": 2.5865712270849354e-05,
      "loss": 0.1274738311767578,
      "step": 3214
    },
    {
      "epoch": 0.44798996725423257,
      "grad_norm": 0.31890591979026794,
      "learning_rate": 2.5856699044316496e-05,
      "loss": 0.08365440368652344,
      "step": 3215
    },
    {
      "epoch": 0.44812931094544695,
      "grad_norm": 0.6423806548118591,
      "learning_rate": 2.5847684516558817e-05,
      "loss": 0.11694145202636719,
      "step": 3216
    },
    {
      "epoch": 0.4482686546366613,
      "grad_norm": 0.5192328095436096,
      "learning_rate": 2.583866868957915e-05,
      "loss": 0.0835113525390625,
      "step": 3217
    },
    {
      "epoch": 0.4484079983278757,
      "grad_norm": 0.4879690408706665,
      "learning_rate": 2.5829651565380598e-05,
      "loss": 0.13028335571289062,
      "step": 3218
    },
    {
      "epoch": 0.4485473420190901,
      "grad_norm": 0.9135570526123047,
      "learning_rate": 2.5820633145966564e-05,
      "loss": 0.15550994873046875,
      "step": 3219
    },
    {
      "epoch": 0.44868668571030446,
      "grad_norm": 0.4413624703884125,
      "learning_rate": 2.581161343334073e-05,
      "loss": 0.10725593566894531,
      "step": 3220
    },
    {
      "epoch": 0.44882602940151883,
      "grad_norm": 0.8122366070747375,
      "learning_rate": 2.5802592429507067e-05,
      "loss": 0.1569499969482422,
      "step": 3221
    },
    {
      "epoch": 0.4489653730927332,
      "grad_norm": 0.38626131415367126,
      "learning_rate": 2.579357013646985e-05,
      "loss": 0.08245468139648438,
      "step": 3222
    },
    {
      "epoch": 0.4491047167839476,
      "grad_norm": 0.4608629047870636,
      "learning_rate": 2.578454655623361e-05,
      "loss": 0.1023550033569336,
      "step": 3223
    },
    {
      "epoch": 0.44924406047516197,
      "grad_norm": 0.40798383951187134,
      "learning_rate": 2.5775521690803197e-05,
      "loss": 0.08950042724609375,
      "step": 3224
    },
    {
      "epoch": 0.44938340416637634,
      "grad_norm": 0.28237906098365784,
      "learning_rate": 2.5766495542183717e-05,
      "loss": 0.06744194030761719,
      "step": 3225
    },
    {
      "epoch": 0.4495227478575907,
      "grad_norm": 0.5146475434303284,
      "learning_rate": 2.575746811238058e-05,
      "loss": 0.10599708557128906,
      "step": 3226
    },
    {
      "epoch": 0.44966209154880515,
      "grad_norm": 0.4172583520412445,
      "learning_rate": 2.574843940339947e-05,
      "loss": 0.08574104309082031,
      "step": 3227
    },
    {
      "epoch": 0.44980143524001953,
      "grad_norm": 0.4712950885295868,
      "learning_rate": 2.5739409417246367e-05,
      "loss": 0.11024093627929688,
      "step": 3228
    },
    {
      "epoch": 0.4499407789312339,
      "grad_norm": 0.5308732986450195,
      "learning_rate": 2.5730378155927524e-05,
      "loss": 0.10855722427368164,
      "step": 3229
    },
    {
      "epoch": 0.4500801226224483,
      "grad_norm": 0.7482237219810486,
      "learning_rate": 2.5721345621449483e-05,
      "loss": 0.14079952239990234,
      "step": 3230
    },
    {
      "epoch": 0.45021946631366266,
      "grad_norm": 0.496581107378006,
      "learning_rate": 2.5712311815819063e-05,
      "loss": 0.08372879028320312,
      "step": 3231
    },
    {
      "epoch": 0.45035881000487704,
      "grad_norm": 0.4173150360584259,
      "learning_rate": 2.570327674104337e-05,
      "loss": 0.0843057632446289,
      "step": 3232
    },
    {
      "epoch": 0.4504981536960914,
      "grad_norm": 0.7348178029060364,
      "learning_rate": 2.56942403991298e-05,
      "loss": 0.1483745574951172,
      "step": 3233
    },
    {
      "epoch": 0.4506374973873058,
      "grad_norm": 0.3747888207435608,
      "learning_rate": 2.568520279208601e-05,
      "loss": 0.08870506286621094,
      "step": 3234
    },
    {
      "epoch": 0.4507768410785202,
      "grad_norm": 0.39361846446990967,
      "learning_rate": 2.5676163921919955e-05,
      "loss": 0.08873748779296875,
      "step": 3235
    },
    {
      "epoch": 0.45091618476973455,
      "grad_norm": 0.4448181986808777,
      "learning_rate": 2.566712379063987e-05,
      "loss": 0.11276054382324219,
      "step": 3236
    },
    {
      "epoch": 0.45105552846094893,
      "grad_norm": 0.5968166589736938,
      "learning_rate": 2.565808240025425e-05,
      "loss": 0.13665008544921875,
      "step": 3237
    },
    {
      "epoch": 0.4511948721521633,
      "grad_norm": 0.27689939737319946,
      "learning_rate": 2.5649039752771914e-05,
      "loss": 0.06540107727050781,
      "step": 3238
    },
    {
      "epoch": 0.4513342158433777,
      "grad_norm": 0.5549759864807129,
      "learning_rate": 2.5639995850201902e-05,
      "loss": 0.11178779602050781,
      "step": 3239
    },
    {
      "epoch": 0.45147355953459206,
      "grad_norm": 0.2442561835050583,
      "learning_rate": 2.5630950694553582e-05,
      "loss": 0.0769805908203125,
      "step": 3240
    },
    {
      "epoch": 0.45161290322580644,
      "grad_norm": 0.4681416451931,
      "learning_rate": 2.5621904287836568e-05,
      "loss": 0.11967658996582031,
      "step": 3241
    },
    {
      "epoch": 0.4517522469170208,
      "grad_norm": 0.3008041977882385,
      "learning_rate": 2.5612856632060776e-05,
      "loss": 0.07324504852294922,
      "step": 3242
    },
    {
      "epoch": 0.4518915906082352,
      "grad_norm": 0.5438188910484314,
      "learning_rate": 2.5603807729236387e-05,
      "loss": 0.11035728454589844,
      "step": 3243
    },
    {
      "epoch": 0.45203093429944957,
      "grad_norm": 0.485194593667984,
      "learning_rate": 2.559475758137385e-05,
      "loss": 0.0882568359375,
      "step": 3244
    },
    {
      "epoch": 0.45217027799066395,
      "grad_norm": 0.5395824909210205,
      "learning_rate": 2.5585706190483914e-05,
      "loss": 0.10652351379394531,
      "step": 3245
    },
    {
      "epoch": 0.4523096216818783,
      "grad_norm": 0.4361737370491028,
      "learning_rate": 2.5576653558577588e-05,
      "loss": 0.10880851745605469,
      "step": 3246
    },
    {
      "epoch": 0.45244896537309276,
      "grad_norm": 0.31356266140937805,
      "learning_rate": 2.556759968766615e-05,
      "loss": 0.09412574768066406,
      "step": 3247
    },
    {
      "epoch": 0.45258830906430714,
      "grad_norm": 0.3259497582912445,
      "learning_rate": 2.5558544579761177e-05,
      "loss": 0.08284187316894531,
      "step": 3248
    },
    {
      "epoch": 0.4527276527555215,
      "grad_norm": 0.4353976547718048,
      "learning_rate": 2.5549488236874506e-05,
      "loss": 0.07864761352539062,
      "step": 3249
    },
    {
      "epoch": 0.4528669964467359,
      "grad_norm": 0.7550287246704102,
      "learning_rate": 2.554043066101824e-05,
      "loss": 0.10595560073852539,
      "step": 3250
    },
    {
      "epoch": 0.45300634013795027,
      "grad_norm": 0.4789970815181732,
      "learning_rate": 2.5531371854204773e-05,
      "loss": 0.08000564575195312,
      "step": 3251
    },
    {
      "epoch": 0.45314568382916465,
      "grad_norm": 0.4191497564315796,
      "learning_rate": 2.5522311818446762e-05,
      "loss": 0.08799362182617188,
      "step": 3252
    },
    {
      "epoch": 0.453285027520379,
      "grad_norm": 0.6698973178863525,
      "learning_rate": 2.5513250555757143e-05,
      "loss": 0.10775947570800781,
      "step": 3253
    },
    {
      "epoch": 0.4534243712115934,
      "grad_norm": 0.8357587456703186,
      "learning_rate": 2.5504188068149126e-05,
      "loss": 0.1771373748779297,
      "step": 3254
    },
    {
      "epoch": 0.4535637149028078,
      "grad_norm": 0.3086981177330017,
      "learning_rate": 2.5495124357636174e-05,
      "loss": 0.07584953308105469,
      "step": 3255
    },
    {
      "epoch": 0.45370305859402216,
      "grad_norm": 0.37314149737358093,
      "learning_rate": 2.5486059426232052e-05,
      "loss": 0.0926198959350586,
      "step": 3256
    },
    {
      "epoch": 0.45384240228523653,
      "grad_norm": 0.5016579031944275,
      "learning_rate": 2.547699327595077e-05,
      "loss": 0.12589454650878906,
      "step": 3257
    },
    {
      "epoch": 0.4539817459764509,
      "grad_norm": 0.3545801043510437,
      "learning_rate": 2.5467925908806622e-05,
      "loss": 0.10574722290039062,
      "step": 3258
    },
    {
      "epoch": 0.4541210896676653,
      "grad_norm": 0.24271726608276367,
      "learning_rate": 2.5458857326814178e-05,
      "loss": 0.07344865798950195,
      "step": 3259
    },
    {
      "epoch": 0.45426043335887967,
      "grad_norm": 0.3427760899066925,
      "learning_rate": 2.5449787531988258e-05,
      "loss": 0.10113716125488281,
      "step": 3260
    },
    {
      "epoch": 0.45439977705009404,
      "grad_norm": 0.33136647939682007,
      "learning_rate": 2.5440716526343972e-05,
      "loss": 0.07993078231811523,
      "step": 3261
    },
    {
      "epoch": 0.4545391207413084,
      "grad_norm": 0.45315349102020264,
      "learning_rate": 2.543164431189669e-05,
      "loss": 0.10439491271972656,
      "step": 3262
    },
    {
      "epoch": 0.4546784644325228,
      "grad_norm": 0.43442025780677795,
      "learning_rate": 2.5422570890662046e-05,
      "loss": 0.10935020446777344,
      "step": 3263
    },
    {
      "epoch": 0.4548178081237372,
      "grad_norm": 0.4548247158527374,
      "learning_rate": 2.541349626465595e-05,
      "loss": 0.1125335693359375,
      "step": 3264
    },
    {
      "epoch": 0.45495715181495155,
      "grad_norm": 0.5460521578788757,
      "learning_rate": 2.5404420435894578e-05,
      "loss": 0.14475631713867188,
      "step": 3265
    },
    {
      "epoch": 0.45509649550616593,
      "grad_norm": 0.3531234860420227,
      "learning_rate": 2.539534340639436e-05,
      "loss": 0.08663749694824219,
      "step": 3266
    },
    {
      "epoch": 0.45523583919738037,
      "grad_norm": 0.5236177444458008,
      "learning_rate": 2.538626517817203e-05,
      "loss": 0.08452033996582031,
      "step": 3267
    },
    {
      "epoch": 0.45537518288859474,
      "grad_norm": 0.7299370169639587,
      "learning_rate": 2.5377185753244537e-05,
      "loss": 0.13871002197265625,
      "step": 3268
    },
    {
      "epoch": 0.4555145265798091,
      "grad_norm": 0.4558258056640625,
      "learning_rate": 2.5368105133629143e-05,
      "loss": 0.10540771484375,
      "step": 3269
    },
    {
      "epoch": 0.4556538702710235,
      "grad_norm": 0.3420353829860687,
      "learning_rate": 2.5359023321343336e-05,
      "loss": 0.07732009887695312,
      "step": 3270
    },
    {
      "epoch": 0.4557932139622379,
      "grad_norm": 0.4567525088787079,
      "learning_rate": 2.5349940318404895e-05,
      "loss": 0.060706138610839844,
      "step": 3271
    },
    {
      "epoch": 0.45593255765345225,
      "grad_norm": 0.4598923623561859,
      "learning_rate": 2.5340856126831864e-05,
      "loss": 0.10048866271972656,
      "step": 3272
    },
    {
      "epoch": 0.45607190134466663,
      "grad_norm": 0.3988206386566162,
      "learning_rate": 2.5331770748642527e-05,
      "loss": 0.09199714660644531,
      "step": 3273
    },
    {
      "epoch": 0.456211245035881,
      "grad_norm": 0.7231907844543457,
      "learning_rate": 2.5322684185855458e-05,
      "loss": 0.1343231201171875,
      "step": 3274
    },
    {
      "epoch": 0.4563505887270954,
      "grad_norm": 0.4518740773200989,
      "learning_rate": 2.5313596440489483e-05,
      "loss": 0.09477806091308594,
      "step": 3275
    },
    {
      "epoch": 0.45648993241830976,
      "grad_norm": 0.5364910364151001,
      "learning_rate": 2.530450751456369e-05,
      "loss": 0.08585739135742188,
      "step": 3276
    },
    {
      "epoch": 0.45662927610952414,
      "grad_norm": 0.36743879318237305,
      "learning_rate": 2.5295417410097437e-05,
      "loss": 0.07013893127441406,
      "step": 3277
    },
    {
      "epoch": 0.4567686198007385,
      "grad_norm": 0.6568399667739868,
      "learning_rate": 2.5286326129110325e-05,
      "loss": 0.11915779113769531,
      "step": 3278
    },
    {
      "epoch": 0.4569079634919529,
      "grad_norm": 0.49057039618492126,
      "learning_rate": 2.5277233673622236e-05,
      "loss": 0.10801506042480469,
      "step": 3279
    },
    {
      "epoch": 0.4570473071831673,
      "grad_norm": 0.5719925761222839,
      "learning_rate": 2.526814004565331e-05,
      "loss": 0.09646224975585938,
      "step": 3280
    },
    {
      "epoch": 0.45718665087438165,
      "grad_norm": 0.2587944269180298,
      "learning_rate": 2.5259045247223933e-05,
      "loss": 0.07669258117675781,
      "step": 3281
    },
    {
      "epoch": 0.45732599456559603,
      "grad_norm": 0.5176771879196167,
      "learning_rate": 2.524994928035477e-05,
      "loss": 0.09971141815185547,
      "step": 3282
    },
    {
      "epoch": 0.4574653382568104,
      "grad_norm": 0.4955536723136902,
      "learning_rate": 2.5240852147066742e-05,
      "loss": 0.10239791870117188,
      "step": 3283
    },
    {
      "epoch": 0.4576046819480248,
      "grad_norm": 0.5372061729431152,
      "learning_rate": 2.5231753849381013e-05,
      "loss": 0.13330459594726562,
      "step": 3284
    },
    {
      "epoch": 0.45774402563923916,
      "grad_norm": 0.43841353058815,
      "learning_rate": 2.5222654389319025e-05,
      "loss": 0.11880302429199219,
      "step": 3285
    },
    {
      "epoch": 0.45788336933045354,
      "grad_norm": 0.42088252305984497,
      "learning_rate": 2.5213553768902466e-05,
      "loss": 0.10557174682617188,
      "step": 3286
    },
    {
      "epoch": 0.45802271302166797,
      "grad_norm": 0.3761953115463257,
      "learning_rate": 2.520445199015328e-05,
      "loss": 0.08219718933105469,
      "step": 3287
    },
    {
      "epoch": 0.45816205671288235,
      "grad_norm": 0.31508657336235046,
      "learning_rate": 2.5195349055093693e-05,
      "loss": 0.06741714477539062,
      "step": 3288
    },
    {
      "epoch": 0.4583014004040967,
      "grad_norm": 0.4324892461299896,
      "learning_rate": 2.5186244965746146e-05,
      "loss": 0.07382488250732422,
      "step": 3289
    },
    {
      "epoch": 0.4584407440953111,
      "grad_norm": 0.20893768966197968,
      "learning_rate": 2.5177139724133376e-05,
      "loss": 0.05457782745361328,
      "step": 3290
    },
    {
      "epoch": 0.4585800877865255,
      "grad_norm": 0.4441015422344208,
      "learning_rate": 2.5168033332278358e-05,
      "loss": 0.10797500610351562,
      "step": 3291
    },
    {
      "epoch": 0.45871943147773986,
      "grad_norm": 0.31421732902526855,
      "learning_rate": 2.5158925792204317e-05,
      "loss": 0.07995796203613281,
      "step": 3292
    },
    {
      "epoch": 0.45885877516895424,
      "grad_norm": 0.3070082664489746,
      "learning_rate": 2.514981710593475e-05,
      "loss": 0.08426856994628906,
      "step": 3293
    },
    {
      "epoch": 0.4589981188601686,
      "grad_norm": 0.3476617634296417,
      "learning_rate": 2.5140707275493394e-05,
      "loss": 0.11507225036621094,
      "step": 3294
    },
    {
      "epoch": 0.459137462551383,
      "grad_norm": 0.33254557847976685,
      "learning_rate": 2.5131596302904245e-05,
      "loss": 0.08539772033691406,
      "step": 3295
    },
    {
      "epoch": 0.45927680624259737,
      "grad_norm": 0.2930243909358978,
      "learning_rate": 2.5122484190191553e-05,
      "loss": 0.07838249206542969,
      "step": 3296
    },
    {
      "epoch": 0.45941614993381175,
      "grad_norm": 0.4825838804244995,
      "learning_rate": 2.511337093937982e-05,
      "loss": 0.10203170776367188,
      "step": 3297
    },
    {
      "epoch": 0.4595554936250261,
      "grad_norm": 0.3848988711833954,
      "learning_rate": 2.510425655249381e-05,
      "loss": 0.08478736877441406,
      "step": 3298
    },
    {
      "epoch": 0.4596948373162405,
      "grad_norm": 0.4111996293067932,
      "learning_rate": 2.509514103155852e-05,
      "loss": 0.0934610366821289,
      "step": 3299
    },
    {
      "epoch": 0.4598341810074549,
      "grad_norm": 0.6232131123542786,
      "learning_rate": 2.5086024378599217e-05,
      "loss": 0.11145782470703125,
      "step": 3300
    },
    {
      "epoch": 0.45997352469866926,
      "grad_norm": 0.2702008783817291,
      "learning_rate": 2.5076906595641422e-05,
      "loss": 0.06834125518798828,
      "step": 3301
    },
    {
      "epoch": 0.46011286838988363,
      "grad_norm": 0.43404415249824524,
      "learning_rate": 2.5067787684710886e-05,
      "loss": 0.10466384887695312,
      "step": 3302
    },
    {
      "epoch": 0.460252212081098,
      "grad_norm": 0.47960561513900757,
      "learning_rate": 2.5058667647833615e-05,
      "loss": 0.09465217590332031,
      "step": 3303
    },
    {
      "epoch": 0.4603915557723124,
      "grad_norm": 0.2939611077308655,
      "learning_rate": 2.50495464870359e-05,
      "loss": 0.07997322082519531,
      "step": 3304
    },
    {
      "epoch": 0.46053089946352677,
      "grad_norm": 0.4728347063064575,
      "learning_rate": 2.5040424204344226e-05,
      "loss": 0.09991073608398438,
      "step": 3305
    },
    {
      "epoch": 0.46067024315474114,
      "grad_norm": 0.44260019063949585,
      "learning_rate": 2.5031300801785374e-05,
      "loss": 0.09527587890625,
      "step": 3306
    },
    {
      "epoch": 0.4608095868459556,
      "grad_norm": 0.2764643430709839,
      "learning_rate": 2.502217628138635e-05,
      "loss": 0.07491493225097656,
      "step": 3307
    },
    {
      "epoch": 0.46094893053716995,
      "grad_norm": 0.30959954857826233,
      "learning_rate": 2.5013050645174414e-05,
      "loss": 0.08568954467773438,
      "step": 3308
    },
    {
      "epoch": 0.46108827422838433,
      "grad_norm": 0.383931964635849,
      "learning_rate": 2.5003923895177073e-05,
      "loss": 0.10343742370605469,
      "step": 3309
    },
    {
      "epoch": 0.4612276179195987,
      "grad_norm": 0.6403789520263672,
      "learning_rate": 2.499479603342209e-05,
      "loss": 0.10701179504394531,
      "step": 3310
    },
    {
      "epoch": 0.4613669616108131,
      "grad_norm": 0.33281171321868896,
      "learning_rate": 2.4985667061937458e-05,
      "loss": 0.07886123657226562,
      "step": 3311
    },
    {
      "epoch": 0.46150630530202746,
      "grad_norm": 0.5058579444885254,
      "learning_rate": 2.4976536982751426e-05,
      "loss": 0.12371253967285156,
      "step": 3312
    },
    {
      "epoch": 0.46164564899324184,
      "grad_norm": 0.365314245223999,
      "learning_rate": 2.4967405797892498e-05,
      "loss": 0.07683372497558594,
      "step": 3313
    },
    {
      "epoch": 0.4617849926844562,
      "grad_norm": 0.5600185990333557,
      "learning_rate": 2.4958273509389406e-05,
      "loss": 0.1122598648071289,
      "step": 3314
    },
    {
      "epoch": 0.4619243363756706,
      "grad_norm": 0.3521411418914795,
      "learning_rate": 2.4949140119271144e-05,
      "loss": 0.09858131408691406,
      "step": 3315
    },
    {
      "epoch": 0.462063680066885,
      "grad_norm": 0.3245527744293213,
      "learning_rate": 2.4940005629566927e-05,
      "loss": 0.08791160583496094,
      "step": 3316
    },
    {
      "epoch": 0.46220302375809935,
      "grad_norm": 0.4042893052101135,
      "learning_rate": 2.4930870042306253e-05,
      "loss": 0.08312606811523438,
      "step": 3317
    },
    {
      "epoch": 0.46234236744931373,
      "grad_norm": 0.4990641176700592,
      "learning_rate": 2.4921733359518824e-05,
      "loss": 0.11503791809082031,
      "step": 3318
    },
    {
      "epoch": 0.4624817111405281,
      "grad_norm": 0.33785480260849,
      "learning_rate": 2.4912595583234608e-05,
      "loss": 0.08199119567871094,
      "step": 3319
    },
    {
      "epoch": 0.4626210548317425,
      "grad_norm": 0.48627999424934387,
      "learning_rate": 2.4903456715483817e-05,
      "loss": 0.10497856140136719,
      "step": 3320
    },
    {
      "epoch": 0.46276039852295686,
      "grad_norm": 0.2803114354610443,
      "learning_rate": 2.489431675829688e-05,
      "loss": 0.06209278106689453,
      "step": 3321
    },
    {
      "epoch": 0.46289974221417124,
      "grad_norm": 0.3343244194984436,
      "learning_rate": 2.488517571370451e-05,
      "loss": 0.08159255981445312,
      "step": 3322
    },
    {
      "epoch": 0.4630390859053856,
      "grad_norm": 0.5613617300987244,
      "learning_rate": 2.487603358373762e-05,
      "loss": 0.12671852111816406,
      "step": 3323
    },
    {
      "epoch": 0.4631784295966,
      "grad_norm": 0.4465530812740326,
      "learning_rate": 2.486689037042739e-05,
      "loss": 0.10179710388183594,
      "step": 3324
    },
    {
      "epoch": 0.46331777328781437,
      "grad_norm": 0.17726410925388336,
      "learning_rate": 2.485774607580523e-05,
      "loss": 0.05734825134277344,
      "step": 3325
    },
    {
      "epoch": 0.46345711697902875,
      "grad_norm": 0.46700364351272583,
      "learning_rate": 2.4848600701902804e-05,
      "loss": 0.12024307250976562,
      "step": 3326
    },
    {
      "epoch": 0.4635964606702432,
      "grad_norm": 0.685096025466919,
      "learning_rate": 2.483945425075199e-05,
      "loss": 0.09981346130371094,
      "step": 3327
    },
    {
      "epoch": 0.46373580436145756,
      "grad_norm": 0.32713958621025085,
      "learning_rate": 2.4830306724384933e-05,
      "loss": 0.0828104019165039,
      "step": 3328
    },
    {
      "epoch": 0.46387514805267194,
      "grad_norm": 0.4482177793979645,
      "learning_rate": 2.4821158124834e-05,
      "loss": 0.10051441192626953,
      "step": 3329
    },
    {
      "epoch": 0.4640144917438863,
      "grad_norm": 0.26099684834480286,
      "learning_rate": 2.4812008454131796e-05,
      "loss": 0.07974624633789062,
      "step": 3330
    },
    {
      "epoch": 0.4641538354351007,
      "grad_norm": 0.2743476629257202,
      "learning_rate": 2.4802857714311177e-05,
      "loss": 0.07523345947265625,
      "step": 3331
    },
    {
      "epoch": 0.46429317912631507,
      "grad_norm": 0.576599657535553,
      "learning_rate": 2.479370590740522e-05,
      "loss": 0.12535667419433594,
      "step": 3332
    },
    {
      "epoch": 0.46443252281752945,
      "grad_norm": 0.23570410907268524,
      "learning_rate": 2.478455303544726e-05,
      "loss": 0.0752716064453125,
      "step": 3333
    },
    {
      "epoch": 0.4645718665087438,
      "grad_norm": 0.2721812427043915,
      "learning_rate": 2.4775399100470837e-05,
      "loss": 0.07195281982421875,
      "step": 3334
    },
    {
      "epoch": 0.4647112101999582,
      "grad_norm": 0.28208863735198975,
      "learning_rate": 2.4766244104509775e-05,
      "loss": 0.08131122589111328,
      "step": 3335
    },
    {
      "epoch": 0.4648505538911726,
      "grad_norm": 0.47009608149528503,
      "learning_rate": 2.475708804959808e-05,
      "loss": 0.11071634292602539,
      "step": 3336
    },
    {
      "epoch": 0.46498989758238696,
      "grad_norm": 0.423000305891037,
      "learning_rate": 2.474793093777002e-05,
      "loss": 0.08918476104736328,
      "step": 3337
    },
    {
      "epoch": 0.46512924127360133,
      "grad_norm": 0.4123391807079315,
      "learning_rate": 2.473877277106011e-05,
      "loss": 0.11799430847167969,
      "step": 3338
    },
    {
      "epoch": 0.4652685849648157,
      "grad_norm": 0.42980530858039856,
      "learning_rate": 2.4729613551503074e-05,
      "loss": 0.0882415771484375,
      "step": 3339
    },
    {
      "epoch": 0.4654079286560301,
      "grad_norm": 0.8886896967887878,
      "learning_rate": 2.472045328113389e-05,
      "loss": 0.11425971984863281,
      "step": 3340
    },
    {
      "epoch": 0.46554727234724447,
      "grad_norm": 0.33905407786369324,
      "learning_rate": 2.4711291961987756e-05,
      "loss": 0.11098670959472656,
      "step": 3341
    },
    {
      "epoch": 0.46568661603845884,
      "grad_norm": 0.4354941248893738,
      "learning_rate": 2.470212959610011e-05,
      "loss": 0.09681320190429688,
      "step": 3342
    },
    {
      "epoch": 0.4658259597296732,
      "grad_norm": 0.34834063053131104,
      "learning_rate": 2.4692966185506615e-05,
      "loss": 0.10825729370117188,
      "step": 3343
    },
    {
      "epoch": 0.4659653034208876,
      "grad_norm": 0.6264306902885437,
      "learning_rate": 2.468380173224318e-05,
      "loss": 0.14245986938476562,
      "step": 3344
    },
    {
      "epoch": 0.466104647112102,
      "grad_norm": 0.3684813380241394,
      "learning_rate": 2.467463623834593e-05,
      "loss": 0.08417892456054688,
      "step": 3345
    },
    {
      "epoch": 0.46624399080331635,
      "grad_norm": 0.39931419491767883,
      "learning_rate": 2.4665469705851232e-05,
      "loss": 0.09123802185058594,
      "step": 3346
    },
    {
      "epoch": 0.46638333449453073,
      "grad_norm": 0.21815811097621918,
      "learning_rate": 2.465630213679568e-05,
      "loss": 0.06405448913574219,
      "step": 3347
    },
    {
      "epoch": 0.46652267818574517,
      "grad_norm": 0.5298275947570801,
      "learning_rate": 2.4647133533216097e-05,
      "loss": 0.11026382446289062,
      "step": 3348
    },
    {
      "epoch": 0.46666202187695954,
      "grad_norm": 0.4270552098751068,
      "learning_rate": 2.4637963897149545e-05,
      "loss": 0.08842086791992188,
      "step": 3349
    },
    {
      "epoch": 0.4668013655681739,
      "grad_norm": 0.37219056487083435,
      "learning_rate": 2.4628793230633293e-05,
      "loss": 0.08679008483886719,
      "step": 3350
    },
    {
      "epoch": 0.4669407092593883,
      "grad_norm": 0.43733301758766174,
      "learning_rate": 2.461962153570487e-05,
      "loss": 0.09819269180297852,
      "step": 3351
    },
    {
      "epoch": 0.4670800529506027,
      "grad_norm": 0.5456830263137817,
      "learning_rate": 2.4610448814402008e-05,
      "loss": 0.12476158142089844,
      "step": 3352
    },
    {
      "epoch": 0.46721939664181705,
      "grad_norm": 0.2934306263923645,
      "learning_rate": 2.4601275068762673e-05,
      "loss": 0.06586647033691406,
      "step": 3353
    },
    {
      "epoch": 0.46735874033303143,
      "grad_norm": 0.38268908858299255,
      "learning_rate": 2.459210030082507e-05,
      "loss": 0.09622001647949219,
      "step": 3354
    },
    {
      "epoch": 0.4674980840242458,
      "grad_norm": 0.39008960127830505,
      "learning_rate": 2.4582924512627616e-05,
      "loss": 0.083526611328125,
      "step": 3355
    },
    {
      "epoch": 0.4676374277154602,
      "grad_norm": 0.3850474953651428,
      "learning_rate": 2.4573747706208966e-05,
      "loss": 0.08421707153320312,
      "step": 3356
    },
    {
      "epoch": 0.46777677140667456,
      "grad_norm": 0.475953072309494,
      "learning_rate": 2.4564569883608003e-05,
      "loss": 0.09879684448242188,
      "step": 3357
    },
    {
      "epoch": 0.46791611509788894,
      "grad_norm": 0.4203357696533203,
      "learning_rate": 2.455539104686382e-05,
      "loss": 0.07872486114501953,
      "step": 3358
    },
    {
      "epoch": 0.4680554587891033,
      "grad_norm": 0.34013116359710693,
      "learning_rate": 2.4546211198015746e-05,
      "loss": 0.09143447875976562,
      "step": 3359
    },
    {
      "epoch": 0.4681948024803177,
      "grad_norm": 0.19097289443016052,
      "learning_rate": 2.4537030339103333e-05,
      "loss": 0.058040618896484375,
      "step": 3360
    },
    {
      "epoch": 0.4683341461715321,
      "grad_norm": 0.5390450954437256,
      "learning_rate": 2.4527848472166364e-05,
      "loss": 0.09769821166992188,
      "step": 3361
    },
    {
      "epoch": 0.46847348986274645,
      "grad_norm": 0.36776474118232727,
      "learning_rate": 2.4518665599244836e-05,
      "loss": 0.07756280899047852,
      "step": 3362
    },
    {
      "epoch": 0.46861283355396083,
      "grad_norm": 0.4971564710140228,
      "learning_rate": 2.450948172237898e-05,
      "loss": 0.1048574447631836,
      "step": 3363
    },
    {
      "epoch": 0.4687521772451752,
      "grad_norm": 0.2971155643463135,
      "learning_rate": 2.450029684360923e-05,
      "loss": 0.08499717712402344,
      "step": 3364
    },
    {
      "epoch": 0.4688915209363896,
      "grad_norm": 0.4844661355018616,
      "learning_rate": 2.449111096497627e-05,
      "loss": 0.1285991668701172,
      "step": 3365
    },
    {
      "epoch": 0.46903086462760396,
      "grad_norm": 0.6175223588943481,
      "learning_rate": 2.448192408852098e-05,
      "loss": 0.11851310729980469,
      "step": 3366
    },
    {
      "epoch": 0.46917020831881834,
      "grad_norm": 0.26451897621154785,
      "learning_rate": 2.4472736216284495e-05,
      "loss": 0.06792259216308594,
      "step": 3367
    },
    {
      "epoch": 0.46930955201003277,
      "grad_norm": 0.26346156001091003,
      "learning_rate": 2.4463547350308123e-05,
      "loss": 0.069915771484375,
      "step": 3368
    },
    {
      "epoch": 0.46944889570124715,
      "grad_norm": 0.3335268497467041,
      "learning_rate": 2.4454357492633444e-05,
      "loss": 0.09439277648925781,
      "step": 3369
    },
    {
      "epoch": 0.4695882393924615,
      "grad_norm": 0.6052480340003967,
      "learning_rate": 2.4445166645302223e-05,
      "loss": 0.11198806762695312,
      "step": 3370
    },
    {
      "epoch": 0.4697275830836759,
      "grad_norm": 0.4253769814968109,
      "learning_rate": 2.4435974810356455e-05,
      "loss": 0.09080696105957031,
      "step": 3371
    },
    {
      "epoch": 0.4698669267748903,
      "grad_norm": 0.5820679068565369,
      "learning_rate": 2.4426781989838365e-05,
      "loss": 0.08974742889404297,
      "step": 3372
    },
    {
      "epoch": 0.47000627046610466,
      "grad_norm": 0.3703373074531555,
      "learning_rate": 2.4417588185790374e-05,
      "loss": 0.08946990966796875,
      "step": 3373
    },
    {
      "epoch": 0.47014561415731904,
      "grad_norm": 0.3728737533092499,
      "learning_rate": 2.4408393400255146e-05,
      "loss": 0.07477188110351562,
      "step": 3374
    },
    {
      "epoch": 0.4702849578485334,
      "grad_norm": 0.4379877746105194,
      "learning_rate": 2.4399197635275554e-05,
      "loss": 0.09031295776367188,
      "step": 3375
    },
    {
      "epoch": 0.4704243015397478,
      "grad_norm": 0.46233823895454407,
      "learning_rate": 2.4390000892894677e-05,
      "loss": 0.09814929962158203,
      "step": 3376
    },
    {
      "epoch": 0.47056364523096217,
      "grad_norm": 0.39544522762298584,
      "learning_rate": 2.4380803175155833e-05,
      "loss": 0.087890625,
      "step": 3377
    },
    {
      "epoch": 0.47070298892217655,
      "grad_norm": 0.31529501080513,
      "learning_rate": 2.4371604484102535e-05,
      "loss": 0.0864248275756836,
      "step": 3378
    },
    {
      "epoch": 0.4708423326133909,
      "grad_norm": 0.497875839471817,
      "learning_rate": 2.4362404821778528e-05,
      "loss": 0.1032094955444336,
      "step": 3379
    },
    {
      "epoch": 0.4709816763046053,
      "grad_norm": 0.517224907875061,
      "learning_rate": 2.4353204190227767e-05,
      "loss": 0.1325225830078125,
      "step": 3380
    },
    {
      "epoch": 0.4711210199958197,
      "grad_norm": 0.4223402738571167,
      "learning_rate": 2.4344002591494415e-05,
      "loss": 0.09404182434082031,
      "step": 3381
    },
    {
      "epoch": 0.47126036368703406,
      "grad_norm": 0.24303072690963745,
      "learning_rate": 2.433480002762286e-05,
      "loss": 0.07674026489257812,
      "step": 3382
    },
    {
      "epoch": 0.47139970737824843,
      "grad_norm": 0.265598863363266,
      "learning_rate": 2.4325596500657714e-05,
      "loss": 0.06712627410888672,
      "step": 3383
    },
    {
      "epoch": 0.4715390510694628,
      "grad_norm": 0.33699771761894226,
      "learning_rate": 2.431639201264377e-05,
      "loss": 0.09324264526367188,
      "step": 3384
    },
    {
      "epoch": 0.4716783947606772,
      "grad_norm": 0.20662163197994232,
      "learning_rate": 2.4307186565626073e-05,
      "loss": 0.07404327392578125,
      "step": 3385
    },
    {
      "epoch": 0.47181773845189157,
      "grad_norm": 0.9201899766921997,
      "learning_rate": 2.429798016164986e-05,
      "loss": 0.1257781982421875,
      "step": 3386
    },
    {
      "epoch": 0.47195708214310594,
      "grad_norm": 0.3018714487552643,
      "learning_rate": 2.428877280276057e-05,
      "loss": 0.08494949340820312,
      "step": 3387
    },
    {
      "epoch": 0.4720964258343204,
      "grad_norm": 0.3003727197647095,
      "learning_rate": 2.4279564491003883e-05,
      "loss": 0.0684213638305664,
      "step": 3388
    },
    {
      "epoch": 0.47223576952553475,
      "grad_norm": 0.6570353507995605,
      "learning_rate": 2.4270355228425664e-05,
      "loss": 0.10650253295898438,
      "step": 3389
    },
    {
      "epoch": 0.47237511321674913,
      "grad_norm": 0.4527633786201477,
      "learning_rate": 2.4261145017072014e-05,
      "loss": 0.10768508911132812,
      "step": 3390
    },
    {
      "epoch": 0.4725144569079635,
      "grad_norm": 0.6783212423324585,
      "learning_rate": 2.425193385898922e-05,
      "loss": 0.12325477600097656,
      "step": 3391
    },
    {
      "epoch": 0.4726538005991779,
      "grad_norm": 0.2917686998844147,
      "learning_rate": 2.42427217562238e-05,
      "loss": 0.06637763977050781,
      "step": 3392
    },
    {
      "epoch": 0.47279314429039226,
      "grad_norm": 0.38302546739578247,
      "learning_rate": 2.4233508710822466e-05,
      "loss": 0.10655784606933594,
      "step": 3393
    },
    {
      "epoch": 0.47293248798160664,
      "grad_norm": 0.5236696600914001,
      "learning_rate": 2.4224294724832152e-05,
      "loss": 0.0955648422241211,
      "step": 3394
    },
    {
      "epoch": 0.473071831672821,
      "grad_norm": 0.41216036677360535,
      "learning_rate": 2.421507980029999e-05,
      "loss": 0.09142303466796875,
      "step": 3395
    },
    {
      "epoch": 0.4732111753640354,
      "grad_norm": 0.32322224974632263,
      "learning_rate": 2.4205863939273328e-05,
      "loss": 0.0830078125,
      "step": 3396
    },
    {
      "epoch": 0.4733505190552498,
      "grad_norm": 0.2783767282962799,
      "learning_rate": 2.4196647143799723e-05,
      "loss": 0.08205986022949219,
      "step": 3397
    },
    {
      "epoch": 0.47348986274646415,
      "grad_norm": 0.505244791507721,
      "learning_rate": 2.4187429415926927e-05,
      "loss": 0.08932113647460938,
      "step": 3398
    },
    {
      "epoch": 0.47362920643767853,
      "grad_norm": 0.4565367102622986,
      "learning_rate": 2.4178210757702924e-05,
      "loss": 0.10303115844726562,
      "step": 3399
    },
    {
      "epoch": 0.4737685501288929,
      "grad_norm": 0.6338425278663635,
      "learning_rate": 2.4168991171175872e-05,
      "loss": 0.15367507934570312,
      "step": 3400
    },
    {
      "epoch": 0.4739078938201073,
      "grad_norm": 0.36610090732574463,
      "learning_rate": 2.415977065839417e-05,
      "loss": 0.07883644104003906,
      "step": 3401
    },
    {
      "epoch": 0.47404723751132166,
      "grad_norm": 0.5593715310096741,
      "learning_rate": 2.4150549221406395e-05,
      "loss": 0.10924339294433594,
      "step": 3402
    },
    {
      "epoch": 0.47418658120253604,
      "grad_norm": 1.0157338380813599,
      "learning_rate": 2.4141326862261332e-05,
      "loss": 0.16391944885253906,
      "step": 3403
    },
    {
      "epoch": 0.4743259248937504,
      "grad_norm": 0.3761114180088043,
      "learning_rate": 2.4132103583008008e-05,
      "loss": 0.09770870208740234,
      "step": 3404
    },
    {
      "epoch": 0.4744652685849648,
      "grad_norm": 0.26143428683280945,
      "learning_rate": 2.4122879385695587e-05,
      "loss": 0.07155418395996094,
      "step": 3405
    },
    {
      "epoch": 0.47460461227617917,
      "grad_norm": 0.4475577175617218,
      "learning_rate": 2.41136542723735e-05,
      "loss": 0.08676719665527344,
      "step": 3406
    },
    {
      "epoch": 0.47474395596739355,
      "grad_norm": 1.1987353563308716,
      "learning_rate": 2.410442824509135e-05,
      "loss": 0.13416099548339844,
      "step": 3407
    },
    {
      "epoch": 0.474883299658608,
      "grad_norm": 0.32180583477020264,
      "learning_rate": 2.409520130589895e-05,
      "loss": 0.08172798156738281,
      "step": 3408
    },
    {
      "epoch": 0.47502264334982236,
      "grad_norm": 0.2820810377597809,
      "learning_rate": 2.4085973456846318e-05,
      "loss": 0.07158565521240234,
      "step": 3409
    },
    {
      "epoch": 0.47516198704103674,
      "grad_norm": 0.20304766297340393,
      "learning_rate": 2.4076744699983663e-05,
      "loss": 0.06934547424316406,
      "step": 3410
    },
    {
      "epoch": 0.4753013307322511,
      "grad_norm": 0.4546390175819397,
      "learning_rate": 2.4067515037361408e-05,
      "loss": 0.10280418395996094,
      "step": 3411
    },
    {
      "epoch": 0.4754406744234655,
      "grad_norm": 0.35618364810943604,
      "learning_rate": 2.405828447103018e-05,
      "loss": 0.07711029052734375,
      "step": 3412
    },
    {
      "epoch": 0.47558001811467987,
      "grad_norm": 0.7602665424346924,
      "learning_rate": 2.4049053003040795e-05,
      "loss": 0.12159347534179688,
      "step": 3413
    },
    {
      "epoch": 0.47571936180589425,
      "grad_norm": 0.5211936831474304,
      "learning_rate": 2.4039820635444264e-05,
      "loss": 0.1286754608154297,
      "step": 3414
    },
    {
      "epoch": 0.4758587054971086,
      "grad_norm": 0.35267195105552673,
      "learning_rate": 2.4030587370291835e-05,
      "loss": 0.0835113525390625,
      "step": 3415
    },
    {
      "epoch": 0.475998049188323,
      "grad_norm": 0.40632766485214233,
      "learning_rate": 2.4021353209634896e-05,
      "loss": 0.08858489990234375,
      "step": 3416
    },
    {
      "epoch": 0.4761373928795374,
      "grad_norm": 0.3793032765388489,
      "learning_rate": 2.4012118155525094e-05,
      "loss": 0.10012435913085938,
      "step": 3417
    },
    {
      "epoch": 0.47627673657075176,
      "grad_norm": 0.36834463477134705,
      "learning_rate": 2.4002882210014227e-05,
      "loss": 0.07397651672363281,
      "step": 3418
    },
    {
      "epoch": 0.47641608026196614,
      "grad_norm": 0.4297516644001007,
      "learning_rate": 2.3993645375154328e-05,
      "loss": 0.10202217102050781,
      "step": 3419
    },
    {
      "epoch": 0.4765554239531805,
      "grad_norm": 0.2647695243358612,
      "learning_rate": 2.3984407652997607e-05,
      "loss": 0.06284141540527344,
      "step": 3420
    },
    {
      "epoch": 0.4766947676443949,
      "grad_norm": 0.21692104637622833,
      "learning_rate": 2.397516904559646e-05,
      "loss": 0.05802297592163086,
      "step": 3421
    },
    {
      "epoch": 0.47683411133560927,
      "grad_norm": 0.3594118654727936,
      "learning_rate": 2.3965929555003512e-05,
      "loss": 0.08323097229003906,
      "step": 3422
    },
    {
      "epoch": 0.47697345502682365,
      "grad_norm": 0.4874275326728821,
      "learning_rate": 2.3956689183271557e-05,
      "loss": 0.09844779968261719,
      "step": 3423
    },
    {
      "epoch": 0.477112798718038,
      "grad_norm": 0.5953719615936279,
      "learning_rate": 2.39474479324536e-05,
      "loss": 0.11598968505859375,
      "step": 3424
    },
    {
      "epoch": 0.4772521424092524,
      "grad_norm": 0.47271066904067993,
      "learning_rate": 2.3938205804602835e-05,
      "loss": 0.0986328125,
      "step": 3425
    },
    {
      "epoch": 0.4773914861004668,
      "grad_norm": 0.47366422414779663,
      "learning_rate": 2.392896280177265e-05,
      "loss": 0.08501434326171875,
      "step": 3426
    },
    {
      "epoch": 0.47753082979168116,
      "grad_norm": 0.482148140668869,
      "learning_rate": 2.391971892601663e-05,
      "loss": 0.11348724365234375,
      "step": 3427
    },
    {
      "epoch": 0.4776701734828956,
      "grad_norm": 0.8253277540206909,
      "learning_rate": 2.3910474179388557e-05,
      "loss": 0.18464088439941406,
      "step": 3428
    },
    {
      "epoch": 0.47780951717410997,
      "grad_norm": 0.31773489713668823,
      "learning_rate": 2.3901228563942397e-05,
      "loss": 0.0797109603881836,
      "step": 3429
    },
    {
      "epoch": 0.47794886086532434,
      "grad_norm": 0.5292791128158569,
      "learning_rate": 2.389198208173231e-05,
      "loss": 0.10355854034423828,
      "step": 3430
    },
    {
      "epoch": 0.4780882045565387,
      "grad_norm": 0.4860965311527252,
      "learning_rate": 2.3882734734812673e-05,
      "loss": 0.09836196899414062,
      "step": 3431
    },
    {
      "epoch": 0.4782275482477531,
      "grad_norm": 0.1627788096666336,
      "learning_rate": 2.3873486525238008e-05,
      "loss": 0.06105327606201172,
      "step": 3432
    },
    {
      "epoch": 0.4783668919389675,
      "grad_norm": 0.4386437237262726,
      "learning_rate": 2.3864237455063083e-05,
      "loss": 0.129913330078125,
      "step": 3433
    },
    {
      "epoch": 0.47850623563018185,
      "grad_norm": 0.30330219864845276,
      "learning_rate": 2.3854987526342806e-05,
      "loss": 0.084808349609375,
      "step": 3434
    },
    {
      "epoch": 0.47864557932139623,
      "grad_norm": 0.3240766227245331,
      "learning_rate": 2.3845736741132317e-05,
      "loss": 0.0643014907836914,
      "step": 3435
    },
    {
      "epoch": 0.4787849230126106,
      "grad_norm": 0.6141373515129089,
      "learning_rate": 2.3836485101486928e-05,
      "loss": 0.09175682067871094,
      "step": 3436
    },
    {
      "epoch": 0.478924266703825,
      "grad_norm": 0.5576227903366089,
      "learning_rate": 2.382723260946213e-05,
      "loss": 0.12470054626464844,
      "step": 3437
    },
    {
      "epoch": 0.47906361039503936,
      "grad_norm": 0.32891637086868286,
      "learning_rate": 2.3817979267113633e-05,
      "loss": 0.10100364685058594,
      "step": 3438
    },
    {
      "epoch": 0.47920295408625374,
      "grad_norm": 0.6603897213935852,
      "learning_rate": 2.3808725076497297e-05,
      "loss": 0.10672760009765625,
      "step": 3439
    },
    {
      "epoch": 0.4793422977774681,
      "grad_norm": 0.5592063069343567,
      "learning_rate": 2.3799470039669212e-05,
      "loss": 0.121490478515625,
      "step": 3440
    },
    {
      "epoch": 0.4794816414686825,
      "grad_norm": 0.6322094798088074,
      "learning_rate": 2.3790214158685622e-05,
      "loss": 0.09412765502929688,
      "step": 3441
    },
    {
      "epoch": 0.4796209851598969,
      "grad_norm": 0.26640185713768005,
      "learning_rate": 2.3780957435602984e-05,
      "loss": 0.08119964599609375,
      "step": 3442
    },
    {
      "epoch": 0.47976032885111125,
      "grad_norm": 0.3336607813835144,
      "learning_rate": 2.377169987247792e-05,
      "loss": 0.10359430313110352,
      "step": 3443
    },
    {
      "epoch": 0.47989967254232563,
      "grad_norm": 0.4002796411514282,
      "learning_rate": 2.376244147136726e-05,
      "loss": 0.08101606369018555,
      "step": 3444
    },
    {
      "epoch": 0.48003901623354,
      "grad_norm": 0.42045366764068604,
      "learning_rate": 2.3753182234327994e-05,
      "loss": 0.09898185729980469,
      "step": 3445
    },
    {
      "epoch": 0.4801783599247544,
      "grad_norm": 0.5751110315322876,
      "learning_rate": 2.374392216341733e-05,
      "loss": 0.13056564331054688,
      "step": 3446
    },
    {
      "epoch": 0.48031770361596876,
      "grad_norm": 0.27729788422584534,
      "learning_rate": 2.3734661260692633e-05,
      "loss": 0.07246589660644531,
      "step": 3447
    },
    {
      "epoch": 0.4804570473071832,
      "grad_norm": 0.33470696210861206,
      "learning_rate": 2.3725399528211462e-05,
      "loss": 0.07543277740478516,
      "step": 3448
    },
    {
      "epoch": 0.48059639099839757,
      "grad_norm": 0.3665543794631958,
      "learning_rate": 2.371613696803158e-05,
      "loss": 0.09302711486816406,
      "step": 3449
    },
    {
      "epoch": 0.48073573468961195,
      "grad_norm": 0.5296663045883179,
      "learning_rate": 2.3706873582210893e-05,
      "loss": 0.10642623901367188,
      "step": 3450
    },
    {
      "epoch": 0.4808750783808263,
      "grad_norm": 0.36532270908355713,
      "learning_rate": 2.369760937280753e-05,
      "loss": 0.08485984802246094,
      "step": 3451
    },
    {
      "epoch": 0.4810144220720407,
      "grad_norm": 0.3713747560977936,
      "learning_rate": 2.368834434187979e-05,
      "loss": 0.08110427856445312,
      "step": 3452
    },
    {
      "epoch": 0.4811537657632551,
      "grad_norm": 0.41377294063568115,
      "learning_rate": 2.3679078491486133e-05,
      "loss": 0.08055686950683594,
      "step": 3453
    },
    {
      "epoch": 0.48129310945446946,
      "grad_norm": 0.6762192845344543,
      "learning_rate": 2.3669811823685235e-05,
      "loss": 0.09663581848144531,
      "step": 3454
    },
    {
      "epoch": 0.48143245314568384,
      "grad_norm": 0.6754793524742126,
      "learning_rate": 2.366054434053593e-05,
      "loss": 0.13373947143554688,
      "step": 3455
    },
    {
      "epoch": 0.4815717968368982,
      "grad_norm": 0.4549417197704315,
      "learning_rate": 2.3651276044097246e-05,
      "loss": 0.10376548767089844,
      "step": 3456
    },
    {
      "epoch": 0.4817111405281126,
      "grad_norm": 0.22388340532779694,
      "learning_rate": 2.364200693642839e-05,
      "loss": 0.0695199966430664,
      "step": 3457
    },
    {
      "epoch": 0.48185048421932697,
      "grad_norm": 0.47678929567337036,
      "learning_rate": 2.363273701958873e-05,
      "loss": 0.08212566375732422,
      "step": 3458
    },
    {
      "epoch": 0.48198982791054135,
      "grad_norm": 0.6933940649032593,
      "learning_rate": 2.3623466295637848e-05,
      "loss": 0.11297225952148438,
      "step": 3459
    },
    {
      "epoch": 0.4821291716017557,
      "grad_norm": 0.8080404996871948,
      "learning_rate": 2.3614194766635482e-05,
      "loss": 0.0964670181274414,
      "step": 3460
    },
    {
      "epoch": 0.4822685152929701,
      "grad_norm": 0.4244355261325836,
      "learning_rate": 2.3604922434641545e-05,
      "loss": 0.10590934753417969,
      "step": 3461
    },
    {
      "epoch": 0.4824078589841845,
      "grad_norm": 0.35458946228027344,
      "learning_rate": 2.3595649301716154e-05,
      "loss": 0.08524417877197266,
      "step": 3462
    },
    {
      "epoch": 0.48254720267539886,
      "grad_norm": 0.44528841972351074,
      "learning_rate": 2.3586375369919573e-05,
      "loss": 0.0905294418334961,
      "step": 3463
    },
    {
      "epoch": 0.48268654636661323,
      "grad_norm": 0.4322637617588043,
      "learning_rate": 2.3577100641312258e-05,
      "loss": 0.10003852844238281,
      "step": 3464
    },
    {
      "epoch": 0.4828258900578276,
      "grad_norm": 0.32373663783073425,
      "learning_rate": 2.356782511795486e-05,
      "loss": 0.08164787292480469,
      "step": 3465
    },
    {
      "epoch": 0.482965233749042,
      "grad_norm": 0.3465843200683594,
      "learning_rate": 2.3558548801908164e-05,
      "loss": 0.0884552001953125,
      "step": 3466
    },
    {
      "epoch": 0.48310457744025637,
      "grad_norm": 0.6706493496894836,
      "learning_rate": 2.3549271695233177e-05,
      "loss": 0.1169748306274414,
      "step": 3467
    },
    {
      "epoch": 0.4832439211314708,
      "grad_norm": 0.4178025722503662,
      "learning_rate": 2.353999379999104e-05,
      "loss": 0.08529376983642578,
      "step": 3468
    },
    {
      "epoch": 0.4833832648226852,
      "grad_norm": 0.42493778467178345,
      "learning_rate": 2.3530715118243105e-05,
      "loss": 0.10468292236328125,
      "step": 3469
    },
    {
      "epoch": 0.48352260851389955,
      "grad_norm": 0.4895673394203186,
      "learning_rate": 2.3521435652050886e-05,
      "loss": 0.09110450744628906,
      "step": 3470
    },
    {
      "epoch": 0.48366195220511393,
      "grad_norm": 0.44503864645957947,
      "learning_rate": 2.351215540347605e-05,
      "loss": 0.0815887451171875,
      "step": 3471
    },
    {
      "epoch": 0.4838012958963283,
      "grad_norm": 0.6322842240333557,
      "learning_rate": 2.350287437458047e-05,
      "loss": 0.07789993286132812,
      "step": 3472
    },
    {
      "epoch": 0.4839406395875427,
      "grad_norm": 0.3704487979412079,
      "learning_rate": 2.349359256742618e-05,
      "loss": 0.07791328430175781,
      "step": 3473
    },
    {
      "epoch": 0.48407998327875706,
      "grad_norm": 0.4088389575481415,
      "learning_rate": 2.3484309984075376e-05,
      "loss": 0.08663749694824219,
      "step": 3474
    },
    {
      "epoch": 0.48421932696997144,
      "grad_norm": 0.2975269854068756,
      "learning_rate": 2.3475026626590443e-05,
      "loss": 0.0869293212890625,
      "step": 3475
    },
    {
      "epoch": 0.4843586706611858,
      "grad_norm": 0.4452313482761383,
      "learning_rate": 2.3465742497033932e-05,
      "loss": 0.10309028625488281,
      "step": 3476
    },
    {
      "epoch": 0.4844980143524002,
      "grad_norm": 0.36391445994377136,
      "learning_rate": 2.345645759746856e-05,
      "loss": 0.07858943939208984,
      "step": 3477
    },
    {
      "epoch": 0.4846373580436146,
      "grad_norm": 0.31398671865463257,
      "learning_rate": 2.3447171929957224e-05,
      "loss": 0.07616901397705078,
      "step": 3478
    },
    {
      "epoch": 0.48477670173482895,
      "grad_norm": 0.6435450911521912,
      "learning_rate": 2.3437885496562986e-05,
      "loss": 0.11063575744628906,
      "step": 3479
    },
    {
      "epoch": 0.48491604542604333,
      "grad_norm": 0.501363217830658,
      "learning_rate": 2.3428598299349076e-05,
      "loss": 0.10598373413085938,
      "step": 3480
    },
    {
      "epoch": 0.4850553891172577,
      "grad_norm": 0.4656845033168793,
      "learning_rate": 2.34193103403789e-05,
      "loss": 0.10491943359375,
      "step": 3481
    },
    {
      "epoch": 0.4851947328084721,
      "grad_norm": 0.341849148273468,
      "learning_rate": 2.341002162171603e-05,
      "loss": 0.08456230163574219,
      "step": 3482
    },
    {
      "epoch": 0.48533407649968646,
      "grad_norm": 0.41346633434295654,
      "learning_rate": 2.3400732145424216e-05,
      "loss": 0.08855342864990234,
      "step": 3483
    },
    {
      "epoch": 0.48547342019090084,
      "grad_norm": 0.7435234785079956,
      "learning_rate": 2.339144191356735e-05,
      "loss": 0.11557865142822266,
      "step": 3484
    },
    {
      "epoch": 0.4856127638821152,
      "grad_norm": 0.24933378398418427,
      "learning_rate": 2.3382150928209523e-05,
      "loss": 0.06495094299316406,
      "step": 3485
    },
    {
      "epoch": 0.4857521075733296,
      "grad_norm": 1.3599504232406616,
      "learning_rate": 2.3372859191414978e-05,
      "loss": 0.201904296875,
      "step": 3486
    },
    {
      "epoch": 0.48589145126454397,
      "grad_norm": 0.3012869358062744,
      "learning_rate": 2.3363566705248117e-05,
      "loss": 0.08104515075683594,
      "step": 3487
    },
    {
      "epoch": 0.4860307949557584,
      "grad_norm": 0.33018165826797485,
      "learning_rate": 2.3354273471773534e-05,
      "loss": 0.09753227233886719,
      "step": 3488
    },
    {
      "epoch": 0.4861701386469728,
      "grad_norm": 0.3006400763988495,
      "learning_rate": 2.3344979493055958e-05,
      "loss": 0.07224655151367188,
      "step": 3489
    },
    {
      "epoch": 0.48630948233818716,
      "grad_norm": 0.34737178683280945,
      "learning_rate": 2.333568477116031e-05,
      "loss": 0.07097816467285156,
      "step": 3490
    },
    {
      "epoch": 0.48644882602940154,
      "grad_norm": 0.4151892066001892,
      "learning_rate": 2.3326389308151658e-05,
      "loss": 0.09715557098388672,
      "step": 3491
    },
    {
      "epoch": 0.4865881697206159,
      "grad_norm": 0.8376125693321228,
      "learning_rate": 2.3317093106095246e-05,
      "loss": 0.13702011108398438,
      "step": 3492
    },
    {
      "epoch": 0.4867275134118303,
      "grad_norm": 0.3436349928379059,
      "learning_rate": 2.330779616705648e-05,
      "loss": 0.072509765625,
      "step": 3493
    },
    {
      "epoch": 0.48686685710304467,
      "grad_norm": 0.43082335591316223,
      "learning_rate": 2.329849849310092e-05,
      "loss": 0.08440685272216797,
      "step": 3494
    },
    {
      "epoch": 0.48700620079425905,
      "grad_norm": 0.4261794984340668,
      "learning_rate": 2.3289200086294298e-05,
      "loss": 0.11366844177246094,
      "step": 3495
    },
    {
      "epoch": 0.4871455444854734,
      "grad_norm": 0.4665292203426361,
      "learning_rate": 2.3279900948702516e-05,
      "loss": 0.0989370346069336,
      "step": 3496
    },
    {
      "epoch": 0.4872848881766878,
      "grad_norm": 0.3691820204257965,
      "learning_rate": 2.3270601082391623e-05,
      "loss": 0.08914947509765625,
      "step": 3497
    },
    {
      "epoch": 0.4874242318679022,
      "grad_norm": 0.6231489181518555,
      "learning_rate": 2.3261300489427835e-05,
      "loss": 0.10580062866210938,
      "step": 3498
    },
    {
      "epoch": 0.48756357555911656,
      "grad_norm": 1.090087890625,
      "learning_rate": 2.3251999171877538e-05,
      "loss": 0.14092636108398438,
      "step": 3499
    },
    {
      "epoch": 0.48770291925033094,
      "grad_norm": 0.32475176453590393,
      "learning_rate": 2.3242697131807267e-05,
      "loss": 0.07478141784667969,
      "step": 3500
    },
    {
      "epoch": 0.4878422629415453,
      "grad_norm": 0.35325801372528076,
      "learning_rate": 2.3233394371283727e-05,
      "loss": 0.09547233581542969,
      "step": 3501
    },
    {
      "epoch": 0.4879816066327597,
      "grad_norm": 0.4399093687534332,
      "learning_rate": 2.322409089237378e-05,
      "loss": 0.10532188415527344,
      "step": 3502
    },
    {
      "epoch": 0.48812095032397407,
      "grad_norm": 0.7503423690795898,
      "learning_rate": 2.321478669714444e-05,
      "loss": 0.08897781372070312,
      "step": 3503
    },
    {
      "epoch": 0.48826029401518845,
      "grad_norm": 0.6290386915206909,
      "learning_rate": 2.3205481787662895e-05,
      "loss": 0.12149810791015625,
      "step": 3504
    },
    {
      "epoch": 0.4883996377064028,
      "grad_norm": 0.3666436970233917,
      "learning_rate": 2.3196176165996476e-05,
      "loss": 0.076629638671875,
      "step": 3505
    },
    {
      "epoch": 0.4885389813976172,
      "grad_norm": 0.597728431224823,
      "learning_rate": 2.3186869834212682e-05,
      "loss": 0.0934906005859375,
      "step": 3506
    },
    {
      "epoch": 0.4886783250888316,
      "grad_norm": 0.2929260730743408,
      "learning_rate": 2.3177562794379173e-05,
      "loss": 0.09285354614257812,
      "step": 3507
    },
    {
      "epoch": 0.488817668780046,
      "grad_norm": 0.4453027844429016,
      "learning_rate": 2.3168255048563753e-05,
      "loss": 0.08423042297363281,
      "step": 3508
    },
    {
      "epoch": 0.4889570124712604,
      "grad_norm": 0.5629048347473145,
      "learning_rate": 2.3158946598834393e-05,
      "loss": 0.1123809814453125,
      "step": 3509
    },
    {
      "epoch": 0.48909635616247477,
      "grad_norm": 0.561160683631897,
      "learning_rate": 2.314963744725922e-05,
      "loss": 0.10912704467773438,
      "step": 3510
    },
    {
      "epoch": 0.48923569985368914,
      "grad_norm": 0.3977614939212799,
      "learning_rate": 2.314032759590651e-05,
      "loss": 0.08147430419921875,
      "step": 3511
    },
    {
      "epoch": 0.4893750435449035,
      "grad_norm": 0.40343931317329407,
      "learning_rate": 2.313101704684471e-05,
      "loss": 0.09473228454589844,
      "step": 3512
    },
    {
      "epoch": 0.4895143872361179,
      "grad_norm": 0.5881628394126892,
      "learning_rate": 2.31217058021424e-05,
      "loss": 0.10203838348388672,
      "step": 3513
    },
    {
      "epoch": 0.4896537309273323,
      "grad_norm": 0.497330904006958,
      "learning_rate": 2.3112393863868327e-05,
      "loss": 0.10740089416503906,
      "step": 3514
    },
    {
      "epoch": 0.48979307461854665,
      "grad_norm": 0.556941568851471,
      "learning_rate": 2.3103081234091406e-05,
      "loss": 0.09459686279296875,
      "step": 3515
    },
    {
      "epoch": 0.48993241830976103,
      "grad_norm": 0.3576592803001404,
      "learning_rate": 2.3093767914880668e-05,
      "loss": 0.08832359313964844,
      "step": 3516
    },
    {
      "epoch": 0.4900717620009754,
      "grad_norm": 0.33244749903678894,
      "learning_rate": 2.308445390830534e-05,
      "loss": 0.09192752838134766,
      "step": 3517
    },
    {
      "epoch": 0.4902111056921898,
      "grad_norm": 0.714246928691864,
      "learning_rate": 2.3075139216434762e-05,
      "loss": 0.14908981323242188,
      "step": 3518
    },
    {
      "epoch": 0.49035044938340416,
      "grad_norm": 0.3638019561767578,
      "learning_rate": 2.3065823841338465e-05,
      "loss": 0.08140945434570312,
      "step": 3519
    },
    {
      "epoch": 0.49048979307461854,
      "grad_norm": 0.30502814054489136,
      "learning_rate": 2.3056507785086105e-05,
      "loss": 0.06790924072265625,
      "step": 3520
    },
    {
      "epoch": 0.4906291367658329,
      "grad_norm": 0.24218115210533142,
      "learning_rate": 2.304719104974749e-05,
      "loss": 0.06303119659423828,
      "step": 3521
    },
    {
      "epoch": 0.4907684804570473,
      "grad_norm": 0.2330855280160904,
      "learning_rate": 2.3037873637392596e-05,
      "loss": 0.06697511672973633,
      "step": 3522
    },
    {
      "epoch": 0.4909078241482617,
      "grad_norm": 0.47906139492988586,
      "learning_rate": 2.3028555550091536e-05,
      "loss": 0.07579421997070312,
      "step": 3523
    },
    {
      "epoch": 0.49104716783947605,
      "grad_norm": 0.3973087966442108,
      "learning_rate": 2.3019236789914575e-05,
      "loss": 0.08883953094482422,
      "step": 3524
    },
    {
      "epoch": 0.49118651153069043,
      "grad_norm": 0.6008855104446411,
      "learning_rate": 2.300991735893213e-05,
      "loss": 0.1333789825439453,
      "step": 3525
    },
    {
      "epoch": 0.4913258552219048,
      "grad_norm": 0.5370943546295166,
      "learning_rate": 2.3000597259214765e-05,
      "loss": 0.13360977172851562,
      "step": 3526
    },
    {
      "epoch": 0.4914651989131192,
      "grad_norm": 0.49261510372161865,
      "learning_rate": 2.2991276492833197e-05,
      "loss": 0.11439323425292969,
      "step": 3527
    },
    {
      "epoch": 0.4916045426043336,
      "grad_norm": 0.5018017888069153,
      "learning_rate": 2.2981955061858282e-05,
      "loss": 0.11037635803222656,
      "step": 3528
    },
    {
      "epoch": 0.491743886295548,
      "grad_norm": 0.5720459818840027,
      "learning_rate": 2.297263296836103e-05,
      "loss": 0.11130332946777344,
      "step": 3529
    },
    {
      "epoch": 0.49188322998676237,
      "grad_norm": 0.4461495876312256,
      "learning_rate": 2.2963310214412596e-05,
      "loss": 0.10687637329101562,
      "step": 3530
    },
    {
      "epoch": 0.49202257367797675,
      "grad_norm": 0.4036763608455658,
      "learning_rate": 2.2953986802084293e-05,
      "loss": 0.08428955078125,
      "step": 3531
    },
    {
      "epoch": 0.4921619173691911,
      "grad_norm": 0.4538286626338959,
      "learning_rate": 2.2944662733447557e-05,
      "loss": 0.0961151123046875,
      "step": 3532
    },
    {
      "epoch": 0.4923012610604055,
      "grad_norm": 0.3409772515296936,
      "learning_rate": 2.2935338010573998e-05,
      "loss": 0.08807182312011719,
      "step": 3533
    },
    {
      "epoch": 0.4924406047516199,
      "grad_norm": 0.49719616770744324,
      "learning_rate": 2.292601263553534e-05,
      "loss": 0.09952449798583984,
      "step": 3534
    },
    {
      "epoch": 0.49257994844283426,
      "grad_norm": 0.5508214235305786,
      "learning_rate": 2.2916686610403477e-05,
      "loss": 0.11870574951171875,
      "step": 3535
    },
    {
      "epoch": 0.49271929213404864,
      "grad_norm": 0.36145704984664917,
      "learning_rate": 2.2907359937250445e-05,
      "loss": 0.08869743347167969,
      "step": 3536
    },
    {
      "epoch": 0.492858635825263,
      "grad_norm": 0.3766576051712036,
      "learning_rate": 2.2898032618148403e-05,
      "loss": 0.07381629943847656,
      "step": 3537
    },
    {
      "epoch": 0.4929979795164774,
      "grad_norm": 0.6557148694992065,
      "learning_rate": 2.288870465516968e-05,
      "loss": 0.10042762756347656,
      "step": 3538
    },
    {
      "epoch": 0.49313732320769177,
      "grad_norm": 0.18732362985610962,
      "learning_rate": 2.287937605038673e-05,
      "loss": 0.05557823181152344,
      "step": 3539
    },
    {
      "epoch": 0.49327666689890615,
      "grad_norm": 0.38100650906562805,
      "learning_rate": 2.2870046805872166e-05,
      "loss": 0.09485626220703125,
      "step": 3540
    },
    {
      "epoch": 0.4934160105901205,
      "grad_norm": 0.6137387752532959,
      "learning_rate": 2.286071692369872e-05,
      "loss": 0.13443374633789062,
      "step": 3541
    },
    {
      "epoch": 0.4935553542813349,
      "grad_norm": 0.817206859588623,
      "learning_rate": 2.2851386405939288e-05,
      "loss": 0.14458274841308594,
      "step": 3542
    },
    {
      "epoch": 0.4936946979725493,
      "grad_norm": 0.47277554869651794,
      "learning_rate": 2.284205525466689e-05,
      "loss": 0.10109138488769531,
      "step": 3543
    },
    {
      "epoch": 0.49383404166376366,
      "grad_norm": 0.5495854020118713,
      "learning_rate": 2.2832723471954705e-05,
      "loss": 0.09332656860351562,
      "step": 3544
    },
    {
      "epoch": 0.49397338535497803,
      "grad_norm": 0.29470640420913696,
      "learning_rate": 2.2823391059876032e-05,
      "loss": 0.08363723754882812,
      "step": 3545
    },
    {
      "epoch": 0.4941127290461924,
      "grad_norm": 0.5468565821647644,
      "learning_rate": 2.2814058020504324e-05,
      "loss": 0.12225341796875,
      "step": 3546
    },
    {
      "epoch": 0.4942520727374068,
      "grad_norm": 0.4509151875972748,
      "learning_rate": 2.280472435591318e-05,
      "loss": 0.10212516784667969,
      "step": 3547
    },
    {
      "epoch": 0.49439141642862117,
      "grad_norm": 0.6681642532348633,
      "learning_rate": 2.2795390068176304e-05,
      "loss": 0.10919952392578125,
      "step": 3548
    },
    {
      "epoch": 0.4945307601198356,
      "grad_norm": 0.4352334439754486,
      "learning_rate": 2.2786055159367588e-05,
      "loss": 0.11079978942871094,
      "step": 3549
    },
    {
      "epoch": 0.49467010381105,
      "grad_norm": 0.28221672773361206,
      "learning_rate": 2.277671963156101e-05,
      "loss": 0.08351325988769531,
      "step": 3550
    },
    {
      "epoch": 0.49480944750226435,
      "grad_norm": 0.4246078431606293,
      "learning_rate": 2.2767383486830728e-05,
      "loss": 0.1057891845703125,
      "step": 3551
    },
    {
      "epoch": 0.49494879119347873,
      "grad_norm": 0.2910420000553131,
      "learning_rate": 2.275804672725102e-05,
      "loss": 0.07705116271972656,
      "step": 3552
    },
    {
      "epoch": 0.4950881348846931,
      "grad_norm": 0.31359702348709106,
      "learning_rate": 2.274870935489629e-05,
      "loss": 0.07721710205078125,
      "step": 3553
    },
    {
      "epoch": 0.4952274785759075,
      "grad_norm": 1.4858375787734985,
      "learning_rate": 2.2739371371841103e-05,
      "loss": 0.17850112915039062,
      "step": 3554
    },
    {
      "epoch": 0.49536682226712186,
      "grad_norm": 0.8760536313056946,
      "learning_rate": 2.2730032780160128e-05,
      "loss": 0.1501445770263672,
      "step": 3555
    },
    {
      "epoch": 0.49550616595833624,
      "grad_norm": 0.5283787846565247,
      "learning_rate": 2.27206935819282e-05,
      "loss": 0.1293659210205078,
      "step": 3556
    },
    {
      "epoch": 0.4956455096495506,
      "grad_norm": 0.6187781691551208,
      "learning_rate": 2.2711353779220278e-05,
      "loss": 0.10190963745117188,
      "step": 3557
    },
    {
      "epoch": 0.495784853340765,
      "grad_norm": 0.49607914686203003,
      "learning_rate": 2.2702013374111443e-05,
      "loss": 0.10712909698486328,
      "step": 3558
    },
    {
      "epoch": 0.4959241970319794,
      "grad_norm": 0.42386361956596375,
      "learning_rate": 2.2692672368676925e-05,
      "loss": 0.07897567749023438,
      "step": 3559
    },
    {
      "epoch": 0.49606354072319375,
      "grad_norm": 0.3044247031211853,
      "learning_rate": 2.2683330764992083e-05,
      "loss": 0.0748286247253418,
      "step": 3560
    },
    {
      "epoch": 0.49620288441440813,
      "grad_norm": 0.37392935156822205,
      "learning_rate": 2.2673988565132404e-05,
      "loss": 0.06544876098632812,
      "step": 3561
    },
    {
      "epoch": 0.4963422281056225,
      "grad_norm": 0.49476325511932373,
      "learning_rate": 2.266464577117352e-05,
      "loss": 0.09716796875,
      "step": 3562
    },
    {
      "epoch": 0.4964815717968369,
      "grad_norm": 0.6164884567260742,
      "learning_rate": 2.2655302385191176e-05,
      "loss": 0.11668205261230469,
      "step": 3563
    },
    {
      "epoch": 0.49662091548805126,
      "grad_norm": 0.4746073782444,
      "learning_rate": 2.2645958409261256e-05,
      "loss": 0.11136817932128906,
      "step": 3564
    },
    {
      "epoch": 0.49676025917926564,
      "grad_norm": 0.36123624444007874,
      "learning_rate": 2.2636613845459802e-05,
      "loss": 0.07408714294433594,
      "step": 3565
    },
    {
      "epoch": 0.49689960287048,
      "grad_norm": 0.29045364260673523,
      "learning_rate": 2.262726869586293e-05,
      "loss": 0.080230712890625,
      "step": 3566
    },
    {
      "epoch": 0.4970389465616944,
      "grad_norm": 0.3141564130783081,
      "learning_rate": 2.2617922962546946e-05,
      "loss": 0.06623554229736328,
      "step": 3567
    },
    {
      "epoch": 0.49717829025290877,
      "grad_norm": 0.4020007848739624,
      "learning_rate": 2.2608576647588242e-05,
      "loss": 0.08577537536621094,
      "step": 3568
    },
    {
      "epoch": 0.4973176339441232,
      "grad_norm": 0.43270909786224365,
      "learning_rate": 2.2599229753063368e-05,
      "loss": 0.10297966003417969,
      "step": 3569
    },
    {
      "epoch": 0.4974569776353376,
      "grad_norm": 0.4862155616283417,
      "learning_rate": 2.2589882281048984e-05,
      "loss": 0.09780311584472656,
      "step": 3570
    },
    {
      "epoch": 0.49759632132655196,
      "grad_norm": 0.3469257354736328,
      "learning_rate": 2.258053423362188e-05,
      "loss": 0.08203411102294922,
      "step": 3571
    },
    {
      "epoch": 0.49773566501776634,
      "grad_norm": 0.27453097701072693,
      "learning_rate": 2.2571185612858987e-05,
      "loss": 0.07065439224243164,
      "step": 3572
    },
    {
      "epoch": 0.4978750087089807,
      "grad_norm": 0.40304479002952576,
      "learning_rate": 2.256183642083735e-05,
      "loss": 0.07018470764160156,
      "step": 3573
    },
    {
      "epoch": 0.4980143524001951,
      "grad_norm": 0.4901616871356964,
      "learning_rate": 2.2552486659634148e-05,
      "loss": 0.11204910278320312,
      "step": 3574
    },
    {
      "epoch": 0.49815369609140947,
      "grad_norm": 0.29293447732925415,
      "learning_rate": 2.2543136331326684e-05,
      "loss": 0.08166694641113281,
      "step": 3575
    },
    {
      "epoch": 0.49829303978262385,
      "grad_norm": 0.3444250524044037,
      "learning_rate": 2.2533785437992392e-05,
      "loss": 0.08587265014648438,
      "step": 3576
    },
    {
      "epoch": 0.4984323834738382,
      "grad_norm": 0.581868588924408,
      "learning_rate": 2.2524433981708822e-05,
      "loss": 0.11177444458007812,
      "step": 3577
    },
    {
      "epoch": 0.4985717271650526,
      "grad_norm": 0.3277111351490021,
      "learning_rate": 2.2515081964553655e-05,
      "loss": 0.08993244171142578,
      "step": 3578
    },
    {
      "epoch": 0.498711070856267,
      "grad_norm": 0.45504358410835266,
      "learning_rate": 2.2505729388604692e-05,
      "loss": 0.09328460693359375,
      "step": 3579
    },
    {
      "epoch": 0.49885041454748136,
      "grad_norm": 0.4179115295410156,
      "learning_rate": 2.2496376255939866e-05,
      "loss": 0.09133720397949219,
      "step": 3580
    },
    {
      "epoch": 0.49898975823869574,
      "grad_norm": 0.3786357641220093,
      "learning_rate": 2.2487022568637236e-05,
      "loss": 0.08354568481445312,
      "step": 3581
    },
    {
      "epoch": 0.4991291019299101,
      "grad_norm": 0.37199994921684265,
      "learning_rate": 2.247766832877496e-05,
      "loss": 0.09311962127685547,
      "step": 3582
    },
    {
      "epoch": 0.4992684456211245,
      "grad_norm": 0.3573143184185028,
      "learning_rate": 2.2468313538431355e-05,
      "loss": 0.085205078125,
      "step": 3583
    },
    {
      "epoch": 0.49940778931233887,
      "grad_norm": 0.5203757882118225,
      "learning_rate": 2.245895819968483e-05,
      "loss": 0.11702156066894531,
      "step": 3584
    },
    {
      "epoch": 0.49954713300355325,
      "grad_norm": 0.36471644043922424,
      "learning_rate": 2.2449602314613937e-05,
      "loss": 0.08889389038085938,
      "step": 3585
    },
    {
      "epoch": 0.4996864766947676,
      "grad_norm": 0.47943583130836487,
      "learning_rate": 2.244024588529734e-05,
      "loss": 0.10224151611328125,
      "step": 3586
    },
    {
      "epoch": 0.499825820385982,
      "grad_norm": 0.346510112285614,
      "learning_rate": 2.2430888913813807e-05,
      "loss": 0.08899497985839844,
      "step": 3587
    },
    {
      "epoch": 0.4999651640771964,
      "grad_norm": 0.34265241026878357,
      "learning_rate": 2.242153140224226e-05,
      "loss": 0.08121681213378906,
      "step": 3588
    },
    {
      "epoch": 0.5001045077684108,
      "grad_norm": 0.4229574501514435,
      "learning_rate": 2.2412173352661722e-05,
      "loss": 0.09548187255859375,
      "step": 3589
    },
    {
      "epoch": 0.5002438514596251,
      "grad_norm": 0.49202293157577515,
      "learning_rate": 2.2402814767151333e-05,
      "loss": 0.10205841064453125,
      "step": 3590
    },
    {
      "epoch": 0.5003831951508395,
      "grad_norm": 0.7047402262687683,
      "learning_rate": 2.2393455647790363e-05,
      "loss": 0.13703346252441406,
      "step": 3591
    },
    {
      "epoch": 0.5005225388420539,
      "grad_norm": 0.4008718729019165,
      "learning_rate": 2.2384095996658188e-05,
      "loss": 0.08972358703613281,
      "step": 3592
    },
    {
      "epoch": 0.5006618825332683,
      "grad_norm": 0.28691425919532776,
      "learning_rate": 2.2374735815834315e-05,
      "loss": 0.08510208129882812,
      "step": 3593
    },
    {
      "epoch": 0.5008012262244826,
      "grad_norm": 0.3459254801273346,
      "learning_rate": 2.2365375107398363e-05,
      "loss": 0.09442138671875,
      "step": 3594
    },
    {
      "epoch": 0.500940569915697,
      "grad_norm": 0.313025563955307,
      "learning_rate": 2.2356013873430058e-05,
      "loss": 0.08616447448730469,
      "step": 3595
    },
    {
      "epoch": 0.5010799136069114,
      "grad_norm": 0.4326810836791992,
      "learning_rate": 2.2346652116009256e-05,
      "loss": 0.08737373352050781,
      "step": 3596
    },
    {
      "epoch": 0.5012192572981258,
      "grad_norm": 0.32725492119789124,
      "learning_rate": 2.2337289837215937e-05,
      "loss": 0.0787811279296875,
      "step": 3597
    },
    {
      "epoch": 0.5013586009893402,
      "grad_norm": 0.16523022949695587,
      "learning_rate": 2.232792703913017e-05,
      "loss": 0.05519676208496094,
      "step": 3598
    },
    {
      "epoch": 0.5014979446805546,
      "grad_norm": 0.387765109539032,
      "learning_rate": 2.2318563723832173e-05,
      "loss": 0.07978439331054688,
      "step": 3599
    },
    {
      "epoch": 0.501637288371769,
      "grad_norm": 0.24549956619739532,
      "learning_rate": 2.230919989340224e-05,
      "loss": 0.07455253601074219,
      "step": 3600
    },
    {
      "epoch": 0.5017766320629834,
      "grad_norm": 0.6037759184837341,
      "learning_rate": 2.2299835549920822e-05,
      "loss": 0.157470703125,
      "step": 3601
    },
    {
      "epoch": 0.5019159757541978,
      "grad_norm": 0.5356497764587402,
      "learning_rate": 2.2290470695468443e-05,
      "loss": 0.0875091552734375,
      "step": 3602
    },
    {
      "epoch": 0.5020553194454122,
      "grad_norm": 0.3111777901649475,
      "learning_rate": 2.2281105332125765e-05,
      "loss": 0.07269287109375,
      "step": 3603
    },
    {
      "epoch": 0.5021946631366265,
      "grad_norm": 0.28498589992523193,
      "learning_rate": 2.2271739461973567e-05,
      "loss": 0.08382415771484375,
      "step": 3604
    },
    {
      "epoch": 0.5023340068278409,
      "grad_norm": 0.4115041196346283,
      "learning_rate": 2.2262373087092722e-05,
      "loss": 0.10698318481445312,
      "step": 3605
    },
    {
      "epoch": 0.5024733505190553,
      "grad_norm": 0.42271724343299866,
      "learning_rate": 2.2253006209564233e-05,
      "loss": 0.106201171875,
      "step": 3606
    },
    {
      "epoch": 0.5026126942102697,
      "grad_norm": 0.6989468336105347,
      "learning_rate": 2.2243638831469197e-05,
      "loss": 0.10185718536376953,
      "step": 3607
    },
    {
      "epoch": 0.502752037901484,
      "grad_norm": 0.4073905348777771,
      "learning_rate": 2.2234270954888833e-05,
      "loss": 0.094024658203125,
      "step": 3608
    },
    {
      "epoch": 0.5028913815926984,
      "grad_norm": 0.2625499367713928,
      "learning_rate": 2.2224902581904476e-05,
      "loss": 0.08005332946777344,
      "step": 3609
    },
    {
      "epoch": 0.5030307252839128,
      "grad_norm": 0.6455435752868652,
      "learning_rate": 2.221553371459756e-05,
      "loss": 0.11818695068359375,
      "step": 3610
    },
    {
      "epoch": 0.5031700689751272,
      "grad_norm": 0.27758702635765076,
      "learning_rate": 2.2206164355049634e-05,
      "loss": 0.07831954956054688,
      "step": 3611
    },
    {
      "epoch": 0.5033094126663415,
      "grad_norm": 0.4879065155982971,
      "learning_rate": 2.2196794505342358e-05,
      "loss": 0.09527969360351562,
      "step": 3612
    },
    {
      "epoch": 0.5034487563575559,
      "grad_norm": 0.37370091676712036,
      "learning_rate": 2.2187424167557496e-05,
      "loss": 0.10657215118408203,
      "step": 3613
    },
    {
      "epoch": 0.5035881000487703,
      "grad_norm": 0.4137735664844513,
      "learning_rate": 2.2178053343776912e-05,
      "loss": 0.10835838317871094,
      "step": 3614
    },
    {
      "epoch": 0.5037274437399847,
      "grad_norm": 0.5408356785774231,
      "learning_rate": 2.216868203608262e-05,
      "loss": 0.11116600036621094,
      "step": 3615
    },
    {
      "epoch": 0.5038667874311991,
      "grad_norm": 0.35009750723838806,
      "learning_rate": 2.2159310246556675e-05,
      "loss": 0.08928298950195312,
      "step": 3616
    },
    {
      "epoch": 0.5040061311224134,
      "grad_norm": 0.40930449962615967,
      "learning_rate": 2.2149937977281296e-05,
      "loss": 0.09708404541015625,
      "step": 3617
    },
    {
      "epoch": 0.5041454748136278,
      "grad_norm": 0.21403788030147552,
      "learning_rate": 2.214056523033879e-05,
      "loss": 0.07234525680541992,
      "step": 3618
    },
    {
      "epoch": 0.5042848185048422,
      "grad_norm": 0.5752096772193909,
      "learning_rate": 2.2131192007811552e-05,
      "loss": 0.07707881927490234,
      "step": 3619
    },
    {
      "epoch": 0.5044241621960566,
      "grad_norm": 0.5906797051429749,
      "learning_rate": 2.2121818311782116e-05,
      "loss": 0.08550453186035156,
      "step": 3620
    },
    {
      "epoch": 0.504563505887271,
      "grad_norm": 0.3406551480293274,
      "learning_rate": 2.211244414433308e-05,
      "loss": 0.08963584899902344,
      "step": 3621
    },
    {
      "epoch": 0.5047028495784853,
      "grad_norm": 0.24212178587913513,
      "learning_rate": 2.2103069507547187e-05,
      "loss": 0.0862722396850586,
      "step": 3622
    },
    {
      "epoch": 0.5048421932696997,
      "grad_norm": 0.5994324684143066,
      "learning_rate": 2.2093694403507264e-05,
      "loss": 0.12993812561035156,
      "step": 3623
    },
    {
      "epoch": 0.5049815369609141,
      "grad_norm": 0.3297486901283264,
      "learning_rate": 2.208431883429625e-05,
      "loss": 0.08048820495605469,
      "step": 3624
    },
    {
      "epoch": 0.5051208806521285,
      "grad_norm": 0.33278051018714905,
      "learning_rate": 2.207494280199717e-05,
      "loss": 0.08412647247314453,
      "step": 3625
    },
    {
      "epoch": 0.5052602243433428,
      "grad_norm": 0.4243289530277252,
      "learning_rate": 2.2065566308693173e-05,
      "loss": 0.08528900146484375,
      "step": 3626
    },
    {
      "epoch": 0.5053995680345572,
      "grad_norm": 0.37780314683914185,
      "learning_rate": 2.2056189356467498e-05,
      "loss": 0.07941246032714844,
      "step": 3627
    },
    {
      "epoch": 0.5055389117257716,
      "grad_norm": 0.39797845482826233,
      "learning_rate": 2.2046811947403492e-05,
      "loss": 0.07242012023925781,
      "step": 3628
    },
    {
      "epoch": 0.505678255416986,
      "grad_norm": 0.6362996697425842,
      "learning_rate": 2.2037434083584605e-05,
      "loss": 0.11664772033691406,
      "step": 3629
    },
    {
      "epoch": 0.5058175991082003,
      "grad_norm": 0.34501734375953674,
      "learning_rate": 2.2028055767094372e-05,
      "loss": 0.08011245727539062,
      "step": 3630
    },
    {
      "epoch": 0.5059569427994147,
      "grad_norm": 0.4625336825847626,
      "learning_rate": 2.2018677000016463e-05,
      "loss": 0.09065055847167969,
      "step": 3631
    },
    {
      "epoch": 0.5060962864906291,
      "grad_norm": 0.5450413823127747,
      "learning_rate": 2.2009297784434595e-05,
      "loss": 0.09954833984375,
      "step": 3632
    },
    {
      "epoch": 0.5062356301818435,
      "grad_norm": 0.5648554563522339,
      "learning_rate": 2.199991812243264e-05,
      "loss": 0.08428955078125,
      "step": 3633
    },
    {
      "epoch": 0.5063749738730579,
      "grad_norm": 0.4863109588623047,
      "learning_rate": 2.1990538016094537e-05,
      "loss": 0.07711029052734375,
      "step": 3634
    },
    {
      "epoch": 0.5065143175642722,
      "grad_norm": 0.4763026833534241,
      "learning_rate": 2.1981157467504332e-05,
      "loss": 0.10681724548339844,
      "step": 3635
    },
    {
      "epoch": 0.5066536612554866,
      "grad_norm": 0.6573106050491333,
      "learning_rate": 2.1971776478746176e-05,
      "loss": 0.11796283721923828,
      "step": 3636
    },
    {
      "epoch": 0.506793004946701,
      "grad_norm": 0.3796146512031555,
      "learning_rate": 2.196239505190429e-05,
      "loss": 0.08820915222167969,
      "step": 3637
    },
    {
      "epoch": 0.5069323486379154,
      "grad_norm": 0.3239118754863739,
      "learning_rate": 2.195301318906303e-05,
      "loss": 0.06958198547363281,
      "step": 3638
    },
    {
      "epoch": 0.5070716923291299,
      "grad_norm": 0.4484650492668152,
      "learning_rate": 2.194363089230683e-05,
      "loss": 0.07629203796386719,
      "step": 3639
    },
    {
      "epoch": 0.5072110360203442,
      "grad_norm": 0.38783514499664307,
      "learning_rate": 2.193424816372022e-05,
      "loss": 0.09799766540527344,
      "step": 3640
    },
    {
      "epoch": 0.5073503797115586,
      "grad_norm": 0.31916338205337524,
      "learning_rate": 2.1924865005387822e-05,
      "loss": 0.07931709289550781,
      "step": 3641
    },
    {
      "epoch": 0.507489723402773,
      "grad_norm": 0.4265303909778595,
      "learning_rate": 2.1915481419394373e-05,
      "loss": 0.12880897521972656,
      "step": 3642
    },
    {
      "epoch": 0.5076290670939874,
      "grad_norm": 0.7687326073646545,
      "learning_rate": 2.190609740782468e-05,
      "loss": 0.10861682891845703,
      "step": 3643
    },
    {
      "epoch": 0.5077684107852017,
      "grad_norm": 0.2703242897987366,
      "learning_rate": 2.1896712972763658e-05,
      "loss": 0.07837295532226562,
      "step": 3644
    },
    {
      "epoch": 0.5079077544764161,
      "grad_norm": 0.5065028667449951,
      "learning_rate": 2.1887328116296315e-05,
      "loss": 0.10987091064453125,
      "step": 3645
    },
    {
      "epoch": 0.5080470981676305,
      "grad_norm": 0.2101999968290329,
      "learning_rate": 2.1877942840507752e-05,
      "loss": 0.0772552490234375,
      "step": 3646
    },
    {
      "epoch": 0.5081864418588449,
      "grad_norm": 0.3729335367679596,
      "learning_rate": 2.1868557147483176e-05,
      "loss": 0.10075759887695312,
      "step": 3647
    },
    {
      "epoch": 0.5083257855500593,
      "grad_norm": 0.5413200855255127,
      "learning_rate": 2.1859171039307848e-05,
      "loss": 0.1278533935546875,
      "step": 3648
    },
    {
      "epoch": 0.5084651292412736,
      "grad_norm": 0.7528550028800964,
      "learning_rate": 2.1849784518067172e-05,
      "loss": 0.12228012084960938,
      "step": 3649
    },
    {
      "epoch": 0.508604472932488,
      "grad_norm": 0.18242529034614563,
      "learning_rate": 2.1840397585846594e-05,
      "loss": 0.06589698791503906,
      "step": 3650
    },
    {
      "epoch": 0.5087438166237024,
      "grad_norm": 0.4570610523223877,
      "learning_rate": 2.1831010244731697e-05,
      "loss": 0.10050392150878906,
      "step": 3651
    },
    {
      "epoch": 0.5088831603149168,
      "grad_norm": 0.22792956233024597,
      "learning_rate": 2.182162249680813e-05,
      "loss": 0.06468677520751953,
      "step": 3652
    },
    {
      "epoch": 0.5090225040061311,
      "grad_norm": 0.5405930876731873,
      "learning_rate": 2.1812234344161623e-05,
      "loss": 0.11087608337402344,
      "step": 3653
    },
    {
      "epoch": 0.5091618476973455,
      "grad_norm": 0.4114237427711487,
      "learning_rate": 2.1802845788878027e-05,
      "loss": 0.09141349792480469,
      "step": 3654
    },
    {
      "epoch": 0.5093011913885599,
      "grad_norm": 0.2723290026187897,
      "learning_rate": 2.1793456833043253e-05,
      "loss": 0.06774520874023438,
      "step": 3655
    },
    {
      "epoch": 0.5094405350797743,
      "grad_norm": 0.26907384395599365,
      "learning_rate": 2.1784067478743317e-05,
      "loss": 0.0638885498046875,
      "step": 3656
    },
    {
      "epoch": 0.5095798787709886,
      "grad_norm": 0.3670291304588318,
      "learning_rate": 2.177467772806432e-05,
      "loss": 0.08334732055664062,
      "step": 3657
    },
    {
      "epoch": 0.509719222462203,
      "grad_norm": 0.291133850812912,
      "learning_rate": 2.1765287583092447e-05,
      "loss": 0.0897064208984375,
      "step": 3658
    },
    {
      "epoch": 0.5098585661534174,
      "grad_norm": 0.27694907784461975,
      "learning_rate": 2.1755897045913975e-05,
      "loss": 0.08148193359375,
      "step": 3659
    },
    {
      "epoch": 0.5099979098446318,
      "grad_norm": 0.34353968501091003,
      "learning_rate": 2.1746506118615267e-05,
      "loss": 0.07778263092041016,
      "step": 3660
    },
    {
      "epoch": 0.5101372535358462,
      "grad_norm": 0.6711382865905762,
      "learning_rate": 2.173711480328277e-05,
      "loss": 0.10190963745117188,
      "step": 3661
    },
    {
      "epoch": 0.5102765972270605,
      "grad_norm": 0.41960564255714417,
      "learning_rate": 2.1727723102003023e-05,
      "loss": 0.10080718994140625,
      "step": 3662
    },
    {
      "epoch": 0.5104159409182749,
      "grad_norm": 0.3519643247127533,
      "learning_rate": 2.1718331016862657e-05,
      "loss": 0.08876800537109375,
      "step": 3663
    },
    {
      "epoch": 0.5105552846094893,
      "grad_norm": 0.6020705103874207,
      "learning_rate": 2.1708938549948354e-05,
      "loss": 0.12206268310546875,
      "step": 3664
    },
    {
      "epoch": 0.5106946283007037,
      "grad_norm": 0.27140310406684875,
      "learning_rate": 2.1699545703346934e-05,
      "loss": 0.08456611633300781,
      "step": 3665
    },
    {
      "epoch": 0.510833971991918,
      "grad_norm": 0.27216362953186035,
      "learning_rate": 2.1690152479145254e-05,
      "loss": 0.07582759857177734,
      "step": 3666
    },
    {
      "epoch": 0.5109733156831324,
      "grad_norm": 0.47954046726226807,
      "learning_rate": 2.1680758879430283e-05,
      "loss": 0.08698272705078125,
      "step": 3667
    },
    {
      "epoch": 0.5111126593743468,
      "grad_norm": 0.32324761152267456,
      "learning_rate": 2.1671364906289053e-05,
      "loss": 0.08398056030273438,
      "step": 3668
    },
    {
      "epoch": 0.5112520030655612,
      "grad_norm": 0.3007957935333252,
      "learning_rate": 2.166197056180871e-05,
      "loss": 0.07563591003417969,
      "step": 3669
    },
    {
      "epoch": 0.5113913467567756,
      "grad_norm": 0.214572474360466,
      "learning_rate": 2.1652575848076446e-05,
      "loss": 0.06020927429199219,
      "step": 3670
    },
    {
      "epoch": 0.5115306904479899,
      "grad_norm": 1.1594676971435547,
      "learning_rate": 2.1643180767179558e-05,
      "loss": 0.15116310119628906,
      "step": 3671
    },
    {
      "epoch": 0.5116700341392043,
      "grad_norm": 0.5094106197357178,
      "learning_rate": 2.163378532120542e-05,
      "loss": 0.09107780456542969,
      "step": 3672
    },
    {
      "epoch": 0.5118093778304187,
      "grad_norm": 0.2819201648235321,
      "learning_rate": 2.162438951224148e-05,
      "loss": 0.06962776184082031,
      "step": 3673
    },
    {
      "epoch": 0.5119487215216331,
      "grad_norm": 0.43773022294044495,
      "learning_rate": 2.1614993342375277e-05,
      "loss": 0.08666706085205078,
      "step": 3674
    },
    {
      "epoch": 0.5120880652128474,
      "grad_norm": 0.4743880331516266,
      "learning_rate": 2.1605596813694426e-05,
      "loss": 0.07237815856933594,
      "step": 3675
    },
    {
      "epoch": 0.5122274089040618,
      "grad_norm": 0.3021768033504486,
      "learning_rate": 2.1596199928286618e-05,
      "loss": 0.08869361877441406,
      "step": 3676
    },
    {
      "epoch": 0.5123667525952762,
      "grad_norm": 0.8729799389839172,
      "learning_rate": 2.1586802688239627e-05,
      "loss": 0.1299877166748047,
      "step": 3677
    },
    {
      "epoch": 0.5125060962864906,
      "grad_norm": 0.3595752716064453,
      "learning_rate": 2.1577405095641307e-05,
      "loss": 0.080841064453125,
      "step": 3678
    },
    {
      "epoch": 0.5126454399777051,
      "grad_norm": 0.5820650458335876,
      "learning_rate": 2.156800715257959e-05,
      "loss": 0.11102962493896484,
      "step": 3679
    },
    {
      "epoch": 0.5127847836689194,
      "grad_norm": 0.4247981905937195,
      "learning_rate": 2.1558608861142472e-05,
      "loss": 0.10100936889648438,
      "step": 3680
    },
    {
      "epoch": 0.5129241273601338,
      "grad_norm": 0.37328967452049255,
      "learning_rate": 2.1549210223418063e-05,
      "loss": 0.09741592407226562,
      "step": 3681
    },
    {
      "epoch": 0.5130634710513482,
      "grad_norm": 0.29361581802368164,
      "learning_rate": 2.15398112414945e-05,
      "loss": 0.08204269409179688,
      "step": 3682
    },
    {
      "epoch": 0.5132028147425626,
      "grad_norm": 0.5099249482154846,
      "learning_rate": 2.1530411917460037e-05,
      "loss": 0.07933902740478516,
      "step": 3683
    },
    {
      "epoch": 0.513342158433777,
      "grad_norm": 0.42604923248291016,
      "learning_rate": 2.1521012253402987e-05,
      "loss": 0.07947540283203125,
      "step": 3684
    },
    {
      "epoch": 0.5134815021249913,
      "grad_norm": 0.31628602743148804,
      "learning_rate": 2.151161225141174e-05,
      "loss": 0.08016777038574219,
      "step": 3685
    },
    {
      "epoch": 0.5136208458162057,
      "grad_norm": 0.41677600145339966,
      "learning_rate": 2.1502211913574764e-05,
      "loss": 0.09615898132324219,
      "step": 3686
    },
    {
      "epoch": 0.5137601895074201,
      "grad_norm": 0.4476262032985687,
      "learning_rate": 2.1492811241980595e-05,
      "loss": 0.11306190490722656,
      "step": 3687
    },
    {
      "epoch": 0.5138995331986345,
      "grad_norm": 0.19930614531040192,
      "learning_rate": 2.1483410238717844e-05,
      "loss": 0.06281852722167969,
      "step": 3688
    },
    {
      "epoch": 0.5140388768898488,
      "grad_norm": 0.486012727022171,
      "learning_rate": 2.147400890587521e-05,
      "loss": 0.08540534973144531,
      "step": 3689
    },
    {
      "epoch": 0.5141782205810632,
      "grad_norm": 0.3799509108066559,
      "learning_rate": 2.146460724554145e-05,
      "loss": 0.09944915771484375,
      "step": 3690
    },
    {
      "epoch": 0.5143175642722776,
      "grad_norm": 0.6501029133796692,
      "learning_rate": 2.1455205259805396e-05,
      "loss": 0.11091423034667969,
      "step": 3691
    },
    {
      "epoch": 0.514456907963492,
      "grad_norm": 0.3774292767047882,
      "learning_rate": 2.1445802950755956e-05,
      "loss": 0.09331130981445312,
      "step": 3692
    },
    {
      "epoch": 0.5145962516547063,
      "grad_norm": 0.41339847445487976,
      "learning_rate": 2.143640032048211e-05,
      "loss": 0.07103729248046875,
      "step": 3693
    },
    {
      "epoch": 0.5147355953459207,
      "grad_norm": 0.5329321622848511,
      "learning_rate": 2.1426997371072905e-05,
      "loss": 0.11150932312011719,
      "step": 3694
    },
    {
      "epoch": 0.5148749390371351,
      "grad_norm": 0.4090617597103119,
      "learning_rate": 2.141759410461746e-05,
      "loss": 0.08116626739501953,
      "step": 3695
    },
    {
      "epoch": 0.5150142827283495,
      "grad_norm": 0.3484518527984619,
      "learning_rate": 2.140819052320497e-05,
      "loss": 0.09588623046875,
      "step": 3696
    },
    {
      "epoch": 0.5151536264195639,
      "grad_norm": 0.44340309500694275,
      "learning_rate": 2.1398786628924705e-05,
      "loss": 0.10256767272949219,
      "step": 3697
    },
    {
      "epoch": 0.5152929701107782,
      "grad_norm": 0.8933324813842773,
      "learning_rate": 2.1389382423865973e-05,
      "loss": 0.13751602172851562,
      "step": 3698
    },
    {
      "epoch": 0.5154323138019926,
      "grad_norm": 0.2327093929052353,
      "learning_rate": 2.137997791011819e-05,
      "loss": 0.06946754455566406,
      "step": 3699
    },
    {
      "epoch": 0.515571657493207,
      "grad_norm": 0.4385296702384949,
      "learning_rate": 2.1370573089770823e-05,
      "loss": 0.10528945922851562,
      "step": 3700
    },
    {
      "epoch": 0.5157110011844214,
      "grad_norm": 0.43449580669403076,
      "learning_rate": 2.136116796491341e-05,
      "loss": 0.11512565612792969,
      "step": 3701
    },
    {
      "epoch": 0.5158503448756357,
      "grad_norm": 0.1449153572320938,
      "learning_rate": 2.1351762537635553e-05,
      "loss": 0.05810356140136719,
      "step": 3702
    },
    {
      "epoch": 0.5159896885668501,
      "grad_norm": 0.31845197081565857,
      "learning_rate": 2.134235681002691e-05,
      "loss": 0.08333396911621094,
      "step": 3703
    },
    {
      "epoch": 0.5161290322580645,
      "grad_norm": 0.8715630173683167,
      "learning_rate": 2.1332950784177235e-05,
      "loss": 0.11542701721191406,
      "step": 3704
    },
    {
      "epoch": 0.5162683759492789,
      "grad_norm": 0.3828117847442627,
      "learning_rate": 2.1323544462176325e-05,
      "loss": 0.0932464599609375,
      "step": 3705
    },
    {
      "epoch": 0.5164077196404933,
      "grad_norm": 0.2595888674259186,
      "learning_rate": 2.131413784611406e-05,
      "loss": 0.07486724853515625,
      "step": 3706
    },
    {
      "epoch": 0.5165470633317076,
      "grad_norm": 0.48626959323883057,
      "learning_rate": 2.1304730938080364e-05,
      "loss": 0.11182975769042969,
      "step": 3707
    },
    {
      "epoch": 0.516686407022922,
      "grad_norm": 0.5025616884231567,
      "learning_rate": 2.129532374016524e-05,
      "loss": 0.1324024200439453,
      "step": 3708
    },
    {
      "epoch": 0.5168257507141364,
      "grad_norm": 0.5887278914451599,
      "learning_rate": 2.128591625445876e-05,
      "loss": 0.11343955993652344,
      "step": 3709
    },
    {
      "epoch": 0.5169650944053508,
      "grad_norm": 0.669082760810852,
      "learning_rate": 2.127650848305104e-05,
      "loss": 0.10241508483886719,
      "step": 3710
    },
    {
      "epoch": 0.5171044380965651,
      "grad_norm": 0.38598114252090454,
      "learning_rate": 2.1267100428032276e-05,
      "loss": 0.08838272094726562,
      "step": 3711
    },
    {
      "epoch": 0.5172437817877795,
      "grad_norm": 0.2648531198501587,
      "learning_rate": 2.1257692091492724e-05,
      "loss": 0.08539199829101562,
      "step": 3712
    },
    {
      "epoch": 0.5173831254789939,
      "grad_norm": 0.27213671803474426,
      "learning_rate": 2.1248283475522712e-05,
      "loss": 0.08123111724853516,
      "step": 3713
    },
    {
      "epoch": 0.5175224691702083,
      "grad_norm": 0.4318782687187195,
      "learning_rate": 2.1238874582212602e-05,
      "loss": 0.09605121612548828,
      "step": 3714
    },
    {
      "epoch": 0.5176618128614227,
      "grad_norm": 0.3763803541660309,
      "learning_rate": 2.1229465413652854e-05,
      "loss": 0.08822441101074219,
      "step": 3715
    },
    {
      "epoch": 0.517801156552637,
      "grad_norm": 0.6045380234718323,
      "learning_rate": 2.122005597193395e-05,
      "loss": 0.09681987762451172,
      "step": 3716
    },
    {
      "epoch": 0.5179405002438514,
      "grad_norm": 0.27451291680336,
      "learning_rate": 2.1210646259146466e-05,
      "loss": 0.08495330810546875,
      "step": 3717
    },
    {
      "epoch": 0.5180798439350658,
      "grad_norm": 0.269501268863678,
      "learning_rate": 2.1201236277381028e-05,
      "loss": 0.07447242736816406,
      "step": 3718
    },
    {
      "epoch": 0.5182191876262803,
      "grad_norm": 0.26348036527633667,
      "learning_rate": 2.119182602872831e-05,
      "loss": 0.07519245147705078,
      "step": 3719
    },
    {
      "epoch": 0.5183585313174947,
      "grad_norm": 0.502983570098877,
      "learning_rate": 2.1182415515279056e-05,
      "loss": 0.11710357666015625,
      "step": 3720
    },
    {
      "epoch": 0.518497875008709,
      "grad_norm": 0.3534432053565979,
      "learning_rate": 2.117300473912407e-05,
      "loss": 0.0880136489868164,
      "step": 3721
    },
    {
      "epoch": 0.5186372186999234,
      "grad_norm": 0.3735477030277252,
      "learning_rate": 2.1163593702354213e-05,
      "loss": 0.082489013671875,
      "step": 3722
    },
    {
      "epoch": 0.5187765623911378,
      "grad_norm": 0.762271523475647,
      "learning_rate": 2.11541824070604e-05,
      "loss": 0.10651016235351562,
      "step": 3723
    },
    {
      "epoch": 0.5189159060823522,
      "grad_norm": 0.5533306002616882,
      "learning_rate": 2.114477085533361e-05,
      "loss": 0.12658309936523438,
      "step": 3724
    },
    {
      "epoch": 0.5190552497735665,
      "grad_norm": 0.3389163911342621,
      "learning_rate": 2.1135359049264868e-05,
      "loss": 0.0807638168334961,
      "step": 3725
    },
    {
      "epoch": 0.5191945934647809,
      "grad_norm": 0.4042077362537384,
      "learning_rate": 2.1125946990945264e-05,
      "loss": 0.08956146240234375,
      "step": 3726
    },
    {
      "epoch": 0.5193339371559953,
      "grad_norm": 0.6003642082214355,
      "learning_rate": 2.111653468246595e-05,
      "loss": 0.11186027526855469,
      "step": 3727
    },
    {
      "epoch": 0.5194732808472097,
      "grad_norm": 0.5273147821426392,
      "learning_rate": 2.1107122125918112e-05,
      "loss": 0.10215902328491211,
      "step": 3728
    },
    {
      "epoch": 0.519612624538424,
      "grad_norm": 0.4350491166114807,
      "learning_rate": 2.1097709323393026e-05,
      "loss": 0.11403083801269531,
      "step": 3729
    },
    {
      "epoch": 0.5197519682296384,
      "grad_norm": 0.4470135569572449,
      "learning_rate": 2.1088296276981978e-05,
      "loss": 0.08587646484375,
      "step": 3730
    },
    {
      "epoch": 0.5198913119208528,
      "grad_norm": 0.5227817296981812,
      "learning_rate": 2.1078882988776352e-05,
      "loss": 0.11548233032226562,
      "step": 3731
    },
    {
      "epoch": 0.5200306556120672,
      "grad_norm": 0.573297917842865,
      "learning_rate": 2.1069469460867547e-05,
      "loss": 0.13646888732910156,
      "step": 3732
    },
    {
      "epoch": 0.5201699993032816,
      "grad_norm": 0.32459381222724915,
      "learning_rate": 2.106005569534705e-05,
      "loss": 0.09409523010253906,
      "step": 3733
    },
    {
      "epoch": 0.5203093429944959,
      "grad_norm": 0.38113951683044434,
      "learning_rate": 2.105064169430638e-05,
      "loss": 0.09031295776367188,
      "step": 3734
    },
    {
      "epoch": 0.5204486866857103,
      "grad_norm": 0.28881099820137024,
      "learning_rate": 2.1041227459837112e-05,
      "loss": 0.07215499877929688,
      "step": 3735
    },
    {
      "epoch": 0.5205880303769247,
      "grad_norm": 0.48555436730384827,
      "learning_rate": 2.103181299403088e-05,
      "loss": 0.08647918701171875,
      "step": 3736
    },
    {
      "epoch": 0.5207273740681391,
      "grad_norm": 0.46159493923187256,
      "learning_rate": 2.1022398298979345e-05,
      "loss": 0.1125640869140625,
      "step": 3737
    },
    {
      "epoch": 0.5208667177593534,
      "grad_norm": 0.685918927192688,
      "learning_rate": 2.1012983376774255e-05,
      "loss": 0.11599922180175781,
      "step": 3738
    },
    {
      "epoch": 0.5210060614505678,
      "grad_norm": 0.21120722591876984,
      "learning_rate": 2.100356822950739e-05,
      "loss": 0.06370735168457031,
      "step": 3739
    },
    {
      "epoch": 0.5211454051417822,
      "grad_norm": 0.43901756405830383,
      "learning_rate": 2.099415285927057e-05,
      "loss": 0.09376716613769531,
      "step": 3740
    },
    {
      "epoch": 0.5212847488329966,
      "grad_norm": 0.3358260691165924,
      "learning_rate": 2.0984737268155686e-05,
      "loss": 0.08922576904296875,
      "step": 3741
    },
    {
      "epoch": 0.521424092524211,
      "grad_norm": 0.5238087177276611,
      "learning_rate": 2.097532145825466e-05,
      "loss": 0.09090042114257812,
      "step": 3742
    },
    {
      "epoch": 0.5215634362154253,
      "grad_norm": 0.42208331823349,
      "learning_rate": 2.0965905431659475e-05,
      "loss": 0.08283042907714844,
      "step": 3743
    },
    {
      "epoch": 0.5217027799066397,
      "grad_norm": 0.28850752115249634,
      "learning_rate": 2.0956489190462156e-05,
      "loss": 0.08185482025146484,
      "step": 3744
    },
    {
      "epoch": 0.5218421235978541,
      "grad_norm": 0.7066154479980469,
      "learning_rate": 2.094707273675477e-05,
      "loss": 0.1384868621826172,
      "step": 3745
    },
    {
      "epoch": 0.5219814672890685,
      "grad_norm": 0.43988296389579773,
      "learning_rate": 2.0937656072629444e-05,
      "loss": 0.09201431274414062,
      "step": 3746
    },
    {
      "epoch": 0.5221208109802828,
      "grad_norm": 0.42135465145111084,
      "learning_rate": 2.0928239200178355e-05,
      "loss": 0.08837127685546875,
      "step": 3747
    },
    {
      "epoch": 0.5222601546714972,
      "grad_norm": 0.7186415195465088,
      "learning_rate": 2.0918822121493697e-05,
      "loss": 0.1274890899658203,
      "step": 3748
    },
    {
      "epoch": 0.5223994983627116,
      "grad_norm": 0.3887670338153839,
      "learning_rate": 2.0909404838667746e-05,
      "loss": 0.08230972290039062,
      "step": 3749
    },
    {
      "epoch": 0.522538842053926,
      "grad_norm": 0.30863457918167114,
      "learning_rate": 2.08999873537928e-05,
      "loss": 0.08733367919921875,
      "step": 3750
    },
    {
      "epoch": 0.5226781857451404,
      "grad_norm": 0.3497994840145111,
      "learning_rate": 2.089056966896122e-05,
      "loss": 0.09687423706054688,
      "step": 3751
    },
    {
      "epoch": 0.5228175294363547,
      "grad_norm": 0.5708939433097839,
      "learning_rate": 2.088115178626539e-05,
      "loss": 0.09667587280273438,
      "step": 3752
    },
    {
      "epoch": 0.5229568731275691,
      "grad_norm": 0.2872249186038971,
      "learning_rate": 2.0871733707797738e-05,
      "loss": 0.06650352478027344,
      "step": 3753
    },
    {
      "epoch": 0.5230962168187835,
      "grad_norm": 0.6846075654029846,
      "learning_rate": 2.0862315435650766e-05,
      "loss": 0.1037435531616211,
      "step": 3754
    },
    {
      "epoch": 0.5232355605099979,
      "grad_norm": 0.32103070616722107,
      "learning_rate": 2.085289697191699e-05,
      "loss": 0.08530998229980469,
      "step": 3755
    },
    {
      "epoch": 0.5233749042012122,
      "grad_norm": 0.35416701436042786,
      "learning_rate": 2.0843478318688978e-05,
      "loss": 0.08759307861328125,
      "step": 3756
    },
    {
      "epoch": 0.5235142478924266,
      "grad_norm": 0.3687765896320343,
      "learning_rate": 2.083405947805934e-05,
      "loss": 0.08040428161621094,
      "step": 3757
    },
    {
      "epoch": 0.523653591583641,
      "grad_norm": 0.7897037267684937,
      "learning_rate": 2.082464045212073e-05,
      "loss": 0.10051631927490234,
      "step": 3758
    },
    {
      "epoch": 0.5237929352748554,
      "grad_norm": 0.4318329095840454,
      "learning_rate": 2.0815221242965835e-05,
      "loss": 0.09160995483398438,
      "step": 3759
    },
    {
      "epoch": 0.5239322789660699,
      "grad_norm": 0.4009858965873718,
      "learning_rate": 2.0805801852687396e-05,
      "loss": 0.09589576721191406,
      "step": 3760
    },
    {
      "epoch": 0.5240716226572842,
      "grad_norm": 0.308749258518219,
      "learning_rate": 2.0796382283378183e-05,
      "loss": 0.07955551147460938,
      "step": 3761
    },
    {
      "epoch": 0.5242109663484986,
      "grad_norm": 0.6224801540374756,
      "learning_rate": 2.0786962537131e-05,
      "loss": 0.08612060546875,
      "step": 3762
    },
    {
      "epoch": 0.524350310039713,
      "grad_norm": 0.5004847645759583,
      "learning_rate": 2.0777542616038718e-05,
      "loss": 0.08172130584716797,
      "step": 3763
    },
    {
      "epoch": 0.5244896537309274,
      "grad_norm": 0.7863150835037231,
      "learning_rate": 2.0768122522194208e-05,
      "loss": 0.12912559509277344,
      "step": 3764
    },
    {
      "epoch": 0.5246289974221418,
      "grad_norm": 0.2799774706363678,
      "learning_rate": 2.0758702257690418e-05,
      "loss": 0.07049560546875,
      "step": 3765
    },
    {
      "epoch": 0.5247683411133561,
      "grad_norm": 0.25755569338798523,
      "learning_rate": 2.0749281824620306e-05,
      "loss": 0.06817436218261719,
      "step": 3766
    },
    {
      "epoch": 0.5249076848045705,
      "grad_norm": 0.3820091784000397,
      "learning_rate": 2.073986122507688e-05,
      "loss": 0.08180046081542969,
      "step": 3767
    },
    {
      "epoch": 0.5250470284957849,
      "grad_norm": 0.39923346042633057,
      "learning_rate": 2.0730440461153183e-05,
      "loss": 0.09477615356445312,
      "step": 3768
    },
    {
      "epoch": 0.5251863721869993,
      "grad_norm": 0.4792860746383667,
      "learning_rate": 2.0721019534942285e-05,
      "loss": 0.09557151794433594,
      "step": 3769
    },
    {
      "epoch": 0.5253257158782136,
      "grad_norm": 0.32124465703964233,
      "learning_rate": 2.071159844853731e-05,
      "loss": 0.07466602325439453,
      "step": 3770
    },
    {
      "epoch": 0.525465059569428,
      "grad_norm": 0.5365555286407471,
      "learning_rate": 2.070217720403141e-05,
      "loss": 0.11181354522705078,
      "step": 3771
    },
    {
      "epoch": 0.5256044032606424,
      "grad_norm": 0.623781681060791,
      "learning_rate": 2.0692755803517764e-05,
      "loss": 0.12084197998046875,
      "step": 3772
    },
    {
      "epoch": 0.5257437469518568,
      "grad_norm": 0.25623396039009094,
      "learning_rate": 2.0683334249089593e-05,
      "loss": 0.07038688659667969,
      "step": 3773
    },
    {
      "epoch": 0.5258830906430711,
      "grad_norm": 0.43338853120803833,
      "learning_rate": 2.067391254284015e-05,
      "loss": 0.11022758483886719,
      "step": 3774
    },
    {
      "epoch": 0.5260224343342855,
      "grad_norm": 0.9088030457496643,
      "learning_rate": 2.066449068686273e-05,
      "loss": 0.1260051727294922,
      "step": 3775
    },
    {
      "epoch": 0.5261617780254999,
      "grad_norm": 0.3546534478664398,
      "learning_rate": 2.065506868325065e-05,
      "loss": 0.09360957145690918,
      "step": 3776
    },
    {
      "epoch": 0.5263011217167143,
      "grad_norm": 0.41577398777008057,
      "learning_rate": 2.0645646534097262e-05,
      "loss": 0.10062408447265625,
      "step": 3777
    },
    {
      "epoch": 0.5264404654079287,
      "grad_norm": 0.5167436003684998,
      "learning_rate": 2.0636224241495954e-05,
      "loss": 0.08769798278808594,
      "step": 3778
    },
    {
      "epoch": 0.526579809099143,
      "grad_norm": 0.4334775507450104,
      "learning_rate": 2.0626801807540148e-05,
      "loss": 0.09807777404785156,
      "step": 3779
    },
    {
      "epoch": 0.5267191527903574,
      "grad_norm": 0.2959149181842804,
      "learning_rate": 2.0617379234323285e-05,
      "loss": 0.07808113098144531,
      "step": 3780
    },
    {
      "epoch": 0.5268584964815718,
      "grad_norm": 0.7268750071525574,
      "learning_rate": 2.060795652393886e-05,
      "loss": 0.13753890991210938,
      "step": 3781
    },
    {
      "epoch": 0.5269978401727862,
      "grad_norm": 0.40258875489234924,
      "learning_rate": 2.0598533678480367e-05,
      "loss": 0.10054969787597656,
      "step": 3782
    },
    {
      "epoch": 0.5271371838640005,
      "grad_norm": 0.2940709590911865,
      "learning_rate": 2.0589110700041357e-05,
      "loss": 0.08047103881835938,
      "step": 3783
    },
    {
      "epoch": 0.5272765275552149,
      "grad_norm": 0.3144650161266327,
      "learning_rate": 2.0579687590715404e-05,
      "loss": 0.07663726806640625,
      "step": 3784
    },
    {
      "epoch": 0.5274158712464293,
      "grad_norm": 0.5522154569625854,
      "learning_rate": 2.0570264352596096e-05,
      "loss": 0.12514114379882812,
      "step": 3785
    },
    {
      "epoch": 0.5275552149376437,
      "grad_norm": 0.41386714577674866,
      "learning_rate": 2.0560840987777074e-05,
      "loss": 0.09080696105957031,
      "step": 3786
    },
    {
      "epoch": 0.5276945586288581,
      "grad_norm": 0.3696684241294861,
      "learning_rate": 2.0551417498351985e-05,
      "loss": 0.09041595458984375,
      "step": 3787
    },
    {
      "epoch": 0.5278339023200724,
      "grad_norm": 0.2963676452636719,
      "learning_rate": 2.0541993886414516e-05,
      "loss": 0.07682418823242188,
      "step": 3788
    },
    {
      "epoch": 0.5279732460112868,
      "grad_norm": 0.6746159195899963,
      "learning_rate": 2.0532570154058385e-05,
      "loss": 0.10757827758789062,
      "step": 3789
    },
    {
      "epoch": 0.5281125897025012,
      "grad_norm": 0.5066646933555603,
      "learning_rate": 2.0523146303377318e-05,
      "loss": 0.11199569702148438,
      "step": 3790
    },
    {
      "epoch": 0.5282519333937156,
      "grad_norm": 0.2815934419631958,
      "learning_rate": 2.0513722336465092e-05,
      "loss": 0.07396507263183594,
      "step": 3791
    },
    {
      "epoch": 0.52839127708493,
      "grad_norm": 0.26682940125465393,
      "learning_rate": 2.0504298255415488e-05,
      "loss": 0.07218170166015625,
      "step": 3792
    },
    {
      "epoch": 0.5285306207761443,
      "grad_norm": 0.416168212890625,
      "learning_rate": 2.0494874062322324e-05,
      "loss": 0.07485580444335938,
      "step": 3793
    },
    {
      "epoch": 0.5286699644673587,
      "grad_norm": 0.25325894355773926,
      "learning_rate": 2.0485449759279442e-05,
      "loss": 0.08135509490966797,
      "step": 3794
    },
    {
      "epoch": 0.5288093081585731,
      "grad_norm": 0.25182971358299255,
      "learning_rate": 2.047602534838071e-05,
      "loss": 0.07892608642578125,
      "step": 3795
    },
    {
      "epoch": 0.5289486518497875,
      "grad_norm": 0.6438229084014893,
      "learning_rate": 2.0466600831720006e-05,
      "loss": 0.1346454620361328,
      "step": 3796
    },
    {
      "epoch": 0.5290879955410018,
      "grad_norm": 0.7453251481056213,
      "learning_rate": 2.0457176211391257e-05,
      "loss": 0.13410663604736328,
      "step": 3797
    },
    {
      "epoch": 0.5292273392322162,
      "grad_norm": 0.7843419313430786,
      "learning_rate": 2.0447751489488387e-05,
      "loss": 0.1317424774169922,
      "step": 3798
    },
    {
      "epoch": 0.5293666829234306,
      "grad_norm": 0.4977480471134186,
      "learning_rate": 2.0438326668105364e-05,
      "loss": 0.11980056762695312,
      "step": 3799
    },
    {
      "epoch": 0.5295060266146451,
      "grad_norm": 0.4485436677932739,
      "learning_rate": 2.0428901749336157e-05,
      "loss": 0.12494277954101562,
      "step": 3800
    },
    {
      "epoch": 0.5296453703058595,
      "grad_norm": 0.5202787518501282,
      "learning_rate": 2.0419476735274774e-05,
      "loss": 0.09427452087402344,
      "step": 3801
    },
    {
      "epoch": 0.5297847139970738,
      "grad_norm": 0.4344961643218994,
      "learning_rate": 2.0410051628015247e-05,
      "loss": 0.09791183471679688,
      "step": 3802
    },
    {
      "epoch": 0.5299240576882882,
      "grad_norm": 0.29683443903923035,
      "learning_rate": 2.0400626429651595e-05,
      "loss": 0.07895755767822266,
      "step": 3803
    },
    {
      "epoch": 0.5300634013795026,
      "grad_norm": 0.46187350153923035,
      "learning_rate": 2.0391201142277905e-05,
      "loss": 0.1009368896484375,
      "step": 3804
    },
    {
      "epoch": 0.530202745070717,
      "grad_norm": 0.2928805351257324,
      "learning_rate": 2.038177576798825e-05,
      "loss": 0.07193374633789062,
      "step": 3805
    },
    {
      "epoch": 0.5303420887619313,
      "grad_norm": 0.2958899140357971,
      "learning_rate": 2.0372350308876732e-05,
      "loss": 0.07990264892578125,
      "step": 3806
    },
    {
      "epoch": 0.5304814324531457,
      "grad_norm": 0.22885552048683167,
      "learning_rate": 2.0362924767037485e-05,
      "loss": 0.06717300415039062,
      "step": 3807
    },
    {
      "epoch": 0.5306207761443601,
      "grad_norm": 0.27310827374458313,
      "learning_rate": 2.0353499144564636e-05,
      "loss": 0.06908607482910156,
      "step": 3808
    },
    {
      "epoch": 0.5307601198355745,
      "grad_norm": 0.3984753489494324,
      "learning_rate": 2.0344073443552347e-05,
      "loss": 0.07847404479980469,
      "step": 3809
    },
    {
      "epoch": 0.5308994635267889,
      "grad_norm": 0.5487156510353088,
      "learning_rate": 2.0334647666094796e-05,
      "loss": 0.10645866394042969,
      "step": 3810
    },
    {
      "epoch": 0.5310388072180032,
      "grad_norm": 0.4339144229888916,
      "learning_rate": 2.0325221814286173e-05,
      "loss": 0.0990753173828125,
      "step": 3811
    },
    {
      "epoch": 0.5311781509092176,
      "grad_norm": 0.4616285264492035,
      "learning_rate": 2.031579589022068e-05,
      "loss": 0.08770751953125,
      "step": 3812
    },
    {
      "epoch": 0.531317494600432,
      "grad_norm": 0.3664989173412323,
      "learning_rate": 2.0306369895992564e-05,
      "loss": 0.07766151428222656,
      "step": 3813
    },
    {
      "epoch": 0.5314568382916464,
      "grad_norm": 0.2466147243976593,
      "learning_rate": 2.029694383369604e-05,
      "loss": 0.07488059997558594,
      "step": 3814
    },
    {
      "epoch": 0.5315961819828607,
      "grad_norm": 0.23618072271347046,
      "learning_rate": 2.028751770542538e-05,
      "loss": 0.0716238021850586,
      "step": 3815
    },
    {
      "epoch": 0.5317355256740751,
      "grad_norm": 0.5335721373558044,
      "learning_rate": 2.0278091513274848e-05,
      "loss": 0.11327552795410156,
      "step": 3816
    },
    {
      "epoch": 0.5318748693652895,
      "grad_norm": 0.558652400970459,
      "learning_rate": 2.0268665259338736e-05,
      "loss": 0.11819648742675781,
      "step": 3817
    },
    {
      "epoch": 0.5320142130565039,
      "grad_norm": 0.8747042417526245,
      "learning_rate": 2.025923894571134e-05,
      "loss": 0.16401100158691406,
      "step": 3818
    },
    {
      "epoch": 0.5321535567477182,
      "grad_norm": 0.5491674542427063,
      "learning_rate": 2.0249812574486957e-05,
      "loss": 0.1113128662109375,
      "step": 3819
    },
    {
      "epoch": 0.5322929004389326,
      "grad_norm": 0.5198338627815247,
      "learning_rate": 2.024038614775993e-05,
      "loss": 0.09024429321289062,
      "step": 3820
    },
    {
      "epoch": 0.532432244130147,
      "grad_norm": 0.41959327459335327,
      "learning_rate": 2.0230959667624587e-05,
      "loss": 0.09379768371582031,
      "step": 3821
    },
    {
      "epoch": 0.5325715878213614,
      "grad_norm": 0.3910626769065857,
      "learning_rate": 2.022153313617528e-05,
      "loss": 0.09554386138916016,
      "step": 3822
    },
    {
      "epoch": 0.5327109315125758,
      "grad_norm": 0.3817790746688843,
      "learning_rate": 2.0212106555506364e-05,
      "loss": 0.09573745727539062,
      "step": 3823
    },
    {
      "epoch": 0.5328502752037901,
      "grad_norm": 0.7284618616104126,
      "learning_rate": 2.0202679927712224e-05,
      "loss": 0.14071369171142578,
      "step": 3824
    },
    {
      "epoch": 0.5329896188950045,
      "grad_norm": 0.7249674797058105,
      "learning_rate": 2.0193253254887223e-05,
      "loss": 0.09522438049316406,
      "step": 3825
    },
    {
      "epoch": 0.5331289625862189,
      "grad_norm": 0.5234107375144958,
      "learning_rate": 2.018382653912576e-05,
      "loss": 0.1364154815673828,
      "step": 3826
    },
    {
      "epoch": 0.5332683062774333,
      "grad_norm": 0.3435443639755249,
      "learning_rate": 2.0174399782522242e-05,
      "loss": 0.08286666870117188,
      "step": 3827
    },
    {
      "epoch": 0.5334076499686476,
      "grad_norm": 0.6361182332038879,
      "learning_rate": 2.016497298717107e-05,
      "loss": 0.09104156494140625,
      "step": 3828
    },
    {
      "epoch": 0.533546993659862,
      "grad_norm": 0.27825477719306946,
      "learning_rate": 2.015554615516667e-05,
      "loss": 0.06898307800292969,
      "step": 3829
    },
    {
      "epoch": 0.5336863373510764,
      "grad_norm": 0.5545371770858765,
      "learning_rate": 2.014611928860346e-05,
      "loss": 0.09286117553710938,
      "step": 3830
    },
    {
      "epoch": 0.5338256810422908,
      "grad_norm": 0.2747955918312073,
      "learning_rate": 2.0136692389575892e-05,
      "loss": 0.060009002685546875,
      "step": 3831
    },
    {
      "epoch": 0.5339650247335052,
      "grad_norm": 0.4900788962841034,
      "learning_rate": 2.012726546017838e-05,
      "loss": 0.07957267761230469,
      "step": 3832
    },
    {
      "epoch": 0.5341043684247195,
      "grad_norm": 0.3292944133281708,
      "learning_rate": 2.01178385025054e-05,
      "loss": 0.0825042724609375,
      "step": 3833
    },
    {
      "epoch": 0.5342437121159339,
      "grad_norm": 0.30794966220855713,
      "learning_rate": 2.0108411518651388e-05,
      "loss": 0.07288932800292969,
      "step": 3834
    },
    {
      "epoch": 0.5343830558071483,
      "grad_norm": 0.4722290337085724,
      "learning_rate": 2.0098984510710812e-05,
      "loss": 0.11651802062988281,
      "step": 3835
    },
    {
      "epoch": 0.5345223994983627,
      "grad_norm": 0.39826294779777527,
      "learning_rate": 2.0089557480778144e-05,
      "loss": 0.11958885192871094,
      "step": 3836
    },
    {
      "epoch": 0.534661743189577,
      "grad_norm": 0.29221001267433167,
      "learning_rate": 2.0080130430947842e-05,
      "loss": 0.07336187362670898,
      "step": 3837
    },
    {
      "epoch": 0.5348010868807914,
      "grad_norm": 0.38710153102874756,
      "learning_rate": 2.007070336331439e-05,
      "loss": 0.07342910766601562,
      "step": 3838
    },
    {
      "epoch": 0.5349404305720058,
      "grad_norm": 0.482023686170578,
      "learning_rate": 2.0061276279972265e-05,
      "loss": 0.11854743957519531,
      "step": 3839
    },
    {
      "epoch": 0.5350797742632203,
      "grad_norm": 0.38023141026496887,
      "learning_rate": 2.0051849183015953e-05,
      "loss": 0.0763702392578125,
      "step": 3840
    },
    {
      "epoch": 0.5352191179544347,
      "grad_norm": 0.39840298891067505,
      "learning_rate": 2.004242207453993e-05,
      "loss": 0.0769805908203125,
      "step": 3841
    },
    {
      "epoch": 0.535358461645649,
      "grad_norm": 0.5028513073921204,
      "learning_rate": 2.0032994956638695e-05,
      "loss": 0.13070106506347656,
      "step": 3842
    },
    {
      "epoch": 0.5354978053368634,
      "grad_norm": 0.5282137393951416,
      "learning_rate": 2.0023567831406733e-05,
      "loss": 0.09972667694091797,
      "step": 3843
    },
    {
      "epoch": 0.5356371490280778,
      "grad_norm": 0.49454590678215027,
      "learning_rate": 2.0014140700938532e-05,
      "loss": 0.08092117309570312,
      "step": 3844
    },
    {
      "epoch": 0.5357764927192922,
      "grad_norm": 0.5855965614318848,
      "learning_rate": 2.0004713567328594e-05,
      "loss": 0.07767486572265625,
      "step": 3845
    },
    {
      "epoch": 0.5359158364105066,
      "grad_norm": 0.5014634728431702,
      "learning_rate": 1.9995286432671412e-05,
      "loss": 0.08607864379882812,
      "step": 3846
    },
    {
      "epoch": 0.5360551801017209,
      "grad_norm": 0.482040137052536,
      "learning_rate": 1.9985859299061474e-05,
      "loss": 0.08632278442382812,
      "step": 3847
    },
    {
      "epoch": 0.5361945237929353,
      "grad_norm": 0.8831118941307068,
      "learning_rate": 1.9976432168593273e-05,
      "loss": 0.1425323486328125,
      "step": 3848
    },
    {
      "epoch": 0.5363338674841497,
      "grad_norm": 0.4596536159515381,
      "learning_rate": 1.996700504336131e-05,
      "loss": 0.07222938537597656,
      "step": 3849
    },
    {
      "epoch": 0.5364732111753641,
      "grad_norm": 0.35978060960769653,
      "learning_rate": 1.9957577925460074e-05,
      "loss": 0.08877420425415039,
      "step": 3850
    },
    {
      "epoch": 0.5366125548665784,
      "grad_norm": 0.22182075679302216,
      "learning_rate": 1.994815081698406e-05,
      "loss": 0.07085037231445312,
      "step": 3851
    },
    {
      "epoch": 0.5367518985577928,
      "grad_norm": 0.38891637325286865,
      "learning_rate": 1.9938723720027745e-05,
      "loss": 0.07542800903320312,
      "step": 3852
    },
    {
      "epoch": 0.5368912422490072,
      "grad_norm": 0.33675673604011536,
      "learning_rate": 1.9929296636685615e-05,
      "loss": 0.07526779174804688,
      "step": 3853
    },
    {
      "epoch": 0.5370305859402216,
      "grad_norm": 0.43933358788490295,
      "learning_rate": 1.9919869569052164e-05,
      "loss": 0.09370613098144531,
      "step": 3854
    },
    {
      "epoch": 0.537169929631436,
      "grad_norm": 0.35994234681129456,
      "learning_rate": 1.991044251922186e-05,
      "loss": 0.08220672607421875,
      "step": 3855
    },
    {
      "epoch": 0.5373092733226503,
      "grad_norm": 0.6342858672142029,
      "learning_rate": 1.9901015489289188e-05,
      "loss": 0.12670516967773438,
      "step": 3856
    },
    {
      "epoch": 0.5374486170138647,
      "grad_norm": 0.4870028495788574,
      "learning_rate": 1.989158848134862e-05,
      "loss": 0.11120033264160156,
      "step": 3857
    },
    {
      "epoch": 0.5375879607050791,
      "grad_norm": 0.5184480547904968,
      "learning_rate": 1.988216149749461e-05,
      "loss": 0.10367012023925781,
      "step": 3858
    },
    {
      "epoch": 0.5377273043962935,
      "grad_norm": 0.6856151819229126,
      "learning_rate": 1.9872734539821626e-05,
      "loss": 0.1280965805053711,
      "step": 3859
    },
    {
      "epoch": 0.5378666480875078,
      "grad_norm": 0.22074289619922638,
      "learning_rate": 1.9863307610424115e-05,
      "loss": 0.0680685043334961,
      "step": 3860
    },
    {
      "epoch": 0.5380059917787222,
      "grad_norm": 0.35646674036979675,
      "learning_rate": 1.985388071139654e-05,
      "loss": 0.09196853637695312,
      "step": 3861
    },
    {
      "epoch": 0.5381453354699366,
      "grad_norm": 0.46178749203681946,
      "learning_rate": 1.984445384483334e-05,
      "loss": 0.10436630249023438,
      "step": 3862
    },
    {
      "epoch": 0.538284679161151,
      "grad_norm": 0.5857861042022705,
      "learning_rate": 1.9835027012828937e-05,
      "loss": 0.11117744445800781,
      "step": 3863
    },
    {
      "epoch": 0.5384240228523653,
      "grad_norm": 0.22374816238880157,
      "learning_rate": 1.9825600217477765e-05,
      "loss": 0.0763092041015625,
      "step": 3864
    },
    {
      "epoch": 0.5385633665435797,
      "grad_norm": 0.2297087162733078,
      "learning_rate": 1.9816173460874243e-05,
      "loss": 0.056969642639160156,
      "step": 3865
    },
    {
      "epoch": 0.5387027102347941,
      "grad_norm": 0.7309514284133911,
      "learning_rate": 1.980674674511278e-05,
      "loss": 0.1277790069580078,
      "step": 3866
    },
    {
      "epoch": 0.5388420539260085,
      "grad_norm": 0.3819792866706848,
      "learning_rate": 1.9797320072287786e-05,
      "loss": 0.0969696044921875,
      "step": 3867
    },
    {
      "epoch": 0.5389813976172229,
      "grad_norm": 0.24344579875469208,
      "learning_rate": 1.9787893444493643e-05,
      "loss": 0.08148956298828125,
      "step": 3868
    },
    {
      "epoch": 0.5391207413084372,
      "grad_norm": 0.3158126473426819,
      "learning_rate": 1.9778466863824726e-05,
      "loss": 0.0901031494140625,
      "step": 3869
    },
    {
      "epoch": 0.5392600849996516,
      "grad_norm": 0.2835071384906769,
      "learning_rate": 1.9769040332375416e-05,
      "loss": 0.09578514099121094,
      "step": 3870
    },
    {
      "epoch": 0.539399428690866,
      "grad_norm": 0.42323511838912964,
      "learning_rate": 1.975961385224007e-05,
      "loss": 0.09077262878417969,
      "step": 3871
    },
    {
      "epoch": 0.5395387723820804,
      "grad_norm": 0.27880725264549255,
      "learning_rate": 1.9750187425513053e-05,
      "loss": 0.08073568344116211,
      "step": 3872
    },
    {
      "epoch": 0.5396781160732947,
      "grad_norm": 0.39048072695732117,
      "learning_rate": 1.9740761054288672e-05,
      "loss": 0.0747222900390625,
      "step": 3873
    },
    {
      "epoch": 0.5398174597645091,
      "grad_norm": 0.4747856557369232,
      "learning_rate": 1.973133474066127e-05,
      "loss": 0.11139297485351562,
      "step": 3874
    },
    {
      "epoch": 0.5399568034557235,
      "grad_norm": 0.28921470046043396,
      "learning_rate": 1.9721908486725156e-05,
      "loss": 0.07567119598388672,
      "step": 3875
    },
    {
      "epoch": 0.5400961471469379,
      "grad_norm": 0.2134292870759964,
      "learning_rate": 1.9712482294574622e-05,
      "loss": 0.061011314392089844,
      "step": 3876
    },
    {
      "epoch": 0.5402354908381523,
      "grad_norm": 0.4929518401622772,
      "learning_rate": 1.9703056166303963e-05,
      "loss": 0.10542106628417969,
      "step": 3877
    },
    {
      "epoch": 0.5403748345293666,
      "grad_norm": 0.37527820467948914,
      "learning_rate": 1.9693630104007446e-05,
      "loss": 0.09271430969238281,
      "step": 3878
    },
    {
      "epoch": 0.540514178220581,
      "grad_norm": 0.6134698390960693,
      "learning_rate": 1.9684204109779324e-05,
      "loss": 0.10982704162597656,
      "step": 3879
    },
    {
      "epoch": 0.5406535219117955,
      "grad_norm": 0.372386634349823,
      "learning_rate": 1.9674778185713834e-05,
      "loss": 0.1056985855102539,
      "step": 3880
    },
    {
      "epoch": 0.5407928656030099,
      "grad_norm": 0.3977491557598114,
      "learning_rate": 1.966535233390521e-05,
      "loss": 0.0801534652709961,
      "step": 3881
    },
    {
      "epoch": 0.5409322092942243,
      "grad_norm": 0.33028313517570496,
      "learning_rate": 1.9655926556447656e-05,
      "loss": 0.08061408996582031,
      "step": 3882
    },
    {
      "epoch": 0.5410715529854386,
      "grad_norm": 0.6627756953239441,
      "learning_rate": 1.9646500855435374e-05,
      "loss": 0.11822700500488281,
      "step": 3883
    },
    {
      "epoch": 0.541210896676653,
      "grad_norm": 0.6197908520698547,
      "learning_rate": 1.963707523296252e-05,
      "loss": 0.13428688049316406,
      "step": 3884
    },
    {
      "epoch": 0.5413502403678674,
      "grad_norm": 0.5019155740737915,
      "learning_rate": 1.962764969112327e-05,
      "loss": 0.09509849548339844,
      "step": 3885
    },
    {
      "epoch": 0.5414895840590818,
      "grad_norm": 0.6304894089698792,
      "learning_rate": 1.9618224232011757e-05,
      "loss": 0.12030029296875,
      "step": 3886
    },
    {
      "epoch": 0.5416289277502961,
      "grad_norm": 0.31857267022132874,
      "learning_rate": 1.96087988577221e-05,
      "loss": 0.0790557861328125,
      "step": 3887
    },
    {
      "epoch": 0.5417682714415105,
      "grad_norm": 0.4645424485206604,
      "learning_rate": 1.9599373570348416e-05,
      "loss": 0.0848236083984375,
      "step": 3888
    },
    {
      "epoch": 0.5419076151327249,
      "grad_norm": 0.3319401144981384,
      "learning_rate": 1.9589948371984766e-05,
      "loss": 0.086517333984375,
      "step": 3889
    },
    {
      "epoch": 0.5420469588239393,
      "grad_norm": 0.6384466886520386,
      "learning_rate": 1.958052326472523e-05,
      "loss": 0.10201644897460938,
      "step": 3890
    },
    {
      "epoch": 0.5421863025151537,
      "grad_norm": 0.8067552447319031,
      "learning_rate": 1.957109825066385e-05,
      "loss": 0.09109306335449219,
      "step": 3891
    },
    {
      "epoch": 0.542325646206368,
      "grad_norm": 0.5014331936836243,
      "learning_rate": 1.956167333189464e-05,
      "loss": 0.08418655395507812,
      "step": 3892
    },
    {
      "epoch": 0.5424649898975824,
      "grad_norm": 0.5645598769187927,
      "learning_rate": 1.9552248510511616e-05,
      "loss": 0.10551166534423828,
      "step": 3893
    },
    {
      "epoch": 0.5426043335887968,
      "grad_norm": 0.447004497051239,
      "learning_rate": 1.954282378860875e-05,
      "loss": 0.10762977600097656,
      "step": 3894
    },
    {
      "epoch": 0.5427436772800112,
      "grad_norm": 0.2882435917854309,
      "learning_rate": 1.9533399168279997e-05,
      "loss": 0.085235595703125,
      "step": 3895
    },
    {
      "epoch": 0.5428830209712255,
      "grad_norm": 0.2437857985496521,
      "learning_rate": 1.9523974651619296e-05,
      "loss": 0.07326650619506836,
      "step": 3896
    },
    {
      "epoch": 0.5430223646624399,
      "grad_norm": 0.4714573919773102,
      "learning_rate": 1.951455024072056e-05,
      "loss": 0.10591697692871094,
      "step": 3897
    },
    {
      "epoch": 0.5431617083536543,
      "grad_norm": 0.4197983741760254,
      "learning_rate": 1.950512593767768e-05,
      "loss": 0.1025247573852539,
      "step": 3898
    },
    {
      "epoch": 0.5433010520448687,
      "grad_norm": 0.3259736895561218,
      "learning_rate": 1.9495701744584522e-05,
      "loss": 0.06965160369873047,
      "step": 3899
    },
    {
      "epoch": 0.543440395736083,
      "grad_norm": 0.5673220753669739,
      "learning_rate": 1.9486277663534915e-05,
      "loss": 0.11825752258300781,
      "step": 3900
    },
    {
      "epoch": 0.5435797394272974,
      "grad_norm": 0.3076384961605072,
      "learning_rate": 1.9476853696622686e-05,
      "loss": 0.08717727661132812,
      "step": 3901
    },
    {
      "epoch": 0.5437190831185118,
      "grad_norm": 0.3325883746147156,
      "learning_rate": 1.9467429845941622e-05,
      "loss": 0.0806121826171875,
      "step": 3902
    },
    {
      "epoch": 0.5438584268097262,
      "grad_norm": 0.3728259205818176,
      "learning_rate": 1.9458006113585484e-05,
      "loss": 0.08054351806640625,
      "step": 3903
    },
    {
      "epoch": 0.5439977705009406,
      "grad_norm": 0.32385241985321045,
      "learning_rate": 1.9448582501648025e-05,
      "loss": 0.06166362762451172,
      "step": 3904
    },
    {
      "epoch": 0.5441371141921549,
      "grad_norm": 0.35998374223709106,
      "learning_rate": 1.9439159012222936e-05,
      "loss": 0.07708930969238281,
      "step": 3905
    },
    {
      "epoch": 0.5442764578833693,
      "grad_norm": 0.42142197489738464,
      "learning_rate": 1.9429735647403908e-05,
      "loss": 0.096710205078125,
      "step": 3906
    },
    {
      "epoch": 0.5444158015745837,
      "grad_norm": 0.7051739692687988,
      "learning_rate": 1.9420312409284606e-05,
      "loss": 0.14265060424804688,
      "step": 3907
    },
    {
      "epoch": 0.5445551452657981,
      "grad_norm": 0.5354011058807373,
      "learning_rate": 1.9410889299958643e-05,
      "loss": 0.11473846435546875,
      "step": 3908
    },
    {
      "epoch": 0.5446944889570124,
      "grad_norm": 0.6049163937568665,
      "learning_rate": 1.940146632151964e-05,
      "loss": 0.09719657897949219,
      "step": 3909
    },
    {
      "epoch": 0.5448338326482268,
      "grad_norm": 0.5465180277824402,
      "learning_rate": 1.939204347606115e-05,
      "loss": 0.09403038024902344,
      "step": 3910
    },
    {
      "epoch": 0.5449731763394412,
      "grad_norm": 0.3624514937400818,
      "learning_rate": 1.938262076567672e-05,
      "loss": 0.0856170654296875,
      "step": 3911
    },
    {
      "epoch": 0.5451125200306556,
      "grad_norm": 0.5709749460220337,
      "learning_rate": 1.9373198192459856e-05,
      "loss": 0.09744834899902344,
      "step": 3912
    },
    {
      "epoch": 0.54525186372187,
      "grad_norm": 0.5682340860366821,
      "learning_rate": 1.936377575850405e-05,
      "loss": 0.08945083618164062,
      "step": 3913
    },
    {
      "epoch": 0.5453912074130843,
      "grad_norm": 0.6351409554481506,
      "learning_rate": 1.935435346590274e-05,
      "loss": 0.10314083099365234,
      "step": 3914
    },
    {
      "epoch": 0.5455305511042987,
      "grad_norm": 0.4473152160644531,
      "learning_rate": 1.934493131674936e-05,
      "loss": 0.06962394714355469,
      "step": 3915
    },
    {
      "epoch": 0.5456698947955131,
      "grad_norm": 0.49571606516838074,
      "learning_rate": 1.9335509313137275e-05,
      "loss": 0.10107421875,
      "step": 3916
    },
    {
      "epoch": 0.5458092384867275,
      "grad_norm": 0.7396365404129028,
      "learning_rate": 1.9326087457159856e-05,
      "loss": 0.1080322265625,
      "step": 3917
    },
    {
      "epoch": 0.5459485821779418,
      "grad_norm": 0.47407665848731995,
      "learning_rate": 1.9316665750910414e-05,
      "loss": 0.10187339782714844,
      "step": 3918
    },
    {
      "epoch": 0.5460879258691562,
      "grad_norm": 0.5741749405860901,
      "learning_rate": 1.930724419648224e-05,
      "loss": 0.11918449401855469,
      "step": 3919
    },
    {
      "epoch": 0.5462272695603707,
      "grad_norm": 0.5995572209358215,
      "learning_rate": 1.92978227959686e-05,
      "loss": 0.10688400268554688,
      "step": 3920
    },
    {
      "epoch": 0.5463666132515851,
      "grad_norm": 0.3789908289909363,
      "learning_rate": 1.9288401551462694e-05,
      "loss": 0.08160686492919922,
      "step": 3921
    },
    {
      "epoch": 0.5465059569427995,
      "grad_norm": 0.3713533878326416,
      "learning_rate": 1.9278980465057722e-05,
      "loss": 0.0953073501586914,
      "step": 3922
    },
    {
      "epoch": 0.5466453006340138,
      "grad_norm": 0.3042924702167511,
      "learning_rate": 1.9269559538846823e-05,
      "loss": 0.07583045959472656,
      "step": 3923
    },
    {
      "epoch": 0.5467846443252282,
      "grad_norm": 0.5760781168937683,
      "learning_rate": 1.9260138774923124e-05,
      "loss": 0.11906814575195312,
      "step": 3924
    },
    {
      "epoch": 0.5469239880164426,
      "grad_norm": 0.9579076170921326,
      "learning_rate": 1.9250718175379697e-05,
      "loss": 0.10641288757324219,
      "step": 3925
    },
    {
      "epoch": 0.547063331707657,
      "grad_norm": 0.449503093957901,
      "learning_rate": 1.924129774230959e-05,
      "loss": 0.0883941650390625,
      "step": 3926
    },
    {
      "epoch": 0.5472026753988714,
      "grad_norm": 0.3463016748428345,
      "learning_rate": 1.9231877477805795e-05,
      "loss": 0.0735321044921875,
      "step": 3927
    },
    {
      "epoch": 0.5473420190900857,
      "grad_norm": 0.6760953664779663,
      "learning_rate": 1.922245738396129e-05,
      "loss": 0.14137649536132812,
      "step": 3928
    },
    {
      "epoch": 0.5474813627813001,
      "grad_norm": 0.3450724184513092,
      "learning_rate": 1.9213037462869003e-05,
      "loss": 0.09736061096191406,
      "step": 3929
    },
    {
      "epoch": 0.5476207064725145,
      "grad_norm": 0.19251063466072083,
      "learning_rate": 1.920361771662183e-05,
      "loss": 0.05637550354003906,
      "step": 3930
    },
    {
      "epoch": 0.5477600501637289,
      "grad_norm": 0.25633877515792847,
      "learning_rate": 1.9194198147312614e-05,
      "loss": 0.0731649398803711,
      "step": 3931
    },
    {
      "epoch": 0.5478993938549432,
      "grad_norm": 1.1468250751495361,
      "learning_rate": 1.9184778757034168e-05,
      "loss": 0.11553764343261719,
      "step": 3932
    },
    {
      "epoch": 0.5480387375461576,
      "grad_norm": 0.416185587644577,
      "learning_rate": 1.9175359547879275e-05,
      "loss": 0.0832529067993164,
      "step": 3933
    },
    {
      "epoch": 0.548178081237372,
      "grad_norm": 0.569829523563385,
      "learning_rate": 1.9165940521940667e-05,
      "loss": 0.1107940673828125,
      "step": 3934
    },
    {
      "epoch": 0.5483174249285864,
      "grad_norm": 0.4175695478916168,
      "learning_rate": 1.9156521681311025e-05,
      "loss": 0.08311843872070312,
      "step": 3935
    },
    {
      "epoch": 0.5484567686198007,
      "grad_norm": 0.41599366068840027,
      "learning_rate": 1.914710302808302e-05,
      "loss": 0.08492660522460938,
      "step": 3936
    },
    {
      "epoch": 0.5485961123110151,
      "grad_norm": 0.46967798471450806,
      "learning_rate": 1.9137684564349244e-05,
      "loss": 0.11506271362304688,
      "step": 3937
    },
    {
      "epoch": 0.5487354560022295,
      "grad_norm": 0.3801932632923126,
      "learning_rate": 1.912826629220227e-05,
      "loss": 0.07425689697265625,
      "step": 3938
    },
    {
      "epoch": 0.5488747996934439,
      "grad_norm": 0.23948411643505096,
      "learning_rate": 1.911884821373462e-05,
      "loss": 0.06833267211914062,
      "step": 3939
    },
    {
      "epoch": 0.5490141433846583,
      "grad_norm": 0.34247496724128723,
      "learning_rate": 1.9109430331038784e-05,
      "loss": 0.0769052505493164,
      "step": 3940
    },
    {
      "epoch": 0.5491534870758726,
      "grad_norm": 0.2993677854537964,
      "learning_rate": 1.91000126462072e-05,
      "loss": 0.08601188659667969,
      "step": 3941
    },
    {
      "epoch": 0.549292830767087,
      "grad_norm": 0.42230039834976196,
      "learning_rate": 1.909059516133226e-05,
      "loss": 0.08914375305175781,
      "step": 3942
    },
    {
      "epoch": 0.5494321744583014,
      "grad_norm": 0.3353284001350403,
      "learning_rate": 1.9081177878506306e-05,
      "loss": 0.07427597045898438,
      "step": 3943
    },
    {
      "epoch": 0.5495715181495158,
      "grad_norm": 0.47984886169433594,
      "learning_rate": 1.907176079982165e-05,
      "loss": 0.12317848205566406,
      "step": 3944
    },
    {
      "epoch": 0.5497108618407301,
      "grad_norm": 0.307580828666687,
      "learning_rate": 1.9062343927370556e-05,
      "loss": 0.07515335083007812,
      "step": 3945
    },
    {
      "epoch": 0.5498502055319445,
      "grad_norm": 0.2059253454208374,
      "learning_rate": 1.905292726324524e-05,
      "loss": 0.05857563018798828,
      "step": 3946
    },
    {
      "epoch": 0.5499895492231589,
      "grad_norm": 0.25506484508514404,
      "learning_rate": 1.9043510809537857e-05,
      "loss": 0.0672006607055664,
      "step": 3947
    },
    {
      "epoch": 0.5501288929143733,
      "grad_norm": 0.42814165353775024,
      "learning_rate": 1.9034094568340532e-05,
      "loss": 0.08703994750976562,
      "step": 3948
    },
    {
      "epoch": 0.5502682366055877,
      "grad_norm": 0.4826018512248993,
      "learning_rate": 1.9024678541745343e-05,
      "loss": 0.09036445617675781,
      "step": 3949
    },
    {
      "epoch": 0.550407580296802,
      "grad_norm": 0.2568722069263458,
      "learning_rate": 1.901526273184432e-05,
      "loss": 0.07078981399536133,
      "step": 3950
    },
    {
      "epoch": 0.5505469239880164,
      "grad_norm": 0.5473968386650085,
      "learning_rate": 1.900584714072943e-05,
      "loss": 0.10625076293945312,
      "step": 3951
    },
    {
      "epoch": 0.5506862676792308,
      "grad_norm": 0.6474319696426392,
      "learning_rate": 1.8996431770492622e-05,
      "loss": 0.10700035095214844,
      "step": 3952
    },
    {
      "epoch": 0.5508256113704452,
      "grad_norm": 0.5131282806396484,
      "learning_rate": 1.8987016623225748e-05,
      "loss": 0.0976104736328125,
      "step": 3953
    },
    {
      "epoch": 0.5509649550616595,
      "grad_norm": 0.5160492658615112,
      "learning_rate": 1.897760170102066e-05,
      "loss": 0.11835670471191406,
      "step": 3954
    },
    {
      "epoch": 0.5511042987528739,
      "grad_norm": 0.26706376671791077,
      "learning_rate": 1.8968187005969126e-05,
      "loss": 0.06664276123046875,
      "step": 3955
    },
    {
      "epoch": 0.5512436424440883,
      "grad_norm": 0.30322617292404175,
      "learning_rate": 1.8958772540162887e-05,
      "loss": 0.07867145538330078,
      "step": 3956
    },
    {
      "epoch": 0.5513829861353027,
      "grad_norm": 0.2494422048330307,
      "learning_rate": 1.8949358305693625e-05,
      "loss": 0.05819129943847656,
      "step": 3957
    },
    {
      "epoch": 0.551522329826517,
      "grad_norm": 0.671815812587738,
      "learning_rate": 1.8939944304652952e-05,
      "loss": 0.1355133056640625,
      "step": 3958
    },
    {
      "epoch": 0.5516616735177314,
      "grad_norm": 0.44608765840530396,
      "learning_rate": 1.8930530539132456e-05,
      "loss": 0.09113693237304688,
      "step": 3959
    },
    {
      "epoch": 0.5518010172089458,
      "grad_norm": 0.5076836347579956,
      "learning_rate": 1.8921117011223655e-05,
      "loss": 0.11048030853271484,
      "step": 3960
    },
    {
      "epoch": 0.5519403609001603,
      "grad_norm": 0.403266966342926,
      "learning_rate": 1.8911703723018025e-05,
      "loss": 0.10024070739746094,
      "step": 3961
    },
    {
      "epoch": 0.5520797045913747,
      "grad_norm": 0.40321964025497437,
      "learning_rate": 1.8902290676606987e-05,
      "loss": 0.09119606018066406,
      "step": 3962
    },
    {
      "epoch": 0.552219048282589,
      "grad_norm": 0.40207070112228394,
      "learning_rate": 1.8892877874081895e-05,
      "loss": 0.11675262451171875,
      "step": 3963
    },
    {
      "epoch": 0.5523583919738034,
      "grad_norm": 0.3285926878452301,
      "learning_rate": 1.8883465317534055e-05,
      "loss": 0.08487701416015625,
      "step": 3964
    },
    {
      "epoch": 0.5524977356650178,
      "grad_norm": 0.503216028213501,
      "learning_rate": 1.887405300905474e-05,
      "loss": 0.07838821411132812,
      "step": 3965
    },
    {
      "epoch": 0.5526370793562322,
      "grad_norm": 0.4430154263973236,
      "learning_rate": 1.886464095073514e-05,
      "loss": 0.08891105651855469,
      "step": 3966
    },
    {
      "epoch": 0.5527764230474466,
      "grad_norm": 0.35276368260383606,
      "learning_rate": 1.88552291446664e-05,
      "loss": 0.07384872436523438,
      "step": 3967
    },
    {
      "epoch": 0.5529157667386609,
      "grad_norm": 0.3404218554496765,
      "learning_rate": 1.884581759293961e-05,
      "loss": 0.08827590942382812,
      "step": 3968
    },
    {
      "epoch": 0.5530551104298753,
      "grad_norm": 0.4768119156360626,
      "learning_rate": 1.883640629764579e-05,
      "loss": 0.08662796020507812,
      "step": 3969
    },
    {
      "epoch": 0.5531944541210897,
      "grad_norm": 0.38938379287719727,
      "learning_rate": 1.8826995260875937e-05,
      "loss": 0.07959604263305664,
      "step": 3970
    },
    {
      "epoch": 0.5533337978123041,
      "grad_norm": 0.4274638593196869,
      "learning_rate": 1.8817584484720947e-05,
      "loss": 0.10473442077636719,
      "step": 3971
    },
    {
      "epoch": 0.5534731415035185,
      "grad_norm": 0.2948648929595947,
      "learning_rate": 1.8808173971271695e-05,
      "loss": 0.07672786712646484,
      "step": 3972
    },
    {
      "epoch": 0.5536124851947328,
      "grad_norm": 0.35727739334106445,
      "learning_rate": 1.8798763722618982e-05,
      "loss": 0.06604671478271484,
      "step": 3973
    },
    {
      "epoch": 0.5537518288859472,
      "grad_norm": 0.4010270833969116,
      "learning_rate": 1.878935374085354e-05,
      "loss": 0.0767364501953125,
      "step": 3974
    },
    {
      "epoch": 0.5538911725771616,
      "grad_norm": 0.5224695801734924,
      "learning_rate": 1.8779944028066057e-05,
      "loss": 0.10113525390625,
      "step": 3975
    },
    {
      "epoch": 0.554030516268376,
      "grad_norm": 0.6470497846603394,
      "learning_rate": 1.8770534586347152e-05,
      "loss": 0.11547183990478516,
      "step": 3976
    },
    {
      "epoch": 0.5541698599595903,
      "grad_norm": 0.45190829038619995,
      "learning_rate": 1.8761125417787398e-05,
      "loss": 0.09692192077636719,
      "step": 3977
    },
    {
      "epoch": 0.5543092036508047,
      "grad_norm": 0.6564455032348633,
      "learning_rate": 1.8751716524477298e-05,
      "loss": 0.13176918029785156,
      "step": 3978
    },
    {
      "epoch": 0.5544485473420191,
      "grad_norm": 0.7814648151397705,
      "learning_rate": 1.874230790850728e-05,
      "loss": 0.12874603271484375,
      "step": 3979
    },
    {
      "epoch": 0.5545878910332335,
      "grad_norm": 0.2591119706630707,
      "learning_rate": 1.8732899571967728e-05,
      "loss": 0.07242202758789062,
      "step": 3980
    },
    {
      "epoch": 0.5547272347244478,
      "grad_norm": 0.5433376431465149,
      "learning_rate": 1.8723491516948968e-05,
      "loss": 0.10040903091430664,
      "step": 3981
    },
    {
      "epoch": 0.5548665784156622,
      "grad_norm": 0.5682857632637024,
      "learning_rate": 1.871408374554125e-05,
      "loss": 0.10281181335449219,
      "step": 3982
    },
    {
      "epoch": 0.5550059221068766,
      "grad_norm": 0.2822878658771515,
      "learning_rate": 1.8704676259834768e-05,
      "loss": 0.0699472427368164,
      "step": 3983
    },
    {
      "epoch": 0.555145265798091,
      "grad_norm": 0.25710529088974,
      "learning_rate": 1.8695269061919642e-05,
      "loss": 0.06441307067871094,
      "step": 3984
    },
    {
      "epoch": 0.5552846094893054,
      "grad_norm": 0.40189847350120544,
      "learning_rate": 1.8685862153885947e-05,
      "loss": 0.09173011779785156,
      "step": 3985
    },
    {
      "epoch": 0.5554239531805197,
      "grad_norm": 0.2719210982322693,
      "learning_rate": 1.867645553782368e-05,
      "loss": 0.06520462036132812,
      "step": 3986
    },
    {
      "epoch": 0.5555632968717341,
      "grad_norm": 0.23362241685390472,
      "learning_rate": 1.866704921582277e-05,
      "loss": 0.062259674072265625,
      "step": 3987
    },
    {
      "epoch": 0.5557026405629485,
      "grad_norm": 0.2743144631385803,
      "learning_rate": 1.86576431899731e-05,
      "loss": 0.08186721801757812,
      "step": 3988
    },
    {
      "epoch": 0.5558419842541629,
      "grad_norm": 0.5928040146827698,
      "learning_rate": 1.864823746236446e-05,
      "loss": 0.12438011169433594,
      "step": 3989
    },
    {
      "epoch": 0.5559813279453772,
      "grad_norm": 0.525719165802002,
      "learning_rate": 1.8638832035086598e-05,
      "loss": 0.10658931732177734,
      "step": 3990
    },
    {
      "epoch": 0.5561206716365916,
      "grad_norm": 0.41032901406288147,
      "learning_rate": 1.862942691022918e-05,
      "loss": 0.08048820495605469,
      "step": 3991
    },
    {
      "epoch": 0.556260015327806,
      "grad_norm": 0.6040444374084473,
      "learning_rate": 1.8620022089881812e-05,
      "loss": 0.09832763671875,
      "step": 3992
    },
    {
      "epoch": 0.5563993590190204,
      "grad_norm": 0.2767842710018158,
      "learning_rate": 1.861061757613403e-05,
      "loss": 0.06828975677490234,
      "step": 3993
    },
    {
      "epoch": 0.5565387027102348,
      "grad_norm": 0.3068370819091797,
      "learning_rate": 1.8601213371075308e-05,
      "loss": 0.08157730102539062,
      "step": 3994
    },
    {
      "epoch": 0.5566780464014491,
      "grad_norm": 0.2696853280067444,
      "learning_rate": 1.8591809476795034e-05,
      "loss": 0.08797454833984375,
      "step": 3995
    },
    {
      "epoch": 0.5568173900926635,
      "grad_norm": 0.41997241973876953,
      "learning_rate": 1.8582405895382544e-05,
      "loss": 0.10389137268066406,
      "step": 3996
    },
    {
      "epoch": 0.5569567337838779,
      "grad_norm": 0.3913429081439972,
      "learning_rate": 1.8573002628927102e-05,
      "loss": 0.10517311096191406,
      "step": 3997
    },
    {
      "epoch": 0.5570960774750923,
      "grad_norm": 0.6394292116165161,
      "learning_rate": 1.8563599679517898e-05,
      "loss": 0.10970687866210938,
      "step": 3998
    },
    {
      "epoch": 0.5572354211663066,
      "grad_norm": 0.4419870674610138,
      "learning_rate": 1.8554197049244054e-05,
      "loss": 0.10516548156738281,
      "step": 3999
    },
    {
      "epoch": 0.557374764857521,
      "grad_norm": 0.22656752169132233,
      "learning_rate": 1.854479474019461e-05,
      "loss": 0.06470775604248047,
      "step": 4000
    },
    {
      "epoch": 0.5575141085487355,
      "grad_norm": 0.3976730704307556,
      "learning_rate": 1.8535392754458555e-05,
      "loss": 0.10113143920898438,
      "step": 4001
    },
    {
      "epoch": 0.5576534522399499,
      "grad_norm": 0.3042682409286499,
      "learning_rate": 1.8525991094124795e-05,
      "loss": 0.07716178894042969,
      "step": 4002
    },
    {
      "epoch": 0.5577927959311643,
      "grad_norm": 0.503616213798523,
      "learning_rate": 1.8516589761282155e-05,
      "loss": 0.10811328887939453,
      "step": 4003
    },
    {
      "epoch": 0.5579321396223786,
      "grad_norm": 0.3720170259475708,
      "learning_rate": 1.850718875801942e-05,
      "loss": 0.08322334289550781,
      "step": 4004
    },
    {
      "epoch": 0.558071483313593,
      "grad_norm": 0.26879093050956726,
      "learning_rate": 1.8497788086425243e-05,
      "loss": 0.06410980224609375,
      "step": 4005
    },
    {
      "epoch": 0.5582108270048074,
      "grad_norm": 0.5390329360961914,
      "learning_rate": 1.8488387748588266e-05,
      "loss": 0.10888481140136719,
      "step": 4006
    },
    {
      "epoch": 0.5583501706960218,
      "grad_norm": 0.27489978075027466,
      "learning_rate": 1.8478987746597017e-05,
      "loss": 0.06244993209838867,
      "step": 4007
    },
    {
      "epoch": 0.5584895143872362,
      "grad_norm": 0.5846813321113586,
      "learning_rate": 1.8469588082539963e-05,
      "loss": 0.10392570495605469,
      "step": 4008
    },
    {
      "epoch": 0.5586288580784505,
      "grad_norm": 0.4215579330921173,
      "learning_rate": 1.8460188758505502e-05,
      "loss": 0.08225822448730469,
      "step": 4009
    },
    {
      "epoch": 0.5587682017696649,
      "grad_norm": 0.3585793673992157,
      "learning_rate": 1.8450789776581947e-05,
      "loss": 0.0606536865234375,
      "step": 4010
    },
    {
      "epoch": 0.5589075454608793,
      "grad_norm": 0.3971717655658722,
      "learning_rate": 1.844139113885753e-05,
      "loss": 0.07358407974243164,
      "step": 4011
    },
    {
      "epoch": 0.5590468891520937,
      "grad_norm": 0.43331003189086914,
      "learning_rate": 1.8431992847420418e-05,
      "loss": 0.08680152893066406,
      "step": 4012
    },
    {
      "epoch": 0.559186232843308,
      "grad_norm": 0.5721511244773865,
      "learning_rate": 1.8422594904358696e-05,
      "loss": 0.11266708374023438,
      "step": 4013
    },
    {
      "epoch": 0.5593255765345224,
      "grad_norm": 0.3700628876686096,
      "learning_rate": 1.8413197311760377e-05,
      "loss": 0.08453559875488281,
      "step": 4014
    },
    {
      "epoch": 0.5594649202257368,
      "grad_norm": 0.4493331015110016,
      "learning_rate": 1.8403800071713392e-05,
      "loss": 0.09166145324707031,
      "step": 4015
    },
    {
      "epoch": 0.5596042639169512,
      "grad_norm": 0.3408151865005493,
      "learning_rate": 1.839440318630558e-05,
      "loss": 0.08568382263183594,
      "step": 4016
    },
    {
      "epoch": 0.5597436076081655,
      "grad_norm": 0.5381366610527039,
      "learning_rate": 1.838500665762473e-05,
      "loss": 0.1011343002319336,
      "step": 4017
    },
    {
      "epoch": 0.5598829512993799,
      "grad_norm": 0.4305996894836426,
      "learning_rate": 1.8375610487758527e-05,
      "loss": 0.1070709228515625,
      "step": 4018
    },
    {
      "epoch": 0.5600222949905943,
      "grad_norm": 0.4391328990459442,
      "learning_rate": 1.8366214678794584e-05,
      "loss": 0.109039306640625,
      "step": 4019
    },
    {
      "epoch": 0.5601616386818087,
      "grad_norm": 0.5724929571151733,
      "learning_rate": 1.8356819232820452e-05,
      "loss": 0.10242938995361328,
      "step": 4020
    },
    {
      "epoch": 0.5603009823730231,
      "grad_norm": 0.6282013654708862,
      "learning_rate": 1.834742415192356e-05,
      "loss": 0.13050460815429688,
      "step": 4021
    },
    {
      "epoch": 0.5604403260642374,
      "grad_norm": 0.2273886352777481,
      "learning_rate": 1.8338029438191298e-05,
      "loss": 0.05887031555175781,
      "step": 4022
    },
    {
      "epoch": 0.5605796697554518,
      "grad_norm": 0.3999176621437073,
      "learning_rate": 1.832863509371095e-05,
      "loss": 0.09090042114257812,
      "step": 4023
    },
    {
      "epoch": 0.5607190134466662,
      "grad_norm": 0.22331783175468445,
      "learning_rate": 1.831924112056972e-05,
      "loss": 0.07464981079101562,
      "step": 4024
    },
    {
      "epoch": 0.5608583571378806,
      "grad_norm": 0.6754215359687805,
      "learning_rate": 1.8309847520854753e-05,
      "loss": 0.1148386001586914,
      "step": 4025
    },
    {
      "epoch": 0.560997700829095,
      "grad_norm": 0.6209039688110352,
      "learning_rate": 1.8300454296653076e-05,
      "loss": 0.14847946166992188,
      "step": 4026
    },
    {
      "epoch": 0.5611370445203093,
      "grad_norm": 0.33953696489334106,
      "learning_rate": 1.829106145005165e-05,
      "loss": 0.09502220153808594,
      "step": 4027
    },
    {
      "epoch": 0.5612763882115237,
      "grad_norm": 0.3697056472301483,
      "learning_rate": 1.828166898313735e-05,
      "loss": 0.08188819885253906,
      "step": 4028
    },
    {
      "epoch": 0.5614157319027381,
      "grad_norm": 0.49544212222099304,
      "learning_rate": 1.8272276897996977e-05,
      "loss": 0.09785842895507812,
      "step": 4029
    },
    {
      "epoch": 0.5615550755939525,
      "grad_norm": 0.27257394790649414,
      "learning_rate": 1.8262885196717232e-05,
      "loss": 0.07114028930664062,
      "step": 4030
    },
    {
      "epoch": 0.5616944192851668,
      "grad_norm": 0.44563552737236023,
      "learning_rate": 1.8253493881384743e-05,
      "loss": 0.076019287109375,
      "step": 4031
    },
    {
      "epoch": 0.5618337629763812,
      "grad_norm": 0.5177099108695984,
      "learning_rate": 1.8244102954086032e-05,
      "loss": 0.10845184326171875,
      "step": 4032
    },
    {
      "epoch": 0.5619731066675956,
      "grad_norm": 0.7175803184509277,
      "learning_rate": 1.823471241690756e-05,
      "loss": 0.12463569641113281,
      "step": 4033
    },
    {
      "epoch": 0.56211245035881,
      "grad_norm": 0.5181860327720642,
      "learning_rate": 1.8225322271935686e-05,
      "loss": 0.09370231628417969,
      "step": 4034
    },
    {
      "epoch": 0.5622517940500243,
      "grad_norm": 1.0214121341705322,
      "learning_rate": 1.8215932521256683e-05,
      "loss": 0.12355899810791016,
      "step": 4035
    },
    {
      "epoch": 0.5623911377412387,
      "grad_norm": 0.5820346474647522,
      "learning_rate": 1.8206543166956754e-05,
      "loss": 0.10139846801757812,
      "step": 4036
    },
    {
      "epoch": 0.5625304814324531,
      "grad_norm": 0.19709883630275726,
      "learning_rate": 1.8197154211121976e-05,
      "loss": 0.06758785247802734,
      "step": 4037
    },
    {
      "epoch": 0.5626698251236675,
      "grad_norm": 1.3540277481079102,
      "learning_rate": 1.818776565583838e-05,
      "loss": 0.14343833923339844,
      "step": 4038
    },
    {
      "epoch": 0.5628091688148819,
      "grad_norm": 0.7221106290817261,
      "learning_rate": 1.8178377503191875e-05,
      "loss": 0.11934661865234375,
      "step": 4039
    },
    {
      "epoch": 0.5629485125060962,
      "grad_norm": 0.3034663200378418,
      "learning_rate": 1.8168989755268303e-05,
      "loss": 0.08368492126464844,
      "step": 4040
    },
    {
      "epoch": 0.5630878561973107,
      "grad_norm": 0.32387328147888184,
      "learning_rate": 1.815960241415341e-05,
      "loss": 0.07765388488769531,
      "step": 4041
    },
    {
      "epoch": 0.5632271998885251,
      "grad_norm": 0.2618301808834076,
      "learning_rate": 1.815021548193284e-05,
      "loss": 0.07066631317138672,
      "step": 4042
    },
    {
      "epoch": 0.5633665435797395,
      "grad_norm": 0.4455222189426422,
      "learning_rate": 1.814082896069216e-05,
      "loss": 0.09250068664550781,
      "step": 4043
    },
    {
      "epoch": 0.5635058872709539,
      "grad_norm": 0.2420588731765747,
      "learning_rate": 1.813144285251683e-05,
      "loss": 0.0748443603515625,
      "step": 4044
    },
    {
      "epoch": 0.5636452309621682,
      "grad_norm": 0.2739112973213196,
      "learning_rate": 1.8122057159492248e-05,
      "loss": 0.06808853149414062,
      "step": 4045
    },
    {
      "epoch": 0.5637845746533826,
      "grad_norm": 0.3326997458934784,
      "learning_rate": 1.8112671883703688e-05,
      "loss": 0.08143806457519531,
      "step": 4046
    },
    {
      "epoch": 0.563923918344597,
      "grad_norm": 0.3509039878845215,
      "learning_rate": 1.8103287027236352e-05,
      "loss": 0.09311485290527344,
      "step": 4047
    },
    {
      "epoch": 0.5640632620358114,
      "grad_norm": 0.5972915291786194,
      "learning_rate": 1.8093902592175328e-05,
      "loss": 0.10935211181640625,
      "step": 4048
    },
    {
      "epoch": 0.5642026057270257,
      "grad_norm": 0.2382107973098755,
      "learning_rate": 1.8084518580605634e-05,
      "loss": 0.05924034118652344,
      "step": 4049
    },
    {
      "epoch": 0.5643419494182401,
      "grad_norm": 0.3819197118282318,
      "learning_rate": 1.807513499461218e-05,
      "loss": 0.09655189514160156,
      "step": 4050
    },
    {
      "epoch": 0.5644812931094545,
      "grad_norm": 0.3384595513343811,
      "learning_rate": 1.8065751836279784e-05,
      "loss": 0.08532905578613281,
      "step": 4051
    },
    {
      "epoch": 0.5646206368006689,
      "grad_norm": 0.5121312141418457,
      "learning_rate": 1.805636910769318e-05,
      "loss": 0.10368061065673828,
      "step": 4052
    },
    {
      "epoch": 0.5647599804918833,
      "grad_norm": 0.6872484683990479,
      "learning_rate": 1.8046986810936974e-05,
      "loss": 0.13508224487304688,
      "step": 4053
    },
    {
      "epoch": 0.5648993241830976,
      "grad_norm": 0.5166663527488708,
      "learning_rate": 1.8037604948095714e-05,
      "loss": 0.1068572998046875,
      "step": 4054
    },
    {
      "epoch": 0.565038667874312,
      "grad_norm": 0.3752843737602234,
      "learning_rate": 1.802822352125383e-05,
      "loss": 0.092254638671875,
      "step": 4055
    },
    {
      "epoch": 0.5651780115655264,
      "grad_norm": 0.2951204478740692,
      "learning_rate": 1.8018842532495667e-05,
      "loss": 0.09109878540039062,
      "step": 4056
    },
    {
      "epoch": 0.5653173552567408,
      "grad_norm": 0.4980812966823578,
      "learning_rate": 1.8009461983905466e-05,
      "loss": 0.11092185974121094,
      "step": 4057
    },
    {
      "epoch": 0.5654566989479551,
      "grad_norm": 0.3126406967639923,
      "learning_rate": 1.8000081877567362e-05,
      "loss": 0.0808248519897461,
      "step": 4058
    },
    {
      "epoch": 0.5655960426391695,
      "grad_norm": 0.45106491446495056,
      "learning_rate": 1.799070221556541e-05,
      "loss": 0.09257698059082031,
      "step": 4059
    },
    {
      "epoch": 0.5657353863303839,
      "grad_norm": 0.5763440728187561,
      "learning_rate": 1.7981322999983547e-05,
      "loss": 0.09988784790039062,
      "step": 4060
    },
    {
      "epoch": 0.5658747300215983,
      "grad_norm": 0.4370148777961731,
      "learning_rate": 1.7971944232905627e-05,
      "loss": 0.0983123779296875,
      "step": 4061
    },
    {
      "epoch": 0.5660140737128126,
      "grad_norm": 0.5264624357223511,
      "learning_rate": 1.7962565916415406e-05,
      "loss": 0.11218643188476562,
      "step": 4062
    },
    {
      "epoch": 0.566153417404027,
      "grad_norm": 0.5439543128013611,
      "learning_rate": 1.7953188052596514e-05,
      "loss": 0.10368537902832031,
      "step": 4063
    },
    {
      "epoch": 0.5662927610952414,
      "grad_norm": 0.4733785092830658,
      "learning_rate": 1.7943810643532506e-05,
      "loss": 0.12571144104003906,
      "step": 4064
    },
    {
      "epoch": 0.5664321047864558,
      "grad_norm": 0.5838122963905334,
      "learning_rate": 1.7934433691306834e-05,
      "loss": 0.10927009582519531,
      "step": 4065
    },
    {
      "epoch": 0.5665714484776702,
      "grad_norm": 0.4894682466983795,
      "learning_rate": 1.7925057198002836e-05,
      "loss": 0.11099052429199219,
      "step": 4066
    },
    {
      "epoch": 0.5667107921688845,
      "grad_norm": 0.6455749273300171,
      "learning_rate": 1.7915681165703754e-05,
      "loss": 0.10688972473144531,
      "step": 4067
    },
    {
      "epoch": 0.5668501358600989,
      "grad_norm": 0.30797144770622253,
      "learning_rate": 1.7906305596492747e-05,
      "loss": 0.08696746826171875,
      "step": 4068
    },
    {
      "epoch": 0.5669894795513133,
      "grad_norm": 0.2736218571662903,
      "learning_rate": 1.7896930492452816e-05,
      "loss": 0.07781410217285156,
      "step": 4069
    },
    {
      "epoch": 0.5671288232425277,
      "grad_norm": 0.6503137946128845,
      "learning_rate": 1.788755585566693e-05,
      "loss": 0.10112190246582031,
      "step": 4070
    },
    {
      "epoch": 0.567268166933742,
      "grad_norm": 0.5114457011222839,
      "learning_rate": 1.7878181688217894e-05,
      "loss": 0.115997314453125,
      "step": 4071
    },
    {
      "epoch": 0.5674075106249564,
      "grad_norm": 0.36321181058883667,
      "learning_rate": 1.7868807992188448e-05,
      "loss": 0.06845283508300781,
      "step": 4072
    },
    {
      "epoch": 0.5675468543161708,
      "grad_norm": 0.5832300186157227,
      "learning_rate": 1.7859434769661218e-05,
      "loss": 0.09729957580566406,
      "step": 4073
    },
    {
      "epoch": 0.5676861980073852,
      "grad_norm": 0.3772968351840973,
      "learning_rate": 1.7850062022718708e-05,
      "loss": 0.08876419067382812,
      "step": 4074
    },
    {
      "epoch": 0.5678255416985996,
      "grad_norm": 0.45914870500564575,
      "learning_rate": 1.7840689753443328e-05,
      "loss": 0.09173965454101562,
      "step": 4075
    },
    {
      "epoch": 0.5679648853898139,
      "grad_norm": 0.678042471408844,
      "learning_rate": 1.7831317963917388e-05,
      "loss": 0.11501121520996094,
      "step": 4076
    },
    {
      "epoch": 0.5681042290810283,
      "grad_norm": 0.597774863243103,
      "learning_rate": 1.7821946656223088e-05,
      "loss": 0.09260940551757812,
      "step": 4077
    },
    {
      "epoch": 0.5682435727722427,
      "grad_norm": 0.3199504017829895,
      "learning_rate": 1.7812575832442518e-05,
      "loss": 0.08275985717773438,
      "step": 4078
    },
    {
      "epoch": 0.5683829164634571,
      "grad_norm": 0.26731351017951965,
      "learning_rate": 1.7803205494657652e-05,
      "loss": 0.06479549407958984,
      "step": 4079
    },
    {
      "epoch": 0.5685222601546714,
      "grad_norm": 0.47408196330070496,
      "learning_rate": 1.7793835644950373e-05,
      "loss": 0.0783834457397461,
      "step": 4080
    },
    {
      "epoch": 0.5686616038458859,
      "grad_norm": 1.0532084703445435,
      "learning_rate": 1.7784466285402445e-05,
      "loss": 0.141387939453125,
      "step": 4081
    },
    {
      "epoch": 0.5688009475371003,
      "grad_norm": 0.24096480011940002,
      "learning_rate": 1.777509741809553e-05,
      "loss": 0.06844520568847656,
      "step": 4082
    },
    {
      "epoch": 0.5689402912283147,
      "grad_norm": 0.2607787251472473,
      "learning_rate": 1.7765729045111177e-05,
      "loss": 0.07101249694824219,
      "step": 4083
    },
    {
      "epoch": 0.5690796349195291,
      "grad_norm": 0.6411575078964233,
      "learning_rate": 1.775636116853081e-05,
      "loss": 0.11623764038085938,
      "step": 4084
    },
    {
      "epoch": 0.5692189786107434,
      "grad_norm": 0.4082685112953186,
      "learning_rate": 1.7746993790435777e-05,
      "loss": 0.08844757080078125,
      "step": 4085
    },
    {
      "epoch": 0.5693583223019578,
      "grad_norm": 0.2897104322910309,
      "learning_rate": 1.773762691290728e-05,
      "loss": 0.07357215881347656,
      "step": 4086
    },
    {
      "epoch": 0.5694976659931722,
      "grad_norm": 0.2369295358657837,
      "learning_rate": 1.7728260538026432e-05,
      "loss": 0.05613899230957031,
      "step": 4087
    },
    {
      "epoch": 0.5696370096843866,
      "grad_norm": 0.6680482625961304,
      "learning_rate": 1.7718894667874235e-05,
      "loss": 0.10429096221923828,
      "step": 4088
    },
    {
      "epoch": 0.569776353375601,
      "grad_norm": 0.6774418354034424,
      "learning_rate": 1.7709529304531567e-05,
      "loss": 0.10048294067382812,
      "step": 4089
    },
    {
      "epoch": 0.5699156970668153,
      "grad_norm": 0.22904351353645325,
      "learning_rate": 1.7700164450079188e-05,
      "loss": 0.0669546127319336,
      "step": 4090
    },
    {
      "epoch": 0.5700550407580297,
      "grad_norm": 1.034339427947998,
      "learning_rate": 1.769080010659776e-05,
      "loss": 0.1301860809326172,
      "step": 4091
    },
    {
      "epoch": 0.5701943844492441,
      "grad_norm": 0.4823663532733917,
      "learning_rate": 1.768143627616783e-05,
      "loss": 0.08158111572265625,
      "step": 4092
    },
    {
      "epoch": 0.5703337281404585,
      "grad_norm": 0.4055899977684021,
      "learning_rate": 1.7672072960869828e-05,
      "loss": 0.09107780456542969,
      "step": 4093
    },
    {
      "epoch": 0.5704730718316728,
      "grad_norm": 0.3230876326560974,
      "learning_rate": 1.766271016278407e-05,
      "loss": 0.08565521240234375,
      "step": 4094
    },
    {
      "epoch": 0.5706124155228872,
      "grad_norm": 0.3029000461101532,
      "learning_rate": 1.7653347883990748e-05,
      "loss": 0.06458663940429688,
      "step": 4095
    },
    {
      "epoch": 0.5707517592141016,
      "grad_norm": 0.3183416426181793,
      "learning_rate": 1.764398612656995e-05,
      "loss": 0.0647268295288086,
      "step": 4096
    },
    {
      "epoch": 0.570891102905316,
      "grad_norm": 0.5179814696311951,
      "learning_rate": 1.7634624892601647e-05,
      "loss": 0.09415960311889648,
      "step": 4097
    },
    {
      "epoch": 0.5710304465965303,
      "grad_norm": 0.8599585294723511,
      "learning_rate": 1.762526418416569e-05,
      "loss": 0.1350078582763672,
      "step": 4098
    },
    {
      "epoch": 0.5711697902877447,
      "grad_norm": 0.30145689845085144,
      "learning_rate": 1.7615904003341822e-05,
      "loss": 0.07440376281738281,
      "step": 4099
    },
    {
      "epoch": 0.5713091339789591,
      "grad_norm": 0.21508269011974335,
      "learning_rate": 1.7606544352209644e-05,
      "loss": 0.06680107116699219,
      "step": 4100
    },
    {
      "epoch": 0.5714484776701735,
      "grad_norm": 0.4374901056289673,
      "learning_rate": 1.7597185232848673e-05,
      "loss": 0.10190391540527344,
      "step": 4101
    },
    {
      "epoch": 0.5715878213613879,
      "grad_norm": 0.2996353507041931,
      "learning_rate": 1.7587826647338285e-05,
      "loss": 0.07889366149902344,
      "step": 4102
    },
    {
      "epoch": 0.5717271650526022,
      "grad_norm": 0.6963887214660645,
      "learning_rate": 1.757846859775774e-05,
      "loss": 0.08677482604980469,
      "step": 4103
    },
    {
      "epoch": 0.5718665087438166,
      "grad_norm": 0.5192792415618896,
      "learning_rate": 1.7569111086186196e-05,
      "loss": 0.0999603271484375,
      "step": 4104
    },
    {
      "epoch": 0.572005852435031,
      "grad_norm": 0.7261146903038025,
      "learning_rate": 1.7559754114702672e-05,
      "loss": 0.10765457153320312,
      "step": 4105
    },
    {
      "epoch": 0.5721451961262454,
      "grad_norm": 0.32212814688682556,
      "learning_rate": 1.755039768538607e-05,
      "loss": 0.08531761169433594,
      "step": 4106
    },
    {
      "epoch": 0.5722845398174597,
      "grad_norm": 0.2693021595478058,
      "learning_rate": 1.7541041800315173e-05,
      "loss": 0.08331871032714844,
      "step": 4107
    },
    {
      "epoch": 0.5724238835086741,
      "grad_norm": 0.6828738451004028,
      "learning_rate": 1.7531686461568648e-05,
      "loss": 0.1011962890625,
      "step": 4108
    },
    {
      "epoch": 0.5725632271998885,
      "grad_norm": 0.5507255792617798,
      "learning_rate": 1.752233167122504e-05,
      "loss": 0.1263275146484375,
      "step": 4109
    },
    {
      "epoch": 0.5727025708911029,
      "grad_norm": 0.3005565106868744,
      "learning_rate": 1.7512977431362777e-05,
      "loss": 0.07114219665527344,
      "step": 4110
    },
    {
      "epoch": 0.5728419145823173,
      "grad_norm": 0.3698022663593292,
      "learning_rate": 1.750362374406014e-05,
      "loss": 0.09227180480957031,
      "step": 4111
    },
    {
      "epoch": 0.5729812582735316,
      "grad_norm": 0.2717233896255493,
      "learning_rate": 1.749427061139531e-05,
      "loss": 0.06452369689941406,
      "step": 4112
    },
    {
      "epoch": 0.573120601964746,
      "grad_norm": 0.37814950942993164,
      "learning_rate": 1.7484918035446352e-05,
      "loss": 0.07384681701660156,
      "step": 4113
    },
    {
      "epoch": 0.5732599456559604,
      "grad_norm": 0.46844303607940674,
      "learning_rate": 1.7475566018291185e-05,
      "loss": 0.10261154174804688,
      "step": 4114
    },
    {
      "epoch": 0.5733992893471748,
      "grad_norm": 0.3850955367088318,
      "learning_rate": 1.7466214562007618e-05,
      "loss": 0.08108711242675781,
      "step": 4115
    },
    {
      "epoch": 0.5735386330383891,
      "grad_norm": 0.7322114109992981,
      "learning_rate": 1.745686366867332e-05,
      "loss": 0.11954116821289062,
      "step": 4116
    },
    {
      "epoch": 0.5736779767296035,
      "grad_norm": 0.36043980717658997,
      "learning_rate": 1.7447513340365855e-05,
      "loss": 0.09004974365234375,
      "step": 4117
    },
    {
      "epoch": 0.5738173204208179,
      "grad_norm": 0.303821861743927,
      "learning_rate": 1.7438163579162658e-05,
      "loss": 0.08235931396484375,
      "step": 4118
    },
    {
      "epoch": 0.5739566641120323,
      "grad_norm": 0.29861196875572205,
      "learning_rate": 1.7428814387141016e-05,
      "loss": 0.07965278625488281,
      "step": 4119
    },
    {
      "epoch": 0.5740960078032467,
      "grad_norm": 0.4718315303325653,
      "learning_rate": 1.741946576637813e-05,
      "loss": 0.09920692443847656,
      "step": 4120
    },
    {
      "epoch": 0.5742353514944611,
      "grad_norm": 0.49304017424583435,
      "learning_rate": 1.7410117718951026e-05,
      "loss": 0.1141204833984375,
      "step": 4121
    },
    {
      "epoch": 0.5743746951856755,
      "grad_norm": 0.45490750670433044,
      "learning_rate": 1.740077024693664e-05,
      "loss": 0.08880043029785156,
      "step": 4122
    },
    {
      "epoch": 0.5745140388768899,
      "grad_norm": 0.4145594537258148,
      "learning_rate": 1.739142335241176e-05,
      "loss": 0.08373641967773438,
      "step": 4123
    },
    {
      "epoch": 0.5746533825681043,
      "grad_norm": 0.26137974858283997,
      "learning_rate": 1.7382077037453057e-05,
      "loss": 0.07437324523925781,
      "step": 4124
    },
    {
      "epoch": 0.5747927262593187,
      "grad_norm": 0.3278530240058899,
      "learning_rate": 1.7372731304137072e-05,
      "loss": 0.07338714599609375,
      "step": 4125
    },
    {
      "epoch": 0.574932069950533,
      "grad_norm": 0.2983459234237671,
      "learning_rate": 1.736338615454021e-05,
      "loss": 0.07404756546020508,
      "step": 4126
    },
    {
      "epoch": 0.5750714136417474,
      "grad_norm": 0.420315682888031,
      "learning_rate": 1.7354041590738747e-05,
      "loss": 0.07807540893554688,
      "step": 4127
    },
    {
      "epoch": 0.5752107573329618,
      "grad_norm": 0.3425889313220978,
      "learning_rate": 1.734469761480883e-05,
      "loss": 0.07760047912597656,
      "step": 4128
    },
    {
      "epoch": 0.5753501010241762,
      "grad_norm": 0.30964651703834534,
      "learning_rate": 1.733535422882649e-05,
      "loss": 0.0778341293334961,
      "step": 4129
    },
    {
      "epoch": 0.5754894447153905,
      "grad_norm": 0.4785287082195282,
      "learning_rate": 1.73260114348676e-05,
      "loss": 0.10133552551269531,
      "step": 4130
    },
    {
      "epoch": 0.5756287884066049,
      "grad_norm": 0.33403992652893066,
      "learning_rate": 1.7316669235007927e-05,
      "loss": 0.08171653747558594,
      "step": 4131
    },
    {
      "epoch": 0.5757681320978193,
      "grad_norm": 0.2903982698917389,
      "learning_rate": 1.7307327631323078e-05,
      "loss": 0.07427024841308594,
      "step": 4132
    },
    {
      "epoch": 0.5759074757890337,
      "grad_norm": 0.6040825843811035,
      "learning_rate": 1.7297986625888563e-05,
      "loss": 0.10210132598876953,
      "step": 4133
    },
    {
      "epoch": 0.576046819480248,
      "grad_norm": 0.28627389669418335,
      "learning_rate": 1.728864622077973e-05,
      "loss": 0.07282638549804688,
      "step": 4134
    },
    {
      "epoch": 0.5761861631714624,
      "grad_norm": 0.28188356757164,
      "learning_rate": 1.72793064180718e-05,
      "loss": 0.07866859436035156,
      "step": 4135
    },
    {
      "epoch": 0.5763255068626768,
      "grad_norm": 0.4858262240886688,
      "learning_rate": 1.7269967219839882e-05,
      "loss": 0.09848785400390625,
      "step": 4136
    },
    {
      "epoch": 0.5764648505538912,
      "grad_norm": 0.33621469140052795,
      "learning_rate": 1.7260628628158907e-05,
      "loss": 0.07585906982421875,
      "step": 4137
    },
    {
      "epoch": 0.5766041942451056,
      "grad_norm": 0.3468513786792755,
      "learning_rate": 1.7251290645103716e-05,
      "loss": 0.0803232192993164,
      "step": 4138
    },
    {
      "epoch": 0.5767435379363199,
      "grad_norm": 0.5121119618415833,
      "learning_rate": 1.7241953272748987e-05,
      "loss": 0.1042633056640625,
      "step": 4139
    },
    {
      "epoch": 0.5768828816275343,
      "grad_norm": 0.3060786724090576,
      "learning_rate": 1.7232616513169272e-05,
      "loss": 0.08213043212890625,
      "step": 4140
    },
    {
      "epoch": 0.5770222253187487,
      "grad_norm": 0.3026684522628784,
      "learning_rate": 1.7223280368438993e-05,
      "loss": 0.08101844787597656,
      "step": 4141
    },
    {
      "epoch": 0.5771615690099631,
      "grad_norm": 0.6800536513328552,
      "learning_rate": 1.7213944840632422e-05,
      "loss": 0.1369953155517578,
      "step": 4142
    },
    {
      "epoch": 0.5773009127011774,
      "grad_norm": 0.6473787426948547,
      "learning_rate": 1.7204609931823702e-05,
      "loss": 0.14793968200683594,
      "step": 4143
    },
    {
      "epoch": 0.5774402563923918,
      "grad_norm": 0.3120273947715759,
      "learning_rate": 1.7195275644086827e-05,
      "loss": 0.08077526092529297,
      "step": 4144
    },
    {
      "epoch": 0.5775796000836062,
      "grad_norm": 0.24476101994514465,
      "learning_rate": 1.7185941979495676e-05,
      "loss": 0.06314754486083984,
      "step": 4145
    },
    {
      "epoch": 0.5777189437748206,
      "grad_norm": 0.2898106575012207,
      "learning_rate": 1.717660894012397e-05,
      "loss": 0.07238006591796875,
      "step": 4146
    },
    {
      "epoch": 0.577858287466035,
      "grad_norm": 0.23891085386276245,
      "learning_rate": 1.7167276528045308e-05,
      "loss": 0.07794380187988281,
      "step": 4147
    },
    {
      "epoch": 0.5779976311572493,
      "grad_norm": 0.45459455251693726,
      "learning_rate": 1.7157944745333114e-05,
      "loss": 0.09162521362304688,
      "step": 4148
    },
    {
      "epoch": 0.5781369748484637,
      "grad_norm": 0.6221371293067932,
      "learning_rate": 1.714861359406072e-05,
      "loss": 0.1309833526611328,
      "step": 4149
    },
    {
      "epoch": 0.5782763185396781,
      "grad_norm": 0.6048424243927002,
      "learning_rate": 1.7139283076301287e-05,
      "loss": 0.11709403991699219,
      "step": 4150
    },
    {
      "epoch": 0.5784156622308925,
      "grad_norm": 0.23226934671401978,
      "learning_rate": 1.7129953194127837e-05,
      "loss": 0.05693531036376953,
      "step": 4151
    },
    {
      "epoch": 0.5785550059221068,
      "grad_norm": 0.3674107789993286,
      "learning_rate": 1.712062394961328e-05,
      "loss": 0.08276176452636719,
      "step": 4152
    },
    {
      "epoch": 0.5786943496133212,
      "grad_norm": 0.2901719808578491,
      "learning_rate": 1.7111295344830324e-05,
      "loss": 0.07941055297851562,
      "step": 4153
    },
    {
      "epoch": 0.5788336933045356,
      "grad_norm": 0.5011548399925232,
      "learning_rate": 1.7101967381851604e-05,
      "loss": 0.08764457702636719,
      "step": 4154
    },
    {
      "epoch": 0.57897303699575,
      "grad_norm": 0.5297576189041138,
      "learning_rate": 1.709264006274956e-05,
      "loss": 0.1109609603881836,
      "step": 4155
    },
    {
      "epoch": 0.5791123806869644,
      "grad_norm": 0.6955164670944214,
      "learning_rate": 1.7083313389596523e-05,
      "loss": 0.1135702133178711,
      "step": 4156
    },
    {
      "epoch": 0.5792517243781787,
      "grad_norm": 0.5102815628051758,
      "learning_rate": 1.7073987364464664e-05,
      "loss": 0.1122589111328125,
      "step": 4157
    },
    {
      "epoch": 0.5793910680693931,
      "grad_norm": 0.2615004777908325,
      "learning_rate": 1.7064661989426012e-05,
      "loss": 0.06870841979980469,
      "step": 4158
    },
    {
      "epoch": 0.5795304117606075,
      "grad_norm": 0.34260252118110657,
      "learning_rate": 1.7055337266552446e-05,
      "loss": 0.07820701599121094,
      "step": 4159
    },
    {
      "epoch": 0.5796697554518219,
      "grad_norm": 0.31398245692253113,
      "learning_rate": 1.704601319791571e-05,
      "loss": 0.07948493957519531,
      "step": 4160
    },
    {
      "epoch": 0.5798090991430362,
      "grad_norm": 0.18907572329044342,
      "learning_rate": 1.7036689785587404e-05,
      "loss": 0.0648050308227539,
      "step": 4161
    },
    {
      "epoch": 0.5799484428342507,
      "grad_norm": 0.2440865933895111,
      "learning_rate": 1.7027367031638976e-05,
      "loss": 0.07743167877197266,
      "step": 4162
    },
    {
      "epoch": 0.5800877865254651,
      "grad_norm": 0.3275556266307831,
      "learning_rate": 1.7018044938141728e-05,
      "loss": 0.07736778259277344,
      "step": 4163
    },
    {
      "epoch": 0.5802271302166795,
      "grad_norm": 0.4042894244194031,
      "learning_rate": 1.700872350716681e-05,
      "loss": 0.08747482299804688,
      "step": 4164
    },
    {
      "epoch": 0.5803664739078939,
      "grad_norm": 0.2917661666870117,
      "learning_rate": 1.6999402740785238e-05,
      "loss": 0.08220863342285156,
      "step": 4165
    },
    {
      "epoch": 0.5805058175991082,
      "grad_norm": 0.382846474647522,
      "learning_rate": 1.6990082641067876e-05,
      "loss": 0.08656120300292969,
      "step": 4166
    },
    {
      "epoch": 0.5806451612903226,
      "grad_norm": 0.38001111149787903,
      "learning_rate": 1.6980763210085425e-05,
      "loss": 0.0933380126953125,
      "step": 4167
    },
    {
      "epoch": 0.580784504981537,
      "grad_norm": 0.41645997762680054,
      "learning_rate": 1.6971444449908474e-05,
      "loss": 0.08356094360351562,
      "step": 4168
    },
    {
      "epoch": 0.5809238486727514,
      "grad_norm": 0.30447179079055786,
      "learning_rate": 1.696212636260741e-05,
      "loss": 0.084991455078125,
      "step": 4169
    },
    {
      "epoch": 0.5810631923639658,
      "grad_norm": 0.4125811755657196,
      "learning_rate": 1.6952808950252518e-05,
      "loss": 0.09144783020019531,
      "step": 4170
    },
    {
      "epoch": 0.5812025360551801,
      "grad_norm": 0.4988737106323242,
      "learning_rate": 1.69434922149139e-05,
      "loss": 0.09589958190917969,
      "step": 4171
    },
    {
      "epoch": 0.5813418797463945,
      "grad_norm": 0.5201955437660217,
      "learning_rate": 1.693417615866154e-05,
      "loss": 0.13245582580566406,
      "step": 4172
    },
    {
      "epoch": 0.5814812234376089,
      "grad_norm": 0.4207427203655243,
      "learning_rate": 1.6924860783565245e-05,
      "loss": 0.08881568908691406,
      "step": 4173
    },
    {
      "epoch": 0.5816205671288233,
      "grad_norm": 0.4924398362636566,
      "learning_rate": 1.691554609169467e-05,
      "loss": 0.1154470443725586,
      "step": 4174
    },
    {
      "epoch": 0.5817599108200376,
      "grad_norm": 0.3551323413848877,
      "learning_rate": 1.6906232085119342e-05,
      "loss": 0.08148527145385742,
      "step": 4175
    },
    {
      "epoch": 0.581899254511252,
      "grad_norm": 0.45390087366104126,
      "learning_rate": 1.6896918765908604e-05,
      "loss": 0.10584640502929688,
      "step": 4176
    },
    {
      "epoch": 0.5820385982024664,
      "grad_norm": 0.36133262515068054,
      "learning_rate": 1.6887606136131673e-05,
      "loss": 0.08881950378417969,
      "step": 4177
    },
    {
      "epoch": 0.5821779418936808,
      "grad_norm": 0.376630574464798,
      "learning_rate": 1.687829419785761e-05,
      "loss": 0.09702873229980469,
      "step": 4178
    },
    {
      "epoch": 0.5823172855848952,
      "grad_norm": 0.1584012806415558,
      "learning_rate": 1.68689829531553e-05,
      "loss": 0.05718803405761719,
      "step": 4179
    },
    {
      "epoch": 0.5824566292761095,
      "grad_norm": 0.35900601744651794,
      "learning_rate": 1.6859672404093494e-05,
      "loss": 0.08628273010253906,
      "step": 4180
    },
    {
      "epoch": 0.5825959729673239,
      "grad_norm": 0.32951533794403076,
      "learning_rate": 1.6850362552740786e-05,
      "loss": 0.07745933532714844,
      "step": 4181
    },
    {
      "epoch": 0.5827353166585383,
      "grad_norm": 0.27356448769569397,
      "learning_rate": 1.6841053401165614e-05,
      "loss": 0.06383132934570312,
      "step": 4182
    },
    {
      "epoch": 0.5828746603497527,
      "grad_norm": 0.41488924622535706,
      "learning_rate": 1.683174495143625e-05,
      "loss": 0.09416580200195312,
      "step": 4183
    },
    {
      "epoch": 0.583014004040967,
      "grad_norm": 0.35064804553985596,
      "learning_rate": 1.6822437205620834e-05,
      "loss": 0.08542633056640625,
      "step": 4184
    },
    {
      "epoch": 0.5831533477321814,
      "grad_norm": 0.21964366734027863,
      "learning_rate": 1.681313016578732e-05,
      "loss": 0.06488466262817383,
      "step": 4185
    },
    {
      "epoch": 0.5832926914233958,
      "grad_norm": 0.41592612862586975,
      "learning_rate": 1.680382383400353e-05,
      "loss": 0.097503662109375,
      "step": 4186
    },
    {
      "epoch": 0.5834320351146102,
      "grad_norm": 0.3102315068244934,
      "learning_rate": 1.679451821233711e-05,
      "loss": 0.09441566467285156,
      "step": 4187
    },
    {
      "epoch": 0.5835713788058245,
      "grad_norm": 0.3010697066783905,
      "learning_rate": 1.6785213302855562e-05,
      "loss": 0.08429527282714844,
      "step": 4188
    },
    {
      "epoch": 0.5837107224970389,
      "grad_norm": 0.5403000712394714,
      "learning_rate": 1.6775909107626227e-05,
      "loss": 0.09599685668945312,
      "step": 4189
    },
    {
      "epoch": 0.5838500661882533,
      "grad_norm": 0.24770641326904297,
      "learning_rate": 1.6766605628716277e-05,
      "loss": 0.0584564208984375,
      "step": 4190
    },
    {
      "epoch": 0.5839894098794677,
      "grad_norm": 0.6927403211593628,
      "learning_rate": 1.675730286819274e-05,
      "loss": 0.1335010528564453,
      "step": 4191
    },
    {
      "epoch": 0.5841287535706821,
      "grad_norm": 0.530586302280426,
      "learning_rate": 1.6748000828122465e-05,
      "loss": 0.10285568237304688,
      "step": 4192
    },
    {
      "epoch": 0.5842680972618964,
      "grad_norm": 0.42310282588005066,
      "learning_rate": 1.673869951057217e-05,
      "loss": 0.10139083862304688,
      "step": 4193
    },
    {
      "epoch": 0.5844074409531108,
      "grad_norm": 0.5413349866867065,
      "learning_rate": 1.6729398917608387e-05,
      "loss": 0.11145305633544922,
      "step": 4194
    },
    {
      "epoch": 0.5845467846443252,
      "grad_norm": 0.6309617161750793,
      "learning_rate": 1.6720099051297494e-05,
      "loss": 0.12165641784667969,
      "step": 4195
    },
    {
      "epoch": 0.5846861283355396,
      "grad_norm": 0.39024996757507324,
      "learning_rate": 1.6710799913705706e-05,
      "loss": 0.0751943588256836,
      "step": 4196
    },
    {
      "epoch": 0.584825472026754,
      "grad_norm": 0.5698543190956116,
      "learning_rate": 1.6701501506899087e-05,
      "loss": 0.08866691589355469,
      "step": 4197
    },
    {
      "epoch": 0.5849648157179683,
      "grad_norm": 0.39603039622306824,
      "learning_rate": 1.6692203832943527e-05,
      "loss": 0.08042526245117188,
      "step": 4198
    },
    {
      "epoch": 0.5851041594091827,
      "grad_norm": 0.3647635877132416,
      "learning_rate": 1.6682906893904754e-05,
      "loss": 0.07810211181640625,
      "step": 4199
    },
    {
      "epoch": 0.5852435031003971,
      "grad_norm": 0.23473447561264038,
      "learning_rate": 1.6673610691848346e-05,
      "loss": 0.0657510757446289,
      "step": 4200
    },
    {
      "epoch": 0.5853828467916115,
      "grad_norm": 0.2577277719974518,
      "learning_rate": 1.6664315228839696e-05,
      "loss": 0.06799983978271484,
      "step": 4201
    },
    {
      "epoch": 0.5855221904828259,
      "grad_norm": 0.4780316650867462,
      "learning_rate": 1.6655020506944046e-05,
      "loss": 0.0970001220703125,
      "step": 4202
    },
    {
      "epoch": 0.5856615341740403,
      "grad_norm": 0.4202057421207428,
      "learning_rate": 1.664572652822647e-05,
      "loss": 0.0806121826171875,
      "step": 4203
    },
    {
      "epoch": 0.5858008778652547,
      "grad_norm": 0.4580095112323761,
      "learning_rate": 1.6636433294751883e-05,
      "loss": 0.10182571411132812,
      "step": 4204
    },
    {
      "epoch": 0.5859402215564691,
      "grad_norm": 0.447134792804718,
      "learning_rate": 1.662714080858503e-05,
      "loss": 0.0942983627319336,
      "step": 4205
    },
    {
      "epoch": 0.5860795652476835,
      "grad_norm": 0.4373663067817688,
      "learning_rate": 1.6617849071790484e-05,
      "loss": 0.08539962768554688,
      "step": 4206
    },
    {
      "epoch": 0.5862189089388978,
      "grad_norm": 0.3297964036464691,
      "learning_rate": 1.6608558086432655e-05,
      "loss": 0.08370208740234375,
      "step": 4207
    },
    {
      "epoch": 0.5863582526301122,
      "grad_norm": 0.3247125446796417,
      "learning_rate": 1.6599267854575788e-05,
      "loss": 0.06982040405273438,
      "step": 4208
    },
    {
      "epoch": 0.5864975963213266,
      "grad_norm": 0.2441520094871521,
      "learning_rate": 1.6589978378283967e-05,
      "loss": 0.0726776123046875,
      "step": 4209
    },
    {
      "epoch": 0.586636940012541,
      "grad_norm": 0.5570501089096069,
      "learning_rate": 1.6580689659621106e-05,
      "loss": 0.12085533142089844,
      "step": 4210
    },
    {
      "epoch": 0.5867762837037553,
      "grad_norm": 0.3606414794921875,
      "learning_rate": 1.6571401700650934e-05,
      "loss": 0.07482051849365234,
      "step": 4211
    },
    {
      "epoch": 0.5869156273949697,
      "grad_norm": 0.3009836673736572,
      "learning_rate": 1.6562114503437017e-05,
      "loss": 0.06842041015625,
      "step": 4212
    },
    {
      "epoch": 0.5870549710861841,
      "grad_norm": 0.4016965925693512,
      "learning_rate": 1.6552828070042782e-05,
      "loss": 0.08827400207519531,
      "step": 4213
    },
    {
      "epoch": 0.5871943147773985,
      "grad_norm": 0.5499261021614075,
      "learning_rate": 1.6543542402531446e-05,
      "loss": 0.10329055786132812,
      "step": 4214
    },
    {
      "epoch": 0.5873336584686129,
      "grad_norm": 0.7325122952461243,
      "learning_rate": 1.6534257502966078e-05,
      "loss": 0.11194229125976562,
      "step": 4215
    },
    {
      "epoch": 0.5874730021598272,
      "grad_norm": 0.21600063145160675,
      "learning_rate": 1.6524973373409563e-05,
      "loss": 0.05808258056640625,
      "step": 4216
    },
    {
      "epoch": 0.5876123458510416,
      "grad_norm": 0.4098205268383026,
      "learning_rate": 1.651569001592463e-05,
      "loss": 0.08368682861328125,
      "step": 4217
    },
    {
      "epoch": 0.587751689542256,
      "grad_norm": 0.5150419473648071,
      "learning_rate": 1.6506407432573828e-05,
      "loss": 0.10800552368164062,
      "step": 4218
    },
    {
      "epoch": 0.5878910332334704,
      "grad_norm": 0.3412403166294098,
      "learning_rate": 1.6497125625419533e-05,
      "loss": 0.06986331939697266,
      "step": 4219
    },
    {
      "epoch": 0.5880303769246847,
      "grad_norm": 0.7018589377403259,
      "learning_rate": 1.6487844596523955e-05,
      "loss": 0.12097930908203125,
      "step": 4220
    },
    {
      "epoch": 0.5881697206158991,
      "grad_norm": 0.39494186639785767,
      "learning_rate": 1.6478564347949127e-05,
      "loss": 0.10202980041503906,
      "step": 4221
    },
    {
      "epoch": 0.5883090643071135,
      "grad_norm": 0.2755142152309418,
      "learning_rate": 1.6469284881756898e-05,
      "loss": 0.07446479797363281,
      "step": 4222
    },
    {
      "epoch": 0.5884484079983279,
      "grad_norm": 0.23825016617774963,
      "learning_rate": 1.6460006200008963e-05,
      "loss": 0.06912994384765625,
      "step": 4223
    },
    {
      "epoch": 0.5885877516895422,
      "grad_norm": 0.42187055945396423,
      "learning_rate": 1.645072830476683e-05,
      "loss": 0.07721138000488281,
      "step": 4224
    },
    {
      "epoch": 0.5887270953807566,
      "grad_norm": 0.45180749893188477,
      "learning_rate": 1.644145119809184e-05,
      "loss": 0.0998830795288086,
      "step": 4225
    },
    {
      "epoch": 0.588866439071971,
      "grad_norm": 0.5356190800666809,
      "learning_rate": 1.643217488204515e-05,
      "loss": 0.08208560943603516,
      "step": 4226
    },
    {
      "epoch": 0.5890057827631854,
      "grad_norm": 0.26010578870773315,
      "learning_rate": 1.6422899358687745e-05,
      "loss": 0.0724649429321289,
      "step": 4227
    },
    {
      "epoch": 0.5891451264543998,
      "grad_norm": 0.3078444004058838,
      "learning_rate": 1.641362463008043e-05,
      "loss": 0.06835269927978516,
      "step": 4228
    },
    {
      "epoch": 0.5892844701456141,
      "grad_norm": 0.34110966324806213,
      "learning_rate": 1.6404350698283853e-05,
      "loss": 0.08257293701171875,
      "step": 4229
    },
    {
      "epoch": 0.5894238138368285,
      "grad_norm": 1.0389535427093506,
      "learning_rate": 1.6395077565358458e-05,
      "loss": 0.15961074829101562,
      "step": 4230
    },
    {
      "epoch": 0.5895631575280429,
      "grad_norm": 0.39979010820388794,
      "learning_rate": 1.6385805233364528e-05,
      "loss": 0.10228347778320312,
      "step": 4231
    },
    {
      "epoch": 0.5897025012192573,
      "grad_norm": 0.31581878662109375,
      "learning_rate": 1.6376533704362155e-05,
      "loss": 0.08399009704589844,
      "step": 4232
    },
    {
      "epoch": 0.5898418449104716,
      "grad_norm": 0.443676620721817,
      "learning_rate": 1.6367262980411273e-05,
      "loss": 0.09247589111328125,
      "step": 4233
    },
    {
      "epoch": 0.589981188601686,
      "grad_norm": 0.477750301361084,
      "learning_rate": 1.635799306357162e-05,
      "loss": 0.0908355712890625,
      "step": 4234
    },
    {
      "epoch": 0.5901205322929004,
      "grad_norm": 0.21010470390319824,
      "learning_rate": 1.6348723955902754e-05,
      "loss": 0.06458854675292969,
      "step": 4235
    },
    {
      "epoch": 0.5902598759841148,
      "grad_norm": 0.49260467290878296,
      "learning_rate": 1.6339455659464073e-05,
      "loss": 0.1141519546508789,
      "step": 4236
    },
    {
      "epoch": 0.5903992196753292,
      "grad_norm": 0.42476221919059753,
      "learning_rate": 1.6330188176314772e-05,
      "loss": 0.09370231628417969,
      "step": 4237
    },
    {
      "epoch": 0.5905385633665435,
      "grad_norm": 0.3329406976699829,
      "learning_rate": 1.6320921508513874e-05,
      "loss": 0.07845783233642578,
      "step": 4238
    },
    {
      "epoch": 0.5906779070577579,
      "grad_norm": 0.5825566053390503,
      "learning_rate": 1.6311655658120214e-05,
      "loss": 0.09480857849121094,
      "step": 4239
    },
    {
      "epoch": 0.5908172507489723,
      "grad_norm": 0.6060634255409241,
      "learning_rate": 1.630239062719247e-05,
      "loss": 0.11422157287597656,
      "step": 4240
    },
    {
      "epoch": 0.5909565944401867,
      "grad_norm": 0.2410358488559723,
      "learning_rate": 1.6293126417789107e-05,
      "loss": 0.06637001037597656,
      "step": 4241
    },
    {
      "epoch": 0.5910959381314012,
      "grad_norm": 0.7789146900177002,
      "learning_rate": 1.628386303196843e-05,
      "loss": 0.114776611328125,
      "step": 4242
    },
    {
      "epoch": 0.5912352818226155,
      "grad_norm": 0.4817928075790405,
      "learning_rate": 1.627460047178854e-05,
      "loss": 0.10793685913085938,
      "step": 4243
    },
    {
      "epoch": 0.5913746255138299,
      "grad_norm": 0.7872316241264343,
      "learning_rate": 1.6265338739307374e-05,
      "loss": 0.1177215576171875,
      "step": 4244
    },
    {
      "epoch": 0.5915139692050443,
      "grad_norm": 0.5085355043411255,
      "learning_rate": 1.6256077836582677e-05,
      "loss": 0.1152496337890625,
      "step": 4245
    },
    {
      "epoch": 0.5916533128962587,
      "grad_norm": 0.4017309248447418,
      "learning_rate": 1.624681776567201e-05,
      "loss": 0.08722496032714844,
      "step": 4246
    },
    {
      "epoch": 0.591792656587473,
      "grad_norm": 0.30937114357948303,
      "learning_rate": 1.6237558528632754e-05,
      "loss": 0.08802604675292969,
      "step": 4247
    },
    {
      "epoch": 0.5919320002786874,
      "grad_norm": 0.41937875747680664,
      "learning_rate": 1.6228300127522083e-05,
      "loss": 0.0906362533569336,
      "step": 4248
    },
    {
      "epoch": 0.5920713439699018,
      "grad_norm": 0.259197860956192,
      "learning_rate": 1.6219042564397023e-05,
      "loss": 0.07859420776367188,
      "step": 4249
    },
    {
      "epoch": 0.5922106876611162,
      "grad_norm": 0.38673698902130127,
      "learning_rate": 1.620978584131438e-05,
      "loss": 0.09707832336425781,
      "step": 4250
    },
    {
      "epoch": 0.5923500313523306,
      "grad_norm": 0.43385303020477295,
      "learning_rate": 1.620052996033079e-05,
      "loss": 0.10590362548828125,
      "step": 4251
    },
    {
      "epoch": 0.5924893750435449,
      "grad_norm": 0.2349770963191986,
      "learning_rate": 1.619127492350271e-05,
      "loss": 0.07269668579101562,
      "step": 4252
    },
    {
      "epoch": 0.5926287187347593,
      "grad_norm": 0.7964914441108704,
      "learning_rate": 1.6182020732886377e-05,
      "loss": 0.10938549041748047,
      "step": 4253
    },
    {
      "epoch": 0.5927680624259737,
      "grad_norm": 0.20606297254562378,
      "learning_rate": 1.6172767390537874e-05,
      "loss": 0.06501960754394531,
      "step": 4254
    },
    {
      "epoch": 0.5929074061171881,
      "grad_norm": 0.7406693696975708,
      "learning_rate": 1.6163514898513076e-05,
      "loss": 0.12823486328125,
      "step": 4255
    },
    {
      "epoch": 0.5930467498084024,
      "grad_norm": 0.27187579870224,
      "learning_rate": 1.6154263258867683e-05,
      "loss": 0.07677268981933594,
      "step": 4256
    },
    {
      "epoch": 0.5931860934996168,
      "grad_norm": 0.5245862007141113,
      "learning_rate": 1.6145012473657197e-05,
      "loss": 0.11178302764892578,
      "step": 4257
    },
    {
      "epoch": 0.5933254371908312,
      "grad_norm": 0.35551878809928894,
      "learning_rate": 1.613576254493693e-05,
      "loss": 0.08932876586914062,
      "step": 4258
    },
    {
      "epoch": 0.5934647808820456,
      "grad_norm": 0.42301812767982483,
      "learning_rate": 1.6126513474762e-05,
      "loss": 0.08414077758789062,
      "step": 4259
    },
    {
      "epoch": 0.59360412457326,
      "grad_norm": 0.6257287263870239,
      "learning_rate": 1.6117265265187337e-05,
      "loss": 0.10829544067382812,
      "step": 4260
    },
    {
      "epoch": 0.5937434682644743,
      "grad_norm": 0.36155927181243896,
      "learning_rate": 1.6108017918267692e-05,
      "loss": 0.07869148254394531,
      "step": 4261
    },
    {
      "epoch": 0.5938828119556887,
      "grad_norm": 0.4308112859725952,
      "learning_rate": 1.6098771436057613e-05,
      "loss": 0.09504127502441406,
      "step": 4262
    },
    {
      "epoch": 0.5940221556469031,
      "grad_norm": 0.2474232167005539,
      "learning_rate": 1.6089525820611453e-05,
      "loss": 0.07116317749023438,
      "step": 4263
    },
    {
      "epoch": 0.5941614993381175,
      "grad_norm": 0.39799797534942627,
      "learning_rate": 1.6080281073983375e-05,
      "loss": 0.1151885986328125,
      "step": 4264
    },
    {
      "epoch": 0.5943008430293318,
      "grad_norm": 0.33493489027023315,
      "learning_rate": 1.6071037198227353e-05,
      "loss": 0.0895700454711914,
      "step": 4265
    },
    {
      "epoch": 0.5944401867205462,
      "grad_norm": 0.40401968359947205,
      "learning_rate": 1.606179419539717e-05,
      "loss": 0.09290885925292969,
      "step": 4266
    },
    {
      "epoch": 0.5945795304117606,
      "grad_norm": 0.4128902554512024,
      "learning_rate": 1.60525520675464e-05,
      "loss": 0.08153820037841797,
      "step": 4267
    },
    {
      "epoch": 0.594718874102975,
      "grad_norm": 0.22877612709999084,
      "learning_rate": 1.6043310816728453e-05,
      "loss": 0.06336021423339844,
      "step": 4268
    },
    {
      "epoch": 0.5948582177941893,
      "grad_norm": 0.5616901516914368,
      "learning_rate": 1.6034070444996498e-05,
      "loss": 0.09675312042236328,
      "step": 4269
    },
    {
      "epoch": 0.5949975614854037,
      "grad_norm": 0.2755493223667145,
      "learning_rate": 1.6024830954403547e-05,
      "loss": 0.07352447509765625,
      "step": 4270
    },
    {
      "epoch": 0.5951369051766181,
      "grad_norm": 0.2607648968696594,
      "learning_rate": 1.60155923470024e-05,
      "loss": 0.07292938232421875,
      "step": 4271
    },
    {
      "epoch": 0.5952762488678325,
      "grad_norm": 0.7093551754951477,
      "learning_rate": 1.6006354624845672e-05,
      "loss": 0.10014915466308594,
      "step": 4272
    },
    {
      "epoch": 0.5954155925590469,
      "grad_norm": 0.29763171076774597,
      "learning_rate": 1.5997117789985776e-05,
      "loss": 0.07940959930419922,
      "step": 4273
    },
    {
      "epoch": 0.5955549362502612,
      "grad_norm": 0.29729560017585754,
      "learning_rate": 1.5987881844474916e-05,
      "loss": 0.08623886108398438,
      "step": 4274
    },
    {
      "epoch": 0.5956942799414756,
      "grad_norm": 0.32531899213790894,
      "learning_rate": 1.597864679036511e-05,
      "loss": 0.08503532409667969,
      "step": 4275
    },
    {
      "epoch": 0.59583362363269,
      "grad_norm": 0.3625545799732208,
      "learning_rate": 1.5969412629708175e-05,
      "loss": 0.091552734375,
      "step": 4276
    },
    {
      "epoch": 0.5959729673239044,
      "grad_norm": 0.4198531210422516,
      "learning_rate": 1.5960179364555736e-05,
      "loss": 0.10091495513916016,
      "step": 4277
    },
    {
      "epoch": 0.5961123110151187,
      "grad_norm": 0.6772096157073975,
      "learning_rate": 1.5950946996959215e-05,
      "loss": 0.1438426971435547,
      "step": 4278
    },
    {
      "epoch": 0.5962516547063331,
      "grad_norm": 0.3440890908241272,
      "learning_rate": 1.594171552896983e-05,
      "loss": 0.09016990661621094,
      "step": 4279
    },
    {
      "epoch": 0.5963909983975475,
      "grad_norm": 0.6459162831306458,
      "learning_rate": 1.5932484962638596e-05,
      "loss": 0.1296067237854004,
      "step": 4280
    },
    {
      "epoch": 0.5965303420887619,
      "grad_norm": 0.30490806698799133,
      "learning_rate": 1.5923255300016343e-05,
      "loss": 0.0869755744934082,
      "step": 4281
    },
    {
      "epoch": 0.5966696857799764,
      "grad_norm": 0.41827458143234253,
      "learning_rate": 1.5914026543153692e-05,
      "loss": 0.09578704833984375,
      "step": 4282
    },
    {
      "epoch": 0.5968090294711907,
      "grad_norm": 0.4666815996170044,
      "learning_rate": 1.5904798694101052e-05,
      "loss": 0.10046577453613281,
      "step": 4283
    },
    {
      "epoch": 0.5969483731624051,
      "grad_norm": 0.5194708108901978,
      "learning_rate": 1.5895571754908656e-05,
      "loss": 0.08269500732421875,
      "step": 4284
    },
    {
      "epoch": 0.5970877168536195,
      "grad_norm": 0.41763946413993835,
      "learning_rate": 1.5886345727626506e-05,
      "loss": 0.08837890625,
      "step": 4285
    },
    {
      "epoch": 0.5972270605448339,
      "grad_norm": 0.4195820093154907,
      "learning_rate": 1.587712061430442e-05,
      "loss": 0.08429336547851562,
      "step": 4286
    },
    {
      "epoch": 0.5973664042360483,
      "grad_norm": 0.8402834534645081,
      "learning_rate": 1.5867896416992002e-05,
      "loss": 0.10921764373779297,
      "step": 4287
    },
    {
      "epoch": 0.5975057479272626,
      "grad_norm": 0.5043755769729614,
      "learning_rate": 1.5858673137738664e-05,
      "loss": 0.0875253677368164,
      "step": 4288
    },
    {
      "epoch": 0.597645091618477,
      "grad_norm": 0.38318386673927307,
      "learning_rate": 1.5849450778593615e-05,
      "loss": 0.08261680603027344,
      "step": 4289
    },
    {
      "epoch": 0.5977844353096914,
      "grad_norm": 0.30514639616012573,
      "learning_rate": 1.5840229341605837e-05,
      "loss": 0.07734107971191406,
      "step": 4290
    },
    {
      "epoch": 0.5979237790009058,
      "grad_norm": 0.4954380691051483,
      "learning_rate": 1.5831008828824134e-05,
      "loss": 0.10290718078613281,
      "step": 4291
    },
    {
      "epoch": 0.5980631226921201,
      "grad_norm": 0.6537814736366272,
      "learning_rate": 1.582178924229708e-05,
      "loss": 0.11701202392578125,
      "step": 4292
    },
    {
      "epoch": 0.5982024663833345,
      "grad_norm": 0.44405457377433777,
      "learning_rate": 1.5812570584073076e-05,
      "loss": 0.06870651245117188,
      "step": 4293
    },
    {
      "epoch": 0.5983418100745489,
      "grad_norm": 0.306293785572052,
      "learning_rate": 1.580335285620028e-05,
      "loss": 0.08233642578125,
      "step": 4294
    },
    {
      "epoch": 0.5984811537657633,
      "grad_norm": 0.48364943265914917,
      "learning_rate": 1.5794136060726682e-05,
      "loss": 0.1042642593383789,
      "step": 4295
    },
    {
      "epoch": 0.5986204974569777,
      "grad_norm": 0.39942795038223267,
      "learning_rate": 1.5784920199700015e-05,
      "loss": 0.09087276458740234,
      "step": 4296
    },
    {
      "epoch": 0.598759841148192,
      "grad_norm": 0.43801358342170715,
      "learning_rate": 1.5775705275167854e-05,
      "loss": 0.0872802734375,
      "step": 4297
    },
    {
      "epoch": 0.5988991848394064,
      "grad_norm": 0.40151938796043396,
      "learning_rate": 1.576649128917754e-05,
      "loss": 0.09026050567626953,
      "step": 4298
    },
    {
      "epoch": 0.5990385285306208,
      "grad_norm": 0.5017002820968628,
      "learning_rate": 1.5757278243776203e-05,
      "loss": 0.09078788757324219,
      "step": 4299
    },
    {
      "epoch": 0.5991778722218352,
      "grad_norm": 0.1967606395483017,
      "learning_rate": 1.5748066141010785e-05,
      "loss": 0.05249214172363281,
      "step": 4300
    },
    {
      "epoch": 0.5993172159130495,
      "grad_norm": 0.38431331515312195,
      "learning_rate": 1.5738854982927993e-05,
      "loss": 0.08099746704101562,
      "step": 4301
    },
    {
      "epoch": 0.5994565596042639,
      "grad_norm": 0.4610428512096405,
      "learning_rate": 1.5729644771574343e-05,
      "loss": 0.08088874816894531,
      "step": 4302
    },
    {
      "epoch": 0.5995959032954783,
      "grad_norm": 0.3018493354320526,
      "learning_rate": 1.572043550899612e-05,
      "loss": 0.0682525634765625,
      "step": 4303
    },
    {
      "epoch": 0.5997352469866927,
      "grad_norm": 0.6429982781410217,
      "learning_rate": 1.5711227197239435e-05,
      "loss": 0.08879470825195312,
      "step": 4304
    },
    {
      "epoch": 0.599874590677907,
      "grad_norm": 0.3359735906124115,
      "learning_rate": 1.5702019838350153e-05,
      "loss": 0.06863117218017578,
      "step": 4305
    },
    {
      "epoch": 0.6000139343691214,
      "grad_norm": 0.4144202768802643,
      "learning_rate": 1.5692813434373934e-05,
      "loss": 0.08248519897460938,
      "step": 4306
    },
    {
      "epoch": 0.6001532780603358,
      "grad_norm": 0.3569704294204712,
      "learning_rate": 1.5683607987356236e-05,
      "loss": 0.08209800720214844,
      "step": 4307
    },
    {
      "epoch": 0.6002926217515502,
      "grad_norm": 0.6100709438323975,
      "learning_rate": 1.5674403499342292e-05,
      "loss": 0.09993934631347656,
      "step": 4308
    },
    {
      "epoch": 0.6004319654427646,
      "grad_norm": 0.8249526023864746,
      "learning_rate": 1.566519997237714e-05,
      "loss": 0.09371757507324219,
      "step": 4309
    },
    {
      "epoch": 0.6005713091339789,
      "grad_norm": 0.25579071044921875,
      "learning_rate": 1.5655997408505595e-05,
      "loss": 0.07179069519042969,
      "step": 4310
    },
    {
      "epoch": 0.6007106528251933,
      "grad_norm": 0.5067604780197144,
      "learning_rate": 1.5646795809772246e-05,
      "loss": 0.12005043029785156,
      "step": 4311
    },
    {
      "epoch": 0.6008499965164077,
      "grad_norm": 0.3436828851699829,
      "learning_rate": 1.563759517822148e-05,
      "loss": 0.07510757446289062,
      "step": 4312
    },
    {
      "epoch": 0.6009893402076221,
      "grad_norm": 0.345567911863327,
      "learning_rate": 1.562839551589747e-05,
      "loss": 0.08088111877441406,
      "step": 4313
    },
    {
      "epoch": 0.6011286838988364,
      "grad_norm": 0.4882480204105377,
      "learning_rate": 1.5619196824844174e-05,
      "loss": 0.09399604797363281,
      "step": 4314
    },
    {
      "epoch": 0.6012680275900508,
      "grad_norm": 0.34947770833969116,
      "learning_rate": 1.5609999107105322e-05,
      "loss": 0.09238815307617188,
      "step": 4315
    },
    {
      "epoch": 0.6014073712812652,
      "grad_norm": 0.5666284561157227,
      "learning_rate": 1.5600802364724456e-05,
      "loss": 0.110137939453125,
      "step": 4316
    },
    {
      "epoch": 0.6015467149724796,
      "grad_norm": 0.2855277955532074,
      "learning_rate": 1.559160659974486e-05,
      "loss": 0.06323432922363281,
      "step": 4317
    },
    {
      "epoch": 0.601686058663694,
      "grad_norm": 0.2140055149793625,
      "learning_rate": 1.5582411814209633e-05,
      "loss": 0.07003593444824219,
      "step": 4318
    },
    {
      "epoch": 0.6018254023549083,
      "grad_norm": 0.44561001658439636,
      "learning_rate": 1.5573218010161642e-05,
      "loss": 0.09647274017333984,
      "step": 4319
    },
    {
      "epoch": 0.6019647460461227,
      "grad_norm": 0.3052234351634979,
      "learning_rate": 1.556402518964355e-05,
      "loss": 0.07429122924804688,
      "step": 4320
    },
    {
      "epoch": 0.6021040897373371,
      "grad_norm": 0.4643697738647461,
      "learning_rate": 1.5554833354697787e-05,
      "loss": 0.09038734436035156,
      "step": 4321
    },
    {
      "epoch": 0.6022434334285516,
      "grad_norm": 0.33749672770500183,
      "learning_rate": 1.5545642507366566e-05,
      "loss": 0.09173965454101562,
      "step": 4322
    },
    {
      "epoch": 0.602382777119766,
      "grad_norm": 0.24376077950000763,
      "learning_rate": 1.5536452649691884e-05,
      "loss": 0.06325149536132812,
      "step": 4323
    },
    {
      "epoch": 0.6025221208109803,
      "grad_norm": 0.4440436065196991,
      "learning_rate": 1.5527263783715515e-05,
      "loss": 0.08221960067749023,
      "step": 4324
    },
    {
      "epoch": 0.6026614645021947,
      "grad_norm": 0.37502971291542053,
      "learning_rate": 1.551807591147902e-05,
      "loss": 0.11469650268554688,
      "step": 4325
    },
    {
      "epoch": 0.6028008081934091,
      "grad_norm": 0.26141172647476196,
      "learning_rate": 1.5508889035023738e-05,
      "loss": 0.07856941223144531,
      "step": 4326
    },
    {
      "epoch": 0.6029401518846235,
      "grad_norm": 0.4522486627101898,
      "learning_rate": 1.549970315639078e-05,
      "loss": 0.09422111511230469,
      "step": 4327
    },
    {
      "epoch": 0.6030794955758378,
      "grad_norm": 0.387520432472229,
      "learning_rate": 1.5490518277621028e-05,
      "loss": 0.07677650451660156,
      "step": 4328
    },
    {
      "epoch": 0.6032188392670522,
      "grad_norm": 0.4246736764907837,
      "learning_rate": 1.5481334400755167e-05,
      "loss": 0.09272956848144531,
      "step": 4329
    },
    {
      "epoch": 0.6033581829582666,
      "grad_norm": 0.38624492287635803,
      "learning_rate": 1.547215152783364e-05,
      "loss": 0.10107612609863281,
      "step": 4330
    },
    {
      "epoch": 0.603497526649481,
      "grad_norm": 0.4081531763076782,
      "learning_rate": 1.5462969660896677e-05,
      "loss": 0.10271549224853516,
      "step": 4331
    },
    {
      "epoch": 0.6036368703406954,
      "grad_norm": 0.5028614401817322,
      "learning_rate": 1.545378880198426e-05,
      "loss": 0.11981391906738281,
      "step": 4332
    },
    {
      "epoch": 0.6037762140319097,
      "grad_norm": 0.44842812418937683,
      "learning_rate": 1.544460895313619e-05,
      "loss": 0.1009979248046875,
      "step": 4333
    },
    {
      "epoch": 0.6039155577231241,
      "grad_norm": 0.5853284001350403,
      "learning_rate": 1.5435430116392003e-05,
      "loss": 0.10016822814941406,
      "step": 4334
    },
    {
      "epoch": 0.6040549014143385,
      "grad_norm": 0.5033504962921143,
      "learning_rate": 1.542625229379103e-05,
      "loss": 0.09569168090820312,
      "step": 4335
    },
    {
      "epoch": 0.6041942451055529,
      "grad_norm": 0.6556228399276733,
      "learning_rate": 1.5417075487372384e-05,
      "loss": 0.1215982437133789,
      "step": 4336
    },
    {
      "epoch": 0.6043335887967672,
      "grad_norm": 0.4698899984359741,
      "learning_rate": 1.5407899699174936e-05,
      "loss": 0.0881967544555664,
      "step": 4337
    },
    {
      "epoch": 0.6044729324879816,
      "grad_norm": 0.7146233916282654,
      "learning_rate": 1.5398724931237334e-05,
      "loss": 0.10245323181152344,
      "step": 4338
    },
    {
      "epoch": 0.604612276179196,
      "grad_norm": 0.3442179560661316,
      "learning_rate": 1.5389551185598e-05,
      "loss": 0.08383750915527344,
      "step": 4339
    },
    {
      "epoch": 0.6047516198704104,
      "grad_norm": 0.2657568156719208,
      "learning_rate": 1.5380378464295133e-05,
      "loss": 0.06733894348144531,
      "step": 4340
    },
    {
      "epoch": 0.6048909635616248,
      "grad_norm": 0.2510393261909485,
      "learning_rate": 1.537120676936671e-05,
      "loss": 0.05688667297363281,
      "step": 4341
    },
    {
      "epoch": 0.6050303072528391,
      "grad_norm": 0.26058146357536316,
      "learning_rate": 1.5362036102850465e-05,
      "loss": 0.08105087280273438,
      "step": 4342
    },
    {
      "epoch": 0.6051696509440535,
      "grad_norm": 0.47928401827812195,
      "learning_rate": 1.535286646678391e-05,
      "loss": 0.10331153869628906,
      "step": 4343
    },
    {
      "epoch": 0.6053089946352679,
      "grad_norm": 0.2748623192310333,
      "learning_rate": 1.5343697863204323e-05,
      "loss": 0.07226753234863281,
      "step": 4344
    },
    {
      "epoch": 0.6054483383264823,
      "grad_norm": 0.754459023475647,
      "learning_rate": 1.533453029414877e-05,
      "loss": 0.12672805786132812,
      "step": 4345
    },
    {
      "epoch": 0.6055876820176966,
      "grad_norm": 0.44312620162963867,
      "learning_rate": 1.5325363761654075e-05,
      "loss": 0.08278560638427734,
      "step": 4346
    },
    {
      "epoch": 0.605727025708911,
      "grad_norm": 0.5666711926460266,
      "learning_rate": 1.5316198267756834e-05,
      "loss": 0.08909988403320312,
      "step": 4347
    },
    {
      "epoch": 0.6058663694001254,
      "grad_norm": 0.5665215849876404,
      "learning_rate": 1.5307033814493392e-05,
      "loss": 0.1345987319946289,
      "step": 4348
    },
    {
      "epoch": 0.6060057130913398,
      "grad_norm": 0.6628089547157288,
      "learning_rate": 1.5297870403899898e-05,
      "loss": 0.11827659606933594,
      "step": 4349
    },
    {
      "epoch": 0.6061450567825541,
      "grad_norm": 0.27774930000305176,
      "learning_rate": 1.528870803801225e-05,
      "loss": 0.0630340576171875,
      "step": 4350
    },
    {
      "epoch": 0.6062844004737685,
      "grad_norm": 0.4380408525466919,
      "learning_rate": 1.5279546718866113e-05,
      "loss": 0.10371017456054688,
      "step": 4351
    },
    {
      "epoch": 0.6064237441649829,
      "grad_norm": 0.2545788586139679,
      "learning_rate": 1.5270386448496926e-05,
      "loss": 0.06620025634765625,
      "step": 4352
    },
    {
      "epoch": 0.6065630878561973,
      "grad_norm": 1.0077630281448364,
      "learning_rate": 1.5261227228939896e-05,
      "loss": 0.1563739776611328,
      "step": 4353
    },
    {
      "epoch": 0.6067024315474117,
      "grad_norm": 0.4680543839931488,
      "learning_rate": 1.5252069062229985e-05,
      "loss": 0.09190940856933594,
      "step": 4354
    },
    {
      "epoch": 0.606841775238626,
      "grad_norm": 0.4419260621070862,
      "learning_rate": 1.5242911950401929e-05,
      "loss": 0.08714866638183594,
      "step": 4355
    },
    {
      "epoch": 0.6069811189298404,
      "grad_norm": 0.24258480966091156,
      "learning_rate": 1.5233755895490232e-05,
      "loss": 0.06897163391113281,
      "step": 4356
    },
    {
      "epoch": 0.6071204626210548,
      "grad_norm": 0.25801438093185425,
      "learning_rate": 1.522460089952916e-05,
      "loss": 0.08353519439697266,
      "step": 4357
    },
    {
      "epoch": 0.6072598063122692,
      "grad_norm": 0.30918967723846436,
      "learning_rate": 1.521544696455275e-05,
      "loss": 0.08315086364746094,
      "step": 4358
    },
    {
      "epoch": 0.6073991500034835,
      "grad_norm": 0.432313472032547,
      "learning_rate": 1.520629409259479e-05,
      "loss": 0.10988235473632812,
      "step": 4359
    },
    {
      "epoch": 0.6075384936946979,
      "grad_norm": 0.3778811991214752,
      "learning_rate": 1.5197142285688831e-05,
      "loss": 0.08148479461669922,
      "step": 4360
    },
    {
      "epoch": 0.6076778373859123,
      "grad_norm": 0.3434462249279022,
      "learning_rate": 1.518799154586821e-05,
      "loss": 0.07528877258300781,
      "step": 4361
    },
    {
      "epoch": 0.6078171810771267,
      "grad_norm": 0.4497319459915161,
      "learning_rate": 1.5178841875166008e-05,
      "loss": 0.0948944091796875,
      "step": 4362
    },
    {
      "epoch": 0.6079565247683412,
      "grad_norm": 0.2841293215751648,
      "learning_rate": 1.5169693275615079e-05,
      "loss": 0.06557941436767578,
      "step": 4363
    },
    {
      "epoch": 0.6080958684595555,
      "grad_norm": 0.4516821503639221,
      "learning_rate": 1.5160545749248014e-05,
      "loss": 0.09394645690917969,
      "step": 4364
    },
    {
      "epoch": 0.6082352121507699,
      "grad_norm": 0.39465266466140747,
      "learning_rate": 1.5151399298097204e-05,
      "loss": 0.08049297332763672,
      "step": 4365
    },
    {
      "epoch": 0.6083745558419843,
      "grad_norm": 0.365246057510376,
      "learning_rate": 1.5142253924194774e-05,
      "loss": 0.07372379302978516,
      "step": 4366
    },
    {
      "epoch": 0.6085138995331987,
      "grad_norm": 0.45779022574424744,
      "learning_rate": 1.5133109629572614e-05,
      "loss": 0.08776092529296875,
      "step": 4367
    },
    {
      "epoch": 0.608653243224413,
      "grad_norm": 0.400137335062027,
      "learning_rate": 1.5123966416262392e-05,
      "loss": 0.08966255187988281,
      "step": 4368
    },
    {
      "epoch": 0.6087925869156274,
      "grad_norm": 0.4529951512813568,
      "learning_rate": 1.51148242862955e-05,
      "loss": 0.09659385681152344,
      "step": 4369
    },
    {
      "epoch": 0.6089319306068418,
      "grad_norm": 0.5005354285240173,
      "learning_rate": 1.5105683241703123e-05,
      "loss": 0.10712623596191406,
      "step": 4370
    },
    {
      "epoch": 0.6090712742980562,
      "grad_norm": 0.42617982625961304,
      "learning_rate": 1.5096543284516188e-05,
      "loss": 0.09533500671386719,
      "step": 4371
    },
    {
      "epoch": 0.6092106179892706,
      "grad_norm": 0.21783317625522614,
      "learning_rate": 1.5087404416765392e-05,
      "loss": 0.06021881103515625,
      "step": 4372
    },
    {
      "epoch": 0.6093499616804849,
      "grad_norm": 0.5120546817779541,
      "learning_rate": 1.5078266640481178e-05,
      "loss": 0.10751533508300781,
      "step": 4373
    },
    {
      "epoch": 0.6094893053716993,
      "grad_norm": 0.46297717094421387,
      "learning_rate": 1.5069129957693755e-05,
      "loss": 0.07459831237792969,
      "step": 4374
    },
    {
      "epoch": 0.6096286490629137,
      "grad_norm": 0.39219221472740173,
      "learning_rate": 1.5059994370433078e-05,
      "loss": 0.08032989501953125,
      "step": 4375
    },
    {
      "epoch": 0.6097679927541281,
      "grad_norm": 0.5103345513343811,
      "learning_rate": 1.5050859880728865e-05,
      "loss": 0.10629749298095703,
      "step": 4376
    },
    {
      "epoch": 0.6099073364453425,
      "grad_norm": 0.2514292895793915,
      "learning_rate": 1.50417264906106e-05,
      "loss": 0.07557487487792969,
      "step": 4377
    },
    {
      "epoch": 0.6100466801365568,
      "grad_norm": 0.6229480504989624,
      "learning_rate": 1.5032594202107509e-05,
      "loss": 0.115631103515625,
      "step": 4378
    },
    {
      "epoch": 0.6101860238277712,
      "grad_norm": 0.24010883271694183,
      "learning_rate": 1.5023463017248582e-05,
      "loss": 0.06472015380859375,
      "step": 4379
    },
    {
      "epoch": 0.6103253675189856,
      "grad_norm": 0.35683804750442505,
      "learning_rate": 1.501433293806255e-05,
      "loss": 0.08292293548583984,
      "step": 4380
    },
    {
      "epoch": 0.6104647112102,
      "grad_norm": 0.8884021639823914,
      "learning_rate": 1.5005203966577919e-05,
      "loss": 0.15869712829589844,
      "step": 4381
    },
    {
      "epoch": 0.6106040549014143,
      "grad_norm": 0.3223867416381836,
      "learning_rate": 1.4996076104822929e-05,
      "loss": 0.07839488983154297,
      "step": 4382
    },
    {
      "epoch": 0.6107433985926287,
      "grad_norm": 0.22706077992916107,
      "learning_rate": 1.498694935482559e-05,
      "loss": 0.06637859344482422,
      "step": 4383
    },
    {
      "epoch": 0.6108827422838431,
      "grad_norm": 0.5740638971328735,
      "learning_rate": 1.4977823718613657e-05,
      "loss": 0.11501502990722656,
      "step": 4384
    },
    {
      "epoch": 0.6110220859750575,
      "grad_norm": 0.5040796399116516,
      "learning_rate": 1.4968699198214634e-05,
      "loss": 0.11604690551757812,
      "step": 4385
    },
    {
      "epoch": 0.6111614296662718,
      "grad_norm": 0.2882956564426422,
      "learning_rate": 1.495957579565578e-05,
      "loss": 0.08512687683105469,
      "step": 4386
    },
    {
      "epoch": 0.6113007733574862,
      "grad_norm": 0.5011085271835327,
      "learning_rate": 1.495045351296411e-05,
      "loss": 0.10135269165039062,
      "step": 4387
    },
    {
      "epoch": 0.6114401170487006,
      "grad_norm": 0.3740328252315521,
      "learning_rate": 1.4941332352166385e-05,
      "loss": 0.0727996826171875,
      "step": 4388
    },
    {
      "epoch": 0.611579460739915,
      "grad_norm": 0.33597666025161743,
      "learning_rate": 1.4932212315289123e-05,
      "loss": 0.08660125732421875,
      "step": 4389
    },
    {
      "epoch": 0.6117188044311294,
      "grad_norm": 0.21418540179729462,
      "learning_rate": 1.4923093404358588e-05,
      "loss": 0.059624671936035156,
      "step": 4390
    },
    {
      "epoch": 0.6118581481223437,
      "grad_norm": 0.33718788623809814,
      "learning_rate": 1.4913975621400787e-05,
      "loss": 0.0784912109375,
      "step": 4391
    },
    {
      "epoch": 0.6119974918135581,
      "grad_norm": 0.46394258737564087,
      "learning_rate": 1.4904858968441485e-05,
      "loss": 0.09051132202148438,
      "step": 4392
    },
    {
      "epoch": 0.6121368355047725,
      "grad_norm": 0.5971395373344421,
      "learning_rate": 1.4895743447506196e-05,
      "loss": 0.10840606689453125,
      "step": 4393
    },
    {
      "epoch": 0.6122761791959869,
      "grad_norm": 0.40696173906326294,
      "learning_rate": 1.4886629060620181e-05,
      "loss": 0.07547950744628906,
      "step": 4394
    },
    {
      "epoch": 0.6124155228872012,
      "grad_norm": 0.20598849654197693,
      "learning_rate": 1.4877515809808459e-05,
      "loss": 0.06276321411132812,
      "step": 4395
    },
    {
      "epoch": 0.6125548665784156,
      "grad_norm": 0.2464522123336792,
      "learning_rate": 1.4868403697095764e-05,
      "loss": 0.07397651672363281,
      "step": 4396
    },
    {
      "epoch": 0.61269421026963,
      "grad_norm": 0.33375898003578186,
      "learning_rate": 1.4859292724506613e-05,
      "loss": 0.09561920166015625,
      "step": 4397
    },
    {
      "epoch": 0.6128335539608444,
      "grad_norm": 0.6788848042488098,
      "learning_rate": 1.4850182894065258e-05,
      "loss": 0.08547210693359375,
      "step": 4398
    },
    {
      "epoch": 0.6129728976520588,
      "grad_norm": 0.4669043719768524,
      "learning_rate": 1.4841074207795684e-05,
      "loss": 0.11097908020019531,
      "step": 4399
    },
    {
      "epoch": 0.6131122413432731,
      "grad_norm": 0.473991334438324,
      "learning_rate": 1.483196666772165e-05,
      "loss": 0.10281181335449219,
      "step": 4400
    },
    {
      "epoch": 0.6132515850344875,
      "grad_norm": 0.3439995348453522,
      "learning_rate": 1.482286027586663e-05,
      "loss": 0.07788848876953125,
      "step": 4401
    },
    {
      "epoch": 0.6133909287257019,
      "grad_norm": 0.2826524078845978,
      "learning_rate": 1.4813755034253862e-05,
      "loss": 0.06554794311523438,
      "step": 4402
    },
    {
      "epoch": 0.6135302724169164,
      "grad_norm": 0.4150504171848297,
      "learning_rate": 1.4804650944906316e-05,
      "loss": 0.09797477722167969,
      "step": 4403
    },
    {
      "epoch": 0.6136696161081308,
      "grad_norm": 0.2973020076751709,
      "learning_rate": 1.4795548009846723e-05,
      "loss": 0.07308578491210938,
      "step": 4404
    },
    {
      "epoch": 0.6138089597993451,
      "grad_norm": 0.7574416995048523,
      "learning_rate": 1.4786446231097546e-05,
      "loss": 0.11719512939453125,
      "step": 4405
    },
    {
      "epoch": 0.6139483034905595,
      "grad_norm": 0.21564653515815735,
      "learning_rate": 1.4777345610680987e-05,
      "loss": 0.07436084747314453,
      "step": 4406
    },
    {
      "epoch": 0.6140876471817739,
      "grad_norm": 0.3153612017631531,
      "learning_rate": 1.4768246150618995e-05,
      "loss": 0.07751274108886719,
      "step": 4407
    },
    {
      "epoch": 0.6142269908729883,
      "grad_norm": 1.010072946548462,
      "learning_rate": 1.4759147852933263e-05,
      "loss": 0.1707611083984375,
      "step": 4408
    },
    {
      "epoch": 0.6143663345642026,
      "grad_norm": 0.29089871048927307,
      "learning_rate": 1.4750050719645227e-05,
      "loss": 0.07923316955566406,
      "step": 4409
    },
    {
      "epoch": 0.614505678255417,
      "grad_norm": 0.42160436511039734,
      "learning_rate": 1.4740954752776064e-05,
      "loss": 0.0824432373046875,
      "step": 4410
    },
    {
      "epoch": 0.6146450219466314,
      "grad_norm": 0.4928901195526123,
      "learning_rate": 1.47318599543467e-05,
      "loss": 0.08845138549804688,
      "step": 4411
    },
    {
      "epoch": 0.6147843656378458,
      "grad_norm": 0.6962693929672241,
      "learning_rate": 1.4722766326377769e-05,
      "loss": 0.13092613220214844,
      "step": 4412
    },
    {
      "epoch": 0.6149237093290602,
      "grad_norm": 0.2583491802215576,
      "learning_rate": 1.4713673870889682e-05,
      "loss": 0.05639171600341797,
      "step": 4413
    },
    {
      "epoch": 0.6150630530202745,
      "grad_norm": 0.249579519033432,
      "learning_rate": 1.4704582589902571e-05,
      "loss": 0.0738992691040039,
      "step": 4414
    },
    {
      "epoch": 0.6152023967114889,
      "grad_norm": 0.2860461473464966,
      "learning_rate": 1.4695492485436308e-05,
      "loss": 0.08080863952636719,
      "step": 4415
    },
    {
      "epoch": 0.6153417404027033,
      "grad_norm": 0.37590473890304565,
      "learning_rate": 1.4686403559510522e-05,
      "loss": 0.07728195190429688,
      "step": 4416
    },
    {
      "epoch": 0.6154810840939177,
      "grad_norm": 0.40981897711753845,
      "learning_rate": 1.4677315814144549e-05,
      "loss": 0.07838058471679688,
      "step": 4417
    },
    {
      "epoch": 0.615620427785132,
      "grad_norm": 0.3278336524963379,
      "learning_rate": 1.4668229251357482e-05,
      "loss": 0.10363578796386719,
      "step": 4418
    },
    {
      "epoch": 0.6157597714763464,
      "grad_norm": 0.3095282316207886,
      "learning_rate": 1.4659143873168145e-05,
      "loss": 0.09001827239990234,
      "step": 4419
    },
    {
      "epoch": 0.6158991151675608,
      "grad_norm": 0.3958961069583893,
      "learning_rate": 1.4650059681595109e-05,
      "loss": 0.09400367736816406,
      "step": 4420
    },
    {
      "epoch": 0.6160384588587752,
      "grad_norm": 0.3282521963119507,
      "learning_rate": 1.4640976678656674e-05,
      "loss": 0.09108352661132812,
      "step": 4421
    },
    {
      "epoch": 0.6161778025499896,
      "grad_norm": 0.6200852990150452,
      "learning_rate": 1.463189486637087e-05,
      "loss": 0.12784957885742188,
      "step": 4422
    },
    {
      "epoch": 0.6163171462412039,
      "grad_norm": 0.4753054082393646,
      "learning_rate": 1.4622814246755468e-05,
      "loss": 0.09352493286132812,
      "step": 4423
    },
    {
      "epoch": 0.6164564899324183,
      "grad_norm": 0.5033499002456665,
      "learning_rate": 1.4613734821827976e-05,
      "loss": 0.10152626037597656,
      "step": 4424
    },
    {
      "epoch": 0.6165958336236327,
      "grad_norm": 0.3000742793083191,
      "learning_rate": 1.4604656593605637e-05,
      "loss": 0.07673454284667969,
      "step": 4425
    },
    {
      "epoch": 0.6167351773148471,
      "grad_norm": 0.29957759380340576,
      "learning_rate": 1.4595579564105432e-05,
      "loss": 0.07199954986572266,
      "step": 4426
    },
    {
      "epoch": 0.6168745210060614,
      "grad_norm": 0.8721594214439392,
      "learning_rate": 1.458650373534406e-05,
      "loss": 0.11328887939453125,
      "step": 4427
    },
    {
      "epoch": 0.6170138646972758,
      "grad_norm": 0.28031790256500244,
      "learning_rate": 1.457742910933796e-05,
      "loss": 0.06940746307373047,
      "step": 4428
    },
    {
      "epoch": 0.6171532083884902,
      "grad_norm": 0.5306028723716736,
      "learning_rate": 1.4568355688103318e-05,
      "loss": 0.1288623809814453,
      "step": 4429
    },
    {
      "epoch": 0.6172925520797046,
      "grad_norm": 0.47203969955444336,
      "learning_rate": 1.4559283473656031e-05,
      "loss": 0.09840202331542969,
      "step": 4430
    },
    {
      "epoch": 0.617431895770919,
      "grad_norm": 0.4191051125526428,
      "learning_rate": 1.4550212468011742e-05,
      "loss": 0.11074638366699219,
      "step": 4431
    },
    {
      "epoch": 0.6175712394621333,
      "grad_norm": 0.5333910584449768,
      "learning_rate": 1.454114267318583e-05,
      "loss": 0.09947395324707031,
      "step": 4432
    },
    {
      "epoch": 0.6177105831533477,
      "grad_norm": 0.3333050608634949,
      "learning_rate": 1.4532074091193385e-05,
      "loss": 0.0719146728515625,
      "step": 4433
    },
    {
      "epoch": 0.6178499268445621,
      "grad_norm": 0.42985668778419495,
      "learning_rate": 1.4523006724049238e-05,
      "loss": 0.09824657440185547,
      "step": 4434
    },
    {
      "epoch": 0.6179892705357765,
      "grad_norm": 0.16967526078224182,
      "learning_rate": 1.4513940573767955e-05,
      "loss": 0.057842254638671875,
      "step": 4435
    },
    {
      "epoch": 0.6181286142269908,
      "grad_norm": 0.3704566955566406,
      "learning_rate": 1.450487564236383e-05,
      "loss": 0.07242012023925781,
      "step": 4436
    },
    {
      "epoch": 0.6182679579182052,
      "grad_norm": 0.6515153646469116,
      "learning_rate": 1.4495811931850886e-05,
      "loss": 0.14598846435546875,
      "step": 4437
    },
    {
      "epoch": 0.6184073016094196,
      "grad_norm": 0.44885873794555664,
      "learning_rate": 1.4486749444242862e-05,
      "loss": 0.11011886596679688,
      "step": 4438
    },
    {
      "epoch": 0.618546645300634,
      "grad_norm": 0.3461162745952606,
      "learning_rate": 1.447768818155324e-05,
      "loss": 0.08923149108886719,
      "step": 4439
    },
    {
      "epoch": 0.6186859889918483,
      "grad_norm": 0.42780423164367676,
      "learning_rate": 1.446862814579523e-05,
      "loss": 0.08513259887695312,
      "step": 4440
    },
    {
      "epoch": 0.6188253326830627,
      "grad_norm": 0.2889516353607178,
      "learning_rate": 1.4459569338981765e-05,
      "loss": 0.06806564331054688,
      "step": 4441
    },
    {
      "epoch": 0.6189646763742771,
      "grad_norm": 0.5077807307243347,
      "learning_rate": 1.4450511763125506e-05,
      "loss": 0.09954071044921875,
      "step": 4442
    },
    {
      "epoch": 0.6191040200654916,
      "grad_norm": 0.5910297632217407,
      "learning_rate": 1.444145542023883e-05,
      "loss": 0.11904525756835938,
      "step": 4443
    },
    {
      "epoch": 0.619243363756706,
      "grad_norm": 0.42604178190231323,
      "learning_rate": 1.4432400312333854e-05,
      "loss": 0.10444831848144531,
      "step": 4444
    },
    {
      "epoch": 0.6193827074479203,
      "grad_norm": 0.39210307598114014,
      "learning_rate": 1.4423346441422422e-05,
      "loss": 0.09297657012939453,
      "step": 4445
    },
    {
      "epoch": 0.6195220511391347,
      "grad_norm": 0.5853728652000427,
      "learning_rate": 1.4414293809516094e-05,
      "loss": 0.10874652862548828,
      "step": 4446
    },
    {
      "epoch": 0.6196613948303491,
      "grad_norm": 0.2693627178668976,
      "learning_rate": 1.4405242418626153e-05,
      "loss": 0.0760040283203125,
      "step": 4447
    },
    {
      "epoch": 0.6198007385215635,
      "grad_norm": 0.4167273938655853,
      "learning_rate": 1.4396192270763622e-05,
      "loss": 0.09534454345703125,
      "step": 4448
    },
    {
      "epoch": 0.6199400822127779,
      "grad_norm": 0.3117867708206177,
      "learning_rate": 1.4387143367939231e-05,
      "loss": 0.08014869689941406,
      "step": 4449
    },
    {
      "epoch": 0.6200794259039922,
      "grad_norm": 0.3887786567211151,
      "learning_rate": 1.4378095712163439e-05,
      "loss": 0.07845544815063477,
      "step": 4450
    },
    {
      "epoch": 0.6202187695952066,
      "grad_norm": 0.6843225359916687,
      "learning_rate": 1.4369049305446423e-05,
      "loss": 0.09704971313476562,
      "step": 4451
    },
    {
      "epoch": 0.620358113286421,
      "grad_norm": 0.3312377333641052,
      "learning_rate": 1.4360004149798101e-05,
      "loss": 0.08862686157226562,
      "step": 4452
    },
    {
      "epoch": 0.6204974569776354,
      "grad_norm": 0.47888845205307007,
      "learning_rate": 1.4350960247228096e-05,
      "loss": 0.0956573486328125,
      "step": 4453
    },
    {
      "epoch": 0.6206368006688497,
      "grad_norm": 0.3487328588962555,
      "learning_rate": 1.4341917599745751e-05,
      "loss": 0.08647537231445312,
      "step": 4454
    },
    {
      "epoch": 0.6207761443600641,
      "grad_norm": 0.24733611941337585,
      "learning_rate": 1.4332876209360136e-05,
      "loss": 0.07207107543945312,
      "step": 4455
    },
    {
      "epoch": 0.6209154880512785,
      "grad_norm": 0.3883887827396393,
      "learning_rate": 1.4323836078080046e-05,
      "loss": 0.10354900360107422,
      "step": 4456
    },
    {
      "epoch": 0.6210548317424929,
      "grad_norm": 0.3598092496395111,
      "learning_rate": 1.4314797207913995e-05,
      "loss": 0.07749176025390625,
      "step": 4457
    },
    {
      "epoch": 0.6211941754337073,
      "grad_norm": 0.3052160143852234,
      "learning_rate": 1.4305759600870208e-05,
      "loss": 0.07677555084228516,
      "step": 4458
    },
    {
      "epoch": 0.6213335191249216,
      "grad_norm": 0.6839844584465027,
      "learning_rate": 1.4296723258956635e-05,
      "loss": 0.11998176574707031,
      "step": 4459
    },
    {
      "epoch": 0.621472862816136,
      "grad_norm": 0.6222801208496094,
      "learning_rate": 1.428768818418094e-05,
      "loss": 0.10771846771240234,
      "step": 4460
    },
    {
      "epoch": 0.6216122065073504,
      "grad_norm": 0.3153083026409149,
      "learning_rate": 1.4278654378550522e-05,
      "loss": 0.0909423828125,
      "step": 4461
    },
    {
      "epoch": 0.6217515501985648,
      "grad_norm": 0.38315221667289734,
      "learning_rate": 1.4269621844072481e-05,
      "loss": 0.08524513244628906,
      "step": 4462
    },
    {
      "epoch": 0.6218908938897791,
      "grad_norm": 0.3358972370624542,
      "learning_rate": 1.4260590582753641e-05,
      "loss": 0.078521728515625,
      "step": 4463
    },
    {
      "epoch": 0.6220302375809935,
      "grad_norm": 0.345201313495636,
      "learning_rate": 1.4251560596600536e-05,
      "loss": 0.11107826232910156,
      "step": 4464
    },
    {
      "epoch": 0.6221695812722079,
      "grad_norm": 0.23416604101657867,
      "learning_rate": 1.4242531887619428e-05,
      "loss": 0.06717395782470703,
      "step": 4465
    },
    {
      "epoch": 0.6223089249634223,
      "grad_norm": 0.4715441167354584,
      "learning_rate": 1.4233504457816291e-05,
      "loss": 0.09518623352050781,
      "step": 4466
    },
    {
      "epoch": 0.6224482686546366,
      "grad_norm": 0.44404715299606323,
      "learning_rate": 1.4224478309196808e-05,
      "loss": 0.07884025573730469,
      "step": 4467
    },
    {
      "epoch": 0.622587612345851,
      "grad_norm": 0.5194984078407288,
      "learning_rate": 1.4215453443766391e-05,
      "loss": 0.09008026123046875,
      "step": 4468
    },
    {
      "epoch": 0.6227269560370654,
      "grad_norm": 0.4031558632850647,
      "learning_rate": 1.420642986353016e-05,
      "loss": 0.0775909423828125,
      "step": 4469
    },
    {
      "epoch": 0.6228662997282798,
      "grad_norm": 0.27821147441864014,
      "learning_rate": 1.4197407570492941e-05,
      "loss": 0.0841522216796875,
      "step": 4470
    },
    {
      "epoch": 0.6230056434194942,
      "grad_norm": 0.624203622341156,
      "learning_rate": 1.4188386566659276e-05,
      "loss": 0.10931110382080078,
      "step": 4471
    },
    {
      "epoch": 0.6231449871107085,
      "grad_norm": 0.4313911199569702,
      "learning_rate": 1.4179366854033441e-05,
      "loss": 0.09364795684814453,
      "step": 4472
    },
    {
      "epoch": 0.6232843308019229,
      "grad_norm": 0.32367846369743347,
      "learning_rate": 1.4170348434619405e-05,
      "loss": 0.0687856674194336,
      "step": 4473
    },
    {
      "epoch": 0.6234236744931373,
      "grad_norm": 0.21808283030986786,
      "learning_rate": 1.4161331310420856e-05,
      "loss": 0.05828380584716797,
      "step": 4474
    },
    {
      "epoch": 0.6235630181843517,
      "grad_norm": 0.7091418504714966,
      "learning_rate": 1.4152315483441188e-05,
      "loss": 0.10678863525390625,
      "step": 4475
    },
    {
      "epoch": 0.623702361875566,
      "grad_norm": 0.2861049771308899,
      "learning_rate": 1.414330095568351e-05,
      "loss": 0.08919143676757812,
      "step": 4476
    },
    {
      "epoch": 0.6238417055667804,
      "grad_norm": 0.39125311374664307,
      "learning_rate": 1.4134287729150653e-05,
      "loss": 0.09955406188964844,
      "step": 4477
    },
    {
      "epoch": 0.6239810492579948,
      "grad_norm": 0.2715814411640167,
      "learning_rate": 1.4125275805845147e-05,
      "loss": 0.0682535171508789,
      "step": 4478
    },
    {
      "epoch": 0.6241203929492092,
      "grad_norm": 0.4069819450378418,
      "learning_rate": 1.4116265187769239e-05,
      "loss": 0.10981941223144531,
      "step": 4479
    },
    {
      "epoch": 0.6242597366404236,
      "grad_norm": 0.3776310980319977,
      "learning_rate": 1.4107255876924865e-05,
      "loss": 0.09056663513183594,
      "step": 4480
    },
    {
      "epoch": 0.6243990803316379,
      "grad_norm": 0.419537752866745,
      "learning_rate": 1.409824787531371e-05,
      "loss": 0.08897018432617188,
      "step": 4481
    },
    {
      "epoch": 0.6245384240228523,
      "grad_norm": 0.504173755645752,
      "learning_rate": 1.408924118493714e-05,
      "loss": 0.09207344055175781,
      "step": 4482
    },
    {
      "epoch": 0.6246777677140668,
      "grad_norm": 0.3685634136199951,
      "learning_rate": 1.4080235807796225e-05,
      "loss": 0.0875701904296875,
      "step": 4483
    },
    {
      "epoch": 0.6248171114052812,
      "grad_norm": 0.7345386147499084,
      "learning_rate": 1.4071231745891768e-05,
      "loss": 0.10126495361328125,
      "step": 4484
    },
    {
      "epoch": 0.6249564550964956,
      "grad_norm": 0.6616860032081604,
      "learning_rate": 1.4062229001224268e-05,
      "loss": 0.10773658752441406,
      "step": 4485
    },
    {
      "epoch": 0.6250957987877099,
      "grad_norm": 0.4704466462135315,
      "learning_rate": 1.4053227575793917e-05,
      "loss": 0.09924507141113281,
      "step": 4486
    },
    {
      "epoch": 0.6252351424789243,
      "grad_norm": 0.7378404140472412,
      "learning_rate": 1.4044227471600627e-05,
      "loss": 0.1233367919921875,
      "step": 4487
    },
    {
      "epoch": 0.6253744861701387,
      "grad_norm": 0.39872685074806213,
      "learning_rate": 1.4035228690644023e-05,
      "loss": 0.092498779296875,
      "step": 4488
    },
    {
      "epoch": 0.6255138298613531,
      "grad_norm": 0.8097701668739319,
      "learning_rate": 1.4026231234923429e-05,
      "loss": 0.11079788208007812,
      "step": 4489
    },
    {
      "epoch": 0.6256531735525674,
      "grad_norm": 0.2856326103210449,
      "learning_rate": 1.4017235106437871e-05,
      "loss": 0.07610607147216797,
      "step": 4490
    },
    {
      "epoch": 0.6257925172437818,
      "grad_norm": 0.4556337893009186,
      "learning_rate": 1.4008240307186084e-05,
      "loss": 0.09707832336425781,
      "step": 4491
    },
    {
      "epoch": 0.6259318609349962,
      "grad_norm": 0.47844991087913513,
      "learning_rate": 1.3999246839166499e-05,
      "loss": 0.07442665100097656,
      "step": 4492
    },
    {
      "epoch": 0.6260712046262106,
      "grad_norm": 0.42635342478752136,
      "learning_rate": 1.399025470437727e-05,
      "loss": 0.08244895935058594,
      "step": 4493
    },
    {
      "epoch": 0.626210548317425,
      "grad_norm": 0.6095706224441528,
      "learning_rate": 1.398126390481624e-05,
      "loss": 0.09479618072509766,
      "step": 4494
    },
    {
      "epoch": 0.6263498920086393,
      "grad_norm": 0.5305610299110413,
      "learning_rate": 1.3972274442480971e-05,
      "loss": 0.09630680084228516,
      "step": 4495
    },
    {
      "epoch": 0.6264892356998537,
      "grad_norm": 0.6241780519485474,
      "learning_rate": 1.3963286319368695e-05,
      "loss": 0.09165191650390625,
      "step": 4496
    },
    {
      "epoch": 0.6266285793910681,
      "grad_norm": 0.6201301217079163,
      "learning_rate": 1.395429953747638e-05,
      "loss": 0.08475208282470703,
      "step": 4497
    },
    {
      "epoch": 0.6267679230822825,
      "grad_norm": 0.3653271496295929,
      "learning_rate": 1.3945314098800684e-05,
      "loss": 0.07399368286132812,
      "step": 4498
    },
    {
      "epoch": 0.6269072667734968,
      "grad_norm": 0.5206770896911621,
      "learning_rate": 1.3936330005337959e-05,
      "loss": 0.10593605041503906,
      "step": 4499
    },
    {
      "epoch": 0.6270466104647112,
      "grad_norm": 0.21271751821041107,
      "learning_rate": 1.392734725908428e-05,
      "loss": 0.07289695739746094,
      "step": 4500
    },
    {
      "epoch": 0.6271859541559256,
      "grad_norm": 0.43345266580581665,
      "learning_rate": 1.3918365862035395e-05,
      "loss": 0.07505035400390625,
      "step": 4501
    },
    {
      "epoch": 0.62732529784714,
      "grad_norm": 0.5195086002349854,
      "learning_rate": 1.3909385816186767e-05,
      "loss": 0.08366966247558594,
      "step": 4502
    },
    {
      "epoch": 0.6274646415383544,
      "grad_norm": 0.2996053099632263,
      "learning_rate": 1.390040712353356e-05,
      "loss": 0.072784423828125,
      "step": 4503
    },
    {
      "epoch": 0.6276039852295687,
      "grad_norm": 0.34629565477371216,
      "learning_rate": 1.3891429786070634e-05,
      "loss": 0.0764455795288086,
      "step": 4504
    },
    {
      "epoch": 0.6277433289207831,
      "grad_norm": 0.5912364721298218,
      "learning_rate": 1.3882453805792549e-05,
      "loss": 0.11140251159667969,
      "step": 4505
    },
    {
      "epoch": 0.6278826726119975,
      "grad_norm": 0.325692355632782,
      "learning_rate": 1.3873479184693568e-05,
      "loss": 0.08506584167480469,
      "step": 4506
    },
    {
      "epoch": 0.6280220163032119,
      "grad_norm": 0.5380306839942932,
      "learning_rate": 1.3864505924767637e-05,
      "loss": 0.10743331909179688,
      "step": 4507
    },
    {
      "epoch": 0.6281613599944262,
      "grad_norm": 0.26450660824775696,
      "learning_rate": 1.3855534028008411e-05,
      "loss": 0.06974220275878906,
      "step": 4508
    },
    {
      "epoch": 0.6283007036856406,
      "grad_norm": 0.21581071615219116,
      "learning_rate": 1.3846563496409245e-05,
      "loss": 0.062229156494140625,
      "step": 4509
    },
    {
      "epoch": 0.628440047376855,
      "grad_norm": 0.17179259657859802,
      "learning_rate": 1.383759433196318e-05,
      "loss": 0.054993629455566406,
      "step": 4510
    },
    {
      "epoch": 0.6285793910680694,
      "grad_norm": 0.6132241487503052,
      "learning_rate": 1.3828626536662978e-05,
      "loss": 0.11154747009277344,
      "step": 4511
    },
    {
      "epoch": 0.6287187347592837,
      "grad_norm": 0.2281607687473297,
      "learning_rate": 1.3819660112501054e-05,
      "loss": 0.06328296661376953,
      "step": 4512
    },
    {
      "epoch": 0.6288580784504981,
      "grad_norm": 0.29085204005241394,
      "learning_rate": 1.3810695061469556e-05,
      "loss": 0.0788583755493164,
      "step": 4513
    },
    {
      "epoch": 0.6289974221417125,
      "grad_norm": 0.46856850385665894,
      "learning_rate": 1.3801731385560312e-05,
      "loss": 0.09288787841796875,
      "step": 4514
    },
    {
      "epoch": 0.6291367658329269,
      "grad_norm": 0.8923294544219971,
      "learning_rate": 1.3792769086764839e-05,
      "loss": 0.1098031997680664,
      "step": 4515
    },
    {
      "epoch": 0.6292761095241413,
      "grad_norm": 1.0878287553787231,
      "learning_rate": 1.3783808167074373e-05,
      "loss": 0.13759231567382812,
      "step": 4516
    },
    {
      "epoch": 0.6294154532153556,
      "grad_norm": 0.2399330586194992,
      "learning_rate": 1.3774848628479807e-05,
      "loss": 0.06270408630371094,
      "step": 4517
    },
    {
      "epoch": 0.62955479690657,
      "grad_norm": 0.42587384581565857,
      "learning_rate": 1.3765890472971755e-05,
      "loss": 0.09809303283691406,
      "step": 4518
    },
    {
      "epoch": 0.6296941405977844,
      "grad_norm": 0.8285608887672424,
      "learning_rate": 1.3756933702540506e-05,
      "loss": 0.1540231704711914,
      "step": 4519
    },
    {
      "epoch": 0.6298334842889988,
      "grad_norm": 0.6803591251373291,
      "learning_rate": 1.3747978319176064e-05,
      "loss": 0.13019943237304688,
      "step": 4520
    },
    {
      "epoch": 0.6299728279802131,
      "grad_norm": 0.3261933922767639,
      "learning_rate": 1.3739024324868107e-05,
      "loss": 0.08071708679199219,
      "step": 4521
    },
    {
      "epoch": 0.6301121716714275,
      "grad_norm": 0.3934943974018097,
      "learning_rate": 1.3730071721605999e-05,
      "loss": 0.07923507690429688,
      "step": 4522
    },
    {
      "epoch": 0.6302515153626419,
      "grad_norm": 0.4387754797935486,
      "learning_rate": 1.3721120511378811e-05,
      "loss": 0.10785102844238281,
      "step": 4523
    },
    {
      "epoch": 0.6303908590538564,
      "grad_norm": 0.6049151420593262,
      "learning_rate": 1.3712170696175289e-05,
      "loss": 0.1043853759765625,
      "step": 4524
    },
    {
      "epoch": 0.6305302027450708,
      "grad_norm": 0.4257989823818207,
      "learning_rate": 1.3703222277983892e-05,
      "loss": 0.093017578125,
      "step": 4525
    },
    {
      "epoch": 0.6306695464362851,
      "grad_norm": 0.5541534423828125,
      "learning_rate": 1.3694275258792742e-05,
      "loss": 0.11624908447265625,
      "step": 4526
    },
    {
      "epoch": 0.6308088901274995,
      "grad_norm": 0.5030675530433655,
      "learning_rate": 1.368532964058968e-05,
      "loss": 0.1008453369140625,
      "step": 4527
    },
    {
      "epoch": 0.6309482338187139,
      "grad_norm": 0.45348939299583435,
      "learning_rate": 1.3676385425362193e-05,
      "loss": 0.11217308044433594,
      "step": 4528
    },
    {
      "epoch": 0.6310875775099283,
      "grad_norm": 0.4023277461528778,
      "learning_rate": 1.3667442615097497e-05,
      "loss": 0.09495162963867188,
      "step": 4529
    },
    {
      "epoch": 0.6312269212011427,
      "grad_norm": 0.4912661910057068,
      "learning_rate": 1.3658501211782478e-05,
      "loss": 0.10956382751464844,
      "step": 4530
    },
    {
      "epoch": 0.631366264892357,
      "grad_norm": 0.4192134737968445,
      "learning_rate": 1.3649561217403707e-05,
      "loss": 0.10927963256835938,
      "step": 4531
    },
    {
      "epoch": 0.6315056085835714,
      "grad_norm": 0.3961367607116699,
      "learning_rate": 1.3640622633947459e-05,
      "loss": 0.07685661315917969,
      "step": 4532
    },
    {
      "epoch": 0.6316449522747858,
      "grad_norm": 0.38368305563926697,
      "learning_rate": 1.3631685463399668e-05,
      "loss": 0.09729385375976562,
      "step": 4533
    },
    {
      "epoch": 0.6317842959660002,
      "grad_norm": 0.4135507047176361,
      "learning_rate": 1.3622749707745979e-05,
      "loss": 0.09932136535644531,
      "step": 4534
    },
    {
      "epoch": 0.6319236396572145,
      "grad_norm": 0.4372934401035309,
      "learning_rate": 1.3613815368971705e-05,
      "loss": 0.0998687744140625,
      "step": 4535
    },
    {
      "epoch": 0.6320629833484289,
      "grad_norm": 0.7867633700370789,
      "learning_rate": 1.360488244906186e-05,
      "loss": 0.12827110290527344,
      "step": 4536
    },
    {
      "epoch": 0.6322023270396433,
      "grad_norm": 0.4166977107524872,
      "learning_rate": 1.3595950950001139e-05,
      "loss": 0.0872802734375,
      "step": 4537
    },
    {
      "epoch": 0.6323416707308577,
      "grad_norm": 0.39581069350242615,
      "learning_rate": 1.3587020873773901e-05,
      "loss": 0.08702945709228516,
      "step": 4538
    },
    {
      "epoch": 0.632481014422072,
      "grad_norm": 0.25799399614334106,
      "learning_rate": 1.3578092222364214e-05,
      "loss": 0.07033729553222656,
      "step": 4539
    },
    {
      "epoch": 0.6326203581132864,
      "grad_norm": 0.5845621228218079,
      "learning_rate": 1.3569164997755821e-05,
      "loss": 0.12522125244140625,
      "step": 4540
    },
    {
      "epoch": 0.6327597018045008,
      "grad_norm": 0.31499022245407104,
      "learning_rate": 1.3560239201932151e-05,
      "loss": 0.07937908172607422,
      "step": 4541
    },
    {
      "epoch": 0.6328990454957152,
      "grad_norm": 0.4379984140396118,
      "learning_rate": 1.35513148368763e-05,
      "loss": 0.09891891479492188,
      "step": 4542
    },
    {
      "epoch": 0.6330383891869296,
      "grad_norm": 0.42195138335227966,
      "learning_rate": 1.3542391904571082e-05,
      "loss": 0.09016990661621094,
      "step": 4543
    },
    {
      "epoch": 0.6331777328781439,
      "grad_norm": 0.2361210435628891,
      "learning_rate": 1.3533470406998941e-05,
      "loss": 0.06784629821777344,
      "step": 4544
    },
    {
      "epoch": 0.6333170765693583,
      "grad_norm": 0.2582574188709259,
      "learning_rate": 1.3524550346142044e-05,
      "loss": 0.06952762603759766,
      "step": 4545
    },
    {
      "epoch": 0.6334564202605727,
      "grad_norm": 0.6075373291969299,
      "learning_rate": 1.3515631723982223e-05,
      "loss": 0.10183143615722656,
      "step": 4546
    },
    {
      "epoch": 0.6335957639517871,
      "grad_norm": 0.39811813831329346,
      "learning_rate": 1.3506714542500986e-05,
      "loss": 0.08579254150390625,
      "step": 4547
    },
    {
      "epoch": 0.6337351076430014,
      "grad_norm": 0.3367428183555603,
      "learning_rate": 1.3497798803679547e-05,
      "loss": 0.09622573852539062,
      "step": 4548
    },
    {
      "epoch": 0.6338744513342158,
      "grad_norm": 0.35608991980552673,
      "learning_rate": 1.348888450949876e-05,
      "loss": 0.07709741592407227,
      "step": 4549
    },
    {
      "epoch": 0.6340137950254302,
      "grad_norm": 0.395068883895874,
      "learning_rate": 1.3479971661939183e-05,
      "loss": 0.09599494934082031,
      "step": 4550
    },
    {
      "epoch": 0.6341531387166446,
      "grad_norm": 0.7665305137634277,
      "learning_rate": 1.3471060262981044e-05,
      "loss": 0.10598564147949219,
      "step": 4551
    },
    {
      "epoch": 0.634292482407859,
      "grad_norm": 0.473127543926239,
      "learning_rate": 1.346215031460426e-05,
      "loss": 0.0818023681640625,
      "step": 4552
    },
    {
      "epoch": 0.6344318260990733,
      "grad_norm": 0.4177418649196625,
      "learning_rate": 1.3453241818788421e-05,
      "loss": 0.08900070190429688,
      "step": 4553
    },
    {
      "epoch": 0.6345711697902877,
      "grad_norm": 0.4634590446949005,
      "learning_rate": 1.3444334777512778e-05,
      "loss": 0.10011863708496094,
      "step": 4554
    },
    {
      "epoch": 0.6347105134815021,
      "grad_norm": 0.2542758882045746,
      "learning_rate": 1.3435429192756275e-05,
      "loss": 0.07648658752441406,
      "step": 4555
    },
    {
      "epoch": 0.6348498571727165,
      "grad_norm": 0.4355270564556122,
      "learning_rate": 1.342652506649754e-05,
      "loss": 0.10840129852294922,
      "step": 4556
    },
    {
      "epoch": 0.6349892008639308,
      "grad_norm": 0.4770478904247284,
      "learning_rate": 1.3417622400714859e-05,
      "loss": 0.10806655883789062,
      "step": 4557
    },
    {
      "epoch": 0.6351285445551452,
      "grad_norm": 0.46158450841903687,
      "learning_rate": 1.3408721197386205e-05,
      "loss": 0.09137535095214844,
      "step": 4558
    },
    {
      "epoch": 0.6352678882463596,
      "grad_norm": 0.34236040711402893,
      "learning_rate": 1.3399821458489215e-05,
      "loss": 0.07886123657226562,
      "step": 4559
    },
    {
      "epoch": 0.635407231937574,
      "grad_norm": 0.3457069396972656,
      "learning_rate": 1.339092318600121e-05,
      "loss": 0.08760261535644531,
      "step": 4560
    },
    {
      "epoch": 0.6355465756287884,
      "grad_norm": 0.9576771259307861,
      "learning_rate": 1.3382026381899191e-05,
      "loss": 0.12982749938964844,
      "step": 4561
    },
    {
      "epoch": 0.6356859193200027,
      "grad_norm": 0.48679786920547485,
      "learning_rate": 1.3373131048159817e-05,
      "loss": 0.10580253601074219,
      "step": 4562
    },
    {
      "epoch": 0.6358252630112171,
      "grad_norm": 0.27350515127182007,
      "learning_rate": 1.3364237186759426e-05,
      "loss": 0.07553672790527344,
      "step": 4563
    },
    {
      "epoch": 0.6359646067024316,
      "grad_norm": 0.33626028895378113,
      "learning_rate": 1.3355344799674042e-05,
      "loss": 0.08385276794433594,
      "step": 4564
    },
    {
      "epoch": 0.636103950393646,
      "grad_norm": 0.5877096056938171,
      "learning_rate": 1.3346453888879341e-05,
      "loss": 0.11372756958007812,
      "step": 4565
    },
    {
      "epoch": 0.6362432940848604,
      "grad_norm": 0.44561782479286194,
      "learning_rate": 1.3337564456350682e-05,
      "loss": 0.10261154174804688,
      "step": 4566
    },
    {
      "epoch": 0.6363826377760747,
      "grad_norm": 0.35637980699539185,
      "learning_rate": 1.3328676504063092e-05,
      "loss": 0.06494712829589844,
      "step": 4567
    },
    {
      "epoch": 0.6365219814672891,
      "grad_norm": 0.5346208810806274,
      "learning_rate": 1.3319790033991278e-05,
      "loss": 0.11942672729492188,
      "step": 4568
    },
    {
      "epoch": 0.6366613251585035,
      "grad_norm": 0.5020948052406311,
      "learning_rate": 1.331090504810961e-05,
      "loss": 0.11946678161621094,
      "step": 4569
    },
    {
      "epoch": 0.6368006688497179,
      "grad_norm": 0.4568309783935547,
      "learning_rate": 1.3302021548392122e-05,
      "loss": 0.09709358215332031,
      "step": 4570
    },
    {
      "epoch": 0.6369400125409322,
      "grad_norm": 0.34954169392585754,
      "learning_rate": 1.3293139536812522e-05,
      "loss": 0.08315277099609375,
      "step": 4571
    },
    {
      "epoch": 0.6370793562321466,
      "grad_norm": 0.40242668986320496,
      "learning_rate": 1.3284259015344205e-05,
      "loss": 0.09887409210205078,
      "step": 4572
    },
    {
      "epoch": 0.637218699923361,
      "grad_norm": 0.7618058323860168,
      "learning_rate": 1.327537998596021e-05,
      "loss": 0.12411689758300781,
      "step": 4573
    },
    {
      "epoch": 0.6373580436145754,
      "grad_norm": 0.16569648683071136,
      "learning_rate": 1.326650245063326e-05,
      "loss": 0.056082725524902344,
      "step": 4574
    },
    {
      "epoch": 0.6374973873057898,
      "grad_norm": 0.29899483919143677,
      "learning_rate": 1.3257626411335733e-05,
      "loss": 0.0890045166015625,
      "step": 4575
    },
    {
      "epoch": 0.6376367309970041,
      "grad_norm": 0.8893817663192749,
      "learning_rate": 1.3248751870039682e-05,
      "loss": 0.11517333984375,
      "step": 4576
    },
    {
      "epoch": 0.6377760746882185,
      "grad_norm": 0.47740551829338074,
      "learning_rate": 1.3239878828716837e-05,
      "loss": 0.09613609313964844,
      "step": 4577
    },
    {
      "epoch": 0.6379154183794329,
      "grad_norm": 0.49468278884887695,
      "learning_rate": 1.3231007289338579e-05,
      "loss": 0.08887434005737305,
      "step": 4578
    },
    {
      "epoch": 0.6380547620706473,
      "grad_norm": 0.2836676239967346,
      "learning_rate": 1.322213725387596e-05,
      "loss": 0.07103538513183594,
      "step": 4579
    },
    {
      "epoch": 0.6381941057618616,
      "grad_norm": 0.6250537633895874,
      "learning_rate": 1.321326872429971e-05,
      "loss": 0.0993194580078125,
      "step": 4580
    },
    {
      "epoch": 0.638333449453076,
      "grad_norm": 0.28257840871810913,
      "learning_rate": 1.3204401702580199e-05,
      "loss": 0.07997512817382812,
      "step": 4581
    },
    {
      "epoch": 0.6384727931442904,
      "grad_norm": 0.47230294346809387,
      "learning_rate": 1.3195536190687485e-05,
      "loss": 0.0904378890991211,
      "step": 4582
    },
    {
      "epoch": 0.6386121368355048,
      "grad_norm": 0.2652273178100586,
      "learning_rate": 1.3186672190591279e-05,
      "loss": 0.06971168518066406,
      "step": 4583
    },
    {
      "epoch": 0.6387514805267192,
      "grad_norm": 0.5331162810325623,
      "learning_rate": 1.3177809704260964e-05,
      "loss": 0.0934295654296875,
      "step": 4584
    },
    {
      "epoch": 0.6388908242179335,
      "grad_norm": 0.24835404753684998,
      "learning_rate": 1.3168948733665583e-05,
      "loss": 0.07886505126953125,
      "step": 4585
    },
    {
      "epoch": 0.6390301679091479,
      "grad_norm": 0.5997633934020996,
      "learning_rate": 1.3160089280773834e-05,
      "loss": 0.1280221939086914,
      "step": 4586
    },
    {
      "epoch": 0.6391695116003623,
      "grad_norm": 0.41046106815338135,
      "learning_rate": 1.3151231347554085e-05,
      "loss": 0.0800013542175293,
      "step": 4587
    },
    {
      "epoch": 0.6393088552915767,
      "grad_norm": 0.27543550729751587,
      "learning_rate": 1.3142374935974373e-05,
      "loss": 0.08016014099121094,
      "step": 4588
    },
    {
      "epoch": 0.639448198982791,
      "grad_norm": 0.2125909924507141,
      "learning_rate": 1.313352004800239e-05,
      "loss": 0.06027793884277344,
      "step": 4589
    },
    {
      "epoch": 0.6395875426740054,
      "grad_norm": 0.44318610429763794,
      "learning_rate": 1.312466668560549e-05,
      "loss": 0.1065073013305664,
      "step": 4590
    },
    {
      "epoch": 0.6397268863652198,
      "grad_norm": 0.3339952528476715,
      "learning_rate": 1.3115814850750686e-05,
      "loss": 0.06838703155517578,
      "step": 4591
    },
    {
      "epoch": 0.6398662300564342,
      "grad_norm": 0.3748065233230591,
      "learning_rate": 1.3106964545404645e-05,
      "loss": 0.08554649353027344,
      "step": 4592
    },
    {
      "epoch": 0.6400055737476485,
      "grad_norm": 0.469242125749588,
      "learning_rate": 1.3098115771533718e-05,
      "loss": 0.09850692749023438,
      "step": 4593
    },
    {
      "epoch": 0.6401449174388629,
      "grad_norm": 0.3519827425479889,
      "learning_rate": 1.3089268531103887e-05,
      "loss": 0.07600212097167969,
      "step": 4594
    },
    {
      "epoch": 0.6402842611300773,
      "grad_norm": 0.39018818736076355,
      "learning_rate": 1.3080422826080828e-05,
      "loss": 0.08947944641113281,
      "step": 4595
    },
    {
      "epoch": 0.6404236048212917,
      "grad_norm": 0.4192550778388977,
      "learning_rate": 1.3071578658429828e-05,
      "loss": 0.09604454040527344,
      "step": 4596
    },
    {
      "epoch": 0.6405629485125061,
      "grad_norm": 0.37847158312797546,
      "learning_rate": 1.3062736030115877e-05,
      "loss": 0.07626533508300781,
      "step": 4597
    },
    {
      "epoch": 0.6407022922037204,
      "grad_norm": 0.6921459436416626,
      "learning_rate": 1.3053894943103598e-05,
      "loss": 0.09898185729980469,
      "step": 4598
    },
    {
      "epoch": 0.6408416358949348,
      "grad_norm": 0.3878924250602722,
      "learning_rate": 1.3045055399357276e-05,
      "loss": 0.09195327758789062,
      "step": 4599
    },
    {
      "epoch": 0.6409809795861492,
      "grad_norm": 0.7339143753051758,
      "learning_rate": 1.3036217400840865e-05,
      "loss": 0.10531997680664062,
      "step": 4600
    },
    {
      "epoch": 0.6411203232773636,
      "grad_norm": 0.25794997811317444,
      "learning_rate": 1.3027380949517964e-05,
      "loss": 0.0682382583618164,
      "step": 4601
    },
    {
      "epoch": 0.641259666968578,
      "grad_norm": 0.39499691128730774,
      "learning_rate": 1.3018546047351828e-05,
      "loss": 0.07992744445800781,
      "step": 4602
    },
    {
      "epoch": 0.6413990106597923,
      "grad_norm": 0.3541356027126312,
      "learning_rate": 1.3009712696305363e-05,
      "loss": 0.08928108215332031,
      "step": 4603
    },
    {
      "epoch": 0.6415383543510068,
      "grad_norm": 0.30947551131248474,
      "learning_rate": 1.3000880898341155e-05,
      "loss": 0.08517837524414062,
      "step": 4604
    },
    {
      "epoch": 0.6416776980422212,
      "grad_norm": 0.40523189306259155,
      "learning_rate": 1.2992050655421413e-05,
      "loss": 0.08661079406738281,
      "step": 4605
    },
    {
      "epoch": 0.6418170417334356,
      "grad_norm": 0.3283569812774658,
      "learning_rate": 1.2983221969508028e-05,
      "loss": 0.08480644226074219,
      "step": 4606
    },
    {
      "epoch": 0.64195638542465,
      "grad_norm": 0.43372541666030884,
      "learning_rate": 1.2974394842562523e-05,
      "loss": 0.11783790588378906,
      "step": 4607
    },
    {
      "epoch": 0.6420957291158643,
      "grad_norm": 0.3714885711669922,
      "learning_rate": 1.2965569276546081e-05,
      "loss": 0.08965301513671875,
      "step": 4608
    },
    {
      "epoch": 0.6422350728070787,
      "grad_norm": 0.49691861867904663,
      "learning_rate": 1.2956745273419551e-05,
      "loss": 0.09466361999511719,
      "step": 4609
    },
    {
      "epoch": 0.6423744164982931,
      "grad_norm": 0.42134004831314087,
      "learning_rate": 1.2947922835143415e-05,
      "loss": 0.09312820434570312,
      "step": 4610
    },
    {
      "epoch": 0.6425137601895075,
      "grad_norm": 0.2938251495361328,
      "learning_rate": 1.2939101963677838e-05,
      "loss": 0.07767105102539062,
      "step": 4611
    },
    {
      "epoch": 0.6426531038807218,
      "grad_norm": 0.5558946132659912,
      "learning_rate": 1.2930282660982592e-05,
      "loss": 0.10945892333984375,
      "step": 4612
    },
    {
      "epoch": 0.6427924475719362,
      "grad_norm": 0.5071899890899658,
      "learning_rate": 1.2921464929017134e-05,
      "loss": 0.1343231201171875,
      "step": 4613
    },
    {
      "epoch": 0.6429317912631506,
      "grad_norm": 0.5871200561523438,
      "learning_rate": 1.2912648769740563e-05,
      "loss": 0.11411094665527344,
      "step": 4614
    },
    {
      "epoch": 0.643071134954365,
      "grad_norm": 0.4432167708873749,
      "learning_rate": 1.2903834185111625e-05,
      "loss": 0.083038330078125,
      "step": 4615
    },
    {
      "epoch": 0.6432104786455793,
      "grad_norm": 0.6252923607826233,
      "learning_rate": 1.2895021177088733e-05,
      "loss": 0.12012481689453125,
      "step": 4616
    },
    {
      "epoch": 0.6433498223367937,
      "grad_norm": 0.6448109745979309,
      "learning_rate": 1.2886209747629921e-05,
      "loss": 0.1269359588623047,
      "step": 4617
    },
    {
      "epoch": 0.6434891660280081,
      "grad_norm": 0.41856157779693604,
      "learning_rate": 1.2877399898692892e-05,
      "loss": 0.08974266052246094,
      "step": 4618
    },
    {
      "epoch": 0.6436285097192225,
      "grad_norm": 0.4617769122123718,
      "learning_rate": 1.286859163223499e-05,
      "loss": 0.10261917114257812,
      "step": 4619
    },
    {
      "epoch": 0.6437678534104369,
      "grad_norm": 0.4924931824207306,
      "learning_rate": 1.2859784950213218e-05,
      "loss": 0.096282958984375,
      "step": 4620
    },
    {
      "epoch": 0.6439071971016512,
      "grad_norm": 0.3940213620662689,
      "learning_rate": 1.2850979854584216e-05,
      "loss": 0.08867645263671875,
      "step": 4621
    },
    {
      "epoch": 0.6440465407928656,
      "grad_norm": 0.4150657653808594,
      "learning_rate": 1.2842176347304283e-05,
      "loss": 0.09499549865722656,
      "step": 4622
    },
    {
      "epoch": 0.64418588448408,
      "grad_norm": 0.34833264350891113,
      "learning_rate": 1.2833374430329341e-05,
      "loss": 0.08764076232910156,
      "step": 4623
    },
    {
      "epoch": 0.6443252281752944,
      "grad_norm": 0.1880774050951004,
      "learning_rate": 1.2824574105614983e-05,
      "loss": 0.05837059020996094,
      "step": 4624
    },
    {
      "epoch": 0.6444645718665087,
      "grad_norm": 0.24805030226707458,
      "learning_rate": 1.2815775375116442e-05,
      "loss": 0.0693817138671875,
      "step": 4625
    },
    {
      "epoch": 0.6446039155577231,
      "grad_norm": 0.35106992721557617,
      "learning_rate": 1.280697824078859e-05,
      "loss": 0.08980178833007812,
      "step": 4626
    },
    {
      "epoch": 0.6447432592489375,
      "grad_norm": 0.1822257786989212,
      "learning_rate": 1.2798182704585968e-05,
      "loss": 0.05999755859375,
      "step": 4627
    },
    {
      "epoch": 0.6448826029401519,
      "grad_norm": 0.22310209274291992,
      "learning_rate": 1.2789388768462715e-05,
      "loss": 0.06826400756835938,
      "step": 4628
    },
    {
      "epoch": 0.6450219466313662,
      "grad_norm": 0.4085978865623474,
      "learning_rate": 1.2780596434372663e-05,
      "loss": 0.08998680114746094,
      "step": 4629
    },
    {
      "epoch": 0.6451612903225806,
      "grad_norm": 0.4442988634109497,
      "learning_rate": 1.2771805704269258e-05,
      "loss": 0.08619022369384766,
      "step": 4630
    },
    {
      "epoch": 0.645300634013795,
      "grad_norm": 0.3271266520023346,
      "learning_rate": 1.2763016580105601e-05,
      "loss": 0.08275985717773438,
      "step": 4631
    },
    {
      "epoch": 0.6454399777050094,
      "grad_norm": 0.672417402267456,
      "learning_rate": 1.2754229063834448e-05,
      "loss": 0.11099004745483398,
      "step": 4632
    },
    {
      "epoch": 0.6455793213962238,
      "grad_norm": 0.3289915919303894,
      "learning_rate": 1.2745443157408164e-05,
      "loss": 0.07958316802978516,
      "step": 4633
    },
    {
      "epoch": 0.6457186650874381,
      "grad_norm": 0.7165187001228333,
      "learning_rate": 1.2736658862778788e-05,
      "loss": 0.10890674591064453,
      "step": 4634
    },
    {
      "epoch": 0.6458580087786525,
      "grad_norm": 0.28138500452041626,
      "learning_rate": 1.2727876181897982e-05,
      "loss": 0.07460594177246094,
      "step": 4635
    },
    {
      "epoch": 0.6459973524698669,
      "grad_norm": 0.782300591468811,
      "learning_rate": 1.2719095116717069e-05,
      "loss": 0.1223001480102539,
      "step": 4636
    },
    {
      "epoch": 0.6461366961610813,
      "grad_norm": 0.47515788674354553,
      "learning_rate": 1.2710315669186994e-05,
      "loss": 0.08977508544921875,
      "step": 4637
    },
    {
      "epoch": 0.6462760398522956,
      "grad_norm": 0.3960266709327698,
      "learning_rate": 1.2701537841258358e-05,
      "loss": 0.0918121337890625,
      "step": 4638
    },
    {
      "epoch": 0.64641538354351,
      "grad_norm": 0.7674164175987244,
      "learning_rate": 1.2692761634881377e-05,
      "loss": 0.12964630126953125,
      "step": 4639
    },
    {
      "epoch": 0.6465547272347244,
      "grad_norm": 0.23392018675804138,
      "learning_rate": 1.2683987052005938e-05,
      "loss": 0.06695938110351562,
      "step": 4640
    },
    {
      "epoch": 0.6466940709259388,
      "grad_norm": 0.514554500579834,
      "learning_rate": 1.2675214094581547e-05,
      "loss": 0.08422279357910156,
      "step": 4641
    },
    {
      "epoch": 0.6468334146171532,
      "grad_norm": 0.5896017551422119,
      "learning_rate": 1.2666442764557352e-05,
      "loss": 0.12107276916503906,
      "step": 4642
    },
    {
      "epoch": 0.6469727583083675,
      "grad_norm": 0.33691656589508057,
      "learning_rate": 1.2657673063882161e-05,
      "loss": 0.07555294036865234,
      "step": 4643
    },
    {
      "epoch": 0.647112101999582,
      "grad_norm": 0.9936367869377136,
      "learning_rate": 1.2648904994504374e-05,
      "loss": 0.12989425659179688,
      "step": 4644
    },
    {
      "epoch": 0.6472514456907964,
      "grad_norm": 0.33850833773612976,
      "learning_rate": 1.2640138558372073e-05,
      "loss": 0.09196281433105469,
      "step": 4645
    },
    {
      "epoch": 0.6473907893820108,
      "grad_norm": 0.31961238384246826,
      "learning_rate": 1.2631373757432957e-05,
      "loss": 0.08641815185546875,
      "step": 4646
    },
    {
      "epoch": 0.6475301330732252,
      "grad_norm": 0.45488911867141724,
      "learning_rate": 1.2622610593634356e-05,
      "loss": 0.1076040267944336,
      "step": 4647
    },
    {
      "epoch": 0.6476694767644395,
      "grad_norm": 0.4775375723838806,
      "learning_rate": 1.2613849068923266e-05,
      "loss": 0.11142730712890625,
      "step": 4648
    },
    {
      "epoch": 0.6478088204556539,
      "grad_norm": 0.21747806668281555,
      "learning_rate": 1.2605089185246277e-05,
      "loss": 0.06620121002197266,
      "step": 4649
    },
    {
      "epoch": 0.6479481641468683,
      "grad_norm": 0.6578227877616882,
      "learning_rate": 1.2596330944549642e-05,
      "loss": 0.11129617691040039,
      "step": 4650
    },
    {
      "epoch": 0.6480875078380827,
      "grad_norm": 1.1994502544403076,
      "learning_rate": 1.2587574348779238e-05,
      "loss": 0.13140487670898438,
      "step": 4651
    },
    {
      "epoch": 0.648226851529297,
      "grad_norm": 0.528349757194519,
      "learning_rate": 1.2578819399880591e-05,
      "loss": 0.11322307586669922,
      "step": 4652
    },
    {
      "epoch": 0.6483661952205114,
      "grad_norm": 0.6469501852989197,
      "learning_rate": 1.2570066099798847e-05,
      "loss": 0.10960960388183594,
      "step": 4653
    },
    {
      "epoch": 0.6485055389117258,
      "grad_norm": 0.5228561162948608,
      "learning_rate": 1.2561314450478785e-05,
      "loss": 0.11450767517089844,
      "step": 4654
    },
    {
      "epoch": 0.6486448826029402,
      "grad_norm": 0.6406696438789368,
      "learning_rate": 1.255256445386482e-05,
      "loss": 0.11263084411621094,
      "step": 4655
    },
    {
      "epoch": 0.6487842262941546,
      "grad_norm": 0.48905622959136963,
      "learning_rate": 1.2543816111901008e-05,
      "loss": 0.07970237731933594,
      "step": 4656
    },
    {
      "epoch": 0.6489235699853689,
      "grad_norm": 0.39804914593696594,
      "learning_rate": 1.253506942653103e-05,
      "loss": 0.08022403717041016,
      "step": 4657
    },
    {
      "epoch": 0.6490629136765833,
      "grad_norm": 0.35044676065444946,
      "learning_rate": 1.2526324399698193e-05,
      "loss": 0.08477115631103516,
      "step": 4658
    },
    {
      "epoch": 0.6492022573677977,
      "grad_norm": 0.893737256526947,
      "learning_rate": 1.2517581033345461e-05,
      "loss": 0.10933208465576172,
      "step": 4659
    },
    {
      "epoch": 0.6493416010590121,
      "grad_norm": 0.28521543741226196,
      "learning_rate": 1.2508839329415384e-05,
      "loss": 0.07487678527832031,
      "step": 4660
    },
    {
      "epoch": 0.6494809447502264,
      "grad_norm": 0.540998101234436,
      "learning_rate": 1.2500099289850185e-05,
      "loss": 0.10055732727050781,
      "step": 4661
    },
    {
      "epoch": 0.6496202884414408,
      "grad_norm": 0.5850149393081665,
      "learning_rate": 1.2491360916591697e-05,
      "loss": 0.11150932312011719,
      "step": 4662
    },
    {
      "epoch": 0.6497596321326552,
      "grad_norm": 0.8154765367507935,
      "learning_rate": 1.2482624211581387e-05,
      "loss": 0.093658447265625,
      "step": 4663
    },
    {
      "epoch": 0.6498989758238696,
      "grad_norm": 0.41117632389068604,
      "learning_rate": 1.2473889176760361e-05,
      "loss": 0.10511493682861328,
      "step": 4664
    },
    {
      "epoch": 0.650038319515084,
      "grad_norm": 0.24195298552513123,
      "learning_rate": 1.246515581406933e-05,
      "loss": 0.06624794006347656,
      "step": 4665
    },
    {
      "epoch": 0.6501776632062983,
      "grad_norm": 0.29970526695251465,
      "learning_rate": 1.2456424125448655e-05,
      "loss": 0.0819711685180664,
      "step": 4666
    },
    {
      "epoch": 0.6503170068975127,
      "grad_norm": 0.4756598472595215,
      "learning_rate": 1.2447694112838309e-05,
      "loss": 0.11324310302734375,
      "step": 4667
    },
    {
      "epoch": 0.6504563505887271,
      "grad_norm": 0.3584001660346985,
      "learning_rate": 1.2438965778177918e-05,
      "loss": 0.095367431640625,
      "step": 4668
    },
    {
      "epoch": 0.6505956942799415,
      "grad_norm": 0.5195461511611938,
      "learning_rate": 1.243023912340671e-05,
      "loss": 0.09329605102539062,
      "step": 4669
    },
    {
      "epoch": 0.6507350379711558,
      "grad_norm": 0.7271227836608887,
      "learning_rate": 1.2421514150463546e-05,
      "loss": 0.13684844970703125,
      "step": 4670
    },
    {
      "epoch": 0.6508743816623702,
      "grad_norm": 0.5592740774154663,
      "learning_rate": 1.2412790861286914e-05,
      "loss": 0.11276054382324219,
      "step": 4671
    },
    {
      "epoch": 0.6510137253535846,
      "grad_norm": 0.44399118423461914,
      "learning_rate": 1.2404069257814939e-05,
      "loss": 0.11736106872558594,
      "step": 4672
    },
    {
      "epoch": 0.651153069044799,
      "grad_norm": 0.5937772393226624,
      "learning_rate": 1.2395349341985355e-05,
      "loss": 0.10880851745605469,
      "step": 4673
    },
    {
      "epoch": 0.6512924127360133,
      "grad_norm": 0.3940418064594269,
      "learning_rate": 1.2386631115735525e-05,
      "loss": 0.08755683898925781,
      "step": 4674
    },
    {
      "epoch": 0.6514317564272277,
      "grad_norm": 0.46224796772003174,
      "learning_rate": 1.2377914581002459e-05,
      "loss": 0.09561538696289062,
      "step": 4675
    },
    {
      "epoch": 0.6515711001184421,
      "grad_norm": 0.47966834902763367,
      "learning_rate": 1.2369199739722744e-05,
      "loss": 0.0870046615600586,
      "step": 4676
    },
    {
      "epoch": 0.6517104438096565,
      "grad_norm": 0.3773738145828247,
      "learning_rate": 1.2360486593832639e-05,
      "loss": 0.090057373046875,
      "step": 4677
    },
    {
      "epoch": 0.6518497875008709,
      "grad_norm": 0.7012414336204529,
      "learning_rate": 1.2351775145267996e-05,
      "loss": 0.13275527954101562,
      "step": 4678
    },
    {
      "epoch": 0.6519891311920852,
      "grad_norm": 0.24457143247127533,
      "learning_rate": 1.2343065395964304e-05,
      "loss": 0.06409263610839844,
      "step": 4679
    },
    {
      "epoch": 0.6521284748832996,
      "grad_norm": 0.26543641090393066,
      "learning_rate": 1.2334357347856678e-05,
      "loss": 0.07217979431152344,
      "step": 4680
    },
    {
      "epoch": 0.652267818574514,
      "grad_norm": 0.5168633460998535,
      "learning_rate": 1.2325651002879835e-05,
      "loss": 0.11470794677734375,
      "step": 4681
    },
    {
      "epoch": 0.6524071622657284,
      "grad_norm": 0.3655026853084564,
      "learning_rate": 1.2316946362968129e-05,
      "loss": 0.07867717742919922,
      "step": 4682
    },
    {
      "epoch": 0.6525465059569427,
      "grad_norm": 0.4323385953903198,
      "learning_rate": 1.230824343005553e-05,
      "loss": 0.09076690673828125,
      "step": 4683
    },
    {
      "epoch": 0.6526858496481572,
      "grad_norm": 0.32415640354156494,
      "learning_rate": 1.2299542206075641e-05,
      "loss": 0.07686996459960938,
      "step": 4684
    },
    {
      "epoch": 0.6528251933393716,
      "grad_norm": 0.19900569319725037,
      "learning_rate": 1.2290842692961673e-05,
      "loss": 0.06326007843017578,
      "step": 4685
    },
    {
      "epoch": 0.652964537030586,
      "grad_norm": 0.27827340364456177,
      "learning_rate": 1.2282144892646453e-05,
      "loss": 0.07333660125732422,
      "step": 4686
    },
    {
      "epoch": 0.6531038807218004,
      "grad_norm": 0.3371955454349518,
      "learning_rate": 1.227344880706243e-05,
      "loss": 0.0788259506225586,
      "step": 4687
    },
    {
      "epoch": 0.6532432244130147,
      "grad_norm": 0.2088494449853897,
      "learning_rate": 1.2264754438141684e-05,
      "loss": 0.0638885498046875,
      "step": 4688
    },
    {
      "epoch": 0.6533825681042291,
      "grad_norm": 0.2508700489997864,
      "learning_rate": 1.2256061787815908e-05,
      "loss": 0.07767677307128906,
      "step": 4689
    },
    {
      "epoch": 0.6535219117954435,
      "grad_norm": 0.3616940379142761,
      "learning_rate": 1.2247370858016407e-05,
      "loss": 0.0838165283203125,
      "step": 4690
    },
    {
      "epoch": 0.6536612554866579,
      "grad_norm": 0.9274809956550598,
      "learning_rate": 1.22386816506741e-05,
      "loss": 0.1276092529296875,
      "step": 4691
    },
    {
      "epoch": 0.6538005991778723,
      "grad_norm": 0.25568634271621704,
      "learning_rate": 1.2229994167719537e-05,
      "loss": 0.06070137023925781,
      "step": 4692
    },
    {
      "epoch": 0.6539399428690866,
      "grad_norm": 0.6080480813980103,
      "learning_rate": 1.2221308411082877e-05,
      "loss": 0.1146697998046875,
      "step": 4693
    },
    {
      "epoch": 0.654079286560301,
      "grad_norm": 0.44333502650260925,
      "learning_rate": 1.2212624382693896e-05,
      "loss": 0.09967422485351562,
      "step": 4694
    },
    {
      "epoch": 0.6542186302515154,
      "grad_norm": 0.4627359211444855,
      "learning_rate": 1.220394208448199e-05,
      "loss": 0.10509490966796875,
      "step": 4695
    },
    {
      "epoch": 0.6543579739427298,
      "grad_norm": 0.276304692029953,
      "learning_rate": 1.2195261518376173e-05,
      "loss": 0.0648183822631836,
      "step": 4696
    },
    {
      "epoch": 0.6544973176339441,
      "grad_norm": 0.3660314381122589,
      "learning_rate": 1.2186582686305056e-05,
      "loss": 0.09733772277832031,
      "step": 4697
    },
    {
      "epoch": 0.6546366613251585,
      "grad_norm": 0.2941841185092926,
      "learning_rate": 1.2177905590196884e-05,
      "loss": 0.07593154907226562,
      "step": 4698
    },
    {
      "epoch": 0.6547760050163729,
      "grad_norm": 0.29108163714408875,
      "learning_rate": 1.2169230231979503e-05,
      "loss": 0.0818629264831543,
      "step": 4699
    },
    {
      "epoch": 0.6549153487075873,
      "grad_norm": 0.3985455632209778,
      "learning_rate": 1.216055661358039e-05,
      "loss": 0.0948333740234375,
      "step": 4700
    },
    {
      "epoch": 0.6550546923988017,
      "grad_norm": 0.5750653743743896,
      "learning_rate": 1.215188473692662e-05,
      "loss": 0.11328506469726562,
      "step": 4701
    },
    {
      "epoch": 0.655194036090016,
      "grad_norm": 0.28588059544563293,
      "learning_rate": 1.2143214603944889e-05,
      "loss": 0.06926965713500977,
      "step": 4702
    },
    {
      "epoch": 0.6553333797812304,
      "grad_norm": 0.4526432752609253,
      "learning_rate": 1.213454621656149e-05,
      "loss": 0.09573936462402344,
      "step": 4703
    },
    {
      "epoch": 0.6554727234724448,
      "grad_norm": 0.3729448616504669,
      "learning_rate": 1.2125879576702354e-05,
      "loss": 0.08625030517578125,
      "step": 4704
    },
    {
      "epoch": 0.6556120671636592,
      "grad_norm": 0.176182821393013,
      "learning_rate": 1.211721468629301e-05,
      "loss": 0.05028343200683594,
      "step": 4705
    },
    {
      "epoch": 0.6557514108548735,
      "grad_norm": 0.28093671798706055,
      "learning_rate": 1.2108551547258598e-05,
      "loss": 0.0761423110961914,
      "step": 4706
    },
    {
      "epoch": 0.6558907545460879,
      "grad_norm": 0.49263814091682434,
      "learning_rate": 1.2099890161523864e-05,
      "loss": 0.10873794555664062,
      "step": 4707
    },
    {
      "epoch": 0.6560300982373023,
      "grad_norm": 0.2510289251804352,
      "learning_rate": 1.209123053101317e-05,
      "loss": 0.08315420150756836,
      "step": 4708
    },
    {
      "epoch": 0.6561694419285167,
      "grad_norm": 0.39222031831741333,
      "learning_rate": 1.2082572657650494e-05,
      "loss": 0.09826183319091797,
      "step": 4709
    },
    {
      "epoch": 0.656308785619731,
      "grad_norm": 0.4111989438533783,
      "learning_rate": 1.2073916543359415e-05,
      "loss": 0.09436798095703125,
      "step": 4710
    },
    {
      "epoch": 0.6564481293109454,
      "grad_norm": 0.20286114513874054,
      "learning_rate": 1.2065262190063132e-05,
      "loss": 0.06338882446289062,
      "step": 4711
    },
    {
      "epoch": 0.6565874730021598,
      "grad_norm": 0.4902791678905487,
      "learning_rate": 1.2056609599684426e-05,
      "loss": 0.10645294189453125,
      "step": 4712
    },
    {
      "epoch": 0.6567268166933742,
      "grad_norm": 0.4025605618953705,
      "learning_rate": 1.2047958774145722e-05,
      "loss": 0.08800697326660156,
      "step": 4713
    },
    {
      "epoch": 0.6568661603845886,
      "grad_norm": 0.48924538493156433,
      "learning_rate": 1.2039309715369033e-05,
      "loss": 0.08595657348632812,
      "step": 4714
    },
    {
      "epoch": 0.6570055040758029,
      "grad_norm": 0.4121195375919342,
      "learning_rate": 1.203066242527597e-05,
      "loss": 0.10023880004882812,
      "step": 4715
    },
    {
      "epoch": 0.6571448477670173,
      "grad_norm": 0.8792543411254883,
      "learning_rate": 1.2022016905787779e-05,
      "loss": 0.11711692810058594,
      "step": 4716
    },
    {
      "epoch": 0.6572841914582317,
      "grad_norm": 0.4625588655471802,
      "learning_rate": 1.2013373158825297e-05,
      "loss": 0.10661697387695312,
      "step": 4717
    },
    {
      "epoch": 0.6574235351494461,
      "grad_norm": 0.39131924510002136,
      "learning_rate": 1.2004731186308956e-05,
      "loss": 0.09503936767578125,
      "step": 4718
    },
    {
      "epoch": 0.6575628788406604,
      "grad_norm": 0.371127188205719,
      "learning_rate": 1.1996090990158804e-05,
      "loss": 0.08177375793457031,
      "step": 4719
    },
    {
      "epoch": 0.6577022225318748,
      "grad_norm": 0.3497975766658783,
      "learning_rate": 1.198745257229451e-05,
      "loss": 0.06208324432373047,
      "step": 4720
    },
    {
      "epoch": 0.6578415662230892,
      "grad_norm": 0.1970861554145813,
      "learning_rate": 1.197881593463532e-05,
      "loss": 0.06342267990112305,
      "step": 4721
    },
    {
      "epoch": 0.6579809099143036,
      "grad_norm": 0.31941401958465576,
      "learning_rate": 1.197018107910011e-05,
      "loss": 0.08168792724609375,
      "step": 4722
    },
    {
      "epoch": 0.658120253605518,
      "grad_norm": 0.590983510017395,
      "learning_rate": 1.1961548007607335e-05,
      "loss": 0.1244049072265625,
      "step": 4723
    },
    {
      "epoch": 0.6582595972967323,
      "grad_norm": 0.436781108379364,
      "learning_rate": 1.1952916722075068e-05,
      "loss": 0.07187271118164062,
      "step": 4724
    },
    {
      "epoch": 0.6583989409879468,
      "grad_norm": 0.6713334918022156,
      "learning_rate": 1.1944287224420991e-05,
      "loss": 0.12765121459960938,
      "step": 4725
    },
    {
      "epoch": 0.6585382846791612,
      "grad_norm": 0.32080692052841187,
      "learning_rate": 1.1935659516562375e-05,
      "loss": 0.07808208465576172,
      "step": 4726
    },
    {
      "epoch": 0.6586776283703756,
      "grad_norm": 0.32225891947746277,
      "learning_rate": 1.1927033600416113e-05,
      "loss": 0.07587051391601562,
      "step": 4727
    },
    {
      "epoch": 0.65881697206159,
      "grad_norm": 0.4505733251571655,
      "learning_rate": 1.1918409477898668e-05,
      "loss": 0.08877754211425781,
      "step": 4728
    },
    {
      "epoch": 0.6589563157528043,
      "grad_norm": 0.7813950181007385,
      "learning_rate": 1.1909787150926128e-05,
      "loss": 0.13432025909423828,
      "step": 4729
    },
    {
      "epoch": 0.6590956594440187,
      "grad_norm": 0.32935529947280884,
      "learning_rate": 1.1901166621414184e-05,
      "loss": 0.07494544982910156,
      "step": 4730
    },
    {
      "epoch": 0.6592350031352331,
      "grad_norm": 0.6717613935470581,
      "learning_rate": 1.1892547891278115e-05,
      "loss": 0.1503772735595703,
      "step": 4731
    },
    {
      "epoch": 0.6593743468264475,
      "grad_norm": 0.8573989868164062,
      "learning_rate": 1.1883930962432811e-05,
      "loss": 0.16273117065429688,
      "step": 4732
    },
    {
      "epoch": 0.6595136905176618,
      "grad_norm": 0.5054094791412354,
      "learning_rate": 1.1875315836792755e-05,
      "loss": 0.1061086654663086,
      "step": 4733
    },
    {
      "epoch": 0.6596530342088762,
      "grad_norm": 0.3250320553779602,
      "learning_rate": 1.1866702516272031e-05,
      "loss": 0.07829666137695312,
      "step": 4734
    },
    {
      "epoch": 0.6597923779000906,
      "grad_norm": 0.8544875979423523,
      "learning_rate": 1.1858091002784315e-05,
      "loss": 0.12312126159667969,
      "step": 4735
    },
    {
      "epoch": 0.659931721591305,
      "grad_norm": 0.25590428709983826,
      "learning_rate": 1.18494812982429e-05,
      "loss": 0.07162284851074219,
      "step": 4736
    },
    {
      "epoch": 0.6600710652825194,
      "grad_norm": 0.4091190993785858,
      "learning_rate": 1.1840873404560662e-05,
      "loss": 0.07591819763183594,
      "step": 4737
    },
    {
      "epoch": 0.6602104089737337,
      "grad_norm": 0.46941789984703064,
      "learning_rate": 1.1832267323650081e-05,
      "loss": 0.10377883911132812,
      "step": 4738
    },
    {
      "epoch": 0.6603497526649481,
      "grad_norm": 0.47158774733543396,
      "learning_rate": 1.1823663057423218e-05,
      "loss": 0.08356094360351562,
      "step": 4739
    },
    {
      "epoch": 0.6604890963561625,
      "grad_norm": 0.496063768863678,
      "learning_rate": 1.1815060607791761e-05,
      "loss": 0.09033393859863281,
      "step": 4740
    },
    {
      "epoch": 0.6606284400473769,
      "grad_norm": 0.2974962592124939,
      "learning_rate": 1.1806459976666972e-05,
      "loss": 0.06977653503417969,
      "step": 4741
    },
    {
      "epoch": 0.6607677837385912,
      "grad_norm": 0.3974471390247345,
      "learning_rate": 1.1797861165959707e-05,
      "loss": 0.08163070678710938,
      "step": 4742
    },
    {
      "epoch": 0.6609071274298056,
      "grad_norm": 0.36079445481300354,
      "learning_rate": 1.1789264177580448e-05,
      "loss": 0.08262157440185547,
      "step": 4743
    },
    {
      "epoch": 0.66104647112102,
      "grad_norm": 0.32545045018196106,
      "learning_rate": 1.1780669013439224e-05,
      "loss": 0.078582763671875,
      "step": 4744
    },
    {
      "epoch": 0.6611858148122344,
      "grad_norm": 0.4704546630382538,
      "learning_rate": 1.1772075675445695e-05,
      "loss": 0.09307861328125,
      "step": 4745
    },
    {
      "epoch": 0.6613251585034488,
      "grad_norm": 0.23987379670143127,
      "learning_rate": 1.1763484165509108e-05,
      "loss": 0.07114219665527344,
      "step": 4746
    },
    {
      "epoch": 0.6614645021946631,
      "grad_norm": 0.29970765113830566,
      "learning_rate": 1.1754894485538288e-05,
      "loss": 0.07055854797363281,
      "step": 4747
    },
    {
      "epoch": 0.6616038458858775,
      "grad_norm": 0.4378312826156616,
      "learning_rate": 1.1746306637441684e-05,
      "loss": 0.0856485366821289,
      "step": 4748
    },
    {
      "epoch": 0.6617431895770919,
      "grad_norm": 0.6849618554115295,
      "learning_rate": 1.1737720623127307e-05,
      "loss": 0.10413169860839844,
      "step": 4749
    },
    {
      "epoch": 0.6618825332683063,
      "grad_norm": 0.4879036545753479,
      "learning_rate": 1.1729136444502775e-05,
      "loss": 0.08490133285522461,
      "step": 4750
    },
    {
      "epoch": 0.6620218769595206,
      "grad_norm": 0.5964678525924683,
      "learning_rate": 1.1720554103475297e-05,
      "loss": 0.10919189453125,
      "step": 4751
    },
    {
      "epoch": 0.662161220650735,
      "grad_norm": 0.5085734724998474,
      "learning_rate": 1.1711973601951676e-05,
      "loss": 0.09035682678222656,
      "step": 4752
    },
    {
      "epoch": 0.6623005643419494,
      "grad_norm": 0.889439582824707,
      "learning_rate": 1.1703394941838302e-05,
      "loss": 0.08391761779785156,
      "step": 4753
    },
    {
      "epoch": 0.6624399080331638,
      "grad_norm": 0.6769018173217773,
      "learning_rate": 1.1694818125041163e-05,
      "loss": 0.09373664855957031,
      "step": 4754
    },
    {
      "epoch": 0.6625792517243781,
      "grad_norm": 0.2515867352485657,
      "learning_rate": 1.1686243153465817e-05,
      "loss": 0.07453536987304688,
      "step": 4755
    },
    {
      "epoch": 0.6627185954155925,
      "grad_norm": 0.3582587242126465,
      "learning_rate": 1.1677670029017437e-05,
      "loss": 0.09323883056640625,
      "step": 4756
    },
    {
      "epoch": 0.6628579391068069,
      "grad_norm": 0.589493453502655,
      "learning_rate": 1.1669098753600777e-05,
      "loss": 0.09296131134033203,
      "step": 4757
    },
    {
      "epoch": 0.6629972827980213,
      "grad_norm": 0.7530641555786133,
      "learning_rate": 1.1660529329120173e-05,
      "loss": 0.12426948547363281,
      "step": 4758
    },
    {
      "epoch": 0.6631366264892357,
      "grad_norm": 0.36014506220817566,
      "learning_rate": 1.1651961757479567e-05,
      "loss": 0.0916147232055664,
      "step": 4759
    },
    {
      "epoch": 0.66327597018045,
      "grad_norm": 0.40629875659942627,
      "learning_rate": 1.1643396040582468e-05,
      "loss": 0.09214019775390625,
      "step": 4760
    },
    {
      "epoch": 0.6634153138716644,
      "grad_norm": 0.2881063222885132,
      "learning_rate": 1.1634832180331976e-05,
      "loss": 0.08112525939941406,
      "step": 4761
    },
    {
      "epoch": 0.6635546575628788,
      "grad_norm": 0.5497850179672241,
      "learning_rate": 1.1626270178630796e-05,
      "loss": 0.11384010314941406,
      "step": 4762
    },
    {
      "epoch": 0.6636940012540932,
      "grad_norm": 0.5820058584213257,
      "learning_rate": 1.1617710037381214e-05,
      "loss": 0.10714340209960938,
      "step": 4763
    },
    {
      "epoch": 0.6638333449453075,
      "grad_norm": 0.6494174599647522,
      "learning_rate": 1.1609151758485088e-05,
      "loss": 0.12042045593261719,
      "step": 4764
    },
    {
      "epoch": 0.663972688636522,
      "grad_norm": 0.4189845025539398,
      "learning_rate": 1.160059534384387e-05,
      "loss": 0.08283805847167969,
      "step": 4765
    },
    {
      "epoch": 0.6641120323277364,
      "grad_norm": 0.7069938778877258,
      "learning_rate": 1.1592040795358604e-05,
      "loss": 0.14244842529296875,
      "step": 4766
    },
    {
      "epoch": 0.6642513760189508,
      "grad_norm": 0.3832804560661316,
      "learning_rate": 1.1583488114929924e-05,
      "loss": 0.09211540222167969,
      "step": 4767
    },
    {
      "epoch": 0.6643907197101652,
      "grad_norm": 0.3414803445339203,
      "learning_rate": 1.1574937304458023e-05,
      "loss": 0.08011817932128906,
      "step": 4768
    },
    {
      "epoch": 0.6645300634013795,
      "grad_norm": 0.23895381391048431,
      "learning_rate": 1.1566388365842717e-05,
      "loss": 0.06811141967773438,
      "step": 4769
    },
    {
      "epoch": 0.6646694070925939,
      "grad_norm": 0.18140864372253418,
      "learning_rate": 1.1557841300983363e-05,
      "loss": 0.05545806884765625,
      "step": 4770
    },
    {
      "epoch": 0.6648087507838083,
      "grad_norm": 0.35216325521469116,
      "learning_rate": 1.1549296111778942e-05,
      "loss": 0.07842063903808594,
      "step": 4771
    },
    {
      "epoch": 0.6649480944750227,
      "grad_norm": 0.45460858941078186,
      "learning_rate": 1.1540752800127986e-05,
      "loss": 0.09504318237304688,
      "step": 4772
    },
    {
      "epoch": 0.665087438166237,
      "grad_norm": 0.3307091295719147,
      "learning_rate": 1.1532211367928628e-05,
      "loss": 0.07523250579833984,
      "step": 4773
    },
    {
      "epoch": 0.6652267818574514,
      "grad_norm": 0.3051570653915405,
      "learning_rate": 1.152367181707859e-05,
      "loss": 0.06237220764160156,
      "step": 4774
    },
    {
      "epoch": 0.6653661255486658,
      "grad_norm": 0.20572714507579803,
      "learning_rate": 1.1515134149475156e-05,
      "loss": 0.05329275131225586,
      "step": 4775
    },
    {
      "epoch": 0.6655054692398802,
      "grad_norm": 0.6783502101898193,
      "learning_rate": 1.1506598367015194e-05,
      "loss": 0.10324287414550781,
      "step": 4776
    },
    {
      "epoch": 0.6656448129310946,
      "grad_norm": 0.38690873980522156,
      "learning_rate": 1.1498064471595167e-05,
      "loss": 0.07210350036621094,
      "step": 4777
    },
    {
      "epoch": 0.6657841566223089,
      "grad_norm": 0.49211564660072327,
      "learning_rate": 1.1489532465111122e-05,
      "loss": 0.10010623931884766,
      "step": 4778
    },
    {
      "epoch": 0.6659235003135233,
      "grad_norm": 0.3164088726043701,
      "learning_rate": 1.1481002349458655e-05,
      "loss": 0.08154678344726562,
      "step": 4779
    },
    {
      "epoch": 0.6660628440047377,
      "grad_norm": 0.5095874667167664,
      "learning_rate": 1.1472474126532981e-05,
      "loss": 0.10906410217285156,
      "step": 4780
    },
    {
      "epoch": 0.6662021876959521,
      "grad_norm": 0.18728159368038177,
      "learning_rate": 1.1463947798228871e-05,
      "loss": 0.054947853088378906,
      "step": 4781
    },
    {
      "epoch": 0.6663415313871665,
      "grad_norm": 0.8258336186408997,
      "learning_rate": 1.1455423366440673e-05,
      "loss": 0.10771417617797852,
      "step": 4782
    },
    {
      "epoch": 0.6664808750783808,
      "grad_norm": 0.4428306221961975,
      "learning_rate": 1.1446900833062325e-05,
      "loss": 0.08477020263671875,
      "step": 4783
    },
    {
      "epoch": 0.6666202187695952,
      "grad_norm": 0.3239222466945648,
      "learning_rate": 1.1438380199987341e-05,
      "loss": 0.08702850341796875,
      "step": 4784
    },
    {
      "epoch": 0.6667595624608096,
      "grad_norm": 0.4847688674926758,
      "learning_rate": 1.1429861469108827e-05,
      "loss": 0.09185934066772461,
      "step": 4785
    },
    {
      "epoch": 0.666898906152024,
      "grad_norm": 0.7997563481330872,
      "learning_rate": 1.1421344642319418e-05,
      "loss": 0.11021232604980469,
      "step": 4786
    },
    {
      "epoch": 0.6670382498432383,
      "grad_norm": 0.2723369598388672,
      "learning_rate": 1.1412829721511378e-05,
      "loss": 0.08430862426757812,
      "step": 4787
    },
    {
      "epoch": 0.6671775935344527,
      "grad_norm": 0.492503821849823,
      "learning_rate": 1.140431670857653e-05,
      "loss": 0.10603523254394531,
      "step": 4788
    },
    {
      "epoch": 0.6673169372256671,
      "grad_norm": 0.2270796000957489,
      "learning_rate": 1.1395805605406263e-05,
      "loss": 0.06230449676513672,
      "step": 4789
    },
    {
      "epoch": 0.6674562809168815,
      "grad_norm": 0.2156296968460083,
      "learning_rate": 1.1387296413891551e-05,
      "loss": 0.06845664978027344,
      "step": 4790
    },
    {
      "epoch": 0.6675956246080959,
      "grad_norm": 0.35421469807624817,
      "learning_rate": 1.1378789135922954e-05,
      "loss": 0.08222198486328125,
      "step": 4791
    },
    {
      "epoch": 0.6677349682993102,
      "grad_norm": 0.453458309173584,
      "learning_rate": 1.1370283773390582e-05,
      "loss": 0.1114664077758789,
      "step": 4792
    },
    {
      "epoch": 0.6678743119905246,
      "grad_norm": 0.41679617762565613,
      "learning_rate": 1.136178032818413e-05,
      "loss": 0.08990669250488281,
      "step": 4793
    },
    {
      "epoch": 0.668013655681739,
      "grad_norm": 0.28685396909713745,
      "learning_rate": 1.1353278802192875e-05,
      "loss": 0.08612537384033203,
      "step": 4794
    },
    {
      "epoch": 0.6681529993729534,
      "grad_norm": 0.32269519567489624,
      "learning_rate": 1.1344779197305674e-05,
      "loss": 0.08080482482910156,
      "step": 4795
    },
    {
      "epoch": 0.6682923430641677,
      "grad_norm": 0.6157500147819519,
      "learning_rate": 1.1336281515410927e-05,
      "loss": 0.12169075012207031,
      "step": 4796
    },
    {
      "epoch": 0.6684316867553821,
      "grad_norm": 0.5253986120223999,
      "learning_rate": 1.1327785758396627e-05,
      "loss": 0.11471939086914062,
      "step": 4797
    },
    {
      "epoch": 0.6685710304465965,
      "grad_norm": 0.30481013655662537,
      "learning_rate": 1.131929192815034e-05,
      "loss": 0.08031272888183594,
      "step": 4798
    },
    {
      "epoch": 0.6687103741378109,
      "grad_norm": 0.4991978108882904,
      "learning_rate": 1.1310800026559213e-05,
      "loss": 0.10347175598144531,
      "step": 4799
    },
    {
      "epoch": 0.6688497178290252,
      "grad_norm": 0.41814640164375305,
      "learning_rate": 1.130231005550993e-05,
      "loss": 0.08360099792480469,
      "step": 4800
    },
    {
      "epoch": 0.6689890615202396,
      "grad_norm": 0.49600547552108765,
      "learning_rate": 1.1293822016888792e-05,
      "loss": 0.10889005661010742,
      "step": 4801
    },
    {
      "epoch": 0.669128405211454,
      "grad_norm": 0.5936664938926697,
      "learning_rate": 1.1285335912581628e-05,
      "loss": 0.10289478302001953,
      "step": 4802
    },
    {
      "epoch": 0.6692677489026684,
      "grad_norm": 0.28394562005996704,
      "learning_rate": 1.1276851744473874e-05,
      "loss": 0.06980514526367188,
      "step": 4803
    },
    {
      "epoch": 0.6694070925938828,
      "grad_norm": 0.3888055682182312,
      "learning_rate": 1.12683695144505e-05,
      "loss": 0.07971572875976562,
      "step": 4804
    },
    {
      "epoch": 0.6695464362850972,
      "grad_norm": 0.33685749769210815,
      "learning_rate": 1.1259889224396072e-05,
      "loss": 0.07966232299804688,
      "step": 4805
    },
    {
      "epoch": 0.6696857799763116,
      "grad_norm": 0.6230636239051819,
      "learning_rate": 1.1251410876194729e-05,
      "loss": 0.12115097045898438,
      "step": 4806
    },
    {
      "epoch": 0.669825123667526,
      "grad_norm": 0.37338998913764954,
      "learning_rate": 1.1242934471730153e-05,
      "loss": 0.0861368179321289,
      "step": 4807
    },
    {
      "epoch": 0.6699644673587404,
      "grad_norm": 0.6596747040748596,
      "learning_rate": 1.1234460012885603e-05,
      "loss": 0.1042623519897461,
      "step": 4808
    },
    {
      "epoch": 0.6701038110499548,
      "grad_norm": 0.553026556968689,
      "learning_rate": 1.122598750154392e-05,
      "loss": 0.09097003936767578,
      "step": 4809
    },
    {
      "epoch": 0.6702431547411691,
      "grad_norm": 0.3573400676250458,
      "learning_rate": 1.1217516939587507e-05,
      "loss": 0.08969688415527344,
      "step": 4810
    },
    {
      "epoch": 0.6703824984323835,
      "grad_norm": 0.24873799085617065,
      "learning_rate": 1.1209048328898313e-05,
      "loss": 0.06612205505371094,
      "step": 4811
    },
    {
      "epoch": 0.6705218421235979,
      "grad_norm": 0.42092999815940857,
      "learning_rate": 1.1200581671357886e-05,
      "loss": 0.09557151794433594,
      "step": 4812
    },
    {
      "epoch": 0.6706611858148123,
      "grad_norm": 0.5001912713050842,
      "learning_rate": 1.1192116968847313e-05,
      "loss": 0.09222030639648438,
      "step": 4813
    },
    {
      "epoch": 0.6708005295060266,
      "grad_norm": 0.6193772554397583,
      "learning_rate": 1.1183654223247268e-05,
      "loss": 0.1329517364501953,
      "step": 4814
    },
    {
      "epoch": 0.670939873197241,
      "grad_norm": 0.5607995986938477,
      "learning_rate": 1.1175193436437968e-05,
      "loss": 0.11072921752929688,
      "step": 4815
    },
    {
      "epoch": 0.6710792168884554,
      "grad_norm": 0.3842204213142395,
      "learning_rate": 1.116673461029921e-05,
      "loss": 0.08977127075195312,
      "step": 4816
    },
    {
      "epoch": 0.6712185605796698,
      "grad_norm": 0.4703288674354553,
      "learning_rate": 1.1158277746710373e-05,
      "loss": 0.10358428955078125,
      "step": 4817
    },
    {
      "epoch": 0.6713579042708842,
      "grad_norm": 0.46156299114227295,
      "learning_rate": 1.1149822847550345e-05,
      "loss": 0.108123779296875,
      "step": 4818
    },
    {
      "epoch": 0.6714972479620985,
      "grad_norm": 0.6436176896095276,
      "learning_rate": 1.1141369914697627e-05,
      "loss": 0.1276569366455078,
      "step": 4819
    },
    {
      "epoch": 0.6716365916533129,
      "grad_norm": 0.23828712105751038,
      "learning_rate": 1.1132918950030274e-05,
      "loss": 0.07295036315917969,
      "step": 4820
    },
    {
      "epoch": 0.6717759353445273,
      "grad_norm": 0.2682456970214844,
      "learning_rate": 1.1124469955425885e-05,
      "loss": 0.06686210632324219,
      "step": 4821
    },
    {
      "epoch": 0.6719152790357417,
      "grad_norm": 0.38418298959732056,
      "learning_rate": 1.1116022932761648e-05,
      "loss": 0.09996223449707031,
      "step": 4822
    },
    {
      "epoch": 0.672054622726956,
      "grad_norm": 0.3907872140407562,
      "learning_rate": 1.1107577883914282e-05,
      "loss": 0.08451461791992188,
      "step": 4823
    },
    {
      "epoch": 0.6721939664181704,
      "grad_norm": 0.5194570422172546,
      "learning_rate": 1.10991348107601e-05,
      "loss": 0.08857917785644531,
      "step": 4824
    },
    {
      "epoch": 0.6723333101093848,
      "grad_norm": 0.3620331585407257,
      "learning_rate": 1.1090693715174947e-05,
      "loss": 0.08185386657714844,
      "step": 4825
    },
    {
      "epoch": 0.6724726538005992,
      "grad_norm": 0.8948733806610107,
      "learning_rate": 1.1082254599034248e-05,
      "loss": 0.12167549133300781,
      "step": 4826
    },
    {
      "epoch": 0.6726119974918136,
      "grad_norm": 0.2811999022960663,
      "learning_rate": 1.1073817464212989e-05,
      "loss": 0.07607841491699219,
      "step": 4827
    },
    {
      "epoch": 0.6727513411830279,
      "grad_norm": 0.3258133828639984,
      "learning_rate": 1.1065382312585698e-05,
      "loss": 0.08303260803222656,
      "step": 4828
    },
    {
      "epoch": 0.6728906848742423,
      "grad_norm": 0.3795014023780823,
      "learning_rate": 1.1056949146026472e-05,
      "loss": 0.08002090454101562,
      "step": 4829
    },
    {
      "epoch": 0.6730300285654567,
      "grad_norm": 0.4429343342781067,
      "learning_rate": 1.1048517966408969e-05,
      "loss": 0.07542610168457031,
      "step": 4830
    },
    {
      "epoch": 0.6731693722566711,
      "grad_norm": 0.8556801676750183,
      "learning_rate": 1.104008877560642e-05,
      "loss": 0.09728050231933594,
      "step": 4831
    },
    {
      "epoch": 0.6733087159478854,
      "grad_norm": 0.43739795684814453,
      "learning_rate": 1.1031661575491577e-05,
      "loss": 0.10685539245605469,
      "step": 4832
    },
    {
      "epoch": 0.6734480596390998,
      "grad_norm": 0.6545463800430298,
      "learning_rate": 1.1023236367936789e-05,
      "loss": 0.10175704956054688,
      "step": 4833
    },
    {
      "epoch": 0.6735874033303142,
      "grad_norm": 0.31383681297302246,
      "learning_rate": 1.1014813154813928e-05,
      "loss": 0.07999801635742188,
      "step": 4834
    },
    {
      "epoch": 0.6737267470215286,
      "grad_norm": 0.4482263922691345,
      "learning_rate": 1.1006391937994459e-05,
      "loss": 0.11433219909667969,
      "step": 4835
    },
    {
      "epoch": 0.673866090712743,
      "grad_norm": 0.41278183460235596,
      "learning_rate": 1.0997972719349363e-05,
      "loss": 0.10252952575683594,
      "step": 4836
    },
    {
      "epoch": 0.6740054344039573,
      "grad_norm": 0.4635907709598541,
      "learning_rate": 1.0989555500749211e-05,
      "loss": 0.09451150894165039,
      "step": 4837
    },
    {
      "epoch": 0.6741447780951717,
      "grad_norm": 0.26358306407928467,
      "learning_rate": 1.0981140284064122e-05,
      "loss": 0.07045364379882812,
      "step": 4838
    },
    {
      "epoch": 0.6742841217863861,
      "grad_norm": 0.26832911372184753,
      "learning_rate": 1.097272707116376e-05,
      "loss": 0.08025932312011719,
      "step": 4839
    },
    {
      "epoch": 0.6744234654776005,
      "grad_norm": 0.38217592239379883,
      "learning_rate": 1.0964315863917337e-05,
      "loss": 0.07806777954101562,
      "step": 4840
    },
    {
      "epoch": 0.6745628091688148,
      "grad_norm": 0.6055396795272827,
      "learning_rate": 1.0955906664193641e-05,
      "loss": 0.12038993835449219,
      "step": 4841
    },
    {
      "epoch": 0.6747021528600292,
      "grad_norm": 0.5734759569168091,
      "learning_rate": 1.0947499473861012e-05,
      "loss": 0.12209892272949219,
      "step": 4842
    },
    {
      "epoch": 0.6748414965512436,
      "grad_norm": 0.227107435464859,
      "learning_rate": 1.0939094294787327e-05,
      "loss": 0.0742330551147461,
      "step": 4843
    },
    {
      "epoch": 0.674980840242458,
      "grad_norm": 0.34071043133735657,
      "learning_rate": 1.0930691128840019e-05,
      "loss": 0.08356285095214844,
      "step": 4844
    },
    {
      "epoch": 0.6751201839336725,
      "grad_norm": 0.4039389491081238,
      "learning_rate": 1.0922289977886087e-05,
      "loss": 0.09166908264160156,
      "step": 4845
    },
    {
      "epoch": 0.6752595276248868,
      "grad_norm": 0.39852574467658997,
      "learning_rate": 1.091389084379208e-05,
      "loss": 0.08418750762939453,
      "step": 4846
    },
    {
      "epoch": 0.6753988713161012,
      "grad_norm": 0.5096591114997864,
      "learning_rate": 1.0905493728424079e-05,
      "loss": 0.09003353118896484,
      "step": 4847
    },
    {
      "epoch": 0.6755382150073156,
      "grad_norm": 0.4495689868927002,
      "learning_rate": 1.0897098633647745e-05,
      "loss": 0.09197425842285156,
      "step": 4848
    },
    {
      "epoch": 0.67567755869853,
      "grad_norm": 0.3575702905654907,
      "learning_rate": 1.0888705561328283e-05,
      "loss": 0.08242607116699219,
      "step": 4849
    },
    {
      "epoch": 0.6758169023897443,
      "grad_norm": 0.4976254105567932,
      "learning_rate": 1.088031451333042e-05,
      "loss": 0.10451602935791016,
      "step": 4850
    },
    {
      "epoch": 0.6759562460809587,
      "grad_norm": 0.3875209093093872,
      "learning_rate": 1.0871925491518462e-05,
      "loss": 0.08437156677246094,
      "step": 4851
    },
    {
      "epoch": 0.6760955897721731,
      "grad_norm": 0.3933204114437103,
      "learning_rate": 1.0863538497756263e-05,
      "loss": 0.09020614624023438,
      "step": 4852
    },
    {
      "epoch": 0.6762349334633875,
      "grad_norm": 0.28586864471435547,
      "learning_rate": 1.085515353390723e-05,
      "loss": 0.07935142517089844,
      "step": 4853
    },
    {
      "epoch": 0.6763742771546019,
      "grad_norm": 0.27550914883613586,
      "learning_rate": 1.0846770601834303e-05,
      "loss": 0.07664871215820312,
      "step": 4854
    },
    {
      "epoch": 0.6765136208458162,
      "grad_norm": 0.29771947860717773,
      "learning_rate": 1.0838389703399965e-05,
      "loss": 0.08447647094726562,
      "step": 4855
    },
    {
      "epoch": 0.6766529645370306,
      "grad_norm": 0.38749960064888,
      "learning_rate": 1.0830010840466282e-05,
      "loss": 0.0803384780883789,
      "step": 4856
    },
    {
      "epoch": 0.676792308228245,
      "grad_norm": 0.268987238407135,
      "learning_rate": 1.0821634014894827e-05,
      "loss": 0.07018470764160156,
      "step": 4857
    },
    {
      "epoch": 0.6769316519194594,
      "grad_norm": 0.5637885332107544,
      "learning_rate": 1.0813259228546746e-05,
      "loss": 0.12892723083496094,
      "step": 4858
    },
    {
      "epoch": 0.6770709956106737,
      "grad_norm": 0.46264615654945374,
      "learning_rate": 1.080488648328274e-05,
      "loss": 0.08848762512207031,
      "step": 4859
    },
    {
      "epoch": 0.6772103393018881,
      "grad_norm": 0.37554994225502014,
      "learning_rate": 1.0796515780963026e-05,
      "loss": 0.08361244201660156,
      "step": 4860
    },
    {
      "epoch": 0.6773496829931025,
      "grad_norm": 0.4743032455444336,
      "learning_rate": 1.078814712344738e-05,
      "loss": 0.10505294799804688,
      "step": 4861
    },
    {
      "epoch": 0.6774890266843169,
      "grad_norm": 0.38195300102233887,
      "learning_rate": 1.0779780512595136e-05,
      "loss": 0.09167098999023438,
      "step": 4862
    },
    {
      "epoch": 0.6776283703755313,
      "grad_norm": 0.35630303621292114,
      "learning_rate": 1.0771415950265163e-05,
      "loss": 0.08433341979980469,
      "step": 4863
    },
    {
      "epoch": 0.6777677140667456,
      "grad_norm": 0.5027072429656982,
      "learning_rate": 1.0763053438315876e-05,
      "loss": 0.11568069458007812,
      "step": 4864
    },
    {
      "epoch": 0.67790705775796,
      "grad_norm": 0.7148174047470093,
      "learning_rate": 1.0754692978605226e-05,
      "loss": 0.14002609252929688,
      "step": 4865
    },
    {
      "epoch": 0.6780464014491744,
      "grad_norm": 0.2682296335697174,
      "learning_rate": 1.074633457299072e-05,
      "loss": 0.07629776000976562,
      "step": 4866
    },
    {
      "epoch": 0.6781857451403888,
      "grad_norm": 0.397986501455307,
      "learning_rate": 1.0737978223329413e-05,
      "loss": 0.07889175415039062,
      "step": 4867
    },
    {
      "epoch": 0.6783250888316031,
      "grad_norm": 0.6775274872779846,
      "learning_rate": 1.0729623931477886e-05,
      "loss": 0.11358642578125,
      "step": 4868
    },
    {
      "epoch": 0.6784644325228175,
      "grad_norm": 0.43099498748779297,
      "learning_rate": 1.0721271699292272e-05,
      "loss": 0.08658599853515625,
      "step": 4869
    },
    {
      "epoch": 0.6786037762140319,
      "grad_norm": 0.5844345688819885,
      "learning_rate": 1.0712921528628258e-05,
      "loss": 0.09583854675292969,
      "step": 4870
    },
    {
      "epoch": 0.6787431199052463,
      "grad_norm": 0.4084152579307556,
      "learning_rate": 1.0704573421341053e-05,
      "loss": 0.10770893096923828,
      "step": 4871
    },
    {
      "epoch": 0.6788824635964607,
      "grad_norm": 0.2944040894508362,
      "learning_rate": 1.0696227379285409e-05,
      "loss": 0.08372306823730469,
      "step": 4872
    },
    {
      "epoch": 0.679021807287675,
      "grad_norm": 0.38478100299835205,
      "learning_rate": 1.0687883404315631e-05,
      "loss": 0.09403800964355469,
      "step": 4873
    },
    {
      "epoch": 0.6791611509788894,
      "grad_norm": 0.3841521441936493,
      "learning_rate": 1.0679541498285568e-05,
      "loss": 0.10695457458496094,
      "step": 4874
    },
    {
      "epoch": 0.6793004946701038,
      "grad_norm": 0.45469430088996887,
      "learning_rate": 1.0671201663048595e-05,
      "loss": 0.12291145324707031,
      "step": 4875
    },
    {
      "epoch": 0.6794398383613182,
      "grad_norm": 0.4401262402534485,
      "learning_rate": 1.0662863900457627e-05,
      "loss": 0.08165168762207031,
      "step": 4876
    },
    {
      "epoch": 0.6795791820525325,
      "grad_norm": 0.3655790090560913,
      "learning_rate": 1.0654528212365127e-05,
      "loss": 0.07804679870605469,
      "step": 4877
    },
    {
      "epoch": 0.6797185257437469,
      "grad_norm": 0.4522641599178314,
      "learning_rate": 1.0646194600623106e-05,
      "loss": 0.0957651138305664,
      "step": 4878
    },
    {
      "epoch": 0.6798578694349613,
      "grad_norm": 0.7998565435409546,
      "learning_rate": 1.0637863067083087e-05,
      "loss": 0.14170455932617188,
      "step": 4879
    },
    {
      "epoch": 0.6799972131261757,
      "grad_norm": 0.36890023946762085,
      "learning_rate": 1.0629533613596162e-05,
      "loss": 0.08062934875488281,
      "step": 4880
    },
    {
      "epoch": 0.68013655681739,
      "grad_norm": 0.3289673626422882,
      "learning_rate": 1.0621206242012936e-05,
      "loss": 0.08096885681152344,
      "step": 4881
    },
    {
      "epoch": 0.6802759005086044,
      "grad_norm": 0.3105328679084778,
      "learning_rate": 1.061288095418355e-05,
      "loss": 0.0748286247253418,
      "step": 4882
    },
    {
      "epoch": 0.6804152441998188,
      "grad_norm": 0.39645084738731384,
      "learning_rate": 1.060455775195771e-05,
      "loss": 0.08550834655761719,
      "step": 4883
    },
    {
      "epoch": 0.6805545878910332,
      "grad_norm": 0.6365534663200378,
      "learning_rate": 1.0596236637184631e-05,
      "loss": 0.106964111328125,
      "step": 4884
    },
    {
      "epoch": 0.6806939315822477,
      "grad_norm": 0.5241995453834534,
      "learning_rate": 1.058791761171309e-05,
      "loss": 0.08218955993652344,
      "step": 4885
    },
    {
      "epoch": 0.680833275273462,
      "grad_norm": 0.25713127851486206,
      "learning_rate": 1.0579600677391375e-05,
      "loss": 0.07313919067382812,
      "step": 4886
    },
    {
      "epoch": 0.6809726189646764,
      "grad_norm": 1.0088403224945068,
      "learning_rate": 1.0571285836067308e-05,
      "loss": 0.1389312744140625,
      "step": 4887
    },
    {
      "epoch": 0.6811119626558908,
      "grad_norm": 0.3513859510421753,
      "learning_rate": 1.0562973089588278e-05,
      "loss": 0.07870197296142578,
      "step": 4888
    },
    {
      "epoch": 0.6812513063471052,
      "grad_norm": 0.4524022340774536,
      "learning_rate": 1.0554662439801167e-05,
      "loss": 0.09143352508544922,
      "step": 4889
    },
    {
      "epoch": 0.6813906500383196,
      "grad_norm": 0.2927952706813812,
      "learning_rate": 1.0546353888552418e-05,
      "loss": 0.06948280334472656,
      "step": 4890
    },
    {
      "epoch": 0.6815299937295339,
      "grad_norm": 0.3145082890987396,
      "learning_rate": 1.0538047437688018e-05,
      "loss": 0.08067893981933594,
      "step": 4891
    },
    {
      "epoch": 0.6816693374207483,
      "grad_norm": 0.2631840705871582,
      "learning_rate": 1.0529743089053452e-05,
      "loss": 0.07291603088378906,
      "step": 4892
    },
    {
      "epoch": 0.6818086811119627,
      "grad_norm": 0.4137323796749115,
      "learning_rate": 1.0521440844493758e-05,
      "loss": 0.09266090393066406,
      "step": 4893
    },
    {
      "epoch": 0.6819480248031771,
      "grad_norm": 0.3972265422344208,
      "learning_rate": 1.0513140705853506e-05,
      "loss": 0.09338665008544922,
      "step": 4894
    },
    {
      "epoch": 0.6820873684943914,
      "grad_norm": 0.354187935590744,
      "learning_rate": 1.0504842674976811e-05,
      "loss": 0.0923757553100586,
      "step": 4895
    },
    {
      "epoch": 0.6822267121856058,
      "grad_norm": 0.31886106729507446,
      "learning_rate": 1.0496546753707295e-05,
      "loss": 0.07342910766601562,
      "step": 4896
    },
    {
      "epoch": 0.6823660558768202,
      "grad_norm": 0.35081911087036133,
      "learning_rate": 1.0488252943888115e-05,
      "loss": 0.09543418884277344,
      "step": 4897
    },
    {
      "epoch": 0.6825053995680346,
      "grad_norm": 0.281780868768692,
      "learning_rate": 1.0479961247361974e-05,
      "loss": 0.08587074279785156,
      "step": 4898
    },
    {
      "epoch": 0.682644743259249,
      "grad_norm": 0.2448401004076004,
      "learning_rate": 1.0471671665971104e-05,
      "loss": 0.06395721435546875,
      "step": 4899
    },
    {
      "epoch": 0.6827840869504633,
      "grad_norm": 0.35675546526908875,
      "learning_rate": 1.0463384201557248e-05,
      "loss": 0.08567047119140625,
      "step": 4900
    },
    {
      "epoch": 0.6829234306416777,
      "grad_norm": 0.4261457920074463,
      "learning_rate": 1.0455098855961705e-05,
      "loss": 0.08635807037353516,
      "step": 4901
    },
    {
      "epoch": 0.6830627743328921,
      "grad_norm": 0.3776802122592926,
      "learning_rate": 1.0446815631025276e-05,
      "loss": 0.09564781188964844,
      "step": 4902
    },
    {
      "epoch": 0.6832021180241065,
      "grad_norm": 0.4416855573654175,
      "learning_rate": 1.0438534528588319e-05,
      "loss": 0.07860565185546875,
      "step": 4903
    },
    {
      "epoch": 0.6833414617153208,
      "grad_norm": 0.2753089368343353,
      "learning_rate": 1.0430255550490686e-05,
      "loss": 0.06847572326660156,
      "step": 4904
    },
    {
      "epoch": 0.6834808054065352,
      "grad_norm": 0.23387877643108368,
      "learning_rate": 1.0421978698571791e-05,
      "loss": 0.06739139556884766,
      "step": 4905
    },
    {
      "epoch": 0.6836201490977496,
      "grad_norm": 0.41215506196022034,
      "learning_rate": 1.0413703974670568e-05,
      "loss": 0.10003089904785156,
      "step": 4906
    },
    {
      "epoch": 0.683759492788964,
      "grad_norm": 0.34148383140563965,
      "learning_rate": 1.0405431380625461e-05,
      "loss": 0.08628654479980469,
      "step": 4907
    },
    {
      "epoch": 0.6838988364801784,
      "grad_norm": 0.23727312684059143,
      "learning_rate": 1.0397160918274447e-05,
      "loss": 0.061435699462890625,
      "step": 4908
    },
    {
      "epoch": 0.6840381801713927,
      "grad_norm": 0.3175593316555023,
      "learning_rate": 1.038889258945504e-05,
      "loss": 0.08549118041992188,
      "step": 4909
    },
    {
      "epoch": 0.6841775238626071,
      "grad_norm": 0.49968087673187256,
      "learning_rate": 1.0380626396004282e-05,
      "loss": 0.10831642150878906,
      "step": 4910
    },
    {
      "epoch": 0.6843168675538215,
      "grad_norm": 0.27669012546539307,
      "learning_rate": 1.0372362339758717e-05,
      "loss": 0.05743885040283203,
      "step": 4911
    },
    {
      "epoch": 0.6844562112450359,
      "grad_norm": 0.505900502204895,
      "learning_rate": 1.0364100422554445e-05,
      "loss": 0.09831809997558594,
      "step": 4912
    },
    {
      "epoch": 0.6845955549362502,
      "grad_norm": 0.6580058336257935,
      "learning_rate": 1.0355840646227063e-05,
      "loss": 0.14297866821289062,
      "step": 4913
    },
    {
      "epoch": 0.6847348986274646,
      "grad_norm": 0.25964322686195374,
      "learning_rate": 1.0347583012611713e-05,
      "loss": 0.062328338623046875,
      "step": 4914
    },
    {
      "epoch": 0.684874242318679,
      "grad_norm": 0.6367183327674866,
      "learning_rate": 1.0339327523543043e-05,
      "loss": 0.13318252563476562,
      "step": 4915
    },
    {
      "epoch": 0.6850135860098934,
      "grad_norm": 0.8033723831176758,
      "learning_rate": 1.0331074180855243e-05,
      "loss": 0.1332406997680664,
      "step": 4916
    },
    {
      "epoch": 0.6851529297011077,
      "grad_norm": 0.4216654300689697,
      "learning_rate": 1.0322822986382026e-05,
      "loss": 0.09020042419433594,
      "step": 4917
    },
    {
      "epoch": 0.6852922733923221,
      "grad_norm": 0.33353284001350403,
      "learning_rate": 1.0314573941956593e-05,
      "loss": 0.08016586303710938,
      "step": 4918
    },
    {
      "epoch": 0.6854316170835365,
      "grad_norm": 0.19543087482452393,
      "learning_rate": 1.0306327049411711e-05,
      "loss": 0.06568145751953125,
      "step": 4919
    },
    {
      "epoch": 0.6855709607747509,
      "grad_norm": 0.26474276185035706,
      "learning_rate": 1.0298082310579653e-05,
      "loss": 0.07256889343261719,
      "step": 4920
    },
    {
      "epoch": 0.6857103044659653,
      "grad_norm": 0.48136138916015625,
      "learning_rate": 1.0289839727292202e-05,
      "loss": 0.08955860137939453,
      "step": 4921
    },
    {
      "epoch": 0.6858496481571796,
      "grad_norm": 0.2271389216184616,
      "learning_rate": 1.0281599301380676e-05,
      "loss": 0.06487083435058594,
      "step": 4922
    },
    {
      "epoch": 0.685988991848394,
      "grad_norm": 0.3940986394882202,
      "learning_rate": 1.0273361034675915e-05,
      "loss": 0.08514213562011719,
      "step": 4923
    },
    {
      "epoch": 0.6861283355396084,
      "grad_norm": 0.3297352194786072,
      "learning_rate": 1.0265124929008272e-05,
      "loss": 0.08451652526855469,
      "step": 4924
    },
    {
      "epoch": 0.6862676792308228,
      "grad_norm": 0.4284960925579071,
      "learning_rate": 1.0256890986207612e-05,
      "loss": 0.09431838989257812,
      "step": 4925
    },
    {
      "epoch": 0.6864070229220373,
      "grad_norm": 0.6568474769592285,
      "learning_rate": 1.0248659208103336e-05,
      "loss": 0.11489295959472656,
      "step": 4926
    },
    {
      "epoch": 0.6865463666132516,
      "grad_norm": 0.3984948694705963,
      "learning_rate": 1.0240429596524361e-05,
      "loss": 0.10499000549316406,
      "step": 4927
    },
    {
      "epoch": 0.686685710304466,
      "grad_norm": 0.34218165278434753,
      "learning_rate": 1.0232202153299118e-05,
      "loss": 0.08030128479003906,
      "step": 4928
    },
    {
      "epoch": 0.6868250539956804,
      "grad_norm": 0.5568341612815857,
      "learning_rate": 1.0223976880255546e-05,
      "loss": 0.10932540893554688,
      "step": 4929
    },
    {
      "epoch": 0.6869643976868948,
      "grad_norm": 0.391722708940506,
      "learning_rate": 1.0215753779221119e-05,
      "loss": 0.07804012298583984,
      "step": 4930
    },
    {
      "epoch": 0.6871037413781091,
      "grad_norm": 0.3797610104084015,
      "learning_rate": 1.0207532852022833e-05,
      "loss": 0.08785676956176758,
      "step": 4931
    },
    {
      "epoch": 0.6872430850693235,
      "grad_norm": 0.2856076955795288,
      "learning_rate": 1.0199314100487175e-05,
      "loss": 0.08537960052490234,
      "step": 4932
    },
    {
      "epoch": 0.6873824287605379,
      "grad_norm": 0.7224351763725281,
      "learning_rate": 1.0191097526440177e-05,
      "loss": 0.12424850463867188,
      "step": 4933
    },
    {
      "epoch": 0.6875217724517523,
      "grad_norm": 0.3226775527000427,
      "learning_rate": 1.0182883131707357e-05,
      "loss": 0.08257007598876953,
      "step": 4934
    },
    {
      "epoch": 0.6876611161429667,
      "grad_norm": 0.44007226824760437,
      "learning_rate": 1.017467091811379e-05,
      "loss": 0.1140604019165039,
      "step": 4935
    },
    {
      "epoch": 0.687800459834181,
      "grad_norm": 0.45145511627197266,
      "learning_rate": 1.0166460887484018e-05,
      "loss": 0.08194923400878906,
      "step": 4936
    },
    {
      "epoch": 0.6879398035253954,
      "grad_norm": 0.34159597754478455,
      "learning_rate": 1.0158253041642132e-05,
      "loss": 0.08581352233886719,
      "step": 4937
    },
    {
      "epoch": 0.6880791472166098,
      "grad_norm": 0.3290710151195526,
      "learning_rate": 1.0150047382411738e-05,
      "loss": 0.0875406265258789,
      "step": 4938
    },
    {
      "epoch": 0.6882184909078242,
      "grad_norm": 0.4611169397830963,
      "learning_rate": 1.0141843911615938e-05,
      "loss": 0.1058807373046875,
      "step": 4939
    },
    {
      "epoch": 0.6883578345990385,
      "grad_norm": 0.46143314242362976,
      "learning_rate": 1.0133642631077348e-05,
      "loss": 0.11905670166015625,
      "step": 4940
    },
    {
      "epoch": 0.6884971782902529,
      "grad_norm": 0.46544548869132996,
      "learning_rate": 1.0125443542618112e-05,
      "loss": 0.09161567687988281,
      "step": 4941
    },
    {
      "epoch": 0.6886365219814673,
      "grad_norm": 0.3057534694671631,
      "learning_rate": 1.0117246648059888e-05,
      "loss": 0.07961463928222656,
      "step": 4942
    },
    {
      "epoch": 0.6887758656726817,
      "grad_norm": 0.363075852394104,
      "learning_rate": 1.0109051949223825e-05,
      "loss": 0.08548927307128906,
      "step": 4943
    },
    {
      "epoch": 0.688915209363896,
      "grad_norm": 0.2729121446609497,
      "learning_rate": 1.0100859447930614e-05,
      "loss": 0.06646919250488281,
      "step": 4944
    },
    {
      "epoch": 0.6890545530551104,
      "grad_norm": 0.44278791546821594,
      "learning_rate": 1.0092669146000422e-05,
      "loss": 0.09829521179199219,
      "step": 4945
    },
    {
      "epoch": 0.6891938967463248,
      "grad_norm": 0.23910470306873322,
      "learning_rate": 1.0084481045252965e-05,
      "loss": 0.06440305709838867,
      "step": 4946
    },
    {
      "epoch": 0.6893332404375392,
      "grad_norm": 0.5476571917533875,
      "learning_rate": 1.0076295147507437e-05,
      "loss": 0.10954856872558594,
      "step": 4947
    },
    {
      "epoch": 0.6894725841287536,
      "grad_norm": 0.3782988488674164,
      "learning_rate": 1.0068111454582565e-05,
      "loss": 0.09658432006835938,
      "step": 4948
    },
    {
      "epoch": 0.6896119278199679,
      "grad_norm": 0.323027104139328,
      "learning_rate": 1.0059929968296597e-05,
      "loss": 0.08370208740234375,
      "step": 4949
    },
    {
      "epoch": 0.6897512715111823,
      "grad_norm": 0.31005731225013733,
      "learning_rate": 1.005175069046724e-05,
      "loss": 0.08263778686523438,
      "step": 4950
    },
    {
      "epoch": 0.6898906152023967,
      "grad_norm": 0.35136526823043823,
      "learning_rate": 1.004357362291175e-05,
      "loss": 0.08499526977539062,
      "step": 4951
    },
    {
      "epoch": 0.6900299588936111,
      "grad_norm": 0.2737633287906647,
      "learning_rate": 1.0035398767446897e-05,
      "loss": 0.0638427734375,
      "step": 4952
    },
    {
      "epoch": 0.6901693025848255,
      "grad_norm": 0.2198205292224884,
      "learning_rate": 1.0027226125888951e-05,
      "loss": 0.07228469848632812,
      "step": 4953
    },
    {
      "epoch": 0.6903086462760398,
      "grad_norm": 0.23611025512218475,
      "learning_rate": 1.0019055700053677e-05,
      "loss": 0.06723785400390625,
      "step": 4954
    },
    {
      "epoch": 0.6904479899672542,
      "grad_norm": 0.5975042581558228,
      "learning_rate": 1.0010887491756347e-05,
      "loss": 0.09978961944580078,
      "step": 4955
    },
    {
      "epoch": 0.6905873336584686,
      "grad_norm": 0.3045758008956909,
      "learning_rate": 1.0002721502811772e-05,
      "loss": 0.08376121520996094,
      "step": 4956
    },
    {
      "epoch": 0.690726677349683,
      "grad_norm": 0.4116871654987335,
      "learning_rate": 9.994557735034227e-06,
      "loss": 0.08355426788330078,
      "step": 4957
    },
    {
      "epoch": 0.6908660210408973,
      "grad_norm": 0.22577494382858276,
      "learning_rate": 9.986396190237526e-06,
      "loss": 0.064727783203125,
      "step": 4958
    },
    {
      "epoch": 0.6910053647321117,
      "grad_norm": 0.6208799481391907,
      "learning_rate": 9.978236870234984e-06,
      "loss": 0.11593437194824219,
      "step": 4959
    },
    {
      "epoch": 0.6911447084233261,
      "grad_norm": 0.267959862947464,
      "learning_rate": 9.970079776839412e-06,
      "loss": 0.07221126556396484,
      "step": 4960
    },
    {
      "epoch": 0.6912840521145405,
      "grad_norm": 0.3604404628276825,
      "learning_rate": 9.961924911863117e-06,
      "loss": 0.08163166046142578,
      "step": 4961
    },
    {
      "epoch": 0.6914233958057548,
      "grad_norm": 0.732986569404602,
      "learning_rate": 9.953772277117933e-06,
      "loss": 0.13184356689453125,
      "step": 4962
    },
    {
      "epoch": 0.6915627394969692,
      "grad_norm": 0.29597803950309753,
      "learning_rate": 9.945621874415197e-06,
      "loss": 0.07835578918457031,
      "step": 4963
    },
    {
      "epoch": 0.6917020831881836,
      "grad_norm": 0.21243169903755188,
      "learning_rate": 9.937473705565728e-06,
      "loss": 0.06528472900390625,
      "step": 4964
    },
    {
      "epoch": 0.691841426879398,
      "grad_norm": 0.5123072266578674,
      "learning_rate": 9.929327772379877e-06,
      "loss": 0.0998992919921875,
      "step": 4965
    },
    {
      "epoch": 0.6919807705706125,
      "grad_norm": 0.5419296622276306,
      "learning_rate": 9.921184076667472e-06,
      "loss": 0.10455513000488281,
      "step": 4966
    },
    {
      "epoch": 0.6921201142618268,
      "grad_norm": 0.2582698464393616,
      "learning_rate": 9.913042620237868e-06,
      "loss": 0.061771392822265625,
      "step": 4967
    },
    {
      "epoch": 0.6922594579530412,
      "grad_norm": 0.44821691513061523,
      "learning_rate": 9.904903404899898e-06,
      "loss": 0.11999893188476562,
      "step": 4968
    },
    {
      "epoch": 0.6923988016442556,
      "grad_norm": 0.20230096578598022,
      "learning_rate": 9.896766432461914e-06,
      "loss": 0.05615043640136719,
      "step": 4969
    },
    {
      "epoch": 0.69253814533547,
      "grad_norm": 0.3491203784942627,
      "learning_rate": 9.88863170473178e-06,
      "loss": 0.08565330505371094,
      "step": 4970
    },
    {
      "epoch": 0.6926774890266844,
      "grad_norm": 0.27427130937576294,
      "learning_rate": 9.880499223516831e-06,
      "loss": 0.0642547607421875,
      "step": 4971
    },
    {
      "epoch": 0.6928168327178987,
      "grad_norm": 0.4668982923030853,
      "learning_rate": 9.872368990623915e-06,
      "loss": 0.09759140014648438,
      "step": 4972
    },
    {
      "epoch": 0.6929561764091131,
      "grad_norm": 0.5656536817550659,
      "learning_rate": 9.864241007859392e-06,
      "loss": 0.09711837768554688,
      "step": 4973
    },
    {
      "epoch": 0.6930955201003275,
      "grad_norm": 0.3404300808906555,
      "learning_rate": 9.856115277029123e-06,
      "loss": 0.08415603637695312,
      "step": 4974
    },
    {
      "epoch": 0.6932348637915419,
      "grad_norm": 0.3858882486820221,
      "learning_rate": 9.84799179993845e-06,
      "loss": 0.09690380096435547,
      "step": 4975
    },
    {
      "epoch": 0.6933742074827562,
      "grad_norm": 0.3579595685005188,
      "learning_rate": 9.839870578392216e-06,
      "loss": 0.07880020141601562,
      "step": 4976
    },
    {
      "epoch": 0.6935135511739706,
      "grad_norm": 0.40616241097450256,
      "learning_rate": 9.83175161419478e-06,
      "loss": 0.09547042846679688,
      "step": 4977
    },
    {
      "epoch": 0.693652894865185,
      "grad_norm": 0.32607316970825195,
      "learning_rate": 9.82363490915e-06,
      "loss": 0.08511924743652344,
      "step": 4978
    },
    {
      "epoch": 0.6937922385563994,
      "grad_norm": 0.2519657015800476,
      "learning_rate": 9.815520465061201e-06,
      "loss": 0.06199455261230469,
      "step": 4979
    },
    {
      "epoch": 0.6939315822476138,
      "grad_norm": 0.6344687342643738,
      "learning_rate": 9.807408283731244e-06,
      "loss": 0.09833145141601562,
      "step": 4980
    },
    {
      "epoch": 0.6940709259388281,
      "grad_norm": 0.29017555713653564,
      "learning_rate": 9.799298366962478e-06,
      "loss": 0.07285594940185547,
      "step": 4981
    },
    {
      "epoch": 0.6942102696300425,
      "grad_norm": 0.4909520149230957,
      "learning_rate": 9.791190716556713e-06,
      "loss": 0.09121131896972656,
      "step": 4982
    },
    {
      "epoch": 0.6943496133212569,
      "grad_norm": 0.5514956116676331,
      "learning_rate": 9.7830853343153e-06,
      "loss": 0.10633182525634766,
      "step": 4983
    },
    {
      "epoch": 0.6944889570124713,
      "grad_norm": 0.44479334354400635,
      "learning_rate": 9.774982222039072e-06,
      "loss": 0.09659957885742188,
      "step": 4984
    },
    {
      "epoch": 0.6946283007036856,
      "grad_norm": 0.3229827880859375,
      "learning_rate": 9.766881381528357e-06,
      "loss": 0.07755231857299805,
      "step": 4985
    },
    {
      "epoch": 0.6947676443949,
      "grad_norm": 0.5126659870147705,
      "learning_rate": 9.758782814582977e-06,
      "loss": 0.08667945861816406,
      "step": 4986
    },
    {
      "epoch": 0.6949069880861144,
      "grad_norm": 0.37956535816192627,
      "learning_rate": 9.750686523002233e-06,
      "loss": 0.08022117614746094,
      "step": 4987
    },
    {
      "epoch": 0.6950463317773288,
      "grad_norm": 0.3261547088623047,
      "learning_rate": 9.742592508584958e-06,
      "loss": 0.0773477554321289,
      "step": 4988
    },
    {
      "epoch": 0.6951856754685432,
      "grad_norm": 0.2754902243614197,
      "learning_rate": 9.734500773129438e-06,
      "loss": 0.06836414337158203,
      "step": 4989
    },
    {
      "epoch": 0.6953250191597575,
      "grad_norm": 0.70970219373703,
      "learning_rate": 9.726411318433482e-06,
      "loss": 0.11392402648925781,
      "step": 4990
    },
    {
      "epoch": 0.6954643628509719,
      "grad_norm": 0.24002085626125336,
      "learning_rate": 9.71832414629439e-06,
      "loss": 0.06633949279785156,
      "step": 4991
    },
    {
      "epoch": 0.6956037065421863,
      "grad_norm": 0.281584769487381,
      "learning_rate": 9.710239258508935e-06,
      "loss": 0.08434677124023438,
      "step": 4992
    },
    {
      "epoch": 0.6957430502334007,
      "grad_norm": 0.5980426073074341,
      "learning_rate": 9.702156656873391e-06,
      "loss": 0.11432933807373047,
      "step": 4993
    },
    {
      "epoch": 0.695882393924615,
      "grad_norm": 0.30097121000289917,
      "learning_rate": 9.694076343183534e-06,
      "loss": 0.08535194396972656,
      "step": 4994
    },
    {
      "epoch": 0.6960217376158294,
      "grad_norm": 0.5965678691864014,
      "learning_rate": 9.685998319234634e-06,
      "loss": 0.08912277221679688,
      "step": 4995
    },
    {
      "epoch": 0.6961610813070438,
      "grad_norm": 1.1033999919891357,
      "learning_rate": 9.677922586821434e-06,
      "loss": 0.1808605194091797,
      "step": 4996
    },
    {
      "epoch": 0.6963004249982582,
      "grad_norm": 0.40111514925956726,
      "learning_rate": 9.669849147738171e-06,
      "loss": 0.09754276275634766,
      "step": 4997
    },
    {
      "epoch": 0.6964397686894725,
      "grad_norm": 0.17840033769607544,
      "learning_rate": 9.661778003778583e-06,
      "loss": 0.055922508239746094,
      "step": 4998
    },
    {
      "epoch": 0.6965791123806869,
      "grad_norm": 0.6950664520263672,
      "learning_rate": 9.653709156735908e-06,
      "loss": 0.10890865325927734,
      "step": 4999
    },
    {
      "epoch": 0.6967184560719013,
      "grad_norm": 0.48984822630882263,
      "learning_rate": 9.64564260840284e-06,
      "loss": 0.11248588562011719,
      "step": 5000
    },
    {
      "epoch": 0.6968577997631157,
      "grad_norm": 0.43112874031066895,
      "learning_rate": 9.63757836057159e-06,
      "loss": 0.11424636840820312,
      "step": 5001
    },
    {
      "epoch": 0.6969971434543301,
      "grad_norm": 0.3635101616382599,
      "learning_rate": 9.629516415033859e-06,
      "loss": 0.06601142883300781,
      "step": 5002
    },
    {
      "epoch": 0.6971364871455444,
      "grad_norm": 0.35844650864601135,
      "learning_rate": 9.621456773580817e-06,
      "loss": 0.07467079162597656,
      "step": 5003
    },
    {
      "epoch": 0.6972758308367588,
      "grad_norm": 0.5324035286903381,
      "learning_rate": 9.613399438003128e-06,
      "loss": 0.09906578063964844,
      "step": 5004
    },
    {
      "epoch": 0.6974151745279732,
      "grad_norm": 0.35989972949028015,
      "learning_rate": 9.605344410090954e-06,
      "loss": 0.07703399658203125,
      "step": 5005
    },
    {
      "epoch": 0.6975545182191877,
      "grad_norm": 0.5261275768280029,
      "learning_rate": 9.597291691633942e-06,
      "loss": 0.09986114501953125,
      "step": 5006
    },
    {
      "epoch": 0.6976938619104021,
      "grad_norm": 0.2537088692188263,
      "learning_rate": 9.589241284421221e-06,
      "loss": 0.0699310302734375,
      "step": 5007
    },
    {
      "epoch": 0.6978332056016164,
      "grad_norm": 0.40790513157844543,
      "learning_rate": 9.581193190241398e-06,
      "loss": 0.09140586853027344,
      "step": 5008
    },
    {
      "epoch": 0.6979725492928308,
      "grad_norm": 0.37301063537597656,
      "learning_rate": 9.57314741088258e-06,
      "loss": 0.09044075012207031,
      "step": 5009
    },
    {
      "epoch": 0.6981118929840452,
      "grad_norm": 0.3905682861804962,
      "learning_rate": 9.565103948132368e-06,
      "loss": 0.08413505554199219,
      "step": 5010
    },
    {
      "epoch": 0.6982512366752596,
      "grad_norm": 0.27527716755867004,
      "learning_rate": 9.557062803777817e-06,
      "loss": 0.07195281982421875,
      "step": 5011
    },
    {
      "epoch": 0.698390580366474,
      "grad_norm": 0.2630101144313812,
      "learning_rate": 9.549023979605503e-06,
      "loss": 0.070037841796875,
      "step": 5012
    },
    {
      "epoch": 0.6985299240576883,
      "grad_norm": 0.40293508768081665,
      "learning_rate": 9.540987477401454e-06,
      "loss": 0.07933235168457031,
      "step": 5013
    },
    {
      "epoch": 0.6986692677489027,
      "grad_norm": 0.38914942741394043,
      "learning_rate": 9.53295329895121e-06,
      "loss": 0.08435249328613281,
      "step": 5014
    },
    {
      "epoch": 0.6988086114401171,
      "grad_norm": 0.40008339285850525,
      "learning_rate": 9.52492144603977e-06,
      "loss": 0.07492542266845703,
      "step": 5015
    },
    {
      "epoch": 0.6989479551313315,
      "grad_norm": 0.5118096470832825,
      "learning_rate": 9.516891920451634e-06,
      "loss": 0.10499382019042969,
      "step": 5016
    },
    {
      "epoch": 0.6990872988225458,
      "grad_norm": 0.4209776520729065,
      "learning_rate": 9.50886472397079e-06,
      "loss": 0.09214401245117188,
      "step": 5017
    },
    {
      "epoch": 0.6992266425137602,
      "grad_norm": 0.3492922782897949,
      "learning_rate": 9.500839858380684e-06,
      "loss": 0.08222389221191406,
      "step": 5018
    },
    {
      "epoch": 0.6993659862049746,
      "grad_norm": 0.39044255018234253,
      "learning_rate": 9.492817325464256e-06,
      "loss": 0.08613777160644531,
      "step": 5019
    },
    {
      "epoch": 0.699505329896189,
      "grad_norm": 0.4876403510570526,
      "learning_rate": 9.484797127003942e-06,
      "loss": 0.09149360656738281,
      "step": 5020
    },
    {
      "epoch": 0.6996446735874033,
      "grad_norm": 0.25706058740615845,
      "learning_rate": 9.476779264781633e-06,
      "loss": 0.0632781982421875,
      "step": 5021
    },
    {
      "epoch": 0.6997840172786177,
      "grad_norm": 0.3330318331718445,
      "learning_rate": 9.468763740578721e-06,
      "loss": 0.07349395751953125,
      "step": 5022
    },
    {
      "epoch": 0.6999233609698321,
      "grad_norm": 0.6070712804794312,
      "learning_rate": 9.460750556176085e-06,
      "loss": 0.10130786895751953,
      "step": 5023
    },
    {
      "epoch": 0.7000627046610465,
      "grad_norm": 0.46403488516807556,
      "learning_rate": 9.452739713354055e-06,
      "loss": 0.08974075317382812,
      "step": 5024
    },
    {
      "epoch": 0.7002020483522609,
      "grad_norm": 0.5460727214813232,
      "learning_rate": 9.444731213892458e-06,
      "loss": 0.10877418518066406,
      "step": 5025
    },
    {
      "epoch": 0.7003413920434752,
      "grad_norm": 0.5589982867240906,
      "learning_rate": 9.436725059570605e-06,
      "loss": 0.08188438415527344,
      "step": 5026
    },
    {
      "epoch": 0.7004807357346896,
      "grad_norm": 0.4819416403770447,
      "learning_rate": 9.428721252167286e-06,
      "loss": 0.10689735412597656,
      "step": 5027
    },
    {
      "epoch": 0.700620079425904,
      "grad_norm": 0.8366943001747131,
      "learning_rate": 9.420719793460758e-06,
      "loss": 0.14647865295410156,
      "step": 5028
    },
    {
      "epoch": 0.7007594231171184,
      "grad_norm": 0.5976685285568237,
      "learning_rate": 9.412720685228755e-06,
      "loss": 0.11793899536132812,
      "step": 5029
    },
    {
      "epoch": 0.7008987668083327,
      "grad_norm": 0.273333877325058,
      "learning_rate": 9.404723929248507e-06,
      "loss": 0.07473564147949219,
      "step": 5030
    },
    {
      "epoch": 0.7010381104995471,
      "grad_norm": 0.5806134939193726,
      "learning_rate": 9.396729527296712e-06,
      "loss": 0.08838844299316406,
      "step": 5031
    },
    {
      "epoch": 0.7011774541907615,
      "grad_norm": 0.2642543613910675,
      "learning_rate": 9.388737481149534e-06,
      "loss": 0.06691360473632812,
      "step": 5032
    },
    {
      "epoch": 0.7013167978819759,
      "grad_norm": 0.4547598361968994,
      "learning_rate": 9.380747792582635e-06,
      "loss": 0.08000564575195312,
      "step": 5033
    },
    {
      "epoch": 0.7014561415731903,
      "grad_norm": 0.38699135184288025,
      "learning_rate": 9.372760463371127e-06,
      "loss": 0.0940399169921875,
      "step": 5034
    },
    {
      "epoch": 0.7015954852644046,
      "grad_norm": 0.4875674545764923,
      "learning_rate": 9.364775495289628e-06,
      "loss": 0.0942850112915039,
      "step": 5035
    },
    {
      "epoch": 0.701734828955619,
      "grad_norm": 0.34513965249061584,
      "learning_rate": 9.3567928901122e-06,
      "loss": 0.07867431640625,
      "step": 5036
    },
    {
      "epoch": 0.7018741726468334,
      "grad_norm": 0.5730146169662476,
      "learning_rate": 9.348812649612404e-06,
      "loss": 0.09345674514770508,
      "step": 5037
    },
    {
      "epoch": 0.7020135163380478,
      "grad_norm": 0.46924862265586853,
      "learning_rate": 9.340834775563275e-06,
      "loss": 0.11110115051269531,
      "step": 5038
    },
    {
      "epoch": 0.7021528600292621,
      "grad_norm": 0.41912055015563965,
      "learning_rate": 9.332859269737303e-06,
      "loss": 0.09215927124023438,
      "step": 5039
    },
    {
      "epoch": 0.7022922037204765,
      "grad_norm": 0.3693384826183319,
      "learning_rate": 9.32488613390646e-06,
      "loss": 0.06923675537109375,
      "step": 5040
    },
    {
      "epoch": 0.7024315474116909,
      "grad_norm": 0.34351763129234314,
      "learning_rate": 9.316915369842201e-06,
      "loss": 0.07317924499511719,
      "step": 5041
    },
    {
      "epoch": 0.7025708911029053,
      "grad_norm": 0.3105156421661377,
      "learning_rate": 9.308946979315456e-06,
      "loss": 0.0794219970703125,
      "step": 5042
    },
    {
      "epoch": 0.7027102347941196,
      "grad_norm": 0.3036092221736908,
      "learning_rate": 9.300980964096604e-06,
      "loss": 0.0716257095336914,
      "step": 5043
    },
    {
      "epoch": 0.702849578485334,
      "grad_norm": 0.3073762357234955,
      "learning_rate": 9.293017325955524e-06,
      "loss": 0.07764244079589844,
      "step": 5044
    },
    {
      "epoch": 0.7029889221765484,
      "grad_norm": 0.41103121638298035,
      "learning_rate": 9.285056066661547e-06,
      "loss": 0.09472084045410156,
      "step": 5045
    },
    {
      "epoch": 0.7031282658677629,
      "grad_norm": 0.23458358645439148,
      "learning_rate": 9.277097187983489e-06,
      "loss": 0.07300376892089844,
      "step": 5046
    },
    {
      "epoch": 0.7032676095589773,
      "grad_norm": 0.45855510234832764,
      "learning_rate": 9.269140691689622e-06,
      "loss": 0.09363555908203125,
      "step": 5047
    },
    {
      "epoch": 0.7034069532501916,
      "grad_norm": 0.6549604535102844,
      "learning_rate": 9.261186579547703e-06,
      "loss": 0.09659004211425781,
      "step": 5048
    },
    {
      "epoch": 0.703546296941406,
      "grad_norm": 0.6095008850097656,
      "learning_rate": 9.253234853324968e-06,
      "loss": 0.12381935119628906,
      "step": 5049
    },
    {
      "epoch": 0.7036856406326204,
      "grad_norm": 0.47968077659606934,
      "learning_rate": 9.245285514788082e-06,
      "loss": 0.10022354125976562,
      "step": 5050
    },
    {
      "epoch": 0.7038249843238348,
      "grad_norm": 0.5633729100227356,
      "learning_rate": 9.237338565703222e-06,
      "loss": 0.12637710571289062,
      "step": 5051
    },
    {
      "epoch": 0.7039643280150492,
      "grad_norm": 0.46615126729011536,
      "learning_rate": 9.229394007836017e-06,
      "loss": 0.10266304016113281,
      "step": 5052
    },
    {
      "epoch": 0.7041036717062635,
      "grad_norm": 0.4680469036102295,
      "learning_rate": 9.221451842951572e-06,
      "loss": 0.09732437133789062,
      "step": 5053
    },
    {
      "epoch": 0.7042430153974779,
      "grad_norm": 0.6042613983154297,
      "learning_rate": 9.21351207281445e-06,
      "loss": 0.11707878112792969,
      "step": 5054
    },
    {
      "epoch": 0.7043823590886923,
      "grad_norm": 0.4946025013923645,
      "learning_rate": 9.205574699188677e-06,
      "loss": 0.09771537780761719,
      "step": 5055
    },
    {
      "epoch": 0.7045217027799067,
      "grad_norm": 0.3503205180168152,
      "learning_rate": 9.197639723837775e-06,
      "loss": 0.08000564575195312,
      "step": 5056
    },
    {
      "epoch": 0.704661046471121,
      "grad_norm": 0.3377445340156555,
      "learning_rate": 9.189707148524697e-06,
      "loss": 0.06848335266113281,
      "step": 5057
    },
    {
      "epoch": 0.7048003901623354,
      "grad_norm": 0.46059450507164,
      "learning_rate": 9.181776975011882e-06,
      "loss": 0.1086273193359375,
      "step": 5058
    },
    {
      "epoch": 0.7049397338535498,
      "grad_norm": 0.2698400616645813,
      "learning_rate": 9.173849205061251e-06,
      "loss": 0.0711674690246582,
      "step": 5059
    },
    {
      "epoch": 0.7050790775447642,
      "grad_norm": 0.21385367214679718,
      "learning_rate": 9.165923840434162e-06,
      "loss": 0.06261634826660156,
      "step": 5060
    },
    {
      "epoch": 0.7052184212359786,
      "grad_norm": 0.5526517033576965,
      "learning_rate": 9.15800088289144e-06,
      "loss": 0.1278858184814453,
      "step": 5061
    },
    {
      "epoch": 0.7053577649271929,
      "grad_norm": 0.35767436027526855,
      "learning_rate": 9.150080334193394e-06,
      "loss": 0.09249687194824219,
      "step": 5062
    },
    {
      "epoch": 0.7054971086184073,
      "grad_norm": 0.2869773209095001,
      "learning_rate": 9.142162196099799e-06,
      "loss": 0.07478904724121094,
      "step": 5063
    },
    {
      "epoch": 0.7056364523096217,
      "grad_norm": 0.5399680733680725,
      "learning_rate": 9.134246470369868e-06,
      "loss": 0.09094667434692383,
      "step": 5064
    },
    {
      "epoch": 0.7057757960008361,
      "grad_norm": 0.31981995701789856,
      "learning_rate": 9.126333158762309e-06,
      "loss": 0.0757455825805664,
      "step": 5065
    },
    {
      "epoch": 0.7059151396920504,
      "grad_norm": 0.2379252314567566,
      "learning_rate": 9.118422263035264e-06,
      "loss": 0.07272195816040039,
      "step": 5066
    },
    {
      "epoch": 0.7060544833832648,
      "grad_norm": 0.29845714569091797,
      "learning_rate": 9.110513784946368e-06,
      "loss": 0.0746011734008789,
      "step": 5067
    },
    {
      "epoch": 0.7061938270744792,
      "grad_norm": 0.3006128966808319,
      "learning_rate": 9.102607726252692e-06,
      "loss": 0.07023334503173828,
      "step": 5068
    },
    {
      "epoch": 0.7063331707656936,
      "grad_norm": 0.5126835107803345,
      "learning_rate": 9.094704088710788e-06,
      "loss": 0.11052894592285156,
      "step": 5069
    },
    {
      "epoch": 0.706472514456908,
      "grad_norm": 0.36278772354125977,
      "learning_rate": 9.08680287407667e-06,
      "loss": 0.07686328887939453,
      "step": 5070
    },
    {
      "epoch": 0.7066118581481223,
      "grad_norm": 0.4253590404987335,
      "learning_rate": 9.078904084105802e-06,
      "loss": 0.10579395294189453,
      "step": 5071
    },
    {
      "epoch": 0.7067512018393367,
      "grad_norm": 0.604463517665863,
      "learning_rate": 9.071007720553104e-06,
      "loss": 0.09521102905273438,
      "step": 5072
    },
    {
      "epoch": 0.7068905455305511,
      "grad_norm": 0.4090389311313629,
      "learning_rate": 9.06311378517298e-06,
      "loss": 0.07865142822265625,
      "step": 5073
    },
    {
      "epoch": 0.7070298892217655,
      "grad_norm": 0.2793520987033844,
      "learning_rate": 9.055222279719284e-06,
      "loss": 0.06829833984375,
      "step": 5074
    },
    {
      "epoch": 0.7071692329129798,
      "grad_norm": 0.6008466482162476,
      "learning_rate": 9.047333205945318e-06,
      "loss": 0.12237358093261719,
      "step": 5075
    },
    {
      "epoch": 0.7073085766041942,
      "grad_norm": 0.4721815586090088,
      "learning_rate": 9.039446565603868e-06,
      "loss": 0.09187817573547363,
      "step": 5076
    },
    {
      "epoch": 0.7074479202954086,
      "grad_norm": 0.3796568810939789,
      "learning_rate": 9.03156236044715e-06,
      "loss": 0.0887298583984375,
      "step": 5077
    },
    {
      "epoch": 0.707587263986623,
      "grad_norm": 0.346814900636673,
      "learning_rate": 9.023680592226868e-06,
      "loss": 0.06937313079833984,
      "step": 5078
    },
    {
      "epoch": 0.7077266076778373,
      "grad_norm": 0.32224252820014954,
      "learning_rate": 9.015801262694157e-06,
      "loss": 0.08800125122070312,
      "step": 5079
    },
    {
      "epoch": 0.7078659513690517,
      "grad_norm": 0.3186229467391968,
      "learning_rate": 9.007924373599634e-06,
      "loss": 0.08640098571777344,
      "step": 5080
    },
    {
      "epoch": 0.7080052950602661,
      "grad_norm": 0.2802046239376068,
      "learning_rate": 9.000049926693375e-06,
      "loss": 0.06688499450683594,
      "step": 5081
    },
    {
      "epoch": 0.7081446387514805,
      "grad_norm": 0.46237489581108093,
      "learning_rate": 8.992177923724876e-06,
      "loss": 0.1093902587890625,
      "step": 5082
    },
    {
      "epoch": 0.7082839824426949,
      "grad_norm": 0.6200442314147949,
      "learning_rate": 8.98430836644313e-06,
      "loss": 0.10738563537597656,
      "step": 5083
    },
    {
      "epoch": 0.7084233261339092,
      "grad_norm": 0.42910531163215637,
      "learning_rate": 8.97644125659657e-06,
      "loss": 0.08350181579589844,
      "step": 5084
    },
    {
      "epoch": 0.7085626698251236,
      "grad_norm": 0.3607664108276367,
      "learning_rate": 8.968576595933098e-06,
      "loss": 0.07746124267578125,
      "step": 5085
    },
    {
      "epoch": 0.7087020135163381,
      "grad_norm": 0.3402417004108429,
      "learning_rate": 8.960714386200056e-06,
      "loss": 0.07793235778808594,
      "step": 5086
    },
    {
      "epoch": 0.7088413572075525,
      "grad_norm": 0.45876577496528625,
      "learning_rate": 8.95285462914424e-06,
      "loss": 0.0984954833984375,
      "step": 5087
    },
    {
      "epoch": 0.7089807008987669,
      "grad_norm": 0.36876380443573,
      "learning_rate": 8.94499732651192e-06,
      "loss": 0.08346176147460938,
      "step": 5088
    },
    {
      "epoch": 0.7091200445899812,
      "grad_norm": 0.43128451704978943,
      "learning_rate": 8.937142480048797e-06,
      "loss": 0.08225059509277344,
      "step": 5089
    },
    {
      "epoch": 0.7092593882811956,
      "grad_norm": 0.3870275020599365,
      "learning_rate": 8.929290091500045e-06,
      "loss": 0.09326362609863281,
      "step": 5090
    },
    {
      "epoch": 0.70939873197241,
      "grad_norm": 0.4229520857334137,
      "learning_rate": 8.921440162610295e-06,
      "loss": 0.07924079895019531,
      "step": 5091
    },
    {
      "epoch": 0.7095380756636244,
      "grad_norm": 0.3940028250217438,
      "learning_rate": 8.913592695123613e-06,
      "loss": 0.07867813110351562,
      "step": 5092
    },
    {
      "epoch": 0.7096774193548387,
      "grad_norm": 0.40666428208351135,
      "learning_rate": 8.905747690783517e-06,
      "loss": 0.08480262756347656,
      "step": 5093
    },
    {
      "epoch": 0.7098167630460531,
      "grad_norm": 0.29039615392684937,
      "learning_rate": 8.897905151333002e-06,
      "loss": 0.0776824951171875,
      "step": 5094
    },
    {
      "epoch": 0.7099561067372675,
      "grad_norm": 0.34529373049736023,
      "learning_rate": 8.890065078514503e-06,
      "loss": 0.07745933532714844,
      "step": 5095
    },
    {
      "epoch": 0.7100954504284819,
      "grad_norm": 0.29522383213043213,
      "learning_rate": 8.882227474069892e-06,
      "loss": 0.07621955871582031,
      "step": 5096
    },
    {
      "epoch": 0.7102347941196963,
      "grad_norm": 0.4033275842666626,
      "learning_rate": 8.874392339740518e-06,
      "loss": 0.09445571899414062,
      "step": 5097
    },
    {
      "epoch": 0.7103741378109106,
      "grad_norm": 0.635503888130188,
      "learning_rate": 8.866559677267162e-06,
      "loss": 0.12297248840332031,
      "step": 5098
    },
    {
      "epoch": 0.710513481502125,
      "grad_norm": 0.28513672947883606,
      "learning_rate": 8.858729488390068e-06,
      "loss": 0.06948471069335938,
      "step": 5099
    },
    {
      "epoch": 0.7106528251933394,
      "grad_norm": 0.3871840238571167,
      "learning_rate": 8.850901774848916e-06,
      "loss": 0.10392379760742188,
      "step": 5100
    },
    {
      "epoch": 0.7107921688845538,
      "grad_norm": 0.3118007779121399,
      "learning_rate": 8.843076538382853e-06,
      "loss": 0.06787872314453125,
      "step": 5101
    },
    {
      "epoch": 0.7109315125757681,
      "grad_norm": 0.3747655153274536,
      "learning_rate": 8.835253780730472e-06,
      "loss": 0.0836944580078125,
      "step": 5102
    },
    {
      "epoch": 0.7110708562669825,
      "grad_norm": 0.35966306924819946,
      "learning_rate": 8.827433503629805e-06,
      "loss": 0.08930683135986328,
      "step": 5103
    },
    {
      "epoch": 0.7112101999581969,
      "grad_norm": 0.6194000244140625,
      "learning_rate": 8.819615708818335e-06,
      "loss": 0.10882568359375,
      "step": 5104
    },
    {
      "epoch": 0.7113495436494113,
      "grad_norm": 0.32348307967185974,
      "learning_rate": 8.811800398032999e-06,
      "loss": 0.07853364944458008,
      "step": 5105
    },
    {
      "epoch": 0.7114888873406257,
      "grad_norm": 0.2370113730430603,
      "learning_rate": 8.803987573010191e-06,
      "loss": 0.0627756118774414,
      "step": 5106
    },
    {
      "epoch": 0.71162823103184,
      "grad_norm": 0.6447005867958069,
      "learning_rate": 8.796177235485736e-06,
      "loss": 0.0991220474243164,
      "step": 5107
    },
    {
      "epoch": 0.7117675747230544,
      "grad_norm": 0.46245312690734863,
      "learning_rate": 8.788369387194904e-06,
      "loss": 0.10154342651367188,
      "step": 5108
    },
    {
      "epoch": 0.7119069184142688,
      "grad_norm": 0.5178766250610352,
      "learning_rate": 8.78056402987243e-06,
      "loss": 0.09460639953613281,
      "step": 5109
    },
    {
      "epoch": 0.7120462621054832,
      "grad_norm": 0.2147335708141327,
      "learning_rate": 8.772761165252488e-06,
      "loss": 0.06746482849121094,
      "step": 5110
    },
    {
      "epoch": 0.7121856057966975,
      "grad_norm": 0.3610585927963257,
      "learning_rate": 8.76496079506869e-06,
      "loss": 0.06991767883300781,
      "step": 5111
    },
    {
      "epoch": 0.7123249494879119,
      "grad_norm": 0.544852077960968,
      "learning_rate": 8.757162921054099e-06,
      "loss": 0.10314083099365234,
      "step": 5112
    },
    {
      "epoch": 0.7124642931791263,
      "grad_norm": 0.6036769151687622,
      "learning_rate": 8.749367544941238e-06,
      "loss": 0.09000015258789062,
      "step": 5113
    },
    {
      "epoch": 0.7126036368703407,
      "grad_norm": 0.289544016122818,
      "learning_rate": 8.741574668462053e-06,
      "loss": 0.07390594482421875,
      "step": 5114
    },
    {
      "epoch": 0.712742980561555,
      "grad_norm": 0.7487357258796692,
      "learning_rate": 8.733784293347934e-06,
      "loss": 0.12241554260253906,
      "step": 5115
    },
    {
      "epoch": 0.7128823242527694,
      "grad_norm": 0.361358106136322,
      "learning_rate": 8.725996421329733e-06,
      "loss": 0.08399295806884766,
      "step": 5116
    },
    {
      "epoch": 0.7130216679439838,
      "grad_norm": 0.4703681170940399,
      "learning_rate": 8.718211054137744e-06,
      "loss": 0.07787322998046875,
      "step": 5117
    },
    {
      "epoch": 0.7131610116351982,
      "grad_norm": 0.47227171063423157,
      "learning_rate": 8.710428193501692e-06,
      "loss": 0.10083770751953125,
      "step": 5118
    },
    {
      "epoch": 0.7133003553264126,
      "grad_norm": 0.3163149356842041,
      "learning_rate": 8.702647841150743e-06,
      "loss": 0.07923126220703125,
      "step": 5119
    },
    {
      "epoch": 0.7134396990176269,
      "grad_norm": 0.43398594856262207,
      "learning_rate": 8.694869998813527e-06,
      "loss": 0.08489704132080078,
      "step": 5120
    },
    {
      "epoch": 0.7135790427088413,
      "grad_norm": 0.40075892210006714,
      "learning_rate": 8.68709466821809e-06,
      "loss": 0.07426166534423828,
      "step": 5121
    },
    {
      "epoch": 0.7137183864000557,
      "grad_norm": 0.40390872955322266,
      "learning_rate": 8.67932185109194e-06,
      "loss": 0.08249330520629883,
      "step": 5122
    },
    {
      "epoch": 0.7138577300912701,
      "grad_norm": 0.7273439168930054,
      "learning_rate": 8.671551549162025e-06,
      "loss": 0.12781524658203125,
      "step": 5123
    },
    {
      "epoch": 0.7139970737824844,
      "grad_norm": 0.39760667085647583,
      "learning_rate": 8.663783764154726e-06,
      "loss": 0.07667922973632812,
      "step": 5124
    },
    {
      "epoch": 0.7141364174736988,
      "grad_norm": 0.6939367651939392,
      "learning_rate": 8.656018497795855e-06,
      "loss": 0.11000251770019531,
      "step": 5125
    },
    {
      "epoch": 0.7142757611649132,
      "grad_norm": 0.3765064775943756,
      "learning_rate": 8.648255751810686e-06,
      "loss": 0.09102058410644531,
      "step": 5126
    },
    {
      "epoch": 0.7144151048561277,
      "grad_norm": 0.23896130919456482,
      "learning_rate": 8.640495527923931e-06,
      "loss": 0.06025886535644531,
      "step": 5127
    },
    {
      "epoch": 0.7145544485473421,
      "grad_norm": 0.4112388789653778,
      "learning_rate": 8.632737827859729e-06,
      "loss": 0.08991622924804688,
      "step": 5128
    },
    {
      "epoch": 0.7146937922385564,
      "grad_norm": 0.442043274641037,
      "learning_rate": 8.624982653341656e-06,
      "loss": 0.09262752532958984,
      "step": 5129
    },
    {
      "epoch": 0.7148331359297708,
      "grad_norm": 0.3844919204711914,
      "learning_rate": 8.61723000609274e-06,
      "loss": 0.07954025268554688,
      "step": 5130
    },
    {
      "epoch": 0.7149724796209852,
      "grad_norm": 0.6022707223892212,
      "learning_rate": 8.609479887835453e-06,
      "loss": 0.10756492614746094,
      "step": 5131
    },
    {
      "epoch": 0.7151118233121996,
      "grad_norm": 0.4081715941429138,
      "learning_rate": 8.601732300291674e-06,
      "loss": 0.07703113555908203,
      "step": 5132
    },
    {
      "epoch": 0.715251167003414,
      "grad_norm": 0.7200596332550049,
      "learning_rate": 8.593987245182754e-06,
      "loss": 0.12595748901367188,
      "step": 5133
    },
    {
      "epoch": 0.7153905106946283,
      "grad_norm": 0.4747401773929596,
      "learning_rate": 8.586244724229471e-06,
      "loss": 0.114227294921875,
      "step": 5134
    },
    {
      "epoch": 0.7155298543858427,
      "grad_norm": 0.2698807120323181,
      "learning_rate": 8.57850473915203e-06,
      "loss": 0.07819461822509766,
      "step": 5135
    },
    {
      "epoch": 0.7156691980770571,
      "grad_norm": 0.317402720451355,
      "learning_rate": 8.57076729167007e-06,
      "loss": 0.06857109069824219,
      "step": 5136
    },
    {
      "epoch": 0.7158085417682715,
      "grad_norm": 0.6743189692497253,
      "learning_rate": 8.563032383502685e-06,
      "loss": 0.12783050537109375,
      "step": 5137
    },
    {
      "epoch": 0.7159478854594858,
      "grad_norm": 1.5394914150238037,
      "learning_rate": 8.555300016368403e-06,
      "loss": 0.15388917922973633,
      "step": 5138
    },
    {
      "epoch": 0.7160872291507002,
      "grad_norm": 0.3642120659351349,
      "learning_rate": 8.547570191985168e-06,
      "loss": 0.0774688720703125,
      "step": 5139
    },
    {
      "epoch": 0.7162265728419146,
      "grad_norm": 0.36346563696861267,
      "learning_rate": 8.539842912070367e-06,
      "loss": 0.08487510681152344,
      "step": 5140
    },
    {
      "epoch": 0.716365916533129,
      "grad_norm": 0.31676119565963745,
      "learning_rate": 8.532118178340829e-06,
      "loss": 0.0726633071899414,
      "step": 5141
    },
    {
      "epoch": 0.7165052602243434,
      "grad_norm": 0.4287249743938446,
      "learning_rate": 8.524395992512827e-06,
      "loss": 0.09098434448242188,
      "step": 5142
    },
    {
      "epoch": 0.7166446039155577,
      "grad_norm": 0.32714638113975525,
      "learning_rate": 8.516676356302031e-06,
      "loss": 0.08021354675292969,
      "step": 5143
    },
    {
      "epoch": 0.7167839476067721,
      "grad_norm": 0.42301255464553833,
      "learning_rate": 8.508959271423589e-06,
      "loss": 0.09534835815429688,
      "step": 5144
    },
    {
      "epoch": 0.7169232912979865,
      "grad_norm": 0.7229560017585754,
      "learning_rate": 8.501244739592045e-06,
      "loss": 0.10836982727050781,
      "step": 5145
    },
    {
      "epoch": 0.7170626349892009,
      "grad_norm": 0.650001585483551,
      "learning_rate": 8.493532762521406e-06,
      "loss": 0.10709941387176514,
      "step": 5146
    },
    {
      "epoch": 0.7172019786804152,
      "grad_norm": 0.3265645503997803,
      "learning_rate": 8.485823341925084e-06,
      "loss": 0.07647895812988281,
      "step": 5147
    },
    {
      "epoch": 0.7173413223716296,
      "grad_norm": 0.30761218070983887,
      "learning_rate": 8.47811647951594e-06,
      "loss": 0.08565711975097656,
      "step": 5148
    },
    {
      "epoch": 0.717480666062844,
      "grad_norm": 0.6360039710998535,
      "learning_rate": 8.470412177006281e-06,
      "loss": 0.11336040496826172,
      "step": 5149
    },
    {
      "epoch": 0.7176200097540584,
      "grad_norm": 0.48578715324401855,
      "learning_rate": 8.462710436107796e-06,
      "loss": 0.10680580139160156,
      "step": 5150
    },
    {
      "epoch": 0.7177593534452728,
      "grad_norm": 0.49643251299858093,
      "learning_rate": 8.455011258531653e-06,
      "loss": 0.09367179870605469,
      "step": 5151
    },
    {
      "epoch": 0.7178986971364871,
      "grad_norm": 0.3269549012184143,
      "learning_rate": 8.44731464598843e-06,
      "loss": 0.07808589935302734,
      "step": 5152
    },
    {
      "epoch": 0.7180380408277015,
      "grad_norm": 1.104999303817749,
      "learning_rate": 8.439620600188147e-06,
      "loss": 0.1510181427001953,
      "step": 5153
    },
    {
      "epoch": 0.7181773845189159,
      "grad_norm": 0.5524479150772095,
      "learning_rate": 8.431929122840234e-06,
      "loss": 0.11038398742675781,
      "step": 5154
    },
    {
      "epoch": 0.7183167282101303,
      "grad_norm": 0.3176952302455902,
      "learning_rate": 8.424240215653571e-06,
      "loss": 0.082275390625,
      "step": 5155
    },
    {
      "epoch": 0.7184560719013446,
      "grad_norm": 0.4152560532093048,
      "learning_rate": 8.416553880336456e-06,
      "loss": 0.09180355072021484,
      "step": 5156
    },
    {
      "epoch": 0.718595415592559,
      "grad_norm": 0.3988531827926636,
      "learning_rate": 8.408870118596606e-06,
      "loss": 0.08794403076171875,
      "step": 5157
    },
    {
      "epoch": 0.7187347592837734,
      "grad_norm": 0.48612943291664124,
      "learning_rate": 8.401188932141184e-06,
      "loss": 0.11799430847167969,
      "step": 5158
    },
    {
      "epoch": 0.7188741029749878,
      "grad_norm": 0.2978157103061676,
      "learning_rate": 8.393510322676784e-06,
      "loss": 0.07317352294921875,
      "step": 5159
    },
    {
      "epoch": 0.7190134466662021,
      "grad_norm": 0.4608181118965149,
      "learning_rate": 8.385834291909409e-06,
      "loss": 0.10091590881347656,
      "step": 5160
    },
    {
      "epoch": 0.7191527903574165,
      "grad_norm": 0.2656065821647644,
      "learning_rate": 8.378160841544493e-06,
      "loss": 0.06689071655273438,
      "step": 5161
    },
    {
      "epoch": 0.7192921340486309,
      "grad_norm": 0.5172542929649353,
      "learning_rate": 8.370489973286907e-06,
      "loss": 0.09465599060058594,
      "step": 5162
    },
    {
      "epoch": 0.7194314777398453,
      "grad_norm": 0.5902307629585266,
      "learning_rate": 8.362821688840947e-06,
      "loss": 0.12065887451171875,
      "step": 5163
    },
    {
      "epoch": 0.7195708214310597,
      "grad_norm": 0.47906601428985596,
      "learning_rate": 8.355155989910322e-06,
      "loss": 0.11677932739257812,
      "step": 5164
    },
    {
      "epoch": 0.719710165122274,
      "grad_norm": 0.3854982256889343,
      "learning_rate": 8.347492878198185e-06,
      "loss": 0.08113861083984375,
      "step": 5165
    },
    {
      "epoch": 0.7198495088134884,
      "grad_norm": 0.3745138347148895,
      "learning_rate": 8.339832355407093e-06,
      "loss": 0.07414436340332031,
      "step": 5166
    },
    {
      "epoch": 0.7199888525047029,
      "grad_norm": 0.29264095425605774,
      "learning_rate": 8.332174423239052e-06,
      "loss": 0.07708597183227539,
      "step": 5167
    },
    {
      "epoch": 0.7201281961959173,
      "grad_norm": 0.7210025191307068,
      "learning_rate": 8.324519083395467e-06,
      "loss": 0.10997867584228516,
      "step": 5168
    },
    {
      "epoch": 0.7202675398871317,
      "grad_norm": 0.2720508575439453,
      "learning_rate": 8.316866337577185e-06,
      "loss": 0.07496452331542969,
      "step": 5169
    },
    {
      "epoch": 0.720406883578346,
      "grad_norm": 0.5303398966789246,
      "learning_rate": 8.309216187484482e-06,
      "loss": 0.11944007873535156,
      "step": 5170
    },
    {
      "epoch": 0.7205462272695604,
      "grad_norm": 0.3748815059661865,
      "learning_rate": 8.301568634817034e-06,
      "loss": 0.0949106216430664,
      "step": 5171
    },
    {
      "epoch": 0.7206855709607748,
      "grad_norm": 0.1971912682056427,
      "learning_rate": 8.29392368127395e-06,
      "loss": 0.06019020080566406,
      "step": 5172
    },
    {
      "epoch": 0.7208249146519892,
      "grad_norm": 0.8516234755516052,
      "learning_rate": 8.286281328553769e-06,
      "loss": 0.1330556869506836,
      "step": 5173
    },
    {
      "epoch": 0.7209642583432035,
      "grad_norm": 0.3069489002227783,
      "learning_rate": 8.278641578354453e-06,
      "loss": 0.07458877563476562,
      "step": 5174
    },
    {
      "epoch": 0.7211036020344179,
      "grad_norm": 0.2986977696418762,
      "learning_rate": 8.271004432373372e-06,
      "loss": 0.08624076843261719,
      "step": 5175
    },
    {
      "epoch": 0.7212429457256323,
      "grad_norm": 0.30304375290870667,
      "learning_rate": 8.263369892307334e-06,
      "loss": 0.07644271850585938,
      "step": 5176
    },
    {
      "epoch": 0.7213822894168467,
      "grad_norm": 0.4310969114303589,
      "learning_rate": 8.255737959852548e-06,
      "loss": 0.09969139099121094,
      "step": 5177
    },
    {
      "epoch": 0.7215216331080611,
      "grad_norm": 0.5641813278198242,
      "learning_rate": 8.248108636704666e-06,
      "loss": 0.106292724609375,
      "step": 5178
    },
    {
      "epoch": 0.7216609767992754,
      "grad_norm": 0.5855363607406616,
      "learning_rate": 8.240481924558739e-06,
      "loss": 0.11700439453125,
      "step": 5179
    },
    {
      "epoch": 0.7218003204904898,
      "grad_norm": 0.5434195399284363,
      "learning_rate": 8.232857825109256e-06,
      "loss": 0.08960437774658203,
      "step": 5180
    },
    {
      "epoch": 0.7219396641817042,
      "grad_norm": 0.3627125322818756,
      "learning_rate": 8.225236340050127e-06,
      "loss": 0.08090591430664062,
      "step": 5181
    },
    {
      "epoch": 0.7220790078729186,
      "grad_norm": 0.3675297796726227,
      "learning_rate": 8.217617471074648e-06,
      "loss": 0.09488868713378906,
      "step": 5182
    },
    {
      "epoch": 0.7222183515641329,
      "grad_norm": 0.3407084047794342,
      "learning_rate": 8.210001219875569e-06,
      "loss": 0.08603858947753906,
      "step": 5183
    },
    {
      "epoch": 0.7223576952553473,
      "grad_norm": 0.3816525638103485,
      "learning_rate": 8.202387588145051e-06,
      "loss": 0.0806427001953125,
      "step": 5184
    },
    {
      "epoch": 0.7224970389465617,
      "grad_norm": 0.3510199189186096,
      "learning_rate": 8.194776577574673e-06,
      "loss": 0.07478523254394531,
      "step": 5185
    },
    {
      "epoch": 0.7226363826377761,
      "grad_norm": 0.5290048718452454,
      "learning_rate": 8.187168189855421e-06,
      "loss": 0.1089315414428711,
      "step": 5186
    },
    {
      "epoch": 0.7227757263289905,
      "grad_norm": 0.3169694244861603,
      "learning_rate": 8.179562426677699e-06,
      "loss": 0.07172203063964844,
      "step": 5187
    },
    {
      "epoch": 0.7229150700202048,
      "grad_norm": 0.2740107774734497,
      "learning_rate": 8.171959289731348e-06,
      "loss": 0.06768798828125,
      "step": 5188
    },
    {
      "epoch": 0.7230544137114192,
      "grad_norm": 0.42453867197036743,
      "learning_rate": 8.164358780705596e-06,
      "loss": 0.0925741195678711,
      "step": 5189
    },
    {
      "epoch": 0.7231937574026336,
      "grad_norm": 0.4400254786014557,
      "learning_rate": 8.156760901289111e-06,
      "loss": 0.0936279296875,
      "step": 5190
    },
    {
      "epoch": 0.723333101093848,
      "grad_norm": 0.42008212208747864,
      "learning_rate": 8.149165653169976e-06,
      "loss": 0.09064865112304688,
      "step": 5191
    },
    {
      "epoch": 0.7234724447850623,
      "grad_norm": 0.9223529696464539,
      "learning_rate": 8.141573038035675e-06,
      "loss": 0.10186386108398438,
      "step": 5192
    },
    {
      "epoch": 0.7236117884762767,
      "grad_norm": 0.9249104261398315,
      "learning_rate": 8.133983057573103e-06,
      "loss": 0.14863204956054688,
      "step": 5193
    },
    {
      "epoch": 0.7237511321674911,
      "grad_norm": 0.43765610456466675,
      "learning_rate": 8.12639571346859e-06,
      "loss": 0.10097503662109375,
      "step": 5194
    },
    {
      "epoch": 0.7238904758587055,
      "grad_norm": 0.28605374693870544,
      "learning_rate": 8.118811007407878e-06,
      "loss": 0.06598663330078125,
      "step": 5195
    },
    {
      "epoch": 0.7240298195499199,
      "grad_norm": 0.3929120600223541,
      "learning_rate": 8.111228941076101e-06,
      "loss": 0.07799625396728516,
      "step": 5196
    },
    {
      "epoch": 0.7241691632411342,
      "grad_norm": 0.46199655532836914,
      "learning_rate": 8.103649516157835e-06,
      "loss": 0.0844879150390625,
      "step": 5197
    },
    {
      "epoch": 0.7243085069323486,
      "grad_norm": 0.41551336646080017,
      "learning_rate": 8.096072734337042e-06,
      "loss": 0.08631134033203125,
      "step": 5198
    },
    {
      "epoch": 0.724447850623563,
      "grad_norm": 0.8455735445022583,
      "learning_rate": 8.088498597297121e-06,
      "loss": 0.11363410949707031,
      "step": 5199
    },
    {
      "epoch": 0.7245871943147774,
      "grad_norm": 0.19877387583255768,
      "learning_rate": 8.080927106720862e-06,
      "loss": 0.06332778930664062,
      "step": 5200
    },
    {
      "epoch": 0.7247265380059917,
      "grad_norm": 0.37594079971313477,
      "learning_rate": 8.073358264290483e-06,
      "loss": 0.08067703247070312,
      "step": 5201
    },
    {
      "epoch": 0.7248658816972061,
      "grad_norm": 0.41934120655059814,
      "learning_rate": 8.065792071687615e-06,
      "loss": 0.09765243530273438,
      "step": 5202
    },
    {
      "epoch": 0.7250052253884205,
      "grad_norm": 0.31635111570358276,
      "learning_rate": 8.058228530593283e-06,
      "loss": 0.08084964752197266,
      "step": 5203
    },
    {
      "epoch": 0.7251445690796349,
      "grad_norm": 0.41760557889938354,
      "learning_rate": 8.050667642687933e-06,
      "loss": 0.11375045776367188,
      "step": 5204
    },
    {
      "epoch": 0.7252839127708492,
      "grad_norm": 0.4118405878543854,
      "learning_rate": 8.043109409651424e-06,
      "loss": 0.09345436096191406,
      "step": 5205
    },
    {
      "epoch": 0.7254232564620636,
      "grad_norm": 0.33801305294036865,
      "learning_rate": 8.03555383316303e-06,
      "loss": 0.08185386657714844,
      "step": 5206
    },
    {
      "epoch": 0.7255626001532781,
      "grad_norm": 0.4541131556034088,
      "learning_rate": 8.028000914901422e-06,
      "loss": 0.09766387939453125,
      "step": 5207
    },
    {
      "epoch": 0.7257019438444925,
      "grad_norm": 0.3576096296310425,
      "learning_rate": 8.020450656544679e-06,
      "loss": 0.09553861618041992,
      "step": 5208
    },
    {
      "epoch": 0.7258412875357069,
      "grad_norm": 0.5684376955032349,
      "learning_rate": 8.012903059770301e-06,
      "loss": 0.09291648864746094,
      "step": 5209
    },
    {
      "epoch": 0.7259806312269212,
      "grad_norm": 0.27862709760665894,
      "learning_rate": 8.005358126255199e-06,
      "loss": 0.07385444641113281,
      "step": 5210
    },
    {
      "epoch": 0.7261199749181356,
      "grad_norm": 0.28414681553840637,
      "learning_rate": 7.997815857675673e-06,
      "loss": 0.07109832763671875,
      "step": 5211
    },
    {
      "epoch": 0.72625931860935,
      "grad_norm": 0.5604143142700195,
      "learning_rate": 7.990276255707449e-06,
      "loss": 0.13321542739868164,
      "step": 5212
    },
    {
      "epoch": 0.7263986623005644,
      "grad_norm": 0.41795778274536133,
      "learning_rate": 7.982739322025663e-06,
      "loss": 0.07536888122558594,
      "step": 5213
    },
    {
      "epoch": 0.7265380059917788,
      "grad_norm": 0.2303725779056549,
      "learning_rate": 7.97520505830484e-06,
      "loss": 0.06885910034179688,
      "step": 5214
    },
    {
      "epoch": 0.7266773496829931,
      "grad_norm": 0.49004650115966797,
      "learning_rate": 7.967673466218914e-06,
      "loss": 0.11131477355957031,
      "step": 5215
    },
    {
      "epoch": 0.7268166933742075,
      "grad_norm": 0.37181439995765686,
      "learning_rate": 7.960144547441242e-06,
      "loss": 0.08528709411621094,
      "step": 5216
    },
    {
      "epoch": 0.7269560370654219,
      "grad_norm": 0.3378145396709442,
      "learning_rate": 7.952618303644584e-06,
      "loss": 0.09295272827148438,
      "step": 5217
    },
    {
      "epoch": 0.7270953807566363,
      "grad_norm": 0.2376299798488617,
      "learning_rate": 7.945094736501094e-06,
      "loss": 0.06809425354003906,
      "step": 5218
    },
    {
      "epoch": 0.7272347244478506,
      "grad_norm": 0.24654662609100342,
      "learning_rate": 7.937573847682325e-06,
      "loss": 0.06886863708496094,
      "step": 5219
    },
    {
      "epoch": 0.727374068139065,
      "grad_norm": 0.6324467062950134,
      "learning_rate": 7.930055638859267e-06,
      "loss": 0.10488510131835938,
      "step": 5220
    },
    {
      "epoch": 0.7275134118302794,
      "grad_norm": 0.5221948623657227,
      "learning_rate": 7.922540111702275e-06,
      "loss": 0.10553550720214844,
      "step": 5221
    },
    {
      "epoch": 0.7276527555214938,
      "grad_norm": 0.6176754236221313,
      "learning_rate": 7.915027267881139e-06,
      "loss": 0.10611343383789062,
      "step": 5222
    },
    {
      "epoch": 0.7277920992127082,
      "grad_norm": 0.4901888072490692,
      "learning_rate": 7.907517109065046e-06,
      "loss": 0.09523391723632812,
      "step": 5223
    },
    {
      "epoch": 0.7279314429039225,
      "grad_norm": 0.20954963564872742,
      "learning_rate": 7.900009636922576e-06,
      "loss": 0.06577491760253906,
      "step": 5224
    },
    {
      "epoch": 0.7280707865951369,
      "grad_norm": 0.4891282021999359,
      "learning_rate": 7.89250485312171e-06,
      "loss": 0.10221195220947266,
      "step": 5225
    },
    {
      "epoch": 0.7282101302863513,
      "grad_norm": 0.5454464554786682,
      "learning_rate": 7.885002759329845e-06,
      "loss": 0.1093902587890625,
      "step": 5226
    },
    {
      "epoch": 0.7283494739775657,
      "grad_norm": 0.36035647988319397,
      "learning_rate": 7.877503357213787e-06,
      "loss": 0.08115196228027344,
      "step": 5227
    },
    {
      "epoch": 0.72848881766878,
      "grad_norm": 0.4631879925727844,
      "learning_rate": 7.870006648439712e-06,
      "loss": 0.11802482604980469,
      "step": 5228
    },
    {
      "epoch": 0.7286281613599944,
      "grad_norm": 0.2796304225921631,
      "learning_rate": 7.862512634673237e-06,
      "loss": 0.07593154907226562,
      "step": 5229
    },
    {
      "epoch": 0.7287675050512088,
      "grad_norm": 0.19027690589427948,
      "learning_rate": 7.855021317579341e-06,
      "loss": 0.059413909912109375,
      "step": 5230
    },
    {
      "epoch": 0.7289068487424232,
      "grad_norm": 0.39414161443710327,
      "learning_rate": 7.847532698822442e-06,
      "loss": 0.08272361755371094,
      "step": 5231
    },
    {
      "epoch": 0.7290461924336376,
      "grad_norm": 0.2920358180999756,
      "learning_rate": 7.840046780066325e-06,
      "loss": 0.07503509521484375,
      "step": 5232
    },
    {
      "epoch": 0.7291855361248519,
      "grad_norm": 0.413739413022995,
      "learning_rate": 7.832563562974196e-06,
      "loss": 0.09159469604492188,
      "step": 5233
    },
    {
      "epoch": 0.7293248798160663,
      "grad_norm": 0.41796594858169556,
      "learning_rate": 7.825083049208665e-06,
      "loss": 0.08780097961425781,
      "step": 5234
    },
    {
      "epoch": 0.7294642235072807,
      "grad_norm": 0.3385968506336212,
      "learning_rate": 7.817605240431718e-06,
      "loss": 0.097625732421875,
      "step": 5235
    },
    {
      "epoch": 0.7296035671984951,
      "grad_norm": 0.35310012102127075,
      "learning_rate": 7.810130138304755e-06,
      "loss": 0.08129501342773438,
      "step": 5236
    },
    {
      "epoch": 0.7297429108897094,
      "grad_norm": 0.4573066830635071,
      "learning_rate": 7.802657744488575e-06,
      "loss": 0.09685707092285156,
      "step": 5237
    },
    {
      "epoch": 0.7298822545809238,
      "grad_norm": 0.5055689811706543,
      "learning_rate": 7.79518806064338e-06,
      "loss": 0.11349296569824219,
      "step": 5238
    },
    {
      "epoch": 0.7300215982721382,
      "grad_norm": 0.2029559165239334,
      "learning_rate": 7.78772108842876e-06,
      "loss": 0.06750869750976562,
      "step": 5239
    },
    {
      "epoch": 0.7301609419633526,
      "grad_norm": 0.31361621618270874,
      "learning_rate": 7.780256829503692e-06,
      "loss": 0.07887077331542969,
      "step": 5240
    },
    {
      "epoch": 0.730300285654567,
      "grad_norm": 0.28357189893722534,
      "learning_rate": 7.772795285526578e-06,
      "loss": 0.06754493713378906,
      "step": 5241
    },
    {
      "epoch": 0.7304396293457813,
      "grad_norm": 0.5571112036705017,
      "learning_rate": 7.765336458155205e-06,
      "loss": 0.1217947006225586,
      "step": 5242
    },
    {
      "epoch": 0.7305789730369957,
      "grad_norm": 0.3396696448326111,
      "learning_rate": 7.757880349046742e-06,
      "loss": 0.0818939208984375,
      "step": 5243
    },
    {
      "epoch": 0.7307183167282101,
      "grad_norm": 0.5522879958152771,
      "learning_rate": 7.750426959857782e-06,
      "loss": 0.10689544677734375,
      "step": 5244
    },
    {
      "epoch": 0.7308576604194245,
      "grad_norm": 0.45427289605140686,
      "learning_rate": 7.74297629224428e-06,
      "loss": 0.08866119384765625,
      "step": 5245
    },
    {
      "epoch": 0.7309970041106388,
      "grad_norm": 0.5256122946739197,
      "learning_rate": 7.735528347861623e-06,
      "loss": 0.1161651611328125,
      "step": 5246
    },
    {
      "epoch": 0.7311363478018533,
      "grad_norm": 0.4239473044872284,
      "learning_rate": 7.728083128364555e-06,
      "loss": 0.08913612365722656,
      "step": 5247
    },
    {
      "epoch": 0.7312756914930677,
      "grad_norm": 0.4787594676017761,
      "learning_rate": 7.720640635407244e-06,
      "loss": 0.09527015686035156,
      "step": 5248
    },
    {
      "epoch": 0.7314150351842821,
      "grad_norm": 0.23695620894432068,
      "learning_rate": 7.713200870643246e-06,
      "loss": 0.06842422485351562,
      "step": 5249
    },
    {
      "epoch": 0.7315543788754965,
      "grad_norm": 0.5999666452407837,
      "learning_rate": 7.705763835725507e-06,
      "loss": 0.1345539093017578,
      "step": 5250
    },
    {
      "epoch": 0.7316937225667108,
      "grad_norm": 0.4842648208141327,
      "learning_rate": 7.69832953230635e-06,
      "loss": 0.11662101745605469,
      "step": 5251
    },
    {
      "epoch": 0.7318330662579252,
      "grad_norm": 0.5254838466644287,
      "learning_rate": 7.69089796203752e-06,
      "loss": 0.11694192886352539,
      "step": 5252
    },
    {
      "epoch": 0.7319724099491396,
      "grad_norm": 0.4872169494628906,
      "learning_rate": 7.683469126570152e-06,
      "loss": 0.10805892944335938,
      "step": 5253
    },
    {
      "epoch": 0.732111753640354,
      "grad_norm": 0.6147377490997314,
      "learning_rate": 7.67604302755474e-06,
      "loss": 0.10059547424316406,
      "step": 5254
    },
    {
      "epoch": 0.7322510973315683,
      "grad_norm": 0.7399958968162537,
      "learning_rate": 7.668619666641216e-06,
      "loss": 0.11346578598022461,
      "step": 5255
    },
    {
      "epoch": 0.7323904410227827,
      "grad_norm": 0.3494555950164795,
      "learning_rate": 7.661199045478874e-06,
      "loss": 0.08198165893554688,
      "step": 5256
    },
    {
      "epoch": 0.7325297847139971,
      "grad_norm": 0.20980234444141388,
      "learning_rate": 7.653781165716396e-06,
      "loss": 0.053356170654296875,
      "step": 5257
    },
    {
      "epoch": 0.7326691284052115,
      "grad_norm": 0.3114386200904846,
      "learning_rate": 7.646366029001873e-06,
      "loss": 0.06892585754394531,
      "step": 5258
    },
    {
      "epoch": 0.7328084720964259,
      "grad_norm": 0.5675535202026367,
      "learning_rate": 7.638953636982789e-06,
      "loss": 0.11671257019042969,
      "step": 5259
    },
    {
      "epoch": 0.7329478157876402,
      "grad_norm": 0.6186447739601135,
      "learning_rate": 7.631543991305998e-06,
      "loss": 0.09685134887695312,
      "step": 5260
    },
    {
      "epoch": 0.7330871594788546,
      "grad_norm": 0.5841290354728699,
      "learning_rate": 7.62413709361775e-06,
      "loss": 0.08207225799560547,
      "step": 5261
    },
    {
      "epoch": 0.733226503170069,
      "grad_norm": 0.36281493306159973,
      "learning_rate": 7.616732945563692e-06,
      "loss": 0.07967281341552734,
      "step": 5262
    },
    {
      "epoch": 0.7333658468612834,
      "grad_norm": 0.4248640239238739,
      "learning_rate": 7.609331548788865e-06,
      "loss": 0.08991622924804688,
      "step": 5263
    },
    {
      "epoch": 0.7335051905524977,
      "grad_norm": 0.38816145062446594,
      "learning_rate": 7.601932904937679e-06,
      "loss": 0.0976715087890625,
      "step": 5264
    },
    {
      "epoch": 0.7336445342437121,
      "grad_norm": 0.5123850107192993,
      "learning_rate": 7.594537015653949e-06,
      "loss": 0.09481048583984375,
      "step": 5265
    },
    {
      "epoch": 0.7337838779349265,
      "grad_norm": 0.37643522024154663,
      "learning_rate": 7.5871438825808786e-06,
      "loss": 0.08455467224121094,
      "step": 5266
    },
    {
      "epoch": 0.7339232216261409,
      "grad_norm": 0.28728827834129333,
      "learning_rate": 7.579753507361048e-06,
      "loss": 0.07533073425292969,
      "step": 5267
    },
    {
      "epoch": 0.7340625653173553,
      "grad_norm": 0.33166998624801636,
      "learning_rate": 7.572365891636422e-06,
      "loss": 0.07481765747070312,
      "step": 5268
    },
    {
      "epoch": 0.7342019090085696,
      "grad_norm": 0.7789461612701416,
      "learning_rate": 7.5649810370483666e-06,
      "loss": 0.11207962036132812,
      "step": 5269
    },
    {
      "epoch": 0.734341252699784,
      "grad_norm": 0.36456742882728577,
      "learning_rate": 7.557598945237634e-06,
      "loss": 0.09762763977050781,
      "step": 5270
    },
    {
      "epoch": 0.7344805963909984,
      "grad_norm": 0.2507241368293762,
      "learning_rate": 7.550219617844354e-06,
      "loss": 0.0762176513671875,
      "step": 5271
    },
    {
      "epoch": 0.7346199400822128,
      "grad_norm": 0.5907347202301025,
      "learning_rate": 7.542843056508034e-06,
      "loss": 0.1425609588623047,
      "step": 5272
    },
    {
      "epoch": 0.7347592837734271,
      "grad_norm": 0.22814048826694489,
      "learning_rate": 7.535469262867583e-06,
      "loss": 0.062175750732421875,
      "step": 5273
    },
    {
      "epoch": 0.7348986274646415,
      "grad_norm": 0.3741365969181061,
      "learning_rate": 7.528098238561301e-06,
      "loss": 0.09040260314941406,
      "step": 5274
    },
    {
      "epoch": 0.7350379711558559,
      "grad_norm": 0.20582439005374908,
      "learning_rate": 7.520729985226842e-06,
      "loss": 0.06995010375976562,
      "step": 5275
    },
    {
      "epoch": 0.7351773148470703,
      "grad_norm": 0.35709893703460693,
      "learning_rate": 7.513364504501283e-06,
      "loss": 0.09345054626464844,
      "step": 5276
    },
    {
      "epoch": 0.7353166585382847,
      "grad_norm": 0.5450581312179565,
      "learning_rate": 7.506001798021049e-06,
      "loss": 0.09265708923339844,
      "step": 5277
    },
    {
      "epoch": 0.735456002229499,
      "grad_norm": 0.2907041907310486,
      "learning_rate": 7.498641867421981e-06,
      "loss": 0.07734107971191406,
      "step": 5278
    },
    {
      "epoch": 0.7355953459207134,
      "grad_norm": 0.2761858105659485,
      "learning_rate": 7.4912847143392706e-06,
      "loss": 0.07556962966918945,
      "step": 5279
    },
    {
      "epoch": 0.7357346896119278,
      "grad_norm": 0.4762260615825653,
      "learning_rate": 7.483930340407519e-06,
      "loss": 0.08451175689697266,
      "step": 5280
    },
    {
      "epoch": 0.7358740333031422,
      "grad_norm": 0.3836508095264435,
      "learning_rate": 7.476578747260712e-06,
      "loss": 0.09780693054199219,
      "step": 5281
    },
    {
      "epoch": 0.7360133769943565,
      "grad_norm": 0.4055038094520569,
      "learning_rate": 7.469229936532179e-06,
      "loss": 0.0874786376953125,
      "step": 5282
    },
    {
      "epoch": 0.7361527206855709,
      "grad_norm": 0.539602518081665,
      "learning_rate": 7.46188390985467e-06,
      "loss": 0.1146240234375,
      "step": 5283
    },
    {
      "epoch": 0.7362920643767853,
      "grad_norm": 0.2968486547470093,
      "learning_rate": 7.454540668860309e-06,
      "loss": 0.081634521484375,
      "step": 5284
    },
    {
      "epoch": 0.7364314080679997,
      "grad_norm": 0.20609451830387115,
      "learning_rate": 7.4472002151805985e-06,
      "loss": 0.06349563598632812,
      "step": 5285
    },
    {
      "epoch": 0.736570751759214,
      "grad_norm": 0.3517487049102783,
      "learning_rate": 7.4398625504464105e-06,
      "loss": 0.08644676208496094,
      "step": 5286
    },
    {
      "epoch": 0.7367100954504285,
      "grad_norm": 0.58217853307724,
      "learning_rate": 7.432527676288015e-06,
      "loss": 0.13733863830566406,
      "step": 5287
    },
    {
      "epoch": 0.7368494391416429,
      "grad_norm": 0.4435776174068451,
      "learning_rate": 7.425195594335053e-06,
      "loss": 0.09581756591796875,
      "step": 5288
    },
    {
      "epoch": 0.7369887828328573,
      "grad_norm": 0.19641825556755066,
      "learning_rate": 7.417866306216532e-06,
      "loss": 0.05942535400390625,
      "step": 5289
    },
    {
      "epoch": 0.7371281265240717,
      "grad_norm": 0.2820185422897339,
      "learning_rate": 7.4105398135608645e-06,
      "loss": 0.07469654083251953,
      "step": 5290
    },
    {
      "epoch": 0.737267470215286,
      "grad_norm": 0.510333240032196,
      "learning_rate": 7.403216117995835e-06,
      "loss": 0.09612464904785156,
      "step": 5291
    },
    {
      "epoch": 0.7374068139065004,
      "grad_norm": 0.4178142547607422,
      "learning_rate": 7.395895221148594e-06,
      "loss": 0.09444808959960938,
      "step": 5292
    },
    {
      "epoch": 0.7375461575977148,
      "grad_norm": 0.733722984790802,
      "learning_rate": 7.388577124645671e-06,
      "loss": 0.12704849243164062,
      "step": 5293
    },
    {
      "epoch": 0.7376855012889292,
      "grad_norm": 0.2610227167606354,
      "learning_rate": 7.381261830112989e-06,
      "loss": 0.06534099578857422,
      "step": 5294
    },
    {
      "epoch": 0.7378248449801436,
      "grad_norm": 0.3517918884754181,
      "learning_rate": 7.373949339175843e-06,
      "loss": 0.08024120330810547,
      "step": 5295
    },
    {
      "epoch": 0.7379641886713579,
      "grad_norm": 0.7256416082382202,
      "learning_rate": 7.366639653458889e-06,
      "loss": 0.12233352661132812,
      "step": 5296
    },
    {
      "epoch": 0.7381035323625723,
      "grad_norm": 0.4519415497779846,
      "learning_rate": 7.359332774586188e-06,
      "loss": 0.09293365478515625,
      "step": 5297
    },
    {
      "epoch": 0.7382428760537867,
      "grad_norm": 0.5059886574745178,
      "learning_rate": 7.352028704181145e-06,
      "loss": 0.08743095397949219,
      "step": 5298
    },
    {
      "epoch": 0.7383822197450011,
      "grad_norm": 0.2818610668182373,
      "learning_rate": 7.344727443866573e-06,
      "loss": 0.0724020004272461,
      "step": 5299
    },
    {
      "epoch": 0.7385215634362154,
      "grad_norm": 0.24482236802577972,
      "learning_rate": 7.3374289952646305e-06,
      "loss": 0.06834030151367188,
      "step": 5300
    },
    {
      "epoch": 0.7386609071274298,
      "grad_norm": 0.5065958499908447,
      "learning_rate": 7.330133359996876e-06,
      "loss": 0.10745620727539062,
      "step": 5301
    },
    {
      "epoch": 0.7388002508186442,
      "grad_norm": 0.4421199858188629,
      "learning_rate": 7.322840539684235e-06,
      "loss": 0.09714508056640625,
      "step": 5302
    },
    {
      "epoch": 0.7389395945098586,
      "grad_norm": 0.44000640511512756,
      "learning_rate": 7.3155505359470046e-06,
      "loss": 0.08619213104248047,
      "step": 5303
    },
    {
      "epoch": 0.739078938201073,
      "grad_norm": 0.4587100148200989,
      "learning_rate": 7.308263350404845e-06,
      "loss": 0.1168050765991211,
      "step": 5304
    },
    {
      "epoch": 0.7392182818922873,
      "grad_norm": 0.4899766147136688,
      "learning_rate": 7.3009789846768116e-06,
      "loss": 0.10746002197265625,
      "step": 5305
    },
    {
      "epoch": 0.7393576255835017,
      "grad_norm": 0.4415782392024994,
      "learning_rate": 7.2936974403813336e-06,
      "loss": 0.10408592224121094,
      "step": 5306
    },
    {
      "epoch": 0.7394969692747161,
      "grad_norm": 0.34515702724456787,
      "learning_rate": 7.286418719136186e-06,
      "loss": 0.07592487335205078,
      "step": 5307
    },
    {
      "epoch": 0.7396363129659305,
      "grad_norm": 0.47330015897750854,
      "learning_rate": 7.279142822558549e-06,
      "loss": 0.08856391906738281,
      "step": 5308
    },
    {
      "epoch": 0.7397756566571448,
      "grad_norm": 0.5328090190887451,
      "learning_rate": 7.271869752264949e-06,
      "loss": 0.10736751556396484,
      "step": 5309
    },
    {
      "epoch": 0.7399150003483592,
      "grad_norm": 0.4696916937828064,
      "learning_rate": 7.264599509871309e-06,
      "loss": 0.12309455871582031,
      "step": 5310
    },
    {
      "epoch": 0.7400543440395736,
      "grad_norm": 0.5248294472694397,
      "learning_rate": 7.2573320969928974e-06,
      "loss": 0.1027841567993164,
      "step": 5311
    },
    {
      "epoch": 0.740193687730788,
      "grad_norm": 0.4099537432193756,
      "learning_rate": 7.250067515244373e-06,
      "loss": 0.08274650573730469,
      "step": 5312
    },
    {
      "epoch": 0.7403330314220024,
      "grad_norm": 0.33823150396347046,
      "learning_rate": 7.2428057662397665e-06,
      "loss": 0.07387733459472656,
      "step": 5313
    },
    {
      "epoch": 0.7404723751132167,
      "grad_norm": 0.3604521155357361,
      "learning_rate": 7.235546851592468e-06,
      "loss": 0.08296585083007812,
      "step": 5314
    },
    {
      "epoch": 0.7406117188044311,
      "grad_norm": 0.5736944079399109,
      "learning_rate": 7.228290772915234e-06,
      "loss": 0.12298965454101562,
      "step": 5315
    },
    {
      "epoch": 0.7407510624956455,
      "grad_norm": 0.21735985577106476,
      "learning_rate": 7.22103753182021e-06,
      "loss": 0.054782867431640625,
      "step": 5316
    },
    {
      "epoch": 0.7408904061868599,
      "grad_norm": 0.9880133867263794,
      "learning_rate": 7.213787129918901e-06,
      "loss": 0.11534690856933594,
      "step": 5317
    },
    {
      "epoch": 0.7410297498780742,
      "grad_norm": 0.29176798462867737,
      "learning_rate": 7.206539568822179e-06,
      "loss": 0.07190132141113281,
      "step": 5318
    },
    {
      "epoch": 0.7411690935692886,
      "grad_norm": 0.32569587230682373,
      "learning_rate": 7.199294850140279e-06,
      "loss": 0.08114337921142578,
      "step": 5319
    },
    {
      "epoch": 0.741308437260503,
      "grad_norm": 0.43902289867401123,
      "learning_rate": 7.1920529754828235e-06,
      "loss": 0.09025764465332031,
      "step": 5320
    },
    {
      "epoch": 0.7414477809517174,
      "grad_norm": 0.576722264289856,
      "learning_rate": 7.184813946458782e-06,
      "loss": 0.1326751708984375,
      "step": 5321
    },
    {
      "epoch": 0.7415871246429317,
      "grad_norm": 0.4572266638278961,
      "learning_rate": 7.177577764676504e-06,
      "loss": 0.08825874328613281,
      "step": 5322
    },
    {
      "epoch": 0.7417264683341461,
      "grad_norm": 0.5242393016815186,
      "learning_rate": 7.170344431743707e-06,
      "loss": 0.10158061981201172,
      "step": 5323
    },
    {
      "epoch": 0.7418658120253605,
      "grad_norm": 0.5776848793029785,
      "learning_rate": 7.163113949267484e-06,
      "loss": 0.09824752807617188,
      "step": 5324
    },
    {
      "epoch": 0.7420051557165749,
      "grad_norm": 0.5140780806541443,
      "learning_rate": 7.155886318854257e-06,
      "loss": 0.09934616088867188,
      "step": 5325
    },
    {
      "epoch": 0.7421444994077893,
      "grad_norm": 0.4499206840991974,
      "learning_rate": 7.148661542109854e-06,
      "loss": 0.09918212890625,
      "step": 5326
    },
    {
      "epoch": 0.7422838430990036,
      "grad_norm": 0.2885253131389618,
      "learning_rate": 7.141439620639463e-06,
      "loss": 0.08504486083984375,
      "step": 5327
    },
    {
      "epoch": 0.7424231867902181,
      "grad_norm": 0.3308602273464203,
      "learning_rate": 7.134220556047613e-06,
      "loss": 0.09352684020996094,
      "step": 5328
    },
    {
      "epoch": 0.7425625304814325,
      "grad_norm": 0.40207818150520325,
      "learning_rate": 7.127004349938234e-06,
      "loss": 0.09848403930664062,
      "step": 5329
    },
    {
      "epoch": 0.7427018741726469,
      "grad_norm": 0.4874917268753052,
      "learning_rate": 7.119791003914584e-06,
      "loss": 0.09022808074951172,
      "step": 5330
    },
    {
      "epoch": 0.7428412178638613,
      "grad_norm": 0.3834069073200226,
      "learning_rate": 7.112580519579322e-06,
      "loss": 0.0862579345703125,
      "step": 5331
    },
    {
      "epoch": 0.7429805615550756,
      "grad_norm": 0.4183783233165741,
      "learning_rate": 7.105372898534435e-06,
      "loss": 0.09955406188964844,
      "step": 5332
    },
    {
      "epoch": 0.74311990524629,
      "grad_norm": 0.43529242277145386,
      "learning_rate": 7.098168142381301e-06,
      "loss": 0.08988571166992188,
      "step": 5333
    },
    {
      "epoch": 0.7432592489375044,
      "grad_norm": 0.25895678997039795,
      "learning_rate": 7.090966252720659e-06,
      "loss": 0.07162952423095703,
      "step": 5334
    },
    {
      "epoch": 0.7433985926287188,
      "grad_norm": 0.48741841316223145,
      "learning_rate": 7.083767231152598e-06,
      "loss": 0.09276580810546875,
      "step": 5335
    },
    {
      "epoch": 0.7435379363199331,
      "grad_norm": 0.5888963937759399,
      "learning_rate": 7.076571079276569e-06,
      "loss": 0.10720252990722656,
      "step": 5336
    },
    {
      "epoch": 0.7436772800111475,
      "grad_norm": 0.6804060935974121,
      "learning_rate": 7.069377798691397e-06,
      "loss": 0.10675048828125,
      "step": 5337
    },
    {
      "epoch": 0.7438166237023619,
      "grad_norm": 0.49692052602767944,
      "learning_rate": 7.0621873909952765e-06,
      "loss": 0.10655021667480469,
      "step": 5338
    },
    {
      "epoch": 0.7439559673935763,
      "grad_norm": 0.6393535733222961,
      "learning_rate": 7.05499985778574e-06,
      "loss": 0.11463356018066406,
      "step": 5339
    },
    {
      "epoch": 0.7440953110847907,
      "grad_norm": 0.2779795527458191,
      "learning_rate": 7.047815200659691e-06,
      "loss": 0.07988166809082031,
      "step": 5340
    },
    {
      "epoch": 0.744234654776005,
      "grad_norm": 0.47452548146247864,
      "learning_rate": 7.040633421213401e-06,
      "loss": 0.10791873931884766,
      "step": 5341
    },
    {
      "epoch": 0.7443739984672194,
      "grad_norm": 0.44544273614883423,
      "learning_rate": 7.033454521042502e-06,
      "loss": 0.09237480163574219,
      "step": 5342
    },
    {
      "epoch": 0.7445133421584338,
      "grad_norm": 0.5560184121131897,
      "learning_rate": 7.026278501741972e-06,
      "loss": 0.10832786560058594,
      "step": 5343
    },
    {
      "epoch": 0.7446526858496482,
      "grad_norm": 0.8864823579788208,
      "learning_rate": 7.019105364906165e-06,
      "loss": 0.1255340576171875,
      "step": 5344
    },
    {
      "epoch": 0.7447920295408625,
      "grad_norm": 0.8156023025512695,
      "learning_rate": 7.011935112128791e-06,
      "loss": 0.13353538513183594,
      "step": 5345
    },
    {
      "epoch": 0.7449313732320769,
      "grad_norm": 0.2233741581439972,
      "learning_rate": 7.004767745002916e-06,
      "loss": 0.067474365234375,
      "step": 5346
    },
    {
      "epoch": 0.7450707169232913,
      "grad_norm": 0.39506271481513977,
      "learning_rate": 6.997603265120951e-06,
      "loss": 0.10436248779296875,
      "step": 5347
    },
    {
      "epoch": 0.7452100606145057,
      "grad_norm": 0.7981248497962952,
      "learning_rate": 6.990441674074695e-06,
      "loss": 0.13108444213867188,
      "step": 5348
    },
    {
      "epoch": 0.74534940430572,
      "grad_norm": 0.36387279629707336,
      "learning_rate": 6.98328297345529e-06,
      "loss": 0.09256649017333984,
      "step": 5349
    },
    {
      "epoch": 0.7454887479969344,
      "grad_norm": 0.5401174426078796,
      "learning_rate": 6.9761271648532306e-06,
      "loss": 0.10043525695800781,
      "step": 5350
    },
    {
      "epoch": 0.7456280916881488,
      "grad_norm": 0.5807051062583923,
      "learning_rate": 6.968974249858371e-06,
      "loss": 0.10367202758789062,
      "step": 5351
    },
    {
      "epoch": 0.7457674353793632,
      "grad_norm": 0.4250313937664032,
      "learning_rate": 6.9618242300599284e-06,
      "loss": 0.0880126953125,
      "step": 5352
    },
    {
      "epoch": 0.7459067790705776,
      "grad_norm": 0.3035208582878113,
      "learning_rate": 6.9546771070464815e-06,
      "loss": 0.07814979553222656,
      "step": 5353
    },
    {
      "epoch": 0.7460461227617919,
      "grad_norm": 0.626301646232605,
      "learning_rate": 6.947532882405945e-06,
      "loss": 0.1120452880859375,
      "step": 5354
    },
    {
      "epoch": 0.7461854664530063,
      "grad_norm": 0.3918405771255493,
      "learning_rate": 6.940391557725616e-06,
      "loss": 0.072967529296875,
      "step": 5355
    },
    {
      "epoch": 0.7463248101442207,
      "grad_norm": 0.42057374119758606,
      "learning_rate": 6.933253134592128e-06,
      "loss": 0.09484481811523438,
      "step": 5356
    },
    {
      "epoch": 0.7464641538354351,
      "grad_norm": 0.34041526913642883,
      "learning_rate": 6.9261176145914655e-06,
      "loss": 0.08697509765625,
      "step": 5357
    },
    {
      "epoch": 0.7466034975266495,
      "grad_norm": 0.4116964638233185,
      "learning_rate": 6.9189849993089905e-06,
      "loss": 0.09820365905761719,
      "step": 5358
    },
    {
      "epoch": 0.7467428412178638,
      "grad_norm": 0.3223320245742798,
      "learning_rate": 6.911855290329408e-06,
      "loss": 0.09090805053710938,
      "step": 5359
    },
    {
      "epoch": 0.7468821849090782,
      "grad_norm": 0.5847012996673584,
      "learning_rate": 6.904728489236767e-06,
      "loss": 0.10735130310058594,
      "step": 5360
    },
    {
      "epoch": 0.7470215286002926,
      "grad_norm": 0.631091296672821,
      "learning_rate": 6.897604597614491e-06,
      "loss": 0.11324691772460938,
      "step": 5361
    },
    {
      "epoch": 0.747160872291507,
      "grad_norm": 0.42836451530456543,
      "learning_rate": 6.890483617045336e-06,
      "loss": 0.0932769775390625,
      "step": 5362
    },
    {
      "epoch": 0.7473002159827213,
      "grad_norm": 0.6613325476646423,
      "learning_rate": 6.883365549111432e-06,
      "loss": 0.11774444580078125,
      "step": 5363
    },
    {
      "epoch": 0.7474395596739357,
      "grad_norm": 0.25608140230178833,
      "learning_rate": 6.876250395394237e-06,
      "loss": 0.0735931396484375,
      "step": 5364
    },
    {
      "epoch": 0.7475789033651501,
      "grad_norm": 0.7796933650970459,
      "learning_rate": 6.869138157474586e-06,
      "loss": 0.13457107543945312,
      "step": 5365
    },
    {
      "epoch": 0.7477182470563645,
      "grad_norm": 0.28765279054641724,
      "learning_rate": 6.862028836932659e-06,
      "loss": 0.06716537475585938,
      "step": 5366
    },
    {
      "epoch": 0.7478575907475788,
      "grad_norm": 0.47190096974372864,
      "learning_rate": 6.854922435347979e-06,
      "loss": 0.10978317260742188,
      "step": 5367
    },
    {
      "epoch": 0.7479969344387933,
      "grad_norm": 0.9470019936561584,
      "learning_rate": 6.847818954299421e-06,
      "loss": 0.16739654541015625,
      "step": 5368
    },
    {
      "epoch": 0.7481362781300077,
      "grad_norm": 0.4666794538497925,
      "learning_rate": 6.840718395365222e-06,
      "loss": 0.10582828521728516,
      "step": 5369
    },
    {
      "epoch": 0.7482756218212221,
      "grad_norm": 0.29082751274108887,
      "learning_rate": 6.833620760122972e-06,
      "loss": 0.07102394104003906,
      "step": 5370
    },
    {
      "epoch": 0.7484149655124365,
      "grad_norm": 0.4226193428039551,
      "learning_rate": 6.826526050149594e-06,
      "loss": 0.11211395263671875,
      "step": 5371
    },
    {
      "epoch": 0.7485543092036508,
      "grad_norm": 0.4173780679702759,
      "learning_rate": 6.819434267021366e-06,
      "loss": 0.0889892578125,
      "step": 5372
    },
    {
      "epoch": 0.7486936528948652,
      "grad_norm": 0.45405134558677673,
      "learning_rate": 6.812345412313926e-06,
      "loss": 0.10229873657226562,
      "step": 5373
    },
    {
      "epoch": 0.7488329965860796,
      "grad_norm": 0.31372514367103577,
      "learning_rate": 6.805259487602261e-06,
      "loss": 0.07987022399902344,
      "step": 5374
    },
    {
      "epoch": 0.748972340277294,
      "grad_norm": 0.3155912756919861,
      "learning_rate": 6.798176494460693e-06,
      "loss": 0.08767318725585938,
      "step": 5375
    },
    {
      "epoch": 0.7491116839685084,
      "grad_norm": 0.39213237166404724,
      "learning_rate": 6.791096434462909e-06,
      "loss": 0.0906209945678711,
      "step": 5376
    },
    {
      "epoch": 0.7492510276597227,
      "grad_norm": 0.19547727704048157,
      "learning_rate": 6.7840193091819285e-06,
      "loss": 0.05392146110534668,
      "step": 5377
    },
    {
      "epoch": 0.7493903713509371,
      "grad_norm": 0.3898513615131378,
      "learning_rate": 6.776945120190137e-06,
      "loss": 0.09250068664550781,
      "step": 5378
    },
    {
      "epoch": 0.7495297150421515,
      "grad_norm": 0.2871211767196655,
      "learning_rate": 6.769873869059247e-06,
      "loss": 0.07194328308105469,
      "step": 5379
    },
    {
      "epoch": 0.7496690587333659,
      "grad_norm": 0.6385897397994995,
      "learning_rate": 6.762805557360335e-06,
      "loss": 0.10859298706054688,
      "step": 5380
    },
    {
      "epoch": 0.7498084024245802,
      "grad_norm": 0.2599533796310425,
      "learning_rate": 6.755740186663822e-06,
      "loss": 0.06892967224121094,
      "step": 5381
    },
    {
      "epoch": 0.7499477461157946,
      "grad_norm": 0.3884093165397644,
      "learning_rate": 6.748677758539468e-06,
      "loss": 0.08082294464111328,
      "step": 5382
    },
    {
      "epoch": 0.750087089807009,
      "grad_norm": 0.8340564966201782,
      "learning_rate": 6.741618274556379e-06,
      "loss": 0.15668296813964844,
      "step": 5383
    },
    {
      "epoch": 0.7502264334982234,
      "grad_norm": 0.5126813054084778,
      "learning_rate": 6.734561736283014e-06,
      "loss": 0.08181953430175781,
      "step": 5384
    },
    {
      "epoch": 0.7503657771894378,
      "grad_norm": 0.33062878251075745,
      "learning_rate": 6.727508145287183e-06,
      "loss": 0.07105255126953125,
      "step": 5385
    },
    {
      "epoch": 0.7505051208806521,
      "grad_norm": 0.509175717830658,
      "learning_rate": 6.720457503136017e-06,
      "loss": 0.09186744689941406,
      "step": 5386
    },
    {
      "epoch": 0.7506444645718665,
      "grad_norm": 0.25099271535873413,
      "learning_rate": 6.713409811396028e-06,
      "loss": 0.06648826599121094,
      "step": 5387
    },
    {
      "epoch": 0.7507838082630809,
      "grad_norm": 0.35701096057891846,
      "learning_rate": 6.706365071633037e-06,
      "loss": 0.08265209197998047,
      "step": 5388
    },
    {
      "epoch": 0.7509231519542953,
      "grad_norm": 0.6357024908065796,
      "learning_rate": 6.699323285412222e-06,
      "loss": 0.10497093200683594,
      "step": 5389
    },
    {
      "epoch": 0.7510624956455096,
      "grad_norm": 0.43411293625831604,
      "learning_rate": 6.692284454298115e-06,
      "loss": 0.0875101089477539,
      "step": 5390
    },
    {
      "epoch": 0.751201839336724,
      "grad_norm": 0.21202746033668518,
      "learning_rate": 6.685248579854589e-06,
      "loss": 0.06991004943847656,
      "step": 5391
    },
    {
      "epoch": 0.7513411830279384,
      "grad_norm": 0.8575679063796997,
      "learning_rate": 6.678215663644845e-06,
      "loss": 0.1467437744140625,
      "step": 5392
    },
    {
      "epoch": 0.7514805267191528,
      "grad_norm": 0.3817749619483948,
      "learning_rate": 6.671185707231434e-06,
      "loss": 0.08721923828125,
      "step": 5393
    },
    {
      "epoch": 0.7516198704103672,
      "grad_norm": 0.23752187192440033,
      "learning_rate": 6.664158712176256e-06,
      "loss": 0.07210922241210938,
      "step": 5394
    },
    {
      "epoch": 0.7517592141015815,
      "grad_norm": 0.3807697892189026,
      "learning_rate": 6.657134680040558e-06,
      "loss": 0.09526824951171875,
      "step": 5395
    },
    {
      "epoch": 0.7518985577927959,
      "grad_norm": 0.30172351002693176,
      "learning_rate": 6.650113612384903e-06,
      "loss": 0.07815361022949219,
      "step": 5396
    },
    {
      "epoch": 0.7520379014840103,
      "grad_norm": 0.4551621973514557,
      "learning_rate": 6.643095510769229e-06,
      "loss": 0.09092330932617188,
      "step": 5397
    },
    {
      "epoch": 0.7521772451752247,
      "grad_norm": 0.6545920372009277,
      "learning_rate": 6.63608037675278e-06,
      "loss": 0.12024831771850586,
      "step": 5398
    },
    {
      "epoch": 0.752316588866439,
      "grad_norm": 0.2520396411418915,
      "learning_rate": 6.629068211894176e-06,
      "loss": 0.06812477111816406,
      "step": 5399
    },
    {
      "epoch": 0.7524559325576534,
      "grad_norm": 0.4703008234500885,
      "learning_rate": 6.622059017751346e-06,
      "loss": 0.09744071960449219,
      "step": 5400
    },
    {
      "epoch": 0.7525952762488678,
      "grad_norm": 0.38878002762794495,
      "learning_rate": 6.615052795881576e-06,
      "loss": 0.08088350296020508,
      "step": 5401
    },
    {
      "epoch": 0.7527346199400822,
      "grad_norm": 0.49537014961242676,
      "learning_rate": 6.6080495478415e-06,
      "loss": 0.10373878479003906,
      "step": 5402
    },
    {
      "epoch": 0.7528739636312965,
      "grad_norm": 0.2493787556886673,
      "learning_rate": 6.60104927518707e-06,
      "loss": 0.06769561767578125,
      "step": 5403
    },
    {
      "epoch": 0.7530133073225109,
      "grad_norm": 0.5747022032737732,
      "learning_rate": 6.594051979473582e-06,
      "loss": 0.09014892578125,
      "step": 5404
    },
    {
      "epoch": 0.7531526510137253,
      "grad_norm": 0.4645173251628876,
      "learning_rate": 6.58705766225568e-06,
      "loss": 0.08118724822998047,
      "step": 5405
    },
    {
      "epoch": 0.7532919947049397,
      "grad_norm": 0.41708967089653015,
      "learning_rate": 6.580066325087351e-06,
      "loss": 0.08668899536132812,
      "step": 5406
    },
    {
      "epoch": 0.7534313383961541,
      "grad_norm": 0.2979566156864166,
      "learning_rate": 6.573077969521892e-06,
      "loss": 0.09412384033203125,
      "step": 5407
    },
    {
      "epoch": 0.7535706820873685,
      "grad_norm": 0.29419851303100586,
      "learning_rate": 6.566092597111977e-06,
      "loss": 0.06911849975585938,
      "step": 5408
    },
    {
      "epoch": 0.7537100257785829,
      "grad_norm": 0.32274869084358215,
      "learning_rate": 6.559110209409578e-06,
      "loss": 0.07682609558105469,
      "step": 5409
    },
    {
      "epoch": 0.7538493694697973,
      "grad_norm": 0.3934105336666107,
      "learning_rate": 6.552130807966035e-06,
      "loss": 0.08617210388183594,
      "step": 5410
    },
    {
      "epoch": 0.7539887131610117,
      "grad_norm": 0.5384942293167114,
      "learning_rate": 6.5451543943320005e-06,
      "loss": 0.10007476806640625,
      "step": 5411
    },
    {
      "epoch": 0.7541280568522261,
      "grad_norm": 0.2780991196632385,
      "learning_rate": 6.538180970057482e-06,
      "loss": 0.06848621368408203,
      "step": 5412
    },
    {
      "epoch": 0.7542674005434404,
      "grad_norm": 0.6526579856872559,
      "learning_rate": 6.531210536691819e-06,
      "loss": 0.10422706604003906,
      "step": 5413
    },
    {
      "epoch": 0.7544067442346548,
      "grad_norm": 0.4447813630104065,
      "learning_rate": 6.524243095783675e-06,
      "loss": 0.10326576232910156,
      "step": 5414
    },
    {
      "epoch": 0.7545460879258692,
      "grad_norm": 0.570479154586792,
      "learning_rate": 6.517278648881054e-06,
      "loss": 0.1380290985107422,
      "step": 5415
    },
    {
      "epoch": 0.7546854316170836,
      "grad_norm": 0.3983791768550873,
      "learning_rate": 6.5103171975312995e-06,
      "loss": 0.1024017333984375,
      "step": 5416
    },
    {
      "epoch": 0.754824775308298,
      "grad_norm": 0.35590988397598267,
      "learning_rate": 6.503358743281098e-06,
      "loss": 0.09337806701660156,
      "step": 5417
    },
    {
      "epoch": 0.7549641189995123,
      "grad_norm": 0.23069748282432556,
      "learning_rate": 6.496403287676443e-06,
      "loss": 0.06593894958496094,
      "step": 5418
    },
    {
      "epoch": 0.7551034626907267,
      "grad_norm": 0.32855483889579773,
      "learning_rate": 6.489450832262692e-06,
      "loss": 0.07491683959960938,
      "step": 5419
    },
    {
      "epoch": 0.7552428063819411,
      "grad_norm": 0.5590240955352783,
      "learning_rate": 6.482501378584511e-06,
      "loss": 0.10749435424804688,
      "step": 5420
    },
    {
      "epoch": 0.7553821500731555,
      "grad_norm": 0.35634827613830566,
      "learning_rate": 6.475554928185912e-06,
      "loss": 0.08460807800292969,
      "step": 5421
    },
    {
      "epoch": 0.7555214937643698,
      "grad_norm": 0.5878504514694214,
      "learning_rate": 6.468611482610238e-06,
      "loss": 0.11061286926269531,
      "step": 5422
    },
    {
      "epoch": 0.7556608374555842,
      "grad_norm": 0.26749250292778015,
      "learning_rate": 6.461671043400166e-06,
      "loss": 0.07394790649414062,
      "step": 5423
    },
    {
      "epoch": 0.7558001811467986,
      "grad_norm": 0.5609726309776306,
      "learning_rate": 6.454733612097717e-06,
      "loss": 0.12810325622558594,
      "step": 5424
    },
    {
      "epoch": 0.755939524838013,
      "grad_norm": 0.44277918338775635,
      "learning_rate": 6.4477991902442e-06,
      "loss": 0.10015392303466797,
      "step": 5425
    },
    {
      "epoch": 0.7560788685292273,
      "grad_norm": 0.37568068504333496,
      "learning_rate": 6.440867779380302e-06,
      "loss": 0.08872032165527344,
      "step": 5426
    },
    {
      "epoch": 0.7562182122204417,
      "grad_norm": 0.28825080394744873,
      "learning_rate": 6.43393938104603e-06,
      "loss": 0.06662559509277344,
      "step": 5427
    },
    {
      "epoch": 0.7563575559116561,
      "grad_norm": 0.35220131278038025,
      "learning_rate": 6.427013996780702e-06,
      "loss": 0.08224868774414062,
      "step": 5428
    },
    {
      "epoch": 0.7564968996028705,
      "grad_norm": 0.6465475559234619,
      "learning_rate": 6.420091628122995e-06,
      "loss": 0.1110992431640625,
      "step": 5429
    },
    {
      "epoch": 0.7566362432940849,
      "grad_norm": 0.5498183369636536,
      "learning_rate": 6.413172276610886e-06,
      "loss": 0.10001659393310547,
      "step": 5430
    },
    {
      "epoch": 0.7567755869852992,
      "grad_norm": 0.5067670941352844,
      "learning_rate": 6.406255943781711e-06,
      "loss": 0.09395980834960938,
      "step": 5431
    },
    {
      "epoch": 0.7569149306765136,
      "grad_norm": 0.4883936643600464,
      "learning_rate": 6.3993426311721095e-06,
      "loss": 0.10237503051757812,
      "step": 5432
    },
    {
      "epoch": 0.757054274367728,
      "grad_norm": 0.6049573421478271,
      "learning_rate": 6.3924323403180685e-06,
      "loss": 0.12650299072265625,
      "step": 5433
    },
    {
      "epoch": 0.7571936180589424,
      "grad_norm": 0.3602919578552246,
      "learning_rate": 6.385525072754899e-06,
      "loss": 0.08189773559570312,
      "step": 5434
    },
    {
      "epoch": 0.7573329617501567,
      "grad_norm": 0.3477003574371338,
      "learning_rate": 6.378620830017237e-06,
      "loss": 0.08017921447753906,
      "step": 5435
    },
    {
      "epoch": 0.7574723054413711,
      "grad_norm": 0.44920551776885986,
      "learning_rate": 6.371719613639036e-06,
      "loss": 0.11045646667480469,
      "step": 5436
    },
    {
      "epoch": 0.7576116491325855,
      "grad_norm": 1.0397391319274902,
      "learning_rate": 6.3648214251536e-06,
      "loss": 0.13048744201660156,
      "step": 5437
    },
    {
      "epoch": 0.7577509928237999,
      "grad_norm": 0.6216824650764465,
      "learning_rate": 6.357926266093552e-06,
      "loss": 0.10159492492675781,
      "step": 5438
    },
    {
      "epoch": 0.7578903365150143,
      "grad_norm": 0.44514256715774536,
      "learning_rate": 6.351034137990828e-06,
      "loss": 0.079864501953125,
      "step": 5439
    },
    {
      "epoch": 0.7580296802062286,
      "grad_norm": 0.23278479278087616,
      "learning_rate": 6.344145042376715e-06,
      "loss": 0.05853271484375,
      "step": 5440
    },
    {
      "epoch": 0.758169023897443,
      "grad_norm": 0.5219769477844238,
      "learning_rate": 6.337258980781797e-06,
      "loss": 0.09805488586425781,
      "step": 5441
    },
    {
      "epoch": 0.7583083675886574,
      "grad_norm": 0.5640836358070374,
      "learning_rate": 6.330375954736014e-06,
      "loss": 0.13898849487304688,
      "step": 5442
    },
    {
      "epoch": 0.7584477112798718,
      "grad_norm": 0.41109687089920044,
      "learning_rate": 6.323495965768605e-06,
      "loss": 0.11105728149414062,
      "step": 5443
    },
    {
      "epoch": 0.7585870549710861,
      "grad_norm": 0.5368844270706177,
      "learning_rate": 6.3166190154081545e-06,
      "loss": 0.08562040328979492,
      "step": 5444
    },
    {
      "epoch": 0.7587263986623005,
      "grad_norm": 0.2581464946269989,
      "learning_rate": 6.309745105182566e-06,
      "loss": 0.07268905639648438,
      "step": 5445
    },
    {
      "epoch": 0.7588657423535149,
      "grad_norm": 0.7036799192428589,
      "learning_rate": 6.302874236619061e-06,
      "loss": 0.11418724060058594,
      "step": 5446
    },
    {
      "epoch": 0.7590050860447293,
      "grad_norm": 0.3938368260860443,
      "learning_rate": 6.296006411244184e-06,
      "loss": 0.07325172424316406,
      "step": 5447
    },
    {
      "epoch": 0.7591444297359438,
      "grad_norm": 0.19827838242053986,
      "learning_rate": 6.289141630583815e-06,
      "loss": 0.06301307678222656,
      "step": 5448
    },
    {
      "epoch": 0.7592837734271581,
      "grad_norm": 0.4769246280193329,
      "learning_rate": 6.28227989616316e-06,
      "loss": 0.10567283630371094,
      "step": 5449
    },
    {
      "epoch": 0.7594231171183725,
      "grad_norm": 0.4471933841705322,
      "learning_rate": 6.275421209506729e-06,
      "loss": 0.11755561828613281,
      "step": 5450
    },
    {
      "epoch": 0.7595624608095869,
      "grad_norm": 0.48801907896995544,
      "learning_rate": 6.268565572138361e-06,
      "loss": 0.08877182006835938,
      "step": 5451
    },
    {
      "epoch": 0.7597018045008013,
      "grad_norm": 0.4508819282054901,
      "learning_rate": 6.2617129855812295e-06,
      "loss": 0.10540008544921875,
      "step": 5452
    },
    {
      "epoch": 0.7598411481920156,
      "grad_norm": 0.6613976359367371,
      "learning_rate": 6.254863451357829e-06,
      "loss": 0.11122798919677734,
      "step": 5453
    },
    {
      "epoch": 0.75998049188323,
      "grad_norm": 1.0758857727050781,
      "learning_rate": 6.248016970989957e-06,
      "loss": 0.139434814453125,
      "step": 5454
    },
    {
      "epoch": 0.7601198355744444,
      "grad_norm": 0.47574397921562195,
      "learning_rate": 6.241173545998749e-06,
      "loss": 0.10753250122070312,
      "step": 5455
    },
    {
      "epoch": 0.7602591792656588,
      "grad_norm": 0.2507322132587433,
      "learning_rate": 6.2343331779046745e-06,
      "loss": 0.07076644897460938,
      "step": 5456
    },
    {
      "epoch": 0.7603985229568732,
      "grad_norm": 0.4457326829433441,
      "learning_rate": 6.227495868227477e-06,
      "loss": 0.0998220443725586,
      "step": 5457
    },
    {
      "epoch": 0.7605378666480875,
      "grad_norm": 0.2072775512933731,
      "learning_rate": 6.220661618486268e-06,
      "loss": 0.061641693115234375,
      "step": 5458
    },
    {
      "epoch": 0.7606772103393019,
      "grad_norm": 0.34904319047927856,
      "learning_rate": 6.2138304301994654e-06,
      "loss": 0.079803466796875,
      "step": 5459
    },
    {
      "epoch": 0.7608165540305163,
      "grad_norm": 0.26476773619651794,
      "learning_rate": 6.207002304884793e-06,
      "loss": 0.06981658935546875,
      "step": 5460
    },
    {
      "epoch": 0.7609558977217307,
      "grad_norm": 0.252936989068985,
      "learning_rate": 6.200177244059313e-06,
      "loss": 0.07707786560058594,
      "step": 5461
    },
    {
      "epoch": 0.761095241412945,
      "grad_norm": 0.6309165358543396,
      "learning_rate": 6.19335524923939e-06,
      "loss": 0.11703681945800781,
      "step": 5462
    },
    {
      "epoch": 0.7612345851041594,
      "grad_norm": 0.45011815428733826,
      "learning_rate": 6.186536321940724e-06,
      "loss": 0.08563995361328125,
      "step": 5463
    },
    {
      "epoch": 0.7613739287953738,
      "grad_norm": 0.45311301946640015,
      "learning_rate": 6.179720463678314e-06,
      "loss": 0.10060882568359375,
      "step": 5464
    },
    {
      "epoch": 0.7615132724865882,
      "grad_norm": 0.38116151094436646,
      "learning_rate": 6.172907675966495e-06,
      "loss": 0.07877540588378906,
      "step": 5465
    },
    {
      "epoch": 0.7616526161778026,
      "grad_norm": 0.34794652462005615,
      "learning_rate": 6.16609796031892e-06,
      "loss": 0.06635856628417969,
      "step": 5466
    },
    {
      "epoch": 0.7617919598690169,
      "grad_norm": 0.36350882053375244,
      "learning_rate": 6.159291318248544e-06,
      "loss": 0.07588005065917969,
      "step": 5467
    },
    {
      "epoch": 0.7619313035602313,
      "grad_norm": 0.42620009183883667,
      "learning_rate": 6.152487751267641e-06,
      "loss": 0.0813446044921875,
      "step": 5468
    },
    {
      "epoch": 0.7620706472514457,
      "grad_norm": 0.60691899061203,
      "learning_rate": 6.145687260887818e-06,
      "loss": 0.12911415100097656,
      "step": 5469
    },
    {
      "epoch": 0.7622099909426601,
      "grad_norm": 0.3409835398197174,
      "learning_rate": 6.138889848619991e-06,
      "loss": 0.07182979583740234,
      "step": 5470
    },
    {
      "epoch": 0.7623493346338744,
      "grad_norm": 0.6642006635665894,
      "learning_rate": 6.132095515974385e-06,
      "loss": 0.10474300384521484,
      "step": 5471
    },
    {
      "epoch": 0.7624886783250888,
      "grad_norm": 0.40079253911972046,
      "learning_rate": 6.125304264460541e-06,
      "loss": 0.078277587890625,
      "step": 5472
    },
    {
      "epoch": 0.7626280220163032,
      "grad_norm": 0.31252363324165344,
      "learning_rate": 6.118516095587321e-06,
      "loss": 0.07710742950439453,
      "step": 5473
    },
    {
      "epoch": 0.7627673657075176,
      "grad_norm": 0.3905729353427887,
      "learning_rate": 6.1117310108629156e-06,
      "loss": 0.08337783813476562,
      "step": 5474
    },
    {
      "epoch": 0.762906709398732,
      "grad_norm": 0.3330124020576477,
      "learning_rate": 6.104949011794796e-06,
      "loss": 0.08472251892089844,
      "step": 5475
    },
    {
      "epoch": 0.7630460530899463,
      "grad_norm": 0.4900970458984375,
      "learning_rate": 6.098170099889777e-06,
      "loss": 0.11617851257324219,
      "step": 5476
    },
    {
      "epoch": 0.7631853967811607,
      "grad_norm": 0.3379685878753662,
      "learning_rate": 6.0913942766539855e-06,
      "loss": 0.07949018478393555,
      "step": 5477
    },
    {
      "epoch": 0.7633247404723751,
      "grad_norm": 0.5697566866874695,
      "learning_rate": 6.084621543592846e-06,
      "loss": 0.09378433227539062,
      "step": 5478
    },
    {
      "epoch": 0.7634640841635895,
      "grad_norm": 0.38149791955947876,
      "learning_rate": 6.0778519022110985e-06,
      "loss": 0.09528732299804688,
      "step": 5479
    },
    {
      "epoch": 0.7636034278548038,
      "grad_norm": 0.47144532203674316,
      "learning_rate": 6.071085354012812e-06,
      "loss": 0.07474136352539062,
      "step": 5480
    },
    {
      "epoch": 0.7637427715460182,
      "grad_norm": 0.39604148268699646,
      "learning_rate": 6.064321900501362e-06,
      "loss": 0.08310317993164062,
      "step": 5481
    },
    {
      "epoch": 0.7638821152372326,
      "grad_norm": 0.3993886113166809,
      "learning_rate": 6.057561543179429e-06,
      "loss": 0.08984375,
      "step": 5482
    },
    {
      "epoch": 0.764021458928447,
      "grad_norm": 0.4063778519630432,
      "learning_rate": 6.050804283549005e-06,
      "loss": 0.09253883361816406,
      "step": 5483
    },
    {
      "epoch": 0.7641608026196614,
      "grad_norm": 0.35410287976264954,
      "learning_rate": 6.0440501231114025e-06,
      "loss": 0.08151626586914062,
      "step": 5484
    },
    {
      "epoch": 0.7643001463108757,
      "grad_norm": 0.39233893156051636,
      "learning_rate": 6.03729906336725e-06,
      "loss": 0.07767963409423828,
      "step": 5485
    },
    {
      "epoch": 0.7644394900020901,
      "grad_norm": 0.5650411248207092,
      "learning_rate": 6.030551105816465e-06,
      "loss": 0.10586166381835938,
      "step": 5486
    },
    {
      "epoch": 0.7645788336933045,
      "grad_norm": 0.4234045445919037,
      "learning_rate": 6.0238062519583015e-06,
      "loss": 0.09926033020019531,
      "step": 5487
    },
    {
      "epoch": 0.764718177384519,
      "grad_norm": 0.4568929076194763,
      "learning_rate": 6.017064503291307e-06,
      "loss": 0.09903907775878906,
      "step": 5488
    },
    {
      "epoch": 0.7648575210757333,
      "grad_norm": 0.538196861743927,
      "learning_rate": 6.01032586131334e-06,
      "loss": 0.11121177673339844,
      "step": 5489
    },
    {
      "epoch": 0.7649968647669477,
      "grad_norm": 0.655273973941803,
      "learning_rate": 6.0035903275215755e-06,
      "loss": 0.09779071807861328,
      "step": 5490
    },
    {
      "epoch": 0.7651362084581621,
      "grad_norm": 0.30899253487586975,
      "learning_rate": 5.996857903412503e-06,
      "loss": 0.07186508178710938,
      "step": 5491
    },
    {
      "epoch": 0.7652755521493765,
      "grad_norm": 0.38374000787734985,
      "learning_rate": 5.990128590481907e-06,
      "loss": 0.08440780639648438,
      "step": 5492
    },
    {
      "epoch": 0.7654148958405909,
      "grad_norm": 0.4119897782802582,
      "learning_rate": 5.983402390224882e-06,
      "loss": 0.08262062072753906,
      "step": 5493
    },
    {
      "epoch": 0.7655542395318052,
      "grad_norm": 0.3379344642162323,
      "learning_rate": 5.97667930413584e-06,
      "loss": 0.080535888671875,
      "step": 5494
    },
    {
      "epoch": 0.7656935832230196,
      "grad_norm": 0.45356515049934387,
      "learning_rate": 5.969959333708506e-06,
      "loss": 0.07735633850097656,
      "step": 5495
    },
    {
      "epoch": 0.765832926914234,
      "grad_norm": 0.35990554094314575,
      "learning_rate": 5.963242480435889e-06,
      "loss": 0.09521961212158203,
      "step": 5496
    },
    {
      "epoch": 0.7659722706054484,
      "grad_norm": 0.4882234036922455,
      "learning_rate": 5.95652874581033e-06,
      "loss": 0.09277915954589844,
      "step": 5497
    },
    {
      "epoch": 0.7661116142966627,
      "grad_norm": 0.33397021889686584,
      "learning_rate": 5.9498181313234726e-06,
      "loss": 0.08430719375610352,
      "step": 5498
    },
    {
      "epoch": 0.7662509579878771,
      "grad_norm": 0.430325984954834,
      "learning_rate": 5.943110638466254e-06,
      "loss": 0.09017753601074219,
      "step": 5499
    },
    {
      "epoch": 0.7663903016790915,
      "grad_norm": 0.27272331714630127,
      "learning_rate": 5.9364062687289226e-06,
      "loss": 0.06490516662597656,
      "step": 5500
    },
    {
      "epoch": 0.7665296453703059,
      "grad_norm": 0.6101490259170532,
      "learning_rate": 5.929705023601038e-06,
      "loss": 0.09710025787353516,
      "step": 5501
    },
    {
      "epoch": 0.7666689890615203,
      "grad_norm": 0.6803492307662964,
      "learning_rate": 5.923006904571476e-06,
      "loss": 0.12291812896728516,
      "step": 5502
    },
    {
      "epoch": 0.7668083327527346,
      "grad_norm": 0.1838691532611847,
      "learning_rate": 5.9163119131283966e-06,
      "loss": 0.057074546813964844,
      "step": 5503
    },
    {
      "epoch": 0.766947676443949,
      "grad_norm": 0.5395563244819641,
      "learning_rate": 5.909620050759266e-06,
      "loss": 0.11144733428955078,
      "step": 5504
    },
    {
      "epoch": 0.7670870201351634,
      "grad_norm": 0.31303671002388,
      "learning_rate": 5.902931318950871e-06,
      "loss": 0.07358169555664062,
      "step": 5505
    },
    {
      "epoch": 0.7672263638263778,
      "grad_norm": 0.4697885811328888,
      "learning_rate": 5.896245719189304e-06,
      "loss": 0.0849456787109375,
      "step": 5506
    },
    {
      "epoch": 0.7673657075175921,
      "grad_norm": 0.28624579310417175,
      "learning_rate": 5.889563252959935e-06,
      "loss": 0.07756233215332031,
      "step": 5507
    },
    {
      "epoch": 0.7675050512088065,
      "grad_norm": 0.303877592086792,
      "learning_rate": 5.882883921747473e-06,
      "loss": 0.07081031799316406,
      "step": 5508
    },
    {
      "epoch": 0.7676443949000209,
      "grad_norm": 0.23901784420013428,
      "learning_rate": 5.876207727035897e-06,
      "loss": 0.06409931182861328,
      "step": 5509
    },
    {
      "epoch": 0.7677837385912353,
      "grad_norm": 0.33295851945877075,
      "learning_rate": 5.869534670308519e-06,
      "loss": 0.0933990478515625,
      "step": 5510
    },
    {
      "epoch": 0.7679230822824497,
      "grad_norm": 0.3850051462650299,
      "learning_rate": 5.862864753047926e-06,
      "loss": 0.07532501220703125,
      "step": 5511
    },
    {
      "epoch": 0.768062425973664,
      "grad_norm": 0.47002583742141724,
      "learning_rate": 5.856197976736029e-06,
      "loss": 0.10259819030761719,
      "step": 5512
    },
    {
      "epoch": 0.7682017696648784,
      "grad_norm": 0.49294349551200867,
      "learning_rate": 5.8495343428540395e-06,
      "loss": 0.09721755981445312,
      "step": 5513
    },
    {
      "epoch": 0.7683411133560928,
      "grad_norm": 0.49381691217422485,
      "learning_rate": 5.84287385288246e-06,
      "loss": 0.11520957946777344,
      "step": 5514
    },
    {
      "epoch": 0.7684804570473072,
      "grad_norm": 0.4883812963962555,
      "learning_rate": 5.836216508301089e-06,
      "loss": 0.0957489013671875,
      "step": 5515
    },
    {
      "epoch": 0.7686198007385215,
      "grad_norm": 0.29631319642066956,
      "learning_rate": 5.829562310589048e-06,
      "loss": 0.07316017150878906,
      "step": 5516
    },
    {
      "epoch": 0.7687591444297359,
      "grad_norm": 0.3663982152938843,
      "learning_rate": 5.822911261224748e-06,
      "loss": 0.06248760223388672,
      "step": 5517
    },
    {
      "epoch": 0.7688984881209503,
      "grad_norm": 0.4885948598384857,
      "learning_rate": 5.816263361685892e-06,
      "loss": 0.09225845336914062,
      "step": 5518
    },
    {
      "epoch": 0.7690378318121647,
      "grad_norm": 0.6836559772491455,
      "learning_rate": 5.809618613449503e-06,
      "loss": 0.10099411010742188,
      "step": 5519
    },
    {
      "epoch": 0.769177175503379,
      "grad_norm": 0.4479810297489166,
      "learning_rate": 5.802977017991888e-06,
      "loss": 0.09682273864746094,
      "step": 5520
    },
    {
      "epoch": 0.7693165191945934,
      "grad_norm": 0.4251222312450409,
      "learning_rate": 5.7963385767886496e-06,
      "loss": 0.09038352966308594,
      "step": 5521
    },
    {
      "epoch": 0.7694558628858078,
      "grad_norm": 0.4616117775440216,
      "learning_rate": 5.789703291314703e-06,
      "loss": 0.10788917541503906,
      "step": 5522
    },
    {
      "epoch": 0.7695952065770222,
      "grad_norm": 0.2932809293270111,
      "learning_rate": 5.783071163044259e-06,
      "loss": 0.06887245178222656,
      "step": 5523
    },
    {
      "epoch": 0.7697345502682366,
      "grad_norm": 0.3136294484138489,
      "learning_rate": 5.776442193450835e-06,
      "loss": 0.0786275863647461,
      "step": 5524
    },
    {
      "epoch": 0.7698738939594509,
      "grad_norm": 0.28551772236824036,
      "learning_rate": 5.769816384007216e-06,
      "loss": 0.07081794738769531,
      "step": 5525
    },
    {
      "epoch": 0.7700132376506653,
      "grad_norm": 0.7451992630958557,
      "learning_rate": 5.7631937361855175e-06,
      "loss": 0.1413872241973877,
      "step": 5526
    },
    {
      "epoch": 0.7701525813418797,
      "grad_norm": 0.47857722640037537,
      "learning_rate": 5.756574251457141e-06,
      "loss": 0.08774089813232422,
      "step": 5527
    },
    {
      "epoch": 0.7702919250330941,
      "grad_norm": 0.4239037036895752,
      "learning_rate": 5.74995793129278e-06,
      "loss": 0.09536552429199219,
      "step": 5528
    },
    {
      "epoch": 0.7704312687243086,
      "grad_norm": 0.4977094233036041,
      "learning_rate": 5.743344777162439e-06,
      "loss": 0.09756088256835938,
      "step": 5529
    },
    {
      "epoch": 0.7705706124155229,
      "grad_norm": 0.6402494311332703,
      "learning_rate": 5.736734790535394e-06,
      "loss": 0.10307693481445312,
      "step": 5530
    },
    {
      "epoch": 0.7707099561067373,
      "grad_norm": 0.5160558223724365,
      "learning_rate": 5.7301279728802525e-06,
      "loss": 0.08398151397705078,
      "step": 5531
    },
    {
      "epoch": 0.7708492997979517,
      "grad_norm": 0.5487915277481079,
      "learning_rate": 5.72352432566488e-06,
      "loss": 0.10751914978027344,
      "step": 5532
    },
    {
      "epoch": 0.7709886434891661,
      "grad_norm": 0.3306606411933899,
      "learning_rate": 5.716923850356466e-06,
      "loss": 0.08510017395019531,
      "step": 5533
    },
    {
      "epoch": 0.7711279871803804,
      "grad_norm": 0.5041764974594116,
      "learning_rate": 5.71032654842149e-06,
      "loss": 0.09428596496582031,
      "step": 5534
    },
    {
      "epoch": 0.7712673308715948,
      "grad_norm": 0.24946224689483643,
      "learning_rate": 5.703732421325716e-06,
      "loss": 0.0734100341796875,
      "step": 5535
    },
    {
      "epoch": 0.7714066745628092,
      "grad_norm": 0.7002207636833191,
      "learning_rate": 5.697141470534204e-06,
      "loss": 0.10991287231445312,
      "step": 5536
    },
    {
      "epoch": 0.7715460182540236,
      "grad_norm": 0.45745134353637695,
      "learning_rate": 5.690553697511316e-06,
      "loss": 0.10733795166015625,
      "step": 5537
    },
    {
      "epoch": 0.771685361945238,
      "grad_norm": 0.46072080731391907,
      "learning_rate": 5.683969103720712e-06,
      "loss": 0.0908966064453125,
      "step": 5538
    },
    {
      "epoch": 0.7718247056364523,
      "grad_norm": 0.5528069138526917,
      "learning_rate": 5.677387690625329e-06,
      "loss": 0.11517524719238281,
      "step": 5539
    },
    {
      "epoch": 0.7719640493276667,
      "grad_norm": 0.31519708037376404,
      "learning_rate": 5.670809459687414e-06,
      "loss": 0.07634162902832031,
      "step": 5540
    },
    {
      "epoch": 0.7721033930188811,
      "grad_norm": 0.2139206975698471,
      "learning_rate": 5.66423441236849e-06,
      "loss": 0.06270027160644531,
      "step": 5541
    },
    {
      "epoch": 0.7722427367100955,
      "grad_norm": 0.3998766541481018,
      "learning_rate": 5.657662550129394e-06,
      "loss": 0.08623504638671875,
      "step": 5542
    },
    {
      "epoch": 0.7723820804013098,
      "grad_norm": 0.3840036392211914,
      "learning_rate": 5.6510938744302356e-06,
      "loss": 0.09958648681640625,
      "step": 5543
    },
    {
      "epoch": 0.7725214240925242,
      "grad_norm": 0.38918596506118774,
      "learning_rate": 5.644528386730424e-06,
      "loss": 0.08456611633300781,
      "step": 5544
    },
    {
      "epoch": 0.7726607677837386,
      "grad_norm": 0.4766157269477844,
      "learning_rate": 5.637966088488671e-06,
      "loss": 0.08827400207519531,
      "step": 5545
    },
    {
      "epoch": 0.772800111474953,
      "grad_norm": 0.6479074954986572,
      "learning_rate": 5.631406981162961e-06,
      "loss": 0.10059928894042969,
      "step": 5546
    },
    {
      "epoch": 0.7729394551661674,
      "grad_norm": 0.3935074210166931,
      "learning_rate": 5.624851066210575e-06,
      "loss": 0.07424545288085938,
      "step": 5547
    },
    {
      "epoch": 0.7730787988573817,
      "grad_norm": 0.33431318402290344,
      "learning_rate": 5.618298345088091e-06,
      "loss": 0.08692455291748047,
      "step": 5548
    },
    {
      "epoch": 0.7732181425485961,
      "grad_norm": 0.60147625207901,
      "learning_rate": 5.611748819251382e-06,
      "loss": 0.11832809448242188,
      "step": 5549
    },
    {
      "epoch": 0.7733574862398105,
      "grad_norm": 0.5956194996833801,
      "learning_rate": 5.605202490155588e-06,
      "loss": 0.07928943634033203,
      "step": 5550
    },
    {
      "epoch": 0.7734968299310249,
      "grad_norm": 0.5900894999504089,
      "learning_rate": 5.5986593592551694e-06,
      "loss": 0.10564613342285156,
      "step": 5551
    },
    {
      "epoch": 0.7736361736222392,
      "grad_norm": 0.26753196120262146,
      "learning_rate": 5.592119428003848e-06,
      "loss": 0.07294082641601562,
      "step": 5552
    },
    {
      "epoch": 0.7737755173134536,
      "grad_norm": 0.5651355981826782,
      "learning_rate": 5.585582697854657e-06,
      "loss": 0.10560798645019531,
      "step": 5553
    },
    {
      "epoch": 0.773914861004668,
      "grad_norm": 0.30781468749046326,
      "learning_rate": 5.579049170259896e-06,
      "loss": 0.07196426391601562,
      "step": 5554
    },
    {
      "epoch": 0.7740542046958824,
      "grad_norm": 0.44568315148353577,
      "learning_rate": 5.572518846671176e-06,
      "loss": 0.07974910736083984,
      "step": 5555
    },
    {
      "epoch": 0.7741935483870968,
      "grad_norm": 0.33378368616104126,
      "learning_rate": 5.565991728539395e-06,
      "loss": 0.07940101623535156,
      "step": 5556
    },
    {
      "epoch": 0.7743328920783111,
      "grad_norm": 0.4446316659450531,
      "learning_rate": 5.559467817314705e-06,
      "loss": 0.08727455139160156,
      "step": 5557
    },
    {
      "epoch": 0.7744722357695255,
      "grad_norm": 0.47160443663597107,
      "learning_rate": 5.552947114446583e-06,
      "loss": 0.07864952087402344,
      "step": 5558
    },
    {
      "epoch": 0.7746115794607399,
      "grad_norm": 0.5461838841438293,
      "learning_rate": 5.546429621383786e-06,
      "loss": 0.09097099304199219,
      "step": 5559
    },
    {
      "epoch": 0.7747509231519543,
      "grad_norm": 0.3548069894313812,
      "learning_rate": 5.5399153395743425e-06,
      "loss": 0.08426284790039062,
      "step": 5560
    },
    {
      "epoch": 0.7748902668431686,
      "grad_norm": 0.32090651988983154,
      "learning_rate": 5.533404270465585e-06,
      "loss": 0.07843208312988281,
      "step": 5561
    },
    {
      "epoch": 0.775029610534383,
      "grad_norm": 0.4261753261089325,
      "learning_rate": 5.526896415504115e-06,
      "loss": 0.08635997772216797,
      "step": 5562
    },
    {
      "epoch": 0.7751689542255974,
      "grad_norm": 0.342864453792572,
      "learning_rate": 5.520391776135841e-06,
      "loss": 0.0870370864868164,
      "step": 5563
    },
    {
      "epoch": 0.7753082979168118,
      "grad_norm": 0.42381200194358826,
      "learning_rate": 5.513890353805933e-06,
      "loss": 0.09576988220214844,
      "step": 5564
    },
    {
      "epoch": 0.7754476416080262,
      "grad_norm": 0.6680832505226135,
      "learning_rate": 5.507392149958865e-06,
      "loss": 0.13741111755371094,
      "step": 5565
    },
    {
      "epoch": 0.7755869852992405,
      "grad_norm": 0.7433701753616333,
      "learning_rate": 5.500897166038397e-06,
      "loss": 0.11650466918945312,
      "step": 5566
    },
    {
      "epoch": 0.7757263289904549,
      "grad_norm": 0.4303596019744873,
      "learning_rate": 5.494405403487557e-06,
      "loss": 0.08211326599121094,
      "step": 5567
    },
    {
      "epoch": 0.7758656726816693,
      "grad_norm": 0.36084893345832825,
      "learning_rate": 5.487916863748664e-06,
      "loss": 0.0764923095703125,
      "step": 5568
    },
    {
      "epoch": 0.7760050163728838,
      "grad_norm": 0.6295400857925415,
      "learning_rate": 5.481431548263325e-06,
      "loss": 0.11561870574951172,
      "step": 5569
    },
    {
      "epoch": 0.7761443600640981,
      "grad_norm": 0.4392228424549103,
      "learning_rate": 5.474949458472438e-06,
      "loss": 0.08215045928955078,
      "step": 5570
    },
    {
      "epoch": 0.7762837037553125,
      "grad_norm": 0.3167163133621216,
      "learning_rate": 5.468470595816162e-06,
      "loss": 0.0771026611328125,
      "step": 5571
    },
    {
      "epoch": 0.7764230474465269,
      "grad_norm": 0.47042712569236755,
      "learning_rate": 5.461994961733967e-06,
      "loss": 0.09822845458984375,
      "step": 5572
    },
    {
      "epoch": 0.7765623911377413,
      "grad_norm": 0.25248804688453674,
      "learning_rate": 5.455522557664576e-06,
      "loss": 0.06484031677246094,
      "step": 5573
    },
    {
      "epoch": 0.7767017348289557,
      "grad_norm": 0.43427860736846924,
      "learning_rate": 5.449053385046023e-06,
      "loss": 0.09099197387695312,
      "step": 5574
    },
    {
      "epoch": 0.77684107852017,
      "grad_norm": 0.48409879207611084,
      "learning_rate": 5.442587445315599e-06,
      "loss": 0.08309555053710938,
      "step": 5575
    },
    {
      "epoch": 0.7769804222113844,
      "grad_norm": 0.39322492480278015,
      "learning_rate": 5.436124739909892e-06,
      "loss": 0.08971405029296875,
      "step": 5576
    },
    {
      "epoch": 0.7771197659025988,
      "grad_norm": 0.45462679862976074,
      "learning_rate": 5.429665270264774e-06,
      "loss": 0.09873485565185547,
      "step": 5577
    },
    {
      "epoch": 0.7772591095938132,
      "grad_norm": 0.4180141091346741,
      "learning_rate": 5.423209037815389e-06,
      "loss": 0.08861541748046875,
      "step": 5578
    },
    {
      "epoch": 0.7773984532850275,
      "grad_norm": 0.7758762240409851,
      "learning_rate": 5.416756043996154e-06,
      "loss": 0.12624740600585938,
      "step": 5579
    },
    {
      "epoch": 0.7775377969762419,
      "grad_norm": 0.45777758955955505,
      "learning_rate": 5.4103062902407855e-06,
      "loss": 0.08131980895996094,
      "step": 5580
    },
    {
      "epoch": 0.7776771406674563,
      "grad_norm": 0.41972222924232483,
      "learning_rate": 5.403859777982279e-06,
      "loss": 0.10394477844238281,
      "step": 5581
    },
    {
      "epoch": 0.7778164843586707,
      "grad_norm": 0.758634626865387,
      "learning_rate": 5.397416508652891e-06,
      "loss": 0.12643051147460938,
      "step": 5582
    },
    {
      "epoch": 0.7779558280498851,
      "grad_norm": 0.3732863664627075,
      "learning_rate": 5.390976483684167e-06,
      "loss": 0.08740472793579102,
      "step": 5583
    },
    {
      "epoch": 0.7780951717410994,
      "grad_norm": 0.886250376701355,
      "learning_rate": 5.384539704506939e-06,
      "loss": 0.10259819030761719,
      "step": 5584
    },
    {
      "epoch": 0.7782345154323138,
      "grad_norm": 0.24178580939769745,
      "learning_rate": 5.378106172551319e-06,
      "loss": 0.07463836669921875,
      "step": 5585
    },
    {
      "epoch": 0.7783738591235282,
      "grad_norm": 0.4198436141014099,
      "learning_rate": 5.371675889246677e-06,
      "loss": 0.09327888488769531,
      "step": 5586
    },
    {
      "epoch": 0.7785132028147426,
      "grad_norm": 0.5162523984909058,
      "learning_rate": 5.3652488560216875e-06,
      "loss": 0.07857704162597656,
      "step": 5587
    },
    {
      "epoch": 0.7786525465059569,
      "grad_norm": 0.296465665102005,
      "learning_rate": 5.358825074304286e-06,
      "loss": 0.08316993713378906,
      "step": 5588
    },
    {
      "epoch": 0.7787918901971713,
      "grad_norm": 0.37294113636016846,
      "learning_rate": 5.352404545521683e-06,
      "loss": 0.0830841064453125,
      "step": 5589
    },
    {
      "epoch": 0.7789312338883857,
      "grad_norm": 0.41440248489379883,
      "learning_rate": 5.345987271100381e-06,
      "loss": 0.0898895263671875,
      "step": 5590
    },
    {
      "epoch": 0.7790705775796001,
      "grad_norm": 0.3802894353866577,
      "learning_rate": 5.339573252466155e-06,
      "loss": 0.07274198532104492,
      "step": 5591
    },
    {
      "epoch": 0.7792099212708145,
      "grad_norm": 0.32686641812324524,
      "learning_rate": 5.333162491044044e-06,
      "loss": 0.07315731048583984,
      "step": 5592
    },
    {
      "epoch": 0.7793492649620288,
      "grad_norm": 0.5207377672195435,
      "learning_rate": 5.3267549882583825e-06,
      "loss": 0.100677490234375,
      "step": 5593
    },
    {
      "epoch": 0.7794886086532432,
      "grad_norm": 0.23007819056510925,
      "learning_rate": 5.320350745532761e-06,
      "loss": 0.06632232666015625,
      "step": 5594
    },
    {
      "epoch": 0.7796279523444576,
      "grad_norm": 0.5154330134391785,
      "learning_rate": 5.3139497642900696e-06,
      "loss": 0.0999603271484375,
      "step": 5595
    },
    {
      "epoch": 0.779767296035672,
      "grad_norm": 0.34746482968330383,
      "learning_rate": 5.307552045952445e-06,
      "loss": 0.09080934524536133,
      "step": 5596
    },
    {
      "epoch": 0.7799066397268863,
      "grad_norm": 0.24995028972625732,
      "learning_rate": 5.3011575919413214e-06,
      "loss": 0.07035255432128906,
      "step": 5597
    },
    {
      "epoch": 0.7800459834181007,
      "grad_norm": 0.3719792366027832,
      "learning_rate": 5.294766403677407e-06,
      "loss": 0.08910274505615234,
      "step": 5598
    },
    {
      "epoch": 0.7801853271093151,
      "grad_norm": 0.5692002177238464,
      "learning_rate": 5.288378482580671e-06,
      "loss": 0.09650802612304688,
      "step": 5599
    },
    {
      "epoch": 0.7803246708005295,
      "grad_norm": 0.3977873921394348,
      "learning_rate": 5.281993830070357e-06,
      "loss": 0.08350563049316406,
      "step": 5600
    },
    {
      "epoch": 0.7804640144917439,
      "grad_norm": 0.4569106996059418,
      "learning_rate": 5.275612447565e-06,
      "loss": 0.08640003204345703,
      "step": 5601
    },
    {
      "epoch": 0.7806033581829582,
      "grad_norm": 0.44899600744247437,
      "learning_rate": 5.269234336482396e-06,
      "loss": 0.10199260711669922,
      "step": 5602
    },
    {
      "epoch": 0.7807427018741726,
      "grad_norm": 0.4218426048755646,
      "learning_rate": 5.262859498239614e-06,
      "loss": 0.10093879699707031,
      "step": 5603
    },
    {
      "epoch": 0.780882045565387,
      "grad_norm": 0.7482976913452148,
      "learning_rate": 5.256487934252992e-06,
      "loss": 0.11282920837402344,
      "step": 5604
    },
    {
      "epoch": 0.7810213892566014,
      "grad_norm": 0.6332219243049622,
      "learning_rate": 5.25011964593815e-06,
      "loss": 0.11429595947265625,
      "step": 5605
    },
    {
      "epoch": 0.7811607329478157,
      "grad_norm": 0.4180949628353119,
      "learning_rate": 5.243754634709983e-06,
      "loss": 0.07964134216308594,
      "step": 5606
    },
    {
      "epoch": 0.7813000766390301,
      "grad_norm": 0.3166016638278961,
      "learning_rate": 5.237392901982641e-06,
      "loss": 0.08282470703125,
      "step": 5607
    },
    {
      "epoch": 0.7814394203302445,
      "grad_norm": 0.26847293972969055,
      "learning_rate": 5.2310344491695585e-06,
      "loss": 0.06253623962402344,
      "step": 5608
    },
    {
      "epoch": 0.781578764021459,
      "grad_norm": 0.6265087723731995,
      "learning_rate": 5.2246792776834466e-06,
      "loss": 0.10777664184570312,
      "step": 5609
    },
    {
      "epoch": 0.7817181077126734,
      "grad_norm": 0.5848155617713928,
      "learning_rate": 5.218327388936273e-06,
      "loss": 0.10179519653320312,
      "step": 5610
    },
    {
      "epoch": 0.7818574514038877,
      "grad_norm": 0.22869634628295898,
      "learning_rate": 5.211978784339275e-06,
      "loss": 0.056471824645996094,
      "step": 5611
    },
    {
      "epoch": 0.7819967950951021,
      "grad_norm": 0.639868974685669,
      "learning_rate": 5.205633465302977e-06,
      "loss": 0.11487007141113281,
      "step": 5612
    },
    {
      "epoch": 0.7821361387863165,
      "grad_norm": 0.4403250515460968,
      "learning_rate": 5.199291433237168e-06,
      "loss": 0.08856010437011719,
      "step": 5613
    },
    {
      "epoch": 0.7822754824775309,
      "grad_norm": 0.41663333773612976,
      "learning_rate": 5.192952689550894e-06,
      "loss": 0.08444833755493164,
      "step": 5614
    },
    {
      "epoch": 0.7824148261687452,
      "grad_norm": 0.2434936910867691,
      "learning_rate": 5.186617235652478e-06,
      "loss": 0.06649303436279297,
      "step": 5615
    },
    {
      "epoch": 0.7825541698599596,
      "grad_norm": 0.3980183005332947,
      "learning_rate": 5.180285072949516e-06,
      "loss": 0.09190559387207031,
      "step": 5616
    },
    {
      "epoch": 0.782693513551174,
      "grad_norm": 0.1845494508743286,
      "learning_rate": 5.173956202848878e-06,
      "loss": 0.05714225769042969,
      "step": 5617
    },
    {
      "epoch": 0.7828328572423884,
      "grad_norm": 0.5550017952919006,
      "learning_rate": 5.167630626756681e-06,
      "loss": 0.11956214904785156,
      "step": 5618
    },
    {
      "epoch": 0.7829722009336028,
      "grad_norm": 0.6702423691749573,
      "learning_rate": 5.161308346078335e-06,
      "loss": 0.12084197998046875,
      "step": 5619
    },
    {
      "epoch": 0.7831115446248171,
      "grad_norm": 0.4007166922092438,
      "learning_rate": 5.154989362218501e-06,
      "loss": 0.075897216796875,
      "step": 5620
    },
    {
      "epoch": 0.7832508883160315,
      "grad_norm": 0.4318609833717346,
      "learning_rate": 5.148673676581109e-06,
      "loss": 0.09497261047363281,
      "step": 5621
    },
    {
      "epoch": 0.7833902320072459,
      "grad_norm": 0.39287474751472473,
      "learning_rate": 5.142361290569365e-06,
      "loss": 0.09283447265625,
      "step": 5622
    },
    {
      "epoch": 0.7835295756984603,
      "grad_norm": 0.35759952664375305,
      "learning_rate": 5.136052205585735e-06,
      "loss": 0.08503532409667969,
      "step": 5623
    },
    {
      "epoch": 0.7836689193896746,
      "grad_norm": 0.4242338240146637,
      "learning_rate": 5.129746423031967e-06,
      "loss": 0.0792999267578125,
      "step": 5624
    },
    {
      "epoch": 0.783808263080889,
      "grad_norm": 0.41541731357574463,
      "learning_rate": 5.123443944309039e-06,
      "loss": 0.08974647521972656,
      "step": 5625
    },
    {
      "epoch": 0.7839476067721034,
      "grad_norm": 0.9979515671730042,
      "learning_rate": 5.1171447708172285e-06,
      "loss": 0.14743995666503906,
      "step": 5626
    },
    {
      "epoch": 0.7840869504633178,
      "grad_norm": 0.4515165388584137,
      "learning_rate": 5.110848903956076e-06,
      "loss": 0.07520675659179688,
      "step": 5627
    },
    {
      "epoch": 0.7842262941545322,
      "grad_norm": 0.5661170482635498,
      "learning_rate": 5.104556345124363e-06,
      "loss": 0.1016998291015625,
      "step": 5628
    },
    {
      "epoch": 0.7843656378457465,
      "grad_norm": 0.3119969964027405,
      "learning_rate": 5.098267095720164e-06,
      "loss": 0.0758056640625,
      "step": 5629
    },
    {
      "epoch": 0.7845049815369609,
      "grad_norm": 0.45896095037460327,
      "learning_rate": 5.091981157140808e-06,
      "loss": 0.09566307067871094,
      "step": 5630
    },
    {
      "epoch": 0.7846443252281753,
      "grad_norm": 0.20681026577949524,
      "learning_rate": 5.085698530782885e-06,
      "loss": 0.0582122802734375,
      "step": 5631
    },
    {
      "epoch": 0.7847836689193897,
      "grad_norm": 0.367009699344635,
      "learning_rate": 5.079419218042243e-06,
      "loss": 0.07361793518066406,
      "step": 5632
    },
    {
      "epoch": 0.784923012610604,
      "grad_norm": 0.3429175913333893,
      "learning_rate": 5.073143220314007e-06,
      "loss": 0.0737762451171875,
      "step": 5633
    },
    {
      "epoch": 0.7850623563018184,
      "grad_norm": 0.26564738154411316,
      "learning_rate": 5.066870538992568e-06,
      "loss": 0.06990432739257812,
      "step": 5634
    },
    {
      "epoch": 0.7852016999930328,
      "grad_norm": 0.5979676842689514,
      "learning_rate": 5.060601175471567e-06,
      "loss": 0.09690093994140625,
      "step": 5635
    },
    {
      "epoch": 0.7853410436842472,
      "grad_norm": 0.6289647817611694,
      "learning_rate": 5.054335131143906e-06,
      "loss": 0.11433601379394531,
      "step": 5636
    },
    {
      "epoch": 0.7854803873754616,
      "grad_norm": 0.316021203994751,
      "learning_rate": 5.0480724074017625e-06,
      "loss": 0.08373260498046875,
      "step": 5637
    },
    {
      "epoch": 0.7856197310666759,
      "grad_norm": 0.7286697030067444,
      "learning_rate": 5.041813005636578e-06,
      "loss": 0.09458351135253906,
      "step": 5638
    },
    {
      "epoch": 0.7857590747578903,
      "grad_norm": 0.41320085525512695,
      "learning_rate": 5.035556927239036e-06,
      "loss": 0.08198833465576172,
      "step": 5639
    },
    {
      "epoch": 0.7858984184491047,
      "grad_norm": 0.4740714728832245,
      "learning_rate": 5.029304173599107e-06,
      "loss": 0.083587646484375,
      "step": 5640
    },
    {
      "epoch": 0.7860377621403191,
      "grad_norm": 0.5497187376022339,
      "learning_rate": 5.023054746106e-06,
      "loss": 0.12179946899414062,
      "step": 5641
    },
    {
      "epoch": 0.7861771058315334,
      "grad_norm": 0.4507594108581543,
      "learning_rate": 5.016808646148204e-06,
      "loss": 0.09488868713378906,
      "step": 5642
    },
    {
      "epoch": 0.7863164495227478,
      "grad_norm": 0.28897884488105774,
      "learning_rate": 5.0105658751134464e-06,
      "loss": 0.06509876251220703,
      "step": 5643
    },
    {
      "epoch": 0.7864557932139622,
      "grad_norm": 0.7803561091423035,
      "learning_rate": 5.0043264343887395e-06,
      "loss": 0.12209892272949219,
      "step": 5644
    },
    {
      "epoch": 0.7865951369051766,
      "grad_norm": 0.26965487003326416,
      "learning_rate": 4.998090325360346e-06,
      "loss": 0.07392120361328125,
      "step": 5645
    },
    {
      "epoch": 0.786734480596391,
      "grad_norm": 0.2806442379951477,
      "learning_rate": 4.991857549413784e-06,
      "loss": 0.06663131713867188,
      "step": 5646
    },
    {
      "epoch": 0.7868738242876053,
      "grad_norm": 0.37051138281822205,
      "learning_rate": 4.9856281079338265e-06,
      "loss": 0.08903694152832031,
      "step": 5647
    },
    {
      "epoch": 0.7870131679788197,
      "grad_norm": 0.48712053894996643,
      "learning_rate": 4.979402002304519e-06,
      "loss": 0.08421993255615234,
      "step": 5648
    },
    {
      "epoch": 0.7871525116700342,
      "grad_norm": 0.33820393681526184,
      "learning_rate": 4.973179233909167e-06,
      "loss": 0.07174873352050781,
      "step": 5649
    },
    {
      "epoch": 0.7872918553612486,
      "grad_norm": 0.28863459825515747,
      "learning_rate": 4.966959804130314e-06,
      "loss": 0.07728958129882812,
      "step": 5650
    },
    {
      "epoch": 0.787431199052463,
      "grad_norm": 0.4273664057254791,
      "learning_rate": 4.96074371434979e-06,
      "loss": 0.0938119888305664,
      "step": 5651
    },
    {
      "epoch": 0.7875705427436773,
      "grad_norm": 0.37286117672920227,
      "learning_rate": 4.954530965948654e-06,
      "loss": 0.08338451385498047,
      "step": 5652
    },
    {
      "epoch": 0.7877098864348917,
      "grad_norm": 0.2729094624519348,
      "learning_rate": 4.948321560307248e-06,
      "loss": 0.07269859313964844,
      "step": 5653
    },
    {
      "epoch": 0.7878492301261061,
      "grad_norm": 0.33811935782432556,
      "learning_rate": 4.942115498805151e-06,
      "loss": 0.067626953125,
      "step": 5654
    },
    {
      "epoch": 0.7879885738173205,
      "grad_norm": 0.20715104043483734,
      "learning_rate": 4.9359127828212125e-06,
      "loss": 0.05515241622924805,
      "step": 5655
    },
    {
      "epoch": 0.7881279175085348,
      "grad_norm": 0.19436496496200562,
      "learning_rate": 4.929713413733548e-06,
      "loss": 0.05989646911621094,
      "step": 5656
    },
    {
      "epoch": 0.7882672611997492,
      "grad_norm": 0.26916325092315674,
      "learning_rate": 4.92351739291949e-06,
      "loss": 0.06653213500976562,
      "step": 5657
    },
    {
      "epoch": 0.7884066048909636,
      "grad_norm": 0.4645390510559082,
      "learning_rate": 4.917324721755665e-06,
      "loss": 0.08740425109863281,
      "step": 5658
    },
    {
      "epoch": 0.788545948582178,
      "grad_norm": 0.5124428272247314,
      "learning_rate": 4.911135401617948e-06,
      "loss": 0.09937286376953125,
      "step": 5659
    },
    {
      "epoch": 0.7886852922733923,
      "grad_norm": 0.42513900995254517,
      "learning_rate": 4.904949433881457e-06,
      "loss": 0.08971786499023438,
      "step": 5660
    },
    {
      "epoch": 0.7888246359646067,
      "grad_norm": 0.782625138759613,
      "learning_rate": 4.8987668199205796e-06,
      "loss": 0.114959716796875,
      "step": 5661
    },
    {
      "epoch": 0.7889639796558211,
      "grad_norm": 0.17023731768131256,
      "learning_rate": 4.892587561108942e-06,
      "loss": 0.052082061767578125,
      "step": 5662
    },
    {
      "epoch": 0.7891033233470355,
      "grad_norm": 0.6163641214370728,
      "learning_rate": 4.886411658819448e-06,
      "loss": 0.10512351989746094,
      "step": 5663
    },
    {
      "epoch": 0.7892426670382499,
      "grad_norm": 0.7495293021202087,
      "learning_rate": 4.880239114424226e-06,
      "loss": 0.12883567810058594,
      "step": 5664
    },
    {
      "epoch": 0.7893820107294642,
      "grad_norm": 0.4445473551750183,
      "learning_rate": 4.874069929294685e-06,
      "loss": 0.10071754455566406,
      "step": 5665
    },
    {
      "epoch": 0.7895213544206786,
      "grad_norm": 0.4127316176891327,
      "learning_rate": 4.867904104801477e-06,
      "loss": 0.08897018432617188,
      "step": 5666
    },
    {
      "epoch": 0.789660698111893,
      "grad_norm": 0.45221707224845886,
      "learning_rate": 4.861741642314511e-06,
      "loss": 0.0730133056640625,
      "step": 5667
    },
    {
      "epoch": 0.7898000418031074,
      "grad_norm": 0.35365915298461914,
      "learning_rate": 4.85558254320293e-06,
      "loss": 0.0850381851196289,
      "step": 5668
    },
    {
      "epoch": 0.7899393854943217,
      "grad_norm": 0.22997444868087769,
      "learning_rate": 4.849426808835156e-06,
      "loss": 0.058258056640625,
      "step": 5669
    },
    {
      "epoch": 0.7900787291855361,
      "grad_norm": 0.3983100354671478,
      "learning_rate": 4.843274440578856e-06,
      "loss": 0.0897369384765625,
      "step": 5670
    },
    {
      "epoch": 0.7902180728767505,
      "grad_norm": 0.28290966153144836,
      "learning_rate": 4.837125439800936e-06,
      "loss": 0.07515859603881836,
      "step": 5671
    },
    {
      "epoch": 0.7903574165679649,
      "grad_norm": 0.545756459236145,
      "learning_rate": 4.830979807867577e-06,
      "loss": 0.09769821166992188,
      "step": 5672
    },
    {
      "epoch": 0.7904967602591793,
      "grad_norm": 0.4839855432510376,
      "learning_rate": 4.824837546144183e-06,
      "loss": 0.07847309112548828,
      "step": 5673
    },
    {
      "epoch": 0.7906361039503936,
      "grad_norm": 0.48080193996429443,
      "learning_rate": 4.818698655995437e-06,
      "loss": 0.09499359130859375,
      "step": 5674
    },
    {
      "epoch": 0.790775447641608,
      "grad_norm": 0.47343459725379944,
      "learning_rate": 4.812563138785249e-06,
      "loss": 0.1078948974609375,
      "step": 5675
    },
    {
      "epoch": 0.7909147913328224,
      "grad_norm": 0.32694512605667114,
      "learning_rate": 4.806430995876796e-06,
      "loss": 0.08149242401123047,
      "step": 5676
    },
    {
      "epoch": 0.7910541350240368,
      "grad_norm": 0.22882424294948578,
      "learning_rate": 4.800302228632505e-06,
      "loss": 0.06014060974121094,
      "step": 5677
    },
    {
      "epoch": 0.7911934787152511,
      "grad_norm": 0.3147236108779907,
      "learning_rate": 4.7941768384140465e-06,
      "loss": 0.0701303482055664,
      "step": 5678
    },
    {
      "epoch": 0.7913328224064655,
      "grad_norm": 0.30596208572387695,
      "learning_rate": 4.788054826582334e-06,
      "loss": 0.07063865661621094,
      "step": 5679
    },
    {
      "epoch": 0.7914721660976799,
      "grad_norm": 0.420888751745224,
      "learning_rate": 4.781936194497543e-06,
      "loss": 0.08429145812988281,
      "step": 5680
    },
    {
      "epoch": 0.7916115097888943,
      "grad_norm": 0.46405813097953796,
      "learning_rate": 4.7758209435191e-06,
      "loss": 0.10609626770019531,
      "step": 5681
    },
    {
      "epoch": 0.7917508534801087,
      "grad_norm": 0.34289026260375977,
      "learning_rate": 4.769709075005673e-06,
      "loss": 0.07353401184082031,
      "step": 5682
    },
    {
      "epoch": 0.791890197171323,
      "grad_norm": 0.4633822739124298,
      "learning_rate": 4.763600590315167e-06,
      "loss": 0.08964920043945312,
      "step": 5683
    },
    {
      "epoch": 0.7920295408625374,
      "grad_norm": 0.4054756164550781,
      "learning_rate": 4.757495490804758e-06,
      "loss": 0.08865165710449219,
      "step": 5684
    },
    {
      "epoch": 0.7921688845537518,
      "grad_norm": 0.6909626722335815,
      "learning_rate": 4.751393777830864e-06,
      "loss": 0.13172340393066406,
      "step": 5685
    },
    {
      "epoch": 0.7923082282449662,
      "grad_norm": 0.5144474506378174,
      "learning_rate": 4.745295452749137e-06,
      "loss": 0.09984302520751953,
      "step": 5686
    },
    {
      "epoch": 0.7924475719361805,
      "grad_norm": 0.44301581382751465,
      "learning_rate": 4.739200516914488e-06,
      "loss": 0.0847940444946289,
      "step": 5687
    },
    {
      "epoch": 0.7925869156273949,
      "grad_norm": 0.43679583072662354,
      "learning_rate": 4.7331089716810866e-06,
      "loss": 0.097686767578125,
      "step": 5688
    },
    {
      "epoch": 0.7927262593186094,
      "grad_norm": 0.5174006223678589,
      "learning_rate": 4.727020818402311e-06,
      "loss": 0.09666252136230469,
      "step": 5689
    },
    {
      "epoch": 0.7928656030098238,
      "grad_norm": 0.25460925698280334,
      "learning_rate": 4.720936058430818e-06,
      "loss": 0.07010602951049805,
      "step": 5690
    },
    {
      "epoch": 0.7930049467010382,
      "grad_norm": 0.4205535054206848,
      "learning_rate": 4.714854693118515e-06,
      "loss": 0.09518051147460938,
      "step": 5691
    },
    {
      "epoch": 0.7931442903922525,
      "grad_norm": 0.2479642778635025,
      "learning_rate": 4.708776723816528e-06,
      "loss": 0.06893539428710938,
      "step": 5692
    },
    {
      "epoch": 0.7932836340834669,
      "grad_norm": 0.3316034972667694,
      "learning_rate": 4.702702151875253e-06,
      "loss": 0.08754539489746094,
      "step": 5693
    },
    {
      "epoch": 0.7934229777746813,
      "grad_norm": 0.4822218120098114,
      "learning_rate": 4.696630978644312e-06,
      "loss": 0.09250640869140625,
      "step": 5694
    },
    {
      "epoch": 0.7935623214658957,
      "grad_norm": 0.49893999099731445,
      "learning_rate": 4.690563205472589e-06,
      "loss": 0.10248756408691406,
      "step": 5695
    },
    {
      "epoch": 0.79370166515711,
      "grad_norm": 0.5823110342025757,
      "learning_rate": 4.684498833708198e-06,
      "loss": 0.12551307678222656,
      "step": 5696
    },
    {
      "epoch": 0.7938410088483244,
      "grad_norm": 0.2669805884361267,
      "learning_rate": 4.678437864698507e-06,
      "loss": 0.07390403747558594,
      "step": 5697
    },
    {
      "epoch": 0.7939803525395388,
      "grad_norm": 0.42157742381095886,
      "learning_rate": 4.672380299790131e-06,
      "loss": 0.08240509033203125,
      "step": 5698
    },
    {
      "epoch": 0.7941196962307532,
      "grad_norm": 0.262778103351593,
      "learning_rate": 4.6663261403289165e-06,
      "loss": 0.06632804870605469,
      "step": 5699
    },
    {
      "epoch": 0.7942590399219676,
      "grad_norm": 0.2866041660308838,
      "learning_rate": 4.660275387659954e-06,
      "loss": 0.07683277130126953,
      "step": 5700
    },
    {
      "epoch": 0.7943983836131819,
      "grad_norm": 0.3691703677177429,
      "learning_rate": 4.654228043127589e-06,
      "loss": 0.07423877716064453,
      "step": 5701
    },
    {
      "epoch": 0.7945377273043963,
      "grad_norm": 0.48941630125045776,
      "learning_rate": 4.6481841080754064e-06,
      "loss": 0.09962081909179688,
      "step": 5702
    },
    {
      "epoch": 0.7946770709956107,
      "grad_norm": 0.4583169221878052,
      "learning_rate": 4.642143583846223e-06,
      "loss": 0.09332275390625,
      "step": 5703
    },
    {
      "epoch": 0.7948164146868251,
      "grad_norm": 0.45926913619041443,
      "learning_rate": 4.636106471782116e-06,
      "loss": 0.08583831787109375,
      "step": 5704
    },
    {
      "epoch": 0.7949557583780394,
      "grad_norm": 0.3149871528148651,
      "learning_rate": 4.630072773224381e-06,
      "loss": 0.08464860916137695,
      "step": 5705
    },
    {
      "epoch": 0.7950951020692538,
      "grad_norm": 0.397433340549469,
      "learning_rate": 4.62404248951358e-06,
      "loss": 0.08165359497070312,
      "step": 5706
    },
    {
      "epoch": 0.7952344457604682,
      "grad_norm": 0.2532755732536316,
      "learning_rate": 4.618015621989493e-06,
      "loss": 0.05828380584716797,
      "step": 5707
    },
    {
      "epoch": 0.7953737894516826,
      "grad_norm": 0.4045651853084564,
      "learning_rate": 4.6119921719911596e-06,
      "loss": 0.08685302734375,
      "step": 5708
    },
    {
      "epoch": 0.795513133142897,
      "grad_norm": 0.35737645626068115,
      "learning_rate": 4.605972140856856e-06,
      "loss": 0.08438873291015625,
      "step": 5709
    },
    {
      "epoch": 0.7956524768341113,
      "grad_norm": 0.6652624607086182,
      "learning_rate": 4.599955529924089e-06,
      "loss": 0.09720230102539062,
      "step": 5710
    },
    {
      "epoch": 0.7957918205253257,
      "grad_norm": 0.2874082326889038,
      "learning_rate": 4.593942340529612e-06,
      "loss": 0.07478523254394531,
      "step": 5711
    },
    {
      "epoch": 0.7959311642165401,
      "grad_norm": 0.5558286905288696,
      "learning_rate": 4.587932574009417e-06,
      "loss": 0.117584228515625,
      "step": 5712
    },
    {
      "epoch": 0.7960705079077545,
      "grad_norm": 0.5126045942306519,
      "learning_rate": 4.58192623169875e-06,
      "loss": 0.10787773132324219,
      "step": 5713
    },
    {
      "epoch": 0.7962098515989688,
      "grad_norm": 0.4443994462490082,
      "learning_rate": 4.575923314932072e-06,
      "loss": 0.10480117797851562,
      "step": 5714
    },
    {
      "epoch": 0.7963491952901832,
      "grad_norm": 0.2733587622642517,
      "learning_rate": 4.56992382504309e-06,
      "loss": 0.06936526298522949,
      "step": 5715
    },
    {
      "epoch": 0.7964885389813976,
      "grad_norm": 0.43731874227523804,
      "learning_rate": 4.563927763364759e-06,
      "loss": 0.08162117004394531,
      "step": 5716
    },
    {
      "epoch": 0.796627882672612,
      "grad_norm": 0.4961440861225128,
      "learning_rate": 4.557935131229274e-06,
      "loss": 0.11594009399414062,
      "step": 5717
    },
    {
      "epoch": 0.7967672263638264,
      "grad_norm": 0.6191819906234741,
      "learning_rate": 4.5519459299680465e-06,
      "loss": 0.12314796447753906,
      "step": 5718
    },
    {
      "epoch": 0.7969065700550407,
      "grad_norm": 0.33314236998558044,
      "learning_rate": 4.545960160911758e-06,
      "loss": 0.08730030059814453,
      "step": 5719
    },
    {
      "epoch": 0.7970459137462551,
      "grad_norm": 0.5012186169624329,
      "learning_rate": 4.539977825390296e-06,
      "loss": 0.07854652404785156,
      "step": 5720
    },
    {
      "epoch": 0.7971852574374695,
      "grad_norm": 0.3092936873435974,
      "learning_rate": 4.533998924732801e-06,
      "loss": 0.08112525939941406,
      "step": 5721
    },
    {
      "epoch": 0.7973246011286839,
      "grad_norm": 0.30831006169319153,
      "learning_rate": 4.528023460267648e-06,
      "loss": 0.06740951538085938,
      "step": 5722
    },
    {
      "epoch": 0.7974639448198982,
      "grad_norm": 0.7222568392753601,
      "learning_rate": 4.52205143332245e-06,
      "loss": 0.1507091522216797,
      "step": 5723
    },
    {
      "epoch": 0.7976032885111126,
      "grad_norm": 0.3376224637031555,
      "learning_rate": 4.516082845224063e-06,
      "loss": 0.09464073181152344,
      "step": 5724
    },
    {
      "epoch": 0.797742632202327,
      "grad_norm": 0.4256724417209625,
      "learning_rate": 4.510117697298564e-06,
      "loss": 0.08450889587402344,
      "step": 5725
    },
    {
      "epoch": 0.7978819758935414,
      "grad_norm": 0.3248347043991089,
      "learning_rate": 4.504155990871266e-06,
      "loss": 0.0777425765991211,
      "step": 5726
    },
    {
      "epoch": 0.7980213195847558,
      "grad_norm": 0.21624957025051117,
      "learning_rate": 4.498197727266738e-06,
      "loss": 0.07103157043457031,
      "step": 5727
    },
    {
      "epoch": 0.7981606632759701,
      "grad_norm": 0.38018715381622314,
      "learning_rate": 4.4922429078087545e-06,
      "loss": 0.08751296997070312,
      "step": 5728
    },
    {
      "epoch": 0.7983000069671845,
      "grad_norm": 0.33947092294692993,
      "learning_rate": 4.486291533820351e-06,
      "loss": 0.08021354675292969,
      "step": 5729
    },
    {
      "epoch": 0.798439350658399,
      "grad_norm": 0.2640308737754822,
      "learning_rate": 4.480343606623789e-06,
      "loss": 0.07692337036132812,
      "step": 5730
    },
    {
      "epoch": 0.7985786943496134,
      "grad_norm": 0.3216187059879303,
      "learning_rate": 4.474399127540561e-06,
      "loss": 0.07148361206054688,
      "step": 5731
    },
    {
      "epoch": 0.7987180380408277,
      "grad_norm": 0.2968781590461731,
      "learning_rate": 4.4684580978913815e-06,
      "loss": 0.08005905151367188,
      "step": 5732
    },
    {
      "epoch": 0.7988573817320421,
      "grad_norm": 0.777054488658905,
      "learning_rate": 4.462520518996225e-06,
      "loss": 0.12508010864257812,
      "step": 5733
    },
    {
      "epoch": 0.7989967254232565,
      "grad_norm": 0.2634645402431488,
      "learning_rate": 4.456586392174285e-06,
      "loss": 0.074249267578125,
      "step": 5734
    },
    {
      "epoch": 0.7991360691144709,
      "grad_norm": 0.47465378046035767,
      "learning_rate": 4.450655718743988e-06,
      "loss": 0.09573936462402344,
      "step": 5735
    },
    {
      "epoch": 0.7992754128056853,
      "grad_norm": 0.4410315454006195,
      "learning_rate": 4.444728500022988e-06,
      "loss": 0.07931900024414062,
      "step": 5736
    },
    {
      "epoch": 0.7994147564968996,
      "grad_norm": 0.423197865486145,
      "learning_rate": 4.4388047373281815e-06,
      "loss": 0.10241508483886719,
      "step": 5737
    },
    {
      "epoch": 0.799554100188114,
      "grad_norm": 0.3805861175060272,
      "learning_rate": 4.432884431975699e-06,
      "loss": 0.08723258972167969,
      "step": 5738
    },
    {
      "epoch": 0.7996934438793284,
      "grad_norm": 0.4521794617176056,
      "learning_rate": 4.426967585280888e-06,
      "loss": 0.07951927185058594,
      "step": 5739
    },
    {
      "epoch": 0.7998327875705428,
      "grad_norm": 0.4139499366283417,
      "learning_rate": 4.4210541985583455e-06,
      "loss": 0.08146524429321289,
      "step": 5740
    },
    {
      "epoch": 0.7999721312617571,
      "grad_norm": 0.6790738701820374,
      "learning_rate": 4.415144273121883e-06,
      "loss": 0.10047531127929688,
      "step": 5741
    },
    {
      "epoch": 0.8001114749529715,
      "grad_norm": 0.6247254014015198,
      "learning_rate": 4.409237810284559e-06,
      "loss": 0.10669422149658203,
      "step": 5742
    },
    {
      "epoch": 0.8002508186441859,
      "grad_norm": 0.7914243936538696,
      "learning_rate": 4.4033348113586465e-06,
      "loss": 0.09213638305664062,
      "step": 5743
    },
    {
      "epoch": 0.8003901623354003,
      "grad_norm": 0.2749635875225067,
      "learning_rate": 4.39743527765566e-06,
      "loss": 0.06101489067077637,
      "step": 5744
    },
    {
      "epoch": 0.8005295060266147,
      "grad_norm": 0.33767813444137573,
      "learning_rate": 4.391539210486346e-06,
      "loss": 0.07876014709472656,
      "step": 5745
    },
    {
      "epoch": 0.800668849717829,
      "grad_norm": 0.42112815380096436,
      "learning_rate": 4.385646611160674e-06,
      "loss": 0.10757255554199219,
      "step": 5746
    },
    {
      "epoch": 0.8008081934090434,
      "grad_norm": 0.3489990234375,
      "learning_rate": 4.379757480987836e-06,
      "loss": 0.08538246154785156,
      "step": 5747
    },
    {
      "epoch": 0.8009475371002578,
      "grad_norm": 0.36643949151039124,
      "learning_rate": 4.373871821276272e-06,
      "loss": 0.08438396453857422,
      "step": 5748
    },
    {
      "epoch": 0.8010868807914722,
      "grad_norm": 0.3192828595638275,
      "learning_rate": 4.367989633333642e-06,
      "loss": 0.07179450988769531,
      "step": 5749
    },
    {
      "epoch": 0.8012262244826865,
      "grad_norm": 0.548864483833313,
      "learning_rate": 4.362110918466826e-06,
      "loss": 0.10724449157714844,
      "step": 5750
    },
    {
      "epoch": 0.8013655681739009,
      "grad_norm": 0.3951129913330078,
      "learning_rate": 4.356235677981952e-06,
      "loss": 0.08402442932128906,
      "step": 5751
    },
    {
      "epoch": 0.8015049118651153,
      "grad_norm": 0.5227956771850586,
      "learning_rate": 4.350363913184355e-06,
      "loss": 0.10839080810546875,
      "step": 5752
    },
    {
      "epoch": 0.8016442555563297,
      "grad_norm": 0.45943084359169006,
      "learning_rate": 4.3444956253786044e-06,
      "loss": 0.10425376892089844,
      "step": 5753
    },
    {
      "epoch": 0.801783599247544,
      "grad_norm": 0.35157501697540283,
      "learning_rate": 4.338630815868505e-06,
      "loss": 0.08690738677978516,
      "step": 5754
    },
    {
      "epoch": 0.8019229429387584,
      "grad_norm": 0.4287568926811218,
      "learning_rate": 4.3327694859570824e-06,
      "loss": 0.08146858215332031,
      "step": 5755
    },
    {
      "epoch": 0.8020622866299728,
      "grad_norm": 0.5608795881271362,
      "learning_rate": 4.326911636946603e-06,
      "loss": 0.08159065246582031,
      "step": 5756
    },
    {
      "epoch": 0.8022016303211872,
      "grad_norm": 0.4520722031593323,
      "learning_rate": 4.321057270138525e-06,
      "loss": 0.08691787719726562,
      "step": 5757
    },
    {
      "epoch": 0.8023409740124016,
      "grad_norm": 0.29534855484962463,
      "learning_rate": 4.315206386833562e-06,
      "loss": 0.06525421142578125,
      "step": 5758
    },
    {
      "epoch": 0.8024803177036159,
      "grad_norm": 0.7585322260856628,
      "learning_rate": 4.309358988331658e-06,
      "loss": 0.12059879302978516,
      "step": 5759
    },
    {
      "epoch": 0.8026196613948303,
      "grad_norm": 0.39878302812576294,
      "learning_rate": 4.303515075931957e-06,
      "loss": 0.08762168884277344,
      "step": 5760
    },
    {
      "epoch": 0.8027590050860447,
      "grad_norm": 0.6048856377601624,
      "learning_rate": 4.297674650932848e-06,
      "loss": 0.101287841796875,
      "step": 5761
    },
    {
      "epoch": 0.8028983487772591,
      "grad_norm": 0.34144991636276245,
      "learning_rate": 4.2918377146319505e-06,
      "loss": 0.07674407958984375,
      "step": 5762
    },
    {
      "epoch": 0.8030376924684735,
      "grad_norm": 0.32711613178253174,
      "learning_rate": 4.286004268326085e-06,
      "loss": 0.07323074340820312,
      "step": 5763
    },
    {
      "epoch": 0.8031770361596878,
      "grad_norm": 0.49057304859161377,
      "learning_rate": 4.280174313311311e-06,
      "loss": 0.09510993957519531,
      "step": 5764
    },
    {
      "epoch": 0.8033163798509022,
      "grad_norm": 0.5425952672958374,
      "learning_rate": 4.274347850882916e-06,
      "loss": 0.09586715698242188,
      "step": 5765
    },
    {
      "epoch": 0.8034557235421166,
      "grad_norm": 0.20432499051094055,
      "learning_rate": 4.26852488233541e-06,
      "loss": 0.06329059600830078,
      "step": 5766
    },
    {
      "epoch": 0.803595067233331,
      "grad_norm": 0.3065548837184906,
      "learning_rate": 4.26270540896252e-06,
      "loss": 0.07519149780273438,
      "step": 5767
    },
    {
      "epoch": 0.8037344109245453,
      "grad_norm": 0.3480375111103058,
      "learning_rate": 4.256889432057194e-06,
      "loss": 0.08692550659179688,
      "step": 5768
    },
    {
      "epoch": 0.8038737546157597,
      "grad_norm": 0.5164533853530884,
      "learning_rate": 4.251076952911615e-06,
      "loss": 0.09987831115722656,
      "step": 5769
    },
    {
      "epoch": 0.8040130983069742,
      "grad_norm": 0.6816145181655884,
      "learning_rate": 4.245267972817189e-06,
      "loss": 0.09496116638183594,
      "step": 5770
    },
    {
      "epoch": 0.8041524419981886,
      "grad_norm": 0.45076847076416016,
      "learning_rate": 4.239462493064525e-06,
      "loss": 0.08380699157714844,
      "step": 5771
    },
    {
      "epoch": 0.804291785689403,
      "grad_norm": 0.4026052951812744,
      "learning_rate": 4.233660514943483e-06,
      "loss": 0.07911872863769531,
      "step": 5772
    },
    {
      "epoch": 0.8044311293806173,
      "grad_norm": 0.6147128939628601,
      "learning_rate": 4.227862039743118e-06,
      "loss": 0.09858131408691406,
      "step": 5773
    },
    {
      "epoch": 0.8045704730718317,
      "grad_norm": 0.5074610114097595,
      "learning_rate": 4.2220670687517275e-06,
      "loss": 0.09175968170166016,
      "step": 5774
    },
    {
      "epoch": 0.8047098167630461,
      "grad_norm": 0.22406648099422455,
      "learning_rate": 4.216275603256814e-06,
      "loss": 0.06613349914550781,
      "step": 5775
    },
    {
      "epoch": 0.8048491604542605,
      "grad_norm": 0.4787399172782898,
      "learning_rate": 4.210487644545112e-06,
      "loss": 0.10070323944091797,
      "step": 5776
    },
    {
      "epoch": 0.8049885041454748,
      "grad_norm": 0.3241536021232605,
      "learning_rate": 4.204703193902582e-06,
      "loss": 0.08920478820800781,
      "step": 5777
    },
    {
      "epoch": 0.8051278478366892,
      "grad_norm": 0.399618536233902,
      "learning_rate": 4.198922252614388e-06,
      "loss": 0.10127449035644531,
      "step": 5778
    },
    {
      "epoch": 0.8052671915279036,
      "grad_norm": 0.6698436737060547,
      "learning_rate": 4.193144821964918e-06,
      "loss": 0.08912849426269531,
      "step": 5779
    },
    {
      "epoch": 0.805406535219118,
      "grad_norm": 0.2682548761367798,
      "learning_rate": 4.1873709032377926e-06,
      "loss": 0.07201957702636719,
      "step": 5780
    },
    {
      "epoch": 0.8055458789103324,
      "grad_norm": 0.359358549118042,
      "learning_rate": 4.181600497715852e-06,
      "loss": 0.08260917663574219,
      "step": 5781
    },
    {
      "epoch": 0.8056852226015467,
      "grad_norm": 0.41767463088035583,
      "learning_rate": 4.175833606681132e-06,
      "loss": 0.06744194030761719,
      "step": 5782
    },
    {
      "epoch": 0.8058245662927611,
      "grad_norm": 0.30972811579704285,
      "learning_rate": 4.17007023141492e-06,
      "loss": 0.0800933837890625,
      "step": 5783
    },
    {
      "epoch": 0.8059639099839755,
      "grad_norm": 0.2272789031267166,
      "learning_rate": 4.164310373197693e-06,
      "loss": 0.06789588928222656,
      "step": 5784
    },
    {
      "epoch": 0.8061032536751899,
      "grad_norm": 0.4090959429740906,
      "learning_rate": 4.158554033309172e-06,
      "loss": 0.09096145629882812,
      "step": 5785
    },
    {
      "epoch": 0.8062425973664042,
      "grad_norm": 0.298571914434433,
      "learning_rate": 4.152801213028273e-06,
      "loss": 0.07482147216796875,
      "step": 5786
    },
    {
      "epoch": 0.8063819410576186,
      "grad_norm": 0.21629935503005981,
      "learning_rate": 4.147051913633147e-06,
      "loss": 0.0650339126586914,
      "step": 5787
    },
    {
      "epoch": 0.806521284748833,
      "grad_norm": 0.4418518841266632,
      "learning_rate": 4.1413061364011665e-06,
      "loss": 0.09589195251464844,
      "step": 5788
    },
    {
      "epoch": 0.8066606284400474,
      "grad_norm": 0.29798203706741333,
      "learning_rate": 4.135563882608893e-06,
      "loss": 0.07706546783447266,
      "step": 5789
    },
    {
      "epoch": 0.8067999721312618,
      "grad_norm": 0.4144391715526581,
      "learning_rate": 4.129825153532132e-06,
      "loss": 0.10109138488769531,
      "step": 5790
    },
    {
      "epoch": 0.8069393158224761,
      "grad_norm": 0.35099121928215027,
      "learning_rate": 4.124089950445906e-06,
      "loss": 0.07657432556152344,
      "step": 5791
    },
    {
      "epoch": 0.8070786595136905,
      "grad_norm": 0.4530881643295288,
      "learning_rate": 4.118358274624435e-06,
      "loss": 0.08520889282226562,
      "step": 5792
    },
    {
      "epoch": 0.8072180032049049,
      "grad_norm": 0.3531850576400757,
      "learning_rate": 4.112630127341175e-06,
      "loss": 0.07258892059326172,
      "step": 5793
    },
    {
      "epoch": 0.8073573468961193,
      "grad_norm": 0.2995791733264923,
      "learning_rate": 4.106905509868781e-06,
      "loss": 0.08699226379394531,
      "step": 5794
    },
    {
      "epoch": 0.8074966905873336,
      "grad_norm": 0.3795398771762848,
      "learning_rate": 4.101184423479143e-06,
      "loss": 0.08843040466308594,
      "step": 5795
    },
    {
      "epoch": 0.807636034278548,
      "grad_norm": 0.4340970814228058,
      "learning_rate": 4.0954668694433455e-06,
      "loss": 0.09912586212158203,
      "step": 5796
    },
    {
      "epoch": 0.8077753779697624,
      "grad_norm": 0.33241933584213257,
      "learning_rate": 4.0897528490317025e-06,
      "loss": 0.08241462707519531,
      "step": 5797
    },
    {
      "epoch": 0.8079147216609768,
      "grad_norm": 0.3067690134048462,
      "learning_rate": 4.084042363513745e-06,
      "loss": 0.08090782165527344,
      "step": 5798
    },
    {
      "epoch": 0.8080540653521912,
      "grad_norm": 0.44497787952423096,
      "learning_rate": 4.078335414158206e-06,
      "loss": 0.10407066345214844,
      "step": 5799
    },
    {
      "epoch": 0.8081934090434055,
      "grad_norm": 0.4380345046520233,
      "learning_rate": 4.0726320022330345e-06,
      "loss": 0.09170675277709961,
      "step": 5800
    },
    {
      "epoch": 0.8083327527346199,
      "grad_norm": 0.41976046562194824,
      "learning_rate": 4.066932129005403e-06,
      "loss": 0.1017913818359375,
      "step": 5801
    },
    {
      "epoch": 0.8084720964258343,
      "grad_norm": 0.21358807384967804,
      "learning_rate": 4.061235795741702e-06,
      "loss": 0.06260299682617188,
      "step": 5802
    },
    {
      "epoch": 0.8086114401170487,
      "grad_norm": 0.5995532274246216,
      "learning_rate": 4.055543003707514e-06,
      "loss": 0.11542654037475586,
      "step": 5803
    },
    {
      "epoch": 0.808750783808263,
      "grad_norm": 0.2147154062986374,
      "learning_rate": 4.049853754167656e-06,
      "loss": 0.06442070007324219,
      "step": 5804
    },
    {
      "epoch": 0.8088901274994774,
      "grad_norm": 0.36500343680381775,
      "learning_rate": 4.0441680483861415e-06,
      "loss": 0.09430313110351562,
      "step": 5805
    },
    {
      "epoch": 0.8090294711906918,
      "grad_norm": 0.21613256633281708,
      "learning_rate": 4.038485887626214e-06,
      "loss": 0.06231880187988281,
      "step": 5806
    },
    {
      "epoch": 0.8091688148819062,
      "grad_norm": 0.799733579158783,
      "learning_rate": 4.032807273150308e-06,
      "loss": 0.1477947235107422,
      "step": 5807
    },
    {
      "epoch": 0.8093081585731206,
      "grad_norm": 0.4835410416126251,
      "learning_rate": 4.02713220622009e-06,
      "loss": 0.09343338012695312,
      "step": 5808
    },
    {
      "epoch": 0.8094475022643349,
      "grad_norm": 0.7171555161476135,
      "learning_rate": 4.021460688096435e-06,
      "loss": 0.1707286834716797,
      "step": 5809
    },
    {
      "epoch": 0.8095868459555494,
      "grad_norm": 0.4625450372695923,
      "learning_rate": 4.015792720039418e-06,
      "loss": 0.09865570068359375,
      "step": 5810
    },
    {
      "epoch": 0.8097261896467638,
      "grad_norm": 0.5930670499801636,
      "learning_rate": 4.010128303308327e-06,
      "loss": 0.09847068786621094,
      "step": 5811
    },
    {
      "epoch": 0.8098655333379782,
      "grad_norm": 0.32328855991363525,
      "learning_rate": 4.004467439161672e-06,
      "loss": 0.08299636840820312,
      "step": 5812
    },
    {
      "epoch": 0.8100048770291925,
      "grad_norm": 0.6118909120559692,
      "learning_rate": 3.998810128857174e-06,
      "loss": 0.09056758880615234,
      "step": 5813
    },
    {
      "epoch": 0.8101442207204069,
      "grad_norm": 0.4110959768295288,
      "learning_rate": 3.993156373651752e-06,
      "loss": 0.07280921936035156,
      "step": 5814
    },
    {
      "epoch": 0.8102835644116213,
      "grad_norm": 0.26330631971359253,
      "learning_rate": 3.987506174801536e-06,
      "loss": 0.07263946533203125,
      "step": 5815
    },
    {
      "epoch": 0.8104229081028357,
      "grad_norm": 0.43202245235443115,
      "learning_rate": 3.981859533561876e-06,
      "loss": 0.0966939926147461,
      "step": 5816
    },
    {
      "epoch": 0.8105622517940501,
      "grad_norm": 0.2843722403049469,
      "learning_rate": 3.976216451187334e-06,
      "loss": 0.07382583618164062,
      "step": 5817
    },
    {
      "epoch": 0.8107015954852644,
      "grad_norm": 0.3897741436958313,
      "learning_rate": 3.97057692893166e-06,
      "loss": 0.08265113830566406,
      "step": 5818
    },
    {
      "epoch": 0.8108409391764788,
      "grad_norm": 0.3721672296524048,
      "learning_rate": 3.964940968047835e-06,
      "loss": 0.07163524627685547,
      "step": 5819
    },
    {
      "epoch": 0.8109802828676932,
      "grad_norm": 0.40209370851516724,
      "learning_rate": 3.959308569788052e-06,
      "loss": 0.08085155487060547,
      "step": 5820
    },
    {
      "epoch": 0.8111196265589076,
      "grad_norm": 0.19883792102336884,
      "learning_rate": 3.953679735403677e-06,
      "loss": 0.051334381103515625,
      "step": 5821
    },
    {
      "epoch": 0.811258970250122,
      "grad_norm": 0.19753260910511017,
      "learning_rate": 3.948054466145324e-06,
      "loss": 0.06000041961669922,
      "step": 5822
    },
    {
      "epoch": 0.8113983139413363,
      "grad_norm": 0.2998635172843933,
      "learning_rate": 3.942432763262794e-06,
      "loss": 0.08606910705566406,
      "step": 5823
    },
    {
      "epoch": 0.8115376576325507,
      "grad_norm": 0.6876741051673889,
      "learning_rate": 3.9368146280051104e-06,
      "loss": 0.09685707092285156,
      "step": 5824
    },
    {
      "epoch": 0.8116770013237651,
      "grad_norm": 0.4737551510334015,
      "learning_rate": 3.931200061620486e-06,
      "loss": 0.09328079223632812,
      "step": 5825
    },
    {
      "epoch": 0.8118163450149795,
      "grad_norm": 0.45151373744010925,
      "learning_rate": 3.925589065356346e-06,
      "loss": 0.10296249389648438,
      "step": 5826
    },
    {
      "epoch": 0.8119556887061938,
      "grad_norm": 0.43353471159935,
      "learning_rate": 3.919981640459336e-06,
      "loss": 0.09314918518066406,
      "step": 5827
    },
    {
      "epoch": 0.8120950323974082,
      "grad_norm": 0.20053760707378387,
      "learning_rate": 3.914377788175287e-06,
      "loss": 0.06680107116699219,
      "step": 5828
    },
    {
      "epoch": 0.8122343760886226,
      "grad_norm": 0.623275876045227,
      "learning_rate": 3.908777509749255e-06,
      "loss": 0.11342620849609375,
      "step": 5829
    },
    {
      "epoch": 0.812373719779837,
      "grad_norm": 0.27016758918762207,
      "learning_rate": 3.903180806425495e-06,
      "loss": 0.07428932189941406,
      "step": 5830
    },
    {
      "epoch": 0.8125130634710513,
      "grad_norm": 0.4298363924026489,
      "learning_rate": 3.897587679447463e-06,
      "loss": 0.08160972595214844,
      "step": 5831
    },
    {
      "epoch": 0.8126524071622657,
      "grad_norm": 0.5625234246253967,
      "learning_rate": 3.891998130057819e-06,
      "loss": 0.07708358764648438,
      "step": 5832
    },
    {
      "epoch": 0.8127917508534801,
      "grad_norm": 0.5247209668159485,
      "learning_rate": 3.886412159498439e-06,
      "loss": 0.0969839096069336,
      "step": 5833
    },
    {
      "epoch": 0.8129310945446945,
      "grad_norm": 0.21361549198627472,
      "learning_rate": 3.880829769010402e-06,
      "loss": 0.06264972686767578,
      "step": 5834
    },
    {
      "epoch": 0.8130704382359089,
      "grad_norm": 0.6517264246940613,
      "learning_rate": 3.875250959833982e-06,
      "loss": 0.13391685485839844,
      "step": 5835
    },
    {
      "epoch": 0.8132097819271232,
      "grad_norm": 0.17446625232696533,
      "learning_rate": 3.869675733208662e-06,
      "loss": 0.05457496643066406,
      "step": 5836
    },
    {
      "epoch": 0.8133491256183376,
      "grad_norm": 0.3378855288028717,
      "learning_rate": 3.8641040903731335e-06,
      "loss": 0.07589054107666016,
      "step": 5837
    },
    {
      "epoch": 0.813488469309552,
      "grad_norm": 0.3982596695423126,
      "learning_rate": 3.85853603256529e-06,
      "loss": 0.090057373046875,
      "step": 5838
    },
    {
      "epoch": 0.8136278130007664,
      "grad_norm": 0.3694663941860199,
      "learning_rate": 3.852971561022218e-06,
      "loss": 0.0801544189453125,
      "step": 5839
    },
    {
      "epoch": 0.8137671566919807,
      "grad_norm": 0.39266082644462585,
      "learning_rate": 3.8474106769802255e-06,
      "loss": 0.09895706176757812,
      "step": 5840
    },
    {
      "epoch": 0.8139065003831951,
      "grad_norm": 0.3252917528152466,
      "learning_rate": 3.841853381674814e-06,
      "loss": 0.09617424011230469,
      "step": 5841
    },
    {
      "epoch": 0.8140458440744095,
      "grad_norm": 0.6276628971099854,
      "learning_rate": 3.836299676340684e-06,
      "loss": 0.0860738754272461,
      "step": 5842
    },
    {
      "epoch": 0.8141851877656239,
      "grad_norm": 0.4068092405796051,
      "learning_rate": 3.83074956221174e-06,
      "loss": 0.08965110778808594,
      "step": 5843
    },
    {
      "epoch": 0.8143245314568383,
      "grad_norm": 0.5269186496734619,
      "learning_rate": 3.825203040521091e-06,
      "loss": 0.09769535064697266,
      "step": 5844
    },
    {
      "epoch": 0.8144638751480526,
      "grad_norm": 0.8164353370666504,
      "learning_rate": 3.819660112501053e-06,
      "loss": 0.12249183654785156,
      "step": 5845
    },
    {
      "epoch": 0.814603218839267,
      "grad_norm": 0.3817645013332367,
      "learning_rate": 3.814120779383137e-06,
      "loss": 0.09700870513916016,
      "step": 5846
    },
    {
      "epoch": 0.8147425625304814,
      "grad_norm": 0.5397366881370544,
      "learning_rate": 3.8085850423980475e-06,
      "loss": 0.09310340881347656,
      "step": 5847
    },
    {
      "epoch": 0.8148819062216958,
      "grad_norm": 0.47126224637031555,
      "learning_rate": 3.8030529027757057e-06,
      "loss": 0.10383796691894531,
      "step": 5848
    },
    {
      "epoch": 0.8150212499129101,
      "grad_norm": 0.5539352297782898,
      "learning_rate": 3.797524361745231e-06,
      "loss": 0.10870504379272461,
      "step": 5849
    },
    {
      "epoch": 0.8151605936041246,
      "grad_norm": 0.7726271152496338,
      "learning_rate": 3.7919994205349287e-06,
      "loss": 0.125244140625,
      "step": 5850
    },
    {
      "epoch": 0.815299937295339,
      "grad_norm": 0.4979258179664612,
      "learning_rate": 3.7864780803723267e-06,
      "loss": 0.1098785400390625,
      "step": 5851
    },
    {
      "epoch": 0.8154392809865534,
      "grad_norm": 0.479111909866333,
      "learning_rate": 3.7809603424841346e-06,
      "loss": 0.10226631164550781,
      "step": 5852
    },
    {
      "epoch": 0.8155786246777678,
      "grad_norm": 0.2699643671512604,
      "learning_rate": 3.7754462080962604e-06,
      "loss": 0.06986236572265625,
      "step": 5853
    },
    {
      "epoch": 0.8157179683689821,
      "grad_norm": 0.3852890133857727,
      "learning_rate": 3.769935678433827e-06,
      "loss": 0.07102775573730469,
      "step": 5854
    },
    {
      "epoch": 0.8158573120601965,
      "grad_norm": 0.6873718500137329,
      "learning_rate": 3.7644287547211476e-06,
      "loss": 0.10587120056152344,
      "step": 5855
    },
    {
      "epoch": 0.8159966557514109,
      "grad_norm": 0.2716093063354492,
      "learning_rate": 3.7589254381817397e-06,
      "loss": 0.07733535766601562,
      "step": 5856
    },
    {
      "epoch": 0.8161359994426253,
      "grad_norm": 0.5599513649940491,
      "learning_rate": 3.753425730038307e-06,
      "loss": 0.0971670150756836,
      "step": 5857
    },
    {
      "epoch": 0.8162753431338396,
      "grad_norm": 0.3076510727405548,
      "learning_rate": 3.7479296315127588e-06,
      "loss": 0.07508087158203125,
      "step": 5858
    },
    {
      "epoch": 0.816414686825054,
      "grad_norm": 0.3657916486263275,
      "learning_rate": 3.7424371438262096e-06,
      "loss": 0.08539915084838867,
      "step": 5859
    },
    {
      "epoch": 0.8165540305162684,
      "grad_norm": 0.38651734590530396,
      "learning_rate": 3.7369482681989565e-06,
      "loss": 0.07268714904785156,
      "step": 5860
    },
    {
      "epoch": 0.8166933742074828,
      "grad_norm": 0.3738669455051422,
      "learning_rate": 3.7314630058505063e-06,
      "loss": 0.08064079284667969,
      "step": 5861
    },
    {
      "epoch": 0.8168327178986972,
      "grad_norm": 0.3860386610031128,
      "learning_rate": 3.725981357999562e-06,
      "loss": 0.08741378784179688,
      "step": 5862
    },
    {
      "epoch": 0.8169720615899115,
      "grad_norm": 0.3698016107082367,
      "learning_rate": 3.72050332586402e-06,
      "loss": 0.08467674255371094,
      "step": 5863
    },
    {
      "epoch": 0.8171114052811259,
      "grad_norm": 0.519075334072113,
      "learning_rate": 3.715028910660967e-06,
      "loss": 0.10147666931152344,
      "step": 5864
    },
    {
      "epoch": 0.8172507489723403,
      "grad_norm": 0.3606766164302826,
      "learning_rate": 3.709558113606697e-06,
      "loss": 0.06927299499511719,
      "step": 5865
    },
    {
      "epoch": 0.8173900926635547,
      "grad_norm": 0.8066274523735046,
      "learning_rate": 3.704090935916702e-06,
      "loss": 0.09131526947021484,
      "step": 5866
    },
    {
      "epoch": 0.817529436354769,
      "grad_norm": 0.4588612914085388,
      "learning_rate": 3.6986273788056592e-06,
      "loss": 0.09450054168701172,
      "step": 5867
    },
    {
      "epoch": 0.8176687800459834,
      "grad_norm": 0.29021358489990234,
      "learning_rate": 3.6931674434874397e-06,
      "loss": 0.06563186645507812,
      "step": 5868
    },
    {
      "epoch": 0.8178081237371978,
      "grad_norm": 0.36841708421707153,
      "learning_rate": 3.6877111311751246e-06,
      "loss": 0.0891885757446289,
      "step": 5869
    },
    {
      "epoch": 0.8179474674284122,
      "grad_norm": 0.3166259229183197,
      "learning_rate": 3.682258443080986e-06,
      "loss": 0.08437347412109375,
      "step": 5870
    },
    {
      "epoch": 0.8180868111196266,
      "grad_norm": 0.44120749831199646,
      "learning_rate": 3.676809380416475e-06,
      "loss": 0.0991201400756836,
      "step": 5871
    },
    {
      "epoch": 0.8182261548108409,
      "grad_norm": 0.4555787742137909,
      "learning_rate": 3.671363944392259e-06,
      "loss": 0.0888357162475586,
      "step": 5872
    },
    {
      "epoch": 0.8183654985020553,
      "grad_norm": 0.3031899631023407,
      "learning_rate": 3.6659221362181827e-06,
      "loss": 0.07115364074707031,
      "step": 5873
    },
    {
      "epoch": 0.8185048421932697,
      "grad_norm": 0.3476454019546509,
      "learning_rate": 3.660483957103298e-06,
      "loss": 0.09343528747558594,
      "step": 5874
    },
    {
      "epoch": 0.8186441858844841,
      "grad_norm": 0.5661113858222961,
      "learning_rate": 3.655049408255835e-06,
      "loss": 0.1084747314453125,
      "step": 5875
    },
    {
      "epoch": 0.8187835295756984,
      "grad_norm": 0.34590500593185425,
      "learning_rate": 3.649618490883233e-06,
      "loss": 0.08631324768066406,
      "step": 5876
    },
    {
      "epoch": 0.8189228732669128,
      "grad_norm": 0.65434330701828,
      "learning_rate": 3.6441912061921205e-06,
      "loss": 0.12784957885742188,
      "step": 5877
    },
    {
      "epoch": 0.8190622169581272,
      "grad_norm": 0.3674112856388092,
      "learning_rate": 3.638767555388314e-06,
      "loss": 0.07309913635253906,
      "step": 5878
    },
    {
      "epoch": 0.8192015606493416,
      "grad_norm": 0.6925652623176575,
      "learning_rate": 3.6333475396768168e-06,
      "loss": 0.1035623550415039,
      "step": 5879
    },
    {
      "epoch": 0.819340904340556,
      "grad_norm": 0.45551711320877075,
      "learning_rate": 3.6279311602618416e-06,
      "loss": 0.09308242797851562,
      "step": 5880
    },
    {
      "epoch": 0.8194802480317703,
      "grad_norm": 0.306031733751297,
      "learning_rate": 3.6225184183467856e-06,
      "loss": 0.08002948760986328,
      "step": 5881
    },
    {
      "epoch": 0.8196195917229847,
      "grad_norm": 0.3192896842956543,
      "learning_rate": 3.6171093151342264e-06,
      "loss": 0.0796661376953125,
      "step": 5882
    },
    {
      "epoch": 0.8197589354141991,
      "grad_norm": 0.38856467604637146,
      "learning_rate": 3.611703851825956e-06,
      "loss": 0.08147239685058594,
      "step": 5883
    },
    {
      "epoch": 0.8198982791054135,
      "grad_norm": 0.761904239654541,
      "learning_rate": 3.6063020296229344e-06,
      "loss": 0.11430168151855469,
      "step": 5884
    },
    {
      "epoch": 0.8200376227966278,
      "grad_norm": 0.32405394315719604,
      "learning_rate": 3.60090384972533e-06,
      "loss": 0.07469940185546875,
      "step": 5885
    },
    {
      "epoch": 0.8201769664878422,
      "grad_norm": 0.2449733316898346,
      "learning_rate": 3.595509313332488e-06,
      "loss": 0.06522083282470703,
      "step": 5886
    },
    {
      "epoch": 0.8203163101790566,
      "grad_norm": 0.6892227530479431,
      "learning_rate": 3.5901184216429585e-06,
      "loss": 0.11584281921386719,
      "step": 5887
    },
    {
      "epoch": 0.820455653870271,
      "grad_norm": 0.3779495656490326,
      "learning_rate": 3.584731175854479e-06,
      "loss": 0.07533073425292969,
      "step": 5888
    },
    {
      "epoch": 0.8205949975614854,
      "grad_norm": 0.36501604318618774,
      "learning_rate": 3.5793475771639562e-06,
      "loss": 0.07806777954101562,
      "step": 5889
    },
    {
      "epoch": 0.8207343412526998,
      "grad_norm": 0.4072225093841553,
      "learning_rate": 3.5739676267675115e-06,
      "loss": 0.08334732055664062,
      "step": 5890
    },
    {
      "epoch": 0.8208736849439142,
      "grad_norm": 0.7222934365272522,
      "learning_rate": 3.568591325860453e-06,
      "loss": 0.0991058349609375,
      "step": 5891
    },
    {
      "epoch": 0.8210130286351286,
      "grad_norm": 0.5367545485496521,
      "learning_rate": 3.563218675637261e-06,
      "loss": 0.10298347473144531,
      "step": 5892
    },
    {
      "epoch": 0.821152372326343,
      "grad_norm": 0.6872527003288269,
      "learning_rate": 3.5578496772916205e-06,
      "loss": 0.12026405334472656,
      "step": 5893
    },
    {
      "epoch": 0.8212917160175573,
      "grad_norm": 0.34621474146842957,
      "learning_rate": 3.552484332016408e-06,
      "loss": 0.07516288757324219,
      "step": 5894
    },
    {
      "epoch": 0.8214310597087717,
      "grad_norm": 0.5347920060157776,
      "learning_rate": 3.547122641003671e-06,
      "loss": 0.11132526397705078,
      "step": 5895
    },
    {
      "epoch": 0.8215704033999861,
      "grad_norm": 0.41580092906951904,
      "learning_rate": 3.5417646054446554e-06,
      "loss": 0.09637069702148438,
      "step": 5896
    },
    {
      "epoch": 0.8217097470912005,
      "grad_norm": 0.5229889750480652,
      "learning_rate": 3.536410226529794e-06,
      "loss": 0.09395217895507812,
      "step": 5897
    },
    {
      "epoch": 0.8218490907824149,
      "grad_norm": 0.23738409578800201,
      "learning_rate": 3.5310595054487173e-06,
      "loss": 0.06240701675415039,
      "step": 5898
    },
    {
      "epoch": 0.8219884344736292,
      "grad_norm": 0.41276171803474426,
      "learning_rate": 3.525712443390226e-06,
      "loss": 0.09011459350585938,
      "step": 5899
    },
    {
      "epoch": 0.8221277781648436,
      "grad_norm": 0.43580058217048645,
      "learning_rate": 3.5203690415423086e-06,
      "loss": 0.08883857727050781,
      "step": 5900
    },
    {
      "epoch": 0.822267121856058,
      "grad_norm": 0.43713656067848206,
      "learning_rate": 3.5150293010921543e-06,
      "loss": 0.083587646484375,
      "step": 5901
    },
    {
      "epoch": 0.8224064655472724,
      "grad_norm": 0.5140691995620728,
      "learning_rate": 3.5096932232261384e-06,
      "loss": 0.09360122680664062,
      "step": 5902
    },
    {
      "epoch": 0.8225458092384867,
      "grad_norm": 0.32079729437828064,
      "learning_rate": 3.504360809129801e-06,
      "loss": 0.07270622253417969,
      "step": 5903
    },
    {
      "epoch": 0.8226851529297011,
      "grad_norm": 0.6171124577522278,
      "learning_rate": 3.4990320599878948e-06,
      "loss": 0.09617042541503906,
      "step": 5904
    },
    {
      "epoch": 0.8228244966209155,
      "grad_norm": 0.26502928137779236,
      "learning_rate": 3.493706976984337e-06,
      "loss": 0.06240272521972656,
      "step": 5905
    },
    {
      "epoch": 0.8229638403121299,
      "grad_norm": 0.6324767470359802,
      "learning_rate": 3.4883855613022476e-06,
      "loss": 0.11882781982421875,
      "step": 5906
    },
    {
      "epoch": 0.8231031840033443,
      "grad_norm": 0.4432928264141083,
      "learning_rate": 3.483067814123917e-06,
      "loss": 0.09133720397949219,
      "step": 5907
    },
    {
      "epoch": 0.8232425276945586,
      "grad_norm": 0.3890266418457031,
      "learning_rate": 3.477753736630829e-06,
      "loss": 0.09441375732421875,
      "step": 5908
    },
    {
      "epoch": 0.823381871385773,
      "grad_norm": 0.32935482263565063,
      "learning_rate": 3.4724433300036565e-06,
      "loss": 0.0778961181640625,
      "step": 5909
    },
    {
      "epoch": 0.8235212150769874,
      "grad_norm": 0.33957135677337646,
      "learning_rate": 3.467136595422247e-06,
      "loss": 0.07725143432617188,
      "step": 5910
    },
    {
      "epoch": 0.8236605587682018,
      "grad_norm": 0.6433339715003967,
      "learning_rate": 3.4618335340656263e-06,
      "loss": 0.11888313293457031,
      "step": 5911
    },
    {
      "epoch": 0.8237999024594161,
      "grad_norm": 0.25005996227264404,
      "learning_rate": 3.456534147112023e-06,
      "loss": 0.05613517761230469,
      "step": 5912
    },
    {
      "epoch": 0.8239392461506305,
      "grad_norm": 0.34963458776474,
      "learning_rate": 3.451238435738844e-06,
      "loss": 0.084197998046875,
      "step": 5913
    },
    {
      "epoch": 0.8240785898418449,
      "grad_norm": 0.2656228244304657,
      "learning_rate": 3.445946401122666e-06,
      "loss": 0.07004070281982422,
      "step": 5914
    },
    {
      "epoch": 0.8242179335330593,
      "grad_norm": 0.938654899597168,
      "learning_rate": 3.4406580444392647e-06,
      "loss": 0.1311187744140625,
      "step": 5915
    },
    {
      "epoch": 0.8243572772242737,
      "grad_norm": 0.3312970995903015,
      "learning_rate": 3.435373366863586e-06,
      "loss": 0.09749221801757812,
      "step": 5916
    },
    {
      "epoch": 0.824496620915488,
      "grad_norm": 0.39837902784347534,
      "learning_rate": 3.430092369569773e-06,
      "loss": 0.08923912048339844,
      "step": 5917
    },
    {
      "epoch": 0.8246359646067024,
      "grad_norm": 0.9318960905075073,
      "learning_rate": 3.4248150537311344e-06,
      "loss": 0.15502071380615234,
      "step": 5918
    },
    {
      "epoch": 0.8247753082979168,
      "grad_norm": 0.7180649638175964,
      "learning_rate": 3.4195414205201718e-06,
      "loss": 0.1254749298095703,
      "step": 5919
    },
    {
      "epoch": 0.8249146519891312,
      "grad_norm": 0.41500693559646606,
      "learning_rate": 3.4142714711085765e-06,
      "loss": 0.09228134155273438,
      "step": 5920
    },
    {
      "epoch": 0.8250539956803455,
      "grad_norm": 0.3274748623371124,
      "learning_rate": 3.409005206667193e-06,
      "loss": 0.06975364685058594,
      "step": 5921
    },
    {
      "epoch": 0.8251933393715599,
      "grad_norm": 0.30552950501441956,
      "learning_rate": 3.4037426283660734e-06,
      "loss": 0.08064556121826172,
      "step": 5922
    },
    {
      "epoch": 0.8253326830627743,
      "grad_norm": 0.5385141372680664,
      "learning_rate": 3.3984837373744406e-06,
      "loss": 0.08904647827148438,
      "step": 5923
    },
    {
      "epoch": 0.8254720267539887,
      "grad_norm": 0.4074801802635193,
      "learning_rate": 3.3932285348607108e-06,
      "loss": 0.07433128356933594,
      "step": 5924
    },
    {
      "epoch": 0.825611370445203,
      "grad_norm": 0.39350637793540955,
      "learning_rate": 3.387977021992459e-06,
      "loss": 0.0794367790222168,
      "step": 5925
    },
    {
      "epoch": 0.8257507141364174,
      "grad_norm": 0.2991810142993927,
      "learning_rate": 3.38272919993645e-06,
      "loss": 0.06830692291259766,
      "step": 5926
    },
    {
      "epoch": 0.8258900578276318,
      "grad_norm": 0.7272956371307373,
      "learning_rate": 3.377485069858639e-06,
      "loss": 0.10809326171875,
      "step": 5927
    },
    {
      "epoch": 0.8260294015188462,
      "grad_norm": 0.4145250916481018,
      "learning_rate": 3.372244632924142e-06,
      "loss": 0.07814407348632812,
      "step": 5928
    },
    {
      "epoch": 0.8261687452100606,
      "grad_norm": 0.5727525949478149,
      "learning_rate": 3.3670078902972693e-06,
      "loss": 0.11113357543945312,
      "step": 5929
    },
    {
      "epoch": 0.8263080889012749,
      "grad_norm": 0.4335343837738037,
      "learning_rate": 3.361774843141512e-06,
      "loss": 0.09780502319335938,
      "step": 5930
    },
    {
      "epoch": 0.8264474325924894,
      "grad_norm": 0.3678942024707794,
      "learning_rate": 3.3565454926195252e-06,
      "loss": 0.08487510681152344,
      "step": 5931
    },
    {
      "epoch": 0.8265867762837038,
      "grad_norm": 0.3676747977733612,
      "learning_rate": 3.35131983989315e-06,
      "loss": 0.07985591888427734,
      "step": 5932
    },
    {
      "epoch": 0.8267261199749182,
      "grad_norm": 0.44906941056251526,
      "learning_rate": 3.3460978861234095e-06,
      "loss": 0.09431838989257812,
      "step": 5933
    },
    {
      "epoch": 0.8268654636661326,
      "grad_norm": 0.2738412320613861,
      "learning_rate": 3.3408796324705085e-06,
      "loss": 0.06808280944824219,
      "step": 5934
    },
    {
      "epoch": 0.8270048073573469,
      "grad_norm": 0.46651172637939453,
      "learning_rate": 3.335665080093815e-06,
      "loss": 0.11558246612548828,
      "step": 5935
    },
    {
      "epoch": 0.8271441510485613,
      "grad_norm": 0.3002771735191345,
      "learning_rate": 3.3304542301518915e-06,
      "loss": 0.0759124755859375,
      "step": 5936
    },
    {
      "epoch": 0.8272834947397757,
      "grad_norm": 0.4293699264526367,
      "learning_rate": 3.325247083802463e-06,
      "loss": 0.0855255126953125,
      "step": 5937
    },
    {
      "epoch": 0.8274228384309901,
      "grad_norm": 0.22232414782047272,
      "learning_rate": 3.320043642202444e-06,
      "loss": 0.06422615051269531,
      "step": 5938
    },
    {
      "epoch": 0.8275621821222044,
      "grad_norm": 0.3650669753551483,
      "learning_rate": 3.3148439065079142e-06,
      "loss": 0.08199882507324219,
      "step": 5939
    },
    {
      "epoch": 0.8277015258134188,
      "grad_norm": 0.27184221148490906,
      "learning_rate": 3.309647877874138e-06,
      "loss": 0.07247352600097656,
      "step": 5940
    },
    {
      "epoch": 0.8278408695046332,
      "grad_norm": 0.5894299149513245,
      "learning_rate": 3.304455557455564e-06,
      "loss": 0.11213302612304688,
      "step": 5941
    },
    {
      "epoch": 0.8279802131958476,
      "grad_norm": 0.302002876996994,
      "learning_rate": 3.299266946405797e-06,
      "loss": 0.07868576049804688,
      "step": 5942
    },
    {
      "epoch": 0.828119556887062,
      "grad_norm": 0.30811387300491333,
      "learning_rate": 3.294082045877627e-06,
      "loss": 0.07997322082519531,
      "step": 5943
    },
    {
      "epoch": 0.8282589005782763,
      "grad_norm": 0.44835326075553894,
      "learning_rate": 3.2889008570230228e-06,
      "loss": 0.09256553649902344,
      "step": 5944
    },
    {
      "epoch": 0.8283982442694907,
      "grad_norm": 0.31270575523376465,
      "learning_rate": 3.2837233809931314e-06,
      "loss": 0.07196235656738281,
      "step": 5945
    },
    {
      "epoch": 0.8285375879607051,
      "grad_norm": 0.3924355208873749,
      "learning_rate": 3.278549618938267e-06,
      "loss": 0.09350776672363281,
      "step": 5946
    },
    {
      "epoch": 0.8286769316519195,
      "grad_norm": 0.6258882880210876,
      "learning_rate": 3.2733795720079133e-06,
      "loss": 0.10442447662353516,
      "step": 5947
    },
    {
      "epoch": 0.8288162753431338,
      "grad_norm": 0.386699378490448,
      "learning_rate": 3.268213241350746e-06,
      "loss": 0.08852005004882812,
      "step": 5948
    },
    {
      "epoch": 0.8289556190343482,
      "grad_norm": 0.28927746415138245,
      "learning_rate": 3.263050628114606e-06,
      "loss": 0.06868267059326172,
      "step": 5949
    },
    {
      "epoch": 0.8290949627255626,
      "grad_norm": 0.28275036811828613,
      "learning_rate": 3.2578917334465034e-06,
      "loss": 0.079315185546875,
      "step": 5950
    },
    {
      "epoch": 0.829234306416777,
      "grad_norm": 0.46204715967178345,
      "learning_rate": 3.2527365584926264e-06,
      "loss": 0.09205293655395508,
      "step": 5951
    },
    {
      "epoch": 0.8293736501079914,
      "grad_norm": 0.5025037527084351,
      "learning_rate": 3.2475851043983496e-06,
      "loss": 0.0969247817993164,
      "step": 5952
    },
    {
      "epoch": 0.8295129937992057,
      "grad_norm": 0.7166548371315002,
      "learning_rate": 3.2424373723081892e-06,
      "loss": 0.13725662231445312,
      "step": 5953
    },
    {
      "epoch": 0.8296523374904201,
      "grad_norm": 0.3021635413169861,
      "learning_rate": 3.2372933633658633e-06,
      "loss": 0.06968307495117188,
      "step": 5954
    },
    {
      "epoch": 0.8297916811816345,
      "grad_norm": 0.32865071296691895,
      "learning_rate": 3.2321530787142508e-06,
      "loss": 0.0791463851928711,
      "step": 5955
    },
    {
      "epoch": 0.8299310248728489,
      "grad_norm": 0.28863033652305603,
      "learning_rate": 3.227016519495414e-06,
      "loss": 0.062075138092041016,
      "step": 5956
    },
    {
      "epoch": 0.8300703685640632,
      "grad_norm": 0.6655749678611755,
      "learning_rate": 3.221883686850573e-06,
      "loss": 0.10507583618164062,
      "step": 5957
    },
    {
      "epoch": 0.8302097122552776,
      "grad_norm": 0.233371764421463,
      "learning_rate": 3.2167545819201227e-06,
      "loss": 0.06587982177734375,
      "step": 5958
    },
    {
      "epoch": 0.830349055946492,
      "grad_norm": 0.7336015105247498,
      "learning_rate": 3.2116292058436383e-06,
      "loss": 0.10976982116699219,
      "step": 5959
    },
    {
      "epoch": 0.8304883996377064,
      "grad_norm": 0.847266674041748,
      "learning_rate": 3.2065075597598573e-06,
      "loss": 0.11776447296142578,
      "step": 5960
    },
    {
      "epoch": 0.8306277433289208,
      "grad_norm": 0.23172317445278168,
      "learning_rate": 3.201389644806692e-06,
      "loss": 0.06261444091796875,
      "step": 5961
    },
    {
      "epoch": 0.8307670870201351,
      "grad_norm": 0.2808094024658203,
      "learning_rate": 3.1962754621212345e-06,
      "loss": 0.06742572784423828,
      "step": 5962
    },
    {
      "epoch": 0.8309064307113495,
      "grad_norm": 0.38375911116600037,
      "learning_rate": 3.1911650128397342e-06,
      "loss": 0.09573650360107422,
      "step": 5963
    },
    {
      "epoch": 0.8310457744025639,
      "grad_norm": 0.5335613489151001,
      "learning_rate": 3.1860582980976117e-06,
      "loss": 0.08881568908691406,
      "step": 5964
    },
    {
      "epoch": 0.8311851180937783,
      "grad_norm": 0.3622586727142334,
      "learning_rate": 3.180955319029464e-06,
      "loss": 0.0902862548828125,
      "step": 5965
    },
    {
      "epoch": 0.8313244617849926,
      "grad_norm": 0.4199208915233612,
      "learning_rate": 3.175856076769066e-06,
      "loss": 0.07773017883300781,
      "step": 5966
    },
    {
      "epoch": 0.831463805476207,
      "grad_norm": 0.5161654949188232,
      "learning_rate": 3.170760572449345e-06,
      "loss": 0.11299705505371094,
      "step": 5967
    },
    {
      "epoch": 0.8316031491674214,
      "grad_norm": 0.4099973440170288,
      "learning_rate": 3.1656688072024024e-06,
      "loss": 0.09703731536865234,
      "step": 5968
    },
    {
      "epoch": 0.8317424928586358,
      "grad_norm": 0.3924803137779236,
      "learning_rate": 3.160580782159517e-06,
      "loss": 0.08985328674316406,
      "step": 5969
    },
    {
      "epoch": 0.8318818365498502,
      "grad_norm": 0.305880069732666,
      "learning_rate": 3.155496498451136e-06,
      "loss": 0.07192802429199219,
      "step": 5970
    },
    {
      "epoch": 0.8320211802410646,
      "grad_norm": 1.0590119361877441,
      "learning_rate": 3.1504159572068604e-06,
      "loss": 0.1198577880859375,
      "step": 5971
    },
    {
      "epoch": 0.832160523932279,
      "grad_norm": 0.6996975541114807,
      "learning_rate": 3.1453391595554783e-06,
      "loss": 0.14168930053710938,
      "step": 5972
    },
    {
      "epoch": 0.8322998676234934,
      "grad_norm": 0.4013158082962036,
      "learning_rate": 3.140266106624941e-06,
      "loss": 0.08887386322021484,
      "step": 5973
    },
    {
      "epoch": 0.8324392113147078,
      "grad_norm": 0.296390175819397,
      "learning_rate": 3.1351967995423594e-06,
      "loss": 0.0687713623046875,
      "step": 5974
    },
    {
      "epoch": 0.8325785550059221,
      "grad_norm": 0.5937119722366333,
      "learning_rate": 3.1301312394340157e-06,
      "loss": 0.14336204528808594,
      "step": 5975
    },
    {
      "epoch": 0.8327178986971365,
      "grad_norm": 0.6880470514297485,
      "learning_rate": 3.1250694274253623e-06,
      "loss": 0.11455917358398438,
      "step": 5976
    },
    {
      "epoch": 0.8328572423883509,
      "grad_norm": 0.6203737258911133,
      "learning_rate": 3.120011364641029e-06,
      "loss": 0.13069534301757812,
      "step": 5977
    },
    {
      "epoch": 0.8329965860795653,
      "grad_norm": 0.5486828088760376,
      "learning_rate": 3.114957052204792e-06,
      "loss": 0.09066009521484375,
      "step": 5978
    },
    {
      "epoch": 0.8331359297707797,
      "grad_norm": 0.5083727240562439,
      "learning_rate": 3.1099064912396002e-06,
      "loss": 0.10340404510498047,
      "step": 5979
    },
    {
      "epoch": 0.833275273461994,
      "grad_norm": 0.4876103103160858,
      "learning_rate": 3.1048596828675805e-06,
      "loss": 0.08077812194824219,
      "step": 5980
    },
    {
      "epoch": 0.8334146171532084,
      "grad_norm": 0.336137592792511,
      "learning_rate": 3.0998166282100215e-06,
      "loss": 0.07556915283203125,
      "step": 5981
    },
    {
      "epoch": 0.8335539608444228,
      "grad_norm": 0.30565690994262695,
      "learning_rate": 3.0947773283873638e-06,
      "loss": 0.06825637817382812,
      "step": 5982
    },
    {
      "epoch": 0.8336933045356372,
      "grad_norm": 0.27219176292419434,
      "learning_rate": 3.0897417845192356e-06,
      "loss": 0.07198524475097656,
      "step": 5983
    },
    {
      "epoch": 0.8338326482268515,
      "grad_norm": 0.20084337890148163,
      "learning_rate": 3.084709997724411e-06,
      "loss": 0.05603218078613281,
      "step": 5984
    },
    {
      "epoch": 0.8339719919180659,
      "grad_norm": 0.529218852519989,
      "learning_rate": 3.079681969120849e-06,
      "loss": 0.09210205078125,
      "step": 5985
    },
    {
      "epoch": 0.8341113356092803,
      "grad_norm": 1.1863633394241333,
      "learning_rate": 3.0746576998256495e-06,
      "loss": 0.10213756561279297,
      "step": 5986
    },
    {
      "epoch": 0.8342506793004947,
      "grad_norm": 0.3188323676586151,
      "learning_rate": 3.0696371909550947e-06,
      "loss": 0.08150482177734375,
      "step": 5987
    },
    {
      "epoch": 0.8343900229917091,
      "grad_norm": 0.4841364324092865,
      "learning_rate": 3.0646204436246372e-06,
      "loss": 0.10869598388671875,
      "step": 5988
    },
    {
      "epoch": 0.8345293666829234,
      "grad_norm": 0.3367457091808319,
      "learning_rate": 3.0596074589488743e-06,
      "loss": 0.06162071228027344,
      "step": 5989
    },
    {
      "epoch": 0.8346687103741378,
      "grad_norm": 0.6023036241531372,
      "learning_rate": 3.054598238041575e-06,
      "loss": 0.08916091918945312,
      "step": 5990
    },
    {
      "epoch": 0.8348080540653522,
      "grad_norm": 0.44613492488861084,
      "learning_rate": 3.04959278201568e-06,
      "loss": 0.091094970703125,
      "step": 5991
    },
    {
      "epoch": 0.8349473977565666,
      "grad_norm": 0.28894999623298645,
      "learning_rate": 3.044591091983282e-06,
      "loss": 0.06940841674804688,
      "step": 5992
    },
    {
      "epoch": 0.835086741447781,
      "grad_norm": 0.29688334465026855,
      "learning_rate": 3.0395931690556435e-06,
      "loss": 0.07472991943359375,
      "step": 5993
    },
    {
      "epoch": 0.8352260851389953,
      "grad_norm": 0.34513744711875916,
      "learning_rate": 3.0345990143431938e-06,
      "loss": 0.07462596893310547,
      "step": 5994
    },
    {
      "epoch": 0.8353654288302097,
      "grad_norm": 0.49477240443229675,
      "learning_rate": 3.029608628955518e-06,
      "loss": 0.10788726806640625,
      "step": 5995
    },
    {
      "epoch": 0.8355047725214241,
      "grad_norm": 0.3080471158027649,
      "learning_rate": 3.0246220140013593e-06,
      "loss": 0.06917953491210938,
      "step": 5996
    },
    {
      "epoch": 0.8356441162126385,
      "grad_norm": 0.47105005383491516,
      "learning_rate": 3.019639170588633e-06,
      "loss": 0.08435821533203125,
      "step": 5997
    },
    {
      "epoch": 0.8357834599038528,
      "grad_norm": 0.25112777948379517,
      "learning_rate": 3.0146600998244226e-06,
      "loss": 0.07252120971679688,
      "step": 5998
    },
    {
      "epoch": 0.8359228035950672,
      "grad_norm": 0.591999351978302,
      "learning_rate": 3.009684802814954e-06,
      "loss": 0.12003326416015625,
      "step": 5999
    },
    {
      "epoch": 0.8360621472862816,
      "grad_norm": 0.758327305316925,
      "learning_rate": 3.004713280665621e-06,
      "loss": 0.11583185195922852,
      "step": 6000
    },
    {
      "epoch": 0.836201490977496,
      "grad_norm": 0.31356847286224365,
      "learning_rate": 2.999745534480989e-06,
      "loss": 0.07633399963378906,
      "step": 6001
    },
    {
      "epoch": 0.8363408346687103,
      "grad_norm": 0.3390062153339386,
      "learning_rate": 2.9947815653647816e-06,
      "loss": 0.07587337493896484,
      "step": 6002
    },
    {
      "epoch": 0.8364801783599247,
      "grad_norm": 0.29491111636161804,
      "learning_rate": 2.98982137441987e-06,
      "loss": 0.07812690734863281,
      "step": 6003
    },
    {
      "epoch": 0.8366195220511391,
      "grad_norm": 0.41523683071136475,
      "learning_rate": 2.9848649627483063e-06,
      "loss": 0.09159660339355469,
      "step": 6004
    },
    {
      "epoch": 0.8367588657423535,
      "grad_norm": 0.5694877505302429,
      "learning_rate": 2.97991233145128e-06,
      "loss": 0.11769866943359375,
      "step": 6005
    },
    {
      "epoch": 0.8368982094335679,
      "grad_norm": 0.3769380450248718,
      "learning_rate": 2.974963481629163e-06,
      "loss": 0.08779239654541016,
      "step": 6006
    },
    {
      "epoch": 0.8370375531247822,
      "grad_norm": 0.3078921437263489,
      "learning_rate": 2.970018414381466e-06,
      "loss": 0.08168983459472656,
      "step": 6007
    },
    {
      "epoch": 0.8371768968159966,
      "grad_norm": 0.5745800137519836,
      "learning_rate": 2.9650771308068773e-06,
      "loss": 0.09920406341552734,
      "step": 6008
    },
    {
      "epoch": 0.837316240507211,
      "grad_norm": 0.38986900448799133,
      "learning_rate": 2.960139632003243e-06,
      "loss": 0.09000205993652344,
      "step": 6009
    },
    {
      "epoch": 0.8374555841984254,
      "grad_norm": 0.3203256130218506,
      "learning_rate": 2.955205919067552e-06,
      "loss": 0.07768917083740234,
      "step": 6010
    },
    {
      "epoch": 0.8375949278896399,
      "grad_norm": 0.329010933637619,
      "learning_rate": 2.950275993095966e-06,
      "loss": 0.07488441467285156,
      "step": 6011
    },
    {
      "epoch": 0.8377342715808542,
      "grad_norm": 0.315508097410202,
      "learning_rate": 2.945349855183799e-06,
      "loss": 0.07185077667236328,
      "step": 6012
    },
    {
      "epoch": 0.8378736152720686,
      "grad_norm": 0.6710783839225769,
      "learning_rate": 2.940427506425536e-06,
      "loss": 0.1017303466796875,
      "step": 6013
    },
    {
      "epoch": 0.838012958963283,
      "grad_norm": 0.2549206018447876,
      "learning_rate": 2.9355089479148003e-06,
      "loss": 0.06960105895996094,
      "step": 6014
    },
    {
      "epoch": 0.8381523026544974,
      "grad_norm": 0.4311777949333191,
      "learning_rate": 2.9305941807443903e-06,
      "loss": 0.09122419357299805,
      "step": 6015
    },
    {
      "epoch": 0.8382916463457117,
      "grad_norm": 0.41985729336738586,
      "learning_rate": 2.925683206006249e-06,
      "loss": 0.07822227478027344,
      "step": 6016
    },
    {
      "epoch": 0.8384309900369261,
      "grad_norm": 0.5131974220275879,
      "learning_rate": 2.9207760247914895e-06,
      "loss": 0.10774612426757812,
      "step": 6017
    },
    {
      "epoch": 0.8385703337281405,
      "grad_norm": 0.3060045540332794,
      "learning_rate": 2.915872638190369e-06,
      "loss": 0.07098770141601562,
      "step": 6018
    },
    {
      "epoch": 0.8387096774193549,
      "grad_norm": 0.49415498971939087,
      "learning_rate": 2.9109730472923093e-06,
      "loss": 0.11088943481445312,
      "step": 6019
    },
    {
      "epoch": 0.8388490211105692,
      "grad_norm": 0.25852856040000916,
      "learning_rate": 2.9060772531858996e-06,
      "loss": 0.07455062866210938,
      "step": 6020
    },
    {
      "epoch": 0.8389883648017836,
      "grad_norm": 0.2807832658290863,
      "learning_rate": 2.9011852569588537e-06,
      "loss": 0.08133316040039062,
      "step": 6021
    },
    {
      "epoch": 0.839127708492998,
      "grad_norm": 0.3982226550579071,
      "learning_rate": 2.896297059698072e-06,
      "loss": 0.07769775390625,
      "step": 6022
    },
    {
      "epoch": 0.8392670521842124,
      "grad_norm": 0.6757959723472595,
      "learning_rate": 2.891412662489599e-06,
      "loss": 0.124725341796875,
      "step": 6023
    },
    {
      "epoch": 0.8394063958754268,
      "grad_norm": 0.44980958104133606,
      "learning_rate": 2.8865320664186412e-06,
      "loss": 0.09700393676757812,
      "step": 6024
    },
    {
      "epoch": 0.8395457395666411,
      "grad_norm": 0.25863179564476013,
      "learning_rate": 2.8816552725695524e-06,
      "loss": 0.06827640533447266,
      "step": 6025
    },
    {
      "epoch": 0.8396850832578555,
      "grad_norm": 0.42598164081573486,
      "learning_rate": 2.8767822820258362e-06,
      "loss": 0.09490489959716797,
      "step": 6026
    },
    {
      "epoch": 0.8398244269490699,
      "grad_norm": 0.2794443964958191,
      "learning_rate": 2.8719130958701736e-06,
      "loss": 0.0713658332824707,
      "step": 6027
    },
    {
      "epoch": 0.8399637706402843,
      "grad_norm": 0.42572668194770813,
      "learning_rate": 2.867047715184377e-06,
      "loss": 0.09069061279296875,
      "step": 6028
    },
    {
      "epoch": 0.8401031143314986,
      "grad_norm": 0.4264449179172516,
      "learning_rate": 2.8621861410494234e-06,
      "loss": 0.08712005615234375,
      "step": 6029
    },
    {
      "epoch": 0.840242458022713,
      "grad_norm": 0.4531373679637909,
      "learning_rate": 2.8573283745454515e-06,
      "loss": 0.0946197509765625,
      "step": 6030
    },
    {
      "epoch": 0.8403818017139274,
      "grad_norm": 0.362354576587677,
      "learning_rate": 2.8524744167517427e-06,
      "loss": 0.08356380462646484,
      "step": 6031
    },
    {
      "epoch": 0.8405211454051418,
      "grad_norm": 0.5839719772338867,
      "learning_rate": 2.847624268746727e-06,
      "loss": 0.12095832824707031,
      "step": 6032
    },
    {
      "epoch": 0.8406604890963562,
      "grad_norm": 0.4246070384979248,
      "learning_rate": 2.842777931608005e-06,
      "loss": 0.0864706039428711,
      "step": 6033
    },
    {
      "epoch": 0.8407998327875705,
      "grad_norm": 0.43371450901031494,
      "learning_rate": 2.8379354064123245e-06,
      "loss": 0.09790515899658203,
      "step": 6034
    },
    {
      "epoch": 0.8409391764787849,
      "grad_norm": 0.45481717586517334,
      "learning_rate": 2.8330966942355752e-06,
      "loss": 0.11210823059082031,
      "step": 6035
    },
    {
      "epoch": 0.8410785201699993,
      "grad_norm": 0.28368887305259705,
      "learning_rate": 2.828261796152818e-06,
      "loss": 0.06558036804199219,
      "step": 6036
    },
    {
      "epoch": 0.8412178638612137,
      "grad_norm": 0.438705712556839,
      "learning_rate": 2.8234307132382486e-06,
      "loss": 0.09321022033691406,
      "step": 6037
    },
    {
      "epoch": 0.841357207552428,
      "grad_norm": 0.3374811112880707,
      "learning_rate": 2.818603446565231e-06,
      "loss": 0.07131576538085938,
      "step": 6038
    },
    {
      "epoch": 0.8414965512436424,
      "grad_norm": 0.47786885499954224,
      "learning_rate": 2.813779997206265e-06,
      "loss": 0.08928489685058594,
      "step": 6039
    },
    {
      "epoch": 0.8416358949348568,
      "grad_norm": 0.4252340793609619,
      "learning_rate": 2.8089603662330155e-06,
      "loss": 0.08931732177734375,
      "step": 6040
    },
    {
      "epoch": 0.8417752386260712,
      "grad_norm": 1.1540358066558838,
      "learning_rate": 2.8041445547162994e-06,
      "loss": 0.1344432830810547,
      "step": 6041
    },
    {
      "epoch": 0.8419145823172856,
      "grad_norm": 0.3762575089931488,
      "learning_rate": 2.7993325637260738e-06,
      "loss": 0.0845499038696289,
      "step": 6042
    },
    {
      "epoch": 0.8420539260084999,
      "grad_norm": 0.6358121633529663,
      "learning_rate": 2.79452439433145e-06,
      "loss": 0.09679603576660156,
      "step": 6043
    },
    {
      "epoch": 0.8421932696997143,
      "grad_norm": 0.3084847927093506,
      "learning_rate": 2.7897200476007015e-06,
      "loss": 0.07917404174804688,
      "step": 6044
    },
    {
      "epoch": 0.8423326133909287,
      "grad_norm": 0.6463673710823059,
      "learning_rate": 2.7849195246012417e-06,
      "loss": 0.09175682067871094,
      "step": 6045
    },
    {
      "epoch": 0.8424719570821431,
      "grad_norm": 0.2675143778324127,
      "learning_rate": 2.780122826399634e-06,
      "loss": 0.06274604797363281,
      "step": 6046
    },
    {
      "epoch": 0.8426113007733574,
      "grad_norm": 0.4555037021636963,
      "learning_rate": 2.775329954061603e-06,
      "loss": 0.08306884765625,
      "step": 6047
    },
    {
      "epoch": 0.8427506444645718,
      "grad_norm": 0.3516727685928345,
      "learning_rate": 2.770540908652004e-06,
      "loss": 0.08041667938232422,
      "step": 6048
    },
    {
      "epoch": 0.8428899881557862,
      "grad_norm": 0.6985844373703003,
      "learning_rate": 2.765755691234866e-06,
      "loss": 0.12052536010742188,
      "step": 6049
    },
    {
      "epoch": 0.8430293318470006,
      "grad_norm": 0.3218729496002197,
      "learning_rate": 2.7609743028733427e-06,
      "loss": 0.0764617919921875,
      "step": 6050
    },
    {
      "epoch": 0.8431686755382151,
      "grad_norm": 0.5126708745956421,
      "learning_rate": 2.7561967446297555e-06,
      "loss": 0.09080219268798828,
      "step": 6051
    },
    {
      "epoch": 0.8433080192294294,
      "grad_norm": 0.6177491545677185,
      "learning_rate": 2.75142301756558e-06,
      "loss": 0.12530899047851562,
      "step": 6052
    },
    {
      "epoch": 0.8434473629206438,
      "grad_norm": 0.7910027503967285,
      "learning_rate": 2.7466531227414074e-06,
      "loss": 0.11333656311035156,
      "step": 6053
    },
    {
      "epoch": 0.8435867066118582,
      "grad_norm": 0.2803467810153961,
      "learning_rate": 2.7418870612170123e-06,
      "loss": 0.0684051513671875,
      "step": 6054
    },
    {
      "epoch": 0.8437260503030726,
      "grad_norm": 0.5629757642745972,
      "learning_rate": 2.737124834051301e-06,
      "loss": 0.11083221435546875,
      "step": 6055
    },
    {
      "epoch": 0.843865393994287,
      "grad_norm": 0.3576008379459381,
      "learning_rate": 2.732366442302339e-06,
      "loss": 0.09086227416992188,
      "step": 6056
    },
    {
      "epoch": 0.8440047376855013,
      "grad_norm": 0.3193489909172058,
      "learning_rate": 2.727611887027326e-06,
      "loss": 0.08743476867675781,
      "step": 6057
    },
    {
      "epoch": 0.8441440813767157,
      "grad_norm": 0.34056103229522705,
      "learning_rate": 2.722861169282611e-06,
      "loss": 0.08976459503173828,
      "step": 6058
    },
    {
      "epoch": 0.8442834250679301,
      "grad_norm": 0.6656012535095215,
      "learning_rate": 2.7181142901237077e-06,
      "loss": 0.12055301666259766,
      "step": 6059
    },
    {
      "epoch": 0.8444227687591445,
      "grad_norm": 0.5287493467330933,
      "learning_rate": 2.71337125060525e-06,
      "loss": 0.13140106201171875,
      "step": 6060
    },
    {
      "epoch": 0.8445621124503588,
      "grad_norm": 0.3663991391658783,
      "learning_rate": 2.7086320517810416e-06,
      "loss": 0.08819198608398438,
      "step": 6061
    },
    {
      "epoch": 0.8447014561415732,
      "grad_norm": 0.5506178140640259,
      "learning_rate": 2.7038966947040247e-06,
      "loss": 0.12643051147460938,
      "step": 6062
    },
    {
      "epoch": 0.8448407998327876,
      "grad_norm": 0.3241470754146576,
      "learning_rate": 2.6991651804262886e-06,
      "loss": 0.08289813995361328,
      "step": 6063
    },
    {
      "epoch": 0.844980143524002,
      "grad_norm": 0.35260558128356934,
      "learning_rate": 2.694437509999057e-06,
      "loss": 0.07013130187988281,
      "step": 6064
    },
    {
      "epoch": 0.8451194872152163,
      "grad_norm": 0.46374887228012085,
      "learning_rate": 2.6897136844727214e-06,
      "loss": 0.08963680267333984,
      "step": 6065
    },
    {
      "epoch": 0.8452588309064307,
      "grad_norm": 0.238925501704216,
      "learning_rate": 2.6849937048968056e-06,
      "loss": 0.06758499145507812,
      "step": 6066
    },
    {
      "epoch": 0.8453981745976451,
      "grad_norm": 0.31748783588409424,
      "learning_rate": 2.680277572319978e-06,
      "loss": 0.08249664306640625,
      "step": 6067
    },
    {
      "epoch": 0.8455375182888595,
      "grad_norm": 0.2243543565273285,
      "learning_rate": 2.675565287790063e-06,
      "loss": 0.06714248657226562,
      "step": 6068
    },
    {
      "epoch": 0.8456768619800739,
      "grad_norm": 0.45018714666366577,
      "learning_rate": 2.6708568523540114e-06,
      "loss": 0.09334468841552734,
      "step": 6069
    },
    {
      "epoch": 0.8458162056712882,
      "grad_norm": 0.21868069469928741,
      "learning_rate": 2.6661522670579398e-06,
      "loss": 0.06909561157226562,
      "step": 6070
    },
    {
      "epoch": 0.8459555493625026,
      "grad_norm": 0.24899961054325104,
      "learning_rate": 2.6614515329470923e-06,
      "loss": 0.06824970245361328,
      "step": 6071
    },
    {
      "epoch": 0.846094893053717,
      "grad_norm": 0.22221729159355164,
      "learning_rate": 2.656754651065869e-06,
      "loss": 0.061514854431152344,
      "step": 6072
    },
    {
      "epoch": 0.8462342367449314,
      "grad_norm": 0.2697114646434784,
      "learning_rate": 2.652061622457813e-06,
      "loss": 0.07717323303222656,
      "step": 6073
    },
    {
      "epoch": 0.8463735804361457,
      "grad_norm": 0.5473085045814514,
      "learning_rate": 2.647372448165606e-06,
      "loss": 0.11187553405761719,
      "step": 6074
    },
    {
      "epoch": 0.8465129241273601,
      "grad_norm": 0.30559125542640686,
      "learning_rate": 2.6426871292310675e-06,
      "loss": 0.07525825500488281,
      "step": 6075
    },
    {
      "epoch": 0.8466522678185745,
      "grad_norm": 0.6351538896560669,
      "learning_rate": 2.638005666695176e-06,
      "loss": 0.10077667236328125,
      "step": 6076
    },
    {
      "epoch": 0.8467916115097889,
      "grad_norm": 0.48891040682792664,
      "learning_rate": 2.6333280615980483e-06,
      "loss": 0.09486198425292969,
      "step": 6077
    },
    {
      "epoch": 0.8469309552010033,
      "grad_norm": 0.5667522549629211,
      "learning_rate": 2.6286543149789355e-06,
      "loss": 0.11515045166015625,
      "step": 6078
    },
    {
      "epoch": 0.8470702988922176,
      "grad_norm": 0.4492383301258087,
      "learning_rate": 2.6239844278762384e-06,
      "loss": 0.10705947875976562,
      "step": 6079
    },
    {
      "epoch": 0.847209642583432,
      "grad_norm": 0.7337267994880676,
      "learning_rate": 2.6193184013274975e-06,
      "loss": 0.12817955017089844,
      "step": 6080
    },
    {
      "epoch": 0.8473489862746464,
      "grad_norm": 0.38938257098197937,
      "learning_rate": 2.614656236369406e-06,
      "loss": 0.08687973022460938,
      "step": 6081
    },
    {
      "epoch": 0.8474883299658608,
      "grad_norm": 0.4894862174987793,
      "learning_rate": 2.60999793403778e-06,
      "loss": 0.11002159118652344,
      "step": 6082
    },
    {
      "epoch": 0.8476276736570751,
      "grad_norm": 0.369227796792984,
      "learning_rate": 2.605343495367596e-06,
      "loss": 0.08879852294921875,
      "step": 6083
    },
    {
      "epoch": 0.8477670173482895,
      "grad_norm": 0.47201672196388245,
      "learning_rate": 2.6006929213929576e-06,
      "loss": 0.114013671875,
      "step": 6084
    },
    {
      "epoch": 0.8479063610395039,
      "grad_norm": 0.4520953297615051,
      "learning_rate": 2.5960462131471233e-06,
      "loss": 0.08694267272949219,
      "step": 6085
    },
    {
      "epoch": 0.8480457047307183,
      "grad_norm": 0.4896768629550934,
      "learning_rate": 2.5914033716624754e-06,
      "loss": 0.09000873565673828,
      "step": 6086
    },
    {
      "epoch": 0.8481850484219327,
      "grad_norm": 0.3052300214767456,
      "learning_rate": 2.5867643979705535e-06,
      "loss": 0.06956005096435547,
      "step": 6087
    },
    {
      "epoch": 0.848324392113147,
      "grad_norm": 0.42583993077278137,
      "learning_rate": 2.582129293102038e-06,
      "loss": 0.07254981994628906,
      "step": 6088
    },
    {
      "epoch": 0.8484637358043614,
      "grad_norm": 0.31622254848480225,
      "learning_rate": 2.577498058086736e-06,
      "loss": 0.08102607727050781,
      "step": 6089
    },
    {
      "epoch": 0.8486030794955758,
      "grad_norm": 0.32135188579559326,
      "learning_rate": 2.5728706939535976e-06,
      "loss": 0.07416534423828125,
      "step": 6090
    },
    {
      "epoch": 0.8487424231867903,
      "grad_norm": 0.2672739624977112,
      "learning_rate": 2.568247201730727e-06,
      "loss": 0.07895469665527344,
      "step": 6091
    },
    {
      "epoch": 0.8488817668780047,
      "grad_norm": 0.32579171657562256,
      "learning_rate": 2.5636275824453515e-06,
      "loss": 0.08190727233886719,
      "step": 6092
    },
    {
      "epoch": 0.849021110569219,
      "grad_norm": 0.518416702747345,
      "learning_rate": 2.559011837123846e-06,
      "loss": 0.1014871597290039,
      "step": 6093
    },
    {
      "epoch": 0.8491604542604334,
      "grad_norm": 0.6070149540901184,
      "learning_rate": 2.554399966791732e-06,
      "loss": 0.11040782928466797,
      "step": 6094
    },
    {
      "epoch": 0.8492997979516478,
      "grad_norm": 0.26617398858070374,
      "learning_rate": 2.5497919724736564e-06,
      "loss": 0.0724630355834961,
      "step": 6095
    },
    {
      "epoch": 0.8494391416428622,
      "grad_norm": 0.44468075037002563,
      "learning_rate": 2.545187855193403e-06,
      "loss": 0.11523056030273438,
      "step": 6096
    },
    {
      "epoch": 0.8495784853340765,
      "grad_norm": 0.3417418897151947,
      "learning_rate": 2.5405876159739083e-06,
      "loss": 0.09253311157226562,
      "step": 6097
    },
    {
      "epoch": 0.8497178290252909,
      "grad_norm": 0.6098604202270508,
      "learning_rate": 2.535991255837247e-06,
      "loss": 0.1244049072265625,
      "step": 6098
    },
    {
      "epoch": 0.8498571727165053,
      "grad_norm": 0.42331188917160034,
      "learning_rate": 2.5313987758046164e-06,
      "loss": 0.10073089599609375,
      "step": 6099
    },
    {
      "epoch": 0.8499965164077197,
      "grad_norm": 0.41634684801101685,
      "learning_rate": 2.526810176896357e-06,
      "loss": 0.09232425689697266,
      "step": 6100
    },
    {
      "epoch": 0.850135860098934,
      "grad_norm": 0.48981231451034546,
      "learning_rate": 2.5222254601319595e-06,
      "loss": 0.0906972885131836,
      "step": 6101
    },
    {
      "epoch": 0.8502752037901484,
      "grad_norm": 0.3192691206932068,
      "learning_rate": 2.5176446265300424e-06,
      "loss": 0.07143211364746094,
      "step": 6102
    },
    {
      "epoch": 0.8504145474813628,
      "grad_norm": 0.6801007390022278,
      "learning_rate": 2.5130676771083585e-06,
      "loss": 0.09728527069091797,
      "step": 6103
    },
    {
      "epoch": 0.8505538911725772,
      "grad_norm": 0.3933090567588806,
      "learning_rate": 2.5084946128838007e-06,
      "loss": 0.08660697937011719,
      "step": 6104
    },
    {
      "epoch": 0.8506932348637916,
      "grad_norm": 0.4547611474990845,
      "learning_rate": 2.5039254348724096e-06,
      "loss": 0.11753273010253906,
      "step": 6105
    },
    {
      "epoch": 0.8508325785550059,
      "grad_norm": 0.5446875691413879,
      "learning_rate": 2.499360144089342e-06,
      "loss": 0.08763694763183594,
      "step": 6106
    },
    {
      "epoch": 0.8509719222462203,
      "grad_norm": 0.27666905522346497,
      "learning_rate": 2.4947987415489004e-06,
      "loss": 0.07766246795654297,
      "step": 6107
    },
    {
      "epoch": 0.8511112659374347,
      "grad_norm": 0.5309749841690063,
      "learning_rate": 2.490241228264527e-06,
      "loss": 0.09719657897949219,
      "step": 6108
    },
    {
      "epoch": 0.8512506096286491,
      "grad_norm": 0.46976763010025024,
      "learning_rate": 2.4856876052488032e-06,
      "loss": 0.09152984619140625,
      "step": 6109
    },
    {
      "epoch": 0.8513899533198634,
      "grad_norm": 0.2724052965641022,
      "learning_rate": 2.481137873513435e-06,
      "loss": 0.07703590393066406,
      "step": 6110
    },
    {
      "epoch": 0.8515292970110778,
      "grad_norm": 0.5727488994598389,
      "learning_rate": 2.476592034069265e-06,
      "loss": 0.10336875915527344,
      "step": 6111
    },
    {
      "epoch": 0.8516686407022922,
      "grad_norm": 0.23325027525424957,
      "learning_rate": 2.4720500879262787e-06,
      "loss": 0.06134986877441406,
      "step": 6112
    },
    {
      "epoch": 0.8518079843935066,
      "grad_norm": 0.551236093044281,
      "learning_rate": 2.4675120360935977e-06,
      "loss": 0.08746910095214844,
      "step": 6113
    },
    {
      "epoch": 0.851947328084721,
      "grad_norm": 0.8448876738548279,
      "learning_rate": 2.4629778795794623e-06,
      "loss": 0.11013984680175781,
      "step": 6114
    },
    {
      "epoch": 0.8520866717759353,
      "grad_norm": 0.47471508383750916,
      "learning_rate": 2.458447619391271e-06,
      "loss": 0.10681915283203125,
      "step": 6115
    },
    {
      "epoch": 0.8522260154671497,
      "grad_norm": 0.5640620589256287,
      "learning_rate": 2.453921256535534e-06,
      "loss": 0.12112998962402344,
      "step": 6116
    },
    {
      "epoch": 0.8523653591583641,
      "grad_norm": 0.40197497606277466,
      "learning_rate": 2.449398792017914e-06,
      "loss": 0.09423637390136719,
      "step": 6117
    },
    {
      "epoch": 0.8525047028495785,
      "grad_norm": 0.6447674632072449,
      "learning_rate": 2.4448802268431914e-06,
      "loss": 0.10263633728027344,
      "step": 6118
    },
    {
      "epoch": 0.8526440465407928,
      "grad_norm": 0.19194839894771576,
      "learning_rate": 2.440365562015292e-06,
      "loss": 0.06379890441894531,
      "step": 6119
    },
    {
      "epoch": 0.8527833902320072,
      "grad_norm": 0.3000054955482483,
      "learning_rate": 2.4358547985372806e-06,
      "loss": 0.077667236328125,
      "step": 6120
    },
    {
      "epoch": 0.8529227339232216,
      "grad_norm": 0.5129270553588867,
      "learning_rate": 2.431347937411328e-06,
      "loss": 0.10344505310058594,
      "step": 6121
    },
    {
      "epoch": 0.853062077614436,
      "grad_norm": 0.39717739820480347,
      "learning_rate": 2.426844979638763e-06,
      "loss": 0.09134674072265625,
      "step": 6122
    },
    {
      "epoch": 0.8532014213056504,
      "grad_norm": 0.37627261877059937,
      "learning_rate": 2.4223459262200422e-06,
      "loss": 0.08382034301757812,
      "step": 6123
    },
    {
      "epoch": 0.8533407649968647,
      "grad_norm": 0.32273003458976746,
      "learning_rate": 2.4178507781547577e-06,
      "loss": 0.08122634887695312,
      "step": 6124
    },
    {
      "epoch": 0.8534801086880791,
      "grad_norm": 0.1896367371082306,
      "learning_rate": 2.4133595364416174e-06,
      "loss": 0.0527191162109375,
      "step": 6125
    },
    {
      "epoch": 0.8536194523792935,
      "grad_norm": 0.30182042717933655,
      "learning_rate": 2.4088722020784828e-06,
      "loss": 0.07706832885742188,
      "step": 6126
    },
    {
      "epoch": 0.8537587960705079,
      "grad_norm": 0.4277908205986023,
      "learning_rate": 2.404388776062332e-06,
      "loss": 0.09140205383300781,
      "step": 6127
    },
    {
      "epoch": 0.8538981397617222,
      "grad_norm": 0.4095993638038635,
      "learning_rate": 2.399909259389277e-06,
      "loss": 0.09116363525390625,
      "step": 6128
    },
    {
      "epoch": 0.8540374834529366,
      "grad_norm": 0.3366397023200989,
      "learning_rate": 2.3954336530545684e-06,
      "loss": 0.07817244529724121,
      "step": 6129
    },
    {
      "epoch": 0.854176827144151,
      "grad_norm": 0.7705382704734802,
      "learning_rate": 2.3909619580525867e-06,
      "loss": 0.1641864776611328,
      "step": 6130
    },
    {
      "epoch": 0.8543161708353654,
      "grad_norm": 0.2510642409324646,
      "learning_rate": 2.3864941753768345e-06,
      "loss": 0.07448959350585938,
      "step": 6131
    },
    {
      "epoch": 0.8544555145265799,
      "grad_norm": 0.5485810041427612,
      "learning_rate": 2.3820303060199513e-06,
      "loss": 0.0965738296508789,
      "step": 6132
    },
    {
      "epoch": 0.8545948582177942,
      "grad_norm": 0.5430808067321777,
      "learning_rate": 2.3775703509737102e-06,
      "loss": 0.10214042663574219,
      "step": 6133
    },
    {
      "epoch": 0.8547342019090086,
      "grad_norm": 0.47783732414245605,
      "learning_rate": 2.3731143112290124e-06,
      "loss": 0.08200263977050781,
      "step": 6134
    },
    {
      "epoch": 0.854873545600223,
      "grad_norm": 0.5380401611328125,
      "learning_rate": 2.368662187775883e-06,
      "loss": 0.11745071411132812,
      "step": 6135
    },
    {
      "epoch": 0.8550128892914374,
      "grad_norm": 0.4718419909477234,
      "learning_rate": 2.3642139816034872e-06,
      "loss": 0.09151268005371094,
      "step": 6136
    },
    {
      "epoch": 0.8551522329826517,
      "grad_norm": 0.38381996750831604,
      "learning_rate": 2.359769693700111e-06,
      "loss": 0.0989227294921875,
      "step": 6137
    },
    {
      "epoch": 0.8552915766738661,
      "grad_norm": 0.34954148530960083,
      "learning_rate": 2.3553293250531794e-06,
      "loss": 0.08082962036132812,
      "step": 6138
    },
    {
      "epoch": 0.8554309203650805,
      "grad_norm": 0.4856241047382355,
      "learning_rate": 2.350892876649231e-06,
      "loss": 0.08167171478271484,
      "step": 6139
    },
    {
      "epoch": 0.8555702640562949,
      "grad_norm": 0.6339848041534424,
      "learning_rate": 2.3464603494739513e-06,
      "loss": 0.10645866394042969,
      "step": 6140
    },
    {
      "epoch": 0.8557096077475093,
      "grad_norm": 0.8167253732681274,
      "learning_rate": 2.3420317445121497e-06,
      "loss": 0.12325668334960938,
      "step": 6141
    },
    {
      "epoch": 0.8558489514387236,
      "grad_norm": 0.45094019174575806,
      "learning_rate": 2.3376070627477555e-06,
      "loss": 0.09259605407714844,
      "step": 6142
    },
    {
      "epoch": 0.855988295129938,
      "grad_norm": 0.19517521560192108,
      "learning_rate": 2.333186305163828e-06,
      "loss": 0.05629730224609375,
      "step": 6143
    },
    {
      "epoch": 0.8561276388211524,
      "grad_norm": 0.5733916163444519,
      "learning_rate": 2.3287694727425623e-06,
      "loss": 0.09826278686523438,
      "step": 6144
    },
    {
      "epoch": 0.8562669825123668,
      "grad_norm": 0.2823205888271332,
      "learning_rate": 2.3243565664652844e-06,
      "loss": 0.07054615020751953,
      "step": 6145
    },
    {
      "epoch": 0.8564063262035811,
      "grad_norm": 0.5767850279808044,
      "learning_rate": 2.31994758731243e-06,
      "loss": 0.1253185272216797,
      "step": 6146
    },
    {
      "epoch": 0.8565456698947955,
      "grad_norm": 0.4193795323371887,
      "learning_rate": 2.3155425362635863e-06,
      "loss": 0.08545112609863281,
      "step": 6147
    },
    {
      "epoch": 0.8566850135860099,
      "grad_norm": 0.3408593237400055,
      "learning_rate": 2.311141414297442e-06,
      "loss": 0.06340408325195312,
      "step": 6148
    },
    {
      "epoch": 0.8568243572772243,
      "grad_norm": 0.2660634219646454,
      "learning_rate": 2.3067442223918345e-06,
      "loss": 0.0651702880859375,
      "step": 6149
    },
    {
      "epoch": 0.8569637009684387,
      "grad_norm": 0.5275837779045105,
      "learning_rate": 2.3023509615237138e-06,
      "loss": 0.10289573669433594,
      "step": 6150
    },
    {
      "epoch": 0.857103044659653,
      "grad_norm": 0.26569780707359314,
      "learning_rate": 2.2979616326691658e-06,
      "loss": 0.06591224670410156,
      "step": 6151
    },
    {
      "epoch": 0.8572423883508674,
      "grad_norm": 0.21894851326942444,
      "learning_rate": 2.2935762368034075e-06,
      "loss": 0.06441116333007812,
      "step": 6152
    },
    {
      "epoch": 0.8573817320420818,
      "grad_norm": 0.345255047082901,
      "learning_rate": 2.289194774900756e-06,
      "loss": 0.0834512710571289,
      "step": 6153
    },
    {
      "epoch": 0.8575210757332962,
      "grad_norm": 0.23532091081142426,
      "learning_rate": 2.2848172479346806e-06,
      "loss": 0.06847763061523438,
      "step": 6154
    },
    {
      "epoch": 0.8576604194245105,
      "grad_norm": 0.5739054679870605,
      "learning_rate": 2.280443656877769e-06,
      "loss": 0.12067413330078125,
      "step": 6155
    },
    {
      "epoch": 0.8577997631157249,
      "grad_norm": 0.376862496137619,
      "learning_rate": 2.276074002701736e-06,
      "loss": 0.07921409606933594,
      "step": 6156
    },
    {
      "epoch": 0.8579391068069393,
      "grad_norm": 0.3740381896495819,
      "learning_rate": 2.271708286377414e-06,
      "loss": 0.07156181335449219,
      "step": 6157
    },
    {
      "epoch": 0.8580784504981537,
      "grad_norm": 0.46686938405036926,
      "learning_rate": 2.267346508874766e-06,
      "loss": 0.0925140380859375,
      "step": 6158
    },
    {
      "epoch": 0.858217794189368,
      "grad_norm": 0.21648165583610535,
      "learning_rate": 2.262988671162882e-06,
      "loss": 0.05639076232910156,
      "step": 6159
    },
    {
      "epoch": 0.8583571378805824,
      "grad_norm": 0.28644120693206787,
      "learning_rate": 2.258634774209969e-06,
      "loss": 0.06700897216796875,
      "step": 6160
    },
    {
      "epoch": 0.8584964815717968,
      "grad_norm": 0.891989529132843,
      "learning_rate": 2.2542848189833675e-06,
      "loss": 0.10189247131347656,
      "step": 6161
    },
    {
      "epoch": 0.8586358252630112,
      "grad_norm": 0.3060687482357025,
      "learning_rate": 2.249938806449539e-06,
      "loss": 0.06688880920410156,
      "step": 6162
    },
    {
      "epoch": 0.8587751689542256,
      "grad_norm": 0.5004172325134277,
      "learning_rate": 2.2455967375740716e-06,
      "loss": 0.1014251708984375,
      "step": 6163
    },
    {
      "epoch": 0.8589145126454399,
      "grad_norm": 0.2624199688434601,
      "learning_rate": 2.241258613321664e-06,
      "loss": 0.07732963562011719,
      "step": 6164
    },
    {
      "epoch": 0.8590538563366543,
      "grad_norm": 0.7008524537086487,
      "learning_rate": 2.2369244346561516e-06,
      "loss": 0.10513687133789062,
      "step": 6165
    },
    {
      "epoch": 0.8591932000278687,
      "grad_norm": 0.6817530989646912,
      "learning_rate": 2.2325942025404968e-06,
      "loss": 0.09541511535644531,
      "step": 6166
    },
    {
      "epoch": 0.8593325437190831,
      "grad_norm": 0.6515265107154846,
      "learning_rate": 2.2282679179367684e-06,
      "loss": 0.12012481689453125,
      "step": 6167
    },
    {
      "epoch": 0.8594718874102975,
      "grad_norm": 0.35227033495903015,
      "learning_rate": 2.2239455818061793e-06,
      "loss": 0.07697486877441406,
      "step": 6168
    },
    {
      "epoch": 0.8596112311015118,
      "grad_norm": 0.7372114658355713,
      "learning_rate": 2.219627195109042e-06,
      "loss": 0.10674285888671875,
      "step": 6169
    },
    {
      "epoch": 0.8597505747927262,
      "grad_norm": 0.3742528259754181,
      "learning_rate": 2.2153127588048127e-06,
      "loss": 0.08556556701660156,
      "step": 6170
    },
    {
      "epoch": 0.8598899184839406,
      "grad_norm": 0.6937990188598633,
      "learning_rate": 2.2110022738520543e-06,
      "loss": 0.108154296875,
      "step": 6171
    },
    {
      "epoch": 0.8600292621751551,
      "grad_norm": 0.28995242714881897,
      "learning_rate": 2.20669574120846e-06,
      "loss": 0.08116436004638672,
      "step": 6172
    },
    {
      "epoch": 0.8601686058663695,
      "grad_norm": 0.4390193223953247,
      "learning_rate": 2.202393161830849e-06,
      "loss": 0.09386062622070312,
      "step": 6173
    },
    {
      "epoch": 0.8603079495575838,
      "grad_norm": 0.3394027054309845,
      "learning_rate": 2.1980945366751503e-06,
      "loss": 0.08933067321777344,
      "step": 6174
    },
    {
      "epoch": 0.8604472932487982,
      "grad_norm": 1.2987918853759766,
      "learning_rate": 2.1937998666964176e-06,
      "loss": 0.16503620147705078,
      "step": 6175
    },
    {
      "epoch": 0.8605866369400126,
      "grad_norm": 0.4892064034938812,
      "learning_rate": 2.189509152848832e-06,
      "loss": 0.10833740234375,
      "step": 6176
    },
    {
      "epoch": 0.860725980631227,
      "grad_norm": 0.34563905000686646,
      "learning_rate": 2.185222396085698e-06,
      "loss": 0.07393074035644531,
      "step": 6177
    },
    {
      "epoch": 0.8608653243224413,
      "grad_norm": 0.3832445740699768,
      "learning_rate": 2.1809395973594263e-06,
      "loss": 0.07958221435546875,
      "step": 6178
    },
    {
      "epoch": 0.8610046680136557,
      "grad_norm": 0.22865910828113556,
      "learning_rate": 2.176660757621558e-06,
      "loss": 0.05886268615722656,
      "step": 6179
    },
    {
      "epoch": 0.8611440117048701,
      "grad_norm": 0.24472646415233612,
      "learning_rate": 2.1723858778227537e-06,
      "loss": 0.06394648551940918,
      "step": 6180
    },
    {
      "epoch": 0.8612833553960845,
      "grad_norm": 0.4410075545310974,
      "learning_rate": 2.1681149589128016e-06,
      "loss": 0.08817100524902344,
      "step": 6181
    },
    {
      "epoch": 0.8614226990872988,
      "grad_norm": 0.5140257477760315,
      "learning_rate": 2.1638480018405916e-06,
      "loss": 0.0994405746459961,
      "step": 6182
    },
    {
      "epoch": 0.8615620427785132,
      "grad_norm": 0.23163393139839172,
      "learning_rate": 2.15958500755415e-06,
      "loss": 0.06215190887451172,
      "step": 6183
    },
    {
      "epoch": 0.8617013864697276,
      "grad_norm": 0.348198264837265,
      "learning_rate": 2.1553259770006196e-06,
      "loss": 0.07444000244140625,
      "step": 6184
    },
    {
      "epoch": 0.861840730160942,
      "grad_norm": 0.4754301607608795,
      "learning_rate": 2.1510709111262584e-06,
      "loss": 0.09931755065917969,
      "step": 6185
    },
    {
      "epoch": 0.8619800738521564,
      "grad_norm": 0.5701733827590942,
      "learning_rate": 2.1468198108764373e-06,
      "loss": 0.11285781860351562,
      "step": 6186
    },
    {
      "epoch": 0.8621194175433707,
      "grad_norm": 0.25285398960113525,
      "learning_rate": 2.1425726771956578e-06,
      "loss": 0.07250022888183594,
      "step": 6187
    },
    {
      "epoch": 0.8622587612345851,
      "grad_norm": 0.6970971822738647,
      "learning_rate": 2.1383295110275437e-06,
      "loss": 0.12930011749267578,
      "step": 6188
    },
    {
      "epoch": 0.8623981049257995,
      "grad_norm": 0.7598935961723328,
      "learning_rate": 2.1340903133148205e-06,
      "loss": 0.1602611541748047,
      "step": 6189
    },
    {
      "epoch": 0.8625374486170139,
      "grad_norm": 0.3051297962665558,
      "learning_rate": 2.1298550849993437e-06,
      "loss": 0.0764760971069336,
      "step": 6190
    },
    {
      "epoch": 0.8626767923082282,
      "grad_norm": 0.6686179637908936,
      "learning_rate": 2.1256238270220853e-06,
      "loss": 0.09928512573242188,
      "step": 6191
    },
    {
      "epoch": 0.8628161359994426,
      "grad_norm": 0.373225599527359,
      "learning_rate": 2.1213965403231328e-06,
      "loss": 0.10668563842773438,
      "step": 6192
    },
    {
      "epoch": 0.862955479690657,
      "grad_norm": 0.5725539922714233,
      "learning_rate": 2.117173225841691e-06,
      "loss": 0.11554384231567383,
      "step": 6193
    },
    {
      "epoch": 0.8630948233818714,
      "grad_norm": 0.35374873876571655,
      "learning_rate": 2.112953884516091e-06,
      "loss": 0.0678253173828125,
      "step": 6194
    },
    {
      "epoch": 0.8632341670730858,
      "grad_norm": 0.267788827419281,
      "learning_rate": 2.1087385172837705e-06,
      "loss": 0.06675338745117188,
      "step": 6195
    },
    {
      "epoch": 0.8633735107643001,
      "grad_norm": 0.6978381276130676,
      "learning_rate": 2.1045271250812817e-06,
      "loss": 0.12001991271972656,
      "step": 6196
    },
    {
      "epoch": 0.8635128544555145,
      "grad_norm": 0.5822411179542542,
      "learning_rate": 2.100319708844307e-06,
      "loss": 0.118865966796875,
      "step": 6197
    },
    {
      "epoch": 0.8636521981467289,
      "grad_norm": 0.6201586723327637,
      "learning_rate": 2.0961162695076397e-06,
      "loss": 0.11176395416259766,
      "step": 6198
    },
    {
      "epoch": 0.8637915418379433,
      "grad_norm": 0.47711411118507385,
      "learning_rate": 2.091916808005179e-06,
      "loss": 0.11138343811035156,
      "step": 6199
    },
    {
      "epoch": 0.8639308855291576,
      "grad_norm": 0.5287181735038757,
      "learning_rate": 2.08772132526996e-06,
      "loss": 0.09464645385742188,
      "step": 6200
    },
    {
      "epoch": 0.864070229220372,
      "grad_norm": 0.369865357875824,
      "learning_rate": 2.083529822234116e-06,
      "loss": 0.09836196899414062,
      "step": 6201
    },
    {
      "epoch": 0.8642095729115864,
      "grad_norm": 0.40709400177001953,
      "learning_rate": 2.079342299828908e-06,
      "loss": 0.07903099060058594,
      "step": 6202
    },
    {
      "epoch": 0.8643489166028008,
      "grad_norm": 0.500076949596405,
      "learning_rate": 2.075158758984701e-06,
      "loss": 0.08227157592773438,
      "step": 6203
    },
    {
      "epoch": 0.8644882602940152,
      "grad_norm": 0.4871565103530884,
      "learning_rate": 2.070979200630987e-06,
      "loss": 0.08368682861328125,
      "step": 6204
    },
    {
      "epoch": 0.8646276039852295,
      "grad_norm": 0.5697380900382996,
      "learning_rate": 2.0668036256963743e-06,
      "loss": 0.11416244506835938,
      "step": 6205
    },
    {
      "epoch": 0.8647669476764439,
      "grad_norm": 0.22186125814914703,
      "learning_rate": 2.0626320351085716e-06,
      "loss": 0.06265544891357422,
      "step": 6206
    },
    {
      "epoch": 0.8649062913676583,
      "grad_norm": 0.3469395339488983,
      "learning_rate": 2.058464429794409e-06,
      "loss": 0.07046699523925781,
      "step": 6207
    },
    {
      "epoch": 0.8650456350588727,
      "grad_norm": 0.40573573112487793,
      "learning_rate": 2.054300810679839e-06,
      "loss": 0.08606910705566406,
      "step": 6208
    },
    {
      "epoch": 0.865184978750087,
      "grad_norm": 0.2924923300743103,
      "learning_rate": 2.0501411786899263e-06,
      "loss": 0.07188987731933594,
      "step": 6209
    },
    {
      "epoch": 0.8653243224413014,
      "grad_norm": 0.2580891251564026,
      "learning_rate": 2.045985534748842e-06,
      "loss": 0.07619667053222656,
      "step": 6210
    },
    {
      "epoch": 0.8654636661325158,
      "grad_norm": 0.30118241906166077,
      "learning_rate": 2.0418338797798686e-06,
      "loss": 0.08199119567871094,
      "step": 6211
    },
    {
      "epoch": 0.8656030098237303,
      "grad_norm": 0.4043683111667633,
      "learning_rate": 2.0376862147054164e-06,
      "loss": 0.10099411010742188,
      "step": 6212
    },
    {
      "epoch": 0.8657423535149447,
      "grad_norm": 0.4757402539253235,
      "learning_rate": 2.0335425404470045e-06,
      "loss": 0.08879566192626953,
      "step": 6213
    },
    {
      "epoch": 0.865881697206159,
      "grad_norm": 0.3641589879989624,
      "learning_rate": 2.0294028579252557e-06,
      "loss": 0.08553504943847656,
      "step": 6214
    },
    {
      "epoch": 0.8660210408973734,
      "grad_norm": 0.3024021089076996,
      "learning_rate": 2.025267168059919e-06,
      "loss": 0.07017135620117188,
      "step": 6215
    },
    {
      "epoch": 0.8661603845885878,
      "grad_norm": 0.2530957758426666,
      "learning_rate": 2.0211354717698437e-06,
      "loss": 0.06912803649902344,
      "step": 6216
    },
    {
      "epoch": 0.8662997282798022,
      "grad_norm": 0.27706608176231384,
      "learning_rate": 2.017007769973005e-06,
      "loss": 0.07220077514648438,
      "step": 6217
    },
    {
      "epoch": 0.8664390719710166,
      "grad_norm": 0.40846511721611023,
      "learning_rate": 2.01288406358648e-06,
      "loss": 0.08023834228515625,
      "step": 6218
    },
    {
      "epoch": 0.8665784156622309,
      "grad_norm": 0.543289065361023,
      "learning_rate": 2.00876435352646e-06,
      "loss": 0.09698677062988281,
      "step": 6219
    },
    {
      "epoch": 0.8667177593534453,
      "grad_norm": 0.20729734003543854,
      "learning_rate": 2.00464864070826e-06,
      "loss": 0.05886268615722656,
      "step": 6220
    },
    {
      "epoch": 0.8668571030446597,
      "grad_norm": 0.7065600752830505,
      "learning_rate": 2.0005369260462904e-06,
      "loss": 0.09293365478515625,
      "step": 6221
    },
    {
      "epoch": 0.8669964467358741,
      "grad_norm": 0.8685945868492126,
      "learning_rate": 1.996429210454078e-06,
      "loss": 0.1315479278564453,
      "step": 6222
    },
    {
      "epoch": 0.8671357904270884,
      "grad_norm": 0.688685953617096,
      "learning_rate": 1.9923254948442648e-06,
      "loss": 0.12176704406738281,
      "step": 6223
    },
    {
      "epoch": 0.8672751341183028,
      "grad_norm": 0.23695588111877441,
      "learning_rate": 1.98822578012861e-06,
      "loss": 0.060810089111328125,
      "step": 6224
    },
    {
      "epoch": 0.8674144778095172,
      "grad_norm": 0.5983558893203735,
      "learning_rate": 1.9841300672179662e-06,
      "loss": 0.10082054138183594,
      "step": 6225
    },
    {
      "epoch": 0.8675538215007316,
      "grad_norm": 0.6987488269805908,
      "learning_rate": 1.9800383570223157e-06,
      "loss": 0.12230491638183594,
      "step": 6226
    },
    {
      "epoch": 0.867693165191946,
      "grad_norm": 0.21722497045993805,
      "learning_rate": 1.97595065045074e-06,
      "loss": 0.06642532348632812,
      "step": 6227
    },
    {
      "epoch": 0.8678325088831603,
      "grad_norm": 0.36800140142440796,
      "learning_rate": 1.9718669484114315e-06,
      "loss": 0.07820510864257812,
      "step": 6228
    },
    {
      "epoch": 0.8679718525743747,
      "grad_norm": 0.26743701100349426,
      "learning_rate": 1.9677872518116948e-06,
      "loss": 0.07381343841552734,
      "step": 6229
    },
    {
      "epoch": 0.8681111962655891,
      "grad_norm": 0.401845782995224,
      "learning_rate": 1.963711561557955e-06,
      "loss": 0.08802032470703125,
      "step": 6230
    },
    {
      "epoch": 0.8682505399568035,
      "grad_norm": 0.47003045678138733,
      "learning_rate": 1.959639878555728e-06,
      "loss": 0.10158634185791016,
      "step": 6231
    },
    {
      "epoch": 0.8683898836480178,
      "grad_norm": 0.3978874385356903,
      "learning_rate": 1.95557220370965e-06,
      "loss": 0.08491897583007812,
      "step": 6232
    },
    {
      "epoch": 0.8685292273392322,
      "grad_norm": 0.36235830187797546,
      "learning_rate": 1.9515085379234656e-06,
      "loss": 0.07915687561035156,
      "step": 6233
    },
    {
      "epoch": 0.8686685710304466,
      "grad_norm": 0.5926169157028198,
      "learning_rate": 1.9474488821000357e-06,
      "loss": 0.08065223693847656,
      "step": 6234
    },
    {
      "epoch": 0.868807914721661,
      "grad_norm": 0.4906582534313202,
      "learning_rate": 1.943393237141311e-06,
      "loss": 0.10334110260009766,
      "step": 6235
    },
    {
      "epoch": 0.8689472584128753,
      "grad_norm": 0.25246039032936096,
      "learning_rate": 1.9393416039483724e-06,
      "loss": 0.06314659118652344,
      "step": 6236
    },
    {
      "epoch": 0.8690866021040897,
      "grad_norm": 0.5675008893013,
      "learning_rate": 1.9352939834214004e-06,
      "loss": 0.10373306274414062,
      "step": 6237
    },
    {
      "epoch": 0.8692259457953041,
      "grad_norm": 0.593263566493988,
      "learning_rate": 1.931250376459679e-06,
      "loss": 0.13410520553588867,
      "step": 6238
    },
    {
      "epoch": 0.8693652894865185,
      "grad_norm": 0.36081454157829285,
      "learning_rate": 1.9272107839616062e-06,
      "loss": 0.09426689147949219,
      "step": 6239
    },
    {
      "epoch": 0.8695046331777329,
      "grad_norm": 0.32264989614486694,
      "learning_rate": 1.923175206824688e-06,
      "loss": 0.08277511596679688,
      "step": 6240
    },
    {
      "epoch": 0.8696439768689472,
      "grad_norm": 0.4625852406024933,
      "learning_rate": 1.9191436459455406e-06,
      "loss": 0.09335708618164062,
      "step": 6241
    },
    {
      "epoch": 0.8697833205601616,
      "grad_norm": 0.6318019032478333,
      "learning_rate": 1.915116102219883e-06,
      "loss": 0.11864662170410156,
      "step": 6242
    },
    {
      "epoch": 0.869922664251376,
      "grad_norm": 0.39255353808403015,
      "learning_rate": 1.911092576542537e-06,
      "loss": 0.09418296813964844,
      "step": 6243
    },
    {
      "epoch": 0.8700620079425904,
      "grad_norm": 0.3702635169029236,
      "learning_rate": 1.9070730698074458e-06,
      "loss": 0.09469032287597656,
      "step": 6244
    },
    {
      "epoch": 0.8702013516338047,
      "grad_norm": 0.45386284589767456,
      "learning_rate": 1.9030575829076525e-06,
      "loss": 0.0948038101196289,
      "step": 6245
    },
    {
      "epoch": 0.8703406953250191,
      "grad_norm": 0.26332685351371765,
      "learning_rate": 1.8990461167353014e-06,
      "loss": 0.06401777267456055,
      "step": 6246
    },
    {
      "epoch": 0.8704800390162335,
      "grad_norm": 0.311318576335907,
      "learning_rate": 1.8950386721816549e-06,
      "loss": 0.08427047729492188,
      "step": 6247
    },
    {
      "epoch": 0.8706193827074479,
      "grad_norm": 0.40963122248649597,
      "learning_rate": 1.8910352501370677e-06,
      "loss": 0.07628822326660156,
      "step": 6248
    },
    {
      "epoch": 0.8707587263986623,
      "grad_norm": 0.2988976538181305,
      "learning_rate": 1.887035851491017e-06,
      "loss": 0.07598525285720825,
      "step": 6249
    },
    {
      "epoch": 0.8708980700898766,
      "grad_norm": 0.2402399182319641,
      "learning_rate": 1.8830404771320721e-06,
      "loss": 0.06158256530761719,
      "step": 6250
    },
    {
      "epoch": 0.871037413781091,
      "grad_norm": 0.39784547686576843,
      "learning_rate": 1.8790491279479139e-06,
      "loss": 0.08606243133544922,
      "step": 6251
    },
    {
      "epoch": 0.8711767574723055,
      "grad_norm": 0.28068819642066956,
      "learning_rate": 1.8750618048253377e-06,
      "loss": 0.08400154113769531,
      "step": 6252
    },
    {
      "epoch": 0.8713161011635199,
      "grad_norm": 0.3113216459751129,
      "learning_rate": 1.8710785086502237e-06,
      "loss": 0.06360626220703125,
      "step": 6253
    },
    {
      "epoch": 0.8714554448547343,
      "grad_norm": 0.41839560866355896,
      "learning_rate": 1.867099240307575e-06,
      "loss": 0.09745979309082031,
      "step": 6254
    },
    {
      "epoch": 0.8715947885459486,
      "grad_norm": 0.21026763319969177,
      "learning_rate": 1.8631240006814933e-06,
      "loss": 0.06408309936523438,
      "step": 6255
    },
    {
      "epoch": 0.871734132237163,
      "grad_norm": 0.5461362600326538,
      "learning_rate": 1.85915279065519e-06,
      "loss": 0.1246337890625,
      "step": 6256
    },
    {
      "epoch": 0.8718734759283774,
      "grad_norm": 0.42185425758361816,
      "learning_rate": 1.85518561111097e-06,
      "loss": 0.10392951965332031,
      "step": 6257
    },
    {
      "epoch": 0.8720128196195918,
      "grad_norm": 0.5524580478668213,
      "learning_rate": 1.85122246293026e-06,
      "loss": 0.1206512451171875,
      "step": 6258
    },
    {
      "epoch": 0.8721521633108061,
      "grad_norm": 0.5464361906051636,
      "learning_rate": 1.847263346993573e-06,
      "loss": 0.09935855865478516,
      "step": 6259
    },
    {
      "epoch": 0.8722915070020205,
      "grad_norm": 0.39103132486343384,
      "learning_rate": 1.8433082641805323e-06,
      "loss": 0.09817123413085938,
      "step": 6260
    },
    {
      "epoch": 0.8724308506932349,
      "grad_norm": 0.36702439188957214,
      "learning_rate": 1.8393572153698724e-06,
      "loss": 0.09551811218261719,
      "step": 6261
    },
    {
      "epoch": 0.8725701943844493,
      "grad_norm": 0.39921388030052185,
      "learning_rate": 1.835410201439427e-06,
      "loss": 0.08054161071777344,
      "step": 6262
    },
    {
      "epoch": 0.8727095380756636,
      "grad_norm": 0.494232714176178,
      "learning_rate": 1.83146722326613e-06,
      "loss": 0.12225341796875,
      "step": 6263
    },
    {
      "epoch": 0.872848881766878,
      "grad_norm": 0.38751542568206787,
      "learning_rate": 1.8275282817260187e-06,
      "loss": 0.08472251892089844,
      "step": 6264
    },
    {
      "epoch": 0.8729882254580924,
      "grad_norm": 0.6939036846160889,
      "learning_rate": 1.8235933776942394e-06,
      "loss": 0.12839508056640625,
      "step": 6265
    },
    {
      "epoch": 0.8731275691493068,
      "grad_norm": 0.5964553952217102,
      "learning_rate": 1.8196625120450396e-06,
      "loss": 0.0962982177734375,
      "step": 6266
    },
    {
      "epoch": 0.8732669128405212,
      "grad_norm": 0.2522365152835846,
      "learning_rate": 1.8157356856517626e-06,
      "loss": 0.06416702270507812,
      "step": 6267
    },
    {
      "epoch": 0.8734062565317355,
      "grad_norm": 0.2652980387210846,
      "learning_rate": 1.8118128993868667e-06,
      "loss": 0.07312965393066406,
      "step": 6268
    },
    {
      "epoch": 0.8735456002229499,
      "grad_norm": 0.23944240808486938,
      "learning_rate": 1.8078941541218964e-06,
      "loss": 0.0690765380859375,
      "step": 6269
    },
    {
      "epoch": 0.8736849439141643,
      "grad_norm": 0.4796037971973419,
      "learning_rate": 1.8039794507275155e-06,
      "loss": 0.10884475708007812,
      "step": 6270
    },
    {
      "epoch": 0.8738242876053787,
      "grad_norm": 0.39657866954803467,
      "learning_rate": 1.8000687900734748e-06,
      "loss": 0.08995819091796875,
      "step": 6271
    },
    {
      "epoch": 0.873963631296593,
      "grad_norm": 0.40571677684783936,
      "learning_rate": 1.796162173028637e-06,
      "loss": 0.08679962158203125,
      "step": 6272
    },
    {
      "epoch": 0.8741029749878074,
      "grad_norm": 0.32280442118644714,
      "learning_rate": 1.7922596004609682e-06,
      "loss": 0.07275772094726562,
      "step": 6273
    },
    {
      "epoch": 0.8742423186790218,
      "grad_norm": 0.4451875686645508,
      "learning_rate": 1.7883610732375278e-06,
      "loss": 0.10102462768554688,
      "step": 6274
    },
    {
      "epoch": 0.8743816623702362,
      "grad_norm": 0.4532744586467743,
      "learning_rate": 1.784466592224472e-06,
      "loss": 0.10247039794921875,
      "step": 6275
    },
    {
      "epoch": 0.8745210060614506,
      "grad_norm": 0.41182005405426025,
      "learning_rate": 1.7805761582870729e-06,
      "loss": 0.0906524658203125,
      "step": 6276
    },
    {
      "epoch": 0.8746603497526649,
      "grad_norm": 0.3910611867904663,
      "learning_rate": 1.7766897722896993e-06,
      "loss": 0.07802057266235352,
      "step": 6277
    },
    {
      "epoch": 0.8747996934438793,
      "grad_norm": 0.38633018732070923,
      "learning_rate": 1.7728074350958068e-06,
      "loss": 0.0752410888671875,
      "step": 6278
    },
    {
      "epoch": 0.8749390371350937,
      "grad_norm": 0.3701724708080292,
      "learning_rate": 1.7689291475679748e-06,
      "loss": 0.10185432434082031,
      "step": 6279
    },
    {
      "epoch": 0.8750783808263081,
      "grad_norm": 0.2864452600479126,
      "learning_rate": 1.7650549105678583e-06,
      "loss": 0.07678890228271484,
      "step": 6280
    },
    {
      "epoch": 0.8752177245175224,
      "grad_norm": 0.4438377320766449,
      "learning_rate": 1.7611847249562352e-06,
      "loss": 0.09981346130371094,
      "step": 6281
    },
    {
      "epoch": 0.8753570682087368,
      "grad_norm": 0.653642475605011,
      "learning_rate": 1.7573185915929625e-06,
      "loss": 0.08833503723144531,
      "step": 6282
    },
    {
      "epoch": 0.8754964118999512,
      "grad_norm": 0.4506444036960602,
      "learning_rate": 1.7534565113370106e-06,
      "loss": 0.09870147705078125,
      "step": 6283
    },
    {
      "epoch": 0.8756357555911656,
      "grad_norm": 0.2357621043920517,
      "learning_rate": 1.749598485046451e-06,
      "loss": 0.06714439392089844,
      "step": 6284
    },
    {
      "epoch": 0.87577509928238,
      "grad_norm": 0.3442234694957733,
      "learning_rate": 1.7457445135784423e-06,
      "loss": 0.08381462097167969,
      "step": 6285
    },
    {
      "epoch": 0.8759144429735943,
      "grad_norm": 0.4789642095565796,
      "learning_rate": 1.7418945977892488e-06,
      "loss": 0.10666275024414062,
      "step": 6286
    },
    {
      "epoch": 0.8760537866648087,
      "grad_norm": 0.36808258295059204,
      "learning_rate": 1.7380487385342371e-06,
      "loss": 0.06828880310058594,
      "step": 6287
    },
    {
      "epoch": 0.8761931303560231,
      "grad_norm": 0.4764224886894226,
      "learning_rate": 1.7342069366678705e-06,
      "loss": 0.11638832092285156,
      "step": 6288
    },
    {
      "epoch": 0.8763324740472375,
      "grad_norm": 0.5305808186531067,
      "learning_rate": 1.7303691930437062e-06,
      "loss": 0.0950927734375,
      "step": 6289
    },
    {
      "epoch": 0.8764718177384518,
      "grad_norm": 0.24285465478897095,
      "learning_rate": 1.726535508514402e-06,
      "loss": 0.06531906127929688,
      "step": 6290
    },
    {
      "epoch": 0.8766111614296662,
      "grad_norm": 0.6481537818908691,
      "learning_rate": 1.7227058839317213e-06,
      "loss": 0.12899255752563477,
      "step": 6291
    },
    {
      "epoch": 0.8767505051208806,
      "grad_norm": 0.31092897057533264,
      "learning_rate": 1.718880320146512e-06,
      "loss": 0.07834815979003906,
      "step": 6292
    },
    {
      "epoch": 0.8768898488120951,
      "grad_norm": 0.4491005539894104,
      "learning_rate": 1.7150588180087302e-06,
      "loss": 0.10754728317260742,
      "step": 6293
    },
    {
      "epoch": 0.8770291925033095,
      "grad_norm": 0.4724651873111725,
      "learning_rate": 1.7112413783674276e-06,
      "loss": 0.08931875228881836,
      "step": 6294
    },
    {
      "epoch": 0.8771685361945238,
      "grad_norm": 0.26519766449928284,
      "learning_rate": 1.7074280020707568e-06,
      "loss": 0.062496185302734375,
      "step": 6295
    },
    {
      "epoch": 0.8773078798857382,
      "grad_norm": 0.4439862370491028,
      "learning_rate": 1.7036186899659513e-06,
      "loss": 0.07197189331054688,
      "step": 6296
    },
    {
      "epoch": 0.8774472235769526,
      "grad_norm": 0.4077838063240051,
      "learning_rate": 1.6998134428993606e-06,
      "loss": 0.08937931060791016,
      "step": 6297
    },
    {
      "epoch": 0.877586567268167,
      "grad_norm": 0.252949059009552,
      "learning_rate": 1.6960122617164243e-06,
      "loss": 0.06935405731201172,
      "step": 6298
    },
    {
      "epoch": 0.8777259109593814,
      "grad_norm": 0.43165531754493713,
      "learning_rate": 1.6922151472616733e-06,
      "loss": 0.0875539779663086,
      "step": 6299
    },
    {
      "epoch": 0.8778652546505957,
      "grad_norm": 0.2921707332134247,
      "learning_rate": 1.688422100378746e-06,
      "loss": 0.07463264465332031,
      "step": 6300
    },
    {
      "epoch": 0.8780045983418101,
      "grad_norm": 0.8306127786636353,
      "learning_rate": 1.684633121910364e-06,
      "loss": 0.11566734313964844,
      "step": 6301
    },
    {
      "epoch": 0.8781439420330245,
      "grad_norm": 0.4203175902366638,
      "learning_rate": 1.6808482126983584e-06,
      "loss": 0.07790756225585938,
      "step": 6302
    },
    {
      "epoch": 0.8782832857242389,
      "grad_norm": 0.37196820974349976,
      "learning_rate": 1.6770673735836452e-06,
      "loss": 0.08359527587890625,
      "step": 6303
    },
    {
      "epoch": 0.8784226294154532,
      "grad_norm": 0.39255020022392273,
      "learning_rate": 1.6732906054062392e-06,
      "loss": 0.09268379211425781,
      "step": 6304
    },
    {
      "epoch": 0.8785619731066676,
      "grad_norm": 0.4745374023914337,
      "learning_rate": 1.6695179090052604e-06,
      "loss": 0.076812744140625,
      "step": 6305
    },
    {
      "epoch": 0.878701316797882,
      "grad_norm": 0.41142576932907104,
      "learning_rate": 1.6657492852189095e-06,
      "loss": 0.0684356689453125,
      "step": 6306
    },
    {
      "epoch": 0.8788406604890964,
      "grad_norm": 0.6419191956520081,
      "learning_rate": 1.6619847348844854e-06,
      "loss": 0.1318340301513672,
      "step": 6307
    },
    {
      "epoch": 0.8789800041803107,
      "grad_norm": 0.7184076905250549,
      "learning_rate": 1.6582242588383902e-06,
      "loss": 0.12021541595458984,
      "step": 6308
    },
    {
      "epoch": 0.8791193478715251,
      "grad_norm": 0.28911063075065613,
      "learning_rate": 1.6544678579161178e-06,
      "loss": 0.07973861694335938,
      "step": 6309
    },
    {
      "epoch": 0.8792586915627395,
      "grad_norm": 0.5950722098350525,
      "learning_rate": 1.6507155329522517e-06,
      "loss": 0.1248025894165039,
      "step": 6310
    },
    {
      "epoch": 0.8793980352539539,
      "grad_norm": 0.5022260546684265,
      "learning_rate": 1.6469672847804697e-06,
      "loss": 0.1122589111328125,
      "step": 6311
    },
    {
      "epoch": 0.8795373789451683,
      "grad_norm": 0.2946960926055908,
      "learning_rate": 1.6432231142335498e-06,
      "loss": 0.060614585876464844,
      "step": 6312
    },
    {
      "epoch": 0.8796767226363826,
      "grad_norm": 0.4212896525859833,
      "learning_rate": 1.6394830221433643e-06,
      "loss": 0.082183837890625,
      "step": 6313
    },
    {
      "epoch": 0.879816066327597,
      "grad_norm": 0.9693267941474915,
      "learning_rate": 1.635747009340871e-06,
      "loss": 0.15661239624023438,
      "step": 6314
    },
    {
      "epoch": 0.8799554100188114,
      "grad_norm": 0.4627176523208618,
      "learning_rate": 1.6320150766561283e-06,
      "loss": 0.0990142822265625,
      "step": 6315
    },
    {
      "epoch": 0.8800947537100258,
      "grad_norm": 0.2826659381389618,
      "learning_rate": 1.6282872249182923e-06,
      "loss": 0.08464241027832031,
      "step": 6316
    },
    {
      "epoch": 0.8802340974012401,
      "grad_norm": 0.3160756826400757,
      "learning_rate": 1.624563454955601e-06,
      "loss": 0.06738758087158203,
      "step": 6317
    },
    {
      "epoch": 0.8803734410924545,
      "grad_norm": 0.31561124324798584,
      "learning_rate": 1.620843767595388e-06,
      "loss": 0.07369232177734375,
      "step": 6318
    },
    {
      "epoch": 0.8805127847836689,
      "grad_norm": 0.4235861599445343,
      "learning_rate": 1.6171281636640856e-06,
      "loss": 0.10497665405273438,
      "step": 6319
    },
    {
      "epoch": 0.8806521284748833,
      "grad_norm": 0.3496689200401306,
      "learning_rate": 1.6134166439872224e-06,
      "loss": 0.07475948333740234,
      "step": 6320
    },
    {
      "epoch": 0.8807914721660977,
      "grad_norm": 0.5472114682197571,
      "learning_rate": 1.6097092093894074e-06,
      "loss": 0.09774208068847656,
      "step": 6321
    },
    {
      "epoch": 0.880930815857312,
      "grad_norm": 0.4223344326019287,
      "learning_rate": 1.606005860694344e-06,
      "loss": 0.07436847686767578,
      "step": 6322
    },
    {
      "epoch": 0.8810701595485264,
      "grad_norm": 0.293888121843338,
      "learning_rate": 1.6023065987248388e-06,
      "loss": 0.0741119384765625,
      "step": 6323
    },
    {
      "epoch": 0.8812095032397408,
      "grad_norm": 0.3115915060043335,
      "learning_rate": 1.598611424302783e-06,
      "loss": 0.07414817810058594,
      "step": 6324
    },
    {
      "epoch": 0.8813488469309552,
      "grad_norm": 0.39990267157554626,
      "learning_rate": 1.5949203382491529e-06,
      "loss": 0.08307456970214844,
      "step": 6325
    },
    {
      "epoch": 0.8814881906221695,
      "grad_norm": 0.3197343349456787,
      "learning_rate": 1.5912333413840331e-06,
      "loss": 0.07665061950683594,
      "step": 6326
    },
    {
      "epoch": 0.8816275343133839,
      "grad_norm": 0.39891931414604187,
      "learning_rate": 1.587550434526588e-06,
      "loss": 0.08481979370117188,
      "step": 6327
    },
    {
      "epoch": 0.8817668780045983,
      "grad_norm": 0.2505267262458801,
      "learning_rate": 1.5838716184950653e-06,
      "loss": 0.06880569458007812,
      "step": 6328
    },
    {
      "epoch": 0.8819062216958127,
      "grad_norm": 0.46227431297302246,
      "learning_rate": 1.5801968941068247e-06,
      "loss": 0.07881450653076172,
      "step": 6329
    },
    {
      "epoch": 0.882045565387027,
      "grad_norm": 0.3289506137371063,
      "learning_rate": 1.5765262621783062e-06,
      "loss": 0.08876895904541016,
      "step": 6330
    },
    {
      "epoch": 0.8821849090782414,
      "grad_norm": 0.4018467962741852,
      "learning_rate": 1.572859723525031e-06,
      "loss": 0.0957794189453125,
      "step": 6331
    },
    {
      "epoch": 0.8823242527694558,
      "grad_norm": 0.2207142412662506,
      "learning_rate": 1.5691972789616338e-06,
      "loss": 0.06149482727050781,
      "step": 6332
    },
    {
      "epoch": 0.8824635964606703,
      "grad_norm": 0.5938962697982788,
      "learning_rate": 1.565538929301813e-06,
      "loss": 0.08368301391601562,
      "step": 6333
    },
    {
      "epoch": 0.8826029401518847,
      "grad_norm": 0.4678293764591217,
      "learning_rate": 1.5618846753583805e-06,
      "loss": 0.08010482788085938,
      "step": 6334
    },
    {
      "epoch": 0.882742283843099,
      "grad_norm": 0.37277859449386597,
      "learning_rate": 1.558234517943218e-06,
      "loss": 0.08573055267333984,
      "step": 6335
    },
    {
      "epoch": 0.8828816275343134,
      "grad_norm": 0.3740997314453125,
      "learning_rate": 1.5545884578673165e-06,
      "loss": 0.08347320556640625,
      "step": 6336
    },
    {
      "epoch": 0.8830209712255278,
      "grad_norm": 0.5697695016860962,
      "learning_rate": 1.5509464959407438e-06,
      "loss": 0.10097694396972656,
      "step": 6337
    },
    {
      "epoch": 0.8831603149167422,
      "grad_norm": 0.4636383056640625,
      "learning_rate": 1.5473086329726638e-06,
      "loss": 0.09846973419189453,
      "step": 6338
    },
    {
      "epoch": 0.8832996586079566,
      "grad_norm": 0.4265225827693939,
      "learning_rate": 1.543674869771319e-06,
      "loss": 0.0861673355102539,
      "step": 6339
    },
    {
      "epoch": 0.8834390022991709,
      "grad_norm": 0.46911922097206116,
      "learning_rate": 1.540045207144052e-06,
      "loss": 0.09244251251220703,
      "step": 6340
    },
    {
      "epoch": 0.8835783459903853,
      "grad_norm": 0.26285481452941895,
      "learning_rate": 1.5364196458972957e-06,
      "loss": 0.07189369201660156,
      "step": 6341
    },
    {
      "epoch": 0.8837176896815997,
      "grad_norm": 0.5140864253044128,
      "learning_rate": 1.5327981868365638e-06,
      "loss": 0.10577774047851562,
      "step": 6342
    },
    {
      "epoch": 0.8838570333728141,
      "grad_norm": 0.45155325531959534,
      "learning_rate": 1.529180830766459e-06,
      "loss": 0.08824634552001953,
      "step": 6343
    },
    {
      "epoch": 0.8839963770640284,
      "grad_norm": 0.3038387894630432,
      "learning_rate": 1.5255675784906764e-06,
      "loss": 0.07589530944824219,
      "step": 6344
    },
    {
      "epoch": 0.8841357207552428,
      "grad_norm": 0.6385385990142822,
      "learning_rate": 1.5219584308120028e-06,
      "loss": 0.10736274719238281,
      "step": 6345
    },
    {
      "epoch": 0.8842750644464572,
      "grad_norm": 0.37171927094459534,
      "learning_rate": 1.518353388532301e-06,
      "loss": 0.06946468353271484,
      "step": 6346
    },
    {
      "epoch": 0.8844144081376716,
      "grad_norm": 0.317028284072876,
      "learning_rate": 1.5147524524525392e-06,
      "loss": 0.07025146484375,
      "step": 6347
    },
    {
      "epoch": 0.884553751828886,
      "grad_norm": 0.2616197168827057,
      "learning_rate": 1.511155623372751e-06,
      "loss": 0.07137203216552734,
      "step": 6348
    },
    {
      "epoch": 0.8846930955201003,
      "grad_norm": 0.40096405148506165,
      "learning_rate": 1.507562902092079e-06,
      "loss": 0.09584236145019531,
      "step": 6349
    },
    {
      "epoch": 0.8848324392113147,
      "grad_norm": 0.6763895750045776,
      "learning_rate": 1.5039742894087384e-06,
      "loss": 0.10794448852539062,
      "step": 6350
    },
    {
      "epoch": 0.8849717829025291,
      "grad_norm": 0.34371688961982727,
      "learning_rate": 1.50038978612004e-06,
      "loss": 0.07867646217346191,
      "step": 6351
    },
    {
      "epoch": 0.8851111265937435,
      "grad_norm": 0.5387987494468689,
      "learning_rate": 1.4968093930223804e-06,
      "loss": 0.09865760803222656,
      "step": 6352
    },
    {
      "epoch": 0.8852504702849578,
      "grad_norm": 0.2825624942779541,
      "learning_rate": 1.4932331109112387e-06,
      "loss": 0.07334613800048828,
      "step": 6353
    },
    {
      "epoch": 0.8853898139761722,
      "grad_norm": 0.34986135363578796,
      "learning_rate": 1.489660940581179e-06,
      "loss": 0.07717323303222656,
      "step": 6354
    },
    {
      "epoch": 0.8855291576673866,
      "grad_norm": 0.38700202107429504,
      "learning_rate": 1.4860928828258604e-06,
      "loss": 0.0963287353515625,
      "step": 6355
    },
    {
      "epoch": 0.885668501358601,
      "grad_norm": 0.35238394141197205,
      "learning_rate": 1.4825289384380282e-06,
      "loss": 0.07713699340820312,
      "step": 6356
    },
    {
      "epoch": 0.8858078450498154,
      "grad_norm": 0.20589211583137512,
      "learning_rate": 1.4789691082095004e-06,
      "loss": 0.07008171081542969,
      "step": 6357
    },
    {
      "epoch": 0.8859471887410297,
      "grad_norm": 0.5352255702018738,
      "learning_rate": 1.4754133929311975e-06,
      "loss": 0.11149215698242188,
      "step": 6358
    },
    {
      "epoch": 0.8860865324322441,
      "grad_norm": 0.4348975718021393,
      "learning_rate": 1.4718617933931146e-06,
      "loss": 0.08794784545898438,
      "step": 6359
    },
    {
      "epoch": 0.8862258761234585,
      "grad_norm": 0.291655033826828,
      "learning_rate": 1.4683143103843355e-06,
      "loss": 0.061901092529296875,
      "step": 6360
    },
    {
      "epoch": 0.8863652198146729,
      "grad_norm": 0.27522122859954834,
      "learning_rate": 1.464770944693028e-06,
      "loss": 0.07447481155395508,
      "step": 6361
    },
    {
      "epoch": 0.8865045635058872,
      "grad_norm": 0.2338055819272995,
      "learning_rate": 1.4612316971064555e-06,
      "loss": 0.055794715881347656,
      "step": 6362
    },
    {
      "epoch": 0.8866439071971016,
      "grad_norm": 0.9842904210090637,
      "learning_rate": 1.4576965684109534e-06,
      "loss": 0.14532089233398438,
      "step": 6363
    },
    {
      "epoch": 0.886783250888316,
      "grad_norm": 0.3685941696166992,
      "learning_rate": 1.4541655593919402e-06,
      "loss": 0.10296058654785156,
      "step": 6364
    },
    {
      "epoch": 0.8869225945795304,
      "grad_norm": 0.5249600410461426,
      "learning_rate": 1.4506386708339325e-06,
      "loss": 0.10117912292480469,
      "step": 6365
    },
    {
      "epoch": 0.8870619382707448,
      "grad_norm": 0.4400736391544342,
      "learning_rate": 1.4471159035205262e-06,
      "loss": 0.09916877746582031,
      "step": 6366
    },
    {
      "epoch": 0.8872012819619591,
      "grad_norm": 0.4385639429092407,
      "learning_rate": 1.4435972582343948e-06,
      "loss": 0.09350395202636719,
      "step": 6367
    },
    {
      "epoch": 0.8873406256531735,
      "grad_norm": 0.35537055134773254,
      "learning_rate": 1.4400827357573043e-06,
      "loss": 0.08539390563964844,
      "step": 6368
    },
    {
      "epoch": 0.8874799693443879,
      "grad_norm": 0.31984764337539673,
      "learning_rate": 1.4365723368700968e-06,
      "loss": 0.08342933654785156,
      "step": 6369
    },
    {
      "epoch": 0.8876193130356023,
      "grad_norm": 0.22788922488689423,
      "learning_rate": 1.4330660623527081e-06,
      "loss": 0.06122398376464844,
      "step": 6370
    },
    {
      "epoch": 0.8877586567268166,
      "grad_norm": 0.42257726192474365,
      "learning_rate": 1.4295639129841466e-06,
      "loss": 0.08043670654296875,
      "step": 6371
    },
    {
      "epoch": 0.887898000418031,
      "grad_norm": 0.44300416111946106,
      "learning_rate": 1.4260658895425162e-06,
      "loss": 0.06927108764648438,
      "step": 6372
    },
    {
      "epoch": 0.8880373441092455,
      "grad_norm": 0.32648053765296936,
      "learning_rate": 1.4225719928049953e-06,
      "loss": 0.07955360412597656,
      "step": 6373
    },
    {
      "epoch": 0.8881766878004599,
      "grad_norm": 0.4176311790943146,
      "learning_rate": 1.4190822235478496e-06,
      "loss": 0.09214019775390625,
      "step": 6374
    },
    {
      "epoch": 0.8883160314916743,
      "grad_norm": 0.28784963488578796,
      "learning_rate": 1.415596582546419e-06,
      "loss": 0.080810546875,
      "step": 6375
    },
    {
      "epoch": 0.8884553751828886,
      "grad_norm": 0.2369752675294876,
      "learning_rate": 1.4121150705751396e-06,
      "loss": 0.07083320617675781,
      "step": 6376
    },
    {
      "epoch": 0.888594718874103,
      "grad_norm": 0.2049647718667984,
      "learning_rate": 1.4086376884075282e-06,
      "loss": 0.054579734802246094,
      "step": 6377
    },
    {
      "epoch": 0.8887340625653174,
      "grad_norm": 0.30273738503456116,
      "learning_rate": 1.4051644368161688e-06,
      "loss": 0.06805419921875,
      "step": 6378
    },
    {
      "epoch": 0.8888734062565318,
      "grad_norm": 0.3864225447177887,
      "learning_rate": 1.4016953165727487e-06,
      "loss": 0.08341598510742188,
      "step": 6379
    },
    {
      "epoch": 0.8890127499477462,
      "grad_norm": 0.28347402811050415,
      "learning_rate": 1.398230328448018e-06,
      "loss": 0.07316207885742188,
      "step": 6380
    },
    {
      "epoch": 0.8891520936389605,
      "grad_norm": 0.5013516545295715,
      "learning_rate": 1.3947694732118278e-06,
      "loss": 0.08783721923828125,
      "step": 6381
    },
    {
      "epoch": 0.8892914373301749,
      "grad_norm": 0.368417352437973,
      "learning_rate": 1.3913127516330916e-06,
      "loss": 0.07451248168945312,
      "step": 6382
    },
    {
      "epoch": 0.8894307810213893,
      "grad_norm": 0.4261535406112671,
      "learning_rate": 1.3878601644798173e-06,
      "loss": 0.10359954833984375,
      "step": 6383
    },
    {
      "epoch": 0.8895701247126037,
      "grad_norm": 0.3200541138648987,
      "learning_rate": 1.3844117125190982e-06,
      "loss": 0.07808494567871094,
      "step": 6384
    },
    {
      "epoch": 0.889709468403818,
      "grad_norm": 0.3023897111415863,
      "learning_rate": 1.3809673965170923e-06,
      "loss": 0.07169914245605469,
      "step": 6385
    },
    {
      "epoch": 0.8898488120950324,
      "grad_norm": 0.7536265850067139,
      "learning_rate": 1.3775272172390497e-06,
      "loss": 0.12587738037109375,
      "step": 6386
    },
    {
      "epoch": 0.8899881557862468,
      "grad_norm": 0.43995535373687744,
      "learning_rate": 1.3740911754493014e-06,
      "loss": 0.09344100952148438,
      "step": 6387
    },
    {
      "epoch": 0.8901274994774612,
      "grad_norm": 0.3706681728363037,
      "learning_rate": 1.3706592719112588e-06,
      "loss": 0.0832977294921875,
      "step": 6388
    },
    {
      "epoch": 0.8902668431686755,
      "grad_norm": 0.33370307087898254,
      "learning_rate": 1.3672315073874098e-06,
      "loss": 0.08833885192871094,
      "step": 6389
    },
    {
      "epoch": 0.8904061868598899,
      "grad_norm": 0.5277245044708252,
      "learning_rate": 1.3638078826393296e-06,
      "loss": 0.08568191528320312,
      "step": 6390
    },
    {
      "epoch": 0.8905455305511043,
      "grad_norm": 0.3230781555175781,
      "learning_rate": 1.3603883984276656e-06,
      "loss": 0.07376670837402344,
      "step": 6391
    },
    {
      "epoch": 0.8906848742423187,
      "grad_norm": 0.25533026456832886,
      "learning_rate": 1.3569730555121452e-06,
      "loss": 0.06451034545898438,
      "step": 6392
    },
    {
      "epoch": 0.8908242179335331,
      "grad_norm": 0.29474076628685,
      "learning_rate": 1.353561854651586e-06,
      "loss": 0.07569503784179688,
      "step": 6393
    },
    {
      "epoch": 0.8909635616247474,
      "grad_norm": 0.7196462154388428,
      "learning_rate": 1.3501547966038775e-06,
      "loss": 0.10285568237304688,
      "step": 6394
    },
    {
      "epoch": 0.8911029053159618,
      "grad_norm": 0.41512438654899597,
      "learning_rate": 1.3467518821259963e-06,
      "loss": 0.09024667739868164,
      "step": 6395
    },
    {
      "epoch": 0.8912422490071762,
      "grad_norm": 0.3708645701408386,
      "learning_rate": 1.3433531119739794e-06,
      "loss": 0.08479118347167969,
      "step": 6396
    },
    {
      "epoch": 0.8913815926983906,
      "grad_norm": 0.6879599690437317,
      "learning_rate": 1.3399584869029613e-06,
      "loss": 0.1258697509765625,
      "step": 6397
    },
    {
      "epoch": 0.891520936389605,
      "grad_norm": 0.43867599964141846,
      "learning_rate": 1.3365680076671561e-06,
      "loss": 0.07941675186157227,
      "step": 6398
    },
    {
      "epoch": 0.8916602800808193,
      "grad_norm": 0.3768528699874878,
      "learning_rate": 1.3331816750198412e-06,
      "loss": 0.07877159118652344,
      "step": 6399
    },
    {
      "epoch": 0.8917996237720337,
      "grad_norm": 0.2934498190879822,
      "learning_rate": 1.3297994897133927e-06,
      "loss": 0.07696533203125,
      "step": 6400
    },
    {
      "epoch": 0.8919389674632481,
      "grad_norm": 0.23889079689979553,
      "learning_rate": 1.326421452499247e-06,
      "loss": 0.06939268112182617,
      "step": 6401
    },
    {
      "epoch": 0.8920783111544625,
      "grad_norm": 0.3615988492965698,
      "learning_rate": 1.3230475641279306e-06,
      "loss": 0.0897369384765625,
      "step": 6402
    },
    {
      "epoch": 0.8922176548456768,
      "grad_norm": 0.3492756187915802,
      "learning_rate": 1.3196778253490417e-06,
      "loss": 0.0834197998046875,
      "step": 6403
    },
    {
      "epoch": 0.8923569985368912,
      "grad_norm": 0.5397752523422241,
      "learning_rate": 1.3163122369112591e-06,
      "loss": 0.11254501342773438,
      "step": 6404
    },
    {
      "epoch": 0.8924963422281056,
      "grad_norm": 0.44906291365623474,
      "learning_rate": 1.312950799562347e-06,
      "loss": 0.08768463134765625,
      "step": 6405
    },
    {
      "epoch": 0.89263568591932,
      "grad_norm": 0.3079161047935486,
      "learning_rate": 1.3095935140491323e-06,
      "loss": 0.07628440856933594,
      "step": 6406
    },
    {
      "epoch": 0.8927750296105343,
      "grad_norm": 0.3130105435848236,
      "learning_rate": 1.3062403811175272e-06,
      "loss": 0.07228851318359375,
      "step": 6407
    },
    {
      "epoch": 0.8929143733017487,
      "grad_norm": 0.2311532348394394,
      "learning_rate": 1.302891401512525e-06,
      "loss": 0.06795310974121094,
      "step": 6408
    },
    {
      "epoch": 0.8930537169929631,
      "grad_norm": 0.5303028225898743,
      "learning_rate": 1.2995465759781944e-06,
      "loss": 0.09942054748535156,
      "step": 6409
    },
    {
      "epoch": 0.8931930606841775,
      "grad_norm": 0.3070272207260132,
      "learning_rate": 1.2962059052576703e-06,
      "loss": 0.06454277038574219,
      "step": 6410
    },
    {
      "epoch": 0.8933324043753919,
      "grad_norm": 0.4686877429485321,
      "learning_rate": 1.2928693900931856e-06,
      "loss": 0.08547210693359375,
      "step": 6411
    },
    {
      "epoch": 0.8934717480666062,
      "grad_norm": 0.3089117407798767,
      "learning_rate": 1.2895370312260247e-06,
      "loss": 0.08093643188476562,
      "step": 6412
    },
    {
      "epoch": 0.8936110917578207,
      "grad_norm": 0.3711654245853424,
      "learning_rate": 1.2862088293965736e-06,
      "loss": 0.07990455627441406,
      "step": 6413
    },
    {
      "epoch": 0.8937504354490351,
      "grad_norm": 0.376597136259079,
      "learning_rate": 1.2828847853442738e-06,
      "loss": 0.07566642761230469,
      "step": 6414
    },
    {
      "epoch": 0.8938897791402495,
      "grad_norm": 0.30630528926849365,
      "learning_rate": 1.2795648998076572e-06,
      "loss": 0.08663558959960938,
      "step": 6415
    },
    {
      "epoch": 0.8940291228314639,
      "grad_norm": 0.2340475171804428,
      "learning_rate": 1.2762491735243264e-06,
      "loss": 0.06133604049682617,
      "step": 6416
    },
    {
      "epoch": 0.8941684665226782,
      "grad_norm": 0.2934882640838623,
      "learning_rate": 1.2729376072309597e-06,
      "loss": 0.070220947265625,
      "step": 6417
    },
    {
      "epoch": 0.8943078102138926,
      "grad_norm": 0.25584396719932556,
      "learning_rate": 1.2696302016633078e-06,
      "loss": 0.07174873352050781,
      "step": 6418
    },
    {
      "epoch": 0.894447153905107,
      "grad_norm": 0.5982056260108948,
      "learning_rate": 1.266326957556201e-06,
      "loss": 0.0820159912109375,
      "step": 6419
    },
    {
      "epoch": 0.8945864975963214,
      "grad_norm": 0.480836421251297,
      "learning_rate": 1.2630278756435522e-06,
      "loss": 0.10312271118164062,
      "step": 6420
    },
    {
      "epoch": 0.8947258412875357,
      "grad_norm": 0.46907317638397217,
      "learning_rate": 1.2597329566583372e-06,
      "loss": 0.07863807678222656,
      "step": 6421
    },
    {
      "epoch": 0.8948651849787501,
      "grad_norm": 0.8326483368873596,
      "learning_rate": 1.2564422013326083e-06,
      "loss": 0.1148519515991211,
      "step": 6422
    },
    {
      "epoch": 0.8950045286699645,
      "grad_norm": 0.33113378286361694,
      "learning_rate": 1.2531556103974984e-06,
      "loss": 0.079345703125,
      "step": 6423
    },
    {
      "epoch": 0.8951438723611789,
      "grad_norm": 0.41896703839302063,
      "learning_rate": 1.249873184583219e-06,
      "loss": 0.08963680267333984,
      "step": 6424
    },
    {
      "epoch": 0.8952832160523932,
      "grad_norm": 0.31311658024787903,
      "learning_rate": 1.2465949246190422e-06,
      "loss": 0.07894611358642578,
      "step": 6425
    },
    {
      "epoch": 0.8954225597436076,
      "grad_norm": 0.4299217462539673,
      "learning_rate": 1.2433208312333255e-06,
      "loss": 0.0848855972290039,
      "step": 6426
    },
    {
      "epoch": 0.895561903434822,
      "grad_norm": 0.6932771801948547,
      "learning_rate": 1.2400509051535026e-06,
      "loss": 0.1132349967956543,
      "step": 6427
    },
    {
      "epoch": 0.8957012471260364,
      "grad_norm": 0.28347328305244446,
      "learning_rate": 1.2367851471060654e-06,
      "loss": 0.07630443572998047,
      "step": 6428
    },
    {
      "epoch": 0.8958405908172508,
      "grad_norm": 0.2832571268081665,
      "learning_rate": 1.2335235578165983e-06,
      "loss": 0.07648468017578125,
      "step": 6429
    },
    {
      "epoch": 0.8959799345084651,
      "grad_norm": 0.44226112961769104,
      "learning_rate": 1.2302661380097547e-06,
      "loss": 0.10201454162597656,
      "step": 6430
    },
    {
      "epoch": 0.8961192781996795,
      "grad_norm": 0.4169379472732544,
      "learning_rate": 1.227012888409249e-06,
      "loss": 0.08756446838378906,
      "step": 6431
    },
    {
      "epoch": 0.8962586218908939,
      "grad_norm": 0.5970325469970703,
      "learning_rate": 1.2237638097378902e-06,
      "loss": 0.10465049743652344,
      "step": 6432
    },
    {
      "epoch": 0.8963979655821083,
      "grad_norm": 0.3429519236087799,
      "learning_rate": 1.2205189027175402e-06,
      "loss": 0.08566665649414062,
      "step": 6433
    },
    {
      "epoch": 0.8965373092733226,
      "grad_norm": 0.7479395270347595,
      "learning_rate": 1.2172781680691515e-06,
      "loss": 0.10310554504394531,
      "step": 6434
    },
    {
      "epoch": 0.896676652964537,
      "grad_norm": 0.48586395382881165,
      "learning_rate": 1.2140416065127324e-06,
      "loss": 0.09621620178222656,
      "step": 6435
    },
    {
      "epoch": 0.8968159966557514,
      "grad_norm": 0.4328007102012634,
      "learning_rate": 1.2108092187673791e-06,
      "loss": 0.10470008850097656,
      "step": 6436
    },
    {
      "epoch": 0.8969553403469658,
      "grad_norm": 0.4776395857334137,
      "learning_rate": 1.2075810055512549e-06,
      "loss": 0.10046195983886719,
      "step": 6437
    },
    {
      "epoch": 0.8970946840381802,
      "grad_norm": 0.33546799421310425,
      "learning_rate": 1.2043569675815924e-06,
      "loss": 0.07385635375976562,
      "step": 6438
    },
    {
      "epoch": 0.8972340277293945,
      "grad_norm": 0.47450485825538635,
      "learning_rate": 1.201137105574699e-06,
      "loss": 0.0881948471069336,
      "step": 6439
    },
    {
      "epoch": 0.8973733714206089,
      "grad_norm": 0.7836313247680664,
      "learning_rate": 1.1979214202459532e-06,
      "loss": 0.12908363342285156,
      "step": 6440
    },
    {
      "epoch": 0.8975127151118233,
      "grad_norm": 0.5593071579933167,
      "learning_rate": 1.1947099123098126e-06,
      "loss": 0.11519432067871094,
      "step": 6441
    },
    {
      "epoch": 0.8976520588030377,
      "grad_norm": 0.41252776980400085,
      "learning_rate": 1.1915025824797976e-06,
      "loss": 0.09378814697265625,
      "step": 6442
    },
    {
      "epoch": 0.897791402494252,
      "grad_norm": 0.3600638210773468,
      "learning_rate": 1.1882994314685003e-06,
      "loss": 0.08592414855957031,
      "step": 6443
    },
    {
      "epoch": 0.8979307461854664,
      "grad_norm": 0.23847205936908722,
      "learning_rate": 1.1851004599875915e-06,
      "loss": 0.06051445007324219,
      "step": 6444
    },
    {
      "epoch": 0.8980700898766808,
      "grad_norm": 0.34454572200775146,
      "learning_rate": 1.1819056687478114e-06,
      "loss": 0.08388137817382812,
      "step": 6445
    },
    {
      "epoch": 0.8982094335678952,
      "grad_norm": 0.5135355591773987,
      "learning_rate": 1.1787150584589657e-06,
      "loss": 0.10358238220214844,
      "step": 6446
    },
    {
      "epoch": 0.8983487772591096,
      "grad_norm": 0.5596545338630676,
      "learning_rate": 1.1755286298299339e-06,
      "loss": 0.08795928955078125,
      "step": 6447
    },
    {
      "epoch": 0.8984881209503239,
      "grad_norm": 0.4202527105808258,
      "learning_rate": 1.1723463835686765e-06,
      "loss": 0.08997249603271484,
      "step": 6448
    },
    {
      "epoch": 0.8986274646415383,
      "grad_norm": 0.4047374129295349,
      "learning_rate": 1.169168320382208e-06,
      "loss": 0.09187507629394531,
      "step": 6449
    },
    {
      "epoch": 0.8987668083327527,
      "grad_norm": 0.5031136274337769,
      "learning_rate": 1.165994440976621e-06,
      "loss": 0.08526802062988281,
      "step": 6450
    },
    {
      "epoch": 0.8989061520239671,
      "grad_norm": 0.3582023084163666,
      "learning_rate": 1.162824746057083e-06,
      "loss": 0.06972312927246094,
      "step": 6451
    },
    {
      "epoch": 0.8990454957151814,
      "grad_norm": 0.38129180669784546,
      "learning_rate": 1.159659236327828e-06,
      "loss": 0.07792091369628906,
      "step": 6452
    },
    {
      "epoch": 0.8991848394063959,
      "grad_norm": 0.2468128204345703,
      "learning_rate": 1.1564979124921582e-06,
      "loss": 0.06743526458740234,
      "step": 6453
    },
    {
      "epoch": 0.8993241830976103,
      "grad_norm": 0.7499858736991882,
      "learning_rate": 1.153340775252445e-06,
      "loss": 0.11701011657714844,
      "step": 6454
    },
    {
      "epoch": 0.8994635267888247,
      "grad_norm": 0.6271414756774902,
      "learning_rate": 1.1501878253101362e-06,
      "loss": 0.12547969818115234,
      "step": 6455
    },
    {
      "epoch": 0.8996028704800391,
      "grad_norm": 0.5412514805793762,
      "learning_rate": 1.1470390633657468e-06,
      "loss": 0.09924697875976562,
      "step": 6456
    },
    {
      "epoch": 0.8997422141712534,
      "grad_norm": 0.3636331260204315,
      "learning_rate": 1.1438944901188532e-06,
      "loss": 0.091461181640625,
      "step": 6457
    },
    {
      "epoch": 0.8998815578624678,
      "grad_norm": 0.4750295877456665,
      "learning_rate": 1.1407541062681138e-06,
      "loss": 0.07520675659179688,
      "step": 6458
    },
    {
      "epoch": 0.9000209015536822,
      "grad_norm": 0.8362365961074829,
      "learning_rate": 1.1376179125112508e-06,
      "loss": 0.13773345947265625,
      "step": 6459
    },
    {
      "epoch": 0.9001602452448966,
      "grad_norm": 0.5754382610321045,
      "learning_rate": 1.1344859095450468e-06,
      "loss": 0.09871292114257812,
      "step": 6460
    },
    {
      "epoch": 0.900299588936111,
      "grad_norm": 0.712861180305481,
      "learning_rate": 1.1313580980653671e-06,
      "loss": 0.10795021057128906,
      "step": 6461
    },
    {
      "epoch": 0.9004389326273253,
      "grad_norm": 0.6808586716651917,
      "learning_rate": 1.1282344787671429e-06,
      "loss": 0.12317085266113281,
      "step": 6462
    },
    {
      "epoch": 0.9005782763185397,
      "grad_norm": 0.312374085187912,
      "learning_rate": 1.1251150523443676e-06,
      "loss": 0.07724952697753906,
      "step": 6463
    },
    {
      "epoch": 0.9007176200097541,
      "grad_norm": 0.316883385181427,
      "learning_rate": 1.121999819490105e-06,
      "loss": 0.09416007995605469,
      "step": 6464
    },
    {
      "epoch": 0.9008569637009685,
      "grad_norm": 0.23273472487926483,
      "learning_rate": 1.1188887808964878e-06,
      "loss": 0.05534648895263672,
      "step": 6465
    },
    {
      "epoch": 0.9009963073921828,
      "grad_norm": 0.4550366997718811,
      "learning_rate": 1.1157819372547252e-06,
      "loss": 0.09100723266601562,
      "step": 6466
    },
    {
      "epoch": 0.9011356510833972,
      "grad_norm": 0.43512532114982605,
      "learning_rate": 1.1126792892550786e-06,
      "loss": 0.07414722442626953,
      "step": 6467
    },
    {
      "epoch": 0.9012749947746116,
      "grad_norm": 0.5720616579055786,
      "learning_rate": 1.10958083758689e-06,
      "loss": 0.11523056030273438,
      "step": 6468
    },
    {
      "epoch": 0.901414338465826,
      "grad_norm": 0.3631727695465088,
      "learning_rate": 1.1064865829385662e-06,
      "loss": 0.08643436431884766,
      "step": 6469
    },
    {
      "epoch": 0.9015536821570403,
      "grad_norm": 0.3908413052558899,
      "learning_rate": 1.1033965259975777e-06,
      "loss": 0.07838821411132812,
      "step": 6470
    },
    {
      "epoch": 0.9016930258482547,
      "grad_norm": 0.3303401470184326,
      "learning_rate": 1.1003106674504592e-06,
      "loss": 0.09259796142578125,
      "step": 6471
    },
    {
      "epoch": 0.9018323695394691,
      "grad_norm": 0.4576283395290375,
      "learning_rate": 1.0972290079828252e-06,
      "loss": 0.08506393432617188,
      "step": 6472
    },
    {
      "epoch": 0.9019717132306835,
      "grad_norm": 0.2140074372291565,
      "learning_rate": 1.094151548279352e-06,
      "loss": 0.0700082778930664,
      "step": 6473
    },
    {
      "epoch": 0.9021110569218979,
      "grad_norm": 0.2155413031578064,
      "learning_rate": 1.0910782890237747e-06,
      "loss": 0.05397987365722656,
      "step": 6474
    },
    {
      "epoch": 0.9022504006131122,
      "grad_norm": 0.5524152517318726,
      "learning_rate": 1.088009230898901e-06,
      "loss": 0.1094512939453125,
      "step": 6475
    },
    {
      "epoch": 0.9023897443043266,
      "grad_norm": 0.39625462889671326,
      "learning_rate": 1.0849443745866095e-06,
      "loss": 0.09475898742675781,
      "step": 6476
    },
    {
      "epoch": 0.902529087995541,
      "grad_norm": 0.408661425113678,
      "learning_rate": 1.0818837207678423e-06,
      "loss": 0.08189964294433594,
      "step": 6477
    },
    {
      "epoch": 0.9026684316867554,
      "grad_norm": 0.49468615651130676,
      "learning_rate": 1.0788272701226e-06,
      "loss": 0.10625457763671875,
      "step": 6478
    },
    {
      "epoch": 0.9028077753779697,
      "grad_norm": 0.39627891778945923,
      "learning_rate": 1.0757750233299658e-06,
      "loss": 0.09277153015136719,
      "step": 6479
    },
    {
      "epoch": 0.9029471190691841,
      "grad_norm": 0.24497127532958984,
      "learning_rate": 1.0727269810680685e-06,
      "loss": 0.07392692565917969,
      "step": 6480
    },
    {
      "epoch": 0.9030864627603985,
      "grad_norm": 0.3497641682624817,
      "learning_rate": 1.0696831440141242e-06,
      "loss": 0.06962203979492188,
      "step": 6481
    },
    {
      "epoch": 0.9032258064516129,
      "grad_norm": 0.5436011552810669,
      "learning_rate": 1.0666435128443942e-06,
      "loss": 0.11316108703613281,
      "step": 6482
    },
    {
      "epoch": 0.9033651501428273,
      "grad_norm": 0.3385253846645355,
      "learning_rate": 1.0636080882342203e-06,
      "loss": 0.07769584655761719,
      "step": 6483
    },
    {
      "epoch": 0.9035044938340416,
      "grad_norm": 0.5331652760505676,
      "learning_rate": 1.0605768708580078e-06,
      "loss": 0.11466217041015625,
      "step": 6484
    },
    {
      "epoch": 0.903643837525256,
      "grad_norm": 0.3398861885070801,
      "learning_rate": 1.0575498613892199e-06,
      "loss": 0.0869140625,
      "step": 6485
    },
    {
      "epoch": 0.9037831812164704,
      "grad_norm": 0.6810885667800903,
      "learning_rate": 1.0545270605003855e-06,
      "loss": 0.11503410339355469,
      "step": 6486
    },
    {
      "epoch": 0.9039225249076848,
      "grad_norm": 0.4192269444465637,
      "learning_rate": 1.0515084688631071e-06,
      "loss": 0.10396385192871094,
      "step": 6487
    },
    {
      "epoch": 0.9040618685988991,
      "grad_norm": 0.38606685400009155,
      "learning_rate": 1.0484940871480464e-06,
      "loss": 0.07985305786132812,
      "step": 6488
    },
    {
      "epoch": 0.9042012122901135,
      "grad_norm": 0.32324695587158203,
      "learning_rate": 1.0454839160249274e-06,
      "loss": 0.08354759216308594,
      "step": 6489
    },
    {
      "epoch": 0.9043405559813279,
      "grad_norm": 0.5439515709877014,
      "learning_rate": 1.0424779561625465e-06,
      "loss": 0.11631011962890625,
      "step": 6490
    },
    {
      "epoch": 0.9044798996725423,
      "grad_norm": 0.38752853870391846,
      "learning_rate": 1.0394762082287557e-06,
      "loss": 0.0877838134765625,
      "step": 6491
    },
    {
      "epoch": 0.9046192433637567,
      "grad_norm": 0.275966078042984,
      "learning_rate": 1.036478672890473e-06,
      "loss": 0.07407951354980469,
      "step": 6492
    },
    {
      "epoch": 0.904758587054971,
      "grad_norm": 0.40706509351730347,
      "learning_rate": 1.0334853508136854e-06,
      "loss": 0.08558464050292969,
      "step": 6493
    },
    {
      "epoch": 0.9048979307461855,
      "grad_norm": 0.48518505692481995,
      "learning_rate": 1.030496242663439e-06,
      "loss": 0.08606147766113281,
      "step": 6494
    },
    {
      "epoch": 0.9050372744373999,
      "grad_norm": 0.26346778869628906,
      "learning_rate": 1.0275113491038512e-06,
      "loss": 0.058875083923339844,
      "step": 6495
    },
    {
      "epoch": 0.9051766181286143,
      "grad_norm": 0.4266607463359833,
      "learning_rate": 1.0245306707980873e-06,
      "loss": 0.08748149871826172,
      "step": 6496
    },
    {
      "epoch": 0.9053159618198287,
      "grad_norm": 0.33037081360816956,
      "learning_rate": 1.021554208408393e-06,
      "loss": 0.06938838958740234,
      "step": 6497
    },
    {
      "epoch": 0.905455305511043,
      "grad_norm": 0.5838533043861389,
      "learning_rate": 1.018581962596068e-06,
      "loss": 0.12118339538574219,
      "step": 6498
    },
    {
      "epoch": 0.9055946492022574,
      "grad_norm": 0.3080866038799286,
      "learning_rate": 1.0156139340214755e-06,
      "loss": 0.08479690551757812,
      "step": 6499
    },
    {
      "epoch": 0.9057339928934718,
      "grad_norm": 0.3781090974807739,
      "learning_rate": 1.01265012334405e-06,
      "loss": 0.0875406265258789,
      "step": 6500
    },
    {
      "epoch": 0.9058733365846862,
      "grad_norm": 0.6796485781669617,
      "learning_rate": 1.009690531222276e-06,
      "loss": 0.15683364868164062,
      "step": 6501
    },
    {
      "epoch": 0.9060126802759005,
      "grad_norm": 0.2567310035228729,
      "learning_rate": 1.0067351583137119e-06,
      "loss": 0.07499122619628906,
      "step": 6502
    },
    {
      "epoch": 0.9061520239671149,
      "grad_norm": 0.2932274043560028,
      "learning_rate": 1.0037840052749682e-06,
      "loss": 0.07372474670410156,
      "step": 6503
    },
    {
      "epoch": 0.9062913676583293,
      "grad_norm": 0.2446686029434204,
      "learning_rate": 1.000837072761729e-06,
      "loss": 0.06362724304199219,
      "step": 6504
    },
    {
      "epoch": 0.9064307113495437,
      "grad_norm": 0.33956632018089294,
      "learning_rate": 9.978943614287374e-07,
      "loss": 0.07677841186523438,
      "step": 6505
    },
    {
      "epoch": 0.906570055040758,
      "grad_norm": 0.3824218511581421,
      "learning_rate": 9.949558719297924e-07,
      "loss": 0.07726478576660156,
      "step": 6506
    },
    {
      "epoch": 0.9067093987319724,
      "grad_norm": 0.37389788031578064,
      "learning_rate": 9.920216049177566e-07,
      "loss": 0.0780935287475586,
      "step": 6507
    },
    {
      "epoch": 0.9068487424231868,
      "grad_norm": 0.27759623527526855,
      "learning_rate": 9.890915610445617e-07,
      "loss": 0.07737159729003906,
      "step": 6508
    },
    {
      "epoch": 0.9069880861144012,
      "grad_norm": 0.5292997360229492,
      "learning_rate": 9.861657409611958e-07,
      "loss": 0.08637237548828125,
      "step": 6509
    },
    {
      "epoch": 0.9071274298056156,
      "grad_norm": 0.4656522572040558,
      "learning_rate": 9.832441453177099e-07,
      "loss": 0.09535694122314453,
      "step": 6510
    },
    {
      "epoch": 0.9072667734968299,
      "grad_norm": 0.22645217180252075,
      "learning_rate": 9.803267747632162e-07,
      "loss": 0.055855751037597656,
      "step": 6511
    },
    {
      "epoch": 0.9074061171880443,
      "grad_norm": 0.20605604350566864,
      "learning_rate": 9.774136299458825e-07,
      "loss": 0.06304645538330078,
      "step": 6512
    },
    {
      "epoch": 0.9075454608792587,
      "grad_norm": 0.2640606760978699,
      "learning_rate": 9.745047115129513e-07,
      "loss": 0.077880859375,
      "step": 6513
    },
    {
      "epoch": 0.9076848045704731,
      "grad_norm": 0.3148421049118042,
      "learning_rate": 9.716000201107102e-07,
      "loss": 0.07644271850585938,
      "step": 6514
    },
    {
      "epoch": 0.9078241482616874,
      "grad_norm": 0.5422933101654053,
      "learning_rate": 9.686995563845204e-07,
      "loss": 0.09826850891113281,
      "step": 6515
    },
    {
      "epoch": 0.9079634919529018,
      "grad_norm": 0.7735921740531921,
      "learning_rate": 9.658033209787998e-07,
      "loss": 0.11243438720703125,
      "step": 6516
    },
    {
      "epoch": 0.9081028356441162,
      "grad_norm": 0.32186219096183777,
      "learning_rate": 9.629113145370228e-07,
      "loss": 0.08183002471923828,
      "step": 6517
    },
    {
      "epoch": 0.9082421793353306,
      "grad_norm": 0.6657729744911194,
      "learning_rate": 9.60023537701724e-07,
      "loss": 0.13293075561523438,
      "step": 6518
    },
    {
      "epoch": 0.908381523026545,
      "grad_norm": 0.20923347771167755,
      "learning_rate": 9.57139991114504e-07,
      "loss": 0.0635833740234375,
      "step": 6519
    },
    {
      "epoch": 0.9085208667177593,
      "grad_norm": 0.3430497348308563,
      "learning_rate": 9.542606754160277e-07,
      "loss": 0.0715036392211914,
      "step": 6520
    },
    {
      "epoch": 0.9086602104089737,
      "grad_norm": 0.37958282232284546,
      "learning_rate": 9.513855912460057e-07,
      "loss": 0.08574342727661133,
      "step": 6521
    },
    {
      "epoch": 0.9087995541001881,
      "grad_norm": 0.45350345969200134,
      "learning_rate": 9.485147392432159e-07,
      "loss": 0.08557701110839844,
      "step": 6522
    },
    {
      "epoch": 0.9089388977914025,
      "grad_norm": 0.4867132902145386,
      "learning_rate": 9.456481200454992e-07,
      "loss": 0.0965728759765625,
      "step": 6523
    },
    {
      "epoch": 0.9090782414826168,
      "grad_norm": 0.4705277383327484,
      "learning_rate": 9.427857342897528e-07,
      "loss": 0.09188461303710938,
      "step": 6524
    },
    {
      "epoch": 0.9092175851738312,
      "grad_norm": 0.26971274614334106,
      "learning_rate": 9.399275826119325e-07,
      "loss": 0.07405662536621094,
      "step": 6525
    },
    {
      "epoch": 0.9093569288650456,
      "grad_norm": 0.32735133171081543,
      "learning_rate": 9.370736656470525e-07,
      "loss": 0.08984375,
      "step": 6526
    },
    {
      "epoch": 0.90949627255626,
      "grad_norm": 0.24433533847332,
      "learning_rate": 9.342239840291967e-07,
      "loss": 0.061412811279296875,
      "step": 6527
    },
    {
      "epoch": 0.9096356162474744,
      "grad_norm": 0.2714102566242218,
      "learning_rate": 9.313785383914852e-07,
      "loss": 0.07314586639404297,
      "step": 6528
    },
    {
      "epoch": 0.9097749599386887,
      "grad_norm": 0.3460380434989929,
      "learning_rate": 9.285373293661193e-07,
      "loss": 0.07423973083496094,
      "step": 6529
    },
    {
      "epoch": 0.9099143036299031,
      "grad_norm": 0.47802403569221497,
      "learning_rate": 9.257003575843537e-07,
      "loss": 0.09429073333740234,
      "step": 6530
    },
    {
      "epoch": 0.9100536473211175,
      "grad_norm": 0.6600710153579712,
      "learning_rate": 9.228676236764911e-07,
      "loss": 0.11326217651367188,
      "step": 6531
    },
    {
      "epoch": 0.9101929910123319,
      "grad_norm": 0.304179847240448,
      "learning_rate": 9.200391282719079e-07,
      "loss": 0.07192230224609375,
      "step": 6532
    },
    {
      "epoch": 0.9103323347035462,
      "grad_norm": 0.2895108163356781,
      "learning_rate": 9.172148719990237e-07,
      "loss": 0.07944869995117188,
      "step": 6533
    },
    {
      "epoch": 0.9104716783947607,
      "grad_norm": 0.4352632761001587,
      "learning_rate": 9.143948554853299e-07,
      "loss": 0.09868431091308594,
      "step": 6534
    },
    {
      "epoch": 0.9106110220859751,
      "grad_norm": 0.5349122285842896,
      "learning_rate": 9.115790793573653e-07,
      "loss": 0.108612060546875,
      "step": 6535
    },
    {
      "epoch": 0.9107503657771895,
      "grad_norm": 0.30581212043762207,
      "learning_rate": 9.08767544240734e-07,
      "loss": 0.08356475830078125,
      "step": 6536
    },
    {
      "epoch": 0.9108897094684039,
      "grad_norm": 0.6149179339408875,
      "learning_rate": 9.059602507600962e-07,
      "loss": 0.10852622985839844,
      "step": 6537
    },
    {
      "epoch": 0.9110290531596182,
      "grad_norm": 0.40292200446128845,
      "learning_rate": 9.031571995391664e-07,
      "loss": 0.09441184997558594,
      "step": 6538
    },
    {
      "epoch": 0.9111683968508326,
      "grad_norm": 0.4705089032649994,
      "learning_rate": 9.003583912007152e-07,
      "loss": 0.105224609375,
      "step": 6539
    },
    {
      "epoch": 0.911307740542047,
      "grad_norm": 0.3757724165916443,
      "learning_rate": 8.975638263665787e-07,
      "loss": 0.09456825256347656,
      "step": 6540
    },
    {
      "epoch": 0.9114470842332614,
      "grad_norm": 0.3695238530635834,
      "learning_rate": 8.947735056576468e-07,
      "loss": 0.09360694885253906,
      "step": 6541
    },
    {
      "epoch": 0.9115864279244758,
      "grad_norm": 0.4511104226112366,
      "learning_rate": 8.919874296938569e-07,
      "loss": 0.0967559814453125,
      "step": 6542
    },
    {
      "epoch": 0.9117257716156901,
      "grad_norm": 0.2884587347507477,
      "learning_rate": 8.892055990942228e-07,
      "loss": 0.07970237731933594,
      "step": 6543
    },
    {
      "epoch": 0.9118651153069045,
      "grad_norm": 0.3478052020072937,
      "learning_rate": 8.864280144767923e-07,
      "loss": 0.07880020141601562,
      "step": 6544
    },
    {
      "epoch": 0.9120044589981189,
      "grad_norm": 0.4460470378398895,
      "learning_rate": 8.836546764586895e-07,
      "loss": 0.07903194427490234,
      "step": 6545
    },
    {
      "epoch": 0.9121438026893333,
      "grad_norm": 0.4114014804363251,
      "learning_rate": 8.808855856560838e-07,
      "loss": 0.10237884521484375,
      "step": 6546
    },
    {
      "epoch": 0.9122831463805476,
      "grad_norm": 0.2861233651638031,
      "learning_rate": 8.781207426842031e-07,
      "loss": 0.06640911102294922,
      "step": 6547
    },
    {
      "epoch": 0.912422490071762,
      "grad_norm": 0.4342323839664459,
      "learning_rate": 8.753601481573337e-07,
      "loss": 0.0813751220703125,
      "step": 6548
    },
    {
      "epoch": 0.9125618337629764,
      "grad_norm": 0.40865620970726013,
      "learning_rate": 8.726038026888184e-07,
      "loss": 0.09746932983398438,
      "step": 6549
    },
    {
      "epoch": 0.9127011774541908,
      "grad_norm": 0.2707406282424927,
      "learning_rate": 8.698517068910495e-07,
      "loss": 0.07116508483886719,
      "step": 6550
    },
    {
      "epoch": 0.9128405211454051,
      "grad_norm": 0.5059632658958435,
      "learning_rate": 8.671038613754846e-07,
      "loss": 0.10668182373046875,
      "step": 6551
    },
    {
      "epoch": 0.9129798648366195,
      "grad_norm": 0.4219966232776642,
      "learning_rate": 8.643602667526307e-07,
      "loss": 0.0897216796875,
      "step": 6552
    },
    {
      "epoch": 0.9131192085278339,
      "grad_norm": 0.5468851923942566,
      "learning_rate": 8.616209236320517e-07,
      "loss": 0.08156299591064453,
      "step": 6553
    },
    {
      "epoch": 0.9132585522190483,
      "grad_norm": 0.28750577569007874,
      "learning_rate": 8.588858326223625e-07,
      "loss": 0.05863666534423828,
      "step": 6554
    },
    {
      "epoch": 0.9133978959102627,
      "grad_norm": 0.5559219121932983,
      "learning_rate": 8.561549943312442e-07,
      "loss": 0.09984970092773438,
      "step": 6555
    },
    {
      "epoch": 0.913537239601477,
      "grad_norm": 0.3423694968223572,
      "learning_rate": 8.534284093654288e-07,
      "loss": 0.08151626586914062,
      "step": 6556
    },
    {
      "epoch": 0.9136765832926914,
      "grad_norm": 0.32039275765419006,
      "learning_rate": 8.507060783306919e-07,
      "loss": 0.07769775390625,
      "step": 6557
    },
    {
      "epoch": 0.9138159269839058,
      "grad_norm": 0.20852826535701752,
      "learning_rate": 8.479880018318831e-07,
      "loss": 0.058948516845703125,
      "step": 6558
    },
    {
      "epoch": 0.9139552706751202,
      "grad_norm": 0.360268235206604,
      "learning_rate": 8.452741804728904e-07,
      "loss": 0.08536529541015625,
      "step": 6559
    },
    {
      "epoch": 0.9140946143663345,
      "grad_norm": 0.42852532863616943,
      "learning_rate": 8.425646148566624e-07,
      "loss": 0.09309864044189453,
      "step": 6560
    },
    {
      "epoch": 0.9142339580575489,
      "grad_norm": 0.603644609451294,
      "learning_rate": 8.398593055852067e-07,
      "loss": 0.09510612487792969,
      "step": 6561
    },
    {
      "epoch": 0.9143733017487633,
      "grad_norm": 0.23036816716194153,
      "learning_rate": 8.371582532595823e-07,
      "loss": 0.06501007080078125,
      "step": 6562
    },
    {
      "epoch": 0.9145126454399777,
      "grad_norm": 0.6221429109573364,
      "learning_rate": 8.34461458479896e-07,
      "loss": 0.0954437255859375,
      "step": 6563
    },
    {
      "epoch": 0.9146519891311921,
      "grad_norm": 0.7127260565757751,
      "learning_rate": 8.317689218453196e-07,
      "loss": 0.10004425048828125,
      "step": 6564
    },
    {
      "epoch": 0.9147913328224064,
      "grad_norm": 0.33584344387054443,
      "learning_rate": 8.290806439540677e-07,
      "loss": 0.08370208740234375,
      "step": 6565
    },
    {
      "epoch": 0.9149306765136208,
      "grad_norm": 0.2507660984992981,
      "learning_rate": 8.263966254034206e-07,
      "loss": 0.06017112731933594,
      "step": 6566
    },
    {
      "epoch": 0.9150700202048352,
      "grad_norm": 0.3566182553768158,
      "learning_rate": 8.237168667897011e-07,
      "loss": 0.08274078369140625,
      "step": 6567
    },
    {
      "epoch": 0.9152093638960496,
      "grad_norm": 0.2487594038248062,
      "learning_rate": 8.210413687082885e-07,
      "loss": 0.06760787963867188,
      "step": 6568
    },
    {
      "epoch": 0.9153487075872639,
      "grad_norm": 0.213240385055542,
      "learning_rate": 8.183701317536253e-07,
      "loss": 0.05800342559814453,
      "step": 6569
    },
    {
      "epoch": 0.9154880512784783,
      "grad_norm": 0.2985682487487793,
      "learning_rate": 8.157031565191941e-07,
      "loss": 0.07796287536621094,
      "step": 6570
    },
    {
      "epoch": 0.9156273949696927,
      "grad_norm": 0.6000349521636963,
      "learning_rate": 8.130404435975348e-07,
      "loss": 0.09581947326660156,
      "step": 6571
    },
    {
      "epoch": 0.9157667386609071,
      "grad_norm": 0.3141961097717285,
      "learning_rate": 8.103819935802426e-07,
      "loss": 0.08697319030761719,
      "step": 6572
    },
    {
      "epoch": 0.9159060823521215,
      "grad_norm": 0.7443259358406067,
      "learning_rate": 8.077278070579675e-07,
      "loss": 0.1261310577392578,
      "step": 6573
    },
    {
      "epoch": 0.9160454260433359,
      "grad_norm": 0.42007237672805786,
      "learning_rate": 8.050778846204066e-07,
      "loss": 0.0881195068359375,
      "step": 6574
    },
    {
      "epoch": 0.9161847697345503,
      "grad_norm": 0.8759194612503052,
      "learning_rate": 8.024322268563112e-07,
      "loss": 0.17675399780273438,
      "step": 6575
    },
    {
      "epoch": 0.9163241134257647,
      "grad_norm": 0.6751231551170349,
      "learning_rate": 7.997908343534844e-07,
      "loss": 0.089111328125,
      "step": 6576
    },
    {
      "epoch": 0.9164634571169791,
      "grad_norm": 0.8199658989906311,
      "learning_rate": 7.971537076987901e-07,
      "loss": 0.09866905212402344,
      "step": 6577
    },
    {
      "epoch": 0.9166028008081935,
      "grad_norm": 0.29679831862449646,
      "learning_rate": 7.945208474781307e-07,
      "loss": 0.0630502700805664,
      "step": 6578
    },
    {
      "epoch": 0.9167421444994078,
      "grad_norm": 0.40699079632759094,
      "learning_rate": 7.918922542764717e-07,
      "loss": 0.09285354614257812,
      "step": 6579
    },
    {
      "epoch": 0.9168814881906222,
      "grad_norm": 0.3590869605541229,
      "learning_rate": 7.89267928677826e-07,
      "loss": 0.08063507080078125,
      "step": 6580
    },
    {
      "epoch": 0.9170208318818366,
      "grad_norm": 0.5593793988227844,
      "learning_rate": 7.866478712652581e-07,
      "loss": 0.1119527816772461,
      "step": 6581
    },
    {
      "epoch": 0.917160175573051,
      "grad_norm": 0.6665065288543701,
      "learning_rate": 7.840320826208825e-07,
      "loss": 0.1184549331665039,
      "step": 6582
    },
    {
      "epoch": 0.9172995192642653,
      "grad_norm": 0.27801838517189026,
      "learning_rate": 7.81420563325872e-07,
      "loss": 0.06329822540283203,
      "step": 6583
    },
    {
      "epoch": 0.9174388629554797,
      "grad_norm": 0.3614061772823334,
      "learning_rate": 7.788133139604448e-07,
      "loss": 0.08771896362304688,
      "step": 6584
    },
    {
      "epoch": 0.9175782066466941,
      "grad_norm": 0.2301931530237198,
      "learning_rate": 7.76210335103873e-07,
      "loss": 0.06080055236816406,
      "step": 6585
    },
    {
      "epoch": 0.9177175503379085,
      "grad_norm": 0.37533503770828247,
      "learning_rate": 7.736116273344763e-07,
      "loss": 0.07808113098144531,
      "step": 6586
    },
    {
      "epoch": 0.9178568940291228,
      "grad_norm": 0.45541253685951233,
      "learning_rate": 7.710171912296305e-07,
      "loss": 0.08830451965332031,
      "step": 6587
    },
    {
      "epoch": 0.9179962377203372,
      "grad_norm": 0.5162289142608643,
      "learning_rate": 7.684270273657612e-07,
      "loss": 0.09355926513671875,
      "step": 6588
    },
    {
      "epoch": 0.9181355814115516,
      "grad_norm": 0.5937817692756653,
      "learning_rate": 7.658411363183393e-07,
      "loss": 0.1232309341430664,
      "step": 6589
    },
    {
      "epoch": 0.918274925102766,
      "grad_norm": 0.4314119517803192,
      "learning_rate": 7.632595186618963e-07,
      "loss": 0.09695911407470703,
      "step": 6590
    },
    {
      "epoch": 0.9184142687939804,
      "grad_norm": 0.5070757865905762,
      "learning_rate": 7.606821749700045e-07,
      "loss": 0.08881378173828125,
      "step": 6591
    },
    {
      "epoch": 0.9185536124851947,
      "grad_norm": 0.2695625126361847,
      "learning_rate": 7.581091058152923e-07,
      "loss": 0.06934928894042969,
      "step": 6592
    },
    {
      "epoch": 0.9186929561764091,
      "grad_norm": 0.34121841192245483,
      "learning_rate": 7.55540311769436e-07,
      "loss": 0.08825492858886719,
      "step": 6593
    },
    {
      "epoch": 0.9188322998676235,
      "grad_norm": 0.37000909447669983,
      "learning_rate": 7.529757934031634e-07,
      "loss": 0.0848236083984375,
      "step": 6594
    },
    {
      "epoch": 0.9189716435588379,
      "grad_norm": 0.7696034908294678,
      "learning_rate": 7.504155512862587e-07,
      "loss": 0.11981010437011719,
      "step": 6595
    },
    {
      "epoch": 0.9191109872500522,
      "grad_norm": 0.33245086669921875,
      "learning_rate": 7.478595859875381e-07,
      "loss": 0.08704757690429688,
      "step": 6596
    },
    {
      "epoch": 0.9192503309412666,
      "grad_norm": 0.4866340756416321,
      "learning_rate": 7.453078980748829e-07,
      "loss": 0.09759283065795898,
      "step": 6597
    },
    {
      "epoch": 0.919389674632481,
      "grad_norm": 0.3704664409160614,
      "learning_rate": 7.427604881152239e-07,
      "loss": 0.08462715148925781,
      "step": 6598
    },
    {
      "epoch": 0.9195290183236954,
      "grad_norm": 0.7151444554328918,
      "learning_rate": 7.402173566745308e-07,
      "loss": 0.12769508361816406,
      "step": 6599
    },
    {
      "epoch": 0.9196683620149098,
      "grad_norm": 0.31409314274787903,
      "learning_rate": 7.376785043178314e-07,
      "loss": 0.06514549255371094,
      "step": 6600
    },
    {
      "epoch": 0.9198077057061241,
      "grad_norm": 0.34350916743278503,
      "learning_rate": 7.35143931609208e-07,
      "loss": 0.07901954650878906,
      "step": 6601
    },
    {
      "epoch": 0.9199470493973385,
      "grad_norm": 0.2665935456752777,
      "learning_rate": 7.326136391117766e-07,
      "loss": 0.07143783569335938,
      "step": 6602
    },
    {
      "epoch": 0.9200863930885529,
      "grad_norm": 0.3795723617076874,
      "learning_rate": 7.3008762738771e-07,
      "loss": 0.0830841064453125,
      "step": 6603
    },
    {
      "epoch": 0.9202257367797673,
      "grad_norm": 0.2776455879211426,
      "learning_rate": 7.275658969982324e-07,
      "loss": 0.07346248626708984,
      "step": 6604
    },
    {
      "epoch": 0.9203650804709816,
      "grad_norm": 0.4205428659915924,
      "learning_rate": 7.250484485036202e-07,
      "loss": 0.09348106384277344,
      "step": 6605
    },
    {
      "epoch": 0.920504424162196,
      "grad_norm": 0.2679228186607361,
      "learning_rate": 7.225352824631859e-07,
      "loss": 0.06442070007324219,
      "step": 6606
    },
    {
      "epoch": 0.9206437678534104,
      "grad_norm": 0.42654886841773987,
      "learning_rate": 7.200263994352985e-07,
      "loss": 0.08454322814941406,
      "step": 6607
    },
    {
      "epoch": 0.9207831115446248,
      "grad_norm": 0.36222341656684875,
      "learning_rate": 7.175217999773765e-07,
      "loss": 0.0687551498413086,
      "step": 6608
    },
    {
      "epoch": 0.9209224552358392,
      "grad_norm": 0.4494735896587372,
      "learning_rate": 7.150214846458859e-07,
      "loss": 0.07894515991210938,
      "step": 6609
    },
    {
      "epoch": 0.9210617989270535,
      "grad_norm": 0.5438541173934937,
      "learning_rate": 7.125254539963356e-07,
      "loss": 0.10592079162597656,
      "step": 6610
    },
    {
      "epoch": 0.9212011426182679,
      "grad_norm": 0.17529365420341492,
      "learning_rate": 7.100337085832909e-07,
      "loss": 0.05667686462402344,
      "step": 6611
    },
    {
      "epoch": 0.9213404863094823,
      "grad_norm": 0.3781813979148865,
      "learning_rate": 7.075462489603557e-07,
      "loss": 0.08823299407958984,
      "step": 6612
    },
    {
      "epoch": 0.9214798300006967,
      "grad_norm": 0.37499719858169556,
      "learning_rate": 7.050630756801924e-07,
      "loss": 0.08816289901733398,
      "step": 6613
    },
    {
      "epoch": 0.9216191736919112,
      "grad_norm": 0.34485799074172974,
      "learning_rate": 7.025841892945018e-07,
      "loss": 0.07259178161621094,
      "step": 6614
    },
    {
      "epoch": 0.9217585173831255,
      "grad_norm": 0.18768461048603058,
      "learning_rate": 7.001095903540345e-07,
      "loss": 0.05909156799316406,
      "step": 6615
    },
    {
      "epoch": 0.9218978610743399,
      "grad_norm": 0.5603020191192627,
      "learning_rate": 6.976392794085973e-07,
      "loss": 0.11426544189453125,
      "step": 6616
    },
    {
      "epoch": 0.9220372047655543,
      "grad_norm": 0.3104931712150574,
      "learning_rate": 6.951732570070291e-07,
      "loss": 0.08048248291015625,
      "step": 6617
    },
    {
      "epoch": 0.9221765484567687,
      "grad_norm": 0.5593743324279785,
      "learning_rate": 6.92711523697227e-07,
      "loss": 0.10905838012695312,
      "step": 6618
    },
    {
      "epoch": 0.922315892147983,
      "grad_norm": 0.5056818127632141,
      "learning_rate": 6.902540800261292e-07,
      "loss": 0.1168522834777832,
      "step": 6619
    },
    {
      "epoch": 0.9224552358391974,
      "grad_norm": 0.34235528111457825,
      "learning_rate": 6.878009265397301e-07,
      "loss": 0.07930946350097656,
      "step": 6620
    },
    {
      "epoch": 0.9225945795304118,
      "grad_norm": 0.5320193767547607,
      "learning_rate": 6.853520637830557e-07,
      "loss": 0.11658096313476562,
      "step": 6621
    },
    {
      "epoch": 0.9227339232216262,
      "grad_norm": 0.6496424674987793,
      "learning_rate": 6.829074923001955e-07,
      "loss": 0.12418460845947266,
      "step": 6622
    },
    {
      "epoch": 0.9228732669128406,
      "grad_norm": 0.385430246591568,
      "learning_rate": 6.804672126342704e-07,
      "loss": 0.09539985656738281,
      "step": 6623
    },
    {
      "epoch": 0.9230126106040549,
      "grad_norm": 0.3858891427516937,
      "learning_rate": 6.780312253274624e-07,
      "loss": 0.08365726470947266,
      "step": 6624
    },
    {
      "epoch": 0.9231519542952693,
      "grad_norm": 0.41163238883018494,
      "learning_rate": 6.755995309209851e-07,
      "loss": 0.09324836730957031,
      "step": 6625
    },
    {
      "epoch": 0.9232912979864837,
      "grad_norm": 0.2914195656776428,
      "learning_rate": 6.731721299551064e-07,
      "loss": 0.06652259826660156,
      "step": 6626
    },
    {
      "epoch": 0.9234306416776981,
      "grad_norm": 0.3604721426963806,
      "learning_rate": 6.707490229691483e-07,
      "loss": 0.0698084831237793,
      "step": 6627
    },
    {
      "epoch": 0.9235699853689124,
      "grad_norm": 0.3797692060470581,
      "learning_rate": 6.683302105014577e-07,
      "loss": 0.08304214477539062,
      "step": 6628
    },
    {
      "epoch": 0.9237093290601268,
      "grad_norm": 0.4249309003353119,
      "learning_rate": 6.65915693089445e-07,
      "loss": 0.09428596496582031,
      "step": 6629
    },
    {
      "epoch": 0.9238486727513412,
      "grad_norm": 0.2175358086824417,
      "learning_rate": 6.635054712695632e-07,
      "loss": 0.05502510070800781,
      "step": 6630
    },
    {
      "epoch": 0.9239880164425556,
      "grad_norm": 0.32141923904418945,
      "learning_rate": 6.610995455773017e-07,
      "loss": 0.0782012939453125,
      "step": 6631
    },
    {
      "epoch": 0.92412736013377,
      "grad_norm": 0.5654904842376709,
      "learning_rate": 6.586979165472107e-07,
      "loss": 0.11055564880371094,
      "step": 6632
    },
    {
      "epoch": 0.9242667038249843,
      "grad_norm": 0.5399138331413269,
      "learning_rate": 6.563005847128701e-07,
      "loss": 0.10218429565429688,
      "step": 6633
    },
    {
      "epoch": 0.9244060475161987,
      "grad_norm": 0.35986974835395813,
      "learning_rate": 6.539075506069181e-07,
      "loss": 0.08703899383544922,
      "step": 6634
    },
    {
      "epoch": 0.9245453912074131,
      "grad_norm": 0.4902538061141968,
      "learning_rate": 6.515188147610274e-07,
      "loss": 0.10908699035644531,
      "step": 6635
    },
    {
      "epoch": 0.9246847348986275,
      "grad_norm": 0.47024884819984436,
      "learning_rate": 6.491343777059245e-07,
      "loss": 0.11682510375976562,
      "step": 6636
    },
    {
      "epoch": 0.9248240785898418,
      "grad_norm": 0.3190925419330597,
      "learning_rate": 6.467542399713744e-07,
      "loss": 0.08687400817871094,
      "step": 6637
    },
    {
      "epoch": 0.9249634222810562,
      "grad_norm": 0.37971243262290955,
      "learning_rate": 6.443784020861921e-07,
      "loss": 0.08101654052734375,
      "step": 6638
    },
    {
      "epoch": 0.9251027659722706,
      "grad_norm": 0.5773796439170837,
      "learning_rate": 6.420068645782285e-07,
      "loss": 0.08248615264892578,
      "step": 6639
    },
    {
      "epoch": 0.925242109663485,
      "grad_norm": 0.49288609623908997,
      "learning_rate": 6.396396279743911e-07,
      "loss": 0.10174751281738281,
      "step": 6640
    },
    {
      "epoch": 0.9253814533546993,
      "grad_norm": 0.33727696537971497,
      "learning_rate": 6.372766928006236e-07,
      "loss": 0.07999420166015625,
      "step": 6641
    },
    {
      "epoch": 0.9255207970459137,
      "grad_norm": 0.5195273160934448,
      "learning_rate": 6.349180595819171e-07,
      "loss": 0.08247184753417969,
      "step": 6642
    },
    {
      "epoch": 0.9256601407371281,
      "grad_norm": 0.2792443633079529,
      "learning_rate": 6.325637288423059e-07,
      "loss": 0.08041191101074219,
      "step": 6643
    },
    {
      "epoch": 0.9257994844283425,
      "grad_norm": 0.344217449426651,
      "learning_rate": 6.302137011048648e-07,
      "loss": 0.08887672424316406,
      "step": 6644
    },
    {
      "epoch": 0.9259388281195569,
      "grad_norm": 0.2827705442905426,
      "learning_rate": 6.278679768917229e-07,
      "loss": 0.06989479064941406,
      "step": 6645
    },
    {
      "epoch": 0.9260781718107712,
      "grad_norm": 0.38775885105133057,
      "learning_rate": 6.25526556724041e-07,
      "loss": 0.08710765838623047,
      "step": 6646
    },
    {
      "epoch": 0.9262175155019856,
      "grad_norm": 0.5543617010116577,
      "learning_rate": 6.231894411220319e-07,
      "loss": 0.10077857971191406,
      "step": 6647
    },
    {
      "epoch": 0.9263568591932,
      "grad_norm": 0.5051101446151733,
      "learning_rate": 6.208566306049513e-07,
      "loss": 0.09556198120117188,
      "step": 6648
    },
    {
      "epoch": 0.9264962028844144,
      "grad_norm": 0.3866461515426636,
      "learning_rate": 6.185281256910936e-07,
      "loss": 0.09577178955078125,
      "step": 6649
    },
    {
      "epoch": 0.9266355465756287,
      "grad_norm": 0.21760503947734833,
      "learning_rate": 6.162039268977981e-07,
      "loss": 0.059696197509765625,
      "step": 6650
    },
    {
      "epoch": 0.9267748902668431,
      "grad_norm": 0.562396764755249,
      "learning_rate": 6.138840347414498e-07,
      "loss": 0.10908699035644531,
      "step": 6651
    },
    {
      "epoch": 0.9269142339580575,
      "grad_norm": 0.3615497648715973,
      "learning_rate": 6.115684497374786e-07,
      "loss": 0.0806131362915039,
      "step": 6652
    },
    {
      "epoch": 0.9270535776492719,
      "grad_norm": 0.44118282198905945,
      "learning_rate": 6.092571724003527e-07,
      "loss": 0.08130264282226562,
      "step": 6653
    },
    {
      "epoch": 0.9271929213404864,
      "grad_norm": 0.395177960395813,
      "learning_rate": 6.069502032435814e-07,
      "loss": 0.08608627319335938,
      "step": 6654
    },
    {
      "epoch": 0.9273322650317007,
      "grad_norm": 0.25216373801231384,
      "learning_rate": 6.04647542779726e-07,
      "loss": 0.07448005676269531,
      "step": 6655
    },
    {
      "epoch": 0.9274716087229151,
      "grad_norm": 0.6371353268623352,
      "learning_rate": 6.023491915203839e-07,
      "loss": 0.10519790649414062,
      "step": 6656
    },
    {
      "epoch": 0.9276109524141295,
      "grad_norm": 0.3372279405593872,
      "learning_rate": 6.000551499761953e-07,
      "loss": 0.08106422424316406,
      "step": 6657
    },
    {
      "epoch": 0.9277502961053439,
      "grad_norm": 0.6367672681808472,
      "learning_rate": 5.977654186568438e-07,
      "loss": 0.11925888061523438,
      "step": 6658
    },
    {
      "epoch": 0.9278896397965583,
      "grad_norm": 0.3356008529663086,
      "learning_rate": 5.954799980710579e-07,
      "loss": 0.06897258758544922,
      "step": 6659
    },
    {
      "epoch": 0.9280289834877726,
      "grad_norm": 0.3450528383255005,
      "learning_rate": 5.931988887266005e-07,
      "loss": 0.07631921768188477,
      "step": 6660
    },
    {
      "epoch": 0.928168327178987,
      "grad_norm": 0.5043894052505493,
      "learning_rate": 5.90922091130286e-07,
      "loss": 0.09424209594726562,
      "step": 6661
    },
    {
      "epoch": 0.9283076708702014,
      "grad_norm": 0.4736936092376709,
      "learning_rate": 5.886496057879676e-07,
      "loss": 0.10711097717285156,
      "step": 6662
    },
    {
      "epoch": 0.9284470145614158,
      "grad_norm": 0.5441661477088928,
      "learning_rate": 5.863814332045347e-07,
      "loss": 0.09080123901367188,
      "step": 6663
    },
    {
      "epoch": 0.9285863582526301,
      "grad_norm": 0.44081375002861023,
      "learning_rate": 5.841175738839311e-07,
      "loss": 0.1002349853515625,
      "step": 6664
    },
    {
      "epoch": 0.9287257019438445,
      "grad_norm": 0.48935723304748535,
      "learning_rate": 5.818580283291276e-07,
      "loss": 0.09604644775390625,
      "step": 6665
    },
    {
      "epoch": 0.9288650456350589,
      "grad_norm": 0.31389591097831726,
      "learning_rate": 5.796027970421492e-07,
      "loss": 0.0885152816772461,
      "step": 6666
    },
    {
      "epoch": 0.9290043893262733,
      "grad_norm": 0.34090209007263184,
      "learning_rate": 5.773518805240508e-07,
      "loss": 0.08242607116699219,
      "step": 6667
    },
    {
      "epoch": 0.9291437330174876,
      "grad_norm": 0.3995712697505951,
      "learning_rate": 5.75105279274939e-07,
      "loss": 0.07183647155761719,
      "step": 6668
    },
    {
      "epoch": 0.929283076708702,
      "grad_norm": 0.32118910551071167,
      "learning_rate": 5.728629937939568e-07,
      "loss": 0.07657337188720703,
      "step": 6669
    },
    {
      "epoch": 0.9294224203999164,
      "grad_norm": 0.52314293384552,
      "learning_rate": 5.706250245792899e-07,
      "loss": 0.07912826538085938,
      "step": 6670
    },
    {
      "epoch": 0.9295617640911308,
      "grad_norm": 0.3489578366279602,
      "learning_rate": 5.683913721281586e-07,
      "loss": 0.08083343505859375,
      "step": 6671
    },
    {
      "epoch": 0.9297011077823452,
      "grad_norm": 0.46099725365638733,
      "learning_rate": 5.661620369368348e-07,
      "loss": 0.09641265869140625,
      "step": 6672
    },
    {
      "epoch": 0.9298404514735595,
      "grad_norm": 0.4630115330219269,
      "learning_rate": 5.639370195006266e-07,
      "loss": 0.08951950073242188,
      "step": 6673
    },
    {
      "epoch": 0.9299797951647739,
      "grad_norm": 0.6780856251716614,
      "learning_rate": 5.617163203138765e-07,
      "loss": 0.11866950988769531,
      "step": 6674
    },
    {
      "epoch": 0.9301191388559883,
      "grad_norm": 0.516190230846405,
      "learning_rate": 5.594999398699785e-07,
      "loss": 0.08120203018188477,
      "step": 6675
    },
    {
      "epoch": 0.9302584825472027,
      "grad_norm": 0.3080327808856964,
      "learning_rate": 5.572878786613589e-07,
      "loss": 0.06950187683105469,
      "step": 6676
    },
    {
      "epoch": 0.930397826238417,
      "grad_norm": 0.5138849020004272,
      "learning_rate": 5.55080137179489e-07,
      "loss": 0.10778236389160156,
      "step": 6677
    },
    {
      "epoch": 0.9305371699296314,
      "grad_norm": 0.3800298869609833,
      "learning_rate": 5.528767159148762e-07,
      "loss": 0.0932769775390625,
      "step": 6678
    },
    {
      "epoch": 0.9306765136208458,
      "grad_norm": 0.3422374725341797,
      "learning_rate": 5.506776153570714e-07,
      "loss": 0.07755470275878906,
      "step": 6679
    },
    {
      "epoch": 0.9308158573120602,
      "grad_norm": 0.3924712538719177,
      "learning_rate": 5.484828359946682e-07,
      "loss": 0.09271240234375,
      "step": 6680
    },
    {
      "epoch": 0.9309552010032746,
      "grad_norm": 0.2326449304819107,
      "learning_rate": 5.462923783152918e-07,
      "loss": 0.06902503967285156,
      "step": 6681
    },
    {
      "epoch": 0.9310945446944889,
      "grad_norm": 0.28155168890953064,
      "learning_rate": 5.44106242805611e-07,
      "loss": 0.07817268371582031,
      "step": 6682
    },
    {
      "epoch": 0.9312338883857033,
      "grad_norm": 0.23045183718204498,
      "learning_rate": 5.419244299513371e-07,
      "loss": 0.06684207916259766,
      "step": 6683
    },
    {
      "epoch": 0.9313732320769177,
      "grad_norm": 0.25650131702423096,
      "learning_rate": 5.397469402372224e-07,
      "loss": 0.06693553924560547,
      "step": 6684
    },
    {
      "epoch": 0.9315125757681321,
      "grad_norm": 0.4884472191333771,
      "learning_rate": 5.375737741470511e-07,
      "loss": 0.09341812133789062,
      "step": 6685
    },
    {
      "epoch": 0.9316519194593464,
      "grad_norm": 0.19392023980617523,
      "learning_rate": 5.354049321636523e-07,
      "loss": 0.0544891357421875,
      "step": 6686
    },
    {
      "epoch": 0.9317912631505608,
      "grad_norm": 0.5219374299049377,
      "learning_rate": 5.332404147688919e-07,
      "loss": 0.08742904663085938,
      "step": 6687
    },
    {
      "epoch": 0.9319306068417752,
      "grad_norm": 0.37558475136756897,
      "learning_rate": 5.310802224436806e-07,
      "loss": 0.09094047546386719,
      "step": 6688
    },
    {
      "epoch": 0.9320699505329896,
      "grad_norm": 0.5372804999351501,
      "learning_rate": 5.28924355667959e-07,
      "loss": 0.088897705078125,
      "step": 6689
    },
    {
      "epoch": 0.932209294224204,
      "grad_norm": 0.608688235282898,
      "learning_rate": 5.267728149207152e-07,
      "loss": 0.12084197998046875,
      "step": 6690
    },
    {
      "epoch": 0.9323486379154183,
      "grad_norm": 0.5047572255134583,
      "learning_rate": 5.246256006799689e-07,
      "loss": 0.10957527160644531,
      "step": 6691
    },
    {
      "epoch": 0.9324879816066327,
      "grad_norm": 0.33723902702331543,
      "learning_rate": 5.22482713422785e-07,
      "loss": 0.07510757446289062,
      "step": 6692
    },
    {
      "epoch": 0.9326273252978471,
      "grad_norm": 0.3445902168750763,
      "learning_rate": 5.203441536252607e-07,
      "loss": 0.0868844985961914,
      "step": 6693
    },
    {
      "epoch": 0.9327666689890615,
      "grad_norm": 0.4786708950996399,
      "learning_rate": 5.182099217625381e-07,
      "loss": 0.10385608673095703,
      "step": 6694
    },
    {
      "epoch": 0.932906012680276,
      "grad_norm": 0.4463580846786499,
      "learning_rate": 5.160800183087955e-07,
      "loss": 0.07473182678222656,
      "step": 6695
    },
    {
      "epoch": 0.9330453563714903,
      "grad_norm": 0.2888081669807434,
      "learning_rate": 5.139544437372457e-07,
      "loss": 0.06249237060546875,
      "step": 6696
    },
    {
      "epoch": 0.9331847000627047,
      "grad_norm": 0.3051902949810028,
      "learning_rate": 5.118331985201441e-07,
      "loss": 0.08365345001220703,
      "step": 6697
    },
    {
      "epoch": 0.9333240437539191,
      "grad_norm": 0.3894302546977997,
      "learning_rate": 5.097162831287872e-07,
      "loss": 0.08506298065185547,
      "step": 6698
    },
    {
      "epoch": 0.9334633874451335,
      "grad_norm": 0.680584728717804,
      "learning_rate": 5.076036980334964e-07,
      "loss": 0.08860015869140625,
      "step": 6699
    },
    {
      "epoch": 0.9336027311363478,
      "grad_norm": 0.28569626808166504,
      "learning_rate": 5.054954437036475e-07,
      "loss": 0.0767364501953125,
      "step": 6700
    },
    {
      "epoch": 0.9337420748275622,
      "grad_norm": 0.35863956809043884,
      "learning_rate": 5.033915206076456e-07,
      "loss": 0.0759124755859375,
      "step": 6701
    },
    {
      "epoch": 0.9338814185187766,
      "grad_norm": 0.4638131558895111,
      "learning_rate": 5.012919292129303e-07,
      "loss": 0.09745597839355469,
      "step": 6702
    },
    {
      "epoch": 0.934020762209991,
      "grad_norm": 0.4212966561317444,
      "learning_rate": 4.991966699859863e-07,
      "loss": 0.09281730651855469,
      "step": 6703
    },
    {
      "epoch": 0.9341601059012054,
      "grad_norm": 0.5626142024993896,
      "learning_rate": 4.971057433923298e-07,
      "loss": 0.09919357299804688,
      "step": 6704
    },
    {
      "epoch": 0.9342994495924197,
      "grad_norm": 0.43106403946876526,
      "learning_rate": 4.950191498965207e-07,
      "loss": 0.07739830017089844,
      "step": 6705
    },
    {
      "epoch": 0.9344387932836341,
      "grad_norm": 0.4531414806842804,
      "learning_rate": 4.929368899621479e-07,
      "loss": 0.08102798461914062,
      "step": 6706
    },
    {
      "epoch": 0.9345781369748485,
      "grad_norm": 0.3744693100452423,
      "learning_rate": 4.908589640518458e-07,
      "loss": 0.09240055084228516,
      "step": 6707
    },
    {
      "epoch": 0.9347174806660629,
      "grad_norm": 0.7656886577606201,
      "learning_rate": 4.887853726272785e-07,
      "loss": 0.13350677490234375,
      "step": 6708
    },
    {
      "epoch": 0.9348568243572772,
      "grad_norm": 0.376514196395874,
      "learning_rate": 4.867161161491551e-07,
      "loss": 0.07762527465820312,
      "step": 6709
    },
    {
      "epoch": 0.9349961680484916,
      "grad_norm": 0.2048104703426361,
      "learning_rate": 4.846511950772148e-07,
      "loss": 0.06313323974609375,
      "step": 6710
    },
    {
      "epoch": 0.935135511739706,
      "grad_norm": 0.39675194025039673,
      "learning_rate": 4.825906098702348e-07,
      "loss": 0.09535980224609375,
      "step": 6711
    },
    {
      "epoch": 0.9352748554309204,
      "grad_norm": 0.37547704577445984,
      "learning_rate": 4.805343609860314e-07,
      "loss": 0.09307098388671875,
      "step": 6712
    },
    {
      "epoch": 0.9354141991221347,
      "grad_norm": 0.32007312774658203,
      "learning_rate": 4.784824488814588e-07,
      "loss": 0.07244300842285156,
      "step": 6713
    },
    {
      "epoch": 0.9355535428133491,
      "grad_norm": 0.3219676613807678,
      "learning_rate": 4.7643487401239917e-07,
      "loss": 0.09319114685058594,
      "step": 6714
    },
    {
      "epoch": 0.9356928865045635,
      "grad_norm": 0.6558754444122314,
      "learning_rate": 4.7439163683377975e-07,
      "loss": 0.11772727966308594,
      "step": 6715
    },
    {
      "epoch": 0.9358322301957779,
      "grad_norm": 0.7904849052429199,
      "learning_rate": 4.723527377995618e-07,
      "loss": 0.13175582885742188,
      "step": 6716
    },
    {
      "epoch": 0.9359715738869923,
      "grad_norm": 0.5510526895523071,
      "learning_rate": 4.7031817736274297e-07,
      "loss": 0.09984874725341797,
      "step": 6717
    },
    {
      "epoch": 0.9361109175782066,
      "grad_norm": 0.26091140508651733,
      "learning_rate": 4.68287955975355e-07,
      "loss": 0.06839561462402344,
      "step": 6718
    },
    {
      "epoch": 0.936250261269421,
      "grad_norm": 0.24571335315704346,
      "learning_rate": 4.662620740884638e-07,
      "loss": 0.06809425354003906,
      "step": 6719
    },
    {
      "epoch": 0.9363896049606354,
      "grad_norm": 0.8018134236335754,
      "learning_rate": 4.642405321521803e-07,
      "loss": 0.11262702941894531,
      "step": 6720
    },
    {
      "epoch": 0.9365289486518498,
      "grad_norm": 0.6890111565589905,
      "learning_rate": 4.622233306156387e-07,
      "loss": 0.09411048889160156,
      "step": 6721
    },
    {
      "epoch": 0.9366682923430641,
      "grad_norm": 0.3756306767463684,
      "learning_rate": 4.6021046992702046e-07,
      "loss": 0.07724571228027344,
      "step": 6722
    },
    {
      "epoch": 0.9368076360342785,
      "grad_norm": 0.3384687602519989,
      "learning_rate": 4.5820195053353445e-07,
      "loss": 0.07736015319824219,
      "step": 6723
    },
    {
      "epoch": 0.9369469797254929,
      "grad_norm": 0.22445981204509735,
      "learning_rate": 4.561977728814282e-07,
      "loss": 0.05788135528564453,
      "step": 6724
    },
    {
      "epoch": 0.9370863234167073,
      "grad_norm": 0.49907374382019043,
      "learning_rate": 4.541979374159833e-07,
      "loss": 0.10427284240722656,
      "step": 6725
    },
    {
      "epoch": 0.9372256671079217,
      "grad_norm": 0.27394819259643555,
      "learning_rate": 4.522024445815176e-07,
      "loss": 0.06667280197143555,
      "step": 6726
    },
    {
      "epoch": 0.937365010799136,
      "grad_norm": 0.361652672290802,
      "learning_rate": 4.502112948213899e-07,
      "loss": 0.07003211975097656,
      "step": 6727
    },
    {
      "epoch": 0.9375043544903504,
      "grad_norm": 0.31202757358551025,
      "learning_rate": 4.482244885779774e-07,
      "loss": 0.08358573913574219,
      "step": 6728
    },
    {
      "epoch": 0.9376436981815648,
      "grad_norm": 0.47156238555908203,
      "learning_rate": 4.462420262927114e-07,
      "loss": 0.10611915588378906,
      "step": 6729
    },
    {
      "epoch": 0.9377830418727792,
      "grad_norm": 0.23631687462329865,
      "learning_rate": 4.442639084060463e-07,
      "loss": 0.05667877197265625,
      "step": 6730
    },
    {
      "epoch": 0.9379223855639935,
      "grad_norm": 0.38064736127853394,
      "learning_rate": 4.422901353574771e-07,
      "loss": 0.08551788330078125,
      "step": 6731
    },
    {
      "epoch": 0.9380617292552079,
      "grad_norm": 0.5650538802146912,
      "learning_rate": 4.403207075855265e-07,
      "loss": 0.0947718620300293,
      "step": 6732
    },
    {
      "epoch": 0.9382010729464223,
      "grad_norm": 0.4957156479358673,
      "learning_rate": 4.3835562552776434e-07,
      "loss": 0.09737992286682129,
      "step": 6733
    },
    {
      "epoch": 0.9383404166376367,
      "grad_norm": 0.6571094989776611,
      "learning_rate": 4.3639488962077923e-07,
      "loss": 0.11236286163330078,
      "step": 6734
    },
    {
      "epoch": 0.9384797603288512,
      "grad_norm": 0.3780471682548523,
      "learning_rate": 4.3443850030020497e-07,
      "loss": 0.07454299926757812,
      "step": 6735
    },
    {
      "epoch": 0.9386191040200655,
      "grad_norm": 0.479097843170166,
      "learning_rate": 4.324864580007071e-07,
      "loss": 0.10063743591308594,
      "step": 6736
    },
    {
      "epoch": 0.9387584477112799,
      "grad_norm": 0.5277379155158997,
      "learning_rate": 4.305387631559854e-07,
      "loss": 0.09780406951904297,
      "step": 6737
    },
    {
      "epoch": 0.9388977914024943,
      "grad_norm": 0.3923250734806061,
      "learning_rate": 4.2859541619877154e-07,
      "loss": 0.07928085327148438,
      "step": 6738
    },
    {
      "epoch": 0.9390371350937087,
      "grad_norm": 0.3509894609451294,
      "learning_rate": 4.2665641756083344e-07,
      "loss": 0.09309768676757812,
      "step": 6739
    },
    {
      "epoch": 0.939176478784923,
      "grad_norm": 0.5545886158943176,
      "learning_rate": 4.247217676729709e-07,
      "loss": 0.11981391906738281,
      "step": 6740
    },
    {
      "epoch": 0.9393158224761374,
      "grad_norm": 0.2218812257051468,
      "learning_rate": 4.227914669650224e-07,
      "loss": 0.06036949157714844,
      "step": 6741
    },
    {
      "epoch": 0.9394551661673518,
      "grad_norm": 0.7205094695091248,
      "learning_rate": 4.2086551586585144e-07,
      "loss": 0.13799285888671875,
      "step": 6742
    },
    {
      "epoch": 0.9395945098585662,
      "grad_norm": 0.5799086689949036,
      "learning_rate": 4.1894391480336694e-07,
      "loss": 0.10459136962890625,
      "step": 6743
    },
    {
      "epoch": 0.9397338535497806,
      "grad_norm": 0.17685124278068542,
      "learning_rate": 4.1702666420450064e-07,
      "loss": 0.05365562438964844,
      "step": 6744
    },
    {
      "epoch": 0.9398731972409949,
      "grad_norm": 0.3525027334690094,
      "learning_rate": 4.15113764495223e-07,
      "loss": 0.08859920501708984,
      "step": 6745
    },
    {
      "epoch": 0.9400125409322093,
      "grad_norm": 0.5000352263450623,
      "learning_rate": 4.1320521610053624e-07,
      "loss": 0.09018325805664062,
      "step": 6746
    },
    {
      "epoch": 0.9401518846234237,
      "grad_norm": 0.6123389005661011,
      "learning_rate": 4.113010194444744e-07,
      "loss": 0.12257766723632812,
      "step": 6747
    },
    {
      "epoch": 0.9402912283146381,
      "grad_norm": 0.29779455065727234,
      "learning_rate": 4.094011749501103e-07,
      "loss": 0.06926155090332031,
      "step": 6748
    },
    {
      "epoch": 0.9404305720058524,
      "grad_norm": 0.248356431722641,
      "learning_rate": 4.075056830395441e-07,
      "loss": 0.0585484504699707,
      "step": 6749
    },
    {
      "epoch": 0.9405699156970668,
      "grad_norm": 0.31627774238586426,
      "learning_rate": 4.056145441339099e-07,
      "loss": 0.07529830932617188,
      "step": 6750
    },
    {
      "epoch": 0.9407092593882812,
      "grad_norm": 0.3172943592071533,
      "learning_rate": 4.037277586533761e-07,
      "loss": 0.0702676773071289,
      "step": 6751
    },
    {
      "epoch": 0.9408486030794956,
      "grad_norm": 0.25905534625053406,
      "learning_rate": 4.018453270171474e-07,
      "loss": 0.06768226623535156,
      "step": 6752
    },
    {
      "epoch": 0.94098794677071,
      "grad_norm": 0.2098427563905716,
      "learning_rate": 3.9996724964344924e-07,
      "loss": 0.050777435302734375,
      "step": 6753
    },
    {
      "epoch": 0.9411272904619243,
      "grad_norm": 0.7403552532196045,
      "learning_rate": 3.980935269495545e-07,
      "loss": 0.11927223205566406,
      "step": 6754
    },
    {
      "epoch": 0.9412666341531387,
      "grad_norm": 0.5926846265792847,
      "learning_rate": 3.9622415935175683e-07,
      "loss": 0.11301231384277344,
      "step": 6755
    },
    {
      "epoch": 0.9414059778443531,
      "grad_norm": 0.23571304976940155,
      "learning_rate": 3.943591472653929e-07,
      "loss": 0.06699562072753906,
      "step": 6756
    },
    {
      "epoch": 0.9415453215355675,
      "grad_norm": 0.33697178959846497,
      "learning_rate": 3.92498491104818e-07,
      "loss": 0.06890869140625,
      "step": 6757
    },
    {
      "epoch": 0.9416846652267818,
      "grad_norm": 0.286998987197876,
      "learning_rate": 3.906421912834324e-07,
      "loss": 0.08141326904296875,
      "step": 6758
    },
    {
      "epoch": 0.9418240089179962,
      "grad_norm": 0.5283071994781494,
      "learning_rate": 3.887902482136663e-07,
      "loss": 0.09610557556152344,
      "step": 6759
    },
    {
      "epoch": 0.9419633526092106,
      "grad_norm": 0.4568100869655609,
      "learning_rate": 3.8694266230697053e-07,
      "loss": 0.08820056915283203,
      "step": 6760
    },
    {
      "epoch": 0.942102696300425,
      "grad_norm": 0.2876872420310974,
      "learning_rate": 3.850994339738434e-07,
      "loss": 0.07981109619140625,
      "step": 6761
    },
    {
      "epoch": 0.9422420399916394,
      "grad_norm": 0.5309542417526245,
      "learning_rate": 3.8326056362380846e-07,
      "loss": 0.0925445556640625,
      "step": 6762
    },
    {
      "epoch": 0.9423813836828537,
      "grad_norm": 0.34627702832221985,
      "learning_rate": 3.814260516654145e-07,
      "loss": 0.07609367370605469,
      "step": 6763
    },
    {
      "epoch": 0.9425207273740681,
      "grad_norm": 0.4071417450904846,
      "learning_rate": 3.795958985062553e-07,
      "loss": 0.07770729064941406,
      "step": 6764
    },
    {
      "epoch": 0.9426600710652825,
      "grad_norm": 0.5057757496833801,
      "learning_rate": 3.777701045529436e-07,
      "loss": 0.08545494079589844,
      "step": 6765
    },
    {
      "epoch": 0.9427994147564969,
      "grad_norm": 0.30591559410095215,
      "learning_rate": 3.759486702111348e-07,
      "loss": 0.07672119140625,
      "step": 6766
    },
    {
      "epoch": 0.9429387584477112,
      "grad_norm": 0.2544185221195221,
      "learning_rate": 3.7413159588550295e-07,
      "loss": 0.0711965560913086,
      "step": 6767
    },
    {
      "epoch": 0.9430781021389256,
      "grad_norm": 0.4555186629295349,
      "learning_rate": 3.723188819797652e-07,
      "loss": 0.08280754089355469,
      "step": 6768
    },
    {
      "epoch": 0.94321744583014,
      "grad_norm": 0.40748047828674316,
      "learning_rate": 3.7051052889666596e-07,
      "loss": 0.08355522155761719,
      "step": 6769
    },
    {
      "epoch": 0.9433567895213544,
      "grad_norm": 0.19781652092933655,
      "learning_rate": 3.6870653703797943e-07,
      "loss": 0.06423664093017578,
      "step": 6770
    },
    {
      "epoch": 0.9434961332125688,
      "grad_norm": 0.29825934767723083,
      "learning_rate": 3.6690690680450723e-07,
      "loss": 0.07646465301513672,
      "step": 6771
    },
    {
      "epoch": 0.9436354769037831,
      "grad_norm": 0.46692174673080444,
      "learning_rate": 3.6511163859608957e-07,
      "loss": 0.083953857421875,
      "step": 6772
    },
    {
      "epoch": 0.9437748205949975,
      "grad_norm": 0.3504297137260437,
      "learning_rate": 3.6332073281159394e-07,
      "loss": 0.05820512771606445,
      "step": 6773
    },
    {
      "epoch": 0.9439141642862119,
      "grad_norm": 0.3720932900905609,
      "learning_rate": 3.6153418984891996e-07,
      "loss": 0.08486652374267578,
      "step": 6774
    },
    {
      "epoch": 0.9440535079774264,
      "grad_norm": 0.24168632924556732,
      "learning_rate": 3.597520101049945e-07,
      "loss": 0.0670480728149414,
      "step": 6775
    },
    {
      "epoch": 0.9441928516686408,
      "grad_norm": 0.43814966082572937,
      "learning_rate": 3.579741939757764e-07,
      "loss": 0.08017730712890625,
      "step": 6776
    },
    {
      "epoch": 0.9443321953598551,
      "grad_norm": 0.33266839385032654,
      "learning_rate": 3.5620074185626075e-07,
      "loss": 0.09765338897705078,
      "step": 6777
    },
    {
      "epoch": 0.9444715390510695,
      "grad_norm": 0.387081116437912,
      "learning_rate": 3.544316541404613e-07,
      "loss": 0.07778739929199219,
      "step": 6778
    },
    {
      "epoch": 0.9446108827422839,
      "grad_norm": 0.30255016684532166,
      "learning_rate": 3.526669312214326e-07,
      "loss": 0.07753181457519531,
      "step": 6779
    },
    {
      "epoch": 0.9447502264334983,
      "grad_norm": 0.2982569932937622,
      "learning_rate": 3.5090657349125647e-07,
      "loss": 0.07930564880371094,
      "step": 6780
    },
    {
      "epoch": 0.9448895701247126,
      "grad_norm": 0.5392690300941467,
      "learning_rate": 3.491505813410445e-07,
      "loss": 0.10487604141235352,
      "step": 6781
    },
    {
      "epoch": 0.945028913815927,
      "grad_norm": 0.521966278553009,
      "learning_rate": 3.473989551609358e-07,
      "loss": 0.09293174743652344,
      "step": 6782
    },
    {
      "epoch": 0.9451682575071414,
      "grad_norm": 0.24884885549545288,
      "learning_rate": 3.4565169534010123e-07,
      "loss": 0.06592941284179688,
      "step": 6783
    },
    {
      "epoch": 0.9453076011983558,
      "grad_norm": 0.4902370870113373,
      "learning_rate": 3.439088022667458e-07,
      "loss": 0.08084869384765625,
      "step": 6784
    },
    {
      "epoch": 0.9454469448895702,
      "grad_norm": 0.47224441170692444,
      "learning_rate": 3.421702763280976e-07,
      "loss": 0.09177112579345703,
      "step": 6785
    },
    {
      "epoch": 0.9455862885807845,
      "grad_norm": 0.3201057016849518,
      "learning_rate": 3.4043611791041874e-07,
      "loss": 0.07168960571289062,
      "step": 6786
    },
    {
      "epoch": 0.9457256322719989,
      "grad_norm": 0.216248020529747,
      "learning_rate": 3.387063273989966e-07,
      "loss": 0.06376266479492188,
      "step": 6787
    },
    {
      "epoch": 0.9458649759632133,
      "grad_norm": 0.37640872597694397,
      "learning_rate": 3.3698090517815696e-07,
      "loss": 0.08612728118896484,
      "step": 6788
    },
    {
      "epoch": 0.9460043196544277,
      "grad_norm": 0.812937319278717,
      "learning_rate": 3.352598516312422e-07,
      "loss": 0.13990020751953125,
      "step": 6789
    },
    {
      "epoch": 0.946143663345642,
      "grad_norm": 0.47826844453811646,
      "learning_rate": 3.3354316714063527e-07,
      "loss": 0.09345436096191406,
      "step": 6790
    },
    {
      "epoch": 0.9462830070368564,
      "grad_norm": 0.2711336612701416,
      "learning_rate": 3.318308520877489e-07,
      "loss": 0.07067680358886719,
      "step": 6791
    },
    {
      "epoch": 0.9464223507280708,
      "grad_norm": 0.4487355947494507,
      "learning_rate": 3.301229068530098e-07,
      "loss": 0.09762763977050781,
      "step": 6792
    },
    {
      "epoch": 0.9465616944192852,
      "grad_norm": 0.3978444039821625,
      "learning_rate": 3.2841933181589234e-07,
      "loss": 0.07670211791992188,
      "step": 6793
    },
    {
      "epoch": 0.9467010381104995,
      "grad_norm": 0.4419565498828888,
      "learning_rate": 3.26720127354887e-07,
      "loss": 0.08962249755859375,
      "step": 6794
    },
    {
      "epoch": 0.9468403818017139,
      "grad_norm": 0.18686196208000183,
      "learning_rate": 3.250252938475229e-07,
      "loss": 0.06031608581542969,
      "step": 6795
    },
    {
      "epoch": 0.9469797254929283,
      "grad_norm": 0.3549850881099701,
      "learning_rate": 3.2333483167035217e-07,
      "loss": 0.07960891723632812,
      "step": 6796
    },
    {
      "epoch": 0.9471190691841427,
      "grad_norm": 0.4807749390602112,
      "learning_rate": 3.216487411989544e-07,
      "loss": 0.08895015716552734,
      "step": 6797
    },
    {
      "epoch": 0.9472584128753571,
      "grad_norm": 0.2885565459728241,
      "learning_rate": 3.19967022807941e-07,
      "loss": 0.0731344223022461,
      "step": 6798
    },
    {
      "epoch": 0.9473977565665714,
      "grad_norm": 0.4322466254234314,
      "learning_rate": 3.182896768709531e-07,
      "loss": 0.1116037368774414,
      "step": 6799
    },
    {
      "epoch": 0.9475371002577858,
      "grad_norm": 0.32523635029792786,
      "learning_rate": 3.166167037606571e-07,
      "loss": 0.08742141723632812,
      "step": 6800
    },
    {
      "epoch": 0.9476764439490002,
      "grad_norm": 0.8179355263710022,
      "learning_rate": 3.1494810384875343e-07,
      "loss": 0.13885116577148438,
      "step": 6801
    },
    {
      "epoch": 0.9478157876402146,
      "grad_norm": 0.38150760531425476,
      "learning_rate": 3.132838775059632e-07,
      "loss": 0.08342647552490234,
      "step": 6802
    },
    {
      "epoch": 0.947955131331429,
      "grad_norm": 0.6485220789909363,
      "learning_rate": 3.116240251020375e-07,
      "loss": 0.12058830261230469,
      "step": 6803
    },
    {
      "epoch": 0.9480944750226433,
      "grad_norm": 0.34870508313179016,
      "learning_rate": 3.0996854700575896e-07,
      "loss": 0.0844268798828125,
      "step": 6804
    },
    {
      "epoch": 0.9482338187138577,
      "grad_norm": 0.48343950510025024,
      "learning_rate": 3.083174435849423e-07,
      "loss": 0.08814430236816406,
      "step": 6805
    },
    {
      "epoch": 0.9483731624050721,
      "grad_norm": 0.5554733276367188,
      "learning_rate": 3.0667071520641857e-07,
      "loss": 0.11525154113769531,
      "step": 6806
    },
    {
      "epoch": 0.9485125060962865,
      "grad_norm": 0.2892366647720337,
      "learning_rate": 3.05028362236055e-07,
      "loss": 0.0774850845336914,
      "step": 6807
    },
    {
      "epoch": 0.9486518497875008,
      "grad_norm": 0.32199710607528687,
      "learning_rate": 3.033903850387465e-07,
      "loss": 0.07614517211914062,
      "step": 6808
    },
    {
      "epoch": 0.9487911934787152,
      "grad_norm": 0.3980802595615387,
      "learning_rate": 3.01756783978413e-07,
      "loss": 0.0816202163696289,
      "step": 6809
    },
    {
      "epoch": 0.9489305371699296,
      "grad_norm": 0.28441715240478516,
      "learning_rate": 3.001275594180042e-07,
      "loss": 0.0677957534790039,
      "step": 6810
    },
    {
      "epoch": 0.949069880861144,
      "grad_norm": 0.3296477794647217,
      "learning_rate": 2.9850271171949495e-07,
      "loss": 0.07610130310058594,
      "step": 6811
    },
    {
      "epoch": 0.9492092245523583,
      "grad_norm": 0.5645394921302795,
      "learning_rate": 2.968822412438921e-07,
      "loss": 0.0933837890625,
      "step": 6812
    },
    {
      "epoch": 0.9493485682435727,
      "grad_norm": 0.37131574749946594,
      "learning_rate": 2.9526614835122314e-07,
      "loss": 0.08467864990234375,
      "step": 6813
    },
    {
      "epoch": 0.9494879119347871,
      "grad_norm": 0.22064082324504852,
      "learning_rate": 2.936544334005498e-07,
      "loss": 0.054790496826171875,
      "step": 6814
    },
    {
      "epoch": 0.9496272556260016,
      "grad_norm": 0.5203630924224854,
      "learning_rate": 2.920470967499589e-07,
      "loss": 0.0867919921875,
      "step": 6815
    },
    {
      "epoch": 0.949766599317216,
      "grad_norm": 0.3252600431442261,
      "learning_rate": 2.9044413875656266e-07,
      "loss": 0.07757568359375,
      "step": 6816
    },
    {
      "epoch": 0.9499059430084303,
      "grad_norm": 0.47037890553474426,
      "learning_rate": 2.8884555977650277e-07,
      "loss": 0.09355545043945312,
      "step": 6817
    },
    {
      "epoch": 0.9500452866996447,
      "grad_norm": 0.4034830331802368,
      "learning_rate": 2.8725136016494404e-07,
      "loss": 0.08321380615234375,
      "step": 6818
    },
    {
      "epoch": 0.9501846303908591,
      "grad_norm": 0.290571928024292,
      "learning_rate": 2.856615402760832e-07,
      "loss": 0.07594108581542969,
      "step": 6819
    },
    {
      "epoch": 0.9503239740820735,
      "grad_norm": 0.3489253520965576,
      "learning_rate": 2.8407610046314425e-07,
      "loss": 0.07201194763183594,
      "step": 6820
    },
    {
      "epoch": 0.9504633177732879,
      "grad_norm": 0.26834601163864136,
      "learning_rate": 2.824950410783722e-07,
      "loss": 0.0742497444152832,
      "step": 6821
    },
    {
      "epoch": 0.9506026614645022,
      "grad_norm": 0.34987637400627136,
      "learning_rate": 2.8091836247304603e-07,
      "loss": 0.085662841796875,
      "step": 6822
    },
    {
      "epoch": 0.9507420051557166,
      "grad_norm": 0.3450709581375122,
      "learning_rate": 2.7934606499746106e-07,
      "loss": 0.08451986312866211,
      "step": 6823
    },
    {
      "epoch": 0.950881348846931,
      "grad_norm": 0.4613151550292969,
      "learning_rate": 2.7777814900095344e-07,
      "loss": 0.10283851623535156,
      "step": 6824
    },
    {
      "epoch": 0.9510206925381454,
      "grad_norm": 0.33864906430244446,
      "learning_rate": 2.7621461483187563e-07,
      "loss": 0.07989501953125,
      "step": 6825
    },
    {
      "epoch": 0.9511600362293597,
      "grad_norm": 0.5372464656829834,
      "learning_rate": 2.7465546283760526e-07,
      "loss": 0.09317207336425781,
      "step": 6826
    },
    {
      "epoch": 0.9512993799205741,
      "grad_norm": 0.32699865102767944,
      "learning_rate": 2.731006933645586e-07,
      "loss": 0.07907390594482422,
      "step": 6827
    },
    {
      "epoch": 0.9514387236117885,
      "grad_norm": 0.2476688027381897,
      "learning_rate": 2.7155030675816153e-07,
      "loss": 0.06430625915527344,
      "step": 6828
    },
    {
      "epoch": 0.9515780673030029,
      "grad_norm": 0.23487898707389832,
      "learning_rate": 2.7000430336287855e-07,
      "loss": 0.07407283782958984,
      "step": 6829
    },
    {
      "epoch": 0.9517174109942172,
      "grad_norm": 0.632836639881134,
      "learning_rate": 2.684626835221971e-07,
      "loss": 0.10158824920654297,
      "step": 6830
    },
    {
      "epoch": 0.9518567546854316,
      "grad_norm": 0.782696545124054,
      "learning_rate": 2.669254475786276e-07,
      "loss": 0.12182998657226562,
      "step": 6831
    },
    {
      "epoch": 0.951996098376646,
      "grad_norm": 0.25316810607910156,
      "learning_rate": 2.6539259587371026e-07,
      "loss": 0.07417678833007812,
      "step": 6832
    },
    {
      "epoch": 0.9521354420678604,
      "grad_norm": 0.6412493586540222,
      "learning_rate": 2.638641287480104e-07,
      "loss": 0.10609626770019531,
      "step": 6833
    },
    {
      "epoch": 0.9522747857590748,
      "grad_norm": 0.4874313771724701,
      "learning_rate": 2.6234004654111854e-07,
      "loss": 0.09543275833129883,
      "step": 6834
    },
    {
      "epoch": 0.9524141294502891,
      "grad_norm": 0.6039149761199951,
      "learning_rate": 2.6082034959164613e-07,
      "loss": 0.11219596862792969,
      "step": 6835
    },
    {
      "epoch": 0.9525534731415035,
      "grad_norm": 0.3766322731971741,
      "learning_rate": 2.5930503823724086e-07,
      "loss": 0.09813117980957031,
      "step": 6836
    },
    {
      "epoch": 0.9526928168327179,
      "grad_norm": 0.7744880318641663,
      "learning_rate": 2.577941128145689e-07,
      "loss": 0.12305259704589844,
      "step": 6837
    },
    {
      "epoch": 0.9528321605239323,
      "grad_norm": 0.6358051896095276,
      "learning_rate": 2.562875736593218e-07,
      "loss": 0.11170005798339844,
      "step": 6838
    },
    {
      "epoch": 0.9529715042151466,
      "grad_norm": 0.33489784598350525,
      "learning_rate": 2.547854211062206e-07,
      "loss": 0.07693099975585938,
      "step": 6839
    },
    {
      "epoch": 0.953110847906361,
      "grad_norm": 0.2715928852558136,
      "learning_rate": 2.532876554890051e-07,
      "loss": 0.0626983642578125,
      "step": 6840
    },
    {
      "epoch": 0.9532501915975754,
      "grad_norm": 0.5176957845687866,
      "learning_rate": 2.5179427714045136e-07,
      "loss": 0.08530426025390625,
      "step": 6841
    },
    {
      "epoch": 0.9533895352887898,
      "grad_norm": 0.5964969396591187,
      "learning_rate": 2.5030528639234717e-07,
      "loss": 0.08558845520019531,
      "step": 6842
    },
    {
      "epoch": 0.9535288789800042,
      "grad_norm": 0.3663705885410309,
      "learning_rate": 2.488206835755147e-07,
      "loss": 0.07744073867797852,
      "step": 6843
    },
    {
      "epoch": 0.9536682226712185,
      "grad_norm": 0.505815863609314,
      "learning_rate": 2.4734046901980114e-07,
      "loss": 0.11165618896484375,
      "step": 6844
    },
    {
      "epoch": 0.9538075663624329,
      "grad_norm": 0.655243456363678,
      "learning_rate": 2.4586464305407454e-07,
      "loss": 0.1357421875,
      "step": 6845
    },
    {
      "epoch": 0.9539469100536473,
      "grad_norm": 0.6102237105369568,
      "learning_rate": 2.443932060062282e-07,
      "loss": 0.10542774200439453,
      "step": 6846
    },
    {
      "epoch": 0.9540862537448617,
      "grad_norm": 0.5520254373550415,
      "learning_rate": 2.429261582031828e-07,
      "loss": 0.10726737976074219,
      "step": 6847
    },
    {
      "epoch": 0.954225597436076,
      "grad_norm": 0.6612182259559631,
      "learning_rate": 2.4146349997088646e-07,
      "loss": 0.1026763916015625,
      "step": 6848
    },
    {
      "epoch": 0.9543649411272904,
      "grad_norm": 0.2598809599876404,
      "learning_rate": 2.400052316343038e-07,
      "loss": 0.07125663757324219,
      "step": 6849
    },
    {
      "epoch": 0.9545042848185048,
      "grad_norm": 0.3196321725845337,
      "learning_rate": 2.385513535174289e-07,
      "loss": 0.09238624572753906,
      "step": 6850
    },
    {
      "epoch": 0.9546436285097192,
      "grad_norm": 0.34134194254875183,
      "learning_rate": 2.3710186594328333e-07,
      "loss": 0.08108663558959961,
      "step": 6851
    },
    {
      "epoch": 0.9547829722009336,
      "grad_norm": 0.28955012559890747,
      "learning_rate": 2.3565676923390734e-07,
      "loss": 0.07465744018554688,
      "step": 6852
    },
    {
      "epoch": 0.9549223158921479,
      "grad_norm": 0.7093157172203064,
      "learning_rate": 2.3421606371037075e-07,
      "loss": 0.081024169921875,
      "step": 6853
    },
    {
      "epoch": 0.9550616595833623,
      "grad_norm": 0.5355100035667419,
      "learning_rate": 2.3277974969276417e-07,
      "loss": 0.10746383666992188,
      "step": 6854
    },
    {
      "epoch": 0.9552010032745768,
      "grad_norm": 0.3892204463481903,
      "learning_rate": 2.3134782750020347e-07,
      "loss": 0.0892801284790039,
      "step": 6855
    },
    {
      "epoch": 0.9553403469657912,
      "grad_norm": 0.2920844852924347,
      "learning_rate": 2.2992029745082966e-07,
      "loss": 0.07392501831054688,
      "step": 6856
    },
    {
      "epoch": 0.9554796906570056,
      "grad_norm": 0.2566264569759369,
      "learning_rate": 2.2849715986180688e-07,
      "loss": 0.06862640380859375,
      "step": 6857
    },
    {
      "epoch": 0.9556190343482199,
      "grad_norm": 0.7659746408462524,
      "learning_rate": 2.2707841504932438e-07,
      "loss": 0.14964866638183594,
      "step": 6858
    },
    {
      "epoch": 0.9557583780394343,
      "grad_norm": 0.5824476480484009,
      "learning_rate": 2.2566406332859449e-07,
      "loss": 0.12337684631347656,
      "step": 6859
    },
    {
      "epoch": 0.9558977217306487,
      "grad_norm": 0.4194205403327942,
      "learning_rate": 2.242541050138547e-07,
      "loss": 0.07881641387939453,
      "step": 6860
    },
    {
      "epoch": 0.9560370654218631,
      "grad_norm": 0.5804741382598877,
      "learning_rate": 2.2284854041836335e-07,
      "loss": 0.1168060302734375,
      "step": 6861
    },
    {
      "epoch": 0.9561764091130774,
      "grad_norm": 0.3460279405117035,
      "learning_rate": 2.214473698544084e-07,
      "loss": 0.0838165283203125,
      "step": 6862
    },
    {
      "epoch": 0.9563157528042918,
      "grad_norm": 0.5177403092384338,
      "learning_rate": 2.2005059363329196e-07,
      "loss": 0.08751773834228516,
      "step": 6863
    },
    {
      "epoch": 0.9564550964955062,
      "grad_norm": 0.5308154225349426,
      "learning_rate": 2.1865821206535243e-07,
      "loss": 0.1054840087890625,
      "step": 6864
    },
    {
      "epoch": 0.9565944401867206,
      "grad_norm": 0.6004311442375183,
      "learning_rate": 2.1727022545994237e-07,
      "loss": 0.09744071960449219,
      "step": 6865
    },
    {
      "epoch": 0.956733783877935,
      "grad_norm": 0.46168020367622375,
      "learning_rate": 2.1588663412544174e-07,
      "loss": 0.08893585205078125,
      "step": 6866
    },
    {
      "epoch": 0.9568731275691493,
      "grad_norm": 0.7353544235229492,
      "learning_rate": 2.1450743836925136e-07,
      "loss": 0.11985588073730469,
      "step": 6867
    },
    {
      "epoch": 0.9570124712603637,
      "grad_norm": 0.25411537289619446,
      "learning_rate": 2.1313263849779498e-07,
      "loss": 0.07079029083251953,
      "step": 6868
    },
    {
      "epoch": 0.9571518149515781,
      "grad_norm": 0.4066571891307831,
      "learning_rate": 2.1176223481652824e-07,
      "loss": 0.092071533203125,
      "step": 6869
    },
    {
      "epoch": 0.9572911586427925,
      "grad_norm": 0.6468311548233032,
      "learning_rate": 2.1039622762991874e-07,
      "loss": 0.10663414001464844,
      "step": 6870
    },
    {
      "epoch": 0.9574305023340068,
      "grad_norm": 0.36678484082221985,
      "learning_rate": 2.0903461724146146e-07,
      "loss": 0.08179473876953125,
      "step": 6871
    },
    {
      "epoch": 0.9575698460252212,
      "grad_norm": 0.26636847853660583,
      "learning_rate": 2.0767740395367886e-07,
      "loss": 0.06438827514648438,
      "step": 6872
    },
    {
      "epoch": 0.9577091897164356,
      "grad_norm": 0.631134569644928,
      "learning_rate": 2.0632458806810974e-07,
      "loss": 0.1323070526123047,
      "step": 6873
    },
    {
      "epoch": 0.95784853340765,
      "grad_norm": 0.3878439962863922,
      "learning_rate": 2.0497616988532032e-07,
      "loss": 0.08051109313964844,
      "step": 6874
    },
    {
      "epoch": 0.9579878770988643,
      "grad_norm": 0.37732556462287903,
      "learning_rate": 2.0363214970489763e-07,
      "loss": 0.08594512939453125,
      "step": 6875
    },
    {
      "epoch": 0.9581272207900787,
      "grad_norm": 0.2774856686592102,
      "learning_rate": 2.0229252782545171e-07,
      "loss": 0.07093238830566406,
      "step": 6876
    },
    {
      "epoch": 0.9582665644812931,
      "grad_norm": 0.5948699712753296,
      "learning_rate": 2.0095730454461781e-07,
      "loss": 0.10755538940429688,
      "step": 6877
    },
    {
      "epoch": 0.9584059081725075,
      "grad_norm": 0.5790180563926697,
      "learning_rate": 1.9962648015904972e-07,
      "loss": 0.10915184020996094,
      "step": 6878
    },
    {
      "epoch": 0.9585452518637219,
      "grad_norm": 0.7054398655891418,
      "learning_rate": 1.9830005496442873e-07,
      "loss": 0.12053871154785156,
      "step": 6879
    },
    {
      "epoch": 0.9586845955549362,
      "grad_norm": 0.40910354256629944,
      "learning_rate": 1.969780292554546e-07,
      "loss": 0.0960693359375,
      "step": 6880
    },
    {
      "epoch": 0.9588239392461506,
      "grad_norm": 0.41224390268325806,
      "learning_rate": 1.9566040332585246e-07,
      "loss": 0.09960746765136719,
      "step": 6881
    },
    {
      "epoch": 0.958963282937365,
      "grad_norm": 0.26176291704177856,
      "learning_rate": 1.9434717746836805e-07,
      "loss": 0.07081985473632812,
      "step": 6882
    },
    {
      "epoch": 0.9591026266285794,
      "grad_norm": 0.44478657841682434,
      "learning_rate": 1.9303835197476804e-07,
      "loss": 0.08673095703125,
      "step": 6883
    },
    {
      "epoch": 0.9592419703197937,
      "grad_norm": 0.5774131417274475,
      "learning_rate": 1.917339271358465e-07,
      "loss": 0.10587501525878906,
      "step": 6884
    },
    {
      "epoch": 0.9593813140110081,
      "grad_norm": 0.30005010962486267,
      "learning_rate": 1.9043390324141597e-07,
      "loss": 0.07809829711914062,
      "step": 6885
    },
    {
      "epoch": 0.9595206577022225,
      "grad_norm": 0.39908862113952637,
      "learning_rate": 1.8913828058031436e-07,
      "loss": 0.08977985382080078,
      "step": 6886
    },
    {
      "epoch": 0.9596600013934369,
      "grad_norm": 0.31218811869621277,
      "learning_rate": 1.878470594403936e-07,
      "loss": 0.077484130859375,
      "step": 6887
    },
    {
      "epoch": 0.9597993450846513,
      "grad_norm": 0.20497344434261322,
      "learning_rate": 1.865602401085398e-07,
      "loss": 0.060100555419921875,
      "step": 6888
    },
    {
      "epoch": 0.9599386887758656,
      "grad_norm": 0.6906037926673889,
      "learning_rate": 1.852778228706509e-07,
      "loss": 0.10911750793457031,
      "step": 6889
    },
    {
      "epoch": 0.96007803246708,
      "grad_norm": 0.29766973853111267,
      "learning_rate": 1.8399980801165006e-07,
      "loss": 0.07554864883422852,
      "step": 6890
    },
    {
      "epoch": 0.9602173761582944,
      "grad_norm": 0.24392370879650116,
      "learning_rate": 1.8272619581549022e-07,
      "loss": 0.06373119354248047,
      "step": 6891
    },
    {
      "epoch": 0.9603567198495088,
      "grad_norm": 0.37702521681785583,
      "learning_rate": 1.8145698656512943e-07,
      "loss": 0.08230781555175781,
      "step": 6892
    },
    {
      "epoch": 0.9604960635407231,
      "grad_norm": 0.3420012891292572,
      "learning_rate": 1.8019218054256216e-07,
      "loss": 0.07683563232421875,
      "step": 6893
    },
    {
      "epoch": 0.9606354072319375,
      "grad_norm": 0.31027016043663025,
      "learning_rate": 1.7893177802879692e-07,
      "loss": 0.06746292114257812,
      "step": 6894
    },
    {
      "epoch": 0.9607747509231519,
      "grad_norm": 0.533224880695343,
      "learning_rate": 1.776757793038697e-07,
      "loss": 0.08683204650878906,
      "step": 6895
    },
    {
      "epoch": 0.9609140946143664,
      "grad_norm": 0.26800382137298584,
      "learning_rate": 1.7642418464683287e-07,
      "loss": 0.07262229919433594,
      "step": 6896
    },
    {
      "epoch": 0.9610534383055808,
      "grad_norm": 0.2904371917247772,
      "learning_rate": 1.7517699433576173e-07,
      "loss": 0.07321834564208984,
      "step": 6897
    },
    {
      "epoch": 0.9611927819967951,
      "grad_norm": 0.3037748336791992,
      "learning_rate": 1.7393420864775467e-07,
      "loss": 0.08370590209960938,
      "step": 6898
    },
    {
      "epoch": 0.9613321256880095,
      "grad_norm": 0.4292869567871094,
      "learning_rate": 1.7269582785892858e-07,
      "loss": 0.10738945007324219,
      "step": 6899
    },
    {
      "epoch": 0.9614714693792239,
      "grad_norm": 0.28763148188591003,
      "learning_rate": 1.7146185224442557e-07,
      "loss": 0.06386375427246094,
      "step": 6900
    },
    {
      "epoch": 0.9616108130704383,
      "grad_norm": 0.5367462038993835,
      "learning_rate": 1.7023228207840637e-07,
      "loss": 0.1044921875,
      "step": 6901
    },
    {
      "epoch": 0.9617501567616527,
      "grad_norm": 0.42610830068588257,
      "learning_rate": 1.6900711763405242e-07,
      "loss": 0.09033393859863281,
      "step": 6902
    },
    {
      "epoch": 0.961889500452867,
      "grad_norm": 0.5485749840736389,
      "learning_rate": 1.67786359183566e-07,
      "loss": 0.1175689697265625,
      "step": 6903
    },
    {
      "epoch": 0.9620288441440814,
      "grad_norm": 0.37856075167655945,
      "learning_rate": 1.665700069981746e-07,
      "loss": 0.08183479309082031,
      "step": 6904
    },
    {
      "epoch": 0.9621681878352958,
      "grad_norm": 0.3508590757846832,
      "learning_rate": 1.6535806134812427e-07,
      "loss": 0.06269598007202148,
      "step": 6905
    },
    {
      "epoch": 0.9623075315265102,
      "grad_norm": 0.3254605233669281,
      "learning_rate": 1.6415052250267738e-07,
      "loss": 0.06852245330810547,
      "step": 6906
    },
    {
      "epoch": 0.9624468752177245,
      "grad_norm": 0.40261203050613403,
      "learning_rate": 1.62947390730126e-07,
      "loss": 0.08795356750488281,
      "step": 6907
    },
    {
      "epoch": 0.9625862189089389,
      "grad_norm": 0.48366087675094604,
      "learning_rate": 1.617486662977763e-07,
      "loss": 0.09106731414794922,
      "step": 6908
    },
    {
      "epoch": 0.9627255626001533,
      "grad_norm": 0.3535245954990387,
      "learning_rate": 1.6055434947195746e-07,
      "loss": 0.08889389038085938,
      "step": 6909
    },
    {
      "epoch": 0.9628649062913677,
      "grad_norm": 0.3078405559062958,
      "learning_rate": 1.5936444051801947e-07,
      "loss": 0.07130241394042969,
      "step": 6910
    },
    {
      "epoch": 0.963004249982582,
      "grad_norm": 0.4257046580314636,
      "learning_rate": 1.5817893970033305e-07,
      "loss": 0.09334564208984375,
      "step": 6911
    },
    {
      "epoch": 0.9631435936737964,
      "grad_norm": 0.4736168384552002,
      "learning_rate": 1.5699784728229196e-07,
      "loss": 0.08681297302246094,
      "step": 6912
    },
    {
      "epoch": 0.9632829373650108,
      "grad_norm": 0.5674563050270081,
      "learning_rate": 1.5582116352630626e-07,
      "loss": 0.08890533447265625,
      "step": 6913
    },
    {
      "epoch": 0.9634222810562252,
      "grad_norm": 0.45424142479896545,
      "learning_rate": 1.5464888869380468e-07,
      "loss": 0.08562278747558594,
      "step": 6914
    },
    {
      "epoch": 0.9635616247474396,
      "grad_norm": 0.7406771779060364,
      "learning_rate": 1.5348102304524548e-07,
      "loss": 0.13175678253173828,
      "step": 6915
    },
    {
      "epoch": 0.9637009684386539,
      "grad_norm": 0.3858097493648529,
      "learning_rate": 1.523175668400989e-07,
      "loss": 0.0803537368774414,
      "step": 6916
    },
    {
      "epoch": 0.9638403121298683,
      "grad_norm": 0.3306906521320343,
      "learning_rate": 1.511585203368582e-07,
      "loss": 0.08117866516113281,
      "step": 6917
    },
    {
      "epoch": 0.9639796558210827,
      "grad_norm": 0.3828243017196655,
      "learning_rate": 1.5000388379303732e-07,
      "loss": 0.06913185119628906,
      "step": 6918
    },
    {
      "epoch": 0.9641189995122971,
      "grad_norm": 0.4043271243572235,
      "learning_rate": 1.4885365746516889e-07,
      "loss": 0.08068084716796875,
      "step": 6919
    },
    {
      "epoch": 0.9642583432035114,
      "grad_norm": 0.2773694097995758,
      "learning_rate": 1.477078416088107e-07,
      "loss": 0.06998443603515625,
      "step": 6920
    },
    {
      "epoch": 0.9643976868947258,
      "grad_norm": 0.746692419052124,
      "learning_rate": 1.465664364785324e-07,
      "loss": 0.10826492309570312,
      "step": 6921
    },
    {
      "epoch": 0.9645370305859402,
      "grad_norm": 0.6899163126945496,
      "learning_rate": 1.454294423279312e-07,
      "loss": 0.12300872802734375,
      "step": 6922
    },
    {
      "epoch": 0.9646763742771546,
      "grad_norm": 0.5337240695953369,
      "learning_rate": 1.4429685940962278e-07,
      "loss": 0.11635398864746094,
      "step": 6923
    },
    {
      "epoch": 0.964815717968369,
      "grad_norm": 0.7090213298797607,
      "learning_rate": 1.4316868797523697e-07,
      "loss": 0.10947799682617188,
      "step": 6924
    },
    {
      "epoch": 0.9649550616595833,
      "grad_norm": 0.22038039565086365,
      "learning_rate": 1.420449282754288e-07,
      "loss": 0.061023712158203125,
      "step": 6925
    },
    {
      "epoch": 0.9650944053507977,
      "grad_norm": 0.5109238028526306,
      "learning_rate": 1.4092558055987193e-07,
      "loss": 0.10279083251953125,
      "step": 6926
    },
    {
      "epoch": 0.9652337490420121,
      "grad_norm": 0.27294161915779114,
      "learning_rate": 1.3981064507726295e-07,
      "loss": 0.07410430908203125,
      "step": 6927
    },
    {
      "epoch": 0.9653730927332265,
      "grad_norm": 0.2468087524175644,
      "learning_rate": 1.387001220753126e-07,
      "loss": 0.06290435791015625,
      "step": 6928
    },
    {
      "epoch": 0.9655124364244408,
      "grad_norm": 0.520156979560852,
      "learning_rate": 1.3759401180075239e-07,
      "loss": 0.07934188842773438,
      "step": 6929
    },
    {
      "epoch": 0.9656517801156552,
      "grad_norm": 0.7201396226882935,
      "learning_rate": 1.3649231449933686e-07,
      "loss": 0.14395713806152344,
      "step": 6930
    },
    {
      "epoch": 0.9657911238068696,
      "grad_norm": 0.4271526038646698,
      "learning_rate": 1.3539503041583913e-07,
      "loss": 0.09188652038574219,
      "step": 6931
    },
    {
      "epoch": 0.965930467498084,
      "grad_norm": 0.481665700674057,
      "learning_rate": 1.3430215979404638e-07,
      "loss": 0.10725021362304688,
      "step": 6932
    },
    {
      "epoch": 0.9660698111892984,
      "grad_norm": 0.6855002045631409,
      "learning_rate": 1.3321370287677328e-07,
      "loss": 0.11236763000488281,
      "step": 6933
    },
    {
      "epoch": 0.9662091548805127,
      "grad_norm": 0.5029910802841187,
      "learning_rate": 1.321296599058508e-07,
      "loss": 0.10214042663574219,
      "step": 6934
    },
    {
      "epoch": 0.9663484985717271,
      "grad_norm": 0.7182873487472534,
      "learning_rate": 1.310500311221241e-07,
      "loss": 0.08667182922363281,
      "step": 6935
    },
    {
      "epoch": 0.9664878422629416,
      "grad_norm": 1.0259249210357666,
      "learning_rate": 1.2997481676546576e-07,
      "loss": 0.11879348754882812,
      "step": 6936
    },
    {
      "epoch": 0.966627185954156,
      "grad_norm": 0.5986254215240479,
      "learning_rate": 1.2890401707476242e-07,
      "loss": 0.12112808227539062,
      "step": 6937
    },
    {
      "epoch": 0.9667665296453704,
      "grad_norm": 0.5382232666015625,
      "learning_rate": 1.2783763228792156e-07,
      "loss": 0.09486103057861328,
      "step": 6938
    },
    {
      "epoch": 0.9669058733365847,
      "grad_norm": 0.6517792344093323,
      "learning_rate": 1.2677566264186925e-07,
      "loss": 0.11126899719238281,
      "step": 6939
    },
    {
      "epoch": 0.9670452170277991,
      "grad_norm": 0.22945058345794678,
      "learning_rate": 1.2571810837255228e-07,
      "loss": 0.05762887001037598,
      "step": 6940
    },
    {
      "epoch": 0.9671845607190135,
      "grad_norm": 0.4503476023674011,
      "learning_rate": 1.246649697149338e-07,
      "loss": 0.0834808349609375,
      "step": 6941
    },
    {
      "epoch": 0.9673239044102279,
      "grad_norm": 0.41739916801452637,
      "learning_rate": 1.2361624690299557e-07,
      "loss": 0.08879852294921875,
      "step": 6942
    },
    {
      "epoch": 0.9674632481014422,
      "grad_norm": 0.38998496532440186,
      "learning_rate": 1.225719401697445e-07,
      "loss": 0.08797454833984375,
      "step": 6943
    },
    {
      "epoch": 0.9676025917926566,
      "grad_norm": 0.45006734132766724,
      "learning_rate": 1.2153204974719722e-07,
      "loss": 0.06791305541992188,
      "step": 6944
    },
    {
      "epoch": 0.967741935483871,
      "grad_norm": 0.37120741605758667,
      "learning_rate": 1.2049657586639786e-07,
      "loss": 0.07333755493164062,
      "step": 6945
    },
    {
      "epoch": 0.9678812791750854,
      "grad_norm": 0.48564016819000244,
      "learning_rate": 1.194655187574001e-07,
      "loss": 0.07888603210449219,
      "step": 6946
    },
    {
      "epoch": 0.9680206228662998,
      "grad_norm": 0.705226719379425,
      "learning_rate": 1.1843887864928294e-07,
      "loss": 0.11122679710388184,
      "step": 6947
    },
    {
      "epoch": 0.9681599665575141,
      "grad_norm": 0.44755661487579346,
      "learning_rate": 1.1741665577014393e-07,
      "loss": 0.077423095703125,
      "step": 6948
    },
    {
      "epoch": 0.9682993102487285,
      "grad_norm": 0.6297992467880249,
      "learning_rate": 1.163988503470992e-07,
      "loss": 0.11162757873535156,
      "step": 6949
    },
    {
      "epoch": 0.9684386539399429,
      "grad_norm": 0.5026305913925171,
      "learning_rate": 1.153854626062767e-07,
      "loss": 0.10066604614257812,
      "step": 6950
    },
    {
      "epoch": 0.9685779976311573,
      "grad_norm": 0.5859058499336243,
      "learning_rate": 1.1437649277283191e-07,
      "loss": 0.10268115997314453,
      "step": 6951
    },
    {
      "epoch": 0.9687173413223716,
      "grad_norm": 0.25809770822525024,
      "learning_rate": 1.133719410709344e-07,
      "loss": 0.05733203887939453,
      "step": 6952
    },
    {
      "epoch": 0.968856685013586,
      "grad_norm": 0.2935693860054016,
      "learning_rate": 1.1237180772377233e-07,
      "loss": 0.07291603088378906,
      "step": 6953
    },
    {
      "epoch": 0.9689960287048004,
      "grad_norm": 0.42765146493911743,
      "learning_rate": 1.1137609295355234e-07,
      "loss": 0.09762382507324219,
      "step": 6954
    },
    {
      "epoch": 0.9691353723960148,
      "grad_norm": 0.6709672212600708,
      "learning_rate": 1.1038479698149752e-07,
      "loss": 0.08826065063476562,
      "step": 6955
    },
    {
      "epoch": 0.9692747160872291,
      "grad_norm": 0.2507948875427246,
      "learning_rate": 1.093979200278561e-07,
      "loss": 0.07392692565917969,
      "step": 6956
    },
    {
      "epoch": 0.9694140597784435,
      "grad_norm": 0.44603589177131653,
      "learning_rate": 1.0841546231188382e-07,
      "loss": 0.09691047668457031,
      "step": 6957
    },
    {
      "epoch": 0.9695534034696579,
      "grad_norm": 0.3231179118156433,
      "learning_rate": 1.0743742405186385e-07,
      "loss": 0.07532978057861328,
      "step": 6958
    },
    {
      "epoch": 0.9696927471608723,
      "grad_norm": 0.3740464746952057,
      "learning_rate": 1.0646380546509572e-07,
      "loss": 0.09311771392822266,
      "step": 6959
    },
    {
      "epoch": 0.9698320908520867,
      "grad_norm": 0.46767061948776245,
      "learning_rate": 1.054946067678908e-07,
      "loss": 0.08648872375488281,
      "step": 6960
    },
    {
      "epoch": 0.969971434543301,
      "grad_norm": 0.35739776492118835,
      "learning_rate": 1.0452982817558577e-07,
      "loss": 0.08187484741210938,
      "step": 6961
    },
    {
      "epoch": 0.9701107782345154,
      "grad_norm": 0.3432977497577667,
      "learning_rate": 1.0356946990253137e-07,
      "loss": 0.07628822326660156,
      "step": 6962
    },
    {
      "epoch": 0.9702501219257298,
      "grad_norm": 0.2109745889902115,
      "learning_rate": 1.0261353216209691e-07,
      "loss": 0.06323432922363281,
      "step": 6963
    },
    {
      "epoch": 0.9703894656169442,
      "grad_norm": 0.46725064516067505,
      "learning_rate": 1.0166201516667029e-07,
      "loss": 0.09242820739746094,
      "step": 6964
    },
    {
      "epoch": 0.9705288093081585,
      "grad_norm": 0.47335028648376465,
      "learning_rate": 1.0071491912766018e-07,
      "loss": 0.09148979187011719,
      "step": 6965
    },
    {
      "epoch": 0.9706681529993729,
      "grad_norm": 0.3854447901248932,
      "learning_rate": 9.977224425548271e-08,
      "loss": 0.09890365600585938,
      "step": 6966
    },
    {
      "epoch": 0.9708074966905873,
      "grad_norm": 0.30972781777381897,
      "learning_rate": 9.883399075958589e-08,
      "loss": 0.07860946655273438,
      "step": 6967
    },
    {
      "epoch": 0.9709468403818017,
      "grad_norm": 0.31210437417030334,
      "learning_rate": 9.790015884842297e-08,
      "loss": 0.0736083984375,
      "step": 6968
    },
    {
      "epoch": 0.9710861840730161,
      "grad_norm": 0.2317143976688385,
      "learning_rate": 9.697074872947242e-08,
      "loss": 0.06217765808105469,
      "step": 6969
    },
    {
      "epoch": 0.9712255277642304,
      "grad_norm": 0.44886812567710876,
      "learning_rate": 9.604576060922687e-08,
      "loss": 0.08819389343261719,
      "step": 6970
    },
    {
      "epoch": 0.9713648714554448,
      "grad_norm": 0.5474607348442078,
      "learning_rate": 9.51251946931997e-08,
      "loss": 0.09588050842285156,
      "step": 6971
    },
    {
      "epoch": 0.9715042151466592,
      "grad_norm": 0.38972100615501404,
      "learning_rate": 9.420905118591617e-08,
      "loss": 0.08313179016113281,
      "step": 6972
    },
    {
      "epoch": 0.9716435588378736,
      "grad_norm": 0.2803761065006256,
      "learning_rate": 9.32973302909268e-08,
      "loss": 0.0762186050415039,
      "step": 6973
    },
    {
      "epoch": 0.9717829025290879,
      "grad_norm": 0.2522527277469635,
      "learning_rate": 9.239003221079179e-08,
      "loss": 0.06608009338378906,
      "step": 6974
    },
    {
      "epoch": 0.9719222462203023,
      "grad_norm": 0.38832926750183105,
      "learning_rate": 9.148715714709433e-08,
      "loss": 0.0764017105102539,
      "step": 6975
    },
    {
      "epoch": 0.9720615899115168,
      "grad_norm": 0.551490068435669,
      "learning_rate": 9.058870530042952e-08,
      "loss": 0.09401130676269531,
      "step": 6976
    },
    {
      "epoch": 0.9722009336027312,
      "grad_norm": 0.5891680717468262,
      "learning_rate": 8.969467687041766e-08,
      "loss": 0.11402702331542969,
      "step": 6977
    },
    {
      "epoch": 0.9723402772939456,
      "grad_norm": 0.40731725096702576,
      "learning_rate": 8.880507205568656e-08,
      "loss": 0.09615612030029297,
      "step": 6978
    },
    {
      "epoch": 0.9724796209851599,
      "grad_norm": 0.385280579328537,
      "learning_rate": 8.791989105388699e-08,
      "loss": 0.08094978332519531,
      "step": 6979
    },
    {
      "epoch": 0.9726189646763743,
      "grad_norm": 0.5281490087509155,
      "learning_rate": 8.70391340616883e-08,
      "loss": 0.13261032104492188,
      "step": 6980
    },
    {
      "epoch": 0.9727583083675887,
      "grad_norm": 0.668847382068634,
      "learning_rate": 8.616280127477395e-08,
      "loss": 0.10655689239501953,
      "step": 6981
    },
    {
      "epoch": 0.9728976520588031,
      "grad_norm": 0.5761567950248718,
      "learning_rate": 8.529089288784376e-08,
      "loss": 0.08992767333984375,
      "step": 6982
    },
    {
      "epoch": 0.9730369957500175,
      "grad_norm": 0.5830804705619812,
      "learning_rate": 8.442340909461832e-08,
      "loss": 0.10387802124023438,
      "step": 6983
    },
    {
      "epoch": 0.9731763394412318,
      "grad_norm": 0.3646184802055359,
      "learning_rate": 8.356035008783014e-08,
      "loss": 0.08025169372558594,
      "step": 6984
    },
    {
      "epoch": 0.9733156831324462,
      "grad_norm": 0.39931198954582214,
      "learning_rate": 8.270171605923027e-08,
      "loss": 0.07467174530029297,
      "step": 6985
    },
    {
      "epoch": 0.9734550268236606,
      "grad_norm": 0.6173413395881653,
      "learning_rate": 8.184750719959278e-08,
      "loss": 0.09091377258300781,
      "step": 6986
    },
    {
      "epoch": 0.973594370514875,
      "grad_norm": 0.41381680965423584,
      "learning_rate": 8.099772369869696e-08,
      "loss": 0.0939178466796875,
      "step": 6987
    },
    {
      "epoch": 0.9737337142060893,
      "grad_norm": 0.34412848949432373,
      "learning_rate": 8.015236574534957e-08,
      "loss": 0.08616828918457031,
      "step": 6988
    },
    {
      "epoch": 0.9738730578973037,
      "grad_norm": 0.41478678584098816,
      "learning_rate": 7.931143352736925e-08,
      "loss": 0.09993362426757812,
      "step": 6989
    },
    {
      "epoch": 0.9740124015885181,
      "grad_norm": 0.43842586874961853,
      "learning_rate": 7.8474927231591e-08,
      "loss": 0.07330989837646484,
      "step": 6990
    },
    {
      "epoch": 0.9741517452797325,
      "grad_norm": 0.41159167885780334,
      "learning_rate": 7.764284704386837e-08,
      "loss": 0.07631874084472656,
      "step": 6991
    },
    {
      "epoch": 0.9742910889709469,
      "grad_norm": 0.23821572959423065,
      "learning_rate": 7.681519314906904e-08,
      "loss": 0.06789970397949219,
      "step": 6992
    },
    {
      "epoch": 0.9744304326621612,
      "grad_norm": 0.5896190404891968,
      "learning_rate": 7.599196573107925e-08,
      "loss": 0.09914016723632812,
      "step": 6993
    },
    {
      "epoch": 0.9745697763533756,
      "grad_norm": 0.3216404318809509,
      "learning_rate": 7.517316497280158e-08,
      "loss": 0.07890510559082031,
      "step": 6994
    },
    {
      "epoch": 0.97470912004459,
      "grad_norm": 0.31259527802467346,
      "learning_rate": 7.435879105615718e-08,
      "loss": 0.08390045166015625,
      "step": 6995
    },
    {
      "epoch": 0.9748484637358044,
      "grad_norm": 0.47789040207862854,
      "learning_rate": 7.354884416207686e-08,
      "loss": 0.10335922241210938,
      "step": 6996
    },
    {
      "epoch": 0.9749878074270187,
      "grad_norm": 0.2896292805671692,
      "learning_rate": 7.274332447051668e-08,
      "loss": 0.06148719787597656,
      "step": 6997
    },
    {
      "epoch": 0.9751271511182331,
      "grad_norm": 0.5316546559333801,
      "learning_rate": 7.194223216044238e-08,
      "loss": 0.10186767578125,
      "step": 6998
    },
    {
      "epoch": 0.9752664948094475,
      "grad_norm": 0.39263755083084106,
      "learning_rate": 7.114556740983824e-08,
      "loss": 0.09210681915283203,
      "step": 6999
    },
    {
      "epoch": 0.9754058385006619,
      "grad_norm": 0.3270455002784729,
      "learning_rate": 7.035333039570492e-08,
      "loss": 0.08107757568359375,
      "step": 7000
    },
    {
      "epoch": 0.9755451821918762,
      "grad_norm": 0.46385714411735535,
      "learning_rate": 6.956552129406158e-08,
      "loss": 0.08921623229980469,
      "step": 7001
    },
    {
      "epoch": 0.9756845258830906,
      "grad_norm": 0.5873894095420837,
      "learning_rate": 6.878214027993935e-08,
      "loss": 0.10069561004638672,
      "step": 7002
    },
    {
      "epoch": 0.975823869574305,
      "grad_norm": 0.4363866150379181,
      "learning_rate": 6.800318752738788e-08,
      "loss": 0.10394477844238281,
      "step": 7003
    },
    {
      "epoch": 0.9759632132655194,
      "grad_norm": 0.6883102655410767,
      "learning_rate": 6.722866320947319e-08,
      "loss": 0.10693740844726562,
      "step": 7004
    },
    {
      "epoch": 0.9761025569567338,
      "grad_norm": 0.3237481117248535,
      "learning_rate": 6.64585674982754e-08,
      "loss": 0.07271003723144531,
      "step": 7005
    },
    {
      "epoch": 0.9762419006479481,
      "grad_norm": 0.6260445713996887,
      "learning_rate": 6.569290056489542e-08,
      "loss": 0.10754013061523438,
      "step": 7006
    },
    {
      "epoch": 0.9763812443391625,
      "grad_norm": 0.38082200288772583,
      "learning_rate": 6.493166257944384e-08,
      "loss": 0.07172489166259766,
      "step": 7007
    },
    {
      "epoch": 0.9765205880303769,
      "grad_norm": 0.37063854932785034,
      "learning_rate": 6.417485371105204e-08,
      "loss": 0.07721710205078125,
      "step": 7008
    },
    {
      "epoch": 0.9766599317215913,
      "grad_norm": 0.26536422967910767,
      "learning_rate": 6.342247412786329e-08,
      "loss": 0.07077693939208984,
      "step": 7009
    },
    {
      "epoch": 0.9767992754128056,
      "grad_norm": 0.4128875732421875,
      "learning_rate": 6.267452399704387e-08,
      "loss": 0.09178829193115234,
      "step": 7010
    },
    {
      "epoch": 0.97693861910402,
      "grad_norm": 0.3108527362346649,
      "learning_rate": 6.193100348476533e-08,
      "loss": 0.060512542724609375,
      "step": 7011
    },
    {
      "epoch": 0.9770779627952344,
      "grad_norm": 0.7662290334701538,
      "learning_rate": 6.11919127562266e-08,
      "loss": 0.11783027648925781,
      "step": 7012
    },
    {
      "epoch": 0.9772173064864488,
      "grad_norm": 0.3592942953109741,
      "learning_rate": 6.045725197563413e-08,
      "loss": 0.07145118713378906,
      "step": 7013
    },
    {
      "epoch": 0.9773566501776632,
      "grad_norm": 0.46367043256759644,
      "learning_rate": 5.972702130621067e-08,
      "loss": 0.0959005355834961,
      "step": 7014
    },
    {
      "epoch": 0.9774959938688775,
      "grad_norm": 0.665275514125824,
      "learning_rate": 5.900122091019977e-08,
      "loss": 0.10320186614990234,
      "step": 7015
    },
    {
      "epoch": 0.977635337560092,
      "grad_norm": 0.7338187098503113,
      "learning_rate": 5.827985094885691e-08,
      "loss": 0.11705780029296875,
      "step": 7016
    },
    {
      "epoch": 0.9777746812513064,
      "grad_norm": 0.38060909509658813,
      "learning_rate": 5.756291158245386e-08,
      "loss": 0.08435726165771484,
      "step": 7017
    },
    {
      "epoch": 0.9779140249425208,
      "grad_norm": 0.4880305826663971,
      "learning_rate": 5.6850402970278774e-08,
      "loss": 0.09760475158691406,
      "step": 7018
    },
    {
      "epoch": 0.9780533686337352,
      "grad_norm": 0.2773900628089905,
      "learning_rate": 5.6142325270633904e-08,
      "loss": 0.06842613220214844,
      "step": 7019
    },
    {
      "epoch": 0.9781927123249495,
      "grad_norm": 0.7557621598243713,
      "learning_rate": 5.543867864083785e-08,
      "loss": 0.13771915435791016,
      "step": 7020
    },
    {
      "epoch": 0.9783320560161639,
      "grad_norm": 0.2838873565196991,
      "learning_rate": 5.473946323722556e-08,
      "loss": 0.07871818542480469,
      "step": 7021
    },
    {
      "epoch": 0.9784713997073783,
      "grad_norm": 0.3506816029548645,
      "learning_rate": 5.404467921514611e-08,
      "loss": 0.0852508544921875,
      "step": 7022
    },
    {
      "epoch": 0.9786107433985927,
      "grad_norm": 0.4458865523338318,
      "learning_rate": 5.335432672896712e-08,
      "loss": 0.09770965576171875,
      "step": 7023
    },
    {
      "epoch": 0.978750087089807,
      "grad_norm": 0.31380510330200195,
      "learning_rate": 5.26684059320659e-08,
      "loss": 0.07405471801757812,
      "step": 7024
    },
    {
      "epoch": 0.9788894307810214,
      "grad_norm": 0.2875730097293854,
      "learning_rate": 5.198691697683833e-08,
      "loss": 0.06359291076660156,
      "step": 7025
    },
    {
      "epoch": 0.9790287744722358,
      "grad_norm": 0.2847253680229187,
      "learning_rate": 5.130986001469884e-08,
      "loss": 0.06658935546875,
      "step": 7026
    },
    {
      "epoch": 0.9791681181634502,
      "grad_norm": 0.6496864557266235,
      "learning_rate": 5.0637235196071555e-08,
      "loss": 0.12758827209472656,
      "step": 7027
    },
    {
      "epoch": 0.9793074618546646,
      "grad_norm": 0.27724024653434753,
      "learning_rate": 4.996904267039693e-08,
      "loss": 0.06882858276367188,
      "step": 7028
    },
    {
      "epoch": 0.9794468055458789,
      "grad_norm": 0.3606931269168854,
      "learning_rate": 4.9305282586136206e-08,
      "loss": 0.0836334228515625,
      "step": 7029
    },
    {
      "epoch": 0.9795861492370933,
      "grad_norm": 0.7000800371170044,
      "learning_rate": 4.864595509076031e-08,
      "loss": 0.10184764862060547,
      "step": 7030
    },
    {
      "epoch": 0.9797254929283077,
      "grad_norm": 0.4327346384525299,
      "learning_rate": 4.799106033075429e-08,
      "loss": 0.09087944030761719,
      "step": 7031
    },
    {
      "epoch": 0.9798648366195221,
      "grad_norm": 0.3313043713569641,
      "learning_rate": 4.734059845162175e-08,
      "loss": 0.0842437744140625,
      "step": 7032
    },
    {
      "epoch": 0.9800041803107364,
      "grad_norm": 0.2857504189014435,
      "learning_rate": 4.669456959788265e-08,
      "loss": 0.0750732421875,
      "step": 7033
    },
    {
      "epoch": 0.9801435240019508,
      "grad_norm": 0.23580637574195862,
      "learning_rate": 4.6052973913068844e-08,
      "loss": 0.07046031951904297,
      "step": 7034
    },
    {
      "epoch": 0.9802828676931652,
      "grad_norm": 0.45814013481140137,
      "learning_rate": 4.541581153972852e-08,
      "loss": 0.09259319305419922,
      "step": 7035
    },
    {
      "epoch": 0.9804222113843796,
      "grad_norm": 0.24139456450939178,
      "learning_rate": 4.478308261942177e-08,
      "loss": 0.06894397735595703,
      "step": 7036
    },
    {
      "epoch": 0.980561555075594,
      "grad_norm": 0.3430197536945343,
      "learning_rate": 4.415478729272949e-08,
      "loss": 0.07888317108154297,
      "step": 7037
    },
    {
      "epoch": 0.9807008987668083,
      "grad_norm": 0.3584829866886139,
      "learning_rate": 4.353092569924444e-08,
      "loss": 0.07424736022949219,
      "step": 7038
    },
    {
      "epoch": 0.9808402424580227,
      "grad_norm": 0.3139784038066864,
      "learning_rate": 4.2911497977573545e-08,
      "loss": 0.07613945007324219,
      "step": 7039
    },
    {
      "epoch": 0.9809795861492371,
      "grad_norm": 0.6756899952888489,
      "learning_rate": 4.2296504265340044e-08,
      "loss": 0.09279060363769531,
      "step": 7040
    },
    {
      "epoch": 0.9811189298404515,
      "grad_norm": 0.4448457956314087,
      "learning_rate": 4.1685944699181304e-08,
      "loss": 0.08668327331542969,
      "step": 7041
    },
    {
      "epoch": 0.9812582735316658,
      "grad_norm": 0.3114352524280548,
      "learning_rate": 4.107981941474881e-08,
      "loss": 0.07505226135253906,
      "step": 7042
    },
    {
      "epoch": 0.9813976172228802,
      "grad_norm": 0.2929372787475586,
      "learning_rate": 4.04781285467104e-08,
      "loss": 0.0663747787475586,
      "step": 7043
    },
    {
      "epoch": 0.9815369609140946,
      "grad_norm": 0.38055741786956787,
      "learning_rate": 3.9880872228748034e-08,
      "loss": 0.07517147064208984,
      "step": 7044
    },
    {
      "epoch": 0.981676304605309,
      "grad_norm": 0.5596561431884766,
      "learning_rate": 3.928805059356e-08,
      "loss": 0.09222793579101562,
      "step": 7045
    },
    {
      "epoch": 0.9818156482965233,
      "grad_norm": 0.24990394711494446,
      "learning_rate": 3.869966377285428e-08,
      "loss": 0.07199287414550781,
      "step": 7046
    },
    {
      "epoch": 0.9819549919877377,
      "grad_norm": 0.3485788106918335,
      "learning_rate": 3.8115711897359634e-08,
      "loss": 0.07266807556152344,
      "step": 7047
    },
    {
      "epoch": 0.9820943356789521,
      "grad_norm": 0.5842549204826355,
      "learning_rate": 3.753619509681672e-08,
      "loss": 0.101531982421875,
      "step": 7048
    },
    {
      "epoch": 0.9822336793701665,
      "grad_norm": 0.1712196171283722,
      "learning_rate": 3.696111349998255e-08,
      "loss": 0.05659675598144531,
      "step": 7049
    },
    {
      "epoch": 0.9823730230613809,
      "grad_norm": 0.4386426508426666,
      "learning_rate": 3.6390467234621585e-08,
      "loss": 0.10433197021484375,
      "step": 7050
    },
    {
      "epoch": 0.9825123667525952,
      "grad_norm": 0.33179235458374023,
      "learning_rate": 3.582425642752352e-08,
      "loss": 0.0924997329711914,
      "step": 7051
    },
    {
      "epoch": 0.9826517104438096,
      "grad_norm": 0.3954107165336609,
      "learning_rate": 3.52624812044855e-08,
      "loss": 0.08159446716308594,
      "step": 7052
    },
    {
      "epoch": 0.982791054135024,
      "grad_norm": 0.5345085859298706,
      "learning_rate": 3.470514169032102e-08,
      "loss": 0.08497428894042969,
      "step": 7053
    },
    {
      "epoch": 0.9829303978262384,
      "grad_norm": 0.29185932874679565,
      "learning_rate": 3.4152238008859915e-08,
      "loss": 0.07545089721679688,
      "step": 7054
    },
    {
      "epoch": 0.9830697415174527,
      "grad_norm": 0.3108827471733093,
      "learning_rate": 3.360377028294171e-08,
      "loss": 0.08596038818359375,
      "step": 7055
    },
    {
      "epoch": 0.9832090852086672,
      "grad_norm": 0.5638288855552673,
      "learning_rate": 3.305973863442669e-08,
      "loss": 0.08717155456542969,
      "step": 7056
    },
    {
      "epoch": 0.9833484288998816,
      "grad_norm": 0.3348013758659363,
      "learning_rate": 3.252014318418484e-08,
      "loss": 0.07555675506591797,
      "step": 7057
    },
    {
      "epoch": 0.983487772591096,
      "grad_norm": 0.2989587187767029,
      "learning_rate": 3.198498405210027e-08,
      "loss": 0.0787515640258789,
      "step": 7058
    },
    {
      "epoch": 0.9836271162823104,
      "grad_norm": 0.4358947277069092,
      "learning_rate": 3.145426135707563e-08,
      "loss": 0.0921783447265625,
      "step": 7059
    },
    {
      "epoch": 0.9837664599735247,
      "grad_norm": 0.2488207370042801,
      "learning_rate": 3.092797521702551e-08,
      "loss": 0.07392692565917969,
      "step": 7060
    },
    {
      "epoch": 0.9839058036647391,
      "grad_norm": 0.27952393889427185,
      "learning_rate": 3.040612574887636e-08,
      "loss": 0.06281471252441406,
      "step": 7061
    },
    {
      "epoch": 0.9840451473559535,
      "grad_norm": 0.46728116273880005,
      "learning_rate": 2.9888713068573215e-08,
      "loss": 0.09207344055175781,
      "step": 7062
    },
    {
      "epoch": 0.9841844910471679,
      "grad_norm": 0.7059627771377563,
      "learning_rate": 2.9375737291070795e-08,
      "loss": 0.1212167739868164,
      "step": 7063
    },
    {
      "epoch": 0.9843238347383823,
      "grad_norm": 0.4854995310306549,
      "learning_rate": 2.8867198530344587e-08,
      "loss": 0.09108161926269531,
      "step": 7064
    },
    {
      "epoch": 0.9844631784295966,
      "grad_norm": 0.21682116389274597,
      "learning_rate": 2.836309689937755e-08,
      "loss": 0.05723285675048828,
      "step": 7065
    },
    {
      "epoch": 0.984602522120811,
      "grad_norm": 0.33756276965141296,
      "learning_rate": 2.7863432510168988e-08,
      "loss": 0.07539939880371094,
      "step": 7066
    },
    {
      "epoch": 0.9847418658120254,
      "grad_norm": 0.514140248298645,
      "learning_rate": 2.7368205473734532e-08,
      "loss": 0.09599113464355469,
      "step": 7067
    },
    {
      "epoch": 0.9848812095032398,
      "grad_norm": 0.4274795651435852,
      "learning_rate": 2.6877415900103955e-08,
      "loss": 0.07802772521972656,
      "step": 7068
    },
    {
      "epoch": 0.9850205531944541,
      "grad_norm": 0.3368118107318878,
      "learning_rate": 2.6391063898314474e-08,
      "loss": 0.08564567565917969,
      "step": 7069
    },
    {
      "epoch": 0.9851598968856685,
      "grad_norm": 0.4978190064430237,
      "learning_rate": 2.5909149576428537e-08,
      "loss": 0.1035308837890625,
      "step": 7070
    },
    {
      "epoch": 0.9852992405768829,
      "grad_norm": 0.9163516759872437,
      "learning_rate": 2.5431673041509396e-08,
      "loss": 0.12821292877197266,
      "step": 7071
    },
    {
      "epoch": 0.9854385842680973,
      "grad_norm": 0.25151243805885315,
      "learning_rate": 2.495863439964774e-08,
      "loss": 0.07075381278991699,
      "step": 7072
    },
    {
      "epoch": 0.9855779279593117,
      "grad_norm": 0.3855830729007721,
      "learning_rate": 2.44900337559395e-08,
      "loss": 0.07012176513671875,
      "step": 7073
    },
    {
      "epoch": 0.985717271650526,
      "grad_norm": 0.6811702847480774,
      "learning_rate": 2.4025871214496954e-08,
      "loss": 0.12108612060546875,
      "step": 7074
    },
    {
      "epoch": 0.9858566153417404,
      "grad_norm": 0.44098055362701416,
      "learning_rate": 2.3566146878446495e-08,
      "loss": 0.08254623413085938,
      "step": 7075
    },
    {
      "epoch": 0.9859959590329548,
      "grad_norm": 0.6748990416526794,
      "learning_rate": 2.311086084992864e-08,
      "loss": 0.10392284393310547,
      "step": 7076
    },
    {
      "epoch": 0.9861353027241692,
      "grad_norm": 0.46223184466362,
      "learning_rate": 2.2660013230098032e-08,
      "loss": 0.09841346740722656,
      "step": 7077
    },
    {
      "epoch": 0.9862746464153835,
      "grad_norm": 0.2507038712501526,
      "learning_rate": 2.2213604119121214e-08,
      "loss": 0.06400299072265625,
      "step": 7078
    },
    {
      "epoch": 0.9864139901065979,
      "grad_norm": 0.5277556777000427,
      "learning_rate": 2.1771633616181066e-08,
      "loss": 0.10231876373291016,
      "step": 7079
    },
    {
      "epoch": 0.9865533337978123,
      "grad_norm": 0.27427002787590027,
      "learning_rate": 2.1334101819472375e-08,
      "loss": 0.07265853881835938,
      "step": 7080
    },
    {
      "epoch": 0.9866926774890267,
      "grad_norm": 0.4197658598423004,
      "learning_rate": 2.0901008826206272e-08,
      "loss": 0.07384395599365234,
      "step": 7081
    },
    {
      "epoch": 0.986832021180241,
      "grad_norm": 0.7394188642501831,
      "learning_rate": 2.047235473260578e-08,
      "loss": 0.10232353210449219,
      "step": 7082
    },
    {
      "epoch": 0.9869713648714554,
      "grad_norm": 0.37359046936035156,
      "learning_rate": 2.004813963390584e-08,
      "loss": 0.09153366088867188,
      "step": 7083
    },
    {
      "epoch": 0.9871107085626698,
      "grad_norm": 0.4021826386451721,
      "learning_rate": 1.9628363624362155e-08,
      "loss": 0.08776664733886719,
      "step": 7084
    },
    {
      "epoch": 0.9872500522538842,
      "grad_norm": 0.3616896867752075,
      "learning_rate": 1.921302679723569e-08,
      "loss": 0.07695770263671875,
      "step": 7085
    },
    {
      "epoch": 0.9873893959450986,
      "grad_norm": 0.33554959297180176,
      "learning_rate": 1.8802129244803735e-08,
      "loss": 0.08467292785644531,
      "step": 7086
    },
    {
      "epoch": 0.9875287396363129,
      "grad_norm": 0.4439508616924286,
      "learning_rate": 1.839567105836215e-08,
      "loss": 0.10301589965820312,
      "step": 7087
    },
    {
      "epoch": 0.9876680833275273,
      "grad_norm": 0.38142141699790955,
      "learning_rate": 1.7993652328214263e-08,
      "loss": 0.0683135986328125,
      "step": 7088
    },
    {
      "epoch": 0.9878074270187417,
      "grad_norm": 0.6153779625892639,
      "learning_rate": 1.7596073143677505e-08,
      "loss": 0.10979461669921875,
      "step": 7089
    },
    {
      "epoch": 0.9879467707099561,
      "grad_norm": 0.4699292480945587,
      "learning_rate": 1.720293359309011e-08,
      "loss": 0.08847618103027344,
      "step": 7090
    },
    {
      "epoch": 0.9880861144011704,
      "grad_norm": 0.5092649459838867,
      "learning_rate": 1.681423376379554e-08,
      "loss": 0.09864997863769531,
      "step": 7091
    },
    {
      "epoch": 0.9882254580923848,
      "grad_norm": 0.30132609605789185,
      "learning_rate": 1.6429973742153606e-08,
      "loss": 0.08182907104492188,
      "step": 7092
    },
    {
      "epoch": 0.9883648017835992,
      "grad_norm": 0.2984306216239929,
      "learning_rate": 1.6050153613538234e-08,
      "loss": 0.0735321044921875,
      "step": 7093
    },
    {
      "epoch": 0.9885041454748136,
      "grad_norm": 0.6130729913711548,
      "learning_rate": 1.567477346233748e-08,
      "loss": 0.11347103118896484,
      "step": 7094
    },
    {
      "epoch": 0.988643489166028,
      "grad_norm": 0.37042757868766785,
      "learning_rate": 1.5303833371953514e-08,
      "loss": 0.08262252807617188,
      "step": 7095
    },
    {
      "epoch": 0.9887828328572423,
      "grad_norm": 0.2572517395019531,
      "learning_rate": 1.4937333424798194e-08,
      "loss": 0.07491302490234375,
      "step": 7096
    },
    {
      "epoch": 0.9889221765484568,
      "grad_norm": 0.23394711315631866,
      "learning_rate": 1.457527370230194e-08,
      "loss": 0.0747365951538086,
      "step": 7097
    },
    {
      "epoch": 0.9890615202396712,
      "grad_norm": 0.5420699715614319,
      "learning_rate": 1.4217654284904846e-08,
      "loss": 0.09369659423828125,
      "step": 7098
    },
    {
      "epoch": 0.9892008639308856,
      "grad_norm": 0.44607287645339966,
      "learning_rate": 1.3864475252058918e-08,
      "loss": 0.0908660888671875,
      "step": 7099
    },
    {
      "epoch": 0.9893402076221,
      "grad_norm": 0.4076556861400604,
      "learning_rate": 1.3515736682239156e-08,
      "loss": 0.09599685668945312,
      "step": 7100
    },
    {
      "epoch": 0.9894795513133143,
      "grad_norm": 0.4245949983596802,
      "learning_rate": 1.3171438652921365e-08,
      "loss": 0.08612251281738281,
      "step": 7101
    },
    {
      "epoch": 0.9896188950045287,
      "grad_norm": 0.46599850058555603,
      "learning_rate": 1.2831581240602131e-08,
      "loss": 0.09012413024902344,
      "step": 7102
    },
    {
      "epoch": 0.9897582386957431,
      "grad_norm": 0.6813873648643494,
      "learning_rate": 1.2496164520792165e-08,
      "loss": 0.12961578369140625,
      "step": 7103
    },
    {
      "epoch": 0.9898975823869575,
      "grad_norm": 0.30176764726638794,
      "learning_rate": 1.2165188568011854e-08,
      "loss": 0.08404541015625,
      "step": 7104
    },
    {
      "epoch": 0.9900369260781718,
      "grad_norm": 0.47852760553359985,
      "learning_rate": 1.1838653455797933e-08,
      "loss": 0.08382797241210938,
      "step": 7105
    },
    {
      "epoch": 0.9901762697693862,
      "grad_norm": 0.3482668399810791,
      "learning_rate": 1.1516559256694594e-08,
      "loss": 0.08401203155517578,
      "step": 7106
    },
    {
      "epoch": 0.9903156134606006,
      "grad_norm": 0.5583547949790955,
      "learning_rate": 1.1198906042269032e-08,
      "loss": 0.11604690551757812,
      "step": 7107
    },
    {
      "epoch": 0.990454957151815,
      "grad_norm": 0.32300105690956116,
      "learning_rate": 1.0885693883093685e-08,
      "loss": 0.06453895568847656,
      "step": 7108
    },
    {
      "epoch": 0.9905943008430294,
      "grad_norm": 0.4005473554134369,
      "learning_rate": 1.0576922848759552e-08,
      "loss": 0.07660126686096191,
      "step": 7109
    },
    {
      "epoch": 0.9907336445342437,
      "grad_norm": 0.3001633286476135,
      "learning_rate": 1.0272593007865094e-08,
      "loss": 0.07167816162109375,
      "step": 7110
    },
    {
      "epoch": 0.9908729882254581,
      "grad_norm": 0.35614943504333496,
      "learning_rate": 9.972704428027335e-09,
      "loss": 0.07771682739257812,
      "step": 7111
    },
    {
      "epoch": 0.9910123319166725,
      "grad_norm": 0.3928835988044739,
      "learning_rate": 9.677257175875199e-09,
      "loss": 0.08368492126464844,
      "step": 7112
    },
    {
      "epoch": 0.9911516756078869,
      "grad_norm": 0.3923397660255432,
      "learning_rate": 9.38625131704951e-09,
      "loss": 0.08047294616699219,
      "step": 7113
    },
    {
      "epoch": 0.9912910192991012,
      "grad_norm": 0.4559655487537384,
      "learning_rate": 9.099686916205219e-09,
      "loss": 0.10528564453125,
      "step": 7114
    },
    {
      "epoch": 0.9914303629903156,
      "grad_norm": 0.3142716884613037,
      "learning_rate": 8.817564037009174e-09,
      "loss": 0.09048271179199219,
      "step": 7115
    },
    {
      "epoch": 0.99156970668153,
      "grad_norm": 0.6222264766693115,
      "learning_rate": 8.539882742146788e-09,
      "loss": 0.1301126480102539,
      "step": 7116
    },
    {
      "epoch": 0.9917090503727444,
      "grad_norm": 0.3332425057888031,
      "learning_rate": 8.266643093306492e-09,
      "loss": 0.07227134704589844,
      "step": 7117
    },
    {
      "epoch": 0.9918483940639587,
      "grad_norm": 0.4975399374961853,
      "learning_rate": 7.997845151201944e-09,
      "loss": 0.09442520141601562,
      "step": 7118
    },
    {
      "epoch": 0.9919877377551731,
      "grad_norm": 0.5111963152885437,
      "learning_rate": 7.733488975549819e-09,
      "loss": 0.08022308349609375,
      "step": 7119
    },
    {
      "epoch": 0.9921270814463875,
      "grad_norm": 0.30542102456092834,
      "learning_rate": 7.473574625085355e-09,
      "loss": 0.07353878021240234,
      "step": 7120
    },
    {
      "epoch": 0.9922664251376019,
      "grad_norm": 0.5740736722946167,
      "learning_rate": 7.218102157555695e-09,
      "loss": 0.10625457763671875,
      "step": 7121
    },
    {
      "epoch": 0.9924057688288163,
      "grad_norm": 0.2600778341293335,
      "learning_rate": 6.967071629719879e-09,
      "loss": 0.07552909851074219,
      "step": 7122
    },
    {
      "epoch": 0.9925451125200306,
      "grad_norm": 0.3793385624885559,
      "learning_rate": 6.720483097353292e-09,
      "loss": 0.08315372467041016,
      "step": 7123
    },
    {
      "epoch": 0.992684456211245,
      "grad_norm": 0.42303621768951416,
      "learning_rate": 6.478336615238779e-09,
      "loss": 0.08804702758789062,
      "step": 7124
    },
    {
      "epoch": 0.9928237999024594,
      "grad_norm": 0.38640889525413513,
      "learning_rate": 6.240632237179967e-09,
      "loss": 0.080322265625,
      "step": 7125
    },
    {
      "epoch": 0.9929631435936738,
      "grad_norm": 0.31561338901519775,
      "learning_rate": 6.0073700159879455e-09,
      "loss": 0.08009529113769531,
      "step": 7126
    },
    {
      "epoch": 0.9931024872848881,
      "grad_norm": 0.3362888991832733,
      "learning_rate": 5.778550003485706e-09,
      "loss": 0.08336448669433594,
      "step": 7127
    },
    {
      "epoch": 0.9932418309761025,
      "grad_norm": 0.42725372314453125,
      "learning_rate": 5.5541722505148e-09,
      "loss": 0.09050273895263672,
      "step": 7128
    },
    {
      "epoch": 0.9933811746673169,
      "grad_norm": 0.3879239559173584,
      "learning_rate": 5.334236806926462e-09,
      "loss": 0.09314918518066406,
      "step": 7129
    },
    {
      "epoch": 0.9935205183585313,
      "grad_norm": 0.3044133484363556,
      "learning_rate": 5.11874372158383e-09,
      "loss": 0.0707387924194336,
      "step": 7130
    },
    {
      "epoch": 0.9936598620497457,
      "grad_norm": 0.2373647540807724,
      "learning_rate": 4.907693042366379e-09,
      "loss": 0.06121635437011719,
      "step": 7131
    },
    {
      "epoch": 0.99379920574096,
      "grad_norm": 0.7730281949043274,
      "learning_rate": 4.70108481616327e-09,
      "loss": 0.11113739013671875,
      "step": 7132
    },
    {
      "epoch": 0.9939385494321744,
      "grad_norm": 0.3169896602630615,
      "learning_rate": 4.498919088877784e-09,
      "loss": 0.08699417114257812,
      "step": 7133
    },
    {
      "epoch": 0.9940778931233888,
      "grad_norm": 0.5029552578926086,
      "learning_rate": 4.301195905427324e-09,
      "loss": 0.08768463134765625,
      "step": 7134
    },
    {
      "epoch": 0.9942172368146032,
      "grad_norm": 0.6178176403045654,
      "learning_rate": 4.107915309743416e-09,
      "loss": 0.11826133728027344,
      "step": 7135
    },
    {
      "epoch": 0.9943565805058175,
      "grad_norm": 0.26564306020736694,
      "learning_rate": 3.919077344765043e-09,
      "loss": 0.0717763900756836,
      "step": 7136
    },
    {
      "epoch": 0.994495924197032,
      "grad_norm": 0.5027791261672974,
      "learning_rate": 3.734682052449756e-09,
      "loss": 0.10825157165527344,
      "step": 7137
    },
    {
      "epoch": 0.9946352678882464,
      "grad_norm": 0.5332139134407043,
      "learning_rate": 3.5547294737670046e-09,
      "loss": 0.10050201416015625,
      "step": 7138
    },
    {
      "epoch": 0.9947746115794608,
      "grad_norm": 0.4502524137496948,
      "learning_rate": 3.3792196486959193e-09,
      "loss": 0.09029960632324219,
      "step": 7139
    },
    {
      "epoch": 0.9949139552706752,
      "grad_norm": 0.588246762752533,
      "learning_rate": 3.2081526162319744e-09,
      "loss": 0.119659423828125,
      "step": 7140
    },
    {
      "epoch": 0.9950532989618895,
      "grad_norm": 0.5456023812294006,
      "learning_rate": 3.0415284143803234e-09,
      "loss": 0.09521961212158203,
      "step": 7141
    },
    {
      "epoch": 0.9951926426531039,
      "grad_norm": 0.5365489721298218,
      "learning_rate": 2.879347080164685e-09,
      "loss": 0.0863037109375,
      "step": 7142
    },
    {
      "epoch": 0.9953319863443183,
      "grad_norm": 0.48316916823387146,
      "learning_rate": 2.721608649614016e-09,
      "loss": 0.08998489379882812,
      "step": 7143
    },
    {
      "epoch": 0.9954713300355327,
      "grad_norm": 0.6606982350349426,
      "learning_rate": 2.5683131577780574e-09,
      "loss": 0.10013771057128906,
      "step": 7144
    },
    {
      "epoch": 0.995610673726747,
      "grad_norm": 0.22666063904762268,
      "learning_rate": 2.4194606387140106e-09,
      "loss": 0.06587696075439453,
      "step": 7145
    },
    {
      "epoch": 0.9957500174179614,
      "grad_norm": 0.27906206250190735,
      "learning_rate": 2.275051125490979e-09,
      "loss": 0.07310104370117188,
      "step": 7146
    },
    {
      "epoch": 0.9958893611091758,
      "grad_norm": 0.3456067740917206,
      "learning_rate": 2.1350846501988487e-09,
      "loss": 0.07865715026855469,
      "step": 7147
    },
    {
      "epoch": 0.9960287048003902,
      "grad_norm": 0.3336813747882843,
      "learning_rate": 1.9995612439305255e-09,
      "loss": 0.06795501708984375,
      "step": 7148
    },
    {
      "epoch": 0.9961680484916046,
      "grad_norm": 0.716064453125,
      "learning_rate": 1.8684809367974786e-09,
      "loss": 0.1285228729248047,
      "step": 7149
    },
    {
      "epoch": 0.9963073921828189,
      "grad_norm": 0.6322847008705139,
      "learning_rate": 1.741843757920858e-09,
      "loss": 0.12994861602783203,
      "step": 7150
    },
    {
      "epoch": 0.9964467358740333,
      "grad_norm": 0.3212963938713074,
      "learning_rate": 1.6196497354403761e-09,
      "loss": 0.08428096771240234,
      "step": 7151
    },
    {
      "epoch": 0.9965860795652477,
      "grad_norm": 0.32924893498420715,
      "learning_rate": 1.5018988965032067e-09,
      "loss": 0.07504653930664062,
      "step": 7152
    },
    {
      "epoch": 0.9967254232564621,
      "grad_norm": 0.3272203505039215,
      "learning_rate": 1.3885912672706447e-09,
      "loss": 0.07892417907714844,
      "step": 7153
    },
    {
      "epoch": 0.9968647669476765,
      "grad_norm": 0.3179364502429962,
      "learning_rate": 1.279726872918108e-09,
      "loss": 0.07855224609375,
      "step": 7154
    },
    {
      "epoch": 0.9970041106388908,
      "grad_norm": 0.3892156481742859,
      "learning_rate": 1.1753057376306942e-09,
      "loss": 0.08907508850097656,
      "step": 7155
    },
    {
      "epoch": 0.9971434543301052,
      "grad_norm": 0.462395578622818,
      "learning_rate": 1.0753278846076242e-09,
      "loss": 0.10816287994384766,
      "step": 7156
    },
    {
      "epoch": 0.9972827980213196,
      "grad_norm": 0.25442808866500854,
      "learning_rate": 9.79793336066681e-10,
      "loss": 0.06982994079589844,
      "step": 7157
    },
    {
      "epoch": 0.997422141712534,
      "grad_norm": 0.4455505609512329,
      "learning_rate": 8.887021132286677e-10,
      "loss": 0.08577728271484375,
      "step": 7158
    },
    {
      "epoch": 0.9975614854037483,
      "grad_norm": 0.5211091041564941,
      "learning_rate": 8.020542363329497e-10,
      "loss": 0.1019287109375,
      "step": 7159
    },
    {
      "epoch": 0.9977008290949627,
      "grad_norm": 0.5843652486801147,
      "learning_rate": 7.198497246307945e-10,
      "loss": 0.10460472106933594,
      "step": 7160
    },
    {
      "epoch": 0.9978401727861771,
      "grad_norm": 0.35286614298820496,
      "learning_rate": 6.420885963875912e-10,
      "loss": 0.07641410827636719,
      "step": 7161
    },
    {
      "epoch": 0.9979795164773915,
      "grad_norm": 0.6045248508453369,
      "learning_rate": 5.687708688806304e-10,
      "loss": 0.10607051849365234,
      "step": 7162
    },
    {
      "epoch": 0.9981188601686058,
      "grad_norm": 0.5210958123207092,
      "learning_rate": 4.99896558396884e-10,
      "loss": 0.10666561126708984,
      "step": 7163
    },
    {
      "epoch": 0.9982582038598202,
      "grad_norm": 0.41777899861335754,
      "learning_rate": 4.3546568023966616e-10,
      "loss": 0.08942031860351562,
      "step": 7164
    },
    {
      "epoch": 0.9983975475510346,
      "grad_norm": 0.1885034143924713,
      "learning_rate": 3.754782487241926e-10,
      "loss": 0.05783271789550781,
      "step": 7165
    },
    {
      "epoch": 0.998536891242249,
      "grad_norm": 0.3365144729614258,
      "learning_rate": 3.1993427717758043e-10,
      "loss": 0.07682609558105469,
      "step": 7166
    },
    {
      "epoch": 0.9986762349334634,
      "grad_norm": 0.3610086739063263,
      "learning_rate": 2.688337779410688e-10,
      "loss": 0.07254791259765625,
      "step": 7167
    },
    {
      "epoch": 0.9988155786246777,
      "grad_norm": 0.18554919958114624,
      "learning_rate": 2.2217676237001883e-10,
      "loss": 0.052176475524902344,
      "step": 7168
    },
    {
      "epoch": 0.9989549223158921,
      "grad_norm": 0.7821488976478577,
      "learning_rate": 1.7996324082725224e-10,
      "loss": 0.14793014526367188,
      "step": 7169
    },
    {
      "epoch": 0.9990942660071065,
      "grad_norm": 1.0546404123306274,
      "learning_rate": 1.4219322269193313e-10,
      "loss": 0.14149093627929688,
      "step": 7170
    },
    {
      "epoch": 0.9992336096983209,
      "grad_norm": 0.34660789370536804,
      "learning_rate": 1.0886671635956803e-10,
      "loss": 0.08515548706054688,
      "step": 7171
    },
    {
      "epoch": 0.9993729533895352,
      "grad_norm": 0.23722459375858307,
      "learning_rate": 7.998372922868313e-11,
      "loss": 0.06633567810058594,
      "step": 7172
    },
    {
      "epoch": 0.9995122970807496,
      "grad_norm": 0.6010448932647705,
      "learning_rate": 5.554426772080845e-11,
      "loss": 0.10683441162109375,
      "step": 7173
    },
    {
      "epoch": 0.999651640771964,
      "grad_norm": 0.356526643037796,
      "learning_rate": 3.554833726493456e-11,
      "loss": 0.0800333023071289,
      "step": 7174
    },
    {
      "epoch": 0.9997909844631784,
      "grad_norm": 0.5378692150115967,
      "learning_rate": 1.999594230417401e-11,
      "loss": 0.09458732604980469,
      "step": 7175
    },
    {
      "epoch": 0.9999303281543928,
      "grad_norm": 0.48508283495903015,
      "learning_rate": 8.887086291320401e-12,
      "loss": 0.10258102416992188,
      "step": 7176
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1569210290908813,
      "learning_rate": 2.2217716955097445e-12,
      "loss": 0.14859390258789062,
      "step": 7177
    },
    {
      "epoch": 1.0,
      "step": 7177,
      "total_flos": 5.326107668090192e+19,
      "train_loss": 0.10390519324295099,
      "train_runtime": 71823.5105,
      "train_samples_per_second": 25.579,
      "train_steps_per_second": 0.1
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 7177,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.326107668090192e+19,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}