{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 7397,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00013519679583593868,
      "grad_norm": 10.255340576171875,
      "learning_rate": 0.0,
      "loss": 0.69140625,
      "step": 1
    },
    {
      "epoch": 0.00027039359167187736,
      "grad_norm": 10.892648696899414,
      "learning_rate": 5.859375e-08,
      "loss": 0.69140625,
      "step": 2
    },
    {
      "epoch": 0.0004055903875078161,
      "grad_norm": 10.60339641571045,
      "learning_rate": 1.171875e-07,
      "loss": 0.69140625,
      "step": 3
    },
    {
      "epoch": 0.0005407871833437547,
      "grad_norm": 10.201833724975586,
      "learning_rate": 1.7578125e-07,
      "loss": 0.69140625,
      "step": 4
    },
    {
      "epoch": 0.0006759839791796934,
      "grad_norm": 10.66333293914795,
      "learning_rate": 2.34375e-07,
      "loss": 0.69140625,
      "step": 5
    },
    {
      "epoch": 0.0008111807750156322,
      "grad_norm": 10.464648246765137,
      "learning_rate": 2.9296875000000003e-07,
      "loss": 0.69140625,
      "step": 6
    },
    {
      "epoch": 0.0009463775708515709,
      "grad_norm": 10.621415138244629,
      "learning_rate": 3.515625e-07,
      "loss": 0.69140625,
      "step": 7
    },
    {
      "epoch": 0.0010815743666875094,
      "grad_norm": 10.700018882751465,
      "learning_rate": 4.1015625e-07,
      "loss": 0.69140625,
      "step": 8
    },
    {
      "epoch": 0.0012167711625234482,
      "grad_norm": 10.476875305175781,
      "learning_rate": 4.6875e-07,
      "loss": 0.69140625,
      "step": 9
    },
    {
      "epoch": 0.0013519679583593868,
      "grad_norm": 10.614548683166504,
      "learning_rate": 5.2734375e-07,
      "loss": 0.69140625,
      "step": 10
    },
    {
      "epoch": 0.0014871647541953256,
      "grad_norm": 10.9356050491333,
      "learning_rate": 5.859375000000001e-07,
      "loss": 0.69140625,
      "step": 11
    },
    {
      "epoch": 0.0016223615500312644,
      "grad_norm": 11.010387420654297,
      "learning_rate": 6.4453125e-07,
      "loss": 0.69140625,
      "step": 12
    },
    {
      "epoch": 0.001757558345867203,
      "grad_norm": 10.474565505981445,
      "learning_rate": 7.03125e-07,
      "loss": 0.6888427734375,
      "step": 13
    },
    {
      "epoch": 0.0018927551417031417,
      "grad_norm": 11.14912223815918,
      "learning_rate": 7.6171875e-07,
      "loss": 0.68798828125,
      "step": 14
    },
    {
      "epoch": 0.0020279519375390803,
      "grad_norm": 10.89780330657959,
      "learning_rate": 8.203125e-07,
      "loss": 0.68792724609375,
      "step": 15
    },
    {
      "epoch": 0.002163148733375019,
      "grad_norm": 10.518672943115234,
      "learning_rate": 8.7890625e-07,
      "loss": 0.68798828125,
      "step": 16
    },
    {
      "epoch": 0.002298345529210958,
      "grad_norm": 10.299038887023926,
      "learning_rate": 9.375e-07,
      "loss": 0.6878662109375,
      "step": 17
    },
    {
      "epoch": 0.0024335423250468964,
      "grad_norm": 10.66930866241455,
      "learning_rate": 9.9609375e-07,
      "loss": 0.683837890625,
      "step": 18
    },
    {
      "epoch": 0.002568739120882835,
      "grad_norm": 10.663459777832031,
      "learning_rate": 1.0546875e-06,
      "loss": 0.68377685546875,
      "step": 19
    },
    {
      "epoch": 0.0027039359167187736,
      "grad_norm": 10.752281188964844,
      "learning_rate": 1.11328125e-06,
      "loss": 0.68316650390625,
      "step": 20
    },
    {
      "epoch": 0.0028391327125547126,
      "grad_norm": 10.501089096069336,
      "learning_rate": 1.1718750000000001e-06,
      "loss": 0.68310546875,
      "step": 21
    },
    {
      "epoch": 0.002974329508390651,
      "grad_norm": 10.665032386779785,
      "learning_rate": 1.23046875e-06,
      "loss": 0.68096923828125,
      "step": 22
    },
    {
      "epoch": 0.0031095263042265897,
      "grad_norm": 9.829651832580566,
      "learning_rate": 1.2890625e-06,
      "loss": 0.68170166015625,
      "step": 23
    },
    {
      "epoch": 0.0032447231000625287,
      "grad_norm": 10.657003402709961,
      "learning_rate": 1.34765625e-06,
      "loss": 0.68011474609375,
      "step": 24
    },
    {
      "epoch": 0.0033799198958984673,
      "grad_norm": 10.138802528381348,
      "learning_rate": 1.40625e-06,
      "loss": 0.66058349609375,
      "step": 25
    },
    {
      "epoch": 0.003515116691734406,
      "grad_norm": 9.759446144104004,
      "learning_rate": 1.46484375e-06,
      "loss": 0.66119384765625,
      "step": 26
    },
    {
      "epoch": 0.0036503134875703444,
      "grad_norm": 10.706598281860352,
      "learning_rate": 1.5234375e-06,
      "loss": 0.65679931640625,
      "step": 27
    },
    {
      "epoch": 0.0037855102834062834,
      "grad_norm": 9.85043716430664,
      "learning_rate": 1.5820312500000001e-06,
      "loss": 0.65850830078125,
      "step": 28
    },
    {
      "epoch": 0.003920707079242222,
      "grad_norm": 9.268957138061523,
      "learning_rate": 1.640625e-06,
      "loss": 0.6593017578125,
      "step": 29
    },
    {
      "epoch": 0.004055903875078161,
      "grad_norm": 9.965956687927246,
      "learning_rate": 1.69921875e-06,
      "loss": 0.6513671875,
      "step": 30
    },
    {
      "epoch": 0.004191100670914099,
      "grad_norm": 9.81229305267334,
      "learning_rate": 1.7578125e-06,
      "loss": 0.65045166015625,
      "step": 31
    },
    {
      "epoch": 0.004326297466750038,
      "grad_norm": 10.022581100463867,
      "learning_rate": 1.81640625e-06,
      "loss": 0.6463623046875,
      "step": 32
    },
    {
      "epoch": 0.004461494262585977,
      "grad_norm": 9.77592658996582,
      "learning_rate": 1.875e-06,
      "loss": 0.63909912109375,
      "step": 33
    },
    {
      "epoch": 0.004596691058421916,
      "grad_norm": 9.745993614196777,
      "learning_rate": 1.93359375e-06,
      "loss": 0.61114501953125,
      "step": 34
    },
    {
      "epoch": 0.004731887854257854,
      "grad_norm": 9.962318420410156,
      "learning_rate": 1.9921875e-06,
      "loss": 0.596435546875,
      "step": 35
    },
    {
      "epoch": 0.004867084650093793,
      "grad_norm": 9.316697120666504,
      "learning_rate": 2.05078125e-06,
      "loss": 0.59722900390625,
      "step": 36
    },
    {
      "epoch": 0.005002281445929731,
      "grad_norm": 10.021162986755371,
      "learning_rate": 2.109375e-06,
      "loss": 0.583251953125,
      "step": 37
    },
    {
      "epoch": 0.00513747824176567,
      "grad_norm": 8.996946334838867,
      "learning_rate": 2.16796875e-06,
      "loss": 0.58868408203125,
      "step": 38
    },
    {
      "epoch": 0.0052726750376016086,
      "grad_norm": 9.7518310546875,
      "learning_rate": 2.2265625e-06,
      "loss": 0.57568359375,
      "step": 39
    },
    {
      "epoch": 0.005407871833437547,
      "grad_norm": 8.269822120666504,
      "learning_rate": 2.28515625e-06,
      "loss": 0.5845947265625,
      "step": 40
    },
    {
      "epoch": 0.005543068629273487,
      "grad_norm": 8.733393669128418,
      "learning_rate": 2.3437500000000002e-06,
      "loss": 0.5721435546875,
      "step": 41
    },
    {
      "epoch": 0.005678265425109425,
      "grad_norm": 8.683788299560547,
      "learning_rate": 2.40234375e-06,
      "loss": 0.56158447265625,
      "step": 42
    },
    {
      "epoch": 0.005813462220945364,
      "grad_norm": 8.420989036560059,
      "learning_rate": 2.4609375e-06,
      "loss": 0.55609130859375,
      "step": 43
    },
    {
      "epoch": 0.005948659016781302,
      "grad_norm": 9.148038864135742,
      "learning_rate": 2.5195312500000003e-06,
      "loss": 0.53692626953125,
      "step": 44
    },
    {
      "epoch": 0.006083855812617241,
      "grad_norm": 8.447755813598633,
      "learning_rate": 2.578125e-06,
      "loss": 0.5377197265625,
      "step": 45
    },
    {
      "epoch": 0.006219052608453179,
      "grad_norm": 8.374534606933594,
      "learning_rate": 2.63671875e-06,
      "loss": 0.531005859375,
      "step": 46
    },
    {
      "epoch": 0.006354249404289118,
      "grad_norm": 8.3541841506958,
      "learning_rate": 2.6953125e-06,
      "loss": 0.51959228515625,
      "step": 47
    },
    {
      "epoch": 0.006489446200125057,
      "grad_norm": 8.071650505065918,
      "learning_rate": 2.75390625e-06,
      "loss": 0.51409912109375,
      "step": 48
    },
    {
      "epoch": 0.006624642995960996,
      "grad_norm": 8.372121810913086,
      "learning_rate": 2.8125e-06,
      "loss": 0.49664306640625,
      "step": 49
    },
    {
      "epoch": 0.006759839791796935,
      "grad_norm": 7.649506568908691,
      "learning_rate": 2.87109375e-06,
      "loss": 0.501007080078125,
      "step": 50
    },
    {
      "epoch": 0.006895036587632873,
      "grad_norm": 7.1636643409729,
      "learning_rate": 2.9296875e-06,
      "loss": 0.4979248046875,
      "step": 51
    },
    {
      "epoch": 0.007030233383468812,
      "grad_norm": 8.19225788116455,
      "learning_rate": 2.9882812500000002e-06,
      "loss": 0.463165283203125,
      "step": 52
    },
    {
      "epoch": 0.00716543017930475,
      "grad_norm": 7.3633198738098145,
      "learning_rate": 3.046875e-06,
      "loss": 0.46734619140625,
      "step": 53
    },
    {
      "epoch": 0.007300626975140689,
      "grad_norm": 6.5898332595825195,
      "learning_rate": 3.10546875e-06,
      "loss": 0.47491455078125,
      "step": 54
    },
    {
      "epoch": 0.007435823770976628,
      "grad_norm": 7.3989129066467285,
      "learning_rate": 3.1640625000000003e-06,
      "loss": 0.4415283203125,
      "step": 55
    },
    {
      "epoch": 0.007571020566812567,
      "grad_norm": 7.007622241973877,
      "learning_rate": 3.22265625e-06,
      "loss": 0.43731689453125,
      "step": 56
    },
    {
      "epoch": 0.007706217362648505,
      "grad_norm": 6.6982245445251465,
      "learning_rate": 3.28125e-06,
      "loss": 0.43072509765625,
      "step": 57
    },
    {
      "epoch": 0.007841414158484444,
      "grad_norm": 6.32157039642334,
      "learning_rate": 3.3398437500000003e-06,
      "loss": 0.426788330078125,
      "step": 58
    },
    {
      "epoch": 0.007976610954320383,
      "grad_norm": 6.236541748046875,
      "learning_rate": 3.3984375e-06,
      "loss": 0.415924072265625,
      "step": 59
    },
    {
      "epoch": 0.008111807750156321,
      "grad_norm": 5.945745468139648,
      "learning_rate": 3.45703125e-06,
      "loss": 0.411834716796875,
      "step": 60
    },
    {
      "epoch": 0.00824700454599226,
      "grad_norm": 6.02635383605957,
      "learning_rate": 3.515625e-06,
      "loss": 0.395233154296875,
      "step": 61
    },
    {
      "epoch": 0.008382201341828198,
      "grad_norm": 5.314901351928711,
      "learning_rate": 3.57421875e-06,
      "loss": 0.409942626953125,
      "step": 62
    },
    {
      "epoch": 0.008517398137664138,
      "grad_norm": 6.0076003074646,
      "learning_rate": 3.6328125e-06,
      "loss": 0.370849609375,
      "step": 63
    },
    {
      "epoch": 0.008652594933500075,
      "grad_norm": 5.681765556335449,
      "learning_rate": 3.69140625e-06,
      "loss": 0.368499755859375,
      "step": 64
    },
    {
      "epoch": 0.008787791729336015,
      "grad_norm": 5.089432239532471,
      "learning_rate": 3.75e-06,
      "loss": 0.378570556640625,
      "step": 65
    },
    {
      "epoch": 0.008922988525171954,
      "grad_norm": 5.389379978179932,
      "learning_rate": 3.8085937500000002e-06,
      "loss": 0.35345458984375,
      "step": 66
    },
    {
      "epoch": 0.009058185321007892,
      "grad_norm": 4.493364334106445,
      "learning_rate": 3.8671875e-06,
      "loss": 0.376190185546875,
      "step": 67
    },
    {
      "epoch": 0.009193382116843831,
      "grad_norm": 4.8557562828063965,
      "learning_rate": 3.92578125e-06,
      "loss": 0.345947265625,
      "step": 68
    },
    {
      "epoch": 0.009328578912679769,
      "grad_norm": 3.7139086723327637,
      "learning_rate": 3.984375e-06,
      "loss": 0.381378173828125,
      "step": 69
    },
    {
      "epoch": 0.009463775708515709,
      "grad_norm": 4.213521480560303,
      "learning_rate": 4.0429687500000004e-06,
      "loss": 0.346466064453125,
      "step": 70
    },
    {
      "epoch": 0.009598972504351646,
      "grad_norm": 3.739476442337036,
      "learning_rate": 4.1015625e-06,
      "loss": 0.356231689453125,
      "step": 71
    },
    {
      "epoch": 0.009734169300187586,
      "grad_norm": 4.2015557289123535,
      "learning_rate": 4.16015625e-06,
      "loss": 0.319122314453125,
      "step": 72
    },
    {
      "epoch": 0.009869366096023523,
      "grad_norm": 3.780320882797241,
      "learning_rate": 4.21875e-06,
      "loss": 0.3252105712890625,
      "step": 73
    },
    {
      "epoch": 0.010004562891859463,
      "grad_norm": 3.6442320346832275,
      "learning_rate": 4.27734375e-06,
      "loss": 0.3150482177734375,
      "step": 74
    },
    {
      "epoch": 0.010139759687695402,
      "grad_norm": 3.2569968700408936,
      "learning_rate": 4.3359375e-06,
      "loss": 0.3176116943359375,
      "step": 75
    },
    {
      "epoch": 0.01027495648353134,
      "grad_norm": 3.287576675415039,
      "learning_rate": 4.3945312500000005e-06,
      "loss": 0.3000030517578125,
      "step": 76
    },
    {
      "epoch": 0.01041015327936728,
      "grad_norm": 3.3834593296051025,
      "learning_rate": 4.453125e-06,
      "loss": 0.2772979736328125,
      "step": 77
    },
    {
      "epoch": 0.010545350075203217,
      "grad_norm": 2.0733957290649414,
      "learning_rate": 4.51171875e-06,
      "loss": 0.3407135009765625,
      "step": 78
    },
    {
      "epoch": 0.010680546871039157,
      "grad_norm": 3.333418369293213,
      "learning_rate": 4.5703125e-06,
      "loss": 0.246673583984375,
      "step": 79
    },
    {
      "epoch": 0.010815743666875094,
      "grad_norm": 3.1509034633636475,
      "learning_rate": 4.62890625e-06,
      "loss": 0.2439117431640625,
      "step": 80
    },
    {
      "epoch": 0.010950940462711034,
      "grad_norm": 2.4727230072021484,
      "learning_rate": 4.6875000000000004e-06,
      "loss": 0.2735137939453125,
      "step": 81
    },
    {
      "epoch": 0.011086137258546973,
      "grad_norm": 2.4484498500823975,
      "learning_rate": 4.74609375e-06,
      "loss": 0.259796142578125,
      "step": 82
    },
    {
      "epoch": 0.01122133405438291,
      "grad_norm": 2.03958797454834,
      "learning_rate": 4.8046875e-06,
      "loss": 0.2721710205078125,
      "step": 83
    },
    {
      "epoch": 0.01135653085021885,
      "grad_norm": 1.9284204244613647,
      "learning_rate": 4.86328125e-06,
      "loss": 0.266845703125,
      "step": 84
    },
    {
      "epoch": 0.011491727646054788,
      "grad_norm": 2.418253183364868,
      "learning_rate": 4.921875e-06,
      "loss": 0.216827392578125,
      "step": 85
    },
    {
      "epoch": 0.011626924441890727,
      "grad_norm": 1.4608980417251587,
      "learning_rate": 4.98046875e-06,
      "loss": 0.2739410400390625,
      "step": 86
    },
    {
      "epoch": 0.011762121237726665,
      "grad_norm": 1.9801719188690186,
      "learning_rate": 5.0390625000000005e-06,
      "loss": 0.2206573486328125,
      "step": 87
    },
    {
      "epoch": 0.011897318033562605,
      "grad_norm": 1.5869640111923218,
      "learning_rate": 5.09765625e-06,
      "loss": 0.2376556396484375,
      "step": 88
    },
    {
      "epoch": 0.012032514829398544,
      "grad_norm": 1.680525302886963,
      "learning_rate": 5.15625e-06,
      "loss": 0.215789794921875,
      "step": 89
    },
    {
      "epoch": 0.012167711625234482,
      "grad_norm": 0.4875149726867676,
      "learning_rate": 5.21484375e-06,
      "loss": 0.301513671875,
      "step": 90
    },
    {
      "epoch": 0.012302908421070421,
      "grad_norm": 0.4136301875114441,
      "learning_rate": 5.2734375e-06,
      "loss": 0.29526519775390625,
      "step": 91
    },
    {
      "epoch": 0.012438105216906359,
      "grad_norm": 1.1960359811782837,
      "learning_rate": 5.3320312500000004e-06,
      "loss": 0.219207763671875,
      "step": 92
    },
    {
      "epoch": 0.012573302012742298,
      "grad_norm": 0.6939961314201355,
      "learning_rate": 5.390625e-06,
      "loss": 0.25130462646484375,
      "step": 93
    },
    {
      "epoch": 0.012708498808578236,
      "grad_norm": 0.48938074707984924,
      "learning_rate": 5.44921875e-06,
      "loss": 0.25917816162109375,
      "step": 94
    },
    {
      "epoch": 0.012843695604414175,
      "grad_norm": 1.3375189304351807,
      "learning_rate": 5.5078125e-06,
      "loss": 0.17076873779296875,
      "step": 95
    },
    {
      "epoch": 0.012978892400250115,
      "grad_norm": 0.21868669986724854,
      "learning_rate": 5.56640625e-06,
      "loss": 0.26607513427734375,
      "step": 96
    },
    {
      "epoch": 0.013114089196086053,
      "grad_norm": 0.37352919578552246,
      "learning_rate": 5.625e-06,
      "loss": 0.243377685546875,
      "step": 97
    },
    {
      "epoch": 0.013249285991921992,
      "grad_norm": 0.1982581615447998,
      "learning_rate": 5.6835937500000005e-06,
      "loss": 0.25643157958984375,
      "step": 98
    },
    {
      "epoch": 0.01338448278775793,
      "grad_norm": 0.8889334201812744,
      "learning_rate": 5.7421875e-06,
      "loss": 0.18682861328125,
      "step": 99
    },
    {
      "epoch": 0.01351967958359387,
      "grad_norm": 0.14042750000953674,
      "learning_rate": 5.80078125e-06,
      "loss": 0.2624664306640625,
      "step": 100
    },
    {
      "epoch": 0.013654876379429807,
      "grad_norm": 0.462993860244751,
      "learning_rate": 5.859375e-06,
      "loss": 0.21479034423828125,
      "step": 101
    },
    {
      "epoch": 0.013790073175265746,
      "grad_norm": 0.6529134511947632,
      "learning_rate": 5.91796875e-06,
      "loss": 0.191650390625,
      "step": 102
    },
    {
      "epoch": 0.013925269971101686,
      "grad_norm": 1.1363776922225952,
      "learning_rate": 5.9765625000000004e-06,
      "loss": 0.1736297607421875,
      "step": 103
    },
    {
      "epoch": 0.014060466766937623,
      "grad_norm": 0.6305656433105469,
      "learning_rate": 6.03515625e-06,
      "loss": 0.245880126953125,
      "step": 104
    },
    {
      "epoch": 0.014195663562773563,
      "grad_norm": 1.309865117073059,
      "learning_rate": 6.09375e-06,
      "loss": 0.16274261474609375,
      "step": 105
    },
    {
      "epoch": 0.0143308603586095,
      "grad_norm": 7.6630096435546875,
      "learning_rate": 6.15234375e-06,
      "loss": 0.1968994140625,
      "step": 106
    },
    {
      "epoch": 0.01446605715444544,
      "grad_norm": 6.000809192657471,
      "learning_rate": 6.2109375e-06,
      "loss": 0.23690032958984375,
      "step": 107
    },
    {
      "epoch": 0.014601253950281378,
      "grad_norm": 1.0207101106643677,
      "learning_rate": 6.26953125e-06,
      "loss": 0.21165084838867188,
      "step": 108
    },
    {
      "epoch": 0.014736450746117317,
      "grad_norm": 1.7974658012390137,
      "learning_rate": 6.3281250000000005e-06,
      "loss": 0.20142364501953125,
      "step": 109
    },
    {
      "epoch": 0.014871647541953257,
      "grad_norm": 1.8257349729537964,
      "learning_rate": 6.38671875e-06,
      "loss": 0.13892745971679688,
      "step": 110
    },
    {
      "epoch": 0.015006844337789194,
      "grad_norm": 1.1191425323486328,
      "learning_rate": 6.4453125e-06,
      "loss": 0.18349456787109375,
      "step": 111
    },
    {
      "epoch": 0.015142041133625134,
      "grad_norm": 1.1117126941680908,
      "learning_rate": 6.50390625e-06,
      "loss": 0.17523574829101562,
      "step": 112
    },
    {
      "epoch": 0.015277237929461071,
      "grad_norm": 1.6576142311096191,
      "learning_rate": 6.5625e-06,
      "loss": 0.21729660034179688,
      "step": 113
    },
    {
      "epoch": 0.01541243472529701,
      "grad_norm": 1.931712031364441,
      "learning_rate": 6.6210937500000004e-06,
      "loss": 0.16382980346679688,
      "step": 114
    },
    {
      "epoch": 0.015547631521132949,
      "grad_norm": 2.039163827896118,
      "learning_rate": 6.679687500000001e-06,
      "loss": 0.20510101318359375,
      "step": 115
    },
    {
      "epoch": 0.015682828316968888,
      "grad_norm": 0.7490442991256714,
      "learning_rate": 6.73828125e-06,
      "loss": 0.14907455444335938,
      "step": 116
    },
    {
      "epoch": 0.015818025112804827,
      "grad_norm": 1.7908422946929932,
      "learning_rate": 6.796875e-06,
      "loss": 0.14217758178710938,
      "step": 117
    },
    {
      "epoch": 0.015953221908640767,
      "grad_norm": 2.6014528274536133,
      "learning_rate": 6.85546875e-06,
      "loss": 0.17675018310546875,
      "step": 118
    },
    {
      "epoch": 0.016088418704476703,
      "grad_norm": 1.5167361497879028,
      "learning_rate": 6.9140625e-06,
      "loss": 0.1981201171875,
      "step": 119
    },
    {
      "epoch": 0.016223615500312642,
      "grad_norm": 3.2530808448791504,
      "learning_rate": 6.9726562500000005e-06,
      "loss": 0.15254592895507812,
      "step": 120
    },
    {
      "epoch": 0.01635881229614858,
      "grad_norm": 1.5909086465835571,
      "learning_rate": 7.03125e-06,
      "loss": 0.12428665161132812,
      "step": 121
    },
    {
      "epoch": 0.01649400909198452,
      "grad_norm": 0.829672634601593,
      "learning_rate": 7.08984375e-06,
      "loss": 0.18768692016601562,
      "step": 122
    },
    {
      "epoch": 0.016629205887820457,
      "grad_norm": 5.443108081817627,
      "learning_rate": 7.1484375e-06,
      "loss": 0.19033050537109375,
      "step": 123
    },
    {
      "epoch": 0.016764402683656397,
      "grad_norm": 5.772006034851074,
      "learning_rate": 7.20703125e-06,
      "loss": 0.25315093994140625,
      "step": 124
    },
    {
      "epoch": 0.016899599479492336,
      "grad_norm": 2.46826171875,
      "learning_rate": 7.265625e-06,
      "loss": 0.17133712768554688,
      "step": 125
    },
    {
      "epoch": 0.017034796275328275,
      "grad_norm": 4.126870155334473,
      "learning_rate": 7.3242187500000006e-06,
      "loss": 0.22769737243652344,
      "step": 126
    },
    {
      "epoch": 0.017169993071164215,
      "grad_norm": 4.511805057525635,
      "learning_rate": 7.3828125e-06,
      "loss": 0.20490646362304688,
      "step": 127
    },
    {
      "epoch": 0.01730518986700015,
      "grad_norm": 1.9656882286071777,
      "learning_rate": 7.44140625e-06,
      "loss": 0.16280364990234375,
      "step": 128
    },
    {
      "epoch": 0.01744038666283609,
      "grad_norm": 2.4700119495391846,
      "learning_rate": 7.5e-06,
      "loss": 0.12662124633789062,
      "step": 129
    },
    {
      "epoch": 0.01757558345867203,
      "grad_norm": 2.5934202671051025,
      "learning_rate": 7.55859375e-06,
      "loss": 0.22777557373046875,
      "step": 130
    },
    {
      "epoch": 0.01771078025450797,
      "grad_norm": 1.3021607398986816,
      "learning_rate": 7.6171875000000005e-06,
      "loss": 0.15658187866210938,
      "step": 131
    },
    {
      "epoch": 0.01784597705034391,
      "grad_norm": 1.5999529361724854,
      "learning_rate": 7.67578125e-06,
      "loss": 0.16962051391601562,
      "step": 132
    },
    {
      "epoch": 0.017981173846179845,
      "grad_norm": 2.5243165493011475,
      "learning_rate": 7.734375e-06,
      "loss": 0.20207977294921875,
      "step": 133
    },
    {
      "epoch": 0.018116370642015784,
      "grad_norm": 0.7855330109596252,
      "learning_rate": 7.792968750000001e-06,
      "loss": 0.10043716430664062,
      "step": 134
    },
    {
      "epoch": 0.018251567437851723,
      "grad_norm": 0.9127803444862366,
      "learning_rate": 7.8515625e-06,
      "loss": 0.15300750732421875,
      "step": 135
    },
    {
      "epoch": 0.018386764233687663,
      "grad_norm": 1.8121592998504639,
      "learning_rate": 7.91015625e-06,
      "loss": 0.1618938446044922,
      "step": 136
    },
    {
      "epoch": 0.0185219610295236,
      "grad_norm": 1.502724051475525,
      "learning_rate": 7.96875e-06,
      "loss": 0.1136932373046875,
      "step": 137
    },
    {
      "epoch": 0.018657157825359538,
      "grad_norm": 4.455524444580078,
      "learning_rate": 8.02734375e-06,
      "loss": 0.1859893798828125,
      "step": 138
    },
    {
      "epoch": 0.018792354621195478,
      "grad_norm": 3.528498888015747,
      "learning_rate": 8.085937500000001e-06,
      "loss": 0.13478469848632812,
      "step": 139
    },
    {
      "epoch": 0.018927551417031417,
      "grad_norm": 0.6664129495620728,
      "learning_rate": 8.14453125e-06,
      "loss": 0.1552581787109375,
      "step": 140
    },
    {
      "epoch": 0.019062748212867357,
      "grad_norm": 1.9160336256027222,
      "learning_rate": 8.203125e-06,
      "loss": 0.11435508728027344,
      "step": 141
    },
    {
      "epoch": 0.019197945008703293,
      "grad_norm": 2.0927252769470215,
      "learning_rate": 8.26171875e-06,
      "loss": 0.14061355590820312,
      "step": 142
    },
    {
      "epoch": 0.019333141804539232,
      "grad_norm": 0.8855465650558472,
      "learning_rate": 8.3203125e-06,
      "loss": 0.17259597778320312,
      "step": 143
    },
    {
      "epoch": 0.01946833860037517,
      "grad_norm": 2.003345012664795,
      "learning_rate": 8.37890625e-06,
      "loss": 0.16485595703125,
      "step": 144
    },
    {
      "epoch": 0.01960353539621111,
      "grad_norm": 1.8325819969177246,
      "learning_rate": 8.4375e-06,
      "loss": 0.17378997802734375,
      "step": 145
    },
    {
      "epoch": 0.019738732192047047,
      "grad_norm": 0.9648417830467224,
      "learning_rate": 8.49609375e-06,
      "loss": 0.1458587646484375,
      "step": 146
    },
    {
      "epoch": 0.019873928987882986,
      "grad_norm": 1.092206597328186,
      "learning_rate": 8.5546875e-06,
      "loss": 0.139373779296875,
      "step": 147
    },
    {
      "epoch": 0.020009125783718926,
      "grad_norm": 2.4173755645751953,
      "learning_rate": 8.61328125e-06,
      "loss": 0.17518234252929688,
      "step": 148
    },
    {
      "epoch": 0.020144322579554865,
      "grad_norm": 1.0320264101028442,
      "learning_rate": 8.671875e-06,
      "loss": 0.19220733642578125,
      "step": 149
    },
    {
      "epoch": 0.020279519375390805,
      "grad_norm": 1.2767515182495117,
      "learning_rate": 8.73046875e-06,
      "loss": 0.183380126953125,
      "step": 150
    },
    {
      "epoch": 0.02041471617122674,
      "grad_norm": 1.3617839813232422,
      "learning_rate": 8.789062500000001e-06,
      "loss": 0.14226531982421875,
      "step": 151
    },
    {
      "epoch": 0.02054991296706268,
      "grad_norm": 1.1166024208068848,
      "learning_rate": 8.84765625e-06,
      "loss": 0.07677078247070312,
      "step": 152
    },
    {
      "epoch": 0.02068510976289862,
      "grad_norm": 1.617315411567688,
      "learning_rate": 8.90625e-06,
      "loss": 0.13544845581054688,
      "step": 153
    },
    {
      "epoch": 0.02082030655873456,
      "grad_norm": 1.0335773229599,
      "learning_rate": 8.96484375e-06,
      "loss": 0.12308311462402344,
      "step": 154
    },
    {
      "epoch": 0.0209555033545705,
      "grad_norm": 0.4307231903076172,
      "learning_rate": 9.0234375e-06,
      "loss": 0.06521034240722656,
      "step": 155
    },
    {
      "epoch": 0.021090700150406434,
      "grad_norm": 2.0309338569641113,
      "learning_rate": 9.082031250000001e-06,
      "loss": 0.12531280517578125,
      "step": 156
    },
    {
      "epoch": 0.021225896946242374,
      "grad_norm": 0.3681405186653137,
      "learning_rate": 9.140625e-06,
      "loss": 0.11372756958007812,
      "step": 157
    },
    {
      "epoch": 0.021361093742078313,
      "grad_norm": 0.9422040581703186,
      "learning_rate": 9.19921875e-06,
      "loss": 0.1118621826171875,
      "step": 158
    },
    {
      "epoch": 0.021496290537914253,
      "grad_norm": 1.0494142770767212,
      "learning_rate": 9.2578125e-06,
      "loss": 0.14452552795410156,
      "step": 159
    },
    {
      "epoch": 0.02163148733375019,
      "grad_norm": 0.5977887511253357,
      "learning_rate": 9.31640625e-06,
      "loss": 0.15737152099609375,
      "step": 160
    },
    {
      "epoch": 0.021766684129586128,
      "grad_norm": 1.815451979637146,
      "learning_rate": 9.375000000000001e-06,
      "loss": 0.15938377380371094,
      "step": 161
    },
    {
      "epoch": 0.021901880925422067,
      "grad_norm": 1.7669764757156372,
      "learning_rate": 9.43359375e-06,
      "loss": 0.13509750366210938,
      "step": 162
    },
    {
      "epoch": 0.022037077721258007,
      "grad_norm": 0.8248637914657593,
      "learning_rate": 9.4921875e-06,
      "loss": 0.11051368713378906,
      "step": 163
    },
    {
      "epoch": 0.022172274517093946,
      "grad_norm": 0.9761935472488403,
      "learning_rate": 9.55078125e-06,
      "loss": 0.0999603271484375,
      "step": 164
    },
    {
      "epoch": 0.022307471312929882,
      "grad_norm": 1.2425165176391602,
      "learning_rate": 9.609375e-06,
      "loss": 0.10451698303222656,
      "step": 165
    },
    {
      "epoch": 0.02244266810876582,
      "grad_norm": 1.2928447723388672,
      "learning_rate": 9.66796875e-06,
      "loss": 0.14514541625976562,
      "step": 166
    },
    {
      "epoch": 0.02257786490460176,
      "grad_norm": 0.67860347032547,
      "learning_rate": 9.7265625e-06,
      "loss": 0.1896514892578125,
      "step": 167
    },
    {
      "epoch": 0.0227130617004377,
      "grad_norm": 0.5783755779266357,
      "learning_rate": 9.78515625e-06,
      "loss": 0.1671886444091797,
      "step": 168
    },
    {
      "epoch": 0.02284825849627364,
      "grad_norm": 2.328864574432373,
      "learning_rate": 9.84375e-06,
      "loss": 0.19832992553710938,
      "step": 169
    },
    {
      "epoch": 0.022983455292109576,
      "grad_norm": 0.7465437650680542,
      "learning_rate": 9.90234375e-06,
      "loss": 0.097137451171875,
      "step": 170
    },
    {
      "epoch": 0.023118652087945515,
      "grad_norm": 0.47173887491226196,
      "learning_rate": 9.9609375e-06,
      "loss": 0.09764289855957031,
      "step": 171
    },
    {
      "epoch": 0.023253848883781455,
      "grad_norm": 0.4709882438182831,
      "learning_rate": 1.001953125e-05,
      "loss": 0.0984792709350586,
      "step": 172
    },
    {
      "epoch": 0.023389045679617394,
      "grad_norm": 1.0793781280517578,
      "learning_rate": 1.0078125000000001e-05,
      "loss": 0.16312408447265625,
      "step": 173
    },
    {
      "epoch": 0.02352424247545333,
      "grad_norm": 0.39833131432533264,
      "learning_rate": 1.013671875e-05,
      "loss": 0.08628463745117188,
      "step": 174
    },
    {
      "epoch": 0.02365943927128927,
      "grad_norm": 1.002116084098816,
      "learning_rate": 1.01953125e-05,
      "loss": 0.125457763671875,
      "step": 175
    },
    {
      "epoch": 0.02379463606712521,
      "grad_norm": 1.1118419170379639,
      "learning_rate": 1.025390625e-05,
      "loss": 0.09708786010742188,
      "step": 176
    },
    {
      "epoch": 0.02392983286296115,
      "grad_norm": 0.7531958818435669,
      "learning_rate": 1.03125e-05,
      "loss": 0.1278400421142578,
      "step": 177
    },
    {
      "epoch": 0.024065029658797088,
      "grad_norm": 0.866471529006958,
      "learning_rate": 1.0371093750000001e-05,
      "loss": 0.14974689483642578,
      "step": 178
    },
    {
      "epoch": 0.024200226454633024,
      "grad_norm": 1.445363163948059,
      "learning_rate": 1.04296875e-05,
      "loss": 0.15588951110839844,
      "step": 179
    },
    {
      "epoch": 0.024335423250468963,
      "grad_norm": 0.8952563405036926,
      "learning_rate": 1.048828125e-05,
      "loss": 0.11579322814941406,
      "step": 180
    },
    {
      "epoch": 0.024470620046304903,
      "grad_norm": 1.4956169128417969,
      "learning_rate": 1.0546875e-05,
      "loss": 0.16865921020507812,
      "step": 181
    },
    {
      "epoch": 0.024605816842140842,
      "grad_norm": 0.9510613679885864,
      "learning_rate": 1.060546875e-05,
      "loss": 0.11063385009765625,
      "step": 182
    },
    {
      "epoch": 0.024741013637976778,
      "grad_norm": 0.6214854121208191,
      "learning_rate": 1.0664062500000001e-05,
      "loss": 0.1398639678955078,
      "step": 183
    },
    {
      "epoch": 0.024876210433812718,
      "grad_norm": 0.7311033010482788,
      "learning_rate": 1.072265625e-05,
      "loss": 0.11818122863769531,
      "step": 184
    },
    {
      "epoch": 0.025011407229648657,
      "grad_norm": 1.4142619371414185,
      "learning_rate": 1.078125e-05,
      "loss": 0.1476421356201172,
      "step": 185
    },
    {
      "epoch": 0.025146604025484597,
      "grad_norm": 1.758416771888733,
      "learning_rate": 1.083984375e-05,
      "loss": 0.13826560974121094,
      "step": 186
    },
    {
      "epoch": 0.025281800821320536,
      "grad_norm": 2.2945053577423096,
      "learning_rate": 1.08984375e-05,
      "loss": 0.1462249755859375,
      "step": 187
    },
    {
      "epoch": 0.025416997617156472,
      "grad_norm": 1.3065208196640015,
      "learning_rate": 1.095703125e-05,
      "loss": 0.12948226928710938,
      "step": 188
    },
    {
      "epoch": 0.02555219441299241,
      "grad_norm": 2.0528178215026855,
      "learning_rate": 1.1015625e-05,
      "loss": 0.1477680206298828,
      "step": 189
    },
    {
      "epoch": 0.02568739120882835,
      "grad_norm": 1.039638876914978,
      "learning_rate": 1.1074218750000001e-05,
      "loss": 0.08852577209472656,
      "step": 190
    },
    {
      "epoch": 0.02582258800466429,
      "grad_norm": 0.7299044728279114,
      "learning_rate": 1.11328125e-05,
      "loss": 0.10076332092285156,
      "step": 191
    },
    {
      "epoch": 0.02595778480050023,
      "grad_norm": 1.6996442079544067,
      "learning_rate": 1.119140625e-05,
      "loss": 0.14342498779296875,
      "step": 192
    },
    {
      "epoch": 0.026092981596336166,
      "grad_norm": 0.8143086433410645,
      "learning_rate": 1.125e-05,
      "loss": 0.09293174743652344,
      "step": 193
    },
    {
      "epoch": 0.026228178392172105,
      "grad_norm": 1.3359870910644531,
      "learning_rate": 1.130859375e-05,
      "loss": 0.12241172790527344,
      "step": 194
    },
    {
      "epoch": 0.026363375188008045,
      "grad_norm": 1.9629666805267334,
      "learning_rate": 1.1367187500000001e-05,
      "loss": 0.15043258666992188,
      "step": 195
    },
    {
      "epoch": 0.026498571983843984,
      "grad_norm": 0.39924514293670654,
      "learning_rate": 1.142578125e-05,
      "loss": 0.08169937133789062,
      "step": 196
    },
    {
      "epoch": 0.02663376877967992,
      "grad_norm": 1.4615709781646729,
      "learning_rate": 1.1484375e-05,
      "loss": 0.08629798889160156,
      "step": 197
    },
    {
      "epoch": 0.02676896557551586,
      "grad_norm": 0.9599554538726807,
      "learning_rate": 1.154296875e-05,
      "loss": 0.11198616027832031,
      "step": 198
    },
    {
      "epoch": 0.0269041623713518,
      "grad_norm": 2.5499606132507324,
      "learning_rate": 1.16015625e-05,
      "loss": 0.13975954055786133,
      "step": 199
    },
    {
      "epoch": 0.02703935916718774,
      "grad_norm": 3.666128158569336,
      "learning_rate": 1.1660156250000001e-05,
      "loss": 0.22464752197265625,
      "step": 200
    },
    {
      "epoch": 0.027174555963023678,
      "grad_norm": 0.8818196654319763,
      "learning_rate": 1.171875e-05,
      "loss": 0.11071205139160156,
      "step": 201
    },
    {
      "epoch": 0.027309752758859614,
      "grad_norm": 0.9403067827224731,
      "learning_rate": 1.177734375e-05,
      "loss": 0.11906719207763672,
      "step": 202
    },
    {
      "epoch": 0.027444949554695553,
      "grad_norm": 0.9900426864624023,
      "learning_rate": 1.18359375e-05,
      "loss": 0.07979202270507812,
      "step": 203
    },
    {
      "epoch": 0.027580146350531493,
      "grad_norm": 0.8115234971046448,
      "learning_rate": 1.189453125e-05,
      "loss": 0.11782455444335938,
      "step": 204
    },
    {
      "epoch": 0.027715343146367432,
      "grad_norm": 0.45454850792884827,
      "learning_rate": 1.1953125000000001e-05,
      "loss": 0.09505510330200195,
      "step": 205
    },
    {
      "epoch": 0.02785053994220337,
      "grad_norm": 1.3852804899215698,
      "learning_rate": 1.201171875e-05,
      "loss": 0.07553958892822266,
      "step": 206
    },
    {
      "epoch": 0.027985736738039307,
      "grad_norm": 2.505516767501831,
      "learning_rate": 1.20703125e-05,
      "loss": 0.2158184051513672,
      "step": 207
    },
    {
      "epoch": 0.028120933533875247,
      "grad_norm": 1.5190612077713013,
      "learning_rate": 1.212890625e-05,
      "loss": 0.14534378051757812,
      "step": 208
    },
    {
      "epoch": 0.028256130329711186,
      "grad_norm": 0.30262491106987,
      "learning_rate": 1.21875e-05,
      "loss": 0.11007308959960938,
      "step": 209
    },
    {
      "epoch": 0.028391327125547126,
      "grad_norm": 2.6057796478271484,
      "learning_rate": 1.224609375e-05,
      "loss": 0.20543670654296875,
      "step": 210
    },
    {
      "epoch": 0.02852652392138306,
      "grad_norm": 0.42562001943588257,
      "learning_rate": 1.23046875e-05,
      "loss": 0.10055351257324219,
      "step": 211
    },
    {
      "epoch": 0.028661720717219,
      "grad_norm": 0.7556141018867493,
      "learning_rate": 1.2363281250000001e-05,
      "loss": 0.08016586303710938,
      "step": 212
    },
    {
      "epoch": 0.02879691751305494,
      "grad_norm": 0.3173273205757141,
      "learning_rate": 1.2421875e-05,
      "loss": 0.11756706237792969,
      "step": 213
    },
    {
      "epoch": 0.02893211430889088,
      "grad_norm": 0.25473323464393616,
      "learning_rate": 1.248046875e-05,
      "loss": 0.07790184020996094,
      "step": 214
    },
    {
      "epoch": 0.02906731110472682,
      "grad_norm": 0.4120863378047943,
      "learning_rate": 1.25390625e-05,
      "loss": 0.10120487213134766,
      "step": 215
    },
    {
      "epoch": 0.029202507900562755,
      "grad_norm": 0.38646212220191956,
      "learning_rate": 1.259765625e-05,
      "loss": 0.1260519027709961,
      "step": 216
    },
    {
      "epoch": 0.029337704696398695,
      "grad_norm": 0.46892306208610535,
      "learning_rate": 1.2656250000000001e-05,
      "loss": 0.10096931457519531,
      "step": 217
    },
    {
      "epoch": 0.029472901492234634,
      "grad_norm": 1.1855638027191162,
      "learning_rate": 1.271484375e-05,
      "loss": 0.08810997009277344,
      "step": 218
    },
    {
      "epoch": 0.029608098288070574,
      "grad_norm": 0.3958824574947357,
      "learning_rate": 1.27734375e-05,
      "loss": 0.09965038299560547,
      "step": 219
    },
    {
      "epoch": 0.029743295083906513,
      "grad_norm": 0.7300817370414734,
      "learning_rate": 1.283203125e-05,
      "loss": 0.07414054870605469,
      "step": 220
    },
    {
      "epoch": 0.02987849187974245,
      "grad_norm": 1.615120768547058,
      "learning_rate": 1.2890625e-05,
      "loss": 0.10107898712158203,
      "step": 221
    },
    {
      "epoch": 0.03001368867557839,
      "grad_norm": 0.8337105512619019,
      "learning_rate": 1.2949218750000001e-05,
      "loss": 0.1640777587890625,
      "step": 222
    },
    {
      "epoch": 0.030148885471414328,
      "grad_norm": 0.8477329015731812,
      "learning_rate": 1.30078125e-05,
      "loss": 0.07065868377685547,
      "step": 223
    },
    {
      "epoch": 0.030284082267250267,
      "grad_norm": 0.8389713764190674,
      "learning_rate": 1.306640625e-05,
      "loss": 0.08960819244384766,
      "step": 224
    },
    {
      "epoch": 0.030419279063086203,
      "grad_norm": 0.8623062968254089,
      "learning_rate": 1.3125e-05,
      "loss": 0.09574508666992188,
      "step": 225
    },
    {
      "epoch": 0.030554475858922143,
      "grad_norm": 1.7276952266693115,
      "learning_rate": 1.318359375e-05,
      "loss": 0.1452322006225586,
      "step": 226
    },
    {
      "epoch": 0.030689672654758082,
      "grad_norm": 2.4967522621154785,
      "learning_rate": 1.3242187500000001e-05,
      "loss": 0.15061664581298828,
      "step": 227
    },
    {
      "epoch": 0.03082486945059402,
      "grad_norm": 0.8809515237808228,
      "learning_rate": 1.330078125e-05,
      "loss": 0.11009407043457031,
      "step": 228
    },
    {
      "epoch": 0.03096006624642996,
      "grad_norm": 1.7495862245559692,
      "learning_rate": 1.3359375000000001e-05,
      "loss": 0.10504150390625,
      "step": 229
    },
    {
      "epoch": 0.031095263042265897,
      "grad_norm": 1.010521411895752,
      "learning_rate": 1.341796875e-05,
      "loss": 0.06667327880859375,
      "step": 230
    },
    {
      "epoch": 0.031230459838101837,
      "grad_norm": 0.7836189270019531,
      "learning_rate": 1.34765625e-05,
      "loss": 0.09214496612548828,
      "step": 231
    },
    {
      "epoch": 0.031365656633937776,
      "grad_norm": 0.7970775961875916,
      "learning_rate": 1.353515625e-05,
      "loss": 0.13988971710205078,
      "step": 232
    },
    {
      "epoch": 0.03150085342977371,
      "grad_norm": 3.22719407081604,
      "learning_rate": 1.359375e-05,
      "loss": 0.1107487678527832,
      "step": 233
    },
    {
      "epoch": 0.031636050225609655,
      "grad_norm": 4.931375503540039,
      "learning_rate": 1.3652343750000001e-05,
      "loss": 0.14489173889160156,
      "step": 234
    },
    {
      "epoch": 0.03177124702144559,
      "grad_norm": 1.9380743503570557,
      "learning_rate": 1.37109375e-05,
      "loss": 0.1752605438232422,
      "step": 235
    },
    {
      "epoch": 0.031906443817281534,
      "grad_norm": 0.9338816404342651,
      "learning_rate": 1.376953125e-05,
      "loss": 0.11135673522949219,
      "step": 236
    },
    {
      "epoch": 0.03204164061311747,
      "grad_norm": 0.9291431307792664,
      "learning_rate": 1.3828125e-05,
      "loss": 0.11193370819091797,
      "step": 237
    },
    {
      "epoch": 0.032176837408953406,
      "grad_norm": 0.6870521306991577,
      "learning_rate": 1.388671875e-05,
      "loss": 0.10838890075683594,
      "step": 238
    },
    {
      "epoch": 0.03231203420478935,
      "grad_norm": 0.7145580053329468,
      "learning_rate": 1.3945312500000001e-05,
      "loss": 0.09445381164550781,
      "step": 239
    },
    {
      "epoch": 0.032447231000625285,
      "grad_norm": 0.21662800014019012,
      "learning_rate": 1.400390625e-05,
      "loss": 0.0929708480834961,
      "step": 240
    },
    {
      "epoch": 0.03258242779646122,
      "grad_norm": 1.7928955554962158,
      "learning_rate": 1.40625e-05,
      "loss": 0.1270885467529297,
      "step": 241
    },
    {
      "epoch": 0.03271762459229716,
      "grad_norm": 1.30325186252594,
      "learning_rate": 1.412109375e-05,
      "loss": 0.08766555786132812,
      "step": 242
    },
    {
      "epoch": 0.0328528213881331,
      "grad_norm": 5.986721515655518,
      "learning_rate": 1.41796875e-05,
      "loss": 0.09464168548583984,
      "step": 243
    },
    {
      "epoch": 0.03298801818396904,
      "grad_norm": 8.504470825195312,
      "learning_rate": 1.4238281250000001e-05,
      "loss": 0.1564788818359375,
      "step": 244
    },
    {
      "epoch": 0.03312321497980498,
      "grad_norm": 1.1269668340682983,
      "learning_rate": 1.4296875e-05,
      "loss": 0.06866836547851562,
      "step": 245
    },
    {
      "epoch": 0.033258411775640914,
      "grad_norm": 2.1340441703796387,
      "learning_rate": 1.435546875e-05,
      "loss": 0.1367034912109375,
      "step": 246
    },
    {
      "epoch": 0.03339360857147686,
      "grad_norm": 1.495689034461975,
      "learning_rate": 1.44140625e-05,
      "loss": 0.11866950988769531,
      "step": 247
    },
    {
      "epoch": 0.03352880536731279,
      "grad_norm": 0.45152899622917175,
      "learning_rate": 1.447265625e-05,
      "loss": 0.0792856216430664,
      "step": 248
    },
    {
      "epoch": 0.033664002163148736,
      "grad_norm": 0.5372416973114014,
      "learning_rate": 1.453125e-05,
      "loss": 0.08746719360351562,
      "step": 249
    },
    {
      "epoch": 0.03379919895898467,
      "grad_norm": 0.32935744524002075,
      "learning_rate": 1.458984375e-05,
      "loss": 0.10184097290039062,
      "step": 250
    },
    {
      "epoch": 0.03393439575482061,
      "grad_norm": 0.4165351688861847,
      "learning_rate": 1.4648437500000001e-05,
      "loss": 0.10245704650878906,
      "step": 251
    },
    {
      "epoch": 0.03406959255065655,
      "grad_norm": 0.3766869604587555,
      "learning_rate": 1.470703125e-05,
      "loss": 0.0938405990600586,
      "step": 252
    },
    {
      "epoch": 0.03420478934649249,
      "grad_norm": 0.4120754301548004,
      "learning_rate": 1.4765625e-05,
      "loss": 0.13604450225830078,
      "step": 253
    },
    {
      "epoch": 0.03433998614232843,
      "grad_norm": 0.5975791811943054,
      "learning_rate": 1.482421875e-05,
      "loss": 0.08067655563354492,
      "step": 254
    },
    {
      "epoch": 0.034475182938164366,
      "grad_norm": 0.8303422331809998,
      "learning_rate": 1.48828125e-05,
      "loss": 0.11191844940185547,
      "step": 255
    },
    {
      "epoch": 0.0346103797340003,
      "grad_norm": 0.38064515590667725,
      "learning_rate": 1.4941406250000001e-05,
      "loss": 0.1566314697265625,
      "step": 256
    },
    {
      "epoch": 0.034745576529836245,
      "grad_norm": 2.072690010070801,
      "learning_rate": 1.5e-05,
      "loss": 0.10210418701171875,
      "step": 257
    },
    {
      "epoch": 0.03488077332567218,
      "grad_norm": 0.5118513703346252,
      "learning_rate": 1.505859375e-05,
      "loss": 0.11151313781738281,
      "step": 258
    },
    {
      "epoch": 0.03501597012150812,
      "grad_norm": 0.8528406620025635,
      "learning_rate": 1.51171875e-05,
      "loss": 0.10950851440429688,
      "step": 259
    },
    {
      "epoch": 0.03515116691734406,
      "grad_norm": 0.3790138363838196,
      "learning_rate": 1.517578125e-05,
      "loss": 0.07526636123657227,
      "step": 260
    },
    {
      "epoch": 0.035286363713179995,
      "grad_norm": 1.4999076128005981,
      "learning_rate": 1.5234375000000001e-05,
      "loss": 0.1217947006225586,
      "step": 261
    },
    {
      "epoch": 0.03542156050901594,
      "grad_norm": 1.1162285804748535,
      "learning_rate": 1.529296875e-05,
      "loss": 0.15423822402954102,
      "step": 262
    },
    {
      "epoch": 0.035556757304851874,
      "grad_norm": 0.17001789808273315,
      "learning_rate": 1.53515625e-05,
      "loss": 0.08667278289794922,
      "step": 263
    },
    {
      "epoch": 0.03569195410068782,
      "grad_norm": 1.3194137811660767,
      "learning_rate": 1.541015625e-05,
      "loss": 0.11114215850830078,
      "step": 264
    },
    {
      "epoch": 0.03582715089652375,
      "grad_norm": 0.6932324171066284,
      "learning_rate": 1.546875e-05,
      "loss": 0.10363578796386719,
      "step": 265
    },
    {
      "epoch": 0.03596234769235969,
      "grad_norm": 0.8319248557090759,
      "learning_rate": 1.552734375e-05,
      "loss": 0.1459178924560547,
      "step": 266
    },
    {
      "epoch": 0.03609754448819563,
      "grad_norm": 1.287670373916626,
      "learning_rate": 1.5585937500000002e-05,
      "loss": 0.08773994445800781,
      "step": 267
    },
    {
      "epoch": 0.03623274128403157,
      "grad_norm": 0.494658887386322,
      "learning_rate": 1.564453125e-05,
      "loss": 0.08005809783935547,
      "step": 268
    },
    {
      "epoch": 0.036367938079867504,
      "grad_norm": 0.7535601258277893,
      "learning_rate": 1.5703125e-05,
      "loss": 0.1199045181274414,
      "step": 269
    },
    {
      "epoch": 0.03650313487570345,
      "grad_norm": 0.422181636095047,
      "learning_rate": 1.576171875e-05,
      "loss": 0.08992600440979004,
      "step": 270
    },
    {
      "epoch": 0.03663833167153938,
      "grad_norm": 0.5136080980300903,
      "learning_rate": 1.58203125e-05,
      "loss": 0.08193111419677734,
      "step": 271
    },
    {
      "epoch": 0.036773528467375326,
      "grad_norm": 0.3550981283187866,
      "learning_rate": 1.587890625e-05,
      "loss": 0.0819244384765625,
      "step": 272
    },
    {
      "epoch": 0.03690872526321126,
      "grad_norm": 0.8455483317375183,
      "learning_rate": 1.59375e-05,
      "loss": 0.10983848571777344,
      "step": 273
    },
    {
      "epoch": 0.0370439220590472,
      "grad_norm": 0.4411819279193878,
      "learning_rate": 1.599609375e-05,
      "loss": 0.06752872467041016,
      "step": 274
    },
    {
      "epoch": 0.03717911885488314,
      "grad_norm": 1.496036171913147,
      "learning_rate": 1.60546875e-05,
      "loss": 0.13500022888183594,
      "step": 275
    },
    {
      "epoch": 0.037314315650719077,
      "grad_norm": 0.26362380385398865,
      "learning_rate": 1.611328125e-05,
      "loss": 0.07960224151611328,
      "step": 276
    },
    {
      "epoch": 0.03744951244655502,
      "grad_norm": 0.9426568150520325,
      "learning_rate": 1.6171875000000002e-05,
      "loss": 0.13000869750976562,
      "step": 277
    },
    {
      "epoch": 0.037584709242390955,
      "grad_norm": 0.37183016538619995,
      "learning_rate": 1.623046875e-05,
      "loss": 0.13940811157226562,
      "step": 278
    },
    {
      "epoch": 0.03771990603822689,
      "grad_norm": 1.4855290651321411,
      "learning_rate": 1.62890625e-05,
      "loss": 0.13234710693359375,
      "step": 279
    },
    {
      "epoch": 0.037855102834062834,
      "grad_norm": 0.38084182143211365,
      "learning_rate": 1.634765625e-05,
      "loss": 0.07734251022338867,
      "step": 280
    },
    {
      "epoch": 0.03799029962989877,
      "grad_norm": 0.6273846626281738,
      "learning_rate": 1.640625e-05,
      "loss": 0.10966873168945312,
      "step": 281
    },
    {
      "epoch": 0.03812549642573471,
      "grad_norm": 0.2489965260028839,
      "learning_rate": 1.646484375e-05,
      "loss": 0.1035623550415039,
      "step": 282
    },
    {
      "epoch": 0.03826069322157065,
      "grad_norm": 0.47680971026420593,
      "learning_rate": 1.65234375e-05,
      "loss": 0.12607383728027344,
      "step": 283
    },
    {
      "epoch": 0.038395890017406585,
      "grad_norm": 0.4228799343109131,
      "learning_rate": 1.6582031250000002e-05,
      "loss": 0.10958576202392578,
      "step": 284
    },
    {
      "epoch": 0.03853108681324253,
      "grad_norm": 0.27856719493865967,
      "learning_rate": 1.6640625e-05,
      "loss": 0.12144136428833008,
      "step": 285
    },
    {
      "epoch": 0.038666283609078464,
      "grad_norm": 0.6729941964149475,
      "learning_rate": 1.669921875e-05,
      "loss": 0.10310649871826172,
      "step": 286
    },
    {
      "epoch": 0.03880148040491441,
      "grad_norm": 0.21593409776687622,
      "learning_rate": 1.67578125e-05,
      "loss": 0.12016677856445312,
      "step": 287
    },
    {
      "epoch": 0.03893667720075034,
      "grad_norm": 0.40790697932243347,
      "learning_rate": 1.681640625e-05,
      "loss": 0.09090614318847656,
      "step": 288
    },
    {
      "epoch": 0.03907187399658628,
      "grad_norm": 0.2896917760372162,
      "learning_rate": 1.6875e-05,
      "loss": 0.040282249450683594,
      "step": 289
    },
    {
      "epoch": 0.03920707079242222,
      "grad_norm": 0.3372872471809387,
      "learning_rate": 1.693359375e-05,
      "loss": 0.09852027893066406,
      "step": 290
    },
    {
      "epoch": 0.03934226758825816,
      "grad_norm": 1.4609146118164062,
      "learning_rate": 1.69921875e-05,
      "loss": 0.13547325134277344,
      "step": 291
    },
    {
      "epoch": 0.039477464384094094,
      "grad_norm": 0.720673143863678,
      "learning_rate": 1.705078125e-05,
      "loss": 0.0884857177734375,
      "step": 292
    },
    {
      "epoch": 0.03961266117993004,
      "grad_norm": 0.6555048227310181,
      "learning_rate": 1.7109375e-05,
      "loss": 0.1395282745361328,
      "step": 293
    },
    {
      "epoch": 0.03974785797576597,
      "grad_norm": 0.16606144607067108,
      "learning_rate": 1.7167968750000002e-05,
      "loss": 0.10356521606445312,
      "step": 294
    },
    {
      "epoch": 0.039883054771601915,
      "grad_norm": 0.3708697259426117,
      "learning_rate": 1.72265625e-05,
      "loss": 0.07172584533691406,
      "step": 295
    },
    {
      "epoch": 0.04001825156743785,
      "grad_norm": 1.0216091871261597,
      "learning_rate": 1.728515625e-05,
      "loss": 0.13167762756347656,
      "step": 296
    },
    {
      "epoch": 0.04015344836327379,
      "grad_norm": 0.7634631991386414,
      "learning_rate": 1.734375e-05,
      "loss": 0.08865737915039062,
      "step": 297
    },
    {
      "epoch": 0.04028864515910973,
      "grad_norm": 1.1382628679275513,
      "learning_rate": 1.740234375e-05,
      "loss": 0.11255073547363281,
      "step": 298
    },
    {
      "epoch": 0.040423841954945666,
      "grad_norm": 0.41765883564949036,
      "learning_rate": 1.74609375e-05,
      "loss": 0.07486391067504883,
      "step": 299
    },
    {
      "epoch": 0.04055903875078161,
      "grad_norm": 0.7868423461914062,
      "learning_rate": 1.751953125e-05,
      "loss": 0.10659217834472656,
      "step": 300
    },
    {
      "epoch": 0.040694235546617545,
      "grad_norm": 0.7546616792678833,
      "learning_rate": 1.7578125000000002e-05,
      "loss": 0.08617591857910156,
      "step": 301
    },
    {
      "epoch": 0.04082943234245348,
      "grad_norm": 0.6418368220329285,
      "learning_rate": 1.763671875e-05,
      "loss": 0.1351461410522461,
      "step": 302
    },
    {
      "epoch": 0.040964629138289424,
      "grad_norm": 0.463878870010376,
      "learning_rate": 1.76953125e-05,
      "loss": 0.12404346466064453,
      "step": 303
    },
    {
      "epoch": 0.04109982593412536,
      "grad_norm": 0.5326356291770935,
      "learning_rate": 1.775390625e-05,
      "loss": 0.08386039733886719,
      "step": 304
    },
    {
      "epoch": 0.0412350227299613,
      "grad_norm": 0.3665894865989685,
      "learning_rate": 1.78125e-05,
      "loss": 0.08456230163574219,
      "step": 305
    },
    {
      "epoch": 0.04137021952579724,
      "grad_norm": 0.6875031590461731,
      "learning_rate": 1.787109375e-05,
      "loss": 0.09281349182128906,
      "step": 306
    },
    {
      "epoch": 0.041505416321633175,
      "grad_norm": 0.9184910655021667,
      "learning_rate": 1.79296875e-05,
      "loss": 0.0856485366821289,
      "step": 307
    },
    {
      "epoch": 0.04164061311746912,
      "grad_norm": 0.6181513071060181,
      "learning_rate": 1.798828125e-05,
      "loss": 0.0825490951538086,
      "step": 308
    },
    {
      "epoch": 0.041775809913305054,
      "grad_norm": 0.8047035336494446,
      "learning_rate": 1.8046875e-05,
      "loss": 0.0860300064086914,
      "step": 309
    },
    {
      "epoch": 0.041911006709141,
      "grad_norm": 1.3716039657592773,
      "learning_rate": 1.810546875e-05,
      "loss": 0.09758186340332031,
      "step": 310
    },
    {
      "epoch": 0.04204620350497693,
      "grad_norm": 0.2594820559024811,
      "learning_rate": 1.8164062500000002e-05,
      "loss": 0.042504310607910156,
      "step": 311
    },
    {
      "epoch": 0.04218140030081287,
      "grad_norm": 0.5145635008811951,
      "learning_rate": 1.822265625e-05,
      "loss": 0.13232803344726562,
      "step": 312
    },
    {
      "epoch": 0.04231659709664881,
      "grad_norm": 0.4917304217815399,
      "learning_rate": 1.828125e-05,
      "loss": 0.08692359924316406,
      "step": 313
    },
    {
      "epoch": 0.04245179389248475,
      "grad_norm": 0.3924074172973633,
      "learning_rate": 1.833984375e-05,
      "loss": 0.12353849411010742,
      "step": 314
    },
    {
      "epoch": 0.04258699068832068,
      "grad_norm": 0.33571773767471313,
      "learning_rate": 1.83984375e-05,
      "loss": 0.0854043960571289,
      "step": 315
    },
    {
      "epoch": 0.042722187484156626,
      "grad_norm": 0.21847468614578247,
      "learning_rate": 1.845703125e-05,
      "loss": 0.08770465850830078,
      "step": 316
    },
    {
      "epoch": 0.04285738427999256,
      "grad_norm": 0.6506348252296448,
      "learning_rate": 1.8515625e-05,
      "loss": 0.11607933044433594,
      "step": 317
    },
    {
      "epoch": 0.042992581075828505,
      "grad_norm": 1.0390257835388184,
      "learning_rate": 1.8574218750000002e-05,
      "loss": 0.10783767700195312,
      "step": 318
    },
    {
      "epoch": 0.04312777787166444,
      "grad_norm": 0.21384234726428986,
      "learning_rate": 1.86328125e-05,
      "loss": 0.09239387512207031,
      "step": 319
    },
    {
      "epoch": 0.04326297466750038,
      "grad_norm": 0.2513214349746704,
      "learning_rate": 1.869140625e-05,
      "loss": 0.0860748291015625,
      "step": 320
    },
    {
      "epoch": 0.04339817146333632,
      "grad_norm": 0.13531465828418732,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 0.06541728973388672,
      "step": 321
    },
    {
      "epoch": 0.043533368259172256,
      "grad_norm": 0.5940647721290588,
      "learning_rate": 1.880859375e-05,
      "loss": 0.09384822845458984,
      "step": 322
    },
    {
      "epoch": 0.0436685650550082,
      "grad_norm": 0.24893070757389069,
      "learning_rate": 1.88671875e-05,
      "loss": 0.0641794204711914,
      "step": 323
    },
    {
      "epoch": 0.043803761850844135,
      "grad_norm": 0.24023190140724182,
      "learning_rate": 1.892578125e-05,
      "loss": 0.1150503158569336,
      "step": 324
    },
    {
      "epoch": 0.04393895864668007,
      "grad_norm": 0.36042916774749756,
      "learning_rate": 1.8984375e-05,
      "loss": 0.09014892578125,
      "step": 325
    },
    {
      "epoch": 0.044074155442516014,
      "grad_norm": 2.0375735759735107,
      "learning_rate": 1.904296875e-05,
      "loss": 0.10724353790283203,
      "step": 326
    },
    {
      "epoch": 0.04420935223835195,
      "grad_norm": 0.5884652733802795,
      "learning_rate": 1.91015625e-05,
      "loss": 0.08040904998779297,
      "step": 327
    },
    {
      "epoch": 0.04434454903418789,
      "grad_norm": 0.2396320104598999,
      "learning_rate": 1.9160156250000002e-05,
      "loss": 0.0972442626953125,
      "step": 328
    },
    {
      "epoch": 0.04447974583002383,
      "grad_norm": 0.6353582143783569,
      "learning_rate": 1.921875e-05,
      "loss": 0.09917926788330078,
      "step": 329
    },
    {
      "epoch": 0.044614942625859765,
      "grad_norm": 0.4164353907108307,
      "learning_rate": 1.927734375e-05,
      "loss": 0.1053004264831543,
      "step": 330
    },
    {
      "epoch": 0.04475013942169571,
      "grad_norm": 0.8356876969337463,
      "learning_rate": 1.93359375e-05,
      "loss": 0.10482120513916016,
      "step": 331
    },
    {
      "epoch": 0.04488533621753164,
      "grad_norm": 0.21361364424228668,
      "learning_rate": 1.939453125e-05,
      "loss": 0.11343526840209961,
      "step": 332
    },
    {
      "epoch": 0.045020533013367586,
      "grad_norm": 0.14705726504325867,
      "learning_rate": 1.9453125e-05,
      "loss": 0.052163124084472656,
      "step": 333
    },
    {
      "epoch": 0.04515572980920352,
      "grad_norm": 0.46907126903533936,
      "learning_rate": 1.951171875e-05,
      "loss": 0.08867025375366211,
      "step": 334
    },
    {
      "epoch": 0.04529092660503946,
      "grad_norm": 0.7444183230400085,
      "learning_rate": 1.95703125e-05,
      "loss": 0.1376323699951172,
      "step": 335
    },
    {
      "epoch": 0.0454261234008754,
      "grad_norm": 0.8058706521987915,
      "learning_rate": 1.962890625e-05,
      "loss": 0.0719766616821289,
      "step": 336
    },
    {
      "epoch": 0.04556132019671134,
      "grad_norm": 0.5213733315467834,
      "learning_rate": 1.96875e-05,
      "loss": 0.11151313781738281,
      "step": 337
    },
    {
      "epoch": 0.04569651699254728,
      "grad_norm": 0.5198162794113159,
      "learning_rate": 1.9746093750000002e-05,
      "loss": 0.15637779235839844,
      "step": 338
    },
    {
      "epoch": 0.045831713788383216,
      "grad_norm": 0.8097032308578491,
      "learning_rate": 1.98046875e-05,
      "loss": 0.1443328857421875,
      "step": 339
    },
    {
      "epoch": 0.04596691058421915,
      "grad_norm": 0.24106748402118683,
      "learning_rate": 1.986328125e-05,
      "loss": 0.08120203018188477,
      "step": 340
    },
    {
      "epoch": 0.046102107380055095,
      "grad_norm": 0.39901643991470337,
      "learning_rate": 1.9921875e-05,
      "loss": 0.07796096801757812,
      "step": 341
    },
    {
      "epoch": 0.04623730417589103,
      "grad_norm": 0.5847902297973633,
      "learning_rate": 1.998046875e-05,
      "loss": 0.10563230514526367,
      "step": 342
    },
    {
      "epoch": 0.04637250097172697,
      "grad_norm": 0.5063965916633606,
      "learning_rate": 2.00390625e-05,
      "loss": 0.07470321655273438,
      "step": 343
    },
    {
      "epoch": 0.04650769776756291,
      "grad_norm": 0.1668115109205246,
      "learning_rate": 2.009765625e-05,
      "loss": 0.0937337875366211,
      "step": 344
    },
    {
      "epoch": 0.046642894563398846,
      "grad_norm": 0.5499696731567383,
      "learning_rate": 2.0156250000000002e-05,
      "loss": 0.11548519134521484,
      "step": 345
    },
    {
      "epoch": 0.04677809135923479,
      "grad_norm": 0.7989301085472107,
      "learning_rate": 2.021484375e-05,
      "loss": 0.09680986404418945,
      "step": 346
    },
    {
      "epoch": 0.046913288155070725,
      "grad_norm": 0.5955407619476318,
      "learning_rate": 2.02734375e-05,
      "loss": 0.09231758117675781,
      "step": 347
    },
    {
      "epoch": 0.04704848495090666,
      "grad_norm": 0.4440248906612396,
      "learning_rate": 2.033203125e-05,
      "loss": 0.11703872680664062,
      "step": 348
    },
    {
      "epoch": 0.0471836817467426,
      "grad_norm": 0.34554049372673035,
      "learning_rate": 2.0390625e-05,
      "loss": 0.07292795181274414,
      "step": 349
    },
    {
      "epoch": 0.04731887854257854,
      "grad_norm": 0.6034355163574219,
      "learning_rate": 2.044921875e-05,
      "loss": 0.07951879501342773,
      "step": 350
    },
    {
      "epoch": 0.04745407533841448,
      "grad_norm": 1.147836685180664,
      "learning_rate": 2.05078125e-05,
      "loss": 0.1187601089477539,
      "step": 351
    },
    {
      "epoch": 0.04758927213425042,
      "grad_norm": 0.35816168785095215,
      "learning_rate": 2.056640625e-05,
      "loss": 0.1359539031982422,
      "step": 352
    },
    {
      "epoch": 0.047724468930086354,
      "grad_norm": 0.25260066986083984,
      "learning_rate": 2.0625e-05,
      "loss": 0.08533668518066406,
      "step": 353
    },
    {
      "epoch": 0.0478596657259223,
      "grad_norm": 0.15749254822731018,
      "learning_rate": 2.068359375e-05,
      "loss": 0.1195363998413086,
      "step": 354
    },
    {
      "epoch": 0.04799486252175823,
      "grad_norm": 0.18505169451236725,
      "learning_rate": 2.0742187500000002e-05,
      "loss": 0.09904670715332031,
      "step": 355
    },
    {
      "epoch": 0.048130059317594176,
      "grad_norm": 0.16219796240329742,
      "learning_rate": 2.080078125e-05,
      "loss": 0.0858449935913086,
      "step": 356
    },
    {
      "epoch": 0.04826525611343011,
      "grad_norm": 0.19502174854278564,
      "learning_rate": 2.0859375e-05,
      "loss": 0.09736824035644531,
      "step": 357
    },
    {
      "epoch": 0.04840045290926605,
      "grad_norm": 0.43384259939193726,
      "learning_rate": 2.091796875e-05,
      "loss": 0.126800537109375,
      "step": 358
    },
    {
      "epoch": 0.04853564970510199,
      "grad_norm": 0.2482476532459259,
      "learning_rate": 2.09765625e-05,
      "loss": 0.09348392486572266,
      "step": 359
    },
    {
      "epoch": 0.04867084650093793,
      "grad_norm": 0.20153117179870605,
      "learning_rate": 2.103515625e-05,
      "loss": 0.0736231803894043,
      "step": 360
    },
    {
      "epoch": 0.04880604329677387,
      "grad_norm": 0.41558927297592163,
      "learning_rate": 2.109375e-05,
      "loss": 0.10034418106079102,
      "step": 361
    },
    {
      "epoch": 0.048941240092609806,
      "grad_norm": 0.3068966567516327,
      "learning_rate": 2.1152343750000002e-05,
      "loss": 0.1041402816772461,
      "step": 362
    },
    {
      "epoch": 0.04907643688844574,
      "grad_norm": 0.27594423294067383,
      "learning_rate": 2.12109375e-05,
      "loss": 0.09015178680419922,
      "step": 363
    },
    {
      "epoch": 0.049211633684281685,
      "grad_norm": 0.7957682609558105,
      "learning_rate": 2.126953125e-05,
      "loss": 0.09404563903808594,
      "step": 364
    },
    {
      "epoch": 0.04934683048011762,
      "grad_norm": 1.0707272291183472,
      "learning_rate": 2.1328125000000002e-05,
      "loss": 0.06875801086425781,
      "step": 365
    },
    {
      "epoch": 0.049482027275953556,
      "grad_norm": 0.48567718267440796,
      "learning_rate": 2.138671875e-05,
      "loss": 0.08456039428710938,
      "step": 366
    },
    {
      "epoch": 0.0496172240717895,
      "grad_norm": 0.870097815990448,
      "learning_rate": 2.14453125e-05,
      "loss": 0.10132884979248047,
      "step": 367
    },
    {
      "epoch": 0.049752420867625435,
      "grad_norm": 0.6377391219139099,
      "learning_rate": 2.150390625e-05,
      "loss": 0.07217788696289062,
      "step": 368
    },
    {
      "epoch": 0.04988761766346138,
      "grad_norm": 0.788158655166626,
      "learning_rate": 2.15625e-05,
      "loss": 0.11425590515136719,
      "step": 369
    },
    {
      "epoch": 0.050022814459297314,
      "grad_norm": 0.21216727793216705,
      "learning_rate": 2.162109375e-05,
      "loss": 0.07319450378417969,
      "step": 370
    },
    {
      "epoch": 0.05015801125513325,
      "grad_norm": 0.25803834199905396,
      "learning_rate": 2.16796875e-05,
      "loss": 0.12201118469238281,
      "step": 371
    },
    {
      "epoch": 0.05029320805096919,
      "grad_norm": 0.41703903675079346,
      "learning_rate": 2.1738281250000002e-05,
      "loss": 0.08775997161865234,
      "step": 372
    },
    {
      "epoch": 0.05042840484680513,
      "grad_norm": 0.20716947317123413,
      "learning_rate": 2.1796875e-05,
      "loss": 0.11542510986328125,
      "step": 373
    },
    {
      "epoch": 0.05056360164264107,
      "grad_norm": 0.18524552881717682,
      "learning_rate": 2.185546875e-05,
      "loss": 0.06212902069091797,
      "step": 374
    },
    {
      "epoch": 0.05069879843847701,
      "grad_norm": 0.12163855135440826,
      "learning_rate": 2.19140625e-05,
      "loss": 0.05814361572265625,
      "step": 375
    },
    {
      "epoch": 0.050833995234312944,
      "grad_norm": 0.15177761018276215,
      "learning_rate": 2.197265625e-05,
      "loss": 0.0757136344909668,
      "step": 376
    },
    {
      "epoch": 0.05096919203014889,
      "grad_norm": 0.14696864783763885,
      "learning_rate": 2.203125e-05,
      "loss": 0.06543827056884766,
      "step": 377
    },
    {
      "epoch": 0.05110438882598482,
      "grad_norm": 0.6678949594497681,
      "learning_rate": 2.208984375e-05,
      "loss": 0.14141273498535156,
      "step": 378
    },
    {
      "epoch": 0.051239585621820766,
      "grad_norm": 0.386575311422348,
      "learning_rate": 2.2148437500000002e-05,
      "loss": 0.08747196197509766,
      "step": 379
    },
    {
      "epoch": 0.0513747824176567,
      "grad_norm": 0.9145892858505249,
      "learning_rate": 2.220703125e-05,
      "loss": 0.09427738189697266,
      "step": 380
    },
    {
      "epoch": 0.05150997921349264,
      "grad_norm": 0.6047827005386353,
      "learning_rate": 2.2265625e-05,
      "loss": 0.09445476531982422,
      "step": 381
    },
    {
      "epoch": 0.05164517600932858,
      "grad_norm": 0.830271303653717,
      "learning_rate": 2.2324218750000002e-05,
      "loss": 0.10351085662841797,
      "step": 382
    },
    {
      "epoch": 0.051780372805164517,
      "grad_norm": 0.48638877272605896,
      "learning_rate": 2.23828125e-05,
      "loss": 0.07607555389404297,
      "step": 383
    },
    {
      "epoch": 0.05191556960100046,
      "grad_norm": 3.4073305130004883,
      "learning_rate": 2.244140625e-05,
      "loss": 0.1930704116821289,
      "step": 384
    },
    {
      "epoch": 0.052050766396836395,
      "grad_norm": 1.4241673946380615,
      "learning_rate": 2.25e-05,
      "loss": 0.14117431640625,
      "step": 385
    },
    {
      "epoch": 0.05218596319267233,
      "grad_norm": 0.2762816250324249,
      "learning_rate": 2.255859375e-05,
      "loss": 0.0757436752319336,
      "step": 386
    },
    {
      "epoch": 0.052321159988508274,
      "grad_norm": 0.9581100940704346,
      "learning_rate": 2.26171875e-05,
      "loss": 0.09415817260742188,
      "step": 387
    },
    {
      "epoch": 0.05245635678434421,
      "grad_norm": 0.7780214548110962,
      "learning_rate": 2.267578125e-05,
      "loss": 0.10392951965332031,
      "step": 388
    },
    {
      "epoch": 0.05259155358018015,
      "grad_norm": 0.34639620780944824,
      "learning_rate": 2.2734375000000002e-05,
      "loss": 0.06629467010498047,
      "step": 389
    },
    {
      "epoch": 0.05272675037601609,
      "grad_norm": 0.5023885369300842,
      "learning_rate": 2.279296875e-05,
      "loss": 0.11465740203857422,
      "step": 390
    },
    {
      "epoch": 0.052861947171852025,
      "grad_norm": 0.725202202796936,
      "learning_rate": 2.28515625e-05,
      "loss": 0.10161781311035156,
      "step": 391
    },
    {
      "epoch": 0.05299714396768797,
      "grad_norm": 0.33798712491989136,
      "learning_rate": 2.291015625e-05,
      "loss": 0.10046005249023438,
      "step": 392
    },
    {
      "epoch": 0.053132340763523904,
      "grad_norm": 0.5291942358016968,
      "learning_rate": 2.296875e-05,
      "loss": 0.08743858337402344,
      "step": 393
    },
    {
      "epoch": 0.05326753755935984,
      "grad_norm": 0.9972317814826965,
      "learning_rate": 2.302734375e-05,
      "loss": 0.14546585083007812,
      "step": 394
    },
    {
      "epoch": 0.05340273435519578,
      "grad_norm": 0.8500605225563049,
      "learning_rate": 2.30859375e-05,
      "loss": 0.09073972702026367,
      "step": 395
    },
    {
      "epoch": 0.05353793115103172,
      "grad_norm": 1.0873994827270508,
      "learning_rate": 2.3144531250000002e-05,
      "loss": 0.0867767333984375,
      "step": 396
    },
    {
      "epoch": 0.05367312794686766,
      "grad_norm": 0.30353739857673645,
      "learning_rate": 2.3203125e-05,
      "loss": 0.12094783782958984,
      "step": 397
    },
    {
      "epoch": 0.0538083247427036,
      "grad_norm": 0.642541229724884,
      "learning_rate": 2.326171875e-05,
      "loss": 0.13100242614746094,
      "step": 398
    },
    {
      "epoch": 0.053943521538539534,
      "grad_norm": 0.37056803703308105,
      "learning_rate": 2.3320312500000002e-05,
      "loss": 0.12386131286621094,
      "step": 399
    },
    {
      "epoch": 0.05407871833437548,
      "grad_norm": 0.26606371998786926,
      "learning_rate": 2.337890625e-05,
      "loss": 0.10099029541015625,
      "step": 400
    },
    {
      "epoch": 0.05421391513021141,
      "grad_norm": 0.598811686038971,
      "learning_rate": 2.34375e-05,
      "loss": 0.10195493698120117,
      "step": 401
    },
    {
      "epoch": 0.054349111926047355,
      "grad_norm": 0.23437465727329254,
      "learning_rate": 2.349609375e-05,
      "loss": 0.10207271575927734,
      "step": 402
    },
    {
      "epoch": 0.05448430872188329,
      "grad_norm": 0.7463119029998779,
      "learning_rate": 2.35546875e-05,
      "loss": 0.11443519592285156,
      "step": 403
    },
    {
      "epoch": 0.05461950551771923,
      "grad_norm": 0.23719719052314758,
      "learning_rate": 2.361328125e-05,
      "loss": 0.06537270545959473,
      "step": 404
    },
    {
      "epoch": 0.05475470231355517,
      "grad_norm": 0.2601531445980072,
      "learning_rate": 2.3671875e-05,
      "loss": 0.08099842071533203,
      "step": 405
    },
    {
      "epoch": 0.054889899109391106,
      "grad_norm": 0.8800279498100281,
      "learning_rate": 2.3730468750000002e-05,
      "loss": 0.08411026000976562,
      "step": 406
    },
    {
      "epoch": 0.05502509590522705,
      "grad_norm": 0.14519470930099487,
      "learning_rate": 2.37890625e-05,
      "loss": 0.059525489807128906,
      "step": 407
    },
    {
      "epoch": 0.055160292701062985,
      "grad_norm": 0.17120572924613953,
      "learning_rate": 2.384765625e-05,
      "loss": 0.06172466278076172,
      "step": 408
    },
    {
      "epoch": 0.05529548949689892,
      "grad_norm": 0.5239396095275879,
      "learning_rate": 2.3906250000000002e-05,
      "loss": 0.09552001953125,
      "step": 409
    },
    {
      "epoch": 0.055430686292734864,
      "grad_norm": 0.1637723594903946,
      "learning_rate": 2.396484375e-05,
      "loss": 0.059935569763183594,
      "step": 410
    },
    {
      "epoch": 0.0555658830885708,
      "grad_norm": 0.5600265860557556,
      "learning_rate": 2.40234375e-05,
      "loss": 0.09963130950927734,
      "step": 411
    },
    {
      "epoch": 0.05570107988440674,
      "grad_norm": 0.3125258982181549,
      "learning_rate": 2.408203125e-05,
      "loss": 0.08180046081542969,
      "step": 412
    },
    {
      "epoch": 0.05583627668024268,
      "grad_norm": 0.14712771773338318,
      "learning_rate": 2.4140625e-05,
      "loss": 0.1001739501953125,
      "step": 413
    },
    {
      "epoch": 0.055971473476078615,
      "grad_norm": 0.2574838399887085,
      "learning_rate": 2.419921875e-05,
      "loss": 0.07106208801269531,
      "step": 414
    },
    {
      "epoch": 0.05610667027191456,
      "grad_norm": 0.41405096650123596,
      "learning_rate": 2.42578125e-05,
      "loss": 0.0865478515625,
      "step": 415
    },
    {
      "epoch": 0.056241867067750494,
      "grad_norm": 0.08951804786920547,
      "learning_rate": 2.4316406250000002e-05,
      "loss": 0.06560754776000977,
      "step": 416
    },
    {
      "epoch": 0.05637706386358643,
      "grad_norm": 0.24774175882339478,
      "learning_rate": 2.4375e-05,
      "loss": 0.09889507293701172,
      "step": 417
    },
    {
      "epoch": 0.05651226065942237,
      "grad_norm": 0.42182087898254395,
      "learning_rate": 2.443359375e-05,
      "loss": 0.12457895278930664,
      "step": 418
    },
    {
      "epoch": 0.05664745745525831,
      "grad_norm": 0.21307718753814697,
      "learning_rate": 2.44921875e-05,
      "loss": 0.08582401275634766,
      "step": 419
    },
    {
      "epoch": 0.05678265425109425,
      "grad_norm": 0.32010015845298767,
      "learning_rate": 2.455078125e-05,
      "loss": 0.12688255310058594,
      "step": 420
    },
    {
      "epoch": 0.05691785104693019,
      "grad_norm": 0.2119021713733673,
      "learning_rate": 2.4609375e-05,
      "loss": 0.09719276428222656,
      "step": 421
    },
    {
      "epoch": 0.05705304784276612,
      "grad_norm": 0.20008584856987,
      "learning_rate": 2.466796875e-05,
      "loss": 0.08666133880615234,
      "step": 422
    },
    {
      "epoch": 0.057188244638602066,
      "grad_norm": 0.1773962676525116,
      "learning_rate": 2.4726562500000002e-05,
      "loss": 0.06006050109863281,
      "step": 423
    },
    {
      "epoch": 0.057323441434438,
      "grad_norm": 0.1694437563419342,
      "learning_rate": 2.478515625e-05,
      "loss": 0.07963371276855469,
      "step": 424
    },
    {
      "epoch": 0.057458638230273945,
      "grad_norm": 0.3612242341041565,
      "learning_rate": 2.484375e-05,
      "loss": 0.12116003036499023,
      "step": 425
    },
    {
      "epoch": 0.05759383502610988,
      "grad_norm": 0.18567639589309692,
      "learning_rate": 2.4902343750000002e-05,
      "loss": 0.09574651718139648,
      "step": 426
    },
    {
      "epoch": 0.05772903182194582,
      "grad_norm": 0.21031694114208221,
      "learning_rate": 2.49609375e-05,
      "loss": 0.10147476196289062,
      "step": 427
    },
    {
      "epoch": 0.05786422861778176,
      "grad_norm": 0.34142056107521057,
      "learning_rate": 2.501953125e-05,
      "loss": 0.09202718734741211,
      "step": 428
    },
    {
      "epoch": 0.057999425413617696,
      "grad_norm": 0.15303567051887512,
      "learning_rate": 2.5078125e-05,
      "loss": 0.06485652923583984,
      "step": 429
    },
    {
      "epoch": 0.05813462220945364,
      "grad_norm": 0.5797934532165527,
      "learning_rate": 2.513671875e-05,
      "loss": 0.13192272186279297,
      "step": 430
    },
    {
      "epoch": 0.058269819005289575,
      "grad_norm": 0.5082610845565796,
      "learning_rate": 2.51953125e-05,
      "loss": 0.053334712982177734,
      "step": 431
    },
    {
      "epoch": 0.05840501580112551,
      "grad_norm": 0.3315902352333069,
      "learning_rate": 2.525390625e-05,
      "loss": 0.09614801406860352,
      "step": 432
    },
    {
      "epoch": 0.058540212596961454,
      "grad_norm": 0.2403307408094406,
      "learning_rate": 2.5312500000000002e-05,
      "loss": 0.07897329330444336,
      "step": 433
    },
    {
      "epoch": 0.05867540939279739,
      "grad_norm": 0.6379460692405701,
      "learning_rate": 2.537109375e-05,
      "loss": 0.1441669464111328,
      "step": 434
    },
    {
      "epoch": 0.05881060618863333,
      "grad_norm": 0.46444088220596313,
      "learning_rate": 2.54296875e-05,
      "loss": 0.12706565856933594,
      "step": 435
    },
    {
      "epoch": 0.05894580298446927,
      "grad_norm": 0.6619452238082886,
      "learning_rate": 2.548828125e-05,
      "loss": 0.10677719116210938,
      "step": 436
    },
    {
      "epoch": 0.059080999780305204,
      "grad_norm": 0.3971615731716156,
      "learning_rate": 2.5546875e-05,
      "loss": 0.07642936706542969,
      "step": 437
    },
    {
      "epoch": 0.05921619657614115,
      "grad_norm": 0.3409765362739563,
      "learning_rate": 2.560546875e-05,
      "loss": 0.09028291702270508,
      "step": 438
    },
    {
      "epoch": 0.05935139337197708,
      "grad_norm": 0.4880635440349579,
      "learning_rate": 2.56640625e-05,
      "loss": 0.0882730484008789,
      "step": 439
    },
    {
      "epoch": 0.059486590167813026,
      "grad_norm": 0.2580116391181946,
      "learning_rate": 2.5722656250000002e-05,
      "loss": 0.11551690101623535,
      "step": 440
    },
    {
      "epoch": 0.05962178696364896,
      "grad_norm": 0.28127238154411316,
      "learning_rate": 2.578125e-05,
      "loss": 0.08358478546142578,
      "step": 441
    },
    {
      "epoch": 0.0597569837594849,
      "grad_norm": 0.36428654193878174,
      "learning_rate": 2.583984375e-05,
      "loss": 0.10225105285644531,
      "step": 442
    },
    {
      "epoch": 0.05989218055532084,
      "grad_norm": 0.15446828305721283,
      "learning_rate": 2.5898437500000002e-05,
      "loss": 0.07058906555175781,
      "step": 443
    },
    {
      "epoch": 0.06002737735115678,
      "grad_norm": 0.4826681911945343,
      "learning_rate": 2.595703125e-05,
      "loss": 0.1385936737060547,
      "step": 444
    },
    {
      "epoch": 0.06016257414699271,
      "grad_norm": 0.17592550814151764,
      "learning_rate": 2.6015625e-05,
      "loss": 0.0698246955871582,
      "step": 445
    },
    {
      "epoch": 0.060297770942828656,
      "grad_norm": 0.31240314245224,
      "learning_rate": 2.607421875e-05,
      "loss": 0.08679771423339844,
      "step": 446
    },
    {
      "epoch": 0.06043296773866459,
      "grad_norm": 0.6398367285728455,
      "learning_rate": 2.61328125e-05,
      "loss": 0.1202545166015625,
      "step": 447
    },
    {
      "epoch": 0.060568164534500535,
      "grad_norm": 0.3890048861503601,
      "learning_rate": 2.619140625e-05,
      "loss": 0.05002260208129883,
      "step": 448
    },
    {
      "epoch": 0.06070336133033647,
      "grad_norm": 0.46764808893203735,
      "learning_rate": 2.625e-05,
      "loss": 0.07270145416259766,
      "step": 449
    },
    {
      "epoch": 0.06083855812617241,
      "grad_norm": 0.3413294553756714,
      "learning_rate": 2.6308593750000002e-05,
      "loss": 0.08463382720947266,
      "step": 450
    },
    {
      "epoch": 0.06097375492200835,
      "grad_norm": 0.2559318244457245,
      "learning_rate": 2.63671875e-05,
      "loss": 0.12294769287109375,
      "step": 451
    },
    {
      "epoch": 0.061108951717844286,
      "grad_norm": 0.7449362874031067,
      "learning_rate": 2.642578125e-05,
      "loss": 0.07621002197265625,
      "step": 452
    },
    {
      "epoch": 0.06124414851368023,
      "grad_norm": 0.3746051788330078,
      "learning_rate": 2.6484375000000002e-05,
      "loss": 0.09185314178466797,
      "step": 453
    },
    {
      "epoch": 0.061379345309516165,
      "grad_norm": 0.5289967656135559,
      "learning_rate": 2.654296875e-05,
      "loss": 0.11911392211914062,
      "step": 454
    },
    {
      "epoch": 0.0615145421053521,
      "grad_norm": 0.22747808694839478,
      "learning_rate": 2.66015625e-05,
      "loss": 0.07048225402832031,
      "step": 455
    },
    {
      "epoch": 0.06164973890118804,
      "grad_norm": 0.42176374793052673,
      "learning_rate": 2.666015625e-05,
      "loss": 0.09965133666992188,
      "step": 456
    },
    {
      "epoch": 0.06178493569702398,
      "grad_norm": 0.21731840074062347,
      "learning_rate": 2.6718750000000002e-05,
      "loss": 0.08211088180541992,
      "step": 457
    },
    {
      "epoch": 0.06192013249285992,
      "grad_norm": 0.4568416476249695,
      "learning_rate": 2.677734375e-05,
      "loss": 0.1116476058959961,
      "step": 458
    },
    {
      "epoch": 0.06205532928869586,
      "grad_norm": 0.22233718633651733,
      "learning_rate": 2.68359375e-05,
      "loss": 0.05214977264404297,
      "step": 459
    },
    {
      "epoch": 0.062190526084531794,
      "grad_norm": 0.43788546323776245,
      "learning_rate": 2.6894531250000002e-05,
      "loss": 0.09244871139526367,
      "step": 460
    },
    {
      "epoch": 0.06232572288036774,
      "grad_norm": 0.5135902166366577,
      "learning_rate": 2.6953125e-05,
      "loss": 0.10364055633544922,
      "step": 461
    },
    {
      "epoch": 0.06246091967620367,
      "grad_norm": 0.2793516218662262,
      "learning_rate": 2.701171875e-05,
      "loss": 0.05248737335205078,
      "step": 462
    },
    {
      "epoch": 0.06259611647203961,
      "grad_norm": 0.406421422958374,
      "learning_rate": 2.70703125e-05,
      "loss": 0.08672618865966797,
      "step": 463
    },
    {
      "epoch": 0.06273131326787555,
      "grad_norm": 0.12528029084205627,
      "learning_rate": 2.712890625e-05,
      "loss": 0.029212474822998047,
      "step": 464
    },
    {
      "epoch": 0.0628665100637115,
      "grad_norm": 0.3743863105773926,
      "learning_rate": 2.71875e-05,
      "loss": 0.0905771255493164,
      "step": 465
    },
    {
      "epoch": 0.06300170685954742,
      "grad_norm": 0.4710961580276489,
      "learning_rate": 2.724609375e-05,
      "loss": 0.08863449096679688,
      "step": 466
    },
    {
      "epoch": 0.06313690365538337,
      "grad_norm": 0.2887037396430969,
      "learning_rate": 2.7304687500000002e-05,
      "loss": 0.07930278778076172,
      "step": 467
    },
    {
      "epoch": 0.06327210045121931,
      "grad_norm": 0.3721965551376343,
      "learning_rate": 2.736328125e-05,
      "loss": 0.07076358795166016,
      "step": 468
    },
    {
      "epoch": 0.06340729724705524,
      "grad_norm": 0.1717955321073532,
      "learning_rate": 2.7421875e-05,
      "loss": 0.07027316093444824,
      "step": 469
    },
    {
      "epoch": 0.06354249404289118,
      "grad_norm": 0.24996301531791687,
      "learning_rate": 2.7480468750000002e-05,
      "loss": 0.12811613082885742,
      "step": 470
    },
    {
      "epoch": 0.06367769083872712,
      "grad_norm": 0.30291178822517395,
      "learning_rate": 2.75390625e-05,
      "loss": 0.16044998168945312,
      "step": 471
    },
    {
      "epoch": 0.06381288763456307,
      "grad_norm": 0.2250363528728485,
      "learning_rate": 2.759765625e-05,
      "loss": 0.07297325134277344,
      "step": 472
    },
    {
      "epoch": 0.063948084430399,
      "grad_norm": 0.2488800436258316,
      "learning_rate": 2.765625e-05,
      "loss": 0.07302236557006836,
      "step": 473
    },
    {
      "epoch": 0.06408328122623494,
      "grad_norm": 0.14107242226600647,
      "learning_rate": 2.7714843750000002e-05,
      "loss": 0.06640434265136719,
      "step": 474
    },
    {
      "epoch": 0.06421847802207088,
      "grad_norm": 0.16244164109230042,
      "learning_rate": 2.77734375e-05,
      "loss": 0.07704639434814453,
      "step": 475
    },
    {
      "epoch": 0.06435367481790681,
      "grad_norm": 0.2833482623100281,
      "learning_rate": 2.783203125e-05,
      "loss": 0.10689687728881836,
      "step": 476
    },
    {
      "epoch": 0.06448887161374275,
      "grad_norm": 0.3703344166278839,
      "learning_rate": 2.7890625000000002e-05,
      "loss": 0.1478271484375,
      "step": 477
    },
    {
      "epoch": 0.0646240684095787,
      "grad_norm": 0.38901495933532715,
      "learning_rate": 2.794921875e-05,
      "loss": 0.09612178802490234,
      "step": 478
    },
    {
      "epoch": 0.06475926520541463,
      "grad_norm": 0.2891184389591217,
      "learning_rate": 2.80078125e-05,
      "loss": 0.06588220596313477,
      "step": 479
    },
    {
      "epoch": 0.06489446200125057,
      "grad_norm": 0.1334463655948639,
      "learning_rate": 2.806640625e-05,
      "loss": 0.07320594787597656,
      "step": 480
    },
    {
      "epoch": 0.06502965879708651,
      "grad_norm": 0.23207786679267883,
      "learning_rate": 2.8125e-05,
      "loss": 0.05840778350830078,
      "step": 481
    },
    {
      "epoch": 0.06516485559292244,
      "grad_norm": 0.4443608820438385,
      "learning_rate": 2.818359375e-05,
      "loss": 0.0740361213684082,
      "step": 482
    },
    {
      "epoch": 0.06530005238875838,
      "grad_norm": 0.3327508568763733,
      "learning_rate": 2.82421875e-05,
      "loss": 0.10940694808959961,
      "step": 483
    },
    {
      "epoch": 0.06543524918459433,
      "grad_norm": 0.32395318150520325,
      "learning_rate": 2.8300781250000002e-05,
      "loss": 0.08982467651367188,
      "step": 484
    },
    {
      "epoch": 0.06557044598043027,
      "grad_norm": 0.7124205231666565,
      "learning_rate": 2.8359375e-05,
      "loss": 0.121795654296875,
      "step": 485
    },
    {
      "epoch": 0.0657056427762662,
      "grad_norm": 0.521129310131073,
      "learning_rate": 2.841796875e-05,
      "loss": 0.09668445587158203,
      "step": 486
    },
    {
      "epoch": 0.06584083957210214,
      "grad_norm": 0.31626927852630615,
      "learning_rate": 2.8476562500000002e-05,
      "loss": 0.05739116668701172,
      "step": 487
    },
    {
      "epoch": 0.06597603636793808,
      "grad_norm": 0.17424233257770538,
      "learning_rate": 2.853515625e-05,
      "loss": 0.08448123931884766,
      "step": 488
    },
    {
      "epoch": 0.06611123316377401,
      "grad_norm": 0.22088785469532013,
      "learning_rate": 2.859375e-05,
      "loss": 0.08600044250488281,
      "step": 489
    },
    {
      "epoch": 0.06624642995960996,
      "grad_norm": 0.14359068870544434,
      "learning_rate": 2.865234375e-05,
      "loss": 0.06694495677947998,
      "step": 490
    },
    {
      "epoch": 0.0663816267554459,
      "grad_norm": 0.5119166970252991,
      "learning_rate": 2.87109375e-05,
      "loss": 0.13225746154785156,
      "step": 491
    },
    {
      "epoch": 0.06651682355128183,
      "grad_norm": 0.11170045286417007,
      "learning_rate": 2.876953125e-05,
      "loss": 0.0768272876739502,
      "step": 492
    },
    {
      "epoch": 0.06665202034711777,
      "grad_norm": 0.24550208449363708,
      "learning_rate": 2.8828125e-05,
      "loss": 0.09083175659179688,
      "step": 493
    },
    {
      "epoch": 0.06678721714295371,
      "grad_norm": 0.49827250838279724,
      "learning_rate": 2.8886718750000002e-05,
      "loss": 0.12553882598876953,
      "step": 494
    },
    {
      "epoch": 0.06692241393878966,
      "grad_norm": 0.1018979400396347,
      "learning_rate": 2.89453125e-05,
      "loss": 0.0836629867553711,
      "step": 495
    },
    {
      "epoch": 0.06705761073462559,
      "grad_norm": 0.7441481351852417,
      "learning_rate": 2.900390625e-05,
      "loss": 0.09560012817382812,
      "step": 496
    },
    {
      "epoch": 0.06719280753046153,
      "grad_norm": 0.16811202466487885,
      "learning_rate": 2.90625e-05,
      "loss": 0.0786895751953125,
      "step": 497
    },
    {
      "epoch": 0.06732800432629747,
      "grad_norm": 0.3336220979690552,
      "learning_rate": 2.912109375e-05,
      "loss": 0.08087158203125,
      "step": 498
    },
    {
      "epoch": 0.0674632011221334,
      "grad_norm": 0.32104796171188354,
      "learning_rate": 2.91796875e-05,
      "loss": 0.09322166442871094,
      "step": 499
    },
    {
      "epoch": 0.06759839791796934,
      "grad_norm": 0.2070852816104889,
      "learning_rate": 2.923828125e-05,
      "loss": 0.09871625900268555,
      "step": 500
    },
    {
      "epoch": 0.06773359471380529,
      "grad_norm": 0.215023472905159,
      "learning_rate": 2.9296875000000002e-05,
      "loss": 0.10586929321289062,
      "step": 501
    },
    {
      "epoch": 0.06786879150964122,
      "grad_norm": 0.4073386490345001,
      "learning_rate": 2.935546875e-05,
      "loss": 0.12042808532714844,
      "step": 502
    },
    {
      "epoch": 0.06800398830547716,
      "grad_norm": 0.16417402029037476,
      "learning_rate": 2.94140625e-05,
      "loss": 0.08030986785888672,
      "step": 503
    },
    {
      "epoch": 0.0681391851013131,
      "grad_norm": 0.18574321269989014,
      "learning_rate": 2.9472656250000002e-05,
      "loss": 0.10937690734863281,
      "step": 504
    },
    {
      "epoch": 0.06827438189714903,
      "grad_norm": 0.2531127333641052,
      "learning_rate": 2.953125e-05,
      "loss": 0.09542155265808105,
      "step": 505
    },
    {
      "epoch": 0.06840957869298497,
      "grad_norm": 0.2664662003517151,
      "learning_rate": 2.958984375e-05,
      "loss": 0.08902454376220703,
      "step": 506
    },
    {
      "epoch": 0.06854477548882092,
      "grad_norm": 0.25008097290992737,
      "learning_rate": 2.96484375e-05,
      "loss": 0.10931873321533203,
      "step": 507
    },
    {
      "epoch": 0.06867997228465686,
      "grad_norm": 0.38765203952789307,
      "learning_rate": 2.970703125e-05,
      "loss": 0.09329700469970703,
      "step": 508
    },
    {
      "epoch": 0.06881516908049279,
      "grad_norm": 0.26607415080070496,
      "learning_rate": 2.9765625e-05,
      "loss": 0.07357215881347656,
      "step": 509
    },
    {
      "epoch": 0.06895036587632873,
      "grad_norm": 0.17761532962322235,
      "learning_rate": 2.982421875e-05,
      "loss": 0.07666778564453125,
      "step": 510
    },
    {
      "epoch": 0.06908556267216467,
      "grad_norm": 0.26230624318122864,
      "learning_rate": 2.9882812500000002e-05,
      "loss": 0.10649490356445312,
      "step": 511
    },
    {
      "epoch": 0.0692207594680006,
      "grad_norm": 0.1139485239982605,
      "learning_rate": 2.994140625e-05,
      "loss": 0.07679367065429688,
      "step": 512
    },
    {
      "epoch": 0.06935595626383655,
      "grad_norm": 0.2930458188056946,
      "learning_rate": 3e-05,
      "loss": 0.0900421142578125,
      "step": 513
    },
    {
      "epoch": 0.06949115305967249,
      "grad_norm": 0.3094085454940796,
      "learning_rate": 2.9999998438460004e-05,
      "loss": 0.06854057312011719,
      "step": 514
    },
    {
      "epoch": 0.06962634985550842,
      "grad_norm": 0.1503412425518036,
      "learning_rate": 2.9999993753840344e-05,
      "loss": 0.08107662200927734,
      "step": 515
    },
    {
      "epoch": 0.06976154665134436,
      "grad_norm": 0.8429868817329407,
      "learning_rate": 2.9999985946141995e-05,
      "loss": 0.11447715759277344,
      "step": 516
    },
    {
      "epoch": 0.0698967434471803,
      "grad_norm": 0.34427618980407715,
      "learning_rate": 2.9999975015366586e-05,
      "loss": 0.12001323699951172,
      "step": 517
    },
    {
      "epoch": 0.07003194024301625,
      "grad_norm": 0.5228518843650818,
      "learning_rate": 2.9999960961516384e-05,
      "loss": 0.11302566528320312,
      "step": 518
    },
    {
      "epoch": 0.07016713703885218,
      "grad_norm": 0.2178574502468109,
      "learning_rate": 2.9999943784594325e-05,
      "loss": 0.08069896697998047,
      "step": 519
    },
    {
      "epoch": 0.07030233383468812,
      "grad_norm": 0.2845768332481384,
      "learning_rate": 2.9999923484603975e-05,
      "loss": 0.09895515441894531,
      "step": 520
    },
    {
      "epoch": 0.07043753063052406,
      "grad_norm": 0.13979288935661316,
      "learning_rate": 2.999990006154957e-05,
      "loss": 0.09073543548583984,
      "step": 521
    },
    {
      "epoch": 0.07057272742635999,
      "grad_norm": 0.14410565793514252,
      "learning_rate": 2.9999873515435977e-05,
      "loss": 0.0753030776977539,
      "step": 522
    },
    {
      "epoch": 0.07070792422219593,
      "grad_norm": 0.39464297890663147,
      "learning_rate": 2.9999843846268735e-05,
      "loss": 0.10400676727294922,
      "step": 523
    },
    {
      "epoch": 0.07084312101803188,
      "grad_norm": 0.2655804753303528,
      "learning_rate": 2.9999811054054018e-05,
      "loss": 0.06646251678466797,
      "step": 524
    },
    {
      "epoch": 0.0709783178138678,
      "grad_norm": 0.19320404529571533,
      "learning_rate": 2.9999775138798646e-05,
      "loss": 0.09278392791748047,
      "step": 525
    },
    {
      "epoch": 0.07111351460970375,
      "grad_norm": 0.11904240399599075,
      "learning_rate": 2.99997361005101e-05,
      "loss": 0.08670663833618164,
      "step": 526
    },
    {
      "epoch": 0.07124871140553969,
      "grad_norm": 0.3094334602355957,
      "learning_rate": 2.9999693939196513e-05,
      "loss": 0.0844879150390625,
      "step": 527
    },
    {
      "epoch": 0.07138390820137563,
      "grad_norm": 0.36753204464912415,
      "learning_rate": 2.999964865486666e-05,
      "loss": 0.07069790363311768,
      "step": 528
    },
    {
      "epoch": 0.07151910499721156,
      "grad_norm": 0.31463855504989624,
      "learning_rate": 2.999960024752997e-05,
      "loss": 0.12663936614990234,
      "step": 529
    },
    {
      "epoch": 0.0716543017930475,
      "grad_norm": 0.16702313721179962,
      "learning_rate": 2.9999548717196516e-05,
      "loss": 0.08238887786865234,
      "step": 530
    },
    {
      "epoch": 0.07178949858888345,
      "grad_norm": 0.11972741037607193,
      "learning_rate": 2.999949406387703e-05,
      "loss": 0.08971595764160156,
      "step": 531
    },
    {
      "epoch": 0.07192469538471938,
      "grad_norm": 0.2606874406337738,
      "learning_rate": 2.9999436287582903e-05,
      "loss": 0.10538005828857422,
      "step": 532
    },
    {
      "epoch": 0.07205989218055532,
      "grad_norm": 0.19562457501888275,
      "learning_rate": 2.9999375388326145e-05,
      "loss": 0.11307525634765625,
      "step": 533
    },
    {
      "epoch": 0.07219508897639126,
      "grad_norm": 0.1900082677602768,
      "learning_rate": 2.9999311366119447e-05,
      "loss": 0.06357097625732422,
      "step": 534
    },
    {
      "epoch": 0.07233028577222719,
      "grad_norm": 0.35355332493782043,
      "learning_rate": 2.9999244220976137e-05,
      "loss": 0.12403202056884766,
      "step": 535
    },
    {
      "epoch": 0.07246548256806314,
      "grad_norm": 0.1257934272289276,
      "learning_rate": 2.9999173952910197e-05,
      "loss": 0.0693352222442627,
      "step": 536
    },
    {
      "epoch": 0.07260067936389908,
      "grad_norm": 0.3972124755382538,
      "learning_rate": 2.9999100561936252e-05,
      "loss": 0.10654115676879883,
      "step": 537
    },
    {
      "epoch": 0.07273587615973501,
      "grad_norm": 0.21748265624046326,
      "learning_rate": 2.9999024048069585e-05,
      "loss": 0.08458232879638672,
      "step": 538
    },
    {
      "epoch": 0.07287107295557095,
      "grad_norm": 0.22351795434951782,
      "learning_rate": 2.9998944411326127e-05,
      "loss": 0.060929298400878906,
      "step": 539
    },
    {
      "epoch": 0.0730062697514069,
      "grad_norm": 0.3372923731803894,
      "learning_rate": 2.999886165172246e-05,
      "loss": 0.08164787292480469,
      "step": 540
    },
    {
      "epoch": 0.07314146654724284,
      "grad_norm": 0.1859789341688156,
      "learning_rate": 2.9998775769275814e-05,
      "loss": 0.10498714447021484,
      "step": 541
    },
    {
      "epoch": 0.07327666334307877,
      "grad_norm": 0.18278458714485168,
      "learning_rate": 2.9998686764004067e-05,
      "loss": 0.09410762786865234,
      "step": 542
    },
    {
      "epoch": 0.07341186013891471,
      "grad_norm": 0.2038678228855133,
      "learning_rate": 2.9998594635925755e-05,
      "loss": 0.08800554275512695,
      "step": 543
    },
    {
      "epoch": 0.07354705693475065,
      "grad_norm": 0.32873788475990295,
      "learning_rate": 2.999849938506005e-05,
      "loss": 0.08473777770996094,
      "step": 544
    },
    {
      "epoch": 0.07368225373058658,
      "grad_norm": 0.21102842688560486,
      "learning_rate": 2.99984010114268e-05,
      "loss": 0.07731151580810547,
      "step": 545
    },
    {
      "epoch": 0.07381745052642252,
      "grad_norm": 0.37139272689819336,
      "learning_rate": 2.9998299515046475e-05,
      "loss": 0.10659599304199219,
      "step": 546
    },
    {
      "epoch": 0.07395264732225847,
      "grad_norm": 0.4046103358268738,
      "learning_rate": 2.9998194895940213e-05,
      "loss": 0.08731555938720703,
      "step": 547
    },
    {
      "epoch": 0.0740878441180944,
      "grad_norm": 0.09351778030395508,
      "learning_rate": 2.9998087154129792e-05,
      "loss": 0.07852745056152344,
      "step": 548
    },
    {
      "epoch": 0.07422304091393034,
      "grad_norm": 0.3234085440635681,
      "learning_rate": 2.9997976289637645e-05,
      "loss": 0.0915689468383789,
      "step": 549
    },
    {
      "epoch": 0.07435823770976628,
      "grad_norm": 0.20921041071414948,
      "learning_rate": 2.9997862302486855e-05,
      "loss": 0.07576560974121094,
      "step": 550
    },
    {
      "epoch": 0.07449343450560222,
      "grad_norm": 0.3197173774242401,
      "learning_rate": 2.9997745192701153e-05,
      "loss": 0.08384895324707031,
      "step": 551
    },
    {
      "epoch": 0.07462863130143815,
      "grad_norm": 0.11288613080978394,
      "learning_rate": 2.9997624960304926e-05,
      "loss": 0.061016082763671875,
      "step": 552
    },
    {
      "epoch": 0.0747638280972741,
      "grad_norm": 0.08520299196243286,
      "learning_rate": 2.9997501605323214e-05,
      "loss": 0.06275606155395508,
      "step": 553
    },
    {
      "epoch": 0.07489902489311004,
      "grad_norm": 0.30760088562965393,
      "learning_rate": 2.999737512778168e-05,
      "loss": 0.07462167739868164,
      "step": 554
    },
    {
      "epoch": 0.07503422168894597,
      "grad_norm": 0.3377489447593689,
      "learning_rate": 2.9997245527706674e-05,
      "loss": 0.08373498916625977,
      "step": 555
    },
    {
      "epoch": 0.07516941848478191,
      "grad_norm": 0.37358325719833374,
      "learning_rate": 2.999711280512517e-05,
      "loss": 0.0783548355102539,
      "step": 556
    },
    {
      "epoch": 0.07530461528061785,
      "grad_norm": 0.2781330347061157,
      "learning_rate": 2.9996976960064807e-05,
      "loss": 0.07808208465576172,
      "step": 557
    },
    {
      "epoch": 0.07543981207645378,
      "grad_norm": 0.3915441036224365,
      "learning_rate": 2.999683799255387e-05,
      "loss": 0.08591938018798828,
      "step": 558
    },
    {
      "epoch": 0.07557500887228973,
      "grad_norm": 0.35622817277908325,
      "learning_rate": 2.999669590262129e-05,
      "loss": 0.047222137451171875,
      "step": 559
    },
    {
      "epoch": 0.07571020566812567,
      "grad_norm": 0.27966177463531494,
      "learning_rate": 2.999655069029665e-05,
      "loss": 0.10179281234741211,
      "step": 560
    },
    {
      "epoch": 0.0758454024639616,
      "grad_norm": 0.13936161994934082,
      "learning_rate": 2.9996402355610183e-05,
      "loss": 0.057047367095947266,
      "step": 561
    },
    {
      "epoch": 0.07598059925979754,
      "grad_norm": 0.11600785702466965,
      "learning_rate": 2.9996250898592777e-05,
      "loss": 0.07817554473876953,
      "step": 562
    },
    {
      "epoch": 0.07611579605563348,
      "grad_norm": 0.3167460560798645,
      "learning_rate": 2.9996096319275962e-05,
      "loss": 0.06467565894126892,
      "step": 563
    },
    {
      "epoch": 0.07625099285146943,
      "grad_norm": 0.40994006395339966,
      "learning_rate": 2.9995938617691925e-05,
      "loss": 0.12283134460449219,
      "step": 564
    },
    {
      "epoch": 0.07638618964730536,
      "grad_norm": 0.25212064385414124,
      "learning_rate": 2.9995777793873504e-05,
      "loss": 0.08774852752685547,
      "step": 565
    },
    {
      "epoch": 0.0765213864431413,
      "grad_norm": 0.21666939556598663,
      "learning_rate": 2.9995613847854176e-05,
      "loss": 0.06520652770996094,
      "step": 566
    },
    {
      "epoch": 0.07665658323897724,
      "grad_norm": 0.27924904227256775,
      "learning_rate": 2.9995446779668078e-05,
      "loss": 0.10445094108581543,
      "step": 567
    },
    {
      "epoch": 0.07679178003481317,
      "grad_norm": 0.141977921128273,
      "learning_rate": 2.9995276589349992e-05,
      "loss": 0.10594844818115234,
      "step": 568
    },
    {
      "epoch": 0.07692697683064911,
      "grad_norm": 0.06773504614830017,
      "learning_rate": 2.9995103276935357e-05,
      "loss": 0.0640711784362793,
      "step": 569
    },
    {
      "epoch": 0.07706217362648506,
      "grad_norm": 0.5213485956192017,
      "learning_rate": 2.9994926842460258e-05,
      "loss": 0.08524489402770996,
      "step": 570
    },
    {
      "epoch": 0.07719737042232099,
      "grad_norm": 0.10157053172588348,
      "learning_rate": 2.9994747285961428e-05,
      "loss": 0.06107282638549805,
      "step": 571
    },
    {
      "epoch": 0.07733256721815693,
      "grad_norm": 0.09224223345518112,
      "learning_rate": 2.9994564607476255e-05,
      "loss": 0.08637428283691406,
      "step": 572
    },
    {
      "epoch": 0.07746776401399287,
      "grad_norm": 0.3050849735736847,
      "learning_rate": 2.9994378807042762e-05,
      "loss": 0.1074533462524414,
      "step": 573
    },
    {
      "epoch": 0.07760296080982881,
      "grad_norm": 0.22146910429000854,
      "learning_rate": 2.9994189884699647e-05,
      "loss": 0.1300792694091797,
      "step": 574
    },
    {
      "epoch": 0.07773815760566474,
      "grad_norm": 0.07144106179475784,
      "learning_rate": 2.9993997840486233e-05,
      "loss": 0.04799842834472656,
      "step": 575
    },
    {
      "epoch": 0.07787335440150069,
      "grad_norm": 0.32296156883239746,
      "learning_rate": 2.9993802674442516e-05,
      "loss": 0.07802867889404297,
      "step": 576
    },
    {
      "epoch": 0.07800855119733663,
      "grad_norm": 0.11064080893993378,
      "learning_rate": 2.999360438660913e-05,
      "loss": 0.06294059753417969,
      "step": 577
    },
    {
      "epoch": 0.07814374799317256,
      "grad_norm": 0.2089330106973648,
      "learning_rate": 2.9993402977027346e-05,
      "loss": 0.054758548736572266,
      "step": 578
    },
    {
      "epoch": 0.0782789447890085,
      "grad_norm": 0.13595914840698242,
      "learning_rate": 2.999319844573911e-05,
      "loss": 0.042527198791503906,
      "step": 579
    },
    {
      "epoch": 0.07841414158484444,
      "grad_norm": 0.3079105019569397,
      "learning_rate": 2.9992990792787007e-05,
      "loss": 0.07813072204589844,
      "step": 580
    },
    {
      "epoch": 0.07854933838068037,
      "grad_norm": 0.25272977352142334,
      "learning_rate": 2.999278001821427e-05,
      "loss": 0.11185455322265625,
      "step": 581
    },
    {
      "epoch": 0.07868453517651632,
      "grad_norm": 0.26442980766296387,
      "learning_rate": 2.9992566122064775e-05,
      "loss": 0.09828448295593262,
      "step": 582
    },
    {
      "epoch": 0.07881973197235226,
      "grad_norm": 0.2732599377632141,
      "learning_rate": 2.999234910438307e-05,
      "loss": 0.07195329666137695,
      "step": 583
    },
    {
      "epoch": 0.07895492876818819,
      "grad_norm": 0.05204978585243225,
      "learning_rate": 2.999212896521433e-05,
      "loss": 0.039618730545043945,
      "step": 584
    },
    {
      "epoch": 0.07909012556402413,
      "grad_norm": 0.12640728056430817,
      "learning_rate": 2.999190570460439e-05,
      "loss": 0.05958151817321777,
      "step": 585
    },
    {
      "epoch": 0.07922532235986007,
      "grad_norm": 0.32790225744247437,
      "learning_rate": 2.9991679322599734e-05,
      "loss": 0.06865406036376953,
      "step": 586
    },
    {
      "epoch": 0.07936051915569602,
      "grad_norm": 0.15171629190444946,
      "learning_rate": 2.9991449819247505e-05,
      "loss": 0.04877662658691406,
      "step": 587
    },
    {
      "epoch": 0.07949571595153195,
      "grad_norm": 0.10906587541103363,
      "learning_rate": 2.9991217194595474e-05,
      "loss": 0.06815218925476074,
      "step": 588
    },
    {
      "epoch": 0.07963091274736789,
      "grad_norm": 0.2120455801486969,
      "learning_rate": 2.9990981448692078e-05,
      "loss": 0.0928192138671875,
      "step": 589
    },
    {
      "epoch": 0.07976610954320383,
      "grad_norm": 0.1074172630906105,
      "learning_rate": 2.999074258158641e-05,
      "loss": 0.09325790405273438,
      "step": 590
    },
    {
      "epoch": 0.07990130633903976,
      "grad_norm": 0.19498956203460693,
      "learning_rate": 2.9990500593328192e-05,
      "loss": 0.05649304389953613,
      "step": 591
    },
    {
      "epoch": 0.0800365031348757,
      "grad_norm": 0.14760342240333557,
      "learning_rate": 2.999025548396781e-05,
      "loss": 0.10672998428344727,
      "step": 592
    },
    {
      "epoch": 0.08017169993071165,
      "grad_norm": 0.21737149357795715,
      "learning_rate": 2.9990007253556302e-05,
      "loss": 0.07566475868225098,
      "step": 593
    },
    {
      "epoch": 0.08030689672654757,
      "grad_norm": 0.2974933385848999,
      "learning_rate": 2.9989755902145345e-05,
      "loss": 0.11906719207763672,
      "step": 594
    },
    {
      "epoch": 0.08044209352238352,
      "grad_norm": 0.11337985843420029,
      "learning_rate": 2.9989501429787273e-05,
      "loss": 0.06615829467773438,
      "step": 595
    },
    {
      "epoch": 0.08057729031821946,
      "grad_norm": 0.21340131759643555,
      "learning_rate": 2.9989243836535073e-05,
      "loss": 0.06207704544067383,
      "step": 596
    },
    {
      "epoch": 0.0807124871140554,
      "grad_norm": 0.2859881818294525,
      "learning_rate": 2.998898312244237e-05,
      "loss": 0.07950067520141602,
      "step": 597
    },
    {
      "epoch": 0.08084768390989133,
      "grad_norm": 0.16309234499931335,
      "learning_rate": 2.9988719287563452e-05,
      "loss": 0.0576624870300293,
      "step": 598
    },
    {
      "epoch": 0.08098288070572728,
      "grad_norm": 0.10070134699344635,
      "learning_rate": 2.998845233195325e-05,
      "loss": 0.08256387710571289,
      "step": 599
    },
    {
      "epoch": 0.08111807750156322,
      "grad_norm": 0.15067686140537262,
      "learning_rate": 2.998818225566734e-05,
      "loss": 0.09409618377685547,
      "step": 600
    },
    {
      "epoch": 0.08125327429739915,
      "grad_norm": 0.10668405145406723,
      "learning_rate": 2.998790905876196e-05,
      "loss": 0.11132514476776123,
      "step": 601
    },
    {
      "epoch": 0.08138847109323509,
      "grad_norm": 0.16517981886863708,
      "learning_rate": 2.9987632741293987e-05,
      "loss": 0.07035708427429199,
      "step": 602
    },
    {
      "epoch": 0.08152366788907103,
      "grad_norm": 0.09393040090799332,
      "learning_rate": 2.998735330332096e-05,
      "loss": 0.07950484752655029,
      "step": 603
    },
    {
      "epoch": 0.08165886468490696,
      "grad_norm": 0.12008967995643616,
      "learning_rate": 2.9987070744901046e-05,
      "loss": 0.06570839881896973,
      "step": 604
    },
    {
      "epoch": 0.0817940614807429,
      "grad_norm": 0.2272833287715912,
      "learning_rate": 2.9986785066093084e-05,
      "loss": 0.09181094169616699,
      "step": 605
    },
    {
      "epoch": 0.08192925827657885,
      "grad_norm": 0.25803667306900024,
      "learning_rate": 2.9986496266956556e-05,
      "loss": 0.09886741638183594,
      "step": 606
    },
    {
      "epoch": 0.08206445507241478,
      "grad_norm": 0.4905322790145874,
      "learning_rate": 2.9986204347551583e-05,
      "loss": 0.12009239196777344,
      "step": 607
    },
    {
      "epoch": 0.08219965186825072,
      "grad_norm": 0.0775560736656189,
      "learning_rate": 2.9985909307938948e-05,
      "loss": 0.06000089645385742,
      "step": 608
    },
    {
      "epoch": 0.08233484866408666,
      "grad_norm": 0.2988452613353729,
      "learning_rate": 2.9985611148180082e-05,
      "loss": 0.08237457275390625,
      "step": 609
    },
    {
      "epoch": 0.0824700454599226,
      "grad_norm": 0.31842100620269775,
      "learning_rate": 2.9985309868337063e-05,
      "loss": 0.0988306999206543,
      "step": 610
    },
    {
      "epoch": 0.08260524225575853,
      "grad_norm": 0.1052919328212738,
      "learning_rate": 2.9985005468472617e-05,
      "loss": 0.030002593994140625,
      "step": 611
    },
    {
      "epoch": 0.08274043905159448,
      "grad_norm": 0.28633493185043335,
      "learning_rate": 2.9984697948650124e-05,
      "loss": 0.10738992691040039,
      "step": 612
    },
    {
      "epoch": 0.08287563584743042,
      "grad_norm": 0.28117629885673523,
      "learning_rate": 2.998438730893361e-05,
      "loss": 0.08930015563964844,
      "step": 613
    },
    {
      "epoch": 0.08301083264326635,
      "grad_norm": 0.1861264705657959,
      "learning_rate": 2.9984073549387747e-05,
      "loss": 0.05366230010986328,
      "step": 614
    },
    {
      "epoch": 0.08314602943910229,
      "grad_norm": 0.41998764872550964,
      "learning_rate": 2.998375667007787e-05,
      "loss": 0.07553625106811523,
      "step": 615
    },
    {
      "epoch": 0.08328122623493824,
      "grad_norm": 0.19615380465984344,
      "learning_rate": 2.998343667106995e-05,
      "loss": 0.06767797470092773,
      "step": 616
    },
    {
      "epoch": 0.08341642303077416,
      "grad_norm": 0.18898122012615204,
      "learning_rate": 2.9983113552430616e-05,
      "loss": 0.0434575080871582,
      "step": 617
    },
    {
      "epoch": 0.08355161982661011,
      "grad_norm": 0.45470932126045227,
      "learning_rate": 2.9982787314227134e-05,
      "loss": 0.07708358764648438,
      "step": 618
    },
    {
      "epoch": 0.08368681662244605,
      "grad_norm": 0.22009018063545227,
      "learning_rate": 2.998245795652744e-05,
      "loss": 0.09606266021728516,
      "step": 619
    },
    {
      "epoch": 0.083822013418282,
      "grad_norm": 0.18823456764221191,
      "learning_rate": 2.9982125479400106e-05,
      "loss": 0.06135225296020508,
      "step": 620
    },
    {
      "epoch": 0.08395721021411792,
      "grad_norm": 0.5377417802810669,
      "learning_rate": 2.9981789882914352e-05,
      "loss": 0.08295488357543945,
      "step": 621
    },
    {
      "epoch": 0.08409240700995387,
      "grad_norm": 0.4763088822364807,
      "learning_rate": 2.9981451167140048e-05,
      "loss": 0.0911707878112793,
      "step": 622
    },
    {
      "epoch": 0.08422760380578981,
      "grad_norm": 0.22370657324790955,
      "learning_rate": 2.9981109332147722e-05,
      "loss": 0.08115291595458984,
      "step": 623
    },
    {
      "epoch": 0.08436280060162574,
      "grad_norm": 0.2826088070869446,
      "learning_rate": 2.9980764378008545e-05,
      "loss": 0.06379127502441406,
      "step": 624
    },
    {
      "epoch": 0.08449799739746168,
      "grad_norm": 0.9952225089073181,
      "learning_rate": 2.9980416304794332e-05,
      "loss": 0.11029529571533203,
      "step": 625
    },
    {
      "epoch": 0.08463319419329762,
      "grad_norm": 0.18860769271850586,
      "learning_rate": 2.9980065112577565e-05,
      "loss": 0.099334716796875,
      "step": 626
    },
    {
      "epoch": 0.08476839098913355,
      "grad_norm": 0.28185638785362244,
      "learning_rate": 2.9979710801431357e-05,
      "loss": 0.08804464340209961,
      "step": 627
    },
    {
      "epoch": 0.0849035877849695,
      "grad_norm": 0.14749231934547424,
      "learning_rate": 2.997935337142948e-05,
      "loss": 0.05631279945373535,
      "step": 628
    },
    {
      "epoch": 0.08503878458080544,
      "grad_norm": 0.4178594946861267,
      "learning_rate": 2.9978992822646347e-05,
      "loss": 0.06901311874389648,
      "step": 629
    },
    {
      "epoch": 0.08517398137664137,
      "grad_norm": 0.1803293079137802,
      "learning_rate": 2.9978629155157036e-05,
      "loss": 0.08359909057617188,
      "step": 630
    },
    {
      "epoch": 0.08530917817247731,
      "grad_norm": 0.19294004142284393,
      "learning_rate": 2.9978262369037252e-05,
      "loss": 0.07879948616027832,
      "step": 631
    },
    {
      "epoch": 0.08544437496831325,
      "grad_norm": 0.1940276175737381,
      "learning_rate": 2.9977892464363375e-05,
      "loss": 0.05239057540893555,
      "step": 632
    },
    {
      "epoch": 0.0855795717641492,
      "grad_norm": 0.2245698869228363,
      "learning_rate": 2.9977519441212412e-05,
      "loss": 0.09803438186645508,
      "step": 633
    },
    {
      "epoch": 0.08571476855998512,
      "grad_norm": 0.23050887882709503,
      "learning_rate": 2.9977143299662034e-05,
      "loss": 0.08642864227294922,
      "step": 634
    },
    {
      "epoch": 0.08584996535582107,
      "grad_norm": 0.09768661856651306,
      "learning_rate": 2.997676403979055e-05,
      "loss": 0.03159141540527344,
      "step": 635
    },
    {
      "epoch": 0.08598516215165701,
      "grad_norm": 0.09331204742193222,
      "learning_rate": 2.997638166167693e-05,
      "loss": 0.07317638397216797,
      "step": 636
    },
    {
      "epoch": 0.08612035894749294,
      "grad_norm": 0.06804801523685455,
      "learning_rate": 2.9975996165400786e-05,
      "loss": 0.044899582862854004,
      "step": 637
    },
    {
      "epoch": 0.08625555574332888,
      "grad_norm": 0.3908451199531555,
      "learning_rate": 2.9975607551042373e-05,
      "loss": 0.0752716064453125,
      "step": 638
    },
    {
      "epoch": 0.08639075253916483,
      "grad_norm": 0.12349463999271393,
      "learning_rate": 2.9975215818682607e-05,
      "loss": 0.07203435897827148,
      "step": 639
    },
    {
      "epoch": 0.08652594933500075,
      "grad_norm": 0.21345306932926178,
      "learning_rate": 2.9974820968403056e-05,
      "loss": 0.08788847923278809,
      "step": 640
    },
    {
      "epoch": 0.0866611461308367,
      "grad_norm": 0.1577584445476532,
      "learning_rate": 2.9974423000285923e-05,
      "loss": 0.07614803314208984,
      "step": 641
    },
    {
      "epoch": 0.08679634292667264,
      "grad_norm": 0.09714977443218231,
      "learning_rate": 2.9974021914414068e-05,
      "loss": 0.04539823532104492,
      "step": 642
    },
    {
      "epoch": 0.08693153972250858,
      "grad_norm": 0.11560861021280289,
      "learning_rate": 2.9973617710871e-05,
      "loss": 0.0796351432800293,
      "step": 643
    },
    {
      "epoch": 0.08706673651834451,
      "grad_norm": 0.21095207333564758,
      "learning_rate": 2.997321038974087e-05,
      "loss": 0.07535362243652344,
      "step": 644
    },
    {
      "epoch": 0.08720193331418045,
      "grad_norm": 0.13098455965518951,
      "learning_rate": 2.997279995110849e-05,
      "loss": 0.08330440521240234,
      "step": 645
    },
    {
      "epoch": 0.0873371301100164,
      "grad_norm": 0.2241521030664444,
      "learning_rate": 2.997238639505932e-05,
      "loss": 0.05736207962036133,
      "step": 646
    },
    {
      "epoch": 0.08747232690585233,
      "grad_norm": 0.08733268082141876,
      "learning_rate": 2.997196972167946e-05,
      "loss": 0.07244205474853516,
      "step": 647
    },
    {
      "epoch": 0.08760752370168827,
      "grad_norm": 0.19124197959899902,
      "learning_rate": 2.9971549931055665e-05,
      "loss": 0.09659004211425781,
      "step": 648
    },
    {
      "epoch": 0.08774272049752421,
      "grad_norm": 0.3027733266353607,
      "learning_rate": 2.997112702327533e-05,
      "loss": 0.11511707305908203,
      "step": 649
    },
    {
      "epoch": 0.08787791729336014,
      "grad_norm": 0.07912140339612961,
      "learning_rate": 2.9970700998426518e-05,
      "loss": 0.05508708953857422,
      "step": 650
    },
    {
      "epoch": 0.08801311408919608,
      "grad_norm": 0.27462825179100037,
      "learning_rate": 2.9970271856597925e-05,
      "loss": 0.08562397956848145,
      "step": 651
    },
    {
      "epoch": 0.08814831088503203,
      "grad_norm": 0.1728072464466095,
      "learning_rate": 2.9969839597878896e-05,
      "loss": 0.09009361267089844,
      "step": 652
    },
    {
      "epoch": 0.08828350768086797,
      "grad_norm": 0.08717627823352814,
      "learning_rate": 2.9969404222359436e-05,
      "loss": 0.05742001533508301,
      "step": 653
    },
    {
      "epoch": 0.0884187044767039,
      "grad_norm": 0.228158637881279,
      "learning_rate": 2.9968965730130188e-05,
      "loss": 0.07795238494873047,
      "step": 654
    },
    {
      "epoch": 0.08855390127253984,
      "grad_norm": 0.11197508126497269,
      "learning_rate": 2.9968524121282455e-05,
      "loss": 0.06445789337158203,
      "step": 655
    },
    {
      "epoch": 0.08868909806837579,
      "grad_norm": 0.2086934596300125,
      "learning_rate": 2.9968079395908178e-05,
      "loss": 0.10188674926757812,
      "step": 656
    },
    {
      "epoch": 0.08882429486421171,
      "grad_norm": 0.22176457941532135,
      "learning_rate": 2.9967631554099947e-05,
      "loss": 0.1122288703918457,
      "step": 657
    },
    {
      "epoch": 0.08895949166004766,
      "grad_norm": 0.1395748257637024,
      "learning_rate": 2.996718059595101e-05,
      "loss": 0.0919804573059082,
      "step": 658
    },
    {
      "epoch": 0.0890946884558836,
      "grad_norm": 0.20750685036182404,
      "learning_rate": 2.9966726521555265e-05,
      "loss": 0.07076454162597656,
      "step": 659
    },
    {
      "epoch": 0.08922988525171953,
      "grad_norm": 0.1286686807870865,
      "learning_rate": 2.996626933100724e-05,
      "loss": 0.052811622619628906,
      "step": 660
    },
    {
      "epoch": 0.08936508204755547,
      "grad_norm": 0.44623395800590515,
      "learning_rate": 2.996580902440213e-05,
      "loss": 0.09715652465820312,
      "step": 661
    },
    {
      "epoch": 0.08950027884339141,
      "grad_norm": 0.16575433313846588,
      "learning_rate": 2.9965345601835773e-05,
      "loss": 0.08658599853515625,
      "step": 662
    },
    {
      "epoch": 0.08963547563922734,
      "grad_norm": 0.29066911339759827,
      "learning_rate": 2.996487906340466e-05,
      "loss": 0.07785797119140625,
      "step": 663
    },
    {
      "epoch": 0.08977067243506329,
      "grad_norm": 0.2852063477039337,
      "learning_rate": 2.996440940920592e-05,
      "loss": 0.08462285995483398,
      "step": 664
    },
    {
      "epoch": 0.08990586923089923,
      "grad_norm": 0.1689836084842682,
      "learning_rate": 2.996393663933735e-05,
      "loss": 0.06775045394897461,
      "step": 665
    },
    {
      "epoch": 0.09004106602673517,
      "grad_norm": 0.12417086213827133,
      "learning_rate": 2.9963460753897364e-05,
      "loss": 0.05731785297393799,
      "step": 666
    },
    {
      "epoch": 0.0901762628225711,
      "grad_norm": 0.244721457362175,
      "learning_rate": 2.996298175298506e-05,
      "loss": 0.09905529022216797,
      "step": 667
    },
    {
      "epoch": 0.09031145961840704,
      "grad_norm": 0.18479038774967194,
      "learning_rate": 2.996249963670016e-05,
      "loss": 0.0888524055480957,
      "step": 668
    },
    {
      "epoch": 0.09044665641424299,
      "grad_norm": 0.10865657031536102,
      "learning_rate": 2.9962014405143042e-05,
      "loss": 0.06245851516723633,
      "step": 669
    },
    {
      "epoch": 0.09058185321007892,
      "grad_norm": 0.27855703234672546,
      "learning_rate": 2.9961526058414745e-05,
      "loss": 0.11337447166442871,
      "step": 670
    },
    {
      "epoch": 0.09071705000591486,
      "grad_norm": 0.10137587040662766,
      "learning_rate": 2.9961034596616936e-05,
      "loss": 0.06521320343017578,
      "step": 671
    },
    {
      "epoch": 0.0908522468017508,
      "grad_norm": 0.24208283424377441,
      "learning_rate": 2.996054001985194e-05,
      "loss": 0.095703125,
      "step": 672
    },
    {
      "epoch": 0.09098744359758673,
      "grad_norm": 0.08551014214754105,
      "learning_rate": 2.9960042328222732e-05,
      "loss": 0.06468367576599121,
      "step": 673
    },
    {
      "epoch": 0.09112264039342267,
      "grad_norm": 0.21869981288909912,
      "learning_rate": 2.995954152183294e-05,
      "loss": 0.05954790115356445,
      "step": 674
    },
    {
      "epoch": 0.09125783718925862,
      "grad_norm": 0.31491169333457947,
      "learning_rate": 2.9959037600786822e-05,
      "loss": 0.06922221183776855,
      "step": 675
    },
    {
      "epoch": 0.09139303398509456,
      "grad_norm": 0.28219854831695557,
      "learning_rate": 2.9958530565189307e-05,
      "loss": 0.08938407897949219,
      "step": 676
    },
    {
      "epoch": 0.09152823078093049,
      "grad_norm": 0.09948960691690445,
      "learning_rate": 2.995802041514596e-05,
      "loss": 0.056900978088378906,
      "step": 677
    },
    {
      "epoch": 0.09166342757676643,
      "grad_norm": 0.2522660195827484,
      "learning_rate": 2.9957507150762996e-05,
      "loss": 0.047379493713378906,
      "step": 678
    },
    {
      "epoch": 0.09179862437260237,
      "grad_norm": 0.2659372389316559,
      "learning_rate": 2.9956990772147283e-05,
      "loss": 0.06755733489990234,
      "step": 679
    },
    {
      "epoch": 0.0919338211684383,
      "grad_norm": 0.10989851504564285,
      "learning_rate": 2.9956471279406324e-05,
      "loss": 0.060033321380615234,
      "step": 680
    },
    {
      "epoch": 0.09206901796427425,
      "grad_norm": 0.09935111552476883,
      "learning_rate": 2.9955948672648298e-05,
      "loss": 0.07506513595581055,
      "step": 681
    },
    {
      "epoch": 0.09220421476011019,
      "grad_norm": 0.1358737349510193,
      "learning_rate": 2.9955422951981994e-05,
      "loss": 0.08691072463989258,
      "step": 682
    },
    {
      "epoch": 0.09233941155594612,
      "grad_norm": 0.3513760566711426,
      "learning_rate": 2.995489411751688e-05,
      "loss": 0.07443714141845703,
      "step": 683
    },
    {
      "epoch": 0.09247460835178206,
      "grad_norm": 0.09077506512403488,
      "learning_rate": 2.9954362169363064e-05,
      "loss": 0.05679333209991455,
      "step": 684
    },
    {
      "epoch": 0.092609805147618,
      "grad_norm": 0.21559298038482666,
      "learning_rate": 2.99538271076313e-05,
      "loss": 0.10883903503417969,
      "step": 685
    },
    {
      "epoch": 0.09274500194345393,
      "grad_norm": 0.15053609013557434,
      "learning_rate": 2.9953288932432985e-05,
      "loss": 0.06873083114624023,
      "step": 686
    },
    {
      "epoch": 0.09288019873928988,
      "grad_norm": 0.10975361615419388,
      "learning_rate": 2.995274764388018e-05,
      "loss": 0.05933570861816406,
      "step": 687
    },
    {
      "epoch": 0.09301539553512582,
      "grad_norm": 0.1789533495903015,
      "learning_rate": 2.9952203242085566e-05,
      "loss": 0.08390617370605469,
      "step": 688
    },
    {
      "epoch": 0.09315059233096176,
      "grad_norm": 0.07746067643165588,
      "learning_rate": 2.995165572716251e-05,
      "loss": 0.049941062927246094,
      "step": 689
    },
    {
      "epoch": 0.09328578912679769,
      "grad_norm": 0.24971401691436768,
      "learning_rate": 2.9951105099225003e-05,
      "loss": 0.07831621170043945,
      "step": 690
    },
    {
      "epoch": 0.09342098592263363,
      "grad_norm": 0.36833053827285767,
      "learning_rate": 2.995055135838768e-05,
      "loss": 0.0910797119140625,
      "step": 691
    },
    {
      "epoch": 0.09355618271846958,
      "grad_norm": 0.10221574455499649,
      "learning_rate": 2.994999450476584e-05,
      "loss": 0.05471158027648926,
      "step": 692
    },
    {
      "epoch": 0.0936913795143055,
      "grad_norm": 0.1148780956864357,
      "learning_rate": 2.9949434538475414e-05,
      "loss": 0.06614893674850464,
      "step": 693
    },
    {
      "epoch": 0.09382657631014145,
      "grad_norm": 0.2983362078666687,
      "learning_rate": 2.9948871459633008e-05,
      "loss": 0.07979011535644531,
      "step": 694
    },
    {
      "epoch": 0.09396177310597739,
      "grad_norm": 0.2829235792160034,
      "learning_rate": 2.994830526835584e-05,
      "loss": 0.08431506156921387,
      "step": 695
    },
    {
      "epoch": 0.09409696990181332,
      "grad_norm": 0.13727952539920807,
      "learning_rate": 2.9947735964761803e-05,
      "loss": 0.05151629447937012,
      "step": 696
    },
    {
      "epoch": 0.09423216669764926,
      "grad_norm": 0.6994669437408447,
      "learning_rate": 2.9947163548969428e-05,
      "loss": 0.1470623016357422,
      "step": 697
    },
    {
      "epoch": 0.0943673634934852,
      "grad_norm": 0.1264650821685791,
      "learning_rate": 2.9946588021097893e-05,
      "loss": 0.06728172302246094,
      "step": 698
    },
    {
      "epoch": 0.09450256028932115,
      "grad_norm": 0.15813982486724854,
      "learning_rate": 2.9946009381267028e-05,
      "loss": 0.06923913955688477,
      "step": 699
    },
    {
      "epoch": 0.09463775708515708,
      "grad_norm": 0.3901388943195343,
      "learning_rate": 2.9945427629597306e-05,
      "loss": 0.08095240592956543,
      "step": 700
    },
    {
      "epoch": 0.09477295388099302,
      "grad_norm": 0.7803894877433777,
      "learning_rate": 2.9944842766209853e-05,
      "loss": 0.05688786506652832,
      "step": 701
    },
    {
      "epoch": 0.09490815067682896,
      "grad_norm": 0.5690038800239563,
      "learning_rate": 2.9944254791226444e-05,
      "loss": 0.10587501525878906,
      "step": 702
    },
    {
      "epoch": 0.0950433474726649,
      "grad_norm": 0.4294602870941162,
      "learning_rate": 2.994366370476949e-05,
      "loss": 0.10778427124023438,
      "step": 703
    },
    {
      "epoch": 0.09517854426850084,
      "grad_norm": 0.4204708933830261,
      "learning_rate": 2.9943069506962067e-05,
      "loss": 0.09755897521972656,
      "step": 704
    },
    {
      "epoch": 0.09531374106433678,
      "grad_norm": 0.46001702547073364,
      "learning_rate": 2.9942472197927886e-05,
      "loss": 0.06331729888916016,
      "step": 705
    },
    {
      "epoch": 0.09544893786017271,
      "grad_norm": 0.5181862711906433,
      "learning_rate": 2.994187177779131e-05,
      "loss": 0.11969566345214844,
      "step": 706
    },
    {
      "epoch": 0.09558413465600865,
      "grad_norm": 0.12053836137056351,
      "learning_rate": 2.9941268246677353e-05,
      "loss": 0.0626363754272461,
      "step": 707
    },
    {
      "epoch": 0.0957193314518446,
      "grad_norm": 0.15417523682117462,
      "learning_rate": 2.9940661604711664e-05,
      "loss": 0.07751274108886719,
      "step": 708
    },
    {
      "epoch": 0.09585452824768052,
      "grad_norm": 0.47470811009407043,
      "learning_rate": 2.994005185202056e-05,
      "loss": 0.12705039978027344,
      "step": 709
    },
    {
      "epoch": 0.09598972504351647,
      "grad_norm": 0.27987006306648254,
      "learning_rate": 2.9939438988730986e-05,
      "loss": 0.07154560089111328,
      "step": 710
    },
    {
      "epoch": 0.09612492183935241,
      "grad_norm": 0.15930283069610596,
      "learning_rate": 2.9938823014970553e-05,
      "loss": 0.10855388641357422,
      "step": 711
    },
    {
      "epoch": 0.09626011863518835,
      "grad_norm": 0.257501482963562,
      "learning_rate": 2.99382039308675e-05,
      "loss": 0.05958271026611328,
      "step": 712
    },
    {
      "epoch": 0.09639531543102428,
      "grad_norm": 0.7364037036895752,
      "learning_rate": 2.993758173655073e-05,
      "loss": 0.12431597709655762,
      "step": 713
    },
    {
      "epoch": 0.09653051222686022,
      "grad_norm": 0.17351634800434113,
      "learning_rate": 2.993695643214979e-05,
      "loss": 0.0854029655456543,
      "step": 714
    },
    {
      "epoch": 0.09666570902269617,
      "grad_norm": 0.298469603061676,
      "learning_rate": 2.9936328017794864e-05,
      "loss": 0.07146835327148438,
      "step": 715
    },
    {
      "epoch": 0.0968009058185321,
      "grad_norm": 0.3269720673561096,
      "learning_rate": 2.9935696493616796e-05,
      "loss": 0.08095455169677734,
      "step": 716
    },
    {
      "epoch": 0.09693610261436804,
      "grad_norm": 1.599771499633789,
      "learning_rate": 2.9935061859747065e-05,
      "loss": 0.0824437141418457,
      "step": 717
    },
    {
      "epoch": 0.09707129941020398,
      "grad_norm": 0.1393277794122696,
      "learning_rate": 2.993442411631782e-05,
      "loss": 0.0891885757446289,
      "step": 718
    },
    {
      "epoch": 0.09720649620603991,
      "grad_norm": 0.0796637013554573,
      "learning_rate": 2.9933783263461827e-05,
      "loss": 0.09085512161254883,
      "step": 719
    },
    {
      "epoch": 0.09734169300187585,
      "grad_norm": 0.3463320732116699,
      "learning_rate": 2.9933139301312526e-05,
      "loss": 0.09558963775634766,
      "step": 720
    },
    {
      "epoch": 0.0974768897977118,
      "grad_norm": 0.29950204491615295,
      "learning_rate": 2.9932492230003984e-05,
      "loss": 0.061611175537109375,
      "step": 721
    },
    {
      "epoch": 0.09761208659354774,
      "grad_norm": 0.29633015394210815,
      "learning_rate": 2.993184204967094e-05,
      "loss": 0.08249521255493164,
      "step": 722
    },
    {
      "epoch": 0.09774728338938367,
      "grad_norm": 0.4554542005062103,
      "learning_rate": 2.9931188760448748e-05,
      "loss": 0.10138177871704102,
      "step": 723
    },
    {
      "epoch": 0.09788248018521961,
      "grad_norm": 0.06369206309318542,
      "learning_rate": 2.9930532362473433e-05,
      "loss": 0.056447505950927734,
      "step": 724
    },
    {
      "epoch": 0.09801767698105555,
      "grad_norm": 0.26517704129219055,
      "learning_rate": 2.9929872855881663e-05,
      "loss": 0.08902454376220703,
      "step": 725
    },
    {
      "epoch": 0.09815287377689148,
      "grad_norm": 0.1894245445728302,
      "learning_rate": 2.9929210240810744e-05,
      "loss": 0.051939964294433594,
      "step": 726
    },
    {
      "epoch": 0.09828807057272743,
      "grad_norm": 0.10839513689279556,
      "learning_rate": 2.9928544517398644e-05,
      "loss": 0.06834030151367188,
      "step": 727
    },
    {
      "epoch": 0.09842326736856337,
      "grad_norm": 0.13607367873191833,
      "learning_rate": 2.9927875685783966e-05,
      "loss": 0.0683584213256836,
      "step": 728
    },
    {
      "epoch": 0.0985584641643993,
      "grad_norm": 0.1251877248287201,
      "learning_rate": 2.9927203746105968e-05,
      "loss": 0.08129453659057617,
      "step": 729
    },
    {
      "epoch": 0.09869366096023524,
      "grad_norm": 0.24035900831222534,
      "learning_rate": 2.9926528698504546e-05,
      "loss": 0.09148168563842773,
      "step": 730
    },
    {
      "epoch": 0.09882885775607118,
      "grad_norm": 0.3103516399860382,
      "learning_rate": 2.992585054312025e-05,
      "loss": 0.0979318618774414,
      "step": 731
    },
    {
      "epoch": 0.09896405455190711,
      "grad_norm": 0.23814600706100464,
      "learning_rate": 2.9925169280094278e-05,
      "loss": 0.11593031883239746,
      "step": 732
    },
    {
      "epoch": 0.09909925134774306,
      "grad_norm": 0.20139487087726593,
      "learning_rate": 2.9924484909568472e-05,
      "loss": 0.07658267021179199,
      "step": 733
    },
    {
      "epoch": 0.099234448143579,
      "grad_norm": 0.13799434900283813,
      "learning_rate": 2.9923797431685322e-05,
      "loss": 0.08754158020019531,
      "step": 734
    },
    {
      "epoch": 0.09936964493941494,
      "grad_norm": 0.16388531029224396,
      "learning_rate": 2.992310684658796e-05,
      "loss": 0.08730697631835938,
      "step": 735
    },
    {
      "epoch": 0.09950484173525087,
      "grad_norm": 0.21970367431640625,
      "learning_rate": 2.9922413154420173e-05,
      "loss": 0.07097339630126953,
      "step": 736
    },
    {
      "epoch": 0.09964003853108681,
      "grad_norm": 0.3876591622829437,
      "learning_rate": 2.9921716355326393e-05,
      "loss": 0.10553598403930664,
      "step": 737
    },
    {
      "epoch": 0.09977523532692276,
      "grad_norm": 0.2511739730834961,
      "learning_rate": 2.9921016449451695e-05,
      "loss": 0.0774528980255127,
      "step": 738
    },
    {
      "epoch": 0.09991043212275869,
      "grad_norm": 0.3090797960758209,
      "learning_rate": 2.9920313436941805e-05,
      "loss": 0.06734907627105713,
      "step": 739
    },
    {
      "epoch": 0.10004562891859463,
      "grad_norm": 0.16804896295070648,
      "learning_rate": 2.991960731794309e-05,
      "loss": 0.07374244928359985,
      "step": 740
    },
    {
      "epoch": 0.10018082571443057,
      "grad_norm": 0.12314272671937943,
      "learning_rate": 2.991889809260257e-05,
      "loss": 0.03489363193511963,
      "step": 741
    },
    {
      "epoch": 0.1003160225102665,
      "grad_norm": 0.15799863636493683,
      "learning_rate": 2.9918185761067912e-05,
      "loss": 0.0667719841003418,
      "step": 742
    },
    {
      "epoch": 0.10045121930610244,
      "grad_norm": 0.12534253299236298,
      "learning_rate": 2.9917470323487423e-05,
      "loss": 0.05903005599975586,
      "step": 743
    },
    {
      "epoch": 0.10058641610193839,
      "grad_norm": 0.10759688913822174,
      "learning_rate": 2.9916751780010063e-05,
      "loss": 0.0698390007019043,
      "step": 744
    },
    {
      "epoch": 0.10072161289777433,
      "grad_norm": 0.12864547967910767,
      "learning_rate": 2.9916030130785436e-05,
      "loss": 0.09099912643432617,
      "step": 745
    },
    {
      "epoch": 0.10085680969361026,
      "grad_norm": 0.20595256984233856,
      "learning_rate": 2.99153053759638e-05,
      "loss": 0.08784770965576172,
      "step": 746
    },
    {
      "epoch": 0.1009920064894462,
      "grad_norm": 0.17962120473384857,
      "learning_rate": 2.991457751569604e-05,
      "loss": 0.09574031829833984,
      "step": 747
    },
    {
      "epoch": 0.10112720328528214,
      "grad_norm": 0.17928659915924072,
      "learning_rate": 2.991384655013371e-05,
      "loss": 0.09390830993652344,
      "step": 748
    },
    {
      "epoch": 0.10126240008111807,
      "grad_norm": 0.3044493794441223,
      "learning_rate": 2.9913112479429e-05,
      "loss": 0.09512901306152344,
      "step": 749
    },
    {
      "epoch": 0.10139759687695402,
      "grad_norm": 0.19055142998695374,
      "learning_rate": 2.991237530373474e-05,
      "loss": 0.09835147857666016,
      "step": 750
    },
    {
      "epoch": 0.10153279367278996,
      "grad_norm": 0.08671168237924576,
      "learning_rate": 2.9911635023204423e-05,
      "loss": 0.05415821075439453,
      "step": 751
    },
    {
      "epoch": 0.10166799046862589,
      "grad_norm": 0.21877916157245636,
      "learning_rate": 2.9910891637992172e-05,
      "loss": 0.08973956108093262,
      "step": 752
    },
    {
      "epoch": 0.10180318726446183,
      "grad_norm": 0.1237107366323471,
      "learning_rate": 2.991014514825277e-05,
      "loss": 0.0618128776550293,
      "step": 753
    },
    {
      "epoch": 0.10193838406029777,
      "grad_norm": 0.20009958744049072,
      "learning_rate": 2.9909395554141638e-05,
      "loss": 0.06196999549865723,
      "step": 754
    },
    {
      "epoch": 0.10207358085613372,
      "grad_norm": 0.2767101228237152,
      "learning_rate": 2.9908642855814844e-05,
      "loss": 0.0626363754272461,
      "step": 755
    },
    {
      "epoch": 0.10220877765196965,
      "grad_norm": 0.12315303832292557,
      "learning_rate": 2.9907887053429107e-05,
      "loss": 0.10156726837158203,
      "step": 756
    },
    {
      "epoch": 0.10234397444780559,
      "grad_norm": 0.3240935802459717,
      "learning_rate": 2.9907128147141783e-05,
      "loss": 0.06920528411865234,
      "step": 757
    },
    {
      "epoch": 0.10247917124364153,
      "grad_norm": 0.3710269629955292,
      "learning_rate": 2.990636613711089e-05,
      "loss": 0.10161972045898438,
      "step": 758
    },
    {
      "epoch": 0.10261436803947746,
      "grad_norm": 0.29096660017967224,
      "learning_rate": 2.990560102349507e-05,
      "loss": 0.10406351089477539,
      "step": 759
    },
    {
      "epoch": 0.1027495648353134,
      "grad_norm": 0.23447288572788239,
      "learning_rate": 2.9904832806453635e-05,
      "loss": 0.07109212875366211,
      "step": 760
    },
    {
      "epoch": 0.10288476163114935,
      "grad_norm": 0.35149598121643066,
      "learning_rate": 2.9904061486146524e-05,
      "loss": 0.09916567802429199,
      "step": 761
    },
    {
      "epoch": 0.10301995842698528,
      "grad_norm": 0.18237677216529846,
      "learning_rate": 2.9903287062734333e-05,
      "loss": 0.06481671333312988,
      "step": 762
    },
    {
      "epoch": 0.10315515522282122,
      "grad_norm": 0.17816005647182465,
      "learning_rate": 2.990250953637831e-05,
      "loss": 0.08998489379882812,
      "step": 763
    },
    {
      "epoch": 0.10329035201865716,
      "grad_norm": 0.19520476460456848,
      "learning_rate": 2.9901728907240326e-05,
      "loss": 0.07070064544677734,
      "step": 764
    },
    {
      "epoch": 0.10342554881449309,
      "grad_norm": 0.4232277572154999,
      "learning_rate": 2.9900945175482916e-05,
      "loss": 0.11032295227050781,
      "step": 765
    },
    {
      "epoch": 0.10356074561032903,
      "grad_norm": 0.16400767862796783,
      "learning_rate": 2.990015834126926e-05,
      "loss": 0.09066009521484375,
      "step": 766
    },
    {
      "epoch": 0.10369594240616498,
      "grad_norm": 0.08228255063295364,
      "learning_rate": 2.989936840476318e-05,
      "loss": 0.06066131591796875,
      "step": 767
    },
    {
      "epoch": 0.10383113920200092,
      "grad_norm": 0.2542920410633087,
      "learning_rate": 2.9898575366129145e-05,
      "loss": 0.07512092590332031,
      "step": 768
    },
    {
      "epoch": 0.10396633599783685,
      "grad_norm": 0.2543547749519348,
      "learning_rate": 2.9897779225532273e-05,
      "loss": 0.08338165283203125,
      "step": 769
    },
    {
      "epoch": 0.10410153279367279,
      "grad_norm": 0.11839622259140015,
      "learning_rate": 2.989697998313832e-05,
      "loss": 0.0728607177734375,
      "step": 770
    },
    {
      "epoch": 0.10423672958950873,
      "grad_norm": 0.08058282732963562,
      "learning_rate": 2.989617763911369e-05,
      "loss": 0.07061290740966797,
      "step": 771
    },
    {
      "epoch": 0.10437192638534466,
      "grad_norm": 0.23767045140266418,
      "learning_rate": 2.9895372193625442e-05,
      "loss": 0.09198951721191406,
      "step": 772
    },
    {
      "epoch": 0.1045071231811806,
      "grad_norm": 0.12013694643974304,
      "learning_rate": 2.9894563646841273e-05,
      "loss": 0.06464433670043945,
      "step": 773
    },
    {
      "epoch": 0.10464231997701655,
      "grad_norm": 0.12468311935663223,
      "learning_rate": 2.9893751998929523e-05,
      "loss": 0.05020427703857422,
      "step": 774
    },
    {
      "epoch": 0.10477751677285248,
      "grad_norm": 0.19225336611270905,
      "learning_rate": 2.9892937250059187e-05,
      "loss": 0.07311725616455078,
      "step": 775
    },
    {
      "epoch": 0.10491271356868842,
      "grad_norm": 0.11506348848342896,
      "learning_rate": 2.9892119400399894e-05,
      "loss": 0.07399535179138184,
      "step": 776
    },
    {
      "epoch": 0.10504791036452436,
      "grad_norm": 0.052996713668107986,
      "learning_rate": 2.989129845012193e-05,
      "loss": 0.04266500473022461,
      "step": 777
    },
    {
      "epoch": 0.1051831071603603,
      "grad_norm": 0.41333791613578796,
      "learning_rate": 2.989047439939621e-05,
      "loss": 0.11240839958190918,
      "step": 778
    },
    {
      "epoch": 0.10531830395619624,
      "grad_norm": 0.16591086983680725,
      "learning_rate": 2.9889647248394324e-05,
      "loss": 0.08731245994567871,
      "step": 779
    },
    {
      "epoch": 0.10545350075203218,
      "grad_norm": 0.39323684573173523,
      "learning_rate": 2.9888816997288475e-05,
      "loss": 0.09854698181152344,
      "step": 780
    },
    {
      "epoch": 0.10558869754786812,
      "grad_norm": 0.18708565831184387,
      "learning_rate": 2.988798364625153e-05,
      "loss": 0.0652015209197998,
      "step": 781
    },
    {
      "epoch": 0.10572389434370405,
      "grad_norm": 0.1845155507326126,
      "learning_rate": 2.9887147195457002e-05,
      "loss": 0.10908269882202148,
      "step": 782
    },
    {
      "epoch": 0.10585909113953999,
      "grad_norm": 0.2472580224275589,
      "learning_rate": 2.9886307645079037e-05,
      "loss": 0.08659076690673828,
      "step": 783
    },
    {
      "epoch": 0.10599428793537594,
      "grad_norm": 0.12436066567897797,
      "learning_rate": 2.9885464995292436e-05,
      "loss": 0.09352922439575195,
      "step": 784
    },
    {
      "epoch": 0.10612948473121186,
      "grad_norm": 0.08763474971055984,
      "learning_rate": 2.9884619246272648e-05,
      "loss": 0.059453725814819336,
      "step": 785
    },
    {
      "epoch": 0.10626468152704781,
      "grad_norm": 0.22112756967544556,
      "learning_rate": 2.988377039819575e-05,
      "loss": 0.07128620147705078,
      "step": 786
    },
    {
      "epoch": 0.10639987832288375,
      "grad_norm": 0.2052437663078308,
      "learning_rate": 2.9882918451238494e-05,
      "loss": 0.059258460998535156,
      "step": 787
    },
    {
      "epoch": 0.10653507511871968,
      "grad_norm": 0.24719718098640442,
      "learning_rate": 2.988206340557825e-05,
      "loss": 0.08119010925292969,
      "step": 788
    },
    {
      "epoch": 0.10667027191455562,
      "grad_norm": 0.1153087466955185,
      "learning_rate": 2.9881205261393037e-05,
      "loss": 0.08741474151611328,
      "step": 789
    },
    {
      "epoch": 0.10680546871039157,
      "grad_norm": 0.07043509185314178,
      "learning_rate": 2.988034401886154e-05,
      "loss": 0.0507049560546875,
      "step": 790
    },
    {
      "epoch": 0.10694066550622751,
      "grad_norm": 0.15033923089504242,
      "learning_rate": 2.9879479678163065e-05,
      "loss": 0.09711217880249023,
      "step": 791
    },
    {
      "epoch": 0.10707586230206344,
      "grad_norm": 0.11357401311397552,
      "learning_rate": 2.9878612239477568e-05,
      "loss": 0.07277107238769531,
      "step": 792
    },
    {
      "epoch": 0.10721105909789938,
      "grad_norm": 0.3398059010505676,
      "learning_rate": 2.9877741702985666e-05,
      "loss": 0.08458948135375977,
      "step": 793
    },
    {
      "epoch": 0.10734625589373532,
      "grad_norm": 0.3241231143474579,
      "learning_rate": 2.98768680688686e-05,
      "loss": 0.08054161071777344,
      "step": 794
    },
    {
      "epoch": 0.10748145268957125,
      "grad_norm": 0.09487687796354294,
      "learning_rate": 2.9875991337308274e-05,
      "loss": 0.06821727752685547,
      "step": 795
    },
    {
      "epoch": 0.1076166494854072,
      "grad_norm": 0.2654559314250946,
      "learning_rate": 2.987511150848722e-05,
      "loss": 0.09479951858520508,
      "step": 796
    },
    {
      "epoch": 0.10775184628124314,
      "grad_norm": 0.1296280026435852,
      "learning_rate": 2.9874228582588627e-05,
      "loss": 0.06629657745361328,
      "step": 797
    },
    {
      "epoch": 0.10788704307707907,
      "grad_norm": 0.328866571187973,
      "learning_rate": 2.9873342559796325e-05,
      "loss": 0.0834050178527832,
      "step": 798
    },
    {
      "epoch": 0.10802223987291501,
      "grad_norm": 0.07810252904891968,
      "learning_rate": 2.9872453440294787e-05,
      "loss": 0.04237055778503418,
      "step": 799
    },
    {
      "epoch": 0.10815743666875095,
      "grad_norm": 0.18273776769638062,
      "learning_rate": 2.9871561224269134e-05,
      "loss": 0.08586692810058594,
      "step": 800
    },
    {
      "epoch": 0.1082926334645869,
      "grad_norm": 0.1715489625930786,
      "learning_rate": 2.9870665911905127e-05,
      "loss": 0.04636788368225098,
      "step": 801
    },
    {
      "epoch": 0.10842783026042283,
      "grad_norm": 0.301963746547699,
      "learning_rate": 2.9869767503389176e-05,
      "loss": 0.08504486083984375,
      "step": 802
    },
    {
      "epoch": 0.10856302705625877,
      "grad_norm": 0.13930180668830872,
      "learning_rate": 2.986886599890834e-05,
      "loss": 0.060051679611206055,
      "step": 803
    },
    {
      "epoch": 0.10869822385209471,
      "grad_norm": 0.18955019116401672,
      "learning_rate": 2.9867961398650306e-05,
      "loss": 0.0805504322052002,
      "step": 804
    },
    {
      "epoch": 0.10883342064793064,
      "grad_norm": 0.18851809203624725,
      "learning_rate": 2.9867053702803425e-05,
      "loss": 0.08764410018920898,
      "step": 805
    },
    {
      "epoch": 0.10896861744376658,
      "grad_norm": 0.0662231594324112,
      "learning_rate": 2.9866142911556685e-05,
      "loss": 0.056791067123413086,
      "step": 806
    },
    {
      "epoch": 0.10910381423960253,
      "grad_norm": 0.08169761300086975,
      "learning_rate": 2.9865229025099713e-05,
      "loss": 0.0470128059387207,
      "step": 807
    },
    {
      "epoch": 0.10923901103543845,
      "grad_norm": 0.15803854167461395,
      "learning_rate": 2.986431204362279e-05,
      "loss": 0.061837196350097656,
      "step": 808
    },
    {
      "epoch": 0.1093742078312744,
      "grad_norm": 0.3536088466644287,
      "learning_rate": 2.9863391967316835e-05,
      "loss": 0.11529064178466797,
      "step": 809
    },
    {
      "epoch": 0.10950940462711034,
      "grad_norm": 0.40793943405151367,
      "learning_rate": 2.9862468796373404e-05,
      "loss": 0.11717033386230469,
      "step": 810
    },
    {
      "epoch": 0.10964460142294627,
      "grad_norm": 0.2415672242641449,
      "learning_rate": 2.9861542530984718e-05,
      "loss": 0.05767989158630371,
      "step": 811
    },
    {
      "epoch": 0.10977979821878221,
      "grad_norm": 0.7584882378578186,
      "learning_rate": 2.9860613171343624e-05,
      "loss": 0.10846614837646484,
      "step": 812
    },
    {
      "epoch": 0.10991499501461816,
      "grad_norm": 0.516636073589325,
      "learning_rate": 2.9859680717643623e-05,
      "loss": 0.10421156883239746,
      "step": 813
    },
    {
      "epoch": 0.1100501918104541,
      "grad_norm": 0.13690637052059174,
      "learning_rate": 2.985874517007885e-05,
      "loss": 0.09915876388549805,
      "step": 814
    },
    {
      "epoch": 0.11018538860629003,
      "grad_norm": 0.32932132482528687,
      "learning_rate": 2.98578065288441e-05,
      "loss": 0.08774566650390625,
      "step": 815
    },
    {
      "epoch": 0.11032058540212597,
      "grad_norm": 0.3442085385322571,
      "learning_rate": 2.9856864794134798e-05,
      "loss": 0.0969090461730957,
      "step": 816
    },
    {
      "epoch": 0.11045578219796191,
      "grad_norm": 0.13868650794029236,
      "learning_rate": 2.9855919966147025e-05,
      "loss": 0.054688215255737305,
      "step": 817
    },
    {
      "epoch": 0.11059097899379784,
      "grad_norm": 0.19820687174797058,
      "learning_rate": 2.9854972045077485e-05,
      "loss": 0.07607817649841309,
      "step": 818
    },
    {
      "epoch": 0.11072617578963379,
      "grad_norm": 0.2938211262226105,
      "learning_rate": 2.9854021031123555e-05,
      "loss": 0.08491277694702148,
      "step": 819
    },
    {
      "epoch": 0.11086137258546973,
      "grad_norm": 0.1576143354177475,
      "learning_rate": 2.9853066924483232e-05,
      "loss": 0.09095048904418945,
      "step": 820
    },
    {
      "epoch": 0.11099656938130566,
      "grad_norm": 0.1545390784740448,
      "learning_rate": 2.9852109725355173e-05,
      "loss": 0.09976053237915039,
      "step": 821
    },
    {
      "epoch": 0.1111317661771416,
      "grad_norm": 0.2253318727016449,
      "learning_rate": 2.9851149433938662e-05,
      "loss": 0.08770465850830078,
      "step": 822
    },
    {
      "epoch": 0.11126696297297754,
      "grad_norm": 0.20439563691616058,
      "learning_rate": 2.9850186050433645e-05,
      "loss": 0.08408737182617188,
      "step": 823
    },
    {
      "epoch": 0.11140215976881349,
      "grad_norm": 0.21274667978286743,
      "learning_rate": 2.9849219575040708e-05,
      "loss": 0.06603193283081055,
      "step": 824
    },
    {
      "epoch": 0.11153735656464941,
      "grad_norm": 0.20621870458126068,
      "learning_rate": 2.984825000796106e-05,
      "loss": 0.057804107666015625,
      "step": 825
    },
    {
      "epoch": 0.11167255336048536,
      "grad_norm": 0.0705530196428299,
      "learning_rate": 2.9847277349396586e-05,
      "loss": 0.043094635009765625,
      "step": 826
    },
    {
      "epoch": 0.1118077501563213,
      "grad_norm": 0.4527442753314972,
      "learning_rate": 2.984630159954979e-05,
      "loss": 0.10841941833496094,
      "step": 827
    },
    {
      "epoch": 0.11194294695215723,
      "grad_norm": 0.08340879529714584,
      "learning_rate": 2.9845322758623833e-05,
      "loss": 0.0617976188659668,
      "step": 828
    },
    {
      "epoch": 0.11207814374799317,
      "grad_norm": 0.19706197082996368,
      "learning_rate": 2.984434082682251e-05,
      "loss": 0.07596111297607422,
      "step": 829
    },
    {
      "epoch": 0.11221334054382912,
      "grad_norm": 0.1969340741634369,
      "learning_rate": 2.984335580435027e-05,
      "loss": 0.06574010848999023,
      "step": 830
    },
    {
      "epoch": 0.11234853733966504,
      "grad_norm": 0.08985655009746552,
      "learning_rate": 2.9842367691412192e-05,
      "loss": 0.06956326961517334,
      "step": 831
    },
    {
      "epoch": 0.11248373413550099,
      "grad_norm": 0.21356859803199768,
      "learning_rate": 2.9841376488214015e-05,
      "loss": 0.1064915657043457,
      "step": 832
    },
    {
      "epoch": 0.11261893093133693,
      "grad_norm": 0.12347124516963959,
      "learning_rate": 2.984038219496211e-05,
      "loss": 0.07301950454711914,
      "step": 833
    },
    {
      "epoch": 0.11275412772717286,
      "grad_norm": 0.1629743129014969,
      "learning_rate": 2.9839384811863493e-05,
      "loss": 0.044960975646972656,
      "step": 834
    },
    {
      "epoch": 0.1128893245230088,
      "grad_norm": 0.0998326987028122,
      "learning_rate": 2.9838384339125824e-05,
      "loss": 0.08784270286560059,
      "step": 835
    },
    {
      "epoch": 0.11302452131884475,
      "grad_norm": 0.3424812853336334,
      "learning_rate": 2.9837380776957405e-05,
      "loss": 0.09475421905517578,
      "step": 836
    },
    {
      "epoch": 0.11315971811468069,
      "grad_norm": 0.11877866834402084,
      "learning_rate": 2.9836374125567193e-05,
      "loss": 0.07374238967895508,
      "step": 837
    },
    {
      "epoch": 0.11329491491051662,
      "grad_norm": 0.11771272122859955,
      "learning_rate": 2.9835364385164764e-05,
      "loss": 0.0850377082824707,
      "step": 838
    },
    {
      "epoch": 0.11343011170635256,
      "grad_norm": 0.18212012946605682,
      "learning_rate": 2.983435155596036e-05,
      "loss": 0.07221508026123047,
      "step": 839
    },
    {
      "epoch": 0.1135653085021885,
      "grad_norm": 0.2089482694864273,
      "learning_rate": 2.9833335638164858e-05,
      "loss": 0.06322050094604492,
      "step": 840
    },
    {
      "epoch": 0.11370050529802443,
      "grad_norm": 0.11180892586708069,
      "learning_rate": 2.9832316631989774e-05,
      "loss": 0.08978652954101562,
      "step": 841
    },
    {
      "epoch": 0.11383570209386037,
      "grad_norm": 0.10422135889530182,
      "learning_rate": 2.9831294537647272e-05,
      "loss": 0.0949702262878418,
      "step": 842
    },
    {
      "epoch": 0.11397089888969632,
      "grad_norm": 0.0864141434431076,
      "learning_rate": 2.9830269355350155e-05,
      "loss": 0.040509819984436035,
      "step": 843
    },
    {
      "epoch": 0.11410609568553225,
      "grad_norm": 0.10074340552091599,
      "learning_rate": 2.9829241085311872e-05,
      "loss": 0.08821964263916016,
      "step": 844
    },
    {
      "epoch": 0.11424129248136819,
      "grad_norm": 0.10178901255130768,
      "learning_rate": 2.9828209727746522e-05,
      "loss": 0.08177757263183594,
      "step": 845
    },
    {
      "epoch": 0.11437648927720413,
      "grad_norm": 0.40882372856140137,
      "learning_rate": 2.982717528286883e-05,
      "loss": 0.09736502170562744,
      "step": 846
    },
    {
      "epoch": 0.11451168607304008,
      "grad_norm": 0.20394711196422577,
      "learning_rate": 2.9826137750894176e-05,
      "loss": 0.06695842742919922,
      "step": 847
    },
    {
      "epoch": 0.114646882868876,
      "grad_norm": 0.2837764620780945,
      "learning_rate": 2.9825097132038578e-05,
      "loss": 0.11022567749023438,
      "step": 848
    },
    {
      "epoch": 0.11478207966471195,
      "grad_norm": 0.17595382034778595,
      "learning_rate": 2.9824053426518703e-05,
      "loss": 0.06178617477416992,
      "step": 849
    },
    {
      "epoch": 0.11491727646054789,
      "grad_norm": 0.08936288952827454,
      "learning_rate": 2.9823006634551848e-05,
      "loss": 0.042883872985839844,
      "step": 850
    },
    {
      "epoch": 0.11505247325638382,
      "grad_norm": 0.0923553854227066,
      "learning_rate": 2.9821956756355973e-05,
      "loss": 0.052956581115722656,
      "step": 851
    },
    {
      "epoch": 0.11518767005221976,
      "grad_norm": 0.15883170068264008,
      "learning_rate": 2.9820903792149653e-05,
      "loss": 0.0631551742553711,
      "step": 852
    },
    {
      "epoch": 0.1153228668480557,
      "grad_norm": 0.12394951283931732,
      "learning_rate": 2.981984774215214e-05,
      "loss": 0.056340694427490234,
      "step": 853
    },
    {
      "epoch": 0.11545806364389163,
      "grad_norm": 0.16585946083068848,
      "learning_rate": 2.9818788606583286e-05,
      "loss": 0.0750889778137207,
      "step": 854
    },
    {
      "epoch": 0.11559326043972758,
      "grad_norm": 0.11545179039239883,
      "learning_rate": 2.9817726385663627e-05,
      "loss": 0.06773948669433594,
      "step": 855
    },
    {
      "epoch": 0.11572845723556352,
      "grad_norm": 0.06577248126268387,
      "learning_rate": 2.9816661079614316e-05,
      "loss": 0.03995513916015625,
      "step": 856
    },
    {
      "epoch": 0.11586365403139945,
      "grad_norm": 0.09831870347261429,
      "learning_rate": 2.9815592688657154e-05,
      "loss": 0.05192971229553223,
      "step": 857
    },
    {
      "epoch": 0.11599885082723539,
      "grad_norm": 0.2290714532136917,
      "learning_rate": 2.9814521213014588e-05,
      "loss": 0.0670008659362793,
      "step": 858
    },
    {
      "epoch": 0.11613404762307133,
      "grad_norm": 0.07488477975130081,
      "learning_rate": 2.9813446652909707e-05,
      "loss": 0.05266571044921875,
      "step": 859
    },
    {
      "epoch": 0.11626924441890728,
      "grad_norm": 0.07828672975301743,
      "learning_rate": 2.981236900856624e-05,
      "loss": 0.08294129371643066,
      "step": 860
    },
    {
      "epoch": 0.1164044412147432,
      "grad_norm": 0.15161457657814026,
      "learning_rate": 2.9811288280208552e-05,
      "loss": 0.07205438613891602,
      "step": 861
    },
    {
      "epoch": 0.11653963801057915,
      "grad_norm": 0.24158193171024323,
      "learning_rate": 2.9810204468061664e-05,
      "loss": 0.06822800636291504,
      "step": 862
    },
    {
      "epoch": 0.11667483480641509,
      "grad_norm": 0.09941220283508301,
      "learning_rate": 2.9809117572351223e-05,
      "loss": 0.09187650680541992,
      "step": 863
    },
    {
      "epoch": 0.11681003160225102,
      "grad_norm": 0.2299419641494751,
      "learning_rate": 2.9808027593303537e-05,
      "loss": 0.0739750862121582,
      "step": 864
    },
    {
      "epoch": 0.11694522839808696,
      "grad_norm": 0.24607360363006592,
      "learning_rate": 2.980693453114554e-05,
      "loss": 0.09745121002197266,
      "step": 865
    },
    {
      "epoch": 0.11708042519392291,
      "grad_norm": 0.09835226088762283,
      "learning_rate": 2.980583838610481e-05,
      "loss": 0.06097698211669922,
      "step": 866
    },
    {
      "epoch": 0.11721562198975884,
      "grad_norm": 0.1356116682291031,
      "learning_rate": 2.980473915840957e-05,
      "loss": 0.048302650451660156,
      "step": 867
    },
    {
      "epoch": 0.11735081878559478,
      "grad_norm": 0.09727887064218521,
      "learning_rate": 2.9803636848288696e-05,
      "loss": 0.062195539474487305,
      "step": 868
    },
    {
      "epoch": 0.11748601558143072,
      "grad_norm": 0.09921512007713318,
      "learning_rate": 2.9802531455971686e-05,
      "loss": 0.11292636394500732,
      "step": 869
    },
    {
      "epoch": 0.11762121237726667,
      "grad_norm": 0.13198482990264893,
      "learning_rate": 2.980142298168869e-05,
      "loss": 0.0755453109741211,
      "step": 870
    },
    {
      "epoch": 0.1177564091731026,
      "grad_norm": 0.1393927037715912,
      "learning_rate": 2.9800311425670495e-05,
      "loss": 0.08868122100830078,
      "step": 871
    },
    {
      "epoch": 0.11789160596893854,
      "grad_norm": 0.1366884410381317,
      "learning_rate": 2.9799196788148538e-05,
      "loss": 0.07724571228027344,
      "step": 872
    },
    {
      "epoch": 0.11802680276477448,
      "grad_norm": 0.19587735831737518,
      "learning_rate": 2.9798079069354893e-05,
      "loss": 0.06855130195617676,
      "step": 873
    },
    {
      "epoch": 0.11816199956061041,
      "grad_norm": 0.10926695168018341,
      "learning_rate": 2.9796958269522273e-05,
      "loss": 0.0879659652709961,
      "step": 874
    },
    {
      "epoch": 0.11829719635644635,
      "grad_norm": 0.09635842591524124,
      "learning_rate": 2.9795834388884034e-05,
      "loss": 0.08609294891357422,
      "step": 875
    },
    {
      "epoch": 0.1184323931522823,
      "grad_norm": 0.2814229726791382,
      "learning_rate": 2.979470742767417e-05,
      "loss": 0.10121011734008789,
      "step": 876
    },
    {
      "epoch": 0.11856758994811822,
      "grad_norm": 0.06718067824840546,
      "learning_rate": 2.9793577386127327e-05,
      "loss": 0.048627495765686035,
      "step": 877
    },
    {
      "epoch": 0.11870278674395417,
      "grad_norm": 0.2509063482284546,
      "learning_rate": 2.9792444264478784e-05,
      "loss": 0.09848976135253906,
      "step": 878
    },
    {
      "epoch": 0.11883798353979011,
      "grad_norm": 0.1740858256816864,
      "learning_rate": 2.979130806296446e-05,
      "loss": 0.0972909927368164,
      "step": 879
    },
    {
      "epoch": 0.11897318033562605,
      "grad_norm": 0.1094849482178688,
      "learning_rate": 2.9790168781820925e-05,
      "loss": 0.10734844207763672,
      "step": 880
    },
    {
      "epoch": 0.11910837713146198,
      "grad_norm": 0.11823206394910812,
      "learning_rate": 2.9789026421285375e-05,
      "loss": 0.046347856521606445,
      "step": 881
    },
    {
      "epoch": 0.11924357392729792,
      "grad_norm": 0.17739428579807281,
      "learning_rate": 2.9787880981595663e-05,
      "loss": 0.056365013122558594,
      "step": 882
    },
    {
      "epoch": 0.11937877072313387,
      "grad_norm": 0.08121839165687561,
      "learning_rate": 2.9786732462990267e-05,
      "loss": 0.03412294387817383,
      "step": 883
    },
    {
      "epoch": 0.1195139675189698,
      "grad_norm": 0.10083824396133423,
      "learning_rate": 2.9785580865708323e-05,
      "loss": 0.08484792709350586,
      "step": 884
    },
    {
      "epoch": 0.11964916431480574,
      "grad_norm": 0.0700761154294014,
      "learning_rate": 2.97844261899896e-05,
      "loss": 0.07140254974365234,
      "step": 885
    },
    {
      "epoch": 0.11978436111064168,
      "grad_norm": 0.19332964718341827,
      "learning_rate": 2.9783268436074495e-05,
      "loss": 0.0855860710144043,
      "step": 886
    },
    {
      "epoch": 0.11991955790647761,
      "grad_norm": 0.07933145016431808,
      "learning_rate": 2.978210760420407e-05,
      "loss": 0.06059074401855469,
      "step": 887
    },
    {
      "epoch": 0.12005475470231355,
      "grad_norm": 0.26722756028175354,
      "learning_rate": 2.978094369462002e-05,
      "loss": 0.05805253982543945,
      "step": 888
    },
    {
      "epoch": 0.1201899514981495,
      "grad_norm": 0.1603512465953827,
      "learning_rate": 2.977977670756467e-05,
      "loss": 0.09630918502807617,
      "step": 889
    },
    {
      "epoch": 0.12032514829398543,
      "grad_norm": 0.10066097229719162,
      "learning_rate": 2.9778606643280987e-05,
      "loss": 0.05611252784729004,
      "step": 890
    },
    {
      "epoch": 0.12046034508982137,
      "grad_norm": 0.5782535076141357,
      "learning_rate": 2.97774335020126e-05,
      "loss": 0.1215205192565918,
      "step": 891
    },
    {
      "epoch": 0.12059554188565731,
      "grad_norm": 0.10206296294927597,
      "learning_rate": 2.9776257284003748e-05,
      "loss": 0.06757831573486328,
      "step": 892
    },
    {
      "epoch": 0.12073073868149325,
      "grad_norm": 0.14233049750328064,
      "learning_rate": 2.9775077989499338e-05,
      "loss": 0.057381629943847656,
      "step": 893
    },
    {
      "epoch": 0.12086593547732918,
      "grad_norm": 0.11968830227851868,
      "learning_rate": 2.97738956187449e-05,
      "loss": 0.06141304969787598,
      "step": 894
    },
    {
      "epoch": 0.12100113227316513,
      "grad_norm": 0.35995978116989136,
      "learning_rate": 2.9772710171986605e-05,
      "loss": 0.0654524564743042,
      "step": 895
    },
    {
      "epoch": 0.12113632906900107,
      "grad_norm": 0.44974932074546814,
      "learning_rate": 2.977152164947128e-05,
      "loss": 0.10231733322143555,
      "step": 896
    },
    {
      "epoch": 0.121271525864837,
      "grad_norm": 0.17205770313739777,
      "learning_rate": 2.9770330051446373e-05,
      "loss": 0.08052492141723633,
      "step": 897
    },
    {
      "epoch": 0.12140672266067294,
      "grad_norm": 0.1110287681221962,
      "learning_rate": 2.976913537815999e-05,
      "loss": 0.0387725830078125,
      "step": 898
    },
    {
      "epoch": 0.12154191945650888,
      "grad_norm": 0.1932668685913086,
      "learning_rate": 2.9767937629860853e-05,
      "loss": 0.06683921813964844,
      "step": 899
    },
    {
      "epoch": 0.12167711625234481,
      "grad_norm": 0.3327488601207733,
      "learning_rate": 2.9766736806798353e-05,
      "loss": 0.11077547073364258,
      "step": 900
    },
    {
      "epoch": 0.12181231304818076,
      "grad_norm": 0.3324854373931885,
      "learning_rate": 2.9765532909222512e-05,
      "loss": 0.11431884765625,
      "step": 901
    },
    {
      "epoch": 0.1219475098440167,
      "grad_norm": 0.16411514580249786,
      "learning_rate": 2.976432593738397e-05,
      "loss": 0.07039260864257812,
      "step": 902
    },
    {
      "epoch": 0.12208270663985264,
      "grad_norm": 0.19517670571804047,
      "learning_rate": 2.9763115891534036e-05,
      "loss": 0.07362890243530273,
      "step": 903
    },
    {
      "epoch": 0.12221790343568857,
      "grad_norm": 0.09529386460781097,
      "learning_rate": 2.9761902771924648e-05,
      "loss": 0.07908391952514648,
      "step": 904
    },
    {
      "epoch": 0.12235310023152451,
      "grad_norm": 0.47957566380500793,
      "learning_rate": 2.9760686578808387e-05,
      "loss": 0.11242246627807617,
      "step": 905
    },
    {
      "epoch": 0.12248829702736046,
      "grad_norm": 0.08849037438631058,
      "learning_rate": 2.9759467312438462e-05,
      "loss": 0.05784595012664795,
      "step": 906
    },
    {
      "epoch": 0.12262349382319639,
      "grad_norm": 0.2830328643321991,
      "learning_rate": 2.975824497306874e-05,
      "loss": 0.07963132858276367,
      "step": 907
    },
    {
      "epoch": 0.12275869061903233,
      "grad_norm": 0.08956493437290192,
      "learning_rate": 2.9757019560953707e-05,
      "loss": 0.055341243743896484,
      "step": 908
    },
    {
      "epoch": 0.12289388741486827,
      "grad_norm": 0.20775741338729858,
      "learning_rate": 2.9755791076348517e-05,
      "loss": 0.08177757263183594,
      "step": 909
    },
    {
      "epoch": 0.1230290842107042,
      "grad_norm": 0.1353548914194107,
      "learning_rate": 2.9754559519508924e-05,
      "loss": 0.05702543258666992,
      "step": 910
    },
    {
      "epoch": 0.12316428100654014,
      "grad_norm": 0.20207025110721588,
      "learning_rate": 2.975332489069137e-05,
      "loss": 0.11015200614929199,
      "step": 911
    },
    {
      "epoch": 0.12329947780237609,
      "grad_norm": 0.13142721354961395,
      "learning_rate": 2.9752087190152893e-05,
      "loss": 0.06446266174316406,
      "step": 912
    },
    {
      "epoch": 0.12343467459821202,
      "grad_norm": 0.061199259012937546,
      "learning_rate": 2.97508464181512e-05,
      "loss": 0.04959380626678467,
      "step": 913
    },
    {
      "epoch": 0.12356987139404796,
      "grad_norm": 0.05604036897420883,
      "learning_rate": 2.9749602574944615e-05,
      "loss": 0.06218576431274414,
      "step": 914
    },
    {
      "epoch": 0.1237050681898839,
      "grad_norm": 0.050905805081129074,
      "learning_rate": 2.9748355660792125e-05,
      "loss": 0.058411598205566406,
      "step": 915
    },
    {
      "epoch": 0.12384026498571984,
      "grad_norm": 0.07387355715036392,
      "learning_rate": 2.9747105675953338e-05,
      "loss": 0.06812095642089844,
      "step": 916
    },
    {
      "epoch": 0.12397546178155577,
      "grad_norm": 0.09652038663625717,
      "learning_rate": 2.9745852620688506e-05,
      "loss": 0.06022357940673828,
      "step": 917
    },
    {
      "epoch": 0.12411065857739172,
      "grad_norm": 0.09810265898704529,
      "learning_rate": 2.974459649525853e-05,
      "loss": 0.09165668487548828,
      "step": 918
    },
    {
      "epoch": 0.12424585537322766,
      "grad_norm": 0.1119033694267273,
      "learning_rate": 2.9743337299924925e-05,
      "loss": 0.06702995300292969,
      "step": 919
    },
    {
      "epoch": 0.12438105216906359,
      "grad_norm": 0.1399790346622467,
      "learning_rate": 2.9742075034949883e-05,
      "loss": 0.0867147445678711,
      "step": 920
    },
    {
      "epoch": 0.12451624896489953,
      "grad_norm": 0.07810115814208984,
      "learning_rate": 2.97408097005962e-05,
      "loss": 0.046642303466796875,
      "step": 921
    },
    {
      "epoch": 0.12465144576073547,
      "grad_norm": 0.06395487487316132,
      "learning_rate": 2.973954129712733e-05,
      "loss": 0.06725311279296875,
      "step": 922
    },
    {
      "epoch": 0.1247866425565714,
      "grad_norm": 0.11322113126516342,
      "learning_rate": 2.973826982480736e-05,
      "loss": 0.08335018157958984,
      "step": 923
    },
    {
      "epoch": 0.12492183935240735,
      "grad_norm": 0.11602670699357986,
      "learning_rate": 2.9736995283901022e-05,
      "loss": 0.06998157501220703,
      "step": 924
    },
    {
      "epoch": 0.1250570361482433,
      "grad_norm": 0.08187684416770935,
      "learning_rate": 2.9735717674673676e-05,
      "loss": 0.05464744567871094,
      "step": 925
    },
    {
      "epoch": 0.12519223294407922,
      "grad_norm": 0.06290099024772644,
      "learning_rate": 2.973443699739133e-05,
      "loss": 0.06893730163574219,
      "step": 926
    },
    {
      "epoch": 0.12532742973991517,
      "grad_norm": 0.08969343453645706,
      "learning_rate": 2.973315325232063e-05,
      "loss": 0.07574081420898438,
      "step": 927
    },
    {
      "epoch": 0.1254626265357511,
      "grad_norm": 0.060889434069395065,
      "learning_rate": 2.9731866439728853e-05,
      "loss": 0.05339348316192627,
      "step": 928
    },
    {
      "epoch": 0.12559782333158703,
      "grad_norm": 0.1132717877626419,
      "learning_rate": 2.9730576559883924e-05,
      "loss": 0.07853174209594727,
      "step": 929
    },
    {
      "epoch": 0.125733020127423,
      "grad_norm": 0.18988798558712006,
      "learning_rate": 2.97292836130544e-05,
      "loss": 0.10422897338867188,
      "step": 930
    },
    {
      "epoch": 0.12586821692325892,
      "grad_norm": 0.24295906722545624,
      "learning_rate": 2.9727987599509485e-05,
      "loss": 0.10706615447998047,
      "step": 931
    },
    {
      "epoch": 0.12600341371909485,
      "grad_norm": 0.08881915360689163,
      "learning_rate": 2.972668851951901e-05,
      "loss": 0.08465337753295898,
      "step": 932
    },
    {
      "epoch": 0.1261386105149308,
      "grad_norm": 0.14923670887947083,
      "learning_rate": 2.9725386373353455e-05,
      "loss": 0.08785533905029297,
      "step": 933
    },
    {
      "epoch": 0.12627380731076673,
      "grad_norm": 0.15341024100780487,
      "learning_rate": 2.972408116128393e-05,
      "loss": 0.0878748893737793,
      "step": 934
    },
    {
      "epoch": 0.12640900410660266,
      "grad_norm": 0.11445598304271698,
      "learning_rate": 2.972277288358219e-05,
      "loss": 0.04725363850593567,
      "step": 935
    },
    {
      "epoch": 0.12654420090243862,
      "grad_norm": 0.1122552752494812,
      "learning_rate": 2.9721461540520628e-05,
      "loss": 0.05743694305419922,
      "step": 936
    },
    {
      "epoch": 0.12667939769827455,
      "grad_norm": 0.09951417148113251,
      "learning_rate": 2.9720147132372265e-05,
      "loss": 0.06026816368103027,
      "step": 937
    },
    {
      "epoch": 0.12681459449411048,
      "grad_norm": 0.22975791990756989,
      "learning_rate": 2.9718829659410772e-05,
      "loss": 0.07405209541320801,
      "step": 938
    },
    {
      "epoch": 0.12694979128994643,
      "grad_norm": 0.33335110545158386,
      "learning_rate": 2.9717509121910453e-05,
      "loss": 0.09210205078125,
      "step": 939
    },
    {
      "epoch": 0.12708498808578236,
      "grad_norm": 0.1875367909669876,
      "learning_rate": 2.971618552014625e-05,
      "loss": 0.09831809997558594,
      "step": 940
    },
    {
      "epoch": 0.1272201848816183,
      "grad_norm": 0.16147367656230927,
      "learning_rate": 2.971485885439375e-05,
      "loss": 0.08818912506103516,
      "step": 941
    },
    {
      "epoch": 0.12735538167745425,
      "grad_norm": 0.2637520134449005,
      "learning_rate": 2.9713529124929163e-05,
      "loss": 0.07030582427978516,
      "step": 942
    },
    {
      "epoch": 0.12749057847329018,
      "grad_norm": 0.17659729719161987,
      "learning_rate": 2.9712196332029352e-05,
      "loss": 0.08257389068603516,
      "step": 943
    },
    {
      "epoch": 0.12762577526912613,
      "grad_norm": 0.1788230687379837,
      "learning_rate": 2.971086047597181e-05,
      "loss": 0.05866813659667969,
      "step": 944
    },
    {
      "epoch": 0.12776097206496206,
      "grad_norm": 0.09143399447202682,
      "learning_rate": 2.9709521557034668e-05,
      "loss": 0.07986783981323242,
      "step": 945
    },
    {
      "epoch": 0.127896168860798,
      "grad_norm": 0.05168140307068825,
      "learning_rate": 2.9708179575496696e-05,
      "loss": 0.04710984230041504,
      "step": 946
    },
    {
      "epoch": 0.12803136565663395,
      "grad_norm": 0.17141243815422058,
      "learning_rate": 2.9706834531637303e-05,
      "loss": 0.07384252548217773,
      "step": 947
    },
    {
      "epoch": 0.12816656245246988,
      "grad_norm": 0.09293242543935776,
      "learning_rate": 2.9705486425736537e-05,
      "loss": 0.05687761306762695,
      "step": 948
    },
    {
      "epoch": 0.1283017592483058,
      "grad_norm": 0.13287778198719025,
      "learning_rate": 2.9704135258075077e-05,
      "loss": 0.09968948364257812,
      "step": 949
    },
    {
      "epoch": 0.12843695604414176,
      "grad_norm": 0.17102405428886414,
      "learning_rate": 2.970278102893424e-05,
      "loss": 0.08391761779785156,
      "step": 950
    },
    {
      "epoch": 0.1285721528399777,
      "grad_norm": 0.07498197257518768,
      "learning_rate": 2.9701423738595992e-05,
      "loss": 0.05570793151855469,
      "step": 951
    },
    {
      "epoch": 0.12870734963581362,
      "grad_norm": 0.2092883437871933,
      "learning_rate": 2.9700063387342925e-05,
      "loss": 0.08389425277709961,
      "step": 952
    },
    {
      "epoch": 0.12884254643164958,
      "grad_norm": 0.11228356510400772,
      "learning_rate": 2.969869997545827e-05,
      "loss": 0.06782150268554688,
      "step": 953
    },
    {
      "epoch": 0.1289777432274855,
      "grad_norm": 0.31297019124031067,
      "learning_rate": 2.9697333503225897e-05,
      "loss": 0.09388470649719238,
      "step": 954
    },
    {
      "epoch": 0.12911294002332144,
      "grad_norm": 0.1070636734366417,
      "learning_rate": 2.969596397093031e-05,
      "loss": 0.04721426963806152,
      "step": 955
    },
    {
      "epoch": 0.1292481368191574,
      "grad_norm": 0.4006432294845581,
      "learning_rate": 2.969459137885666e-05,
      "loss": 0.09582674503326416,
      "step": 956
    },
    {
      "epoch": 0.12938333361499332,
      "grad_norm": 0.05496162921190262,
      "learning_rate": 2.969321572729072e-05,
      "loss": 0.057157039642333984,
      "step": 957
    },
    {
      "epoch": 0.12951853041082925,
      "grad_norm": 0.1691799759864807,
      "learning_rate": 2.9691837016518915e-05,
      "loss": 0.10239124298095703,
      "step": 958
    },
    {
      "epoch": 0.1296537272066652,
      "grad_norm": 0.11953835934400558,
      "learning_rate": 2.9690455246828294e-05,
      "loss": 0.0597691535949707,
      "step": 959
    },
    {
      "epoch": 0.12978892400250114,
      "grad_norm": 0.10983915627002716,
      "learning_rate": 2.968907041850655e-05,
      "loss": 0.08515787124633789,
      "step": 960
    },
    {
      "epoch": 0.12992412079833707,
      "grad_norm": 0.4553511440753937,
      "learning_rate": 2.968768253184202e-05,
      "loss": 0.10332679748535156,
      "step": 961
    },
    {
      "epoch": 0.13005931759417302,
      "grad_norm": 0.20180925726890564,
      "learning_rate": 2.9686291587123655e-05,
      "loss": 0.06601858139038086,
      "step": 962
    },
    {
      "epoch": 0.13019451439000895,
      "grad_norm": 0.2608233690261841,
      "learning_rate": 2.968489758464107e-05,
      "loss": 0.08244514465332031,
      "step": 963
    },
    {
      "epoch": 0.13032971118584488,
      "grad_norm": 0.16600099205970764,
      "learning_rate": 2.9683500524684494e-05,
      "loss": 0.09819841384887695,
      "step": 964
    },
    {
      "epoch": 0.13046490798168084,
      "grad_norm": 0.10266660153865814,
      "learning_rate": 2.9682100407544812e-05,
      "loss": 0.09746980667114258,
      "step": 965
    },
    {
      "epoch": 0.13060010477751677,
      "grad_norm": 0.21525733172893524,
      "learning_rate": 2.9680697233513526e-05,
      "loss": 0.07324165105819702,
      "step": 966
    },
    {
      "epoch": 0.13073530157335272,
      "grad_norm": 0.0983826220035553,
      "learning_rate": 2.9679291002882793e-05,
      "loss": 0.04398226737976074,
      "step": 967
    },
    {
      "epoch": 0.13087049836918865,
      "grad_norm": 0.08117310702800751,
      "learning_rate": 2.967788171594539e-05,
      "loss": 0.06145191192626953,
      "step": 968
    },
    {
      "epoch": 0.13100569516502458,
      "grad_norm": 0.06245778501033783,
      "learning_rate": 2.967646937299474e-05,
      "loss": 0.057662010192871094,
      "step": 969
    },
    {
      "epoch": 0.13114089196086054,
      "grad_norm": 0.1285935938358307,
      "learning_rate": 2.9675053974324907e-05,
      "loss": 0.05845296382904053,
      "step": 970
    },
    {
      "epoch": 0.13127608875669647,
      "grad_norm": 0.25850266218185425,
      "learning_rate": 2.9673635520230576e-05,
      "loss": 0.09070491790771484,
      "step": 971
    },
    {
      "epoch": 0.1314112855525324,
      "grad_norm": 0.10657283663749695,
      "learning_rate": 2.9672214011007087e-05,
      "loss": 0.06935310363769531,
      "step": 972
    },
    {
      "epoch": 0.13154648234836835,
      "grad_norm": 0.14891095459461212,
      "learning_rate": 2.9670789446950396e-05,
      "loss": 0.07125735282897949,
      "step": 973
    },
    {
      "epoch": 0.13168167914420428,
      "grad_norm": 0.06608948856592178,
      "learning_rate": 2.9669361828357105e-05,
      "loss": 0.07768774032592773,
      "step": 974
    },
    {
      "epoch": 0.1318168759400402,
      "grad_norm": 0.4589276611804962,
      "learning_rate": 2.9667931155524454e-05,
      "loss": 0.12001943588256836,
      "step": 975
    },
    {
      "epoch": 0.13195207273587617,
      "grad_norm": 0.11258334666490555,
      "learning_rate": 2.966649742875032e-05,
      "loss": 0.06550455093383789,
      "step": 976
    },
    {
      "epoch": 0.1320872695317121,
      "grad_norm": 0.11503694951534271,
      "learning_rate": 2.9665060648333206e-05,
      "loss": 0.06824827194213867,
      "step": 977
    },
    {
      "epoch": 0.13222246632754803,
      "grad_norm": 0.10666774213314056,
      "learning_rate": 2.9663620814572266e-05,
      "loss": 0.06955242156982422,
      "step": 978
    },
    {
      "epoch": 0.13235766312338398,
      "grad_norm": 0.2170712649822235,
      "learning_rate": 2.966217792776728e-05,
      "loss": 0.10490036010742188,
      "step": 979
    },
    {
      "epoch": 0.1324928599192199,
      "grad_norm": 0.17030830681324005,
      "learning_rate": 2.9660731988218652e-05,
      "loss": 0.1233377456665039,
      "step": 980
    },
    {
      "epoch": 0.13262805671505584,
      "grad_norm": 0.22178776562213898,
      "learning_rate": 2.965928299622745e-05,
      "loss": 0.11655044555664062,
      "step": 981
    },
    {
      "epoch": 0.1327632535108918,
      "grad_norm": 0.17455221712589264,
      "learning_rate": 2.965783095209535e-05,
      "loss": 0.07438206672668457,
      "step": 982
    },
    {
      "epoch": 0.13289845030672773,
      "grad_norm": 0.10594431310892105,
      "learning_rate": 2.965637585612469e-05,
      "loss": 0.06237363815307617,
      "step": 983
    },
    {
      "epoch": 0.13303364710256366,
      "grad_norm": 0.22808212041854858,
      "learning_rate": 2.965491770861841e-05,
      "loss": 0.06763434410095215,
      "step": 984
    },
    {
      "epoch": 0.1331688438983996,
      "grad_norm": 0.09387168288230896,
      "learning_rate": 2.965345650988012e-05,
      "loss": 0.06642758846282959,
      "step": 985
    },
    {
      "epoch": 0.13330404069423554,
      "grad_norm": 0.20906609296798706,
      "learning_rate": 2.9651992260214035e-05,
      "loss": 0.055144548416137695,
      "step": 986
    },
    {
      "epoch": 0.13343923749007147,
      "grad_norm": 0.12606149911880493,
      "learning_rate": 2.9650524959925037e-05,
      "loss": 0.06200456619262695,
      "step": 987
    },
    {
      "epoch": 0.13357443428590743,
      "grad_norm": 0.19453994929790497,
      "learning_rate": 2.9649054609318607e-05,
      "loss": 0.05778694152832031,
      "step": 988
    },
    {
      "epoch": 0.13370963108174336,
      "grad_norm": 0.0841677114367485,
      "learning_rate": 2.9647581208700894e-05,
      "loss": 0.04187750816345215,
      "step": 989
    },
    {
      "epoch": 0.13384482787757931,
      "grad_norm": 0.06964724510908127,
      "learning_rate": 2.9646104758378666e-05,
      "loss": 0.040574073791503906,
      "step": 990
    },
    {
      "epoch": 0.13398002467341524,
      "grad_norm": 0.14623387157917023,
      "learning_rate": 2.964462525865932e-05,
      "loss": 0.08375215530395508,
      "step": 991
    },
    {
      "epoch": 0.13411522146925117,
      "grad_norm": 0.12396002560853958,
      "learning_rate": 2.96431427098509e-05,
      "loss": 0.042246341705322266,
      "step": 992
    },
    {
      "epoch": 0.13425041826508713,
      "grad_norm": 0.07503268122673035,
      "learning_rate": 2.9641657112262084e-05,
      "loss": 0.07427215576171875,
      "step": 993
    },
    {
      "epoch": 0.13438561506092306,
      "grad_norm": 0.18358756601810455,
      "learning_rate": 2.9640168466202174e-05,
      "loss": 0.0798649787902832,
      "step": 994
    },
    {
      "epoch": 0.134520811856759,
      "grad_norm": 0.06345956027507782,
      "learning_rate": 2.9638676771981124e-05,
      "loss": 0.07236003875732422,
      "step": 995
    },
    {
      "epoch": 0.13465600865259494,
      "grad_norm": 0.13814058899879456,
      "learning_rate": 2.9637182029909508e-05,
      "loss": 0.074737548828125,
      "step": 996
    },
    {
      "epoch": 0.13479120544843087,
      "grad_norm": 0.19358345866203308,
      "learning_rate": 2.9635684240298532e-05,
      "loss": 0.08238029479980469,
      "step": 997
    },
    {
      "epoch": 0.1349264022442668,
      "grad_norm": 0.1957382708787918,
      "learning_rate": 2.9634183403460053e-05,
      "loss": 0.09319698810577393,
      "step": 998
    },
    {
      "epoch": 0.13506159904010276,
      "grad_norm": 0.25816795229911804,
      "learning_rate": 2.9632679519706553e-05,
      "loss": 0.05713081359863281,
      "step": 999
    },
    {
      "epoch": 0.1351967958359387,
      "grad_norm": 0.08890276402235031,
      "learning_rate": 2.9631172589351137e-05,
      "loss": 0.06299352645874023,
      "step": 1000
    },
    {
      "epoch": 0.13533199263177462,
      "grad_norm": 0.08693654090166092,
      "learning_rate": 2.962966261270758e-05,
      "loss": 0.07686614990234375,
      "step": 1001
    },
    {
      "epoch": 0.13546718942761057,
      "grad_norm": 0.22788509726524353,
      "learning_rate": 2.962814959009024e-05,
      "loss": 0.07744884490966797,
      "step": 1002
    },
    {
      "epoch": 0.1356023862234465,
      "grad_norm": 0.07897160202264786,
      "learning_rate": 2.962663352181415e-05,
      "loss": 0.07285690307617188,
      "step": 1003
    },
    {
      "epoch": 0.13573758301928243,
      "grad_norm": 0.08798522502183914,
      "learning_rate": 2.9625114408194966e-05,
      "loss": 0.08118867874145508,
      "step": 1004
    },
    {
      "epoch": 0.1358727798151184,
      "grad_norm": 0.058871470391750336,
      "learning_rate": 2.962359224954897e-05,
      "loss": 0.07728958129882812,
      "step": 1005
    },
    {
      "epoch": 0.13600797661095432,
      "grad_norm": 0.10668406635522842,
      "learning_rate": 2.9622067046193086e-05,
      "loss": 0.07386445999145508,
      "step": 1006
    },
    {
      "epoch": 0.13614317340679025,
      "grad_norm": 0.06813656538724899,
      "learning_rate": 2.9620538798444867e-05,
      "loss": 0.05845308303833008,
      "step": 1007
    },
    {
      "epoch": 0.1362783702026262,
      "grad_norm": 0.06310522556304932,
      "learning_rate": 2.9619007506622506e-05,
      "loss": 0.05810260772705078,
      "step": 1008
    },
    {
      "epoch": 0.13641356699846213,
      "grad_norm": 0.10126934200525284,
      "learning_rate": 2.961747317104482e-05,
      "loss": 0.05495619773864746,
      "step": 1009
    },
    {
      "epoch": 0.13654876379429806,
      "grad_norm": 0.09578578919172287,
      "learning_rate": 2.9615935792031274e-05,
      "loss": 0.06580543518066406,
      "step": 1010
    },
    {
      "epoch": 0.13668396059013402,
      "grad_norm": 0.05667395517230034,
      "learning_rate": 2.9614395369901953e-05,
      "loss": 0.05591392517089844,
      "step": 1011
    },
    {
      "epoch": 0.13681915738596995,
      "grad_norm": 0.15220583975315094,
      "learning_rate": 2.9612851904977582e-05,
      "loss": 0.07385396957397461,
      "step": 1012
    },
    {
      "epoch": 0.1369543541818059,
      "grad_norm": 0.08858183771371841,
      "learning_rate": 2.9611305397579518e-05,
      "loss": 0.058804988861083984,
      "step": 1013
    },
    {
      "epoch": 0.13708955097764183,
      "grad_norm": 0.192491352558136,
      "learning_rate": 2.9609755848029755e-05,
      "loss": 0.08931732177734375,
      "step": 1014
    },
    {
      "epoch": 0.13722474777347776,
      "grad_norm": 0.08235607296228409,
      "learning_rate": 2.9608203256650916e-05,
      "loss": 0.06106531620025635,
      "step": 1015
    },
    {
      "epoch": 0.13735994456931372,
      "grad_norm": 0.05607544630765915,
      "learning_rate": 2.9606647623766257e-05,
      "loss": 0.06089353561401367,
      "step": 1016
    },
    {
      "epoch": 0.13749514136514965,
      "grad_norm": 0.09464474767446518,
      "learning_rate": 2.9605088949699672e-05,
      "loss": 0.05215764045715332,
      "step": 1017
    },
    {
      "epoch": 0.13763033816098558,
      "grad_norm": 0.23505446314811707,
      "learning_rate": 2.9603527234775682e-05,
      "loss": 0.0972222089767456,
      "step": 1018
    },
    {
      "epoch": 0.13776553495682153,
      "grad_norm": 0.17249640822410583,
      "learning_rate": 2.960196247931945e-05,
      "loss": 0.07334613800048828,
      "step": 1019
    },
    {
      "epoch": 0.13790073175265746,
      "grad_norm": 0.13054914772510529,
      "learning_rate": 2.960039468365676e-05,
      "loss": 0.06658458709716797,
      "step": 1020
    },
    {
      "epoch": 0.1380359285484934,
      "grad_norm": 0.08485963940620422,
      "learning_rate": 2.959882384811404e-05,
      "loss": 0.053723812103271484,
      "step": 1021
    },
    {
      "epoch": 0.13817112534432935,
      "grad_norm": 0.1172117218375206,
      "learning_rate": 2.9597249973018343e-05,
      "loss": 0.07741475105285645,
      "step": 1022
    },
    {
      "epoch": 0.13830632214016528,
      "grad_norm": 0.10905864089727402,
      "learning_rate": 2.959567305869736e-05,
      "loss": 0.075042724609375,
      "step": 1023
    },
    {
      "epoch": 0.1384415189360012,
      "grad_norm": 0.23235218226909637,
      "learning_rate": 2.9594093105479413e-05,
      "loss": 0.08209609985351562,
      "step": 1024
    },
    {
      "epoch": 0.13857671573183716,
      "grad_norm": 0.1621515452861786,
      "learning_rate": 2.959251011369345e-05,
      "loss": 0.07379913330078125,
      "step": 1025
    },
    {
      "epoch": 0.1387119125276731,
      "grad_norm": 0.13144813477993011,
      "learning_rate": 2.959092408366907e-05,
      "loss": 0.06781506538391113,
      "step": 1026
    },
    {
      "epoch": 0.13884710932350902,
      "grad_norm": 0.17865842580795288,
      "learning_rate": 2.958933501573649e-05,
      "loss": 0.10501575469970703,
      "step": 1027
    },
    {
      "epoch": 0.13898230611934498,
      "grad_norm": 0.18086294829845428,
      "learning_rate": 2.9587742910226555e-05,
      "loss": 0.06138896942138672,
      "step": 1028
    },
    {
      "epoch": 0.1391175029151809,
      "grad_norm": 0.24017727375030518,
      "learning_rate": 2.958614776747076e-05,
      "loss": 0.09089183807373047,
      "step": 1029
    },
    {
      "epoch": 0.13925269971101684,
      "grad_norm": 0.163410484790802,
      "learning_rate": 2.9584549587801213e-05,
      "loss": 0.10097527503967285,
      "step": 1030
    },
    {
      "epoch": 0.1393878965068528,
      "grad_norm": 0.12211813032627106,
      "learning_rate": 2.958294837155067e-05,
      "loss": 0.08792757987976074,
      "step": 1031
    },
    {
      "epoch": 0.13952309330268872,
      "grad_norm": 0.08845590800046921,
      "learning_rate": 2.9581344119052508e-05,
      "loss": 0.05278968811035156,
      "step": 1032
    },
    {
      "epoch": 0.13965829009852465,
      "grad_norm": 0.2417311668395996,
      "learning_rate": 2.957973683064074e-05,
      "loss": 0.08494710922241211,
      "step": 1033
    },
    {
      "epoch": 0.1397934868943606,
      "grad_norm": 0.09509695321321487,
      "learning_rate": 2.957812650665002e-05,
      "loss": 0.08342361450195312,
      "step": 1034
    },
    {
      "epoch": 0.13992868369019654,
      "grad_norm": 0.08689370006322861,
      "learning_rate": 2.957651314741562e-05,
      "loss": 0.06669139862060547,
      "step": 1035
    },
    {
      "epoch": 0.1400638804860325,
      "grad_norm": 0.14130592346191406,
      "learning_rate": 2.9574896753273454e-05,
      "loss": 0.07112407684326172,
      "step": 1036
    },
    {
      "epoch": 0.14019907728186842,
      "grad_norm": 0.10611394792795181,
      "learning_rate": 2.9573277324560058e-05,
      "loss": 0.09284114837646484,
      "step": 1037
    },
    {
      "epoch": 0.14033427407770435,
      "grad_norm": 0.09717437624931335,
      "learning_rate": 2.9571654861612608e-05,
      "loss": 0.06455516815185547,
      "step": 1038
    },
    {
      "epoch": 0.1404694708735403,
      "grad_norm": 0.09705877304077148,
      "learning_rate": 2.957002936476891e-05,
      "loss": 0.10404205322265625,
      "step": 1039
    },
    {
      "epoch": 0.14060466766937624,
      "grad_norm": 0.07513929158449173,
      "learning_rate": 2.9568400834367406e-05,
      "loss": 0.043250322341918945,
      "step": 1040
    },
    {
      "epoch": 0.14073986446521217,
      "grad_norm": 0.2907352149486542,
      "learning_rate": 2.9566769270747158e-05,
      "loss": 0.07195138931274414,
      "step": 1041
    },
    {
      "epoch": 0.14087506126104812,
      "grad_norm": 0.2050797939300537,
      "learning_rate": 2.9565134674247864e-05,
      "loss": 0.0687749981880188,
      "step": 1042
    },
    {
      "epoch": 0.14101025805688405,
      "grad_norm": 0.39528849720954895,
      "learning_rate": 2.9563497045209866e-05,
      "loss": 0.09494972229003906,
      "step": 1043
    },
    {
      "epoch": 0.14114545485271998,
      "grad_norm": 0.07782010734081268,
      "learning_rate": 2.9561856383974118e-05,
      "loss": 0.06363105773925781,
      "step": 1044
    },
    {
      "epoch": 0.14128065164855594,
      "grad_norm": 0.1474486142396927,
      "learning_rate": 2.9560212690882218e-05,
      "loss": 0.05806732177734375,
      "step": 1045
    },
    {
      "epoch": 0.14141584844439187,
      "grad_norm": 0.10805214196443558,
      "learning_rate": 2.9558565966276395e-05,
      "loss": 0.07311058044433594,
      "step": 1046
    },
    {
      "epoch": 0.1415510452402278,
      "grad_norm": 0.4037792384624481,
      "learning_rate": 2.9556916210499497e-05,
      "loss": 0.09949612617492676,
      "step": 1047
    },
    {
      "epoch": 0.14168624203606375,
      "grad_norm": 0.1432960480451584,
      "learning_rate": 2.9555263423895016e-05,
      "loss": 0.0778653621673584,
      "step": 1048
    },
    {
      "epoch": 0.14182143883189968,
      "grad_norm": 0.056540507823228836,
      "learning_rate": 2.955360760680708e-05,
      "loss": 0.0507662296295166,
      "step": 1049
    },
    {
      "epoch": 0.1419566356277356,
      "grad_norm": 0.09269817918539047,
      "learning_rate": 2.9551948759580423e-05,
      "loss": 0.06925535202026367,
      "step": 1050
    },
    {
      "epoch": 0.14209183242357157,
      "grad_norm": 0.23423974215984344,
      "learning_rate": 2.9550286882560435e-05,
      "loss": 0.14203643798828125,
      "step": 1051
    },
    {
      "epoch": 0.1422270292194075,
      "grad_norm": 0.04959813877940178,
      "learning_rate": 2.9548621976093126e-05,
      "loss": 0.05577993392944336,
      "step": 1052
    },
    {
      "epoch": 0.14236222601524343,
      "grad_norm": 0.2520340383052826,
      "learning_rate": 2.9546954040525144e-05,
      "loss": 0.08503293991088867,
      "step": 1053
    },
    {
      "epoch": 0.14249742281107938,
      "grad_norm": 0.12083239108324051,
      "learning_rate": 2.9545283076203753e-05,
      "loss": 0.05936098098754883,
      "step": 1054
    },
    {
      "epoch": 0.1426326196069153,
      "grad_norm": 0.2398243099451065,
      "learning_rate": 2.954360908347686e-05,
      "loss": 0.0962214469909668,
      "step": 1055
    },
    {
      "epoch": 0.14276781640275127,
      "grad_norm": 0.09604228287935257,
      "learning_rate": 2.9541932062693e-05,
      "loss": 0.06707572937011719,
      "step": 1056
    },
    {
      "epoch": 0.1429030131985872,
      "grad_norm": 0.08386726677417755,
      "learning_rate": 2.954025201420134e-05,
      "loss": 0.07052421569824219,
      "step": 1057
    },
    {
      "epoch": 0.14303820999442313,
      "grad_norm": 0.5823044180870056,
      "learning_rate": 2.9538568938351672e-05,
      "loss": 0.10803794860839844,
      "step": 1058
    },
    {
      "epoch": 0.14317340679025908,
      "grad_norm": 0.0700690820813179,
      "learning_rate": 2.953688283549442e-05,
      "loss": 0.08406209945678711,
      "step": 1059
    },
    {
      "epoch": 0.143308603586095,
      "grad_norm": 0.09271769225597382,
      "learning_rate": 2.9535193705980642e-05,
      "loss": 0.09754276275634766,
      "step": 1060
    },
    {
      "epoch": 0.14344380038193094,
      "grad_norm": 0.18816334009170532,
      "learning_rate": 2.9533501550162028e-05,
      "loss": 0.12133526802062988,
      "step": 1061
    },
    {
      "epoch": 0.1435789971777669,
      "grad_norm": 0.06169494241476059,
      "learning_rate": 2.9531806368390882e-05,
      "loss": 0.04880106449127197,
      "step": 1062
    },
    {
      "epoch": 0.14371419397360283,
      "grad_norm": 0.11677113175392151,
      "learning_rate": 2.953010816102016e-05,
      "loss": 0.0968770980834961,
      "step": 1063
    },
    {
      "epoch": 0.14384939076943876,
      "grad_norm": 0.34681445360183716,
      "learning_rate": 2.952840692840343e-05,
      "loss": 0.08517265319824219,
      "step": 1064
    },
    {
      "epoch": 0.1439845875652747,
      "grad_norm": 0.19578534364700317,
      "learning_rate": 2.9526702670894914e-05,
      "loss": 0.09182238578796387,
      "step": 1065
    },
    {
      "epoch": 0.14411978436111064,
      "grad_norm": 0.06605084240436554,
      "learning_rate": 2.952499538884943e-05,
      "loss": 0.04653024673461914,
      "step": 1066
    },
    {
      "epoch": 0.14425498115694657,
      "grad_norm": 0.08458984643220901,
      "learning_rate": 2.9523285082622448e-05,
      "loss": 0.07516765594482422,
      "step": 1067
    },
    {
      "epoch": 0.14439017795278253,
      "grad_norm": 0.23083150386810303,
      "learning_rate": 2.9521571752570064e-05,
      "loss": 0.0717320442199707,
      "step": 1068
    },
    {
      "epoch": 0.14452537474861846,
      "grad_norm": 0.1745014637708664,
      "learning_rate": 2.9519855399049004e-05,
      "loss": 0.08561468124389648,
      "step": 1069
    },
    {
      "epoch": 0.14466057154445439,
      "grad_norm": 0.09659496694803238,
      "learning_rate": 2.951813602241662e-05,
      "loss": 0.07809066772460938,
      "step": 1070
    },
    {
      "epoch": 0.14479576834029034,
      "grad_norm": 0.2586137056350708,
      "learning_rate": 2.9516413623030896e-05,
      "loss": 0.09674263000488281,
      "step": 1071
    },
    {
      "epoch": 0.14493096513612627,
      "grad_norm": 0.10419290512800217,
      "learning_rate": 2.951468820125045e-05,
      "loss": 0.0826728343963623,
      "step": 1072
    },
    {
      "epoch": 0.1450661619319622,
      "grad_norm": 0.12646038830280304,
      "learning_rate": 2.9512959757434508e-05,
      "loss": 0.07571268081665039,
      "step": 1073
    },
    {
      "epoch": 0.14520135872779816,
      "grad_norm": 0.1377960443496704,
      "learning_rate": 2.951122829194296e-05,
      "loss": 0.057172298431396484,
      "step": 1074
    },
    {
      "epoch": 0.1453365555236341,
      "grad_norm": 0.24179475009441376,
      "learning_rate": 2.9509493805136296e-05,
      "loss": 0.05516242980957031,
      "step": 1075
    },
    {
      "epoch": 0.14547175231947002,
      "grad_norm": 0.251004695892334,
      "learning_rate": 2.9507756297375648e-05,
      "loss": 0.09329891204833984,
      "step": 1076
    },
    {
      "epoch": 0.14560694911530597,
      "grad_norm": 0.12214822322130203,
      "learning_rate": 2.9506015769022778e-05,
      "loss": 0.07416343688964844,
      "step": 1077
    },
    {
      "epoch": 0.1457421459111419,
      "grad_norm": 0.11316990107297897,
      "learning_rate": 2.950427222044006e-05,
      "loss": 0.06862330436706543,
      "step": 1078
    },
    {
      "epoch": 0.14587734270697786,
      "grad_norm": 0.10652310401201248,
      "learning_rate": 2.9502525651990525e-05,
      "loss": 0.07209396362304688,
      "step": 1079
    },
    {
      "epoch": 0.1460125395028138,
      "grad_norm": 0.14423726499080658,
      "learning_rate": 2.9500776064037813e-05,
      "loss": 0.058995723724365234,
      "step": 1080
    },
    {
      "epoch": 0.14614773629864972,
      "grad_norm": 0.16166767477989197,
      "learning_rate": 2.9499023456946194e-05,
      "loss": 0.0825796127319336,
      "step": 1081
    },
    {
      "epoch": 0.14628293309448567,
      "grad_norm": 0.2889252305030823,
      "learning_rate": 2.9497267831080575e-05,
      "loss": 0.08362865447998047,
      "step": 1082
    },
    {
      "epoch": 0.1464181298903216,
      "grad_norm": 0.11203325539827347,
      "learning_rate": 2.949550918680649e-05,
      "loss": 0.0511401891708374,
      "step": 1083
    },
    {
      "epoch": 0.14655332668615753,
      "grad_norm": 0.09853683412075043,
      "learning_rate": 2.9493747524490086e-05,
      "loss": 0.049675941467285156,
      "step": 1084
    },
    {
      "epoch": 0.1466885234819935,
      "grad_norm": 0.0966378003358841,
      "learning_rate": 2.9491982844498156e-05,
      "loss": 0.04838848114013672,
      "step": 1085
    },
    {
      "epoch": 0.14682372027782942,
      "grad_norm": 0.21662436425685883,
      "learning_rate": 2.949021514719812e-05,
      "loss": 0.07179498672485352,
      "step": 1086
    },
    {
      "epoch": 0.14695891707366535,
      "grad_norm": 0.08794362843036652,
      "learning_rate": 2.948844443295802e-05,
      "loss": 0.0547785758972168,
      "step": 1087
    },
    {
      "epoch": 0.1470941138695013,
      "grad_norm": 0.1866329461336136,
      "learning_rate": 2.9486670702146526e-05,
      "loss": 0.07671117782592773,
      "step": 1088
    },
    {
      "epoch": 0.14722931066533723,
      "grad_norm": 0.1171988844871521,
      "learning_rate": 2.948489395513294e-05,
      "loss": 0.051657676696777344,
      "step": 1089
    },
    {
      "epoch": 0.14736450746117316,
      "grad_norm": 0.17662739753723145,
      "learning_rate": 2.948311419228719e-05,
      "loss": 0.047513484954833984,
      "step": 1090
    },
    {
      "epoch": 0.14749970425700912,
      "grad_norm": 0.1720883548259735,
      "learning_rate": 2.948133141397983e-05,
      "loss": 0.07854557037353516,
      "step": 1091
    },
    {
      "epoch": 0.14763490105284505,
      "grad_norm": 0.09083975851535797,
      "learning_rate": 2.9479545620582047e-05,
      "loss": 0.067718505859375,
      "step": 1092
    },
    {
      "epoch": 0.14777009784868098,
      "grad_norm": 0.19623054563999176,
      "learning_rate": 2.9477756812465652e-05,
      "loss": 0.09048688411712646,
      "step": 1093
    },
    {
      "epoch": 0.14790529464451693,
      "grad_norm": 0.11081378161907196,
      "learning_rate": 2.9475964990003085e-05,
      "loss": 0.07282352447509766,
      "step": 1094
    },
    {
      "epoch": 0.14804049144035286,
      "grad_norm": 0.21468673646450043,
      "learning_rate": 2.9474170153567406e-05,
      "loss": 0.09254264831542969,
      "step": 1095
    },
    {
      "epoch": 0.1481756882361888,
      "grad_norm": 0.14591306447982788,
      "learning_rate": 2.947237230353232e-05,
      "loss": 0.05003070831298828,
      "step": 1096
    },
    {
      "epoch": 0.14831088503202475,
      "grad_norm": 0.10723714530467987,
      "learning_rate": 2.9470571440272147e-05,
      "loss": 0.07667946815490723,
      "step": 1097
    },
    {
      "epoch": 0.14844608182786068,
      "grad_norm": 0.11086497455835342,
      "learning_rate": 2.946876756416183e-05,
      "loss": 0.08315134048461914,
      "step": 1098
    },
    {
      "epoch": 0.1485812786236966,
      "grad_norm": 0.17267684638500214,
      "learning_rate": 2.946696067557695e-05,
      "loss": 0.0909261703491211,
      "step": 1099
    },
    {
      "epoch": 0.14871647541953256,
      "grad_norm": 0.2377883642911911,
      "learning_rate": 2.9465150774893706e-05,
      "loss": 0.08050012588500977,
      "step": 1100
    },
    {
      "epoch": 0.1488516722153685,
      "grad_norm": 0.16752363741397858,
      "learning_rate": 2.9463337862488938e-05,
      "loss": 0.05095905065536499,
      "step": 1101
    },
    {
      "epoch": 0.14898686901120445,
      "grad_norm": 0.2799507975578308,
      "learning_rate": 2.9461521938740096e-05,
      "loss": 0.0885920524597168,
      "step": 1102
    },
    {
      "epoch": 0.14912206580704038,
      "grad_norm": 0.06536594033241272,
      "learning_rate": 2.9459703004025273e-05,
      "loss": 0.050568580627441406,
      "step": 1103
    },
    {
      "epoch": 0.1492572626028763,
      "grad_norm": 0.06007133796811104,
      "learning_rate": 2.9457881058723174e-05,
      "loss": 0.06733274459838867,
      "step": 1104
    },
    {
      "epoch": 0.14939245939871226,
      "grad_norm": 0.2571377158164978,
      "learning_rate": 2.9456056103213137e-05,
      "loss": 0.07812309265136719,
      "step": 1105
    },
    {
      "epoch": 0.1495276561945482,
      "grad_norm": 0.4716612994670868,
      "learning_rate": 2.945422813787513e-05,
      "loss": 0.11354351043701172,
      "step": 1106
    },
    {
      "epoch": 0.14966285299038412,
      "grad_norm": 0.3563525676727295,
      "learning_rate": 2.9452397163089748e-05,
      "loss": 0.06363677978515625,
      "step": 1107
    },
    {
      "epoch": 0.14979804978622008,
      "grad_norm": 0.07930450141429901,
      "learning_rate": 2.9450563179238207e-05,
      "loss": 0.07386207580566406,
      "step": 1108
    },
    {
      "epoch": 0.149933246582056,
      "grad_norm": 0.15830406546592712,
      "learning_rate": 2.9448726186702354e-05,
      "loss": 0.06419610977172852,
      "step": 1109
    },
    {
      "epoch": 0.15006844337789194,
      "grad_norm": 0.41483670473098755,
      "learning_rate": 2.9446886185864652e-05,
      "loss": 0.07947468757629395,
      "step": 1110
    },
    {
      "epoch": 0.1502036401737279,
      "grad_norm": 0.23662076890468597,
      "learning_rate": 2.944504317710821e-05,
      "loss": 0.06407642364501953,
      "step": 1111
    },
    {
      "epoch": 0.15033883696956382,
      "grad_norm": 0.26178011298179626,
      "learning_rate": 2.944319716081675e-05,
      "loss": 0.048445701599121094,
      "step": 1112
    },
    {
      "epoch": 0.15047403376539975,
      "grad_norm": 0.20574241876602173,
      "learning_rate": 2.944134813737462e-05,
      "loss": 0.0809546709060669,
      "step": 1113
    },
    {
      "epoch": 0.1506092305612357,
      "grad_norm": 0.14405415952205658,
      "learning_rate": 2.9439496107166796e-05,
      "loss": 0.09885573387145996,
      "step": 1114
    },
    {
      "epoch": 0.15074442735707164,
      "grad_norm": 0.07743999361991882,
      "learning_rate": 2.943764107057888e-05,
      "loss": 0.033701419830322266,
      "step": 1115
    },
    {
      "epoch": 0.15087962415290757,
      "grad_norm": 0.36796656250953674,
      "learning_rate": 2.9435783027997106e-05,
      "loss": 0.06601905822753906,
      "step": 1116
    },
    {
      "epoch": 0.15101482094874352,
      "grad_norm": 0.17643453180789948,
      "learning_rate": 2.9433921979808323e-05,
      "loss": 0.06286144256591797,
      "step": 1117
    },
    {
      "epoch": 0.15115001774457945,
      "grad_norm": 0.3632143437862396,
      "learning_rate": 2.9432057926400014e-05,
      "loss": 0.07956409454345703,
      "step": 1118
    },
    {
      "epoch": 0.15128521454041538,
      "grad_norm": 0.23792868852615356,
      "learning_rate": 2.943019086816028e-05,
      "loss": 0.05762052536010742,
      "step": 1119
    },
    {
      "epoch": 0.15142041133625134,
      "grad_norm": 0.12486688047647476,
      "learning_rate": 2.9428320805477855e-05,
      "loss": 0.06322145462036133,
      "step": 1120
    },
    {
      "epoch": 0.15155560813208727,
      "grad_norm": 0.22102205455303192,
      "learning_rate": 2.9426447738742104e-05,
      "loss": 0.07183337211608887,
      "step": 1121
    },
    {
      "epoch": 0.1516908049279232,
      "grad_norm": 0.17263171076774597,
      "learning_rate": 2.9424571668343e-05,
      "loss": 0.06249523162841797,
      "step": 1122
    },
    {
      "epoch": 0.15182600172375915,
      "grad_norm": 0.17273354530334473,
      "learning_rate": 2.942269259467115e-05,
      "loss": 0.060373783111572266,
      "step": 1123
    },
    {
      "epoch": 0.15196119851959508,
      "grad_norm": 0.1259409636259079,
      "learning_rate": 2.9420810518117794e-05,
      "loss": 0.08330488204956055,
      "step": 1124
    },
    {
      "epoch": 0.15209639531543104,
      "grad_norm": 0.2565913498401642,
      "learning_rate": 2.9418925439074784e-05,
      "loss": 0.053517818450927734,
      "step": 1125
    },
    {
      "epoch": 0.15223159211126697,
      "grad_norm": 0.09067431092262268,
      "learning_rate": 2.9417037357934606e-05,
      "loss": 0.06913185119628906,
      "step": 1126
    },
    {
      "epoch": 0.1523667889071029,
      "grad_norm": 0.06976161897182465,
      "learning_rate": 2.9415146275090373e-05,
      "loss": 0.04553675651550293,
      "step": 1127
    },
    {
      "epoch": 0.15250198570293885,
      "grad_norm": 0.0911419540643692,
      "learning_rate": 2.9413252190935813e-05,
      "loss": 0.07584905624389648,
      "step": 1128
    },
    {
      "epoch": 0.15263718249877478,
      "grad_norm": 0.07555969804525375,
      "learning_rate": 2.9411355105865286e-05,
      "loss": 0.08169174194335938,
      "step": 1129
    },
    {
      "epoch": 0.1527723792946107,
      "grad_norm": 0.13209667801856995,
      "learning_rate": 2.9409455020273775e-05,
      "loss": 0.059699058532714844,
      "step": 1130
    },
    {
      "epoch": 0.15290757609044667,
      "grad_norm": 0.07090654969215393,
      "learning_rate": 2.940755193455689e-05,
      "loss": 0.06534671783447266,
      "step": 1131
    },
    {
      "epoch": 0.1530427728862826,
      "grad_norm": 0.12138225883245468,
      "learning_rate": 2.940564584911086e-05,
      "loss": 0.10901832580566406,
      "step": 1132
    },
    {
      "epoch": 0.15317796968211853,
      "grad_norm": 0.13750989735126495,
      "learning_rate": 2.9403736764332543e-05,
      "loss": 0.0717926025390625,
      "step": 1133
    },
    {
      "epoch": 0.15331316647795448,
      "grad_norm": 0.19666792452335358,
      "learning_rate": 2.9401824680619423e-05,
      "loss": 0.03712654113769531,
      "step": 1134
    },
    {
      "epoch": 0.1534483632737904,
      "grad_norm": 0.09003041684627533,
      "learning_rate": 2.9399909598369604e-05,
      "loss": 0.0582585334777832,
      "step": 1135
    },
    {
      "epoch": 0.15358356006962634,
      "grad_norm": 0.09080123901367188,
      "learning_rate": 2.939799151798182e-05,
      "loss": 0.0539020299911499,
      "step": 1136
    },
    {
      "epoch": 0.1537187568654623,
      "grad_norm": 0.21130475401878357,
      "learning_rate": 2.9396070439855417e-05,
      "loss": 0.0970611572265625,
      "step": 1137
    },
    {
      "epoch": 0.15385395366129823,
      "grad_norm": 0.2650197446346283,
      "learning_rate": 2.9394146364390382e-05,
      "loss": 0.0815286636352539,
      "step": 1138
    },
    {
      "epoch": 0.15398915045713416,
      "grad_norm": 0.2304464727640152,
      "learning_rate": 2.9392219291987315e-05,
      "loss": 0.06618404388427734,
      "step": 1139
    },
    {
      "epoch": 0.1541243472529701,
      "grad_norm": 0.10913332551717758,
      "learning_rate": 2.939028922304744e-05,
      "loss": 0.06919169425964355,
      "step": 1140
    },
    {
      "epoch": 0.15425954404880604,
      "grad_norm": 0.17635436356067657,
      "learning_rate": 2.9388356157972615e-05,
      "loss": 0.0587315559387207,
      "step": 1141
    },
    {
      "epoch": 0.15439474084464197,
      "grad_norm": 0.1624658703804016,
      "learning_rate": 2.938642009716531e-05,
      "loss": 0.06929469108581543,
      "step": 1142
    },
    {
      "epoch": 0.15452993764047793,
      "grad_norm": 0.09626515954732895,
      "learning_rate": 2.938448104102862e-05,
      "loss": 0.05512237548828125,
      "step": 1143
    },
    {
      "epoch": 0.15466513443631386,
      "grad_norm": 0.35350921750068665,
      "learning_rate": 2.9382538989966267e-05,
      "loss": 0.08420944213867188,
      "step": 1144
    },
    {
      "epoch": 0.15480033123214978,
      "grad_norm": 0.055737074464559555,
      "learning_rate": 2.9380593944382605e-05,
      "loss": 0.06031656265258789,
      "step": 1145
    },
    {
      "epoch": 0.15493552802798574,
      "grad_norm": 0.09223190695047379,
      "learning_rate": 2.9378645904682596e-05,
      "loss": 0.06427526473999023,
      "step": 1146
    },
    {
      "epoch": 0.15507072482382167,
      "grad_norm": 0.22569283843040466,
      "learning_rate": 2.937669487127183e-05,
      "loss": 0.061815738677978516,
      "step": 1147
    },
    {
      "epoch": 0.15520592161965763,
      "grad_norm": 0.11823631823062897,
      "learning_rate": 2.9374740844556532e-05,
      "loss": 0.09103202819824219,
      "step": 1148
    },
    {
      "epoch": 0.15534111841549356,
      "grad_norm": 0.274277925491333,
      "learning_rate": 2.937278382494353e-05,
      "loss": 0.09007656574249268,
      "step": 1149
    },
    {
      "epoch": 0.15547631521132949,
      "grad_norm": 0.15700848400592804,
      "learning_rate": 2.9370823812840287e-05,
      "loss": 0.08530592918395996,
      "step": 1150
    },
    {
      "epoch": 0.15561151200716544,
      "grad_norm": 0.09494221955537796,
      "learning_rate": 2.93688608086549e-05,
      "loss": 0.07379150390625,
      "step": 1151
    },
    {
      "epoch": 0.15574670880300137,
      "grad_norm": 0.08665430545806885,
      "learning_rate": 2.9366894812796064e-05,
      "loss": 0.07693290710449219,
      "step": 1152
    },
    {
      "epoch": 0.1558819055988373,
      "grad_norm": 0.17276400327682495,
      "learning_rate": 2.9364925825673117e-05,
      "loss": 0.06704044342041016,
      "step": 1153
    },
    {
      "epoch": 0.15601710239467326,
      "grad_norm": 0.1022331714630127,
      "learning_rate": 2.9362953847696006e-05,
      "loss": 0.06880712509155273,
      "step": 1154
    },
    {
      "epoch": 0.1561522991905092,
      "grad_norm": 0.10129193961620331,
      "learning_rate": 2.9360978879275313e-05,
      "loss": 0.06425905227661133,
      "step": 1155
    },
    {
      "epoch": 0.15628749598634512,
      "grad_norm": 0.1395132690668106,
      "learning_rate": 2.9359000920822237e-05,
      "loss": 0.09534454345703125,
      "step": 1156
    },
    {
      "epoch": 0.15642269278218107,
      "grad_norm": 0.05284448713064194,
      "learning_rate": 2.9357019972748594e-05,
      "loss": 0.04510068893432617,
      "step": 1157
    },
    {
      "epoch": 0.156557889578017,
      "grad_norm": 0.10269604623317719,
      "learning_rate": 2.9355036035466836e-05,
      "loss": 0.07177257537841797,
      "step": 1158
    },
    {
      "epoch": 0.15669308637385293,
      "grad_norm": 0.19277019798755646,
      "learning_rate": 2.935304910939002e-05,
      "loss": 0.0703129768371582,
      "step": 1159
    },
    {
      "epoch": 0.1568282831696889,
      "grad_norm": 0.09898892045021057,
      "learning_rate": 2.935105919493184e-05,
      "loss": 0.0472712516784668,
      "step": 1160
    },
    {
      "epoch": 0.15696347996552482,
      "grad_norm": 0.09241020679473877,
      "learning_rate": 2.9349066292506613e-05,
      "loss": 0.049954891204833984,
      "step": 1161
    },
    {
      "epoch": 0.15709867676136074,
      "grad_norm": 0.0797291249036789,
      "learning_rate": 2.934707040252926e-05,
      "loss": 0.04801750183105469,
      "step": 1162
    },
    {
      "epoch": 0.1572338735571967,
      "grad_norm": 0.184931218624115,
      "learning_rate": 2.9345071525415342e-05,
      "loss": 0.10203933715820312,
      "step": 1163
    },
    {
      "epoch": 0.15736907035303263,
      "grad_norm": 0.2322974056005478,
      "learning_rate": 2.9343069661581035e-05,
      "loss": 0.09958314895629883,
      "step": 1164
    },
    {
      "epoch": 0.15750426714886856,
      "grad_norm": 0.2699923813343048,
      "learning_rate": 2.9341064811443138e-05,
      "loss": 0.13122177124023438,
      "step": 1165
    },
    {
      "epoch": 0.15763946394470452,
      "grad_norm": 0.06598154455423355,
      "learning_rate": 2.9339056975419078e-05,
      "loss": 0.07757329940795898,
      "step": 1166
    },
    {
      "epoch": 0.15777466074054045,
      "grad_norm": 0.12984755635261536,
      "learning_rate": 2.9337046153926882e-05,
      "loss": 0.06386709213256836,
      "step": 1167
    },
    {
      "epoch": 0.15790985753637637,
      "grad_norm": 0.4202303886413574,
      "learning_rate": 2.9335032347385224e-05,
      "loss": 0.12419319152832031,
      "step": 1168
    },
    {
      "epoch": 0.15804505433221233,
      "grad_norm": 0.24104250967502594,
      "learning_rate": 2.933301555621339e-05,
      "loss": 0.08004891872406006,
      "step": 1169
    },
    {
      "epoch": 0.15818025112804826,
      "grad_norm": 0.08389753848314285,
      "learning_rate": 2.933099578083128e-05,
      "loss": 0.07021570205688477,
      "step": 1170
    },
    {
      "epoch": 0.15831544792388422,
      "grad_norm": 0.27718332409858704,
      "learning_rate": 2.932897302165943e-05,
      "loss": 0.06672859191894531,
      "step": 1171
    },
    {
      "epoch": 0.15845064471972015,
      "grad_norm": 0.17038077116012573,
      "learning_rate": 2.9326947279118983e-05,
      "loss": 0.06165933609008789,
      "step": 1172
    },
    {
      "epoch": 0.15858584151555608,
      "grad_norm": 0.1964806616306305,
      "learning_rate": 2.9324918553631716e-05,
      "loss": 0.05494403839111328,
      "step": 1173
    },
    {
      "epoch": 0.15872103831139203,
      "grad_norm": 0.1367480605840683,
      "learning_rate": 2.9322886845620013e-05,
      "loss": 0.04713630676269531,
      "step": 1174
    },
    {
      "epoch": 0.15885623510722796,
      "grad_norm": 0.3066956698894501,
      "learning_rate": 2.932085215550689e-05,
      "loss": 0.08920001983642578,
      "step": 1175
    },
    {
      "epoch": 0.1589914319030639,
      "grad_norm": 0.21093346178531647,
      "learning_rate": 2.9318814483715982e-05,
      "loss": 0.08451461791992188,
      "step": 1176
    },
    {
      "epoch": 0.15912662869889985,
      "grad_norm": 0.11341755092144012,
      "learning_rate": 2.9316773830671537e-05,
      "loss": 0.07023680210113525,
      "step": 1177
    },
    {
      "epoch": 0.15926182549473578,
      "grad_norm": 0.29472970962524414,
      "learning_rate": 2.9314730196798437e-05,
      "loss": 0.07318401336669922,
      "step": 1178
    },
    {
      "epoch": 0.1593970222905717,
      "grad_norm": 0.1765746772289276,
      "learning_rate": 2.9312683582522178e-05,
      "loss": 0.07141256332397461,
      "step": 1179
    },
    {
      "epoch": 0.15953221908640766,
      "grad_norm": 0.1467924416065216,
      "learning_rate": 2.9310633988268868e-05,
      "loss": 0.0755157470703125,
      "step": 1180
    },
    {
      "epoch": 0.1596674158822436,
      "grad_norm": 0.08082759380340576,
      "learning_rate": 2.9308581414465246e-05,
      "loss": 0.05016326904296875,
      "step": 1181
    },
    {
      "epoch": 0.15980261267807952,
      "grad_norm": 0.17234677076339722,
      "learning_rate": 2.9306525861538674e-05,
      "loss": 0.07890582084655762,
      "step": 1182
    },
    {
      "epoch": 0.15993780947391548,
      "grad_norm": 0.21386606991291046,
      "learning_rate": 2.9304467329917127e-05,
      "loss": 0.09691619873046875,
      "step": 1183
    },
    {
      "epoch": 0.1600730062697514,
      "grad_norm": 0.3143528997898102,
      "learning_rate": 2.9302405820029198e-05,
      "loss": 0.09788703918457031,
      "step": 1184
    },
    {
      "epoch": 0.16020820306558733,
      "grad_norm": 0.07750120759010315,
      "learning_rate": 2.9300341332304114e-05,
      "loss": 0.04978680610656738,
      "step": 1185
    },
    {
      "epoch": 0.1603433998614233,
      "grad_norm": 0.10350902378559113,
      "learning_rate": 2.9298273867171697e-05,
      "loss": 0.08543968200683594,
      "step": 1186
    },
    {
      "epoch": 0.16047859665725922,
      "grad_norm": 0.06081010401248932,
      "learning_rate": 2.929620342506242e-05,
      "loss": 0.05190324783325195,
      "step": 1187
    },
    {
      "epoch": 0.16061379345309515,
      "grad_norm": 0.14381834864616394,
      "learning_rate": 2.929413000640735e-05,
      "loss": 0.05739545822143555,
      "step": 1188
    },
    {
      "epoch": 0.1607489902489311,
      "grad_norm": 0.11380499601364136,
      "learning_rate": 2.9292053611638187e-05,
      "loss": 0.06984281539916992,
      "step": 1189
    },
    {
      "epoch": 0.16088418704476704,
      "grad_norm": 0.16667023301124573,
      "learning_rate": 2.928997424118725e-05,
      "loss": 0.07844972610473633,
      "step": 1190
    },
    {
      "epoch": 0.16101938384060296,
      "grad_norm": 0.15189296007156372,
      "learning_rate": 2.928789189548747e-05,
      "loss": 0.05687522888183594,
      "step": 1191
    },
    {
      "epoch": 0.16115458063643892,
      "grad_norm": 0.24288663268089294,
      "learning_rate": 2.9285806574972405e-05,
      "loss": 0.0876169204711914,
      "step": 1192
    },
    {
      "epoch": 0.16128977743227485,
      "grad_norm": 0.13709843158721924,
      "learning_rate": 2.928371828007623e-05,
      "loss": 0.0818643569946289,
      "step": 1193
    },
    {
      "epoch": 0.1614249742281108,
      "grad_norm": 0.2224666327238083,
      "learning_rate": 2.928162701123374e-05,
      "loss": 0.04541301727294922,
      "step": 1194
    },
    {
      "epoch": 0.16156017102394674,
      "grad_norm": 0.37950173020362854,
      "learning_rate": 2.9279532768880345e-05,
      "loss": 0.09860467910766602,
      "step": 1195
    },
    {
      "epoch": 0.16169536781978266,
      "grad_norm": 0.25968003273010254,
      "learning_rate": 2.9277435553452084e-05,
      "loss": 0.0796588659286499,
      "step": 1196
    },
    {
      "epoch": 0.16183056461561862,
      "grad_norm": 0.15009881556034088,
      "learning_rate": 2.9275335365385602e-05,
      "loss": 0.09694814682006836,
      "step": 1197
    },
    {
      "epoch": 0.16196576141145455,
      "grad_norm": 0.07834871113300323,
      "learning_rate": 2.927323220511817e-05,
      "loss": 0.04359889030456543,
      "step": 1198
    },
    {
      "epoch": 0.16210095820729048,
      "grad_norm": 0.07751213759183884,
      "learning_rate": 2.9271126073087684e-05,
      "loss": 0.06636476516723633,
      "step": 1199
    },
    {
      "epoch": 0.16223615500312644,
      "grad_norm": 0.2132425308227539,
      "learning_rate": 2.926901696973264e-05,
      "loss": 0.06555485725402832,
      "step": 1200
    },
    {
      "epoch": 0.16237135179896237,
      "grad_norm": 0.24026857316493988,
      "learning_rate": 2.9266904895492177e-05,
      "loss": 0.08239245414733887,
      "step": 1201
    },
    {
      "epoch": 0.1625065485947983,
      "grad_norm": 0.10026822984218597,
      "learning_rate": 2.926478985080603e-05,
      "loss": 0.08301210403442383,
      "step": 1202
    },
    {
      "epoch": 0.16264174539063425,
      "grad_norm": 0.0927710011601448,
      "learning_rate": 2.9262671836114568e-05,
      "loss": 0.05316925048828125,
      "step": 1203
    },
    {
      "epoch": 0.16277694218647018,
      "grad_norm": 0.053466539829969406,
      "learning_rate": 2.9260550851858774e-05,
      "loss": 0.07049942016601562,
      "step": 1204
    },
    {
      "epoch": 0.1629121389823061,
      "grad_norm": 0.23009340465068817,
      "learning_rate": 2.9258426898480243e-05,
      "loss": 0.0935511589050293,
      "step": 1205
    },
    {
      "epoch": 0.16304733577814207,
      "grad_norm": 0.15983623266220093,
      "learning_rate": 2.9256299976421198e-05,
      "loss": 0.08137989044189453,
      "step": 1206
    },
    {
      "epoch": 0.163182532573978,
      "grad_norm": 0.16412276029586792,
      "learning_rate": 2.9254170086124474e-05,
      "loss": 0.06665325164794922,
      "step": 1207
    },
    {
      "epoch": 0.16331772936981392,
      "grad_norm": 0.23292100429534912,
      "learning_rate": 2.9252037228033526e-05,
      "loss": 0.06276392936706543,
      "step": 1208
    },
    {
      "epoch": 0.16345292616564988,
      "grad_norm": 0.10810549557209015,
      "learning_rate": 2.9249901402592424e-05,
      "loss": 0.07190942764282227,
      "step": 1209
    },
    {
      "epoch": 0.1635881229614858,
      "grad_norm": 0.1365727186203003,
      "learning_rate": 2.9247762610245863e-05,
      "loss": 0.05400681495666504,
      "step": 1210
    },
    {
      "epoch": 0.16372331975732174,
      "grad_norm": 0.05971809849143028,
      "learning_rate": 2.9245620851439146e-05,
      "loss": 0.037708282470703125,
      "step": 1211
    },
    {
      "epoch": 0.1638585165531577,
      "grad_norm": 0.12509702146053314,
      "learning_rate": 2.92434761266182e-05,
      "loss": 0.08895158767700195,
      "step": 1212
    },
    {
      "epoch": 0.16399371334899362,
      "grad_norm": 0.08663970232009888,
      "learning_rate": 2.924132843622957e-05,
      "loss": 0.07162857055664062,
      "step": 1213
    },
    {
      "epoch": 0.16412891014482955,
      "grad_norm": 0.08429182320833206,
      "learning_rate": 2.9239177780720418e-05,
      "loss": 0.06822454929351807,
      "step": 1214
    },
    {
      "epoch": 0.1642641069406655,
      "grad_norm": 0.12882648408412933,
      "learning_rate": 2.923702416053852e-05,
      "loss": 0.08619308471679688,
      "step": 1215
    },
    {
      "epoch": 0.16439930373650144,
      "grad_norm": 0.13457289338111877,
      "learning_rate": 2.9234867576132268e-05,
      "loss": 0.08995437622070312,
      "step": 1216
    },
    {
      "epoch": 0.1645345005323374,
      "grad_norm": 0.10224489867687225,
      "learning_rate": 2.923270802795068e-05,
      "loss": 0.07589328289031982,
      "step": 1217
    },
    {
      "epoch": 0.16466969732817333,
      "grad_norm": 0.19573554396629333,
      "learning_rate": 2.9230545516443378e-05,
      "loss": 0.11031937599182129,
      "step": 1218
    },
    {
      "epoch": 0.16480489412400925,
      "grad_norm": 0.08408115804195404,
      "learning_rate": 2.9228380042060615e-05,
      "loss": 0.04436063766479492,
      "step": 1219
    },
    {
      "epoch": 0.1649400909198452,
      "grad_norm": 0.059222813695669174,
      "learning_rate": 2.9226211605253252e-05,
      "loss": 0.06900215148925781,
      "step": 1220
    },
    {
      "epoch": 0.16507528771568114,
      "grad_norm": 0.11509162187576294,
      "learning_rate": 2.922404020647277e-05,
      "loss": 0.07071709632873535,
      "step": 1221
    },
    {
      "epoch": 0.16521048451151707,
      "grad_norm": 0.15091605484485626,
      "learning_rate": 2.9221865846171264e-05,
      "loss": 0.07544474303722382,
      "step": 1222
    },
    {
      "epoch": 0.16534568130735303,
      "grad_norm": 0.09125815331935883,
      "learning_rate": 2.9219688524801446e-05,
      "loss": 0.049719929695129395,
      "step": 1223
    },
    {
      "epoch": 0.16548087810318896,
      "grad_norm": 0.14215806126594543,
      "learning_rate": 2.9217508242816653e-05,
      "loss": 0.07222175598144531,
      "step": 1224
    },
    {
      "epoch": 0.16561607489902488,
      "grad_norm": 0.0726814940571785,
      "learning_rate": 2.921532500067083e-05,
      "loss": 0.046596527099609375,
      "step": 1225
    },
    {
      "epoch": 0.16575127169486084,
      "grad_norm": 0.0951012521982193,
      "learning_rate": 2.9213138798818528e-05,
      "loss": 0.06709766387939453,
      "step": 1226
    },
    {
      "epoch": 0.16588646849069677,
      "grad_norm": 0.14965760707855225,
      "learning_rate": 2.921094963771494e-05,
      "loss": 0.09727096557617188,
      "step": 1227
    },
    {
      "epoch": 0.1660216652865327,
      "grad_norm": 0.08207011967897415,
      "learning_rate": 2.9208757517815855e-05,
      "loss": 0.08413529396057129,
      "step": 1228
    },
    {
      "epoch": 0.16615686208236866,
      "grad_norm": 0.0920029953122139,
      "learning_rate": 2.9206562439577684e-05,
      "loss": 0.06987190246582031,
      "step": 1229
    },
    {
      "epoch": 0.16629205887820458,
      "grad_norm": 0.04928401857614517,
      "learning_rate": 2.9204364403457452e-05,
      "loss": 0.054015159606933594,
      "step": 1230
    },
    {
      "epoch": 0.16642725567404051,
      "grad_norm": 0.0512600913643837,
      "learning_rate": 2.9202163409912808e-05,
      "loss": 0.04779958724975586,
      "step": 1231
    },
    {
      "epoch": 0.16656245246987647,
      "grad_norm": 0.07469185441732407,
      "learning_rate": 2.9199959459402003e-05,
      "loss": 0.0555112361907959,
      "step": 1232
    },
    {
      "epoch": 0.1666976492657124,
      "grad_norm": 0.07954122126102448,
      "learning_rate": 2.919775255238392e-05,
      "loss": 0.05206298828125,
      "step": 1233
    },
    {
      "epoch": 0.16683284606154833,
      "grad_norm": 0.042790018022060394,
      "learning_rate": 2.919554268931804e-05,
      "loss": 0.0326237678527832,
      "step": 1234
    },
    {
      "epoch": 0.16696804285738429,
      "grad_norm": 0.09722354263067245,
      "learning_rate": 2.9193329870664475e-05,
      "loss": 0.07882452011108398,
      "step": 1235
    },
    {
      "epoch": 0.16710323965322021,
      "grad_norm": 0.13021519780158997,
      "learning_rate": 2.9191114096883938e-05,
      "loss": 0.08313465118408203,
      "step": 1236
    },
    {
      "epoch": 0.16723843644905614,
      "grad_norm": 0.07478612661361694,
      "learning_rate": 2.9188895368437774e-05,
      "loss": 0.05598902702331543,
      "step": 1237
    },
    {
      "epoch": 0.1673736332448921,
      "grad_norm": 0.18403148651123047,
      "learning_rate": 2.9186673685787926e-05,
      "loss": 0.0759739875793457,
      "step": 1238
    },
    {
      "epoch": 0.16750883004072803,
      "grad_norm": 0.08487872779369354,
      "learning_rate": 2.918444904939697e-05,
      "loss": 0.06282567977905273,
      "step": 1239
    },
    {
      "epoch": 0.167644026836564,
      "grad_norm": 0.08897240459918976,
      "learning_rate": 2.9182221459728078e-05,
      "loss": 0.06490278244018555,
      "step": 1240
    },
    {
      "epoch": 0.16777922363239992,
      "grad_norm": 0.1542828381061554,
      "learning_rate": 2.917999091724505e-05,
      "loss": 0.0795431137084961,
      "step": 1241
    },
    {
      "epoch": 0.16791442042823584,
      "grad_norm": 0.11625979095697403,
      "learning_rate": 2.9177757422412294e-05,
      "loss": 0.08052492141723633,
      "step": 1242
    },
    {
      "epoch": 0.1680496172240718,
      "grad_norm": 0.09489346295595169,
      "learning_rate": 2.917552097569484e-05,
      "loss": 0.05592775344848633,
      "step": 1243
    },
    {
      "epoch": 0.16818481401990773,
      "grad_norm": 0.158501997590065,
      "learning_rate": 2.917328157755832e-05,
      "loss": 0.06336051225662231,
      "step": 1244
    },
    {
      "epoch": 0.16832001081574366,
      "grad_norm": 0.07722340524196625,
      "learning_rate": 2.9171039228469003e-05,
      "loss": 0.06006908416748047,
      "step": 1245
    },
    {
      "epoch": 0.16845520761157962,
      "grad_norm": 0.07774420827627182,
      "learning_rate": 2.9168793928893747e-05,
      "loss": 0.06841564178466797,
      "step": 1246
    },
    {
      "epoch": 0.16859040440741555,
      "grad_norm": 0.15228503942489624,
      "learning_rate": 2.9166545679300036e-05,
      "loss": 0.0656576156616211,
      "step": 1247
    },
    {
      "epoch": 0.16872560120325147,
      "grad_norm": 0.16187550127506256,
      "learning_rate": 2.9164294480155966e-05,
      "loss": 0.054795265197753906,
      "step": 1248
    },
    {
      "epoch": 0.16886079799908743,
      "grad_norm": 0.1031121239066124,
      "learning_rate": 2.9162040331930256e-05,
      "loss": 0.0848090648651123,
      "step": 1249
    },
    {
      "epoch": 0.16899599479492336,
      "grad_norm": 0.21679864823818207,
      "learning_rate": 2.915978323509223e-05,
      "loss": 0.06322550773620605,
      "step": 1250
    },
    {
      "epoch": 0.1691311915907593,
      "grad_norm": 0.24488388001918793,
      "learning_rate": 2.915752319011182e-05,
      "loss": 0.10214900970458984,
      "step": 1251
    },
    {
      "epoch": 0.16926638838659525,
      "grad_norm": 0.16754993796348572,
      "learning_rate": 2.9155260197459588e-05,
      "loss": 0.10164070129394531,
      "step": 1252
    },
    {
      "epoch": 0.16940158518243117,
      "grad_norm": 0.15058457851409912,
      "learning_rate": 2.91529942576067e-05,
      "loss": 0.09105300903320312,
      "step": 1253
    },
    {
      "epoch": 0.1695367819782671,
      "grad_norm": 0.06617345660924911,
      "learning_rate": 2.915072537102493e-05,
      "loss": 0.05120724439620972,
      "step": 1254
    },
    {
      "epoch": 0.16967197877410306,
      "grad_norm": 0.062166474759578705,
      "learning_rate": 2.914845353818668e-05,
      "loss": 0.04924631118774414,
      "step": 1255
    },
    {
      "epoch": 0.169807175569939,
      "grad_norm": 0.34717512130737305,
      "learning_rate": 2.9146178759564953e-05,
      "loss": 0.0993661880493164,
      "step": 1256
    },
    {
      "epoch": 0.16994237236577492,
      "grad_norm": 0.04060515761375427,
      "learning_rate": 2.914390103563337e-05,
      "loss": 0.0339890718460083,
      "step": 1257
    },
    {
      "epoch": 0.17007756916161088,
      "grad_norm": 0.19279055297374725,
      "learning_rate": 2.914162036686617e-05,
      "loss": 0.07509946823120117,
      "step": 1258
    },
    {
      "epoch": 0.1702127659574468,
      "grad_norm": 0.10772058367729187,
      "learning_rate": 2.9139336753738196e-05,
      "loss": 0.10341405868530273,
      "step": 1259
    },
    {
      "epoch": 0.17034796275328273,
      "grad_norm": 0.11588366329669952,
      "learning_rate": 2.913705019672491e-05,
      "loss": 0.06786322593688965,
      "step": 1260
    },
    {
      "epoch": 0.1704831595491187,
      "grad_norm": 0.34414902329444885,
      "learning_rate": 2.9134760696302386e-05,
      "loss": 0.09893321990966797,
      "step": 1261
    },
    {
      "epoch": 0.17061835634495462,
      "grad_norm": 0.180582657456398,
      "learning_rate": 2.9132468252947306e-05,
      "loss": 0.08387088775634766,
      "step": 1262
    },
    {
      "epoch": 0.17075355314079058,
      "grad_norm": 0.4320274889469147,
      "learning_rate": 2.9130172867136974e-05,
      "loss": 0.09543085098266602,
      "step": 1263
    },
    {
      "epoch": 0.1708887499366265,
      "grad_norm": 0.17885689437389374,
      "learning_rate": 2.91278745393493e-05,
      "loss": 0.07803922891616821,
      "step": 1264
    },
    {
      "epoch": 0.17102394673246243,
      "grad_norm": 0.13092473149299622,
      "learning_rate": 2.9125573270062812e-05,
      "loss": 0.06229686737060547,
      "step": 1265
    },
    {
      "epoch": 0.1711591435282984,
      "grad_norm": 0.18804676830768585,
      "learning_rate": 2.9123269059756634e-05,
      "loss": 0.060498714447021484,
      "step": 1266
    },
    {
      "epoch": 0.17129434032413432,
      "grad_norm": 0.08068740367889404,
      "learning_rate": 2.9120961908910528e-05,
      "loss": 0.07163596153259277,
      "step": 1267
    },
    {
      "epoch": 0.17142953711997025,
      "grad_norm": 0.22152014076709747,
      "learning_rate": 2.911865181800485e-05,
      "loss": 0.10860919952392578,
      "step": 1268
    },
    {
      "epoch": 0.1715647339158062,
      "grad_norm": 0.11795254796743393,
      "learning_rate": 2.9116338787520577e-05,
      "loss": 0.10941553115844727,
      "step": 1269
    },
    {
      "epoch": 0.17169993071164213,
      "grad_norm": 0.10287696123123169,
      "learning_rate": 2.9114022817939283e-05,
      "loss": 0.08036231994628906,
      "step": 1270
    },
    {
      "epoch": 0.17183512750747806,
      "grad_norm": 0.07889505475759506,
      "learning_rate": 2.911170390974318e-05,
      "loss": 0.09151744842529297,
      "step": 1271
    },
    {
      "epoch": 0.17197032430331402,
      "grad_norm": 0.13890427350997925,
      "learning_rate": 2.9109382063415067e-05,
      "loss": 0.04506111145019531,
      "step": 1272
    },
    {
      "epoch": 0.17210552109914995,
      "grad_norm": 0.08674082905054092,
      "learning_rate": 2.9107057279438372e-05,
      "loss": 0.04581022262573242,
      "step": 1273
    },
    {
      "epoch": 0.17224071789498588,
      "grad_norm": 0.12049552798271179,
      "learning_rate": 2.910472955829712e-05,
      "loss": 0.11568403244018555,
      "step": 1274
    },
    {
      "epoch": 0.17237591469082184,
      "grad_norm": 0.08607936650514603,
      "learning_rate": 2.9102398900475958e-05,
      "loss": 0.08767271041870117,
      "step": 1275
    },
    {
      "epoch": 0.17251111148665776,
      "grad_norm": 0.13958725333213806,
      "learning_rate": 2.910006530646014e-05,
      "loss": 0.05032539367675781,
      "step": 1276
    },
    {
      "epoch": 0.1726463082824937,
      "grad_norm": 0.09221465140581131,
      "learning_rate": 2.909772877673554e-05,
      "loss": 0.04073596000671387,
      "step": 1277
    },
    {
      "epoch": 0.17278150507832965,
      "grad_norm": 0.07107503712177277,
      "learning_rate": 2.9095389311788626e-05,
      "loss": 0.05596303939819336,
      "step": 1278
    },
    {
      "epoch": 0.17291670187416558,
      "grad_norm": 0.07603446394205093,
      "learning_rate": 2.9093046912106494e-05,
      "loss": 0.07087564468383789,
      "step": 1279
    },
    {
      "epoch": 0.1730518986700015,
      "grad_norm": 0.141872376203537,
      "learning_rate": 2.909070157817684e-05,
      "loss": 0.06925201416015625,
      "step": 1280
    },
    {
      "epoch": 0.17318709546583747,
      "grad_norm": 0.16083335876464844,
      "learning_rate": 2.9088353310487976e-05,
      "loss": 0.09833264350891113,
      "step": 1281
    },
    {
      "epoch": 0.1733222922616734,
      "grad_norm": 0.35496050119400024,
      "learning_rate": 2.9086002109528825e-05,
      "loss": 0.09937858581542969,
      "step": 1282
    },
    {
      "epoch": 0.17345748905750935,
      "grad_norm": 0.16016283631324768,
      "learning_rate": 2.908364797578892e-05,
      "loss": 0.0793687105178833,
      "step": 1283
    },
    {
      "epoch": 0.17359268585334528,
      "grad_norm": 0.05883108451962471,
      "learning_rate": 2.9081290909758405e-05,
      "loss": 0.060964226722717285,
      "step": 1284
    },
    {
      "epoch": 0.1737278826491812,
      "grad_norm": 0.1897485852241516,
      "learning_rate": 2.9078930911928033e-05,
      "loss": 0.07701396942138672,
      "step": 1285
    },
    {
      "epoch": 0.17386307944501717,
      "grad_norm": 0.11912494897842407,
      "learning_rate": 2.907656798278916e-05,
      "loss": 0.06920289993286133,
      "step": 1286
    },
    {
      "epoch": 0.1739982762408531,
      "grad_norm": 0.07339036464691162,
      "learning_rate": 2.9074202122833773e-05,
      "loss": 0.0572047233581543,
      "step": 1287
    },
    {
      "epoch": 0.17413347303668902,
      "grad_norm": 0.06731192022562027,
      "learning_rate": 2.907183333255445e-05,
      "loss": 0.057007551193237305,
      "step": 1288
    },
    {
      "epoch": 0.17426866983252498,
      "grad_norm": 0.1837085336446762,
      "learning_rate": 2.9069461612444384e-05,
      "loss": 0.10115432739257812,
      "step": 1289
    },
    {
      "epoch": 0.1744038666283609,
      "grad_norm": 0.06417826563119888,
      "learning_rate": 2.9067086962997385e-05,
      "loss": 0.03990578651428223,
      "step": 1290
    },
    {
      "epoch": 0.17453906342419684,
      "grad_norm": 0.11471592634916306,
      "learning_rate": 2.9064709384707868e-05,
      "loss": 0.0711812973022461,
      "step": 1291
    },
    {
      "epoch": 0.1746742602200328,
      "grad_norm": 0.24925322830677032,
      "learning_rate": 2.9062328878070855e-05,
      "loss": 0.0804452896118164,
      "step": 1292
    },
    {
      "epoch": 0.17480945701586872,
      "grad_norm": 0.17708346247673035,
      "learning_rate": 2.905994544358198e-05,
      "loss": 0.07260704040527344,
      "step": 1293
    },
    {
      "epoch": 0.17494465381170465,
      "grad_norm": 0.07037805765867233,
      "learning_rate": 2.9057559081737482e-05,
      "loss": 0.061341166496276855,
      "step": 1294
    },
    {
      "epoch": 0.1750798506075406,
      "grad_norm": 0.1026465892791748,
      "learning_rate": 2.9055169793034225e-05,
      "loss": 0.06517291069030762,
      "step": 1295
    },
    {
      "epoch": 0.17521504740337654,
      "grad_norm": 0.13157184422016144,
      "learning_rate": 2.9052777577969656e-05,
      "loss": 0.04810833930969238,
      "step": 1296
    },
    {
      "epoch": 0.17535024419921247,
      "grad_norm": 0.09227099269628525,
      "learning_rate": 2.9050382437041868e-05,
      "loss": 0.05074000358581543,
      "step": 1297
    },
    {
      "epoch": 0.17548544099504843,
      "grad_norm": 0.17749840021133423,
      "learning_rate": 2.9047984370749526e-05,
      "loss": 0.0745927095413208,
      "step": 1298
    },
    {
      "epoch": 0.17562063779088435,
      "grad_norm": 0.10106880217790604,
      "learning_rate": 2.9045583379591925e-05,
      "loss": 0.055890798568725586,
      "step": 1299
    },
    {
      "epoch": 0.17575583458672028,
      "grad_norm": 0.18339994549751282,
      "learning_rate": 2.9043179464068965e-05,
      "loss": 0.08118295669555664,
      "step": 1300
    },
    {
      "epoch": 0.17589103138255624,
      "grad_norm": 0.10667243599891663,
      "learning_rate": 2.9040772624681152e-05,
      "loss": 0.06749486923217773,
      "step": 1301
    },
    {
      "epoch": 0.17602622817839217,
      "grad_norm": 0.10810138285160065,
      "learning_rate": 2.9038362861929603e-05,
      "loss": 0.0656580924987793,
      "step": 1302
    },
    {
      "epoch": 0.1761614249742281,
      "grad_norm": 0.08462870866060257,
      "learning_rate": 2.903595017631605e-05,
      "loss": 0.05225372314453125,
      "step": 1303
    },
    {
      "epoch": 0.17629662177006405,
      "grad_norm": 0.069978728890419,
      "learning_rate": 2.903353456834282e-05,
      "loss": 0.06925821304321289,
      "step": 1304
    },
    {
      "epoch": 0.17643181856589998,
      "grad_norm": 0.11681298911571503,
      "learning_rate": 2.903111603851285e-05,
      "loss": 0.05601310729980469,
      "step": 1305
    },
    {
      "epoch": 0.17656701536173594,
      "grad_norm": 0.06930729001760483,
      "learning_rate": 2.9028694587329704e-05,
      "loss": 0.06746578216552734,
      "step": 1306
    },
    {
      "epoch": 0.17670221215757187,
      "grad_norm": 0.21610063314437866,
      "learning_rate": 2.902627021529753e-05,
      "loss": 0.0843818187713623,
      "step": 1307
    },
    {
      "epoch": 0.1768374089534078,
      "grad_norm": 0.06469037383794785,
      "learning_rate": 2.9023842922921105e-05,
      "loss": 0.05560588836669922,
      "step": 1308
    },
    {
      "epoch": 0.17697260574924376,
      "grad_norm": 0.1363135576248169,
      "learning_rate": 2.90214127107058e-05,
      "loss": 0.10736942291259766,
      "step": 1309
    },
    {
      "epoch": 0.17710780254507968,
      "grad_norm": 0.1837257593870163,
      "learning_rate": 2.9018979579157592e-05,
      "loss": 0.06760454177856445,
      "step": 1310
    },
    {
      "epoch": 0.1772429993409156,
      "grad_norm": 0.11503897607326508,
      "learning_rate": 2.901654352878308e-05,
      "loss": 0.059558868408203125,
      "step": 1311
    },
    {
      "epoch": 0.17737819613675157,
      "grad_norm": 0.13690738379955292,
      "learning_rate": 2.9014104560089462e-05,
      "loss": 0.06150054931640625,
      "step": 1312
    },
    {
      "epoch": 0.1775133929325875,
      "grad_norm": 0.07230895012617111,
      "learning_rate": 2.9011662673584538e-05,
      "loss": 0.05124711990356445,
      "step": 1313
    },
    {
      "epoch": 0.17764858972842343,
      "grad_norm": 0.12711846828460693,
      "learning_rate": 2.900921786977673e-05,
      "loss": 0.08394670486450195,
      "step": 1314
    },
    {
      "epoch": 0.17778378652425939,
      "grad_norm": 0.22098760306835175,
      "learning_rate": 2.900677014917505e-05,
      "loss": 0.06563472747802734,
      "step": 1315
    },
    {
      "epoch": 0.17791898332009531,
      "grad_norm": 0.1604573130607605,
      "learning_rate": 2.9004319512289136e-05,
      "loss": 0.07813453674316406,
      "step": 1316
    },
    {
      "epoch": 0.17805418011593124,
      "grad_norm": 0.06745977699756622,
      "learning_rate": 2.9001865959629222e-05,
      "loss": 0.046000003814697266,
      "step": 1317
    },
    {
      "epoch": 0.1781893769117672,
      "grad_norm": 0.12523406744003296,
      "learning_rate": 2.8999409491706143e-05,
      "loss": 0.07711029052734375,
      "step": 1318
    },
    {
      "epoch": 0.17832457370760313,
      "grad_norm": 0.2040608525276184,
      "learning_rate": 2.8996950109031355e-05,
      "loss": 0.07084465026855469,
      "step": 1319
    },
    {
      "epoch": 0.17845977050343906,
      "grad_norm": 0.09971045702695847,
      "learning_rate": 2.8994487812116917e-05,
      "loss": 0.06867599487304688,
      "step": 1320
    },
    {
      "epoch": 0.17859496729927501,
      "grad_norm": 0.09231485426425934,
      "learning_rate": 2.8992022601475483e-05,
      "loss": 0.03790569305419922,
      "step": 1321
    },
    {
      "epoch": 0.17873016409511094,
      "grad_norm": 0.0515068955719471,
      "learning_rate": 2.8989554477620332e-05,
      "loss": 0.058396339416503906,
      "step": 1322
    },
    {
      "epoch": 0.17886536089094687,
      "grad_norm": 0.1407320201396942,
      "learning_rate": 2.8987083441065335e-05,
      "loss": 0.09112024307250977,
      "step": 1323
    },
    {
      "epoch": 0.17900055768678283,
      "grad_norm": 0.08635195344686508,
      "learning_rate": 2.8984609492324983e-05,
      "loss": 0.06437969207763672,
      "step": 1324
    },
    {
      "epoch": 0.17913575448261876,
      "grad_norm": 0.08393722027540207,
      "learning_rate": 2.8982132631914357e-05,
      "loss": 0.07030361890792847,
      "step": 1325
    },
    {
      "epoch": 0.1792709512784547,
      "grad_norm": 0.22234483063220978,
      "learning_rate": 2.8979652860349154e-05,
      "loss": 0.044274330139160156,
      "step": 1326
    },
    {
      "epoch": 0.17940614807429064,
      "grad_norm": 0.07304778695106506,
      "learning_rate": 2.8977170178145675e-05,
      "loss": 0.05028223991394043,
      "step": 1327
    },
    {
      "epoch": 0.17954134487012657,
      "grad_norm": 0.08771605044603348,
      "learning_rate": 2.8974684585820833e-05,
      "loss": 0.060231685638427734,
      "step": 1328
    },
    {
      "epoch": 0.17967654166596253,
      "grad_norm": 0.05200042948126793,
      "learning_rate": 2.8972196083892138e-05,
      "loss": 0.057005882263183594,
      "step": 1329
    },
    {
      "epoch": 0.17981173846179846,
      "grad_norm": 0.12458130717277527,
      "learning_rate": 2.8969704672877707e-05,
      "loss": 0.051749229431152344,
      "step": 1330
    },
    {
      "epoch": 0.1799469352576344,
      "grad_norm": 0.2294543981552124,
      "learning_rate": 2.896721035329627e-05,
      "loss": 0.07013463973999023,
      "step": 1331
    },
    {
      "epoch": 0.18008213205347035,
      "grad_norm": 0.09830637276172638,
      "learning_rate": 2.8964713125667153e-05,
      "loss": 0.06429815292358398,
      "step": 1332
    },
    {
      "epoch": 0.18021732884930627,
      "grad_norm": 0.113348089158535,
      "learning_rate": 2.8962212990510294e-05,
      "loss": 0.06415843963623047,
      "step": 1333
    },
    {
      "epoch": 0.1803525256451422,
      "grad_norm": 0.2105209082365036,
      "learning_rate": 2.8959709948346237e-05,
      "loss": 0.08428645133972168,
      "step": 1334
    },
    {
      "epoch": 0.18048772244097816,
      "grad_norm": 0.09588108956813812,
      "learning_rate": 2.8957203999696124e-05,
      "loss": 0.04796886444091797,
      "step": 1335
    },
    {
      "epoch": 0.1806229192368141,
      "grad_norm": 0.09961932897567749,
      "learning_rate": 2.8954695145081713e-05,
      "loss": 0.08334970474243164,
      "step": 1336
    },
    {
      "epoch": 0.18075811603265002,
      "grad_norm": 0.06924933195114136,
      "learning_rate": 2.8952183385025356e-05,
      "loss": 0.07975625991821289,
      "step": 1337
    },
    {
      "epoch": 0.18089331282848597,
      "grad_norm": 0.22378431260585785,
      "learning_rate": 2.8949668720050014e-05,
      "loss": 0.10448598861694336,
      "step": 1338
    },
    {
      "epoch": 0.1810285096243219,
      "grad_norm": 0.1825861632823944,
      "learning_rate": 2.8947151150679256e-05,
      "loss": 0.06799077987670898,
      "step": 1339
    },
    {
      "epoch": 0.18116370642015783,
      "grad_norm": 0.08274604380130768,
      "learning_rate": 2.8944630677437255e-05,
      "loss": 0.0810018926858902,
      "step": 1340
    },
    {
      "epoch": 0.1812989032159938,
      "grad_norm": 0.10907977819442749,
      "learning_rate": 2.8942107300848784e-05,
      "loss": 0.06739604473114014,
      "step": 1341
    },
    {
      "epoch": 0.18143410001182972,
      "grad_norm": 0.10027384012937546,
      "learning_rate": 2.8939581021439225e-05,
      "loss": 0.09553384780883789,
      "step": 1342
    },
    {
      "epoch": 0.18156929680766565,
      "grad_norm": 0.1585473120212555,
      "learning_rate": 2.8937051839734563e-05,
      "loss": 0.05302238464355469,
      "step": 1343
    },
    {
      "epoch": 0.1817044936035016,
      "grad_norm": 0.19858047366142273,
      "learning_rate": 2.8934519756261384e-05,
      "loss": 0.056659698486328125,
      "step": 1344
    },
    {
      "epoch": 0.18183969039933753,
      "grad_norm": 0.19681096076965332,
      "learning_rate": 2.8931984771546885e-05,
      "loss": 0.05824720859527588,
      "step": 1345
    },
    {
      "epoch": 0.18197488719517346,
      "grad_norm": 0.1715867668390274,
      "learning_rate": 2.8929446886118866e-05,
      "loss": 0.08165168762207031,
      "step": 1346
    },
    {
      "epoch": 0.18211008399100942,
      "grad_norm": 0.14102637767791748,
      "learning_rate": 2.892690610050572e-05,
      "loss": 0.06902551651000977,
      "step": 1347
    },
    {
      "epoch": 0.18224528078684535,
      "grad_norm": 0.15581168234348297,
      "learning_rate": 2.892436241523646e-05,
      "loss": 0.05586886405944824,
      "step": 1348
    },
    {
      "epoch": 0.18238047758268128,
      "grad_norm": 0.09487874060869217,
      "learning_rate": 2.8921815830840685e-05,
      "loss": 0.09083080291748047,
      "step": 1349
    },
    {
      "epoch": 0.18251567437851723,
      "grad_norm": 0.1292032152414322,
      "learning_rate": 2.891926634784862e-05,
      "loss": 0.07888293266296387,
      "step": 1350
    },
    {
      "epoch": 0.18265087117435316,
      "grad_norm": 0.3102639615535736,
      "learning_rate": 2.8916713966791076e-05,
      "loss": 0.07295083999633789,
      "step": 1351
    },
    {
      "epoch": 0.18278606797018912,
      "grad_norm": 0.3073422610759735,
      "learning_rate": 2.8914158688199464e-05,
      "loss": 0.09141802787780762,
      "step": 1352
    },
    {
      "epoch": 0.18292126476602505,
      "grad_norm": 0.0924450159072876,
      "learning_rate": 2.891160051260582e-05,
      "loss": 0.07750105857849121,
      "step": 1353
    },
    {
      "epoch": 0.18305646156186098,
      "grad_norm": 0.07807210087776184,
      "learning_rate": 2.8909039440542758e-05,
      "loss": 0.0388340950012207,
      "step": 1354
    },
    {
      "epoch": 0.18319165835769693,
      "grad_norm": 0.2586558163166046,
      "learning_rate": 2.890647547254352e-05,
      "loss": 0.0616002082824707,
      "step": 1355
    },
    {
      "epoch": 0.18332685515353286,
      "grad_norm": 0.0976000651717186,
      "learning_rate": 2.8903908609141923e-05,
      "loss": 0.04634904861450195,
      "step": 1356
    },
    {
      "epoch": 0.1834620519493688,
      "grad_norm": 0.3379673957824707,
      "learning_rate": 2.8901338850872413e-05,
      "loss": 0.08740568161010742,
      "step": 1357
    },
    {
      "epoch": 0.18359724874520475,
      "grad_norm": 0.11325690150260925,
      "learning_rate": 2.8898766198270022e-05,
      "loss": 0.059988975524902344,
      "step": 1358
    },
    {
      "epoch": 0.18373244554104068,
      "grad_norm": 0.09347891062498093,
      "learning_rate": 2.8896190651870392e-05,
      "loss": 0.0576169490814209,
      "step": 1359
    },
    {
      "epoch": 0.1838676423368766,
      "grad_norm": 0.22027359902858734,
      "learning_rate": 2.8893612212209763e-05,
      "loss": 0.07803654670715332,
      "step": 1360
    },
    {
      "epoch": 0.18400283913271256,
      "grad_norm": 0.27932098507881165,
      "learning_rate": 2.8891030879824985e-05,
      "loss": 0.08052492141723633,
      "step": 1361
    },
    {
      "epoch": 0.1841380359285485,
      "grad_norm": 0.1708294004201889,
      "learning_rate": 2.88884466552535e-05,
      "loss": 0.062349557876586914,
      "step": 1362
    },
    {
      "epoch": 0.18427323272438442,
      "grad_norm": 0.06322871893644333,
      "learning_rate": 2.888585953903336e-05,
      "loss": 0.0391697883605957,
      "step": 1363
    },
    {
      "epoch": 0.18440842952022038,
      "grad_norm": 0.09236955642700195,
      "learning_rate": 2.888326953170321e-05,
      "loss": 0.09978437423706055,
      "step": 1364
    },
    {
      "epoch": 0.1845436263160563,
      "grad_norm": 0.08766787499189377,
      "learning_rate": 2.8880676633802314e-05,
      "loss": 0.05164957046508789,
      "step": 1365
    },
    {
      "epoch": 0.18467882311189224,
      "grad_norm": 0.11937934905290604,
      "learning_rate": 2.8878080845870522e-05,
      "loss": 0.06482076644897461,
      "step": 1366
    },
    {
      "epoch": 0.1848140199077282,
      "grad_norm": 0.03756975382566452,
      "learning_rate": 2.887548216844829e-05,
      "loss": 0.04184436798095703,
      "step": 1367
    },
    {
      "epoch": 0.18494921670356412,
      "grad_norm": 0.3251025974750519,
      "learning_rate": 2.8872880602076675e-05,
      "loss": 0.10559749603271484,
      "step": 1368
    },
    {
      "epoch": 0.18508441349940005,
      "grad_norm": 0.1538354903459549,
      "learning_rate": 2.8870276147297344e-05,
      "loss": 0.09036731719970703,
      "step": 1369
    },
    {
      "epoch": 0.185219610295236,
      "grad_norm": 0.10339432209730148,
      "learning_rate": 2.8867668804652552e-05,
      "loss": 0.0590212345123291,
      "step": 1370
    },
    {
      "epoch": 0.18535480709107194,
      "grad_norm": 0.08854861557483673,
      "learning_rate": 2.886505857468516e-05,
      "loss": 0.05467057228088379,
      "step": 1371
    },
    {
      "epoch": 0.18549000388690787,
      "grad_norm": 0.10698610544204712,
      "learning_rate": 2.8862445457938642e-05,
      "loss": 0.07664823532104492,
      "step": 1372
    },
    {
      "epoch": 0.18562520068274382,
      "grad_norm": 0.24214357137680054,
      "learning_rate": 2.8859829454957053e-05,
      "loss": 0.0495753288269043,
      "step": 1373
    },
    {
      "epoch": 0.18576039747857975,
      "grad_norm": 0.14994406700134277,
      "learning_rate": 2.8857210566285062e-05,
      "loss": 0.04423844814300537,
      "step": 1374
    },
    {
      "epoch": 0.1858955942744157,
      "grad_norm": 0.05552583560347557,
      "learning_rate": 2.8854588792467932e-05,
      "loss": 0.051290273666381836,
      "step": 1375
    },
    {
      "epoch": 0.18603079107025164,
      "grad_norm": 0.17275118827819824,
      "learning_rate": 2.8851964134051535e-05,
      "loss": 0.06133460998535156,
      "step": 1376
    },
    {
      "epoch": 0.18616598786608757,
      "grad_norm": 0.10736674070358276,
      "learning_rate": 2.884933659158234e-05,
      "loss": 0.0592951774597168,
      "step": 1377
    },
    {
      "epoch": 0.18630118466192352,
      "grad_norm": 0.14024773240089417,
      "learning_rate": 2.8846706165607415e-05,
      "loss": 0.09714174270629883,
      "step": 1378
    },
    {
      "epoch": 0.18643638145775945,
      "grad_norm": 0.08408856391906738,
      "learning_rate": 2.8844072856674422e-05,
      "loss": 0.07675886154174805,
      "step": 1379
    },
    {
      "epoch": 0.18657157825359538,
      "grad_norm": 0.0785374715924263,
      "learning_rate": 2.8841436665331634e-05,
      "loss": 0.057961463928222656,
      "step": 1380
    },
    {
      "epoch": 0.18670677504943134,
      "grad_norm": 0.15104219317436218,
      "learning_rate": 2.8838797592127927e-05,
      "loss": 0.07774686813354492,
      "step": 1381
    },
    {
      "epoch": 0.18684197184526727,
      "grad_norm": 0.17305879294872284,
      "learning_rate": 2.883615563761276e-05,
      "loss": 0.07398366928100586,
      "step": 1382
    },
    {
      "epoch": 0.1869771686411032,
      "grad_norm": 0.12430399656295776,
      "learning_rate": 2.8833510802336206e-05,
      "loss": 0.06166648864746094,
      "step": 1383
    },
    {
      "epoch": 0.18711236543693915,
      "grad_norm": 0.0911678597331047,
      "learning_rate": 2.883086308684893e-05,
      "loss": 0.06403088569641113,
      "step": 1384
    },
    {
      "epoch": 0.18724756223277508,
      "grad_norm": 0.11092045903205872,
      "learning_rate": 2.882821249170221e-05,
      "loss": 0.058730363845825195,
      "step": 1385
    },
    {
      "epoch": 0.187382759028611,
      "grad_norm": 0.07783066481351852,
      "learning_rate": 2.8825559017447905e-05,
      "loss": 0.05014991760253906,
      "step": 1386
    },
    {
      "epoch": 0.18751795582444697,
      "grad_norm": 0.11592923104763031,
      "learning_rate": 2.8822902664638487e-05,
      "loss": 0.07435894012451172,
      "step": 1387
    },
    {
      "epoch": 0.1876531526202829,
      "grad_norm": 0.08970188349485397,
      "learning_rate": 2.882024343382702e-05,
      "loss": 0.07713985443115234,
      "step": 1388
    },
    {
      "epoch": 0.18778834941611883,
      "grad_norm": 0.22262392938137054,
      "learning_rate": 2.8817581325567174e-05,
      "loss": 0.124847412109375,
      "step": 1389
    },
    {
      "epoch": 0.18792354621195478,
      "grad_norm": 0.056012123823165894,
      "learning_rate": 2.8814916340413205e-05,
      "loss": 0.0842580795288086,
      "step": 1390
    },
    {
      "epoch": 0.1880587430077907,
      "grad_norm": 0.09239956736564636,
      "learning_rate": 2.881224847891999e-05,
      "loss": 0.06169247627258301,
      "step": 1391
    },
    {
      "epoch": 0.18819393980362664,
      "grad_norm": 0.09727323800325394,
      "learning_rate": 2.8809577741642987e-05,
      "loss": 0.08878540992736816,
      "step": 1392
    },
    {
      "epoch": 0.1883291365994626,
      "grad_norm": 0.138835608959198,
      "learning_rate": 2.8806904129138255e-05,
      "loss": 0.0812673568725586,
      "step": 1393
    },
    {
      "epoch": 0.18846433339529853,
      "grad_norm": 0.15790875256061554,
      "learning_rate": 2.8804227641962457e-05,
      "loss": 0.0831451416015625,
      "step": 1394
    },
    {
      "epoch": 0.18859953019113446,
      "grad_norm": 0.15054289996623993,
      "learning_rate": 2.8801548280672847e-05,
      "loss": 0.11885452270507812,
      "step": 1395
    },
    {
      "epoch": 0.1887347269869704,
      "grad_norm": 0.11854302138090134,
      "learning_rate": 2.8798866045827288e-05,
      "loss": 0.037550926208496094,
      "step": 1396
    },
    {
      "epoch": 0.18886992378280634,
      "grad_norm": 0.07879561185836792,
      "learning_rate": 2.8796180937984234e-05,
      "loss": 0.06037092208862305,
      "step": 1397
    },
    {
      "epoch": 0.1890051205786423,
      "grad_norm": 0.06782495975494385,
      "learning_rate": 2.8793492957702738e-05,
      "loss": 0.06937313079833984,
      "step": 1398
    },
    {
      "epoch": 0.18914031737447823,
      "grad_norm": 0.22681939601898193,
      "learning_rate": 2.8790802105542454e-05,
      "loss": 0.08480256795883179,
      "step": 1399
    },
    {
      "epoch": 0.18927551417031416,
      "grad_norm": 0.10194907337427139,
      "learning_rate": 2.8788108382063628e-05,
      "loss": 0.05944967269897461,
      "step": 1400
    },
    {
      "epoch": 0.18941071096615011,
      "grad_norm": 0.2543809115886688,
      "learning_rate": 2.878541178782711e-05,
      "loss": 0.12061214447021484,
      "step": 1401
    },
    {
      "epoch": 0.18954590776198604,
      "grad_norm": 0.15377728641033173,
      "learning_rate": 2.8782712323394344e-05,
      "loss": 0.11813163757324219,
      "step": 1402
    },
    {
      "epoch": 0.18968110455782197,
      "grad_norm": 0.15461359918117523,
      "learning_rate": 2.878000998932738e-05,
      "loss": 0.05657768249511719,
      "step": 1403
    },
    {
      "epoch": 0.18981630135365793,
      "grad_norm": 0.16069062054157257,
      "learning_rate": 2.8777304786188847e-05,
      "loss": 0.10333633422851562,
      "step": 1404
    },
    {
      "epoch": 0.18995149814949386,
      "grad_norm": 0.06819060444831848,
      "learning_rate": 2.8774596714541988e-05,
      "loss": 0.0556105375289917,
      "step": 1405
    },
    {
      "epoch": 0.1900866949453298,
      "grad_norm": 0.0878991186618805,
      "learning_rate": 2.8771885774950637e-05,
      "loss": 0.04519295692443848,
      "step": 1406
    },
    {
      "epoch": 0.19022189174116574,
      "grad_norm": 0.17676198482513428,
      "learning_rate": 2.876917196797923e-05,
      "loss": 0.07409811019897461,
      "step": 1407
    },
    {
      "epoch": 0.19035708853700167,
      "grad_norm": 0.1823957860469818,
      "learning_rate": 2.876645529419279e-05,
      "loss": 0.08322715759277344,
      "step": 1408
    },
    {
      "epoch": 0.1904922853328376,
      "grad_norm": 0.09615840762853622,
      "learning_rate": 2.876373575415695e-05,
      "loss": 0.0773468017578125,
      "step": 1409
    },
    {
      "epoch": 0.19062748212867356,
      "grad_norm": 0.13884080946445465,
      "learning_rate": 2.8761013348437926e-05,
      "loss": 0.05861163139343262,
      "step": 1410
    },
    {
      "epoch": 0.1907626789245095,
      "grad_norm": 0.12990429997444153,
      "learning_rate": 2.875828807760254e-05,
      "loss": 0.07942724227905273,
      "step": 1411
    },
    {
      "epoch": 0.19089787572034542,
      "grad_norm": 0.21610304713249207,
      "learning_rate": 2.875555994221821e-05,
      "loss": 0.06644105911254883,
      "step": 1412
    },
    {
      "epoch": 0.19103307251618137,
      "grad_norm": 0.11897508800029755,
      "learning_rate": 2.8752828942852943e-05,
      "loss": 0.06753742694854736,
      "step": 1413
    },
    {
      "epoch": 0.1911682693120173,
      "grad_norm": 0.17719510197639465,
      "learning_rate": 2.875009508007535e-05,
      "loss": 0.08179759979248047,
      "step": 1414
    },
    {
      "epoch": 0.19130346610785323,
      "grad_norm": 0.062041688710451126,
      "learning_rate": 2.8747358354454642e-05,
      "loss": 0.06464147567749023,
      "step": 1415
    },
    {
      "epoch": 0.1914386629036892,
      "grad_norm": 0.09665444493293762,
      "learning_rate": 2.8744618766560614e-05,
      "loss": 0.06567001342773438,
      "step": 1416
    },
    {
      "epoch": 0.19157385969952512,
      "grad_norm": 0.13068965077400208,
      "learning_rate": 2.8741876316963664e-05,
      "loss": 0.09388256072998047,
      "step": 1417
    },
    {
      "epoch": 0.19170905649536105,
      "grad_norm": 0.10729022324085236,
      "learning_rate": 2.873913100623478e-05,
      "loss": 0.06386137008666992,
      "step": 1418
    },
    {
      "epoch": 0.191844253291197,
      "grad_norm": 0.3651251792907715,
      "learning_rate": 2.873638283494556e-05,
      "loss": 0.0839681625366211,
      "step": 1419
    },
    {
      "epoch": 0.19197945008703293,
      "grad_norm": 0.18678618967533112,
      "learning_rate": 2.8733631803668178e-05,
      "loss": 0.05196845531463623,
      "step": 1420
    },
    {
      "epoch": 0.1921146468828689,
      "grad_norm": 0.16412784159183502,
      "learning_rate": 2.8730877912975418e-05,
      "loss": 0.050917625427246094,
      "step": 1421
    },
    {
      "epoch": 0.19224984367870482,
      "grad_norm": 0.09918685257434845,
      "learning_rate": 2.8728121163440656e-05,
      "loss": 0.06438601016998291,
      "step": 1422
    },
    {
      "epoch": 0.19238504047454075,
      "grad_norm": 0.17637163400650024,
      "learning_rate": 2.8725361555637863e-05,
      "loss": 0.07499885559082031,
      "step": 1423
    },
    {
      "epoch": 0.1925202372703767,
      "grad_norm": 0.30195796489715576,
      "learning_rate": 2.8722599090141598e-05,
      "loss": 0.08899223804473877,
      "step": 1424
    },
    {
      "epoch": 0.19265543406621263,
      "grad_norm": 0.23839521408081055,
      "learning_rate": 2.8719833767527026e-05,
      "loss": 0.09548377990722656,
      "step": 1425
    },
    {
      "epoch": 0.19279063086204856,
      "grad_norm": 0.07339231669902802,
      "learning_rate": 2.8717065588369896e-05,
      "loss": 0.05073881149291992,
      "step": 1426
    },
    {
      "epoch": 0.19292582765788452,
      "grad_norm": 0.11924371868371964,
      "learning_rate": 2.871429455324657e-05,
      "loss": 0.06584513187408447,
      "step": 1427
    },
    {
      "epoch": 0.19306102445372045,
      "grad_norm": 0.14208443462848663,
      "learning_rate": 2.871152066273398e-05,
      "loss": 0.08660650253295898,
      "step": 1428
    },
    {
      "epoch": 0.19319622124955638,
      "grad_norm": 0.3181878924369812,
      "learning_rate": 2.870874391740967e-05,
      "loss": 0.08827018737792969,
      "step": 1429
    },
    {
      "epoch": 0.19333141804539233,
      "grad_norm": 0.07801058143377304,
      "learning_rate": 2.8705964317851774e-05,
      "loss": 0.07984018325805664,
      "step": 1430
    },
    {
      "epoch": 0.19346661484122826,
      "grad_norm": 0.09154821932315826,
      "learning_rate": 2.8703181864639013e-05,
      "loss": 0.0673375129699707,
      "step": 1431
    },
    {
      "epoch": 0.1936018116370642,
      "grad_norm": 0.11852426826953888,
      "learning_rate": 2.870039655835072e-05,
      "loss": 0.06897473335266113,
      "step": 1432
    },
    {
      "epoch": 0.19373700843290015,
      "grad_norm": 0.07511452585458755,
      "learning_rate": 2.8697608399566796e-05,
      "loss": 0.06895732879638672,
      "step": 1433
    },
    {
      "epoch": 0.19387220522873608,
      "grad_norm": 0.11842640489339828,
      "learning_rate": 2.869481738886777e-05,
      "loss": 0.058078765869140625,
      "step": 1434
    },
    {
      "epoch": 0.194007402024572,
      "grad_norm": 0.09408573806285858,
      "learning_rate": 2.8692023526834725e-05,
      "loss": 0.07669878005981445,
      "step": 1435
    },
    {
      "epoch": 0.19414259882040796,
      "grad_norm": 0.198813796043396,
      "learning_rate": 2.8689226814049367e-05,
      "loss": 0.07886314392089844,
      "step": 1436
    },
    {
      "epoch": 0.1942777956162439,
      "grad_norm": 0.0880727544426918,
      "learning_rate": 2.868642725109399e-05,
      "loss": 0.06202530860900879,
      "step": 1437
    },
    {
      "epoch": 0.19441299241207982,
      "grad_norm": 0.06944162398576736,
      "learning_rate": 2.868362483855147e-05,
      "loss": 0.05888009071350098,
      "step": 1438
    },
    {
      "epoch": 0.19454818920791578,
      "grad_norm": 0.1189267709851265,
      "learning_rate": 2.8680819577005295e-05,
      "loss": 0.0660104751586914,
      "step": 1439
    },
    {
      "epoch": 0.1946833860037517,
      "grad_norm": 0.07902384549379349,
      "learning_rate": 2.8678011467039526e-05,
      "loss": 0.05221271514892578,
      "step": 1440
    },
    {
      "epoch": 0.19481858279958764,
      "grad_norm": 0.09434419870376587,
      "learning_rate": 2.867520050923883e-05,
      "loss": 0.05070090293884277,
      "step": 1441
    },
    {
      "epoch": 0.1949537795954236,
      "grad_norm": 0.061128754168748856,
      "learning_rate": 2.8672386704188466e-05,
      "loss": 0.0540614128112793,
      "step": 1442
    },
    {
      "epoch": 0.19508897639125952,
      "grad_norm": 0.07600434869527817,
      "learning_rate": 2.8669570052474273e-05,
      "loss": 0.06619977951049805,
      "step": 1443
    },
    {
      "epoch": 0.19522417318709548,
      "grad_norm": 0.11401189118623734,
      "learning_rate": 2.86667505546827e-05,
      "loss": 0.06352663040161133,
      "step": 1444
    },
    {
      "epoch": 0.1953593699829314,
      "grad_norm": 0.11938532441854477,
      "learning_rate": 2.866392821140079e-05,
      "loss": 0.04743194580078125,
      "step": 1445
    },
    {
      "epoch": 0.19549456677876734,
      "grad_norm": 0.18245841562747955,
      "learning_rate": 2.8661103023216154e-05,
      "loss": 0.07476139068603516,
      "step": 1446
    },
    {
      "epoch": 0.1956297635746033,
      "grad_norm": 0.16927899420261383,
      "learning_rate": 2.8658274990717018e-05,
      "loss": 0.08357441425323486,
      "step": 1447
    },
    {
      "epoch": 0.19576496037043922,
      "grad_norm": 0.08636972308158875,
      "learning_rate": 2.86554441144922e-05,
      "loss": 0.06247091293334961,
      "step": 1448
    },
    {
      "epoch": 0.19590015716627515,
      "grad_norm": 0.08359581232070923,
      "learning_rate": 2.8652610395131097e-05,
      "loss": 0.05119967460632324,
      "step": 1449
    },
    {
      "epoch": 0.1960353539621111,
      "grad_norm": 0.21092578768730164,
      "learning_rate": 2.8649773833223702e-05,
      "loss": 0.06998217105865479,
      "step": 1450
    },
    {
      "epoch": 0.19617055075794704,
      "grad_norm": 0.11341535300016403,
      "learning_rate": 2.8646934429360606e-05,
      "loss": 0.08960509300231934,
      "step": 1451
    },
    {
      "epoch": 0.19630574755378297,
      "grad_norm": 0.09743883460760117,
      "learning_rate": 2.8644092184132986e-05,
      "loss": 0.05595874786376953,
      "step": 1452
    },
    {
      "epoch": 0.19644094434961892,
      "grad_norm": 0.17838837206363678,
      "learning_rate": 2.864124709813262e-05,
      "loss": 0.08580327033996582,
      "step": 1453
    },
    {
      "epoch": 0.19657614114545485,
      "grad_norm": 0.13772979378700256,
      "learning_rate": 2.8638399171951856e-05,
      "loss": 0.09108620882034302,
      "step": 1454
    },
    {
      "epoch": 0.19671133794129078,
      "grad_norm": 0.2179628610610962,
      "learning_rate": 2.8635548406183664e-05,
      "loss": 0.12022209167480469,
      "step": 1455
    },
    {
      "epoch": 0.19684653473712674,
      "grad_norm": 0.3471464216709137,
      "learning_rate": 2.8632694801421576e-05,
      "loss": 0.0930701494216919,
      "step": 1456
    },
    {
      "epoch": 0.19698173153296267,
      "grad_norm": 0.056388743221759796,
      "learning_rate": 2.862983835825973e-05,
      "loss": 0.05864667892456055,
      "step": 1457
    },
    {
      "epoch": 0.1971169283287986,
      "grad_norm": 0.16650868952274323,
      "learning_rate": 2.8626979077292856e-05,
      "loss": 0.11506080627441406,
      "step": 1458
    },
    {
      "epoch": 0.19725212512463455,
      "grad_norm": 0.05247064307332039,
      "learning_rate": 2.862411695911627e-05,
      "loss": 0.0590968132019043,
      "step": 1459
    },
    {
      "epoch": 0.19738732192047048,
      "grad_norm": 0.16531048715114594,
      "learning_rate": 2.862125200432588e-05,
      "loss": 0.06190013885498047,
      "step": 1460
    },
    {
      "epoch": 0.1975225187163064,
      "grad_norm": 0.06729508191347122,
      "learning_rate": 2.8618384213518188e-05,
      "loss": 0.03719806671142578,
      "step": 1461
    },
    {
      "epoch": 0.19765771551214237,
      "grad_norm": 0.15682785212993622,
      "learning_rate": 2.861551358729028e-05,
      "loss": 0.06908249855041504,
      "step": 1462
    },
    {
      "epoch": 0.1977929123079783,
      "grad_norm": 0.08799882233142853,
      "learning_rate": 2.8612640126239836e-05,
      "loss": 0.08551311492919922,
      "step": 1463
    },
    {
      "epoch": 0.19792810910381423,
      "grad_norm": 0.06565811485052109,
      "learning_rate": 2.8609763830965126e-05,
      "loss": 0.06384038925170898,
      "step": 1464
    },
    {
      "epoch": 0.19806330589965018,
      "grad_norm": 0.05718689411878586,
      "learning_rate": 2.860688470206501e-05,
      "loss": 0.050319671630859375,
      "step": 1465
    },
    {
      "epoch": 0.1981985026954861,
      "grad_norm": 0.1384674608707428,
      "learning_rate": 2.8604002740138936e-05,
      "loss": 0.09149646759033203,
      "step": 1466
    },
    {
      "epoch": 0.19833369949132207,
      "grad_norm": 0.09817821532487869,
      "learning_rate": 2.860111794578695e-05,
      "loss": 0.07784509658813477,
      "step": 1467
    },
    {
      "epoch": 0.198468896287158,
      "grad_norm": 0.09491139650344849,
      "learning_rate": 2.8598230319609677e-05,
      "loss": 0.06491518020629883,
      "step": 1468
    },
    {
      "epoch": 0.19860409308299393,
      "grad_norm": 0.09622706472873688,
      "learning_rate": 2.8595339862208336e-05,
      "loss": 0.07344484329223633,
      "step": 1469
    },
    {
      "epoch": 0.19873928987882988,
      "grad_norm": 0.22665034234523773,
      "learning_rate": 2.8592446574184733e-05,
      "loss": 0.08306217193603516,
      "step": 1470
    },
    {
      "epoch": 0.1988744866746658,
      "grad_norm": 0.08944015204906464,
      "learning_rate": 2.8589550456141274e-05,
      "loss": 0.07159876823425293,
      "step": 1471
    },
    {
      "epoch": 0.19900968347050174,
      "grad_norm": 0.07798139750957489,
      "learning_rate": 2.8586651508680942e-05,
      "loss": 0.05242156982421875,
      "step": 1472
    },
    {
      "epoch": 0.1991448802663377,
      "grad_norm": 0.13667482137680054,
      "learning_rate": 2.8583749732407312e-05,
      "loss": 0.07505989074707031,
      "step": 1473
    },
    {
      "epoch": 0.19928007706217363,
      "grad_norm": 0.22467473149299622,
      "learning_rate": 2.8580845127924546e-05,
      "loss": 0.09481024742126465,
      "step": 1474
    },
    {
      "epoch": 0.19941527385800956,
      "grad_norm": 0.09221498668193817,
      "learning_rate": 2.8577937695837408e-05,
      "loss": 0.04923224449157715,
      "step": 1475
    },
    {
      "epoch": 0.1995504706538455,
      "grad_norm": 0.1812954843044281,
      "learning_rate": 2.8575027436751235e-05,
      "loss": 0.08632588386535645,
      "step": 1476
    },
    {
      "epoch": 0.19968566744968144,
      "grad_norm": 0.08311624079942703,
      "learning_rate": 2.8572114351271955e-05,
      "loss": 0.06156265735626221,
      "step": 1477
    },
    {
      "epoch": 0.19982086424551737,
      "grad_norm": 0.14101848006248474,
      "learning_rate": 2.85691984400061e-05,
      "loss": 0.053627967834472656,
      "step": 1478
    },
    {
      "epoch": 0.19995606104135333,
      "grad_norm": 0.14146101474761963,
      "learning_rate": 2.8566279703560762e-05,
      "loss": 0.08534765243530273,
      "step": 1479
    },
    {
      "epoch": 0.20009125783718926,
      "grad_norm": 0.33410540223121643,
      "learning_rate": 2.8563358142543648e-05,
      "loss": 0.09288358688354492,
      "step": 1480
    },
    {
      "epoch": 0.20022645463302519,
      "grad_norm": 0.0630439817905426,
      "learning_rate": 2.856043375756304e-05,
      "loss": 0.04931831359863281,
      "step": 1481
    },
    {
      "epoch": 0.20036165142886114,
      "grad_norm": 0.10115737468004227,
      "learning_rate": 2.855750654922781e-05,
      "loss": 0.07011294364929199,
      "step": 1482
    },
    {
      "epoch": 0.20049684822469707,
      "grad_norm": 0.0963035374879837,
      "learning_rate": 2.855457651814742e-05,
      "loss": 0.06272411346435547,
      "step": 1483
    },
    {
      "epoch": 0.200632045020533,
      "grad_norm": 0.0937761664390564,
      "learning_rate": 2.8551643664931916e-05,
      "loss": 0.061197757720947266,
      "step": 1484
    },
    {
      "epoch": 0.20076724181636896,
      "grad_norm": 0.1477745622396469,
      "learning_rate": 2.8548707990191933e-05,
      "loss": 0.09209823608398438,
      "step": 1485
    },
    {
      "epoch": 0.2009024386122049,
      "grad_norm": 0.08705328404903412,
      "learning_rate": 2.8545769494538698e-05,
      "loss": 0.06943559646606445,
      "step": 1486
    },
    {
      "epoch": 0.20103763540804082,
      "grad_norm": 0.07988999783992767,
      "learning_rate": 2.854282817858402e-05,
      "loss": 0.053441524505615234,
      "step": 1487
    },
    {
      "epoch": 0.20117283220387677,
      "grad_norm": 0.06395494192838669,
      "learning_rate": 2.85398840429403e-05,
      "loss": 0.0539705753326416,
      "step": 1488
    },
    {
      "epoch": 0.2013080289997127,
      "grad_norm": 0.06291565299034119,
      "learning_rate": 2.853693708822051e-05,
      "loss": 0.05599689483642578,
      "step": 1489
    },
    {
      "epoch": 0.20144322579554866,
      "grad_norm": 0.15285834670066833,
      "learning_rate": 2.8533987315038234e-05,
      "loss": 0.0469818115234375,
      "step": 1490
    },
    {
      "epoch": 0.2015784225913846,
      "grad_norm": 0.21539804339408875,
      "learning_rate": 2.8531034724007627e-05,
      "loss": 0.07901370525360107,
      "step": 1491
    },
    {
      "epoch": 0.20171361938722052,
      "grad_norm": 0.11085191369056702,
      "learning_rate": 2.8528079315743435e-05,
      "loss": 0.08513593673706055,
      "step": 1492
    },
    {
      "epoch": 0.20184881618305647,
      "grad_norm": 0.1706719994544983,
      "learning_rate": 2.852512109086099e-05,
      "loss": 0.05638456344604492,
      "step": 1493
    },
    {
      "epoch": 0.2019840129788924,
      "grad_norm": 0.07389377802610397,
      "learning_rate": 2.8522160049976208e-05,
      "loss": 0.047602057456970215,
      "step": 1494
    },
    {
      "epoch": 0.20211920977472833,
      "grad_norm": 0.05763942003250122,
      "learning_rate": 2.8519196193705595e-05,
      "loss": 0.039682626724243164,
      "step": 1495
    },
    {
      "epoch": 0.2022544065705643,
      "grad_norm": 0.08368533104658127,
      "learning_rate": 2.8516229522666243e-05,
      "loss": 0.0820918083190918,
      "step": 1496
    },
    {
      "epoch": 0.20238960336640022,
      "grad_norm": 0.09992201626300812,
      "learning_rate": 2.8513260037475825e-05,
      "loss": 0.07529139518737793,
      "step": 1497
    },
    {
      "epoch": 0.20252480016223615,
      "grad_norm": 0.0575963519513607,
      "learning_rate": 2.8510287738752604e-05,
      "loss": 0.05639004707336426,
      "step": 1498
    },
    {
      "epoch": 0.2026599969580721,
      "grad_norm": 0.16548512876033783,
      "learning_rate": 2.8507312627115435e-05,
      "loss": 0.11516666412353516,
      "step": 1499
    },
    {
      "epoch": 0.20279519375390803,
      "grad_norm": 0.12798194587230682,
      "learning_rate": 2.850433470318374e-05,
      "loss": 0.05388355255126953,
      "step": 1500
    },
    {
      "epoch": 0.20293039054974396,
      "grad_norm": 0.09383414685726166,
      "learning_rate": 2.8501353967577556e-05,
      "loss": 0.06741315126419067,
      "step": 1501
    },
    {
      "epoch": 0.20306558734557992,
      "grad_norm": 0.09627026319503784,
      "learning_rate": 2.8498370420917468e-05,
      "loss": 0.07947826385498047,
      "step": 1502
    },
    {
      "epoch": 0.20320078414141585,
      "grad_norm": 0.09149527549743652,
      "learning_rate": 2.8495384063824683e-05,
      "loss": 0.047959089279174805,
      "step": 1503
    },
    {
      "epoch": 0.20333598093725178,
      "grad_norm": 0.17404185235500336,
      "learning_rate": 2.8492394896920964e-05,
      "loss": 0.06344318389892578,
      "step": 1504
    },
    {
      "epoch": 0.20347117773308773,
      "grad_norm": 0.12634052336215973,
      "learning_rate": 2.848940292082868e-05,
      "loss": 0.0624239444732666,
      "step": 1505
    },
    {
      "epoch": 0.20360637452892366,
      "grad_norm": 0.2699243426322937,
      "learning_rate": 2.8486408136170772e-05,
      "loss": 0.09841346740722656,
      "step": 1506
    },
    {
      "epoch": 0.2037415713247596,
      "grad_norm": 0.1997794210910797,
      "learning_rate": 2.8483410543570776e-05,
      "loss": 0.09163236618041992,
      "step": 1507
    },
    {
      "epoch": 0.20387676812059555,
      "grad_norm": 0.06674224138259888,
      "learning_rate": 2.8480410143652803e-05,
      "loss": 0.047437191009521484,
      "step": 1508
    },
    {
      "epoch": 0.20401196491643148,
      "grad_norm": 0.09612417221069336,
      "learning_rate": 2.8477406937041547e-05,
      "loss": 0.08894634246826172,
      "step": 1509
    },
    {
      "epoch": 0.20414716171226743,
      "grad_norm": 0.06677885353565216,
      "learning_rate": 2.8474400924362298e-05,
      "loss": 0.04779243469238281,
      "step": 1510
    },
    {
      "epoch": 0.20428235850810336,
      "grad_norm": 0.17527706921100616,
      "learning_rate": 2.847139210624092e-05,
      "loss": 0.0888051986694336,
      "step": 1511
    },
    {
      "epoch": 0.2044175553039393,
      "grad_norm": 0.1711282730102539,
      "learning_rate": 2.8468380483303873e-05,
      "loss": 0.08749103546142578,
      "step": 1512
    },
    {
      "epoch": 0.20455275209977525,
      "grad_norm": 0.1732027381658554,
      "learning_rate": 2.8465366056178183e-05,
      "loss": 0.0960836410522461,
      "step": 1513
    },
    {
      "epoch": 0.20468794889561118,
      "grad_norm": 0.21484826505184174,
      "learning_rate": 2.8462348825491475e-05,
      "loss": 0.0897359848022461,
      "step": 1514
    },
    {
      "epoch": 0.2048231456914471,
      "grad_norm": 0.08397260308265686,
      "learning_rate": 2.8459328791871953e-05,
      "loss": 0.07047891616821289,
      "step": 1515
    },
    {
      "epoch": 0.20495834248728306,
      "grad_norm": 0.09662534296512604,
      "learning_rate": 2.8456305955948402e-05,
      "loss": 0.06568765640258789,
      "step": 1516
    },
    {
      "epoch": 0.205093539283119,
      "grad_norm": 0.19887784123420715,
      "learning_rate": 2.845328031835019e-05,
      "loss": 0.11615180969238281,
      "step": 1517
    },
    {
      "epoch": 0.20522873607895492,
      "grad_norm": 0.1498584747314453,
      "learning_rate": 2.8450251879707277e-05,
      "loss": 0.05876421928405762,
      "step": 1518
    },
    {
      "epoch": 0.20536393287479088,
      "grad_norm": 0.1901555061340332,
      "learning_rate": 2.8447220640650194e-05,
      "loss": 0.07042741775512695,
      "step": 1519
    },
    {
      "epoch": 0.2054991296706268,
      "grad_norm": 0.06683945655822754,
      "learning_rate": 2.8444186601810068e-05,
      "loss": 0.0695185661315918,
      "step": 1520
    },
    {
      "epoch": 0.20563432646646274,
      "grad_norm": 0.1322280764579773,
      "learning_rate": 2.84411497638186e-05,
      "loss": 0.06537187099456787,
      "step": 1521
    },
    {
      "epoch": 0.2057695232622987,
      "grad_norm": 0.12220027297735214,
      "learning_rate": 2.843811012730807e-05,
      "loss": 0.07801580429077148,
      "step": 1522
    },
    {
      "epoch": 0.20590472005813462,
      "grad_norm": 0.20445573329925537,
      "learning_rate": 2.8435067692911353e-05,
      "loss": 0.10744094848632812,
      "step": 1523
    },
    {
      "epoch": 0.20603991685397055,
      "grad_norm": 0.22539056837558746,
      "learning_rate": 2.8432022461261897e-05,
      "loss": 0.0722203254699707,
      "step": 1524
    },
    {
      "epoch": 0.2061751136498065,
      "grad_norm": 0.23824705183506012,
      "learning_rate": 2.8428974432993736e-05,
      "loss": 0.10843801498413086,
      "step": 1525
    },
    {
      "epoch": 0.20631031044564244,
      "grad_norm": 0.1543579399585724,
      "learning_rate": 2.8425923608741486e-05,
      "loss": 0.05908036231994629,
      "step": 1526
    },
    {
      "epoch": 0.20644550724147837,
      "grad_norm": 0.07272074371576309,
      "learning_rate": 2.8422869989140343e-05,
      "loss": 0.06726193428039551,
      "step": 1527
    },
    {
      "epoch": 0.20658070403731432,
      "grad_norm": 0.15433360636234283,
      "learning_rate": 2.8419813574826093e-05,
      "loss": 0.08235502243041992,
      "step": 1528
    },
    {
      "epoch": 0.20671590083315025,
      "grad_norm": 0.10479071736335754,
      "learning_rate": 2.8416754366435092e-05,
      "loss": 0.07231855392456055,
      "step": 1529
    },
    {
      "epoch": 0.20685109762898618,
      "grad_norm": 0.17646753787994385,
      "learning_rate": 2.8413692364604285e-05,
      "loss": 0.09379863739013672,
      "step": 1530
    },
    {
      "epoch": 0.20698629442482214,
      "grad_norm": 0.06437936425209045,
      "learning_rate": 2.8410627569971197e-05,
      "loss": 0.05408430099487305,
      "step": 1531
    },
    {
      "epoch": 0.20712149122065807,
      "grad_norm": 0.11157441139221191,
      "learning_rate": 2.8407559983173934e-05,
      "loss": 0.08189725875854492,
      "step": 1532
    },
    {
      "epoch": 0.20725668801649402,
      "grad_norm": 0.07680966705083847,
      "learning_rate": 2.8404489604851186e-05,
      "loss": 0.061344146728515625,
      "step": 1533
    },
    {
      "epoch": 0.20739188481232995,
      "grad_norm": 0.132106751203537,
      "learning_rate": 2.840141643564222e-05,
      "loss": 0.07763051986694336,
      "step": 1534
    },
    {
      "epoch": 0.20752708160816588,
      "grad_norm": 0.06416943669319153,
      "learning_rate": 2.8398340476186885e-05,
      "loss": 0.07701873779296875,
      "step": 1535
    },
    {
      "epoch": 0.20766227840400184,
      "grad_norm": 0.09628237038850784,
      "learning_rate": 2.8395261727125617e-05,
      "loss": 0.06579113006591797,
      "step": 1536
    },
    {
      "epoch": 0.20779747519983777,
      "grad_norm": 0.10490274429321289,
      "learning_rate": 2.8392180189099425e-05,
      "loss": 0.08664131164550781,
      "step": 1537
    },
    {
      "epoch": 0.2079326719956737,
      "grad_norm": 0.11617899686098099,
      "learning_rate": 2.83890958627499e-05,
      "loss": 0.06887149810791016,
      "step": 1538
    },
    {
      "epoch": 0.20806786879150965,
      "grad_norm": 0.09263032674789429,
      "learning_rate": 2.8386008748719216e-05,
      "loss": 0.0717477798461914,
      "step": 1539
    },
    {
      "epoch": 0.20820306558734558,
      "grad_norm": 0.16455240547657013,
      "learning_rate": 2.838291884765013e-05,
      "loss": 0.05478811264038086,
      "step": 1540
    },
    {
      "epoch": 0.2083382623831815,
      "grad_norm": 0.08099474757909775,
      "learning_rate": 2.8379826160185975e-05,
      "loss": 0.054837703704833984,
      "step": 1541
    },
    {
      "epoch": 0.20847345917901747,
      "grad_norm": 0.2022770792245865,
      "learning_rate": 2.8376730686970664e-05,
      "loss": 0.07261061668395996,
      "step": 1542
    },
    {
      "epoch": 0.2086086559748534,
      "grad_norm": 0.12341496348381042,
      "learning_rate": 2.8373632428648683e-05,
      "loss": 0.06386590003967285,
      "step": 1543
    },
    {
      "epoch": 0.20874385277068933,
      "grad_norm": 0.27008938789367676,
      "learning_rate": 2.8370531385865124e-05,
      "loss": 0.08893513679504395,
      "step": 1544
    },
    {
      "epoch": 0.20887904956652528,
      "grad_norm": 0.04684915021061897,
      "learning_rate": 2.8367427559265622e-05,
      "loss": 0.04062134027481079,
      "step": 1545
    },
    {
      "epoch": 0.2090142463623612,
      "grad_norm": 0.08418384194374084,
      "learning_rate": 2.836432094949642e-05,
      "loss": 0.06748557090759277,
      "step": 1546
    },
    {
      "epoch": 0.20914944315819714,
      "grad_norm": 0.13064037263393402,
      "learning_rate": 2.836121155720433e-05,
      "loss": 0.07818770408630371,
      "step": 1547
    },
    {
      "epoch": 0.2092846399540331,
      "grad_norm": 0.13980351388454437,
      "learning_rate": 2.8358099383036745e-05,
      "loss": 0.08222794532775879,
      "step": 1548
    },
    {
      "epoch": 0.20941983674986903,
      "grad_norm": 0.07518435269594193,
      "learning_rate": 2.8354984427641634e-05,
      "loss": 0.06541681289672852,
      "step": 1549
    },
    {
      "epoch": 0.20955503354570496,
      "grad_norm": 0.26029059290885925,
      "learning_rate": 2.8351866691667544e-05,
      "loss": 0.06605732440948486,
      "step": 1550
    },
    {
      "epoch": 0.2096902303415409,
      "grad_norm": 0.08158840984106064,
      "learning_rate": 2.8348746175763613e-05,
      "loss": 0.051041364669799805,
      "step": 1551
    },
    {
      "epoch": 0.20982542713737684,
      "grad_norm": 0.09598598629236221,
      "learning_rate": 2.8345622880579537e-05,
      "loss": 0.10406064987182617,
      "step": 1552
    },
    {
      "epoch": 0.20996062393321277,
      "grad_norm": 0.13267220556735992,
      "learning_rate": 2.8342496806765615e-05,
      "loss": 0.070159912109375,
      "step": 1553
    },
    {
      "epoch": 0.21009582072904873,
      "grad_norm": 0.10742221772670746,
      "learning_rate": 2.833936795497271e-05,
      "loss": 0.08212882280349731,
      "step": 1554
    },
    {
      "epoch": 0.21023101752488466,
      "grad_norm": 0.13372552394866943,
      "learning_rate": 2.8336236325852257e-05,
      "loss": 0.07744884490966797,
      "step": 1555
    },
    {
      "epoch": 0.2103662143207206,
      "grad_norm": 0.19794897735118866,
      "learning_rate": 2.8333101920056285e-05,
      "loss": 0.10175323486328125,
      "step": 1556
    },
    {
      "epoch": 0.21050141111655654,
      "grad_norm": 0.12407051026821136,
      "learning_rate": 2.8329964738237392e-05,
      "loss": 0.0694122314453125,
      "step": 1557
    },
    {
      "epoch": 0.21063660791239247,
      "grad_norm": 0.17577140033245087,
      "learning_rate": 2.8326824781048756e-05,
      "loss": 0.06978273391723633,
      "step": 1558
    },
    {
      "epoch": 0.21077180470822843,
      "grad_norm": 0.22986391186714172,
      "learning_rate": 2.8323682049144135e-05,
      "loss": 0.07682514190673828,
      "step": 1559
    },
    {
      "epoch": 0.21090700150406436,
      "grad_norm": 0.07227930426597595,
      "learning_rate": 2.832053654317786e-05,
      "loss": 0.07820320129394531,
      "step": 1560
    },
    {
      "epoch": 0.21104219829990029,
      "grad_norm": 0.17465704679489136,
      "learning_rate": 2.8317388263804842e-05,
      "loss": 0.07083892822265625,
      "step": 1561
    },
    {
      "epoch": 0.21117739509573624,
      "grad_norm": 0.08385077863931656,
      "learning_rate": 2.8314237211680573e-05,
      "loss": 0.06954383850097656,
      "step": 1562
    },
    {
      "epoch": 0.21131259189157217,
      "grad_norm": 0.11462143063545227,
      "learning_rate": 2.8311083387461118e-05,
      "loss": 0.08948302268981934,
      "step": 1563
    },
    {
      "epoch": 0.2114477886874081,
      "grad_norm": 0.22417183220386505,
      "learning_rate": 2.8307926791803114e-05,
      "loss": 0.060292720794677734,
      "step": 1564
    },
    {
      "epoch": 0.21158298548324406,
      "grad_norm": 0.09488919377326965,
      "learning_rate": 2.8304767425363785e-05,
      "loss": 0.06700277328491211,
      "step": 1565
    },
    {
      "epoch": 0.21171818227907999,
      "grad_norm": 0.09614776819944382,
      "learning_rate": 2.830160528880093e-05,
      "loss": 0.0829324722290039,
      "step": 1566
    },
    {
      "epoch": 0.21185337907491592,
      "grad_norm": 0.07649706304073334,
      "learning_rate": 2.829844038277292e-05,
      "loss": 0.07271313667297363,
      "step": 1567
    },
    {
      "epoch": 0.21198857587075187,
      "grad_norm": 0.10743039846420288,
      "learning_rate": 2.8295272707938706e-05,
      "loss": 0.07613086700439453,
      "step": 1568
    },
    {
      "epoch": 0.2121237726665878,
      "grad_norm": 0.03896505758166313,
      "learning_rate": 2.8292102264957817e-05,
      "loss": 0.036650657653808594,
      "step": 1569
    },
    {
      "epoch": 0.21225896946242373,
      "grad_norm": 0.0936877578496933,
      "learning_rate": 2.8288929054490357e-05,
      "loss": 0.05343818664550781,
      "step": 1570
    },
    {
      "epoch": 0.2123941662582597,
      "grad_norm": 0.07594258338212967,
      "learning_rate": 2.8285753077196998e-05,
      "loss": 0.08158373832702637,
      "step": 1571
    },
    {
      "epoch": 0.21252936305409562,
      "grad_norm": 0.21740159392356873,
      "learning_rate": 2.8282574333739006e-05,
      "loss": 0.0598607063293457,
      "step": 1572
    },
    {
      "epoch": 0.21266455984993154,
      "grad_norm": 0.1019560918211937,
      "learning_rate": 2.8279392824778197e-05,
      "loss": 0.08047676086425781,
      "step": 1573
    },
    {
      "epoch": 0.2127997566457675,
      "grad_norm": 0.08105242997407913,
      "learning_rate": 2.8276208550976993e-05,
      "loss": 0.06468057632446289,
      "step": 1574
    },
    {
      "epoch": 0.21293495344160343,
      "grad_norm": 0.1682303249835968,
      "learning_rate": 2.8273021512998372e-05,
      "loss": 0.05381631851196289,
      "step": 1575
    },
    {
      "epoch": 0.21307015023743936,
      "grad_norm": 0.09697219729423523,
      "learning_rate": 2.826983171150589e-05,
      "loss": 0.07071447372436523,
      "step": 1576
    },
    {
      "epoch": 0.21320534703327532,
      "grad_norm": 0.060206107795238495,
      "learning_rate": 2.826663914716368e-05,
      "loss": 0.06472301483154297,
      "step": 1577
    },
    {
      "epoch": 0.21334054382911125,
      "grad_norm": 0.1292990893125534,
      "learning_rate": 2.826344382063646e-05,
      "loss": 0.0739588737487793,
      "step": 1578
    },
    {
      "epoch": 0.2134757406249472,
      "grad_norm": 0.1430521160364151,
      "learning_rate": 2.8260245732589503e-05,
      "loss": 0.08129310607910156,
      "step": 1579
    },
    {
      "epoch": 0.21361093742078313,
      "grad_norm": 0.06377974152565002,
      "learning_rate": 2.8257044883688672e-05,
      "loss": 0.07360696792602539,
      "step": 1580
    },
    {
      "epoch": 0.21374613421661906,
      "grad_norm": 0.13742199540138245,
      "learning_rate": 2.82538412746004e-05,
      "loss": 0.09415674209594727,
      "step": 1581
    },
    {
      "epoch": 0.21388133101245502,
      "grad_norm": 0.09372340887784958,
      "learning_rate": 2.8250634905991695e-05,
      "loss": 0.08354854583740234,
      "step": 1582
    },
    {
      "epoch": 0.21401652780829095,
      "grad_norm": 0.1757609099149704,
      "learning_rate": 2.824742577853015e-05,
      "loss": 0.07148551940917969,
      "step": 1583
    },
    {
      "epoch": 0.21415172460412688,
      "grad_norm": 0.13692936301231384,
      "learning_rate": 2.8244213892883907e-05,
      "loss": 0.06240952014923096,
      "step": 1584
    },
    {
      "epoch": 0.21428692139996283,
      "grad_norm": 0.15134233236312866,
      "learning_rate": 2.82409992497217e-05,
      "loss": 0.09677934646606445,
      "step": 1585
    },
    {
      "epoch": 0.21442211819579876,
      "grad_norm": 0.13773484528064728,
      "learning_rate": 2.8237781849712852e-05,
      "loss": 0.10232770442962646,
      "step": 1586
    },
    {
      "epoch": 0.2145573149916347,
      "grad_norm": 0.28322187066078186,
      "learning_rate": 2.8234561693527222e-05,
      "loss": 0.08209753036499023,
      "step": 1587
    },
    {
      "epoch": 0.21469251178747065,
      "grad_norm": 0.21161141991615295,
      "learning_rate": 2.8231338781835275e-05,
      "loss": 0.06932401657104492,
      "step": 1588
    },
    {
      "epoch": 0.21482770858330658,
      "grad_norm": 0.05349447578191757,
      "learning_rate": 2.8228113115308032e-05,
      "loss": 0.0672307014465332,
      "step": 1589
    },
    {
      "epoch": 0.2149629053791425,
      "grad_norm": 0.12604087591171265,
      "learning_rate": 2.82248846946171e-05,
      "loss": 0.07590246200561523,
      "step": 1590
    },
    {
      "epoch": 0.21509810217497846,
      "grad_norm": 0.09108059108257294,
      "learning_rate": 2.822165352043465e-05,
      "loss": 0.03805732727050781,
      "step": 1591
    },
    {
      "epoch": 0.2152332989708144,
      "grad_norm": 0.07555746287107468,
      "learning_rate": 2.8218419593433437e-05,
      "loss": 0.06487464904785156,
      "step": 1592
    },
    {
      "epoch": 0.21536849576665032,
      "grad_norm": 0.14201481640338898,
      "learning_rate": 2.8215182914286768e-05,
      "loss": 0.0627593994140625,
      "step": 1593
    },
    {
      "epoch": 0.21550369256248628,
      "grad_norm": 0.16735677421092987,
      "learning_rate": 2.8211943483668546e-05,
      "loss": 0.08746910095214844,
      "step": 1594
    },
    {
      "epoch": 0.2156388893583222,
      "grad_norm": 0.12074435502290726,
      "learning_rate": 2.8208701302253237e-05,
      "loss": 0.06325006484985352,
      "step": 1595
    },
    {
      "epoch": 0.21577408615415813,
      "grad_norm": 0.09936917573213577,
      "learning_rate": 2.820545637071588e-05,
      "loss": 0.09252023696899414,
      "step": 1596
    },
    {
      "epoch": 0.2159092829499941,
      "grad_norm": 0.13411439955234528,
      "learning_rate": 2.8202208689732083e-05,
      "loss": 0.07288265228271484,
      "step": 1597
    },
    {
      "epoch": 0.21604447974583002,
      "grad_norm": 0.16384102404117584,
      "learning_rate": 2.819895825997804e-05,
      "loss": 0.09059333801269531,
      "step": 1598
    },
    {
      "epoch": 0.21617967654166595,
      "grad_norm": 0.15895752608776093,
      "learning_rate": 2.81957050821305e-05,
      "loss": 0.07207655906677246,
      "step": 1599
    },
    {
      "epoch": 0.2163148733375019,
      "grad_norm": 0.11737911403179169,
      "learning_rate": 2.8192449156866787e-05,
      "loss": 0.07565879821777344,
      "step": 1600
    },
    {
      "epoch": 0.21645007013333784,
      "grad_norm": 0.16184231638908386,
      "learning_rate": 2.8189190484864814e-05,
      "loss": 0.08757781982421875,
      "step": 1601
    },
    {
      "epoch": 0.2165852669291738,
      "grad_norm": 0.09004601836204529,
      "learning_rate": 2.8185929066803052e-05,
      "loss": 0.05258011817932129,
      "step": 1602
    },
    {
      "epoch": 0.21672046372500972,
      "grad_norm": 0.1932934671640396,
      "learning_rate": 2.818266490336054e-05,
      "loss": 0.06732559204101562,
      "step": 1603
    },
    {
      "epoch": 0.21685566052084565,
      "grad_norm": 0.10448861867189407,
      "learning_rate": 2.817939799521689e-05,
      "loss": 0.056088924407958984,
      "step": 1604
    },
    {
      "epoch": 0.2169908573166816,
      "grad_norm": 0.049659907817840576,
      "learning_rate": 2.8176128343052304e-05,
      "loss": 0.05062723159790039,
      "step": 1605
    },
    {
      "epoch": 0.21712605411251754,
      "grad_norm": 0.08144520968198776,
      "learning_rate": 2.817285594754753e-05,
      "loss": 0.07903766632080078,
      "step": 1606
    },
    {
      "epoch": 0.21726125090835346,
      "grad_norm": 0.056168362498283386,
      "learning_rate": 2.8169580809383902e-05,
      "loss": 0.05120849609375,
      "step": 1607
    },
    {
      "epoch": 0.21739644770418942,
      "grad_norm": 0.09560943394899368,
      "learning_rate": 2.8166302929243326e-05,
      "loss": 0.07111024856567383,
      "step": 1608
    },
    {
      "epoch": 0.21753164450002535,
      "grad_norm": 0.07586546242237091,
      "learning_rate": 2.8163022307808264e-05,
      "loss": 0.07851409912109375,
      "step": 1609
    },
    {
      "epoch": 0.21766684129586128,
      "grad_norm": 0.05134633183479309,
      "learning_rate": 2.8159738945761764e-05,
      "loss": 0.05237221717834473,
      "step": 1610
    },
    {
      "epoch": 0.21780203809169724,
      "grad_norm": 0.10992392152547836,
      "learning_rate": 2.8156452843787438e-05,
      "loss": 0.10461616516113281,
      "step": 1611
    },
    {
      "epoch": 0.21793723488753317,
      "grad_norm": 0.12636183202266693,
      "learning_rate": 2.815316400256947e-05,
      "loss": 0.09352779388427734,
      "step": 1612
    },
    {
      "epoch": 0.2180724316833691,
      "grad_norm": 0.23528893291950226,
      "learning_rate": 2.814987242279262e-05,
      "loss": 0.09353137016296387,
      "step": 1613
    },
    {
      "epoch": 0.21820762847920505,
      "grad_norm": 0.13521729409694672,
      "learning_rate": 2.8146578105142202e-05,
      "loss": 0.08751773834228516,
      "step": 1614
    },
    {
      "epoch": 0.21834282527504098,
      "grad_norm": 0.12088946253061295,
      "learning_rate": 2.814328105030412e-05,
      "loss": 0.0687868595123291,
      "step": 1615
    },
    {
      "epoch": 0.2184780220708769,
      "grad_norm": 0.2687852680683136,
      "learning_rate": 2.8139981258964836e-05,
      "loss": 0.09131598472595215,
      "step": 1616
    },
    {
      "epoch": 0.21861321886671287,
      "grad_norm": 0.08263131976127625,
      "learning_rate": 2.8136678731811385e-05,
      "loss": 0.06813740730285645,
      "step": 1617
    },
    {
      "epoch": 0.2187484156625488,
      "grad_norm": 0.29229700565338135,
      "learning_rate": 2.8133373469531362e-05,
      "loss": 0.07965850830078125,
      "step": 1618
    },
    {
      "epoch": 0.21888361245838472,
      "grad_norm": 0.051193296909332275,
      "learning_rate": 2.8130065472812952e-05,
      "loss": 0.0567936897277832,
      "step": 1619
    },
    {
      "epoch": 0.21901880925422068,
      "grad_norm": 0.10434012115001678,
      "learning_rate": 2.812675474234489e-05,
      "loss": 0.07630383968353271,
      "step": 1620
    },
    {
      "epoch": 0.2191540060500566,
      "grad_norm": 0.08696451783180237,
      "learning_rate": 2.812344127881649e-05,
      "loss": 0.0768289566040039,
      "step": 1621
    },
    {
      "epoch": 0.21928920284589254,
      "grad_norm": 0.1323765516281128,
      "learning_rate": 2.8120125082917638e-05,
      "loss": 0.07840633392333984,
      "step": 1622
    },
    {
      "epoch": 0.2194243996417285,
      "grad_norm": 0.2015657126903534,
      "learning_rate": 2.8116806155338773e-05,
      "loss": 0.07006978988647461,
      "step": 1623
    },
    {
      "epoch": 0.21955959643756442,
      "grad_norm": 0.08281800895929337,
      "learning_rate": 2.8113484496770923e-05,
      "loss": 0.0648653507232666,
      "step": 1624
    },
    {
      "epoch": 0.21969479323340038,
      "grad_norm": 0.34033703804016113,
      "learning_rate": 2.811016010790567e-05,
      "loss": 0.10746574401855469,
      "step": 1625
    },
    {
      "epoch": 0.2198299900292363,
      "grad_norm": 0.08292002230882645,
      "learning_rate": 2.8106832989435165e-05,
      "loss": 0.05260574817657471,
      "step": 1626
    },
    {
      "epoch": 0.21996518682507224,
      "grad_norm": 0.14075486361980438,
      "learning_rate": 2.8103503142052146e-05,
      "loss": 0.04801654815673828,
      "step": 1627
    },
    {
      "epoch": 0.2201003836209082,
      "grad_norm": 0.365215539932251,
      "learning_rate": 2.8100170566449892e-05,
      "loss": 0.07610559463500977,
      "step": 1628
    },
    {
      "epoch": 0.22023558041674413,
      "grad_norm": 0.14373613893985748,
      "learning_rate": 2.8096835263322266e-05,
      "loss": 0.04710793495178223,
      "step": 1629
    },
    {
      "epoch": 0.22037077721258005,
      "grad_norm": 0.10747268050909042,
      "learning_rate": 2.8093497233363702e-05,
      "loss": 0.0658864974975586,
      "step": 1630
    },
    {
      "epoch": 0.220505974008416,
      "grad_norm": 0.1373828947544098,
      "learning_rate": 2.8090156477269185e-05,
      "loss": 0.06066489219665527,
      "step": 1631
    },
    {
      "epoch": 0.22064117080425194,
      "grad_norm": 0.0827164426445961,
      "learning_rate": 2.808681299573429e-05,
      "loss": 0.054712772369384766,
      "step": 1632
    },
    {
      "epoch": 0.22077636760008787,
      "grad_norm": 0.10212383419275284,
      "learning_rate": 2.8083466789455137e-05,
      "loss": 0.026240110397338867,
      "step": 1633
    },
    {
      "epoch": 0.22091156439592383,
      "grad_norm": 0.14644062519073486,
      "learning_rate": 2.808011785912843e-05,
      "loss": 0.06903481483459473,
      "step": 1634
    },
    {
      "epoch": 0.22104676119175976,
      "grad_norm": 0.1701095849275589,
      "learning_rate": 2.8076766205451435e-05,
      "loss": 0.03863692283630371,
      "step": 1635
    },
    {
      "epoch": 0.22118195798759568,
      "grad_norm": 0.2067553699016571,
      "learning_rate": 2.8073411829121983e-05,
      "loss": 0.09719610214233398,
      "step": 1636
    },
    {
      "epoch": 0.22131715478343164,
      "grad_norm": 0.1486196517944336,
      "learning_rate": 2.8070054730838467e-05,
      "loss": 0.0491485595703125,
      "step": 1637
    },
    {
      "epoch": 0.22145235157926757,
      "grad_norm": 0.07885777950286865,
      "learning_rate": 2.8066694911299865e-05,
      "loss": 0.0641927719116211,
      "step": 1638
    },
    {
      "epoch": 0.2215875483751035,
      "grad_norm": 0.09383964538574219,
      "learning_rate": 2.8063332371205698e-05,
      "loss": 0.10130882263183594,
      "step": 1639
    },
    {
      "epoch": 0.22172274517093946,
      "grad_norm": 0.10606039315462112,
      "learning_rate": 2.8059967111256072e-05,
      "loss": 0.07357358932495117,
      "step": 1640
    },
    {
      "epoch": 0.22185794196677538,
      "grad_norm": 0.09118998795747757,
      "learning_rate": 2.8056599132151647e-05,
      "loss": 0.06270885467529297,
      "step": 1641
    },
    {
      "epoch": 0.22199313876261131,
      "grad_norm": 0.11135542392730713,
      "learning_rate": 2.8053228434593656e-05,
      "loss": 0.06854462623596191,
      "step": 1642
    },
    {
      "epoch": 0.22212833555844727,
      "grad_norm": 0.25021982192993164,
      "learning_rate": 2.8049855019283895e-05,
      "loss": 0.09043657779693604,
      "step": 1643
    },
    {
      "epoch": 0.2222635323542832,
      "grad_norm": 0.16723372042179108,
      "learning_rate": 2.8046478886924736e-05,
      "loss": 0.11274099349975586,
      "step": 1644
    },
    {
      "epoch": 0.22239872915011913,
      "grad_norm": 0.07573637366294861,
      "learning_rate": 2.804310003821909e-05,
      "loss": 0.044938087463378906,
      "step": 1645
    },
    {
      "epoch": 0.22253392594595509,
      "grad_norm": 0.07393091917037964,
      "learning_rate": 2.8039718473870473e-05,
      "loss": 0.06756353378295898,
      "step": 1646
    },
    {
      "epoch": 0.22266912274179101,
      "grad_norm": 0.07173236459493637,
      "learning_rate": 2.8036334194582924e-05,
      "loss": 0.05489706993103027,
      "step": 1647
    },
    {
      "epoch": 0.22280431953762697,
      "grad_norm": 0.11214042454957962,
      "learning_rate": 2.8032947201061084e-05,
      "loss": 0.07085013389587402,
      "step": 1648
    },
    {
      "epoch": 0.2229395163334629,
      "grad_norm": 0.18437223136425018,
      "learning_rate": 2.8029557494010132e-05,
      "loss": 0.09244060516357422,
      "step": 1649
    },
    {
      "epoch": 0.22307471312929883,
      "grad_norm": 0.24310776591300964,
      "learning_rate": 2.802616507413583e-05,
      "loss": 0.0722799301147461,
      "step": 1650
    },
    {
      "epoch": 0.2232099099251348,
      "grad_norm": 0.08807925134897232,
      "learning_rate": 2.8022769942144492e-05,
      "loss": 0.07807731628417969,
      "step": 1651
    },
    {
      "epoch": 0.22334510672097072,
      "grad_norm": 0.2650485336780548,
      "learning_rate": 2.801937209874301e-05,
      "loss": 0.0627899169921875,
      "step": 1652
    },
    {
      "epoch": 0.22348030351680664,
      "grad_norm": 0.08361349254846573,
      "learning_rate": 2.8015971544638832e-05,
      "loss": 0.0717768669128418,
      "step": 1653
    },
    {
      "epoch": 0.2236155003126426,
      "grad_norm": 0.19367912411689758,
      "learning_rate": 2.8012568280539964e-05,
      "loss": 0.07652783393859863,
      "step": 1654
    },
    {
      "epoch": 0.22375069710847853,
      "grad_norm": 0.2210891842842102,
      "learning_rate": 2.800916230715499e-05,
      "loss": 0.07922744750976562,
      "step": 1655
    },
    {
      "epoch": 0.22388589390431446,
      "grad_norm": 0.15765532851219177,
      "learning_rate": 2.800575362519305e-05,
      "loss": 0.07997322082519531,
      "step": 1656
    },
    {
      "epoch": 0.22402109070015042,
      "grad_norm": 0.05100607872009277,
      "learning_rate": 2.800234223536385e-05,
      "loss": 0.04888582229614258,
      "step": 1657
    },
    {
      "epoch": 0.22415628749598634,
      "grad_norm": 0.12813875079154968,
      "learning_rate": 2.799892813837766e-05,
      "loss": 0.0952460765838623,
      "step": 1658
    },
    {
      "epoch": 0.22429148429182227,
      "grad_norm": 0.22915862500667572,
      "learning_rate": 2.7995511334945315e-05,
      "loss": 0.06994915008544922,
      "step": 1659
    },
    {
      "epoch": 0.22442668108765823,
      "grad_norm": 0.11997871100902557,
      "learning_rate": 2.7992091825778202e-05,
      "loss": 0.08458495140075684,
      "step": 1660
    },
    {
      "epoch": 0.22456187788349416,
      "grad_norm": 0.07525476068258286,
      "learning_rate": 2.7988669611588295e-05,
      "loss": 0.05947160720825195,
      "step": 1661
    },
    {
      "epoch": 0.2246970746793301,
      "grad_norm": 0.0761144831776619,
      "learning_rate": 2.7985244693088112e-05,
      "loss": 0.0955662727355957,
      "step": 1662
    },
    {
      "epoch": 0.22483227147516605,
      "grad_norm": 0.10482840240001678,
      "learning_rate": 2.7981817070990736e-05,
      "loss": 0.06867051124572754,
      "step": 1663
    },
    {
      "epoch": 0.22496746827100197,
      "grad_norm": 0.06475368142127991,
      "learning_rate": 2.7978386746009813e-05,
      "loss": 0.06558990478515625,
      "step": 1664
    },
    {
      "epoch": 0.2251026650668379,
      "grad_norm": 0.08630302548408508,
      "learning_rate": 2.797495371885957e-05,
      "loss": 0.09059739112854004,
      "step": 1665
    },
    {
      "epoch": 0.22523786186267386,
      "grad_norm": 0.14821258187294006,
      "learning_rate": 2.7971517990254768e-05,
      "loss": 0.06691479682922363,
      "step": 1666
    },
    {
      "epoch": 0.2253730586585098,
      "grad_norm": 0.21616770327091217,
      "learning_rate": 2.7968079560910744e-05,
      "loss": 0.06915140151977539,
      "step": 1667
    },
    {
      "epoch": 0.22550825545434572,
      "grad_norm": 0.11089816689491272,
      "learning_rate": 2.7964638431543402e-05,
      "loss": 0.04687166213989258,
      "step": 1668
    },
    {
      "epoch": 0.22564345225018168,
      "grad_norm": 0.2505188286304474,
      "learning_rate": 2.7961194602869208e-05,
      "loss": 0.09011673927307129,
      "step": 1669
    },
    {
      "epoch": 0.2257786490460176,
      "grad_norm": 0.11089015007019043,
      "learning_rate": 2.7957748075605178e-05,
      "loss": 0.06541275978088379,
      "step": 1670
    },
    {
      "epoch": 0.22591384584185356,
      "grad_norm": 0.17884966731071472,
      "learning_rate": 2.7954298850468898e-05,
      "loss": 0.09997940063476562,
      "step": 1671
    },
    {
      "epoch": 0.2260490426376895,
      "grad_norm": 0.15689659118652344,
      "learning_rate": 2.7950846928178517e-05,
      "loss": 0.09926700592041016,
      "step": 1672
    },
    {
      "epoch": 0.22618423943352542,
      "grad_norm": 0.12514255940914154,
      "learning_rate": 2.7947392309452744e-05,
      "loss": 0.06580209732055664,
      "step": 1673
    },
    {
      "epoch": 0.22631943622936138,
      "grad_norm": 0.10486900061368942,
      "learning_rate": 2.7943934995010845e-05,
      "loss": 0.06351089477539062,
      "step": 1674
    },
    {
      "epoch": 0.2264546330251973,
      "grad_norm": 0.08569465577602386,
      "learning_rate": 2.7940474985572657e-05,
      "loss": 0.05024516582489014,
      "step": 1675
    },
    {
      "epoch": 0.22658982982103323,
      "grad_norm": 0.06809850037097931,
      "learning_rate": 2.793701228185857e-05,
      "loss": 0.039927005767822266,
      "step": 1676
    },
    {
      "epoch": 0.2267250266168692,
      "grad_norm": 0.08198157697916031,
      "learning_rate": 2.7933546884589536e-05,
      "loss": 0.0691232681274414,
      "step": 1677
    },
    {
      "epoch": 0.22686022341270512,
      "grad_norm": 0.15173587203025818,
      "learning_rate": 2.7930078794487077e-05,
      "loss": 0.05842554569244385,
      "step": 1678
    },
    {
      "epoch": 0.22699542020854105,
      "grad_norm": 0.11979150772094727,
      "learning_rate": 2.7926608012273253e-05,
      "loss": 0.046627044677734375,
      "step": 1679
    },
    {
      "epoch": 0.227130617004377,
      "grad_norm": 0.1184156984090805,
      "learning_rate": 2.7923134538670715e-05,
      "loss": 0.11591625213623047,
      "step": 1680
    },
    {
      "epoch": 0.22726581380021293,
      "grad_norm": 0.07041200995445251,
      "learning_rate": 2.7919658374402645e-05,
      "loss": 0.05948638916015625,
      "step": 1681
    },
    {
      "epoch": 0.22740101059604886,
      "grad_norm": 0.16953514516353607,
      "learning_rate": 2.7916179520192807e-05,
      "loss": 0.057387351989746094,
      "step": 1682
    },
    {
      "epoch": 0.22753620739188482,
      "grad_norm": 0.09883497655391693,
      "learning_rate": 2.7912697976765516e-05,
      "loss": 0.06579351425170898,
      "step": 1683
    },
    {
      "epoch": 0.22767140418772075,
      "grad_norm": 0.15086331963539124,
      "learning_rate": 2.790921374484565e-05,
      "loss": 0.06711924076080322,
      "step": 1684
    },
    {
      "epoch": 0.22780660098355668,
      "grad_norm": 0.1059052050113678,
      "learning_rate": 2.7905726825158637e-05,
      "loss": 0.07807683944702148,
      "step": 1685
    },
    {
      "epoch": 0.22794179777939264,
      "grad_norm": 0.12642516195774078,
      "learning_rate": 2.7902237218430485e-05,
      "loss": 0.05850076675415039,
      "step": 1686
    },
    {
      "epoch": 0.22807699457522856,
      "grad_norm": 0.17913424968719482,
      "learning_rate": 2.7898744925387735e-05,
      "loss": 0.10469627380371094,
      "step": 1687
    },
    {
      "epoch": 0.2282121913710645,
      "grad_norm": 0.12199673801660538,
      "learning_rate": 2.7895249946757505e-05,
      "loss": 0.06017148494720459,
      "step": 1688
    },
    {
      "epoch": 0.22834738816690045,
      "grad_norm": 0.0852661207318306,
      "learning_rate": 2.7891752283267474e-05,
      "loss": 0.0633392333984375,
      "step": 1689
    },
    {
      "epoch": 0.22848258496273638,
      "grad_norm": 0.07933986186981201,
      "learning_rate": 2.788825193564587e-05,
      "loss": 0.056157827377319336,
      "step": 1690
    },
    {
      "epoch": 0.2286177817585723,
      "grad_norm": 0.08577398955821991,
      "learning_rate": 2.7884748904621483e-05,
      "loss": 0.09413909912109375,
      "step": 1691
    },
    {
      "epoch": 0.22875297855440826,
      "grad_norm": 0.07432276755571365,
      "learning_rate": 2.7881243190923667e-05,
      "loss": 0.06516695022583008,
      "step": 1692
    },
    {
      "epoch": 0.2288881753502442,
      "grad_norm": 0.12504330277442932,
      "learning_rate": 2.7877734795282326e-05,
      "loss": 0.07417774200439453,
      "step": 1693
    },
    {
      "epoch": 0.22902337214608015,
      "grad_norm": 0.09034009277820587,
      "learning_rate": 2.7874223718427926e-05,
      "loss": 0.05020713806152344,
      "step": 1694
    },
    {
      "epoch": 0.22915856894191608,
      "grad_norm": 0.1191539391875267,
      "learning_rate": 2.78707099610915e-05,
      "loss": 0.08720779418945312,
      "step": 1695
    },
    {
      "epoch": 0.229293765737752,
      "grad_norm": 0.18190699815750122,
      "learning_rate": 2.7867193524004618e-05,
      "loss": 0.0926237404346466,
      "step": 1696
    },
    {
      "epoch": 0.22942896253358797,
      "grad_norm": 0.1577400416135788,
      "learning_rate": 2.786367440789943e-05,
      "loss": 0.07935047149658203,
      "step": 1697
    },
    {
      "epoch": 0.2295641593294239,
      "grad_norm": 0.07673060148954391,
      "learning_rate": 2.7860152613508634e-05,
      "loss": 0.06094551086425781,
      "step": 1698
    },
    {
      "epoch": 0.22969935612525982,
      "grad_norm": 0.617973804473877,
      "learning_rate": 2.7856628141565484e-05,
      "loss": 0.1449728012084961,
      "step": 1699
    },
    {
      "epoch": 0.22983455292109578,
      "grad_norm": 0.07696376740932465,
      "learning_rate": 2.7853100992803797e-05,
      "loss": 0.06810665130615234,
      "step": 1700
    },
    {
      "epoch": 0.2299697497169317,
      "grad_norm": 0.2503134608268738,
      "learning_rate": 2.7849571167957942e-05,
      "loss": 0.08841514587402344,
      "step": 1701
    },
    {
      "epoch": 0.23010494651276764,
      "grad_norm": 0.16584907472133636,
      "learning_rate": 2.784603866776285e-05,
      "loss": 0.06377124786376953,
      "step": 1702
    },
    {
      "epoch": 0.2302401433086036,
      "grad_norm": 0.3064413070678711,
      "learning_rate": 2.7842503492953996e-05,
      "loss": 0.06874585151672363,
      "step": 1703
    },
    {
      "epoch": 0.23037534010443952,
      "grad_norm": 0.33040449023246765,
      "learning_rate": 2.7838965644267435e-05,
      "loss": 0.09757280349731445,
      "step": 1704
    },
    {
      "epoch": 0.23051053690027545,
      "grad_norm": 0.1609157770872116,
      "learning_rate": 2.7835425122439764e-05,
      "loss": 0.10168588161468506,
      "step": 1705
    },
    {
      "epoch": 0.2306457336961114,
      "grad_norm": 0.06962718069553375,
      "learning_rate": 2.7831881928208128e-05,
      "loss": 0.08070611953735352,
      "step": 1706
    },
    {
      "epoch": 0.23078093049194734,
      "grad_norm": 0.26125577092170715,
      "learning_rate": 2.7828336062310252e-05,
      "loss": 0.08952713012695312,
      "step": 1707
    },
    {
      "epoch": 0.23091612728778327,
      "grad_norm": 0.14154115319252014,
      "learning_rate": 2.7824787525484403e-05,
      "loss": 0.03807520866394043,
      "step": 1708
    },
    {
      "epoch": 0.23105132408361923,
      "grad_norm": 0.08921989798545837,
      "learning_rate": 2.7821236318469395e-05,
      "loss": 0.041365981101989746,
      "step": 1709
    },
    {
      "epoch": 0.23118652087945515,
      "grad_norm": 0.13391892611980438,
      "learning_rate": 2.7817682442004615e-05,
      "loss": 0.07260739803314209,
      "step": 1710
    },
    {
      "epoch": 0.23132171767529108,
      "grad_norm": 0.2577361464500427,
      "learning_rate": 2.781412589683e-05,
      "loss": 0.09927582740783691,
      "step": 1711
    },
    {
      "epoch": 0.23145691447112704,
      "grad_norm": 0.07660982012748718,
      "learning_rate": 2.781056668368604e-05,
      "loss": 0.0736246109008789,
      "step": 1712
    },
    {
      "epoch": 0.23159211126696297,
      "grad_norm": 0.09655215591192245,
      "learning_rate": 2.780700480331378e-05,
      "loss": 0.0680195689201355,
      "step": 1713
    },
    {
      "epoch": 0.2317273080627989,
      "grad_norm": 0.10156284272670746,
      "learning_rate": 2.7803440256454825e-05,
      "loss": 0.0675501823425293,
      "step": 1714
    },
    {
      "epoch": 0.23186250485863485,
      "grad_norm": 0.19986282289028168,
      "learning_rate": 2.7799873043851337e-05,
      "loss": 0.10214805603027344,
      "step": 1715
    },
    {
      "epoch": 0.23199770165447078,
      "grad_norm": 0.13760405778884888,
      "learning_rate": 2.7796303166246016e-05,
      "loss": 0.08921337127685547,
      "step": 1716
    },
    {
      "epoch": 0.23213289845030674,
      "grad_norm": 0.07674062997102737,
      "learning_rate": 2.7792730624382142e-05,
      "loss": 0.07658171653747559,
      "step": 1717
    },
    {
      "epoch": 0.23226809524614267,
      "grad_norm": 0.0891207680106163,
      "learning_rate": 2.778915541900353e-05,
      "loss": 0.08002948760986328,
      "step": 1718
    },
    {
      "epoch": 0.2324032920419786,
      "grad_norm": 0.11375489830970764,
      "learning_rate": 2.7785577550854566e-05,
      "loss": 0.05949282646179199,
      "step": 1719
    },
    {
      "epoch": 0.23253848883781456,
      "grad_norm": 0.09317008405923843,
      "learning_rate": 2.778199702068017e-05,
      "loss": 0.05891084671020508,
      "step": 1720
    },
    {
      "epoch": 0.23267368563365048,
      "grad_norm": 0.1997774839401245,
      "learning_rate": 2.777841382922583e-05,
      "loss": 0.11283183097839355,
      "step": 1721
    },
    {
      "epoch": 0.2328088824294864,
      "grad_norm": 0.23426946997642517,
      "learning_rate": 2.7774827977237596e-05,
      "loss": 0.08265495300292969,
      "step": 1722
    },
    {
      "epoch": 0.23294407922532237,
      "grad_norm": 0.2385103702545166,
      "learning_rate": 2.777123946546205e-05,
      "loss": 0.07417219877243042,
      "step": 1723
    },
    {
      "epoch": 0.2330792760211583,
      "grad_norm": 0.1510564684867859,
      "learning_rate": 2.776764829464634e-05,
      "loss": 0.09803891181945801,
      "step": 1724
    },
    {
      "epoch": 0.23321447281699423,
      "grad_norm": 0.1913069486618042,
      "learning_rate": 2.7764054465538173e-05,
      "loss": 0.07884073257446289,
      "step": 1725
    },
    {
      "epoch": 0.23334966961283019,
      "grad_norm": 0.1414082646369934,
      "learning_rate": 2.7760457978885794e-05,
      "loss": 0.06635475158691406,
      "step": 1726
    },
    {
      "epoch": 0.23348486640866611,
      "grad_norm": 0.2202560156583786,
      "learning_rate": 2.7756858835438022e-05,
      "loss": 0.06752020120620728,
      "step": 1727
    },
    {
      "epoch": 0.23362006320450204,
      "grad_norm": 0.056420207023620605,
      "learning_rate": 2.7753257035944216e-05,
      "loss": 0.052390098571777344,
      "step": 1728
    },
    {
      "epoch": 0.233755260000338,
      "grad_norm": 0.15214776992797852,
      "learning_rate": 2.7749652581154277e-05,
      "loss": 0.056015610694885254,
      "step": 1729
    },
    {
      "epoch": 0.23389045679617393,
      "grad_norm": 0.18461285531520844,
      "learning_rate": 2.7746045471818685e-05,
      "loss": 0.08249831199645996,
      "step": 1730
    },
    {
      "epoch": 0.23402565359200986,
      "grad_norm": 0.16687346994876862,
      "learning_rate": 2.7742435708688458e-05,
      "loss": 0.06353402137756348,
      "step": 1731
    },
    {
      "epoch": 0.23416085038784581,
      "grad_norm": 0.11642317473888397,
      "learning_rate": 2.7738823292515167e-05,
      "loss": 0.10229349136352539,
      "step": 1732
    },
    {
      "epoch": 0.23429604718368174,
      "grad_norm": 0.10276017338037491,
      "learning_rate": 2.773520822405093e-05,
      "loss": 0.10089659690856934,
      "step": 1733
    },
    {
      "epoch": 0.23443124397951767,
      "grad_norm": 0.12194877117872238,
      "learning_rate": 2.7731590504048433e-05,
      "loss": 0.043982505798339844,
      "step": 1734
    },
    {
      "epoch": 0.23456644077535363,
      "grad_norm": 0.07328354567289352,
      "learning_rate": 2.7727970133260896e-05,
      "loss": 0.05307340621948242,
      "step": 1735
    },
    {
      "epoch": 0.23470163757118956,
      "grad_norm": 0.11488433927297592,
      "learning_rate": 2.7724347112442106e-05,
      "loss": 0.05113840103149414,
      "step": 1736
    },
    {
      "epoch": 0.2348368343670255,
      "grad_norm": 0.2745373249053955,
      "learning_rate": 2.772072144234639e-05,
      "loss": 0.07746362686157227,
      "step": 1737
    },
    {
      "epoch": 0.23497203116286144,
      "grad_norm": 0.16850346326828003,
      "learning_rate": 2.7717093123728634e-05,
      "loss": 0.08483743667602539,
      "step": 1738
    },
    {
      "epoch": 0.23510722795869737,
      "grad_norm": 0.06206810846924782,
      "learning_rate": 2.771346215734428e-05,
      "loss": 0.04123663902282715,
      "step": 1739
    },
    {
      "epoch": 0.23524242475453333,
      "grad_norm": 0.10241566598415375,
      "learning_rate": 2.7709828543949302e-05,
      "loss": 0.11449801921844482,
      "step": 1740
    },
    {
      "epoch": 0.23537762155036926,
      "grad_norm": 0.11740782111883163,
      "learning_rate": 2.770619228430025e-05,
      "loss": 0.0947728157043457,
      "step": 1741
    },
    {
      "epoch": 0.2355128183462052,
      "grad_norm": 0.10658842325210571,
      "learning_rate": 2.77025533791542e-05,
      "loss": 0.07320404052734375,
      "step": 1742
    },
    {
      "epoch": 0.23564801514204115,
      "grad_norm": 0.2130248099565506,
      "learning_rate": 2.76989118292688e-05,
      "loss": 0.0629202127456665,
      "step": 1743
    },
    {
      "epoch": 0.23578321193787707,
      "grad_norm": 0.08386167883872986,
      "learning_rate": 2.7695267635402242e-05,
      "loss": 0.06380367279052734,
      "step": 1744
    },
    {
      "epoch": 0.235918408733713,
      "grad_norm": 0.053069718182086945,
      "learning_rate": 2.7691620798313258e-05,
      "loss": 0.0423736572265625,
      "step": 1745
    },
    {
      "epoch": 0.23605360552954896,
      "grad_norm": 0.07481296360492706,
      "learning_rate": 2.7687971318761145e-05,
      "loss": 0.05146193504333496,
      "step": 1746
    },
    {
      "epoch": 0.2361888023253849,
      "grad_norm": 0.10211343318223953,
      "learning_rate": 2.7684319197505746e-05,
      "loss": 0.054984331130981445,
      "step": 1747
    },
    {
      "epoch": 0.23632399912122082,
      "grad_norm": 0.0420459620654583,
      "learning_rate": 2.7680664435307446e-05,
      "loss": 0.042733073234558105,
      "step": 1748
    },
    {
      "epoch": 0.23645919591705677,
      "grad_norm": 0.09920414537191391,
      "learning_rate": 2.767700703292719e-05,
      "loss": 0.08068704605102539,
      "step": 1749
    },
    {
      "epoch": 0.2365943927128927,
      "grad_norm": 0.056447435170412064,
      "learning_rate": 2.767334699112647e-05,
      "loss": 0.061174869537353516,
      "step": 1750
    },
    {
      "epoch": 0.23672958950872863,
      "grad_norm": 0.1442887783050537,
      "learning_rate": 2.7669684310667318e-05,
      "loss": 0.10539007186889648,
      "step": 1751
    },
    {
      "epoch": 0.2368647863045646,
      "grad_norm": 0.0931338518857956,
      "learning_rate": 2.7666018992312333e-05,
      "loss": 0.04785966873168945,
      "step": 1752
    },
    {
      "epoch": 0.23699998310040052,
      "grad_norm": 0.045554425567388535,
      "learning_rate": 2.7662351036824653e-05,
      "loss": 0.04404497146606445,
      "step": 1753
    },
    {
      "epoch": 0.23713517989623645,
      "grad_norm": 0.04117034003138542,
      "learning_rate": 2.7658680444967964e-05,
      "loss": 0.036290884017944336,
      "step": 1754
    },
    {
      "epoch": 0.2372703766920724,
      "grad_norm": 0.11715850234031677,
      "learning_rate": 2.76550072175065e-05,
      "loss": 0.09999275207519531,
      "step": 1755
    },
    {
      "epoch": 0.23740557348790833,
      "grad_norm": 0.1328911930322647,
      "learning_rate": 2.7651331355205044e-05,
      "loss": 0.0979456901550293,
      "step": 1756
    },
    {
      "epoch": 0.23754077028374426,
      "grad_norm": 0.055276934057474136,
      "learning_rate": 2.7647652858828936e-05,
      "loss": 0.05973386764526367,
      "step": 1757
    },
    {
      "epoch": 0.23767596707958022,
      "grad_norm": 0.05377735570073128,
      "learning_rate": 2.764397172914406e-05,
      "loss": 0.06414568424224854,
      "step": 1758
    },
    {
      "epoch": 0.23781116387541615,
      "grad_norm": 0.1497291475534439,
      "learning_rate": 2.7640287966916845e-05,
      "loss": 0.07093620300292969,
      "step": 1759
    },
    {
      "epoch": 0.2379463606712521,
      "grad_norm": 0.10229865461587906,
      "learning_rate": 2.7636601572914266e-05,
      "loss": 0.07892036437988281,
      "step": 1760
    },
    {
      "epoch": 0.23808155746708803,
      "grad_norm": 0.11082065105438232,
      "learning_rate": 2.7632912547903855e-05,
      "loss": 0.06773090362548828,
      "step": 1761
    },
    {
      "epoch": 0.23821675426292396,
      "grad_norm": 0.0587751530110836,
      "learning_rate": 2.7629220892653685e-05,
      "loss": 0.06294918060302734,
      "step": 1762
    },
    {
      "epoch": 0.23835195105875992,
      "grad_norm": 0.051163315773010254,
      "learning_rate": 2.7625526607932378e-05,
      "loss": 0.06062507629394531,
      "step": 1763
    },
    {
      "epoch": 0.23848714785459585,
      "grad_norm": 0.07822441309690475,
      "learning_rate": 2.76218296945091e-05,
      "loss": 0.09968280792236328,
      "step": 1764
    },
    {
      "epoch": 0.23862234465043178,
      "grad_norm": 0.09619243443012238,
      "learning_rate": 2.7618130153153577e-05,
      "loss": 0.06036710739135742,
      "step": 1765
    },
    {
      "epoch": 0.23875754144626773,
      "grad_norm": 0.057991448789834976,
      "learning_rate": 2.7614427984636063e-05,
      "loss": 0.047690629959106445,
      "step": 1766
    },
    {
      "epoch": 0.23889273824210366,
      "grad_norm": 0.144164577126503,
      "learning_rate": 2.7610723189727377e-05,
      "loss": 0.0775446891784668,
      "step": 1767
    },
    {
      "epoch": 0.2390279350379396,
      "grad_norm": 0.35411012172698975,
      "learning_rate": 2.760701576919888e-05,
      "loss": 0.10361731052398682,
      "step": 1768
    },
    {
      "epoch": 0.23916313183377555,
      "grad_norm": 0.20320117473602295,
      "learning_rate": 2.760330572382246e-05,
      "loss": 0.10682868957519531,
      "step": 1769
    },
    {
      "epoch": 0.23929832862961148,
      "grad_norm": 0.09519585967063904,
      "learning_rate": 2.7599593054370584e-05,
      "loss": 0.03622698783874512,
      "step": 1770
    },
    {
      "epoch": 0.2394335254254474,
      "grad_norm": 0.22248098254203796,
      "learning_rate": 2.7595877761616246e-05,
      "loss": 0.06809139251708984,
      "step": 1771
    },
    {
      "epoch": 0.23956872222128336,
      "grad_norm": 0.20105934143066406,
      "learning_rate": 2.759215984633299e-05,
      "loss": 0.061814308166503906,
      "step": 1772
    },
    {
      "epoch": 0.2397039190171193,
      "grad_norm": 0.20335231721401215,
      "learning_rate": 2.7588439309294902e-05,
      "loss": 0.08295965194702148,
      "step": 1773
    },
    {
      "epoch": 0.23983911581295522,
      "grad_norm": 0.0795859843492508,
      "learning_rate": 2.7584716151276623e-05,
      "loss": 0.03898775577545166,
      "step": 1774
    },
    {
      "epoch": 0.23997431260879118,
      "grad_norm": 0.13627904653549194,
      "learning_rate": 2.7580990373053325e-05,
      "loss": 0.0668020248413086,
      "step": 1775
    },
    {
      "epoch": 0.2401095094046271,
      "grad_norm": 0.47427457571029663,
      "learning_rate": 2.7577261975400747e-05,
      "loss": 0.10455322265625,
      "step": 1776
    },
    {
      "epoch": 0.24024470620046304,
      "grad_norm": 0.1331705003976822,
      "learning_rate": 2.7573530959095154e-05,
      "loss": 0.07956361770629883,
      "step": 1777
    },
    {
      "epoch": 0.240379902996299,
      "grad_norm": 0.2390034794807434,
      "learning_rate": 2.756979732491336e-05,
      "loss": 0.08353757858276367,
      "step": 1778
    },
    {
      "epoch": 0.24051509979213492,
      "grad_norm": 0.25375720858573914,
      "learning_rate": 2.756606107363274e-05,
      "loss": 0.07422256469726562,
      "step": 1779
    },
    {
      "epoch": 0.24065029658797085,
      "grad_norm": 0.20521622896194458,
      "learning_rate": 2.7562322206031192e-05,
      "loss": 0.07655715942382812,
      "step": 1780
    },
    {
      "epoch": 0.2407854933838068,
      "grad_norm": 0.08148027211427689,
      "learning_rate": 2.7558580722887166e-05,
      "loss": 0.05353736877441406,
      "step": 1781
    },
    {
      "epoch": 0.24092069017964274,
      "grad_norm": 0.07796908169984818,
      "learning_rate": 2.7554836624979666e-05,
      "loss": 0.060524940490722656,
      "step": 1782
    },
    {
      "epoch": 0.2410558869754787,
      "grad_norm": 0.21814604103565216,
      "learning_rate": 2.7551089913088233e-05,
      "loss": 0.062492966651916504,
      "step": 1783
    },
    {
      "epoch": 0.24119108377131462,
      "grad_norm": 0.10739132016897202,
      "learning_rate": 2.7547340587992948e-05,
      "loss": 0.06865286827087402,
      "step": 1784
    },
    {
      "epoch": 0.24132628056715055,
      "grad_norm": 0.14468367397785187,
      "learning_rate": 2.754358865047444e-05,
      "loss": 0.05918312072753906,
      "step": 1785
    },
    {
      "epoch": 0.2414614773629865,
      "grad_norm": 0.2825700342655182,
      "learning_rate": 2.7539834101313885e-05,
      "loss": 0.09009218215942383,
      "step": 1786
    },
    {
      "epoch": 0.24159667415882244,
      "grad_norm": 0.07572701573371887,
      "learning_rate": 2.7536076941293003e-05,
      "loss": 0.047743797302246094,
      "step": 1787
    },
    {
      "epoch": 0.24173187095465837,
      "grad_norm": 0.07999354600906372,
      "learning_rate": 2.753231717119405e-05,
      "loss": 0.100921630859375,
      "step": 1788
    },
    {
      "epoch": 0.24186706775049432,
      "grad_norm": 0.12332987040281296,
      "learning_rate": 2.7528554791799826e-05,
      "loss": 0.06456756591796875,
      "step": 1789
    },
    {
      "epoch": 0.24200226454633025,
      "grad_norm": 0.1570417284965515,
      "learning_rate": 2.7524789803893686e-05,
      "loss": 0.06658697128295898,
      "step": 1790
    },
    {
      "epoch": 0.24213746134216618,
      "grad_norm": 0.16550730168819427,
      "learning_rate": 2.7521022208259526e-05,
      "loss": 0.05148625373840332,
      "step": 1791
    },
    {
      "epoch": 0.24227265813800214,
      "grad_norm": 0.08395404368638992,
      "learning_rate": 2.7517252005681762e-05,
      "loss": 0.07175612449645996,
      "step": 1792
    },
    {
      "epoch": 0.24240785493383807,
      "grad_norm": 0.07427235692739487,
      "learning_rate": 2.7513479196945385e-05,
      "loss": 0.042111873626708984,
      "step": 1793
    },
    {
      "epoch": 0.242543051729674,
      "grad_norm": 0.10649193823337555,
      "learning_rate": 2.750970378283591e-05,
      "loss": 0.06889820098876953,
      "step": 1794
    },
    {
      "epoch": 0.24267824852550995,
      "grad_norm": 0.07281965017318726,
      "learning_rate": 2.7505925764139398e-05,
      "loss": 0.06787073612213135,
      "step": 1795
    },
    {
      "epoch": 0.24281344532134588,
      "grad_norm": 0.1479542851448059,
      "learning_rate": 2.7502145141642447e-05,
      "loss": 0.10392379760742188,
      "step": 1796
    },
    {
      "epoch": 0.2429486421171818,
      "grad_norm": 0.17237550020217896,
      "learning_rate": 2.7498361916132212e-05,
      "loss": 0.08661484718322754,
      "step": 1797
    },
    {
      "epoch": 0.24308383891301777,
      "grad_norm": 0.10848338901996613,
      "learning_rate": 2.7494576088396376e-05,
      "loss": 0.058625221252441406,
      "step": 1798
    },
    {
      "epoch": 0.2432190357088537,
      "grad_norm": 0.19653986394405365,
      "learning_rate": 2.749078765922317e-05,
      "loss": 0.10975790023803711,
      "step": 1799
    },
    {
      "epoch": 0.24335423250468963,
      "grad_norm": 0.1427716165781021,
      "learning_rate": 2.7486996629401366e-05,
      "loss": 0.0874488353729248,
      "step": 1800
    },
    {
      "epoch": 0.24348942930052558,
      "grad_norm": 0.1089082807302475,
      "learning_rate": 2.7483202999720272e-05,
      "loss": 0.0858912467956543,
      "step": 1801
    },
    {
      "epoch": 0.2436246260963615,
      "grad_norm": 0.07230830192565918,
      "learning_rate": 2.7479406770969747e-05,
      "loss": 0.0698235034942627,
      "step": 1802
    },
    {
      "epoch": 0.24375982289219744,
      "grad_norm": 0.07871078699827194,
      "learning_rate": 2.7475607943940182e-05,
      "loss": 0.07205390930175781,
      "step": 1803
    },
    {
      "epoch": 0.2438950196880334,
      "grad_norm": 0.19898729026317596,
      "learning_rate": 2.7471806519422514e-05,
      "loss": 0.07661724090576172,
      "step": 1804
    },
    {
      "epoch": 0.24403021648386933,
      "grad_norm": 0.07905599474906921,
      "learning_rate": 2.746800249820822e-05,
      "loss": 0.06638145446777344,
      "step": 1805
    },
    {
      "epoch": 0.24416541327970528,
      "grad_norm": 0.043868523091077805,
      "learning_rate": 2.7464195881089323e-05,
      "loss": 0.0452427864074707,
      "step": 1806
    },
    {
      "epoch": 0.2443006100755412,
      "grad_norm": 0.12468072772026062,
      "learning_rate": 2.746038666885837e-05,
      "loss": 0.0823965072631836,
      "step": 1807
    },
    {
      "epoch": 0.24443580687137714,
      "grad_norm": 0.17508931457996368,
      "learning_rate": 2.7456574862308474e-05,
      "loss": 0.0991206169128418,
      "step": 1808
    },
    {
      "epoch": 0.2445710036672131,
      "grad_norm": 0.15082964301109314,
      "learning_rate": 2.745276046223326e-05,
      "loss": 0.08705997467041016,
      "step": 1809
    },
    {
      "epoch": 0.24470620046304903,
      "grad_norm": 0.18758554756641388,
      "learning_rate": 2.744894346942691e-05,
      "loss": 0.09409761428833008,
      "step": 1810
    },
    {
      "epoch": 0.24484139725888496,
      "grad_norm": 0.1278875768184662,
      "learning_rate": 2.744512388468415e-05,
      "loss": 0.0665130615234375,
      "step": 1811
    },
    {
      "epoch": 0.24497659405472091,
      "grad_norm": 0.1592629849910736,
      "learning_rate": 2.7441301708800227e-05,
      "loss": 0.08577775955200195,
      "step": 1812
    },
    {
      "epoch": 0.24511179085055684,
      "grad_norm": 0.17943112552165985,
      "learning_rate": 2.7437476942570942e-05,
      "loss": 0.07355308532714844,
      "step": 1813
    },
    {
      "epoch": 0.24524698764639277,
      "grad_norm": 0.09428275376558304,
      "learning_rate": 2.7433649586792637e-05,
      "loss": 0.06973576545715332,
      "step": 1814
    },
    {
      "epoch": 0.24538218444222873,
      "grad_norm": 0.10559109598398209,
      "learning_rate": 2.7429819642262178e-05,
      "loss": 0.07807350158691406,
      "step": 1815
    },
    {
      "epoch": 0.24551738123806466,
      "grad_norm": 0.05933111533522606,
      "learning_rate": 2.7425987109776994e-05,
      "loss": 0.046124935150146484,
      "step": 1816
    },
    {
      "epoch": 0.2456525780339006,
      "grad_norm": 0.08133486658334732,
      "learning_rate": 2.7422151990135022e-05,
      "loss": 0.08216977119445801,
      "step": 1817
    },
    {
      "epoch": 0.24578777482973654,
      "grad_norm": 0.08142583072185516,
      "learning_rate": 2.741831428413477e-05,
      "loss": 0.050116539001464844,
      "step": 1818
    },
    {
      "epoch": 0.24592297162557247,
      "grad_norm": 0.11490466445684433,
      "learning_rate": 2.7414473992575257e-05,
      "loss": 0.0796046257019043,
      "step": 1819
    },
    {
      "epoch": 0.2460581684214084,
      "grad_norm": 0.08164677023887634,
      "learning_rate": 2.7410631116256054e-05,
      "loss": 0.05349743366241455,
      "step": 1820
    },
    {
      "epoch": 0.24619336521724436,
      "grad_norm": 0.09565792977809906,
      "learning_rate": 2.7406785655977275e-05,
      "loss": 0.055932044982910156,
      "step": 1821
    },
    {
      "epoch": 0.2463285620130803,
      "grad_norm": 0.05333436280488968,
      "learning_rate": 2.7402937612539563e-05,
      "loss": 0.061859846115112305,
      "step": 1822
    },
    {
      "epoch": 0.24646375880891622,
      "grad_norm": 0.044519152492284775,
      "learning_rate": 2.7399086986744095e-05,
      "loss": 0.061359405517578125,
      "step": 1823
    },
    {
      "epoch": 0.24659895560475217,
      "grad_norm": 0.13185469806194305,
      "learning_rate": 2.7395233779392598e-05,
      "loss": 0.06690573692321777,
      "step": 1824
    },
    {
      "epoch": 0.2467341524005881,
      "grad_norm": 0.09143054485321045,
      "learning_rate": 2.739137799128733e-05,
      "loss": 0.09738707542419434,
      "step": 1825
    },
    {
      "epoch": 0.24686934919642403,
      "grad_norm": 0.06961528211832047,
      "learning_rate": 2.7387519623231085e-05,
      "loss": 0.0673065185546875,
      "step": 1826
    },
    {
      "epoch": 0.24700454599226,
      "grad_norm": 0.06789089739322662,
      "learning_rate": 2.7383658676027195e-05,
      "loss": 0.07415652275085449,
      "step": 1827
    },
    {
      "epoch": 0.24713974278809592,
      "grad_norm": 0.1685425192117691,
      "learning_rate": 2.7379795150479535e-05,
      "loss": 0.08176422119140625,
      "step": 1828
    },
    {
      "epoch": 0.24727493958393187,
      "grad_norm": 0.08949195593595505,
      "learning_rate": 2.73759290473925e-05,
      "loss": 0.05508732795715332,
      "step": 1829
    },
    {
      "epoch": 0.2474101363797678,
      "grad_norm": 0.0691029280424118,
      "learning_rate": 2.7372060367571044e-05,
      "loss": 0.06569278240203857,
      "step": 1830
    },
    {
      "epoch": 0.24754533317560373,
      "grad_norm": 0.15121932327747345,
      "learning_rate": 2.7368189111820648e-05,
      "loss": 0.08872222900390625,
      "step": 1831
    },
    {
      "epoch": 0.2476805299714397,
      "grad_norm": 0.08026376366615295,
      "learning_rate": 2.736431528094732e-05,
      "loss": 0.051239728927612305,
      "step": 1832
    },
    {
      "epoch": 0.24781572676727562,
      "grad_norm": 0.04872721806168556,
      "learning_rate": 2.7360438875757614e-05,
      "loss": 0.04367685317993164,
      "step": 1833
    },
    {
      "epoch": 0.24795092356311155,
      "grad_norm": 0.09968399256467819,
      "learning_rate": 2.7356559897058624e-05,
      "loss": 0.07347869873046875,
      "step": 1834
    },
    {
      "epoch": 0.2480861203589475,
      "grad_norm": 0.17475005984306335,
      "learning_rate": 2.735267834565797e-05,
      "loss": 0.10741519927978516,
      "step": 1835
    },
    {
      "epoch": 0.24822131715478343,
      "grad_norm": 0.15285226702690125,
      "learning_rate": 2.734879422236381e-05,
      "loss": 0.07441902160644531,
      "step": 1836
    },
    {
      "epoch": 0.24835651395061936,
      "grad_norm": 0.119447261095047,
      "learning_rate": 2.734490752798484e-05,
      "loss": 0.051962852478027344,
      "step": 1837
    },
    {
      "epoch": 0.24849171074645532,
      "grad_norm": 0.22703321278095245,
      "learning_rate": 2.7341018263330296e-05,
      "loss": 0.09919500350952148,
      "step": 1838
    },
    {
      "epoch": 0.24862690754229125,
      "grad_norm": 0.14595769345760345,
      "learning_rate": 2.7337126429209935e-05,
      "loss": 0.0747079849243164,
      "step": 1839
    },
    {
      "epoch": 0.24876210433812718,
      "grad_norm": 0.0487702377140522,
      "learning_rate": 2.7333232026434064e-05,
      "loss": 0.06485152244567871,
      "step": 1840
    },
    {
      "epoch": 0.24889730113396313,
      "grad_norm": 0.14004072546958923,
      "learning_rate": 2.7329335055813517e-05,
      "loss": 0.07010078430175781,
      "step": 1841
    },
    {
      "epoch": 0.24903249792979906,
      "grad_norm": 0.05502966418862343,
      "learning_rate": 2.732543551815966e-05,
      "loss": 0.05611228942871094,
      "step": 1842
    },
    {
      "epoch": 0.249167694725635,
      "grad_norm": 0.15478190779685974,
      "learning_rate": 2.7321533414284404e-05,
      "loss": 0.07949399948120117,
      "step": 1843
    },
    {
      "epoch": 0.24930289152147095,
      "grad_norm": 0.043452560901641846,
      "learning_rate": 2.731762874500018e-05,
      "loss": 0.045624732971191406,
      "step": 1844
    },
    {
      "epoch": 0.24943808831730688,
      "grad_norm": 0.08242925256490707,
      "learning_rate": 2.7313721511119972e-05,
      "loss": 0.07660722732543945,
      "step": 1845
    },
    {
      "epoch": 0.2495732851131428,
      "grad_norm": 0.1302908957004547,
      "learning_rate": 2.7309811713457275e-05,
      "loss": 0.07224464416503906,
      "step": 1846
    },
    {
      "epoch": 0.24970848190897876,
      "grad_norm": 0.16624121367931366,
      "learning_rate": 2.730589935282614e-05,
      "loss": 0.09778642654418945,
      "step": 1847
    },
    {
      "epoch": 0.2498436787048147,
      "grad_norm": 0.09746264666318893,
      "learning_rate": 2.7301984430041135e-05,
      "loss": 0.058812618255615234,
      "step": 1848
    },
    {
      "epoch": 0.24997887550065062,
      "grad_norm": 0.08061551302671432,
      "learning_rate": 2.7298066945917368e-05,
      "loss": 0.07266318798065186,
      "step": 1849
    },
    {
      "epoch": 0.2501140722964866,
      "grad_norm": 0.18297156691551208,
      "learning_rate": 2.7294146901270482e-05,
      "loss": 0.09313583374023438,
      "step": 1850
    },
    {
      "epoch": 0.25024926909232253,
      "grad_norm": 0.07570967078208923,
      "learning_rate": 2.7290224296916653e-05,
      "loss": 0.0724782943725586,
      "step": 1851
    },
    {
      "epoch": 0.25038446588815844,
      "grad_norm": 0.10806790739297867,
      "learning_rate": 2.7286299133672584e-05,
      "loss": 0.05780839920043945,
      "step": 1852
    },
    {
      "epoch": 0.2505196626839944,
      "grad_norm": 0.05563180893659592,
      "learning_rate": 2.728237141235552e-05,
      "loss": 0.06669878959655762,
      "step": 1853
    },
    {
      "epoch": 0.25065485947983035,
      "grad_norm": 0.12812623381614685,
      "learning_rate": 2.727844113378322e-05,
      "loss": 0.07734537124633789,
      "step": 1854
    },
    {
      "epoch": 0.25079005627566625,
      "grad_norm": 0.11334457993507385,
      "learning_rate": 2.7274508298774013e-05,
      "loss": 0.06672239303588867,
      "step": 1855
    },
    {
      "epoch": 0.2509252530715022,
      "grad_norm": 0.10986971855163574,
      "learning_rate": 2.727057290814672e-05,
      "loss": 0.0865480899810791,
      "step": 1856
    },
    {
      "epoch": 0.25106044986733816,
      "grad_norm": 0.2216511070728302,
      "learning_rate": 2.7266634962720704e-05,
      "loss": 0.07958316802978516,
      "step": 1857
    },
    {
      "epoch": 0.25119564666317407,
      "grad_norm": 0.33249416947364807,
      "learning_rate": 2.726269446331588e-05,
      "loss": 0.09619903564453125,
      "step": 1858
    },
    {
      "epoch": 0.25133084345901,
      "grad_norm": 0.05294548347592354,
      "learning_rate": 2.7258751410752676e-05,
      "loss": 0.046959877014160156,
      "step": 1859
    },
    {
      "epoch": 0.251466040254846,
      "grad_norm": 0.12228503823280334,
      "learning_rate": 2.725480580585206e-05,
      "loss": 0.05336642265319824,
      "step": 1860
    },
    {
      "epoch": 0.2516012370506819,
      "grad_norm": 0.11407537758350372,
      "learning_rate": 2.7250857649435522e-05,
      "loss": 0.05225372314453125,
      "step": 1861
    },
    {
      "epoch": 0.25173643384651784,
      "grad_norm": 0.2380487620830536,
      "learning_rate": 2.724690694232509e-05,
      "loss": 0.07039976119995117,
      "step": 1862
    },
    {
      "epoch": 0.2518716306423538,
      "grad_norm": 0.03274841979146004,
      "learning_rate": 2.7242953685343327e-05,
      "loss": 0.036340951919555664,
      "step": 1863
    },
    {
      "epoch": 0.2520068274381897,
      "grad_norm": 0.29131993651390076,
      "learning_rate": 2.723899787931332e-05,
      "loss": 0.09333992004394531,
      "step": 1864
    },
    {
      "epoch": 0.25214202423402565,
      "grad_norm": 0.10917803645133972,
      "learning_rate": 2.7235039525058684e-05,
      "loss": 0.09343409538269043,
      "step": 1865
    },
    {
      "epoch": 0.2522772210298616,
      "grad_norm": 0.07211418449878693,
      "learning_rate": 2.7231078623403575e-05,
      "loss": 0.06926798820495605,
      "step": 1866
    },
    {
      "epoch": 0.2524124178256975,
      "grad_norm": 0.1154571995139122,
      "learning_rate": 2.722711517517267e-05,
      "loss": 0.06325149536132812,
      "step": 1867
    },
    {
      "epoch": 0.25254761462153347,
      "grad_norm": 0.22004513442516327,
      "learning_rate": 2.7223149181191187e-05,
      "loss": 0.08453893661499023,
      "step": 1868
    },
    {
      "epoch": 0.2526828114173694,
      "grad_norm": 0.08724702894687653,
      "learning_rate": 2.7219180642284864e-05,
      "loss": 0.07194232940673828,
      "step": 1869
    },
    {
      "epoch": 0.2528180082132053,
      "grad_norm": 0.05822712555527687,
      "learning_rate": 2.721520955927997e-05,
      "loss": 0.06417250633239746,
      "step": 1870
    },
    {
      "epoch": 0.2529532050090413,
      "grad_norm": 0.18000958859920502,
      "learning_rate": 2.7211235933003302e-05,
      "loss": 0.06889104843139648,
      "step": 1871
    },
    {
      "epoch": 0.25308840180487724,
      "grad_norm": 0.10869405418634415,
      "learning_rate": 2.72072597642822e-05,
      "loss": 0.08119058609008789,
      "step": 1872
    },
    {
      "epoch": 0.25322359860071314,
      "grad_norm": 0.13462482392787933,
      "learning_rate": 2.7203281053944512e-05,
      "loss": 0.08239626884460449,
      "step": 1873
    },
    {
      "epoch": 0.2533587953965491,
      "grad_norm": 0.15795378386974335,
      "learning_rate": 2.719929980281864e-05,
      "loss": 0.06319999694824219,
      "step": 1874
    },
    {
      "epoch": 0.25349399219238505,
      "grad_norm": 0.16120566427707672,
      "learning_rate": 2.719531601173349e-05,
      "loss": 0.06544899940490723,
      "step": 1875
    },
    {
      "epoch": 0.25362918898822095,
      "grad_norm": 0.12590977549552917,
      "learning_rate": 2.7191329681518512e-05,
      "loss": 0.0908041000366211,
      "step": 1876
    },
    {
      "epoch": 0.2537643857840569,
      "grad_norm": 0.07940160483121872,
      "learning_rate": 2.7187340813003682e-05,
      "loss": 0.05701804161071777,
      "step": 1877
    },
    {
      "epoch": 0.25389958257989287,
      "grad_norm": 0.1196286678314209,
      "learning_rate": 2.718334940701951e-05,
      "loss": 0.05937361717224121,
      "step": 1878
    },
    {
      "epoch": 0.25403477937572877,
      "grad_norm": 0.1276506930589676,
      "learning_rate": 2.7179355464397014e-05,
      "loss": 0.10046029090881348,
      "step": 1879
    },
    {
      "epoch": 0.2541699761715647,
      "grad_norm": 0.0757499560713768,
      "learning_rate": 2.7175358985967763e-05,
      "loss": 0.032299041748046875,
      "step": 1880
    },
    {
      "epoch": 0.2543051729674007,
      "grad_norm": 0.1111755296587944,
      "learning_rate": 2.717135997256385e-05,
      "loss": 0.061296701431274414,
      "step": 1881
    },
    {
      "epoch": 0.2544403697632366,
      "grad_norm": 0.1633571982383728,
      "learning_rate": 2.7167358425017882e-05,
      "loss": 0.08982467651367188,
      "step": 1882
    },
    {
      "epoch": 0.25457556655907254,
      "grad_norm": 0.23074117302894592,
      "learning_rate": 2.7163354344163004e-05,
      "loss": 0.08429047465324402,
      "step": 1883
    },
    {
      "epoch": 0.2547107633549085,
      "grad_norm": 0.09363125264644623,
      "learning_rate": 2.715934773083289e-05,
      "loss": 0.07416868209838867,
      "step": 1884
    },
    {
      "epoch": 0.25484596015074445,
      "grad_norm": 0.09598194062709808,
      "learning_rate": 2.715533858586174e-05,
      "loss": 0.07480204105377197,
      "step": 1885
    },
    {
      "epoch": 0.25498115694658036,
      "grad_norm": 0.1794494241476059,
      "learning_rate": 2.715132691008427e-05,
      "loss": 0.06553339958190918,
      "step": 1886
    },
    {
      "epoch": 0.2551163537424163,
      "grad_norm": 0.25364989042282104,
      "learning_rate": 2.714731270433574e-05,
      "loss": 0.07598268985748291,
      "step": 1887
    },
    {
      "epoch": 0.25525155053825227,
      "grad_norm": 0.08078136295080185,
      "learning_rate": 2.7143295969451933e-05,
      "loss": 0.05204680562019348,
      "step": 1888
    },
    {
      "epoch": 0.25538674733408817,
      "grad_norm": 0.1530763804912567,
      "learning_rate": 2.7139276706269147e-05,
      "loss": 0.08506584167480469,
      "step": 1889
    },
    {
      "epoch": 0.25552194412992413,
      "grad_norm": 0.06039128080010414,
      "learning_rate": 2.7135254915624213e-05,
      "loss": 0.06613683700561523,
      "step": 1890
    },
    {
      "epoch": 0.2556571409257601,
      "grad_norm": 0.11038939654827118,
      "learning_rate": 2.7131230598354497e-05,
      "loss": 0.11463785171508789,
      "step": 1891
    },
    {
      "epoch": 0.255792337721596,
      "grad_norm": 0.047232795506715775,
      "learning_rate": 2.712720375529787e-05,
      "loss": 0.04717755317687988,
      "step": 1892
    },
    {
      "epoch": 0.25592753451743194,
      "grad_norm": 0.058000437915325165,
      "learning_rate": 2.7123174387292758e-05,
      "loss": 0.05485057830810547,
      "step": 1893
    },
    {
      "epoch": 0.2560627313132679,
      "grad_norm": 0.1676839143037796,
      "learning_rate": 2.7119142495178088e-05,
      "loss": 0.06298685073852539,
      "step": 1894
    },
    {
      "epoch": 0.2561979281091038,
      "grad_norm": 0.13864421844482422,
      "learning_rate": 2.711510807979333e-05,
      "loss": 0.06916332244873047,
      "step": 1895
    },
    {
      "epoch": 0.25633312490493976,
      "grad_norm": 0.10523370653390884,
      "learning_rate": 2.7111071141978452e-05,
      "loss": 0.06897759437561035,
      "step": 1896
    },
    {
      "epoch": 0.2564683217007757,
      "grad_norm": 0.07564353197813034,
      "learning_rate": 2.7107031682573987e-05,
      "loss": 0.08072948455810547,
      "step": 1897
    },
    {
      "epoch": 0.2566035184966116,
      "grad_norm": 0.07633915543556213,
      "learning_rate": 2.710298970242096e-05,
      "loss": 0.08800125122070312,
      "step": 1898
    },
    {
      "epoch": 0.2567387152924476,
      "grad_norm": 0.13839472830295563,
      "learning_rate": 2.7098945202360937e-05,
      "loss": 0.11099767684936523,
      "step": 1899
    },
    {
      "epoch": 0.25687391208828353,
      "grad_norm": 0.17224784195423126,
      "learning_rate": 2.7094898183236e-05,
      "loss": 0.08221054077148438,
      "step": 1900
    },
    {
      "epoch": 0.25700910888411943,
      "grad_norm": 0.07429969310760498,
      "learning_rate": 2.709084864588877e-05,
      "loss": 0.05411934852600098,
      "step": 1901
    },
    {
      "epoch": 0.2571443056799554,
      "grad_norm": 0.12401656061410904,
      "learning_rate": 2.708679659116237e-05,
      "loss": 0.07936358451843262,
      "step": 1902
    },
    {
      "epoch": 0.25727950247579134,
      "grad_norm": 0.04505674168467522,
      "learning_rate": 2.708274201990047e-05,
      "loss": 0.035658836364746094,
      "step": 1903
    },
    {
      "epoch": 0.25741469927162725,
      "grad_norm": 0.06026989594101906,
      "learning_rate": 2.7078684932947247e-05,
      "loss": 0.06824731826782227,
      "step": 1904
    },
    {
      "epoch": 0.2575498960674632,
      "grad_norm": 0.0819808766245842,
      "learning_rate": 2.7074625331147407e-05,
      "loss": 0.0772542953491211,
      "step": 1905
    },
    {
      "epoch": 0.25768509286329916,
      "grad_norm": 0.07260015606880188,
      "learning_rate": 2.7070563215346184e-05,
      "loss": 0.04888296127319336,
      "step": 1906
    },
    {
      "epoch": 0.25782028965913506,
      "grad_norm": 0.13175158202648163,
      "learning_rate": 2.7066498586389332e-05,
      "loss": 0.0486409068107605,
      "step": 1907
    },
    {
      "epoch": 0.257955486454971,
      "grad_norm": 0.04404719918966293,
      "learning_rate": 2.7062431445123127e-05,
      "loss": 0.04677295684814453,
      "step": 1908
    },
    {
      "epoch": 0.258090683250807,
      "grad_norm": 0.07209489494562149,
      "learning_rate": 2.705836179239437e-05,
      "loss": 0.06144356727600098,
      "step": 1909
    },
    {
      "epoch": 0.2582258800466429,
      "grad_norm": 0.06582480669021606,
      "learning_rate": 2.705428962905039e-05,
      "loss": 0.06948471069335938,
      "step": 1910
    },
    {
      "epoch": 0.25836107684247883,
      "grad_norm": 0.05781620740890503,
      "learning_rate": 2.705021495593902e-05,
      "loss": 0.04116058349609375,
      "step": 1911
    },
    {
      "epoch": 0.2584962736383148,
      "grad_norm": 0.08257302641868591,
      "learning_rate": 2.704613777390864e-05,
      "loss": 0.06554484367370605,
      "step": 1912
    },
    {
      "epoch": 0.2586314704341507,
      "grad_norm": 0.07617247104644775,
      "learning_rate": 2.7042058083808135e-05,
      "loss": 0.08389806747436523,
      "step": 1913
    },
    {
      "epoch": 0.25876666722998665,
      "grad_norm": 0.11628446727991104,
      "learning_rate": 2.7037975886486928e-05,
      "loss": 0.08635663986206055,
      "step": 1914
    },
    {
      "epoch": 0.2589018640258226,
      "grad_norm": 0.08471303433179855,
      "learning_rate": 2.7033891182794942e-05,
      "loss": 0.07380294799804688,
      "step": 1915
    },
    {
      "epoch": 0.2590370608216585,
      "grad_norm": 0.08547315746545792,
      "learning_rate": 2.7029803973582642e-05,
      "loss": 0.050653934478759766,
      "step": 1916
    },
    {
      "epoch": 0.25917225761749446,
      "grad_norm": 0.055937085300683975,
      "learning_rate": 2.7025714259701e-05,
      "loss": 0.03486669063568115,
      "step": 1917
    },
    {
      "epoch": 0.2593074544133304,
      "grad_norm": 0.13840137422084808,
      "learning_rate": 2.7021622042001524e-05,
      "loss": 0.0895574688911438,
      "step": 1918
    },
    {
      "epoch": 0.2594426512091663,
      "grad_norm": 0.04367328807711601,
      "learning_rate": 2.701752732133623e-05,
      "loss": 0.04321324825286865,
      "step": 1919
    },
    {
      "epoch": 0.2595778480050023,
      "grad_norm": 0.15681852400302887,
      "learning_rate": 2.7013430098557664e-05,
      "loss": 0.08592557907104492,
      "step": 1920
    },
    {
      "epoch": 0.25971304480083823,
      "grad_norm": 0.1759447455406189,
      "learning_rate": 2.7009330374518885e-05,
      "loss": 0.07019585371017456,
      "step": 1921
    },
    {
      "epoch": 0.25984824159667413,
      "grad_norm": 0.1534302830696106,
      "learning_rate": 2.7005228150073483e-05,
      "loss": 0.06428885459899902,
      "step": 1922
    },
    {
      "epoch": 0.2599834383925101,
      "grad_norm": 0.21793781220912933,
      "learning_rate": 2.7001123426075558e-05,
      "loss": 0.11581873893737793,
      "step": 1923
    },
    {
      "epoch": 0.26011863518834605,
      "grad_norm": 0.09259770065546036,
      "learning_rate": 2.699701620337974e-05,
      "loss": 0.061139583587646484,
      "step": 1924
    },
    {
      "epoch": 0.26025383198418195,
      "grad_norm": 0.2145368456840515,
      "learning_rate": 2.699290648284117e-05,
      "loss": 0.05533432960510254,
      "step": 1925
    },
    {
      "epoch": 0.2603890287800179,
      "grad_norm": 0.13483993709087372,
      "learning_rate": 2.6988794265315522e-05,
      "loss": 0.06428718566894531,
      "step": 1926
    },
    {
      "epoch": 0.26052422557585386,
      "grad_norm": 0.15750481188297272,
      "learning_rate": 2.698467955165897e-05,
      "loss": 0.06235980987548828,
      "step": 1927
    },
    {
      "epoch": 0.26065942237168976,
      "grad_norm": 0.17170867323875427,
      "learning_rate": 2.6980562342728226e-05,
      "loss": 0.0690385103225708,
      "step": 1928
    },
    {
      "epoch": 0.2607946191675257,
      "grad_norm": 0.055610764771699905,
      "learning_rate": 2.6976442639380516e-05,
      "loss": 0.04764509201049805,
      "step": 1929
    },
    {
      "epoch": 0.2609298159633617,
      "grad_norm": 0.16008204221725464,
      "learning_rate": 2.6972320442473583e-05,
      "loss": 0.058400630950927734,
      "step": 1930
    },
    {
      "epoch": 0.26106501275919763,
      "grad_norm": 0.09385678172111511,
      "learning_rate": 2.6968195752865686e-05,
      "loss": 0.0539708137512207,
      "step": 1931
    },
    {
      "epoch": 0.26120020955503354,
      "grad_norm": 0.09797768294811249,
      "learning_rate": 2.6964068571415613e-05,
      "loss": 0.06435680389404297,
      "step": 1932
    },
    {
      "epoch": 0.2613354063508695,
      "grad_norm": 0.10132727026939392,
      "learning_rate": 2.6959938898982667e-05,
      "loss": 0.033181071281433105,
      "step": 1933
    },
    {
      "epoch": 0.26147060314670545,
      "grad_norm": 0.17216293513774872,
      "learning_rate": 2.6955806736426657e-05,
      "loss": 0.07015323638916016,
      "step": 1934
    },
    {
      "epoch": 0.26160579994254135,
      "grad_norm": 0.08743895590305328,
      "learning_rate": 2.6951672084607937e-05,
      "loss": 0.06944751739501953,
      "step": 1935
    },
    {
      "epoch": 0.2617409967383773,
      "grad_norm": 0.050038017332553864,
      "learning_rate": 2.694753494438735e-05,
      "loss": 0.05537307262420654,
      "step": 1936
    },
    {
      "epoch": 0.26187619353421326,
      "grad_norm": 0.14453940093517303,
      "learning_rate": 2.6943395316626272e-05,
      "loss": 0.07186472415924072,
      "step": 1937
    },
    {
      "epoch": 0.26201139033004917,
      "grad_norm": 0.12734849750995636,
      "learning_rate": 2.69392532021866e-05,
      "loss": 0.08086848258972168,
      "step": 1938
    },
    {
      "epoch": 0.2621465871258851,
      "grad_norm": 0.05402849242091179,
      "learning_rate": 2.693510860193075e-05,
      "loss": 0.03654217720031738,
      "step": 1939
    },
    {
      "epoch": 0.2622817839217211,
      "grad_norm": 0.07365751266479492,
      "learning_rate": 2.6930961516721638e-05,
      "loss": 0.05237889289855957,
      "step": 1940
    },
    {
      "epoch": 0.262416980717557,
      "grad_norm": 0.12777666747570038,
      "learning_rate": 2.6926811947422717e-05,
      "loss": 0.05820131301879883,
      "step": 1941
    },
    {
      "epoch": 0.26255217751339294,
      "grad_norm": 0.13499997556209564,
      "learning_rate": 2.6922659894897946e-05,
      "loss": 0.06490612030029297,
      "step": 1942
    },
    {
      "epoch": 0.2626873743092289,
      "grad_norm": 0.1302352249622345,
      "learning_rate": 2.6918505360011805e-05,
      "loss": 0.09477472305297852,
      "step": 1943
    },
    {
      "epoch": 0.2628225711050648,
      "grad_norm": 0.13706472516059875,
      "learning_rate": 2.6914348343629292e-05,
      "loss": 0.07878828048706055,
      "step": 1944
    },
    {
      "epoch": 0.26295776790090075,
      "grad_norm": 0.0754261240363121,
      "learning_rate": 2.6910188846615918e-05,
      "loss": 0.05060696601867676,
      "step": 1945
    },
    {
      "epoch": 0.2630929646967367,
      "grad_norm": 0.1018480509519577,
      "learning_rate": 2.6906026869837714e-05,
      "loss": 0.07848453521728516,
      "step": 1946
    },
    {
      "epoch": 0.2632281614925726,
      "grad_norm": 0.12182067334651947,
      "learning_rate": 2.6901862414161222e-05,
      "loss": 0.06874632835388184,
      "step": 1947
    },
    {
      "epoch": 0.26336335828840857,
      "grad_norm": 0.1469702124595642,
      "learning_rate": 2.689769548045351e-05,
      "loss": 0.04393577575683594,
      "step": 1948
    },
    {
      "epoch": 0.2634985550842445,
      "grad_norm": 0.29771187901496887,
      "learning_rate": 2.6893526069582154e-05,
      "loss": 0.08208227157592773,
      "step": 1949
    },
    {
      "epoch": 0.2636337518800804,
      "grad_norm": 0.15200366079807281,
      "learning_rate": 2.6889354182415245e-05,
      "loss": 0.07265377044677734,
      "step": 1950
    },
    {
      "epoch": 0.2637689486759164,
      "grad_norm": 0.19280299544334412,
      "learning_rate": 2.688517981982139e-05,
      "loss": 0.066864013671875,
      "step": 1951
    },
    {
      "epoch": 0.26390414547175234,
      "grad_norm": 0.08240275830030441,
      "learning_rate": 2.6881002982669723e-05,
      "loss": 0.10303783416748047,
      "step": 1952
    },
    {
      "epoch": 0.26403934226758824,
      "grad_norm": 0.08139554411172867,
      "learning_rate": 2.6876823671829874e-05,
      "loss": 0.06015300750732422,
      "step": 1953
    },
    {
      "epoch": 0.2641745390634242,
      "grad_norm": 0.2022184580564499,
      "learning_rate": 2.6872641888172e-05,
      "loss": 0.07303333282470703,
      "step": 1954
    },
    {
      "epoch": 0.26430973585926015,
      "grad_norm": 0.0518563911318779,
      "learning_rate": 2.6868457632566774e-05,
      "loss": 0.04127919673919678,
      "step": 1955
    },
    {
      "epoch": 0.26444493265509605,
      "grad_norm": 0.14683027565479279,
      "learning_rate": 2.6864270905885377e-05,
      "loss": 0.06519961357116699,
      "step": 1956
    },
    {
      "epoch": 0.264580129450932,
      "grad_norm": 0.1502714604139328,
      "learning_rate": 2.6860081708999515e-05,
      "loss": 0.06728219985961914,
      "step": 1957
    },
    {
      "epoch": 0.26471532624676797,
      "grad_norm": 0.09536710381507874,
      "learning_rate": 2.685589004278139e-05,
      "loss": 0.06737709045410156,
      "step": 1958
    },
    {
      "epoch": 0.26485052304260387,
      "grad_norm": 0.19764938950538635,
      "learning_rate": 2.6851695908103737e-05,
      "loss": 0.0974421501159668,
      "step": 1959
    },
    {
      "epoch": 0.2649857198384398,
      "grad_norm": 0.16358472406864166,
      "learning_rate": 2.6847499305839796e-05,
      "loss": 0.07037496566772461,
      "step": 1960
    },
    {
      "epoch": 0.2651209166342758,
      "grad_norm": 0.04844284802675247,
      "learning_rate": 2.684330023686332e-05,
      "loss": 0.048337459564208984,
      "step": 1961
    },
    {
      "epoch": 0.2652561134301117,
      "grad_norm": 0.029073940590023994,
      "learning_rate": 2.6839098702048577e-05,
      "loss": 0.020238637924194336,
      "step": 1962
    },
    {
      "epoch": 0.26539131022594764,
      "grad_norm": 0.10952551662921906,
      "learning_rate": 2.683489470227035e-05,
      "loss": 0.07552814483642578,
      "step": 1963
    },
    {
      "epoch": 0.2655265070217836,
      "grad_norm": 0.1424301117658615,
      "learning_rate": 2.6830688238403936e-05,
      "loss": 0.07334160804748535,
      "step": 1964
    },
    {
      "epoch": 0.2656617038176195,
      "grad_norm": 0.10241356492042542,
      "learning_rate": 2.682647931132514e-05,
      "loss": 0.05843830108642578,
      "step": 1965
    },
    {
      "epoch": 0.26579690061345546,
      "grad_norm": 0.037290506064891815,
      "learning_rate": 2.682226792191029e-05,
      "loss": 0.03623199462890625,
      "step": 1966
    },
    {
      "epoch": 0.2659320974092914,
      "grad_norm": 0.127027690410614,
      "learning_rate": 2.681805407103621e-05,
      "loss": 0.046491026878356934,
      "step": 1967
    },
    {
      "epoch": 0.2660672942051273,
      "grad_norm": 0.07609870284795761,
      "learning_rate": 2.6813837759580253e-05,
      "loss": 0.08343672752380371,
      "step": 1968
    },
    {
      "epoch": 0.26620249100096327,
      "grad_norm": 0.03872406855225563,
      "learning_rate": 2.6809618988420274e-05,
      "loss": 0.04909372329711914,
      "step": 1969
    },
    {
      "epoch": 0.2663376877967992,
      "grad_norm": 0.10477709770202637,
      "learning_rate": 2.6805397758434647e-05,
      "loss": 0.07259082794189453,
      "step": 1970
    },
    {
      "epoch": 0.26647288459263513,
      "grad_norm": 0.06550008058547974,
      "learning_rate": 2.6801174070502248e-05,
      "loss": 0.05088019371032715,
      "step": 1971
    },
    {
      "epoch": 0.2666080813884711,
      "grad_norm": 0.1912948042154312,
      "learning_rate": 2.679694792550248e-05,
      "loss": 0.0713052749633789,
      "step": 1972
    },
    {
      "epoch": 0.26674327818430704,
      "grad_norm": 0.06793061643838882,
      "learning_rate": 2.6792719324315248e-05,
      "loss": 0.06447601318359375,
      "step": 1973
    },
    {
      "epoch": 0.26687847498014294,
      "grad_norm": 0.11101162433624268,
      "learning_rate": 2.678848826782096e-05,
      "loss": 0.07129859924316406,
      "step": 1974
    },
    {
      "epoch": 0.2670136717759789,
      "grad_norm": 0.08733350038528442,
      "learning_rate": 2.678425475690055e-05,
      "loss": 0.05308079719543457,
      "step": 1975
    },
    {
      "epoch": 0.26714886857181486,
      "grad_norm": 0.1475624442100525,
      "learning_rate": 2.6780018792435464e-05,
      "loss": 0.06163835525512695,
      "step": 1976
    },
    {
      "epoch": 0.2672840653676508,
      "grad_norm": 0.061899274587631226,
      "learning_rate": 2.6775780375307645e-05,
      "loss": 0.0669260025024414,
      "step": 1977
    },
    {
      "epoch": 0.2674192621634867,
      "grad_norm": 0.08895330876111984,
      "learning_rate": 2.6771539506399555e-05,
      "loss": 0.062021493911743164,
      "step": 1978
    },
    {
      "epoch": 0.26755445895932267,
      "grad_norm": 0.07559031248092651,
      "learning_rate": 2.6767296186594165e-05,
      "loss": 0.06453144550323486,
      "step": 1979
    },
    {
      "epoch": 0.26768965575515863,
      "grad_norm": 0.0834704115986824,
      "learning_rate": 2.676305041677496e-05,
      "loss": 0.0607830286026001,
      "step": 1980
    },
    {
      "epoch": 0.26782485255099453,
      "grad_norm": 0.050367336720228195,
      "learning_rate": 2.675880219782593e-05,
      "loss": 0.031048297882080078,
      "step": 1981
    },
    {
      "epoch": 0.2679600493468305,
      "grad_norm": 0.045134637504816055,
      "learning_rate": 2.6754551530631575e-05,
      "loss": 0.05095529556274414,
      "step": 1982
    },
    {
      "epoch": 0.26809524614266644,
      "grad_norm": 0.11680918186903,
      "learning_rate": 2.6750298416076907e-05,
      "loss": 0.06729412078857422,
      "step": 1983
    },
    {
      "epoch": 0.26823044293850234,
      "grad_norm": 0.14327314496040344,
      "learning_rate": 2.674604285504745e-05,
      "loss": 0.10972738265991211,
      "step": 1984
    },
    {
      "epoch": 0.2683656397343383,
      "grad_norm": 0.05404512584209442,
      "learning_rate": 2.6741784848429235e-05,
      "loss": 0.043704986572265625,
      "step": 1985
    },
    {
      "epoch": 0.26850083653017426,
      "grad_norm": 0.1642342209815979,
      "learning_rate": 2.67375243971088e-05,
      "loss": 0.0786275863647461,
      "step": 1986
    },
    {
      "epoch": 0.26863603332601016,
      "grad_norm": 0.1660015881061554,
      "learning_rate": 2.6733261501973192e-05,
      "loss": 0.06278562545776367,
      "step": 1987
    },
    {
      "epoch": 0.2687712301218461,
      "grad_norm": 0.164725661277771,
      "learning_rate": 2.672899616390997e-05,
      "loss": 0.09509825706481934,
      "step": 1988
    },
    {
      "epoch": 0.2689064269176821,
      "grad_norm": 0.07612603902816772,
      "learning_rate": 2.67247283838072e-05,
      "loss": 0.07907819747924805,
      "step": 1989
    },
    {
      "epoch": 0.269041623713518,
      "grad_norm": 0.13096463680267334,
      "learning_rate": 2.6720458162553457e-05,
      "loss": 0.08989977836608887,
      "step": 1990
    },
    {
      "epoch": 0.26917682050935393,
      "grad_norm": 0.09332142025232315,
      "learning_rate": 2.6716185501037822e-05,
      "loss": 0.04581499099731445,
      "step": 1991
    },
    {
      "epoch": 0.2693120173051899,
      "grad_norm": 0.12690138816833496,
      "learning_rate": 2.671191040014989e-05,
      "loss": 0.1052560806274414,
      "step": 1992
    },
    {
      "epoch": 0.2694472141010258,
      "grad_norm": 0.0809553861618042,
      "learning_rate": 2.6707632860779756e-05,
      "loss": 0.06932902336120605,
      "step": 1993
    },
    {
      "epoch": 0.26958241089686175,
      "grad_norm": 0.07745625078678131,
      "learning_rate": 2.6703352883818024e-05,
      "loss": 0.050919532775878906,
      "step": 1994
    },
    {
      "epoch": 0.2697176076926977,
      "grad_norm": 0.18288680911064148,
      "learning_rate": 2.6699070470155816e-05,
      "loss": 0.07959508895874023,
      "step": 1995
    },
    {
      "epoch": 0.2698528044885336,
      "grad_norm": 0.10290535539388657,
      "learning_rate": 2.669478562068475e-05,
      "loss": 0.06593942642211914,
      "step": 1996
    },
    {
      "epoch": 0.26998800128436956,
      "grad_norm": 0.10326408594846725,
      "learning_rate": 2.6690498336296955e-05,
      "loss": 0.06656599044799805,
      "step": 1997
    },
    {
      "epoch": 0.2701231980802055,
      "grad_norm": 0.2663053572177887,
      "learning_rate": 2.6686208617885057e-05,
      "loss": 0.06684589385986328,
      "step": 1998
    },
    {
      "epoch": 0.2702583948760414,
      "grad_norm": 0.100665383040905,
      "learning_rate": 2.668191646634221e-05,
      "loss": 0.08602237701416016,
      "step": 1999
    },
    {
      "epoch": 0.2703935916718774,
      "grad_norm": 0.10618358105421066,
      "learning_rate": 2.667762188256206e-05,
      "loss": 0.052654385566711426,
      "step": 2000
    },
    {
      "epoch": 0.27052878846771333,
      "grad_norm": 0.2265174388885498,
      "learning_rate": 2.6673324867438764e-05,
      "loss": 0.08537483215332031,
      "step": 2001
    },
    {
      "epoch": 0.27066398526354923,
      "grad_norm": 0.12552383542060852,
      "learning_rate": 2.666902542186698e-05,
      "loss": 0.05460166931152344,
      "step": 2002
    },
    {
      "epoch": 0.2707991820593852,
      "grad_norm": 0.2202092409133911,
      "learning_rate": 2.666472354674187e-05,
      "loss": 0.08910870552062988,
      "step": 2003
    },
    {
      "epoch": 0.27093437885522115,
      "grad_norm": 0.03565860167145729,
      "learning_rate": 2.666041924295912e-05,
      "loss": 0.03602790832519531,
      "step": 2004
    },
    {
      "epoch": 0.27106957565105705,
      "grad_norm": 0.03264477103948593,
      "learning_rate": 2.6656112511414902e-05,
      "loss": 0.040204644203186035,
      "step": 2005
    },
    {
      "epoch": 0.271204772446893,
      "grad_norm": 0.0596921443939209,
      "learning_rate": 2.6651803353005896e-05,
      "loss": 0.05192756652832031,
      "step": 2006
    },
    {
      "epoch": 0.27133996924272896,
      "grad_norm": 0.08647216856479645,
      "learning_rate": 2.66474917686293e-05,
      "loss": 0.050931692123413086,
      "step": 2007
    },
    {
      "epoch": 0.27147516603856486,
      "grad_norm": 0.15114521980285645,
      "learning_rate": 2.664317775918281e-05,
      "loss": 0.06802058219909668,
      "step": 2008
    },
    {
      "epoch": 0.2716103628344008,
      "grad_norm": 0.3004302382469177,
      "learning_rate": 2.6638861325564615e-05,
      "loss": 0.0746757984161377,
      "step": 2009
    },
    {
      "epoch": 0.2717455596302368,
      "grad_norm": 0.11647263914346695,
      "learning_rate": 2.6634542468673432e-05,
      "loss": 0.04262232780456543,
      "step": 2010
    },
    {
      "epoch": 0.2718807564260727,
      "grad_norm": 0.10886501520872116,
      "learning_rate": 2.663022118940846e-05,
      "loss": 0.04444265365600586,
      "step": 2011
    },
    {
      "epoch": 0.27201595322190864,
      "grad_norm": 0.17771105468273163,
      "learning_rate": 2.662589748866942e-05,
      "loss": 0.05081295967102051,
      "step": 2012
    },
    {
      "epoch": 0.2721511500177446,
      "grad_norm": 0.09182816743850708,
      "learning_rate": 2.6621571367356522e-05,
      "loss": 0.03714632987976074,
      "step": 2013
    },
    {
      "epoch": 0.2722863468135805,
      "grad_norm": 0.1519690454006195,
      "learning_rate": 2.6617242826370495e-05,
      "loss": 0.08922791481018066,
      "step": 2014
    },
    {
      "epoch": 0.27242154360941645,
      "grad_norm": 0.08940640836954117,
      "learning_rate": 2.661291186661256e-05,
      "loss": 0.08684539794921875,
      "step": 2015
    },
    {
      "epoch": 0.2725567404052524,
      "grad_norm": 0.0886605754494667,
      "learning_rate": 2.6608578488984444e-05,
      "loss": 0.07495689392089844,
      "step": 2016
    },
    {
      "epoch": 0.2726919372010883,
      "grad_norm": 0.08003921806812286,
      "learning_rate": 2.6604242694388388e-05,
      "loss": 0.04020333290100098,
      "step": 2017
    },
    {
      "epoch": 0.27282713399692426,
      "grad_norm": 0.08868551254272461,
      "learning_rate": 2.6599904483727116e-05,
      "loss": 0.09149885177612305,
      "step": 2018
    },
    {
      "epoch": 0.2729623307927602,
      "grad_norm": 0.07572377473115921,
      "learning_rate": 2.6595563857903872e-05,
      "loss": 0.05034780502319336,
      "step": 2019
    },
    {
      "epoch": 0.2730975275885961,
      "grad_norm": 0.07474635541439056,
      "learning_rate": 2.6591220817822405e-05,
      "loss": 0.06969404220581055,
      "step": 2020
    },
    {
      "epoch": 0.2732327243844321,
      "grad_norm": 0.03147522360086441,
      "learning_rate": 2.658687536438694e-05,
      "loss": 0.03863334655761719,
      "step": 2021
    },
    {
      "epoch": 0.27336792118026804,
      "grad_norm": 0.13000202178955078,
      "learning_rate": 2.6582527498502243e-05,
      "loss": 0.06570768356323242,
      "step": 2022
    },
    {
      "epoch": 0.273503117976104,
      "grad_norm": 0.07879548519849777,
      "learning_rate": 2.6578177221073556e-05,
      "loss": 0.0561366081237793,
      "step": 2023
    },
    {
      "epoch": 0.2736383147719399,
      "grad_norm": 0.04455554485321045,
      "learning_rate": 2.6573824533006628e-05,
      "loss": 0.028705358505249023,
      "step": 2024
    },
    {
      "epoch": 0.27377351156777585,
      "grad_norm": 0.1452604979276657,
      "learning_rate": 2.6569469435207712e-05,
      "loss": 0.07401388883590698,
      "step": 2025
    },
    {
      "epoch": 0.2739087083636118,
      "grad_norm": 0.14333850145339966,
      "learning_rate": 2.656511192858356e-05,
      "loss": 0.05373525619506836,
      "step": 2026
    },
    {
      "epoch": 0.2740439051594477,
      "grad_norm": 0.07900907844305038,
      "learning_rate": 2.6560752014041438e-05,
      "loss": 0.051888227462768555,
      "step": 2027
    },
    {
      "epoch": 0.27417910195528367,
      "grad_norm": 0.07668299973011017,
      "learning_rate": 2.6556389692489098e-05,
      "loss": 0.07629776000976562,
      "step": 2028
    },
    {
      "epoch": 0.2743142987511196,
      "grad_norm": 0.09356392920017242,
      "learning_rate": 2.6552024964834795e-05,
      "loss": 0.05437636375427246,
      "step": 2029
    },
    {
      "epoch": 0.2744494955469555,
      "grad_norm": 0.10345875471830368,
      "learning_rate": 2.6547657831987286e-05,
      "loss": 0.07483243942260742,
      "step": 2030
    },
    {
      "epoch": 0.2745846923427915,
      "grad_norm": 0.09706155210733414,
      "learning_rate": 2.6543288294855843e-05,
      "loss": 0.02883601188659668,
      "step": 2031
    },
    {
      "epoch": 0.27471988913862744,
      "grad_norm": 0.09148816019296646,
      "learning_rate": 2.653891635435022e-05,
      "loss": 0.06745243072509766,
      "step": 2032
    },
    {
      "epoch": 0.27485508593446334,
      "grad_norm": 0.1024070531129837,
      "learning_rate": 2.653454201138068e-05,
      "loss": 0.10385990142822266,
      "step": 2033
    },
    {
      "epoch": 0.2749902827302993,
      "grad_norm": 0.17602460086345673,
      "learning_rate": 2.653016526685798e-05,
      "loss": 0.07130289077758789,
      "step": 2034
    },
    {
      "epoch": 0.27512547952613525,
      "grad_norm": 0.16487474739551544,
      "learning_rate": 2.6525786121693387e-05,
      "loss": 0.08155179023742676,
      "step": 2035
    },
    {
      "epoch": 0.27526067632197115,
      "grad_norm": 0.09475169330835342,
      "learning_rate": 2.652140457679866e-05,
      "loss": 0.05621039867401123,
      "step": 2036
    },
    {
      "epoch": 0.2753958731178071,
      "grad_norm": 0.08501619100570679,
      "learning_rate": 2.6517020633086064e-05,
      "loss": 0.06403207778930664,
      "step": 2037
    },
    {
      "epoch": 0.27553106991364307,
      "grad_norm": 0.2924579381942749,
      "learning_rate": 2.6512634291468354e-05,
      "loss": 0.09018754959106445,
      "step": 2038
    },
    {
      "epoch": 0.27566626670947897,
      "grad_norm": 0.07076648622751236,
      "learning_rate": 2.6508245552858792e-05,
      "loss": 0.06429815292358398,
      "step": 2039
    },
    {
      "epoch": 0.2758014635053149,
      "grad_norm": 0.06613663583993912,
      "learning_rate": 2.6503854418171133e-05,
      "loss": 0.057242393493652344,
      "step": 2040
    },
    {
      "epoch": 0.2759366603011509,
      "grad_norm": 0.23619668185710907,
      "learning_rate": 2.6499460888319644e-05,
      "loss": 0.09474372863769531,
      "step": 2041
    },
    {
      "epoch": 0.2760718570969868,
      "grad_norm": 0.1373104751110077,
      "learning_rate": 2.6495064964219073e-05,
      "loss": 0.06896734237670898,
      "step": 2042
    },
    {
      "epoch": 0.27620705389282274,
      "grad_norm": 0.10663674026727676,
      "learning_rate": 2.649066664678467e-05,
      "loss": 0.08289837837219238,
      "step": 2043
    },
    {
      "epoch": 0.2763422506886587,
      "grad_norm": 0.13968206942081451,
      "learning_rate": 2.6486265936932205e-05,
      "loss": 0.08006739616394043,
      "step": 2044
    },
    {
      "epoch": 0.2764774474844946,
      "grad_norm": 0.1566438525915146,
      "learning_rate": 2.6481862835577915e-05,
      "loss": 0.06683707237243652,
      "step": 2045
    },
    {
      "epoch": 0.27661264428033056,
      "grad_norm": 0.09505102038383484,
      "learning_rate": 2.6477457343638557e-05,
      "loss": 0.059329986572265625,
      "step": 2046
    },
    {
      "epoch": 0.2767478410761665,
      "grad_norm": 0.06063215434551239,
      "learning_rate": 2.647304946203137e-05,
      "loss": 0.06640112400054932,
      "step": 2047
    },
    {
      "epoch": 0.2768830378720024,
      "grad_norm": 0.06275895982980728,
      "learning_rate": 2.6468639191674106e-05,
      "loss": 0.03875541687011719,
      "step": 2048
    },
    {
      "epoch": 0.27701823466783837,
      "grad_norm": 0.1204524114727974,
      "learning_rate": 2.6464226533485007e-05,
      "loss": 0.0769195556640625,
      "step": 2049
    },
    {
      "epoch": 0.2771534314636743,
      "grad_norm": 0.1370803713798523,
      "learning_rate": 2.6459811488382806e-05,
      "loss": 0.05951547622680664,
      "step": 2050
    },
    {
      "epoch": 0.27728862825951023,
      "grad_norm": 0.21443584561347961,
      "learning_rate": 2.645539405728674e-05,
      "loss": 0.06662940979003906,
      "step": 2051
    },
    {
      "epoch": 0.2774238250553462,
      "grad_norm": 0.08277695626020432,
      "learning_rate": 2.6450974241116545e-05,
      "loss": 0.04816639423370361,
      "step": 2052
    },
    {
      "epoch": 0.27755902185118214,
      "grad_norm": 0.09988467395305634,
      "learning_rate": 2.644655204079245e-05,
      "loss": 0.09002542495727539,
      "step": 2053
    },
    {
      "epoch": 0.27769421864701804,
      "grad_norm": 0.05024226754903793,
      "learning_rate": 2.6442127457235177e-05,
      "loss": 0.05195474624633789,
      "step": 2054
    },
    {
      "epoch": 0.277829415442854,
      "grad_norm": 0.12617287039756775,
      "learning_rate": 2.6437700491365957e-05,
      "loss": 0.05921363830566406,
      "step": 2055
    },
    {
      "epoch": 0.27796461223868996,
      "grad_norm": 0.08306661993265152,
      "learning_rate": 2.6433271144106495e-05,
      "loss": 0.04903411865234375,
      "step": 2056
    },
    {
      "epoch": 0.27809980903452586,
      "grad_norm": 0.1661880761384964,
      "learning_rate": 2.6428839416379015e-05,
      "loss": 0.09780478477478027,
      "step": 2057
    },
    {
      "epoch": 0.2782350058303618,
      "grad_norm": 0.057911623269319534,
      "learning_rate": 2.642440530910622e-05,
      "loss": 0.06536006927490234,
      "step": 2058
    },
    {
      "epoch": 0.27837020262619777,
      "grad_norm": 0.1030687466263771,
      "learning_rate": 2.6419968823211318e-05,
      "loss": 0.06716156005859375,
      "step": 2059
    },
    {
      "epoch": 0.2785053994220337,
      "grad_norm": 0.11096462607383728,
      "learning_rate": 2.641552995961801e-05,
      "loss": 0.05932283401489258,
      "step": 2060
    },
    {
      "epoch": 0.27864059621786963,
      "grad_norm": 0.06314587593078613,
      "learning_rate": 2.6411088719250484e-05,
      "loss": 0.049739837646484375,
      "step": 2061
    },
    {
      "epoch": 0.2787757930137056,
      "grad_norm": 0.12002673745155334,
      "learning_rate": 2.6406645103033442e-05,
      "loss": 0.0627899169921875,
      "step": 2062
    },
    {
      "epoch": 0.2789109898095415,
      "grad_norm": 0.1444939821958542,
      "learning_rate": 2.640219911189206e-05,
      "loss": 0.0700833797454834,
      "step": 2063
    },
    {
      "epoch": 0.27904618660537744,
      "grad_norm": 0.07030951231718063,
      "learning_rate": 2.6397750746752015e-05,
      "loss": 0.07575082778930664,
      "step": 2064
    },
    {
      "epoch": 0.2791813834012134,
      "grad_norm": 0.049706388264894485,
      "learning_rate": 2.6393300008539488e-05,
      "loss": 0.050824880599975586,
      "step": 2065
    },
    {
      "epoch": 0.2793165801970493,
      "grad_norm": 0.08290249854326248,
      "learning_rate": 2.6388846898181143e-05,
      "loss": 0.0519561767578125,
      "step": 2066
    },
    {
      "epoch": 0.27945177699288526,
      "grad_norm": 0.11108475923538208,
      "learning_rate": 2.6384391416604142e-05,
      "loss": 0.07414007186889648,
      "step": 2067
    },
    {
      "epoch": 0.2795869737887212,
      "grad_norm": 0.07047361135482788,
      "learning_rate": 2.6379933564736136e-05,
      "loss": 0.07421302795410156,
      "step": 2068
    },
    {
      "epoch": 0.2797221705845572,
      "grad_norm": 0.10987763106822968,
      "learning_rate": 2.637547334350528e-05,
      "loss": 0.08148670196533203,
      "step": 2069
    },
    {
      "epoch": 0.2798573673803931,
      "grad_norm": 0.06859488040208817,
      "learning_rate": 2.637101075384021e-05,
      "loss": 0.05065727233886719,
      "step": 2070
    },
    {
      "epoch": 0.27999256417622903,
      "grad_norm": 0.12862250208854675,
      "learning_rate": 2.636654579667006e-05,
      "loss": 0.06699180603027344,
      "step": 2071
    },
    {
      "epoch": 0.280127760972065,
      "grad_norm": 0.1253284066915512,
      "learning_rate": 2.6362078472924467e-05,
      "loss": 0.0641794204711914,
      "step": 2072
    },
    {
      "epoch": 0.2802629577679009,
      "grad_norm": 0.19095271825790405,
      "learning_rate": 2.6357608783533545e-05,
      "loss": 0.06796979904174805,
      "step": 2073
    },
    {
      "epoch": 0.28039815456373685,
      "grad_norm": 0.07041297107934952,
      "learning_rate": 2.6353136729427907e-05,
      "loss": 0.05645179748535156,
      "step": 2074
    },
    {
      "epoch": 0.2805333513595728,
      "grad_norm": 0.07707662135362625,
      "learning_rate": 2.6348662311538657e-05,
      "loss": 0.05318903923034668,
      "step": 2075
    },
    {
      "epoch": 0.2806685481554087,
      "grad_norm": 0.0824110209941864,
      "learning_rate": 2.6344185530797398e-05,
      "loss": 0.09498566389083862,
      "step": 2076
    },
    {
      "epoch": 0.28080374495124466,
      "grad_norm": 0.06920803338289261,
      "learning_rate": 2.633970638813622e-05,
      "loss": 0.06720471382141113,
      "step": 2077
    },
    {
      "epoch": 0.2809389417470806,
      "grad_norm": 0.1455577164888382,
      "learning_rate": 2.6335224884487698e-05,
      "loss": 0.07756519317626953,
      "step": 2078
    },
    {
      "epoch": 0.2810741385429165,
      "grad_norm": 0.08363092690706253,
      "learning_rate": 2.6330741020784905e-05,
      "loss": 0.07347440719604492,
      "step": 2079
    },
    {
      "epoch": 0.2812093353387525,
      "grad_norm": 0.06588733941316605,
      "learning_rate": 2.6326254797961415e-05,
      "loss": 0.03544259071350098,
      "step": 2080
    },
    {
      "epoch": 0.28134453213458843,
      "grad_norm": 0.0798463225364685,
      "learning_rate": 2.6321766216951273e-05,
      "loss": 0.03764486312866211,
      "step": 2081
    },
    {
      "epoch": 0.28147972893042433,
      "grad_norm": 0.09690330177545547,
      "learning_rate": 2.631727527868903e-05,
      "loss": 0.062256574630737305,
      "step": 2082
    },
    {
      "epoch": 0.2816149257262603,
      "grad_norm": 0.0920526310801506,
      "learning_rate": 2.6312781984109727e-05,
      "loss": 0.07936477661132812,
      "step": 2083
    },
    {
      "epoch": 0.28175012252209625,
      "grad_norm": 0.05120435729622841,
      "learning_rate": 2.6308286334148882e-05,
      "loss": 0.06283760070800781,
      "step": 2084
    },
    {
      "epoch": 0.28188531931793215,
      "grad_norm": 0.05456714332103729,
      "learning_rate": 2.630378832974252e-05,
      "loss": 0.054611384868621826,
      "step": 2085
    },
    {
      "epoch": 0.2820205161137681,
      "grad_norm": 0.037975460290908813,
      "learning_rate": 2.6299287971827154e-05,
      "loss": 0.04564058780670166,
      "step": 2086
    },
    {
      "epoch": 0.28215571290960406,
      "grad_norm": 0.22946839034557343,
      "learning_rate": 2.629478526133977e-05,
      "loss": 0.08869409561157227,
      "step": 2087
    },
    {
      "epoch": 0.28229090970543996,
      "grad_norm": 0.1571892946958542,
      "learning_rate": 2.6290280199217867e-05,
      "loss": 0.07628417015075684,
      "step": 2088
    },
    {
      "epoch": 0.2824261065012759,
      "grad_norm": 0.07472819089889526,
      "learning_rate": 2.6285772786399424e-05,
      "loss": 0.04982256889343262,
      "step": 2089
    },
    {
      "epoch": 0.2825613032971119,
      "grad_norm": 0.2132357954978943,
      "learning_rate": 2.6281263023822894e-05,
      "loss": 0.07956933975219727,
      "step": 2090
    },
    {
      "epoch": 0.2826965000929478,
      "grad_norm": 0.18320582807064056,
      "learning_rate": 2.627675091242725e-05,
      "loss": 0.06894779205322266,
      "step": 2091
    },
    {
      "epoch": 0.28283169688878373,
      "grad_norm": 0.08101629465818405,
      "learning_rate": 2.627223645315193e-05,
      "loss": 0.03517794609069824,
      "step": 2092
    },
    {
      "epoch": 0.2829668936846197,
      "grad_norm": 0.13307827711105347,
      "learning_rate": 2.6267719646936868e-05,
      "loss": 0.057976484298706055,
      "step": 2093
    },
    {
      "epoch": 0.2831020904804556,
      "grad_norm": 0.2228143811225891,
      "learning_rate": 2.626320049472249e-05,
      "loss": 0.06149768829345703,
      "step": 2094
    },
    {
      "epoch": 0.28323728727629155,
      "grad_norm": 0.21607151627540588,
      "learning_rate": 2.6258678997449705e-05,
      "loss": 0.05923795700073242,
      "step": 2095
    },
    {
      "epoch": 0.2833724840721275,
      "grad_norm": 0.28659555315971375,
      "learning_rate": 2.6254155156059912e-05,
      "loss": 0.04944181442260742,
      "step": 2096
    },
    {
      "epoch": 0.2835076808679634,
      "grad_norm": 0.2129606157541275,
      "learning_rate": 2.6249628971495006e-05,
      "loss": 0.05794954299926758,
      "step": 2097
    },
    {
      "epoch": 0.28364287766379936,
      "grad_norm": 0.12535381317138672,
      "learning_rate": 2.6245100444697353e-05,
      "loss": 0.06516098976135254,
      "step": 2098
    },
    {
      "epoch": 0.2837780744596353,
      "grad_norm": 0.06816662847995758,
      "learning_rate": 2.6240569576609824e-05,
      "loss": 0.036916375160217285,
      "step": 2099
    },
    {
      "epoch": 0.2839132712554712,
      "grad_norm": 0.07636474072933197,
      "learning_rate": 2.623603636817577e-05,
      "loss": 0.05437588691711426,
      "step": 2100
    },
    {
      "epoch": 0.2840484680513072,
      "grad_norm": 0.22799153625965118,
      "learning_rate": 2.6231500820339024e-05,
      "loss": 0.05071878433227539,
      "step": 2101
    },
    {
      "epoch": 0.28418366484714314,
      "grad_norm": 0.17223896086215973,
      "learning_rate": 2.6226962934043913e-05,
      "loss": 0.046317100524902344,
      "step": 2102
    },
    {
      "epoch": 0.28431886164297904,
      "grad_norm": 0.21222834289073944,
      "learning_rate": 2.622242271023525e-05,
      "loss": 0.041506290435791016,
      "step": 2103
    },
    {
      "epoch": 0.284454058438815,
      "grad_norm": 0.10336212813854218,
      "learning_rate": 2.6217880149858333e-05,
      "loss": 0.07502317428588867,
      "step": 2104
    },
    {
      "epoch": 0.28458925523465095,
      "grad_norm": 0.17416946589946747,
      "learning_rate": 2.621333525385895e-05,
      "loss": 0.08292198181152344,
      "step": 2105
    },
    {
      "epoch": 0.28472445203048685,
      "grad_norm": 0.04412782937288284,
      "learning_rate": 2.6208788023183366e-05,
      "loss": 0.035071492195129395,
      "step": 2106
    },
    {
      "epoch": 0.2848596488263228,
      "grad_norm": 0.26852941513061523,
      "learning_rate": 2.6204238458778346e-05,
      "loss": 0.09268230199813843,
      "step": 2107
    },
    {
      "epoch": 0.28499484562215877,
      "grad_norm": 0.08535538613796234,
      "learning_rate": 2.619968656159113e-05,
      "loss": 0.05893373489379883,
      "step": 2108
    },
    {
      "epoch": 0.28513004241799467,
      "grad_norm": 0.12854574620723724,
      "learning_rate": 2.6195132332569445e-05,
      "loss": 0.04909944534301758,
      "step": 2109
    },
    {
      "epoch": 0.2852652392138306,
      "grad_norm": 0.12075182795524597,
      "learning_rate": 2.619057577266151e-05,
      "loss": 0.0672757625579834,
      "step": 2110
    },
    {
      "epoch": 0.2854004360096666,
      "grad_norm": 0.09140174090862274,
      "learning_rate": 2.6186016882816027e-05,
      "loss": 0.09554576873779297,
      "step": 2111
    },
    {
      "epoch": 0.28553563280550254,
      "grad_norm": 0.1659955084323883,
      "learning_rate": 2.6181455663982175e-05,
      "loss": 0.09116029739379883,
      "step": 2112
    },
    {
      "epoch": 0.28567082960133844,
      "grad_norm": 0.1251361072063446,
      "learning_rate": 2.6176892117109628e-05,
      "loss": 0.09169983863830566,
      "step": 2113
    },
    {
      "epoch": 0.2858060263971744,
      "grad_norm": 0.06685018539428711,
      "learning_rate": 2.617232624314854e-05,
      "loss": 0.05655074119567871,
      "step": 2114
    },
    {
      "epoch": 0.28594122319301035,
      "grad_norm": 0.14993202686309814,
      "learning_rate": 2.616775804304955e-05,
      "loss": 0.06606578826904297,
      "step": 2115
    },
    {
      "epoch": 0.28607641998884625,
      "grad_norm": 0.08795338869094849,
      "learning_rate": 2.616318751776378e-05,
      "loss": 0.04561328887939453,
      "step": 2116
    },
    {
      "epoch": 0.2862116167846822,
      "grad_norm": 0.15838433802127838,
      "learning_rate": 2.615861466824284e-05,
      "loss": 0.07107336819171906,
      "step": 2117
    },
    {
      "epoch": 0.28634681358051817,
      "grad_norm": 0.2660566568374634,
      "learning_rate": 2.6154039495438825e-05,
      "loss": 0.12222003936767578,
      "step": 2118
    },
    {
      "epoch": 0.28648201037635407,
      "grad_norm": 0.09853076189756393,
      "learning_rate": 2.6149462000304302e-05,
      "loss": 0.08576822280883789,
      "step": 2119
    },
    {
      "epoch": 0.28661720717219,
      "grad_norm": 0.07233782857656479,
      "learning_rate": 2.6144882183792335e-05,
      "loss": 0.059578895568847656,
      "step": 2120
    },
    {
      "epoch": 0.286752403968026,
      "grad_norm": 0.06667614728212357,
      "learning_rate": 2.6140300046856468e-05,
      "loss": 0.04423332214355469,
      "step": 2121
    },
    {
      "epoch": 0.2868876007638619,
      "grad_norm": 0.2102232575416565,
      "learning_rate": 2.6135715590450722e-05,
      "loss": 0.10194778442382812,
      "step": 2122
    },
    {
      "epoch": 0.28702279755969784,
      "grad_norm": 0.19450493156909943,
      "learning_rate": 2.6131128815529608e-05,
      "loss": 0.06612181663513184,
      "step": 2123
    },
    {
      "epoch": 0.2871579943555338,
      "grad_norm": 0.0883248820900917,
      "learning_rate": 2.6126539723048115e-05,
      "loss": 0.07953023910522461,
      "step": 2124
    },
    {
      "epoch": 0.2872931911513697,
      "grad_norm": 0.10256035625934601,
      "learning_rate": 2.612194831396172e-05,
      "loss": 0.04761248826980591,
      "step": 2125
    },
    {
      "epoch": 0.28742838794720565,
      "grad_norm": 0.27658361196517944,
      "learning_rate": 2.611735458922637e-05,
      "loss": 0.07846355438232422,
      "step": 2126
    },
    {
      "epoch": 0.2875635847430416,
      "grad_norm": 0.10063400119543076,
      "learning_rate": 2.6112758549798515e-05,
      "loss": 0.06652474403381348,
      "step": 2127
    },
    {
      "epoch": 0.2876987815388775,
      "grad_norm": 0.1689261645078659,
      "learning_rate": 2.610816019663507e-05,
      "loss": 0.12515521049499512,
      "step": 2128
    },
    {
      "epoch": 0.28783397833471347,
      "grad_norm": 0.15385380387306213,
      "learning_rate": 2.6103559530693436e-05,
      "loss": 0.07243585586547852,
      "step": 2129
    },
    {
      "epoch": 0.2879691751305494,
      "grad_norm": 0.11074637621641159,
      "learning_rate": 2.6098956552931495e-05,
      "loss": 0.10162067413330078,
      "step": 2130
    },
    {
      "epoch": 0.2881043719263853,
      "grad_norm": 0.3546871542930603,
      "learning_rate": 2.6094351264307613e-05,
      "loss": 0.09630537033081055,
      "step": 2131
    },
    {
      "epoch": 0.2882395687222213,
      "grad_norm": 0.07562275975942612,
      "learning_rate": 2.6089743665780635e-05,
      "loss": 0.057579994201660156,
      "step": 2132
    },
    {
      "epoch": 0.28837476551805724,
      "grad_norm": 0.054605837911367416,
      "learning_rate": 2.6085133758309887e-05,
      "loss": 0.056827545166015625,
      "step": 2133
    },
    {
      "epoch": 0.28850996231389314,
      "grad_norm": 0.0396919809281826,
      "learning_rate": 2.6080521542855182e-05,
      "loss": 0.0452117919921875,
      "step": 2134
    },
    {
      "epoch": 0.2886451591097291,
      "grad_norm": 0.10526531934738159,
      "learning_rate": 2.60759070203768e-05,
      "loss": 0.07203531265258789,
      "step": 2135
    },
    {
      "epoch": 0.28878035590556506,
      "grad_norm": 0.15222521126270294,
      "learning_rate": 2.607129019183551e-05,
      "loss": 0.07175898551940918,
      "step": 2136
    },
    {
      "epoch": 0.28891555270140096,
      "grad_norm": 0.06815645843744278,
      "learning_rate": 2.6066671058192566e-05,
      "loss": 0.08519530296325684,
      "step": 2137
    },
    {
      "epoch": 0.2890507494972369,
      "grad_norm": 0.04716069996356964,
      "learning_rate": 2.606204962040969e-05,
      "loss": 0.05622053146362305,
      "step": 2138
    },
    {
      "epoch": 0.28918594629307287,
      "grad_norm": 0.08457861840724945,
      "learning_rate": 2.6057425879449095e-05,
      "loss": 0.08150696754455566,
      "step": 2139
    },
    {
      "epoch": 0.28932114308890877,
      "grad_norm": 0.19944559037685394,
      "learning_rate": 2.605279983627347e-05,
      "loss": 0.07221436500549316,
      "step": 2140
    },
    {
      "epoch": 0.28945633988474473,
      "grad_norm": 0.1339990496635437,
      "learning_rate": 2.6048171491845974e-05,
      "loss": 0.06955194473266602,
      "step": 2141
    },
    {
      "epoch": 0.2895915366805807,
      "grad_norm": 0.06655459105968475,
      "learning_rate": 2.604354084713026e-05,
      "loss": 0.07123279571533203,
      "step": 2142
    },
    {
      "epoch": 0.2897267334764166,
      "grad_norm": 0.0839993804693222,
      "learning_rate": 2.6038907903090446e-05,
      "loss": 0.05715513229370117,
      "step": 2143
    },
    {
      "epoch": 0.28986193027225254,
      "grad_norm": 0.06809302419424057,
      "learning_rate": 2.6034272660691143e-05,
      "loss": 0.046620845794677734,
      "step": 2144
    },
    {
      "epoch": 0.2899971270680885,
      "grad_norm": 0.10806578397750854,
      "learning_rate": 2.6029635120897434e-05,
      "loss": 0.08280181884765625,
      "step": 2145
    },
    {
      "epoch": 0.2901323238639244,
      "grad_norm": 0.18974973261356354,
      "learning_rate": 2.6024995284674867e-05,
      "loss": 0.08384335041046143,
      "step": 2146
    },
    {
      "epoch": 0.29026752065976036,
      "grad_norm": 0.09509654343128204,
      "learning_rate": 2.6020353152989496e-05,
      "loss": 0.05357968807220459,
      "step": 2147
    },
    {
      "epoch": 0.2904027174555963,
      "grad_norm": 0.2146265208721161,
      "learning_rate": 2.601570872680783e-05,
      "loss": 0.07716226577758789,
      "step": 2148
    },
    {
      "epoch": 0.2905379142514322,
      "grad_norm": 0.16212603449821472,
      "learning_rate": 2.6011062007096857e-05,
      "loss": 0.05240631103515625,
      "step": 2149
    },
    {
      "epoch": 0.2906731110472682,
      "grad_norm": 0.05886950343847275,
      "learning_rate": 2.6006412994824067e-05,
      "loss": 0.058757781982421875,
      "step": 2150
    },
    {
      "epoch": 0.29080830784310413,
      "grad_norm": 0.08821522444486618,
      "learning_rate": 2.6001761690957388e-05,
      "loss": 0.06379985809326172,
      "step": 2151
    },
    {
      "epoch": 0.29094350463894003,
      "grad_norm": 0.05830740928649902,
      "learning_rate": 2.5997108096465263e-05,
      "loss": 0.051024019718170166,
      "step": 2152
    },
    {
      "epoch": 0.291078701434776,
      "grad_norm": 0.13905581831932068,
      "learning_rate": 2.599245221231659e-05,
      "loss": 0.059331655502319336,
      "step": 2153
    },
    {
      "epoch": 0.29121389823061195,
      "grad_norm": 0.2339545488357544,
      "learning_rate": 2.5987794039480743e-05,
      "loss": 0.0910719633102417,
      "step": 2154
    },
    {
      "epoch": 0.29134909502644785,
      "grad_norm": 0.3189767003059387,
      "learning_rate": 2.5983133578927584e-05,
      "loss": 0.08815765380859375,
      "step": 2155
    },
    {
      "epoch": 0.2914842918222838,
      "grad_norm": 0.06492186337709427,
      "learning_rate": 2.5978470831627444e-05,
      "loss": 0.05967903137207031,
      "step": 2156
    },
    {
      "epoch": 0.29161948861811976,
      "grad_norm": 0.14014960825443268,
      "learning_rate": 2.597380579855113e-05,
      "loss": 0.0846705436706543,
      "step": 2157
    },
    {
      "epoch": 0.2917546854139557,
      "grad_norm": 0.10657402127981186,
      "learning_rate": 2.5969138480669936e-05,
      "loss": 0.07187724113464355,
      "step": 2158
    },
    {
      "epoch": 0.2918898822097916,
      "grad_norm": 0.24730943143367767,
      "learning_rate": 2.5964468878955614e-05,
      "loss": 0.07778358459472656,
      "step": 2159
    },
    {
      "epoch": 0.2920250790056276,
      "grad_norm": 0.19430230557918549,
      "learning_rate": 2.5959796994380397e-05,
      "loss": 0.05954170227050781,
      "step": 2160
    },
    {
      "epoch": 0.29216027580146353,
      "grad_norm": 0.09123359620571136,
      "learning_rate": 2.5955122827917004e-05,
      "loss": 0.06579208374023438,
      "step": 2161
    },
    {
      "epoch": 0.29229547259729943,
      "grad_norm": 0.06193273887038231,
      "learning_rate": 2.595044638053862e-05,
      "loss": 0.05091738700866699,
      "step": 2162
    },
    {
      "epoch": 0.2924306693931354,
      "grad_norm": 0.057271480560302734,
      "learning_rate": 2.59457676532189e-05,
      "loss": 0.05841255187988281,
      "step": 2163
    },
    {
      "epoch": 0.29256586618897135,
      "grad_norm": 0.3168664276599884,
      "learning_rate": 2.594108664693199e-05,
      "loss": 0.12117242813110352,
      "step": 2164
    },
    {
      "epoch": 0.29270106298480725,
      "grad_norm": 0.09441503137350082,
      "learning_rate": 2.5936403362652494e-05,
      "loss": 0.08960413932800293,
      "step": 2165
    },
    {
      "epoch": 0.2928362597806432,
      "grad_norm": 0.07435912638902664,
      "learning_rate": 2.5931717801355497e-05,
      "loss": 0.07468914985656738,
      "step": 2166
    },
    {
      "epoch": 0.29297145657647916,
      "grad_norm": 0.08187311887741089,
      "learning_rate": 2.5927029964016556e-05,
      "loss": 0.05230635404586792,
      "step": 2167
    },
    {
      "epoch": 0.29310665337231506,
      "grad_norm": 0.17817473411560059,
      "learning_rate": 2.592233985161171e-05,
      "loss": 0.09858036041259766,
      "step": 2168
    },
    {
      "epoch": 0.293241850168151,
      "grad_norm": 0.11683443933725357,
      "learning_rate": 2.5917647465117463e-05,
      "loss": 0.0915069580078125,
      "step": 2169
    },
    {
      "epoch": 0.293377046963987,
      "grad_norm": 0.29508888721466064,
      "learning_rate": 2.591295280551079e-05,
      "loss": 0.07468843460083008,
      "step": 2170
    },
    {
      "epoch": 0.2935122437598229,
      "grad_norm": 0.07180826365947723,
      "learning_rate": 2.590825587376915e-05,
      "loss": 0.06031370162963867,
      "step": 2171
    },
    {
      "epoch": 0.29364744055565883,
      "grad_norm": 0.06169212982058525,
      "learning_rate": 2.5903556670870464e-05,
      "loss": 0.0592656135559082,
      "step": 2172
    },
    {
      "epoch": 0.2937826373514948,
      "grad_norm": 0.06674641370773315,
      "learning_rate": 2.589885519779314e-05,
      "loss": 0.042227745056152344,
      "step": 2173
    },
    {
      "epoch": 0.2939178341473307,
      "grad_norm": 0.1836625039577484,
      "learning_rate": 2.5894151455516043e-05,
      "loss": 0.06458330154418945,
      "step": 2174
    },
    {
      "epoch": 0.29405303094316665,
      "grad_norm": 0.14718295633792877,
      "learning_rate": 2.5889445445018513e-05,
      "loss": 0.08520126342773438,
      "step": 2175
    },
    {
      "epoch": 0.2941882277390026,
      "grad_norm": 0.09587886184453964,
      "learning_rate": 2.5884737167280375e-05,
      "loss": 0.05840873718261719,
      "step": 2176
    },
    {
      "epoch": 0.2943234245348385,
      "grad_norm": 0.09669416397809982,
      "learning_rate": 2.5880026623281914e-05,
      "loss": 0.055062055587768555,
      "step": 2177
    },
    {
      "epoch": 0.29445862133067446,
      "grad_norm": 0.19759581983089447,
      "learning_rate": 2.5875313814003892e-05,
      "loss": 0.08341217041015625,
      "step": 2178
    },
    {
      "epoch": 0.2945938181265104,
      "grad_norm": 0.05674462392926216,
      "learning_rate": 2.587059874042754e-05,
      "loss": 0.042471885681152344,
      "step": 2179
    },
    {
      "epoch": 0.2947290149223463,
      "grad_norm": 0.049224868416786194,
      "learning_rate": 2.5865881403534557e-05,
      "loss": 0.04879045486450195,
      "step": 2180
    },
    {
      "epoch": 0.2948642117181823,
      "grad_norm": 0.09328607469797134,
      "learning_rate": 2.5861161804307124e-05,
      "loss": 0.06924581527709961,
      "step": 2181
    },
    {
      "epoch": 0.29499940851401824,
      "grad_norm": 0.15414372086524963,
      "learning_rate": 2.5856439943727886e-05,
      "loss": 0.10968542098999023,
      "step": 2182
    },
    {
      "epoch": 0.29513460530985414,
      "grad_norm": 0.06725557893514633,
      "learning_rate": 2.5851715822779954e-05,
      "loss": 0.05676007270812988,
      "step": 2183
    },
    {
      "epoch": 0.2952698021056901,
      "grad_norm": 0.23364093899726868,
      "learning_rate": 2.5846989442446926e-05,
      "loss": 0.06983709335327148,
      "step": 2184
    },
    {
      "epoch": 0.29540499890152605,
      "grad_norm": 0.1450379192829132,
      "learning_rate": 2.584226080371285e-05,
      "loss": 0.06979644298553467,
      "step": 2185
    },
    {
      "epoch": 0.29554019569736195,
      "grad_norm": 0.27563992142677307,
      "learning_rate": 2.5837529907562258e-05,
      "loss": 0.08946192264556885,
      "step": 2186
    },
    {
      "epoch": 0.2956753924931979,
      "grad_norm": 0.14683540165424347,
      "learning_rate": 2.5832796754980138e-05,
      "loss": 0.08282756805419922,
      "step": 2187
    },
    {
      "epoch": 0.29581058928903387,
      "grad_norm": 0.058543555438518524,
      "learning_rate": 2.5828061346951974e-05,
      "loss": 0.06068229675292969,
      "step": 2188
    },
    {
      "epoch": 0.29594578608486977,
      "grad_norm": 0.09753567725419998,
      "learning_rate": 2.5823323684463693e-05,
      "loss": 0.0587306022644043,
      "step": 2189
    },
    {
      "epoch": 0.2960809828807057,
      "grad_norm": 0.17675673961639404,
      "learning_rate": 2.5818583768501708e-05,
      "loss": 0.06389808654785156,
      "step": 2190
    },
    {
      "epoch": 0.2962161796765417,
      "grad_norm": 0.1600918173789978,
      "learning_rate": 2.5813841600052887e-05,
      "loss": 0.07679128646850586,
      "step": 2191
    },
    {
      "epoch": 0.2963513764723776,
      "grad_norm": 0.09405505657196045,
      "learning_rate": 2.580909718010458e-05,
      "loss": 0.056090354919433594,
      "step": 2192
    },
    {
      "epoch": 0.29648657326821354,
      "grad_norm": 0.07425408065319061,
      "learning_rate": 2.58043505096446e-05,
      "loss": 0.07295727729797363,
      "step": 2193
    },
    {
      "epoch": 0.2966217700640495,
      "grad_norm": 0.0767381340265274,
      "learning_rate": 2.5799601589661223e-05,
      "loss": 0.05618906021118164,
      "step": 2194
    },
    {
      "epoch": 0.2967569668598854,
      "grad_norm": 0.06611032783985138,
      "learning_rate": 2.579485042114321e-05,
      "loss": 0.05595207214355469,
      "step": 2195
    },
    {
      "epoch": 0.29689216365572135,
      "grad_norm": 0.09716714173555374,
      "learning_rate": 2.5790097005079766e-05,
      "loss": 0.0463939905166626,
      "step": 2196
    },
    {
      "epoch": 0.2970273604515573,
      "grad_norm": 0.0926392525434494,
      "learning_rate": 2.5785341342460595e-05,
      "loss": 0.05385637283325195,
      "step": 2197
    },
    {
      "epoch": 0.2971625572473932,
      "grad_norm": 0.04031377285718918,
      "learning_rate": 2.5780583434275837e-05,
      "loss": 0.05214262008666992,
      "step": 2198
    },
    {
      "epoch": 0.29729775404322917,
      "grad_norm": 0.05670323222875595,
      "learning_rate": 2.577582328151612e-05,
      "loss": 0.07280492782592773,
      "step": 2199
    },
    {
      "epoch": 0.2974329508390651,
      "grad_norm": 0.1620096117258072,
      "learning_rate": 2.5771060885172532e-05,
      "loss": 0.11318397521972656,
      "step": 2200
    },
    {
      "epoch": 0.297568147634901,
      "grad_norm": 0.07859797030687332,
      "learning_rate": 2.5766296246236628e-05,
      "loss": 0.06444263458251953,
      "step": 2201
    },
    {
      "epoch": 0.297703344430737,
      "grad_norm": 0.2573566138744354,
      "learning_rate": 2.5761529365700437e-05,
      "loss": 0.10247278213500977,
      "step": 2202
    },
    {
      "epoch": 0.29783854122657294,
      "grad_norm": 0.15654997527599335,
      "learning_rate": 2.5756760244556445e-05,
      "loss": 0.0964818000793457,
      "step": 2203
    },
    {
      "epoch": 0.2979737380224089,
      "grad_norm": 0.09162845462560654,
      "learning_rate": 2.5751988883797603e-05,
      "loss": 0.07657527923583984,
      "step": 2204
    },
    {
      "epoch": 0.2981089348182448,
      "grad_norm": 0.09175653010606766,
      "learning_rate": 2.574721528441734e-05,
      "loss": 0.040541887283325195,
      "step": 2205
    },
    {
      "epoch": 0.29824413161408075,
      "grad_norm": 0.1088048443198204,
      "learning_rate": 2.5742439447409545e-05,
      "loss": 0.045897722244262695,
      "step": 2206
    },
    {
      "epoch": 0.2983793284099167,
      "grad_norm": 0.11155156791210175,
      "learning_rate": 2.5737661373768568e-05,
      "loss": 0.05272102355957031,
      "step": 2207
    },
    {
      "epoch": 0.2985145252057526,
      "grad_norm": 0.03990326076745987,
      "learning_rate": 2.5732881064489237e-05,
      "loss": 0.06196737289428711,
      "step": 2208
    },
    {
      "epoch": 0.29864972200158857,
      "grad_norm": 0.23571626842021942,
      "learning_rate": 2.572809852056683e-05,
      "loss": 0.12282919883728027,
      "step": 2209
    },
    {
      "epoch": 0.2987849187974245,
      "grad_norm": 0.09622597694396973,
      "learning_rate": 2.572331374299711e-05,
      "loss": 0.091552734375,
      "step": 2210
    },
    {
      "epoch": 0.2989201155932604,
      "grad_norm": 0.1908119171857834,
      "learning_rate": 2.5718526732776276e-05,
      "loss": 0.09036707878112793,
      "step": 2211
    },
    {
      "epoch": 0.2990553123890964,
      "grad_norm": 0.13202886283397675,
      "learning_rate": 2.5713737490901023e-05,
      "loss": 0.0438840389251709,
      "step": 2212
    },
    {
      "epoch": 0.29919050918493234,
      "grad_norm": 0.12403219193220139,
      "learning_rate": 2.570894601836849e-05,
      "loss": 0.06312799453735352,
      "step": 2213
    },
    {
      "epoch": 0.29932570598076824,
      "grad_norm": 0.0662514716386795,
      "learning_rate": 2.5704152316176287e-05,
      "loss": 0.046802520751953125,
      "step": 2214
    },
    {
      "epoch": 0.2994609027766042,
      "grad_norm": 0.22417013347148895,
      "learning_rate": 2.5699356385322487e-05,
      "loss": 0.10979652404785156,
      "step": 2215
    },
    {
      "epoch": 0.29959609957244016,
      "grad_norm": 0.11969805508852005,
      "learning_rate": 2.5694558226805643e-05,
      "loss": 0.0529780387878418,
      "step": 2216
    },
    {
      "epoch": 0.29973129636827606,
      "grad_norm": 0.07081670314073563,
      "learning_rate": 2.568975784162474e-05,
      "loss": 0.07811403274536133,
      "step": 2217
    },
    {
      "epoch": 0.299866493164112,
      "grad_norm": 0.0757928416132927,
      "learning_rate": 2.5684955230779245e-05,
      "loss": 0.03953409194946289,
      "step": 2218
    },
    {
      "epoch": 0.30000168995994797,
      "grad_norm": 0.10730839520692825,
      "learning_rate": 2.5680150395269096e-05,
      "loss": 0.11053705215454102,
      "step": 2219
    },
    {
      "epoch": 0.30013688675578387,
      "grad_norm": 0.16285882890224457,
      "learning_rate": 2.5675343336094683e-05,
      "loss": 0.08858633041381836,
      "step": 2220
    },
    {
      "epoch": 0.30027208355161983,
      "grad_norm": 0.06795356422662735,
      "learning_rate": 2.5670534054256855e-05,
      "loss": 0.061885833740234375,
      "step": 2221
    },
    {
      "epoch": 0.3004072803474558,
      "grad_norm": 0.3278854489326477,
      "learning_rate": 2.5665722550756937e-05,
      "loss": 0.10415267944335938,
      "step": 2222
    },
    {
      "epoch": 0.3005424771432917,
      "grad_norm": 0.11619860678911209,
      "learning_rate": 2.5660908826596707e-05,
      "loss": 0.07813048362731934,
      "step": 2223
    },
    {
      "epoch": 0.30067767393912764,
      "grad_norm": 0.14091181755065918,
      "learning_rate": 2.5656092882778413e-05,
      "loss": 0.0955878496170044,
      "step": 2224
    },
    {
      "epoch": 0.3008128707349636,
      "grad_norm": 0.18230250477790833,
      "learning_rate": 2.565127472030475e-05,
      "loss": 0.06585180759429932,
      "step": 2225
    },
    {
      "epoch": 0.3009480675307995,
      "grad_norm": 0.20425178110599518,
      "learning_rate": 2.5646454340178894e-05,
      "loss": 0.06609654426574707,
      "step": 2226
    },
    {
      "epoch": 0.30108326432663546,
      "grad_norm": 0.11512142419815063,
      "learning_rate": 2.564163174340447e-05,
      "loss": 0.053675174713134766,
      "step": 2227
    },
    {
      "epoch": 0.3012184611224714,
      "grad_norm": 0.21589672565460205,
      "learning_rate": 2.5636806930985565e-05,
      "loss": 0.12194061279296875,
      "step": 2228
    },
    {
      "epoch": 0.3013536579183073,
      "grad_norm": 0.16532501578330994,
      "learning_rate": 2.5631979903926738e-05,
      "loss": 0.08331567049026489,
      "step": 2229
    },
    {
      "epoch": 0.3014888547141433,
      "grad_norm": 0.21849635243415833,
      "learning_rate": 2.5627150663233e-05,
      "loss": 0.055123329162597656,
      "step": 2230
    },
    {
      "epoch": 0.30162405150997923,
      "grad_norm": 0.1428455114364624,
      "learning_rate": 2.5622319209909817e-05,
      "loss": 0.055088043212890625,
      "step": 2231
    },
    {
      "epoch": 0.30175924830581513,
      "grad_norm": 0.47405576705932617,
      "learning_rate": 2.5617485544963135e-05,
      "loss": 0.09007501602172852,
      "step": 2232
    },
    {
      "epoch": 0.3018944451016511,
      "grad_norm": 0.1740884780883789,
      "learning_rate": 2.561264966939934e-05,
      "loss": 0.07647299766540527,
      "step": 2233
    },
    {
      "epoch": 0.30202964189748704,
      "grad_norm": 0.1323907971382141,
      "learning_rate": 2.5607811584225294e-05,
      "loss": 0.0838174819946289,
      "step": 2234
    },
    {
      "epoch": 0.30216483869332295,
      "grad_norm": 0.11019740253686905,
      "learning_rate": 2.5602971290448305e-05,
      "loss": 0.05591773986816406,
      "step": 2235
    },
    {
      "epoch": 0.3023000354891589,
      "grad_norm": 0.06115534156560898,
      "learning_rate": 2.5598128789076152e-05,
      "loss": 0.05503726005554199,
      "step": 2236
    },
    {
      "epoch": 0.30243523228499486,
      "grad_norm": 0.37740057706832886,
      "learning_rate": 2.559328408111707e-05,
      "loss": 0.0873861312866211,
      "step": 2237
    },
    {
      "epoch": 0.30257042908083076,
      "grad_norm": 0.23444798588752747,
      "learning_rate": 2.5588437167579755e-05,
      "loss": 0.061660319566726685,
      "step": 2238
    },
    {
      "epoch": 0.3027056258766667,
      "grad_norm": 0.10367490351200104,
      "learning_rate": 2.558358804947335e-05,
      "loss": 0.05496370792388916,
      "step": 2239
    },
    {
      "epoch": 0.3028408226725027,
      "grad_norm": 0.10610882192850113,
      "learning_rate": 2.557873672780748e-05,
      "loss": 0.09772443771362305,
      "step": 2240
    },
    {
      "epoch": 0.3029760194683386,
      "grad_norm": 0.0653059259057045,
      "learning_rate": 2.557388320359221e-05,
      "loss": 0.04547691345214844,
      "step": 2241
    },
    {
      "epoch": 0.30311121626417453,
      "grad_norm": 0.16897399723529816,
      "learning_rate": 2.5569027477838068e-05,
      "loss": 0.06138944625854492,
      "step": 2242
    },
    {
      "epoch": 0.3032464130600105,
      "grad_norm": 0.18280315399169922,
      "learning_rate": 2.5564169551556044e-05,
      "loss": 0.049242258071899414,
      "step": 2243
    },
    {
      "epoch": 0.3033816098558464,
      "grad_norm": 0.1963411420583725,
      "learning_rate": 2.5559309425757586e-05,
      "loss": 0.07587909698486328,
      "step": 2244
    },
    {
      "epoch": 0.30351680665168235,
      "grad_norm": 0.1461852341890335,
      "learning_rate": 2.5554447101454597e-05,
      "loss": 0.052052974700927734,
      "step": 2245
    },
    {
      "epoch": 0.3036520034475183,
      "grad_norm": 0.10520715266466141,
      "learning_rate": 2.554958257965944e-05,
      "loss": 0.06604957580566406,
      "step": 2246
    },
    {
      "epoch": 0.3037872002433542,
      "grad_norm": 0.11993642151355743,
      "learning_rate": 2.554471586138493e-05,
      "loss": 0.06366157531738281,
      "step": 2247
    },
    {
      "epoch": 0.30392239703919016,
      "grad_norm": 0.08100903034210205,
      "learning_rate": 2.5539846947644342e-05,
      "loss": 0.05422258377075195,
      "step": 2248
    },
    {
      "epoch": 0.3040575938350261,
      "grad_norm": 0.3330111801624298,
      "learning_rate": 2.5534975839451416e-05,
      "loss": 0.07745647430419922,
      "step": 2249
    },
    {
      "epoch": 0.3041927906308621,
      "grad_norm": 0.18529288470745087,
      "learning_rate": 2.5530102537820348e-05,
      "loss": 0.07784557342529297,
      "step": 2250
    },
    {
      "epoch": 0.304327987426698,
      "grad_norm": 0.11477598547935486,
      "learning_rate": 2.5525227043765774e-05,
      "loss": 0.10215377807617188,
      "step": 2251
    },
    {
      "epoch": 0.30446318422253393,
      "grad_norm": 0.10381592810153961,
      "learning_rate": 2.55203493583028e-05,
      "loss": 0.0888218879699707,
      "step": 2252
    },
    {
      "epoch": 0.3045983810183699,
      "grad_norm": 0.23445124924182892,
      "learning_rate": 2.551546948244699e-05,
      "loss": 0.11539363861083984,
      "step": 2253
    },
    {
      "epoch": 0.3047335778142058,
      "grad_norm": 0.18332181870937347,
      "learning_rate": 2.551058741721436e-05,
      "loss": 0.0734856128692627,
      "step": 2254
    },
    {
      "epoch": 0.30486877461004175,
      "grad_norm": 0.16522622108459473,
      "learning_rate": 2.550570316362138e-05,
      "loss": 0.10046911239624023,
      "step": 2255
    },
    {
      "epoch": 0.3050039714058777,
      "grad_norm": 0.09568028151988983,
      "learning_rate": 2.5500816722684975e-05,
      "loss": 0.07169008255004883,
      "step": 2256
    },
    {
      "epoch": 0.3051391682017136,
      "grad_norm": 0.051216576248407364,
      "learning_rate": 2.549592809542253e-05,
      "loss": 0.05153989791870117,
      "step": 2257
    },
    {
      "epoch": 0.30527436499754956,
      "grad_norm": 0.10579536110162735,
      "learning_rate": 2.549103728285189e-05,
      "loss": 0.08163261413574219,
      "step": 2258
    },
    {
      "epoch": 0.3054095617933855,
      "grad_norm": 0.0329514779150486,
      "learning_rate": 2.548614428599134e-05,
      "loss": 0.03496980667114258,
      "step": 2259
    },
    {
      "epoch": 0.3055447585892214,
      "grad_norm": 0.10999783873558044,
      "learning_rate": 2.5481249105859633e-05,
      "loss": 0.05714726448059082,
      "step": 2260
    },
    {
      "epoch": 0.3056799553850574,
      "grad_norm": 0.03867647051811218,
      "learning_rate": 2.5476351743475964e-05,
      "loss": 0.041094183921813965,
      "step": 2261
    },
    {
      "epoch": 0.30581515218089333,
      "grad_norm": 0.33016112446784973,
      "learning_rate": 2.547145219986e-05,
      "loss": 0.10236454010009766,
      "step": 2262
    },
    {
      "epoch": 0.30595034897672924,
      "grad_norm": 0.11598561704158783,
      "learning_rate": 2.5466550476031846e-05,
      "loss": 0.051011085510253906,
      "step": 2263
    },
    {
      "epoch": 0.3060855457725652,
      "grad_norm": 0.22420819103717804,
      "learning_rate": 2.5461646573012072e-05,
      "loss": 0.07788515090942383,
      "step": 2264
    },
    {
      "epoch": 0.30622074256840115,
      "grad_norm": 0.1614769548177719,
      "learning_rate": 2.5456740491821687e-05,
      "loss": 0.0509333610534668,
      "step": 2265
    },
    {
      "epoch": 0.30635593936423705,
      "grad_norm": 0.04684172198176384,
      "learning_rate": 2.5451832233482172e-05,
      "loss": 0.033049583435058594,
      "step": 2266
    },
    {
      "epoch": 0.306491136160073,
      "grad_norm": 0.21184338629245758,
      "learning_rate": 2.544692179901545e-05,
      "loss": 0.07379531860351562,
      "step": 2267
    },
    {
      "epoch": 0.30662633295590896,
      "grad_norm": 0.09742343425750732,
      "learning_rate": 2.5442009189443902e-05,
      "loss": 0.04856705665588379,
      "step": 2268
    },
    {
      "epoch": 0.30676152975174487,
      "grad_norm": 0.06952012330293655,
      "learning_rate": 2.5437094405790355e-05,
      "loss": 0.059391021728515625,
      "step": 2269
    },
    {
      "epoch": 0.3068967265475808,
      "grad_norm": 0.11396628618240356,
      "learning_rate": 2.5432177449078096e-05,
      "loss": 0.09135818481445312,
      "step": 2270
    },
    {
      "epoch": 0.3070319233434168,
      "grad_norm": 0.030981941148638725,
      "learning_rate": 2.5427258320330857e-05,
      "loss": 0.029935359954833984,
      "step": 2271
    },
    {
      "epoch": 0.3071671201392527,
      "grad_norm": 0.05656053125858307,
      "learning_rate": 2.5422337020572835e-05,
      "loss": 0.0642547607421875,
      "step": 2272
    },
    {
      "epoch": 0.30730231693508864,
      "grad_norm": 0.08299758285284042,
      "learning_rate": 2.5417413550828667e-05,
      "loss": 0.06255435943603516,
      "step": 2273
    },
    {
      "epoch": 0.3074375137309246,
      "grad_norm": 0.12040698528289795,
      "learning_rate": 2.5412487912123444e-05,
      "loss": 0.10099029541015625,
      "step": 2274
    },
    {
      "epoch": 0.3075727105267605,
      "grad_norm": 0.06496715545654297,
      "learning_rate": 2.5407560105482708e-05,
      "loss": 0.04317343235015869,
      "step": 2275
    },
    {
      "epoch": 0.30770790732259645,
      "grad_norm": 0.0502430722117424,
      "learning_rate": 2.540263013193246e-05,
      "loss": 0.047867536544799805,
      "step": 2276
    },
    {
      "epoch": 0.3078431041184324,
      "grad_norm": 0.1314105987548828,
      "learning_rate": 2.539769799249915e-05,
      "loss": 0.07472860813140869,
      "step": 2277
    },
    {
      "epoch": 0.3079783009142683,
      "grad_norm": 0.0870327427983284,
      "learning_rate": 2.5392763688209666e-05,
      "loss": 0.08386802673339844,
      "step": 2278
    },
    {
      "epoch": 0.30811349771010427,
      "grad_norm": 0.08614157140254974,
      "learning_rate": 2.5387827220091362e-05,
      "loss": 0.07688188552856445,
      "step": 2279
    },
    {
      "epoch": 0.3082486945059402,
      "grad_norm": 0.1127915009856224,
      "learning_rate": 2.538288858917204e-05,
      "loss": 0.10305452346801758,
      "step": 2280
    },
    {
      "epoch": 0.3083838913017761,
      "grad_norm": 0.11062409728765488,
      "learning_rate": 2.5377947796479936e-05,
      "loss": 0.08144855499267578,
      "step": 2281
    },
    {
      "epoch": 0.3085190880976121,
      "grad_norm": 0.05871828645467758,
      "learning_rate": 2.537300484304377e-05,
      "loss": 0.05708503723144531,
      "step": 2282
    },
    {
      "epoch": 0.30865428489344804,
      "grad_norm": 0.0803714320063591,
      "learning_rate": 2.536805972989267e-05,
      "loss": 0.06974267959594727,
      "step": 2283
    },
    {
      "epoch": 0.30878948168928394,
      "grad_norm": 0.08624116331338882,
      "learning_rate": 2.5363112458056252e-05,
      "loss": 0.04434061050415039,
      "step": 2284
    },
    {
      "epoch": 0.3089246784851199,
      "grad_norm": 0.08888224512338638,
      "learning_rate": 2.5358163028564552e-05,
      "loss": 0.045476555824279785,
      "step": 2285
    },
    {
      "epoch": 0.30905987528095585,
      "grad_norm": 0.09790392220020294,
      "learning_rate": 2.535321144244808e-05,
      "loss": 0.07386910915374756,
      "step": 2286
    },
    {
      "epoch": 0.30919507207679175,
      "grad_norm": 0.0618215836584568,
      "learning_rate": 2.534825770073777e-05,
      "loss": 0.09366035461425781,
      "step": 2287
    },
    {
      "epoch": 0.3093302688726277,
      "grad_norm": 0.2710959017276764,
      "learning_rate": 2.5343301804465026e-05,
      "loss": 0.07016468048095703,
      "step": 2288
    },
    {
      "epoch": 0.30946546566846367,
      "grad_norm": 0.09551559388637543,
      "learning_rate": 2.533834375466169e-05,
      "loss": 0.08010244369506836,
      "step": 2289
    },
    {
      "epoch": 0.30960066246429957,
      "grad_norm": 0.06421948969364166,
      "learning_rate": 2.533338355236005e-05,
      "loss": 0.06645393371582031,
      "step": 2290
    },
    {
      "epoch": 0.3097358592601355,
      "grad_norm": 0.03430100530385971,
      "learning_rate": 2.532842119859285e-05,
      "loss": 0.03856325149536133,
      "step": 2291
    },
    {
      "epoch": 0.3098710560559715,
      "grad_norm": 0.15019206702709198,
      "learning_rate": 2.532345669439328e-05,
      "loss": 0.07943582534790039,
      "step": 2292
    },
    {
      "epoch": 0.3100062528518074,
      "grad_norm": 0.08374845236539841,
      "learning_rate": 2.5318490040794975e-05,
      "loss": 0.04987788200378418,
      "step": 2293
    },
    {
      "epoch": 0.31014144964764334,
      "grad_norm": 0.14791464805603027,
      "learning_rate": 2.531352123883202e-05,
      "loss": 0.07583856582641602,
      "step": 2294
    },
    {
      "epoch": 0.3102766464434793,
      "grad_norm": 0.12051131576299667,
      "learning_rate": 2.530855028953894e-05,
      "loss": 0.10768318176269531,
      "step": 2295
    },
    {
      "epoch": 0.31041184323931525,
      "grad_norm": 0.06165115162730217,
      "learning_rate": 2.5303577193950724e-05,
      "loss": 0.04842996597290039,
      "step": 2296
    },
    {
      "epoch": 0.31054704003515116,
      "grad_norm": 0.04354666918516159,
      "learning_rate": 2.5298601953102785e-05,
      "loss": 0.041533470153808594,
      "step": 2297
    },
    {
      "epoch": 0.3106822368309871,
      "grad_norm": 0.05833108723163605,
      "learning_rate": 2.5293624568031008e-05,
      "loss": 0.04864239692687988,
      "step": 2298
    },
    {
      "epoch": 0.31081743362682307,
      "grad_norm": 0.31188562512397766,
      "learning_rate": 2.5288645039771697e-05,
      "loss": 0.11146879196166992,
      "step": 2299
    },
    {
      "epoch": 0.31095263042265897,
      "grad_norm": 0.040503934025764465,
      "learning_rate": 2.5283663369361624e-05,
      "loss": 0.04791879653930664,
      "step": 2300
    },
    {
      "epoch": 0.31108782721849493,
      "grad_norm": 0.05054056644439697,
      "learning_rate": 2.5278679557837998e-05,
      "loss": 0.047451019287109375,
      "step": 2301
    },
    {
      "epoch": 0.3112230240143309,
      "grad_norm": 0.09091488271951675,
      "learning_rate": 2.5273693606238474e-05,
      "loss": 0.08942079544067383,
      "step": 2302
    },
    {
      "epoch": 0.3113582208101668,
      "grad_norm": 0.06451370567083359,
      "learning_rate": 2.5268705515601164e-05,
      "loss": 0.048095703125,
      "step": 2303
    },
    {
      "epoch": 0.31149341760600274,
      "grad_norm": 0.19290512800216675,
      "learning_rate": 2.5263715286964596e-05,
      "loss": 0.1023675799369812,
      "step": 2304
    },
    {
      "epoch": 0.3116286144018387,
      "grad_norm": 0.062214192003011703,
      "learning_rate": 2.525872292136778e-05,
      "loss": 0.04499626159667969,
      "step": 2305
    },
    {
      "epoch": 0.3117638111976746,
      "grad_norm": 0.11830825358629227,
      "learning_rate": 2.525372841985014e-05,
      "loss": 0.07111459970474243,
      "step": 2306
    },
    {
      "epoch": 0.31189900799351056,
      "grad_norm": 0.11798103898763657,
      "learning_rate": 2.5248731783451567e-05,
      "loss": 0.0844111442565918,
      "step": 2307
    },
    {
      "epoch": 0.3120342047893465,
      "grad_norm": 0.0817176103591919,
      "learning_rate": 2.524373301321238e-05,
      "loss": 0.06750679016113281,
      "step": 2308
    },
    {
      "epoch": 0.3121694015851824,
      "grad_norm": 0.18090811371803284,
      "learning_rate": 2.5238732110173356e-05,
      "loss": 0.0880126953125,
      "step": 2309
    },
    {
      "epoch": 0.3123045983810184,
      "grad_norm": 0.12871138751506805,
      "learning_rate": 2.5233729075375708e-05,
      "loss": 0.05259895324707031,
      "step": 2310
    },
    {
      "epoch": 0.31243979517685433,
      "grad_norm": 0.10063536465167999,
      "learning_rate": 2.522872390986109e-05,
      "loss": 0.05362749099731445,
      "step": 2311
    },
    {
      "epoch": 0.31257499197269023,
      "grad_norm": 0.12158272415399551,
      "learning_rate": 2.522371661467161e-05,
      "loss": 0.07718062400817871,
      "step": 2312
    },
    {
      "epoch": 0.3127101887685262,
      "grad_norm": 0.06021657586097717,
      "learning_rate": 2.521870719084981e-05,
      "loss": 0.04691481590270996,
      "step": 2313
    },
    {
      "epoch": 0.31284538556436214,
      "grad_norm": 0.1747627556324005,
      "learning_rate": 2.5213695639438686e-05,
      "loss": 0.08052420616149902,
      "step": 2314
    },
    {
      "epoch": 0.31298058236019805,
      "grad_norm": 0.1785389631986618,
      "learning_rate": 2.5208681961481657e-05,
      "loss": 0.06527280807495117,
      "step": 2315
    },
    {
      "epoch": 0.313115779156034,
      "grad_norm": 0.17502829432487488,
      "learning_rate": 2.5203666158022607e-05,
      "loss": 0.07204568386077881,
      "step": 2316
    },
    {
      "epoch": 0.31325097595186996,
      "grad_norm": 0.08006101846694946,
      "learning_rate": 2.519864823010585e-05,
      "loss": 0.08469772338867188,
      "step": 2317
    },
    {
      "epoch": 0.31338617274770586,
      "grad_norm": 0.05874093621969223,
      "learning_rate": 2.5193628178776148e-05,
      "loss": 0.049321889877319336,
      "step": 2318
    },
    {
      "epoch": 0.3135213695435418,
      "grad_norm": 0.10377375036478043,
      "learning_rate": 2.5188606005078695e-05,
      "loss": 0.06112253665924072,
      "step": 2319
    },
    {
      "epoch": 0.3136565663393778,
      "grad_norm": 0.21512280404567719,
      "learning_rate": 2.518358171005914e-05,
      "loss": 0.09794676303863525,
      "step": 2320
    },
    {
      "epoch": 0.3137917631352137,
      "grad_norm": 0.11540849506855011,
      "learning_rate": 2.517855529476357e-05,
      "loss": 0.05525970458984375,
      "step": 2321
    },
    {
      "epoch": 0.31392695993104963,
      "grad_norm": 0.23029382526874542,
      "learning_rate": 2.517352676023851e-05,
      "loss": 0.07313638925552368,
      "step": 2322
    },
    {
      "epoch": 0.3140621567268856,
      "grad_norm": 0.08203113079071045,
      "learning_rate": 2.5168496107530925e-05,
      "loss": 0.09019255638122559,
      "step": 2323
    },
    {
      "epoch": 0.3141973535227215,
      "grad_norm": 0.1450749784708023,
      "learning_rate": 2.5163463337688224e-05,
      "loss": 0.069061279296875,
      "step": 2324
    },
    {
      "epoch": 0.31433255031855745,
      "grad_norm": 0.0744481086730957,
      "learning_rate": 2.515842845175826e-05,
      "loss": 0.052683472633361816,
      "step": 2325
    },
    {
      "epoch": 0.3144677471143934,
      "grad_norm": 0.14945203065872192,
      "learning_rate": 2.5153391450789326e-05,
      "loss": 0.08378362655639648,
      "step": 2326
    },
    {
      "epoch": 0.3146029439102293,
      "grad_norm": 0.06944262236356735,
      "learning_rate": 2.514835233583014e-05,
      "loss": 0.07415914535522461,
      "step": 2327
    },
    {
      "epoch": 0.31473814070606526,
      "grad_norm": 0.04126359522342682,
      "learning_rate": 2.514331110792988e-05,
      "loss": 0.05480104684829712,
      "step": 2328
    },
    {
      "epoch": 0.3148733375019012,
      "grad_norm": 0.09607455879449844,
      "learning_rate": 2.513826776813816e-05,
      "loss": 0.06066393852233887,
      "step": 2329
    },
    {
      "epoch": 0.3150085342977371,
      "grad_norm": 0.16600649058818817,
      "learning_rate": 2.5133222317505024e-05,
      "loss": 0.0763096809387207,
      "step": 2330
    },
    {
      "epoch": 0.3151437310935731,
      "grad_norm": 0.08547329902648926,
      "learning_rate": 2.5128174757080965e-05,
      "loss": 0.043604135513305664,
      "step": 2331
    },
    {
      "epoch": 0.31527892788940903,
      "grad_norm": 0.0659867525100708,
      "learning_rate": 2.5123125087916916e-05,
      "loss": 0.08096981048583984,
      "step": 2332
    },
    {
      "epoch": 0.31541412468524493,
      "grad_norm": 0.13382531702518463,
      "learning_rate": 2.5118073311064236e-05,
      "loss": 0.07327628135681152,
      "step": 2333
    },
    {
      "epoch": 0.3155493214810809,
      "grad_norm": 0.14889389276504517,
      "learning_rate": 2.5113019427574734e-05,
      "loss": 0.08733415603637695,
      "step": 2334
    },
    {
      "epoch": 0.31568451827691685,
      "grad_norm": 0.09740637242794037,
      "learning_rate": 2.5107963438500666e-05,
      "loss": 0.06246376037597656,
      "step": 2335
    },
    {
      "epoch": 0.31581971507275275,
      "grad_norm": 0.16206662356853485,
      "learning_rate": 2.51029053448947e-05,
      "loss": 0.07365751266479492,
      "step": 2336
    },
    {
      "epoch": 0.3159549118685887,
      "grad_norm": 0.07863498479127884,
      "learning_rate": 2.509784514780997e-05,
      "loss": 0.04028058052062988,
      "step": 2337
    },
    {
      "epoch": 0.31609010866442466,
      "grad_norm": 0.10697831958532333,
      "learning_rate": 2.5092782848300033e-05,
      "loss": 0.06625747680664062,
      "step": 2338
    },
    {
      "epoch": 0.3162253054602606,
      "grad_norm": 0.11231585592031479,
      "learning_rate": 2.5087718447418886e-05,
      "loss": 0.08422279357910156,
      "step": 2339
    },
    {
      "epoch": 0.3163605022560965,
      "grad_norm": 0.058244891464710236,
      "learning_rate": 2.5082651946220958e-05,
      "loss": 0.05232381820678711,
      "step": 2340
    },
    {
      "epoch": 0.3164956990519325,
      "grad_norm": 0.075192391872406,
      "learning_rate": 2.507758334576113e-05,
      "loss": 0.05902707576751709,
      "step": 2341
    },
    {
      "epoch": 0.31663089584776843,
      "grad_norm": 0.05973155051469803,
      "learning_rate": 2.5072512647094713e-05,
      "loss": 0.06212282180786133,
      "step": 2342
    },
    {
      "epoch": 0.31676609264360434,
      "grad_norm": 0.1123841404914856,
      "learning_rate": 2.506743985127745e-05,
      "loss": 0.06532955169677734,
      "step": 2343
    },
    {
      "epoch": 0.3169012894394403,
      "grad_norm": 0.0680035874247551,
      "learning_rate": 2.506236495936552e-05,
      "loss": 0.06137585639953613,
      "step": 2344
    },
    {
      "epoch": 0.31703648623527625,
      "grad_norm": 0.16798603534698486,
      "learning_rate": 2.5057287972415547e-05,
      "loss": 0.062421560287475586,
      "step": 2345
    },
    {
      "epoch": 0.31717168303111215,
      "grad_norm": 0.095027394592762,
      "learning_rate": 2.5052208891484588e-05,
      "loss": 0.054033756256103516,
      "step": 2346
    },
    {
      "epoch": 0.3173068798269481,
      "grad_norm": 0.07161493599414825,
      "learning_rate": 2.504712771763013e-05,
      "loss": 0.04774975776672363,
      "step": 2347
    },
    {
      "epoch": 0.31744207662278406,
      "grad_norm": 0.10778544098138809,
      "learning_rate": 2.5042044451910108e-05,
      "loss": 0.07732677459716797,
      "step": 2348
    },
    {
      "epoch": 0.31757727341861997,
      "grad_norm": 0.083549365401268,
      "learning_rate": 2.5036959095382875e-05,
      "loss": 0.049724578857421875,
      "step": 2349
    },
    {
      "epoch": 0.3177124702144559,
      "grad_norm": 0.0668855607509613,
      "learning_rate": 2.5031871649107233e-05,
      "loss": 0.04806923866271973,
      "step": 2350
    },
    {
      "epoch": 0.3178476670102919,
      "grad_norm": 0.07492992281913757,
      "learning_rate": 2.5026782114142426e-05,
      "loss": 0.05648326873779297,
      "step": 2351
    },
    {
      "epoch": 0.3179828638061278,
      "grad_norm": 0.07489626854658127,
      "learning_rate": 2.5021690491548107e-05,
      "loss": 0.0668635368347168,
      "step": 2352
    },
    {
      "epoch": 0.31811806060196374,
      "grad_norm": 0.03294162079691887,
      "learning_rate": 2.5016596782384387e-05,
      "loss": 0.035373568534851074,
      "step": 2353
    },
    {
      "epoch": 0.3182532573977997,
      "grad_norm": 0.12333255261182785,
      "learning_rate": 2.5011500987711804e-05,
      "loss": 0.06703853607177734,
      "step": 2354
    },
    {
      "epoch": 0.3183884541936356,
      "grad_norm": 0.07464449852705002,
      "learning_rate": 2.5006403108591325e-05,
      "loss": 0.04925203323364258,
      "step": 2355
    },
    {
      "epoch": 0.31852365098947155,
      "grad_norm": 0.040240079164505005,
      "learning_rate": 2.500130314608436e-05,
      "loss": 0.040866851806640625,
      "step": 2356
    },
    {
      "epoch": 0.3186588477853075,
      "grad_norm": 0.09679000824689865,
      "learning_rate": 2.4996201101252742e-05,
      "loss": 0.10651779174804688,
      "step": 2357
    },
    {
      "epoch": 0.3187940445811434,
      "grad_norm": 0.10267443209886551,
      "learning_rate": 2.4991096975158757e-05,
      "loss": 0.1043238639831543,
      "step": 2358
    },
    {
      "epoch": 0.31892924137697937,
      "grad_norm": 0.11303116381168365,
      "learning_rate": 2.4985990768865095e-05,
      "loss": 0.10076355934143066,
      "step": 2359
    },
    {
      "epoch": 0.3190644381728153,
      "grad_norm": 0.0692182406783104,
      "learning_rate": 2.4980882483434904e-05,
      "loss": 0.06462359428405762,
      "step": 2360
    },
    {
      "epoch": 0.3191996349686512,
      "grad_norm": 0.0846063494682312,
      "learning_rate": 2.497577211993176e-05,
      "loss": 0.04369497299194336,
      "step": 2361
    },
    {
      "epoch": 0.3193348317644872,
      "grad_norm": 0.022632962092757225,
      "learning_rate": 2.4970659679419658e-05,
      "loss": 0.025502681732177734,
      "step": 2362
    },
    {
      "epoch": 0.31947002856032314,
      "grad_norm": 0.11357161402702332,
      "learning_rate": 2.496554516296304e-05,
      "loss": 0.08234643936157227,
      "step": 2363
    },
    {
      "epoch": 0.31960522535615904,
      "grad_norm": 0.06699955463409424,
      "learning_rate": 2.4960428571626784e-05,
      "loss": 0.04887890815734863,
      "step": 2364
    },
    {
      "epoch": 0.319740422151995,
      "grad_norm": 0.10973644256591797,
      "learning_rate": 2.4955309906476177e-05,
      "loss": 0.08873987197875977,
      "step": 2365
    },
    {
      "epoch": 0.31987561894783095,
      "grad_norm": 0.16547797620296478,
      "learning_rate": 2.495018916857696e-05,
      "loss": 0.06520485877990723,
      "step": 2366
    },
    {
      "epoch": 0.32001081574366685,
      "grad_norm": 0.10857252776622772,
      "learning_rate": 2.4945066358995304e-05,
      "loss": 0.045226216316223145,
      "step": 2367
    },
    {
      "epoch": 0.3201460125395028,
      "grad_norm": 0.2102724313735962,
      "learning_rate": 2.493994147879779e-05,
      "loss": 0.07881355285644531,
      "step": 2368
    },
    {
      "epoch": 0.32028120933533877,
      "grad_norm": 0.09104569256305695,
      "learning_rate": 2.4934814529051458e-05,
      "loss": 0.05008077621459961,
      "step": 2369
    },
    {
      "epoch": 0.32041640613117467,
      "grad_norm": 0.06916666030883789,
      "learning_rate": 2.4929685510823763e-05,
      "loss": 0.06512880325317383,
      "step": 2370
    },
    {
      "epoch": 0.3205516029270106,
      "grad_norm": 0.08778955787420273,
      "learning_rate": 2.492455442518259e-05,
      "loss": 0.08102905750274658,
      "step": 2371
    },
    {
      "epoch": 0.3206867997228466,
      "grad_norm": 0.09968572854995728,
      "learning_rate": 2.4919421273196262e-05,
      "loss": 0.07334661483764648,
      "step": 2372
    },
    {
      "epoch": 0.3208219965186825,
      "grad_norm": 0.07584325224161148,
      "learning_rate": 2.4914286055933527e-05,
      "loss": 0.03999125957489014,
      "step": 2373
    },
    {
      "epoch": 0.32095719331451844,
      "grad_norm": 0.07697533816099167,
      "learning_rate": 2.4909148774463572e-05,
      "loss": 0.061720848083496094,
      "step": 2374
    },
    {
      "epoch": 0.3210923901103544,
      "grad_norm": 0.06482286006212234,
      "learning_rate": 2.4904009429855992e-05,
      "loss": 0.049068570137023926,
      "step": 2375
    },
    {
      "epoch": 0.3212275869061903,
      "grad_norm": 0.12293051183223724,
      "learning_rate": 2.4898868023180844e-05,
      "loss": 0.040645599365234375,
      "step": 2376
    },
    {
      "epoch": 0.32136278370202626,
      "grad_norm": 0.03160116448998451,
      "learning_rate": 2.4893724555508575e-05,
      "loss": 0.03198385238647461,
      "step": 2377
    },
    {
      "epoch": 0.3214979804978622,
      "grad_norm": 0.1189572736620903,
      "learning_rate": 2.4888579027910105e-05,
      "loss": 0.052210330963134766,
      "step": 2378
    },
    {
      "epoch": 0.3216331772936981,
      "grad_norm": 0.057750605046749115,
      "learning_rate": 2.4883431441456738e-05,
      "loss": 0.05643749237060547,
      "step": 2379
    },
    {
      "epoch": 0.32176837408953407,
      "grad_norm": 0.08462224155664444,
      "learning_rate": 2.4878281797220244e-05,
      "loss": 0.08265113830566406,
      "step": 2380
    },
    {
      "epoch": 0.32190357088537,
      "grad_norm": 0.1276833564043045,
      "learning_rate": 2.4873130096272805e-05,
      "loss": 0.061980247497558594,
      "step": 2381
    },
    {
      "epoch": 0.32203876768120593,
      "grad_norm": 0.2143513560295105,
      "learning_rate": 2.4867976339687026e-05,
      "loss": 0.08031105995178223,
      "step": 2382
    },
    {
      "epoch": 0.3221739644770419,
      "grad_norm": 0.03482074290513992,
      "learning_rate": 2.4862820528535955e-05,
      "loss": 0.04183077812194824,
      "step": 2383
    },
    {
      "epoch": 0.32230916127287784,
      "grad_norm": 0.07347599416971207,
      "learning_rate": 2.4857662663893054e-05,
      "loss": 0.09386968612670898,
      "step": 2384
    },
    {
      "epoch": 0.3224443580687138,
      "grad_norm": 0.09324394911527634,
      "learning_rate": 2.485250274683222e-05,
      "loss": 0.06085968017578125,
      "step": 2385
    },
    {
      "epoch": 0.3225795548645497,
      "grad_norm": 0.05152599513530731,
      "learning_rate": 2.4847340778427772e-05,
      "loss": 0.05889081954956055,
      "step": 2386
    },
    {
      "epoch": 0.32271475166038566,
      "grad_norm": 0.0555427223443985,
      "learning_rate": 2.484217675975446e-05,
      "loss": 0.039900779724121094,
      "step": 2387
    },
    {
      "epoch": 0.3228499484562216,
      "grad_norm": 0.07489274442195892,
      "learning_rate": 2.4837010691887466e-05,
      "loss": 0.05003166198730469,
      "step": 2388
    },
    {
      "epoch": 0.3229851452520575,
      "grad_norm": 0.06637747585773468,
      "learning_rate": 2.4831842575902383e-05,
      "loss": 0.08872222900390625,
      "step": 2389
    },
    {
      "epoch": 0.32312034204789347,
      "grad_norm": 0.05388886108994484,
      "learning_rate": 2.482667241287525e-05,
      "loss": 0.05496954917907715,
      "step": 2390
    },
    {
      "epoch": 0.32325553884372943,
      "grad_norm": 0.059940699487924576,
      "learning_rate": 2.4821500203882517e-05,
      "loss": 0.0766594409942627,
      "step": 2391
    },
    {
      "epoch": 0.32339073563956533,
      "grad_norm": 0.11315230280160904,
      "learning_rate": 2.4816325950001067e-05,
      "loss": 0.1268310546875,
      "step": 2392
    },
    {
      "epoch": 0.3235259324354013,
      "grad_norm": 0.09025148302316666,
      "learning_rate": 2.4811149652308205e-05,
      "loss": 0.06718683242797852,
      "step": 2393
    },
    {
      "epoch": 0.32366112923123724,
      "grad_norm": 0.1144208163022995,
      "learning_rate": 2.480597131188167e-05,
      "loss": 0.08830738067626953,
      "step": 2394
    },
    {
      "epoch": 0.32379632602707314,
      "grad_norm": 0.10639499872922897,
      "learning_rate": 2.4800790929799614e-05,
      "loss": 0.07596683502197266,
      "step": 2395
    },
    {
      "epoch": 0.3239315228229091,
      "grad_norm": 0.05590733513236046,
      "learning_rate": 2.4795608507140623e-05,
      "loss": 0.03989553451538086,
      "step": 2396
    },
    {
      "epoch": 0.32406671961874506,
      "grad_norm": 0.178768128156662,
      "learning_rate": 2.4790424044983705e-05,
      "loss": 0.0841226577758789,
      "step": 2397
    },
    {
      "epoch": 0.32420191641458096,
      "grad_norm": 0.09750822186470032,
      "learning_rate": 2.4785237544408288e-05,
      "loss": 0.08683586120605469,
      "step": 2398
    },
    {
      "epoch": 0.3243371132104169,
      "grad_norm": 0.2906476557254791,
      "learning_rate": 2.478004900649424e-05,
      "loss": 0.07411885261535645,
      "step": 2399
    },
    {
      "epoch": 0.3244723100062529,
      "grad_norm": 0.1681055724620819,
      "learning_rate": 2.477485843232183e-05,
      "loss": 0.05577874183654785,
      "step": 2400
    },
    {
      "epoch": 0.3246075068020888,
      "grad_norm": 0.055169619619846344,
      "learning_rate": 2.476966582297177e-05,
      "loss": 0.045427799224853516,
      "step": 2401
    },
    {
      "epoch": 0.32474270359792473,
      "grad_norm": 0.09917508810758591,
      "learning_rate": 2.4764471179525188e-05,
      "loss": 0.06054329872131348,
      "step": 2402
    },
    {
      "epoch": 0.3248779003937607,
      "grad_norm": 0.17921628057956696,
      "learning_rate": 2.4759274503063632e-05,
      "loss": 0.12224006652832031,
      "step": 2403
    },
    {
      "epoch": 0.3250130971895966,
      "grad_norm": 0.10945697873830795,
      "learning_rate": 2.4754075794669088e-05,
      "loss": 0.06194591522216797,
      "step": 2404
    },
    {
      "epoch": 0.32514829398543255,
      "grad_norm": 0.14536145329475403,
      "learning_rate": 2.4748875055423942e-05,
      "loss": 0.09374499320983887,
      "step": 2405
    },
    {
      "epoch": 0.3252834907812685,
      "grad_norm": 0.17457795143127441,
      "learning_rate": 2.4743672286411027e-05,
      "loss": 0.055704593658447266,
      "step": 2406
    },
    {
      "epoch": 0.3254186875771044,
      "grad_norm": 0.07643047720193863,
      "learning_rate": 2.4738467488713582e-05,
      "loss": 0.03910708427429199,
      "step": 2407
    },
    {
      "epoch": 0.32555388437294036,
      "grad_norm": 0.07738658785820007,
      "learning_rate": 2.473326066341527e-05,
      "loss": 0.06092846393585205,
      "step": 2408
    },
    {
      "epoch": 0.3256890811687763,
      "grad_norm": 0.19982022047042847,
      "learning_rate": 2.4728051811600184e-05,
      "loss": 0.10982227325439453,
      "step": 2409
    },
    {
      "epoch": 0.3258242779646122,
      "grad_norm": 0.09709753096103668,
      "learning_rate": 2.4722840934352838e-05,
      "loss": 0.08084249496459961,
      "step": 2410
    },
    {
      "epoch": 0.3259594747604482,
      "grad_norm": 0.1849898248910904,
      "learning_rate": 2.471762803275816e-05,
      "loss": 0.08124637603759766,
      "step": 2411
    },
    {
      "epoch": 0.32609467155628413,
      "grad_norm": 0.35742610692977905,
      "learning_rate": 2.4712413107901504e-05,
      "loss": 0.08498001098632812,
      "step": 2412
    },
    {
      "epoch": 0.32622986835212003,
      "grad_norm": 0.15750548243522644,
      "learning_rate": 2.470719616086865e-05,
      "loss": 0.08361291885375977,
      "step": 2413
    },
    {
      "epoch": 0.326365065147956,
      "grad_norm": 0.0812111347913742,
      "learning_rate": 2.4701977192745785e-05,
      "loss": 0.0506134033203125,
      "step": 2414
    },
    {
      "epoch": 0.32650026194379195,
      "grad_norm": 0.08652845025062561,
      "learning_rate": 2.4696756204619535e-05,
      "loss": 0.06915473937988281,
      "step": 2415
    },
    {
      "epoch": 0.32663545873962785,
      "grad_norm": 0.0650532990694046,
      "learning_rate": 2.469153319757693e-05,
      "loss": 0.05527293682098389,
      "step": 2416
    },
    {
      "epoch": 0.3267706555354638,
      "grad_norm": 0.06082440912723541,
      "learning_rate": 2.4686308172705433e-05,
      "loss": 0.06781196594238281,
      "step": 2417
    },
    {
      "epoch": 0.32690585233129976,
      "grad_norm": 0.10656819492578506,
      "learning_rate": 2.4681081131092926e-05,
      "loss": 0.044648170471191406,
      "step": 2418
    },
    {
      "epoch": 0.32704104912713566,
      "grad_norm": 0.07517208158969879,
      "learning_rate": 2.467585207382769e-05,
      "loss": 0.07143878936767578,
      "step": 2419
    },
    {
      "epoch": 0.3271762459229716,
      "grad_norm": 0.07864128798246384,
      "learning_rate": 2.4670621001998467e-05,
      "loss": 0.0643090009689331,
      "step": 2420
    },
    {
      "epoch": 0.3273114427188076,
      "grad_norm": 0.09423500299453735,
      "learning_rate": 2.466538791669437e-05,
      "loss": 0.050605058670043945,
      "step": 2421
    },
    {
      "epoch": 0.3274466395146435,
      "grad_norm": 0.08596308529376984,
      "learning_rate": 2.4660152819004973e-05,
      "loss": 0.06726741790771484,
      "step": 2422
    },
    {
      "epoch": 0.32758183631047944,
      "grad_norm": 0.08023789525032043,
      "learning_rate": 2.4654915710020246e-05,
      "loss": 0.0673479437828064,
      "step": 2423
    },
    {
      "epoch": 0.3277170331063154,
      "grad_norm": 0.06163188815116882,
      "learning_rate": 2.464967659083058e-05,
      "loss": 0.06528234481811523,
      "step": 2424
    },
    {
      "epoch": 0.3278522299021513,
      "grad_norm": 0.05535079538822174,
      "learning_rate": 2.464443546252679e-05,
      "loss": 0.0523221492767334,
      "step": 2425
    },
    {
      "epoch": 0.32798742669798725,
      "grad_norm": 0.08049971610307693,
      "learning_rate": 2.4639192326200104e-05,
      "loss": 0.07275652885437012,
      "step": 2426
    },
    {
      "epoch": 0.3281226234938232,
      "grad_norm": 0.2276969701051712,
      "learning_rate": 2.463394718294218e-05,
      "loss": 0.07629859447479248,
      "step": 2427
    },
    {
      "epoch": 0.3282578202896591,
      "grad_norm": 0.11948155611753464,
      "learning_rate": 2.4628700033845072e-05,
      "loss": 0.06624412536621094,
      "step": 2428
    },
    {
      "epoch": 0.32839301708549506,
      "grad_norm": 0.16820286214351654,
      "learning_rate": 2.4623450880001268e-05,
      "loss": 0.04695701599121094,
      "step": 2429
    },
    {
      "epoch": 0.328528213881331,
      "grad_norm": 0.07818910479545593,
      "learning_rate": 2.4618199722503676e-05,
      "loss": 0.05118513107299805,
      "step": 2430
    },
    {
      "epoch": 0.328663410677167,
      "grad_norm": 0.15611349046230316,
      "learning_rate": 2.4612946562445613e-05,
      "loss": 0.07812213897705078,
      "step": 2431
    },
    {
      "epoch": 0.3287986074730029,
      "grad_norm": 0.07964304089546204,
      "learning_rate": 2.460769140092081e-05,
      "loss": 0.05789899826049805,
      "step": 2432
    },
    {
      "epoch": 0.32893380426883884,
      "grad_norm": 0.16050387918949127,
      "learning_rate": 2.460243423902342e-05,
      "loss": 0.06040549278259277,
      "step": 2433
    },
    {
      "epoch": 0.3290690010646748,
      "grad_norm": 0.06264235824346542,
      "learning_rate": 2.459717507784802e-05,
      "loss": 0.05280160903930664,
      "step": 2434
    },
    {
      "epoch": 0.3292041978605107,
      "grad_norm": 0.16657042503356934,
      "learning_rate": 2.459191391848959e-05,
      "loss": 0.08240318298339844,
      "step": 2435
    },
    {
      "epoch": 0.32933939465634665,
      "grad_norm": 0.09569832682609558,
      "learning_rate": 2.4586650762043538e-05,
      "loss": 0.06727290153503418,
      "step": 2436
    },
    {
      "epoch": 0.3294745914521826,
      "grad_norm": 0.07165045291185379,
      "learning_rate": 2.4581385609605665e-05,
      "loss": 0.060944557189941406,
      "step": 2437
    },
    {
      "epoch": 0.3296097882480185,
      "grad_norm": 0.06087822467088699,
      "learning_rate": 2.4576118462272218e-05,
      "loss": 0.06873607635498047,
      "step": 2438
    },
    {
      "epoch": 0.32974498504385447,
      "grad_norm": 0.24831995368003845,
      "learning_rate": 2.4570849321139836e-05,
      "loss": 0.09560632705688477,
      "step": 2439
    },
    {
      "epoch": 0.3298801818396904,
      "grad_norm": 0.10931834578514099,
      "learning_rate": 2.4565578187305596e-05,
      "loss": 0.05682849884033203,
      "step": 2440
    },
    {
      "epoch": 0.3300153786355263,
      "grad_norm": 0.08153682947158813,
      "learning_rate": 2.456030506186696e-05,
      "loss": 0.07766056060791016,
      "step": 2441
    },
    {
      "epoch": 0.3301505754313623,
      "grad_norm": 0.06826280057430267,
      "learning_rate": 2.4555029945921832e-05,
      "loss": 0.07746005058288574,
      "step": 2442
    },
    {
      "epoch": 0.33028577222719824,
      "grad_norm": 0.08454151451587677,
      "learning_rate": 2.4549752840568516e-05,
      "loss": 0.07732677459716797,
      "step": 2443
    },
    {
      "epoch": 0.33042096902303414,
      "grad_norm": 0.12010058760643005,
      "learning_rate": 2.4544473746905733e-05,
      "loss": 0.06531429290771484,
      "step": 2444
    },
    {
      "epoch": 0.3305561658188701,
      "grad_norm": 0.09992405027151108,
      "learning_rate": 2.4539192666032617e-05,
      "loss": 0.07818913459777832,
      "step": 2445
    },
    {
      "epoch": 0.33069136261470605,
      "grad_norm": 0.18263866007328033,
      "learning_rate": 2.4533909599048718e-05,
      "loss": 0.090087890625,
      "step": 2446
    },
    {
      "epoch": 0.33082655941054195,
      "grad_norm": 0.28081822395324707,
      "learning_rate": 2.4528624547054003e-05,
      "loss": 0.10488128662109375,
      "step": 2447
    },
    {
      "epoch": 0.3309617562063779,
      "grad_norm": 0.1193016767501831,
      "learning_rate": 2.4523337511148843e-05,
      "loss": 0.05753660202026367,
      "step": 2448
    },
    {
      "epoch": 0.33109695300221387,
      "grad_norm": 0.1351795643568039,
      "learning_rate": 2.4518048492434028e-05,
      "loss": 0.0748896598815918,
      "step": 2449
    },
    {
      "epoch": 0.33123214979804977,
      "grad_norm": 0.07759327441453934,
      "learning_rate": 2.4512757492010762e-05,
      "loss": 0.04728102684020996,
      "step": 2450
    },
    {
      "epoch": 0.3313673465938857,
      "grad_norm": 0.07855939120054245,
      "learning_rate": 2.4507464510980652e-05,
      "loss": 0.037827253341674805,
      "step": 2451
    },
    {
      "epoch": 0.3315025433897217,
      "grad_norm": 0.04127610847353935,
      "learning_rate": 2.450216955044574e-05,
      "loss": 0.05664873123168945,
      "step": 2452
    },
    {
      "epoch": 0.3316377401855576,
      "grad_norm": 0.11055218428373337,
      "learning_rate": 2.449687261150845e-05,
      "loss": 0.07512664794921875,
      "step": 2453
    },
    {
      "epoch": 0.33177293698139354,
      "grad_norm": 0.06807762384414673,
      "learning_rate": 2.449157369527164e-05,
      "loss": 0.03923296928405762,
      "step": 2454
    },
    {
      "epoch": 0.3319081337772295,
      "grad_norm": 0.07681751996278763,
      "learning_rate": 2.448627280283857e-05,
      "loss": 0.05240321159362793,
      "step": 2455
    },
    {
      "epoch": 0.3320433305730654,
      "grad_norm": 0.053800538182258606,
      "learning_rate": 2.4480969935312917e-05,
      "loss": 0.04906940460205078,
      "step": 2456
    },
    {
      "epoch": 0.33217852736890136,
      "grad_norm": 0.12979096174240112,
      "learning_rate": 2.4475665093798766e-05,
      "loss": 0.08299636840820312,
      "step": 2457
    },
    {
      "epoch": 0.3323137241647373,
      "grad_norm": 0.05039919912815094,
      "learning_rate": 2.447035827940061e-05,
      "loss": 0.04596900939941406,
      "step": 2458
    },
    {
      "epoch": 0.3324489209605732,
      "grad_norm": 0.35557109117507935,
      "learning_rate": 2.4465049493223356e-05,
      "loss": 0.09499406814575195,
      "step": 2459
    },
    {
      "epoch": 0.33258411775640917,
      "grad_norm": 0.058588236570358276,
      "learning_rate": 2.4459738736372327e-05,
      "loss": 0.06646537780761719,
      "step": 2460
    },
    {
      "epoch": 0.3327193145522451,
      "grad_norm": 0.22273430228233337,
      "learning_rate": 2.4454426009953252e-05,
      "loss": 0.09099411964416504,
      "step": 2461
    },
    {
      "epoch": 0.33285451134808103,
      "grad_norm": 0.02887437492609024,
      "learning_rate": 2.4449111315072254e-05,
      "loss": 0.029601454734802246,
      "step": 2462
    },
    {
      "epoch": 0.332989708143917,
      "grad_norm": 0.03981492668390274,
      "learning_rate": 2.44437946528359e-05,
      "loss": 0.02615034580230713,
      "step": 2463
    },
    {
      "epoch": 0.33312490493975294,
      "grad_norm": 0.12644590437412262,
      "learning_rate": 2.4438476024351138e-05,
      "loss": 0.049398183822631836,
      "step": 2464
    },
    {
      "epoch": 0.33326010173558884,
      "grad_norm": 0.19335535168647766,
      "learning_rate": 2.4433155430725333e-05,
      "loss": 0.07953166961669922,
      "step": 2465
    },
    {
      "epoch": 0.3333952985314248,
      "grad_norm": 0.09825469553470612,
      "learning_rate": 2.4427832873066262e-05,
      "loss": 0.11213040351867676,
      "step": 2466
    },
    {
      "epoch": 0.33353049532726076,
      "grad_norm": 0.07603967934846878,
      "learning_rate": 2.4422508352482113e-05,
      "loss": 0.06043732166290283,
      "step": 2467
    },
    {
      "epoch": 0.33366569212309666,
      "grad_norm": 0.039789531379938126,
      "learning_rate": 2.441718187008148e-05,
      "loss": 0.037857770919799805,
      "step": 2468
    },
    {
      "epoch": 0.3338008889189326,
      "grad_norm": 0.07418537139892578,
      "learning_rate": 2.441185342697336e-05,
      "loss": 0.043689846992492676,
      "step": 2469
    },
    {
      "epoch": 0.33393608571476857,
      "grad_norm": 0.09936017543077469,
      "learning_rate": 2.440652302426717e-05,
      "loss": 0.08456897735595703,
      "step": 2470
    },
    {
      "epoch": 0.3340712825106045,
      "grad_norm": 0.0806683599948883,
      "learning_rate": 2.440119066307272e-05,
      "loss": 0.06488752365112305,
      "step": 2471
    },
    {
      "epoch": 0.33420647930644043,
      "grad_norm": 0.08230933547019958,
      "learning_rate": 2.4395856344500244e-05,
      "loss": 0.08169841766357422,
      "step": 2472
    },
    {
      "epoch": 0.3343416761022764,
      "grad_norm": 0.08933542668819427,
      "learning_rate": 2.4390520069660377e-05,
      "loss": 0.07372283935546875,
      "step": 2473
    },
    {
      "epoch": 0.3344768728981123,
      "grad_norm": 0.18069317936897278,
      "learning_rate": 2.4385181839664146e-05,
      "loss": 0.07696413993835449,
      "step": 2474
    },
    {
      "epoch": 0.33461206969394824,
      "grad_norm": 0.041141998022794724,
      "learning_rate": 2.437984165562301e-05,
      "loss": 0.05206632614135742,
      "step": 2475
    },
    {
      "epoch": 0.3347472664897842,
      "grad_norm": 0.10608562082052231,
      "learning_rate": 2.4374499518648827e-05,
      "loss": 0.08230805397033691,
      "step": 2476
    },
    {
      "epoch": 0.33488246328562016,
      "grad_norm": 0.13384272158145905,
      "learning_rate": 2.436915542985385e-05,
      "loss": 0.09208157658576965,
      "step": 2477
    },
    {
      "epoch": 0.33501766008145606,
      "grad_norm": 0.1401708722114563,
      "learning_rate": 2.436380939035075e-05,
      "loss": 0.0727243423461914,
      "step": 2478
    },
    {
      "epoch": 0.335152856877292,
      "grad_norm": 0.0704093724489212,
      "learning_rate": 2.43584614012526e-05,
      "loss": 0.05974757671356201,
      "step": 2479
    },
    {
      "epoch": 0.335288053673128,
      "grad_norm": 0.08460766077041626,
      "learning_rate": 2.4353111463672882e-05,
      "loss": 0.054102540016174316,
      "step": 2480
    },
    {
      "epoch": 0.3354232504689639,
      "grad_norm": 0.0492757223546505,
      "learning_rate": 2.4347759578725482e-05,
      "loss": 0.040642738342285156,
      "step": 2481
    },
    {
      "epoch": 0.33555844726479983,
      "grad_norm": 0.07080303132534027,
      "learning_rate": 2.4342405747524685e-05,
      "loss": 0.06881427764892578,
      "step": 2482
    },
    {
      "epoch": 0.3356936440606358,
      "grad_norm": 0.1043875589966774,
      "learning_rate": 2.4337049971185194e-05,
      "loss": 0.09662342071533203,
      "step": 2483
    },
    {
      "epoch": 0.3358288408564717,
      "grad_norm": 0.07296701520681381,
      "learning_rate": 2.433169225082211e-05,
      "loss": 0.057660818099975586,
      "step": 2484
    },
    {
      "epoch": 0.33596403765230765,
      "grad_norm": 0.07778462767601013,
      "learning_rate": 2.432633258755093e-05,
      "loss": 0.06607341766357422,
      "step": 2485
    },
    {
      "epoch": 0.3360992344481436,
      "grad_norm": 0.09076784551143646,
      "learning_rate": 2.432097098248758e-05,
      "loss": 0.06475687026977539,
      "step": 2486
    },
    {
      "epoch": 0.3362344312439795,
      "grad_norm": 0.09413054585456848,
      "learning_rate": 2.4315607436748362e-05,
      "loss": 0.053586483001708984,
      "step": 2487
    },
    {
      "epoch": 0.33636962803981546,
      "grad_norm": 0.08732490241527557,
      "learning_rate": 2.4310241951449997e-05,
      "loss": 0.049224853515625,
      "step": 2488
    },
    {
      "epoch": 0.3365048248356514,
      "grad_norm": 0.09103147685527802,
      "learning_rate": 2.4304874527709614e-05,
      "loss": 0.05713611841201782,
      "step": 2489
    },
    {
      "epoch": 0.3366400216314873,
      "grad_norm": 0.06709442287683487,
      "learning_rate": 2.429950516664473e-05,
      "loss": 0.05126011371612549,
      "step": 2490
    },
    {
      "epoch": 0.3367752184273233,
      "grad_norm": 0.12592142820358276,
      "learning_rate": 2.4294133869373284e-05,
      "loss": 0.05962419509887695,
      "step": 2491
    },
    {
      "epoch": 0.33691041522315923,
      "grad_norm": 0.055713240057229996,
      "learning_rate": 2.42887606370136e-05,
      "loss": 0.03740692138671875,
      "step": 2492
    },
    {
      "epoch": 0.33704561201899513,
      "grad_norm": 0.10131179541349411,
      "learning_rate": 2.428338547068442e-05,
      "loss": 0.057768821716308594,
      "step": 2493
    },
    {
      "epoch": 0.3371808088148311,
      "grad_norm": 0.0491848848760128,
      "learning_rate": 2.427800837150488e-05,
      "loss": 0.03424358367919922,
      "step": 2494
    },
    {
      "epoch": 0.33731600561066705,
      "grad_norm": 0.09536954760551453,
      "learning_rate": 2.4272629340594518e-05,
      "loss": 0.07404160499572754,
      "step": 2495
    },
    {
      "epoch": 0.33745120240650295,
      "grad_norm": 0.05238153040409088,
      "learning_rate": 2.426724837907328e-05,
      "loss": 0.06235027313232422,
      "step": 2496
    },
    {
      "epoch": 0.3375863992023389,
      "grad_norm": 0.037997812032699585,
      "learning_rate": 2.4261865488061512e-05,
      "loss": 0.03674960136413574,
      "step": 2497
    },
    {
      "epoch": 0.33772159599817486,
      "grad_norm": 0.08411721885204315,
      "learning_rate": 2.4256480668679958e-05,
      "loss": 0.031391799449920654,
      "step": 2498
    },
    {
      "epoch": 0.33785679279401076,
      "grad_norm": 0.06932103633880615,
      "learning_rate": 2.4251093922049766e-05,
      "loss": 0.052609920501708984,
      "step": 2499
    },
    {
      "epoch": 0.3379919895898467,
      "grad_norm": 0.08212162554264069,
      "learning_rate": 2.4245705249292494e-05,
      "loss": 0.06226414442062378,
      "step": 2500
    },
    {
      "epoch": 0.3381271863856827,
      "grad_norm": 0.17662891745567322,
      "learning_rate": 2.4240314651530073e-05,
      "loss": 0.07156133651733398,
      "step": 2501
    },
    {
      "epoch": 0.3382623831815186,
      "grad_norm": 0.07636614143848419,
      "learning_rate": 2.4234922129884873e-05,
      "loss": 0.06734883785247803,
      "step": 2502
    },
    {
      "epoch": 0.33839757997735453,
      "grad_norm": 0.09068209677934647,
      "learning_rate": 2.4229527685479644e-05,
      "loss": 0.058286190032958984,
      "step": 2503
    },
    {
      "epoch": 0.3385327767731905,
      "grad_norm": 0.10690157860517502,
      "learning_rate": 2.4224131319437523e-05,
      "loss": 0.09006214141845703,
      "step": 2504
    },
    {
      "epoch": 0.3386679735690264,
      "grad_norm": 0.15770073235034943,
      "learning_rate": 2.421873303288208e-05,
      "loss": 0.07337117195129395,
      "step": 2505
    },
    {
      "epoch": 0.33880317036486235,
      "grad_norm": 0.20079083740711212,
      "learning_rate": 2.4213332826937255e-05,
      "loss": 0.07252883911132812,
      "step": 2506
    },
    {
      "epoch": 0.3389383671606983,
      "grad_norm": 0.06770540773868561,
      "learning_rate": 2.4207930702727404e-05,
      "loss": 0.06073760986328125,
      "step": 2507
    },
    {
      "epoch": 0.3390735639565342,
      "grad_norm": 0.23615293204784393,
      "learning_rate": 2.420252666137728e-05,
      "loss": 0.09350407123565674,
      "step": 2508
    },
    {
      "epoch": 0.33920876075237016,
      "grad_norm": 0.08327589184045792,
      "learning_rate": 2.419712070401203e-05,
      "loss": 0.06562519073486328,
      "step": 2509
    },
    {
      "epoch": 0.3393439575482061,
      "grad_norm": 0.072517529129982,
      "learning_rate": 2.4191712831757203e-05,
      "loss": 0.06211280822753906,
      "step": 2510
    },
    {
      "epoch": 0.339479154344042,
      "grad_norm": 0.21241813898086548,
      "learning_rate": 2.418630304573875e-05,
      "loss": 0.0859375,
      "step": 2511
    },
    {
      "epoch": 0.339614351139878,
      "grad_norm": 0.11263559013605118,
      "learning_rate": 2.418089134708302e-05,
      "loss": 0.05048847198486328,
      "step": 2512
    },
    {
      "epoch": 0.33974954793571394,
      "grad_norm": 0.13258202373981476,
      "learning_rate": 2.4175477736916743e-05,
      "loss": 0.05666661262512207,
      "step": 2513
    },
    {
      "epoch": 0.33988474473154984,
      "grad_norm": 0.09894821047782898,
      "learning_rate": 2.4170062216367082e-05,
      "loss": 0.059726715087890625,
      "step": 2514
    },
    {
      "epoch": 0.3400199415273858,
      "grad_norm": 0.1272960901260376,
      "learning_rate": 2.416464478656156e-05,
      "loss": 0.07559537887573242,
      "step": 2515
    },
    {
      "epoch": 0.34015513832322175,
      "grad_norm": 0.2235560417175293,
      "learning_rate": 2.4159225448628123e-05,
      "loss": 0.0636606216430664,
      "step": 2516
    },
    {
      "epoch": 0.34029033511905765,
      "grad_norm": 0.11577052623033524,
      "learning_rate": 2.4153804203695103e-05,
      "loss": 0.07288074493408203,
      "step": 2517
    },
    {
      "epoch": 0.3404255319148936,
      "grad_norm": 0.22524523735046387,
      "learning_rate": 2.4148381052891236e-05,
      "loss": 0.08819818496704102,
      "step": 2518
    },
    {
      "epoch": 0.34056072871072957,
      "grad_norm": 0.2137891799211502,
      "learning_rate": 2.4142955997345648e-05,
      "loss": 0.07626008987426758,
      "step": 2519
    },
    {
      "epoch": 0.34069592550656547,
      "grad_norm": 0.08030815422534943,
      "learning_rate": 2.4137529038187864e-05,
      "loss": 0.05479228496551514,
      "step": 2520
    },
    {
      "epoch": 0.3408311223024014,
      "grad_norm": 0.20790722966194153,
      "learning_rate": 2.413210017654781e-05,
      "loss": 0.07788324356079102,
      "step": 2521
    },
    {
      "epoch": 0.3409663190982374,
      "grad_norm": 0.24542152881622314,
      "learning_rate": 2.4126669413555802e-05,
      "loss": 0.0837564468383789,
      "step": 2522
    },
    {
      "epoch": 0.34110151589407334,
      "grad_norm": 0.12429172545671463,
      "learning_rate": 2.4121236750342548e-05,
      "loss": 0.038735151290893555,
      "step": 2523
    },
    {
      "epoch": 0.34123671268990924,
      "grad_norm": 0.256369411945343,
      "learning_rate": 2.4115802188039165e-05,
      "loss": 0.07303023338317871,
      "step": 2524
    },
    {
      "epoch": 0.3413719094857452,
      "grad_norm": 0.1187695786356926,
      "learning_rate": 2.4110365727777156e-05,
      "loss": 0.037609100341796875,
      "step": 2525
    },
    {
      "epoch": 0.34150710628158115,
      "grad_norm": 0.11538417637348175,
      "learning_rate": 2.410492737068842e-05,
      "loss": 0.07685530185699463,
      "step": 2526
    },
    {
      "epoch": 0.34164230307741705,
      "grad_norm": 0.09548041969537735,
      "learning_rate": 2.409948711790525e-05,
      "loss": 0.06582450866699219,
      "step": 2527
    },
    {
      "epoch": 0.341777499873253,
      "grad_norm": 0.3541737198829651,
      "learning_rate": 2.4094044970560336e-05,
      "loss": 0.08763790130615234,
      "step": 2528
    },
    {
      "epoch": 0.34191269666908897,
      "grad_norm": 0.25250816345214844,
      "learning_rate": 2.4088600929786767e-05,
      "loss": 0.07401752471923828,
      "step": 2529
    },
    {
      "epoch": 0.34204789346492487,
      "grad_norm": 0.05376897752285004,
      "learning_rate": 2.408315499671802e-05,
      "loss": 0.041963815689086914,
      "step": 2530
    },
    {
      "epoch": 0.3421830902607608,
      "grad_norm": 0.05698872357606888,
      "learning_rate": 2.407770717248796e-05,
      "loss": 0.06352472305297852,
      "step": 2531
    },
    {
      "epoch": 0.3423182870565968,
      "grad_norm": 0.2583656907081604,
      "learning_rate": 2.407225745823086e-05,
      "loss": 0.07767629623413086,
      "step": 2532
    },
    {
      "epoch": 0.3424534838524327,
      "grad_norm": 0.11859942227602005,
      "learning_rate": 2.4066805855081378e-05,
      "loss": 0.0802459716796875,
      "step": 2533
    },
    {
      "epoch": 0.34258868064826864,
      "grad_norm": 0.20530535280704498,
      "learning_rate": 2.406135236417457e-05,
      "loss": 0.056284546852111816,
      "step": 2534
    },
    {
      "epoch": 0.3427238774441046,
      "grad_norm": 0.2839643955230713,
      "learning_rate": 2.4055896986645875e-05,
      "loss": 0.07913827896118164,
      "step": 2535
    },
    {
      "epoch": 0.3428590742399405,
      "grad_norm": 0.27108123898506165,
      "learning_rate": 2.4050439723631136e-05,
      "loss": 0.07863044738769531,
      "step": 2536
    },
    {
      "epoch": 0.34299427103577645,
      "grad_norm": 0.14025717973709106,
      "learning_rate": 2.404498057626659e-05,
      "loss": 0.07633376121520996,
      "step": 2537
    },
    {
      "epoch": 0.3431294678316124,
      "grad_norm": 0.21891368925571442,
      "learning_rate": 2.4039519545688848e-05,
      "loss": 0.07482755184173584,
      "step": 2538
    },
    {
      "epoch": 0.3432646646274483,
      "grad_norm": 0.07436855882406235,
      "learning_rate": 2.4034056633034932e-05,
      "loss": 0.06477475166320801,
      "step": 2539
    },
    {
      "epoch": 0.34339986142328427,
      "grad_norm": 0.11428110301494598,
      "learning_rate": 2.402859183944225e-05,
      "loss": 0.07420492172241211,
      "step": 2540
    },
    {
      "epoch": 0.3435350582191202,
      "grad_norm": 0.11608269065618515,
      "learning_rate": 2.4023125166048597e-05,
      "loss": 0.07308387756347656,
      "step": 2541
    },
    {
      "epoch": 0.3436702550149561,
      "grad_norm": 0.3174440860748291,
      "learning_rate": 2.401765661399218e-05,
      "loss": 0.08926451206207275,
      "step": 2542
    },
    {
      "epoch": 0.3438054518107921,
      "grad_norm": 0.13375575840473175,
      "learning_rate": 2.4012186184411556e-05,
      "loss": 0.0831151008605957,
      "step": 2543
    },
    {
      "epoch": 0.34394064860662804,
      "grad_norm": 0.12427858263254166,
      "learning_rate": 2.400671387844571e-05,
      "loss": 0.05663442611694336,
      "step": 2544
    },
    {
      "epoch": 0.34407584540246394,
      "grad_norm": 0.07700292766094208,
      "learning_rate": 2.4001239697234008e-05,
      "loss": 0.06965017318725586,
      "step": 2545
    },
    {
      "epoch": 0.3442110421982999,
      "grad_norm": 0.09709370881319046,
      "learning_rate": 2.3995763641916205e-05,
      "loss": 0.038106679916381836,
      "step": 2546
    },
    {
      "epoch": 0.34434623899413586,
      "grad_norm": 0.184953510761261,
      "learning_rate": 2.3990285713632436e-05,
      "loss": 0.0747835636138916,
      "step": 2547
    },
    {
      "epoch": 0.34448143578997176,
      "grad_norm": 0.1404159814119339,
      "learning_rate": 2.398480591352324e-05,
      "loss": 0.0558781623840332,
      "step": 2548
    },
    {
      "epoch": 0.3446166325858077,
      "grad_norm": 0.0346747562289238,
      "learning_rate": 2.3979324242729537e-05,
      "loss": 0.03399914503097534,
      "step": 2549
    },
    {
      "epoch": 0.34475182938164367,
      "grad_norm": 0.08864232152700424,
      "learning_rate": 2.3973840702392646e-05,
      "loss": 0.09303855895996094,
      "step": 2550
    },
    {
      "epoch": 0.34488702617747957,
      "grad_norm": 0.154584139585495,
      "learning_rate": 2.3968355293654267e-05,
      "loss": 0.06567645072937012,
      "step": 2551
    },
    {
      "epoch": 0.34502222297331553,
      "grad_norm": 0.17533770203590393,
      "learning_rate": 2.396286801765649e-05,
      "loss": 0.07946348190307617,
      "step": 2552
    },
    {
      "epoch": 0.3451574197691515,
      "grad_norm": 0.11433087289333344,
      "learning_rate": 2.3957378875541795e-05,
      "loss": 0.06172299385070801,
      "step": 2553
    },
    {
      "epoch": 0.3452926165649874,
      "grad_norm": 0.08917225152254105,
      "learning_rate": 2.395188786845305e-05,
      "loss": 0.05142402648925781,
      "step": 2554
    },
    {
      "epoch": 0.34542781336082334,
      "grad_norm": 0.11143828928470612,
      "learning_rate": 2.3946394997533516e-05,
      "loss": 0.0695028305053711,
      "step": 2555
    },
    {
      "epoch": 0.3455630101566593,
      "grad_norm": 0.19000935554504395,
      "learning_rate": 2.3940900263926833e-05,
      "loss": 0.11290311813354492,
      "step": 2556
    },
    {
      "epoch": 0.3456982069524952,
      "grad_norm": 0.09674094617366791,
      "learning_rate": 2.393540366877704e-05,
      "loss": 0.06610798835754395,
      "step": 2557
    },
    {
      "epoch": 0.34583340374833116,
      "grad_norm": 0.057298630475997925,
      "learning_rate": 2.392990521322855e-05,
      "loss": 0.04901587963104248,
      "step": 2558
    },
    {
      "epoch": 0.3459686005441671,
      "grad_norm": 0.09472622722387314,
      "learning_rate": 2.392440489842618e-05,
      "loss": 0.053554654121398926,
      "step": 2559
    },
    {
      "epoch": 0.346103797340003,
      "grad_norm": 0.2285137176513672,
      "learning_rate": 2.3918902725515118e-05,
      "loss": 0.04955458641052246,
      "step": 2560
    },
    {
      "epoch": 0.346238994135839,
      "grad_norm": 0.10635851323604584,
      "learning_rate": 2.391339869564094e-05,
      "loss": 0.09121417999267578,
      "step": 2561
    },
    {
      "epoch": 0.34637419093167493,
      "grad_norm": 0.19390016794204712,
      "learning_rate": 2.3907892809949628e-05,
      "loss": 0.079193115234375,
      "step": 2562
    },
    {
      "epoch": 0.34650938772751083,
      "grad_norm": 0.07630050927400589,
      "learning_rate": 2.390238506958753e-05,
      "loss": 0.0725393295288086,
      "step": 2563
    },
    {
      "epoch": 0.3466445845233468,
      "grad_norm": 0.05220980942249298,
      "learning_rate": 2.3896875475701387e-05,
      "loss": 0.043163299560546875,
      "step": 2564
    },
    {
      "epoch": 0.34677978131918274,
      "grad_norm": 0.14900819957256317,
      "learning_rate": 2.3891364029438323e-05,
      "loss": 0.06634807586669922,
      "step": 2565
    },
    {
      "epoch": 0.3469149781150187,
      "grad_norm": 0.30286872386932373,
      "learning_rate": 2.3885850731945857e-05,
      "loss": 0.09769010543823242,
      "step": 2566
    },
    {
      "epoch": 0.3470501749108546,
      "grad_norm": 0.2032707780599594,
      "learning_rate": 2.3880335584371884e-05,
      "loss": 0.08674049377441406,
      "step": 2567
    },
    {
      "epoch": 0.34718537170669056,
      "grad_norm": 0.06512822210788727,
      "learning_rate": 2.387481858786468e-05,
      "loss": 0.06437945365905762,
      "step": 2568
    },
    {
      "epoch": 0.3473205685025265,
      "grad_norm": 0.06642600893974304,
      "learning_rate": 2.386929974357293e-05,
      "loss": 0.0353083610534668,
      "step": 2569
    },
    {
      "epoch": 0.3474557652983624,
      "grad_norm": 0.22962428629398346,
      "learning_rate": 2.386377905264567e-05,
      "loss": 0.07955026626586914,
      "step": 2570
    },
    {
      "epoch": 0.3475909620941984,
      "grad_norm": 0.09827073663473129,
      "learning_rate": 2.3858256516232346e-05,
      "loss": 0.054482460021972656,
      "step": 2571
    },
    {
      "epoch": 0.34772615889003433,
      "grad_norm": 0.17681315541267395,
      "learning_rate": 2.3852732135482775e-05,
      "loss": 0.04503798484802246,
      "step": 2572
    },
    {
      "epoch": 0.34786135568587023,
      "grad_norm": 0.20490522682666779,
      "learning_rate": 2.3847205911547166e-05,
      "loss": 0.06481099128723145,
      "step": 2573
    },
    {
      "epoch": 0.3479965524817062,
      "grad_norm": 0.04753294214606285,
      "learning_rate": 2.3841677845576108e-05,
      "loss": 0.04919910430908203,
      "step": 2574
    },
    {
      "epoch": 0.34813174927754215,
      "grad_norm": 0.07276739925146103,
      "learning_rate": 2.383614793872057e-05,
      "loss": 0.042105913162231445,
      "step": 2575
    },
    {
      "epoch": 0.34826694607337805,
      "grad_norm": 0.05481953173875809,
      "learning_rate": 2.3830616192131913e-05,
      "loss": 0.06608033180236816,
      "step": 2576
    },
    {
      "epoch": 0.348402142869214,
      "grad_norm": 0.04325982928276062,
      "learning_rate": 2.3825082606961876e-05,
      "loss": 0.0615384578704834,
      "step": 2577
    },
    {
      "epoch": 0.34853733966504996,
      "grad_norm": 0.08279349654912949,
      "learning_rate": 2.3819547184362575e-05,
      "loss": 0.06657028198242188,
      "step": 2578
    },
    {
      "epoch": 0.34867253646088586,
      "grad_norm": 0.16935347020626068,
      "learning_rate": 2.3814009925486522e-05,
      "loss": 0.1102151870727539,
      "step": 2579
    },
    {
      "epoch": 0.3488077332567218,
      "grad_norm": 0.07301969826221466,
      "learning_rate": 2.38084708314866e-05,
      "loss": 0.04360055923461914,
      "step": 2580
    },
    {
      "epoch": 0.3489429300525578,
      "grad_norm": 0.2518143951892853,
      "learning_rate": 2.380292990351608e-05,
      "loss": 0.07262802124023438,
      "step": 2581
    },
    {
      "epoch": 0.3490781268483937,
      "grad_norm": 0.2129613310098648,
      "learning_rate": 2.3797387142728607e-05,
      "loss": 0.05638575553894043,
      "step": 2582
    },
    {
      "epoch": 0.34921332364422963,
      "grad_norm": 0.05965558439493179,
      "learning_rate": 2.379184255027822e-05,
      "loss": 0.06928062438964844,
      "step": 2583
    },
    {
      "epoch": 0.3493485204400656,
      "grad_norm": 0.04601344093680382,
      "learning_rate": 2.378629612731933e-05,
      "loss": 0.0615992546081543,
      "step": 2584
    },
    {
      "epoch": 0.3494837172359015,
      "grad_norm": 0.2759677767753601,
      "learning_rate": 2.3780747875006735e-05,
      "loss": 0.05584144592285156,
      "step": 2585
    },
    {
      "epoch": 0.34961891403173745,
      "grad_norm": 0.0999409556388855,
      "learning_rate": 2.37751977944956e-05,
      "loss": 0.07807350158691406,
      "step": 2586
    },
    {
      "epoch": 0.3497541108275734,
      "grad_norm": 0.3165963590145111,
      "learning_rate": 2.3769645886941497e-05,
      "loss": 0.09059429168701172,
      "step": 2587
    },
    {
      "epoch": 0.3498893076234093,
      "grad_norm": 0.1861148327589035,
      "learning_rate": 2.376409215350035e-05,
      "loss": 0.06132638454437256,
      "step": 2588
    },
    {
      "epoch": 0.35002450441924526,
      "grad_norm": 0.10965070128440857,
      "learning_rate": 2.3758536595328486e-05,
      "loss": 0.07873296737670898,
      "step": 2589
    },
    {
      "epoch": 0.3501597012150812,
      "grad_norm": 0.048266466706991196,
      "learning_rate": 2.375297921358259e-05,
      "loss": 0.046037644147872925,
      "step": 2590
    },
    {
      "epoch": 0.3502948980109171,
      "grad_norm": 0.06304934620857239,
      "learning_rate": 2.3747420009419745e-05,
      "loss": 0.07021570205688477,
      "step": 2591
    },
    {
      "epoch": 0.3504300948067531,
      "grad_norm": 0.0976991280913353,
      "learning_rate": 2.3741858983997415e-05,
      "loss": 0.04964160919189453,
      "step": 2592
    },
    {
      "epoch": 0.35056529160258904,
      "grad_norm": 0.12907104194164276,
      "learning_rate": 2.373629613847342e-05,
      "loss": 0.05902290344238281,
      "step": 2593
    },
    {
      "epoch": 0.35070048839842494,
      "grad_norm": 0.06459382176399231,
      "learning_rate": 2.3730731474005988e-05,
      "loss": 0.06478548049926758,
      "step": 2594
    },
    {
      "epoch": 0.3508356851942609,
      "grad_norm": 0.11330688744783401,
      "learning_rate": 2.37251649917537e-05,
      "loss": 0.04386448860168457,
      "step": 2595
    },
    {
      "epoch": 0.35097088199009685,
      "grad_norm": 0.19156739115715027,
      "learning_rate": 2.3719596692875534e-05,
      "loss": 0.05406102538108826,
      "step": 2596
    },
    {
      "epoch": 0.35110607878593275,
      "grad_norm": 0.09353367239236832,
      "learning_rate": 2.3714026578530836e-05,
      "loss": 0.05425453186035156,
      "step": 2597
    },
    {
      "epoch": 0.3512412755817687,
      "grad_norm": 0.09591637551784515,
      "learning_rate": 2.370845464987934e-05,
      "loss": 0.06257152557373047,
      "step": 2598
    },
    {
      "epoch": 0.35137647237760467,
      "grad_norm": 0.07256006449460983,
      "learning_rate": 2.370288090808114e-05,
      "loss": 0.0731973648071289,
      "step": 2599
    },
    {
      "epoch": 0.35151166917344057,
      "grad_norm": 0.08736106753349304,
      "learning_rate": 2.369730535429673e-05,
      "loss": 0.0936741828918457,
      "step": 2600
    },
    {
      "epoch": 0.3516468659692765,
      "grad_norm": 0.10256905853748322,
      "learning_rate": 2.369172798968697e-05,
      "loss": 0.0741281509399414,
      "step": 2601
    },
    {
      "epoch": 0.3517820627651125,
      "grad_norm": 0.14048469066619873,
      "learning_rate": 2.3686148815413083e-05,
      "loss": 0.0610661506652832,
      "step": 2602
    },
    {
      "epoch": 0.3519172595609484,
      "grad_norm": 0.05986129119992256,
      "learning_rate": 2.3680567832636695e-05,
      "loss": 0.04454541206359863,
      "step": 2603
    },
    {
      "epoch": 0.35205245635678434,
      "grad_norm": 0.10789672285318375,
      "learning_rate": 2.3674985042519795e-05,
      "loss": 0.08755874633789062,
      "step": 2604
    },
    {
      "epoch": 0.3521876531526203,
      "grad_norm": 0.05408640578389168,
      "learning_rate": 2.366940044622475e-05,
      "loss": 0.0374140739440918,
      "step": 2605
    },
    {
      "epoch": 0.3523228499484562,
      "grad_norm": 0.0820254236459732,
      "learning_rate": 2.3663814044914302e-05,
      "loss": 0.06505441665649414,
      "step": 2606
    },
    {
      "epoch": 0.35245804674429215,
      "grad_norm": 0.0746568962931633,
      "learning_rate": 2.3658225839751566e-05,
      "loss": 0.08930015563964844,
      "step": 2607
    },
    {
      "epoch": 0.3525932435401281,
      "grad_norm": 0.05777743086218834,
      "learning_rate": 2.3652635831900043e-05,
      "loss": 0.06343793869018555,
      "step": 2608
    },
    {
      "epoch": 0.352728440335964,
      "grad_norm": 0.08560241013765335,
      "learning_rate": 2.3647044022523595e-05,
      "loss": 0.09074974060058594,
      "step": 2609
    },
    {
      "epoch": 0.35286363713179997,
      "grad_norm": 0.08882021903991699,
      "learning_rate": 2.364145041278647e-05,
      "loss": 0.050927162170410156,
      "step": 2610
    },
    {
      "epoch": 0.3529988339276359,
      "grad_norm": 0.1459065079689026,
      "learning_rate": 2.3635855003853287e-05,
      "loss": 0.0890188217163086,
      "step": 2611
    },
    {
      "epoch": 0.3531340307234719,
      "grad_norm": 0.09583209455013275,
      "learning_rate": 2.363025779688904e-05,
      "loss": 0.05753755569458008,
      "step": 2612
    },
    {
      "epoch": 0.3532692275193078,
      "grad_norm": 0.05305046588182449,
      "learning_rate": 2.3624658793059103e-05,
      "loss": 0.051485300064086914,
      "step": 2613
    },
    {
      "epoch": 0.35340442431514374,
      "grad_norm": 0.05574241280555725,
      "learning_rate": 2.3619057993529204e-05,
      "loss": 0.041806817054748535,
      "step": 2614
    },
    {
      "epoch": 0.3535396211109797,
      "grad_norm": 0.04217883571982384,
      "learning_rate": 2.3613455399465475e-05,
      "loss": 0.04663550853729248,
      "step": 2615
    },
    {
      "epoch": 0.3536748179068156,
      "grad_norm": 0.0355115570127964,
      "learning_rate": 2.3607851012034394e-05,
      "loss": 0.04209566116333008,
      "step": 2616
    },
    {
      "epoch": 0.35381001470265155,
      "grad_norm": 0.10890758037567139,
      "learning_rate": 2.3602244832402838e-05,
      "loss": 0.06583833694458008,
      "step": 2617
    },
    {
      "epoch": 0.3539452114984875,
      "grad_norm": 0.14254949986934662,
      "learning_rate": 2.3596636861738024e-05,
      "loss": 0.062408447265625,
      "step": 2618
    },
    {
      "epoch": 0.3540804082943234,
      "grad_norm": 0.09131710231304169,
      "learning_rate": 2.3591027101207578e-05,
      "loss": 0.048073530197143555,
      "step": 2619
    },
    {
      "epoch": 0.35421560509015937,
      "grad_norm": 0.08882056921720505,
      "learning_rate": 2.3585415551979476e-05,
      "loss": 0.09334373474121094,
      "step": 2620
    },
    {
      "epoch": 0.3543508018859953,
      "grad_norm": 0.1009046882390976,
      "learning_rate": 2.3579802215222076e-05,
      "loss": 0.10957920551300049,
      "step": 2621
    },
    {
      "epoch": 0.3544859986818312,
      "grad_norm": 0.15729114413261414,
      "learning_rate": 2.35741870921041e-05,
      "loss": 0.06731593608856201,
      "step": 2622
    },
    {
      "epoch": 0.3546211954776672,
      "grad_norm": 0.19666698575019836,
      "learning_rate": 2.3568570183794645e-05,
      "loss": 0.04717528820037842,
      "step": 2623
    },
    {
      "epoch": 0.35475639227350314,
      "grad_norm": 0.09813804924488068,
      "learning_rate": 2.356295149146319e-05,
      "loss": 0.05095100402832031,
      "step": 2624
    },
    {
      "epoch": 0.35489158906933904,
      "grad_norm": 0.07537435740232468,
      "learning_rate": 2.3557331016279567e-05,
      "loss": 0.04773664474487305,
      "step": 2625
    },
    {
      "epoch": 0.355026785865175,
      "grad_norm": 0.10022012889385223,
      "learning_rate": 2.3551708759413998e-05,
      "loss": 0.086090087890625,
      "step": 2626
    },
    {
      "epoch": 0.35516198266101096,
      "grad_norm": 0.16503126919269562,
      "learning_rate": 2.354608472203706e-05,
      "loss": 0.06345498561859131,
      "step": 2627
    },
    {
      "epoch": 0.35529717945684686,
      "grad_norm": 0.07295048981904984,
      "learning_rate": 2.3540458905319705e-05,
      "loss": 0.04761457443237305,
      "step": 2628
    },
    {
      "epoch": 0.3554323762526828,
      "grad_norm": 0.08839529752731323,
      "learning_rate": 2.3534831310433264e-05,
      "loss": 0.07560676336288452,
      "step": 2629
    },
    {
      "epoch": 0.35556757304851877,
      "grad_norm": 0.10200095176696777,
      "learning_rate": 2.3529201938549434e-05,
      "loss": 0.0702185332775116,
      "step": 2630
    },
    {
      "epoch": 0.35570276984435467,
      "grad_norm": 0.05184308439493179,
      "learning_rate": 2.3523570790840274e-05,
      "loss": 0.04102891683578491,
      "step": 2631
    },
    {
      "epoch": 0.35583796664019063,
      "grad_norm": 0.23401685059070587,
      "learning_rate": 2.3517937868478228e-05,
      "loss": 0.05937385559082031,
      "step": 2632
    },
    {
      "epoch": 0.3559731634360266,
      "grad_norm": 0.15099912881851196,
      "learning_rate": 2.3512303172636092e-05,
      "loss": 0.06755036115646362,
      "step": 2633
    },
    {
      "epoch": 0.3561083602318625,
      "grad_norm": 0.07012380659580231,
      "learning_rate": 2.3506666704487033e-05,
      "loss": 0.0710906982421875,
      "step": 2634
    },
    {
      "epoch": 0.35624355702769844,
      "grad_norm": 0.14317156374454498,
      "learning_rate": 2.3501028465204614e-05,
      "loss": 0.08348512649536133,
      "step": 2635
    },
    {
      "epoch": 0.3563787538235344,
      "grad_norm": 0.0987994521856308,
      "learning_rate": 2.3495388455962734e-05,
      "loss": 0.06573677062988281,
      "step": 2636
    },
    {
      "epoch": 0.3565139506193703,
      "grad_norm": 0.12739720940589905,
      "learning_rate": 2.3489746677935673e-05,
      "loss": 0.0580441951751709,
      "step": 2637
    },
    {
      "epoch": 0.35664914741520626,
      "grad_norm": 0.23972631990909576,
      "learning_rate": 2.3484103132298082e-05,
      "loss": 0.06209278106689453,
      "step": 2638
    },
    {
      "epoch": 0.3567843442110422,
      "grad_norm": 0.08867891877889633,
      "learning_rate": 2.347845782022497e-05,
      "loss": 0.08517313003540039,
      "step": 2639
    },
    {
      "epoch": 0.3569195410068781,
      "grad_norm": 0.06530512869358063,
      "learning_rate": 2.3472810742891734e-05,
      "loss": 0.08191490173339844,
      "step": 2640
    },
    {
      "epoch": 0.3570547378027141,
      "grad_norm": 0.15433338284492493,
      "learning_rate": 2.3467161901474118e-05,
      "loss": 0.067047119140625,
      "step": 2641
    },
    {
      "epoch": 0.35718993459855003,
      "grad_norm": 0.06282784044742584,
      "learning_rate": 2.346151129714824e-05,
      "loss": 0.057749390602111816,
      "step": 2642
    },
    {
      "epoch": 0.35732513139438593,
      "grad_norm": 0.2099207192659378,
      "learning_rate": 2.3455858931090588e-05,
      "loss": 0.10221624374389648,
      "step": 2643
    },
    {
      "epoch": 0.3574603281902219,
      "grad_norm": 0.18579956889152527,
      "learning_rate": 2.3450204804478014e-05,
      "loss": 0.0904233455657959,
      "step": 2644
    },
    {
      "epoch": 0.35759552498605784,
      "grad_norm": 0.10013622045516968,
      "learning_rate": 2.344454891848774e-05,
      "loss": 0.055353641510009766,
      "step": 2645
    },
    {
      "epoch": 0.35773072178189375,
      "grad_norm": 0.07909790426492691,
      "learning_rate": 2.3438891274297348e-05,
      "loss": 0.07140171527862549,
      "step": 2646
    },
    {
      "epoch": 0.3578659185777297,
      "grad_norm": 0.0854388102889061,
      "learning_rate": 2.343323187308479e-05,
      "loss": 0.060881614685058594,
      "step": 2647
    },
    {
      "epoch": 0.35800111537356566,
      "grad_norm": 0.09499748051166534,
      "learning_rate": 2.342757071602839e-05,
      "loss": 0.08330488204956055,
      "step": 2648
    },
    {
      "epoch": 0.35813631216940156,
      "grad_norm": 0.11566922068595886,
      "learning_rate": 2.3421907804306816e-05,
      "loss": 0.10874366760253906,
      "step": 2649
    },
    {
      "epoch": 0.3582715089652375,
      "grad_norm": 0.15448719263076782,
      "learning_rate": 2.341624313909913e-05,
      "loss": 0.07314777374267578,
      "step": 2650
    },
    {
      "epoch": 0.3584067057610735,
      "grad_norm": 0.10072079300880432,
      "learning_rate": 2.3410576721584742e-05,
      "loss": 0.08783245086669922,
      "step": 2651
    },
    {
      "epoch": 0.3585419025569094,
      "grad_norm": 0.08502735942602158,
      "learning_rate": 2.3404908552943435e-05,
      "loss": 0.06700992584228516,
      "step": 2652
    },
    {
      "epoch": 0.35867709935274533,
      "grad_norm": 0.06627336144447327,
      "learning_rate": 2.339923863435534e-05,
      "loss": 0.05844402313232422,
      "step": 2653
    },
    {
      "epoch": 0.3588122961485813,
      "grad_norm": 0.06782086938619614,
      "learning_rate": 2.3393566967000974e-05,
      "loss": 0.04925274848937988,
      "step": 2654
    },
    {
      "epoch": 0.3589474929444172,
      "grad_norm": 0.0751691684126854,
      "learning_rate": 2.3387893552061202e-05,
      "loss": 0.06341791152954102,
      "step": 2655
    },
    {
      "epoch": 0.35908268974025315,
      "grad_norm": 0.09506940841674805,
      "learning_rate": 2.3382218390717268e-05,
      "loss": 0.0672452449798584,
      "step": 2656
    },
    {
      "epoch": 0.3592178865360891,
      "grad_norm": 0.1252928376197815,
      "learning_rate": 2.3376541484150762e-05,
      "loss": 0.08710122108459473,
      "step": 2657
    },
    {
      "epoch": 0.35935308333192506,
      "grad_norm": 0.08160462230443954,
      "learning_rate": 2.3370862833543652e-05,
      "loss": 0.09289407730102539,
      "step": 2658
    },
    {
      "epoch": 0.35948828012776096,
      "grad_norm": 0.06463818997144699,
      "learning_rate": 2.336518244007826e-05,
      "loss": 0.06690812110900879,
      "step": 2659
    },
    {
      "epoch": 0.3596234769235969,
      "grad_norm": 0.13034532964229584,
      "learning_rate": 2.3359500304937274e-05,
      "loss": 0.07425403594970703,
      "step": 2660
    },
    {
      "epoch": 0.3597586737194329,
      "grad_norm": 0.08017157763242722,
      "learning_rate": 2.335381642930375e-05,
      "loss": 0.08252096176147461,
      "step": 2661
    },
    {
      "epoch": 0.3598938705152688,
      "grad_norm": 0.10218983888626099,
      "learning_rate": 2.3348130814361094e-05,
      "loss": 0.06869041919708252,
      "step": 2662
    },
    {
      "epoch": 0.36002906731110473,
      "grad_norm": 0.05752688646316528,
      "learning_rate": 2.334244346129309e-05,
      "loss": 0.06130063533782959,
      "step": 2663
    },
    {
      "epoch": 0.3601642641069407,
      "grad_norm": 0.13094431161880493,
      "learning_rate": 2.3336754371283862e-05,
      "loss": 0.07358217239379883,
      "step": 2664
    },
    {
      "epoch": 0.3602994609027766,
      "grad_norm": 0.08713006973266602,
      "learning_rate": 2.333106354551792e-05,
      "loss": 0.05613231658935547,
      "step": 2665
    },
    {
      "epoch": 0.36043465769861255,
      "grad_norm": 0.08167698234319687,
      "learning_rate": 2.332537098518012e-05,
      "loss": 0.0674431324005127,
      "step": 2666
    },
    {
      "epoch": 0.3605698544944485,
      "grad_norm": 0.11628971248865128,
      "learning_rate": 2.3319676691455686e-05,
      "loss": 0.11261987686157227,
      "step": 2667
    },
    {
      "epoch": 0.3607050512902844,
      "grad_norm": 0.06777145713567734,
      "learning_rate": 2.3313980665530205e-05,
      "loss": 0.06354594230651855,
      "step": 2668
    },
    {
      "epoch": 0.36084024808612036,
      "grad_norm": 0.07502219080924988,
      "learning_rate": 2.3308282908589606e-05,
      "loss": 0.0854799747467041,
      "step": 2669
    },
    {
      "epoch": 0.3609754448819563,
      "grad_norm": 0.07987609505653381,
      "learning_rate": 2.330258342182021e-05,
      "loss": 0.07422924041748047,
      "step": 2670
    },
    {
      "epoch": 0.3611106416777922,
      "grad_norm": 0.05155773088335991,
      "learning_rate": 2.329688220640866e-05,
      "loss": 0.053259849548339844,
      "step": 2671
    },
    {
      "epoch": 0.3612458384736282,
      "grad_norm": 0.05814781412482262,
      "learning_rate": 2.329117926354199e-05,
      "loss": 0.08957481384277344,
      "step": 2672
    },
    {
      "epoch": 0.36138103526946413,
      "grad_norm": 0.09278398007154465,
      "learning_rate": 2.3285474594407588e-05,
      "loss": 0.06488543748855591,
      "step": 2673
    },
    {
      "epoch": 0.36151623206530004,
      "grad_norm": 0.08003925532102585,
      "learning_rate": 2.327976820019319e-05,
      "loss": 0.04104804992675781,
      "step": 2674
    },
    {
      "epoch": 0.361651428861136,
      "grad_norm": 0.0690421387553215,
      "learning_rate": 2.32740600820869e-05,
      "loss": 0.06993722915649414,
      "step": 2675
    },
    {
      "epoch": 0.36178662565697195,
      "grad_norm": 0.07715563476085663,
      "learning_rate": 2.326835024127718e-05,
      "loss": 0.06545495986938477,
      "step": 2676
    },
    {
      "epoch": 0.36192182245280785,
      "grad_norm": 0.11081211268901825,
      "learning_rate": 2.326263867895285e-05,
      "loss": 0.06154513359069824,
      "step": 2677
    },
    {
      "epoch": 0.3620570192486438,
      "grad_norm": 0.07325959205627441,
      "learning_rate": 2.3256925396303076e-05,
      "loss": 0.05494523048400879,
      "step": 2678
    },
    {
      "epoch": 0.36219221604447976,
      "grad_norm": 0.09612903743982315,
      "learning_rate": 2.3251210394517412e-05,
      "loss": 0.043145179748535156,
      "step": 2679
    },
    {
      "epoch": 0.36232741284031567,
      "grad_norm": 0.12016204744577408,
      "learning_rate": 2.3245493674785742e-05,
      "loss": 0.06176042556762695,
      "step": 2680
    },
    {
      "epoch": 0.3624626096361516,
      "grad_norm": 0.1272999346256256,
      "learning_rate": 2.3239775238298316e-05,
      "loss": 0.04294252395629883,
      "step": 2681
    },
    {
      "epoch": 0.3625978064319876,
      "grad_norm": 0.130104660987854,
      "learning_rate": 2.3234055086245744e-05,
      "loss": 0.07342720031738281,
      "step": 2682
    },
    {
      "epoch": 0.3627330032278235,
      "grad_norm": 0.18558073043823242,
      "learning_rate": 2.3228333219818998e-05,
      "loss": 0.06855964660644531,
      "step": 2683
    },
    {
      "epoch": 0.36286820002365944,
      "grad_norm": 0.09637975692749023,
      "learning_rate": 2.3222609640209397e-05,
      "loss": 0.05062079429626465,
      "step": 2684
    },
    {
      "epoch": 0.3630033968194954,
      "grad_norm": 0.10081814229488373,
      "learning_rate": 2.3216884348608614e-05,
      "loss": 0.07904863357543945,
      "step": 2685
    },
    {
      "epoch": 0.3631385936153313,
      "grad_norm": 0.08016641438007355,
      "learning_rate": 2.32111573462087e-05,
      "loss": 0.07809066772460938,
      "step": 2686
    },
    {
      "epoch": 0.36327379041116725,
      "grad_norm": 0.16184598207473755,
      "learning_rate": 2.3205428634202028e-05,
      "loss": 0.10135173797607422,
      "step": 2687
    },
    {
      "epoch": 0.3634089872070032,
      "grad_norm": 0.08866973221302032,
      "learning_rate": 2.3199698213781367e-05,
      "loss": 0.0719752311706543,
      "step": 2688
    },
    {
      "epoch": 0.3635441840028391,
      "grad_norm": 0.10324577242136002,
      "learning_rate": 2.319396608613981e-05,
      "loss": 0.050694942474365234,
      "step": 2689
    },
    {
      "epoch": 0.36367938079867507,
      "grad_norm": 0.04577205330133438,
      "learning_rate": 2.318823225247082e-05,
      "loss": 0.035382986068725586,
      "step": 2690
    },
    {
      "epoch": 0.363814577594511,
      "grad_norm": 0.09056251496076584,
      "learning_rate": 2.3182496713968208e-05,
      "loss": 0.05240058898925781,
      "step": 2691
    },
    {
      "epoch": 0.3639497743903469,
      "grad_norm": 0.08505533635616302,
      "learning_rate": 2.3176759471826143e-05,
      "loss": 0.05746889114379883,
      "step": 2692
    },
    {
      "epoch": 0.3640849711861829,
      "grad_norm": 0.2081272006034851,
      "learning_rate": 2.3171020527239155e-05,
      "loss": 0.07700604200363159,
      "step": 2693
    },
    {
      "epoch": 0.36422016798201884,
      "grad_norm": 0.0531369186937809,
      "learning_rate": 2.316527988140212e-05,
      "loss": 0.030387282371520996,
      "step": 2694
    },
    {
      "epoch": 0.36435536477785474,
      "grad_norm": 0.06013937667012215,
      "learning_rate": 2.315953753551027e-05,
      "loss": 0.055182039737701416,
      "step": 2695
    },
    {
      "epoch": 0.3644905615736907,
      "grad_norm": 0.10666341334581375,
      "learning_rate": 2.3153793490759197e-05,
      "loss": 0.06256580352783203,
      "step": 2696
    },
    {
      "epoch": 0.36462575836952665,
      "grad_norm": 0.08365563303232193,
      "learning_rate": 2.3148047748344835e-05,
      "loss": 0.06514883041381836,
      "step": 2697
    },
    {
      "epoch": 0.36476095516536255,
      "grad_norm": 0.08651472628116608,
      "learning_rate": 2.314230030946348e-05,
      "loss": 0.06864029169082642,
      "step": 2698
    },
    {
      "epoch": 0.3648961519611985,
      "grad_norm": 0.17710311710834503,
      "learning_rate": 2.3136551175311782e-05,
      "loss": 0.05901694297790527,
      "step": 2699
    },
    {
      "epoch": 0.36503134875703447,
      "grad_norm": 0.060504451394081116,
      "learning_rate": 2.313080034708674e-05,
      "loss": 0.05349087715148926,
      "step": 2700
    },
    {
      "epoch": 0.36516654555287037,
      "grad_norm": 0.06128257140517235,
      "learning_rate": 2.312504782598571e-05,
      "loss": 0.06322497129440308,
      "step": 2701
    },
    {
      "epoch": 0.3653017423487063,
      "grad_norm": 0.05907202884554863,
      "learning_rate": 2.311929361320639e-05,
      "loss": 0.03665876388549805,
      "step": 2702
    },
    {
      "epoch": 0.3654369391445423,
      "grad_norm": 0.09939534962177277,
      "learning_rate": 2.311353770994684e-05,
      "loss": 0.060498714447021484,
      "step": 2703
    },
    {
      "epoch": 0.36557213594037824,
      "grad_norm": 0.1539541482925415,
      "learning_rate": 2.310778011740548e-05,
      "loss": 0.04438591003417969,
      "step": 2704
    },
    {
      "epoch": 0.36570733273621414,
      "grad_norm": 0.07016557455062866,
      "learning_rate": 2.310202083678106e-05,
      "loss": 0.07504618167877197,
      "step": 2705
    },
    {
      "epoch": 0.3658425295320501,
      "grad_norm": 0.07573091238737106,
      "learning_rate": 2.3096259869272694e-05,
      "loss": 0.07481127977371216,
      "step": 2706
    },
    {
      "epoch": 0.36597772632788605,
      "grad_norm": 0.06745566427707672,
      "learning_rate": 2.309049721607985e-05,
      "loss": 0.05385398864746094,
      "step": 2707
    },
    {
      "epoch": 0.36611292312372196,
      "grad_norm": 0.15015491843223572,
      "learning_rate": 2.3084732878402342e-05,
      "loss": 0.0795435905456543,
      "step": 2708
    },
    {
      "epoch": 0.3662481199195579,
      "grad_norm": 0.09348881244659424,
      "learning_rate": 2.307896685744034e-05,
      "loss": 0.08438968658447266,
      "step": 2709
    },
    {
      "epoch": 0.36638331671539387,
      "grad_norm": 0.044678203761577606,
      "learning_rate": 2.3073199154394352e-05,
      "loss": 0.03957223892211914,
      "step": 2710
    },
    {
      "epoch": 0.36651851351122977,
      "grad_norm": 0.1181914433836937,
      "learning_rate": 2.3067429770465246e-05,
      "loss": 0.032999396324157715,
      "step": 2711
    },
    {
      "epoch": 0.3666537103070657,
      "grad_norm": 0.2017289400100708,
      "learning_rate": 2.3061658706854244e-05,
      "loss": 0.07780218124389648,
      "step": 2712
    },
    {
      "epoch": 0.3667889071029017,
      "grad_norm": 0.21630126237869263,
      "learning_rate": 2.3055885964762907e-05,
      "loss": 0.10248970985412598,
      "step": 2713
    },
    {
      "epoch": 0.3669241038987376,
      "grad_norm": 0.16515401005744934,
      "learning_rate": 2.3050111545393156e-05,
      "loss": 0.06472444534301758,
      "step": 2714
    },
    {
      "epoch": 0.36705930069457354,
      "grad_norm": 0.08267875760793686,
      "learning_rate": 2.304433544994725e-05,
      "loss": 0.06518292427062988,
      "step": 2715
    },
    {
      "epoch": 0.3671944974904095,
      "grad_norm": 0.09502299875020981,
      "learning_rate": 2.303855767962781e-05,
      "loss": 0.07845473289489746,
      "step": 2716
    },
    {
      "epoch": 0.3673296942862454,
      "grad_norm": 0.42943859100341797,
      "learning_rate": 2.303277823563779e-05,
      "loss": 0.08218860626220703,
      "step": 2717
    },
    {
      "epoch": 0.36746489108208136,
      "grad_norm": 0.16932396590709686,
      "learning_rate": 2.3026997119180507e-05,
      "loss": 0.0467226505279541,
      "step": 2718
    },
    {
      "epoch": 0.3676000878779173,
      "grad_norm": 0.05949682369828224,
      "learning_rate": 2.3021214331459616e-05,
      "loss": 0.06003618240356445,
      "step": 2719
    },
    {
      "epoch": 0.3677352846737532,
      "grad_norm": 0.19413945078849792,
      "learning_rate": 2.301542987367913e-05,
      "loss": 0.09903526306152344,
      "step": 2720
    },
    {
      "epoch": 0.3678704814695892,
      "grad_norm": 0.22097532451152802,
      "learning_rate": 2.3009643747043403e-05,
      "loss": 0.09497189521789551,
      "step": 2721
    },
    {
      "epoch": 0.36800567826542513,
      "grad_norm": 0.06210830435156822,
      "learning_rate": 2.3003855952757132e-05,
      "loss": 0.06505823135375977,
      "step": 2722
    },
    {
      "epoch": 0.36814087506126103,
      "grad_norm": 0.12661142647266388,
      "learning_rate": 2.2998066492025372e-05,
      "loss": 0.04744744300842285,
      "step": 2723
    },
    {
      "epoch": 0.368276071857097,
      "grad_norm": 0.24423353374004364,
      "learning_rate": 2.2992275366053513e-05,
      "loss": 0.11527371406555176,
      "step": 2724
    },
    {
      "epoch": 0.36841126865293294,
      "grad_norm": 0.1709631085395813,
      "learning_rate": 2.2986482576047305e-05,
      "loss": 0.051657795906066895,
      "step": 2725
    },
    {
      "epoch": 0.36854646544876885,
      "grad_norm": 0.17682389914989471,
      "learning_rate": 2.298068812321284e-05,
      "loss": 0.07861995697021484,
      "step": 2726
    },
    {
      "epoch": 0.3686816622446048,
      "grad_norm": 0.18474675714969635,
      "learning_rate": 2.297489200875654e-05,
      "loss": 0.08978843688964844,
      "step": 2727
    },
    {
      "epoch": 0.36881685904044076,
      "grad_norm": 0.16826172173023224,
      "learning_rate": 2.2969094233885204e-05,
      "loss": 0.07134509086608887,
      "step": 2728
    },
    {
      "epoch": 0.36895205583627666,
      "grad_norm": 0.09160614758729935,
      "learning_rate": 2.296329479980595e-05,
      "loss": 0.07286214828491211,
      "step": 2729
    },
    {
      "epoch": 0.3690872526321126,
      "grad_norm": 0.29583054780960083,
      "learning_rate": 2.2957493707726252e-05,
      "loss": 0.08674812316894531,
      "step": 2730
    },
    {
      "epoch": 0.3692224494279486,
      "grad_norm": 0.11967773735523224,
      "learning_rate": 2.2951690958853932e-05,
      "loss": 0.04410243034362793,
      "step": 2731
    },
    {
      "epoch": 0.3693576462237845,
      "grad_norm": 0.18442955613136292,
      "learning_rate": 2.2945886554397154e-05,
      "loss": 0.05983924865722656,
      "step": 2732
    },
    {
      "epoch": 0.36949284301962043,
      "grad_norm": 0.11077980697154999,
      "learning_rate": 2.294008049556441e-05,
      "loss": 0.04616779088973999,
      "step": 2733
    },
    {
      "epoch": 0.3696280398154564,
      "grad_norm": 0.10775760561227798,
      "learning_rate": 2.2934272783564577e-05,
      "loss": 0.036651611328125,
      "step": 2734
    },
    {
      "epoch": 0.3697632366112923,
      "grad_norm": 0.0744224488735199,
      "learning_rate": 2.2928463419606835e-05,
      "loss": 0.07147765159606934,
      "step": 2735
    },
    {
      "epoch": 0.36989843340712825,
      "grad_norm": 0.08086686581373215,
      "learning_rate": 2.292265240490073e-05,
      "loss": 0.06716775894165039,
      "step": 2736
    },
    {
      "epoch": 0.3700336302029642,
      "grad_norm": 0.08919425308704376,
      "learning_rate": 2.2916839740656154e-05,
      "loss": 0.07674074172973633,
      "step": 2737
    },
    {
      "epoch": 0.3701688269988001,
      "grad_norm": 0.07416735589504242,
      "learning_rate": 2.2911025428083316e-05,
      "loss": 0.03160429000854492,
      "step": 2738
    },
    {
      "epoch": 0.37030402379463606,
      "grad_norm": 0.12528888881206512,
      "learning_rate": 2.2905209468392798e-05,
      "loss": 0.06074643135070801,
      "step": 2739
    },
    {
      "epoch": 0.370439220590472,
      "grad_norm": 0.13408605754375458,
      "learning_rate": 2.2899391862795514e-05,
      "loss": 0.07269096374511719,
      "step": 2740
    },
    {
      "epoch": 0.3705744173863079,
      "grad_norm": 0.06715600937604904,
      "learning_rate": 2.2893572612502718e-05,
      "loss": 0.05186152458190918,
      "step": 2741
    },
    {
      "epoch": 0.3707096141821439,
      "grad_norm": 0.05913464352488518,
      "learning_rate": 2.2887751718726013e-05,
      "loss": 0.06685709953308105,
      "step": 2742
    },
    {
      "epoch": 0.37084481097797983,
      "grad_norm": 0.10640189796686172,
      "learning_rate": 2.288192918267734e-05,
      "loss": 0.0781707763671875,
      "step": 2743
    },
    {
      "epoch": 0.37098000777381573,
      "grad_norm": 0.1281871199607849,
      "learning_rate": 2.2876105005568974e-05,
      "loss": 0.06212592124938965,
      "step": 2744
    },
    {
      "epoch": 0.3711152045696517,
      "grad_norm": 0.08705922216176987,
      "learning_rate": 2.287027918861355e-05,
      "loss": 0.05160844326019287,
      "step": 2745
    },
    {
      "epoch": 0.37125040136548765,
      "grad_norm": 0.13004440069198608,
      "learning_rate": 2.2864451733024024e-05,
      "loss": 0.06316614151000977,
      "step": 2746
    },
    {
      "epoch": 0.37138559816132355,
      "grad_norm": 0.07214105129241943,
      "learning_rate": 2.2858622640013716e-05,
      "loss": 0.09713459014892578,
      "step": 2747
    },
    {
      "epoch": 0.3715207949571595,
      "grad_norm": 0.047702912241220474,
      "learning_rate": 2.285279191079626e-05,
      "loss": 0.05330824851989746,
      "step": 2748
    },
    {
      "epoch": 0.37165599175299546,
      "grad_norm": 0.12142839282751083,
      "learning_rate": 2.2846959546585656e-05,
      "loss": 0.0802164077758789,
      "step": 2749
    },
    {
      "epoch": 0.3717911885488314,
      "grad_norm": 0.07923062890768051,
      "learning_rate": 2.2841125548596225e-05,
      "loss": 0.058669328689575195,
      "step": 2750
    },
    {
      "epoch": 0.3719263853446673,
      "grad_norm": 0.1896916776895523,
      "learning_rate": 2.2835289918042648e-05,
      "loss": 0.06959748268127441,
      "step": 2751
    },
    {
      "epoch": 0.3720615821405033,
      "grad_norm": 0.0663958266377449,
      "learning_rate": 2.282945265613992e-05,
      "loss": 0.08017897605895996,
      "step": 2752
    },
    {
      "epoch": 0.37219677893633923,
      "grad_norm": 0.10044600814580917,
      "learning_rate": 2.2823613764103406e-05,
      "loss": 0.08410584926605225,
      "step": 2753
    },
    {
      "epoch": 0.37233197573217514,
      "grad_norm": 0.09321119636297226,
      "learning_rate": 2.2817773243148776e-05,
      "loss": 0.060943603515625,
      "step": 2754
    },
    {
      "epoch": 0.3724671725280111,
      "grad_norm": 0.06502675265073776,
      "learning_rate": 2.2811931094492074e-05,
      "loss": 0.04653823375701904,
      "step": 2755
    },
    {
      "epoch": 0.37260236932384705,
      "grad_norm": 0.14461436867713928,
      "learning_rate": 2.280608731934966e-05,
      "loss": 0.07020258903503418,
      "step": 2756
    },
    {
      "epoch": 0.37273756611968295,
      "grad_norm": 0.1345045417547226,
      "learning_rate": 2.280024191893823e-05,
      "loss": 0.08864402770996094,
      "step": 2757
    },
    {
      "epoch": 0.3728727629155189,
      "grad_norm": 0.100283183157444,
      "learning_rate": 2.279439489447485e-05,
      "loss": 0.09058094024658203,
      "step": 2758
    },
    {
      "epoch": 0.37300795971135486,
      "grad_norm": 0.09580633044242859,
      "learning_rate": 2.278854624717688e-05,
      "loss": 0.06604033708572388,
      "step": 2759
    },
    {
      "epoch": 0.37314315650719077,
      "grad_norm": 0.061638735234737396,
      "learning_rate": 2.2782695978262045e-05,
      "loss": 0.0608983039855957,
      "step": 2760
    },
    {
      "epoch": 0.3732783533030267,
      "grad_norm": 0.07221037894487381,
      "learning_rate": 2.2776844088948406e-05,
      "loss": 0.05460381507873535,
      "step": 2761
    },
    {
      "epoch": 0.3734135500988627,
      "grad_norm": 0.03720799461007118,
      "learning_rate": 2.2770990580454364e-05,
      "loss": 0.045108795166015625,
      "step": 2762
    },
    {
      "epoch": 0.3735487468946986,
      "grad_norm": 0.08576427400112152,
      "learning_rate": 2.276513545399864e-05,
      "loss": 0.06876599788665771,
      "step": 2763
    },
    {
      "epoch": 0.37368394369053454,
      "grad_norm": 0.08200296014547348,
      "learning_rate": 2.2759278710800306e-05,
      "loss": 0.040184736251831055,
      "step": 2764
    },
    {
      "epoch": 0.3738191404863705,
      "grad_norm": 0.07570790499448776,
      "learning_rate": 2.275342035207876e-05,
      "loss": 0.06896781921386719,
      "step": 2765
    },
    {
      "epoch": 0.3739543372822064,
      "grad_norm": 0.15543131530284882,
      "learning_rate": 2.2747560379053752e-05,
      "loss": 0.08238428831100464,
      "step": 2766
    },
    {
      "epoch": 0.37408953407804235,
      "grad_norm": 0.07087230682373047,
      "learning_rate": 2.2741698792945364e-05,
      "loss": 0.03446626663208008,
      "step": 2767
    },
    {
      "epoch": 0.3742247308738783,
      "grad_norm": 0.2016008049249649,
      "learning_rate": 2.2735835594974003e-05,
      "loss": 0.0862741470336914,
      "step": 2768
    },
    {
      "epoch": 0.3743599276697142,
      "grad_norm": 0.10248856246471405,
      "learning_rate": 2.272997078636042e-05,
      "loss": 0.041033267974853516,
      "step": 2769
    },
    {
      "epoch": 0.37449512446555017,
      "grad_norm": 0.10012223571538925,
      "learning_rate": 2.272410436832569e-05,
      "loss": 0.06709051132202148,
      "step": 2770
    },
    {
      "epoch": 0.3746303212613861,
      "grad_norm": 0.09525300562381744,
      "learning_rate": 2.2718236342091248e-05,
      "loss": 0.06319558620452881,
      "step": 2771
    },
    {
      "epoch": 0.374765518057222,
      "grad_norm": 0.0983513593673706,
      "learning_rate": 2.2712366708878838e-05,
      "loss": 0.07823848724365234,
      "step": 2772
    },
    {
      "epoch": 0.374900714853058,
      "grad_norm": 0.2814671993255615,
      "learning_rate": 2.2706495469910552e-05,
      "loss": 0.1147909164428711,
      "step": 2773
    },
    {
      "epoch": 0.37503591164889394,
      "grad_norm": 0.08254552632570267,
      "learning_rate": 2.2700622626408814e-05,
      "loss": 0.08213996887207031,
      "step": 2774
    },
    {
      "epoch": 0.37517110844472984,
      "grad_norm": 0.06676315516233444,
      "learning_rate": 2.2694748179596375e-05,
      "loss": 0.0569148063659668,
      "step": 2775
    },
    {
      "epoch": 0.3753063052405658,
      "grad_norm": 0.09155144542455673,
      "learning_rate": 2.2688872130696342e-05,
      "loss": 0.054955482482910156,
      "step": 2776
    },
    {
      "epoch": 0.37544150203640175,
      "grad_norm": 0.06272590160369873,
      "learning_rate": 2.268299448093212e-05,
      "loss": 0.05836343765258789,
      "step": 2777
    },
    {
      "epoch": 0.37557669883223765,
      "grad_norm": 0.14821811020374298,
      "learning_rate": 2.2677115231527482e-05,
      "loss": 0.09092545509338379,
      "step": 2778
    },
    {
      "epoch": 0.3757118956280736,
      "grad_norm": 0.08862105011940002,
      "learning_rate": 2.267123438370651e-05,
      "loss": 0.04809236526489258,
      "step": 2779
    },
    {
      "epoch": 0.37584709242390957,
      "grad_norm": 0.05294622853398323,
      "learning_rate": 2.266535193869363e-05,
      "loss": 0.05145597457885742,
      "step": 2780
    },
    {
      "epoch": 0.37598228921974547,
      "grad_norm": 0.1998557597398758,
      "learning_rate": 2.2659467897713604e-05,
      "loss": 0.07789409160614014,
      "step": 2781
    },
    {
      "epoch": 0.3761174860155814,
      "grad_norm": 0.06989549845457077,
      "learning_rate": 2.2653582261991516e-05,
      "loss": 0.04813957214355469,
      "step": 2782
    },
    {
      "epoch": 0.3762526828114174,
      "grad_norm": 0.05387307330965996,
      "learning_rate": 2.2647695032752785e-05,
      "loss": 0.0516359806060791,
      "step": 2783
    },
    {
      "epoch": 0.3763878796072533,
      "grad_norm": 0.10231353342533112,
      "learning_rate": 2.264180621122317e-05,
      "loss": 0.06594467163085938,
      "step": 2784
    },
    {
      "epoch": 0.37652307640308924,
      "grad_norm": 0.12177782505750656,
      "learning_rate": 2.2635915798628747e-05,
      "loss": 0.08141422271728516,
      "step": 2785
    },
    {
      "epoch": 0.3766582731989252,
      "grad_norm": 0.05815134942531586,
      "learning_rate": 2.2630023796195932e-05,
      "loss": 0.046163082122802734,
      "step": 2786
    },
    {
      "epoch": 0.3767934699947611,
      "grad_norm": 0.10425441712141037,
      "learning_rate": 2.262413020515148e-05,
      "loss": 0.08174657821655273,
      "step": 2787
    },
    {
      "epoch": 0.37692866679059706,
      "grad_norm": 0.18714912235736847,
      "learning_rate": 2.261823502672246e-05,
      "loss": 0.06075096130371094,
      "step": 2788
    },
    {
      "epoch": 0.377063863586433,
      "grad_norm": 0.24350875616073608,
      "learning_rate": 2.261233826213628e-05,
      "loss": 0.07392144203186035,
      "step": 2789
    },
    {
      "epoch": 0.3771990603822689,
      "grad_norm": 0.08834494650363922,
      "learning_rate": 2.2606439912620688e-05,
      "loss": 0.07562696933746338,
      "step": 2790
    },
    {
      "epoch": 0.37733425717810487,
      "grad_norm": 0.10013694316148758,
      "learning_rate": 2.2600539979403734e-05,
      "loss": 0.06572079658508301,
      "step": 2791
    },
    {
      "epoch": 0.3774694539739408,
      "grad_norm": 0.10506601631641388,
      "learning_rate": 2.259463846371383e-05,
      "loss": 0.0860595703125,
      "step": 2792
    },
    {
      "epoch": 0.3776046507697768,
      "grad_norm": 0.14454716444015503,
      "learning_rate": 2.2588735366779698e-05,
      "loss": 0.05851936340332031,
      "step": 2793
    },
    {
      "epoch": 0.3777398475656127,
      "grad_norm": 0.23578503727912903,
      "learning_rate": 2.2582830689830394e-05,
      "loss": 0.08013248443603516,
      "step": 2794
    },
    {
      "epoch": 0.37787504436144864,
      "grad_norm": 0.41214001178741455,
      "learning_rate": 2.2576924434095305e-05,
      "loss": 0.10672998428344727,
      "step": 2795
    },
    {
      "epoch": 0.3780102411572846,
      "grad_norm": 0.04705068841576576,
      "learning_rate": 2.257101660080414e-05,
      "loss": 0.047440528869628906,
      "step": 2796
    },
    {
      "epoch": 0.3781454379531205,
      "grad_norm": 0.08299770951271057,
      "learning_rate": 2.256510719118695e-05,
      "loss": 0.03977060317993164,
      "step": 2797
    },
    {
      "epoch": 0.37828063474895646,
      "grad_norm": 0.050721246749162674,
      "learning_rate": 2.2559196206474094e-05,
      "loss": 0.0630655288696289,
      "step": 2798
    },
    {
      "epoch": 0.3784158315447924,
      "grad_norm": 0.12276268750429153,
      "learning_rate": 2.2553283647896287e-05,
      "loss": 0.07132959365844727,
      "step": 2799
    },
    {
      "epoch": 0.3785510283406283,
      "grad_norm": 0.1889687329530716,
      "learning_rate": 2.254736951668454e-05,
      "loss": 0.09444999694824219,
      "step": 2800
    },
    {
      "epoch": 0.37868622513646427,
      "grad_norm": 0.11569591611623764,
      "learning_rate": 2.2541453814070212e-05,
      "loss": 0.06716674566268921,
      "step": 2801
    },
    {
      "epoch": 0.37882142193230023,
      "grad_norm": 0.20026740431785583,
      "learning_rate": 2.2535536541284983e-05,
      "loss": 0.09140205383300781,
      "step": 2802
    },
    {
      "epoch": 0.37895661872813613,
      "grad_norm": 0.2150438278913498,
      "learning_rate": 2.2529617699560857e-05,
      "loss": 0.06071138381958008,
      "step": 2803
    },
    {
      "epoch": 0.3790918155239721,
      "grad_norm": 0.1123405247926712,
      "learning_rate": 2.2523697290130185e-05,
      "loss": 0.049558281898498535,
      "step": 2804
    },
    {
      "epoch": 0.37922701231980804,
      "grad_norm": 0.061004046350717545,
      "learning_rate": 2.251777531422561e-05,
      "loss": 0.03796124458312988,
      "step": 2805
    },
    {
      "epoch": 0.37936220911564394,
      "grad_norm": 0.15213656425476074,
      "learning_rate": 2.2511851773080127e-05,
      "loss": 0.09109020233154297,
      "step": 2806
    },
    {
      "epoch": 0.3794974059114799,
      "grad_norm": 0.0711393654346466,
      "learning_rate": 2.2505926667927043e-05,
      "loss": 0.04447793960571289,
      "step": 2807
    },
    {
      "epoch": 0.37963260270731586,
      "grad_norm": 0.1593368947505951,
      "learning_rate": 2.25e-05,
      "loss": 0.08230447769165039,
      "step": 2808
    },
    {
      "epoch": 0.37976779950315176,
      "grad_norm": 0.06413017958402634,
      "learning_rate": 2.2494071770532966e-05,
      "loss": 0.10065078735351562,
      "step": 2809
    },
    {
      "epoch": 0.3799029962989877,
      "grad_norm": 0.11518452316522598,
      "learning_rate": 2.2488141980760223e-05,
      "loss": 0.038037776947021484,
      "step": 2810
    },
    {
      "epoch": 0.3800381930948237,
      "grad_norm": 0.05485963076353073,
      "learning_rate": 2.248221063191639e-05,
      "loss": 0.06107282638549805,
      "step": 2811
    },
    {
      "epoch": 0.3801733898906596,
      "grad_norm": 0.0578949972987175,
      "learning_rate": 2.24762777252364e-05,
      "loss": 0.05760788917541504,
      "step": 2812
    },
    {
      "epoch": 0.38030858668649553,
      "grad_norm": 0.09842949360609055,
      "learning_rate": 2.2470343261955525e-05,
      "loss": 0.10003852844238281,
      "step": 2813
    },
    {
      "epoch": 0.3804437834823315,
      "grad_norm": 0.06713572144508362,
      "learning_rate": 2.246440724330934e-05,
      "loss": 0.04203915596008301,
      "step": 2814
    },
    {
      "epoch": 0.3805789802781674,
      "grad_norm": 0.07310368120670319,
      "learning_rate": 2.2458469670533765e-05,
      "loss": 0.04777121543884277,
      "step": 2815
    },
    {
      "epoch": 0.38071417707400335,
      "grad_norm": 0.10394813865423203,
      "learning_rate": 2.2452530544865034e-05,
      "loss": 0.04501032829284668,
      "step": 2816
    },
    {
      "epoch": 0.3808493738698393,
      "grad_norm": 0.043789658695459366,
      "learning_rate": 2.24465898675397e-05,
      "loss": 0.05106019973754883,
      "step": 2817
    },
    {
      "epoch": 0.3809845706656752,
      "grad_norm": 0.1555924117565155,
      "learning_rate": 2.244064763979464e-05,
      "loss": 0.07743453979492188,
      "step": 2818
    },
    {
      "epoch": 0.38111976746151116,
      "grad_norm": 0.19059306383132935,
      "learning_rate": 2.2434703862867068e-05,
      "loss": 0.09229755401611328,
      "step": 2819
    },
    {
      "epoch": 0.3812549642573471,
      "grad_norm": 0.14083994925022125,
      "learning_rate": 2.2428758537994504e-05,
      "loss": 0.05932426452636719,
      "step": 2820
    },
    {
      "epoch": 0.381390161053183,
      "grad_norm": 0.07788024097681046,
      "learning_rate": 2.24228116664148e-05,
      "loss": 0.04738330841064453,
      "step": 2821
    },
    {
      "epoch": 0.381525357849019,
      "grad_norm": 0.19101087749004364,
      "learning_rate": 2.2416863249366125e-05,
      "loss": 0.06658792495727539,
      "step": 2822
    },
    {
      "epoch": 0.38166055464485493,
      "grad_norm": 0.12705843150615692,
      "learning_rate": 2.241091328808696e-05,
      "loss": 0.0597684383392334,
      "step": 2823
    },
    {
      "epoch": 0.38179575144069083,
      "grad_norm": 0.05747111886739731,
      "learning_rate": 2.240496178381614e-05,
      "loss": 0.04332387447357178,
      "step": 2824
    },
    {
      "epoch": 0.3819309482365268,
      "grad_norm": 0.12303254753351212,
      "learning_rate": 2.239900873779278e-05,
      "loss": 0.06458282470703125,
      "step": 2825
    },
    {
      "epoch": 0.38206614503236275,
      "grad_norm": 0.06540141254663467,
      "learning_rate": 2.2393054151256352e-05,
      "loss": 0.06014728546142578,
      "step": 2826
    },
    {
      "epoch": 0.38220134182819865,
      "grad_norm": 0.11612020432949066,
      "learning_rate": 2.238709802544662e-05,
      "loss": 0.08864045143127441,
      "step": 2827
    },
    {
      "epoch": 0.3823365386240346,
      "grad_norm": 0.0544077605009079,
      "learning_rate": 2.2381140361603686e-05,
      "loss": 0.05488467216491699,
      "step": 2828
    },
    {
      "epoch": 0.38247173541987056,
      "grad_norm": 0.05459945276379585,
      "learning_rate": 2.237518116096797e-05,
      "loss": 0.06433665752410889,
      "step": 2829
    },
    {
      "epoch": 0.38260693221570646,
      "grad_norm": 0.16710463166236877,
      "learning_rate": 2.2369220424780203e-05,
      "loss": 0.05536776781082153,
      "step": 2830
    },
    {
      "epoch": 0.3827421290115424,
      "grad_norm": 0.10125366598367691,
      "learning_rate": 2.2363258154281452e-05,
      "loss": 0.09418916702270508,
      "step": 2831
    },
    {
      "epoch": 0.3828773258073784,
      "grad_norm": 0.10701412707567215,
      "learning_rate": 2.2357294350713088e-05,
      "loss": 0.06345701217651367,
      "step": 2832
    },
    {
      "epoch": 0.3830125226032143,
      "grad_norm": 0.06601959466934204,
      "learning_rate": 2.2351329015316802e-05,
      "loss": 0.05744194984436035,
      "step": 2833
    },
    {
      "epoch": 0.38314771939905024,
      "grad_norm": 0.20217901468276978,
      "learning_rate": 2.2345362149334613e-05,
      "loss": 0.1077275276184082,
      "step": 2834
    },
    {
      "epoch": 0.3832829161948862,
      "grad_norm": 0.10598672926425934,
      "learning_rate": 2.2339393754008854e-05,
      "loss": 0.05259418487548828,
      "step": 2835
    },
    {
      "epoch": 0.3834181129907221,
      "grad_norm": 0.10022184252738953,
      "learning_rate": 2.233342383058218e-05,
      "loss": 0.0843663215637207,
      "step": 2836
    },
    {
      "epoch": 0.38355330978655805,
      "grad_norm": 0.21557487547397614,
      "learning_rate": 2.2327452380297554e-05,
      "loss": 0.0895071029663086,
      "step": 2837
    },
    {
      "epoch": 0.383688506582394,
      "grad_norm": 0.18631607294082642,
      "learning_rate": 2.232147940439827e-05,
      "loss": 0.06493210792541504,
      "step": 2838
    },
    {
      "epoch": 0.38382370337822996,
      "grad_norm": 0.14699497818946838,
      "learning_rate": 2.2315504904127936e-05,
      "loss": 0.09984707832336426,
      "step": 2839
    },
    {
      "epoch": 0.38395890017406586,
      "grad_norm": 0.08688609302043915,
      "learning_rate": 2.2309528880730463e-05,
      "loss": 0.08707427978515625,
      "step": 2840
    },
    {
      "epoch": 0.3840940969699018,
      "grad_norm": 0.09528959542512894,
      "learning_rate": 2.2303551335450096e-05,
      "loss": 0.06581449508666992,
      "step": 2841
    },
    {
      "epoch": 0.3842292937657378,
      "grad_norm": 0.05205417424440384,
      "learning_rate": 2.2297572269531398e-05,
      "loss": 0.04139566421508789,
      "step": 2842
    },
    {
      "epoch": 0.3843644905615737,
      "grad_norm": 0.07317586243152618,
      "learning_rate": 2.2291591684219243e-05,
      "loss": 0.05797839164733887,
      "step": 2843
    },
    {
      "epoch": 0.38449968735740964,
      "grad_norm": 0.1700967699289322,
      "learning_rate": 2.2285609580758806e-05,
      "loss": 0.0629042387008667,
      "step": 2844
    },
    {
      "epoch": 0.3846348841532456,
      "grad_norm": 0.38466769456863403,
      "learning_rate": 2.227962596039561e-05,
      "loss": 0.10962724685668945,
      "step": 2845
    },
    {
      "epoch": 0.3847700809490815,
      "grad_norm": 0.07835380733013153,
      "learning_rate": 2.2273640824375462e-05,
      "loss": 0.06114912033081055,
      "step": 2846
    },
    {
      "epoch": 0.38490527774491745,
      "grad_norm": 0.11597051471471786,
      "learning_rate": 2.2267654173944515e-05,
      "loss": 0.0951838493347168,
      "step": 2847
    },
    {
      "epoch": 0.3850404745407534,
      "grad_norm": 0.1270502507686615,
      "learning_rate": 2.2261666010349212e-05,
      "loss": 0.09619283676147461,
      "step": 2848
    },
    {
      "epoch": 0.3851756713365893,
      "grad_norm": 0.17541272938251495,
      "learning_rate": 2.2255676334836317e-05,
      "loss": 0.0588226318359375,
      "step": 2849
    },
    {
      "epoch": 0.38531086813242527,
      "grad_norm": 0.16797445714473724,
      "learning_rate": 2.2249685148652917e-05,
      "loss": 0.07074284553527832,
      "step": 2850
    },
    {
      "epoch": 0.3854460649282612,
      "grad_norm": 0.12969140708446503,
      "learning_rate": 2.224369245304641e-05,
      "loss": 0.08632183074951172,
      "step": 2851
    },
    {
      "epoch": 0.3855812617240971,
      "grad_norm": 0.0639800950884819,
      "learning_rate": 2.2237698249264507e-05,
      "loss": 0.04711794853210449,
      "step": 2852
    },
    {
      "epoch": 0.3857164585199331,
      "grad_norm": 0.2360343337059021,
      "learning_rate": 2.2231702538555235e-05,
      "loss": 0.0767362117767334,
      "step": 2853
    },
    {
      "epoch": 0.38585165531576904,
      "grad_norm": 0.05243297666311264,
      "learning_rate": 2.2225705322166928e-05,
      "loss": 0.07010650634765625,
      "step": 2854
    },
    {
      "epoch": 0.38598685211160494,
      "grad_norm": 0.16941840946674347,
      "learning_rate": 2.2219706601348242e-05,
      "loss": 0.10315513610839844,
      "step": 2855
    },
    {
      "epoch": 0.3861220489074409,
      "grad_norm": 0.07535373419523239,
      "learning_rate": 2.221370637734814e-05,
      "loss": 0.048774003982543945,
      "step": 2856
    },
    {
      "epoch": 0.38625724570327685,
      "grad_norm": 0.09462258219718933,
      "learning_rate": 2.22077046514159e-05,
      "loss": 0.07574272155761719,
      "step": 2857
    },
    {
      "epoch": 0.38639244249911275,
      "grad_norm": 0.11887585371732712,
      "learning_rate": 2.220170142480112e-05,
      "loss": 0.06749486923217773,
      "step": 2858
    },
    {
      "epoch": 0.3865276392949487,
      "grad_norm": 0.07648269832134247,
      "learning_rate": 2.2195696698753695e-05,
      "loss": 0.06076592206954956,
      "step": 2859
    },
    {
      "epoch": 0.38666283609078467,
      "grad_norm": 0.09027384966611862,
      "learning_rate": 2.2189690474523844e-05,
      "loss": 0.05429410934448242,
      "step": 2860
    },
    {
      "epoch": 0.38679803288662057,
      "grad_norm": 0.06682781130075455,
      "learning_rate": 2.21836827533621e-05,
      "loss": 0.054702043533325195,
      "step": 2861
    },
    {
      "epoch": 0.3869332296824565,
      "grad_norm": 0.1171586737036705,
      "learning_rate": 2.2177673536519297e-05,
      "loss": 0.07994914054870605,
      "step": 2862
    },
    {
      "epoch": 0.3870684264782925,
      "grad_norm": 0.10911373794078827,
      "learning_rate": 2.217166282524659e-05,
      "loss": 0.07723045349121094,
      "step": 2863
    },
    {
      "epoch": 0.3872036232741284,
      "grad_norm": 0.09084512293338776,
      "learning_rate": 2.216565062079544e-05,
      "loss": 0.06727838516235352,
      "step": 2864
    },
    {
      "epoch": 0.38733882006996434,
      "grad_norm": 0.09167740494012833,
      "learning_rate": 2.2159636924417612e-05,
      "loss": 0.0527501106262207,
      "step": 2865
    },
    {
      "epoch": 0.3874740168658003,
      "grad_norm": 0.23680470883846283,
      "learning_rate": 2.2153621737365205e-05,
      "loss": 0.08391380310058594,
      "step": 2866
    },
    {
      "epoch": 0.3876092136616362,
      "grad_norm": 0.10274053364992142,
      "learning_rate": 2.2147605060890598e-05,
      "loss": 0.08497810363769531,
      "step": 2867
    },
    {
      "epoch": 0.38774441045747216,
      "grad_norm": 0.17132432758808136,
      "learning_rate": 2.2141586896246503e-05,
      "loss": 0.09068691730499268,
      "step": 2868
    },
    {
      "epoch": 0.3878796072533081,
      "grad_norm": 0.11979129910469055,
      "learning_rate": 2.2135567244685933e-05,
      "loss": 0.07145345211029053,
      "step": 2869
    },
    {
      "epoch": 0.388014804049144,
      "grad_norm": 0.16250404715538025,
      "learning_rate": 2.2129546107462214e-05,
      "loss": 0.10976696014404297,
      "step": 2870
    },
    {
      "epoch": 0.38815000084497997,
      "grad_norm": 0.1045694574713707,
      "learning_rate": 2.212352348582897e-05,
      "loss": 0.08321857452392578,
      "step": 2871
    },
    {
      "epoch": 0.3882851976408159,
      "grad_norm": 0.16664868593215942,
      "learning_rate": 2.2117499381040157e-05,
      "loss": 0.061040401458740234,
      "step": 2872
    },
    {
      "epoch": 0.38842039443665183,
      "grad_norm": 0.08886837959289551,
      "learning_rate": 2.211147379435001e-05,
      "loss": 0.05218172073364258,
      "step": 2873
    },
    {
      "epoch": 0.3885555912324878,
      "grad_norm": 0.08197692036628723,
      "learning_rate": 2.2105446727013098e-05,
      "loss": 0.0597538948059082,
      "step": 2874
    },
    {
      "epoch": 0.38869078802832374,
      "grad_norm": 0.06315159797668457,
      "learning_rate": 2.209941818028429e-05,
      "loss": 0.06333255767822266,
      "step": 2875
    },
    {
      "epoch": 0.38882598482415964,
      "grad_norm": 0.09284429997205734,
      "learning_rate": 2.2093388155418757e-05,
      "loss": 0.05210423469543457,
      "step": 2876
    },
    {
      "epoch": 0.3889611816199956,
      "grad_norm": 0.0909295529127121,
      "learning_rate": 2.2087356653671982e-05,
      "loss": 0.05983614921569824,
      "step": 2877
    },
    {
      "epoch": 0.38909637841583156,
      "grad_norm": 0.1092667207121849,
      "learning_rate": 2.2081323676299756e-05,
      "loss": 0.05205661058425903,
      "step": 2878
    },
    {
      "epoch": 0.38923157521166746,
      "grad_norm": 0.11321708559989929,
      "learning_rate": 2.207528922455818e-05,
      "loss": 0.05749708414077759,
      "step": 2879
    },
    {
      "epoch": 0.3893667720075034,
      "grad_norm": 0.1219281479716301,
      "learning_rate": 2.206925329970366e-05,
      "loss": 0.060623884201049805,
      "step": 2880
    },
    {
      "epoch": 0.38950196880333937,
      "grad_norm": 0.07548338174819946,
      "learning_rate": 2.20632159029929e-05,
      "loss": 0.07660472393035889,
      "step": 2881
    },
    {
      "epoch": 0.3896371655991753,
      "grad_norm": 0.07393660396337509,
      "learning_rate": 2.2057177035682926e-05,
      "loss": 0.06603860855102539,
      "step": 2882
    },
    {
      "epoch": 0.38977236239501123,
      "grad_norm": 0.11670587211847305,
      "learning_rate": 2.2051136699031058e-05,
      "loss": 0.06527531147003174,
      "step": 2883
    },
    {
      "epoch": 0.3899075591908472,
      "grad_norm": 0.18693673610687256,
      "learning_rate": 2.2045094894294933e-05,
      "loss": 0.055972933769226074,
      "step": 2884
    },
    {
      "epoch": 0.39004275598668314,
      "grad_norm": 0.04825974255800247,
      "learning_rate": 2.203905162273248e-05,
      "loss": 0.04258167743682861,
      "step": 2885
    },
    {
      "epoch": 0.39017795278251904,
      "grad_norm": 0.05295131728053093,
      "learning_rate": 2.203300688560194e-05,
      "loss": 0.05836498737335205,
      "step": 2886
    },
    {
      "epoch": 0.390313149578355,
      "grad_norm": 0.14171721041202545,
      "learning_rate": 2.2026960684161862e-05,
      "loss": 0.07374858856201172,
      "step": 2887
    },
    {
      "epoch": 0.39044834637419096,
      "grad_norm": 0.08113564550876617,
      "learning_rate": 2.2020913019671097e-05,
      "loss": 0.05691087245941162,
      "step": 2888
    },
    {
      "epoch": 0.39058354317002686,
      "grad_norm": 0.05424632877111435,
      "learning_rate": 2.20148638933888e-05,
      "loss": 0.05599355697631836,
      "step": 2889
    },
    {
      "epoch": 0.3907187399658628,
      "grad_norm": 0.09617909789085388,
      "learning_rate": 2.2008813306574438e-05,
      "loss": 0.03911733627319336,
      "step": 2890
    },
    {
      "epoch": 0.3908539367616988,
      "grad_norm": 0.07980953902006149,
      "learning_rate": 2.200276126048777e-05,
      "loss": 0.09441471099853516,
      "step": 2891
    },
    {
      "epoch": 0.3909891335575347,
      "grad_norm": 0.0935797244310379,
      "learning_rate": 2.199670775638886e-05,
      "loss": 0.07928848266601562,
      "step": 2892
    },
    {
      "epoch": 0.39112433035337063,
      "grad_norm": 0.28658780455589294,
      "learning_rate": 2.1990652795538085e-05,
      "loss": 0.0719062089920044,
      "step": 2893
    },
    {
      "epoch": 0.3912595271492066,
      "grad_norm": 0.09717074036598206,
      "learning_rate": 2.1984596379196117e-05,
      "loss": 0.07060098648071289,
      "step": 2894
    },
    {
      "epoch": 0.3913947239450425,
      "grad_norm": 0.0727180764079094,
      "learning_rate": 2.1978538508623942e-05,
      "loss": 0.05077362060546875,
      "step": 2895
    },
    {
      "epoch": 0.39152992074087845,
      "grad_norm": 0.07455969601869583,
      "learning_rate": 2.197247918508283e-05,
      "loss": 0.037695884704589844,
      "step": 2896
    },
    {
      "epoch": 0.3916651175367144,
      "grad_norm": 0.1373928189277649,
      "learning_rate": 2.1966418409834374e-05,
      "loss": 0.06282782554626465,
      "step": 2897
    },
    {
      "epoch": 0.3918003143325503,
      "grad_norm": 0.08033270388841629,
      "learning_rate": 2.1960356184140453e-05,
      "loss": 0.07048797607421875,
      "step": 2898
    },
    {
      "epoch": 0.39193551112838626,
      "grad_norm": 0.2078501284122467,
      "learning_rate": 2.1954292509263258e-05,
      "loss": 0.07110238075256348,
      "step": 2899
    },
    {
      "epoch": 0.3920707079242222,
      "grad_norm": 0.08781122416257858,
      "learning_rate": 2.194822738646528e-05,
      "loss": 0.06780123710632324,
      "step": 2900
    },
    {
      "epoch": 0.3922059047200581,
      "grad_norm": 0.04870763048529625,
      "learning_rate": 2.1942160817009304e-05,
      "loss": 0.04882550239562988,
      "step": 2901
    },
    {
      "epoch": 0.3923411015158941,
      "grad_norm": 0.0994347333908081,
      "learning_rate": 2.193609280215843e-05,
      "loss": 0.07555484771728516,
      "step": 2902
    },
    {
      "epoch": 0.39247629831173003,
      "grad_norm": 0.12155533581972122,
      "learning_rate": 2.1930023343176044e-05,
      "loss": 0.061284780502319336,
      "step": 2903
    },
    {
      "epoch": 0.39261149510756593,
      "grad_norm": 0.04636900871992111,
      "learning_rate": 2.1923952441325837e-05,
      "loss": 0.053002357482910156,
      "step": 2904
    },
    {
      "epoch": 0.3927466919034019,
      "grad_norm": 0.06653093546628952,
      "learning_rate": 2.191788009787182e-05,
      "loss": 0.06837129592895508,
      "step": 2905
    },
    {
      "epoch": 0.39288188869923785,
      "grad_norm": 0.10176833719015121,
      "learning_rate": 2.1911806314078267e-05,
      "loss": 0.060544490814208984,
      "step": 2906
    },
    {
      "epoch": 0.39301708549507375,
      "grad_norm": 0.07731485366821289,
      "learning_rate": 2.1905731091209786e-05,
      "loss": 0.0642235279083252,
      "step": 2907
    },
    {
      "epoch": 0.3931522822909097,
      "grad_norm": 0.20840750634670258,
      "learning_rate": 2.1899654430531262e-05,
      "loss": 0.09018063545227051,
      "step": 2908
    },
    {
      "epoch": 0.39328747908674566,
      "grad_norm": 0.1256875842809677,
      "learning_rate": 2.18935763333079e-05,
      "loss": 0.04741537570953369,
      "step": 2909
    },
    {
      "epoch": 0.39342267588258156,
      "grad_norm": 0.09193024039268494,
      "learning_rate": 2.1887496800805175e-05,
      "loss": 0.07258749008178711,
      "step": 2910
    },
    {
      "epoch": 0.3935578726784175,
      "grad_norm": 0.05182574689388275,
      "learning_rate": 2.188141583428889e-05,
      "loss": 0.05204415321350098,
      "step": 2911
    },
    {
      "epoch": 0.3936930694742535,
      "grad_norm": 0.07833902537822723,
      "learning_rate": 2.1875333435025138e-05,
      "loss": 0.06772053241729736,
      "step": 2912
    },
    {
      "epoch": 0.3938282662700894,
      "grad_norm": 0.20408417284488678,
      "learning_rate": 2.1869249604280296e-05,
      "loss": 0.056893348693847656,
      "step": 2913
    },
    {
      "epoch": 0.39396346306592533,
      "grad_norm": 0.1315690279006958,
      "learning_rate": 2.1863164343321057e-05,
      "loss": 0.061517953872680664,
      "step": 2914
    },
    {
      "epoch": 0.3940986598617613,
      "grad_norm": 0.048998698592185974,
      "learning_rate": 2.1857077653414397e-05,
      "loss": 0.04927492141723633,
      "step": 2915
    },
    {
      "epoch": 0.3942338566575972,
      "grad_norm": 0.10806699842214584,
      "learning_rate": 2.185098953582761e-05,
      "loss": 0.06411170959472656,
      "step": 2916
    },
    {
      "epoch": 0.39436905345343315,
      "grad_norm": 0.07757361978292465,
      "learning_rate": 2.1844899991828265e-05,
      "loss": 0.06878209114074707,
      "step": 2917
    },
    {
      "epoch": 0.3945042502492691,
      "grad_norm": 0.0717562660574913,
      "learning_rate": 2.1838809022684247e-05,
      "loss": 0.07520771026611328,
      "step": 2918
    },
    {
      "epoch": 0.394639447045105,
      "grad_norm": 0.07770061492919922,
      "learning_rate": 2.1832716629663712e-05,
      "loss": 0.0619356632232666,
      "step": 2919
    },
    {
      "epoch": 0.39477464384094096,
      "grad_norm": 0.19647294282913208,
      "learning_rate": 2.1826622814035138e-05,
      "loss": 0.09703421592712402,
      "step": 2920
    },
    {
      "epoch": 0.3949098406367769,
      "grad_norm": 0.21972203254699707,
      "learning_rate": 2.1820527577067293e-05,
      "loss": 0.06958174705505371,
      "step": 2921
    },
    {
      "epoch": 0.3950450374326128,
      "grad_norm": 0.15243977308273315,
      "learning_rate": 2.1814430920029238e-05,
      "loss": 0.07746529579162598,
      "step": 2922
    },
    {
      "epoch": 0.3951802342284488,
      "grad_norm": 0.15289008617401123,
      "learning_rate": 2.1808332844190325e-05,
      "loss": 0.05689287185668945,
      "step": 2923
    },
    {
      "epoch": 0.39531543102428474,
      "grad_norm": 0.0488365963101387,
      "learning_rate": 2.1802233350820203e-05,
      "loss": 0.047292470932006836,
      "step": 2924
    },
    {
      "epoch": 0.39545062782012064,
      "grad_norm": 0.10398261994123459,
      "learning_rate": 2.179613244118883e-05,
      "loss": 0.06886279582977295,
      "step": 2925
    },
    {
      "epoch": 0.3955858246159566,
      "grad_norm": 0.09976505488157272,
      "learning_rate": 2.1790030116566436e-05,
      "loss": 0.03988444805145264,
      "step": 2926
    },
    {
      "epoch": 0.39572102141179255,
      "grad_norm": 0.09861832857131958,
      "learning_rate": 2.1783926378223563e-05,
      "loss": 0.075400710105896,
      "step": 2927
    },
    {
      "epoch": 0.39585621820762845,
      "grad_norm": 0.16030851006507874,
      "learning_rate": 2.1777821227431048e-05,
      "loss": 0.05587649345397949,
      "step": 2928
    },
    {
      "epoch": 0.3959914150034644,
      "grad_norm": 0.061079803854227066,
      "learning_rate": 2.1771714665460005e-05,
      "loss": 0.07207679748535156,
      "step": 2929
    },
    {
      "epoch": 0.39612661179930037,
      "grad_norm": 0.29675906896591187,
      "learning_rate": 2.1765606693581857e-05,
      "loss": 0.10237550735473633,
      "step": 2930
    },
    {
      "epoch": 0.3962618085951363,
      "grad_norm": 0.1858254224061966,
      "learning_rate": 2.1759497313068316e-05,
      "loss": 0.07250165939331055,
      "step": 2931
    },
    {
      "epoch": 0.3963970053909722,
      "grad_norm": 0.0911153256893158,
      "learning_rate": 2.175338652519139e-05,
      "loss": 0.08556985855102539,
      "step": 2932
    },
    {
      "epoch": 0.3965322021868082,
      "grad_norm": 0.10770085453987122,
      "learning_rate": 2.1747274331223377e-05,
      "loss": 0.08372950553894043,
      "step": 2933
    },
    {
      "epoch": 0.39666739898264414,
      "grad_norm": 0.07733134180307388,
      "learning_rate": 2.1741160732436865e-05,
      "loss": 0.044725894927978516,
      "step": 2934
    },
    {
      "epoch": 0.39680259577848004,
      "grad_norm": 0.2189149707555771,
      "learning_rate": 2.1735045730104746e-05,
      "loss": 0.07374417781829834,
      "step": 2935
    },
    {
      "epoch": 0.396937792574316,
      "grad_norm": 0.12623330950737,
      "learning_rate": 2.1728929325500183e-05,
      "loss": 0.08131217956542969,
      "step": 2936
    },
    {
      "epoch": 0.39707298937015195,
      "grad_norm": 0.08285139501094818,
      "learning_rate": 2.1722811519896654e-05,
      "loss": 0.08355093002319336,
      "step": 2937
    },
    {
      "epoch": 0.39720818616598785,
      "grad_norm": 0.08076751232147217,
      "learning_rate": 2.171669231456792e-05,
      "loss": 0.07858467102050781,
      "step": 2938
    },
    {
      "epoch": 0.3973433829618238,
      "grad_norm": 0.051190897822380066,
      "learning_rate": 2.1710571710788025e-05,
      "loss": 0.0547715425491333,
      "step": 2939
    },
    {
      "epoch": 0.39747857975765977,
      "grad_norm": 0.07591031491756439,
      "learning_rate": 2.1704449709831312e-05,
      "loss": 0.07062768936157227,
      "step": 2940
    },
    {
      "epoch": 0.39761377655349567,
      "grad_norm": 0.12319187074899673,
      "learning_rate": 2.1698326312972423e-05,
      "loss": 0.05763721466064453,
      "step": 2941
    },
    {
      "epoch": 0.3977489733493316,
      "grad_norm": 0.07787314802408218,
      "learning_rate": 2.1692201521486268e-05,
      "loss": 0.045623779296875,
      "step": 2942
    },
    {
      "epoch": 0.3978841701451676,
      "grad_norm": 0.1041032075881958,
      "learning_rate": 2.1686075336648075e-05,
      "loss": 0.07411837577819824,
      "step": 2943
    },
    {
      "epoch": 0.3980193669410035,
      "grad_norm": 0.09960998594760895,
      "learning_rate": 2.167994775973334e-05,
      "loss": 0.1112370491027832,
      "step": 2944
    },
    {
      "epoch": 0.39815456373683944,
      "grad_norm": 0.03396177664399147,
      "learning_rate": 2.167381879201786e-05,
      "loss": 0.03413820266723633,
      "step": 2945
    },
    {
      "epoch": 0.3982897605326754,
      "grad_norm": 0.16394257545471191,
      "learning_rate": 2.166768843477772e-05,
      "loss": 0.07999277114868164,
      "step": 2946
    },
    {
      "epoch": 0.3984249573285113,
      "grad_norm": 0.11818178743124008,
      "learning_rate": 2.166155668928929e-05,
      "loss": 0.05094313621520996,
      "step": 2947
    },
    {
      "epoch": 0.39856015412434725,
      "grad_norm": 0.05330347269773483,
      "learning_rate": 2.1655423556829233e-05,
      "loss": 0.06194496154785156,
      "step": 2948
    },
    {
      "epoch": 0.3986953509201832,
      "grad_norm": 0.05445524677634239,
      "learning_rate": 2.1649289038674504e-05,
      "loss": 0.06589770317077637,
      "step": 2949
    },
    {
      "epoch": 0.3988305477160191,
      "grad_norm": 0.11717861145734787,
      "learning_rate": 2.1643153136102333e-05,
      "loss": 0.08285784721374512,
      "step": 2950
    },
    {
      "epoch": 0.39896574451185507,
      "grad_norm": 0.09164632111787796,
      "learning_rate": 2.1637015850390255e-05,
      "loss": 0.0628896951675415,
      "step": 2951
    },
    {
      "epoch": 0.399100941307691,
      "grad_norm": 0.15552611649036407,
      "learning_rate": 2.1630877182816087e-05,
      "loss": 0.12502717971801758,
      "step": 2952
    },
    {
      "epoch": 0.3992361381035269,
      "grad_norm": 0.04887017980217934,
      "learning_rate": 2.162473713465793e-05,
      "loss": 0.05484938621520996,
      "step": 2953
    },
    {
      "epoch": 0.3993713348993629,
      "grad_norm": 0.1604534387588501,
      "learning_rate": 2.161859570719417e-05,
      "loss": 0.07022333145141602,
      "step": 2954
    },
    {
      "epoch": 0.39950653169519884,
      "grad_norm": 0.08292780071496964,
      "learning_rate": 2.161245290170349e-05,
      "loss": 0.06511497497558594,
      "step": 2955
    },
    {
      "epoch": 0.39964172849103474,
      "grad_norm": 0.09012793749570847,
      "learning_rate": 2.1606308719464858e-05,
      "loss": 0.07941818237304688,
      "step": 2956
    },
    {
      "epoch": 0.3997769252868707,
      "grad_norm": 0.06618606299161911,
      "learning_rate": 2.160016316175752e-05,
      "loss": 0.07085943222045898,
      "step": 2957
    },
    {
      "epoch": 0.39991212208270666,
      "grad_norm": 0.1177380159497261,
      "learning_rate": 2.159401622986101e-05,
      "loss": 0.0714254379272461,
      "step": 2958
    },
    {
      "epoch": 0.40004731887854256,
      "grad_norm": 0.18334154784679413,
      "learning_rate": 2.1587867925055165e-05,
      "loss": 0.0803985595703125,
      "step": 2959
    },
    {
      "epoch": 0.4001825156743785,
      "grad_norm": 0.12085236608982086,
      "learning_rate": 2.158171824862008e-05,
      "loss": 0.07714378833770752,
      "step": 2960
    },
    {
      "epoch": 0.40031771247021447,
      "grad_norm": 0.08710359036922455,
      "learning_rate": 2.157556720183616e-05,
      "loss": 0.04966241121292114,
      "step": 2961
    },
    {
      "epoch": 0.40045290926605037,
      "grad_norm": 0.05405594781041145,
      "learning_rate": 2.156941478598409e-05,
      "loss": 0.036066532135009766,
      "step": 2962
    },
    {
      "epoch": 0.40058810606188633,
      "grad_norm": 0.13081663846969604,
      "learning_rate": 2.156326100234482e-05,
      "loss": 0.0903012752532959,
      "step": 2963
    },
    {
      "epoch": 0.4007233028577223,
      "grad_norm": 0.08093271404504776,
      "learning_rate": 2.1557105852199612e-05,
      "loss": 0.10352659225463867,
      "step": 2964
    },
    {
      "epoch": 0.4008584996535582,
      "grad_norm": 0.07003793865442276,
      "learning_rate": 2.155094933683e-05,
      "loss": 0.0554126501083374,
      "step": 2965
    },
    {
      "epoch": 0.40099369644939414,
      "grad_norm": 0.1738732010126114,
      "learning_rate": 2.1544791457517802e-05,
      "loss": 0.07842874526977539,
      "step": 2966
    },
    {
      "epoch": 0.4011288932452301,
      "grad_norm": 0.06995101273059845,
      "learning_rate": 2.1538632215545126e-05,
      "loss": 0.06881844997406006,
      "step": 2967
    },
    {
      "epoch": 0.401264090041066,
      "grad_norm": 0.03176918253302574,
      "learning_rate": 2.153247161219435e-05,
      "loss": 0.03398263454437256,
      "step": 2968
    },
    {
      "epoch": 0.40139928683690196,
      "grad_norm": 0.07758156210184097,
      "learning_rate": 2.1526309648748147e-05,
      "loss": 0.04697573184967041,
      "step": 2969
    },
    {
      "epoch": 0.4015344836327379,
      "grad_norm": 0.05357353761792183,
      "learning_rate": 2.1520146326489476e-05,
      "loss": 0.049936771392822266,
      "step": 2970
    },
    {
      "epoch": 0.4016696804285738,
      "grad_norm": 0.08340733498334885,
      "learning_rate": 2.151398164670157e-05,
      "loss": 0.06646072864532471,
      "step": 2971
    },
    {
      "epoch": 0.4018048772244098,
      "grad_norm": 0.14851519465446472,
      "learning_rate": 2.1507815610667948e-05,
      "loss": 0.0814967155456543,
      "step": 2972
    },
    {
      "epoch": 0.40194007402024573,
      "grad_norm": 0.08107221871614456,
      "learning_rate": 2.1501648219672407e-05,
      "loss": 0.059036970138549805,
      "step": 2973
    },
    {
      "epoch": 0.40207527081608163,
      "grad_norm": 0.074784055352211,
      "learning_rate": 2.149547947499904e-05,
      "loss": 0.04662632942199707,
      "step": 2974
    },
    {
      "epoch": 0.4022104676119176,
      "grad_norm": 0.22925560176372528,
      "learning_rate": 2.1489309377932212e-05,
      "loss": 0.07279229164123535,
      "step": 2975
    },
    {
      "epoch": 0.40234566440775354,
      "grad_norm": 0.06163569912314415,
      "learning_rate": 2.1483137929756562e-05,
      "loss": 0.06318807601928711,
      "step": 2976
    },
    {
      "epoch": 0.4024808612035895,
      "grad_norm": 0.15652644634246826,
      "learning_rate": 2.147696513175702e-05,
      "loss": 0.07147181034088135,
      "step": 2977
    },
    {
      "epoch": 0.4026160579994254,
      "grad_norm": 0.14739644527435303,
      "learning_rate": 2.1470790985218804e-05,
      "loss": 0.04479563236236572,
      "step": 2978
    },
    {
      "epoch": 0.40275125479526136,
      "grad_norm": 0.0683426484465599,
      "learning_rate": 2.1464615491427393e-05,
      "loss": 0.06860971450805664,
      "step": 2979
    },
    {
      "epoch": 0.4028864515910973,
      "grad_norm": 0.20358186960220337,
      "learning_rate": 2.1458438651668567e-05,
      "loss": 0.05639910697937012,
      "step": 2980
    },
    {
      "epoch": 0.4030216483869332,
      "grad_norm": 0.07618564367294312,
      "learning_rate": 2.1452260467228376e-05,
      "loss": 0.07398104667663574,
      "step": 2981
    },
    {
      "epoch": 0.4031568451827692,
      "grad_norm": 0.10755223780870438,
      "learning_rate": 2.144608093939314e-05,
      "loss": 0.08925628662109375,
      "step": 2982
    },
    {
      "epoch": 0.40329204197860513,
      "grad_norm": 0.10787951201200485,
      "learning_rate": 2.1439900069449483e-05,
      "loss": 0.06861186027526855,
      "step": 2983
    },
    {
      "epoch": 0.40342723877444103,
      "grad_norm": 0.12498243898153305,
      "learning_rate": 2.1433717858684286e-05,
      "loss": 0.0492253303527832,
      "step": 2984
    },
    {
      "epoch": 0.403562435570277,
      "grad_norm": 0.15943868458271027,
      "learning_rate": 2.1427534308384724e-05,
      "loss": 0.07301950454711914,
      "step": 2985
    },
    {
      "epoch": 0.40369763236611295,
      "grad_norm": 0.31179332733154297,
      "learning_rate": 2.1421349419838245e-05,
      "loss": 0.10101604461669922,
      "step": 2986
    },
    {
      "epoch": 0.40383282916194885,
      "grad_norm": 0.06472857296466827,
      "learning_rate": 2.1415163194332574e-05,
      "loss": 0.0504913330078125,
      "step": 2987
    },
    {
      "epoch": 0.4039680259577848,
      "grad_norm": 0.10403521358966827,
      "learning_rate": 2.1408975633155715e-05,
      "loss": 0.06492233276367188,
      "step": 2988
    },
    {
      "epoch": 0.40410322275362076,
      "grad_norm": 0.22043445706367493,
      "learning_rate": 2.140278673759595e-05,
      "loss": 0.064263254404068,
      "step": 2989
    },
    {
      "epoch": 0.40423841954945666,
      "grad_norm": 0.024387631565332413,
      "learning_rate": 2.1396596508941847e-05,
      "loss": 0.023431196808815002,
      "step": 2990
    },
    {
      "epoch": 0.4043736163452926,
      "grad_norm": 0.09069811552762985,
      "learning_rate": 2.1390404948482238e-05,
      "loss": 0.057845115661621094,
      "step": 2991
    },
    {
      "epoch": 0.4045088131411286,
      "grad_norm": 0.0715799331665039,
      "learning_rate": 2.1384212057506243e-05,
      "loss": 0.06095367670059204,
      "step": 2992
    },
    {
      "epoch": 0.4046440099369645,
      "grad_norm": 0.10685287415981293,
      "learning_rate": 2.137801783730325e-05,
      "loss": 0.07075166702270508,
      "step": 2993
    },
    {
      "epoch": 0.40477920673280043,
      "grad_norm": 0.09337018430233002,
      "learning_rate": 2.137182228916293e-05,
      "loss": 0.06181478500366211,
      "step": 2994
    },
    {
      "epoch": 0.4049144035286364,
      "grad_norm": 0.09772127866744995,
      "learning_rate": 2.136562541437523e-05,
      "loss": 0.06584453582763672,
      "step": 2995
    },
    {
      "epoch": 0.4050496003244723,
      "grad_norm": 0.09671685099601746,
      "learning_rate": 2.135942721423038e-05,
      "loss": 0.06240963935852051,
      "step": 2996
    },
    {
      "epoch": 0.40518479712030825,
      "grad_norm": 0.03251372277736664,
      "learning_rate": 2.1353227690018865e-05,
      "loss": 0.03559398651123047,
      "step": 2997
    },
    {
      "epoch": 0.4053199939161442,
      "grad_norm": 0.07866518944501877,
      "learning_rate": 2.1347026843031467e-05,
      "loss": 0.05701172351837158,
      "step": 2998
    },
    {
      "epoch": 0.4054551907119801,
      "grad_norm": 0.061463821679353714,
      "learning_rate": 2.1340824674559238e-05,
      "loss": 0.06513738632202148,
      "step": 2999
    },
    {
      "epoch": 0.40559038750781606,
      "grad_norm": 0.06971432268619537,
      "learning_rate": 2.133462118589349e-05,
      "loss": 0.05636489391326904,
      "step": 3000
    },
    {
      "epoch": 0.405725584303652,
      "grad_norm": 0.08442901819944382,
      "learning_rate": 2.1328416378325837e-05,
      "loss": 0.0807790756225586,
      "step": 3001
    },
    {
      "epoch": 0.4058607810994879,
      "grad_norm": 0.07448405772447586,
      "learning_rate": 2.1322210253148144e-05,
      "loss": 0.05698251724243164,
      "step": 3002
    },
    {
      "epoch": 0.4059959778953239,
      "grad_norm": 0.05302705615758896,
      "learning_rate": 2.131600281165257e-05,
      "loss": 0.041414737701416016,
      "step": 3003
    },
    {
      "epoch": 0.40613117469115984,
      "grad_norm": 0.18194712698459625,
      "learning_rate": 2.130979405513152e-05,
      "loss": 0.07084739208221436,
      "step": 3004
    },
    {
      "epoch": 0.40626637148699574,
      "grad_norm": 0.08297908306121826,
      "learning_rate": 2.1303583984877697e-05,
      "loss": 0.05286884307861328,
      "step": 3005
    },
    {
      "epoch": 0.4064015682828317,
      "grad_norm": 0.10639271140098572,
      "learning_rate": 2.1297372602184085e-05,
      "loss": 0.08077192306518555,
      "step": 3006
    },
    {
      "epoch": 0.40653676507866765,
      "grad_norm": 0.14989924430847168,
      "learning_rate": 2.1291159908343907e-05,
      "loss": 0.06470870971679688,
      "step": 3007
    },
    {
      "epoch": 0.40667196187450355,
      "grad_norm": 0.07326214015483856,
      "learning_rate": 2.1284945904650693e-05,
      "loss": 0.04383516311645508,
      "step": 3008
    },
    {
      "epoch": 0.4068071586703395,
      "grad_norm": 0.12765485048294067,
      "learning_rate": 2.127873059239822e-05,
      "loss": 0.06995940208435059,
      "step": 3009
    },
    {
      "epoch": 0.40694235546617546,
      "grad_norm": 0.13427044451236725,
      "learning_rate": 2.127251397288056e-05,
      "loss": 0.081390380859375,
      "step": 3010
    },
    {
      "epoch": 0.40707755226201137,
      "grad_norm": 0.062342721968889236,
      "learning_rate": 2.126629604739204e-05,
      "loss": 0.06158982962369919,
      "step": 3011
    },
    {
      "epoch": 0.4072127490578473,
      "grad_norm": 0.10195564478635788,
      "learning_rate": 2.1260076817227268e-05,
      "loss": 0.09209346771240234,
      "step": 3012
    },
    {
      "epoch": 0.4073479458536833,
      "grad_norm": 0.07800912111997604,
      "learning_rate": 2.1253856283681122e-05,
      "loss": 0.03464120626449585,
      "step": 3013
    },
    {
      "epoch": 0.4074831426495192,
      "grad_norm": 0.08662211149930954,
      "learning_rate": 2.1247634448048743e-05,
      "loss": 0.05074167251586914,
      "step": 3014
    },
    {
      "epoch": 0.40761833944535514,
      "grad_norm": 0.07274608314037323,
      "learning_rate": 2.1241411311625562e-05,
      "loss": 0.05540144443511963,
      "step": 3015
    },
    {
      "epoch": 0.4077535362411911,
      "grad_norm": 0.08997748792171478,
      "learning_rate": 2.1235186875707257e-05,
      "loss": 0.04934501647949219,
      "step": 3016
    },
    {
      "epoch": 0.407888733037027,
      "grad_norm": 0.09159883111715317,
      "learning_rate": 2.1228961141589797e-05,
      "loss": 0.06996488571166992,
      "step": 3017
    },
    {
      "epoch": 0.40802392983286295,
      "grad_norm": 0.12758347392082214,
      "learning_rate": 2.122273411056941e-05,
      "loss": 0.05107855796813965,
      "step": 3018
    },
    {
      "epoch": 0.4081591266286989,
      "grad_norm": 0.09249735623598099,
      "learning_rate": 2.1216505783942592e-05,
      "loss": 0.06343221664428711,
      "step": 3019
    },
    {
      "epoch": 0.40829432342453487,
      "grad_norm": 0.056661177426576614,
      "learning_rate": 2.121027616300613e-05,
      "loss": 0.04540300369262695,
      "step": 3020
    },
    {
      "epoch": 0.40842952022037077,
      "grad_norm": 0.07007092237472534,
      "learning_rate": 2.1204045249057043e-05,
      "loss": 0.043569087982177734,
      "step": 3021
    },
    {
      "epoch": 0.4085647170162067,
      "grad_norm": 0.09861729294061661,
      "learning_rate": 2.119781304339266e-05,
      "loss": 0.041123151779174805,
      "step": 3022
    },
    {
      "epoch": 0.4086999138120427,
      "grad_norm": 0.18184809386730194,
      "learning_rate": 2.1191579547310547e-05,
      "loss": 0.11762666702270508,
      "step": 3023
    },
    {
      "epoch": 0.4088351106078786,
      "grad_norm": 0.09003068506717682,
      "learning_rate": 2.1185344762108556e-05,
      "loss": 0.0914621353149414,
      "step": 3024
    },
    {
      "epoch": 0.40897030740371454,
      "grad_norm": 0.09968771040439606,
      "learning_rate": 2.11791086890848e-05,
      "loss": 0.10293006896972656,
      "step": 3025
    },
    {
      "epoch": 0.4091055041995505,
      "grad_norm": 0.08220849186182022,
      "learning_rate": 2.1172871329537662e-05,
      "loss": 0.0676412582397461,
      "step": 3026
    },
    {
      "epoch": 0.4092407009953864,
      "grad_norm": 0.12748676538467407,
      "learning_rate": 2.1166632684765794e-05,
      "loss": 0.05331754684448242,
      "step": 3027
    },
    {
      "epoch": 0.40937589779122235,
      "grad_norm": 0.1278143674135208,
      "learning_rate": 2.1160392756068124e-05,
      "loss": 0.054271697998046875,
      "step": 3028
    },
    {
      "epoch": 0.4095110945870583,
      "grad_norm": 0.054544832557439804,
      "learning_rate": 2.1154151544743826e-05,
      "loss": 0.05579090118408203,
      "step": 3029
    },
    {
      "epoch": 0.4096462913828942,
      "grad_norm": 0.20833535492420197,
      "learning_rate": 2.114790905209236e-05,
      "loss": 0.09582304954528809,
      "step": 3030
    },
    {
      "epoch": 0.40978148817873017,
      "grad_norm": 0.15027658641338348,
      "learning_rate": 2.1141665279413444e-05,
      "loss": 0.10325431823730469,
      "step": 3031
    },
    {
      "epoch": 0.4099166849745661,
      "grad_norm": 0.10712064057588577,
      "learning_rate": 2.1135420228007062e-05,
      "loss": 0.0791773796081543,
      "step": 3032
    },
    {
      "epoch": 0.410051881770402,
      "grad_norm": 0.159246027469635,
      "learning_rate": 2.1129173899173474e-05,
      "loss": 0.10382843017578125,
      "step": 3033
    },
    {
      "epoch": 0.410187078566238,
      "grad_norm": 0.1293485313653946,
      "learning_rate": 2.11229262942132e-05,
      "loss": 0.09114742279052734,
      "step": 3034
    },
    {
      "epoch": 0.41032227536207394,
      "grad_norm": 0.06148974969983101,
      "learning_rate": 2.1116677414427008e-05,
      "loss": 0.07535743713378906,
      "step": 3035
    },
    {
      "epoch": 0.41045747215790984,
      "grad_norm": 0.17812968790531158,
      "learning_rate": 2.1110427261115972e-05,
      "loss": 0.08387422561645508,
      "step": 3036
    },
    {
      "epoch": 0.4105926689537458,
      "grad_norm": 0.11709120124578476,
      "learning_rate": 2.1104175835581386e-05,
      "loss": 0.0662384033203125,
      "step": 3037
    },
    {
      "epoch": 0.41072786574958176,
      "grad_norm": 0.041039351373910904,
      "learning_rate": 2.1097923139124846e-05,
      "loss": 0.03850674629211426,
      "step": 3038
    },
    {
      "epoch": 0.41086306254541766,
      "grad_norm": 0.0684625580906868,
      "learning_rate": 2.109166917304819e-05,
      "loss": 0.0749208927154541,
      "step": 3039
    },
    {
      "epoch": 0.4109982593412536,
      "grad_norm": 0.05245031788945198,
      "learning_rate": 2.1085413938653532e-05,
      "loss": 0.06904721260070801,
      "step": 3040
    },
    {
      "epoch": 0.41113345613708957,
      "grad_norm": 0.08892738819122314,
      "learning_rate": 2.107915743724323e-05,
      "loss": 0.07693004608154297,
      "step": 3041
    },
    {
      "epoch": 0.41126865293292547,
      "grad_norm": 0.11836464703083038,
      "learning_rate": 2.1072899670119935e-05,
      "loss": 0.07848429679870605,
      "step": 3042
    },
    {
      "epoch": 0.41140384972876143,
      "grad_norm": 0.09845896810293198,
      "learning_rate": 2.1066640638586543e-05,
      "loss": 0.09078264236450195,
      "step": 3043
    },
    {
      "epoch": 0.4115390465245974,
      "grad_norm": 0.04389701783657074,
      "learning_rate": 2.1060380343946223e-05,
      "loss": 0.04076647758483887,
      "step": 3044
    },
    {
      "epoch": 0.4116742433204333,
      "grad_norm": 0.07464593648910522,
      "learning_rate": 2.10541187875024e-05,
      "loss": 0.04718589782714844,
      "step": 3045
    },
    {
      "epoch": 0.41180944011626924,
      "grad_norm": 0.13523316383361816,
      "learning_rate": 2.1047855970558753e-05,
      "loss": 0.07450461387634277,
      "step": 3046
    },
    {
      "epoch": 0.4119446369121052,
      "grad_norm": 0.0772295892238617,
      "learning_rate": 2.1041591894419244e-05,
      "loss": 0.05384349822998047,
      "step": 3047
    },
    {
      "epoch": 0.4120798337079411,
      "grad_norm": 0.06032535433769226,
      "learning_rate": 2.1035326560388087e-05,
      "loss": 0.06676864624023438,
      "step": 3048
    },
    {
      "epoch": 0.41221503050377706,
      "grad_norm": 0.09851662069559097,
      "learning_rate": 2.1029059969769756e-05,
      "loss": 0.08440446853637695,
      "step": 3049
    },
    {
      "epoch": 0.412350227299613,
      "grad_norm": 0.0773877426981926,
      "learning_rate": 2.1022792123868986e-05,
      "loss": 0.05698060989379883,
      "step": 3050
    },
    {
      "epoch": 0.4124854240954489,
      "grad_norm": 0.19656893610954285,
      "learning_rate": 2.1016523023990783e-05,
      "loss": 0.10703563690185547,
      "step": 3051
    },
    {
      "epoch": 0.4126206208912849,
      "grad_norm": 0.07784394919872284,
      "learning_rate": 2.1010252671440398e-05,
      "loss": 0.06422150135040283,
      "step": 3052
    },
    {
      "epoch": 0.41275581768712083,
      "grad_norm": 0.05273876711726189,
      "learning_rate": 2.1003981067523358e-05,
      "loss": 0.04862689971923828,
      "step": 3053
    },
    {
      "epoch": 0.41289101448295673,
      "grad_norm": 0.11838555335998535,
      "learning_rate": 2.099770821354544e-05,
      "loss": 0.08938407897949219,
      "step": 3054
    },
    {
      "epoch": 0.4130262112787927,
      "grad_norm": 0.057645510882139206,
      "learning_rate": 2.0991434110812692e-05,
      "loss": 0.05691719055175781,
      "step": 3055
    },
    {
      "epoch": 0.41316140807462864,
      "grad_norm": 0.04171757772564888,
      "learning_rate": 2.0985158760631415e-05,
      "loss": 0.0449824333190918,
      "step": 3056
    },
    {
      "epoch": 0.41329660487046455,
      "grad_norm": 0.2051403671503067,
      "learning_rate": 2.0978882164308157e-05,
      "loss": 0.10453391075134277,
      "step": 3057
    },
    {
      "epoch": 0.4134318016663005,
      "grad_norm": 0.17624957859516144,
      "learning_rate": 2.0972604323149755e-05,
      "loss": 0.0975494384765625,
      "step": 3058
    },
    {
      "epoch": 0.41356699846213646,
      "grad_norm": 0.060458704829216,
      "learning_rate": 2.0966325238463283e-05,
      "loss": 0.061492919921875,
      "step": 3059
    },
    {
      "epoch": 0.41370219525797236,
      "grad_norm": 0.039016854017972946,
      "learning_rate": 2.096004491155608e-05,
      "loss": 0.0387115478515625,
      "step": 3060
    },
    {
      "epoch": 0.4138373920538083,
      "grad_norm": 0.05725531652569771,
      "learning_rate": 2.0953763343735746e-05,
      "loss": 0.06633472442626953,
      "step": 3061
    },
    {
      "epoch": 0.4139725888496443,
      "grad_norm": 0.1099138855934143,
      "learning_rate": 2.0947480536310133e-05,
      "loss": 0.03399014472961426,
      "step": 3062
    },
    {
      "epoch": 0.4141077856454802,
      "grad_norm": 0.1203671395778656,
      "learning_rate": 2.0941196490587352e-05,
      "loss": 0.04614830017089844,
      "step": 3063
    },
    {
      "epoch": 0.41424298244131613,
      "grad_norm": 0.08886481821537018,
      "learning_rate": 2.0934911207875782e-05,
      "loss": 0.05822920799255371,
      "step": 3064
    },
    {
      "epoch": 0.4143781792371521,
      "grad_norm": 0.05011489987373352,
      "learning_rate": 2.092862468948405e-05,
      "loss": 0.05697751045227051,
      "step": 3065
    },
    {
      "epoch": 0.41451337603298805,
      "grad_norm": 0.1723695546388626,
      "learning_rate": 2.0922336936721044e-05,
      "loss": 0.07545614242553711,
      "step": 3066
    },
    {
      "epoch": 0.41464857282882395,
      "grad_norm": 0.11021208763122559,
      "learning_rate": 2.0916047950895907e-05,
      "loss": 0.1034536361694336,
      "step": 3067
    },
    {
      "epoch": 0.4147837696246599,
      "grad_norm": 0.05389266088604927,
      "learning_rate": 2.0909757733318035e-05,
      "loss": 0.03635275363922119,
      "step": 3068
    },
    {
      "epoch": 0.41491896642049586,
      "grad_norm": 0.15880133211612701,
      "learning_rate": 2.090346628529709e-05,
      "loss": 0.08521342277526855,
      "step": 3069
    },
    {
      "epoch": 0.41505416321633176,
      "grad_norm": 0.05554840713739395,
      "learning_rate": 2.089717360814298e-05,
      "loss": 0.056429386138916016,
      "step": 3070
    },
    {
      "epoch": 0.4151893600121677,
      "grad_norm": 0.15616926550865173,
      "learning_rate": 2.0890879703165885e-05,
      "loss": 0.11056041717529297,
      "step": 3071
    },
    {
      "epoch": 0.4153245568080037,
      "grad_norm": 0.11596349626779556,
      "learning_rate": 2.0884584571676217e-05,
      "loss": 0.04823112487792969,
      "step": 3072
    },
    {
      "epoch": 0.4154597536038396,
      "grad_norm": 0.07576422393321991,
      "learning_rate": 2.0878288214984657e-05,
      "loss": 0.07398319244384766,
      "step": 3073
    },
    {
      "epoch": 0.41559495039967553,
      "grad_norm": 0.06998582184314728,
      "learning_rate": 2.0871990634402147e-05,
      "loss": 0.06373214721679688,
      "step": 3074
    },
    {
      "epoch": 0.4157301471955115,
      "grad_norm": 0.07030367106199265,
      "learning_rate": 2.0865691831239877e-05,
      "loss": 0.06789588928222656,
      "step": 3075
    },
    {
      "epoch": 0.4158653439913474,
      "grad_norm": 0.12691596150398254,
      "learning_rate": 2.0859391806809285e-05,
      "loss": 0.06635773181915283,
      "step": 3076
    },
    {
      "epoch": 0.41600054078718335,
      "grad_norm": 0.11216427385807037,
      "learning_rate": 2.0853090562422072e-05,
      "loss": 0.09018993377685547,
      "step": 3077
    },
    {
      "epoch": 0.4161357375830193,
      "grad_norm": 0.07276362180709839,
      "learning_rate": 2.084678809939019e-05,
      "loss": 0.06460797786712646,
      "step": 3078
    },
    {
      "epoch": 0.4162709343788552,
      "grad_norm": 0.057856395840644836,
      "learning_rate": 2.084048441902585e-05,
      "loss": 0.07272624969482422,
      "step": 3079
    },
    {
      "epoch": 0.41640613117469116,
      "grad_norm": 0.08151490241289139,
      "learning_rate": 2.0834179522641508e-05,
      "loss": 0.05999279022216797,
      "step": 3080
    },
    {
      "epoch": 0.4165413279705271,
      "grad_norm": 0.07579455524682999,
      "learning_rate": 2.0827873411549877e-05,
      "loss": 0.054164767265319824,
      "step": 3081
    },
    {
      "epoch": 0.416676524766363,
      "grad_norm": 0.04379725083708763,
      "learning_rate": 2.0821566087063926e-05,
      "loss": 0.04554009437561035,
      "step": 3082
    },
    {
      "epoch": 0.416811721562199,
      "grad_norm": 0.04158473014831543,
      "learning_rate": 2.081525755049687e-05,
      "loss": 0.04493856430053711,
      "step": 3083
    },
    {
      "epoch": 0.41694691835803493,
      "grad_norm": 0.06890937685966492,
      "learning_rate": 2.0808947803162182e-05,
      "loss": 0.0759425163269043,
      "step": 3084
    },
    {
      "epoch": 0.41708211515387084,
      "grad_norm": 0.12793216109275818,
      "learning_rate": 2.0802636846373578e-05,
      "loss": 0.060241103172302246,
      "step": 3085
    },
    {
      "epoch": 0.4172173119497068,
      "grad_norm": 0.06333158165216446,
      "learning_rate": 2.0796324681445045e-05,
      "loss": 0.05533003807067871,
      "step": 3086
    },
    {
      "epoch": 0.41735250874554275,
      "grad_norm": 0.04647061973810196,
      "learning_rate": 2.0790011309690806e-05,
      "loss": 0.05628824234008789,
      "step": 3087
    },
    {
      "epoch": 0.41748770554137865,
      "grad_norm": 0.08324585109949112,
      "learning_rate": 2.0783696732425332e-05,
      "loss": 0.07022237777709961,
      "step": 3088
    },
    {
      "epoch": 0.4176229023372146,
      "grad_norm": 0.08278854936361313,
      "learning_rate": 2.0777380950963355e-05,
      "loss": 0.05677366256713867,
      "step": 3089
    },
    {
      "epoch": 0.41775809913305056,
      "grad_norm": 0.11014974862337112,
      "learning_rate": 2.0771063966619854e-05,
      "loss": 0.059134483337402344,
      "step": 3090
    },
    {
      "epoch": 0.41789329592888647,
      "grad_norm": 0.0685177892446518,
      "learning_rate": 2.0764745780710065e-05,
      "loss": 0.06299197673797607,
      "step": 3091
    },
    {
      "epoch": 0.4180284927247224,
      "grad_norm": 0.06992512196302414,
      "learning_rate": 2.075842639454946e-05,
      "loss": 0.04918193817138672,
      "step": 3092
    },
    {
      "epoch": 0.4181636895205584,
      "grad_norm": 0.07779309898614883,
      "learning_rate": 2.075210580945378e-05,
      "loss": 0.042852044105529785,
      "step": 3093
    },
    {
      "epoch": 0.4182988863163943,
      "grad_norm": 0.10292708873748779,
      "learning_rate": 2.0745784026738984e-05,
      "loss": 0.08461833000183105,
      "step": 3094
    },
    {
      "epoch": 0.41843408311223024,
      "grad_norm": 0.06557980179786682,
      "learning_rate": 2.073946104772132e-05,
      "loss": 0.056448280811309814,
      "step": 3095
    },
    {
      "epoch": 0.4185692799080662,
      "grad_norm": 0.13443908095359802,
      "learning_rate": 2.0733136873717258e-05,
      "loss": 0.0901947021484375,
      "step": 3096
    },
    {
      "epoch": 0.4187044767039021,
      "grad_norm": 0.09556220471858978,
      "learning_rate": 2.0726811506043527e-05,
      "loss": 0.06488943099975586,
      "step": 3097
    },
    {
      "epoch": 0.41883967349973805,
      "grad_norm": 0.05614485219120979,
      "learning_rate": 2.0720484946017104e-05,
      "loss": 0.04627394676208496,
      "step": 3098
    },
    {
      "epoch": 0.418974870295574,
      "grad_norm": 0.06654493510723114,
      "learning_rate": 2.0714157194955202e-05,
      "loss": 0.07684820890426636,
      "step": 3099
    },
    {
      "epoch": 0.4191100670914099,
      "grad_norm": 0.1605784296989441,
      "learning_rate": 2.070782825417531e-05,
      "loss": 0.08035087585449219,
      "step": 3100
    },
    {
      "epoch": 0.41924526388724587,
      "grad_norm": 0.042471446096897125,
      "learning_rate": 2.0701498124995127e-05,
      "loss": 0.037381887435913086,
      "step": 3101
    },
    {
      "epoch": 0.4193804606830818,
      "grad_norm": 0.08313640207052231,
      "learning_rate": 2.069516680873264e-05,
      "loss": 0.08044379949569702,
      "step": 3102
    },
    {
      "epoch": 0.4195156574789177,
      "grad_norm": 0.08323860913515091,
      "learning_rate": 2.0688834306706047e-05,
      "loss": 0.07319927215576172,
      "step": 3103
    },
    {
      "epoch": 0.4196508542747537,
      "grad_norm": 0.14642830193042755,
      "learning_rate": 2.0682500620233815e-05,
      "loss": 0.06599903106689453,
      "step": 3104
    },
    {
      "epoch": 0.41978605107058964,
      "grad_norm": 0.06447198241949081,
      "learning_rate": 2.0676165750634656e-05,
      "loss": 0.06340980529785156,
      "step": 3105
    },
    {
      "epoch": 0.41992124786642554,
      "grad_norm": 0.11369089037179947,
      "learning_rate": 2.0669829699227513e-05,
      "loss": 0.049335360527038574,
      "step": 3106
    },
    {
      "epoch": 0.4200564446622615,
      "grad_norm": 0.12140129506587982,
      "learning_rate": 2.06634924673316e-05,
      "loss": 0.0637502670288086,
      "step": 3107
    },
    {
      "epoch": 0.42019164145809745,
      "grad_norm": 0.06305120140314102,
      "learning_rate": 2.0657154056266346e-05,
      "loss": 0.09333229064941406,
      "step": 3108
    },
    {
      "epoch": 0.42032683825393335,
      "grad_norm": 0.06813465803861618,
      "learning_rate": 2.0650814467351452e-05,
      "loss": 0.06344270706176758,
      "step": 3109
    },
    {
      "epoch": 0.4204620350497693,
      "grad_norm": 0.06209234148263931,
      "learning_rate": 2.064447370190685e-05,
      "loss": 0.06512045860290527,
      "step": 3110
    },
    {
      "epoch": 0.42059723184560527,
      "grad_norm": 0.07012920081615448,
      "learning_rate": 2.0638131761252724e-05,
      "loss": 0.053237199783325195,
      "step": 3111
    },
    {
      "epoch": 0.4207324286414412,
      "grad_norm": 0.056594476103782654,
      "learning_rate": 2.06317886467095e-05,
      "loss": 0.07568359375,
      "step": 3112
    },
    {
      "epoch": 0.4208676254372771,
      "grad_norm": 0.058603040874004364,
      "learning_rate": 2.0625444359597847e-05,
      "loss": 0.05091691017150879,
      "step": 3113
    },
    {
      "epoch": 0.4210028222331131,
      "grad_norm": 0.109632208943367,
      "learning_rate": 2.0619098901238684e-05,
      "loss": 0.11093711853027344,
      "step": 3114
    },
    {
      "epoch": 0.42113801902894904,
      "grad_norm": 0.11282766610383987,
      "learning_rate": 2.0612752272953158e-05,
      "loss": 0.10464859008789062,
      "step": 3115
    },
    {
      "epoch": 0.42127321582478494,
      "grad_norm": 0.03744662553071976,
      "learning_rate": 2.060640447606268e-05,
      "loss": 0.04370832443237305,
      "step": 3116
    },
    {
      "epoch": 0.4214084126206209,
      "grad_norm": 0.0604991540312767,
      "learning_rate": 2.0600055511888895e-05,
      "loss": 0.05349850654602051,
      "step": 3117
    },
    {
      "epoch": 0.42154360941645685,
      "grad_norm": 0.07266171276569366,
      "learning_rate": 2.059370538175369e-05,
      "loss": 0.055145978927612305,
      "step": 3118
    },
    {
      "epoch": 0.42167880621229276,
      "grad_norm": 0.044544413685798645,
      "learning_rate": 2.0587354086979194e-05,
      "loss": 0.06481552124023438,
      "step": 3119
    },
    {
      "epoch": 0.4218140030081287,
      "grad_norm": 0.0836453065276146,
      "learning_rate": 2.0581001628887785e-05,
      "loss": 0.06637883186340332,
      "step": 3120
    },
    {
      "epoch": 0.42194919980396467,
      "grad_norm": 0.04422401636838913,
      "learning_rate": 2.057464800880207e-05,
      "loss": 0.03454303741455078,
      "step": 3121
    },
    {
      "epoch": 0.42208439659980057,
      "grad_norm": 0.10634706169366837,
      "learning_rate": 2.0568293228044914e-05,
      "loss": 0.05202054977416992,
      "step": 3122
    },
    {
      "epoch": 0.4222195933956365,
      "grad_norm": 0.10487055778503418,
      "learning_rate": 2.0561937287939413e-05,
      "loss": 0.056520938873291016,
      "step": 3123
    },
    {
      "epoch": 0.4223547901914725,
      "grad_norm": 0.044058021157979965,
      "learning_rate": 2.055558018980891e-05,
      "loss": 0.04020202159881592,
      "step": 3124
    },
    {
      "epoch": 0.4224899869873084,
      "grad_norm": 0.03371758759021759,
      "learning_rate": 2.0549221934976987e-05,
      "loss": 0.03473806381225586,
      "step": 3125
    },
    {
      "epoch": 0.42262518378314434,
      "grad_norm": 0.10781261324882507,
      "learning_rate": 2.054286252476746e-05,
      "loss": 0.0739436149597168,
      "step": 3126
    },
    {
      "epoch": 0.4227603805789803,
      "grad_norm": 0.06343436241149902,
      "learning_rate": 2.05365019605044e-05,
      "loss": 0.05147576332092285,
      "step": 3127
    },
    {
      "epoch": 0.4228955773748162,
      "grad_norm": 0.10348464548587799,
      "learning_rate": 2.053014024351211e-05,
      "loss": 0.06627631187438965,
      "step": 3128
    },
    {
      "epoch": 0.42303077417065216,
      "grad_norm": 0.06274273991584778,
      "learning_rate": 2.0523777375115133e-05,
      "loss": 0.053835391998291016,
      "step": 3129
    },
    {
      "epoch": 0.4231659709664881,
      "grad_norm": 0.07987461984157562,
      "learning_rate": 2.0517413356638245e-05,
      "loss": 0.05454719066619873,
      "step": 3130
    },
    {
      "epoch": 0.423301167762324,
      "grad_norm": 0.08394049853086472,
      "learning_rate": 2.0511048189406472e-05,
      "loss": 0.04127311706542969,
      "step": 3131
    },
    {
      "epoch": 0.42343636455815997,
      "grad_norm": 0.08531752228736877,
      "learning_rate": 2.0504681874745082e-05,
      "loss": 0.07720017433166504,
      "step": 3132
    },
    {
      "epoch": 0.42357156135399593,
      "grad_norm": 0.13521486520767212,
      "learning_rate": 2.049831441397957e-05,
      "loss": 0.09921646118164062,
      "step": 3133
    },
    {
      "epoch": 0.42370675814983183,
      "grad_norm": 0.0751941055059433,
      "learning_rate": 2.0491945808435674e-05,
      "loss": 0.04781484603881836,
      "step": 3134
    },
    {
      "epoch": 0.4238419549456678,
      "grad_norm": 0.03657032176852226,
      "learning_rate": 2.048557605943938e-05,
      "loss": 0.040822505950927734,
      "step": 3135
    },
    {
      "epoch": 0.42397715174150374,
      "grad_norm": 0.17417150735855103,
      "learning_rate": 2.047920516831689e-05,
      "loss": 0.07330513000488281,
      "step": 3136
    },
    {
      "epoch": 0.42411234853733965,
      "grad_norm": 0.07247629016637802,
      "learning_rate": 2.047283313639467e-05,
      "loss": 0.05410957336425781,
      "step": 3137
    },
    {
      "epoch": 0.4242475453331756,
      "grad_norm": 0.13214074075222015,
      "learning_rate": 2.0466459964999408e-05,
      "loss": 0.06297755241394043,
      "step": 3138
    },
    {
      "epoch": 0.42438274212901156,
      "grad_norm": 0.0893910676240921,
      "learning_rate": 2.0460085655458025e-05,
      "loss": 0.05625462532043457,
      "step": 3139
    },
    {
      "epoch": 0.42451793892484746,
      "grad_norm": 0.07778891921043396,
      "learning_rate": 2.0453710209097697e-05,
      "loss": 0.0769956111907959,
      "step": 3140
    },
    {
      "epoch": 0.4246531357206834,
      "grad_norm": 0.16175420582294464,
      "learning_rate": 2.044733362724582e-05,
      "loss": 0.05966007709503174,
      "step": 3141
    },
    {
      "epoch": 0.4247883325165194,
      "grad_norm": 0.07387705147266388,
      "learning_rate": 2.0440955911230028e-05,
      "loss": 0.03690910339355469,
      "step": 3142
    },
    {
      "epoch": 0.4249235293123553,
      "grad_norm": 0.11351793259382248,
      "learning_rate": 2.0434577062378203e-05,
      "loss": 0.06286334991455078,
      "step": 3143
    },
    {
      "epoch": 0.42505872610819123,
      "grad_norm": 0.16056591272354126,
      "learning_rate": 2.0428197082018458e-05,
      "loss": 0.05270254611968994,
      "step": 3144
    },
    {
      "epoch": 0.4251939229040272,
      "grad_norm": 0.08143115788698196,
      "learning_rate": 2.042181597147913e-05,
      "loss": 0.05297136306762695,
      "step": 3145
    },
    {
      "epoch": 0.4253291196998631,
      "grad_norm": 0.11065438389778137,
      "learning_rate": 2.0415433732088806e-05,
      "loss": 0.08269381523132324,
      "step": 3146
    },
    {
      "epoch": 0.42546431649569905,
      "grad_norm": 0.053662966936826706,
      "learning_rate": 2.0409050365176294e-05,
      "loss": 0.049225687980651855,
      "step": 3147
    },
    {
      "epoch": 0.425599513291535,
      "grad_norm": 0.18104280531406403,
      "learning_rate": 2.0402665872070656e-05,
      "loss": 0.09186267852783203,
      "step": 3148
    },
    {
      "epoch": 0.4257347100873709,
      "grad_norm": 0.0901026725769043,
      "learning_rate": 2.0396280254101172e-05,
      "loss": 0.04329252243041992,
      "step": 3149
    },
    {
      "epoch": 0.42586990688320686,
      "grad_norm": 0.2154507040977478,
      "learning_rate": 2.0389893512597364e-05,
      "loss": 0.07400131225585938,
      "step": 3150
    },
    {
      "epoch": 0.4260051036790428,
      "grad_norm": 0.16561947762966156,
      "learning_rate": 2.0383505648888986e-05,
      "loss": 0.055444300174713135,
      "step": 3151
    },
    {
      "epoch": 0.4261403004748787,
      "grad_norm": 0.10395420342683792,
      "learning_rate": 2.037711666430602e-05,
      "loss": 0.07625532150268555,
      "step": 3152
    },
    {
      "epoch": 0.4262754972707147,
      "grad_norm": 0.09603693336248398,
      "learning_rate": 2.0370726560178693e-05,
      "loss": 0.04709172248840332,
      "step": 3153
    },
    {
      "epoch": 0.42641069406655063,
      "grad_norm": 0.06602743268013,
      "learning_rate": 2.036433533783745e-05,
      "loss": 0.05215644836425781,
      "step": 3154
    },
    {
      "epoch": 0.42654589086238653,
      "grad_norm": 0.10673072934150696,
      "learning_rate": 2.0357942998612988e-05,
      "loss": 0.06460332870483398,
      "step": 3155
    },
    {
      "epoch": 0.4266810876582225,
      "grad_norm": 0.08656342327594757,
      "learning_rate": 2.0351549543836224e-05,
      "loss": 0.07187271118164062,
      "step": 3156
    },
    {
      "epoch": 0.42681628445405845,
      "grad_norm": 0.11255121976137161,
      "learning_rate": 2.0345154974838307e-05,
      "loss": 0.047976016998291016,
      "step": 3157
    },
    {
      "epoch": 0.4269514812498944,
      "grad_norm": 0.08434443920850754,
      "learning_rate": 2.0338759292950618e-05,
      "loss": 0.049024343490600586,
      "step": 3158
    },
    {
      "epoch": 0.4270866780457303,
      "grad_norm": 0.2571490705013275,
      "learning_rate": 2.033236249950477e-05,
      "loss": 0.07988381385803223,
      "step": 3159
    },
    {
      "epoch": 0.42722187484156626,
      "grad_norm": 0.05089796707034111,
      "learning_rate": 2.0325964595832618e-05,
      "loss": 0.06427013874053955,
      "step": 3160
    },
    {
      "epoch": 0.4273570716374022,
      "grad_norm": 0.1525297909975052,
      "learning_rate": 2.031956558326624e-05,
      "loss": 0.08524465560913086,
      "step": 3161
    },
    {
      "epoch": 0.4274922684332381,
      "grad_norm": 0.0746731162071228,
      "learning_rate": 2.0313165463137935e-05,
      "loss": 0.07829952239990234,
      "step": 3162
    },
    {
      "epoch": 0.4276274652290741,
      "grad_norm": 0.08205119520425797,
      "learning_rate": 2.030676423678025e-05,
      "loss": 0.0600128173828125,
      "step": 3163
    },
    {
      "epoch": 0.42776266202491003,
      "grad_norm": 0.09793049097061157,
      "learning_rate": 2.030036190552595e-05,
      "loss": 0.03784537315368652,
      "step": 3164
    },
    {
      "epoch": 0.42789785882074594,
      "grad_norm": 0.1096339076757431,
      "learning_rate": 2.029395847070803e-05,
      "loss": 0.05259275436401367,
      "step": 3165
    },
    {
      "epoch": 0.4280330556165819,
      "grad_norm": 0.15999886393547058,
      "learning_rate": 2.0287553933659735e-05,
      "loss": 0.06393766403198242,
      "step": 3166
    },
    {
      "epoch": 0.42816825241241785,
      "grad_norm": 0.1600424200296402,
      "learning_rate": 2.0281148295714512e-05,
      "loss": 0.0844869613647461,
      "step": 3167
    },
    {
      "epoch": 0.42830344920825375,
      "grad_norm": 0.16639955341815948,
      "learning_rate": 2.027474155820605e-05,
      "loss": 0.06110811233520508,
      "step": 3168
    },
    {
      "epoch": 0.4284386460040897,
      "grad_norm": 0.06218131259083748,
      "learning_rate": 2.026833372246827e-05,
      "loss": 0.09666824340820312,
      "step": 3169
    },
    {
      "epoch": 0.42857384279992566,
      "grad_norm": 0.058496762067079544,
      "learning_rate": 2.026192478983531e-05,
      "loss": 0.05681562423706055,
      "step": 3170
    },
    {
      "epoch": 0.42870903959576157,
      "grad_norm": 0.10382314026355743,
      "learning_rate": 2.0255514761641555e-05,
      "loss": 0.07334446907043457,
      "step": 3171
    },
    {
      "epoch": 0.4288442363915975,
      "grad_norm": 0.1503184586763382,
      "learning_rate": 2.0249103639221597e-05,
      "loss": 0.08085298538208008,
      "step": 3172
    },
    {
      "epoch": 0.4289794331874335,
      "grad_norm": 0.13700683414936066,
      "learning_rate": 2.024269142391027e-05,
      "loss": 0.06449460983276367,
      "step": 3173
    },
    {
      "epoch": 0.4291146299832694,
      "grad_norm": 0.1608564853668213,
      "learning_rate": 2.023627811704263e-05,
      "loss": 0.08159399032592773,
      "step": 3174
    },
    {
      "epoch": 0.42924982677910534,
      "grad_norm": 0.12306592613458633,
      "learning_rate": 2.0229863719953963e-05,
      "loss": 0.07640337944030762,
      "step": 3175
    },
    {
      "epoch": 0.4293850235749413,
      "grad_norm": 0.0657508447766304,
      "learning_rate": 2.0223448233979785e-05,
      "loss": 0.050688326358795166,
      "step": 3176
    },
    {
      "epoch": 0.4295202203707772,
      "grad_norm": 0.22151495516300201,
      "learning_rate": 2.0217031660455825e-05,
      "loss": 0.09482574462890625,
      "step": 3177
    },
    {
      "epoch": 0.42965541716661315,
      "grad_norm": 0.09951341897249222,
      "learning_rate": 2.0210614000718054e-05,
      "loss": 0.0622866153717041,
      "step": 3178
    },
    {
      "epoch": 0.4297906139624491,
      "grad_norm": 0.12187718600034714,
      "learning_rate": 2.020419525610266e-05,
      "loss": 0.08787405490875244,
      "step": 3179
    },
    {
      "epoch": 0.429925810758285,
      "grad_norm": 0.09063244611024857,
      "learning_rate": 2.0197775427946066e-05,
      "loss": 0.09056663513183594,
      "step": 3180
    },
    {
      "epoch": 0.43006100755412097,
      "grad_norm": 0.04932304844260216,
      "learning_rate": 2.0191354517584902e-05,
      "loss": 0.06115436553955078,
      "step": 3181
    },
    {
      "epoch": 0.4301962043499569,
      "grad_norm": 0.20993757247924805,
      "learning_rate": 2.018493252635605e-05,
      "loss": 0.08200836181640625,
      "step": 3182
    },
    {
      "epoch": 0.4303314011457928,
      "grad_norm": 0.15816952288150787,
      "learning_rate": 2.0178509455596598e-05,
      "loss": 0.06627660989761353,
      "step": 3183
    },
    {
      "epoch": 0.4304665979416288,
      "grad_norm": 0.04997331649065018,
      "learning_rate": 2.017208530664386e-05,
      "loss": 0.057770729064941406,
      "step": 3184
    },
    {
      "epoch": 0.43060179473746474,
      "grad_norm": 0.06886984407901764,
      "learning_rate": 2.016566008083538e-05,
      "loss": 0.07120418548583984,
      "step": 3185
    },
    {
      "epoch": 0.43073699153330064,
      "grad_norm": 0.1024257019162178,
      "learning_rate": 2.0159233779508923e-05,
      "loss": 0.07142853736877441,
      "step": 3186
    },
    {
      "epoch": 0.4308721883291366,
      "grad_norm": 0.06449129432439804,
      "learning_rate": 2.0152806404002482e-05,
      "loss": 0.053707122802734375,
      "step": 3187
    },
    {
      "epoch": 0.43100738512497255,
      "grad_norm": 0.1538398265838623,
      "learning_rate": 2.014637795565427e-05,
      "loss": 0.08022022247314453,
      "step": 3188
    },
    {
      "epoch": 0.43114258192080845,
      "grad_norm": 0.1447172462940216,
      "learning_rate": 2.0139948435802722e-05,
      "loss": 0.0802949070930481,
      "step": 3189
    },
    {
      "epoch": 0.4312777787166444,
      "grad_norm": 0.05420884117484093,
      "learning_rate": 2.0133517845786504e-05,
      "loss": 0.06741786003112793,
      "step": 3190
    },
    {
      "epoch": 0.43141297551248037,
      "grad_norm": 0.0832863375544548,
      "learning_rate": 2.012708618694449e-05,
      "loss": 0.07708120346069336,
      "step": 3191
    },
    {
      "epoch": 0.43154817230831627,
      "grad_norm": 0.046726614236831665,
      "learning_rate": 2.0120653460615795e-05,
      "loss": 0.051119327545166016,
      "step": 3192
    },
    {
      "epoch": 0.4316833691041522,
      "grad_norm": 0.06133299320936203,
      "learning_rate": 2.011421966813974e-05,
      "loss": 0.04952883720397949,
      "step": 3193
    },
    {
      "epoch": 0.4318185658999882,
      "grad_norm": 0.10175460577011108,
      "learning_rate": 2.0107784810855882e-05,
      "loss": 0.06792628765106201,
      "step": 3194
    },
    {
      "epoch": 0.4319537626958241,
      "grad_norm": 0.0926133245229721,
      "learning_rate": 2.0101348890103985e-05,
      "loss": 0.054627180099487305,
      "step": 3195
    },
    {
      "epoch": 0.43208895949166004,
      "grad_norm": 0.05594244226813316,
      "learning_rate": 2.0094911907224043e-05,
      "loss": 0.06363487243652344,
      "step": 3196
    },
    {
      "epoch": 0.432224156287496,
      "grad_norm": 0.1287541687488556,
      "learning_rate": 2.008847386355628e-05,
      "loss": 0.07038569450378418,
      "step": 3197
    },
    {
      "epoch": 0.4323593530833319,
      "grad_norm": 0.1956535279750824,
      "learning_rate": 2.008203476044112e-05,
      "loss": 0.07847857475280762,
      "step": 3198
    },
    {
      "epoch": 0.43249454987916786,
      "grad_norm": 0.13023318350315094,
      "learning_rate": 2.007559459921922e-05,
      "loss": 0.07769918441772461,
      "step": 3199
    },
    {
      "epoch": 0.4326297466750038,
      "grad_norm": 0.12397795170545578,
      "learning_rate": 2.0069153381231456e-05,
      "loss": 0.060346126556396484,
      "step": 3200
    },
    {
      "epoch": 0.4327649434708397,
      "grad_norm": 0.08094791322946548,
      "learning_rate": 2.0062711107818933e-05,
      "loss": 0.06024932861328125,
      "step": 3201
    },
    {
      "epoch": 0.43290014026667567,
      "grad_norm": 0.0785946398973465,
      "learning_rate": 2.0056267780322953e-05,
      "loss": 0.046599388122558594,
      "step": 3202
    },
    {
      "epoch": 0.4330353370625116,
      "grad_norm": 0.08903083950281143,
      "learning_rate": 2.004982340008506e-05,
      "loss": 0.05499982833862305,
      "step": 3203
    },
    {
      "epoch": 0.4331705338583476,
      "grad_norm": 0.1369883418083191,
      "learning_rate": 2.004337796844701e-05,
      "loss": 0.04588675498962402,
      "step": 3204
    },
    {
      "epoch": 0.4333057306541835,
      "grad_norm": 0.09771862626075745,
      "learning_rate": 2.003693148675077e-05,
      "loss": 0.04064750671386719,
      "step": 3205
    },
    {
      "epoch": 0.43344092745001944,
      "grad_norm": 0.032162558287382126,
      "learning_rate": 2.003048395633853e-05,
      "loss": 0.03018641471862793,
      "step": 3206
    },
    {
      "epoch": 0.4335761242458554,
      "grad_norm": 0.10231100022792816,
      "learning_rate": 2.0024035378552708e-05,
      "loss": 0.11081218719482422,
      "step": 3207
    },
    {
      "epoch": 0.4337113210416913,
      "grad_norm": 0.08728881180286407,
      "learning_rate": 2.001758575473593e-05,
      "loss": 0.06610870361328125,
      "step": 3208
    },
    {
      "epoch": 0.43384651783752726,
      "grad_norm": 0.09966468065977097,
      "learning_rate": 2.0011135086231042e-05,
      "loss": 0.08336687088012695,
      "step": 3209
    },
    {
      "epoch": 0.4339817146333632,
      "grad_norm": 0.0684678927063942,
      "learning_rate": 2.0004683374381104e-05,
      "loss": 0.08991813659667969,
      "step": 3210
    },
    {
      "epoch": 0.4341169114291991,
      "grad_norm": 0.130214661359787,
      "learning_rate": 1.9998230620529395e-05,
      "loss": 0.07419872283935547,
      "step": 3211
    },
    {
      "epoch": 0.43425210822503507,
      "grad_norm": 0.16386649012565613,
      "learning_rate": 1.999177682601942e-05,
      "loss": 0.07621908187866211,
      "step": 3212
    },
    {
      "epoch": 0.43438730502087103,
      "grad_norm": 0.20553043484687805,
      "learning_rate": 1.9985321992194896e-05,
      "loss": 0.11010169982910156,
      "step": 3213
    },
    {
      "epoch": 0.43452250181670693,
      "grad_norm": 0.03922649845480919,
      "learning_rate": 1.9978866120399746e-05,
      "loss": 0.04037795960903168,
      "step": 3214
    },
    {
      "epoch": 0.4346576986125429,
      "grad_norm": 0.03622576594352722,
      "learning_rate": 1.9972409211978116e-05,
      "loss": 0.04348158836364746,
      "step": 3215
    },
    {
      "epoch": 0.43479289540837884,
      "grad_norm": 0.08675309270620346,
      "learning_rate": 1.9965951268274373e-05,
      "loss": 0.07869338989257812,
      "step": 3216
    },
    {
      "epoch": 0.43492809220421474,
      "grad_norm": 0.08813906461000443,
      "learning_rate": 1.9959492290633093e-05,
      "loss": 0.051762521266937256,
      "step": 3217
    },
    {
      "epoch": 0.4350632890000507,
      "grad_norm": 0.07535872608423233,
      "learning_rate": 1.995303228039907e-05,
      "loss": 0.06920510530471802,
      "step": 3218
    },
    {
      "epoch": 0.43519848579588666,
      "grad_norm": 0.09170832484960556,
      "learning_rate": 1.994657123891732e-05,
      "loss": 0.07438278198242188,
      "step": 3219
    },
    {
      "epoch": 0.43533368259172256,
      "grad_norm": 0.07532388716936111,
      "learning_rate": 1.9940109167533055e-05,
      "loss": 0.06286758184432983,
      "step": 3220
    },
    {
      "epoch": 0.4354688793875585,
      "grad_norm": 0.17018432915210724,
      "learning_rate": 1.9933646067591716e-05,
      "loss": 0.07802581787109375,
      "step": 3221
    },
    {
      "epoch": 0.4356040761833945,
      "grad_norm": 0.08761341124773026,
      "learning_rate": 1.992718194043896e-05,
      "loss": 0.06822872161865234,
      "step": 3222
    },
    {
      "epoch": 0.4357392729792304,
      "grad_norm": 0.1064404621720314,
      "learning_rate": 1.9920716787420643e-05,
      "loss": 0.06493830680847168,
      "step": 3223
    },
    {
      "epoch": 0.43587446977506633,
      "grad_norm": 0.0616588331758976,
      "learning_rate": 1.9914250609882858e-05,
      "loss": 0.06285524368286133,
      "step": 3224
    },
    {
      "epoch": 0.4360096665709023,
      "grad_norm": 0.19811654090881348,
      "learning_rate": 1.9907783409171885e-05,
      "loss": 0.09122323989868164,
      "step": 3225
    },
    {
      "epoch": 0.4361448633667382,
      "grad_norm": 0.0374748669564724,
      "learning_rate": 1.990131518663424e-05,
      "loss": 0.036103010177612305,
      "step": 3226
    },
    {
      "epoch": 0.43628006016257415,
      "grad_norm": 0.06356500834226608,
      "learning_rate": 1.9894845943616632e-05,
      "loss": 0.05052900314331055,
      "step": 3227
    },
    {
      "epoch": 0.4364152569584101,
      "grad_norm": 0.09023895114660263,
      "learning_rate": 1.988837568146599e-05,
      "loss": 0.057244300842285156,
      "step": 3228
    },
    {
      "epoch": 0.436550453754246,
      "grad_norm": 0.08295368403196335,
      "learning_rate": 1.988190440152947e-05,
      "loss": 0.07534408569335938,
      "step": 3229
    },
    {
      "epoch": 0.43668565055008196,
      "grad_norm": 0.17162348330020905,
      "learning_rate": 1.9875432105154424e-05,
      "loss": 0.08234882354736328,
      "step": 3230
    },
    {
      "epoch": 0.4368208473459179,
      "grad_norm": 0.1472318470478058,
      "learning_rate": 1.9868958793688412e-05,
      "loss": 0.09856009483337402,
      "step": 3231
    },
    {
      "epoch": 0.4369560441417538,
      "grad_norm": 0.11385153234004974,
      "learning_rate": 1.9862484468479213e-05,
      "loss": 0.07649481296539307,
      "step": 3232
    },
    {
      "epoch": 0.4370912409375898,
      "grad_norm": 0.0535292886197567,
      "learning_rate": 1.985600913087482e-05,
      "loss": 0.03593182563781738,
      "step": 3233
    },
    {
      "epoch": 0.43722643773342573,
      "grad_norm": 0.05281436815857887,
      "learning_rate": 1.9849532782223425e-05,
      "loss": 0.04095196723937988,
      "step": 3234
    },
    {
      "epoch": 0.43736163452926163,
      "grad_norm": 0.08454808592796326,
      "learning_rate": 1.9843055423873447e-05,
      "loss": 0.0487523078918457,
      "step": 3235
    },
    {
      "epoch": 0.4374968313250976,
      "grad_norm": 0.04947862774133682,
      "learning_rate": 1.9836577057173507e-05,
      "loss": 0.05003643035888672,
      "step": 3236
    },
    {
      "epoch": 0.43763202812093355,
      "grad_norm": 0.12157605588436127,
      "learning_rate": 1.9830097683472427e-05,
      "loss": 0.07188990712165833,
      "step": 3237
    },
    {
      "epoch": 0.43776722491676945,
      "grad_norm": 0.1331806778907776,
      "learning_rate": 1.9823617304119252e-05,
      "loss": 0.06160330772399902,
      "step": 3238
    },
    {
      "epoch": 0.4379024217126054,
      "grad_norm": 0.03333786129951477,
      "learning_rate": 1.9817135920463232e-05,
      "loss": 0.025844454765319824,
      "step": 3239
    },
    {
      "epoch": 0.43803761850844136,
      "grad_norm": 0.0759047269821167,
      "learning_rate": 1.9810653533853826e-05,
      "loss": 0.07703280448913574,
      "step": 3240
    },
    {
      "epoch": 0.43817281530427726,
      "grad_norm": 0.043087806552648544,
      "learning_rate": 1.9804170145640706e-05,
      "loss": 0.05115079879760742,
      "step": 3241
    },
    {
      "epoch": 0.4383080121001132,
      "grad_norm": 0.09004665911197662,
      "learning_rate": 1.9797685757173737e-05,
      "loss": 0.06872403621673584,
      "step": 3242
    },
    {
      "epoch": 0.4384432088959492,
      "grad_norm": 0.18874123692512512,
      "learning_rate": 1.979120036980301e-05,
      "loss": 0.06737768650054932,
      "step": 3243
    },
    {
      "epoch": 0.4385784056917851,
      "grad_norm": 0.11173944920301437,
      "learning_rate": 1.9784713984878814e-05,
      "loss": 0.04859170317649841,
      "step": 3244
    },
    {
      "epoch": 0.43871360248762103,
      "grad_norm": 0.046070925891399384,
      "learning_rate": 1.9778226603751652e-05,
      "loss": 0.05103862285614014,
      "step": 3245
    },
    {
      "epoch": 0.438848799283457,
      "grad_norm": 0.15152396261692047,
      "learning_rate": 1.9771738227772235e-05,
      "loss": 0.07271099090576172,
      "step": 3246
    },
    {
      "epoch": 0.4389839960792929,
      "grad_norm": 0.06952129304409027,
      "learning_rate": 1.976524885829147e-05,
      "loss": 0.05684375762939453,
      "step": 3247
    },
    {
      "epoch": 0.43911919287512885,
      "grad_norm": 0.08088558167219162,
      "learning_rate": 1.975875849666048e-05,
      "loss": 0.0633234977722168,
      "step": 3248
    },
    {
      "epoch": 0.4392543896709648,
      "grad_norm": 0.07711949944496155,
      "learning_rate": 1.9752267144230595e-05,
      "loss": 0.0728921890258789,
      "step": 3249
    },
    {
      "epoch": 0.43938958646680076,
      "grad_norm": 0.07124711573123932,
      "learning_rate": 1.9745774802353347e-05,
      "loss": 0.06157541275024414,
      "step": 3250
    },
    {
      "epoch": 0.43952478326263666,
      "grad_norm": 0.159086674451828,
      "learning_rate": 1.973928147238048e-05,
      "loss": 0.05113077163696289,
      "step": 3251
    },
    {
      "epoch": 0.4396599800584726,
      "grad_norm": 0.10507015138864517,
      "learning_rate": 1.973278715566394e-05,
      "loss": 0.049019575119018555,
      "step": 3252
    },
    {
      "epoch": 0.4397951768543086,
      "grad_norm": 0.07731273770332336,
      "learning_rate": 1.972629185355587e-05,
      "loss": 0.06448245048522949,
      "step": 3253
    },
    {
      "epoch": 0.4399303736501445,
      "grad_norm": 0.07900913059711456,
      "learning_rate": 1.971979556740864e-05,
      "loss": 0.05887889862060547,
      "step": 3254
    },
    {
      "epoch": 0.44006557044598044,
      "grad_norm": 0.07361288368701935,
      "learning_rate": 1.971329829857479e-05,
      "loss": 0.07146072387695312,
      "step": 3255
    },
    {
      "epoch": 0.4402007672418164,
      "grad_norm": 0.06809293478727341,
      "learning_rate": 1.9706800048407112e-05,
      "loss": 0.03605484962463379,
      "step": 3256
    },
    {
      "epoch": 0.4403359640376523,
      "grad_norm": 0.21496723592281342,
      "learning_rate": 1.9700300818258566e-05,
      "loss": 0.072593092918396,
      "step": 3257
    },
    {
      "epoch": 0.44047116083348825,
      "grad_norm": 0.05230354145169258,
      "learning_rate": 1.9693800609482318e-05,
      "loss": 0.05963575839996338,
      "step": 3258
    },
    {
      "epoch": 0.4406063576293242,
      "grad_norm": 0.056048981845378876,
      "learning_rate": 1.9687299423431754e-05,
      "loss": 0.05493319034576416,
      "step": 3259
    },
    {
      "epoch": 0.4407415544251601,
      "grad_norm": 0.12063869088888168,
      "learning_rate": 1.968079726146045e-05,
      "loss": 0.07966184616088867,
      "step": 3260
    },
    {
      "epoch": 0.44087675122099607,
      "grad_norm": 0.12758807837963104,
      "learning_rate": 1.9674294124922204e-05,
      "loss": 0.08736658096313477,
      "step": 3261
    },
    {
      "epoch": 0.441011948016832,
      "grad_norm": 0.05945262312889099,
      "learning_rate": 1.966779001517099e-05,
      "loss": 0.06276941299438477,
      "step": 3262
    },
    {
      "epoch": 0.4411471448126679,
      "grad_norm": 0.04829312488436699,
      "learning_rate": 1.9661284933561006e-05,
      "loss": 0.047865867614746094,
      "step": 3263
    },
    {
      "epoch": 0.4412823416085039,
      "grad_norm": 0.13132239878177643,
      "learning_rate": 1.9654778881446636e-05,
      "loss": 0.07096004486083984,
      "step": 3264
    },
    {
      "epoch": 0.44141753840433984,
      "grad_norm": 0.0730091780424118,
      "learning_rate": 1.9648271860182487e-05,
      "loss": 0.07902050018310547,
      "step": 3265
    },
    {
      "epoch": 0.44155273520017574,
      "grad_norm": 0.05608237534761429,
      "learning_rate": 1.9641763871123345e-05,
      "loss": 0.07141232490539551,
      "step": 3266
    },
    {
      "epoch": 0.4416879319960117,
      "grad_norm": 0.09248071163892746,
      "learning_rate": 1.963525491562421e-05,
      "loss": 0.07359836250543594,
      "step": 3267
    },
    {
      "epoch": 0.44182312879184765,
      "grad_norm": 0.0771331861615181,
      "learning_rate": 1.9628744995040287e-05,
      "loss": 0.0667567253112793,
      "step": 3268
    },
    {
      "epoch": 0.44195832558768355,
      "grad_norm": 0.099677175283432,
      "learning_rate": 1.9622234110726976e-05,
      "loss": 0.06462430953979492,
      "step": 3269
    },
    {
      "epoch": 0.4420935223835195,
      "grad_norm": 0.07637973129749298,
      "learning_rate": 1.9615722264039868e-05,
      "loss": 0.06608521938323975,
      "step": 3270
    },
    {
      "epoch": 0.44222871917935547,
      "grad_norm": 0.08768682926893234,
      "learning_rate": 1.9609209456334772e-05,
      "loss": 0.046433210372924805,
      "step": 3271
    },
    {
      "epoch": 0.44236391597519137,
      "grad_norm": 0.07879055291414261,
      "learning_rate": 1.960269568896769e-05,
      "loss": 0.06598556041717529,
      "step": 3272
    },
    {
      "epoch": 0.4424991127710273,
      "grad_norm": 0.062198273837566376,
      "learning_rate": 1.9596180963294822e-05,
      "loss": 0.04973268508911133,
      "step": 3273
    },
    {
      "epoch": 0.4426343095668633,
      "grad_norm": 0.06544417887926102,
      "learning_rate": 1.9589665280672564e-05,
      "loss": 0.04590177536010742,
      "step": 3274
    },
    {
      "epoch": 0.4427695063626992,
      "grad_norm": 0.07249431312084198,
      "learning_rate": 1.958314864245752e-05,
      "loss": 0.05933821201324463,
      "step": 3275
    },
    {
      "epoch": 0.44290470315853514,
      "grad_norm": 0.03899597004055977,
      "learning_rate": 1.957663105000649e-05,
      "loss": 0.03536057472229004,
      "step": 3276
    },
    {
      "epoch": 0.4430398999543711,
      "grad_norm": 0.11802103370428085,
      "learning_rate": 1.957011250467647e-05,
      "loss": 0.06655168533325195,
      "step": 3277
    },
    {
      "epoch": 0.443175096750207,
      "grad_norm": 0.06084742397069931,
      "learning_rate": 1.9563593007824658e-05,
      "loss": 0.0634080171585083,
      "step": 3278
    },
    {
      "epoch": 0.44331029354604296,
      "grad_norm": 0.05009220540523529,
      "learning_rate": 1.9557072560808442e-05,
      "loss": 0.0459136962890625,
      "step": 3279
    },
    {
      "epoch": 0.4434454903418789,
      "grad_norm": 0.1192793995141983,
      "learning_rate": 1.9550551164985418e-05,
      "loss": 0.04971790313720703,
      "step": 3280
    },
    {
      "epoch": 0.4435806871377148,
      "grad_norm": 0.1543719321489334,
      "learning_rate": 1.9544028821713372e-05,
      "loss": 0.07663488388061523,
      "step": 3281
    },
    {
      "epoch": 0.44371588393355077,
      "grad_norm": 0.04591715335845947,
      "learning_rate": 1.9537505532350298e-05,
      "loss": 0.043121337890625,
      "step": 3282
    },
    {
      "epoch": 0.4438510807293867,
      "grad_norm": 0.21003416180610657,
      "learning_rate": 1.9530981298254376e-05,
      "loss": 0.06580352783203125,
      "step": 3283
    },
    {
      "epoch": 0.44398627752522263,
      "grad_norm": 0.06348355859518051,
      "learning_rate": 1.9524456120783983e-05,
      "loss": 0.04552507400512695,
      "step": 3284
    },
    {
      "epoch": 0.4441214743210586,
      "grad_norm": 0.12992259860038757,
      "learning_rate": 1.95179300012977e-05,
      "loss": 0.08132648468017578,
      "step": 3285
    },
    {
      "epoch": 0.44425667111689454,
      "grad_norm": 0.06173097714781761,
      "learning_rate": 1.9511402941154296e-05,
      "loss": 0.05788033455610275,
      "step": 3286
    },
    {
      "epoch": 0.44439186791273044,
      "grad_norm": 0.09726239740848541,
      "learning_rate": 1.950487494171274e-05,
      "loss": 0.07055425643920898,
      "step": 3287
    },
    {
      "epoch": 0.4445270647085664,
      "grad_norm": 0.08492317795753479,
      "learning_rate": 1.9498346004332203e-05,
      "loss": 0.04618768393993378,
      "step": 3288
    },
    {
      "epoch": 0.44466226150440236,
      "grad_norm": 0.18086524307727814,
      "learning_rate": 1.949181613037204e-05,
      "loss": 0.0637044906616211,
      "step": 3289
    },
    {
      "epoch": 0.44479745830023826,
      "grad_norm": 0.18768538534641266,
      "learning_rate": 1.9485285321191804e-05,
      "loss": 0.07714366912841797,
      "step": 3290
    },
    {
      "epoch": 0.4449326550960742,
      "grad_norm": 0.06693324446678162,
      "learning_rate": 1.9478753578151244e-05,
      "loss": 0.05836009979248047,
      "step": 3291
    },
    {
      "epoch": 0.44506785189191017,
      "grad_norm": 0.17425455152988434,
      "learning_rate": 1.9472220902610304e-05,
      "loss": 0.06835460662841797,
      "step": 3292
    },
    {
      "epoch": 0.44520304868774613,
      "grad_norm": 0.24480070173740387,
      "learning_rate": 1.9465687295929127e-05,
      "loss": 0.0700838565826416,
      "step": 3293
    },
    {
      "epoch": 0.44533824548358203,
      "grad_norm": 0.12210825830698013,
      "learning_rate": 1.945915275946804e-05,
      "loss": 0.04314112663269043,
      "step": 3294
    },
    {
      "epoch": 0.445473442279418,
      "grad_norm": 0.06020324304699898,
      "learning_rate": 1.9452617294587573e-05,
      "loss": 0.05254793167114258,
      "step": 3295
    },
    {
      "epoch": 0.44560863907525394,
      "grad_norm": 0.12664490938186646,
      "learning_rate": 1.9446080902648435e-05,
      "loss": 0.07109546661376953,
      "step": 3296
    },
    {
      "epoch": 0.44574383587108984,
      "grad_norm": 0.038578931242227554,
      "learning_rate": 1.943954358501154e-05,
      "loss": 0.035285115242004395,
      "step": 3297
    },
    {
      "epoch": 0.4458790326669258,
      "grad_norm": 0.054506756365299225,
      "learning_rate": 1.943300534303801e-05,
      "loss": 0.0599825382232666,
      "step": 3298
    },
    {
      "epoch": 0.44601422946276176,
      "grad_norm": 0.08307227492332458,
      "learning_rate": 1.9426466178089116e-05,
      "loss": 0.07606077194213867,
      "step": 3299
    },
    {
      "epoch": 0.44614942625859766,
      "grad_norm": 0.06323505192995071,
      "learning_rate": 1.9419926091526367e-05,
      "loss": 0.08138561248779297,
      "step": 3300
    },
    {
      "epoch": 0.4462846230544336,
      "grad_norm": 0.04589042812585831,
      "learning_rate": 1.9413385084711425e-05,
      "loss": 0.06054115295410156,
      "step": 3301
    },
    {
      "epoch": 0.4464198198502696,
      "grad_norm": 0.09724168479442596,
      "learning_rate": 1.9406843159006183e-05,
      "loss": 0.07733547687530518,
      "step": 3302
    },
    {
      "epoch": 0.4465550166461055,
      "grad_norm": 0.10368669778108597,
      "learning_rate": 1.940030031577269e-05,
      "loss": 0.06560420989990234,
      "step": 3303
    },
    {
      "epoch": 0.44669021344194143,
      "grad_norm": 0.052629485726356506,
      "learning_rate": 1.9393756556373212e-05,
      "loss": 0.0666041374206543,
      "step": 3304
    },
    {
      "epoch": 0.4468254102377774,
      "grad_norm": 0.09566310793161392,
      "learning_rate": 1.9387211882170184e-05,
      "loss": 0.07320308685302734,
      "step": 3305
    },
    {
      "epoch": 0.4469606070336133,
      "grad_norm": 0.16783958673477173,
      "learning_rate": 1.9380666294526243e-05,
      "loss": 0.0665731430053711,
      "step": 3306
    },
    {
      "epoch": 0.44709580382944925,
      "grad_norm": 0.08109188824892044,
      "learning_rate": 1.9374119794804228e-05,
      "loss": 0.05227309465408325,
      "step": 3307
    },
    {
      "epoch": 0.4472310006252852,
      "grad_norm": 0.058383334428071976,
      "learning_rate": 1.936757238436714e-05,
      "loss": 0.0552821159362793,
      "step": 3308
    },
    {
      "epoch": 0.4473661974211211,
      "grad_norm": 0.10600022226572037,
      "learning_rate": 1.936102406457819e-05,
      "loss": 0.07664573192596436,
      "step": 3309
    },
    {
      "epoch": 0.44750139421695706,
      "grad_norm": 0.08078666031360626,
      "learning_rate": 1.935447483680078e-05,
      "loss": 0.05944967269897461,
      "step": 3310
    },
    {
      "epoch": 0.447636591012793,
      "grad_norm": 0.10073341429233551,
      "learning_rate": 1.9347924702398484e-05,
      "loss": 0.08360576629638672,
      "step": 3311
    },
    {
      "epoch": 0.4477717878086289,
      "grad_norm": 0.15922707319259644,
      "learning_rate": 1.9341373662735075e-05,
      "loss": 0.051326751708984375,
      "step": 3312
    },
    {
      "epoch": 0.4479069846044649,
      "grad_norm": 0.12030304223299026,
      "learning_rate": 1.9334821719174524e-05,
      "loss": 0.06579971313476562,
      "step": 3313
    },
    {
      "epoch": 0.44804218140030083,
      "grad_norm": 0.08145484328269958,
      "learning_rate": 1.9328268873080974e-05,
      "loss": 0.05148029327392578,
      "step": 3314
    },
    {
      "epoch": 0.44817737819613673,
      "grad_norm": 0.10939681529998779,
      "learning_rate": 1.9321715125818765e-05,
      "loss": 0.08198165893554688,
      "step": 3315
    },
    {
      "epoch": 0.4483125749919727,
      "grad_norm": 0.10112543404102325,
      "learning_rate": 1.931516047875242e-05,
      "loss": 0.04317474365234375,
      "step": 3316
    },
    {
      "epoch": 0.44844777178780865,
      "grad_norm": 0.163516566157341,
      "learning_rate": 1.930860493324665e-05,
      "loss": 0.10313796997070312,
      "step": 3317
    },
    {
      "epoch": 0.44858296858364455,
      "grad_norm": 0.24030812084674835,
      "learning_rate": 1.9302048490666356e-05,
      "loss": 0.14154672622680664,
      "step": 3318
    },
    {
      "epoch": 0.4487181653794805,
      "grad_norm": 0.07341719418764114,
      "learning_rate": 1.9295491152376623e-05,
      "loss": 0.06919002532958984,
      "step": 3319
    },
    {
      "epoch": 0.44885336217531646,
      "grad_norm": 0.1363794058561325,
      "learning_rate": 1.928893291974273e-05,
      "loss": 0.07803583145141602,
      "step": 3320
    },
    {
      "epoch": 0.44898855897115236,
      "grad_norm": 0.1386910378932953,
      "learning_rate": 1.9282373794130132e-05,
      "loss": 0.05220353603363037,
      "step": 3321
    },
    {
      "epoch": 0.4491237557669883,
      "grad_norm": 0.15536145865917206,
      "learning_rate": 1.9275813776904472e-05,
      "loss": 0.09165549278259277,
      "step": 3322
    },
    {
      "epoch": 0.4492589525628243,
      "grad_norm": 0.14397242665290833,
      "learning_rate": 1.9269252869431582e-05,
      "loss": 0.05765819549560547,
      "step": 3323
    },
    {
      "epoch": 0.4493941493586602,
      "grad_norm": 0.17620156705379486,
      "learning_rate": 1.9262691073077476e-05,
      "loss": 0.055515289306640625,
      "step": 3324
    },
    {
      "epoch": 0.44952934615449613,
      "grad_norm": 0.0667574480175972,
      "learning_rate": 1.9256128389208363e-05,
      "loss": 0.07583379745483398,
      "step": 3325
    },
    {
      "epoch": 0.4496645429503321,
      "grad_norm": 0.12930706143379211,
      "learning_rate": 1.924956481919062e-05,
      "loss": 0.052909016609191895,
      "step": 3326
    },
    {
      "epoch": 0.449799739746168,
      "grad_norm": 0.06159701943397522,
      "learning_rate": 1.9243000364390825e-05,
      "loss": 0.0617833137512207,
      "step": 3327
    },
    {
      "epoch": 0.44993493654200395,
      "grad_norm": 0.09984344989061356,
      "learning_rate": 1.9236435026175717e-05,
      "loss": 0.048177242279052734,
      "step": 3328
    },
    {
      "epoch": 0.4500701333378399,
      "grad_norm": 0.11977959424257278,
      "learning_rate": 1.9229868805912248e-05,
      "loss": 0.11803674697875977,
      "step": 3329
    },
    {
      "epoch": 0.4502053301336758,
      "grad_norm": 0.06340879201889038,
      "learning_rate": 1.9223301704967543e-05,
      "loss": 0.07313919067382812,
      "step": 3330
    },
    {
      "epoch": 0.45034052692951176,
      "grad_norm": 0.12955889105796814,
      "learning_rate": 1.92167337247089e-05,
      "loss": 0.08098793029785156,
      "step": 3331
    },
    {
      "epoch": 0.4504757237253477,
      "grad_norm": 0.15105193853378296,
      "learning_rate": 1.9210164866503808e-05,
      "loss": 0.07820415496826172,
      "step": 3332
    },
    {
      "epoch": 0.4506109205211836,
      "grad_norm": 0.061245497316122055,
      "learning_rate": 1.9203595131719935e-05,
      "loss": 0.06598091125488281,
      "step": 3333
    },
    {
      "epoch": 0.4507461173170196,
      "grad_norm": 0.08076949417591095,
      "learning_rate": 1.9197024521725148e-05,
      "loss": 0.10294008255004883,
      "step": 3334
    },
    {
      "epoch": 0.45088131411285554,
      "grad_norm": 0.14071492850780487,
      "learning_rate": 1.9190453037887465e-05,
      "loss": 0.07048368453979492,
      "step": 3335
    },
    {
      "epoch": 0.45101651090869144,
      "grad_norm": 0.15181340277194977,
      "learning_rate": 1.918388068157512e-05,
      "loss": 0.0891256332397461,
      "step": 3336
    },
    {
      "epoch": 0.4511517077045274,
      "grad_norm": 0.07215507328510284,
      "learning_rate": 1.9177307454156507e-05,
      "loss": 0.08237147331237793,
      "step": 3337
    },
    {
      "epoch": 0.45128690450036335,
      "grad_norm": 0.10343208909034729,
      "learning_rate": 1.9170733357000202e-05,
      "loss": 0.08262085914611816,
      "step": 3338
    },
    {
      "epoch": 0.4514221012961993,
      "grad_norm": 0.13351555168628693,
      "learning_rate": 1.916415839147497e-05,
      "loss": 0.048540592193603516,
      "step": 3339
    },
    {
      "epoch": 0.4515572980920352,
      "grad_norm": 0.07054270058870316,
      "learning_rate": 1.9157582558949756e-05,
      "loss": 0.0576019287109375,
      "step": 3340
    },
    {
      "epoch": 0.45169249488787117,
      "grad_norm": 0.14163419604301453,
      "learning_rate": 1.9151005860793682e-05,
      "loss": 0.07925033569335938,
      "step": 3341
    },
    {
      "epoch": 0.4518276916837071,
      "grad_norm": 0.07599087804555893,
      "learning_rate": 1.9144428298376056e-05,
      "loss": 0.06356644630432129,
      "step": 3342
    },
    {
      "epoch": 0.451962888479543,
      "grad_norm": 0.06390077620744705,
      "learning_rate": 1.9137849873066355e-05,
      "loss": 0.05053520202636719,
      "step": 3343
    },
    {
      "epoch": 0.452098085275379,
      "grad_norm": 0.11575238406658173,
      "learning_rate": 1.9131270586234243e-05,
      "loss": 0.09288692474365234,
      "step": 3344
    },
    {
      "epoch": 0.45223328207121494,
      "grad_norm": 0.28957614302635193,
      "learning_rate": 1.9124690439249564e-05,
      "loss": 0.08353424072265625,
      "step": 3345
    },
    {
      "epoch": 0.45236847886705084,
      "grad_norm": 0.07142768800258636,
      "learning_rate": 1.9118109433482342e-05,
      "loss": 0.06046032905578613,
      "step": 3346
    },
    {
      "epoch": 0.4525036756628868,
      "grad_norm": 0.1332925707101822,
      "learning_rate": 1.911152757030278e-05,
      "loss": 0.08690881729125977,
      "step": 3347
    },
    {
      "epoch": 0.45263887245872275,
      "grad_norm": 0.13871270418167114,
      "learning_rate": 1.9104944851081247e-05,
      "loss": 0.07982838153839111,
      "step": 3348
    },
    {
      "epoch": 0.45277406925455865,
      "grad_norm": 0.12735576927661896,
      "learning_rate": 1.9098361277188303e-05,
      "loss": 0.05886983871459961,
      "step": 3349
    },
    {
      "epoch": 0.4529092660503946,
      "grad_norm": 0.10669447481632233,
      "learning_rate": 1.909177684999469e-05,
      "loss": 0.0580788254737854,
      "step": 3350
    },
    {
      "epoch": 0.45304446284623057,
      "grad_norm": 0.05876397714018822,
      "learning_rate": 1.9085191570871316e-05,
      "loss": 0.030082225799560547,
      "step": 3351
    },
    {
      "epoch": 0.45317965964206647,
      "grad_norm": 0.06846889108419418,
      "learning_rate": 1.9078605441189275e-05,
      "loss": 0.056444644927978516,
      "step": 3352
    },
    {
      "epoch": 0.4533148564379024,
      "grad_norm": 0.10122275352478027,
      "learning_rate": 1.9072018462319828e-05,
      "loss": 0.07421910762786865,
      "step": 3353
    },
    {
      "epoch": 0.4534500532337384,
      "grad_norm": 0.044947437942028046,
      "learning_rate": 1.9065430635634422e-05,
      "loss": 0.04962897300720215,
      "step": 3354
    },
    {
      "epoch": 0.4535852500295743,
      "grad_norm": 0.05820563808083534,
      "learning_rate": 1.9058841962504677e-05,
      "loss": 0.048013925552368164,
      "step": 3355
    },
    {
      "epoch": 0.45372044682541024,
      "grad_norm": 0.055406201630830765,
      "learning_rate": 1.9052252444302394e-05,
      "loss": 0.06353473663330078,
      "step": 3356
    },
    {
      "epoch": 0.4538556436212462,
      "grad_norm": 0.08920296281576157,
      "learning_rate": 1.904566208239954e-05,
      "loss": 0.07845258712768555,
      "step": 3357
    },
    {
      "epoch": 0.4539908404170821,
      "grad_norm": 0.056996092200279236,
      "learning_rate": 1.903907087816827e-05,
      "loss": 0.056680917739868164,
      "step": 3358
    },
    {
      "epoch": 0.45412603721291805,
      "grad_norm": 0.08179796487092972,
      "learning_rate": 1.9032478832980902e-05,
      "loss": 0.060242652893066406,
      "step": 3359
    },
    {
      "epoch": 0.454261234008754,
      "grad_norm": 0.05451581999659538,
      "learning_rate": 1.9025885948209938e-05,
      "loss": 0.04308128356933594,
      "step": 3360
    },
    {
      "epoch": 0.4543964308045899,
      "grad_norm": 0.054087620228528976,
      "learning_rate": 1.901929222522805e-05,
      "loss": 0.058830857276916504,
      "step": 3361
    },
    {
      "epoch": 0.45453162760042587,
      "grad_norm": 0.14267101883888245,
      "learning_rate": 1.901269766540809e-05,
      "loss": 0.07298135757446289,
      "step": 3362
    },
    {
      "epoch": 0.4546668243962618,
      "grad_norm": 0.04137654975056648,
      "learning_rate": 1.9006102270123076e-05,
      "loss": 0.06249809265136719,
      "step": 3363
    },
    {
      "epoch": 0.4548020211920977,
      "grad_norm": 0.05987751483917236,
      "learning_rate": 1.8999506040746208e-05,
      "loss": 0.045018911361694336,
      "step": 3364
    },
    {
      "epoch": 0.4549372179879337,
      "grad_norm": 0.05372031778097153,
      "learning_rate": 1.899290897865085e-05,
      "loss": 0.05090665817260742,
      "step": 3365
    },
    {
      "epoch": 0.45507241478376964,
      "grad_norm": 0.0519489049911499,
      "learning_rate": 1.898631108521055e-05,
      "loss": 0.03508281707763672,
      "step": 3366
    },
    {
      "epoch": 0.45520761157960554,
      "grad_norm": 0.07589711993932724,
      "learning_rate": 1.8979712361799027e-05,
      "loss": 0.04430168867111206,
      "step": 3367
    },
    {
      "epoch": 0.4553428083754415,
      "grad_norm": 0.09291955828666687,
      "learning_rate": 1.8973112809790168e-05,
      "loss": 0.08121395111083984,
      "step": 3368
    },
    {
      "epoch": 0.45547800517127746,
      "grad_norm": 0.10260295867919922,
      "learning_rate": 1.8966512430558036e-05,
      "loss": 0.0785675048828125,
      "step": 3369
    },
    {
      "epoch": 0.45561320196711336,
      "grad_norm": 0.09186156094074249,
      "learning_rate": 1.8959911225476858e-05,
      "loss": 0.054290771484375,
      "step": 3370
    },
    {
      "epoch": 0.4557483987629493,
      "grad_norm": 0.11577325314283371,
      "learning_rate": 1.895330919592105e-05,
      "loss": 0.07844400405883789,
      "step": 3371
    },
    {
      "epoch": 0.45588359555878527,
      "grad_norm": 0.18349677324295044,
      "learning_rate": 1.8946706343265187e-05,
      "loss": 0.07850754261016846,
      "step": 3372
    },
    {
      "epoch": 0.45601879235462117,
      "grad_norm": 0.11858109384775162,
      "learning_rate": 1.8940102668884016e-05,
      "loss": 0.0690622329711914,
      "step": 3373
    },
    {
      "epoch": 0.45615398915045713,
      "grad_norm": 0.1004151925444603,
      "learning_rate": 1.893349817415246e-05,
      "loss": 0.05154871940612793,
      "step": 3374
    },
    {
      "epoch": 0.4562891859462931,
      "grad_norm": 0.07658327370882034,
      "learning_rate": 1.8926892860445607e-05,
      "loss": 0.0634160041809082,
      "step": 3375
    },
    {
      "epoch": 0.456424382742129,
      "grad_norm": 0.12076064944267273,
      "learning_rate": 1.8920286729138718e-05,
      "loss": 0.06859207153320312,
      "step": 3376
    },
    {
      "epoch": 0.45655957953796494,
      "grad_norm": 0.059752751141786575,
      "learning_rate": 1.8913679781607225e-05,
      "loss": 0.04063749313354492,
      "step": 3377
    },
    {
      "epoch": 0.4566947763338009,
      "grad_norm": 0.14601118862628937,
      "learning_rate": 1.8907072019226734e-05,
      "loss": 0.05247354507446289,
      "step": 3378
    },
    {
      "epoch": 0.4568299731296368,
      "grad_norm": 0.16539974510669708,
      "learning_rate": 1.8900463443373015e-05,
      "loss": 0.07305121421813965,
      "step": 3379
    },
    {
      "epoch": 0.45696516992547276,
      "grad_norm": 0.05158930644392967,
      "learning_rate": 1.889385405542201e-05,
      "loss": 0.058272361755371094,
      "step": 3380
    },
    {
      "epoch": 0.4571003667213087,
      "grad_norm": 0.05201015993952751,
      "learning_rate": 1.8887243856749816e-05,
      "loss": 0.055187225341796875,
      "step": 3381
    },
    {
      "epoch": 0.4572355635171446,
      "grad_norm": 0.06832972168922424,
      "learning_rate": 1.8880632848732723e-05,
      "loss": 0.06260538101196289,
      "step": 3382
    },
    {
      "epoch": 0.4573707603129806,
      "grad_norm": 0.10209842771291733,
      "learning_rate": 1.8874021032747185e-05,
      "loss": 0.07155990600585938,
      "step": 3383
    },
    {
      "epoch": 0.45750595710881653,
      "grad_norm": 0.20631210505962372,
      "learning_rate": 1.8867408410169803e-05,
      "loss": 0.13002777099609375,
      "step": 3384
    },
    {
      "epoch": 0.4576411539046525,
      "grad_norm": 0.17883168160915375,
      "learning_rate": 1.886079498237737e-05,
      "loss": 0.0935506820678711,
      "step": 3385
    },
    {
      "epoch": 0.4577763507004884,
      "grad_norm": 0.10258541256189346,
      "learning_rate": 1.885418075074683e-05,
      "loss": 0.045658111572265625,
      "step": 3386
    },
    {
      "epoch": 0.45791154749632434,
      "grad_norm": 0.14190711081027985,
      "learning_rate": 1.884756571665531e-05,
      "loss": 0.09167361259460449,
      "step": 3387
    },
    {
      "epoch": 0.4580467442921603,
      "grad_norm": 0.05444727838039398,
      "learning_rate": 1.8840949881480085e-05,
      "loss": 0.05582749843597412,
      "step": 3388
    },
    {
      "epoch": 0.4581819410879962,
      "grad_norm": 0.05539333075284958,
      "learning_rate": 1.8834333246598613e-05,
      "loss": 0.037569522857666016,
      "step": 3389
    },
    {
      "epoch": 0.45831713788383216,
      "grad_norm": 0.07443052530288696,
      "learning_rate": 1.8827715813388514e-05,
      "loss": 0.04635262489318848,
      "step": 3390
    },
    {
      "epoch": 0.4584523346796681,
      "grad_norm": 0.09747791290283203,
      "learning_rate": 1.8821097583227572e-05,
      "loss": 0.051010966300964355,
      "step": 3391
    },
    {
      "epoch": 0.458587531475504,
      "grad_norm": 0.09465256333351135,
      "learning_rate": 1.8814478557493732e-05,
      "loss": 0.06458258628845215,
      "step": 3392
    },
    {
      "epoch": 0.45872272827134,
      "grad_norm": 0.06467480957508087,
      "learning_rate": 1.8807858737565118e-05,
      "loss": 0.04518604278564453,
      "step": 3393
    },
    {
      "epoch": 0.45885792506717593,
      "grad_norm": 0.07349219918251038,
      "learning_rate": 1.880123812482001e-05,
      "loss": 0.06886053085327148,
      "step": 3394
    },
    {
      "epoch": 0.45899312186301183,
      "grad_norm": 0.11295178532600403,
      "learning_rate": 1.8794616720636853e-05,
      "loss": 0.07073509693145752,
      "step": 3395
    },
    {
      "epoch": 0.4591283186588478,
      "grad_norm": 0.08168105781078339,
      "learning_rate": 1.8787994526394257e-05,
      "loss": 0.08412361145019531,
      "step": 3396
    },
    {
      "epoch": 0.45926351545468375,
      "grad_norm": 0.10572415590286255,
      "learning_rate": 1.8781371543471e-05,
      "loss": 0.08240842819213867,
      "step": 3397
    },
    {
      "epoch": 0.45939871225051965,
      "grad_norm": 0.07295472174882889,
      "learning_rate": 1.8774747773246024e-05,
      "loss": 0.05265235900878906,
      "step": 3398
    },
    {
      "epoch": 0.4595339090463556,
      "grad_norm": 0.06520816683769226,
      "learning_rate": 1.8768123217098438e-05,
      "loss": 0.06548810005187988,
      "step": 3399
    },
    {
      "epoch": 0.45966910584219156,
      "grad_norm": 0.05021342262625694,
      "learning_rate": 1.8761497876407496e-05,
      "loss": 0.039754629135131836,
      "step": 3400
    },
    {
      "epoch": 0.45980430263802746,
      "grad_norm": 0.0390787348151207,
      "learning_rate": 1.8754871752552646e-05,
      "loss": 0.04110407829284668,
      "step": 3401
    },
    {
      "epoch": 0.4599394994338634,
      "grad_norm": 0.05425737425684929,
      "learning_rate": 1.8748244846913463e-05,
      "loss": 0.04517549276351929,
      "step": 3402
    },
    {
      "epoch": 0.4600746962296994,
      "grad_norm": 0.07912790775299072,
      "learning_rate": 1.874161716086972e-05,
      "loss": 0.056559622287750244,
      "step": 3403
    },
    {
      "epoch": 0.4602098930255353,
      "grad_norm": 0.10625887662172318,
      "learning_rate": 1.8734988695801333e-05,
      "loss": 0.07645535469055176,
      "step": 3404
    },
    {
      "epoch": 0.46034508982137123,
      "grad_norm": 0.05065944790840149,
      "learning_rate": 1.8728359453088382e-05,
      "loss": 0.042622387409210205,
      "step": 3405
    },
    {
      "epoch": 0.4604802866172072,
      "grad_norm": 0.16429084539413452,
      "learning_rate": 1.8721729434111108e-05,
      "loss": 0.10288619995117188,
      "step": 3406
    },
    {
      "epoch": 0.4606154834130431,
      "grad_norm": 0.06927911937236786,
      "learning_rate": 1.871509864024992e-05,
      "loss": 0.05532026290893555,
      "step": 3407
    },
    {
      "epoch": 0.46075068020887905,
      "grad_norm": 0.03308272361755371,
      "learning_rate": 1.8708467072885385e-05,
      "loss": 0.04797959327697754,
      "step": 3408
    },
    {
      "epoch": 0.460885877004715,
      "grad_norm": 0.07664691656827927,
      "learning_rate": 1.8701834733398227e-05,
      "loss": 0.04942655563354492,
      "step": 3409
    },
    {
      "epoch": 0.4610210738005509,
      "grad_norm": 0.0725528746843338,
      "learning_rate": 1.8695201623169335e-05,
      "loss": 0.05355715751647949,
      "step": 3410
    },
    {
      "epoch": 0.46115627059638686,
      "grad_norm": 0.08797306567430496,
      "learning_rate": 1.868856774357977e-05,
      "loss": 0.06735944747924805,
      "step": 3411
    },
    {
      "epoch": 0.4612914673922228,
      "grad_norm": 0.0751928761601448,
      "learning_rate": 1.868193309601072e-05,
      "loss": 0.042236328125,
      "step": 3412
    },
    {
      "epoch": 0.4614266641880587,
      "grad_norm": 0.1532754749059677,
      "learning_rate": 1.867529768184357e-05,
      "loss": 0.08893823623657227,
      "step": 3413
    },
    {
      "epoch": 0.4615618609838947,
      "grad_norm": 0.06271111220121384,
      "learning_rate": 1.8668661502459842e-05,
      "loss": 0.04842948913574219,
      "step": 3414
    },
    {
      "epoch": 0.46169705777973064,
      "grad_norm": 0.06832994520664215,
      "learning_rate": 1.866202455924123e-05,
      "loss": 0.02930617332458496,
      "step": 3415
    },
    {
      "epoch": 0.46183225457556654,
      "grad_norm": 0.0676761344075203,
      "learning_rate": 1.865538685356957e-05,
      "loss": 0.06079220771789551,
      "step": 3416
    },
    {
      "epoch": 0.4619674513714025,
      "grad_norm": 0.16096584498882294,
      "learning_rate": 1.864874838682688e-05,
      "loss": 0.08106279373168945,
      "step": 3417
    },
    {
      "epoch": 0.46210264816723845,
      "grad_norm": 0.06340809911489487,
      "learning_rate": 1.864210916039531e-05,
      "loss": 0.044806480407714844,
      "step": 3418
    },
    {
      "epoch": 0.46223784496307435,
      "grad_norm": 0.1108318641781807,
      "learning_rate": 1.86354691756572e-05,
      "loss": 0.08450126647949219,
      "step": 3419
    },
    {
      "epoch": 0.4623730417589103,
      "grad_norm": 0.0868648886680603,
      "learning_rate": 1.8628828433995013e-05,
      "loss": 0.06555521488189697,
      "step": 3420
    },
    {
      "epoch": 0.46250823855474626,
      "grad_norm": 0.13067664206027985,
      "learning_rate": 1.86221869367914e-05,
      "loss": 0.05691194534301758,
      "step": 3421
    },
    {
      "epoch": 0.46264343535058217,
      "grad_norm": 0.03764400631189346,
      "learning_rate": 1.8615544685429153e-05,
      "loss": 0.043553829193115234,
      "step": 3422
    },
    {
      "epoch": 0.4627786321464181,
      "grad_norm": 0.10746897757053375,
      "learning_rate": 1.860890168129122e-05,
      "loss": 0.04911398887634277,
      "step": 3423
    },
    {
      "epoch": 0.4629138289422541,
      "grad_norm": 0.0851382240653038,
      "learning_rate": 1.8602257925760708e-05,
      "loss": 0.07423734664916992,
      "step": 3424
    },
    {
      "epoch": 0.46304902573809,
      "grad_norm": 0.12360157072544098,
      "learning_rate": 1.859561342022089e-05,
      "loss": 0.08611297607421875,
      "step": 3425
    },
    {
      "epoch": 0.46318422253392594,
      "grad_norm": 0.09106816351413727,
      "learning_rate": 1.8588968166055185e-05,
      "loss": 0.07781457901000977,
      "step": 3426
    },
    {
      "epoch": 0.4633194193297619,
      "grad_norm": 0.05281224474310875,
      "learning_rate": 1.8582322164647166e-05,
      "loss": 0.06018543243408203,
      "step": 3427
    },
    {
      "epoch": 0.4634546161255978,
      "grad_norm": 0.14227604866027832,
      "learning_rate": 1.8575675417380568e-05,
      "loss": 0.07758688926696777,
      "step": 3428
    },
    {
      "epoch": 0.46358981292143375,
      "grad_norm": 0.17344899475574493,
      "learning_rate": 1.856902792563928e-05,
      "loss": 0.0871438980102539,
      "step": 3429
    },
    {
      "epoch": 0.4637250097172697,
      "grad_norm": 0.07698340713977814,
      "learning_rate": 1.856237969080735e-05,
      "loss": 0.06049633026123047,
      "step": 3430
    },
    {
      "epoch": 0.46386020651310567,
      "grad_norm": 0.05003563314676285,
      "learning_rate": 1.8555730714268967e-05,
      "loss": 0.04724407196044922,
      "step": 3431
    },
    {
      "epoch": 0.46399540330894157,
      "grad_norm": 0.15106461942195892,
      "learning_rate": 1.8549080997408492e-05,
      "loss": 0.09495067596435547,
      "step": 3432
    },
    {
      "epoch": 0.4641306001047775,
      "grad_norm": 0.041310567408800125,
      "learning_rate": 1.8542430541610426e-05,
      "loss": 0.03312993049621582,
      "step": 3433
    },
    {
      "epoch": 0.4642657969006135,
      "grad_norm": 0.1053837388753891,
      "learning_rate": 1.8535779348259428e-05,
      "loss": 0.06680011749267578,
      "step": 3434
    },
    {
      "epoch": 0.4644009936964494,
      "grad_norm": 0.047986049205064774,
      "learning_rate": 1.852912741874032e-05,
      "loss": 0.04587411880493164,
      "step": 3435
    },
    {
      "epoch": 0.46453619049228534,
      "grad_norm": 0.06248097121715546,
      "learning_rate": 1.8522474754438056e-05,
      "loss": 0.06231224536895752,
      "step": 3436
    },
    {
      "epoch": 0.4646713872881213,
      "grad_norm": 0.10099022090435028,
      "learning_rate": 1.851582135673777e-05,
      "loss": 0.0876932144165039,
      "step": 3437
    },
    {
      "epoch": 0.4648065840839572,
      "grad_norm": 0.09039631485939026,
      "learning_rate": 1.850916722702473e-05,
      "loss": 0.10105228424072266,
      "step": 3438
    },
    {
      "epoch": 0.46494178087979315,
      "grad_norm": 0.040435608476400375,
      "learning_rate": 1.8502512366684355e-05,
      "loss": 0.038915395736694336,
      "step": 3439
    },
    {
      "epoch": 0.4650769776756291,
      "grad_norm": 0.04880201071500778,
      "learning_rate": 1.8495856777102232e-05,
      "loss": 0.052845001220703125,
      "step": 3440
    },
    {
      "epoch": 0.465212174471465,
      "grad_norm": 0.07616909593343735,
      "learning_rate": 1.848920045966408e-05,
      "loss": 0.07064557075500488,
      "step": 3441
    },
    {
      "epoch": 0.46534737126730097,
      "grad_norm": 0.06871594488620758,
      "learning_rate": 1.8482543415755797e-05,
      "loss": 0.07067060470581055,
      "step": 3442
    },
    {
      "epoch": 0.4654825680631369,
      "grad_norm": 0.05334244668483734,
      "learning_rate": 1.8475885646763394e-05,
      "loss": 0.04635286331176758,
      "step": 3443
    },
    {
      "epoch": 0.4656177648589728,
      "grad_norm": 0.054341871291399,
      "learning_rate": 1.8469227154073064e-05,
      "loss": 0.059173583984375,
      "step": 3444
    },
    {
      "epoch": 0.4657529616548088,
      "grad_norm": 0.04777482897043228,
      "learning_rate": 1.8462567939071142e-05,
      "loss": 0.03973746299743652,
      "step": 3445
    },
    {
      "epoch": 0.46588815845064474,
      "grad_norm": 0.0668109804391861,
      "learning_rate": 1.845590800314411e-05,
      "loss": 0.054836273193359375,
      "step": 3446
    },
    {
      "epoch": 0.46602335524648064,
      "grad_norm": 0.14163623750209808,
      "learning_rate": 1.8449247347678607e-05,
      "loss": 0.06424188613891602,
      "step": 3447
    },
    {
      "epoch": 0.4661585520423166,
      "grad_norm": 0.0577327236533165,
      "learning_rate": 1.8442585974061405e-05,
      "loss": 0.06545388698577881,
      "step": 3448
    },
    {
      "epoch": 0.46629374883815256,
      "grad_norm": 0.03324039280414581,
      "learning_rate": 1.8435923883679452e-05,
      "loss": 0.04532194137573242,
      "step": 3449
    },
    {
      "epoch": 0.46642894563398846,
      "grad_norm": 0.09412823617458344,
      "learning_rate": 1.8429261077919818e-05,
      "loss": 0.08692502975463867,
      "step": 3450
    },
    {
      "epoch": 0.4665641424298244,
      "grad_norm": 0.05564672872424126,
      "learning_rate": 1.8422597558169742e-05,
      "loss": 0.055829405784606934,
      "step": 3451
    },
    {
      "epoch": 0.46669933922566037,
      "grad_norm": 0.06166764348745346,
      "learning_rate": 1.84159333258166e-05,
      "loss": 0.056166648864746094,
      "step": 3452
    },
    {
      "epoch": 0.46683453602149627,
      "grad_norm": 0.0592268742620945,
      "learning_rate": 1.8409268382247933e-05,
      "loss": 0.03865456581115723,
      "step": 3453
    },
    {
      "epoch": 0.46696973281733223,
      "grad_norm": 0.0823439359664917,
      "learning_rate": 1.8402602728851405e-05,
      "loss": 0.06898117065429688,
      "step": 3454
    },
    {
      "epoch": 0.4671049296131682,
      "grad_norm": 0.0611744150519371,
      "learning_rate": 1.839593636701484e-05,
      "loss": 0.038504183292388916,
      "step": 3455
    },
    {
      "epoch": 0.4672401264090041,
      "grad_norm": 0.054461125284433365,
      "learning_rate": 1.8389269298126214e-05,
      "loss": 0.047418832778930664,
      "step": 3456
    },
    {
      "epoch": 0.46737532320484004,
      "grad_norm": 0.058121565729379654,
      "learning_rate": 1.838260152357365e-05,
      "loss": 0.042220115661621094,
      "step": 3457
    },
    {
      "epoch": 0.467510520000676,
      "grad_norm": 0.08581023663282394,
      "learning_rate": 1.837593304474541e-05,
      "loss": 0.07930755615234375,
      "step": 3458
    },
    {
      "epoch": 0.4676457167965119,
      "grad_norm": 0.09980616718530655,
      "learning_rate": 1.836926386302991e-05,
      "loss": 0.08956289291381836,
      "step": 3459
    },
    {
      "epoch": 0.46778091359234786,
      "grad_norm": 0.07131228595972061,
      "learning_rate": 1.8362593979815696e-05,
      "loss": 0.06785130500793457,
      "step": 3460
    },
    {
      "epoch": 0.4679161103881838,
      "grad_norm": 0.10368010401725769,
      "learning_rate": 1.8355923396491496e-05,
      "loss": 0.10187339782714844,
      "step": 3461
    },
    {
      "epoch": 0.4680513071840197,
      "grad_norm": 0.08152949064970016,
      "learning_rate": 1.8349252114446138e-05,
      "loss": 0.056446075439453125,
      "step": 3462
    },
    {
      "epoch": 0.4681865039798557,
      "grad_norm": 0.04941851273179054,
      "learning_rate": 1.834258013506864e-05,
      "loss": 0.0626382827758789,
      "step": 3463
    },
    {
      "epoch": 0.46832170077569163,
      "grad_norm": 0.05576416850090027,
      "learning_rate": 1.833590745974813e-05,
      "loss": 0.058524370193481445,
      "step": 3464
    },
    {
      "epoch": 0.46845689757152753,
      "grad_norm": 0.10252301394939423,
      "learning_rate": 1.8329234089873892e-05,
      "loss": 0.04757833480834961,
      "step": 3465
    },
    {
      "epoch": 0.4685920943673635,
      "grad_norm": 0.028037160634994507,
      "learning_rate": 1.8322560026835366e-05,
      "loss": 0.023073315620422363,
      "step": 3466
    },
    {
      "epoch": 0.46872729116319944,
      "grad_norm": 0.11344929039478302,
      "learning_rate": 1.8315885272022125e-05,
      "loss": 0.08417391777038574,
      "step": 3467
    },
    {
      "epoch": 0.46886248795903535,
      "grad_norm": 0.05986790731549263,
      "learning_rate": 1.830920982682389e-05,
      "loss": 0.05260425806045532,
      "step": 3468
    },
    {
      "epoch": 0.4689976847548713,
      "grad_norm": 0.08218964189291,
      "learning_rate": 1.830253369263052e-05,
      "loss": 0.06722700595855713,
      "step": 3469
    },
    {
      "epoch": 0.46913288155070726,
      "grad_norm": 0.10893671959638596,
      "learning_rate": 1.8295856870832024e-05,
      "loss": 0.06206321716308594,
      "step": 3470
    },
    {
      "epoch": 0.46926807834654316,
      "grad_norm": 0.05701784789562225,
      "learning_rate": 1.828917936281855e-05,
      "loss": 0.050049543380737305,
      "step": 3471
    },
    {
      "epoch": 0.4694032751423791,
      "grad_norm": 0.11329787969589233,
      "learning_rate": 1.8282501169980396e-05,
      "loss": 0.0606992244720459,
      "step": 3472
    },
    {
      "epoch": 0.4695384719382151,
      "grad_norm": 0.09615732729434967,
      "learning_rate": 1.8275822293707992e-05,
      "loss": 0.07668781280517578,
      "step": 3473
    },
    {
      "epoch": 0.469673668734051,
      "grad_norm": 0.11335486173629761,
      "learning_rate": 1.8269142735391917e-05,
      "loss": 0.06651115417480469,
      "step": 3474
    },
    {
      "epoch": 0.46980886552988693,
      "grad_norm": 0.09174930304288864,
      "learning_rate": 1.8262462496422893e-05,
      "loss": 0.04760622978210449,
      "step": 3475
    },
    {
      "epoch": 0.4699440623257229,
      "grad_norm": 0.06920750439167023,
      "learning_rate": 1.8255781578191778e-05,
      "loss": 0.048304349184036255,
      "step": 3476
    },
    {
      "epoch": 0.47007925912155885,
      "grad_norm": 0.06450843065977097,
      "learning_rate": 1.824909998208958e-05,
      "loss": 0.05025315284729004,
      "step": 3477
    },
    {
      "epoch": 0.47021445591739475,
      "grad_norm": 0.20177200436592102,
      "learning_rate": 1.8242417709507434e-05,
      "loss": 0.07547140121459961,
      "step": 3478
    },
    {
      "epoch": 0.4703496527132307,
      "grad_norm": 0.08519097417593002,
      "learning_rate": 1.8235734761836637e-05,
      "loss": 0.06819629669189453,
      "step": 3479
    },
    {
      "epoch": 0.47048484950906666,
      "grad_norm": 0.09071063250303268,
      "learning_rate": 1.82290511404686e-05,
      "loss": 0.04694056510925293,
      "step": 3480
    },
    {
      "epoch": 0.47062004630490256,
      "grad_norm": 0.17191484570503235,
      "learning_rate": 1.8222366846794904e-05,
      "loss": 0.0582888126373291,
      "step": 3481
    },
    {
      "epoch": 0.4707552431007385,
      "grad_norm": 0.08369534462690353,
      "learning_rate": 1.8215681882207238e-05,
      "loss": 0.0454862117767334,
      "step": 3482
    },
    {
      "epoch": 0.4708904398965745,
      "grad_norm": 0.06966706365346909,
      "learning_rate": 1.8208996248097462e-05,
      "loss": 0.0385441780090332,
      "step": 3483
    },
    {
      "epoch": 0.4710256366924104,
      "grad_norm": 0.07693798840045929,
      "learning_rate": 1.8202309945857557e-05,
      "loss": 0.07320630550384521,
      "step": 3484
    },
    {
      "epoch": 0.47116083348824633,
      "grad_norm": 0.10992797464132309,
      "learning_rate": 1.8195622976879638e-05,
      "loss": 0.06183671951293945,
      "step": 3485
    },
    {
      "epoch": 0.4712960302840823,
      "grad_norm": 0.12138800323009491,
      "learning_rate": 1.8188935342555977e-05,
      "loss": 0.06736171245574951,
      "step": 3486
    },
    {
      "epoch": 0.4714312270799182,
      "grad_norm": 0.07128792256116867,
      "learning_rate": 1.818224704427897e-05,
      "loss": 0.054497480392456055,
      "step": 3487
    },
    {
      "epoch": 0.47156642387575415,
      "grad_norm": 0.06149104982614517,
      "learning_rate": 1.8175558083441162e-05,
      "loss": 0.06778943538665771,
      "step": 3488
    },
    {
      "epoch": 0.4717016206715901,
      "grad_norm": 0.2007308006286621,
      "learning_rate": 1.8168868461435225e-05,
      "loss": 0.05882704257965088,
      "step": 3489
    },
    {
      "epoch": 0.471836817467426,
      "grad_norm": 0.07799633592367172,
      "learning_rate": 1.8162178179653977e-05,
      "loss": 0.059824466705322266,
      "step": 3490
    },
    {
      "epoch": 0.47197201426326196,
      "grad_norm": 0.07294801622629166,
      "learning_rate": 1.815548723949037e-05,
      "loss": 0.06339430809020996,
      "step": 3491
    },
    {
      "epoch": 0.4721072110590979,
      "grad_norm": 0.08424577116966248,
      "learning_rate": 1.814879564233749e-05,
      "loss": 0.08004307746887207,
      "step": 3492
    },
    {
      "epoch": 0.4722424078549338,
      "grad_norm": 0.0833587795495987,
      "learning_rate": 1.8142103389588567e-05,
      "loss": 0.07729113101959229,
      "step": 3493
    },
    {
      "epoch": 0.4723776046507698,
      "grad_norm": 0.06454747915267944,
      "learning_rate": 1.813541048263696e-05,
      "loss": 0.05300617218017578,
      "step": 3494
    },
    {
      "epoch": 0.47251280144660573,
      "grad_norm": 0.1339472234249115,
      "learning_rate": 1.8128716922876178e-05,
      "loss": 0.08754396438598633,
      "step": 3495
    },
    {
      "epoch": 0.47264799824244164,
      "grad_norm": 0.09679068624973297,
      "learning_rate": 1.812202271169984e-05,
      "loss": 0.06735682487487793,
      "step": 3496
    },
    {
      "epoch": 0.4727831950382776,
      "grad_norm": 0.0711110383272171,
      "learning_rate": 1.8115327850501726e-05,
      "loss": 0.04534256458282471,
      "step": 3497
    },
    {
      "epoch": 0.47291839183411355,
      "grad_norm": 0.08141111582517624,
      "learning_rate": 1.8108632340675746e-05,
      "loss": 0.0625767707824707,
      "step": 3498
    },
    {
      "epoch": 0.47305358862994945,
      "grad_norm": 0.05167558416724205,
      "learning_rate": 1.810193618361593e-05,
      "loss": 0.055295467376708984,
      "step": 3499
    },
    {
      "epoch": 0.4731887854257854,
      "grad_norm": 0.12220493704080582,
      "learning_rate": 1.8095239380716464e-05,
      "loss": 0.04735064506530762,
      "step": 3500
    },
    {
      "epoch": 0.47332398222162136,
      "grad_norm": 0.06282928586006165,
      "learning_rate": 1.808854193337165e-05,
      "loss": 0.07062315940856934,
      "step": 3501
    },
    {
      "epoch": 0.47345917901745727,
      "grad_norm": 0.05195547640323639,
      "learning_rate": 1.8081843842975935e-05,
      "loss": 0.07199633121490479,
      "step": 3502
    },
    {
      "epoch": 0.4735943758132932,
      "grad_norm": 0.23887105286121368,
      "learning_rate": 1.80751451109239e-05,
      "loss": 0.08484089374542236,
      "step": 3503
    },
    {
      "epoch": 0.4737295726091292,
      "grad_norm": 0.05381393805146217,
      "learning_rate": 1.806844573861025e-05,
      "loss": 0.06839513778686523,
      "step": 3504
    },
    {
      "epoch": 0.4738647694049651,
      "grad_norm": 0.10352759063243866,
      "learning_rate": 1.806174572742984e-05,
      "loss": 0.06512832641601562,
      "step": 3505
    },
    {
      "epoch": 0.47399996620080104,
      "grad_norm": 0.0745445191860199,
      "learning_rate": 1.8055045078777634e-05,
      "loss": 0.08244776725769043,
      "step": 3506
    },
    {
      "epoch": 0.474135162996637,
      "grad_norm": 0.08982892334461212,
      "learning_rate": 1.8048343794048762e-05,
      "loss": 0.06277763843536377,
      "step": 3507
    },
    {
      "epoch": 0.4742703597924729,
      "grad_norm": 0.15886647999286652,
      "learning_rate": 1.8041641874638445e-05,
      "loss": 0.07700347900390625,
      "step": 3508
    },
    {
      "epoch": 0.47440555658830885,
      "grad_norm": 0.12980806827545166,
      "learning_rate": 1.8034939321942077e-05,
      "loss": 0.05527544021606445,
      "step": 3509
    },
    {
      "epoch": 0.4745407533841448,
      "grad_norm": 0.11058919876813889,
      "learning_rate": 1.8028236137355154e-05,
      "loss": 0.0814218521118164,
      "step": 3510
    },
    {
      "epoch": 0.4746759501799807,
      "grad_norm": 0.07259859889745712,
      "learning_rate": 1.8021532322273327e-05,
      "loss": 0.08646726608276367,
      "step": 3511
    },
    {
      "epoch": 0.47481114697581667,
      "grad_norm": 0.051616255193948746,
      "learning_rate": 1.8014827878092347e-05,
      "loss": 0.059804439544677734,
      "step": 3512
    },
    {
      "epoch": 0.4749463437716526,
      "grad_norm": 0.11216377466917038,
      "learning_rate": 1.800812280620813e-05,
      "loss": 0.09200799465179443,
      "step": 3513
    },
    {
      "epoch": 0.4750815405674885,
      "grad_norm": 0.07416464388370514,
      "learning_rate": 1.80014171080167e-05,
      "loss": 0.07598209381103516,
      "step": 3514
    },
    {
      "epoch": 0.4752167373633245,
      "grad_norm": 0.08573921769857407,
      "learning_rate": 1.7994710784914227e-05,
      "loss": 0.0655217170715332,
      "step": 3515
    },
    {
      "epoch": 0.47535193415916044,
      "grad_norm": 0.0943194255232811,
      "learning_rate": 1.7988003838297e-05,
      "loss": 0.04941558837890625,
      "step": 3516
    },
    {
      "epoch": 0.47548713095499634,
      "grad_norm": 0.13027246296405792,
      "learning_rate": 1.7981296269561438e-05,
      "loss": 0.0620427131652832,
      "step": 3517
    },
    {
      "epoch": 0.4756223277508323,
      "grad_norm": 0.065853551030159,
      "learning_rate": 1.7974588080104095e-05,
      "loss": 0.061742305755615234,
      "step": 3518
    },
    {
      "epoch": 0.47575752454666825,
      "grad_norm": 0.09798743575811386,
      "learning_rate": 1.7967879271321648e-05,
      "loss": 0.049071550369262695,
      "step": 3519
    },
    {
      "epoch": 0.4758927213425042,
      "grad_norm": 0.12141052633523941,
      "learning_rate": 1.7961169844610913e-05,
      "loss": 0.048730552196502686,
      "step": 3520
    },
    {
      "epoch": 0.4760279181383401,
      "grad_norm": 0.09053508192300797,
      "learning_rate": 1.795445980136883e-05,
      "loss": 0.05276894569396973,
      "step": 3521
    },
    {
      "epoch": 0.47616311493417607,
      "grad_norm": 0.048913341015577316,
      "learning_rate": 1.794774914299245e-05,
      "loss": 0.045694589614868164,
      "step": 3522
    },
    {
      "epoch": 0.476298311730012,
      "grad_norm": 0.08369025588035583,
      "learning_rate": 1.794103787087899e-05,
      "loss": 0.08330965042114258,
      "step": 3523
    },
    {
      "epoch": 0.4764335085258479,
      "grad_norm": 0.07855888456106186,
      "learning_rate": 1.7934325986425755e-05,
      "loss": 0.07700634002685547,
      "step": 3524
    },
    {
      "epoch": 0.4765687053216839,
      "grad_norm": 0.10310858488082886,
      "learning_rate": 1.7927613491030204e-05,
      "loss": 0.10139870643615723,
      "step": 3525
    },
    {
      "epoch": 0.47670390211751984,
      "grad_norm": 0.09178508818149567,
      "learning_rate": 1.7920900386089913e-05,
      "loss": 0.06978440284729004,
      "step": 3526
    },
    {
      "epoch": 0.47683909891335574,
      "grad_norm": 0.09025003015995026,
      "learning_rate": 1.7914186673002588e-05,
      "loss": 0.0522613525390625,
      "step": 3527
    },
    {
      "epoch": 0.4769742957091917,
      "grad_norm": 0.2030298262834549,
      "learning_rate": 1.790747235316605e-05,
      "loss": 0.05517292022705078,
      "step": 3528
    },
    {
      "epoch": 0.47710949250502765,
      "grad_norm": 0.09993169456720352,
      "learning_rate": 1.790075742797827e-05,
      "loss": 0.051212310791015625,
      "step": 3529
    },
    {
      "epoch": 0.47724468930086356,
      "grad_norm": 0.04517374932765961,
      "learning_rate": 1.789404189883732e-05,
      "loss": 0.05431985855102539,
      "step": 3530
    },
    {
      "epoch": 0.4773798860966995,
      "grad_norm": 0.10662773251533508,
      "learning_rate": 1.7887325767141415e-05,
      "loss": 0.05180820822715759,
      "step": 3531
    },
    {
      "epoch": 0.47751508289253547,
      "grad_norm": 0.1368734985589981,
      "learning_rate": 1.7880609034288894e-05,
      "loss": 0.06817197799682617,
      "step": 3532
    },
    {
      "epoch": 0.47765027968837137,
      "grad_norm": 0.06638755649328232,
      "learning_rate": 1.7873891701678208e-05,
      "loss": 0.05412650108337402,
      "step": 3533
    },
    {
      "epoch": 0.4777854764842073,
      "grad_norm": 0.08016375452280045,
      "learning_rate": 1.786717377070794e-05,
      "loss": 0.04256296157836914,
      "step": 3534
    },
    {
      "epoch": 0.4779206732800433,
      "grad_norm": 0.0387137345969677,
      "learning_rate": 1.7860455242776803e-05,
      "loss": 0.0393376350402832,
      "step": 3535
    },
    {
      "epoch": 0.4780558700758792,
      "grad_norm": 0.14212824404239655,
      "learning_rate": 1.7853736119283635e-05,
      "loss": 0.06944620609283447,
      "step": 3536
    },
    {
      "epoch": 0.47819106687171514,
      "grad_norm": 0.1294069141149521,
      "learning_rate": 1.7847016401627388e-05,
      "loss": 0.05991697311401367,
      "step": 3537
    },
    {
      "epoch": 0.4783262636675511,
      "grad_norm": 0.1399858444929123,
      "learning_rate": 1.7840296091207144e-05,
      "loss": 0.08116912841796875,
      "step": 3538
    },
    {
      "epoch": 0.478461460463387,
      "grad_norm": 0.1316068172454834,
      "learning_rate": 1.7833575189422107e-05,
      "loss": 0.0764765739440918,
      "step": 3539
    },
    {
      "epoch": 0.47859665725922296,
      "grad_norm": 0.04513904079794884,
      "learning_rate": 1.7826853697671604e-05,
      "loss": 0.0444035530090332,
      "step": 3540
    },
    {
      "epoch": 0.4787318540550589,
      "grad_norm": 0.06743033230304718,
      "learning_rate": 1.782013161735509e-05,
      "loss": 0.07162857055664062,
      "step": 3541
    },
    {
      "epoch": 0.4788670508508948,
      "grad_norm": 0.07776404917240143,
      "learning_rate": 1.781340894987213e-05,
      "loss": 0.05133962631225586,
      "step": 3542
    },
    {
      "epoch": 0.47900224764673077,
      "grad_norm": 0.08006428927183151,
      "learning_rate": 1.7806685696622426e-05,
      "loss": 0.0925908088684082,
      "step": 3543
    },
    {
      "epoch": 0.47913744444256673,
      "grad_norm": 0.06711641699075699,
      "learning_rate": 1.779996185900579e-05,
      "loss": 0.0480729341506958,
      "step": 3544
    },
    {
      "epoch": 0.47927264123840263,
      "grad_norm": 0.15275880694389343,
      "learning_rate": 1.7793237438422165e-05,
      "loss": 0.0650796890258789,
      "step": 3545
    },
    {
      "epoch": 0.4794078380342386,
      "grad_norm": 0.04742910712957382,
      "learning_rate": 1.7786512436271617e-05,
      "loss": 0.04808545112609863,
      "step": 3546
    },
    {
      "epoch": 0.47954303483007454,
      "grad_norm": 0.08539284020662308,
      "learning_rate": 1.777978685395431e-05,
      "loss": 0.07293939590454102,
      "step": 3547
    },
    {
      "epoch": 0.47967823162591045,
      "grad_norm": 0.05164426192641258,
      "learning_rate": 1.7773060692870564e-05,
      "loss": 0.024924039840698242,
      "step": 3548
    },
    {
      "epoch": 0.4798134284217464,
      "grad_norm": 0.07167036831378937,
      "learning_rate": 1.7766333954420794e-05,
      "loss": 0.05931425094604492,
      "step": 3549
    },
    {
      "epoch": 0.47994862521758236,
      "grad_norm": 0.1040789857506752,
      "learning_rate": 1.775960664000554e-05,
      "loss": 0.08452892303466797,
      "step": 3550
    },
    {
      "epoch": 0.48008382201341826,
      "grad_norm": 0.06301531940698624,
      "learning_rate": 1.7752878751025463e-05,
      "loss": 0.06488990783691406,
      "step": 3551
    },
    {
      "epoch": 0.4802190188092542,
      "grad_norm": 0.04951043426990509,
      "learning_rate": 1.7746150288881352e-05,
      "loss": 0.040463805198669434,
      "step": 3552
    },
    {
      "epoch": 0.4803542156050902,
      "grad_norm": 0.0803549736738205,
      "learning_rate": 1.7739421254974114e-05,
      "loss": 0.06554222106933594,
      "step": 3553
    },
    {
      "epoch": 0.4804894124009261,
      "grad_norm": 0.15805961191654205,
      "learning_rate": 1.7732691650704756e-05,
      "loss": 0.05463051795959473,
      "step": 3554
    },
    {
      "epoch": 0.48062460919676203,
      "grad_norm": 0.07525870949029922,
      "learning_rate": 1.7725961477474423e-05,
      "loss": 0.06330084800720215,
      "step": 3555
    },
    {
      "epoch": 0.480759805992598,
      "grad_norm": 0.06652756035327911,
      "learning_rate": 1.7719230736684375e-05,
      "loss": 0.06137204170227051,
      "step": 3556
    },
    {
      "epoch": 0.4808950027884339,
      "grad_norm": 0.05889417231082916,
      "learning_rate": 1.771249942973599e-05,
      "loss": 0.04092574119567871,
      "step": 3557
    },
    {
      "epoch": 0.48103019958426985,
      "grad_norm": 0.0636635571718216,
      "learning_rate": 1.7705767558030756e-05,
      "loss": 0.05011940002441406,
      "step": 3558
    },
    {
      "epoch": 0.4811653963801058,
      "grad_norm": 0.14814883470535278,
      "learning_rate": 1.769903512297029e-05,
      "loss": 0.0930180549621582,
      "step": 3559
    },
    {
      "epoch": 0.4813005931759417,
      "grad_norm": 0.053254399448633194,
      "learning_rate": 1.7692302125956315e-05,
      "loss": 0.04150569438934326,
      "step": 3560
    },
    {
      "epoch": 0.48143578997177766,
      "grad_norm": 0.17485874891281128,
      "learning_rate": 1.768556856839068e-05,
      "loss": 0.05882751941680908,
      "step": 3561
    },
    {
      "epoch": 0.4815709867676136,
      "grad_norm": 0.07868581265211105,
      "learning_rate": 1.767883445167535e-05,
      "loss": 0.04897665977478027,
      "step": 3562
    },
    {
      "epoch": 0.4817061835634495,
      "grad_norm": 0.10659178346395493,
      "learning_rate": 1.7672099777212398e-05,
      "loss": 0.07756972312927246,
      "step": 3563
    },
    {
      "epoch": 0.4818413803592855,
      "grad_norm": 0.20663602650165558,
      "learning_rate": 1.7665364546404034e-05,
      "loss": 0.08997774124145508,
      "step": 3564
    },
    {
      "epoch": 0.48197657715512143,
      "grad_norm": 0.13580411672592163,
      "learning_rate": 1.7658628760652548e-05,
      "loss": 0.06884914636611938,
      "step": 3565
    },
    {
      "epoch": 0.4821117739509574,
      "grad_norm": 0.09598405659198761,
      "learning_rate": 1.765189242136038e-05,
      "loss": 0.06521987915039062,
      "step": 3566
    },
    {
      "epoch": 0.4822469707467933,
      "grad_norm": 0.21622607111930847,
      "learning_rate": 1.7645155529930065e-05,
      "loss": 0.07096123695373535,
      "step": 3567
    },
    {
      "epoch": 0.48238216754262925,
      "grad_norm": 0.05366864427924156,
      "learning_rate": 1.763841808776426e-05,
      "loss": 0.03513383865356445,
      "step": 3568
    },
    {
      "epoch": 0.4825173643384652,
      "grad_norm": 0.1085478886961937,
      "learning_rate": 1.763168009626575e-05,
      "loss": 0.06385660171508789,
      "step": 3569
    },
    {
      "epoch": 0.4826525611343011,
      "grad_norm": 0.0757722407579422,
      "learning_rate": 1.7624941556837406e-05,
      "loss": 0.07055854797363281,
      "step": 3570
    },
    {
      "epoch": 0.48278775793013706,
      "grad_norm": 0.08150186389684677,
      "learning_rate": 1.7618202470882233e-05,
      "loss": 0.08629035949707031,
      "step": 3571
    },
    {
      "epoch": 0.482922954725973,
      "grad_norm": 0.08880028873682022,
      "learning_rate": 1.7611462839803336e-05,
      "loss": 0.08350419998168945,
      "step": 3572
    },
    {
      "epoch": 0.4830581515218089,
      "grad_norm": 0.07939331978559494,
      "learning_rate": 1.760472266500396e-05,
      "loss": 0.0688624382019043,
      "step": 3573
    },
    {
      "epoch": 0.4831933483176449,
      "grad_norm": 0.075597383081913,
      "learning_rate": 1.759798194788743e-05,
      "loss": 0.07290935516357422,
      "step": 3574
    },
    {
      "epoch": 0.48332854511348083,
      "grad_norm": 0.04727097600698471,
      "learning_rate": 1.75912406898572e-05,
      "loss": 0.057602882385253906,
      "step": 3575
    },
    {
      "epoch": 0.48346374190931674,
      "grad_norm": 0.13879527151584625,
      "learning_rate": 1.758449889231685e-05,
      "loss": 0.06453084945678711,
      "step": 3576
    },
    {
      "epoch": 0.4835989387051527,
      "grad_norm": 0.07641217857599258,
      "learning_rate": 1.757775655667004e-05,
      "loss": 0.08345484733581543,
      "step": 3577
    },
    {
      "epoch": 0.48373413550098865,
      "grad_norm": 0.07564567029476166,
      "learning_rate": 1.757101368432057e-05,
      "loss": 0.06930351257324219,
      "step": 3578
    },
    {
      "epoch": 0.48386933229682455,
      "grad_norm": 0.08055280148983002,
      "learning_rate": 1.7564270276672343e-05,
      "loss": 0.0822610855102539,
      "step": 3579
    },
    {
      "epoch": 0.4840045290926605,
      "grad_norm": 0.14254067838191986,
      "learning_rate": 1.7557526335129372e-05,
      "loss": 0.08722805976867676,
      "step": 3580
    },
    {
      "epoch": 0.48413972588849646,
      "grad_norm": 0.1177094429731369,
      "learning_rate": 1.7550781861095774e-05,
      "loss": 0.06356048583984375,
      "step": 3581
    },
    {
      "epoch": 0.48427492268433237,
      "grad_norm": 0.10753617435693741,
      "learning_rate": 1.754403685597579e-05,
      "loss": 0.05353355407714844,
      "step": 3582
    },
    {
      "epoch": 0.4844101194801683,
      "grad_norm": 0.09060823917388916,
      "learning_rate": 1.7537291321173773e-05,
      "loss": 0.05819261074066162,
      "step": 3583
    },
    {
      "epoch": 0.4845453162760043,
      "grad_norm": 0.10979875922203064,
      "learning_rate": 1.7530545258094165e-05,
      "loss": 0.08277839422225952,
      "step": 3584
    },
    {
      "epoch": 0.4846805130718402,
      "grad_norm": 0.058302171528339386,
      "learning_rate": 1.7523798668141548e-05,
      "loss": 0.049314022064208984,
      "step": 3585
    },
    {
      "epoch": 0.48481570986767614,
      "grad_norm": 0.03541146591305733,
      "learning_rate": 1.7517051552720584e-05,
      "loss": 0.03382599353790283,
      "step": 3586
    },
    {
      "epoch": 0.4849509066635121,
      "grad_norm": 0.048140544444322586,
      "learning_rate": 1.7510303913236066e-05,
      "loss": 0.06548905372619629,
      "step": 3587
    },
    {
      "epoch": 0.485086103459348,
      "grad_norm": 0.05853385105729103,
      "learning_rate": 1.7503555751092883e-05,
      "loss": 0.05188095569610596,
      "step": 3588
    },
    {
      "epoch": 0.48522130025518395,
      "grad_norm": 0.06466836482286453,
      "learning_rate": 1.7496807067696046e-05,
      "loss": 0.05787205696105957,
      "step": 3589
    },
    {
      "epoch": 0.4853564970510199,
      "grad_norm": 0.13970918953418732,
      "learning_rate": 1.7490057864450665e-05,
      "loss": 0.07719898223876953,
      "step": 3590
    },
    {
      "epoch": 0.4854916938468558,
      "grad_norm": 0.041163310408592224,
      "learning_rate": 1.748330814276195e-05,
      "loss": 0.04050576686859131,
      "step": 3591
    },
    {
      "epoch": 0.48562689064269177,
      "grad_norm": 0.09408541023731232,
      "learning_rate": 1.7476557904035243e-05,
      "loss": 0.06473207473754883,
      "step": 3592
    },
    {
      "epoch": 0.4857620874385277,
      "grad_norm": 0.08710946887731552,
      "learning_rate": 1.7469807149675973e-05,
      "loss": 0.08213043212890625,
      "step": 3593
    },
    {
      "epoch": 0.4858972842343636,
      "grad_norm": 0.05058041960000992,
      "learning_rate": 1.7463055881089685e-05,
      "loss": 0.02990889549255371,
      "step": 3594
    },
    {
      "epoch": 0.4860324810301996,
      "grad_norm": 0.05483940988779068,
      "learning_rate": 1.7456304099682024e-05,
      "loss": 0.05761528015136719,
      "step": 3595
    },
    {
      "epoch": 0.48616767782603554,
      "grad_norm": 0.10008472949266434,
      "learning_rate": 1.7449551806858756e-05,
      "loss": 0.0710608959197998,
      "step": 3596
    },
    {
      "epoch": 0.48630287462187144,
      "grad_norm": 0.06381729990243912,
      "learning_rate": 1.7442799004025733e-05,
      "loss": 0.06379079818725586,
      "step": 3597
    },
    {
      "epoch": 0.4864380714177074,
      "grad_norm": 0.06842538714408875,
      "learning_rate": 1.7436045692588934e-05,
      "loss": 0.038065433502197266,
      "step": 3598
    },
    {
      "epoch": 0.48657326821354335,
      "grad_norm": 0.08415386825799942,
      "learning_rate": 1.742929187395443e-05,
      "loss": 0.08133077621459961,
      "step": 3599
    },
    {
      "epoch": 0.48670846500937925,
      "grad_norm": 0.14112652838230133,
      "learning_rate": 1.7422537549528402e-05,
      "loss": 0.07901191711425781,
      "step": 3600
    },
    {
      "epoch": 0.4868436618052152,
      "grad_norm": 0.08802326023578644,
      "learning_rate": 1.7415782720717147e-05,
      "loss": 0.059496402740478516,
      "step": 3601
    },
    {
      "epoch": 0.48697885860105117,
      "grad_norm": 0.06313256919384003,
      "learning_rate": 1.740902738892704e-05,
      "loss": 0.04536402225494385,
      "step": 3602
    },
    {
      "epoch": 0.48711405539688707,
      "grad_norm": 0.05303634703159332,
      "learning_rate": 1.7402271555564585e-05,
      "loss": 0.059635162353515625,
      "step": 3603
    },
    {
      "epoch": 0.487249252192723,
      "grad_norm": 0.11761408299207687,
      "learning_rate": 1.739551522203638e-05,
      "loss": 0.06266450881958008,
      "step": 3604
    },
    {
      "epoch": 0.487384448988559,
      "grad_norm": 0.13394463062286377,
      "learning_rate": 1.738875838974913e-05,
      "loss": 0.06246042251586914,
      "step": 3605
    },
    {
      "epoch": 0.4875196457843949,
      "grad_norm": 0.07512262463569641,
      "learning_rate": 1.7382001060109652e-05,
      "loss": 0.05531167984008789,
      "step": 3606
    },
    {
      "epoch": 0.48765484258023084,
      "grad_norm": 0.04697803407907486,
      "learning_rate": 1.7375243234524843e-05,
      "loss": 0.05104684829711914,
      "step": 3607
    },
    {
      "epoch": 0.4877900393760668,
      "grad_norm": 0.06704489141702652,
      "learning_rate": 1.736848491440173e-05,
      "loss": 0.06655192375183105,
      "step": 3608
    },
    {
      "epoch": 0.4879252361719027,
      "grad_norm": 0.08053556829690933,
      "learning_rate": 1.7361726101147424e-05,
      "loss": 0.057255566120147705,
      "step": 3609
    },
    {
      "epoch": 0.48806043296773866,
      "grad_norm": 0.1130124032497406,
      "learning_rate": 1.7354966796169157e-05,
      "loss": 0.08051919937133789,
      "step": 3610
    },
    {
      "epoch": 0.4881956297635746,
      "grad_norm": 0.03626268357038498,
      "learning_rate": 1.734820700087424e-05,
      "loss": 0.03832656145095825,
      "step": 3611
    },
    {
      "epoch": 0.48833082655941057,
      "grad_norm": 0.05131611227989197,
      "learning_rate": 1.7341446716670103e-05,
      "loss": 0.05534648895263672,
      "step": 3612
    },
    {
      "epoch": 0.48846602335524647,
      "grad_norm": 0.04659602791070938,
      "learning_rate": 1.7334685944964272e-05,
      "loss": 0.05809736251831055,
      "step": 3613
    },
    {
      "epoch": 0.4886012201510824,
      "grad_norm": 0.03651859238743782,
      "learning_rate": 1.732792468716438e-05,
      "loss": 0.0412445068359375,
      "step": 3614
    },
    {
      "epoch": 0.4887364169469184,
      "grad_norm": 0.22989794611930847,
      "learning_rate": 1.7321162944678155e-05,
      "loss": 0.07867467403411865,
      "step": 3615
    },
    {
      "epoch": 0.4888716137427543,
      "grad_norm": 0.07495442777872086,
      "learning_rate": 1.7314400718913425e-05,
      "loss": 0.06592893600463867,
      "step": 3616
    },
    {
      "epoch": 0.48900681053859024,
      "grad_norm": 0.033692941069602966,
      "learning_rate": 1.7307638011278126e-05,
      "loss": 0.03146100044250488,
      "step": 3617
    },
    {
      "epoch": 0.4891420073344262,
      "grad_norm": 0.05789799615740776,
      "learning_rate": 1.7300874823180284e-05,
      "loss": 0.052855849266052246,
      "step": 3618
    },
    {
      "epoch": 0.4892772041302621,
      "grad_norm": 0.06451264023780823,
      "learning_rate": 1.7294111156028034e-05,
      "loss": 0.06437468528747559,
      "step": 3619
    },
    {
      "epoch": 0.48941240092609806,
      "grad_norm": 0.06568501144647598,
      "learning_rate": 1.7287347011229605e-05,
      "loss": 0.0803375244140625,
      "step": 3620
    },
    {
      "epoch": 0.489547597721934,
      "grad_norm": 0.1086663156747818,
      "learning_rate": 1.7280582390193333e-05,
      "loss": 0.06152153015136719,
      "step": 3621
    },
    {
      "epoch": 0.4896827945177699,
      "grad_norm": 0.08800483494997025,
      "learning_rate": 1.7273817294327653e-05,
      "loss": 0.055284976959228516,
      "step": 3622
    },
    {
      "epoch": 0.48981799131360587,
      "grad_norm": 0.04633105918765068,
      "learning_rate": 1.726705172504108e-05,
      "loss": 0.06363499164581299,
      "step": 3623
    },
    {
      "epoch": 0.48995318810944183,
      "grad_norm": 0.0646023079752922,
      "learning_rate": 1.7260285683742248e-05,
      "loss": 0.07406139373779297,
      "step": 3624
    },
    {
      "epoch": 0.49008838490527773,
      "grad_norm": 0.09981629252433777,
      "learning_rate": 1.7253519171839883e-05,
      "loss": 0.05177879333496094,
      "step": 3625
    },
    {
      "epoch": 0.4902235817011137,
      "grad_norm": 0.1757117360830307,
      "learning_rate": 1.724675219074281e-05,
      "loss": 0.06247490644454956,
      "step": 3626
    },
    {
      "epoch": 0.49035877849694964,
      "grad_norm": 0.06993858516216278,
      "learning_rate": 1.7239984741859957e-05,
      "loss": 0.06765270233154297,
      "step": 3627
    },
    {
      "epoch": 0.49049397529278554,
      "grad_norm": 0.12846733629703522,
      "learning_rate": 1.7233216826600324e-05,
      "loss": 0.08030715584754944,
      "step": 3628
    },
    {
      "epoch": 0.4906291720886215,
      "grad_norm": 0.07241996377706528,
      "learning_rate": 1.7226448446373047e-05,
      "loss": 0.05104684829711914,
      "step": 3629
    },
    {
      "epoch": 0.49076436888445746,
      "grad_norm": 0.07053696364164352,
      "learning_rate": 1.7219679602587326e-05,
      "loss": 0.06361579895019531,
      "step": 3630
    },
    {
      "epoch": 0.49089956568029336,
      "grad_norm": 0.1293111890554428,
      "learning_rate": 1.7212910296652476e-05,
      "loss": 0.06764888763427734,
      "step": 3631
    },
    {
      "epoch": 0.4910347624761293,
      "grad_norm": 0.13575351238250732,
      "learning_rate": 1.7206140529977904e-05,
      "loss": 0.06544160842895508,
      "step": 3632
    },
    {
      "epoch": 0.4911699592719653,
      "grad_norm": 0.09848814457654953,
      "learning_rate": 1.719937030397311e-05,
      "loss": 0.07082962989807129,
      "step": 3633
    },
    {
      "epoch": 0.4913051560678012,
      "grad_norm": 0.25757378339767456,
      "learning_rate": 1.7192599620047683e-05,
      "loss": 0.0867757797241211,
      "step": 3634
    },
    {
      "epoch": 0.49144035286363713,
      "grad_norm": 0.1287541389465332,
      "learning_rate": 1.718582847961133e-05,
      "loss": 0.06615257263183594,
      "step": 3635
    },
    {
      "epoch": 0.4915755496594731,
      "grad_norm": 0.1424255669116974,
      "learning_rate": 1.7179056884073826e-05,
      "loss": 0.07771587371826172,
      "step": 3636
    },
    {
      "epoch": 0.491710746455309,
      "grad_norm": 0.08382894843816757,
      "learning_rate": 1.717228483484506e-05,
      "loss": 0.06627535820007324,
      "step": 3637
    },
    {
      "epoch": 0.49184594325114495,
      "grad_norm": 0.09832364320755005,
      "learning_rate": 1.7165512333335013e-05,
      "loss": 0.06449282169342041,
      "step": 3638
    },
    {
      "epoch": 0.4919811400469809,
      "grad_norm": 0.14199407398700714,
      "learning_rate": 1.715873938095374e-05,
      "loss": 0.08585596084594727,
      "step": 3639
    },
    {
      "epoch": 0.4921163368428168,
      "grad_norm": 0.05253595486283302,
      "learning_rate": 1.7151965979111427e-05,
      "loss": 0.03324699401855469,
      "step": 3640
    },
    {
      "epoch": 0.49225153363865276,
      "grad_norm": 0.07695692032575607,
      "learning_rate": 1.7145192129218313e-05,
      "loss": 0.038543701171875,
      "step": 3641
    },
    {
      "epoch": 0.4923867304344887,
      "grad_norm": 0.1592692732810974,
      "learning_rate": 1.7138417832684763e-05,
      "loss": 0.08372068405151367,
      "step": 3642
    },
    {
      "epoch": 0.4925219272303246,
      "grad_norm": 0.04907463863492012,
      "learning_rate": 1.7131643090921216e-05,
      "loss": 0.06655740737915039,
      "step": 3643
    },
    {
      "epoch": 0.4926571240261606,
      "grad_norm": 0.24179919064044952,
      "learning_rate": 1.712486790533821e-05,
      "loss": 0.1024484634399414,
      "step": 3644
    },
    {
      "epoch": 0.49279232082199653,
      "grad_norm": 0.047212354838848114,
      "learning_rate": 1.7118092277346372e-05,
      "loss": 0.04586261510848999,
      "step": 3645
    },
    {
      "epoch": 0.49292751761783243,
      "grad_norm": 0.10214903205633163,
      "learning_rate": 1.7111316208356428e-05,
      "loss": 0.055303215980529785,
      "step": 3646
    },
    {
      "epoch": 0.4930627144136684,
      "grad_norm": 0.21530210971832275,
      "learning_rate": 1.7104539699779192e-05,
      "loss": 0.06785714626312256,
      "step": 3647
    },
    {
      "epoch": 0.49319791120950435,
      "grad_norm": 0.08425609767436981,
      "learning_rate": 1.709776275302557e-05,
      "loss": 0.049014151096343994,
      "step": 3648
    },
    {
      "epoch": 0.49333310800534025,
      "grad_norm": 0.2440614104270935,
      "learning_rate": 1.7090985369506555e-05,
      "loss": 0.07528328895568848,
      "step": 3649
    },
    {
      "epoch": 0.4934683048011762,
      "grad_norm": 0.07584003359079361,
      "learning_rate": 1.708420755063323e-05,
      "loss": 0.07402420043945312,
      "step": 3650
    },
    {
      "epoch": 0.49360350159701216,
      "grad_norm": 0.07362735271453857,
      "learning_rate": 1.707742929781678e-05,
      "loss": 0.05284547805786133,
      "step": 3651
    },
    {
      "epoch": 0.49373869839284806,
      "grad_norm": 0.10645918548107147,
      "learning_rate": 1.707065061246848e-05,
      "loss": 0.06648111343383789,
      "step": 3652
    },
    {
      "epoch": 0.493873895188684,
      "grad_norm": 0.09508925676345825,
      "learning_rate": 1.7063871495999677e-05,
      "loss": 0.07498025894165039,
      "step": 3653
    },
    {
      "epoch": 0.49400909198452,
      "grad_norm": 0.1568058729171753,
      "learning_rate": 1.705709194982182e-05,
      "loss": 0.05193781852722168,
      "step": 3654
    },
    {
      "epoch": 0.4941442887803559,
      "grad_norm": 0.07692322134971619,
      "learning_rate": 1.7050311975346447e-05,
      "loss": 0.06574797630310059,
      "step": 3655
    },
    {
      "epoch": 0.49427948557619183,
      "grad_norm": 0.20145468413829803,
      "learning_rate": 1.704353157398519e-05,
      "loss": 0.0744791030883789,
      "step": 3656
    },
    {
      "epoch": 0.4944146823720278,
      "grad_norm": 0.08423712104558945,
      "learning_rate": 1.7036750747149764e-05,
      "loss": 0.05287003517150879,
      "step": 3657
    },
    {
      "epoch": 0.49454987916786375,
      "grad_norm": 0.20200200378894806,
      "learning_rate": 1.702996949625197e-05,
      "loss": 0.08007049560546875,
      "step": 3658
    },
    {
      "epoch": 0.49468507596369965,
      "grad_norm": 0.10411909222602844,
      "learning_rate": 1.7023187822703702e-05,
      "loss": 0.05687999725341797,
      "step": 3659
    },
    {
      "epoch": 0.4948202727595356,
      "grad_norm": 0.10596667230129242,
      "learning_rate": 1.7016405727916936e-05,
      "loss": 0.08284330368041992,
      "step": 3660
    },
    {
      "epoch": 0.49495546955537156,
      "grad_norm": 0.0667446106672287,
      "learning_rate": 1.700962321330375e-05,
      "loss": 0.05919456481933594,
      "step": 3661
    },
    {
      "epoch": 0.49509066635120746,
      "grad_norm": 0.06044870615005493,
      "learning_rate": 1.700284028027629e-05,
      "loss": 0.04858824610710144,
      "step": 3662
    },
    {
      "epoch": 0.4952258631470434,
      "grad_norm": 0.20683208107948303,
      "learning_rate": 1.6996056930246807e-05,
      "loss": 0.07144784927368164,
      "step": 3663
    },
    {
      "epoch": 0.4953610599428794,
      "grad_norm": 0.11469860374927521,
      "learning_rate": 1.6989273164627626e-05,
      "loss": 0.09452939033508301,
      "step": 3664
    },
    {
      "epoch": 0.4954962567387153,
      "grad_norm": 0.12053488940000534,
      "learning_rate": 1.6982488984831163e-05,
      "loss": 0.06110095977783203,
      "step": 3665
    },
    {
      "epoch": 0.49563145353455124,
      "grad_norm": 0.10469727218151093,
      "learning_rate": 1.697570439226992e-05,
      "loss": 0.06834030151367188,
      "step": 3666
    },
    {
      "epoch": 0.4957666503303872,
      "grad_norm": 0.0829186663031578,
      "learning_rate": 1.6968919388356486e-05,
      "loss": 0.06743144989013672,
      "step": 3667
    },
    {
      "epoch": 0.4959018471262231,
      "grad_norm": 0.12245159596204758,
      "learning_rate": 1.696213397450354e-05,
      "loss": 0.07387471199035645,
      "step": 3668
    },
    {
      "epoch": 0.49603704392205905,
      "grad_norm": 0.06037118658423424,
      "learning_rate": 1.695534815212384e-05,
      "loss": 0.07586240768432617,
      "step": 3669
    },
    {
      "epoch": 0.496172240717895,
      "grad_norm": 0.055406391620635986,
      "learning_rate": 1.6948561922630223e-05,
      "loss": 0.061498045921325684,
      "step": 3670
    },
    {
      "epoch": 0.4963074375137309,
      "grad_norm": 0.0674186646938324,
      "learning_rate": 1.694177528743562e-05,
      "loss": 0.05174517631530762,
      "step": 3671
    },
    {
      "epoch": 0.49644263430956687,
      "grad_norm": 0.06779282540082932,
      "learning_rate": 1.6934988247953053e-05,
      "loss": 0.06182551383972168,
      "step": 3672
    },
    {
      "epoch": 0.4965778311054028,
      "grad_norm": 0.09276220947504044,
      "learning_rate": 1.6928200805595606e-05,
      "loss": 0.07191336154937744,
      "step": 3673
    },
    {
      "epoch": 0.4967130279012387,
      "grad_norm": 0.0914679765701294,
      "learning_rate": 1.6921412961776475e-05,
      "loss": 0.0748758316040039,
      "step": 3674
    },
    {
      "epoch": 0.4968482246970747,
      "grad_norm": 0.11273293942213058,
      "learning_rate": 1.6914624717908922e-05,
      "loss": 0.07414913177490234,
      "step": 3675
    },
    {
      "epoch": 0.49698342149291064,
      "grad_norm": 0.06812527775764465,
      "learning_rate": 1.6907836075406288e-05,
      "loss": 0.07167196273803711,
      "step": 3676
    },
    {
      "epoch": 0.49711861828874654,
      "grad_norm": 0.14902490377426147,
      "learning_rate": 1.690104703568201e-05,
      "loss": 0.05316972732543945,
      "step": 3677
    },
    {
      "epoch": 0.4972538150845825,
      "grad_norm": 0.05334148928523064,
      "learning_rate": 1.68942576001496e-05,
      "loss": 0.05182605981826782,
      "step": 3678
    },
    {
      "epoch": 0.49738901188041845,
      "grad_norm": 0.12150225043296814,
      "learning_rate": 1.6887467770222658e-05,
      "loss": 0.06699976325035095,
      "step": 3679
    },
    {
      "epoch": 0.49752420867625435,
      "grad_norm": 0.0888490304350853,
      "learning_rate": 1.6880677547314865e-05,
      "loss": 0.05602145195007324,
      "step": 3680
    },
    {
      "epoch": 0.4976594054720903,
      "grad_norm": 0.05888473615050316,
      "learning_rate": 1.6873886932839973e-05,
      "loss": 0.0517694354057312,
      "step": 3681
    },
    {
      "epoch": 0.49779460226792627,
      "grad_norm": 0.08135475963354111,
      "learning_rate": 1.686709592821183e-05,
      "loss": 0.08391571044921875,
      "step": 3682
    },
    {
      "epoch": 0.49792979906376217,
      "grad_norm": 0.09950217604637146,
      "learning_rate": 1.6860304534844355e-05,
      "loss": 0.07316875457763672,
      "step": 3683
    },
    {
      "epoch": 0.4980649958595981,
      "grad_norm": 0.1802351474761963,
      "learning_rate": 1.6853512754151556e-05,
      "loss": 0.08319240808486938,
      "step": 3684
    },
    {
      "epoch": 0.4982001926554341,
      "grad_norm": 0.03477051854133606,
      "learning_rate": 1.684672058754752e-05,
      "loss": 0.04135942459106445,
      "step": 3685
    },
    {
      "epoch": 0.49833538945127,
      "grad_norm": 0.10808668285608292,
      "learning_rate": 1.6839928036446416e-05,
      "loss": 0.07411050796508789,
      "step": 3686
    },
    {
      "epoch": 0.49847058624710594,
      "grad_norm": 0.12113899737596512,
      "learning_rate": 1.6833135102262473e-05,
      "loss": 0.07147717475891113,
      "step": 3687
    },
    {
      "epoch": 0.4986057830429419,
      "grad_norm": 0.11671570688486099,
      "learning_rate": 1.682634178641003e-05,
      "loss": 0.045951128005981445,
      "step": 3688
    },
    {
      "epoch": 0.4987409798387778,
      "grad_norm": 0.2194662094116211,
      "learning_rate": 1.6819548090303485e-05,
      "loss": 0.08658981323242188,
      "step": 3689
    },
    {
      "epoch": 0.49887617663461375,
      "grad_norm": 0.06902020424604416,
      "learning_rate": 1.6812754015357328e-05,
      "loss": 0.07514506578445435,
      "step": 3690
    },
    {
      "epoch": 0.4990113734304497,
      "grad_norm": 0.05978160351514816,
      "learning_rate": 1.680595956298612e-05,
      "loss": 0.07438373565673828,
      "step": 3691
    },
    {
      "epoch": 0.4991465702262856,
      "grad_norm": 0.06871736794710159,
      "learning_rate": 1.6799164734604497e-05,
      "loss": 0.07107377052307129,
      "step": 3692
    },
    {
      "epoch": 0.49928176702212157,
      "grad_norm": 0.09108826518058777,
      "learning_rate": 1.6792369531627186e-05,
      "loss": 0.057581424713134766,
      "step": 3693
    },
    {
      "epoch": 0.4994169638179575,
      "grad_norm": 0.12614762783050537,
      "learning_rate": 1.6785573955468974e-05,
      "loss": 0.04407024383544922,
      "step": 3694
    },
    {
      "epoch": 0.49955216061379343,
      "grad_norm": 0.1276492327451706,
      "learning_rate": 1.6778778007544745e-05,
      "loss": 0.06745529174804688,
      "step": 3695
    },
    {
      "epoch": 0.4996873574096294,
      "grad_norm": 0.16838513314723969,
      "learning_rate": 1.6771981689269452e-05,
      "loss": 0.0907735824584961,
      "step": 3696
    },
    {
      "epoch": 0.49982255420546534,
      "grad_norm": 0.092658132314682,
      "learning_rate": 1.6765185002058123e-05,
      "loss": 0.05747032165527344,
      "step": 3697
    },
    {
      "epoch": 0.49995775100130124,
      "grad_norm": 0.025818796828389168,
      "learning_rate": 1.6758387947325856e-05,
      "loss": 0.028505563735961914,
      "step": 3698
    },
    {
      "epoch": 0.5000929477971372,
      "grad_norm": 0.06602969020605087,
      "learning_rate": 1.6751590526487843e-05,
      "loss": 0.07781982421875,
      "step": 3699
    },
    {
      "epoch": 0.5002281445929732,
      "grad_norm": 0.20559480786323547,
      "learning_rate": 1.6744792740959347e-05,
      "loss": 0.08217430114746094,
      "step": 3700
    },
    {
      "epoch": 0.5003633413888091,
      "grad_norm": 0.14475670456886292,
      "learning_rate": 1.6737994592155697e-05,
      "loss": 0.05979299545288086,
      "step": 3701
    },
    {
      "epoch": 0.5004985381846451,
      "grad_norm": 0.08835698664188385,
      "learning_rate": 1.6731196081492307e-05,
      "loss": 0.05028557777404785,
      "step": 3702
    },
    {
      "epoch": 0.5006337349804809,
      "grad_norm": 0.07242413610219955,
      "learning_rate": 1.6724397210384655e-05,
      "loss": 0.05842924118041992,
      "step": 3703
    },
    {
      "epoch": 0.5007689317763169,
      "grad_norm": 0.08495945483446121,
      "learning_rate": 1.6717597980248308e-05,
      "loss": 0.06853437423706055,
      "step": 3704
    },
    {
      "epoch": 0.5009041285721528,
      "grad_norm": 0.13081209361553192,
      "learning_rate": 1.6710798392498904e-05,
      "loss": 0.07030391693115234,
      "step": 3705
    },
    {
      "epoch": 0.5010393253679888,
      "grad_norm": 0.040592074394226074,
      "learning_rate": 1.6703998448552154e-05,
      "loss": 0.04258108139038086,
      "step": 3706
    },
    {
      "epoch": 0.5011745221638247,
      "grad_norm": 0.056849293410778046,
      "learning_rate": 1.669719814982384e-05,
      "loss": 0.05554676055908203,
      "step": 3707
    },
    {
      "epoch": 0.5013097189596607,
      "grad_norm": 0.16090774536132812,
      "learning_rate": 1.6690397497729818e-05,
      "loss": 0.11242866516113281,
      "step": 3708
    },
    {
      "epoch": 0.5014449157554965,
      "grad_norm": 0.12958185374736786,
      "learning_rate": 1.6683596493686028e-05,
      "loss": 0.08212113380432129,
      "step": 3709
    },
    {
      "epoch": 0.5015801125513325,
      "grad_norm": 0.14468182623386383,
      "learning_rate": 1.667679513910846e-05,
      "loss": 0.06597638130187988,
      "step": 3710
    },
    {
      "epoch": 0.5017153093471685,
      "grad_norm": 0.06954637169837952,
      "learning_rate": 1.666999343541321e-05,
      "loss": 0.048244476318359375,
      "step": 3711
    },
    {
      "epoch": 0.5018505061430044,
      "grad_norm": 0.03504554554820061,
      "learning_rate": 1.6663191384016422e-05,
      "loss": 0.04031825065612793,
      "step": 3712
    },
    {
      "epoch": 0.5019857029388404,
      "grad_norm": 0.058307476341724396,
      "learning_rate": 1.6656388986334315e-05,
      "loss": 0.07852506637573242,
      "step": 3713
    },
    {
      "epoch": 0.5021208997346763,
      "grad_norm": 0.08634979277849197,
      "learning_rate": 1.6649586243783186e-05,
      "loss": 0.06403493881225586,
      "step": 3714
    },
    {
      "epoch": 0.5022560965305122,
      "grad_norm": 0.03979789838194847,
      "learning_rate": 1.6642783157779405e-05,
      "loss": 0.04753732681274414,
      "step": 3715
    },
    {
      "epoch": 0.5023912933263481,
      "grad_norm": 0.08048854023218155,
      "learning_rate": 1.6635979729739417e-05,
      "loss": 0.06908845901489258,
      "step": 3716
    },
    {
      "epoch": 0.5025264901221841,
      "grad_norm": 0.07217130064964294,
      "learning_rate": 1.662917596107972e-05,
      "loss": 0.07325267791748047,
      "step": 3717
    },
    {
      "epoch": 0.50266168691802,
      "grad_norm": 0.1316104680299759,
      "learning_rate": 1.6622371853216904e-05,
      "loss": 0.07314467430114746,
      "step": 3718
    },
    {
      "epoch": 0.502796883713856,
      "grad_norm": 0.09273051470518112,
      "learning_rate": 1.661556740756761e-05,
      "loss": 0.056643009185791016,
      "step": 3719
    },
    {
      "epoch": 0.502932080509692,
      "grad_norm": 0.10645793378353119,
      "learning_rate": 1.6608762625548572e-05,
      "loss": 0.060303688049316406,
      "step": 3720
    },
    {
      "epoch": 0.5030672773055278,
      "grad_norm": 0.05880333483219147,
      "learning_rate": 1.6601957508576573e-05,
      "loss": 0.02784597873687744,
      "step": 3721
    },
    {
      "epoch": 0.5032024741013638,
      "grad_norm": 0.051762621849775314,
      "learning_rate": 1.659515205806848e-05,
      "loss": 0.0576329231262207,
      "step": 3722
    },
    {
      "epoch": 0.5033376708971997,
      "grad_norm": 0.0424882210791111,
      "learning_rate": 1.6588346275441224e-05,
      "loss": 0.04112982749938965,
      "step": 3723
    },
    {
      "epoch": 0.5034728676930357,
      "grad_norm": 0.06540235131978989,
      "learning_rate": 1.65815401621118e-05,
      "loss": 0.06955790519714355,
      "step": 3724
    },
    {
      "epoch": 0.5036080644888716,
      "grad_norm": 0.08217894285917282,
      "learning_rate": 1.657473371949728e-05,
      "loss": 0.05295515060424805,
      "step": 3725
    },
    {
      "epoch": 0.5037432612847076,
      "grad_norm": 0.10725366324186325,
      "learning_rate": 1.6567926949014805e-05,
      "loss": 0.09947586059570312,
      "step": 3726
    },
    {
      "epoch": 0.5038784580805435,
      "grad_norm": 0.08663515746593475,
      "learning_rate": 1.6561119852081574e-05,
      "loss": 0.08110237121582031,
      "step": 3727
    },
    {
      "epoch": 0.5040136548763794,
      "grad_norm": 0.046510547399520874,
      "learning_rate": 1.6554312430114868e-05,
      "loss": 0.0476679801940918,
      "step": 3728
    },
    {
      "epoch": 0.5041488516722153,
      "grad_norm": 0.08510484546422958,
      "learning_rate": 1.6547504684532026e-05,
      "loss": 0.07415771484375,
      "step": 3729
    },
    {
      "epoch": 0.5042840484680513,
      "grad_norm": 0.10251069068908691,
      "learning_rate": 1.6540696616750454e-05,
      "loss": 0.07366853952407837,
      "step": 3730
    },
    {
      "epoch": 0.5044192452638873,
      "grad_norm": 0.05395134910941124,
      "learning_rate": 1.6533888228187628e-05,
      "loss": 0.05528545379638672,
      "step": 3731
    },
    {
      "epoch": 0.5045544420597232,
      "grad_norm": 0.06983628869056702,
      "learning_rate": 1.6527079520261103e-05,
      "loss": 0.05981945991516113,
      "step": 3732
    },
    {
      "epoch": 0.5046896388555592,
      "grad_norm": 0.060526005923748016,
      "learning_rate": 1.6520270494388472e-05,
      "loss": 0.0650949478149414,
      "step": 3733
    },
    {
      "epoch": 0.504824835651395,
      "grad_norm": 0.04767141863703728,
      "learning_rate": 1.6513461151987418e-05,
      "loss": 0.05769610404968262,
      "step": 3734
    },
    {
      "epoch": 0.504960032447231,
      "grad_norm": 0.036564599722623825,
      "learning_rate": 1.6506651494475678e-05,
      "loss": 0.038811683654785156,
      "step": 3735
    },
    {
      "epoch": 0.5050952292430669,
      "grad_norm": 0.07894016802310944,
      "learning_rate": 1.6499841523271062e-05,
      "loss": 0.0546497106552124,
      "step": 3736
    },
    {
      "epoch": 0.5052304260389029,
      "grad_norm": 0.10773193836212158,
      "learning_rate": 1.649303123979145e-05,
      "loss": 0.09120464324951172,
      "step": 3737
    },
    {
      "epoch": 0.5053656228347388,
      "grad_norm": 0.060023244470357895,
      "learning_rate": 1.648622064545477e-05,
      "loss": 0.06358909606933594,
      "step": 3738
    },
    {
      "epoch": 0.5055008196305748,
      "grad_norm": 0.0951104462146759,
      "learning_rate": 1.6479409741679025e-05,
      "loss": 0.05770683288574219,
      "step": 3739
    },
    {
      "epoch": 0.5056360164264107,
      "grad_norm": 0.07395204156637192,
      "learning_rate": 1.6472598529882277e-05,
      "loss": 0.06883478164672852,
      "step": 3740
    },
    {
      "epoch": 0.5057712132222466,
      "grad_norm": 0.07054848223924637,
      "learning_rate": 1.646578701148267e-05,
      "loss": 0.07321429252624512,
      "step": 3741
    },
    {
      "epoch": 0.5059064100180826,
      "grad_norm": 0.1394336223602295,
      "learning_rate": 1.6458975187898384e-05,
      "loss": 0.09001421928405762,
      "step": 3742
    },
    {
      "epoch": 0.5060416068139185,
      "grad_norm": 0.04635302722454071,
      "learning_rate": 1.6452163060547687e-05,
      "loss": 0.04274177551269531,
      "step": 3743
    },
    {
      "epoch": 0.5061768036097545,
      "grad_norm": 0.12482871115207672,
      "learning_rate": 1.64453506308489e-05,
      "loss": 0.05527004599571228,
      "step": 3744
    },
    {
      "epoch": 0.5063120004055904,
      "grad_norm": 0.19521453976631165,
      "learning_rate": 1.64385379002204e-05,
      "loss": 0.08733940124511719,
      "step": 3745
    },
    {
      "epoch": 0.5064471972014263,
      "grad_norm": 0.0754537582397461,
      "learning_rate": 1.643172487008064e-05,
      "loss": 0.054754823446273804,
      "step": 3746
    },
    {
      "epoch": 0.5065823939972622,
      "grad_norm": 0.1390322893857956,
      "learning_rate": 1.6424911541848124e-05,
      "loss": 0.05157053470611572,
      "step": 3747
    },
    {
      "epoch": 0.5067175907930982,
      "grad_norm": 0.05547186732292175,
      "learning_rate": 1.641809791694143e-05,
      "loss": 0.04738926887512207,
      "step": 3748
    },
    {
      "epoch": 0.5068527875889342,
      "grad_norm": 0.07509680092334747,
      "learning_rate": 1.6411283996779184e-05,
      "loss": 0.05946412682533264,
      "step": 3749
    },
    {
      "epoch": 0.5069879843847701,
      "grad_norm": 0.06471659243106842,
      "learning_rate": 1.6404469782780088e-05,
      "loss": 0.039231061935424805,
      "step": 3750
    },
    {
      "epoch": 0.5071231811806061,
      "grad_norm": 0.06595001369714737,
      "learning_rate": 1.639765527636289e-05,
      "loss": 0.08158111572265625,
      "step": 3751
    },
    {
      "epoch": 0.5072583779764419,
      "grad_norm": 0.07521500438451767,
      "learning_rate": 1.639084047894641e-05,
      "loss": 0.05309092998504639,
      "step": 3752
    },
    {
      "epoch": 0.5073935747722779,
      "grad_norm": 0.0801013633608818,
      "learning_rate": 1.638402539194953e-05,
      "loss": 0.054356396198272705,
      "step": 3753
    },
    {
      "epoch": 0.5075287715681138,
      "grad_norm": 0.05563773959875107,
      "learning_rate": 1.6377210016791182e-05,
      "loss": 0.033054351806640625,
      "step": 3754
    },
    {
      "epoch": 0.5076639683639498,
      "grad_norm": 0.19027762115001678,
      "learning_rate": 1.6370394354890364e-05,
      "loss": 0.0989912748336792,
      "step": 3755
    },
    {
      "epoch": 0.5077991651597857,
      "grad_norm": 0.07743227481842041,
      "learning_rate": 1.636357840766613e-05,
      "loss": 0.054848670959472656,
      "step": 3756
    },
    {
      "epoch": 0.5079343619556217,
      "grad_norm": 0.07018070667982101,
      "learning_rate": 1.6356762176537606e-05,
      "loss": 0.05977439880371094,
      "step": 3757
    },
    {
      "epoch": 0.5080695587514575,
      "grad_norm": 0.06427748501300812,
      "learning_rate": 1.6349945662923953e-05,
      "loss": 0.0861368179321289,
      "step": 3758
    },
    {
      "epoch": 0.5082047555472935,
      "grad_norm": 0.11053955554962158,
      "learning_rate": 1.634312886824442e-05,
      "loss": 0.05287642776966095,
      "step": 3759
    },
    {
      "epoch": 0.5083399523431295,
      "grad_norm": 0.06569293886423111,
      "learning_rate": 1.6336311793918298e-05,
      "loss": 0.07151281833648682,
      "step": 3760
    },
    {
      "epoch": 0.5084751491389654,
      "grad_norm": 0.07489728182554245,
      "learning_rate": 1.6329494441364925e-05,
      "loss": 0.04281347990036011,
      "step": 3761
    },
    {
      "epoch": 0.5086103459348014,
      "grad_norm": 0.08263158798217773,
      "learning_rate": 1.6322676812003727e-05,
      "loss": 0.06351804733276367,
      "step": 3762
    },
    {
      "epoch": 0.5087455427306373,
      "grad_norm": 0.0663142204284668,
      "learning_rate": 1.631585890725416e-05,
      "loss": 0.06160569190979004,
      "step": 3763
    },
    {
      "epoch": 0.5088807395264732,
      "grad_norm": 0.1277400702238083,
      "learning_rate": 1.630904072853575e-05,
      "loss": 0.09601640701293945,
      "step": 3764
    },
    {
      "epoch": 0.5090159363223091,
      "grad_norm": 0.054404523223638535,
      "learning_rate": 1.6302222277268085e-05,
      "loss": 0.05263924598693848,
      "step": 3765
    },
    {
      "epoch": 0.5091511331181451,
      "grad_norm": 0.04608814790844917,
      "learning_rate": 1.6295403554870794e-05,
      "loss": 0.04669177532196045,
      "step": 3766
    },
    {
      "epoch": 0.509286329913981,
      "grad_norm": 0.06972187012434006,
      "learning_rate": 1.6288584562763572e-05,
      "loss": 0.06536746025085449,
      "step": 3767
    },
    {
      "epoch": 0.509421526709817,
      "grad_norm": 0.07748934626579285,
      "learning_rate": 1.6281765302366176e-05,
      "loss": 0.10449600219726562,
      "step": 3768
    },
    {
      "epoch": 0.509556723505653,
      "grad_norm": 0.0524292066693306,
      "learning_rate": 1.6274945775098412e-05,
      "loss": 0.06367683410644531,
      "step": 3769
    },
    {
      "epoch": 0.5096919203014889,
      "grad_norm": 0.03959411755204201,
      "learning_rate": 1.6268125982380135e-05,
      "loss": 0.045714378356933594,
      "step": 3770
    },
    {
      "epoch": 0.5098271170973248,
      "grad_norm": 0.14590390026569366,
      "learning_rate": 1.626130592563127e-05,
      "loss": 0.06103801727294922,
      "step": 3771
    },
    {
      "epoch": 0.5099623138931607,
      "grad_norm": 0.0810171589255333,
      "learning_rate": 1.6254485606271778e-05,
      "loss": 0.04927778244018555,
      "step": 3772
    },
    {
      "epoch": 0.5100975106889967,
      "grad_norm": 0.057023655623197556,
      "learning_rate": 1.6247665025721698e-05,
      "loss": 0.04679107666015625,
      "step": 3773
    },
    {
      "epoch": 0.5102327074848326,
      "grad_norm": 0.07838837802410126,
      "learning_rate": 1.62408441854011e-05,
      "loss": 0.05614662170410156,
      "step": 3774
    },
    {
      "epoch": 0.5103679042806686,
      "grad_norm": 0.08667189627885818,
      "learning_rate": 1.6234023086730136e-05,
      "loss": 0.06268644332885742,
      "step": 3775
    },
    {
      "epoch": 0.5105031010765045,
      "grad_norm": 0.04647170752286911,
      "learning_rate": 1.622720173112898e-05,
      "loss": 0.06048393249511719,
      "step": 3776
    },
    {
      "epoch": 0.5106382978723404,
      "grad_norm": 0.07768401503562927,
      "learning_rate": 1.6220380120017874e-05,
      "loss": 0.05697762966156006,
      "step": 3777
    },
    {
      "epoch": 0.5107734946681763,
      "grad_norm": 0.11173009872436523,
      "learning_rate": 1.6213558254817128e-05,
      "loss": 0.05269980430603027,
      "step": 3778
    },
    {
      "epoch": 0.5109086914640123,
      "grad_norm": 0.15954984724521637,
      "learning_rate": 1.6206736136947074e-05,
      "loss": 0.08355236053466797,
      "step": 3779
    },
    {
      "epoch": 0.5110438882598483,
      "grad_norm": 0.06876694411039352,
      "learning_rate": 1.6199913767828126e-05,
      "loss": 0.05315274000167847,
      "step": 3780
    },
    {
      "epoch": 0.5111790850556842,
      "grad_norm": 0.04431116208434105,
      "learning_rate": 1.6193091148880733e-05,
      "loss": 0.060036659240722656,
      "step": 3781
    },
    {
      "epoch": 0.5113142818515202,
      "grad_norm": 0.029200371354818344,
      "learning_rate": 1.61862682815254e-05,
      "loss": 0.034078359603881836,
      "step": 3782
    },
    {
      "epoch": 0.511449478647356,
      "grad_norm": 0.042952120304107666,
      "learning_rate": 1.617944516718268e-05,
      "loss": 0.04280126094818115,
      "step": 3783
    },
    {
      "epoch": 0.511584675443192,
      "grad_norm": 0.07698074728250504,
      "learning_rate": 1.617262180727319e-05,
      "loss": 0.06352090835571289,
      "step": 3784
    },
    {
      "epoch": 0.5117198722390279,
      "grad_norm": 0.10133420675992966,
      "learning_rate": 1.6165798203217588e-05,
      "loss": 0.051175594329833984,
      "step": 3785
    },
    {
      "epoch": 0.5118550690348639,
      "grad_norm": 0.06485810875892639,
      "learning_rate": 1.6158974356436585e-05,
      "loss": 0.06676959991455078,
      "step": 3786
    },
    {
      "epoch": 0.5119902658306998,
      "grad_norm": 0.08356793224811554,
      "learning_rate": 1.6152150268350938e-05,
      "loss": 0.06603431701660156,
      "step": 3787
    },
    {
      "epoch": 0.5121254626265358,
      "grad_norm": 0.03928566351532936,
      "learning_rate": 1.6145325940381458e-05,
      "loss": 0.04350090026855469,
      "step": 3788
    },
    {
      "epoch": 0.5122606594223716,
      "grad_norm": 0.1155201643705368,
      "learning_rate": 1.6138501373949018e-05,
      "loss": 0.07770967483520508,
      "step": 3789
    },
    {
      "epoch": 0.5123958562182076,
      "grad_norm": 0.077711321413517,
      "learning_rate": 1.613167657047451e-05,
      "loss": 0.07222151756286621,
      "step": 3790
    },
    {
      "epoch": 0.5125310530140436,
      "grad_norm": 0.10896597802639008,
      "learning_rate": 1.612485153137891e-05,
      "loss": 0.0713735818862915,
      "step": 3791
    },
    {
      "epoch": 0.5126662498098795,
      "grad_norm": 0.09788154065608978,
      "learning_rate": 1.611802625808323e-05,
      "loss": 0.062265872955322266,
      "step": 3792
    },
    {
      "epoch": 0.5128014466057155,
      "grad_norm": 0.05886108800768852,
      "learning_rate": 1.611120075200851e-05,
      "loss": 0.05328443646430969,
      "step": 3793
    },
    {
      "epoch": 0.5129366434015514,
      "grad_norm": 0.058363646268844604,
      "learning_rate": 1.610437501457587e-05,
      "loss": 0.04350912570953369,
      "step": 3794
    },
    {
      "epoch": 0.5130718401973873,
      "grad_norm": 0.14264824986457825,
      "learning_rate": 1.6097549047206464e-05,
      "loss": 0.06175494194030762,
      "step": 3795
    },
    {
      "epoch": 0.5132070369932232,
      "grad_norm": 0.09353962540626526,
      "learning_rate": 1.6090722851321497e-05,
      "loss": 0.07037687301635742,
      "step": 3796
    },
    {
      "epoch": 0.5133422337890592,
      "grad_norm": 0.14305047690868378,
      "learning_rate": 1.6083896428342213e-05,
      "loss": 0.08239507675170898,
      "step": 3797
    },
    {
      "epoch": 0.5134774305848951,
      "grad_norm": 0.0963195413351059,
      "learning_rate": 1.6077069779689915e-05,
      "loss": 0.04875802993774414,
      "step": 3798
    },
    {
      "epoch": 0.5136126273807311,
      "grad_norm": 0.1686539500951767,
      "learning_rate": 1.607024290678594e-05,
      "loss": 0.0822446346282959,
      "step": 3799
    },
    {
      "epoch": 0.5137478241765671,
      "grad_norm": 0.0595044381916523,
      "learning_rate": 1.6063415811051686e-05,
      "loss": 0.061818599700927734,
      "step": 3800
    },
    {
      "epoch": 0.5138830209724029,
      "grad_norm": 0.11236492544412613,
      "learning_rate": 1.6056588493908596e-05,
      "loss": 0.08846747875213623,
      "step": 3801
    },
    {
      "epoch": 0.5140182177682389,
      "grad_norm": 0.049755096435546875,
      "learning_rate": 1.604976095677814e-05,
      "loss": 0.05275154113769531,
      "step": 3802
    },
    {
      "epoch": 0.5141534145640748,
      "grad_norm": 0.06233778968453407,
      "learning_rate": 1.604293320108186e-05,
      "loss": 0.06415486335754395,
      "step": 3803
    },
    {
      "epoch": 0.5142886113599108,
      "grad_norm": 0.11785244941711426,
      "learning_rate": 1.603610522824132e-05,
      "loss": 0.07109951972961426,
      "step": 3804
    },
    {
      "epoch": 0.5144238081557467,
      "grad_norm": 0.06423136591911316,
      "learning_rate": 1.6029277039678153e-05,
      "loss": 0.06672000885009766,
      "step": 3805
    },
    {
      "epoch": 0.5145590049515827,
      "grad_norm": 0.07795750349760056,
      "learning_rate": 1.602244863681401e-05,
      "loss": 0.08669614791870117,
      "step": 3806
    },
    {
      "epoch": 0.5146942017474185,
      "grad_norm": 0.06414157897233963,
      "learning_rate": 1.6015620021070613e-05,
      "loss": 0.06005525588989258,
      "step": 3807
    },
    {
      "epoch": 0.5148293985432545,
      "grad_norm": 0.06245529279112816,
      "learning_rate": 1.6008791193869714e-05,
      "loss": 0.06514453887939453,
      "step": 3808
    },
    {
      "epoch": 0.5149645953390904,
      "grad_norm": 0.043420981615781784,
      "learning_rate": 1.6001962156633102e-05,
      "loss": 0.042061805725097656,
      "step": 3809
    },
    {
      "epoch": 0.5150997921349264,
      "grad_norm": 0.08287721127271652,
      "learning_rate": 1.5995132910782632e-05,
      "loss": 0.061681151390075684,
      "step": 3810
    },
    {
      "epoch": 0.5152349889307624,
      "grad_norm": 0.051789380609989166,
      "learning_rate": 1.5988303457740178e-05,
      "loss": 0.040281474590301514,
      "step": 3811
    },
    {
      "epoch": 0.5153701857265983,
      "grad_norm": 0.06059059873223305,
      "learning_rate": 1.598147379892768e-05,
      "loss": 0.0915842056274414,
      "step": 3812
    },
    {
      "epoch": 0.5155053825224342,
      "grad_norm": 0.12110874056816101,
      "learning_rate": 1.5974643935767098e-05,
      "loss": 0.06841802597045898,
      "step": 3813
    },
    {
      "epoch": 0.5156405793182701,
      "grad_norm": 0.048348553478717804,
      "learning_rate": 1.5967813869680452e-05,
      "loss": 0.059702157974243164,
      "step": 3814
    },
    {
      "epoch": 0.5157757761141061,
      "grad_norm": 0.07379832118749619,
      "learning_rate": 1.59609836020898e-05,
      "loss": 0.053717851638793945,
      "step": 3815
    },
    {
      "epoch": 0.515910972909942,
      "grad_norm": 0.09354014694690704,
      "learning_rate": 1.5954153134417236e-05,
      "loss": 0.08050727844238281,
      "step": 3816
    },
    {
      "epoch": 0.516046169705778,
      "grad_norm": 0.04181956499814987,
      "learning_rate": 1.59473224680849e-05,
      "loss": 0.05160093307495117,
      "step": 3817
    },
    {
      "epoch": 0.516181366501614,
      "grad_norm": 0.08444039523601532,
      "learning_rate": 1.5940491604514976e-05,
      "loss": 0.04850679636001587,
      "step": 3818
    },
    {
      "epoch": 0.5163165632974499,
      "grad_norm": 0.08802709728479385,
      "learning_rate": 1.5933660545129683e-05,
      "loss": 0.055463552474975586,
      "step": 3819
    },
    {
      "epoch": 0.5164517600932857,
      "grad_norm": 0.05439211055636406,
      "learning_rate": 1.5926829291351288e-05,
      "loss": 0.0729532241821289,
      "step": 3820
    },
    {
      "epoch": 0.5165869568891217,
      "grad_norm": 0.07838249206542969,
      "learning_rate": 1.591999784460209e-05,
      "loss": 0.07324886322021484,
      "step": 3821
    },
    {
      "epoch": 0.5167221536849577,
      "grad_norm": 0.06595619022846222,
      "learning_rate": 1.5913166206304435e-05,
      "loss": 0.04148983955383301,
      "step": 3822
    },
    {
      "epoch": 0.5168573504807936,
      "grad_norm": 0.09841666370630264,
      "learning_rate": 1.5906334377880707e-05,
      "loss": 0.09315657615661621,
      "step": 3823
    },
    {
      "epoch": 0.5169925472766296,
      "grad_norm": 0.11421307176351547,
      "learning_rate": 1.589950236075333e-05,
      "loss": 0.0876319408416748,
      "step": 3824
    },
    {
      "epoch": 0.5171277440724655,
      "grad_norm": 0.05589543655514717,
      "learning_rate": 1.5892670156344764e-05,
      "loss": 0.06772851943969727,
      "step": 3825
    },
    {
      "epoch": 0.5172629408683014,
      "grad_norm": 0.052531734108924866,
      "learning_rate": 1.588583776607751e-05,
      "loss": 0.056597232818603516,
      "step": 3826
    },
    {
      "epoch": 0.5173981376641373,
      "grad_norm": 0.1090838611125946,
      "learning_rate": 1.5879005191374106e-05,
      "loss": 0.05065488815307617,
      "step": 3827
    },
    {
      "epoch": 0.5175333344599733,
      "grad_norm": 0.06877259165048599,
      "learning_rate": 1.587217243365714e-05,
      "loss": 0.053522706031799316,
      "step": 3828
    },
    {
      "epoch": 0.5176685312558092,
      "grad_norm": 0.06729893386363983,
      "learning_rate": 1.586533949434922e-05,
      "loss": 0.04121077060699463,
      "step": 3829
    },
    {
      "epoch": 0.5178037280516452,
      "grad_norm": 0.0439026914536953,
      "learning_rate": 1.5858506374872998e-05,
      "loss": 0.04289674758911133,
      "step": 3830
    },
    {
      "epoch": 0.5179389248474812,
      "grad_norm": 0.18009690940380096,
      "learning_rate": 1.5851673076651178e-05,
      "loss": 0.09005510807037354,
      "step": 3831
    },
    {
      "epoch": 0.518074121643317,
      "grad_norm": 0.16789914667606354,
      "learning_rate": 1.5844839601106477e-05,
      "loss": 0.07492780685424805,
      "step": 3832
    },
    {
      "epoch": 0.518209318439153,
      "grad_norm": 0.06930071860551834,
      "learning_rate": 1.583800594966167e-05,
      "loss": 0.05796194076538086,
      "step": 3833
    },
    {
      "epoch": 0.5183445152349889,
      "grad_norm": 0.12920333445072174,
      "learning_rate": 1.583117212373955e-05,
      "loss": 0.0737161636352539,
      "step": 3834
    },
    {
      "epoch": 0.5184797120308249,
      "grad_norm": 0.09299825876951218,
      "learning_rate": 1.5824338124762967e-05,
      "loss": 0.06281715631484985,
      "step": 3835
    },
    {
      "epoch": 0.5186149088266608,
      "grad_norm": 0.10455896705389023,
      "learning_rate": 1.581750395415479e-05,
      "loss": 0.0762324333190918,
      "step": 3836
    },
    {
      "epoch": 0.5187501056224968,
      "grad_norm": 0.09843981266021729,
      "learning_rate": 1.5810669613337922e-05,
      "loss": 0.07065558433532715,
      "step": 3837
    },
    {
      "epoch": 0.5188853024183326,
      "grad_norm": 0.062306541949510574,
      "learning_rate": 1.5803835103735327e-05,
      "loss": 0.059501975774765015,
      "step": 3838
    },
    {
      "epoch": 0.5190204992141686,
      "grad_norm": 0.09388948231935501,
      "learning_rate": 1.5797000426769973e-05,
      "loss": 0.041937828063964844,
      "step": 3839
    },
    {
      "epoch": 0.5191556960100046,
      "grad_norm": 0.08753836154937744,
      "learning_rate": 1.579016558386488e-05,
      "loss": 0.07376432418823242,
      "step": 3840
    },
    {
      "epoch": 0.5192908928058405,
      "grad_norm": 0.05474778264760971,
      "learning_rate": 1.5783330576443096e-05,
      "loss": 0.062407493591308594,
      "step": 3841
    },
    {
      "epoch": 0.5194260896016765,
      "grad_norm": 0.10432812571525574,
      "learning_rate": 1.5776495405927716e-05,
      "loss": 0.06308603286743164,
      "step": 3842
    },
    {
      "epoch": 0.5195612863975124,
      "grad_norm": 0.06874200701713562,
      "learning_rate": 1.5769660073741844e-05,
      "loss": 0.06066441535949707,
      "step": 3843
    },
    {
      "epoch": 0.5196964831933483,
      "grad_norm": 0.098721943795681,
      "learning_rate": 1.5762824581308645e-05,
      "loss": 0.06487464904785156,
      "step": 3844
    },
    {
      "epoch": 0.5198316799891842,
      "grad_norm": 0.14265795052051544,
      "learning_rate": 1.5755988930051304e-05,
      "loss": 0.06925880908966064,
      "step": 3845
    },
    {
      "epoch": 0.5199668767850202,
      "grad_norm": 0.09727565199136734,
      "learning_rate": 1.5749153121393025e-05,
      "loss": 0.07371652126312256,
      "step": 3846
    },
    {
      "epoch": 0.5201020735808561,
      "grad_norm": 0.12147000432014465,
      "learning_rate": 1.574231715675708e-05,
      "loss": 0.06255006790161133,
      "step": 3847
    },
    {
      "epoch": 0.5202372703766921,
      "grad_norm": 0.048076607286930084,
      "learning_rate": 1.573548103756674e-05,
      "loss": 0.04524517059326172,
      "step": 3848
    },
    {
      "epoch": 0.520372467172528,
      "grad_norm": 0.045020125806331635,
      "learning_rate": 1.572864476524533e-05,
      "loss": 0.06437432765960693,
      "step": 3849
    },
    {
      "epoch": 0.5205076639683639,
      "grad_norm": 0.09541922062635422,
      "learning_rate": 1.5721808341216195e-05,
      "loss": 0.08754968643188477,
      "step": 3850
    },
    {
      "epoch": 0.5206428607641999,
      "grad_norm": 0.14167389273643494,
      "learning_rate": 1.571497176690271e-05,
      "loss": 0.07863259315490723,
      "step": 3851
    },
    {
      "epoch": 0.5207780575600358,
      "grad_norm": 0.06314025819301605,
      "learning_rate": 1.570813504372829e-05,
      "loss": 0.07802295684814453,
      "step": 3852
    },
    {
      "epoch": 0.5209132543558718,
      "grad_norm": 0.09422006458044052,
      "learning_rate": 1.570129817311638e-05,
      "loss": 0.057814836502075195,
      "step": 3853
    },
    {
      "epoch": 0.5210484511517077,
      "grad_norm": 0.07615645229816437,
      "learning_rate": 1.5694461156490452e-05,
      "loss": 0.0606694221496582,
      "step": 3854
    },
    {
      "epoch": 0.5211836479475437,
      "grad_norm": 0.07192464917898178,
      "learning_rate": 1.5687623995274008e-05,
      "loss": 0.060402095317840576,
      "step": 3855
    },
    {
      "epoch": 0.5213188447433795,
      "grad_norm": 0.11170166730880737,
      "learning_rate": 1.568078669089058e-05,
      "loss": 0.07771480083465576,
      "step": 3856
    },
    {
      "epoch": 0.5214540415392155,
      "grad_norm": 0.18437263369560242,
      "learning_rate": 1.567394924476373e-05,
      "loss": 0.07514524459838867,
      "step": 3857
    },
    {
      "epoch": 0.5215892383350514,
      "grad_norm": 0.06449896842241287,
      "learning_rate": 1.5667111658317057e-05,
      "loss": 0.057717084884643555,
      "step": 3858
    },
    {
      "epoch": 0.5217244351308874,
      "grad_norm": 0.11347664147615433,
      "learning_rate": 1.5660273932974177e-05,
      "loss": 0.06470084190368652,
      "step": 3859
    },
    {
      "epoch": 0.5218596319267234,
      "grad_norm": 0.12989670038223267,
      "learning_rate": 1.5653436070158743e-05,
      "loss": 0.06696939468383789,
      "step": 3860
    },
    {
      "epoch": 0.5219948287225593,
      "grad_norm": 0.09307479858398438,
      "learning_rate": 1.564659807129444e-05,
      "loss": 0.05035877227783203,
      "step": 3861
    },
    {
      "epoch": 0.5221300255183953,
      "grad_norm": 0.06069663539528847,
      "learning_rate": 1.5639759937804962e-05,
      "loss": 0.049747586250305176,
      "step": 3862
    },
    {
      "epoch": 0.5222652223142311,
      "grad_norm": 0.07519473880529404,
      "learning_rate": 1.5632921671114055e-05,
      "loss": 0.041859984397888184,
      "step": 3863
    },
    {
      "epoch": 0.5224004191100671,
      "grad_norm": 0.16680215299129486,
      "learning_rate": 1.5626083272645485e-05,
      "loss": 0.09029364585876465,
      "step": 3864
    },
    {
      "epoch": 0.522535615905903,
      "grad_norm": 0.059998974204063416,
      "learning_rate": 1.5619244743823038e-05,
      "loss": 0.06173241138458252,
      "step": 3865
    },
    {
      "epoch": 0.522670812701739,
      "grad_norm": 0.07698603719472885,
      "learning_rate": 1.5612406086070534e-05,
      "loss": 0.09070968627929688,
      "step": 3866
    },
    {
      "epoch": 0.5228060094975749,
      "grad_norm": 0.06496132165193558,
      "learning_rate": 1.560556730081181e-05,
      "loss": 0.05671834945678711,
      "step": 3867
    },
    {
      "epoch": 0.5229412062934109,
      "grad_norm": 0.17899605631828308,
      "learning_rate": 1.5598728389470754e-05,
      "loss": 0.08618974685668945,
      "step": 3868
    },
    {
      "epoch": 0.5230764030892467,
      "grad_norm": 0.07173573970794678,
      "learning_rate": 1.5591889353471245e-05,
      "loss": 0.05062144994735718,
      "step": 3869
    },
    {
      "epoch": 0.5232115998850827,
      "grad_norm": 0.07148000597953796,
      "learning_rate": 1.5585050194237226e-05,
      "loss": 0.08269500732421875,
      "step": 3870
    },
    {
      "epoch": 0.5233467966809187,
      "grad_norm": 0.1372024267911911,
      "learning_rate": 1.557821091319263e-05,
      "loss": 0.05329108238220215,
      "step": 3871
    },
    {
      "epoch": 0.5234819934767546,
      "grad_norm": 0.05171942710876465,
      "learning_rate": 1.5571371511761446e-05,
      "loss": 0.0505826473236084,
      "step": 3872
    },
    {
      "epoch": 0.5236171902725906,
      "grad_norm": 0.05921654403209686,
      "learning_rate": 1.5564531991367658e-05,
      "loss": 0.07123398780822754,
      "step": 3873
    },
    {
      "epoch": 0.5237523870684265,
      "grad_norm": 0.04116828739643097,
      "learning_rate": 1.5557692353435302e-05,
      "loss": 0.03347063064575195,
      "step": 3874
    },
    {
      "epoch": 0.5238875838642624,
      "grad_norm": 0.04731554538011551,
      "learning_rate": 1.5550852599388424e-05,
      "loss": 0.05790519714355469,
      "step": 3875
    },
    {
      "epoch": 0.5240227806600983,
      "grad_norm": 0.07882832735776901,
      "learning_rate": 1.5544012730651096e-05,
      "loss": 0.07251739501953125,
      "step": 3876
    },
    {
      "epoch": 0.5241579774559343,
      "grad_norm": 0.16371144354343414,
      "learning_rate": 1.5537172748647422e-05,
      "loss": 0.06999373435974121,
      "step": 3877
    },
    {
      "epoch": 0.5242931742517702,
      "grad_norm": 0.07022348791360855,
      "learning_rate": 1.553033265480151e-05,
      "loss": 0.053800106048583984,
      "step": 3878
    },
    {
      "epoch": 0.5244283710476062,
      "grad_norm": 0.09335823357105255,
      "learning_rate": 1.552349245053752e-05,
      "loss": 0.07693004608154297,
      "step": 3879
    },
    {
      "epoch": 0.5245635678434422,
      "grad_norm": 0.16085992753505707,
      "learning_rate": 1.5516652137279597e-05,
      "loss": 0.0805203914642334,
      "step": 3880
    },
    {
      "epoch": 0.524698764639278,
      "grad_norm": 0.09730314463376999,
      "learning_rate": 1.5509811716451955e-05,
      "loss": 0.051087260246276855,
      "step": 3881
    },
    {
      "epoch": 0.524833961435114,
      "grad_norm": 0.13653609156608582,
      "learning_rate": 1.550297118947879e-05,
      "loss": 0.06461071968078613,
      "step": 3882
    },
    {
      "epoch": 0.5249691582309499,
      "grad_norm": 0.11773184686899185,
      "learning_rate": 1.5496130557784343e-05,
      "loss": 0.06762957572937012,
      "step": 3883
    },
    {
      "epoch": 0.5251043550267859,
      "grad_norm": 0.10872945934534073,
      "learning_rate": 1.5489289822792868e-05,
      "loss": 0.07314491271972656,
      "step": 3884
    },
    {
      "epoch": 0.5252395518226218,
      "grad_norm": 0.14969436824321747,
      "learning_rate": 1.5482448985928645e-05,
      "loss": 0.08587002754211426,
      "step": 3885
    },
    {
      "epoch": 0.5253747486184578,
      "grad_norm": 0.10076700896024704,
      "learning_rate": 1.5475608048615964e-05,
      "loss": 0.07973664999008179,
      "step": 3886
    },
    {
      "epoch": 0.5255099454142936,
      "grad_norm": 0.11089161783456802,
      "learning_rate": 1.546876701227916e-05,
      "loss": 0.05116748809814453,
      "step": 3887
    },
    {
      "epoch": 0.5256451422101296,
      "grad_norm": 0.06001724675297737,
      "learning_rate": 1.5461925878342558e-05,
      "loss": 0.05179119110107422,
      "step": 3888
    },
    {
      "epoch": 0.5257803390059655,
      "grad_norm": 0.07770363241434097,
      "learning_rate": 1.5455084648230527e-05,
      "loss": 0.03802156448364258,
      "step": 3889
    },
    {
      "epoch": 0.5259155358018015,
      "grad_norm": 0.23153872787952423,
      "learning_rate": 1.5448243323367438e-05,
      "loss": 0.0949256420135498,
      "step": 3890
    },
    {
      "epoch": 0.5260507325976375,
      "grad_norm": 0.06408070027828217,
      "learning_rate": 1.544140190517771e-05,
      "loss": 0.059844970703125,
      "step": 3891
    },
    {
      "epoch": 0.5261859293934734,
      "grad_norm": 0.10859911888837814,
      "learning_rate": 1.5434560395085745e-05,
      "loss": 0.06161069869995117,
      "step": 3892
    },
    {
      "epoch": 0.5263211261893093,
      "grad_norm": 0.0758669376373291,
      "learning_rate": 1.542771879451599e-05,
      "loss": 0.06345462799072266,
      "step": 3893
    },
    {
      "epoch": 0.5264563229851452,
      "grad_norm": 0.09149164706468582,
      "learning_rate": 1.54208771048929e-05,
      "loss": 0.07653465867042542,
      "step": 3894
    },
    {
      "epoch": 0.5265915197809812,
      "grad_norm": 0.06464672833681107,
      "learning_rate": 1.5414035327640958e-05,
      "loss": 0.06371927261352539,
      "step": 3895
    },
    {
      "epoch": 0.5267267165768171,
      "grad_norm": 0.10314838588237762,
      "learning_rate": 1.5407193464184644e-05,
      "loss": 0.08294963836669922,
      "step": 3896
    },
    {
      "epoch": 0.5268619133726531,
      "grad_norm": 0.09328769892454147,
      "learning_rate": 1.5400351515948485e-05,
      "loss": 0.037291765213012695,
      "step": 3897
    },
    {
      "epoch": 0.526997110168489,
      "grad_norm": 0.08296113461256027,
      "learning_rate": 1.5393509484357006e-05,
      "loss": 0.05197262763977051,
      "step": 3898
    },
    {
      "epoch": 0.5271323069643249,
      "grad_norm": 0.121526800096035,
      "learning_rate": 1.538666737083475e-05,
      "loss": 0.05584794282913208,
      "step": 3899
    },
    {
      "epoch": 0.5272675037601608,
      "grad_norm": 0.050672899931669235,
      "learning_rate": 1.537982517680629e-05,
      "loss": 0.04543423652648926,
      "step": 3900
    },
    {
      "epoch": 0.5274027005559968,
      "grad_norm": 0.07551522552967072,
      "learning_rate": 1.5372982903696196e-05,
      "loss": 0.0659552812576294,
      "step": 3901
    },
    {
      "epoch": 0.5275378973518328,
      "grad_norm": 0.040404438972473145,
      "learning_rate": 1.536614055292908e-05,
      "loss": 0.0595707893371582,
      "step": 3902
    },
    {
      "epoch": 0.5276730941476687,
      "grad_norm": 0.13261736929416656,
      "learning_rate": 1.535929812592955e-05,
      "loss": 0.05104660987854004,
      "step": 3903
    },
    {
      "epoch": 0.5278082909435047,
      "grad_norm": 0.08313792943954468,
      "learning_rate": 1.5352455624122227e-05,
      "loss": 0.062470436096191406,
      "step": 3904
    },
    {
      "epoch": 0.5279434877393405,
      "grad_norm": 0.058765213936567307,
      "learning_rate": 1.5345613048931765e-05,
      "loss": 0.061493873596191406,
      "step": 3905
    },
    {
      "epoch": 0.5280786845351765,
      "grad_norm": 0.0861872136592865,
      "learning_rate": 1.5338770401782822e-05,
      "loss": 0.08306741714477539,
      "step": 3906
    },
    {
      "epoch": 0.5282138813310124,
      "grad_norm": 0.02905181236565113,
      "learning_rate": 1.5331927684100077e-05,
      "loss": 0.03806662559509277,
      "step": 3907
    },
    {
      "epoch": 0.5283490781268484,
      "grad_norm": 0.0869058221578598,
      "learning_rate": 1.5325084897308218e-05,
      "loss": 0.06251716613769531,
      "step": 3908
    },
    {
      "epoch": 0.5284842749226843,
      "grad_norm": 0.15307335555553436,
      "learning_rate": 1.5318242042831952e-05,
      "loss": 0.0908346176147461,
      "step": 3909
    },
    {
      "epoch": 0.5286194717185203,
      "grad_norm": 0.1766705960035324,
      "learning_rate": 1.5311399122095992e-05,
      "loss": 0.0809323787689209,
      "step": 3910
    },
    {
      "epoch": 0.5287546685143563,
      "grad_norm": 0.11445577442646027,
      "learning_rate": 1.5304556136525074e-05,
      "loss": 0.042017459869384766,
      "step": 3911
    },
    {
      "epoch": 0.5288898653101921,
      "grad_norm": 0.10634942352771759,
      "learning_rate": 1.5297713087543948e-05,
      "loss": 0.06469142436981201,
      "step": 3912
    },
    {
      "epoch": 0.5290250621060281,
      "grad_norm": 0.08895298838615417,
      "learning_rate": 1.5290869976577365e-05,
      "loss": 0.09692764282226562,
      "step": 3913
    },
    {
      "epoch": 0.529160258901864,
      "grad_norm": 0.07691548019647598,
      "learning_rate": 1.5284026805050107e-05,
      "loss": 0.06772613525390625,
      "step": 3914
    },
    {
      "epoch": 0.5292954556977,
      "grad_norm": 0.09204553812742233,
      "learning_rate": 1.5277183574386947e-05,
      "loss": 0.06881999969482422,
      "step": 3915
    },
    {
      "epoch": 0.5294306524935359,
      "grad_norm": 0.06572859734296799,
      "learning_rate": 1.5270340286012694e-05,
      "loss": 0.05770134925842285,
      "step": 3916
    },
    {
      "epoch": 0.5295658492893719,
      "grad_norm": 0.12960654497146606,
      "learning_rate": 1.526349694135215e-05,
      "loss": 0.04240679740905762,
      "step": 3917
    },
    {
      "epoch": 0.5297010460852077,
      "grad_norm": 0.0583483912050724,
      "learning_rate": 1.525665354183014e-05,
      "loss": 0.060099124908447266,
      "step": 3918
    },
    {
      "epoch": 0.5298362428810437,
      "grad_norm": 0.14548785984516144,
      "learning_rate": 1.5249810088871493e-05,
      "loss": 0.08008337020874023,
      "step": 3919
    },
    {
      "epoch": 0.5299714396768797,
      "grad_norm": 0.1373683512210846,
      "learning_rate": 1.5242966583901052e-05,
      "loss": 0.07833051681518555,
      "step": 3920
    },
    {
      "epoch": 0.5301066364727156,
      "grad_norm": 0.14249083399772644,
      "learning_rate": 1.523612302834367e-05,
      "loss": 0.11403656005859375,
      "step": 3921
    },
    {
      "epoch": 0.5302418332685516,
      "grad_norm": 0.07043417543172836,
      "learning_rate": 1.5229279423624217e-05,
      "loss": 0.04156756401062012,
      "step": 3922
    },
    {
      "epoch": 0.5303770300643875,
      "grad_norm": 0.07135765999555588,
      "learning_rate": 1.5222435771167566e-05,
      "loss": 0.06475543975830078,
      "step": 3923
    },
    {
      "epoch": 0.5305122268602234,
      "grad_norm": 0.19948060810565948,
      "learning_rate": 1.5215592072398602e-05,
      "loss": 0.08450007438659668,
      "step": 3924
    },
    {
      "epoch": 0.5306474236560593,
      "grad_norm": 0.07923898100852966,
      "learning_rate": 1.520874832874222e-05,
      "loss": 0.032118797302246094,
      "step": 3925
    },
    {
      "epoch": 0.5307826204518953,
      "grad_norm": 0.0771009624004364,
      "learning_rate": 1.5201904541623318e-05,
      "loss": 0.05563783645629883,
      "step": 3926
    },
    {
      "epoch": 0.5309178172477312,
      "grad_norm": 0.046507254242897034,
      "learning_rate": 1.5195060712466817e-05,
      "loss": 0.05328008532524109,
      "step": 3927
    },
    {
      "epoch": 0.5310530140435672,
      "grad_norm": 0.06295576691627502,
      "learning_rate": 1.5188216842697635e-05,
      "loss": 0.05476522445678711,
      "step": 3928
    },
    {
      "epoch": 0.5311882108394032,
      "grad_norm": 0.1106257289648056,
      "learning_rate": 1.5181372933740703e-05,
      "loss": 0.05213189125061035,
      "step": 3929
    },
    {
      "epoch": 0.531323407635239,
      "grad_norm": 0.10583825409412384,
      "learning_rate": 1.5174528987020958e-05,
      "loss": 0.058868408203125,
      "step": 3930
    },
    {
      "epoch": 0.531458604431075,
      "grad_norm": 0.03729039803147316,
      "learning_rate": 1.5167685003963345e-05,
      "loss": 0.03513073921203613,
      "step": 3931
    },
    {
      "epoch": 0.5315938012269109,
      "grad_norm": 0.05793580785393715,
      "learning_rate": 1.5160840985992824e-05,
      "loss": 0.06871342658996582,
      "step": 3932
    },
    {
      "epoch": 0.5317289980227469,
      "grad_norm": 0.09303639084100723,
      "learning_rate": 1.515399693453435e-05,
      "loss": 0.06128692626953125,
      "step": 3933
    },
    {
      "epoch": 0.5318641948185828,
      "grad_norm": 0.08739953488111496,
      "learning_rate": 1.5147152851012894e-05,
      "loss": 0.05667126178741455,
      "step": 3934
    },
    {
      "epoch": 0.5319993916144188,
      "grad_norm": 0.047449346631765366,
      "learning_rate": 1.514030873685343e-05,
      "loss": 0.04632049798965454,
      "step": 3935
    },
    {
      "epoch": 0.5321345884102546,
      "grad_norm": 0.08154062926769257,
      "learning_rate": 1.513346459348094e-05,
      "loss": 0.05924797058105469,
      "step": 3936
    },
    {
      "epoch": 0.5322697852060906,
      "grad_norm": 0.09296199679374695,
      "learning_rate": 1.5126620422320405e-05,
      "loss": 0.07582998275756836,
      "step": 3937
    },
    {
      "epoch": 0.5324049820019265,
      "grad_norm": 0.0505727082490921,
      "learning_rate": 1.5119776224796823e-05,
      "loss": 0.04465341567993164,
      "step": 3938
    },
    {
      "epoch": 0.5325401787977625,
      "grad_norm": 0.12637023627758026,
      "learning_rate": 1.5112932002335195e-05,
      "loss": 0.061310768127441406,
      "step": 3939
    },
    {
      "epoch": 0.5326753755935985,
      "grad_norm": 0.07522363215684891,
      "learning_rate": 1.5106087756360524e-05,
      "loss": 0.06966352462768555,
      "step": 3940
    },
    {
      "epoch": 0.5328105723894344,
      "grad_norm": 0.10160709172487259,
      "learning_rate": 1.5099243488297816e-05,
      "loss": 0.0876927375793457,
      "step": 3941
    },
    {
      "epoch": 0.5329457691852703,
      "grad_norm": 0.08219432085752487,
      "learning_rate": 1.5092399199572083e-05,
      "loss": 0.0741729736328125,
      "step": 3942
    },
    {
      "epoch": 0.5330809659811062,
      "grad_norm": 0.045017704367637634,
      "learning_rate": 1.5085554891608343e-05,
      "loss": 0.04221534729003906,
      "step": 3943
    },
    {
      "epoch": 0.5332161627769422,
      "grad_norm": 0.07566311955451965,
      "learning_rate": 1.5078710565831616e-05,
      "loss": 0.07285451889038086,
      "step": 3944
    },
    {
      "epoch": 0.5333513595727781,
      "grad_norm": 0.04524838551878929,
      "learning_rate": 1.5071866223666935e-05,
      "loss": 0.03532552719116211,
      "step": 3945
    },
    {
      "epoch": 0.5334865563686141,
      "grad_norm": 0.06698434054851532,
      "learning_rate": 1.5065021866539323e-05,
      "loss": 0.05963277816772461,
      "step": 3946
    },
    {
      "epoch": 0.53362175316445,
      "grad_norm": 0.05195764824748039,
      "learning_rate": 1.5058177495873805e-05,
      "loss": 0.04464384913444519,
      "step": 3947
    },
    {
      "epoch": 0.5337569499602859,
      "grad_norm": 0.0745687261223793,
      "learning_rate": 1.5051333113095429e-05,
      "loss": 0.06804251670837402,
      "step": 3948
    },
    {
      "epoch": 0.5338921467561218,
      "grad_norm": 0.045073360204696655,
      "learning_rate": 1.5044488719629218e-05,
      "loss": 0.04778623580932617,
      "step": 3949
    },
    {
      "epoch": 0.5340273435519578,
      "grad_norm": 0.04042297601699829,
      "learning_rate": 1.5037644316900227e-05,
      "loss": 0.05254650115966797,
      "step": 3950
    },
    {
      "epoch": 0.5341625403477938,
      "grad_norm": 0.07153289765119553,
      "learning_rate": 1.5030799906333484e-05,
      "loss": 0.07042825222015381,
      "step": 3951
    },
    {
      "epoch": 0.5342977371436297,
      "grad_norm": 0.10275546461343765,
      "learning_rate": 1.5023955489354031e-05,
      "loss": 0.05800127983093262,
      "step": 3952
    },
    {
      "epoch": 0.5344329339394657,
      "grad_norm": 0.08324237167835236,
      "learning_rate": 1.5017111067386927e-05,
      "loss": 0.07979631423950195,
      "step": 3953
    },
    {
      "epoch": 0.5345681307353016,
      "grad_norm": 0.041641414165496826,
      "learning_rate": 1.50102666418572e-05,
      "loss": 0.039324164390563965,
      "step": 3954
    },
    {
      "epoch": 0.5347033275311375,
      "grad_norm": 0.10182870179414749,
      "learning_rate": 1.500342221418991e-05,
      "loss": 0.10242462158203125,
      "step": 3955
    },
    {
      "epoch": 0.5348385243269734,
      "grad_norm": 0.07581325620412827,
      "learning_rate": 1.4996577785810094e-05,
      "loss": 0.04960298538208008,
      "step": 3956
    },
    {
      "epoch": 0.5349737211228094,
      "grad_norm": 0.05332183092832565,
      "learning_rate": 1.4989733358142798e-05,
      "loss": 0.0499800443649292,
      "step": 3957
    },
    {
      "epoch": 0.5351089179186453,
      "grad_norm": 0.06088460981845856,
      "learning_rate": 1.498288893261308e-05,
      "loss": 0.05264925956726074,
      "step": 3958
    },
    {
      "epoch": 0.5352441147144813,
      "grad_norm": 0.089680977165699,
      "learning_rate": 1.497604451064597e-05,
      "loss": 0.052727460861206055,
      "step": 3959
    },
    {
      "epoch": 0.5353793115103173,
      "grad_norm": 0.06065080687403679,
      "learning_rate": 1.496920009366652e-05,
      "loss": 0.0589599609375,
      "step": 3960
    },
    {
      "epoch": 0.5355145083061531,
      "grad_norm": 0.03876560553908348,
      "learning_rate": 1.4962355683099777e-05,
      "loss": 0.038787841796875,
      "step": 3961
    },
    {
      "epoch": 0.5356497051019891,
      "grad_norm": 0.06510844081640244,
      "learning_rate": 1.4955511280370782e-05,
      "loss": 0.06097531318664551,
      "step": 3962
    },
    {
      "epoch": 0.535784901897825,
      "grad_norm": 0.11961193382740021,
      "learning_rate": 1.4948666886904579e-05,
      "loss": 0.08705329895019531,
      "step": 3963
    },
    {
      "epoch": 0.535920098693661,
      "grad_norm": 0.03729398921132088,
      "learning_rate": 1.4941822504126199e-05,
      "loss": 0.036171793937683105,
      "step": 3964
    },
    {
      "epoch": 0.5360552954894969,
      "grad_norm": 0.07781950384378433,
      "learning_rate": 1.4934978133460681e-05,
      "loss": 0.07162904739379883,
      "step": 3965
    },
    {
      "epoch": 0.5361904922853329,
      "grad_norm": 0.07768019288778305,
      "learning_rate": 1.4928133776333068e-05,
      "loss": 0.07037782669067383,
      "step": 3966
    },
    {
      "epoch": 0.5363256890811687,
      "grad_norm": 0.11429572105407715,
      "learning_rate": 1.4921289434168386e-05,
      "loss": 0.07644963264465332,
      "step": 3967
    },
    {
      "epoch": 0.5364608858770047,
      "grad_norm": 0.10944870859384537,
      "learning_rate": 1.4914445108391663e-05,
      "loss": 0.0532231330871582,
      "step": 3968
    },
    {
      "epoch": 0.5365960826728406,
      "grad_norm": 0.05272554233670235,
      "learning_rate": 1.4907600800427922e-05,
      "loss": 0.053615570068359375,
      "step": 3969
    },
    {
      "epoch": 0.5367312794686766,
      "grad_norm": 0.09222150593996048,
      "learning_rate": 1.4900756511702188e-05,
      "loss": 0.06598258018493652,
      "step": 3970
    },
    {
      "epoch": 0.5368664762645126,
      "grad_norm": 0.0834721103310585,
      "learning_rate": 1.4893912243639479e-05,
      "loss": 0.0625772476196289,
      "step": 3971
    },
    {
      "epoch": 0.5370016730603485,
      "grad_norm": 0.0897260308265686,
      "learning_rate": 1.4887067997664807e-05,
      "loss": 0.05955678969621658,
      "step": 3972
    },
    {
      "epoch": 0.5371368698561844,
      "grad_norm": 0.08779985457658768,
      "learning_rate": 1.488022377520318e-05,
      "loss": 0.09327077865600586,
      "step": 3973
    },
    {
      "epoch": 0.5372720666520203,
      "grad_norm": 0.09298571199178696,
      "learning_rate": 1.4873379577679599e-05,
      "loss": 0.06790244579315186,
      "step": 3974
    },
    {
      "epoch": 0.5374072634478563,
      "grad_norm": 0.05805501714348793,
      "learning_rate": 1.4866535406519063e-05,
      "loss": 0.06986933946609497,
      "step": 3975
    },
    {
      "epoch": 0.5375424602436922,
      "grad_norm": 0.056761037558317184,
      "learning_rate": 1.4859691263146574e-05,
      "loss": 0.05704069137573242,
      "step": 3976
    },
    {
      "epoch": 0.5376776570395282,
      "grad_norm": 0.08102252334356308,
      "learning_rate": 1.485284714898711e-05,
      "loss": 0.07594895362854004,
      "step": 3977
    },
    {
      "epoch": 0.5378128538353641,
      "grad_norm": 0.09888124465942383,
      "learning_rate": 1.4846003065465653e-05,
      "loss": 0.07868790626525879,
      "step": 3978
    },
    {
      "epoch": 0.5379480506312,
      "grad_norm": 0.12224207073450089,
      "learning_rate": 1.4839159014007179e-05,
      "loss": 0.08986568450927734,
      "step": 3979
    },
    {
      "epoch": 0.538083247427036,
      "grad_norm": 0.06923440843820572,
      "learning_rate": 1.4832314996036653e-05,
      "loss": 0.07303094863891602,
      "step": 3980
    },
    {
      "epoch": 0.5382184442228719,
      "grad_norm": 0.10873255133628845,
      "learning_rate": 1.4825471012979048e-05,
      "loss": 0.0881357192993164,
      "step": 3981
    },
    {
      "epoch": 0.5383536410187079,
      "grad_norm": 0.13190844655036926,
      "learning_rate": 1.4818627066259301e-05,
      "loss": 0.08095932006835938,
      "step": 3982
    },
    {
      "epoch": 0.5384888378145438,
      "grad_norm": 0.08680063486099243,
      "learning_rate": 1.481178315730237e-05,
      "loss": 0.07169675827026367,
      "step": 3983
    },
    {
      "epoch": 0.5386240346103798,
      "grad_norm": 0.13077495992183685,
      "learning_rate": 1.4804939287533184e-05,
      "loss": 0.06654715538024902,
      "step": 3984
    },
    {
      "epoch": 0.5387592314062156,
      "grad_norm": 0.09566811472177505,
      "learning_rate": 1.4798095458376682e-05,
      "loss": 0.0431288480758667,
      "step": 3985
    },
    {
      "epoch": 0.5388944282020516,
      "grad_norm": 0.08555074781179428,
      "learning_rate": 1.4791251671257788e-05,
      "loss": 0.05816817283630371,
      "step": 3986
    },
    {
      "epoch": 0.5390296249978875,
      "grad_norm": 0.09549107402563095,
      "learning_rate": 1.4784407927601401e-05,
      "loss": 0.0589756965637207,
      "step": 3987
    },
    {
      "epoch": 0.5391648217937235,
      "grad_norm": 0.06103307008743286,
      "learning_rate": 1.4777564228832436e-05,
      "loss": 0.058516621589660645,
      "step": 3988
    },
    {
      "epoch": 0.5393000185895594,
      "grad_norm": 0.08083301037549973,
      "learning_rate": 1.4770720576375782e-05,
      "loss": 0.06440114974975586,
      "step": 3989
    },
    {
      "epoch": 0.5394352153853954,
      "grad_norm": 0.16758571565151215,
      "learning_rate": 1.4763876971656334e-05,
      "loss": 0.08042621612548828,
      "step": 3990
    },
    {
      "epoch": 0.5395704121812313,
      "grad_norm": 0.04845372587442398,
      "learning_rate": 1.4757033416098953e-05,
      "loss": 0.04959309101104736,
      "step": 3991
    },
    {
      "epoch": 0.5397056089770672,
      "grad_norm": 0.11092467606067657,
      "learning_rate": 1.4750189911128511e-05,
      "loss": 0.09543991088867188,
      "step": 3992
    },
    {
      "epoch": 0.5398408057729032,
      "grad_norm": 0.07181750237941742,
      "learning_rate": 1.4743346458169863e-05,
      "loss": 0.0502493679523468,
      "step": 3993
    },
    {
      "epoch": 0.5399760025687391,
      "grad_norm": 0.07229512184858322,
      "learning_rate": 1.473650305864785e-05,
      "loss": 0.06245088577270508,
      "step": 3994
    },
    {
      "epoch": 0.5401111993645751,
      "grad_norm": 0.097012959420681,
      "learning_rate": 1.472965971398731e-05,
      "loss": 0.07404017448425293,
      "step": 3995
    },
    {
      "epoch": 0.540246396160411,
      "grad_norm": 0.06427924335002899,
      "learning_rate": 1.4722816425613054e-05,
      "loss": 0.049645185470581055,
      "step": 3996
    },
    {
      "epoch": 0.540381592956247,
      "grad_norm": 0.05618743970990181,
      "learning_rate": 1.4715973194949895e-05,
      "loss": 0.04946005344390869,
      "step": 3997
    },
    {
      "epoch": 0.5405167897520828,
      "grad_norm": 0.08045166730880737,
      "learning_rate": 1.4709130023422636e-05,
      "loss": 0.07694530487060547,
      "step": 3998
    },
    {
      "epoch": 0.5406519865479188,
      "grad_norm": 0.05502136051654816,
      "learning_rate": 1.4702286912456052e-05,
      "loss": 0.056183815002441406,
      "step": 3999
    },
    {
      "epoch": 0.5407871833437548,
      "grad_norm": 0.1206865906715393,
      "learning_rate": 1.4695443863474928e-05,
      "loss": 0.10046792030334473,
      "step": 4000
    },
    {
      "epoch": 0.5409223801395907,
      "grad_norm": 0.19072018563747406,
      "learning_rate": 1.4688600877904012e-05,
      "loss": 0.0827031135559082,
      "step": 4001
    },
    {
      "epoch": 0.5410575769354267,
      "grad_norm": 0.14215311408042908,
      "learning_rate": 1.468175795716805e-05,
      "loss": 0.07501697540283203,
      "step": 4002
    },
    {
      "epoch": 0.5411927737312626,
      "grad_norm": 0.06811357289552689,
      "learning_rate": 1.4674915102691783e-05,
      "loss": 0.05690455436706543,
      "step": 4003
    },
    {
      "epoch": 0.5413279705270985,
      "grad_norm": 0.1141452044248581,
      "learning_rate": 1.4668072315899926e-05,
      "loss": 0.09975820779800415,
      "step": 4004
    },
    {
      "epoch": 0.5414631673229344,
      "grad_norm": 0.07782881706953049,
      "learning_rate": 1.466122959821718e-05,
      "loss": 0.09650778770446777,
      "step": 4005
    },
    {
      "epoch": 0.5415983641187704,
      "grad_norm": 0.05322935804724693,
      "learning_rate": 1.4654386951068239e-05,
      "loss": 0.047977179288864136,
      "step": 4006
    },
    {
      "epoch": 0.5417335609146063,
      "grad_norm": 0.07051298767328262,
      "learning_rate": 1.4647544375877776e-05,
      "loss": 0.03043365478515625,
      "step": 4007
    },
    {
      "epoch": 0.5418687577104423,
      "grad_norm": 0.08823000639677048,
      "learning_rate": 1.4640701874070457e-05,
      "loss": 0.07546186447143555,
      "step": 4008
    },
    {
      "epoch": 0.5420039545062783,
      "grad_norm": 0.08165750652551651,
      "learning_rate": 1.4633859447070922e-05,
      "loss": 0.06728363037109375,
      "step": 4009
    },
    {
      "epoch": 0.5421391513021141,
      "grad_norm": 0.07753219455480576,
      "learning_rate": 1.4627017096303805e-05,
      "loss": 0.06643152236938477,
      "step": 4010
    },
    {
      "epoch": 0.54227434809795,
      "grad_norm": 0.14924810826778412,
      "learning_rate": 1.4620174823193711e-05,
      "loss": 0.07922667264938354,
      "step": 4011
    },
    {
      "epoch": 0.542409544893786,
      "grad_norm": 0.11316506564617157,
      "learning_rate": 1.4613332629165249e-05,
      "loss": 0.08025646209716797,
      "step": 4012
    },
    {
      "epoch": 0.542544741689622,
      "grad_norm": 0.07670755684375763,
      "learning_rate": 1.4606490515642998e-05,
      "loss": 0.07195925712585449,
      "step": 4013
    },
    {
      "epoch": 0.5426799384854579,
      "grad_norm": 0.09826210886240005,
      "learning_rate": 1.4599648484051516e-05,
      "loss": 0.09030818939208984,
      "step": 4014
    },
    {
      "epoch": 0.5428151352812939,
      "grad_norm": 0.05562635138630867,
      "learning_rate": 1.4592806535815358e-05,
      "loss": 0.052507877349853516,
      "step": 4015
    },
    {
      "epoch": 0.5429503320771297,
      "grad_norm": 0.06538919359445572,
      "learning_rate": 1.4585964672359045e-05,
      "loss": 0.06469321250915527,
      "step": 4016
    },
    {
      "epoch": 0.5430855288729657,
      "grad_norm": 0.06262054294347763,
      "learning_rate": 1.4579122895107098e-05,
      "loss": 0.06033515930175781,
      "step": 4017
    },
    {
      "epoch": 0.5432207256688016,
      "grad_norm": 0.12654909491539001,
      "learning_rate": 1.4572281205484012e-05,
      "loss": 0.05745267868041992,
      "step": 4018
    },
    {
      "epoch": 0.5433559224646376,
      "grad_norm": 0.0647176206111908,
      "learning_rate": 1.4565439604914256e-05,
      "loss": 0.060677528381347656,
      "step": 4019
    },
    {
      "epoch": 0.5434911192604736,
      "grad_norm": 0.09649985283613205,
      "learning_rate": 1.4558598094822294e-05,
      "loss": 0.06522369384765625,
      "step": 4020
    },
    {
      "epoch": 0.5436263160563095,
      "grad_norm": 0.14743003249168396,
      "learning_rate": 1.455175667663256e-05,
      "loss": 0.08967161178588867,
      "step": 4021
    },
    {
      "epoch": 0.5437615128521454,
      "grad_norm": 0.0617133192718029,
      "learning_rate": 1.4544915351769476e-05,
      "loss": 0.045554161071777344,
      "step": 4022
    },
    {
      "epoch": 0.5438967096479813,
      "grad_norm": 0.19207578897476196,
      "learning_rate": 1.4538074121657448e-05,
      "loss": 0.07785415649414062,
      "step": 4023
    },
    {
      "epoch": 0.5440319064438173,
      "grad_norm": 0.06239357218146324,
      "learning_rate": 1.4531232987720846e-05,
      "loss": 0.06808662414550781,
      "step": 4024
    },
    {
      "epoch": 0.5441671032396532,
      "grad_norm": 0.10362675786018372,
      "learning_rate": 1.4524391951384037e-05,
      "loss": 0.0351710319519043,
      "step": 4025
    },
    {
      "epoch": 0.5443023000354892,
      "grad_norm": 0.09004382789134979,
      "learning_rate": 1.4517551014071358e-05,
      "loss": 0.061960697174072266,
      "step": 4026
    },
    {
      "epoch": 0.5444374968313251,
      "grad_norm": 0.1107117235660553,
      "learning_rate": 1.4510710177207137e-05,
      "loss": 0.04803276062011719,
      "step": 4027
    },
    {
      "epoch": 0.544572693627161,
      "grad_norm": 0.1254362165927887,
      "learning_rate": 1.450386944221566e-05,
      "loss": 0.06754922866821289,
      "step": 4028
    },
    {
      "epoch": 0.5447078904229969,
      "grad_norm": 0.1771830916404724,
      "learning_rate": 1.449702881052121e-05,
      "loss": 0.08376109600067139,
      "step": 4029
    },
    {
      "epoch": 0.5448430872188329,
      "grad_norm": 0.18193966150283813,
      "learning_rate": 1.4490188283548048e-05,
      "loss": 0.08131217956542969,
      "step": 4030
    },
    {
      "epoch": 0.5449782840146689,
      "grad_norm": 0.14530053734779358,
      "learning_rate": 1.44833478627204e-05,
      "loss": 0.07862281799316406,
      "step": 4031
    },
    {
      "epoch": 0.5451134808105048,
      "grad_norm": 0.0912572368979454,
      "learning_rate": 1.447650754946249e-05,
      "loss": 0.03206348419189453,
      "step": 4032
    },
    {
      "epoch": 0.5452486776063408,
      "grad_norm": 0.15424300730228424,
      "learning_rate": 1.4469667345198492e-05,
      "loss": 0.05220985412597656,
      "step": 4033
    },
    {
      "epoch": 0.5453838744021766,
      "grad_norm": 0.09253767877817154,
      "learning_rate": 1.446282725135258e-05,
      "loss": 0.08451128005981445,
      "step": 4034
    },
    {
      "epoch": 0.5455190711980126,
      "grad_norm": 0.298917680978775,
      "learning_rate": 1.4455987269348904e-05,
      "loss": 0.10218369960784912,
      "step": 4035
    },
    {
      "epoch": 0.5456542679938485,
      "grad_norm": 0.22219686210155487,
      "learning_rate": 1.4449147400611578e-05,
      "loss": 0.09082603454589844,
      "step": 4036
    },
    {
      "epoch": 0.5457894647896845,
      "grad_norm": 0.11093132197856903,
      "learning_rate": 1.4442307646564702e-05,
      "loss": 0.06393581628799438,
      "step": 4037
    },
    {
      "epoch": 0.5459246615855204,
      "grad_norm": 0.134236678481102,
      "learning_rate": 1.4435468008632345e-05,
      "loss": 0.0842595100402832,
      "step": 4038
    },
    {
      "epoch": 0.5460598583813564,
      "grad_norm": 0.07306552678346634,
      "learning_rate": 1.4428628488238557e-05,
      "loss": 0.06939458847045898,
      "step": 4039
    },
    {
      "epoch": 0.5461950551771922,
      "grad_norm": 0.14223705232143402,
      "learning_rate": 1.442178908680737e-05,
      "loss": 0.05612373352050781,
      "step": 4040
    },
    {
      "epoch": 0.5463302519730282,
      "grad_norm": 0.15557502210140228,
      "learning_rate": 1.4414949805762779e-05,
      "loss": 0.07192802429199219,
      "step": 4041
    },
    {
      "epoch": 0.5464654487688642,
      "grad_norm": 0.06720432639122009,
      "learning_rate": 1.4408110646528757e-05,
      "loss": 0.08239865303039551,
      "step": 4042
    },
    {
      "epoch": 0.5466006455647001,
      "grad_norm": 0.0795508548617363,
      "learning_rate": 1.440127161052925e-05,
      "loss": 0.07013583183288574,
      "step": 4043
    },
    {
      "epoch": 0.5467358423605361,
      "grad_norm": 0.07922715693712234,
      "learning_rate": 1.4394432699188188e-05,
      "loss": 0.053696632385253906,
      "step": 4044
    },
    {
      "epoch": 0.546871039156372,
      "grad_norm": 0.1482291966676712,
      "learning_rate": 1.4387593913929472e-05,
      "loss": 0.07376480102539062,
      "step": 4045
    },
    {
      "epoch": 0.547006235952208,
      "grad_norm": 0.12175357341766357,
      "learning_rate": 1.4380755256176968e-05,
      "loss": 0.07642602920532227,
      "step": 4046
    },
    {
      "epoch": 0.5471414327480438,
      "grad_norm": 0.08537746965885162,
      "learning_rate": 1.437391672735452e-05,
      "loss": 0.0711512565612793,
      "step": 4047
    },
    {
      "epoch": 0.5472766295438798,
      "grad_norm": 0.07718467712402344,
      "learning_rate": 1.4367078328885946e-05,
      "loss": 0.07671165466308594,
      "step": 4048
    },
    {
      "epoch": 0.5474118263397157,
      "grad_norm": 0.06549594551324844,
      "learning_rate": 1.4360240062195039e-05,
      "loss": 0.08379745483398438,
      "step": 4049
    },
    {
      "epoch": 0.5475470231355517,
      "grad_norm": 0.09232236444950104,
      "learning_rate": 1.435340192870557e-05,
      "loss": 0.05181431770324707,
      "step": 4050
    },
    {
      "epoch": 0.5476822199313877,
      "grad_norm": 0.15960656106472015,
      "learning_rate": 1.434656392984126e-05,
      "loss": 0.07012367248535156,
      "step": 4051
    },
    {
      "epoch": 0.5478174167272236,
      "grad_norm": 0.06931684911251068,
      "learning_rate": 1.4339726067025828e-05,
      "loss": 0.058058738708496094,
      "step": 4052
    },
    {
      "epoch": 0.5479526135230595,
      "grad_norm": 0.150718092918396,
      "learning_rate": 1.4332888341682947e-05,
      "loss": 0.061393022537231445,
      "step": 4053
    },
    {
      "epoch": 0.5480878103188954,
      "grad_norm": 0.08669254183769226,
      "learning_rate": 1.432605075523627e-05,
      "loss": 0.08437573909759521,
      "step": 4054
    },
    {
      "epoch": 0.5482230071147314,
      "grad_norm": 0.07061182707548141,
      "learning_rate": 1.4319213309109426e-05,
      "loss": 0.07586526870727539,
      "step": 4055
    },
    {
      "epoch": 0.5483582039105673,
      "grad_norm": 0.06959249079227448,
      "learning_rate": 1.4312376004725996e-05,
      "loss": 0.05583477020263672,
      "step": 4056
    },
    {
      "epoch": 0.5484934007064033,
      "grad_norm": 0.059173211455345154,
      "learning_rate": 1.430553884350955e-05,
      "loss": 0.050989627838134766,
      "step": 4057
    },
    {
      "epoch": 0.5486285975022392,
      "grad_norm": 0.0690905749797821,
      "learning_rate": 1.429870182688362e-05,
      "loss": 0.03889751434326172,
      "step": 4058
    },
    {
      "epoch": 0.5487637942980751,
      "grad_norm": 0.11699472367763519,
      "learning_rate": 1.4291864956271713e-05,
      "loss": 0.07867813110351562,
      "step": 4059
    },
    {
      "epoch": 0.548898991093911,
      "grad_norm": 0.05598118528723717,
      "learning_rate": 1.4285028233097293e-05,
      "loss": 0.0629434585571289,
      "step": 4060
    },
    {
      "epoch": 0.549034187889747,
      "grad_norm": 0.05027598887681961,
      "learning_rate": 1.4278191658783809e-05,
      "loss": 0.04721999168395996,
      "step": 4061
    },
    {
      "epoch": 0.549169384685583,
      "grad_norm": 0.0698387622833252,
      "learning_rate": 1.427135523475467e-05,
      "loss": 0.06154298782348633,
      "step": 4062
    },
    {
      "epoch": 0.5493045814814189,
      "grad_norm": 0.18718264997005463,
      "learning_rate": 1.4264518962433258e-05,
      "loss": 0.07340455055236816,
      "step": 4063
    },
    {
      "epoch": 0.5494397782772549,
      "grad_norm": 0.10411976277828217,
      "learning_rate": 1.4257682843242925e-05,
      "loss": 0.0651845932006836,
      "step": 4064
    },
    {
      "epoch": 0.5495749750730907,
      "grad_norm": 0.06220874935388565,
      "learning_rate": 1.4250846878606974e-05,
      "loss": 0.0698246955871582,
      "step": 4065
    },
    {
      "epoch": 0.5497101718689267,
      "grad_norm": 0.0830516442656517,
      "learning_rate": 1.4244011069948702e-05,
      "loss": 0.06204843521118164,
      "step": 4066
    },
    {
      "epoch": 0.5498453686647626,
      "grad_norm": 0.03967805579304695,
      "learning_rate": 1.4237175418691357e-05,
      "loss": 0.04635214805603027,
      "step": 4067
    },
    {
      "epoch": 0.5499805654605986,
      "grad_norm": 0.060160860419273376,
      "learning_rate": 1.4230339926258153e-05,
      "loss": 0.0524139404296875,
      "step": 4068
    },
    {
      "epoch": 0.5501157622564345,
      "grad_norm": 0.12720409035682678,
      "learning_rate": 1.422350459407229e-05,
      "loss": 0.05837535858154297,
      "step": 4069
    },
    {
      "epoch": 0.5502509590522705,
      "grad_norm": 0.11871021240949631,
      "learning_rate": 1.4216669423556903e-05,
      "loss": 0.07346153259277344,
      "step": 4070
    },
    {
      "epoch": 0.5503861558481064,
      "grad_norm": 0.05642274394631386,
      "learning_rate": 1.420983441613512e-05,
      "loss": 0.054018616676330566,
      "step": 4071
    },
    {
      "epoch": 0.5505213526439423,
      "grad_norm": 0.09161009639501572,
      "learning_rate": 1.420299957323003e-05,
      "loss": 0.0765218734741211,
      "step": 4072
    },
    {
      "epoch": 0.5506565494397783,
      "grad_norm": 0.08621351420879364,
      "learning_rate": 1.4196164896264679e-05,
      "loss": 0.057952165603637695,
      "step": 4073
    },
    {
      "epoch": 0.5507917462356142,
      "grad_norm": 0.11336734890937805,
      "learning_rate": 1.418933038666208e-05,
      "loss": 0.04543304443359375,
      "step": 4074
    },
    {
      "epoch": 0.5509269430314502,
      "grad_norm": 0.09643787145614624,
      "learning_rate": 1.4182496045845217e-05,
      "loss": 0.07432794570922852,
      "step": 4075
    },
    {
      "epoch": 0.5510621398272861,
      "grad_norm": 0.06991898268461227,
      "learning_rate": 1.4175661875237036e-05,
      "loss": 0.05778098106384277,
      "step": 4076
    },
    {
      "epoch": 0.551197336623122,
      "grad_norm": 0.05274023860692978,
      "learning_rate": 1.416882787626045e-05,
      "loss": 0.030061841011047363,
      "step": 4077
    },
    {
      "epoch": 0.5513325334189579,
      "grad_norm": 0.10722766071557999,
      "learning_rate": 1.4161994050338334e-05,
      "loss": 0.0763922929763794,
      "step": 4078
    },
    {
      "epoch": 0.5514677302147939,
      "grad_norm": 0.07351508736610413,
      "learning_rate": 1.4155160398893528e-05,
      "loss": 0.07125234603881836,
      "step": 4079
    },
    {
      "epoch": 0.5516029270106299,
      "grad_norm": 0.12987740337848663,
      "learning_rate": 1.4148326923348824e-05,
      "loss": 0.0997319221496582,
      "step": 4080
    },
    {
      "epoch": 0.5517381238064658,
      "grad_norm": 0.07257214933633804,
      "learning_rate": 1.4141493625127e-05,
      "loss": 0.058835387229919434,
      "step": 4081
    },
    {
      "epoch": 0.5518733206023018,
      "grad_norm": 0.10118108987808228,
      "learning_rate": 1.4134660505650786e-05,
      "loss": 0.03758549690246582,
      "step": 4082
    },
    {
      "epoch": 0.5520085173981376,
      "grad_norm": 0.06019473075866699,
      "learning_rate": 1.4127827566342864e-05,
      "loss": 0.04579758644104004,
      "step": 4083
    },
    {
      "epoch": 0.5521437141939736,
      "grad_norm": 0.1526159793138504,
      "learning_rate": 1.4120994808625896e-05,
      "loss": 0.08717477321624756,
      "step": 4084
    },
    {
      "epoch": 0.5522789109898095,
      "grad_norm": 0.09764004498720169,
      "learning_rate": 1.4114162233922494e-05,
      "loss": 0.0828394889831543,
      "step": 4085
    },
    {
      "epoch": 0.5524141077856455,
      "grad_norm": 0.11772321164608002,
      "learning_rate": 1.4107329843655238e-05,
      "loss": 0.08561849594116211,
      "step": 4086
    },
    {
      "epoch": 0.5525493045814814,
      "grad_norm": 0.17571792006492615,
      "learning_rate": 1.4100497639246675e-05,
      "loss": 0.07077193260192871,
      "step": 4087
    },
    {
      "epoch": 0.5526845013773174,
      "grad_norm": 0.05513649433851242,
      "learning_rate": 1.4093665622119294e-05,
      "loss": 0.04854846000671387,
      "step": 4088
    },
    {
      "epoch": 0.5528196981731534,
      "grad_norm": 0.12004321068525314,
      "learning_rate": 1.4086833793695566e-05,
      "loss": 0.06998515129089355,
      "step": 4089
    },
    {
      "epoch": 0.5529548949689892,
      "grad_norm": 0.09957384318113327,
      "learning_rate": 1.408000215539791e-05,
      "loss": 0.07066822052001953,
      "step": 4090
    },
    {
      "epoch": 0.5530900917648252,
      "grad_norm": 0.08457494527101517,
      "learning_rate": 1.4073170708648711e-05,
      "loss": 0.06726408004760742,
      "step": 4091
    },
    {
      "epoch": 0.5532252885606611,
      "grad_norm": 0.05638938397169113,
      "learning_rate": 1.406633945487032e-05,
      "loss": 0.05702662467956543,
      "step": 4092
    },
    {
      "epoch": 0.5533604853564971,
      "grad_norm": 0.0509008951485157,
      "learning_rate": 1.4059508395485026e-05,
      "loss": 0.05689096450805664,
      "step": 4093
    },
    {
      "epoch": 0.553495682152333,
      "grad_norm": 0.06192107871174812,
      "learning_rate": 1.4052677531915102e-05,
      "loss": 0.059439241886138916,
      "step": 4094
    },
    {
      "epoch": 0.553630878948169,
      "grad_norm": 0.08555155992507935,
      "learning_rate": 1.4045846865582765e-05,
      "loss": 0.03972148895263672,
      "step": 4095
    },
    {
      "epoch": 0.5537660757440048,
      "grad_norm": 0.12039042264223099,
      "learning_rate": 1.4039016397910206e-05,
      "loss": 0.059500694274902344,
      "step": 4096
    },
    {
      "epoch": 0.5539012725398408,
      "grad_norm": 0.19533653557300568,
      "learning_rate": 1.403218613031955e-05,
      "loss": 0.07166004180908203,
      "step": 4097
    },
    {
      "epoch": 0.5540364693356767,
      "grad_norm": 0.0606880709528923,
      "learning_rate": 1.4025356064232903e-05,
      "loss": 0.05276799201965332,
      "step": 4098
    },
    {
      "epoch": 0.5541716661315127,
      "grad_norm": 0.05337110534310341,
      "learning_rate": 1.4018526201072324e-05,
      "loss": 0.049727439880371094,
      "step": 4099
    },
    {
      "epoch": 0.5543068629273487,
      "grad_norm": 0.08678770810365677,
      "learning_rate": 1.4011696542259821e-05,
      "loss": 0.08035659790039062,
      "step": 4100
    },
    {
      "epoch": 0.5544420597231846,
      "grad_norm": 0.04927872493863106,
      "learning_rate": 1.4004867089217376e-05,
      "loss": 0.04891204833984375,
      "step": 4101
    },
    {
      "epoch": 0.5545772565190205,
      "grad_norm": 0.0675642341375351,
      "learning_rate": 1.39980378433669e-05,
      "loss": 0.07493162155151367,
      "step": 4102
    },
    {
      "epoch": 0.5547124533148564,
      "grad_norm": 0.04178496077656746,
      "learning_rate": 1.399120880613029e-05,
      "loss": 0.05071115493774414,
      "step": 4103
    },
    {
      "epoch": 0.5548476501106924,
      "grad_norm": 0.09921976178884506,
      "learning_rate": 1.3984379978929388e-05,
      "loss": 0.09384346008300781,
      "step": 4104
    },
    {
      "epoch": 0.5549828469065283,
      "grad_norm": 0.1152753159403801,
      "learning_rate": 1.3977551363185995e-05,
      "loss": 0.059639692306518555,
      "step": 4105
    },
    {
      "epoch": 0.5551180437023643,
      "grad_norm": 0.18421974778175354,
      "learning_rate": 1.3970722960321854e-05,
      "loss": 0.09104394912719727,
      "step": 4106
    },
    {
      "epoch": 0.5552532404982002,
      "grad_norm": 0.1050596535205841,
      "learning_rate": 1.3963894771758682e-05,
      "loss": 0.06557977199554443,
      "step": 4107
    },
    {
      "epoch": 0.5553884372940361,
      "grad_norm": 0.049730006605386734,
      "learning_rate": 1.3957066798918143e-05,
      "loss": 0.0556025505065918,
      "step": 4108
    },
    {
      "epoch": 0.555523634089872,
      "grad_norm": 0.058101147413253784,
      "learning_rate": 1.3950239043221861e-05,
      "loss": 0.06447029113769531,
      "step": 4109
    },
    {
      "epoch": 0.555658830885708,
      "grad_norm": 0.15102678537368774,
      "learning_rate": 1.3943411506091408e-05,
      "loss": 0.09522128105163574,
      "step": 4110
    },
    {
      "epoch": 0.555794027681544,
      "grad_norm": 0.08406145125627518,
      "learning_rate": 1.3936584188948313e-05,
      "loss": 0.08793139457702637,
      "step": 4111
    },
    {
      "epoch": 0.5559292244773799,
      "grad_norm": 0.15618664026260376,
      "learning_rate": 1.3929757093214059e-05,
      "loss": 0.07349300384521484,
      "step": 4112
    },
    {
      "epoch": 0.5560644212732159,
      "grad_norm": 0.09589562565088272,
      "learning_rate": 1.3922930220310085e-05,
      "loss": 0.07369613647460938,
      "step": 4113
    },
    {
      "epoch": 0.5561996180690517,
      "grad_norm": 0.11526025086641312,
      "learning_rate": 1.3916103571657786e-05,
      "loss": 0.06490325927734375,
      "step": 4114
    },
    {
      "epoch": 0.5563348148648877,
      "grad_norm": 0.06394743174314499,
      "learning_rate": 1.3909277148678504e-05,
      "loss": 0.05130791664123535,
      "step": 4115
    },
    {
      "epoch": 0.5564700116607236,
      "grad_norm": 0.08251817524433136,
      "learning_rate": 1.3902450952793536e-05,
      "loss": 0.04411637783050537,
      "step": 4116
    },
    {
      "epoch": 0.5566052084565596,
      "grad_norm": 0.09122715145349503,
      "learning_rate": 1.389562498542413e-05,
      "loss": 0.06601238250732422,
      "step": 4117
    },
    {
      "epoch": 0.5567404052523955,
      "grad_norm": 0.07016460597515106,
      "learning_rate": 1.388879924799149e-05,
      "loss": 0.04880332946777344,
      "step": 4118
    },
    {
      "epoch": 0.5568756020482315,
      "grad_norm": 0.0685984194278717,
      "learning_rate": 1.388197374191678e-05,
      "loss": 0.06895732879638672,
      "step": 4119
    },
    {
      "epoch": 0.5570107988440673,
      "grad_norm": 0.060960136353969574,
      "learning_rate": 1.387514846862109e-05,
      "loss": 0.07156848907470703,
      "step": 4120
    },
    {
      "epoch": 0.5571459956399033,
      "grad_norm": 0.0268340315669775,
      "learning_rate": 1.3868323429525492e-05,
      "loss": 0.03031754493713379,
      "step": 4121
    },
    {
      "epoch": 0.5572811924357393,
      "grad_norm": 0.06605499237775803,
      "learning_rate": 1.3861498626050986e-05,
      "loss": 0.08360886573791504,
      "step": 4122
    },
    {
      "epoch": 0.5574163892315752,
      "grad_norm": 0.0721614882349968,
      "learning_rate": 1.385467405961854e-05,
      "loss": 0.06602191925048828,
      "step": 4123
    },
    {
      "epoch": 0.5575515860274112,
      "grad_norm": 0.13270018994808197,
      "learning_rate": 1.3847849731649066e-05,
      "loss": 0.08553099632263184,
      "step": 4124
    },
    {
      "epoch": 0.5576867828232471,
      "grad_norm": 0.10970934480428696,
      "learning_rate": 1.3841025643563418e-05,
      "loss": 0.07076334953308105,
      "step": 4125
    },
    {
      "epoch": 0.557821979619083,
      "grad_norm": 0.09613852947950363,
      "learning_rate": 1.3834201796782413e-05,
      "loss": 0.07680797576904297,
      "step": 4126
    },
    {
      "epoch": 0.5579571764149189,
      "grad_norm": 0.05349888280034065,
      "learning_rate": 1.3827378192726808e-05,
      "loss": 0.056195735931396484,
      "step": 4127
    },
    {
      "epoch": 0.5580923732107549,
      "grad_norm": 0.06799422204494476,
      "learning_rate": 1.3820554832817324e-05,
      "loss": 0.05526018142700195,
      "step": 4128
    },
    {
      "epoch": 0.5582275700065908,
      "grad_norm": 0.07790925353765488,
      "learning_rate": 1.3813731718474606e-05,
      "loss": 0.05934453010559082,
      "step": 4129
    },
    {
      "epoch": 0.5583627668024268,
      "grad_norm": 0.09974254667758942,
      "learning_rate": 1.380690885111927e-05,
      "loss": 0.059127092361450195,
      "step": 4130
    },
    {
      "epoch": 0.5584979635982628,
      "grad_norm": 0.1049327552318573,
      "learning_rate": 1.3800086232171877e-05,
      "loss": 0.05780649185180664,
      "step": 4131
    },
    {
      "epoch": 0.5586331603940986,
      "grad_norm": 0.2112324982881546,
      "learning_rate": 1.3793263863052926e-05,
      "loss": 0.10281717777252197,
      "step": 4132
    },
    {
      "epoch": 0.5587683571899346,
      "grad_norm": 0.1996569037437439,
      "learning_rate": 1.3786441745182881e-05,
      "loss": 0.08563566207885742,
      "step": 4133
    },
    {
      "epoch": 0.5589035539857705,
      "grad_norm": 0.07548871636390686,
      "learning_rate": 1.3779619879982127e-05,
      "loss": 0.054668426513671875,
      "step": 4134
    },
    {
      "epoch": 0.5590387507816065,
      "grad_norm": 0.09903180599212646,
      "learning_rate": 1.3772798268871025e-05,
      "loss": 0.05505460500717163,
      "step": 4135
    },
    {
      "epoch": 0.5591739475774424,
      "grad_norm": 0.06804535537958145,
      "learning_rate": 1.376597691326987e-05,
      "loss": 0.06219196319580078,
      "step": 4136
    },
    {
      "epoch": 0.5593091443732784,
      "grad_norm": 0.085710808634758,
      "learning_rate": 1.3759155814598898e-05,
      "loss": 0.052351951599121094,
      "step": 4137
    },
    {
      "epoch": 0.5594443411691143,
      "grad_norm": 0.0954032838344574,
      "learning_rate": 1.3752334974278308e-05,
      "loss": 0.056681811809539795,
      "step": 4138
    },
    {
      "epoch": 0.5595795379649502,
      "grad_norm": 0.09635797888040543,
      "learning_rate": 1.3745514393728225e-05,
      "loss": 0.081623375415802,
      "step": 4139
    },
    {
      "epoch": 0.5597147347607861,
      "grad_norm": 0.12712836265563965,
      "learning_rate": 1.3738694074368735e-05,
      "loss": 0.05469036102294922,
      "step": 4140
    },
    {
      "epoch": 0.5598499315566221,
      "grad_norm": 0.045935600996017456,
      "learning_rate": 1.3731874017619868e-05,
      "loss": 0.05090534687042236,
      "step": 4141
    },
    {
      "epoch": 0.5599851283524581,
      "grad_norm": 0.08144697546958923,
      "learning_rate": 1.3725054224901597e-05,
      "loss": 0.06747055053710938,
      "step": 4142
    },
    {
      "epoch": 0.560120325148294,
      "grad_norm": 0.11648909002542496,
      "learning_rate": 1.3718234697633826e-05,
      "loss": 0.05892801284790039,
      "step": 4143
    },
    {
      "epoch": 0.56025552194413,
      "grad_norm": 0.03973710536956787,
      "learning_rate": 1.3711415437236427e-05,
      "loss": 0.04181087017059326,
      "step": 4144
    },
    {
      "epoch": 0.5603907187399658,
      "grad_norm": 0.09953191876411438,
      "learning_rate": 1.3704596445129207e-05,
      "loss": 0.09266901016235352,
      "step": 4145
    },
    {
      "epoch": 0.5605259155358018,
      "grad_norm": 0.06498900800943375,
      "learning_rate": 1.369777772273192e-05,
      "loss": 0.06208086013793945,
      "step": 4146
    },
    {
      "epoch": 0.5606611123316377,
      "grad_norm": 0.162736177444458,
      "learning_rate": 1.369095927146425e-05,
      "loss": 0.07050085067749023,
      "step": 4147
    },
    {
      "epoch": 0.5607963091274737,
      "grad_norm": 0.11763712763786316,
      "learning_rate": 1.3684141092745846e-05,
      "loss": 0.09427165985107422,
      "step": 4148
    },
    {
      "epoch": 0.5609315059233096,
      "grad_norm": 0.0428580567240715,
      "learning_rate": 1.3677323187996276e-05,
      "loss": 0.03928828239440918,
      "step": 4149
    },
    {
      "epoch": 0.5610667027191456,
      "grad_norm": 0.08018385618925095,
      "learning_rate": 1.3670505558635074e-05,
      "loss": 0.061930060386657715,
      "step": 4150
    },
    {
      "epoch": 0.5612018995149815,
      "grad_norm": 0.11412448436021805,
      "learning_rate": 1.366368820608171e-05,
      "loss": 0.07166659832000732,
      "step": 4151
    },
    {
      "epoch": 0.5613370963108174,
      "grad_norm": 0.04032596945762634,
      "learning_rate": 1.365687113175558e-05,
      "loss": 0.044728755950927734,
      "step": 4152
    },
    {
      "epoch": 0.5614722931066534,
      "grad_norm": 0.06026893109083176,
      "learning_rate": 1.3650054337076049e-05,
      "loss": 0.07131719589233398,
      "step": 4153
    },
    {
      "epoch": 0.5616074899024893,
      "grad_norm": 0.05437815561890602,
      "learning_rate": 1.3643237823462398e-05,
      "loss": 0.05131793022155762,
      "step": 4154
    },
    {
      "epoch": 0.5617426866983253,
      "grad_norm": 0.08465273678302765,
      "learning_rate": 1.363642159233387e-05,
      "loss": 0.07694077491760254,
      "step": 4155
    },
    {
      "epoch": 0.5618778834941612,
      "grad_norm": 0.09877727180719376,
      "learning_rate": 1.3629605645109642e-05,
      "loss": 0.05626237392425537,
      "step": 4156
    },
    {
      "epoch": 0.5620130802899971,
      "grad_norm": 0.10937295854091644,
      "learning_rate": 1.362278998320882e-05,
      "loss": 0.08494925498962402,
      "step": 4157
    },
    {
      "epoch": 0.562148277085833,
      "grad_norm": 0.16738057136535645,
      "learning_rate": 1.3615974608050472e-05,
      "loss": 0.08388376235961914,
      "step": 4158
    },
    {
      "epoch": 0.562283473881669,
      "grad_norm": 0.07141170650720596,
      "learning_rate": 1.3609159521053588e-05,
      "loss": 0.06803560256958008,
      "step": 4159
    },
    {
      "epoch": 0.562418670677505,
      "grad_norm": 0.06486458331346512,
      "learning_rate": 1.3602344723637107e-05,
      "loss": 0.040422677993774414,
      "step": 4160
    },
    {
      "epoch": 0.5625538674733409,
      "grad_norm": 0.09215779602527618,
      "learning_rate": 1.3595530217219916e-05,
      "loss": 0.06354379653930664,
      "step": 4161
    },
    {
      "epoch": 0.5626890642691769,
      "grad_norm": 0.12979668378829956,
      "learning_rate": 1.3588716003220815e-05,
      "loss": 0.08090019226074219,
      "step": 4162
    },
    {
      "epoch": 0.5628242610650127,
      "grad_norm": 0.06881259381771088,
      "learning_rate": 1.3581902083058574e-05,
      "loss": 0.07419204711914062,
      "step": 4163
    },
    {
      "epoch": 0.5629594578608487,
      "grad_norm": 0.08449291437864304,
      "learning_rate": 1.3575088458151877e-05,
      "loss": 0.0575709342956543,
      "step": 4164
    },
    {
      "epoch": 0.5630946546566846,
      "grad_norm": 0.16275553405284882,
      "learning_rate": 1.3568275129919367e-05,
      "loss": 0.06701159477233887,
      "step": 4165
    },
    {
      "epoch": 0.5632298514525206,
      "grad_norm": 0.07035147398710251,
      "learning_rate": 1.3561462099779604e-05,
      "loss": 0.06704211235046387,
      "step": 4166
    },
    {
      "epoch": 0.5633650482483565,
      "grad_norm": 0.11745652556419373,
      "learning_rate": 1.3554649369151104e-05,
      "loss": 0.06928348541259766,
      "step": 4167
    },
    {
      "epoch": 0.5635002450441925,
      "grad_norm": 0.16336382925510406,
      "learning_rate": 1.3547836939452315e-05,
      "loss": 0.05987215042114258,
      "step": 4168
    },
    {
      "epoch": 0.5636354418400283,
      "grad_norm": 0.043371912091970444,
      "learning_rate": 1.3541024812101615e-05,
      "loss": 0.04890775680541992,
      "step": 4169
    },
    {
      "epoch": 0.5637706386358643,
      "grad_norm": 0.04410286992788315,
      "learning_rate": 1.3534212988517339e-05,
      "loss": 0.04256725311279297,
      "step": 4170
    },
    {
      "epoch": 0.5639058354317003,
      "grad_norm": 0.11862429976463318,
      "learning_rate": 1.3527401470117726e-05,
      "loss": 0.07047128677368164,
      "step": 4171
    },
    {
      "epoch": 0.5640410322275362,
      "grad_norm": 0.15285752713680267,
      "learning_rate": 1.3520590258320981e-05,
      "loss": 0.06650954484939575,
      "step": 4172
    },
    {
      "epoch": 0.5641762290233722,
      "grad_norm": 0.05751129612326622,
      "learning_rate": 1.3513779354545235e-05,
      "loss": 0.036438584327697754,
      "step": 4173
    },
    {
      "epoch": 0.5643114258192081,
      "grad_norm": 0.0486147366464138,
      "learning_rate": 1.3506968760208557e-05,
      "loss": 0.04313361644744873,
      "step": 4174
    },
    {
      "epoch": 0.564446622615044,
      "grad_norm": 0.10829327255487442,
      "learning_rate": 1.3500158476728938e-05,
      "loss": 0.05553328990936279,
      "step": 4175
    },
    {
      "epoch": 0.5645818194108799,
      "grad_norm": 0.06074419245123863,
      "learning_rate": 1.3493348505524325e-05,
      "loss": 0.05710744857788086,
      "step": 4176
    },
    {
      "epoch": 0.5647170162067159,
      "grad_norm": 0.13050898909568787,
      "learning_rate": 1.3486538848012586e-05,
      "loss": 0.06959700584411621,
      "step": 4177
    },
    {
      "epoch": 0.5648522130025518,
      "grad_norm": 0.06630608439445496,
      "learning_rate": 1.3479729505611532e-05,
      "loss": 0.06532549858093262,
      "step": 4178
    },
    {
      "epoch": 0.5649874097983878,
      "grad_norm": 0.1411292403936386,
      "learning_rate": 1.3472920479738906e-05,
      "loss": 0.10419750213623047,
      "step": 4179
    },
    {
      "epoch": 0.5651226065942238,
      "grad_norm": 0.0672093853354454,
      "learning_rate": 1.346611177181237e-05,
      "loss": 0.06861841678619385,
      "step": 4180
    },
    {
      "epoch": 0.5652578033900597,
      "grad_norm": 0.10425559431314468,
      "learning_rate": 1.3459303383249547e-05,
      "loss": 0.09084463119506836,
      "step": 4181
    },
    {
      "epoch": 0.5653930001858956,
      "grad_norm": 0.04203963279724121,
      "learning_rate": 1.3452495315467975e-05,
      "loss": 0.04257559776306152,
      "step": 4182
    },
    {
      "epoch": 0.5655281969817315,
      "grad_norm": 0.11237596720457077,
      "learning_rate": 1.3445687569885132e-05,
      "loss": 0.05130720138549805,
      "step": 4183
    },
    {
      "epoch": 0.5656633937775675,
      "grad_norm": 0.11588342487812042,
      "learning_rate": 1.3438880147918429e-05,
      "loss": 0.06269454956054688,
      "step": 4184
    },
    {
      "epoch": 0.5657985905734034,
      "grad_norm": 0.05830025300383568,
      "learning_rate": 1.3432073050985201e-05,
      "loss": 0.055112600326538086,
      "step": 4185
    },
    {
      "epoch": 0.5659337873692394,
      "grad_norm": 0.09931790828704834,
      "learning_rate": 1.3425266280502721e-05,
      "loss": 0.05435037612915039,
      "step": 4186
    },
    {
      "epoch": 0.5660689841650753,
      "grad_norm": 0.12272310256958008,
      "learning_rate": 1.3418459837888202e-05,
      "loss": 0.05464887619018555,
      "step": 4187
    },
    {
      "epoch": 0.5662041809609112,
      "grad_norm": 0.07456616312265396,
      "learning_rate": 1.3411653724558784e-05,
      "loss": 0.06966352462768555,
      "step": 4188
    },
    {
      "epoch": 0.5663393777567471,
      "grad_norm": 0.0517207533121109,
      "learning_rate": 1.3404847941931523e-05,
      "loss": 0.05562591552734375,
      "step": 4189
    },
    {
      "epoch": 0.5664745745525831,
      "grad_norm": 0.050687264651060104,
      "learning_rate": 1.339804249142343e-05,
      "loss": 0.040265798568725586,
      "step": 4190
    },
    {
      "epoch": 0.5666097713484191,
      "grad_norm": 0.058872658759355545,
      "learning_rate": 1.3391237374451429e-05,
      "loss": 0.05400204658508301,
      "step": 4191
    },
    {
      "epoch": 0.566744968144255,
      "grad_norm": 0.08813443034887314,
      "learning_rate": 1.3384432592432388e-05,
      "loss": 0.07317638397216797,
      "step": 4192
    },
    {
      "epoch": 0.566880164940091,
      "grad_norm": 0.11647678166627884,
      "learning_rate": 1.3377628146783102e-05,
      "loss": 0.06634187698364258,
      "step": 4193
    },
    {
      "epoch": 0.5670153617359268,
      "grad_norm": 0.10847894847393036,
      "learning_rate": 1.3370824038920281e-05,
      "loss": 0.062330007553100586,
      "step": 4194
    },
    {
      "epoch": 0.5671505585317628,
      "grad_norm": 0.05624714866280556,
      "learning_rate": 1.3364020270260586e-05,
      "loss": 0.054793477058410645,
      "step": 4195
    },
    {
      "epoch": 0.5672857553275987,
      "grad_norm": 0.11303391307592392,
      "learning_rate": 1.335721684222059e-05,
      "loss": 0.07008886337280273,
      "step": 4196
    },
    {
      "epoch": 0.5674209521234347,
      "grad_norm": 0.042140375822782516,
      "learning_rate": 1.3350413756216816e-05,
      "loss": 0.053290605545043945,
      "step": 4197
    },
    {
      "epoch": 0.5675561489192706,
      "grad_norm": 0.05239458754658699,
      "learning_rate": 1.334361101366569e-05,
      "loss": 0.041712045669555664,
      "step": 4198
    },
    {
      "epoch": 0.5676913457151066,
      "grad_norm": 0.16714490950107574,
      "learning_rate": 1.3336808615983582e-05,
      "loss": 0.08261919021606445,
      "step": 4199
    },
    {
      "epoch": 0.5678265425109424,
      "grad_norm": 0.08035875111818314,
      "learning_rate": 1.3330006564586791e-05,
      "loss": 0.07848501205444336,
      "step": 4200
    },
    {
      "epoch": 0.5679617393067784,
      "grad_norm": 0.09957578033208847,
      "learning_rate": 1.3323204860891539e-05,
      "loss": 0.06833028793334961,
      "step": 4201
    },
    {
      "epoch": 0.5680969361026144,
      "grad_norm": 0.08059883117675781,
      "learning_rate": 1.3316403506313981e-05,
      "loss": 0.051741600036621094,
      "step": 4202
    },
    {
      "epoch": 0.5682321328984503,
      "grad_norm": 0.05721670016646385,
      "learning_rate": 1.3309602502270184e-05,
      "loss": 0.04900932312011719,
      "step": 4203
    },
    {
      "epoch": 0.5683673296942863,
      "grad_norm": 0.06999655067920685,
      "learning_rate": 1.3302801850176161e-05,
      "loss": 0.056043148040771484,
      "step": 4204
    },
    {
      "epoch": 0.5685025264901222,
      "grad_norm": 0.063369520008564,
      "learning_rate": 1.3296001551447848e-05,
      "loss": 0.07485222816467285,
      "step": 4205
    },
    {
      "epoch": 0.5686377232859581,
      "grad_norm": 0.06795758008956909,
      "learning_rate": 1.32892016075011e-05,
      "loss": 0.07515382766723633,
      "step": 4206
    },
    {
      "epoch": 0.568772920081794,
      "grad_norm": 0.22410255670547485,
      "learning_rate": 1.3282402019751694e-05,
      "loss": 0.07778263092041016,
      "step": 4207
    },
    {
      "epoch": 0.56890811687763,
      "grad_norm": 0.07383868098258972,
      "learning_rate": 1.327560278961535e-05,
      "loss": 0.05089735984802246,
      "step": 4208
    },
    {
      "epoch": 0.569043313673466,
      "grad_norm": 0.09812065213918686,
      "learning_rate": 1.3268803918507699e-05,
      "loss": 0.07952666282653809,
      "step": 4209
    },
    {
      "epoch": 0.5691785104693019,
      "grad_norm": 0.08895250409841537,
      "learning_rate": 1.3262005407844306e-05,
      "loss": 0.06500053405761719,
      "step": 4210
    },
    {
      "epoch": 0.5693137072651379,
      "grad_norm": 0.11505065858364105,
      "learning_rate": 1.325520725904066e-05,
      "loss": 0.0758066177368164,
      "step": 4211
    },
    {
      "epoch": 0.5694489040609737,
      "grad_norm": 0.08348774164915085,
      "learning_rate": 1.3248409473512158e-05,
      "loss": 0.05151009559631348,
      "step": 4212
    },
    {
      "epoch": 0.5695841008568097,
      "grad_norm": 0.07074927538633347,
      "learning_rate": 1.3241612052674146e-05,
      "loss": 0.05847978591918945,
      "step": 4213
    },
    {
      "epoch": 0.5697192976526456,
      "grad_norm": 0.08385651558637619,
      "learning_rate": 1.3234814997941883e-05,
      "loss": 0.05460691452026367,
      "step": 4214
    },
    {
      "epoch": 0.5698544944484816,
      "grad_norm": 0.11835520714521408,
      "learning_rate": 1.322801831073055e-05,
      "loss": 0.07225751876831055,
      "step": 4215
    },
    {
      "epoch": 0.5699896912443175,
      "grad_norm": 0.0451829731464386,
      "learning_rate": 1.322122199245526e-05,
      "loss": 0.04921305179595947,
      "step": 4216
    },
    {
      "epoch": 0.5701248880401535,
      "grad_norm": 0.09435160458087921,
      "learning_rate": 1.321442604453103e-05,
      "loss": 0.0588705837726593,
      "step": 4217
    },
    {
      "epoch": 0.5702600848359893,
      "grad_norm": 0.11451330780982971,
      "learning_rate": 1.320763046837282e-05,
      "loss": 0.0531536340713501,
      "step": 4218
    },
    {
      "epoch": 0.5703952816318253,
      "grad_norm": 0.1146269217133522,
      "learning_rate": 1.3200835265395504e-05,
      "loss": 0.08288764953613281,
      "step": 4219
    },
    {
      "epoch": 0.5705304784276612,
      "grad_norm": 0.045473698526620865,
      "learning_rate": 1.3194040437013885e-05,
      "loss": 0.04354667663574219,
      "step": 4220
    },
    {
      "epoch": 0.5706656752234972,
      "grad_norm": 0.10906580090522766,
      "learning_rate": 1.3187245984642673e-05,
      "loss": 0.07678627967834473,
      "step": 4221
    },
    {
      "epoch": 0.5708008720193332,
      "grad_norm": 0.04136877506971359,
      "learning_rate": 1.3180451909696517e-05,
      "loss": 0.045319318771362305,
      "step": 4222
    },
    {
      "epoch": 0.5709360688151691,
      "grad_norm": 0.1619584858417511,
      "learning_rate": 1.3173658213589972e-05,
      "loss": 0.10235190391540527,
      "step": 4223
    },
    {
      "epoch": 0.5710712656110051,
      "grad_norm": 0.04617733135819435,
      "learning_rate": 1.3166864897737526e-05,
      "loss": 0.043419480323791504,
      "step": 4224
    },
    {
      "epoch": 0.5712064624068409,
      "grad_norm": 0.056700531393289566,
      "learning_rate": 1.3160071963553593e-05,
      "loss": 0.04490089416503906,
      "step": 4225
    },
    {
      "epoch": 0.5713416592026769,
      "grad_norm": 0.07429137825965881,
      "learning_rate": 1.315327941245248e-05,
      "loss": 0.08240365982055664,
      "step": 4226
    },
    {
      "epoch": 0.5714768559985128,
      "grad_norm": 0.09860310703516006,
      "learning_rate": 1.3146487245848445e-05,
      "loss": 0.05005502700805664,
      "step": 4227
    },
    {
      "epoch": 0.5716120527943488,
      "grad_norm": 0.06659059226512909,
      "learning_rate": 1.3139695465155645e-05,
      "loss": 0.04583883285522461,
      "step": 4228
    },
    {
      "epoch": 0.5717472495901847,
      "grad_norm": 0.09478210657835007,
      "learning_rate": 1.3132904071788177e-05,
      "loss": 0.0897223949432373,
      "step": 4229
    },
    {
      "epoch": 0.5718824463860207,
      "grad_norm": 0.07268854975700378,
      "learning_rate": 1.3126113067160031e-05,
      "loss": 0.052277326583862305,
      "step": 4230
    },
    {
      "epoch": 0.5720176431818565,
      "grad_norm": 0.09079897403717041,
      "learning_rate": 1.3119322452685139e-05,
      "loss": 0.051481008529663086,
      "step": 4231
    },
    {
      "epoch": 0.5721528399776925,
      "grad_norm": 0.08313412219285965,
      "learning_rate": 1.3112532229777344e-05,
      "loss": 0.06950795650482178,
      "step": 4232
    },
    {
      "epoch": 0.5722880367735285,
      "grad_norm": 0.12661270797252655,
      "learning_rate": 1.3105742399850399e-05,
      "loss": 0.05651652812957764,
      "step": 4233
    },
    {
      "epoch": 0.5724232335693644,
      "grad_norm": 0.1941634565591812,
      "learning_rate": 1.3098952964317996e-05,
      "loss": 0.07221412658691406,
      "step": 4234
    },
    {
      "epoch": 0.5725584303652004,
      "grad_norm": 0.05457120016217232,
      "learning_rate": 1.3092163924593717e-05,
      "loss": 0.07674026489257812,
      "step": 4235
    },
    {
      "epoch": 0.5726936271610363,
      "grad_norm": 0.1400396078824997,
      "learning_rate": 1.308537528209108e-05,
      "loss": 0.0699148178100586,
      "step": 4236
    },
    {
      "epoch": 0.5728288239568722,
      "grad_norm": 0.07919779419898987,
      "learning_rate": 1.3078587038223525e-05,
      "loss": 0.07209181785583496,
      "step": 4237
    },
    {
      "epoch": 0.5729640207527081,
      "grad_norm": 0.07076119631528854,
      "learning_rate": 1.3071799194404392e-05,
      "loss": 0.04361701011657715,
      "step": 4238
    },
    {
      "epoch": 0.5730992175485441,
      "grad_norm": 0.05911535769701004,
      "learning_rate": 1.3065011752046955e-05,
      "loss": 0.0661616325378418,
      "step": 4239
    },
    {
      "epoch": 0.57323441434438,
      "grad_norm": 0.11360391229391098,
      "learning_rate": 1.3058224712564382e-05,
      "loss": 0.10242629051208496,
      "step": 4240
    },
    {
      "epoch": 0.573369611140216,
      "grad_norm": 0.19428549706935883,
      "learning_rate": 1.305143807736978e-05,
      "loss": 0.08803272247314453,
      "step": 4241
    },
    {
      "epoch": 0.573504807936052,
      "grad_norm": 0.07120192050933838,
      "learning_rate": 1.3044651847876163e-05,
      "loss": 0.051282644271850586,
      "step": 4242
    },
    {
      "epoch": 0.5736400047318878,
      "grad_norm": 0.06361155956983566,
      "learning_rate": 1.3037866025496466e-05,
      "loss": 0.058825016021728516,
      "step": 4243
    },
    {
      "epoch": 0.5737752015277238,
      "grad_norm": 0.04544292762875557,
      "learning_rate": 1.3031080611643514e-05,
      "loss": 0.061600685119628906,
      "step": 4244
    },
    {
      "epoch": 0.5739103983235597,
      "grad_norm": 0.08656535297632217,
      "learning_rate": 1.3024295607730083e-05,
      "loss": 0.08835315704345703,
      "step": 4245
    },
    {
      "epoch": 0.5740455951193957,
      "grad_norm": 0.13732722401618958,
      "learning_rate": 1.301751101516884e-05,
      "loss": 0.06532704830169678,
      "step": 4246
    },
    {
      "epoch": 0.5741807919152316,
      "grad_norm": 0.08534975349903107,
      "learning_rate": 1.3010726835372377e-05,
      "loss": 0.06534075736999512,
      "step": 4247
    },
    {
      "epoch": 0.5743159887110676,
      "grad_norm": 0.06839689612388611,
      "learning_rate": 1.30039430697532e-05,
      "loss": 0.06026506423950195,
      "step": 4248
    },
    {
      "epoch": 0.5744511855069034,
      "grad_norm": 0.07596047967672348,
      "learning_rate": 1.2997159719723713e-05,
      "loss": 0.04868745803833008,
      "step": 4249
    },
    {
      "epoch": 0.5745863823027394,
      "grad_norm": 0.142812579870224,
      "learning_rate": 1.2990376786696254e-05,
      "loss": 0.06736147403717041,
      "step": 4250
    },
    {
      "epoch": 0.5747215790985754,
      "grad_norm": 0.11198151856660843,
      "learning_rate": 1.2983594272083063e-05,
      "loss": 0.06825399398803711,
      "step": 4251
    },
    {
      "epoch": 0.5748567758944113,
      "grad_norm": 0.06062844395637512,
      "learning_rate": 1.2976812177296307e-05,
      "loss": 0.0415073037147522,
      "step": 4252
    },
    {
      "epoch": 0.5749919726902473,
      "grad_norm": 0.2237558513879776,
      "learning_rate": 1.2970030503748039e-05,
      "loss": 0.08611869812011719,
      "step": 4253
    },
    {
      "epoch": 0.5751271694860832,
      "grad_norm": 0.0790284126996994,
      "learning_rate": 1.2963249252850242e-05,
      "loss": 0.053389549255371094,
      "step": 4254
    },
    {
      "epoch": 0.5752623662819191,
      "grad_norm": 0.08783169835805893,
      "learning_rate": 1.295646842601481e-05,
      "loss": 0.047675132751464844,
      "step": 4255
    },
    {
      "epoch": 0.575397563077755,
      "grad_norm": 0.11812038719654083,
      "learning_rate": 1.294968802465355e-05,
      "loss": 0.04704391956329346,
      "step": 4256
    },
    {
      "epoch": 0.575532759873591,
      "grad_norm": 0.06148853898048401,
      "learning_rate": 1.2942908050178187e-05,
      "loss": 0.05360066890716553,
      "step": 4257
    },
    {
      "epoch": 0.5756679566694269,
      "grad_norm": 0.11899429559707642,
      "learning_rate": 1.293612850400033e-05,
      "loss": 0.0918264389038086,
      "step": 4258
    },
    {
      "epoch": 0.5758031534652629,
      "grad_norm": 0.09223272651433945,
      "learning_rate": 1.2929349387531525e-05,
      "loss": 0.09097796678543091,
      "step": 4259
    },
    {
      "epoch": 0.5759383502610989,
      "grad_norm": 0.17729069292545319,
      "learning_rate": 1.2922570702183217e-05,
      "loss": 0.0759168490767479,
      "step": 4260
    },
    {
      "epoch": 0.5760735470569347,
      "grad_norm": 0.06010180711746216,
      "learning_rate": 1.2915792449366768e-05,
      "loss": 0.05062532424926758,
      "step": 4261
    },
    {
      "epoch": 0.5762087438527707,
      "grad_norm": 0.12164776772260666,
      "learning_rate": 1.2909014630493451e-05,
      "loss": 0.08553028106689453,
      "step": 4262
    },
    {
      "epoch": 0.5763439406486066,
      "grad_norm": 0.07825644314289093,
      "learning_rate": 1.2902237246974432e-05,
      "loss": 0.06135845184326172,
      "step": 4263
    },
    {
      "epoch": 0.5764791374444426,
      "grad_norm": 0.04814029857516289,
      "learning_rate": 1.289546030022081e-05,
      "loss": 0.03315305709838867,
      "step": 4264
    },
    {
      "epoch": 0.5766143342402785,
      "grad_norm": 0.0693979263305664,
      "learning_rate": 1.2888683791643572e-05,
      "loss": 0.04173469543457031,
      "step": 4265
    },
    {
      "epoch": 0.5767495310361145,
      "grad_norm": 0.10865487158298492,
      "learning_rate": 1.2881907722653633e-05,
      "loss": 0.06183052062988281,
      "step": 4266
    },
    {
      "epoch": 0.5768847278319503,
      "grad_norm": 0.09468737989664078,
      "learning_rate": 1.2875132094661796e-05,
      "loss": 0.04216897487640381,
      "step": 4267
    },
    {
      "epoch": 0.5770199246277863,
      "grad_norm": 0.10020890086889267,
      "learning_rate": 1.2868356909078787e-05,
      "loss": 0.07439160346984863,
      "step": 4268
    },
    {
      "epoch": 0.5771551214236222,
      "grad_norm": 0.07731208205223083,
      "learning_rate": 1.286158216731524e-05,
      "loss": 0.04565715789794922,
      "step": 4269
    },
    {
      "epoch": 0.5772903182194582,
      "grad_norm": 0.07695096731185913,
      "learning_rate": 1.2854807870781686e-05,
      "loss": 0.058754920959472656,
      "step": 4270
    },
    {
      "epoch": 0.5774255150152942,
      "grad_norm": 0.0921509712934494,
      "learning_rate": 1.284803402088858e-05,
      "loss": 0.06733107566833496,
      "step": 4271
    },
    {
      "epoch": 0.5775607118111301,
      "grad_norm": 0.17220793664455414,
      "learning_rate": 1.284126061904626e-05,
      "loss": 0.08247411251068115,
      "step": 4272
    },
    {
      "epoch": 0.5776959086069661,
      "grad_norm": 0.04844580218195915,
      "learning_rate": 1.283448766666499e-05,
      "loss": 0.05113554000854492,
      "step": 4273
    },
    {
      "epoch": 0.5778311054028019,
      "grad_norm": 0.0884275957942009,
      "learning_rate": 1.282771516515494e-05,
      "loss": 0.08338689804077148,
      "step": 4274
    },
    {
      "epoch": 0.5779663021986379,
      "grad_norm": 0.05612083524465561,
      "learning_rate": 1.282094311592618e-05,
      "loss": 0.0509493350982666,
      "step": 4275
    },
    {
      "epoch": 0.5781014989944738,
      "grad_norm": 0.0678568035364151,
      "learning_rate": 1.2814171520388676e-05,
      "loss": 0.04532456398010254,
      "step": 4276
    },
    {
      "epoch": 0.5782366957903098,
      "grad_norm": 0.08181630820035934,
      "learning_rate": 1.2807400379952318e-05,
      "loss": 0.05478405952453613,
      "step": 4277
    },
    {
      "epoch": 0.5783718925861457,
      "grad_norm": 0.0537908598780632,
      "learning_rate": 1.2800629696026895e-05,
      "loss": 0.06338620185852051,
      "step": 4278
    },
    {
      "epoch": 0.5785070893819817,
      "grad_norm": 0.15645410120487213,
      "learning_rate": 1.2793859470022098e-05,
      "loss": 0.08153152465820312,
      "step": 4279
    },
    {
      "epoch": 0.5786422861778175,
      "grad_norm": 0.05124976485967636,
      "learning_rate": 1.278708970334753e-05,
      "loss": 0.04568624496459961,
      "step": 4280
    },
    {
      "epoch": 0.5787774829736535,
      "grad_norm": 0.08156215399503708,
      "learning_rate": 1.2780320397412678e-05,
      "loss": 0.062059640884399414,
      "step": 4281
    },
    {
      "epoch": 0.5789126797694895,
      "grad_norm": 0.07532893121242523,
      "learning_rate": 1.2773551553626957e-05,
      "loss": 0.06192612648010254,
      "step": 4282
    },
    {
      "epoch": 0.5790478765653254,
      "grad_norm": 0.11657476425170898,
      "learning_rate": 1.2766783173399675e-05,
      "loss": 0.06608283519744873,
      "step": 4283
    },
    {
      "epoch": 0.5791830733611614,
      "grad_norm": 0.054533701390028,
      "learning_rate": 1.276001525814005e-05,
      "loss": 0.04305529594421387,
      "step": 4284
    },
    {
      "epoch": 0.5793182701569973,
      "grad_norm": 0.04621422290802002,
      "learning_rate": 1.2753247809257192e-05,
      "loss": 0.045116424560546875,
      "step": 4285
    },
    {
      "epoch": 0.5794534669528332,
      "grad_norm": 0.058378081768751144,
      "learning_rate": 1.2746480828160119e-05,
      "loss": 0.06225442886352539,
      "step": 4286
    },
    {
      "epoch": 0.5795886637486691,
      "grad_norm": 0.09055574238300323,
      "learning_rate": 1.2739714316257753e-05,
      "loss": 0.06014537811279297,
      "step": 4287
    },
    {
      "epoch": 0.5797238605445051,
      "grad_norm": 0.09533646702766418,
      "learning_rate": 1.273294827495892e-05,
      "loss": 0.07230591773986816,
      "step": 4288
    },
    {
      "epoch": 0.579859057340341,
      "grad_norm": 0.07357525825500488,
      "learning_rate": 1.2726182705672352e-05,
      "loss": 0.0585169792175293,
      "step": 4289
    },
    {
      "epoch": 0.579994254136177,
      "grad_norm": 0.09403366595506668,
      "learning_rate": 1.271941760980667e-05,
      "loss": 0.06914520263671875,
      "step": 4290
    },
    {
      "epoch": 0.580129450932013,
      "grad_norm": 0.08954782783985138,
      "learning_rate": 1.2712652988770396e-05,
      "loss": 0.07144689559936523,
      "step": 4291
    },
    {
      "epoch": 0.5802646477278488,
      "grad_norm": 0.11628621816635132,
      "learning_rate": 1.2705888843971967e-05,
      "loss": 0.07306385040283203,
      "step": 4292
    },
    {
      "epoch": 0.5803998445236848,
      "grad_norm": 0.07146918028593063,
      "learning_rate": 1.2699125176819717e-05,
      "loss": 0.05750393867492676,
      "step": 4293
    },
    {
      "epoch": 0.5805350413195207,
      "grad_norm": 0.0771651491522789,
      "learning_rate": 1.269236198872188e-05,
      "loss": 0.07769155502319336,
      "step": 4294
    },
    {
      "epoch": 0.5806702381153567,
      "grad_norm": 0.04131464287638664,
      "learning_rate": 1.2685599281086577e-05,
      "loss": 0.04571548104286194,
      "step": 4295
    },
    {
      "epoch": 0.5808054349111926,
      "grad_norm": 0.0727495551109314,
      "learning_rate": 1.2678837055321849e-05,
      "loss": 0.05162614583969116,
      "step": 4296
    },
    {
      "epoch": 0.5809406317070286,
      "grad_norm": 0.08859908580780029,
      "learning_rate": 1.267207531283562e-05,
      "loss": 0.07144522666931152,
      "step": 4297
    },
    {
      "epoch": 0.5810758285028644,
      "grad_norm": 0.12046671658754349,
      "learning_rate": 1.266531405503573e-05,
      "loss": 0.09143280982971191,
      "step": 4298
    },
    {
      "epoch": 0.5812110252987004,
      "grad_norm": 0.05022713169455528,
      "learning_rate": 1.26585532833299e-05,
      "loss": 0.045766353607177734,
      "step": 4299
    },
    {
      "epoch": 0.5813462220945363,
      "grad_norm": 0.05693650245666504,
      "learning_rate": 1.2651792999125763e-05,
      "loss": 0.07350161671638489,
      "step": 4300
    },
    {
      "epoch": 0.5814814188903723,
      "grad_norm": 0.08686058968305588,
      "learning_rate": 1.2645033203830846e-05,
      "loss": 0.06072640419006348,
      "step": 4301
    },
    {
      "epoch": 0.5816166156862083,
      "grad_norm": 0.0378505140542984,
      "learning_rate": 1.2638273898852573e-05,
      "loss": 0.03803896903991699,
      "step": 4302
    },
    {
      "epoch": 0.5817518124820442,
      "grad_norm": 0.07793077826499939,
      "learning_rate": 1.2631515085598275e-05,
      "loss": 0.06520962715148926,
      "step": 4303
    },
    {
      "epoch": 0.5818870092778801,
      "grad_norm": 0.07835107296705246,
      "learning_rate": 1.262475676547516e-05,
      "loss": 0.06730532646179199,
      "step": 4304
    },
    {
      "epoch": 0.582022206073716,
      "grad_norm": 0.04331303760409355,
      "learning_rate": 1.2617998939890352e-05,
      "loss": 0.03670799732208252,
      "step": 4305
    },
    {
      "epoch": 0.582157402869552,
      "grad_norm": 0.061630748212337494,
      "learning_rate": 1.261124161025087e-05,
      "loss": 0.07912278175354004,
      "step": 4306
    },
    {
      "epoch": 0.5822925996653879,
      "grad_norm": 0.08987826108932495,
      "learning_rate": 1.260448477796362e-05,
      "loss": 0.07572793960571289,
      "step": 4307
    },
    {
      "epoch": 0.5824277964612239,
      "grad_norm": 0.07333193719387054,
      "learning_rate": 1.259772844443542e-05,
      "loss": 0.05310225486755371,
      "step": 4308
    },
    {
      "epoch": 0.5825629932570598,
      "grad_norm": 0.07325128465890884,
      "learning_rate": 1.2590972611072964e-05,
      "loss": 0.060450196266174316,
      "step": 4309
    },
    {
      "epoch": 0.5826981900528957,
      "grad_norm": 0.07857516407966614,
      "learning_rate": 1.2584217279282855e-05,
      "loss": 0.06274294853210449,
      "step": 4310
    },
    {
      "epoch": 0.5828333868487316,
      "grad_norm": 0.21415475010871887,
      "learning_rate": 1.2577462450471593e-05,
      "loss": 0.07872295379638672,
      "step": 4311
    },
    {
      "epoch": 0.5829685836445676,
      "grad_norm": 0.06408897042274475,
      "learning_rate": 1.2570708126045574e-05,
      "loss": 0.06525969505310059,
      "step": 4312
    },
    {
      "epoch": 0.5831037804404036,
      "grad_norm": 0.08825203031301498,
      "learning_rate": 1.256395430741107e-05,
      "loss": 0.05959606170654297,
      "step": 4313
    },
    {
      "epoch": 0.5832389772362395,
      "grad_norm": 0.12151241302490234,
      "learning_rate": 1.2557200995974268e-05,
      "loss": 0.08298683166503906,
      "step": 4314
    },
    {
      "epoch": 0.5833741740320755,
      "grad_norm": 0.2041746973991394,
      "learning_rate": 1.2550448193141248e-05,
      "loss": 0.06635129451751709,
      "step": 4315
    },
    {
      "epoch": 0.5835093708279114,
      "grad_norm": 0.04379083216190338,
      "learning_rate": 1.2543695900317977e-05,
      "loss": 0.05576348304748535,
      "step": 4316
    },
    {
      "epoch": 0.5836445676237473,
      "grad_norm": 0.06385776400566101,
      "learning_rate": 1.2536944118910323e-05,
      "loss": 0.055294156074523926,
      "step": 4317
    },
    {
      "epoch": 0.5837797644195832,
      "grad_norm": 0.08738244324922562,
      "learning_rate": 1.2530192850324032e-05,
      "loss": 0.06669628620147705,
      "step": 4318
    },
    {
      "epoch": 0.5839149612154192,
      "grad_norm": 0.1625872254371643,
      "learning_rate": 1.252344209596476e-05,
      "loss": 0.07272851467132568,
      "step": 4319
    },
    {
      "epoch": 0.5840501580112551,
      "grad_norm": 0.05684758350253105,
      "learning_rate": 1.251669185723805e-05,
      "loss": 0.04969900846481323,
      "step": 4320
    },
    {
      "epoch": 0.5841853548070911,
      "grad_norm": 0.06663138419389725,
      "learning_rate": 1.2509942135549344e-05,
      "loss": 0.06862401962280273,
      "step": 4321
    },
    {
      "epoch": 0.5843205516029271,
      "grad_norm": 0.10185294598340988,
      "learning_rate": 1.250319293230396e-05,
      "loss": 0.03975176811218262,
      "step": 4322
    },
    {
      "epoch": 0.5844557483987629,
      "grad_norm": 0.16229085624217987,
      "learning_rate": 1.2496444248907121e-05,
      "loss": 0.053262948989868164,
      "step": 4323
    },
    {
      "epoch": 0.5845909451945989,
      "grad_norm": 0.22272610664367676,
      "learning_rate": 1.2489696086763939e-05,
      "loss": 0.10411542654037476,
      "step": 4324
    },
    {
      "epoch": 0.5847261419904348,
      "grad_norm": 0.15241949260234833,
      "learning_rate": 1.2482948447279417e-05,
      "loss": 0.07978057861328125,
      "step": 4325
    },
    {
      "epoch": 0.5848613387862708,
      "grad_norm": 0.14279307425022125,
      "learning_rate": 1.2476201331858458e-05,
      "loss": 0.07953023910522461,
      "step": 4326
    },
    {
      "epoch": 0.5849965355821067,
      "grad_norm": 0.03669939562678337,
      "learning_rate": 1.2469454741905839e-05,
      "loss": 0.045300424098968506,
      "step": 4327
    },
    {
      "epoch": 0.5851317323779427,
      "grad_norm": 0.06310693919658661,
      "learning_rate": 1.2462708678826233e-05,
      "loss": 0.08255577087402344,
      "step": 4328
    },
    {
      "epoch": 0.5852669291737785,
      "grad_norm": 0.07301875203847885,
      "learning_rate": 1.245596314402421e-05,
      "loss": 0.06145954132080078,
      "step": 4329
    },
    {
      "epoch": 0.5854021259696145,
      "grad_norm": 0.085129015147686,
      "learning_rate": 1.2449218138904225e-05,
      "loss": 0.06589889526367188,
      "step": 4330
    },
    {
      "epoch": 0.5855373227654505,
      "grad_norm": 0.06567202508449554,
      "learning_rate": 1.2442473664870636e-05,
      "loss": 0.06220436096191406,
      "step": 4331
    },
    {
      "epoch": 0.5856725195612864,
      "grad_norm": 0.11292267590761185,
      "learning_rate": 1.2435729723327661e-05,
      "loss": 0.07966041564941406,
      "step": 4332
    },
    {
      "epoch": 0.5858077163571224,
      "grad_norm": 0.11293935775756836,
      "learning_rate": 1.2428986315679433e-05,
      "loss": 0.10083329677581787,
      "step": 4333
    },
    {
      "epoch": 0.5859429131529583,
      "grad_norm": 0.06286844611167908,
      "learning_rate": 1.2422243443329962e-05,
      "loss": 0.07518863677978516,
      "step": 4334
    },
    {
      "epoch": 0.5860781099487942,
      "grad_norm": 0.13119444251060486,
      "learning_rate": 1.241550110768316e-05,
      "loss": 0.08473682403564453,
      "step": 4335
    },
    {
      "epoch": 0.5862133067446301,
      "grad_norm": 0.05801859125494957,
      "learning_rate": 1.2408759310142803e-05,
      "loss": 0.051717281341552734,
      "step": 4336
    },
    {
      "epoch": 0.5863485035404661,
      "grad_norm": 0.14584961533546448,
      "learning_rate": 1.2402018052112576e-05,
      "loss": 0.08333253860473633,
      "step": 4337
    },
    {
      "epoch": 0.586483700336302,
      "grad_norm": 0.08096270263195038,
      "learning_rate": 1.2395277334996045e-05,
      "loss": 0.0552973747253418,
      "step": 4338
    },
    {
      "epoch": 0.586618897132138,
      "grad_norm": 0.07098846137523651,
      "learning_rate": 1.2388537160196663e-05,
      "loss": 0.05774843692779541,
      "step": 4339
    },
    {
      "epoch": 0.586754093927974,
      "grad_norm": 0.06354069709777832,
      "learning_rate": 1.2381797529117776e-05,
      "loss": 0.060511231422424316,
      "step": 4340
    },
    {
      "epoch": 0.5868892907238098,
      "grad_norm": 0.09636250138282776,
      "learning_rate": 1.23750584431626e-05,
      "loss": 0.06908249855041504,
      "step": 4341
    },
    {
      "epoch": 0.5870244875196458,
      "grad_norm": 0.11914902925491333,
      "learning_rate": 1.236831990373425e-05,
      "loss": 0.09679985046386719,
      "step": 4342
    },
    {
      "epoch": 0.5871596843154817,
      "grad_norm": 0.0579880066215992,
      "learning_rate": 1.2361581912235736e-05,
      "loss": 0.03981423377990723,
      "step": 4343
    },
    {
      "epoch": 0.5872948811113177,
      "grad_norm": 0.0891919881105423,
      "learning_rate": 1.235484447006994e-05,
      "loss": 0.059975266456604004,
      "step": 4344
    },
    {
      "epoch": 0.5874300779071536,
      "grad_norm": 0.060457512736320496,
      "learning_rate": 1.2348107578639627e-05,
      "loss": 0.04759836196899414,
      "step": 4345
    },
    {
      "epoch": 0.5875652747029896,
      "grad_norm": 0.040965937077999115,
      "learning_rate": 1.2341371239347454e-05,
      "loss": 0.03705120086669922,
      "step": 4346
    },
    {
      "epoch": 0.5877004714988254,
      "grad_norm": 0.0886962041258812,
      "learning_rate": 1.233463545359597e-05,
      "loss": 0.08275675773620605,
      "step": 4347
    },
    {
      "epoch": 0.5878356682946614,
      "grad_norm": 0.054824456572532654,
      "learning_rate": 1.23279002227876e-05,
      "loss": 0.050445556640625,
      "step": 4348
    },
    {
      "epoch": 0.5879708650904973,
      "grad_norm": 0.06049030274152756,
      "learning_rate": 1.2321165548324655e-05,
      "loss": 0.04991459846496582,
      "step": 4349
    },
    {
      "epoch": 0.5881060618863333,
      "grad_norm": 0.082413449883461,
      "learning_rate": 1.2314431431609323e-05,
      "loss": 0.05001091957092285,
      "step": 4350
    },
    {
      "epoch": 0.5882412586821693,
      "grad_norm": 0.06685661524534225,
      "learning_rate": 1.2307697874043687e-05,
      "loss": 0.0806427001953125,
      "step": 4351
    },
    {
      "epoch": 0.5883764554780052,
      "grad_norm": 0.13062191009521484,
      "learning_rate": 1.2300964877029712e-05,
      "loss": 0.04336404800415039,
      "step": 4352
    },
    {
      "epoch": 0.5885116522738411,
      "grad_norm": 0.19655302166938782,
      "learning_rate": 1.2294232441969246e-05,
      "loss": 0.06818056106567383,
      "step": 4353
    },
    {
      "epoch": 0.588646849069677,
      "grad_norm": 0.12404762208461761,
      "learning_rate": 1.2287500570264017e-05,
      "loss": 0.061942100524902344,
      "step": 4354
    },
    {
      "epoch": 0.588782045865513,
      "grad_norm": 0.0574922151863575,
      "learning_rate": 1.2280769263315628e-05,
      "loss": 0.05584937334060669,
      "step": 4355
    },
    {
      "epoch": 0.5889172426613489,
      "grad_norm": 0.08373738825321198,
      "learning_rate": 1.2274038522525577e-05,
      "loss": 0.0830233097076416,
      "step": 4356
    },
    {
      "epoch": 0.5890524394571849,
      "grad_norm": 0.08427920192480087,
      "learning_rate": 1.2267308349295246e-05,
      "loss": 0.05469787120819092,
      "step": 4357
    },
    {
      "epoch": 0.5891876362530208,
      "grad_norm": 0.18004299700260162,
      "learning_rate": 1.2260578745025892e-05,
      "loss": 0.088470458984375,
      "step": 4358
    },
    {
      "epoch": 0.5893228330488567,
      "grad_norm": 0.04521116614341736,
      "learning_rate": 1.225384971111865e-05,
      "loss": 0.03809046745300293,
      "step": 4359
    },
    {
      "epoch": 0.5894580298446926,
      "grad_norm": 0.07919525355100632,
      "learning_rate": 1.224712124897454e-05,
      "loss": 0.04444409906864166,
      "step": 4360
    },
    {
      "epoch": 0.5895932266405286,
      "grad_norm": 0.0926135927438736,
      "learning_rate": 1.2240393359994466e-05,
      "loss": 0.04586225748062134,
      "step": 4361
    },
    {
      "epoch": 0.5897284234363646,
      "grad_norm": 0.04971746355295181,
      "learning_rate": 1.2233666045579209e-05,
      "loss": 0.040291011333465576,
      "step": 4362
    },
    {
      "epoch": 0.5898636202322005,
      "grad_norm": 0.12292158603668213,
      "learning_rate": 1.222693930712944e-05,
      "loss": 0.05761420726776123,
      "step": 4363
    },
    {
      "epoch": 0.5899988170280365,
      "grad_norm": 0.1523953378200531,
      "learning_rate": 1.2220213146045691e-05,
      "loss": 0.07309556007385254,
      "step": 4364
    },
    {
      "epoch": 0.5901340138238724,
      "grad_norm": 0.13882263004779816,
      "learning_rate": 1.2213487563728389e-05,
      "loss": 0.06578588485717773,
      "step": 4365
    },
    {
      "epoch": 0.5902692106197083,
      "grad_norm": 0.12734755873680115,
      "learning_rate": 1.220676256157783e-05,
      "loss": 0.05518484115600586,
      "step": 4366
    },
    {
      "epoch": 0.5904044074155442,
      "grad_norm": 0.0852704793214798,
      "learning_rate": 1.2200038140994212e-05,
      "loss": 0.06588315963745117,
      "step": 4367
    },
    {
      "epoch": 0.5905396042113802,
      "grad_norm": 0.12388197332620621,
      "learning_rate": 1.2193314303377578e-05,
      "loss": 0.06680738925933838,
      "step": 4368
    },
    {
      "epoch": 0.5906748010072161,
      "grad_norm": 0.07882075756788254,
      "learning_rate": 1.2186591050127874e-05,
      "loss": 0.08265852928161621,
      "step": 4369
    },
    {
      "epoch": 0.5908099978030521,
      "grad_norm": 0.11623174697160721,
      "learning_rate": 1.2179868382644916e-05,
      "loss": 0.11804866790771484,
      "step": 4370
    },
    {
      "epoch": 0.5909451945988881,
      "grad_norm": 0.15952181816101074,
      "learning_rate": 1.2173146302328396e-05,
      "loss": 0.06362748146057129,
      "step": 4371
    },
    {
      "epoch": 0.5910803913947239,
      "grad_norm": 0.12032456696033478,
      "learning_rate": 1.21664248105779e-05,
      "loss": 0.07953834533691406,
      "step": 4372
    },
    {
      "epoch": 0.5912155881905599,
      "grad_norm": 0.13278843462467194,
      "learning_rate": 1.2159703908792858e-05,
      "loss": 0.04867291450500488,
      "step": 4373
    },
    {
      "epoch": 0.5913507849863958,
      "grad_norm": 0.17047610878944397,
      "learning_rate": 1.2152983598372613e-05,
      "loss": 0.050318002700805664,
      "step": 4374
    },
    {
      "epoch": 0.5914859817822318,
      "grad_norm": 0.13333120942115784,
      "learning_rate": 1.2146263880716366e-05,
      "loss": 0.052118778228759766,
      "step": 4375
    },
    {
      "epoch": 0.5916211785780677,
      "grad_norm": 0.05238313227891922,
      "learning_rate": 1.2139544757223194e-05,
      "loss": 0.03557056188583374,
      "step": 4376
    },
    {
      "epoch": 0.5917563753739037,
      "grad_norm": 0.0734664723277092,
      "learning_rate": 1.2132826229292066e-05,
      "loss": 0.07580423355102539,
      "step": 4377
    },
    {
      "epoch": 0.5918915721697395,
      "grad_norm": 0.058580432087183,
      "learning_rate": 1.2126108298321798e-05,
      "loss": 0.04852795600891113,
      "step": 4378
    },
    {
      "epoch": 0.5920267689655755,
      "grad_norm": 0.1077832356095314,
      "learning_rate": 1.2119390965711107e-05,
      "loss": 0.047350525856018066,
      "step": 4379
    },
    {
      "epoch": 0.5921619657614114,
      "grad_norm": 0.12427591532468796,
      "learning_rate": 1.2112674232858582e-05,
      "loss": 0.06860542297363281,
      "step": 4380
    },
    {
      "epoch": 0.5922971625572474,
      "grad_norm": 0.06752648204565048,
      "learning_rate": 1.2105958101162684e-05,
      "loss": 0.0730905532836914,
      "step": 4381
    },
    {
      "epoch": 0.5924323593530834,
      "grad_norm": 0.10141565650701523,
      "learning_rate": 1.2099242572021735e-05,
      "loss": 0.057066917419433594,
      "step": 4382
    },
    {
      "epoch": 0.5925675561489193,
      "grad_norm": 0.05199457332491875,
      "learning_rate": 1.209252764683395e-05,
      "loss": 0.05041339993476868,
      "step": 4383
    },
    {
      "epoch": 0.5927027529447552,
      "grad_norm": 0.08452307432889938,
      "learning_rate": 1.2085813326997414e-05,
      "loss": 0.05943500995635986,
      "step": 4384
    },
    {
      "epoch": 0.5928379497405911,
      "grad_norm": 0.10068266093730927,
      "learning_rate": 1.2079099613910088e-05,
      "loss": 0.07964611053466797,
      "step": 4385
    },
    {
      "epoch": 0.5929731465364271,
      "grad_norm": 0.05352442339062691,
      "learning_rate": 1.20723865089698e-05,
      "loss": 0.051158905029296875,
      "step": 4386
    },
    {
      "epoch": 0.593108343332263,
      "grad_norm": 0.09588709473609924,
      "learning_rate": 1.2065674013574248e-05,
      "loss": 0.04647183418273926,
      "step": 4387
    },
    {
      "epoch": 0.593243540128099,
      "grad_norm": 0.06838831305503845,
      "learning_rate": 1.2058962129121013e-05,
      "loss": 0.0778799057006836,
      "step": 4388
    },
    {
      "epoch": 0.593378736923935,
      "grad_norm": 0.1471521109342575,
      "learning_rate": 1.2052250857007548e-05,
      "loss": 0.07695746421813965,
      "step": 4389
    },
    {
      "epoch": 0.5935139337197708,
      "grad_norm": 0.08080458641052246,
      "learning_rate": 1.2045540198631177e-05,
      "loss": 0.05729961395263672,
      "step": 4390
    },
    {
      "epoch": 0.5936491305156067,
      "grad_norm": 0.056590184569358826,
      "learning_rate": 1.2038830155389091e-05,
      "loss": 0.0617365837097168,
      "step": 4391
    },
    {
      "epoch": 0.5937843273114427,
      "grad_norm": 0.08045302331447601,
      "learning_rate": 1.2032120728678354e-05,
      "loss": 0.05838799476623535,
      "step": 4392
    },
    {
      "epoch": 0.5939195241072787,
      "grad_norm": 0.030840545892715454,
      "learning_rate": 1.2025411919895907e-05,
      "loss": 0.02690201997756958,
      "step": 4393
    },
    {
      "epoch": 0.5940547209031146,
      "grad_norm": 0.07777990400791168,
      "learning_rate": 1.2018703730438561e-05,
      "loss": 0.058644771575927734,
      "step": 4394
    },
    {
      "epoch": 0.5941899176989506,
      "grad_norm": 0.1365945041179657,
      "learning_rate": 1.2011996161703003e-05,
      "loss": 0.08714818954467773,
      "step": 4395
    },
    {
      "epoch": 0.5943251144947864,
      "grad_norm": 0.09489940106868744,
      "learning_rate": 1.2005289215085775e-05,
      "loss": 0.07715433835983276,
      "step": 4396
    },
    {
      "epoch": 0.5944603112906224,
      "grad_norm": 0.06490685045719147,
      "learning_rate": 1.19985828919833e-05,
      "loss": 0.038370609283447266,
      "step": 4397
    },
    {
      "epoch": 0.5945955080864583,
      "grad_norm": 0.06644526869058609,
      "learning_rate": 1.1991877193791872e-05,
      "loss": 0.06057024002075195,
      "step": 4398
    },
    {
      "epoch": 0.5947307048822943,
      "grad_norm": 0.057687729597091675,
      "learning_rate": 1.1985172121907653e-05,
      "loss": 0.06514382362365723,
      "step": 4399
    },
    {
      "epoch": 0.5948659016781302,
      "grad_norm": 0.07421575486660004,
      "learning_rate": 1.1978467677726682e-05,
      "loss": 0.06838130950927734,
      "step": 4400
    },
    {
      "epoch": 0.5950010984739662,
      "grad_norm": 0.06818335503339767,
      "learning_rate": 1.197176386264485e-05,
      "loss": 0.062157630920410156,
      "step": 4401
    },
    {
      "epoch": 0.595136295269802,
      "grad_norm": 0.08384193480014801,
      "learning_rate": 1.1965060678057927e-05,
      "loss": 0.06805944442749023,
      "step": 4402
    },
    {
      "epoch": 0.595271492065638,
      "grad_norm": 0.06513229757547379,
      "learning_rate": 1.1958358125361554e-05,
      "loss": 0.06832408905029297,
      "step": 4403
    },
    {
      "epoch": 0.595406688861474,
      "grad_norm": 0.16560900211334229,
      "learning_rate": 1.1951656205951247e-05,
      "loss": 0.063323974609375,
      "step": 4404
    },
    {
      "epoch": 0.5955418856573099,
      "grad_norm": 0.07044423371553421,
      "learning_rate": 1.1944954921222367e-05,
      "loss": 0.05042409896850586,
      "step": 4405
    },
    {
      "epoch": 0.5956770824531459,
      "grad_norm": 0.08897704631090164,
      "learning_rate": 1.1938254272570167e-05,
      "loss": 0.07789993286132812,
      "step": 4406
    },
    {
      "epoch": 0.5958122792489818,
      "grad_norm": 0.0321781188249588,
      "learning_rate": 1.1931554261389751e-05,
      "loss": 0.03359043598175049,
      "step": 4407
    },
    {
      "epoch": 0.5959474760448178,
      "grad_norm": 0.043969232589006424,
      "learning_rate": 1.1924854889076103e-05,
      "loss": 0.052018165588378906,
      "step": 4408
    },
    {
      "epoch": 0.5960826728406536,
      "grad_norm": 0.07983013987541199,
      "learning_rate": 1.191815615702407e-05,
      "loss": 0.04252278804779053,
      "step": 4409
    },
    {
      "epoch": 0.5962178696364896,
      "grad_norm": 0.08181300759315491,
      "learning_rate": 1.1911458066628353e-05,
      "loss": 0.067130446434021,
      "step": 4410
    },
    {
      "epoch": 0.5963530664323256,
      "grad_norm": 0.08677782863378525,
      "learning_rate": 1.1904760619283537e-05,
      "loss": 0.05368852615356445,
      "step": 4411
    },
    {
      "epoch": 0.5964882632281615,
      "grad_norm": 0.046128302812576294,
      "learning_rate": 1.1898063816384069e-05,
      "loss": 0.039492011070251465,
      "step": 4412
    },
    {
      "epoch": 0.5966234600239975,
      "grad_norm": 0.06046607345342636,
      "learning_rate": 1.189136765932426e-05,
      "loss": 0.0479738712310791,
      "step": 4413
    },
    {
      "epoch": 0.5967586568198334,
      "grad_norm": 0.11654657125473022,
      "learning_rate": 1.1884672149498276e-05,
      "loss": 0.06967872381210327,
      "step": 4414
    },
    {
      "epoch": 0.5968938536156693,
      "grad_norm": 0.08996270596981049,
      "learning_rate": 1.187797728830016e-05,
      "loss": 0.05943727493286133,
      "step": 4415
    },
    {
      "epoch": 0.5970290504115052,
      "grad_norm": 0.18222592771053314,
      "learning_rate": 1.1871283077123823e-05,
      "loss": 0.12794017791748047,
      "step": 4416
    },
    {
      "epoch": 0.5971642472073412,
      "grad_norm": 0.06251989305019379,
      "learning_rate": 1.1864589517363038e-05,
      "loss": 0.0666964054107666,
      "step": 4417
    },
    {
      "epoch": 0.5972994440031771,
      "grad_norm": 0.10987559705972672,
      "learning_rate": 1.185789661041144e-05,
      "loss": 0.06929570436477661,
      "step": 4418
    },
    {
      "epoch": 0.5974346407990131,
      "grad_norm": 0.1608314961194992,
      "learning_rate": 1.1851204357662513e-05,
      "loss": 0.07338333129882812,
      "step": 4419
    },
    {
      "epoch": 0.597569837594849,
      "grad_norm": 0.04701576754450798,
      "learning_rate": 1.1844512760509634e-05,
      "loss": 0.04197359085083008,
      "step": 4420
    },
    {
      "epoch": 0.5977050343906849,
      "grad_norm": 0.07407865673303604,
      "learning_rate": 1.1837821820346022e-05,
      "loss": 0.051410675048828125,
      "step": 4421
    },
    {
      "epoch": 0.5978402311865209,
      "grad_norm": 0.05823945626616478,
      "learning_rate": 1.1831131538564775e-05,
      "loss": 0.040430545806884766,
      "step": 4422
    },
    {
      "epoch": 0.5979754279823568,
      "grad_norm": 0.10350481420755386,
      "learning_rate": 1.1824441916558843e-05,
      "loss": 0.052404165267944336,
      "step": 4423
    },
    {
      "epoch": 0.5981106247781928,
      "grad_norm": 0.08278493583202362,
      "learning_rate": 1.1817752955721031e-05,
      "loss": 0.07575035095214844,
      "step": 4424
    },
    {
      "epoch": 0.5982458215740287,
      "grad_norm": 0.07234026491641998,
      "learning_rate": 1.1811064657444023e-05,
      "loss": 0.04076719284057617,
      "step": 4425
    },
    {
      "epoch": 0.5983810183698647,
      "grad_norm": 0.19123956561088562,
      "learning_rate": 1.1804377023120361e-05,
      "loss": 0.07475078105926514,
      "step": 4426
    },
    {
      "epoch": 0.5985162151657005,
      "grad_norm": 0.046525053679943085,
      "learning_rate": 1.1797690054142451e-05,
      "loss": 0.0392451286315918,
      "step": 4427
    },
    {
      "epoch": 0.5986514119615365,
      "grad_norm": 0.14990352094173431,
      "learning_rate": 1.1791003751902542e-05,
      "loss": 0.057016611099243164,
      "step": 4428
    },
    {
      "epoch": 0.5987866087573724,
      "grad_norm": 0.07029904425144196,
      "learning_rate": 1.1784318117792763e-05,
      "loss": 0.06714379787445068,
      "step": 4429
    },
    {
      "epoch": 0.5989218055532084,
      "grad_norm": 0.07516420632600784,
      "learning_rate": 1.17776331532051e-05,
      "loss": 0.04988747835159302,
      "step": 4430
    },
    {
      "epoch": 0.5990570023490444,
      "grad_norm": 0.061831362545490265,
      "learning_rate": 1.1770948859531397e-05,
      "loss": 0.057512760162353516,
      "step": 4431
    },
    {
      "epoch": 0.5991921991448803,
      "grad_norm": 0.07346495240926743,
      "learning_rate": 1.1764265238163369e-05,
      "loss": 0.05953788757324219,
      "step": 4432
    },
    {
      "epoch": 0.5993273959407162,
      "grad_norm": 0.05526241287589073,
      "learning_rate": 1.1757582290492568e-05,
      "loss": 0.05492663383483887,
      "step": 4433
    },
    {
      "epoch": 0.5994625927365521,
      "grad_norm": 0.04413831979036331,
      "learning_rate": 1.1750900017910425e-05,
      "loss": 0.04280579090118408,
      "step": 4434
    },
    {
      "epoch": 0.5995977895323881,
      "grad_norm": 0.05408065766096115,
      "learning_rate": 1.1744218421808221e-05,
      "loss": 0.052460670471191406,
      "step": 4435
    },
    {
      "epoch": 0.599732986328224,
      "grad_norm": 0.04588092491030693,
      "learning_rate": 1.1737537503577112e-05,
      "loss": 0.043430209159851074,
      "step": 4436
    },
    {
      "epoch": 0.59986818312406,
      "grad_norm": 0.07211904227733612,
      "learning_rate": 1.1730857264608086e-05,
      "loss": 0.05210256576538086,
      "step": 4437
    },
    {
      "epoch": 0.6000033799198959,
      "grad_norm": 0.1481742560863495,
      "learning_rate": 1.1724177706292013e-05,
      "loss": 0.0910649299621582,
      "step": 4438
    },
    {
      "epoch": 0.6001385767157318,
      "grad_norm": 0.09262997657060623,
      "learning_rate": 1.1717498830019607e-05,
      "loss": 0.07965517044067383,
      "step": 4439
    },
    {
      "epoch": 0.6002737735115677,
      "grad_norm": 0.10517065227031708,
      "learning_rate": 1.1710820637181449e-05,
      "loss": 0.06760859489440918,
      "step": 4440
    },
    {
      "epoch": 0.6004089703074037,
      "grad_norm": 0.12969042360782623,
      "learning_rate": 1.170414312916798e-05,
      "loss": 0.10483169555664062,
      "step": 4441
    },
    {
      "epoch": 0.6005441671032397,
      "grad_norm": 0.0682075098156929,
      "learning_rate": 1.1697466307369484e-05,
      "loss": 0.04946744441986084,
      "step": 4442
    },
    {
      "epoch": 0.6006793638990756,
      "grad_norm": 0.09153049439191818,
      "learning_rate": 1.1690790173176116e-05,
      "loss": 0.07303977012634277,
      "step": 4443
    },
    {
      "epoch": 0.6008145606949116,
      "grad_norm": 0.0307060144841671,
      "learning_rate": 1.1684114727977876e-05,
      "loss": 0.02816760540008545,
      "step": 4444
    },
    {
      "epoch": 0.6009497574907474,
      "grad_norm": 0.19880400598049164,
      "learning_rate": 1.167743997316464e-05,
      "loss": 0.08561182022094727,
      "step": 4445
    },
    {
      "epoch": 0.6010849542865834,
      "grad_norm": 0.08696088194847107,
      "learning_rate": 1.1670765910126112e-05,
      "loss": 0.054329514503479004,
      "step": 4446
    },
    {
      "epoch": 0.6012201510824193,
      "grad_norm": 0.09676600247621536,
      "learning_rate": 1.1664092540251877e-05,
      "loss": 0.0719614028930664,
      "step": 4447
    },
    {
      "epoch": 0.6013553478782553,
      "grad_norm": 0.04374127835035324,
      "learning_rate": 1.1657419864931361e-05,
      "loss": 0.04409217834472656,
      "step": 4448
    },
    {
      "epoch": 0.6014905446740912,
      "grad_norm": 0.05726258084177971,
      "learning_rate": 1.165074788555386e-05,
      "loss": 0.04376363754272461,
      "step": 4449
    },
    {
      "epoch": 0.6016257414699272,
      "grad_norm": 0.08017420768737793,
      "learning_rate": 1.1644076603508514e-05,
      "loss": 0.04785299301147461,
      "step": 4450
    },
    {
      "epoch": 0.6017609382657632,
      "grad_norm": 0.08526419848203659,
      "learning_rate": 1.1637406020184305e-05,
      "loss": 0.06665301322937012,
      "step": 4451
    },
    {
      "epoch": 0.601896135061599,
      "grad_norm": 0.08559117466211319,
      "learning_rate": 1.1630736136970097e-05,
      "loss": 0.08066320419311523,
      "step": 4452
    },
    {
      "epoch": 0.602031331857435,
      "grad_norm": 0.03953111544251442,
      "learning_rate": 1.162406695525459e-05,
      "loss": 0.04483795166015625,
      "step": 4453
    },
    {
      "epoch": 0.6021665286532709,
      "grad_norm": 0.036057956516742706,
      "learning_rate": 1.161739847642635e-05,
      "loss": 0.05193614959716797,
      "step": 4454
    },
    {
      "epoch": 0.6023017254491069,
      "grad_norm": 0.13869522511959076,
      "learning_rate": 1.1610730701873788e-05,
      "loss": 0.06614875793457031,
      "step": 4455
    },
    {
      "epoch": 0.6024369222449428,
      "grad_norm": 0.07303797453641891,
      "learning_rate": 1.1604063632985163e-05,
      "loss": 0.09101521968841553,
      "step": 4456
    },
    {
      "epoch": 0.6025721190407788,
      "grad_norm": 0.18419502675533295,
      "learning_rate": 1.1597397271148598e-05,
      "loss": 0.06749451160430908,
      "step": 4457
    },
    {
      "epoch": 0.6027073158366146,
      "grad_norm": 0.06704884022474289,
      "learning_rate": 1.1590731617752067e-05,
      "loss": 0.04151320457458496,
      "step": 4458
    },
    {
      "epoch": 0.6028425126324506,
      "grad_norm": 0.07047439366579056,
      "learning_rate": 1.1584066674183398e-05,
      "loss": 0.051407575607299805,
      "step": 4459
    },
    {
      "epoch": 0.6029777094282865,
      "grad_norm": 0.07154999673366547,
      "learning_rate": 1.1577402441830262e-05,
      "loss": 0.08313798904418945,
      "step": 4460
    },
    {
      "epoch": 0.6031129062241225,
      "grad_norm": 0.10697804391384125,
      "learning_rate": 1.1570738922080185e-05,
      "loss": 0.08496499061584473,
      "step": 4461
    },
    {
      "epoch": 0.6032481030199585,
      "grad_norm": 0.07778322696685791,
      "learning_rate": 1.1564076116320552e-05,
      "loss": 0.06845259666442871,
      "step": 4462
    },
    {
      "epoch": 0.6033832998157944,
      "grad_norm": 0.12543387711048126,
      "learning_rate": 1.1557414025938592e-05,
      "loss": 0.08098649978637695,
      "step": 4463
    },
    {
      "epoch": 0.6035184966116303,
      "grad_norm": 0.06336898356676102,
      "learning_rate": 1.15507526523214e-05,
      "loss": 0.0354960560798645,
      "step": 4464
    },
    {
      "epoch": 0.6036536934074662,
      "grad_norm": 0.2825378477573395,
      "learning_rate": 1.1544091996855895e-05,
      "loss": 0.12302923202514648,
      "step": 4465
    },
    {
      "epoch": 0.6037888902033022,
      "grad_norm": 0.08610116690397263,
      "learning_rate": 1.153743206092886e-05,
      "loss": 0.06659555435180664,
      "step": 4466
    },
    {
      "epoch": 0.6039240869991381,
      "grad_norm": 0.06296218931674957,
      "learning_rate": 1.1530772845926936e-05,
      "loss": 0.03441047668457031,
      "step": 4467
    },
    {
      "epoch": 0.6040592837949741,
      "grad_norm": 0.17660225927829742,
      "learning_rate": 1.1524114353236614e-05,
      "loss": 0.07161283493041992,
      "step": 4468
    },
    {
      "epoch": 0.60419448059081,
      "grad_norm": 0.0970006138086319,
      "learning_rate": 1.151745658424421e-05,
      "loss": 0.07809591293334961,
      "step": 4469
    },
    {
      "epoch": 0.6043296773866459,
      "grad_norm": 0.04494573548436165,
      "learning_rate": 1.151079954033592e-05,
      "loss": 0.029191136360168457,
      "step": 4470
    },
    {
      "epoch": 0.6044648741824818,
      "grad_norm": 0.06157716363668442,
      "learning_rate": 1.150414322289777e-05,
      "loss": 0.05384635925292969,
      "step": 4471
    },
    {
      "epoch": 0.6046000709783178,
      "grad_norm": 0.0951310321688652,
      "learning_rate": 1.1497487633315643e-05,
      "loss": 0.07082986831665039,
      "step": 4472
    },
    {
      "epoch": 0.6047352677741538,
      "grad_norm": 0.08796324580907822,
      "learning_rate": 1.1490832772975275e-05,
      "loss": 0.06610298156738281,
      "step": 4473
    },
    {
      "epoch": 0.6048704645699897,
      "grad_norm": 0.11102192848920822,
      "learning_rate": 1.148417864326223e-05,
      "loss": 0.06828904151916504,
      "step": 4474
    },
    {
      "epoch": 0.6050056613658257,
      "grad_norm": 0.07584069669246674,
      "learning_rate": 1.1477525245561944e-05,
      "loss": 0.049087464809417725,
      "step": 4475
    },
    {
      "epoch": 0.6051408581616615,
      "grad_norm": 0.05231916904449463,
      "learning_rate": 1.1470872581259684e-05,
      "loss": 0.07267236709594727,
      "step": 4476
    },
    {
      "epoch": 0.6052760549574975,
      "grad_norm": 0.1191301941871643,
      "learning_rate": 1.146422065174057e-05,
      "loss": 0.09348361939191818,
      "step": 4477
    },
    {
      "epoch": 0.6054112517533334,
      "grad_norm": 0.12056214362382889,
      "learning_rate": 1.1457569458389578e-05,
      "loss": 0.0738821029663086,
      "step": 4478
    },
    {
      "epoch": 0.6055464485491694,
      "grad_norm": 0.09845026582479477,
      "learning_rate": 1.145091900259151e-05,
      "loss": 0.08072614669799805,
      "step": 4479
    },
    {
      "epoch": 0.6056816453450053,
      "grad_norm": 0.07445214688777924,
      "learning_rate": 1.1444269285731032e-05,
      "loss": 0.04471850395202637,
      "step": 4480
    },
    {
      "epoch": 0.6058168421408413,
      "grad_norm": 0.06066924333572388,
      "learning_rate": 1.1437620309192652e-05,
      "loss": 0.03565335273742676,
      "step": 4481
    },
    {
      "epoch": 0.6059520389366772,
      "grad_norm": 0.07676257938146591,
      "learning_rate": 1.1430972074360722e-05,
      "loss": 0.07148265838623047,
      "step": 4482
    },
    {
      "epoch": 0.6060872357325131,
      "grad_norm": 0.08411075919866562,
      "learning_rate": 1.1424324582619435e-05,
      "loss": 0.06795120239257812,
      "step": 4483
    },
    {
      "epoch": 0.6062224325283491,
      "grad_norm": 0.05896037071943283,
      "learning_rate": 1.1417677835352837e-05,
      "loss": 0.05142498016357422,
      "step": 4484
    },
    {
      "epoch": 0.606357629324185,
      "grad_norm": 0.12102647870779037,
      "learning_rate": 1.1411031833944816e-05,
      "loss": 0.08036601543426514,
      "step": 4485
    },
    {
      "epoch": 0.606492826120021,
      "grad_norm": 0.08629456162452698,
      "learning_rate": 1.1404386579779111e-05,
      "loss": 0.05320119857788086,
      "step": 4486
    },
    {
      "epoch": 0.6066280229158569,
      "grad_norm": 0.12007882446050644,
      "learning_rate": 1.1397742074239296e-05,
      "loss": 0.07620716094970703,
      "step": 4487
    },
    {
      "epoch": 0.6067632197116928,
      "grad_norm": 0.08685746043920517,
      "learning_rate": 1.1391098318708785e-05,
      "loss": 0.07518482208251953,
      "step": 4488
    },
    {
      "epoch": 0.6068984165075287,
      "grad_norm": 0.07086971402168274,
      "learning_rate": 1.1384455314570848e-05,
      "loss": 0.04566526412963867,
      "step": 4489
    },
    {
      "epoch": 0.6070336133033647,
      "grad_norm": 0.06558715552091599,
      "learning_rate": 1.1377813063208596e-05,
      "loss": 0.07185649871826172,
      "step": 4490
    },
    {
      "epoch": 0.6071688100992007,
      "grad_norm": 0.08730585128068924,
      "learning_rate": 1.1371171566004986e-05,
      "loss": 0.0811614990234375,
      "step": 4491
    },
    {
      "epoch": 0.6073040068950366,
      "grad_norm": 0.12361347675323486,
      "learning_rate": 1.1364530824342806e-05,
      "loss": 0.05371090769767761,
      "step": 4492
    },
    {
      "epoch": 0.6074392036908726,
      "grad_norm": 0.05922313779592514,
      "learning_rate": 1.1357890839604688e-05,
      "loss": 0.05019420385360718,
      "step": 4493
    },
    {
      "epoch": 0.6075744004867084,
      "grad_norm": 0.06025974079966545,
      "learning_rate": 1.1351251613173122e-05,
      "loss": 0.055617570877075195,
      "step": 4494
    },
    {
      "epoch": 0.6077095972825444,
      "grad_norm": 0.06857304275035858,
      "learning_rate": 1.1344613146430428e-05,
      "loss": 0.06400120258331299,
      "step": 4495
    },
    {
      "epoch": 0.6078447940783803,
      "grad_norm": 0.04857001453638077,
      "learning_rate": 1.1337975440758775e-05,
      "loss": 0.048676252365112305,
      "step": 4496
    },
    {
      "epoch": 0.6079799908742163,
      "grad_norm": 0.1769881695508957,
      "learning_rate": 1.133133849754016e-05,
      "loss": 0.09433364868164062,
      "step": 4497
    },
    {
      "epoch": 0.6081151876700522,
      "grad_norm": 0.07416752725839615,
      "learning_rate": 1.1324702318156431e-05,
      "loss": 0.06578731536865234,
      "step": 4498
    },
    {
      "epoch": 0.6082503844658882,
      "grad_norm": 0.053321123123168945,
      "learning_rate": 1.1318066903989279e-05,
      "loss": 0.06825423240661621,
      "step": 4499
    },
    {
      "epoch": 0.6083855812617242,
      "grad_norm": 0.10447041690349579,
      "learning_rate": 1.1311432256420232e-05,
      "loss": 0.06466418504714966,
      "step": 4500
    },
    {
      "epoch": 0.60852077805756,
      "grad_norm": 0.04495958983898163,
      "learning_rate": 1.1304798376830664e-05,
      "loss": 0.0367891788482666,
      "step": 4501
    },
    {
      "epoch": 0.608655974853396,
      "grad_norm": 0.11137567460536957,
      "learning_rate": 1.1298165266601778e-05,
      "loss": 0.09295177459716797,
      "step": 4502
    },
    {
      "epoch": 0.6087911716492319,
      "grad_norm": 0.08697705715894699,
      "learning_rate": 1.129153292711462e-05,
      "loss": 0.06599617004394531,
      "step": 4503
    },
    {
      "epoch": 0.6089263684450679,
      "grad_norm": 0.042401522397994995,
      "learning_rate": 1.1284901359750082e-05,
      "loss": 0.04012654721736908,
      "step": 4504
    },
    {
      "epoch": 0.6090615652409038,
      "grad_norm": 0.06264140456914902,
      "learning_rate": 1.1278270565888897e-05,
      "loss": 0.04436802864074707,
      "step": 4505
    },
    {
      "epoch": 0.6091967620367398,
      "grad_norm": 0.13806472718715668,
      "learning_rate": 1.1271640546911624e-05,
      "loss": 0.0663565993309021,
      "step": 4506
    },
    {
      "epoch": 0.6093319588325756,
      "grad_norm": 0.08638143539428711,
      "learning_rate": 1.1265011304198672e-05,
      "loss": 0.08188581466674805,
      "step": 4507
    },
    {
      "epoch": 0.6094671556284116,
      "grad_norm": 0.17412403225898743,
      "learning_rate": 1.1258382839130282e-05,
      "loss": 0.08156490325927734,
      "step": 4508
    },
    {
      "epoch": 0.6096023524242475,
      "grad_norm": 0.04546007141470909,
      "learning_rate": 1.1251755153086536e-05,
      "loss": 0.040462493896484375,
      "step": 4509
    },
    {
      "epoch": 0.6097375492200835,
      "grad_norm": 0.10680299252271652,
      "learning_rate": 1.1245128247447362e-05,
      "loss": 0.07860314846038818,
      "step": 4510
    },
    {
      "epoch": 0.6098727460159195,
      "grad_norm": 0.05734534561634064,
      "learning_rate": 1.1238502123592507e-05,
      "loss": 0.05141758918762207,
      "step": 4511
    },
    {
      "epoch": 0.6100079428117554,
      "grad_norm": 0.08302834630012512,
      "learning_rate": 1.1231876782901568e-05,
      "loss": 0.04714542627334595,
      "step": 4512
    },
    {
      "epoch": 0.6101431396075913,
      "grad_norm": 0.055085211992263794,
      "learning_rate": 1.1225252226753975e-05,
      "loss": 0.05603361129760742,
      "step": 4513
    },
    {
      "epoch": 0.6102783364034272,
      "grad_norm": 0.10334553569555283,
      "learning_rate": 1.1218628456529005e-05,
      "loss": 0.07889223098754883,
      "step": 4514
    },
    {
      "epoch": 0.6104135331992632,
      "grad_norm": 0.08391764014959335,
      "learning_rate": 1.1212005473605746e-05,
      "loss": 0.07118511199951172,
      "step": 4515
    },
    {
      "epoch": 0.6105487299950991,
      "grad_norm": 0.1853853464126587,
      "learning_rate": 1.120538327936315e-05,
      "loss": 0.07809734344482422,
      "step": 4516
    },
    {
      "epoch": 0.6106839267909351,
      "grad_norm": 0.05819907411932945,
      "learning_rate": 1.1198761875179993e-05,
      "loss": 0.04647541046142578,
      "step": 4517
    },
    {
      "epoch": 0.610819123586771,
      "grad_norm": 0.09502635896205902,
      "learning_rate": 1.1192141262434883e-05,
      "loss": 0.05236482620239258,
      "step": 4518
    },
    {
      "epoch": 0.6109543203826069,
      "grad_norm": 0.03877397999167442,
      "learning_rate": 1.1185521442506272e-05,
      "loss": 0.05437469482421875,
      "step": 4519
    },
    {
      "epoch": 0.6110895171784428,
      "grad_norm": 0.0829184502363205,
      "learning_rate": 1.1178902416772432e-05,
      "loss": 0.04833507537841797,
      "step": 4520
    },
    {
      "epoch": 0.6112247139742788,
      "grad_norm": 0.11189968883991241,
      "learning_rate": 1.1172284186611485e-05,
      "loss": 0.06994009017944336,
      "step": 4521
    },
    {
      "epoch": 0.6113599107701148,
      "grad_norm": 0.05897166579961777,
      "learning_rate": 1.1165666753401384e-05,
      "loss": 0.04030972719192505,
      "step": 4522
    },
    {
      "epoch": 0.6114951075659507,
      "grad_norm": 0.09178470820188522,
      "learning_rate": 1.1159050118519914e-05,
      "loss": 0.08080816268920898,
      "step": 4523
    },
    {
      "epoch": 0.6116303043617867,
      "grad_norm": 0.16161687672138214,
      "learning_rate": 1.1152434283344696e-05,
      "loss": 0.0764613151550293,
      "step": 4524
    },
    {
      "epoch": 0.6117655011576225,
      "grad_norm": 0.06385630369186401,
      "learning_rate": 1.114581924925317e-05,
      "loss": 0.056527018547058105,
      "step": 4525
    },
    {
      "epoch": 0.6119006979534585,
      "grad_norm": 0.04277735948562622,
      "learning_rate": 1.113920501762263e-05,
      "loss": 0.041338324546813965,
      "step": 4526
    },
    {
      "epoch": 0.6120358947492944,
      "grad_norm": 0.043355513364076614,
      "learning_rate": 1.1132591589830193e-05,
      "loss": 0.052962303161621094,
      "step": 4527
    },
    {
      "epoch": 0.6121710915451304,
      "grad_norm": 0.09569446742534637,
      "learning_rate": 1.1125978967252818e-05,
      "loss": 0.06783771514892578,
      "step": 4528
    },
    {
      "epoch": 0.6123062883409663,
      "grad_norm": 0.058598004281520844,
      "learning_rate": 1.1119367151267278e-05,
      "loss": 0.046468138694763184,
      "step": 4529
    },
    {
      "epoch": 0.6124414851368023,
      "grad_norm": 0.12105164676904678,
      "learning_rate": 1.1112756143250186e-05,
      "loss": 0.08240771293640137,
      "step": 4530
    },
    {
      "epoch": 0.6125766819326381,
      "grad_norm": 0.07010209560394287,
      "learning_rate": 1.1106145944577995e-05,
      "loss": 0.05855751037597656,
      "step": 4531
    },
    {
      "epoch": 0.6127118787284741,
      "grad_norm": 0.12166358530521393,
      "learning_rate": 1.1099536556626984e-05,
      "loss": 0.05544114112854004,
      "step": 4532
    },
    {
      "epoch": 0.6128470755243101,
      "grad_norm": 0.0898999348282814,
      "learning_rate": 1.1092927980773269e-05,
      "loss": 0.05614280700683594,
      "step": 4533
    },
    {
      "epoch": 0.612982272320146,
      "grad_norm": 0.1359388381242752,
      "learning_rate": 1.1086320218392777e-05,
      "loss": 0.06669783592224121,
      "step": 4534
    },
    {
      "epoch": 0.613117469115982,
      "grad_norm": 0.04480629414319992,
      "learning_rate": 1.1079713270861286e-05,
      "loss": 0.049057722091674805,
      "step": 4535
    },
    {
      "epoch": 0.6132526659118179,
      "grad_norm": 0.13082189857959747,
      "learning_rate": 1.1073107139554395e-05,
      "loss": 0.04634416103363037,
      "step": 4536
    },
    {
      "epoch": 0.6133878627076538,
      "grad_norm": 0.05390363559126854,
      "learning_rate": 1.1066501825847545e-05,
      "loss": 0.054264068603515625,
      "step": 4537
    },
    {
      "epoch": 0.6135230595034897,
      "grad_norm": 0.08407049626111984,
      "learning_rate": 1.1059897331115985e-05,
      "loss": 0.04398465156555176,
      "step": 4538
    },
    {
      "epoch": 0.6136582562993257,
      "grad_norm": 0.1293017566204071,
      "learning_rate": 1.1053293656734816e-05,
      "loss": 0.0730581283569336,
      "step": 4539
    },
    {
      "epoch": 0.6137934530951616,
      "grad_norm": 0.09226568043231964,
      "learning_rate": 1.1046690804078949e-05,
      "loss": 0.05648469924926758,
      "step": 4540
    },
    {
      "epoch": 0.6139286498909976,
      "grad_norm": 0.052317120134830475,
      "learning_rate": 1.1040088774523139e-05,
      "loss": 0.047637939453125,
      "step": 4541
    },
    {
      "epoch": 0.6140638466868336,
      "grad_norm": 0.07901492714881897,
      "learning_rate": 1.1033487569441971e-05,
      "loss": 0.0620880126953125,
      "step": 4542
    },
    {
      "epoch": 0.6141990434826695,
      "grad_norm": 0.09862032532691956,
      "learning_rate": 1.1026887190209834e-05,
      "loss": 0.05237436294555664,
      "step": 4543
    },
    {
      "epoch": 0.6143342402785054,
      "grad_norm": 0.06221865862607956,
      "learning_rate": 1.1020287638200977e-05,
      "loss": 0.0640573501586914,
      "step": 4544
    },
    {
      "epoch": 0.6144694370743413,
      "grad_norm": 0.05767063423991203,
      "learning_rate": 1.1013688914789452e-05,
      "loss": 0.056105852127075195,
      "step": 4545
    },
    {
      "epoch": 0.6146046338701773,
      "grad_norm": 0.055492449551820755,
      "learning_rate": 1.100709102134915e-05,
      "loss": 0.057866811752319336,
      "step": 4546
    },
    {
      "epoch": 0.6147398306660132,
      "grad_norm": 0.06277579069137573,
      "learning_rate": 1.10004939592538e-05,
      "loss": 0.05402016639709473,
      "step": 4547
    },
    {
      "epoch": 0.6148750274618492,
      "grad_norm": 0.0807250514626503,
      "learning_rate": 1.0993897729876927e-05,
      "loss": 0.0560833215713501,
      "step": 4548
    },
    {
      "epoch": 0.6150102242576851,
      "grad_norm": 0.09189426898956299,
      "learning_rate": 1.0987302334591915e-05,
      "loss": 0.04750269651412964,
      "step": 4549
    },
    {
      "epoch": 0.615145421053521,
      "grad_norm": 0.12089020758867264,
      "learning_rate": 1.098070777477195e-05,
      "loss": 0.05227017402648926,
      "step": 4550
    },
    {
      "epoch": 0.615280617849357,
      "grad_norm": 0.09360329061746597,
      "learning_rate": 1.0974114051790067e-05,
      "loss": 0.06433185935020447,
      "step": 4551
    },
    {
      "epoch": 0.6154158146451929,
      "grad_norm": 0.11711576581001282,
      "learning_rate": 1.09675211670191e-05,
      "loss": 0.0613856315612793,
      "step": 4552
    },
    {
      "epoch": 0.6155510114410289,
      "grad_norm": 0.12193598598241806,
      "learning_rate": 1.0960929121831732e-05,
      "loss": 0.044788360595703125,
      "step": 4553
    },
    {
      "epoch": 0.6156862082368648,
      "grad_norm": 0.10480480641126633,
      "learning_rate": 1.095433791760046e-05,
      "loss": 0.05182170867919922,
      "step": 4554
    },
    {
      "epoch": 0.6158214050327008,
      "grad_norm": 0.08455322682857513,
      "learning_rate": 1.0947747555697609e-05,
      "loss": 0.0470576286315918,
      "step": 4555
    },
    {
      "epoch": 0.6159566018285366,
      "grad_norm": 0.06877885013818741,
      "learning_rate": 1.0941158037495328e-05,
      "loss": 0.059427738189697266,
      "step": 4556
    },
    {
      "epoch": 0.6160917986243726,
      "grad_norm": 0.06504160910844803,
      "learning_rate": 1.0934569364365583e-05,
      "loss": 0.06248807907104492,
      "step": 4557
    },
    {
      "epoch": 0.6162269954202085,
      "grad_norm": 0.11980217695236206,
      "learning_rate": 1.0927981537680176e-05,
      "loss": 0.05568826198577881,
      "step": 4558
    },
    {
      "epoch": 0.6163621922160445,
      "grad_norm": 0.10692892968654633,
      "learning_rate": 1.0921394558810726e-05,
      "loss": 0.08085012435913086,
      "step": 4559
    },
    {
      "epoch": 0.6164973890118804,
      "grad_norm": 0.09154141694307327,
      "learning_rate": 1.0914808429128688e-05,
      "loss": 0.054042696952819824,
      "step": 4560
    },
    {
      "epoch": 0.6166325858077164,
      "grad_norm": 0.08877961337566376,
      "learning_rate": 1.0908223150005315e-05,
      "loss": 0.05475306510925293,
      "step": 4561
    },
    {
      "epoch": 0.6167677826035523,
      "grad_norm": 0.14868059754371643,
      "learning_rate": 1.09016387228117e-05,
      "loss": 0.06053924560546875,
      "step": 4562
    },
    {
      "epoch": 0.6169029793993882,
      "grad_norm": 0.08445591479539871,
      "learning_rate": 1.0895055148918758e-05,
      "loss": 0.06797933578491211,
      "step": 4563
    },
    {
      "epoch": 0.6170381761952242,
      "grad_norm": 0.10985692590475082,
      "learning_rate": 1.0888472429697223e-05,
      "loss": 0.07233119010925293,
      "step": 4564
    },
    {
      "epoch": 0.6171733729910601,
      "grad_norm": 0.14809873700141907,
      "learning_rate": 1.088189056651766e-05,
      "loss": 0.08217453956604004,
      "step": 4565
    },
    {
      "epoch": 0.6173085697868961,
      "grad_norm": 0.09050686657428741,
      "learning_rate": 1.0875309560750438e-05,
      "loss": 0.07004356384277344,
      "step": 4566
    },
    {
      "epoch": 0.617443766582732,
      "grad_norm": 0.06252928823232651,
      "learning_rate": 1.086872941376576e-05,
      "loss": 0.057417869567871094,
      "step": 4567
    },
    {
      "epoch": 0.6175789633785679,
      "grad_norm": 0.09707442671060562,
      "learning_rate": 1.0862150126933648e-05,
      "loss": 0.047460198402404785,
      "step": 4568
    },
    {
      "epoch": 0.6177141601744038,
      "grad_norm": 0.08752568811178207,
      "learning_rate": 1.0855571701623942e-05,
      "loss": 0.08448314666748047,
      "step": 4569
    },
    {
      "epoch": 0.6178493569702398,
      "grad_norm": 0.12507696449756622,
      "learning_rate": 1.0848994139206317e-05,
      "loss": 0.09156608581542969,
      "step": 4570
    },
    {
      "epoch": 0.6179845537660758,
      "grad_norm": 0.10488781332969666,
      "learning_rate": 1.0842417441050247e-05,
      "loss": 0.07348823547363281,
      "step": 4571
    },
    {
      "epoch": 0.6181197505619117,
      "grad_norm": 0.07688906788825989,
      "learning_rate": 1.0835841608525031e-05,
      "loss": 0.07751977443695068,
      "step": 4572
    },
    {
      "epoch": 0.6182549473577477,
      "grad_norm": 0.05180235207080841,
      "learning_rate": 1.08292666429998e-05,
      "loss": 0.055185794830322266,
      "step": 4573
    },
    {
      "epoch": 0.6183901441535835,
      "grad_norm": 0.11347298324108124,
      "learning_rate": 1.08226925458435e-05,
      "loss": 0.06383049488067627,
      "step": 4574
    },
    {
      "epoch": 0.6185253409494195,
      "grad_norm": 0.11888216435909271,
      "learning_rate": 1.0816119318424882e-05,
      "loss": 0.10407376289367676,
      "step": 4575
    },
    {
      "epoch": 0.6186605377452554,
      "grad_norm": 0.05760718509554863,
      "learning_rate": 1.0809546962112535e-05,
      "loss": 0.0504530668258667,
      "step": 4576
    },
    {
      "epoch": 0.6187957345410914,
      "grad_norm": 0.061022959649562836,
      "learning_rate": 1.0802975478274856e-05,
      "loss": 0.069732666015625,
      "step": 4577
    },
    {
      "epoch": 0.6189309313369273,
      "grad_norm": 0.07595275342464447,
      "learning_rate": 1.0796404868280062e-05,
      "loss": 0.08083534240722656,
      "step": 4578
    },
    {
      "epoch": 0.6190661281327633,
      "grad_norm": 0.05080515518784523,
      "learning_rate": 1.07898351334962e-05,
      "loss": 0.046851396560668945,
      "step": 4579
    },
    {
      "epoch": 0.6192013249285991,
      "grad_norm": 0.21546253561973572,
      "learning_rate": 1.0783266275291103e-05,
      "loss": 0.09590339660644531,
      "step": 4580
    },
    {
      "epoch": 0.6193365217244351,
      "grad_norm": 0.06499781459569931,
      "learning_rate": 1.077669829503246e-05,
      "loss": 0.04685091972351074,
      "step": 4581
    },
    {
      "epoch": 0.619471718520271,
      "grad_norm": 0.0577646940946579,
      "learning_rate": 1.077013119408775e-05,
      "loss": 0.05171847343444824,
      "step": 4582
    },
    {
      "epoch": 0.619606915316107,
      "grad_norm": 0.0627833753824234,
      "learning_rate": 1.0763564973824289e-05,
      "loss": 0.06477808952331543,
      "step": 4583
    },
    {
      "epoch": 0.619742112111943,
      "grad_norm": 0.08142044395208359,
      "learning_rate": 1.0756999635609185e-05,
      "loss": 0.07486128807067871,
      "step": 4584
    },
    {
      "epoch": 0.6198773089077789,
      "grad_norm": 0.04036731645464897,
      "learning_rate": 1.0750435180809381e-05,
      "loss": 0.033394813537597656,
      "step": 4585
    },
    {
      "epoch": 0.6200125057036148,
      "grad_norm": 0.08377616852521896,
      "learning_rate": 1.074387161079164e-05,
      "loss": 0.07164502143859863,
      "step": 4586
    },
    {
      "epoch": 0.6201477024994507,
      "grad_norm": 0.05724209547042847,
      "learning_rate": 1.0737308926922521e-05,
      "loss": 0.056397438049316406,
      "step": 4587
    },
    {
      "epoch": 0.6202828992952867,
      "grad_norm": 0.06017300859093666,
      "learning_rate": 1.0730747130568424e-05,
      "loss": 0.0733184814453125,
      "step": 4588
    },
    {
      "epoch": 0.6204180960911226,
      "grad_norm": 0.07337550818920135,
      "learning_rate": 1.0724186223095532e-05,
      "loss": 0.056066691875457764,
      "step": 4589
    },
    {
      "epoch": 0.6205532928869586,
      "grad_norm": 0.06339246034622192,
      "learning_rate": 1.071762620586987e-05,
      "loss": 0.06686973571777344,
      "step": 4590
    },
    {
      "epoch": 0.6206884896827946,
      "grad_norm": 0.15413643419742584,
      "learning_rate": 1.0711067080257273e-05,
      "loss": 0.051012277603149414,
      "step": 4591
    },
    {
      "epoch": 0.6208236864786305,
      "grad_norm": 0.09538403153419495,
      "learning_rate": 1.0704508847623374e-05,
      "loss": 0.06360960006713867,
      "step": 4592
    },
    {
      "epoch": 0.6209588832744664,
      "grad_norm": 0.04824884980916977,
      "learning_rate": 1.069795150933365e-05,
      "loss": 0.04549694061279297,
      "step": 4593
    },
    {
      "epoch": 0.6210940800703023,
      "grad_norm": 0.07252566516399384,
      "learning_rate": 1.0691395066753357e-05,
      "loss": 0.08558177947998047,
      "step": 4594
    },
    {
      "epoch": 0.6212292768661383,
      "grad_norm": 0.1259952038526535,
      "learning_rate": 1.0684839521247584e-05,
      "loss": 0.0791010856628418,
      "step": 4595
    },
    {
      "epoch": 0.6213644736619742,
      "grad_norm": 0.07317842543125153,
      "learning_rate": 1.0678284874181234e-05,
      "loss": 0.06471514701843262,
      "step": 4596
    },
    {
      "epoch": 0.6214996704578102,
      "grad_norm": 0.06666328758001328,
      "learning_rate": 1.0671731126919028e-05,
      "loss": 0.048066139221191406,
      "step": 4597
    },
    {
      "epoch": 0.6216348672536461,
      "grad_norm": 0.041755132377147675,
      "learning_rate": 1.066517828082548e-05,
      "loss": 0.027606487274169922,
      "step": 4598
    },
    {
      "epoch": 0.621770064049482,
      "grad_norm": 0.07600902020931244,
      "learning_rate": 1.0658626337264926e-05,
      "loss": 0.04816699028015137,
      "step": 4599
    },
    {
      "epoch": 0.6219052608453179,
      "grad_norm": 0.08347147703170776,
      "learning_rate": 1.0652075297601518e-05,
      "loss": 0.05012869834899902,
      "step": 4600
    },
    {
      "epoch": 0.6220404576411539,
      "grad_norm": 0.032536547631025314,
      "learning_rate": 1.0645525163199222e-05,
      "loss": 0.03523588180541992,
      "step": 4601
    },
    {
      "epoch": 0.6221756544369899,
      "grad_norm": 0.029898855835199356,
      "learning_rate": 1.063897593542181e-05,
      "loss": 0.039261817932128906,
      "step": 4602
    },
    {
      "epoch": 0.6223108512328258,
      "grad_norm": 0.08544182777404785,
      "learning_rate": 1.0632427615632864e-05,
      "loss": 0.05735349655151367,
      "step": 4603
    },
    {
      "epoch": 0.6224460480286618,
      "grad_norm": 0.10907492786645889,
      "learning_rate": 1.0625880205195776e-05,
      "loss": 0.06522655487060547,
      "step": 4604
    },
    {
      "epoch": 0.6225812448244976,
      "grad_norm": 0.07812304049730301,
      "learning_rate": 1.0619333705473754e-05,
      "loss": 0.06149721145629883,
      "step": 4605
    },
    {
      "epoch": 0.6227164416203336,
      "grad_norm": 0.06729117035865784,
      "learning_rate": 1.0612788117829821e-05,
      "loss": 0.06447267532348633,
      "step": 4606
    },
    {
      "epoch": 0.6228516384161695,
      "grad_norm": 0.12640045583248138,
      "learning_rate": 1.0606243443626792e-05,
      "loss": 0.06893444061279297,
      "step": 4607
    },
    {
      "epoch": 0.6229868352120055,
      "grad_norm": 0.0570698007941246,
      "learning_rate": 1.0599699684227313e-05,
      "loss": 0.03769737482070923,
      "step": 4608
    },
    {
      "epoch": 0.6231220320078414,
      "grad_norm": 0.08291788399219513,
      "learning_rate": 1.0593156840993818e-05,
      "loss": 0.07607030868530273,
      "step": 4609
    },
    {
      "epoch": 0.6232572288036774,
      "grad_norm": 0.12094594538211823,
      "learning_rate": 1.0586614915288571e-05,
      "loss": 0.06901097297668457,
      "step": 4610
    },
    {
      "epoch": 0.6233924255995132,
      "grad_norm": 0.12530755996704102,
      "learning_rate": 1.0580073908473641e-05,
      "loss": 0.09278249740600586,
      "step": 4611
    },
    {
      "epoch": 0.6235276223953492,
      "grad_norm": 0.06343226134777069,
      "learning_rate": 1.0573533821910885e-05,
      "loss": 0.05980968475341797,
      "step": 4612
    },
    {
      "epoch": 0.6236628191911852,
      "grad_norm": 0.05784863233566284,
      "learning_rate": 1.0566994656961997e-05,
      "loss": 0.04554760456085205,
      "step": 4613
    },
    {
      "epoch": 0.6237980159870211,
      "grad_norm": 0.07663862407207489,
      "learning_rate": 1.0560456414988456e-05,
      "loss": 0.08053302764892578,
      "step": 4614
    },
    {
      "epoch": 0.6239332127828571,
      "grad_norm": 0.043838586658239365,
      "learning_rate": 1.0553919097351564e-05,
      "loss": 0.03961610794067383,
      "step": 4615
    },
    {
      "epoch": 0.624068409578693,
      "grad_norm": 0.10038360953330994,
      "learning_rate": 1.0547382705412434e-05,
      "loss": 0.07228636741638184,
      "step": 4616
    },
    {
      "epoch": 0.6242036063745289,
      "grad_norm": 0.05642108991742134,
      "learning_rate": 1.054084724053196e-05,
      "loss": 0.044862300157547,
      "step": 4617
    },
    {
      "epoch": 0.6243388031703648,
      "grad_norm": 0.07895555347204208,
      "learning_rate": 1.0534312704070875e-05,
      "loss": 0.08383417129516602,
      "step": 4618
    },
    {
      "epoch": 0.6244739999662008,
      "grad_norm": 0.06675758957862854,
      "learning_rate": 1.0527779097389695e-05,
      "loss": 0.06104004383087158,
      "step": 4619
    },
    {
      "epoch": 0.6246091967620367,
      "grad_norm": 0.09647714346647263,
      "learning_rate": 1.0521246421848762e-05,
      "loss": 0.06209111213684082,
      "step": 4620
    },
    {
      "epoch": 0.6247443935578727,
      "grad_norm": 0.06091001629829407,
      "learning_rate": 1.0514714678808202e-05,
      "loss": 0.04690009355545044,
      "step": 4621
    },
    {
      "epoch": 0.6248795903537087,
      "grad_norm": 0.1445750892162323,
      "learning_rate": 1.0508183869627962e-05,
      "loss": 0.07294130325317383,
      "step": 4622
    },
    {
      "epoch": 0.6250147871495445,
      "grad_norm": 0.11366762965917587,
      "learning_rate": 1.0501653995667798e-05,
      "loss": 0.03745102882385254,
      "step": 4623
    },
    {
      "epoch": 0.6251499839453805,
      "grad_norm": 0.09455694258213043,
      "learning_rate": 1.0495125058287258e-05,
      "loss": 0.07582062482833862,
      "step": 4624
    },
    {
      "epoch": 0.6252851807412164,
      "grad_norm": 0.09008247405290604,
      "learning_rate": 1.0488597058845708e-05,
      "loss": 0.05620574951171875,
      "step": 4625
    },
    {
      "epoch": 0.6254203775370524,
      "grad_norm": 0.07328398525714874,
      "learning_rate": 1.0482069998702304e-05,
      "loss": 0.061090558767318726,
      "step": 4626
    },
    {
      "epoch": 0.6255555743328883,
      "grad_norm": 0.034278471022844315,
      "learning_rate": 1.0475543879216017e-05,
      "loss": 0.034066200256347656,
      "step": 4627
    },
    {
      "epoch": 0.6256907711287243,
      "grad_norm": 0.09474608302116394,
      "learning_rate": 1.0469018701745626e-05,
      "loss": 0.08124363422393799,
      "step": 4628
    },
    {
      "epoch": 0.6258259679245601,
      "grad_norm": 0.08546648919582367,
      "learning_rate": 1.0462494467649704e-05,
      "loss": 0.06053125858306885,
      "step": 4629
    },
    {
      "epoch": 0.6259611647203961,
      "grad_norm": 0.04647248983383179,
      "learning_rate": 1.045597117828663e-05,
      "loss": 0.03316211700439453,
      "step": 4630
    },
    {
      "epoch": 0.626096361516232,
      "grad_norm": 0.07093106210231781,
      "learning_rate": 1.0449448835014586e-05,
      "loss": 0.04970729351043701,
      "step": 4631
    },
    {
      "epoch": 0.626231558312068,
      "grad_norm": 0.1595800518989563,
      "learning_rate": 1.044292743919156e-05,
      "loss": 0.10310888290405273,
      "step": 4632
    },
    {
      "epoch": 0.626366755107904,
      "grad_norm": 0.05669156461954117,
      "learning_rate": 1.0436406992175343e-05,
      "loss": 0.05315113067626953,
      "step": 4633
    },
    {
      "epoch": 0.6265019519037399,
      "grad_norm": 0.07586748152971268,
      "learning_rate": 1.0429887495323532e-05,
      "loss": 0.05605113506317139,
      "step": 4634
    },
    {
      "epoch": 0.6266371486995759,
      "grad_norm": 0.08573278039693832,
      "learning_rate": 1.0423368949993512e-05,
      "loss": 0.07182550430297852,
      "step": 4635
    },
    {
      "epoch": 0.6267723454954117,
      "grad_norm": 0.05033598095178604,
      "learning_rate": 1.041685135754248e-05,
      "loss": 0.04836010932922363,
      "step": 4636
    },
    {
      "epoch": 0.6269075422912477,
      "grad_norm": 0.0657183974981308,
      "learning_rate": 1.0410334719327435e-05,
      "loss": 0.06773090362548828,
      "step": 4637
    },
    {
      "epoch": 0.6270427390870836,
      "grad_norm": 0.047336284071207047,
      "learning_rate": 1.0403819036705177e-05,
      "loss": 0.05188131332397461,
      "step": 4638
    },
    {
      "epoch": 0.6271779358829196,
      "grad_norm": 0.0654551237821579,
      "learning_rate": 1.0397304311032311e-05,
      "loss": 0.056011199951171875,
      "step": 4639
    },
    {
      "epoch": 0.6273131326787555,
      "grad_norm": 0.07031180709600449,
      "learning_rate": 1.039079054366523e-05,
      "loss": 0.051982879638671875,
      "step": 4640
    },
    {
      "epoch": 0.6274483294745915,
      "grad_norm": 0.10409411042928696,
      "learning_rate": 1.0384277735960133e-05,
      "loss": 0.05213010311126709,
      "step": 4641
    },
    {
      "epoch": 0.6275835262704273,
      "grad_norm": 0.09999171644449234,
      "learning_rate": 1.0377765889273025e-05,
      "loss": 0.049260616302490234,
      "step": 4642
    },
    {
      "epoch": 0.6277187230662633,
      "grad_norm": 0.09111247956752777,
      "learning_rate": 1.0371255004959715e-05,
      "loss": 0.042534828186035156,
      "step": 4643
    },
    {
      "epoch": 0.6278539198620993,
      "grad_norm": 0.052284471690654755,
      "learning_rate": 1.036474508437579e-05,
      "loss": 0.03229665756225586,
      "step": 4644
    },
    {
      "epoch": 0.6279891166579352,
      "grad_norm": 0.08931080996990204,
      "learning_rate": 1.035823612887666e-05,
      "loss": 0.05841207504272461,
      "step": 4645
    },
    {
      "epoch": 0.6281243134537712,
      "grad_norm": 0.06335139274597168,
      "learning_rate": 1.0351728139817517e-05,
      "loss": 0.04215601086616516,
      "step": 4646
    },
    {
      "epoch": 0.6282595102496071,
      "grad_norm": 0.04505191743373871,
      "learning_rate": 1.0345221118553362e-05,
      "loss": 0.04367262125015259,
      "step": 4647
    },
    {
      "epoch": 0.628394707045443,
      "grad_norm": 0.10204634815454483,
      "learning_rate": 1.0338715066439002e-05,
      "loss": 0.06146645545959473,
      "step": 4648
    },
    {
      "epoch": 0.6285299038412789,
      "grad_norm": 0.18705184757709503,
      "learning_rate": 1.0332209984829013e-05,
      "loss": 0.07594788074493408,
      "step": 4649
    },
    {
      "epoch": 0.6286651006371149,
      "grad_norm": 0.04279199615120888,
      "learning_rate": 1.03257058750778e-05,
      "loss": 0.04190528392791748,
      "step": 4650
    },
    {
      "epoch": 0.6288002974329508,
      "grad_norm": 0.12377715110778809,
      "learning_rate": 1.0319202738539548e-05,
      "loss": 0.056028127670288086,
      "step": 4651
    },
    {
      "epoch": 0.6289354942287868,
      "grad_norm": 0.04761936143040657,
      "learning_rate": 1.0312700576568253e-05,
      "loss": 0.05487704277038574,
      "step": 4652
    },
    {
      "epoch": 0.6290706910246228,
      "grad_norm": 0.18021810054779053,
      "learning_rate": 1.0306199390517688e-05,
      "loss": 0.08748221397399902,
      "step": 4653
    },
    {
      "epoch": 0.6292058878204586,
      "grad_norm": 0.06973965466022491,
      "learning_rate": 1.0299699181741439e-05,
      "loss": 0.05473470687866211,
      "step": 4654
    },
    {
      "epoch": 0.6293410846162946,
      "grad_norm": 0.06991768628358841,
      "learning_rate": 1.0293199951592889e-05,
      "loss": 0.058176517486572266,
      "step": 4655
    },
    {
      "epoch": 0.6294762814121305,
      "grad_norm": 0.07701300084590912,
      "learning_rate": 1.0286701701425206e-05,
      "loss": 0.06687712669372559,
      "step": 4656
    },
    {
      "epoch": 0.6296114782079665,
      "grad_norm": 0.08925246447324753,
      "learning_rate": 1.0280204432591369e-05,
      "loss": 0.07634258270263672,
      "step": 4657
    },
    {
      "epoch": 0.6297466750038024,
      "grad_norm": 0.05631409212946892,
      "learning_rate": 1.0273708146444133e-05,
      "loss": 0.07979965209960938,
      "step": 4658
    },
    {
      "epoch": 0.6298818717996384,
      "grad_norm": 0.06966651231050491,
      "learning_rate": 1.0267212844336062e-05,
      "loss": 0.07070302963256836,
      "step": 4659
    },
    {
      "epoch": 0.6300170685954742,
      "grad_norm": 0.10738211870193481,
      "learning_rate": 1.026071852761952e-05,
      "loss": 0.09098625183105469,
      "step": 4660
    },
    {
      "epoch": 0.6301522653913102,
      "grad_norm": 0.06925952434539795,
      "learning_rate": 1.025422519764665e-05,
      "loss": 0.043718695640563965,
      "step": 4661
    },
    {
      "epoch": 0.6302874621871462,
      "grad_norm": 0.11434145271778107,
      "learning_rate": 1.024773285576941e-05,
      "loss": 0.06419801712036133,
      "step": 4662
    },
    {
      "epoch": 0.6304226589829821,
      "grad_norm": 0.12941968441009521,
      "learning_rate": 1.0241241503339524e-05,
      "loss": 0.045635223388671875,
      "step": 4663
    },
    {
      "epoch": 0.6305578557788181,
      "grad_norm": 0.04785580933094025,
      "learning_rate": 1.023475114170853e-05,
      "loss": 0.059967994689941406,
      "step": 4664
    },
    {
      "epoch": 0.630693052574654,
      "grad_norm": 0.08671826124191284,
      "learning_rate": 1.0228261772227768e-05,
      "loss": 0.06833600997924805,
      "step": 4665
    },
    {
      "epoch": 0.6308282493704899,
      "grad_norm": 0.0740731880068779,
      "learning_rate": 1.0221773396248349e-05,
      "loss": 0.08113479614257812,
      "step": 4666
    },
    {
      "epoch": 0.6309634461663258,
      "grad_norm": 0.07166953384876251,
      "learning_rate": 1.021528601512119e-05,
      "loss": 0.0814390778541565,
      "step": 4667
    },
    {
      "epoch": 0.6310986429621618,
      "grad_norm": 0.049806226044893265,
      "learning_rate": 1.0208799630196994e-05,
      "loss": 0.052498817443847656,
      "step": 4668
    },
    {
      "epoch": 0.6312338397579977,
      "grad_norm": 0.047655995935201645,
      "learning_rate": 1.0202314242826264e-05,
      "loss": 0.035086870193481445,
      "step": 4669
    },
    {
      "epoch": 0.6313690365538337,
      "grad_norm": 0.09570282697677612,
      "learning_rate": 1.0195829854359299e-05,
      "loss": 0.0708608627319336,
      "step": 4670
    },
    {
      "epoch": 0.6315042333496697,
      "grad_norm": 0.13582776486873627,
      "learning_rate": 1.0189346466146175e-05,
      "loss": 0.08343559503555298,
      "step": 4671
    },
    {
      "epoch": 0.6316394301455055,
      "grad_norm": 0.08645089715719223,
      "learning_rate": 1.018286407953677e-05,
      "loss": 0.06045341491699219,
      "step": 4672
    },
    {
      "epoch": 0.6317746269413415,
      "grad_norm": 0.11129418760538101,
      "learning_rate": 1.017638269588075e-05,
      "loss": 0.062251150608062744,
      "step": 4673
    },
    {
      "epoch": 0.6319098237371774,
      "grad_norm": 0.19079650938510895,
      "learning_rate": 1.0169902316527575e-05,
      "loss": 0.0704948902130127,
      "step": 4674
    },
    {
      "epoch": 0.6320450205330134,
      "grad_norm": 0.06261169165372849,
      "learning_rate": 1.0163422942826502e-05,
      "loss": 0.047031402587890625,
      "step": 4675
    },
    {
      "epoch": 0.6321802173288493,
      "grad_norm": 0.16144797205924988,
      "learning_rate": 1.0156944576126555e-05,
      "loss": 0.06773519515991211,
      "step": 4676
    },
    {
      "epoch": 0.6323154141246853,
      "grad_norm": 0.04278594255447388,
      "learning_rate": 1.0150467217776579e-05,
      "loss": 0.03280520439147949,
      "step": 4677
    },
    {
      "epoch": 0.6324506109205212,
      "grad_norm": 0.05383704602718353,
      "learning_rate": 1.0143990869125185e-05,
      "loss": 0.05450177192687988,
      "step": 4678
    },
    {
      "epoch": 0.6325858077163571,
      "grad_norm": 0.05481462553143501,
      "learning_rate": 1.013751553152079e-05,
      "loss": 0.04588794708251953,
      "step": 4679
    },
    {
      "epoch": 0.632721004512193,
      "grad_norm": 0.12561596930027008,
      "learning_rate": 1.0131041206311594e-05,
      "loss": 0.06728601455688477,
      "step": 4680
    },
    {
      "epoch": 0.632856201308029,
      "grad_norm": 0.07858829200267792,
      "learning_rate": 1.0124567894845578e-05,
      "loss": 0.0653071403503418,
      "step": 4681
    },
    {
      "epoch": 0.632991398103865,
      "grad_norm": 0.12705333530902863,
      "learning_rate": 1.0118095598470528e-05,
      "loss": 0.06161785125732422,
      "step": 4682
    },
    {
      "epoch": 0.6331265948997009,
      "grad_norm": 0.20888195931911469,
      "learning_rate": 1.0111624318534006e-05,
      "loss": 0.08056330680847168,
      "step": 4683
    },
    {
      "epoch": 0.6332617916955369,
      "grad_norm": 0.07739520817995071,
      "learning_rate": 1.0105154056383377e-05,
      "loss": 0.08420372009277344,
      "step": 4684
    },
    {
      "epoch": 0.6333969884913727,
      "grad_norm": 0.07199637591838837,
      "learning_rate": 1.0098684813365764e-05,
      "loss": 0.060961246490478516,
      "step": 4685
    },
    {
      "epoch": 0.6335321852872087,
      "grad_norm": 0.10835880786180496,
      "learning_rate": 1.0092216590828115e-05,
      "loss": 0.050191402435302734,
      "step": 4686
    },
    {
      "epoch": 0.6336673820830446,
      "grad_norm": 0.07479794323444366,
      "learning_rate": 1.0085749390117146e-05,
      "loss": 0.07352566719055176,
      "step": 4687
    },
    {
      "epoch": 0.6338025788788806,
      "grad_norm": 0.08301626145839691,
      "learning_rate": 1.0079283212579354e-05,
      "loss": 0.06337213516235352,
      "step": 4688
    },
    {
      "epoch": 0.6339377756747165,
      "grad_norm": 0.1601741462945938,
      "learning_rate": 1.0072818059561045e-05,
      "loss": 0.05661177635192871,
      "step": 4689
    },
    {
      "epoch": 0.6340729724705525,
      "grad_norm": 0.17615368962287903,
      "learning_rate": 1.0066353932408285e-05,
      "loss": 0.09593796730041504,
      "step": 4690
    },
    {
      "epoch": 0.6342081692663883,
      "grad_norm": 0.0910477340221405,
      "learning_rate": 1.0059890832466948e-05,
      "loss": 0.04729163646697998,
      "step": 4691
    },
    {
      "epoch": 0.6343433660622243,
      "grad_norm": 0.12613417208194733,
      "learning_rate": 1.0053428761082684e-05,
      "loss": 0.07843419909477234,
      "step": 4692
    },
    {
      "epoch": 0.6344785628580603,
      "grad_norm": 0.08855003118515015,
      "learning_rate": 1.0046967719600927e-05,
      "loss": 0.04972577095031738,
      "step": 4693
    },
    {
      "epoch": 0.6346137596538962,
      "grad_norm": 0.06535353511571884,
      "learning_rate": 1.0040507709366912e-05,
      "loss": 0.06455039978027344,
      "step": 4694
    },
    {
      "epoch": 0.6347489564497322,
      "grad_norm": 0.13490630686283112,
      "learning_rate": 1.0034048731725631e-05,
      "loss": 0.056884169578552246,
      "step": 4695
    },
    {
      "epoch": 0.6348841532455681,
      "grad_norm": 0.057094454765319824,
      "learning_rate": 1.0027590788021886e-05,
      "loss": 0.06293630599975586,
      "step": 4696
    },
    {
      "epoch": 0.635019350041404,
      "grad_norm": 0.08362656831741333,
      "learning_rate": 1.0021133879600258e-05,
      "loss": 0.05770599842071533,
      "step": 4697
    },
    {
      "epoch": 0.6351545468372399,
      "grad_norm": 0.04646025970578194,
      "learning_rate": 1.0014678007805108e-05,
      "loss": 0.05965447425842285,
      "step": 4698
    },
    {
      "epoch": 0.6352897436330759,
      "grad_norm": 0.08744431287050247,
      "learning_rate": 1.0008223173980579e-05,
      "loss": 0.06365203857421875,
      "step": 4699
    },
    {
      "epoch": 0.6354249404289118,
      "grad_norm": 0.05192157253623009,
      "learning_rate": 1.0001769379470604e-05,
      "loss": 0.03719377517700195,
      "step": 4700
    },
    {
      "epoch": 0.6355601372247478,
      "grad_norm": 0.05070551484823227,
      "learning_rate": 9.995316625618898e-06,
      "loss": 0.06077384948730469,
      "step": 4701
    },
    {
      "epoch": 0.6356953340205838,
      "grad_norm": 0.16018208861351013,
      "learning_rate": 9.988864913768962e-06,
      "loss": 0.07527637481689453,
      "step": 4702
    },
    {
      "epoch": 0.6358305308164196,
      "grad_norm": 0.0906892940402031,
      "learning_rate": 9.982414245264071e-06,
      "loss": 0.07805728912353516,
      "step": 4703
    },
    {
      "epoch": 0.6359657276122556,
      "grad_norm": 0.05629304423928261,
      "learning_rate": 9.975964621447293e-06,
      "loss": 0.032604217529296875,
      "step": 4704
    },
    {
      "epoch": 0.6361009244080915,
      "grad_norm": 0.07576316595077515,
      "learning_rate": 9.96951604366147e-06,
      "loss": 0.059985995292663574,
      "step": 4705
    },
    {
      "epoch": 0.6362361212039275,
      "grad_norm": 0.051404230296611786,
      "learning_rate": 9.963068513249233e-06,
      "loss": 0.06557893753051758,
      "step": 4706
    },
    {
      "epoch": 0.6363713179997634,
      "grad_norm": 0.08788328617811203,
      "learning_rate": 9.956622031552996e-06,
      "loss": 0.07783031463623047,
      "step": 4707
    },
    {
      "epoch": 0.6365065147955994,
      "grad_norm": 0.06323837488889694,
      "learning_rate": 9.950176599914942e-06,
      "loss": 0.0667886734008789,
      "step": 4708
    },
    {
      "epoch": 0.6366417115914352,
      "grad_norm": 0.09150069952011108,
      "learning_rate": 9.943732219677048e-06,
      "loss": 0.05652618408203125,
      "step": 4709
    },
    {
      "epoch": 0.6367769083872712,
      "grad_norm": 0.0968921035528183,
      "learning_rate": 9.93728889218107e-06,
      "loss": 0.05686187744140625,
      "step": 4710
    },
    {
      "epoch": 0.6369121051831071,
      "grad_norm": 0.08614809066057205,
      "learning_rate": 9.930846618768543e-06,
      "loss": 0.07626223564147949,
      "step": 4711
    },
    {
      "epoch": 0.6370473019789431,
      "grad_norm": 0.04705547168850899,
      "learning_rate": 9.924405400780784e-06,
      "loss": 0.05328822135925293,
      "step": 4712
    },
    {
      "epoch": 0.6371824987747791,
      "grad_norm": 0.05670127645134926,
      "learning_rate": 9.917965239558885e-06,
      "loss": 0.05498456954956055,
      "step": 4713
    },
    {
      "epoch": 0.637317695570615,
      "grad_norm": 0.10341071337461472,
      "learning_rate": 9.911526136443726e-06,
      "loss": 0.07686924934387207,
      "step": 4714
    },
    {
      "epoch": 0.6374528923664509,
      "grad_norm": 0.11412568390369415,
      "learning_rate": 9.905088092775956e-06,
      "loss": 0.08795738220214844,
      "step": 4715
    },
    {
      "epoch": 0.6375880891622868,
      "grad_norm": 0.1375456154346466,
      "learning_rate": 9.898651109896015e-06,
      "loss": 0.07765698432922363,
      "step": 4716
    },
    {
      "epoch": 0.6377232859581228,
      "grad_norm": 0.0671454593539238,
      "learning_rate": 9.892215189144123e-06,
      "loss": 0.03809666633605957,
      "step": 4717
    },
    {
      "epoch": 0.6378584827539587,
      "grad_norm": 0.03875022381544113,
      "learning_rate": 9.88578033186026e-06,
      "loss": 0.036783814430236816,
      "step": 4718
    },
    {
      "epoch": 0.6379936795497947,
      "grad_norm": 0.09829258173704147,
      "learning_rate": 9.879346539384207e-06,
      "loss": 0.05499982833862305,
      "step": 4719
    },
    {
      "epoch": 0.6381288763456306,
      "grad_norm": 0.07232803851366043,
      "learning_rate": 9.87291381305551e-06,
      "loss": 0.037989139556884766,
      "step": 4720
    },
    {
      "epoch": 0.6382640731414665,
      "grad_norm": 0.12916040420532227,
      "learning_rate": 9.866482154213502e-06,
      "loss": 0.0659339427947998,
      "step": 4721
    },
    {
      "epoch": 0.6383992699373024,
      "grad_norm": 0.04350670799612999,
      "learning_rate": 9.86005156419728e-06,
      "loss": 0.03740239143371582,
      "step": 4722
    },
    {
      "epoch": 0.6385344667331384,
      "grad_norm": 0.09592660516500473,
      "learning_rate": 9.853622044345732e-06,
      "loss": 0.08411908149719238,
      "step": 4723
    },
    {
      "epoch": 0.6386696635289744,
      "grad_norm": 0.09439944475889206,
      "learning_rate": 9.847193595997522e-06,
      "loss": 0.08109784126281738,
      "step": 4724
    },
    {
      "epoch": 0.6388048603248103,
      "grad_norm": 0.07170064002275467,
      "learning_rate": 9.840766220491078e-06,
      "loss": 0.07858061790466309,
      "step": 4725
    },
    {
      "epoch": 0.6389400571206463,
      "grad_norm": 0.0726851373910904,
      "learning_rate": 9.834339919164625e-06,
      "loss": 0.07430505752563477,
      "step": 4726
    },
    {
      "epoch": 0.6390752539164822,
      "grad_norm": 0.09529086947441101,
      "learning_rate": 9.827914693356145e-06,
      "loss": 0.06363117694854736,
      "step": 4727
    },
    {
      "epoch": 0.6392104507123181,
      "grad_norm": 0.054350197315216064,
      "learning_rate": 9.821490544403403e-06,
      "loss": 0.05079984664916992,
      "step": 4728
    },
    {
      "epoch": 0.639345647508154,
      "grad_norm": 0.07870441675186157,
      "learning_rate": 9.815067473643951e-06,
      "loss": 0.05780029296875,
      "step": 4729
    },
    {
      "epoch": 0.63948084430399,
      "grad_norm": 0.0771501362323761,
      "learning_rate": 9.808645482415097e-06,
      "loss": 0.06372389942407608,
      "step": 4730
    },
    {
      "epoch": 0.639616041099826,
      "grad_norm": 0.05503477156162262,
      "learning_rate": 9.80222457205394e-06,
      "loss": 0.05171859264373779,
      "step": 4731
    },
    {
      "epoch": 0.6397512378956619,
      "grad_norm": 0.3246934413909912,
      "learning_rate": 9.795804743897341e-06,
      "loss": 0.06792831420898438,
      "step": 4732
    },
    {
      "epoch": 0.6398864346914979,
      "grad_norm": 0.09514765441417694,
      "learning_rate": 9.789385999281948e-06,
      "loss": 0.06976509094238281,
      "step": 4733
    },
    {
      "epoch": 0.6400216314873337,
      "grad_norm": 0.07772054523229599,
      "learning_rate": 9.782968339544179e-06,
      "loss": 0.05619454383850098,
      "step": 4734
    },
    {
      "epoch": 0.6401568282831697,
      "grad_norm": 0.07579687237739563,
      "learning_rate": 9.776551766020219e-06,
      "loss": 0.04684770107269287,
      "step": 4735
    },
    {
      "epoch": 0.6402920250790056,
      "grad_norm": 0.0779886320233345,
      "learning_rate": 9.77013628004604e-06,
      "loss": 0.05618441104888916,
      "step": 4736
    },
    {
      "epoch": 0.6404272218748416,
      "grad_norm": 0.06135810911655426,
      "learning_rate": 9.763721882957371e-06,
      "loss": 0.042951107025146484,
      "step": 4737
    },
    {
      "epoch": 0.6405624186706775,
      "grad_norm": 0.08087282627820969,
      "learning_rate": 9.757308576089732e-06,
      "loss": 0.0594707727432251,
      "step": 4738
    },
    {
      "epoch": 0.6406976154665135,
      "grad_norm": 0.033737912774086,
      "learning_rate": 9.750896360778404e-06,
      "loss": 0.032074689865112305,
      "step": 4739
    },
    {
      "epoch": 0.6408328122623493,
      "grad_norm": 0.04383016377687454,
      "learning_rate": 9.744485238358448e-06,
      "loss": 0.06212806701660156,
      "step": 4740
    },
    {
      "epoch": 0.6409680090581853,
      "grad_norm": 0.0744631439447403,
      "learning_rate": 9.73807521016469e-06,
      "loss": 0.054350078105926514,
      "step": 4741
    },
    {
      "epoch": 0.6411032058540213,
      "grad_norm": 0.106045201420784,
      "learning_rate": 9.731666277531732e-06,
      "loss": 0.05135154724121094,
      "step": 4742
    },
    {
      "epoch": 0.6412384026498572,
      "grad_norm": 0.11143618077039719,
      "learning_rate": 9.725258441793947e-06,
      "loss": 0.060022830963134766,
      "step": 4743
    },
    {
      "epoch": 0.6413735994456932,
      "grad_norm": 0.059104084968566895,
      "learning_rate": 9.71885170428549e-06,
      "loss": 0.04802215099334717,
      "step": 4744
    },
    {
      "epoch": 0.6415087962415291,
      "grad_norm": 0.11476949602365494,
      "learning_rate": 9.712446066340265e-06,
      "loss": 0.06240034103393555,
      "step": 4745
    },
    {
      "epoch": 0.641643993037365,
      "grad_norm": 0.06269510835409164,
      "learning_rate": 9.70604152929197e-06,
      "loss": 0.04493749141693115,
      "step": 4746
    },
    {
      "epoch": 0.6417791898332009,
      "grad_norm": 0.059973087161779404,
      "learning_rate": 9.699638094474054e-06,
      "loss": 0.0621793270111084,
      "step": 4747
    },
    {
      "epoch": 0.6419143866290369,
      "grad_norm": 0.04422726109623909,
      "learning_rate": 9.693235763219752e-06,
      "loss": 0.04668474197387695,
      "step": 4748
    },
    {
      "epoch": 0.6420495834248728,
      "grad_norm": 0.07597751170396805,
      "learning_rate": 9.68683453686207e-06,
      "loss": 0.0742950439453125,
      "step": 4749
    },
    {
      "epoch": 0.6421847802207088,
      "grad_norm": 0.0901431292295456,
      "learning_rate": 9.680434416733763e-06,
      "loss": 0.06837058067321777,
      "step": 4750
    },
    {
      "epoch": 0.6423199770165448,
      "grad_norm": 0.11406173557043076,
      "learning_rate": 9.674035404167381e-06,
      "loss": 0.04448425769805908,
      "step": 4751
    },
    {
      "epoch": 0.6424551738123806,
      "grad_norm": 0.09654321521520615,
      "learning_rate": 9.66763750049523e-06,
      "loss": 0.04701948165893555,
      "step": 4752
    },
    {
      "epoch": 0.6425903706082166,
      "grad_norm": 0.060939475893974304,
      "learning_rate": 9.66124070704939e-06,
      "loss": 0.05481863021850586,
      "step": 4753
    },
    {
      "epoch": 0.6427255674040525,
      "grad_norm": 0.12973052263259888,
      "learning_rate": 9.654845025161699e-06,
      "loss": 0.06844782829284668,
      "step": 4754
    },
    {
      "epoch": 0.6428607641998885,
      "grad_norm": 0.06413166970014572,
      "learning_rate": 9.648450456163777e-06,
      "loss": 0.04809761047363281,
      "step": 4755
    },
    {
      "epoch": 0.6429959609957244,
      "grad_norm": 0.05641203373670578,
      "learning_rate": 9.64205700138701e-06,
      "loss": 0.05434608459472656,
      "step": 4756
    },
    {
      "epoch": 0.6431311577915604,
      "grad_norm": 0.062255438417196274,
      "learning_rate": 9.635664662162548e-06,
      "loss": 0.06367373466491699,
      "step": 4757
    },
    {
      "epoch": 0.6432663545873962,
      "grad_norm": 0.19724592566490173,
      "learning_rate": 9.629273439821315e-06,
      "loss": 0.05742764472961426,
      "step": 4758
    },
    {
      "epoch": 0.6434015513832322,
      "grad_norm": 0.10318677127361298,
      "learning_rate": 9.622883335693984e-06,
      "loss": 0.07251644134521484,
      "step": 4759
    },
    {
      "epoch": 0.6435367481790681,
      "grad_norm": 0.1288115382194519,
      "learning_rate": 9.616494351111017e-06,
      "loss": 0.06653308868408203,
      "step": 4760
    },
    {
      "epoch": 0.6436719449749041,
      "grad_norm": 0.03664591163396835,
      "learning_rate": 9.610106487402637e-06,
      "loss": 0.04697155952453613,
      "step": 4761
    },
    {
      "epoch": 0.64380714177074,
      "grad_norm": 0.04664827510714531,
      "learning_rate": 9.603719745898826e-06,
      "loss": 0.047150611877441406,
      "step": 4762
    },
    {
      "epoch": 0.643942338566576,
      "grad_norm": 0.08848141878843307,
      "learning_rate": 9.597334127929346e-06,
      "loss": 0.07023659348487854,
      "step": 4763
    },
    {
      "epoch": 0.6440775353624119,
      "grad_norm": 0.09126049280166626,
      "learning_rate": 9.590949634823707e-06,
      "loss": 0.07262587547302246,
      "step": 4764
    },
    {
      "epoch": 0.6442127321582478,
      "grad_norm": 0.12153051048517227,
      "learning_rate": 9.584566267911198e-06,
      "loss": 0.05871462821960449,
      "step": 4765
    },
    {
      "epoch": 0.6443479289540838,
      "grad_norm": 0.1530318707227707,
      "learning_rate": 9.578184028520874e-06,
      "loss": 0.05263471603393555,
      "step": 4766
    },
    {
      "epoch": 0.6444831257499197,
      "grad_norm": 0.06433980911970139,
      "learning_rate": 9.571802917981548e-06,
      "loss": 0.05490922927856445,
      "step": 4767
    },
    {
      "epoch": 0.6446183225457557,
      "grad_norm": 0.050201933830976486,
      "learning_rate": 9.565422937621798e-06,
      "loss": 0.03552103042602539,
      "step": 4768
    },
    {
      "epoch": 0.6447535193415916,
      "grad_norm": 0.07437662780284882,
      "learning_rate": 9.559044088769971e-06,
      "loss": 0.07383871078491211,
      "step": 4769
    },
    {
      "epoch": 0.6448887161374276,
      "grad_norm": 0.05894550308585167,
      "learning_rate": 9.552666372754182e-06,
      "loss": 0.05084085464477539,
      "step": 4770
    },
    {
      "epoch": 0.6450239129332634,
      "grad_norm": 0.05779692530632019,
      "learning_rate": 9.546289790902307e-06,
      "loss": 0.03959548473358154,
      "step": 4771
    },
    {
      "epoch": 0.6451591097290994,
      "grad_norm": 0.09428199380636215,
      "learning_rate": 9.539914344541976e-06,
      "loss": 0.07280898094177246,
      "step": 4772
    },
    {
      "epoch": 0.6452943065249354,
      "grad_norm": 0.0767393484711647,
      "learning_rate": 9.533540035000598e-06,
      "loss": 0.055715084075927734,
      "step": 4773
    },
    {
      "epoch": 0.6454295033207713,
      "grad_norm": 0.07694047689437866,
      "learning_rate": 9.52716686360533e-06,
      "loss": 0.0643010139465332,
      "step": 4774
    },
    {
      "epoch": 0.6455647001166073,
      "grad_norm": 0.09121976047754288,
      "learning_rate": 9.520794831683108e-06,
      "loss": 0.0471339225769043,
      "step": 4775
    },
    {
      "epoch": 0.6456998969124432,
      "grad_norm": 0.08535441756248474,
      "learning_rate": 9.514423940560627e-06,
      "loss": 0.07198810577392578,
      "step": 4776
    },
    {
      "epoch": 0.6458350937082791,
      "grad_norm": 0.17966727912425995,
      "learning_rate": 9.508054191564326e-06,
      "loss": 0.07330751419067383,
      "step": 4777
    },
    {
      "epoch": 0.645970290504115,
      "grad_norm": 0.049681130796670914,
      "learning_rate": 9.501685586020434e-06,
      "loss": 0.055103421211242676,
      "step": 4778
    },
    {
      "epoch": 0.646105487299951,
      "grad_norm": 0.07365203648805618,
      "learning_rate": 9.495318125254919e-06,
      "loss": 0.06869173049926758,
      "step": 4779
    },
    {
      "epoch": 0.6462406840957869,
      "grad_norm": 0.15162532031536102,
      "learning_rate": 9.488951810593527e-06,
      "loss": 0.0923604965209961,
      "step": 4780
    },
    {
      "epoch": 0.6463758808916229,
      "grad_norm": 0.04366796091198921,
      "learning_rate": 9.48258664336176e-06,
      "loss": 0.04380369186401367,
      "step": 4781
    },
    {
      "epoch": 0.6465110776874589,
      "grad_norm": 0.1019766628742218,
      "learning_rate": 9.476222624884873e-06,
      "loss": 0.05139732360839844,
      "step": 4782
    },
    {
      "epoch": 0.6466462744832947,
      "grad_norm": 0.11347633600234985,
      "learning_rate": 9.469859756487893e-06,
      "loss": 0.06167197227478027,
      "step": 4783
    },
    {
      "epoch": 0.6467814712791307,
      "grad_norm": 0.11291565746068954,
      "learning_rate": 9.463498039495598e-06,
      "loss": 0.06554603576660156,
      "step": 4784
    },
    {
      "epoch": 0.6469166680749666,
      "grad_norm": 0.21300140023231506,
      "learning_rate": 9.457137475232537e-06,
      "loss": 0.08475494384765625,
      "step": 4785
    },
    {
      "epoch": 0.6470518648708026,
      "grad_norm": 0.1279429942369461,
      "learning_rate": 9.450778065023019e-06,
      "loss": 0.06005287170410156,
      "step": 4786
    },
    {
      "epoch": 0.6471870616666385,
      "grad_norm": 0.05140487477183342,
      "learning_rate": 9.444419810191091e-06,
      "loss": 0.0693502426147461,
      "step": 4787
    },
    {
      "epoch": 0.6473222584624745,
      "grad_norm": 0.10644088685512543,
      "learning_rate": 9.43806271206059e-06,
      "loss": 0.06517815589904785,
      "step": 4788
    },
    {
      "epoch": 0.6474574552583103,
      "grad_norm": 0.07961923629045486,
      "learning_rate": 9.431706771955089e-06,
      "loss": 0.05404472351074219,
      "step": 4789
    },
    {
      "epoch": 0.6475926520541463,
      "grad_norm": 0.07635562121868134,
      "learning_rate": 9.425351991197937e-06,
      "loss": 0.08744347095489502,
      "step": 4790
    },
    {
      "epoch": 0.6477278488499822,
      "grad_norm": 0.16630281507968903,
      "learning_rate": 9.418998371112221e-06,
      "loss": 0.05486869812011719,
      "step": 4791
    },
    {
      "epoch": 0.6478630456458182,
      "grad_norm": 0.10429415106773376,
      "learning_rate": 9.412645913020807e-06,
      "loss": 0.07294023036956787,
      "step": 4792
    },
    {
      "epoch": 0.6479982424416542,
      "grad_norm": 0.07620232552289963,
      "learning_rate": 9.406294618246313e-06,
      "loss": 0.06562268733978271,
      "step": 4793
    },
    {
      "epoch": 0.6481334392374901,
      "grad_norm": 0.10241834074258804,
      "learning_rate": 9.399944488111103e-06,
      "loss": 0.03882145881652832,
      "step": 4794
    },
    {
      "epoch": 0.648268636033326,
      "grad_norm": 0.07540863007307053,
      "learning_rate": 9.39359552393732e-06,
      "loss": 0.056829214096069336,
      "step": 4795
    },
    {
      "epoch": 0.6484038328291619,
      "grad_norm": 0.06442712992429733,
      "learning_rate": 9.387247727046845e-06,
      "loss": 0.06419014930725098,
      "step": 4796
    },
    {
      "epoch": 0.6485390296249979,
      "grad_norm": 0.07219398766756058,
      "learning_rate": 9.380901098761319e-06,
      "loss": 0.05280756950378418,
      "step": 4797
    },
    {
      "epoch": 0.6486742264208338,
      "grad_norm": 0.0553305558860302,
      "learning_rate": 9.374555640402153e-06,
      "loss": 0.05795645713806152,
      "step": 4798
    },
    {
      "epoch": 0.6488094232166698,
      "grad_norm": 0.18707898259162903,
      "learning_rate": 9.368211353290503e-06,
      "loss": 0.07580041885375977,
      "step": 4799
    },
    {
      "epoch": 0.6489446200125057,
      "grad_norm": 0.045232661068439484,
      "learning_rate": 9.36186823874728e-06,
      "loss": 0.03824043273925781,
      "step": 4800
    },
    {
      "epoch": 0.6490798168083416,
      "grad_norm": 0.10402955114841461,
      "learning_rate": 9.355526298093152e-06,
      "loss": 0.05919694900512695,
      "step": 4801
    },
    {
      "epoch": 0.6492150136041775,
      "grad_norm": 0.10305744409561157,
      "learning_rate": 9.34918553264855e-06,
      "loss": 0.07936334609985352,
      "step": 4802
    },
    {
      "epoch": 0.6493502104000135,
      "grad_norm": 0.08143419772386551,
      "learning_rate": 9.342845943733658e-06,
      "loss": 0.07880687713623047,
      "step": 4803
    },
    {
      "epoch": 0.6494854071958495,
      "grad_norm": 0.07584871351718903,
      "learning_rate": 9.336507532668407e-06,
      "loss": 0.05534029006958008,
      "step": 4804
    },
    {
      "epoch": 0.6496206039916854,
      "grad_norm": 0.08206778019666672,
      "learning_rate": 9.33017030077249e-06,
      "loss": 0.06379318237304688,
      "step": 4805
    },
    {
      "epoch": 0.6497558007875214,
      "grad_norm": 0.116139255464077,
      "learning_rate": 9.323834249365346e-06,
      "loss": 0.04954695701599121,
      "step": 4806
    },
    {
      "epoch": 0.6498909975833572,
      "grad_norm": 0.1587880253791809,
      "learning_rate": 9.317499379766183e-06,
      "loss": 0.07483458518981934,
      "step": 4807
    },
    {
      "epoch": 0.6500261943791932,
      "grad_norm": 0.1243368610739708,
      "learning_rate": 9.311165693293954e-06,
      "loss": 0.06823205947875977,
      "step": 4808
    },
    {
      "epoch": 0.6501613911750291,
      "grad_norm": 0.05966208875179291,
      "learning_rate": 9.304833191267364e-06,
      "loss": 0.054343223571777344,
      "step": 4809
    },
    {
      "epoch": 0.6502965879708651,
      "grad_norm": 0.10272936522960663,
      "learning_rate": 9.298501875004874e-06,
      "loss": 0.0591428279876709,
      "step": 4810
    },
    {
      "epoch": 0.650431784766701,
      "grad_norm": 0.0742470920085907,
      "learning_rate": 9.292171745824695e-06,
      "loss": 0.05108910799026489,
      "step": 4811
    },
    {
      "epoch": 0.650566981562537,
      "grad_norm": 0.07919621467590332,
      "learning_rate": 9.285842805044797e-06,
      "loss": 0.07143473625183105,
      "step": 4812
    },
    {
      "epoch": 0.6507021783583729,
      "grad_norm": 0.06798703223466873,
      "learning_rate": 9.279515053982905e-06,
      "loss": 0.054782867431640625,
      "step": 4813
    },
    {
      "epoch": 0.6508373751542088,
      "grad_norm": 0.10217496752738953,
      "learning_rate": 9.273188493956476e-06,
      "loss": 0.043814897537231445,
      "step": 4814
    },
    {
      "epoch": 0.6509725719500448,
      "grad_norm": 0.08980832248926163,
      "learning_rate": 9.266863126282746e-06,
      "loss": 0.060454368591308594,
      "step": 4815
    },
    {
      "epoch": 0.6511077687458807,
      "grad_norm": 0.09616213291883469,
      "learning_rate": 9.260538952278683e-06,
      "loss": 0.03709614276885986,
      "step": 4816
    },
    {
      "epoch": 0.6512429655417167,
      "grad_norm": 0.13874880969524384,
      "learning_rate": 9.254215973261014e-06,
      "loss": 0.09170341491699219,
      "step": 4817
    },
    {
      "epoch": 0.6513781623375526,
      "grad_norm": 0.2109307050704956,
      "learning_rate": 9.247894190546228e-06,
      "loss": 0.06992048025131226,
      "step": 4818
    },
    {
      "epoch": 0.6515133591333886,
      "grad_norm": 0.12446895986795425,
      "learning_rate": 9.241573605450539e-06,
      "loss": 0.07599878311157227,
      "step": 4819
    },
    {
      "epoch": 0.6516485559292244,
      "grad_norm": 0.06491612643003464,
      "learning_rate": 9.235254219289937e-06,
      "loss": 0.04820096492767334,
      "step": 4820
    },
    {
      "epoch": 0.6517837527250604,
      "grad_norm": 0.07671085000038147,
      "learning_rate": 9.228936033380143e-06,
      "loss": 0.07025575637817383,
      "step": 4821
    },
    {
      "epoch": 0.6519189495208964,
      "grad_norm": 0.08806153386831284,
      "learning_rate": 9.222619049036649e-06,
      "loss": 0.0580744743347168,
      "step": 4822
    },
    {
      "epoch": 0.6520541463167323,
      "grad_norm": 0.09657975286245346,
      "learning_rate": 9.216303267574674e-06,
      "loss": 0.07407474517822266,
      "step": 4823
    },
    {
      "epoch": 0.6521893431125683,
      "grad_norm": 0.05526379495859146,
      "learning_rate": 9.209988690309198e-06,
      "loss": 0.050994873046875,
      "step": 4824
    },
    {
      "epoch": 0.6523245399084042,
      "grad_norm": 0.12516328692436218,
      "learning_rate": 9.203675318554956e-06,
      "loss": 0.0677499771118164,
      "step": 4825
    },
    {
      "epoch": 0.6524597367042401,
      "grad_norm": 0.04505413398146629,
      "learning_rate": 9.19736315362642e-06,
      "loss": 0.05249142646789551,
      "step": 4826
    },
    {
      "epoch": 0.652594933500076,
      "grad_norm": 0.10846035927534103,
      "learning_rate": 9.191052196837825e-06,
      "loss": 0.05478870868682861,
      "step": 4827
    },
    {
      "epoch": 0.652730130295912,
      "grad_norm": 0.1005919873714447,
      "learning_rate": 9.184742449503135e-06,
      "loss": 0.05190730094909668,
      "step": 4828
    },
    {
      "epoch": 0.6528653270917479,
      "grad_norm": 0.08012963831424713,
      "learning_rate": 9.178433912936077e-06,
      "loss": 0.0807640552520752,
      "step": 4829
    },
    {
      "epoch": 0.6530005238875839,
      "grad_norm": 0.03739846125245094,
      "learning_rate": 9.172126588450125e-06,
      "loss": 0.035629332065582275,
      "step": 4830
    },
    {
      "epoch": 0.6531357206834199,
      "grad_norm": 0.07629495114088058,
      "learning_rate": 9.165820477358491e-06,
      "loss": 0.052832841873168945,
      "step": 4831
    },
    {
      "epoch": 0.6532709174792557,
      "grad_norm": 0.05624593794345856,
      "learning_rate": 9.159515580974154e-06,
      "loss": 0.04668146371841431,
      "step": 4832
    },
    {
      "epoch": 0.6534061142750917,
      "grad_norm": 0.09051258116960526,
      "learning_rate": 9.15321190060981e-06,
      "loss": 0.05543529987335205,
      "step": 4833
    },
    {
      "epoch": 0.6535413110709276,
      "grad_norm": 0.08451498299837112,
      "learning_rate": 9.14690943757793e-06,
      "loss": 0.05232357978820801,
      "step": 4834
    },
    {
      "epoch": 0.6536765078667636,
      "grad_norm": 0.07455456256866455,
      "learning_rate": 9.14060819319072e-06,
      "loss": 0.0569191575050354,
      "step": 4835
    },
    {
      "epoch": 0.6538117046625995,
      "grad_norm": 0.06281017512083054,
      "learning_rate": 9.134308168760127e-06,
      "loss": 0.051880478858947754,
      "step": 4836
    },
    {
      "epoch": 0.6539469014584355,
      "grad_norm": 0.08092804998159409,
      "learning_rate": 9.128009365597854e-06,
      "loss": 0.0421660840511322,
      "step": 4837
    },
    {
      "epoch": 0.6540820982542713,
      "grad_norm": 0.06607811152935028,
      "learning_rate": 9.121711785015342e-06,
      "loss": 0.04873204231262207,
      "step": 4838
    },
    {
      "epoch": 0.6542172950501073,
      "grad_norm": 0.044848911464214325,
      "learning_rate": 9.115415428323787e-06,
      "loss": 0.042281270027160645,
      "step": 4839
    },
    {
      "epoch": 0.6543524918459432,
      "grad_norm": 0.11027204990386963,
      "learning_rate": 9.109120296834118e-06,
      "loss": 0.08812141418457031,
      "step": 4840
    },
    {
      "epoch": 0.6544876886417792,
      "grad_norm": 0.09283500164747238,
      "learning_rate": 9.10282639185702e-06,
      "loss": 0.07068729400634766,
      "step": 4841
    },
    {
      "epoch": 0.6546228854376152,
      "grad_norm": 0.09839151054620743,
      "learning_rate": 9.096533714702913e-06,
      "loss": 0.060366153717041016,
      "step": 4842
    },
    {
      "epoch": 0.6547580822334511,
      "grad_norm": 0.06751798093318939,
      "learning_rate": 9.090242266681967e-06,
      "loss": 0.05329442024230957,
      "step": 4843
    },
    {
      "epoch": 0.654893279029287,
      "grad_norm": 0.0747271478176117,
      "learning_rate": 9.083952049104094e-06,
      "loss": 0.05246615409851074,
      "step": 4844
    },
    {
      "epoch": 0.6550284758251229,
      "grad_norm": 0.1263909637928009,
      "learning_rate": 9.07766306327896e-06,
      "loss": 0.0943707823753357,
      "step": 4845
    },
    {
      "epoch": 0.6551636726209589,
      "grad_norm": 0.08197560161352158,
      "learning_rate": 9.071375310515949e-06,
      "loss": 0.060854434967041016,
      "step": 4846
    },
    {
      "epoch": 0.6552988694167948,
      "grad_norm": 0.06990229338407516,
      "learning_rate": 9.065088792124219e-06,
      "loss": 0.03728199005126953,
      "step": 4847
    },
    {
      "epoch": 0.6554340662126308,
      "grad_norm": 0.093485027551651,
      "learning_rate": 9.058803509412647e-06,
      "loss": 0.07423925399780273,
      "step": 4848
    },
    {
      "epoch": 0.6555692630084667,
      "grad_norm": 0.09205955266952515,
      "learning_rate": 9.05251946368987e-06,
      "loss": 0.05419516563415527,
      "step": 4849
    },
    {
      "epoch": 0.6557044598043026,
      "grad_norm": 0.05168626829981804,
      "learning_rate": 9.046236656264258e-06,
      "loss": 0.05808258056640625,
      "step": 4850
    },
    {
      "epoch": 0.6558396566001385,
      "grad_norm": 0.0883084386587143,
      "learning_rate": 9.03995508844392e-06,
      "loss": 0.04830360412597656,
      "step": 4851
    },
    {
      "epoch": 0.6559748533959745,
      "grad_norm": 0.06716379523277283,
      "learning_rate": 9.033674761536718e-06,
      "loss": 0.07019400596618652,
      "step": 4852
    },
    {
      "epoch": 0.6561100501918105,
      "grad_norm": 0.03799884393811226,
      "learning_rate": 9.027395676850244e-06,
      "loss": 0.04934489727020264,
      "step": 4853
    },
    {
      "epoch": 0.6562452469876464,
      "grad_norm": 0.06861934065818787,
      "learning_rate": 9.02111783569184e-06,
      "loss": 0.06679368019104004,
      "step": 4854
    },
    {
      "epoch": 0.6563804437834824,
      "grad_norm": 0.09711926430463791,
      "learning_rate": 9.014841239368591e-06,
      "loss": 0.06822347640991211,
      "step": 4855
    },
    {
      "epoch": 0.6565156405793182,
      "grad_norm": 0.05366342514753342,
      "learning_rate": 9.008565889187308e-06,
      "loss": 0.05695343017578125,
      "step": 4856
    },
    {
      "epoch": 0.6566508373751542,
      "grad_norm": 0.11509646475315094,
      "learning_rate": 9.00229178645456e-06,
      "loss": 0.0469319224357605,
      "step": 4857
    },
    {
      "epoch": 0.6567860341709901,
      "grad_norm": 0.0437554195523262,
      "learning_rate": 8.996018932476641e-06,
      "loss": 0.03499004244804382,
      "step": 4858
    },
    {
      "epoch": 0.6569212309668261,
      "grad_norm": 0.06808724999427795,
      "learning_rate": 8.989747328559606e-06,
      "loss": 0.05707836151123047,
      "step": 4859
    },
    {
      "epoch": 0.657056427762662,
      "grad_norm": 0.06786102801561356,
      "learning_rate": 8.98347697600922e-06,
      "loss": 0.06307046115398407,
      "step": 4860
    },
    {
      "epoch": 0.657191624558498,
      "grad_norm": 0.05690736323595047,
      "learning_rate": 8.977207876131013e-06,
      "loss": 0.05515623092651367,
      "step": 4861
    },
    {
      "epoch": 0.657326821354334,
      "grad_norm": 0.09086936712265015,
      "learning_rate": 8.970940030230245e-06,
      "loss": 0.053468942642211914,
      "step": 4862
    },
    {
      "epoch": 0.6574620181501698,
      "grad_norm": 0.05574815720319748,
      "learning_rate": 8.96467343961191e-06,
      "loss": 0.05734062194824219,
      "step": 4863
    },
    {
      "epoch": 0.6575972149460058,
      "grad_norm": 0.12143168598413467,
      "learning_rate": 8.958408105580759e-06,
      "loss": 0.06808352470397949,
      "step": 4864
    },
    {
      "epoch": 0.6577324117418417,
      "grad_norm": 0.09449589997529984,
      "learning_rate": 8.952144029441248e-06,
      "loss": 0.07207393646240234,
      "step": 4865
    },
    {
      "epoch": 0.6578676085376777,
      "grad_norm": 0.12222263962030411,
      "learning_rate": 8.945881212497603e-06,
      "loss": 0.04599571228027344,
      "step": 4866
    },
    {
      "epoch": 0.6580028053335136,
      "grad_norm": 0.06164616718888283,
      "learning_rate": 8.939619656053777e-06,
      "loss": 0.06098055839538574,
      "step": 4867
    },
    {
      "epoch": 0.6581380021293496,
      "grad_norm": 0.1222454309463501,
      "learning_rate": 8.933359361413456e-06,
      "loss": 0.09478718042373657,
      "step": 4868
    },
    {
      "epoch": 0.6582731989251854,
      "grad_norm": 0.09668613970279694,
      "learning_rate": 8.92710032988007e-06,
      "loss": 0.07592487335205078,
      "step": 4869
    },
    {
      "epoch": 0.6584083957210214,
      "grad_norm": 0.09866383671760559,
      "learning_rate": 8.920842562756773e-06,
      "loss": 0.08246517181396484,
      "step": 4870
    },
    {
      "epoch": 0.6585435925168573,
      "grad_norm": 0.055106084793806076,
      "learning_rate": 8.914586061346474e-06,
      "loss": 0.04721784591674805,
      "step": 4871
    },
    {
      "epoch": 0.6586787893126933,
      "grad_norm": 0.06283453851938248,
      "learning_rate": 8.908330826951811e-06,
      "loss": 0.06782341003417969,
      "step": 4872
    },
    {
      "epoch": 0.6588139861085293,
      "grad_norm": 0.0724920928478241,
      "learning_rate": 8.902076860875155e-06,
      "loss": 0.04080760478973389,
      "step": 4873
    },
    {
      "epoch": 0.6589491829043652,
      "grad_norm": 0.1315796822309494,
      "learning_rate": 8.895824164418615e-06,
      "loss": 0.06891417503356934,
      "step": 4874
    },
    {
      "epoch": 0.6590843797002011,
      "grad_norm": 0.15078124403953552,
      "learning_rate": 8.889572738884033e-06,
      "loss": 0.06255936622619629,
      "step": 4875
    },
    {
      "epoch": 0.659219576496037,
      "grad_norm": 0.0863567441701889,
      "learning_rate": 8.88332258557299e-06,
      "loss": 0.07738113403320312,
      "step": 4876
    },
    {
      "epoch": 0.659354773291873,
      "grad_norm": 0.04448531195521355,
      "learning_rate": 8.877073705786806e-06,
      "loss": 0.0442737340927124,
      "step": 4877
    },
    {
      "epoch": 0.6594899700877089,
      "grad_norm": 0.17885813117027283,
      "learning_rate": 8.870826100826527e-06,
      "loss": 0.07945680618286133,
      "step": 4878
    },
    {
      "epoch": 0.6596251668835449,
      "grad_norm": 0.038998864591121674,
      "learning_rate": 8.86457977199294e-06,
      "loss": 0.031073331832885742,
      "step": 4879
    },
    {
      "epoch": 0.6597603636793808,
      "grad_norm": 0.1214047223329544,
      "learning_rate": 8.85833472058656e-06,
      "loss": 0.08460617065429688,
      "step": 4880
    },
    {
      "epoch": 0.6598955604752167,
      "grad_norm": 0.1492588222026825,
      "learning_rate": 8.852090947907643e-06,
      "loss": 0.053884267807006836,
      "step": 4881
    },
    {
      "epoch": 0.6600307572710526,
      "grad_norm": 0.046563033014535904,
      "learning_rate": 8.84584845525618e-06,
      "loss": 0.05501675605773926,
      "step": 4882
    },
    {
      "epoch": 0.6601659540668886,
      "grad_norm": 0.08040779829025269,
      "learning_rate": 8.83960724393188e-06,
      "loss": 0.08896112442016602,
      "step": 4883
    },
    {
      "epoch": 0.6603011508627246,
      "grad_norm": 0.05154357850551605,
      "learning_rate": 8.833367315234206e-06,
      "loss": 0.05368471145629883,
      "step": 4884
    },
    {
      "epoch": 0.6604363476585605,
      "grad_norm": 0.05111148953437805,
      "learning_rate": 8.82712867046234e-06,
      "loss": 0.03934288024902344,
      "step": 4885
    },
    {
      "epoch": 0.6605715444543965,
      "grad_norm": 0.05213287100195885,
      "learning_rate": 8.820891310915203e-06,
      "loss": 0.046380043029785156,
      "step": 4886
    },
    {
      "epoch": 0.6607067412502323,
      "grad_norm": 0.06324321776628494,
      "learning_rate": 8.81465523789145e-06,
      "loss": 0.042552947998046875,
      "step": 4887
    },
    {
      "epoch": 0.6608419380460683,
      "grad_norm": 0.07806697487831116,
      "learning_rate": 8.808420452689455e-06,
      "loss": 0.06439042091369629,
      "step": 4888
    },
    {
      "epoch": 0.6609771348419042,
      "grad_norm": 0.17124710977077484,
      "learning_rate": 8.802186956607344e-06,
      "loss": 0.06840074062347412,
      "step": 4889
    },
    {
      "epoch": 0.6611123316377402,
      "grad_norm": 0.10869485884904861,
      "learning_rate": 8.795954750942954e-06,
      "loss": 0.04557037353515625,
      "step": 4890
    },
    {
      "epoch": 0.6612475284335761,
      "grad_norm": 0.09238927066326141,
      "learning_rate": 8.789723836993878e-06,
      "loss": 0.09480142593383789,
      "step": 4891
    },
    {
      "epoch": 0.6613827252294121,
      "grad_norm": 0.11820916086435318,
      "learning_rate": 8.783494216057407e-06,
      "loss": 0.06707632541656494,
      "step": 4892
    },
    {
      "epoch": 0.661517922025248,
      "grad_norm": 0.033634837716817856,
      "learning_rate": 8.777265889430593e-06,
      "loss": 0.03417396545410156,
      "step": 4893
    },
    {
      "epoch": 0.6616531188210839,
      "grad_norm": 0.15732374787330627,
      "learning_rate": 8.771038858410206e-06,
      "loss": 0.0848923921585083,
      "step": 4894
    },
    {
      "epoch": 0.6617883156169199,
      "grad_norm": 0.0708540603518486,
      "learning_rate": 8.764813124292744e-06,
      "loss": 0.05016779899597168,
      "step": 4895
    },
    {
      "epoch": 0.6619235124127558,
      "grad_norm": 0.058807123452425,
      "learning_rate": 8.758588688374445e-06,
      "loss": 0.054766178131103516,
      "step": 4896
    },
    {
      "epoch": 0.6620587092085918,
      "grad_norm": 0.11723329871892929,
      "learning_rate": 8.752365551951262e-06,
      "loss": 0.1027536392211914,
      "step": 4897
    },
    {
      "epoch": 0.6621939060044277,
      "grad_norm": 0.0490725077688694,
      "learning_rate": 8.74614371631888e-06,
      "loss": 0.03446251153945923,
      "step": 4898
    },
    {
      "epoch": 0.6623291028002636,
      "grad_norm": 0.07183082401752472,
      "learning_rate": 8.739923182772732e-06,
      "loss": 0.05862855911254883,
      "step": 4899
    },
    {
      "epoch": 0.6624642995960995,
      "grad_norm": 0.07355359196662903,
      "learning_rate": 8.733703952607956e-06,
      "loss": 0.0664055347442627,
      "step": 4900
    },
    {
      "epoch": 0.6625994963919355,
      "grad_norm": 0.042284250259399414,
      "learning_rate": 8.727486027119443e-06,
      "loss": 0.04697680473327637,
      "step": 4901
    },
    {
      "epoch": 0.6627346931877715,
      "grad_norm": 0.09456198662519455,
      "learning_rate": 8.721269407601783e-06,
      "loss": 0.08613348007202148,
      "step": 4902
    },
    {
      "epoch": 0.6628698899836074,
      "grad_norm": 0.04701028764247894,
      "learning_rate": 8.71505409534931e-06,
      "loss": 0.042190492153167725,
      "step": 4903
    },
    {
      "epoch": 0.6630050867794434,
      "grad_norm": 0.0853782370686531,
      "learning_rate": 8.708840091656093e-06,
      "loss": 0.05188995599746704,
      "step": 4904
    },
    {
      "epoch": 0.6631402835752793,
      "grad_norm": 0.0897306501865387,
      "learning_rate": 8.70262739781592e-06,
      "loss": 0.09391683340072632,
      "step": 4905
    },
    {
      "epoch": 0.6632754803711152,
      "grad_norm": 0.08579965680837631,
      "learning_rate": 8.696416015122302e-06,
      "loss": 0.05850362777709961,
      "step": 4906
    },
    {
      "epoch": 0.6634106771669511,
      "grad_norm": 0.06838425993919373,
      "learning_rate": 8.690205944868487e-06,
      "loss": 0.04645967483520508,
      "step": 4907
    },
    {
      "epoch": 0.6635458739627871,
      "grad_norm": 0.06270729005336761,
      "learning_rate": 8.683997188347436e-06,
      "loss": 0.04364466667175293,
      "step": 4908
    },
    {
      "epoch": 0.663681070758623,
      "grad_norm": 0.14725305140018463,
      "learning_rate": 8.677789746851855e-06,
      "loss": 0.07762765884399414,
      "step": 4909
    },
    {
      "epoch": 0.663816267554459,
      "grad_norm": 0.07702852040529251,
      "learning_rate": 8.671583621674167e-06,
      "loss": 0.04913806915283203,
      "step": 4910
    },
    {
      "epoch": 0.663951464350295,
      "grad_norm": 0.04927091673016548,
      "learning_rate": 8.665378814106512e-06,
      "loss": 0.049920082092285156,
      "step": 4911
    },
    {
      "epoch": 0.6640866611461308,
      "grad_norm": 0.09376531839370728,
      "learning_rate": 8.65917532544077e-06,
      "loss": 0.057477474212646484,
      "step": 4912
    },
    {
      "epoch": 0.6642218579419668,
      "grad_norm": 0.06192830204963684,
      "learning_rate": 8.652973156968532e-06,
      "loss": 0.04173433780670166,
      "step": 4913
    },
    {
      "epoch": 0.6643570547378027,
      "grad_norm": 0.13013476133346558,
      "learning_rate": 8.646772309981141e-06,
      "loss": 0.08666729927062988,
      "step": 4914
    },
    {
      "epoch": 0.6644922515336387,
      "grad_norm": 0.0691540464758873,
      "learning_rate": 8.640572785769624e-06,
      "loss": 0.07049989700317383,
      "step": 4915
    },
    {
      "epoch": 0.6646274483294746,
      "grad_norm": 0.10392516106367111,
      "learning_rate": 8.63437458562477e-06,
      "loss": 0.037362098693847656,
      "step": 4916
    },
    {
      "epoch": 0.6647626451253106,
      "grad_norm": 0.11799994111061096,
      "learning_rate": 8.628177710837068e-06,
      "loss": 0.09870147705078125,
      "step": 4917
    },
    {
      "epoch": 0.6648978419211464,
      "grad_norm": 0.10603231191635132,
      "learning_rate": 8.621982162696752e-06,
      "loss": 0.08280134201049805,
      "step": 4918
    },
    {
      "epoch": 0.6650330387169824,
      "grad_norm": 0.11542531847953796,
      "learning_rate": 8.615787942493766e-06,
      "loss": 0.03516983985900879,
      "step": 4919
    },
    {
      "epoch": 0.6651682355128183,
      "grad_norm": 0.07931727170944214,
      "learning_rate": 8.609595051517765e-06,
      "loss": 0.07717704772949219,
      "step": 4920
    },
    {
      "epoch": 0.6653034323086543,
      "grad_norm": 0.09853526204824448,
      "learning_rate": 8.603403491058157e-06,
      "loss": 0.0735853910446167,
      "step": 4921
    },
    {
      "epoch": 0.6654386291044903,
      "grad_norm": 0.06230878457427025,
      "learning_rate": 8.597213262404046e-06,
      "loss": 0.05725669860839844,
      "step": 4922
    },
    {
      "epoch": 0.6655738259003262,
      "grad_norm": 0.048624128103256226,
      "learning_rate": 8.591024366844291e-06,
      "loss": 0.044798851013183594,
      "step": 4923
    },
    {
      "epoch": 0.6657090226961621,
      "grad_norm": 0.1617489457130432,
      "learning_rate": 8.584836805667434e-06,
      "loss": 0.06394720077514648,
      "step": 4924
    },
    {
      "epoch": 0.665844219491998,
      "grad_norm": 0.0880943313241005,
      "learning_rate": 8.578650580161754e-06,
      "loss": 0.07224702835083008,
      "step": 4925
    },
    {
      "epoch": 0.665979416287834,
      "grad_norm": 0.10280217230319977,
      "learning_rate": 8.572465691615275e-06,
      "loss": 0.06000781059265137,
      "step": 4926
    },
    {
      "epoch": 0.6661146130836699,
      "grad_norm": 0.055835746228694916,
      "learning_rate": 8.56628214131571e-06,
      "loss": 0.0520939826965332,
      "step": 4927
    },
    {
      "epoch": 0.6662498098795059,
      "grad_norm": 0.1812639683485031,
      "learning_rate": 8.560099930550523e-06,
      "loss": 0.08518028259277344,
      "step": 4928
    },
    {
      "epoch": 0.6663850066753418,
      "grad_norm": 0.07087575644254684,
      "learning_rate": 8.553919060606866e-06,
      "loss": 0.03808313608169556,
      "step": 4929
    },
    {
      "epoch": 0.6665202034711777,
      "grad_norm": 0.0676916167140007,
      "learning_rate": 8.54773953277163e-06,
      "loss": 0.06779193878173828,
      "step": 4930
    },
    {
      "epoch": 0.6666554002670136,
      "grad_norm": 0.12563782930374146,
      "learning_rate": 8.541561348331433e-06,
      "loss": 0.08685779571533203,
      "step": 4931
    },
    {
      "epoch": 0.6667905970628496,
      "grad_norm": 0.1221809834241867,
      "learning_rate": 8.535384508572603e-06,
      "loss": 0.08554840087890625,
      "step": 4932
    },
    {
      "epoch": 0.6669257938586856,
      "grad_norm": 0.06465268135070801,
      "learning_rate": 8.529209014781202e-06,
      "loss": 0.0711984634399414,
      "step": 4933
    },
    {
      "epoch": 0.6670609906545215,
      "grad_norm": 0.0423555001616478,
      "learning_rate": 8.523034868242984e-06,
      "loss": 0.04691123962402344,
      "step": 4934
    },
    {
      "epoch": 0.6671961874503575,
      "grad_norm": 0.07108409702777863,
      "learning_rate": 8.51686207024344e-06,
      "loss": 0.07406044006347656,
      "step": 4935
    },
    {
      "epoch": 0.6673313842461933,
      "grad_norm": 0.04676260054111481,
      "learning_rate": 8.510690622067792e-06,
      "loss": 0.03845500946044922,
      "step": 4936
    },
    {
      "epoch": 0.6674665810420293,
      "grad_norm": 0.09336749464273453,
      "learning_rate": 8.50452052500096e-06,
      "loss": 0.06414878368377686,
      "step": 4937
    },
    {
      "epoch": 0.6676017778378652,
      "grad_norm": 0.0990876704454422,
      "learning_rate": 8.498351780327594e-06,
      "loss": 0.08987903594970703,
      "step": 4938
    },
    {
      "epoch": 0.6677369746337012,
      "grad_norm": 0.05231555551290512,
      "learning_rate": 8.492184389332061e-06,
      "loss": 0.05912339687347412,
      "step": 4939
    },
    {
      "epoch": 0.6678721714295371,
      "grad_norm": 0.061920735985040665,
      "learning_rate": 8.486018353298432e-06,
      "loss": 0.08337268233299255,
      "step": 4940
    },
    {
      "epoch": 0.6680073682253731,
      "grad_norm": 0.044133953750133514,
      "learning_rate": 8.479853673510528e-06,
      "loss": 0.04243624210357666,
      "step": 4941
    },
    {
      "epoch": 0.668142565021209,
      "grad_norm": 0.11150945723056793,
      "learning_rate": 8.473690351251855e-06,
      "loss": 0.04516524076461792,
      "step": 4942
    },
    {
      "epoch": 0.6682777618170449,
      "grad_norm": 0.04301859810948372,
      "learning_rate": 8.467528387805656e-06,
      "loss": 0.03423190116882324,
      "step": 4943
    },
    {
      "epoch": 0.6684129586128809,
      "grad_norm": 0.0725431889295578,
      "learning_rate": 8.461367784454881e-06,
      "loss": 0.050350189208984375,
      "step": 4944
    },
    {
      "epoch": 0.6685481554087168,
      "grad_norm": 0.12706536054611206,
      "learning_rate": 8.455208542482195e-06,
      "loss": 0.06219625473022461,
      "step": 4945
    },
    {
      "epoch": 0.6686833522045528,
      "grad_norm": 0.0744287371635437,
      "learning_rate": 8.449050663170004e-06,
      "loss": 0.07586669921875,
      "step": 4946
    },
    {
      "epoch": 0.6688185490003887,
      "grad_norm": 0.08442191779613495,
      "learning_rate": 8.442894147800387e-06,
      "loss": 0.048914432525634766,
      "step": 4947
    },
    {
      "epoch": 0.6689537457962246,
      "grad_norm": 0.0932261198759079,
      "learning_rate": 8.436738997655184e-06,
      "loss": 0.05852067470550537,
      "step": 4948
    },
    {
      "epoch": 0.6690889425920605,
      "grad_norm": 0.04948500916361809,
      "learning_rate": 8.430585214015918e-06,
      "loss": 0.03929734230041504,
      "step": 4949
    },
    {
      "epoch": 0.6692241393878965,
      "grad_norm": 0.07685403525829315,
      "learning_rate": 8.424432798163838e-06,
      "loss": 0.06699171662330627,
      "step": 4950
    },
    {
      "epoch": 0.6693593361837324,
      "grad_norm": 0.12221293896436691,
      "learning_rate": 8.418281751379926e-06,
      "loss": 0.06083989143371582,
      "step": 4951
    },
    {
      "epoch": 0.6694945329795684,
      "grad_norm": 0.2653440833091736,
      "learning_rate": 8.41213207494484e-06,
      "loss": 0.09709548950195312,
      "step": 4952
    },
    {
      "epoch": 0.6696297297754044,
      "grad_norm": 0.09595558047294617,
      "learning_rate": 8.405983770138992e-06,
      "loss": 0.07072973251342773,
      "step": 4953
    },
    {
      "epoch": 0.6697649265712403,
      "grad_norm": 0.11023281514644623,
      "learning_rate": 8.399836838242479e-06,
      "loss": 0.06504392623901367,
      "step": 4954
    },
    {
      "epoch": 0.6699001233670762,
      "grad_norm": 0.06484043598175049,
      "learning_rate": 8.393691280535143e-06,
      "loss": 0.05286455154418945,
      "step": 4955
    },
    {
      "epoch": 0.6700353201629121,
      "grad_norm": 0.21476531028747559,
      "learning_rate": 8.387547098296516e-06,
      "loss": 0.1045083999633789,
      "step": 4956
    },
    {
      "epoch": 0.6701705169587481,
      "grad_norm": 0.06728379428386688,
      "learning_rate": 8.38140429280583e-06,
      "loss": 0.05539655685424805,
      "step": 4957
    },
    {
      "epoch": 0.670305713754584,
      "grad_norm": 0.11947528272867203,
      "learning_rate": 8.375262865342073e-06,
      "loss": 0.07096648216247559,
      "step": 4958
    },
    {
      "epoch": 0.67044091055042,
      "grad_norm": 0.07799821346998215,
      "learning_rate": 8.36912281718391e-06,
      "loss": 0.05708050727844238,
      "step": 4959
    },
    {
      "epoch": 0.670576107346256,
      "grad_norm": 0.07748235017061234,
      "learning_rate": 8.362984149609748e-06,
      "loss": 0.0420374870300293,
      "step": 4960
    },
    {
      "epoch": 0.6707113041420918,
      "grad_norm": 0.14528732001781464,
      "learning_rate": 8.356846863897672e-06,
      "loss": 0.06159478425979614,
      "step": 4961
    },
    {
      "epoch": 0.6708465009379277,
      "grad_norm": 0.05735364928841591,
      "learning_rate": 8.350710961325498e-06,
      "loss": 0.031278252601623535,
      "step": 4962
    },
    {
      "epoch": 0.6709816977337637,
      "grad_norm": 0.04464547336101532,
      "learning_rate": 8.344576443170768e-06,
      "loss": 0.041180968284606934,
      "step": 4963
    },
    {
      "epoch": 0.6711168945295997,
      "grad_norm": 0.07714742422103882,
      "learning_rate": 8.338443310710708e-06,
      "loss": 0.05145072937011719,
      "step": 4964
    },
    {
      "epoch": 0.6712520913254356,
      "grad_norm": 0.1096891239285469,
      "learning_rate": 8.332311565222284e-06,
      "loss": 0.0773930549621582,
      "step": 4965
    },
    {
      "epoch": 0.6713872881212716,
      "grad_norm": 0.16796119511127472,
      "learning_rate": 8.326181207982145e-06,
      "loss": 0.08898353576660156,
      "step": 4966
    },
    {
      "epoch": 0.6715224849171074,
      "grad_norm": 0.07012584805488586,
      "learning_rate": 8.32005224026666e-06,
      "loss": 0.06497669219970703,
      "step": 4967
    },
    {
      "epoch": 0.6716576817129434,
      "grad_norm": 0.0804951936006546,
      "learning_rate": 8.313924663351927e-06,
      "loss": 0.05633258819580078,
      "step": 4968
    },
    {
      "epoch": 0.6717928785087793,
      "grad_norm": 0.1588028073310852,
      "learning_rate": 8.307798478513733e-06,
      "loss": 0.07164812088012695,
      "step": 4969
    },
    {
      "epoch": 0.6719280753046153,
      "grad_norm": 0.0795266330242157,
      "learning_rate": 8.301673687027583e-06,
      "loss": 0.03998684883117676,
      "step": 4970
    },
    {
      "epoch": 0.6720632721004512,
      "grad_norm": 0.03827924281358719,
      "learning_rate": 8.295550290168692e-06,
      "loss": 0.04121851921081543,
      "step": 4971
    },
    {
      "epoch": 0.6721984688962872,
      "grad_norm": 0.041952770203351974,
      "learning_rate": 8.289428289211977e-06,
      "loss": 0.0473254919052124,
      "step": 4972
    },
    {
      "epoch": 0.672333665692123,
      "grad_norm": 0.12119283527135849,
      "learning_rate": 8.283307685432083e-06,
      "loss": 0.08006525039672852,
      "step": 4973
    },
    {
      "epoch": 0.672468862487959,
      "grad_norm": 0.06962171941995621,
      "learning_rate": 8.277188480103348e-06,
      "loss": 0.08458590507507324,
      "step": 4974
    },
    {
      "epoch": 0.672604059283795,
      "grad_norm": 0.079900823533535,
      "learning_rate": 8.271070674499821e-06,
      "loss": 0.06687116622924805,
      "step": 4975
    },
    {
      "epoch": 0.6727392560796309,
      "grad_norm": 0.08159684389829636,
      "learning_rate": 8.264954269895262e-06,
      "loss": 0.048860788345336914,
      "step": 4976
    },
    {
      "epoch": 0.6728744528754669,
      "grad_norm": 0.07078345119953156,
      "learning_rate": 8.258839267563134e-06,
      "loss": 0.049520134925842285,
      "step": 4977
    },
    {
      "epoch": 0.6730096496713028,
      "grad_norm": 0.049564849585294724,
      "learning_rate": 8.252725668776623e-06,
      "loss": 0.05319809913635254,
      "step": 4978
    },
    {
      "epoch": 0.6731448464671387,
      "grad_norm": 0.06402891129255295,
      "learning_rate": 8.24661347480861e-06,
      "loss": 0.0640106201171875,
      "step": 4979
    },
    {
      "epoch": 0.6732800432629746,
      "grad_norm": 0.07140089571475983,
      "learning_rate": 8.240502686931686e-06,
      "loss": 0.05462515354156494,
      "step": 4980
    },
    {
      "epoch": 0.6734152400588106,
      "grad_norm": 0.038319412618875504,
      "learning_rate": 8.234393306418148e-06,
      "loss": 0.03282272815704346,
      "step": 4981
    },
    {
      "epoch": 0.6735504368546466,
      "grad_norm": 0.14451679587364197,
      "learning_rate": 8.228285334539995e-06,
      "loss": 0.08681714534759521,
      "step": 4982
    },
    {
      "epoch": 0.6736856336504825,
      "grad_norm": 0.1048111841082573,
      "learning_rate": 8.22217877256896e-06,
      "loss": 0.0899057388305664,
      "step": 4983
    },
    {
      "epoch": 0.6738208304463185,
      "grad_norm": 0.08846127241849899,
      "learning_rate": 8.216073621776436e-06,
      "loss": 0.07339954376220703,
      "step": 4984
    },
    {
      "epoch": 0.6739560272421543,
      "grad_norm": 0.06530094891786575,
      "learning_rate": 8.209969883433566e-06,
      "loss": 0.06296944618225098,
      "step": 4985
    },
    {
      "epoch": 0.6740912240379903,
      "grad_norm": 0.06321869790554047,
      "learning_rate": 8.203867558811177e-06,
      "loss": 0.06943702697753906,
      "step": 4986
    },
    {
      "epoch": 0.6742264208338262,
      "grad_norm": 0.0455254502594471,
      "learning_rate": 8.197766649179795e-06,
      "loss": 0.050469398498535156,
      "step": 4987
    },
    {
      "epoch": 0.6743616176296622,
      "grad_norm": 0.051120419055223465,
      "learning_rate": 8.191667155809684e-06,
      "loss": 0.05274152755737305,
      "step": 4988
    },
    {
      "epoch": 0.6744968144254981,
      "grad_norm": 0.16293583810329437,
      "learning_rate": 8.185569079970764e-06,
      "loss": 0.05103790760040283,
      "step": 4989
    },
    {
      "epoch": 0.6746320112213341,
      "grad_norm": 0.055799953639507294,
      "learning_rate": 8.179472422932709e-06,
      "loss": 0.058223724365234375,
      "step": 4990
    },
    {
      "epoch": 0.6747672080171699,
      "grad_norm": 0.09432735294103622,
      "learning_rate": 8.17337718596486e-06,
      "loss": 0.0534052848815918,
      "step": 4991
    },
    {
      "epoch": 0.6749024048130059,
      "grad_norm": 0.09030844271183014,
      "learning_rate": 8.167283370336295e-06,
      "loss": 0.08411085605621338,
      "step": 4992
    },
    {
      "epoch": 0.6750376016088419,
      "grad_norm": 0.05092301219701767,
      "learning_rate": 8.161190977315766e-06,
      "loss": 0.05570089817047119,
      "step": 4993
    },
    {
      "epoch": 0.6751727984046778,
      "grad_norm": 0.10943484306335449,
      "learning_rate": 8.155100008171736e-06,
      "loss": 0.07501763105392456,
      "step": 4994
    },
    {
      "epoch": 0.6753079952005138,
      "grad_norm": 0.04953750595450401,
      "learning_rate": 8.149010464172392e-06,
      "loss": 0.03264760971069336,
      "step": 4995
    },
    {
      "epoch": 0.6754431919963497,
      "grad_norm": 0.06535929441452026,
      "learning_rate": 8.142922346585597e-06,
      "loss": 0.04820537567138672,
      "step": 4996
    },
    {
      "epoch": 0.6755783887921857,
      "grad_norm": 0.08695446699857712,
      "learning_rate": 8.13683565667895e-06,
      "loss": 0.06612396240234375,
      "step": 4997
    },
    {
      "epoch": 0.6757135855880215,
      "grad_norm": 0.05927109345793724,
      "learning_rate": 8.13075039571971e-06,
      "loss": 0.043105125427246094,
      "step": 4998
    },
    {
      "epoch": 0.6758487823838575,
      "grad_norm": 0.04301923140883446,
      "learning_rate": 8.124666564974864e-06,
      "loss": 0.04664754867553711,
      "step": 4999
    },
    {
      "epoch": 0.6759839791796934,
      "grad_norm": 0.11336946487426758,
      "learning_rate": 8.11858416571111e-06,
      "loss": 0.07625558972358704,
      "step": 5000
    },
    {
      "epoch": 0.6761191759755294,
      "grad_norm": 0.08768303692340851,
      "learning_rate": 8.112503199194821e-06,
      "loss": 0.0694434642791748,
      "step": 5001
    },
    {
      "epoch": 0.6762543727713654,
      "grad_norm": 0.0627063512802124,
      "learning_rate": 8.106423666692108e-06,
      "loss": 0.07008993625640869,
      "step": 5002
    },
    {
      "epoch": 0.6763895695672013,
      "grad_norm": 0.07504633814096451,
      "learning_rate": 8.100345569468742e-06,
      "loss": 0.07243967056274414,
      "step": 5003
    },
    {
      "epoch": 0.6765247663630372,
      "grad_norm": 0.05151161924004555,
      "learning_rate": 8.094268908790215e-06,
      "loss": 0.05637025833129883,
      "step": 5004
    },
    {
      "epoch": 0.6766599631588731,
      "grad_norm": 0.04521086439490318,
      "learning_rate": 8.088193685921733e-06,
      "loss": 0.04199337959289551,
      "step": 5005
    },
    {
      "epoch": 0.6767951599547091,
      "grad_norm": 0.06515444070100784,
      "learning_rate": 8.082119902128185e-06,
      "loss": 0.07120251655578613,
      "step": 5006
    },
    {
      "epoch": 0.676930356750545,
      "grad_norm": 0.11023542284965515,
      "learning_rate": 8.076047558674164e-06,
      "loss": 0.09229302406311035,
      "step": 5007
    },
    {
      "epoch": 0.677065553546381,
      "grad_norm": 0.08682873845100403,
      "learning_rate": 8.069976656823964e-06,
      "loss": 0.055080413818359375,
      "step": 5008
    },
    {
      "epoch": 0.6772007503422169,
      "grad_norm": 0.1625545471906662,
      "learning_rate": 8.063907197841574e-06,
      "loss": 0.040492117404937744,
      "step": 5009
    },
    {
      "epoch": 0.6773359471380528,
      "grad_norm": 0.07702849805355072,
      "learning_rate": 8.057839182990698e-06,
      "loss": 0.05395984649658203,
      "step": 5010
    },
    {
      "epoch": 0.6774711439338887,
      "grad_norm": 0.17600449919700623,
      "learning_rate": 8.051772613534725e-06,
      "loss": 0.05693554878234863,
      "step": 5011
    },
    {
      "epoch": 0.6776063407297247,
      "grad_norm": 0.07110217213630676,
      "learning_rate": 8.045707490736745e-06,
      "loss": 0.05278158187866211,
      "step": 5012
    },
    {
      "epoch": 0.6777415375255607,
      "grad_norm": 0.0685405284166336,
      "learning_rate": 8.039643815859552e-06,
      "loss": 0.055429697036743164,
      "step": 5013
    },
    {
      "epoch": 0.6778767343213966,
      "grad_norm": 0.13837657868862152,
      "learning_rate": 8.033581590165627e-06,
      "loss": 0.08815836906433105,
      "step": 5014
    },
    {
      "epoch": 0.6780119311172326,
      "grad_norm": 0.051498956978321075,
      "learning_rate": 8.027520814917175e-06,
      "loss": 0.047371864318847656,
      "step": 5015
    },
    {
      "epoch": 0.6781471279130684,
      "grad_norm": 0.12761028110980988,
      "learning_rate": 8.021461491376064e-06,
      "loss": 0.07303357124328613,
      "step": 5016
    },
    {
      "epoch": 0.6782823247089044,
      "grad_norm": 0.09460553526878357,
      "learning_rate": 8.015403620803885e-06,
      "loss": 0.06608045101165771,
      "step": 5017
    },
    {
      "epoch": 0.6784175215047403,
      "grad_norm": 0.05743256211280823,
      "learning_rate": 8.009347204461922e-06,
      "loss": 0.062085628509521484,
      "step": 5018
    },
    {
      "epoch": 0.6785527183005763,
      "grad_norm": 0.0879201740026474,
      "learning_rate": 8.003292243611143e-06,
      "loss": 0.05473756790161133,
      "step": 5019
    },
    {
      "epoch": 0.6786879150964122,
      "grad_norm": 0.07949728518724442,
      "learning_rate": 7.99723873951224e-06,
      "loss": 0.06790757179260254,
      "step": 5020
    },
    {
      "epoch": 0.6788231118922482,
      "grad_norm": 0.06531021744012833,
      "learning_rate": 7.991186693425563e-06,
      "loss": 0.05360287427902222,
      "step": 5021
    },
    {
      "epoch": 0.678958308688084,
      "grad_norm": 0.08996099978685379,
      "learning_rate": 7.9851361066112e-06,
      "loss": 0.0458034873008728,
      "step": 5022
    },
    {
      "epoch": 0.67909350548392,
      "grad_norm": 0.11217587441205978,
      "learning_rate": 7.979086980328907e-06,
      "loss": 0.05580878257751465,
      "step": 5023
    },
    {
      "epoch": 0.679228702279756,
      "grad_norm": 0.058305565267801285,
      "learning_rate": 7.973039315838137e-06,
      "loss": 0.05259418487548828,
      "step": 5024
    },
    {
      "epoch": 0.6793638990755919,
      "grad_norm": 0.11344636976718903,
      "learning_rate": 7.966993114398067e-06,
      "loss": 0.07076263427734375,
      "step": 5025
    },
    {
      "epoch": 0.6794990958714279,
      "grad_norm": 0.043777696788311005,
      "learning_rate": 7.960948377267524e-06,
      "loss": 0.048508420586586,
      "step": 5026
    },
    {
      "epoch": 0.6796342926672638,
      "grad_norm": 0.06955604255199432,
      "learning_rate": 7.954905105705071e-06,
      "loss": 0.05074763298034668,
      "step": 5027
    },
    {
      "epoch": 0.6797694894630997,
      "grad_norm": 0.08960990607738495,
      "learning_rate": 7.948863300968938e-06,
      "loss": 0.08656072616577148,
      "step": 5028
    },
    {
      "epoch": 0.6799046862589356,
      "grad_norm": 0.10875927656888962,
      "learning_rate": 7.942822964317078e-06,
      "loss": 0.05288103222846985,
      "step": 5029
    },
    {
      "epoch": 0.6800398830547716,
      "grad_norm": 0.09737157076597214,
      "learning_rate": 7.936784097007105e-06,
      "loss": 0.053118228912353516,
      "step": 5030
    },
    {
      "epoch": 0.6801750798506075,
      "grad_norm": 0.10650312155485153,
      "learning_rate": 7.930746700296344e-06,
      "loss": 0.08737468719482422,
      "step": 5031
    },
    {
      "epoch": 0.6803102766464435,
      "grad_norm": 0.07064100354909897,
      "learning_rate": 7.924710775441822e-06,
      "loss": 0.06871414184570312,
      "step": 5032
    },
    {
      "epoch": 0.6804454734422795,
      "grad_norm": 0.057951413094997406,
      "learning_rate": 7.918676323700241e-06,
      "loss": 0.06516039371490479,
      "step": 5033
    },
    {
      "epoch": 0.6805806702381153,
      "grad_norm": 0.07951295375823975,
      "learning_rate": 7.912643346328023e-06,
      "loss": 0.06445860862731934,
      "step": 5034
    },
    {
      "epoch": 0.6807158670339513,
      "grad_norm": 0.045828066766262054,
      "learning_rate": 7.906611844581251e-06,
      "loss": 0.04909694194793701,
      "step": 5035
    },
    {
      "epoch": 0.6808510638297872,
      "grad_norm": 0.17410415410995483,
      "learning_rate": 7.900581819715713e-06,
      "loss": 0.08871793746948242,
      "step": 5036
    },
    {
      "epoch": 0.6809862606256232,
      "grad_norm": 0.04508494213223457,
      "learning_rate": 7.894553272986901e-06,
      "loss": 0.040786564350128174,
      "step": 5037
    },
    {
      "epoch": 0.6811214574214591,
      "grad_norm": 0.06835804134607315,
      "learning_rate": 7.888526205649993e-06,
      "loss": 0.0617070198059082,
      "step": 5038
    },
    {
      "epoch": 0.6812566542172951,
      "grad_norm": 0.10027843713760376,
      "learning_rate": 7.882500618959849e-06,
      "loss": 0.08986115455627441,
      "step": 5039
    },
    {
      "epoch": 0.6813918510131309,
      "grad_norm": 0.06253597140312195,
      "learning_rate": 7.876476514171033e-06,
      "loss": 0.0515742301940918,
      "step": 5040
    },
    {
      "epoch": 0.6815270478089669,
      "grad_norm": 0.07670264691114426,
      "learning_rate": 7.870453892537788e-06,
      "loss": 0.07343602180480957,
      "step": 5041
    },
    {
      "epoch": 0.6816622446048028,
      "grad_norm": 0.06232365965843201,
      "learning_rate": 7.864432755314068e-06,
      "loss": 0.05206248164176941,
      "step": 5042
    },
    {
      "epoch": 0.6817974414006388,
      "grad_norm": 0.059800885617733,
      "learning_rate": 7.858413103753499e-06,
      "loss": 0.037760019302368164,
      "step": 5043
    },
    {
      "epoch": 0.6819326381964748,
      "grad_norm": 0.12328537553548813,
      "learning_rate": 7.852394939109408e-06,
      "loss": 0.054387569427490234,
      "step": 5044
    },
    {
      "epoch": 0.6820678349923107,
      "grad_norm": 0.14490880072116852,
      "learning_rate": 7.846378262634803e-06,
      "loss": 0.061079978942871094,
      "step": 5045
    },
    {
      "epoch": 0.6822030317881467,
      "grad_norm": 0.06602545827627182,
      "learning_rate": 7.840363075582385e-06,
      "loss": 0.07086324691772461,
      "step": 5046
    },
    {
      "epoch": 0.6823382285839825,
      "grad_norm": 0.09575103968381882,
      "learning_rate": 7.834349379204565e-06,
      "loss": 0.07340621948242188,
      "step": 5047
    },
    {
      "epoch": 0.6824734253798185,
      "grad_norm": 0.104802705347538,
      "learning_rate": 7.828337174753411e-06,
      "loss": 0.07211112976074219,
      "step": 5048
    },
    {
      "epoch": 0.6826086221756544,
      "grad_norm": 0.07989746332168579,
      "learning_rate": 7.822326463480703e-06,
      "loss": 0.06185269355773926,
      "step": 5049
    },
    {
      "epoch": 0.6827438189714904,
      "grad_norm": 0.05287535861134529,
      "learning_rate": 7.816317246637901e-06,
      "loss": 0.04005555808544159,
      "step": 5050
    },
    {
      "epoch": 0.6828790157673263,
      "grad_norm": 0.1011776402592659,
      "learning_rate": 7.810309525476152e-06,
      "loss": 0.052993178367614746,
      "step": 5051
    },
    {
      "epoch": 0.6830142125631623,
      "grad_norm": 0.16065378487110138,
      "learning_rate": 7.804303301246311e-06,
      "loss": 0.0652306079864502,
      "step": 5052
    },
    {
      "epoch": 0.6831494093589981,
      "grad_norm": 0.08364777266979218,
      "learning_rate": 7.798298575198884e-06,
      "loss": 0.09554576873779297,
      "step": 5053
    },
    {
      "epoch": 0.6832846061548341,
      "grad_norm": 0.11256513744592667,
      "learning_rate": 7.792295348584103e-06,
      "loss": 0.04108166694641113,
      "step": 5054
    },
    {
      "epoch": 0.6834198029506701,
      "grad_norm": 0.09159068018198013,
      "learning_rate": 7.786293622651866e-06,
      "loss": 0.06287813186645508,
      "step": 5055
    },
    {
      "epoch": 0.683554999746506,
      "grad_norm": 0.08099980652332306,
      "learning_rate": 7.78029339865176e-06,
      "loss": 0.04396533966064453,
      "step": 5056
    },
    {
      "epoch": 0.683690196542342,
      "grad_norm": 0.055911049246788025,
      "learning_rate": 7.774294677833078e-06,
      "loss": 0.05603682994842529,
      "step": 5057
    },
    {
      "epoch": 0.6838253933381779,
      "grad_norm": 0.0674651563167572,
      "learning_rate": 7.768297461444766e-06,
      "loss": 0.05206373333930969,
      "step": 5058
    },
    {
      "epoch": 0.6839605901340138,
      "grad_norm": 0.11200419813394547,
      "learning_rate": 7.762301750735494e-06,
      "loss": 0.07729125022888184,
      "step": 5059
    },
    {
      "epoch": 0.6840957869298497,
      "grad_norm": 0.0715426579117775,
      "learning_rate": 7.756307546953592e-06,
      "loss": 0.046530961990356445,
      "step": 5060
    },
    {
      "epoch": 0.6842309837256857,
      "grad_norm": 0.04778808727860451,
      "learning_rate": 7.750314851347087e-06,
      "loss": 0.03265070915222168,
      "step": 5061
    },
    {
      "epoch": 0.6843661805215216,
      "grad_norm": 0.06253280490636826,
      "learning_rate": 7.74432366516369e-06,
      "loss": 0.05382585525512695,
      "step": 5062
    },
    {
      "epoch": 0.6845013773173576,
      "grad_norm": 0.027975067496299744,
      "learning_rate": 7.738333989650794e-06,
      "loss": 0.022824764251708984,
      "step": 5063
    },
    {
      "epoch": 0.6846365741131936,
      "grad_norm": 0.06765957176685333,
      "learning_rate": 7.732345826055487e-06,
      "loss": 0.058557212352752686,
      "step": 5064
    },
    {
      "epoch": 0.6847717709090294,
      "grad_norm": 0.12224598973989487,
      "learning_rate": 7.726359175624537e-06,
      "loss": 0.07012653350830078,
      "step": 5065
    },
    {
      "epoch": 0.6849069677048654,
      "grad_norm": 0.08704402297735214,
      "learning_rate": 7.720374039604395e-06,
      "loss": 0.0554652214050293,
      "step": 5066
    },
    {
      "epoch": 0.6850421645007013,
      "grad_norm": 0.0807432010769844,
      "learning_rate": 7.714390419241198e-06,
      "loss": 0.06577324867248535,
      "step": 5067
    },
    {
      "epoch": 0.6851773612965373,
      "grad_norm": 0.06077314540743828,
      "learning_rate": 7.70840831578076e-06,
      "loss": 0.04194331169128418,
      "step": 5068
    },
    {
      "epoch": 0.6853125580923732,
      "grad_norm": 0.047380853444337845,
      "learning_rate": 7.702427730468601e-06,
      "loss": 0.05104851722717285,
      "step": 5069
    },
    {
      "epoch": 0.6854477548882092,
      "grad_norm": 0.056827180087566376,
      "learning_rate": 7.696448664549898e-06,
      "loss": 0.06889700889587402,
      "step": 5070
    },
    {
      "epoch": 0.685582951684045,
      "grad_norm": 0.07652460783720016,
      "learning_rate": 7.690471119269541e-06,
      "loss": 0.0346217155456543,
      "step": 5071
    },
    {
      "epoch": 0.685718148479881,
      "grad_norm": 0.06559426337480545,
      "learning_rate": 7.684495095872073e-06,
      "loss": 0.052144765853881836,
      "step": 5072
    },
    {
      "epoch": 0.685853345275717,
      "grad_norm": 0.09552326798439026,
      "learning_rate": 7.678520595601728e-06,
      "loss": 0.06816208362579346,
      "step": 5073
    },
    {
      "epoch": 0.6859885420715529,
      "grad_norm": 0.03336964547634125,
      "learning_rate": 7.672547619702445e-06,
      "loss": 0.033823758363723755,
      "step": 5074
    },
    {
      "epoch": 0.6861237388673889,
      "grad_norm": 0.18504440784454346,
      "learning_rate": 7.666576169417823e-06,
      "loss": 0.0859217643737793,
      "step": 5075
    },
    {
      "epoch": 0.6862589356632248,
      "grad_norm": 0.048577431589365005,
      "learning_rate": 7.660606245991147e-06,
      "loss": 0.04616965353488922,
      "step": 5076
    },
    {
      "epoch": 0.6863941324590607,
      "grad_norm": 0.1842295229434967,
      "learning_rate": 7.654637850665393e-06,
      "loss": 0.07582733035087585,
      "step": 5077
    },
    {
      "epoch": 0.6865293292548966,
      "grad_norm": 0.07542487233877182,
      "learning_rate": 7.648670984683199e-06,
      "loss": 0.05849027633666992,
      "step": 5078
    },
    {
      "epoch": 0.6866645260507326,
      "grad_norm": 0.07645473629236221,
      "learning_rate": 7.642705649286916e-06,
      "loss": 0.049231886863708496,
      "step": 5079
    },
    {
      "epoch": 0.6867997228465685,
      "grad_norm": 0.10021381825208664,
      "learning_rate": 7.63674184571855e-06,
      "loss": 0.06332027912139893,
      "step": 5080
    },
    {
      "epoch": 0.6869349196424045,
      "grad_norm": 0.09316113591194153,
      "learning_rate": 7.630779575219797e-06,
      "loss": 0.046317338943481445,
      "step": 5081
    },
    {
      "epoch": 0.6870701164382405,
      "grad_norm": 0.06710349768400192,
      "learning_rate": 7.6248188390320344e-06,
      "loss": 0.0571141242980957,
      "step": 5082
    },
    {
      "epoch": 0.6872053132340763,
      "grad_norm": 0.051020462065935135,
      "learning_rate": 7.6188596383963135e-06,
      "loss": 0.04548037052154541,
      "step": 5083
    },
    {
      "epoch": 0.6873405100299123,
      "grad_norm": 0.0944100096821785,
      "learning_rate": 7.612901974553388e-06,
      "loss": 0.05563795566558838,
      "step": 5084
    },
    {
      "epoch": 0.6874757068257482,
      "grad_norm": 0.04953437298536301,
      "learning_rate": 7.606945848743653e-06,
      "loss": 0.05012011528015137,
      "step": 5085
    },
    {
      "epoch": 0.6876109036215842,
      "grad_norm": 0.18776196241378784,
      "learning_rate": 7.600991262207221e-06,
      "loss": 0.09292605519294739,
      "step": 5086
    },
    {
      "epoch": 0.6877461004174201,
      "grad_norm": 0.045769114047288895,
      "learning_rate": 7.595038216183867e-06,
      "loss": 0.05554145574569702,
      "step": 5087
    },
    {
      "epoch": 0.6878812972132561,
      "grad_norm": 0.06939801573753357,
      "learning_rate": 7.589086711913037e-06,
      "loss": 0.04562854766845703,
      "step": 5088
    },
    {
      "epoch": 0.688016494009092,
      "grad_norm": 0.0660228580236435,
      "learning_rate": 7.583136750633885e-06,
      "loss": 0.05246230959892273,
      "step": 5089
    },
    {
      "epoch": 0.6881516908049279,
      "grad_norm": 0.1277281939983368,
      "learning_rate": 7.577188333585202e-06,
      "loss": 0.06319618225097656,
      "step": 5090
    },
    {
      "epoch": 0.6882868876007638,
      "grad_norm": 0.14541186392307281,
      "learning_rate": 7.5712414620054975e-06,
      "loss": 0.09457850456237793,
      "step": 5091
    },
    {
      "epoch": 0.6884220843965998,
      "grad_norm": 0.11648689210414886,
      "learning_rate": 7.565296137132935e-06,
      "loss": 0.07001566886901855,
      "step": 5092
    },
    {
      "epoch": 0.6885572811924358,
      "grad_norm": 0.2234610915184021,
      "learning_rate": 7.559352360205357e-06,
      "loss": 0.06101638078689575,
      "step": 5093
    },
    {
      "epoch": 0.6886924779882717,
      "grad_norm": 0.044088516384363174,
      "learning_rate": 7.553410132460308e-06,
      "loss": 0.03093719482421875,
      "step": 5094
    },
    {
      "epoch": 0.6888276747841077,
      "grad_norm": 0.0879405215382576,
      "learning_rate": 7.547469455134968e-06,
      "loss": 0.06865859031677246,
      "step": 5095
    },
    {
      "epoch": 0.6889628715799435,
      "grad_norm": 0.152337446808815,
      "learning_rate": 7.541530329466236e-06,
      "loss": 0.06282424926757812,
      "step": 5096
    },
    {
      "epoch": 0.6890980683757795,
      "grad_norm": 0.12875361740589142,
      "learning_rate": 7.535592756690661e-06,
      "loss": 0.09005355834960938,
      "step": 5097
    },
    {
      "epoch": 0.6892332651716154,
      "grad_norm": 0.07957500219345093,
      "learning_rate": 7.52965673804448e-06,
      "loss": 0.05158579349517822,
      "step": 5098
    },
    {
      "epoch": 0.6893684619674514,
      "grad_norm": 0.16104726493358612,
      "learning_rate": 7.5237222747636025e-06,
      "loss": 0.07291078567504883,
      "step": 5099
    },
    {
      "epoch": 0.6895036587632873,
      "grad_norm": 0.07990613579750061,
      "learning_rate": 7.517789368083611e-06,
      "loss": 0.0559384822845459,
      "step": 5100
    },
    {
      "epoch": 0.6896388555591233,
      "grad_norm": 0.07068870961666107,
      "learning_rate": 7.511858019239778e-06,
      "loss": 0.07599782943725586,
      "step": 5101
    },
    {
      "epoch": 0.6897740523549591,
      "grad_norm": 0.0743715837597847,
      "learning_rate": 7.505928229467038e-06,
      "loss": 0.05978125333786011,
      "step": 5102
    },
    {
      "epoch": 0.6899092491507951,
      "grad_norm": 0.09362413734197617,
      "learning_rate": 7.500000000000004e-06,
      "loss": 0.05211925506591797,
      "step": 5103
    },
    {
      "epoch": 0.6900444459466311,
      "grad_norm": 0.07234194129705429,
      "learning_rate": 7.494073332072963e-06,
      "loss": 0.0369877815246582,
      "step": 5104
    },
    {
      "epoch": 0.690179642742467,
      "grad_norm": 0.08082889765501022,
      "learning_rate": 7.488148226919877e-06,
      "loss": 0.05261269211769104,
      "step": 5105
    },
    {
      "epoch": 0.690314839538303,
      "grad_norm": 0.17888988554477692,
      "learning_rate": 7.482224685774393e-06,
      "loss": 0.09072732925415039,
      "step": 5106
    },
    {
      "epoch": 0.6904500363341389,
      "grad_norm": 0.11682732403278351,
      "learning_rate": 7.4763027098698184e-06,
      "loss": 0.0584789514541626,
      "step": 5107
    },
    {
      "epoch": 0.6905852331299748,
      "grad_norm": 0.07673139125108719,
      "learning_rate": 7.470382300439143e-06,
      "loss": 0.04524785280227661,
      "step": 5108
    },
    {
      "epoch": 0.6907204299258107,
      "grad_norm": 0.08818185329437256,
      "learning_rate": 7.4644634587150225e-06,
      "loss": 0.06822824478149414,
      "step": 5109
    },
    {
      "epoch": 0.6908556267216467,
      "grad_norm": 0.09656547009944916,
      "learning_rate": 7.4585461859297906e-06,
      "loss": 0.04591912031173706,
      "step": 5110
    },
    {
      "epoch": 0.6909908235174826,
      "grad_norm": 0.05788983032107353,
      "learning_rate": 7.452630483315463e-06,
      "loss": 0.04120028018951416,
      "step": 5111
    },
    {
      "epoch": 0.6911260203133186,
      "grad_norm": 0.1802745908498764,
      "learning_rate": 7.4467163521037186e-06,
      "loss": 0.09113550186157227,
      "step": 5112
    },
    {
      "epoch": 0.6912612171091546,
      "grad_norm": 0.12040071934461594,
      "learning_rate": 7.440803793525907e-06,
      "loss": 0.059673309326171875,
      "step": 5113
    },
    {
      "epoch": 0.6913964139049904,
      "grad_norm": 0.11113721132278442,
      "learning_rate": 7.434892808813056e-06,
      "loss": 0.06353449821472168,
      "step": 5114
    },
    {
      "epoch": 0.6915316107008264,
      "grad_norm": 0.13186430931091309,
      "learning_rate": 7.42898339919586e-06,
      "loss": 0.08459639549255371,
      "step": 5115
    },
    {
      "epoch": 0.6916668074966623,
      "grad_norm": 0.07785635441541672,
      "learning_rate": 7.423075565904698e-06,
      "loss": 0.04776430130004883,
      "step": 5116
    },
    {
      "epoch": 0.6918020042924983,
      "grad_norm": 0.08863280713558197,
      "learning_rate": 7.417169310169609e-06,
      "loss": 0.05180215835571289,
      "step": 5117
    },
    {
      "epoch": 0.6919372010883342,
      "grad_norm": 0.07045657187700272,
      "learning_rate": 7.411264633220305e-06,
      "loss": 0.04566240310668945,
      "step": 5118
    },
    {
      "epoch": 0.6920723978841702,
      "grad_norm": 0.04311591014266014,
      "learning_rate": 7.405361536286174e-06,
      "loss": 0.0449676513671875,
      "step": 5119
    },
    {
      "epoch": 0.692207594680006,
      "grad_norm": 0.09630683809518814,
      "learning_rate": 7.399460020596266e-06,
      "loss": 0.08295631408691406,
      "step": 5120
    },
    {
      "epoch": 0.692342791475842,
      "grad_norm": 0.07007215917110443,
      "learning_rate": 7.393560087379322e-06,
      "loss": 0.05119359493255615,
      "step": 5121
    },
    {
      "epoch": 0.692477988271678,
      "grad_norm": 0.06707391887903214,
      "learning_rate": 7.3876617378637195e-06,
      "loss": 0.04883003234863281,
      "step": 5122
    },
    {
      "epoch": 0.6926131850675139,
      "grad_norm": 0.07459626346826553,
      "learning_rate": 7.381764973277543e-06,
      "loss": 0.045946478843688965,
      "step": 5123
    },
    {
      "epoch": 0.6927483818633499,
      "grad_norm": 0.08041670173406601,
      "learning_rate": 7.375869794848525e-06,
      "loss": 0.07621145248413086,
      "step": 5124
    },
    {
      "epoch": 0.6928835786591858,
      "grad_norm": 0.0551561675965786,
      "learning_rate": 7.3699762038040654e-06,
      "loss": 0.05382108688354492,
      "step": 5125
    },
    {
      "epoch": 0.6930187754550217,
      "grad_norm": 0.17855863273143768,
      "learning_rate": 7.364084201371261e-06,
      "loss": 0.1080636978149414,
      "step": 5126
    },
    {
      "epoch": 0.6931539722508576,
      "grad_norm": 0.036853570491075516,
      "learning_rate": 7.3581937887768334e-06,
      "loss": 0.03311949968338013,
      "step": 5127
    },
    {
      "epoch": 0.6932891690466936,
      "grad_norm": 0.08787119388580322,
      "learning_rate": 7.352304967247217e-06,
      "loss": 0.04837656021118164,
      "step": 5128
    },
    {
      "epoch": 0.6934243658425295,
      "grad_norm": 0.056589942425489426,
      "learning_rate": 7.346417738008487e-06,
      "loss": 0.0705728530883789,
      "step": 5129
    },
    {
      "epoch": 0.6935595626383655,
      "grad_norm": 0.15202593803405762,
      "learning_rate": 7.340532102286399e-06,
      "loss": 0.07158017158508301,
      "step": 5130
    },
    {
      "epoch": 0.6936947594342014,
      "grad_norm": 0.0418815053999424,
      "learning_rate": 7.3346480613063725e-06,
      "loss": 0.03671407699584961,
      "step": 5131
    },
    {
      "epoch": 0.6938299562300374,
      "grad_norm": 0.10494381189346313,
      "learning_rate": 7.328765616293491e-06,
      "loss": 0.04256272315979004,
      "step": 5132
    },
    {
      "epoch": 0.6939651530258732,
      "grad_norm": 0.06368820369243622,
      "learning_rate": 7.322884768472521e-06,
      "loss": 0.06847715377807617,
      "step": 5133
    },
    {
      "epoch": 0.6941003498217092,
      "grad_norm": 0.044152189046144485,
      "learning_rate": 7.317005519067881e-06,
      "loss": 0.05760622024536133,
      "step": 5134
    },
    {
      "epoch": 0.6942355466175452,
      "grad_norm": 0.08222747594118118,
      "learning_rate": 7.311127869303665e-06,
      "loss": 0.039581298828125,
      "step": 5135
    },
    {
      "epoch": 0.6943707434133811,
      "grad_norm": 0.07158730924129486,
      "learning_rate": 7.305251820403628e-06,
      "loss": 0.051753997802734375,
      "step": 5136
    },
    {
      "epoch": 0.6945059402092171,
      "grad_norm": 0.08473771065473557,
      "learning_rate": 7.299377373591188e-06,
      "loss": 0.05568647384643555,
      "step": 5137
    },
    {
      "epoch": 0.694641137005053,
      "grad_norm": 0.09255312383174896,
      "learning_rate": 7.29350453008945e-06,
      "loss": 0.06441974639892578,
      "step": 5138
    },
    {
      "epoch": 0.6947763338008889,
      "grad_norm": 0.08644882589578629,
      "learning_rate": 7.287633291121166e-06,
      "loss": 0.0827341079711914,
      "step": 5139
    },
    {
      "epoch": 0.6949115305967248,
      "grad_norm": 0.07177211344242096,
      "learning_rate": 7.281763657908756e-06,
      "loss": 0.07648968696594238,
      "step": 5140
    },
    {
      "epoch": 0.6950467273925608,
      "grad_norm": 0.10879196971654892,
      "learning_rate": 7.275895631674313e-06,
      "loss": 0.0838766098022461,
      "step": 5141
    },
    {
      "epoch": 0.6951819241883967,
      "grad_norm": 0.033908288925886154,
      "learning_rate": 7.2700292136395826e-06,
      "loss": 0.029471397399902344,
      "step": 5142
    },
    {
      "epoch": 0.6953171209842327,
      "grad_norm": 0.11807725578546524,
      "learning_rate": 7.264164405025997e-06,
      "loss": 0.07389593124389648,
      "step": 5143
    },
    {
      "epoch": 0.6954523177800687,
      "grad_norm": 0.05082758143544197,
      "learning_rate": 7.2583012070546364e-06,
      "loss": 0.052376389503479004,
      "step": 5144
    },
    {
      "epoch": 0.6955875145759045,
      "grad_norm": 0.04344399645924568,
      "learning_rate": 7.252439620946247e-06,
      "loss": 0.05011838674545288,
      "step": 5145
    },
    {
      "epoch": 0.6957227113717405,
      "grad_norm": 0.09419479966163635,
      "learning_rate": 7.246579647921243e-06,
      "loss": 0.05389326810836792,
      "step": 5146
    },
    {
      "epoch": 0.6958579081675764,
      "grad_norm": 0.05912099406123161,
      "learning_rate": 7.240721289199699e-06,
      "loss": 0.043886661529541016,
      "step": 5147
    },
    {
      "epoch": 0.6959931049634124,
      "grad_norm": 0.051456380635499954,
      "learning_rate": 7.234864546001364e-06,
      "loss": 0.05203723907470703,
      "step": 5148
    },
    {
      "epoch": 0.6961283017592483,
      "grad_norm": 0.07783391326665878,
      "learning_rate": 7.229009419545638e-06,
      "loss": 0.061969757080078125,
      "step": 5149
    },
    {
      "epoch": 0.6962634985550843,
      "grad_norm": 0.053779236972332,
      "learning_rate": 7.223155911051593e-06,
      "loss": 0.053030967712402344,
      "step": 5150
    },
    {
      "epoch": 0.6963986953509201,
      "grad_norm": 0.12143230438232422,
      "learning_rate": 7.2173040217379575e-06,
      "loss": 0.07530605792999268,
      "step": 5151
    },
    {
      "epoch": 0.6965338921467561,
      "grad_norm": 0.061984140425920486,
      "learning_rate": 7.211453752823122e-06,
      "loss": 0.055178046226501465,
      "step": 5152
    },
    {
      "epoch": 0.696669088942592,
      "grad_norm": 0.11437245458364487,
      "learning_rate": 7.205605105525161e-06,
      "loss": 0.06825780868530273,
      "step": 5153
    },
    {
      "epoch": 0.696804285738428,
      "grad_norm": 0.05304133892059326,
      "learning_rate": 7.19975808106177e-06,
      "loss": 0.050585269927978516,
      "step": 5154
    },
    {
      "epoch": 0.696939482534264,
      "grad_norm": 0.0624159537255764,
      "learning_rate": 7.193912680650346e-06,
      "loss": 0.05103921890258789,
      "step": 5155
    },
    {
      "epoch": 0.6970746793300999,
      "grad_norm": 0.05065198987722397,
      "learning_rate": 7.188068905507931e-06,
      "loss": 0.06281852722167969,
      "step": 5156
    },
    {
      "epoch": 0.6972098761259358,
      "grad_norm": 0.10954926162958145,
      "learning_rate": 7.182226756851223e-06,
      "loss": 0.050206899642944336,
      "step": 5157
    },
    {
      "epoch": 0.6973450729217717,
      "grad_norm": 0.0819028839468956,
      "learning_rate": 7.176386235896603e-06,
      "loss": 0.06219756603240967,
      "step": 5158
    },
    {
      "epoch": 0.6974802697176077,
      "grad_norm": 0.05428193509578705,
      "learning_rate": 7.170547343860079e-06,
      "loss": 0.04502427577972412,
      "step": 5159
    },
    {
      "epoch": 0.6976154665134436,
      "grad_norm": 0.05840172991156578,
      "learning_rate": 7.164710081957355e-06,
      "loss": 0.056485652923583984,
      "step": 5160
    },
    {
      "epoch": 0.6977506633092796,
      "grad_norm": 0.11079628765583038,
      "learning_rate": 7.158874451403777e-06,
      "loss": 0.07830274105072021,
      "step": 5161
    },
    {
      "epoch": 0.6978858601051156,
      "grad_norm": 0.04507223516702652,
      "learning_rate": 7.15304045341435e-06,
      "loss": 0.039655983448028564,
      "step": 5162
    },
    {
      "epoch": 0.6980210569009514,
      "grad_norm": 0.049603331834077835,
      "learning_rate": 7.147208089203745e-06,
      "loss": 0.05478668212890625,
      "step": 5163
    },
    {
      "epoch": 0.6981562536967874,
      "grad_norm": 0.07919736951589584,
      "learning_rate": 7.141377359986288e-06,
      "loss": 0.06793379783630371,
      "step": 5164
    },
    {
      "epoch": 0.6982914504926233,
      "grad_norm": 0.06492088735103607,
      "learning_rate": 7.135548266975978e-06,
      "loss": 0.06708502769470215,
      "step": 5165
    },
    {
      "epoch": 0.6984266472884593,
      "grad_norm": 0.14459918439388275,
      "learning_rate": 7.129720811386456e-06,
      "loss": 0.06500530242919922,
      "step": 5166
    },
    {
      "epoch": 0.6985618440842952,
      "grad_norm": 0.06155581399798393,
      "learning_rate": 7.12389499443103e-06,
      "loss": 0.07036352157592773,
      "step": 5167
    },
    {
      "epoch": 0.6986970408801312,
      "grad_norm": 0.10594793409109116,
      "learning_rate": 7.118070817322668e-06,
      "loss": 0.04835796356201172,
      "step": 5168
    },
    {
      "epoch": 0.698832237675967,
      "grad_norm": 0.06809194386005402,
      "learning_rate": 7.1122482812739885e-06,
      "loss": 0.06902766227722168,
      "step": 5169
    },
    {
      "epoch": 0.698967434471803,
      "grad_norm": 0.07082079350948334,
      "learning_rate": 7.106427387497283e-06,
      "loss": 0.07391858100891113,
      "step": 5170
    },
    {
      "epoch": 0.6991026312676389,
      "grad_norm": 0.05770654231309891,
      "learning_rate": 7.10060813720449e-06,
      "loss": 0.05979514122009277,
      "step": 5171
    },
    {
      "epoch": 0.6992378280634749,
      "grad_norm": 0.10694356262683868,
      "learning_rate": 7.094790531607207e-06,
      "loss": 0.0682300329208374,
      "step": 5172
    },
    {
      "epoch": 0.6993730248593109,
      "grad_norm": 0.06676455587148666,
      "learning_rate": 7.088974571916692e-06,
      "loss": 0.04688286781311035,
      "step": 5173
    },
    {
      "epoch": 0.6995082216551468,
      "grad_norm": 0.09390151500701904,
      "learning_rate": 7.0831602593438515e-06,
      "loss": 0.05468571186065674,
      "step": 5174
    },
    {
      "epoch": 0.6996434184509827,
      "grad_norm": 0.05963120609521866,
      "learning_rate": 7.077347595099269e-06,
      "loss": 0.06016874313354492,
      "step": 5175
    },
    {
      "epoch": 0.6997786152468186,
      "grad_norm": 0.1497652530670166,
      "learning_rate": 7.071536580393166e-06,
      "loss": 0.06478071212768555,
      "step": 5176
    },
    {
      "epoch": 0.6999138120426546,
      "grad_norm": 0.05569586902856827,
      "learning_rate": 7.065727216435426e-06,
      "loss": 0.03982280194759369,
      "step": 5177
    },
    {
      "epoch": 0.7000490088384905,
      "grad_norm": 0.14853443205356598,
      "learning_rate": 7.05991950443559e-06,
      "loss": 0.07564735412597656,
      "step": 5178
    },
    {
      "epoch": 0.7001842056343265,
      "grad_norm": 0.04694928228855133,
      "learning_rate": 7.05411344560285e-06,
      "loss": 0.0429036021232605,
      "step": 5179
    },
    {
      "epoch": 0.7003194024301624,
      "grad_norm": 0.15553955733776093,
      "learning_rate": 7.048309041146069e-06,
      "loss": 0.08684754371643066,
      "step": 5180
    },
    {
      "epoch": 0.7004545992259984,
      "grad_norm": 0.0598907433450222,
      "learning_rate": 7.0425062922737495e-06,
      "loss": 0.07079172134399414,
      "step": 5181
    },
    {
      "epoch": 0.7005897960218342,
      "grad_norm": 0.10334526002407074,
      "learning_rate": 7.036705200194053e-06,
      "loss": 0.05799150466918945,
      "step": 5182
    },
    {
      "epoch": 0.7007249928176702,
      "grad_norm": 0.1398114264011383,
      "learning_rate": 7.0309057661148e-06,
      "loss": 0.07380008697509766,
      "step": 5183
    },
    {
      "epoch": 0.7008601896135062,
      "grad_norm": 0.09317490458488464,
      "learning_rate": 7.0251079912434565e-06,
      "loss": 0.06856918334960938,
      "step": 5184
    },
    {
      "epoch": 0.7009953864093421,
      "grad_norm": 0.09354618936777115,
      "learning_rate": 7.019311876787169e-06,
      "loss": 0.03395271301269531,
      "step": 5185
    },
    {
      "epoch": 0.7011305832051781,
      "grad_norm": 0.08463387936353683,
      "learning_rate": 7.013517423952696e-06,
      "loss": 0.0663614273071289,
      "step": 5186
    },
    {
      "epoch": 0.701265780001014,
      "grad_norm": 0.10186035931110382,
      "learning_rate": 7.0077246339464904e-06,
      "loss": 0.0855398178100586,
      "step": 5187
    },
    {
      "epoch": 0.7014009767968499,
      "grad_norm": 0.08779264986515045,
      "learning_rate": 7.001933507974635e-06,
      "loss": 0.05159878730773926,
      "step": 5188
    },
    {
      "epoch": 0.7015361735926858,
      "grad_norm": 0.07597818225622177,
      "learning_rate": 6.996144047242868e-06,
      "loss": 0.06801390647888184,
      "step": 5189
    },
    {
      "epoch": 0.7016713703885218,
      "grad_norm": 0.08194998651742935,
      "learning_rate": 6.9903562529566044e-06,
      "loss": 0.0679391622543335,
      "step": 5190
    },
    {
      "epoch": 0.7018065671843577,
      "grad_norm": 0.047737594693899155,
      "learning_rate": 6.984570126320869e-06,
      "loss": 0.042282700538635254,
      "step": 5191
    },
    {
      "epoch": 0.7019417639801937,
      "grad_norm": 0.052308209240436554,
      "learning_rate": 6.978785668540384e-06,
      "loss": 0.02779725193977356,
      "step": 5192
    },
    {
      "epoch": 0.7020769607760297,
      "grad_norm": 0.10460575670003891,
      "learning_rate": 6.973002880819496e-06,
      "loss": 0.07366108894348145,
      "step": 5193
    },
    {
      "epoch": 0.7022121575718655,
      "grad_norm": 0.12668466567993164,
      "learning_rate": 6.96722176436221e-06,
      "loss": 0.055776357650756836,
      "step": 5194
    },
    {
      "epoch": 0.7023473543677015,
      "grad_norm": 0.09756961464881897,
      "learning_rate": 6.9614423203721975e-06,
      "loss": 0.06728655099868774,
      "step": 5195
    },
    {
      "epoch": 0.7024825511635374,
      "grad_norm": 0.07483081519603729,
      "learning_rate": 6.955664550052749e-06,
      "loss": 0.04792451858520508,
      "step": 5196
    },
    {
      "epoch": 0.7026177479593734,
      "grad_norm": 0.049773622304201126,
      "learning_rate": 6.949888454606847e-06,
      "loss": 0.05487394332885742,
      "step": 5197
    },
    {
      "epoch": 0.7027529447552093,
      "grad_norm": 0.043582089245319366,
      "learning_rate": 6.944114035237095e-06,
      "loss": 0.041115760803222656,
      "step": 5198
    },
    {
      "epoch": 0.7028881415510453,
      "grad_norm": 0.059224683791399,
      "learning_rate": 6.93834129314576e-06,
      "loss": 0.052294254302978516,
      "step": 5199
    },
    {
      "epoch": 0.7030233383468811,
      "grad_norm": 0.20177409052848816,
      "learning_rate": 6.932570229534759e-06,
      "loss": 0.11620330810546875,
      "step": 5200
    },
    {
      "epoch": 0.7031585351427171,
      "grad_norm": 0.06465593725442886,
      "learning_rate": 6.9268008456056505e-06,
      "loss": 0.032606661319732666,
      "step": 5201
    },
    {
      "epoch": 0.703293731938553,
      "grad_norm": 0.07341928035020828,
      "learning_rate": 6.921033142559664e-06,
      "loss": 0.05399465560913086,
      "step": 5202
    },
    {
      "epoch": 0.703428928734389,
      "grad_norm": 0.08364728093147278,
      "learning_rate": 6.915267121597659e-06,
      "loss": 0.06726408004760742,
      "step": 5203
    },
    {
      "epoch": 0.703564125530225,
      "grad_norm": 0.09734231233596802,
      "learning_rate": 6.909502783920153e-06,
      "loss": 0.07239216566085815,
      "step": 5204
    },
    {
      "epoch": 0.7036993223260609,
      "grad_norm": 0.07396171987056732,
      "learning_rate": 6.903740130727312e-06,
      "loss": 0.07080841064453125,
      "step": 5205
    },
    {
      "epoch": 0.7038345191218968,
      "grad_norm": 0.1027669683098793,
      "learning_rate": 6.8979791632189425e-06,
      "loss": 0.08009147644042969,
      "step": 5206
    },
    {
      "epoch": 0.7039697159177327,
      "grad_norm": 0.15998680889606476,
      "learning_rate": 6.892219882594523e-06,
      "loss": 0.06155216693878174,
      "step": 5207
    },
    {
      "epoch": 0.7041049127135687,
      "grad_norm": 0.10081665217876434,
      "learning_rate": 6.886462290053159e-06,
      "loss": 0.0668749213218689,
      "step": 5208
    },
    {
      "epoch": 0.7042401095094046,
      "grad_norm": 0.06833750009536743,
      "learning_rate": 6.880706386793614e-06,
      "loss": 0.052366167306900024,
      "step": 5209
    },
    {
      "epoch": 0.7043753063052406,
      "grad_norm": 0.17746584117412567,
      "learning_rate": 6.874952174014298e-06,
      "loss": 0.08236157894134521,
      "step": 5210
    },
    {
      "epoch": 0.7045105031010765,
      "grad_norm": 0.05688696727156639,
      "learning_rate": 6.8691996529132585e-06,
      "loss": 0.05448341369628906,
      "step": 5211
    },
    {
      "epoch": 0.7046456998969124,
      "grad_norm": 0.11775568127632141,
      "learning_rate": 6.863448824688217e-06,
      "loss": 0.07089114189147949,
      "step": 5212
    },
    {
      "epoch": 0.7047808966927483,
      "grad_norm": 0.0966481864452362,
      "learning_rate": 6.857699690536521e-06,
      "loss": 0.054990530014038086,
      "step": 5213
    },
    {
      "epoch": 0.7049160934885843,
      "grad_norm": 0.06725800037384033,
      "learning_rate": 6.8519522516551685e-06,
      "loss": 0.05025172233581543,
      "step": 5214
    },
    {
      "epoch": 0.7050512902844203,
      "grad_norm": 0.07729680091142654,
      "learning_rate": 6.846206509240807e-06,
      "loss": 0.08900260925292969,
      "step": 5215
    },
    {
      "epoch": 0.7051864870802562,
      "grad_norm": 0.051595691591501236,
      "learning_rate": 6.840462464489726e-06,
      "loss": 0.04011797904968262,
      "step": 5216
    },
    {
      "epoch": 0.7053216838760922,
      "grad_norm": 0.1215374693274498,
      "learning_rate": 6.834720118597879e-06,
      "loss": 0.06545579433441162,
      "step": 5217
    },
    {
      "epoch": 0.705456880671928,
      "grad_norm": 0.08781389892101288,
      "learning_rate": 6.828979472760846e-06,
      "loss": 0.07096457481384277,
      "step": 5218
    },
    {
      "epoch": 0.705592077467764,
      "grad_norm": 0.06484188139438629,
      "learning_rate": 6.823240528173858e-06,
      "loss": 0.06389856338500977,
      "step": 5219
    },
    {
      "epoch": 0.7057272742635999,
      "grad_norm": 0.16192342340946198,
      "learning_rate": 6.817503286031797e-06,
      "loss": 0.061789512634277344,
      "step": 5220
    },
    {
      "epoch": 0.7058624710594359,
      "grad_norm": 0.05168599635362625,
      "learning_rate": 6.811767747529181e-06,
      "loss": 0.04684257507324219,
      "step": 5221
    },
    {
      "epoch": 0.7059976678552718,
      "grad_norm": 0.0708392783999443,
      "learning_rate": 6.806033913860195e-06,
      "loss": 0.06925821304321289,
      "step": 5222
    },
    {
      "epoch": 0.7061328646511078,
      "grad_norm": 0.08808889985084534,
      "learning_rate": 6.800301786218634e-06,
      "loss": 0.04602479934692383,
      "step": 5223
    },
    {
      "epoch": 0.7062680614469438,
      "grad_norm": 0.08793748915195465,
      "learning_rate": 6.794571365797971e-06,
      "loss": 0.04866921901702881,
      "step": 5224
    },
    {
      "epoch": 0.7064032582427796,
      "grad_norm": 0.1303250938653946,
      "learning_rate": 6.788842653791308e-06,
      "loss": 0.07110881805419922,
      "step": 5225
    },
    {
      "epoch": 0.7065384550386156,
      "grad_norm": 0.05211092531681061,
      "learning_rate": 6.7831156513913864e-06,
      "loss": 0.0505070686340332,
      "step": 5226
    },
    {
      "epoch": 0.7066736518344515,
      "grad_norm": 0.13809941709041595,
      "learning_rate": 6.777390359790614e-06,
      "loss": 0.08930253982543945,
      "step": 5227
    },
    {
      "epoch": 0.7068088486302875,
      "grad_norm": 0.16696114838123322,
      "learning_rate": 6.771666780181004e-06,
      "loss": 0.06232607364654541,
      "step": 5228
    },
    {
      "epoch": 0.7069440454261234,
      "grad_norm": 0.09382566064596176,
      "learning_rate": 6.765944913754258e-06,
      "loss": 0.02843475341796875,
      "step": 5229
    },
    {
      "epoch": 0.7070792422219594,
      "grad_norm": 0.07554718852043152,
      "learning_rate": 6.7602247617016885e-06,
      "loss": 0.053112149238586426,
      "step": 5230
    },
    {
      "epoch": 0.7072144390177952,
      "grad_norm": 0.12718260288238525,
      "learning_rate": 6.754506325214265e-06,
      "loss": 0.05199909210205078,
      "step": 5231
    },
    {
      "epoch": 0.7073496358136312,
      "grad_norm": 0.1223500519990921,
      "learning_rate": 6.748789605482593e-06,
      "loss": 0.08435928821563721,
      "step": 5232
    },
    {
      "epoch": 0.7074848326094672,
      "grad_norm": 0.17492032051086426,
      "learning_rate": 6.743074603696922e-06,
      "loss": 0.06446027755737305,
      "step": 5233
    },
    {
      "epoch": 0.7076200294053031,
      "grad_norm": 0.14729131758213043,
      "learning_rate": 6.737361321047155e-06,
      "loss": 0.0760049819946289,
      "step": 5234
    },
    {
      "epoch": 0.7077552262011391,
      "grad_norm": 0.051025256514549255,
      "learning_rate": 6.731649758722823e-06,
      "loss": 0.03641080856323242,
      "step": 5235
    },
    {
      "epoch": 0.707890422996975,
      "grad_norm": 0.16203363239765167,
      "learning_rate": 6.725939917913102e-06,
      "loss": 0.0928335189819336,
      "step": 5236
    },
    {
      "epoch": 0.7080256197928109,
      "grad_norm": 0.04803748428821564,
      "learning_rate": 6.720231799806814e-06,
      "loss": 0.03930628299713135,
      "step": 5237
    },
    {
      "epoch": 0.7081608165886468,
      "grad_norm": 0.07853997498750687,
      "learning_rate": 6.7145254055924136e-06,
      "loss": 0.056729793548583984,
      "step": 5238
    },
    {
      "epoch": 0.7082960133844828,
      "grad_norm": 0.06909221410751343,
      "learning_rate": 6.70882073645801e-06,
      "loss": 0.06695842742919922,
      "step": 5239
    },
    {
      "epoch": 0.7084312101803187,
      "grad_norm": 0.1401662975549698,
      "learning_rate": 6.703117793591346e-06,
      "loss": 0.06996715068817139,
      "step": 5240
    },
    {
      "epoch": 0.7085664069761547,
      "grad_norm": 0.08216295391321182,
      "learning_rate": 6.6974165781798e-06,
      "loss": 0.08788108825683594,
      "step": 5241
    },
    {
      "epoch": 0.7087016037719907,
      "grad_norm": 0.08670076727867126,
      "learning_rate": 6.691717091410398e-06,
      "loss": 0.08666515350341797,
      "step": 5242
    },
    {
      "epoch": 0.7088368005678265,
      "grad_norm": 0.08967714756727219,
      "learning_rate": 6.686019334469797e-06,
      "loss": 0.06845802068710327,
      "step": 5243
    },
    {
      "epoch": 0.7089719973636625,
      "grad_norm": 0.07060317695140839,
      "learning_rate": 6.680323308544312e-06,
      "loss": 0.06550812721252441,
      "step": 5244
    },
    {
      "epoch": 0.7091071941594984,
      "grad_norm": 0.07658001035451889,
      "learning_rate": 6.674629014819879e-06,
      "loss": 0.055979013442993164,
      "step": 5245
    },
    {
      "epoch": 0.7092423909553344,
      "grad_norm": 0.13820207118988037,
      "learning_rate": 6.668936454482082e-06,
      "loss": 0.07737338542938232,
      "step": 5246
    },
    {
      "epoch": 0.7093775877511703,
      "grad_norm": 0.12385187298059464,
      "learning_rate": 6.6632456287161426e-06,
      "loss": 0.06941461563110352,
      "step": 5247
    },
    {
      "epoch": 0.7095127845470063,
      "grad_norm": 0.14194141328334808,
      "learning_rate": 6.657556538706914e-06,
      "loss": 0.09146249294281006,
      "step": 5248
    },
    {
      "epoch": 0.7096479813428421,
      "grad_norm": 0.07196071743965149,
      "learning_rate": 6.651869185638907e-06,
      "loss": 0.0751657485961914,
      "step": 5249
    },
    {
      "epoch": 0.7097831781386781,
      "grad_norm": 0.07490207999944687,
      "learning_rate": 6.646183570696253e-06,
      "loss": 0.08531475067138672,
      "step": 5250
    },
    {
      "epoch": 0.709918374934514,
      "grad_norm": 0.11638922244310379,
      "learning_rate": 6.6404996950627275e-06,
      "loss": 0.06609213352203369,
      "step": 5251
    },
    {
      "epoch": 0.71005357173035,
      "grad_norm": 0.08221335709095001,
      "learning_rate": 6.634817559921744e-06,
      "loss": 0.06448519229888916,
      "step": 5252
    },
    {
      "epoch": 0.710188768526186,
      "grad_norm": 0.09152976423501968,
      "learning_rate": 6.629137166456348e-06,
      "loss": 0.057424187660217285,
      "step": 5253
    },
    {
      "epoch": 0.7103239653220219,
      "grad_norm": 0.10937793552875519,
      "learning_rate": 6.623458515849244e-06,
      "loss": 0.08165574073791504,
      "step": 5254
    },
    {
      "epoch": 0.7104591621178578,
      "grad_norm": 0.11498243361711502,
      "learning_rate": 6.6177816092827354e-06,
      "loss": 0.04826921224594116,
      "step": 5255
    },
    {
      "epoch": 0.7105943589136937,
      "grad_norm": 0.04066452011466026,
      "learning_rate": 6.6121064479388e-06,
      "loss": 0.03294992446899414,
      "step": 5256
    },
    {
      "epoch": 0.7107295557095297,
      "grad_norm": 0.14823363721370697,
      "learning_rate": 6.606433032999031e-06,
      "loss": 0.07320308685302734,
      "step": 5257
    },
    {
      "epoch": 0.7108647525053656,
      "grad_norm": 0.0708981454372406,
      "learning_rate": 6.60076136564466e-06,
      "loss": 0.04673290252685547,
      "step": 5258
    },
    {
      "epoch": 0.7109999493012016,
      "grad_norm": 0.07480026036500931,
      "learning_rate": 6.595091447056574e-06,
      "loss": 0.07643890380859375,
      "step": 5259
    },
    {
      "epoch": 0.7111351460970375,
      "grad_norm": 0.06935848295688629,
      "learning_rate": 6.589423278415259e-06,
      "loss": 0.052535563707351685,
      "step": 5260
    },
    {
      "epoch": 0.7112703428928734,
      "grad_norm": 0.07488243281841278,
      "learning_rate": 6.583756860900872e-06,
      "loss": 0.061678290367126465,
      "step": 5261
    },
    {
      "epoch": 0.7114055396887093,
      "grad_norm": 0.03968992084264755,
      "learning_rate": 6.578092195693187e-06,
      "loss": 0.031808674335479736,
      "step": 5262
    },
    {
      "epoch": 0.7115407364845453,
      "grad_norm": 0.12007761746644974,
      "learning_rate": 6.572429283971614e-06,
      "loss": 0.08109569549560547,
      "step": 5263
    },
    {
      "epoch": 0.7116759332803813,
      "grad_norm": 0.07446523755788803,
      "learning_rate": 6.566768126915215e-06,
      "loss": 0.06934309005737305,
      "step": 5264
    },
    {
      "epoch": 0.7118111300762172,
      "grad_norm": 0.06147582828998566,
      "learning_rate": 6.561108725702653e-06,
      "loss": 0.06867790222167969,
      "step": 5265
    },
    {
      "epoch": 0.7119463268720532,
      "grad_norm": 0.05361782759428024,
      "learning_rate": 6.555451081512262e-06,
      "loss": 0.03309512138366699,
      "step": 5266
    },
    {
      "epoch": 0.712081523667889,
      "grad_norm": 0.09725134074687958,
      "learning_rate": 6.549795195521988e-06,
      "loss": 0.05037188529968262,
      "step": 5267
    },
    {
      "epoch": 0.712216720463725,
      "grad_norm": 0.06769993156194687,
      "learning_rate": 6.544141068909416e-06,
      "loss": 0.06625199317932129,
      "step": 5268
    },
    {
      "epoch": 0.7123519172595609,
      "grad_norm": 0.09040994197130203,
      "learning_rate": 6.5384887028517645e-06,
      "loss": 0.0701906681060791,
      "step": 5269
    },
    {
      "epoch": 0.7124871140553969,
      "grad_norm": 0.09078021347522736,
      "learning_rate": 6.532838098525883e-06,
      "loss": 0.07329720258712769,
      "step": 5270
    },
    {
      "epoch": 0.7126223108512328,
      "grad_norm": 0.10193806886672974,
      "learning_rate": 6.5271892571082655e-06,
      "loss": 0.05244690179824829,
      "step": 5271
    },
    {
      "epoch": 0.7127575076470688,
      "grad_norm": 0.05036259442567825,
      "learning_rate": 6.521542179775029e-06,
      "loss": 0.061924099922180176,
      "step": 5272
    },
    {
      "epoch": 0.7128927044429048,
      "grad_norm": 0.04565594345331192,
      "learning_rate": 6.515896867701924e-06,
      "loss": 0.033681511878967285,
      "step": 5273
    },
    {
      "epoch": 0.7130279012387406,
      "grad_norm": 0.06286714971065521,
      "learning_rate": 6.510253322064333e-06,
      "loss": 0.057758212089538574,
      "step": 5274
    },
    {
      "epoch": 0.7131630980345766,
      "grad_norm": 0.09807729721069336,
      "learning_rate": 6.504611544037267e-06,
      "loss": 0.07482337951660156,
      "step": 5275
    },
    {
      "epoch": 0.7132982948304125,
      "grad_norm": 0.08780847489833832,
      "learning_rate": 6.498971534795387e-06,
      "loss": 0.06775462627410889,
      "step": 5276
    },
    {
      "epoch": 0.7134334916262485,
      "grad_norm": 0.0908341109752655,
      "learning_rate": 6.493333295512965e-06,
      "loss": 0.07049250602722168,
      "step": 5277
    },
    {
      "epoch": 0.7135686884220844,
      "grad_norm": 0.07731003314256668,
      "learning_rate": 6.487696827363916e-06,
      "loss": 0.06020742654800415,
      "step": 5278
    },
    {
      "epoch": 0.7137038852179204,
      "grad_norm": 0.08584830164909363,
      "learning_rate": 6.48206213152178e-06,
      "loss": 0.04250943660736084,
      "step": 5279
    },
    {
      "epoch": 0.7138390820137562,
      "grad_norm": 0.08685700595378876,
      "learning_rate": 6.476429209159725e-06,
      "loss": 0.08098793029785156,
      "step": 5280
    },
    {
      "epoch": 0.7139742788095922,
      "grad_norm": 0.06242678314447403,
      "learning_rate": 6.470798061450568e-06,
      "loss": 0.0533905029296875,
      "step": 5281
    },
    {
      "epoch": 0.7141094756054281,
      "grad_norm": 0.09067007154226303,
      "learning_rate": 6.465168689566738e-06,
      "loss": 0.07322447001934052,
      "step": 5282
    },
    {
      "epoch": 0.7142446724012641,
      "grad_norm": 0.05833905190229416,
      "learning_rate": 6.4595410946803e-06,
      "loss": 0.062474608421325684,
      "step": 5283
    },
    {
      "epoch": 0.7143798691971001,
      "grad_norm": 0.10698240250349045,
      "learning_rate": 6.453915277962948e-06,
      "loss": 0.052127957344055176,
      "step": 5284
    },
    {
      "epoch": 0.714515065992936,
      "grad_norm": 0.0889761745929718,
      "learning_rate": 6.4482912405860055e-06,
      "loss": 0.05313444137573242,
      "step": 5285
    },
    {
      "epoch": 0.7146502627887719,
      "grad_norm": 0.08653388172388077,
      "learning_rate": 6.442668983720434e-06,
      "loss": 0.08327579498291016,
      "step": 5286
    },
    {
      "epoch": 0.7147854595846078,
      "grad_norm": 0.14286509156227112,
      "learning_rate": 6.437048508536813e-06,
      "loss": 0.08691072463989258,
      "step": 5287
    },
    {
      "epoch": 0.7149206563804438,
      "grad_norm": 0.05450690537691116,
      "learning_rate": 6.431429816205357e-06,
      "loss": 0.0380479097366333,
      "step": 5288
    },
    {
      "epoch": 0.7150558531762797,
      "grad_norm": 0.1602146029472351,
      "learning_rate": 6.425812907895904e-06,
      "loss": 0.07602882385253906,
      "step": 5289
    },
    {
      "epoch": 0.7151910499721157,
      "grad_norm": 0.04617540165781975,
      "learning_rate": 6.420197784777925e-06,
      "loss": 0.036473244428634644,
      "step": 5290
    },
    {
      "epoch": 0.7153262467679516,
      "grad_norm": 0.06446775794029236,
      "learning_rate": 6.414584448020528e-06,
      "loss": 0.046510934829711914,
      "step": 5291
    },
    {
      "epoch": 0.7154614435637875,
      "grad_norm": 0.18855026364326477,
      "learning_rate": 6.408972898792423e-06,
      "loss": 0.07486724853515625,
      "step": 5292
    },
    {
      "epoch": 0.7155966403596234,
      "grad_norm": 0.0894274190068245,
      "learning_rate": 6.4033631382619766e-06,
      "loss": 0.05085396766662598,
      "step": 5293
    },
    {
      "epoch": 0.7157318371554594,
      "grad_norm": 0.09192391484975815,
      "learning_rate": 6.397755167597171e-06,
      "loss": 0.05952167510986328,
      "step": 5294
    },
    {
      "epoch": 0.7158670339512954,
      "grad_norm": 0.07320386916399002,
      "learning_rate": 6.392148987965603e-06,
      "loss": 0.06919527053833008,
      "step": 5295
    },
    {
      "epoch": 0.7160022307471313,
      "grad_norm": 0.16711179912090302,
      "learning_rate": 6.386544600534532e-06,
      "loss": 0.06976819038391113,
      "step": 5296
    },
    {
      "epoch": 0.7161374275429673,
      "grad_norm": 0.07418635487556458,
      "learning_rate": 6.3809420064707965e-06,
      "loss": 0.05153822898864746,
      "step": 5297
    },
    {
      "epoch": 0.7162726243388031,
      "grad_norm": 0.05701493099331856,
      "learning_rate": 6.375341206940902e-06,
      "loss": 0.04898262023925781,
      "step": 5298
    },
    {
      "epoch": 0.7164078211346391,
      "grad_norm": 0.07073450833559036,
      "learning_rate": 6.369742203110962e-06,
      "loss": 0.07201194763183594,
      "step": 5299
    },
    {
      "epoch": 0.716543017930475,
      "grad_norm": 0.05002342909574509,
      "learning_rate": 6.364144996146716e-06,
      "loss": 0.04228997230529785,
      "step": 5300
    },
    {
      "epoch": 0.716678214726311,
      "grad_norm": 0.10059481114149094,
      "learning_rate": 6.358549587213534e-06,
      "loss": 0.05995523929595947,
      "step": 5301
    },
    {
      "epoch": 0.716813411522147,
      "grad_norm": 0.10729820281267166,
      "learning_rate": 6.352955977476405e-06,
      "loss": 0.05074715614318848,
      "step": 5302
    },
    {
      "epoch": 0.7169486083179829,
      "grad_norm": 0.042587559670209885,
      "learning_rate": 6.347364168099959e-06,
      "loss": 0.043564677238464355,
      "step": 5303
    },
    {
      "epoch": 0.7170838051138188,
      "grad_norm": 0.0937381163239479,
      "learning_rate": 6.341774160248435e-06,
      "loss": 0.06353425979614258,
      "step": 5304
    },
    {
      "epoch": 0.7172190019096547,
      "grad_norm": 0.10774174332618713,
      "learning_rate": 6.3361859550857e-06,
      "loss": 0.06760931015014648,
      "step": 5305
    },
    {
      "epoch": 0.7173541987054907,
      "grad_norm": 0.09925224632024765,
      "learning_rate": 6.330599553775252e-06,
      "loss": 0.06612062454223633,
      "step": 5306
    },
    {
      "epoch": 0.7174893955013266,
      "grad_norm": 0.0786658525466919,
      "learning_rate": 6.325014957480203e-06,
      "loss": 0.04143184423446655,
      "step": 5307
    },
    {
      "epoch": 0.7176245922971626,
      "grad_norm": 0.1537683606147766,
      "learning_rate": 6.319432167363305e-06,
      "loss": 0.08494281768798828,
      "step": 5308
    },
    {
      "epoch": 0.7177597890929985,
      "grad_norm": 0.0969962477684021,
      "learning_rate": 6.313851184586918e-06,
      "loss": 0.0869135856628418,
      "step": 5309
    },
    {
      "epoch": 0.7178949858888344,
      "grad_norm": 0.07230785489082336,
      "learning_rate": 6.308272010313037e-06,
      "loss": 0.05916094779968262,
      "step": 5310
    },
    {
      "epoch": 0.7180301826846703,
      "grad_norm": 0.04013494774699211,
      "learning_rate": 6.302694645703273e-06,
      "loss": 0.039933621883392334,
      "step": 5311
    },
    {
      "epoch": 0.7181653794805063,
      "grad_norm": 0.2185848355293274,
      "learning_rate": 6.297119091918857e-06,
      "loss": 0.09819304943084717,
      "step": 5312
    },
    {
      "epoch": 0.7183005762763423,
      "grad_norm": 0.09592147916555405,
      "learning_rate": 6.2915453501206634e-06,
      "loss": 0.04529118537902832,
      "step": 5313
    },
    {
      "epoch": 0.7184357730721782,
      "grad_norm": 0.08197664469480515,
      "learning_rate": 6.285973421469166e-06,
      "loss": 0.07849514484405518,
      "step": 5314
    },
    {
      "epoch": 0.7185709698680142,
      "grad_norm": 0.2049938142299652,
      "learning_rate": 6.28040330712447e-06,
      "loss": 0.08316230773925781,
      "step": 5315
    },
    {
      "epoch": 0.7187061666638501,
      "grad_norm": 0.06143787503242493,
      "learning_rate": 6.274835008246304e-06,
      "loss": 0.04585576057434082,
      "step": 5316
    },
    {
      "epoch": 0.718841363459686,
      "grad_norm": 0.061775028705596924,
      "learning_rate": 6.269268525994013e-06,
      "loss": 0.06304037570953369,
      "step": 5317
    },
    {
      "epoch": 0.7189765602555219,
      "grad_norm": 0.11862292885780334,
      "learning_rate": 6.263703861526578e-06,
      "loss": 0.07857632637023926,
      "step": 5318
    },
    {
      "epoch": 0.7191117570513579,
      "grad_norm": 0.037855781614780426,
      "learning_rate": 6.258141016002587e-06,
      "loss": 0.035109519958496094,
      "step": 5319
    },
    {
      "epoch": 0.7192469538471938,
      "grad_norm": 0.09372851252555847,
      "learning_rate": 6.252579990580254e-06,
      "loss": 0.057793617248535156,
      "step": 5320
    },
    {
      "epoch": 0.7193821506430298,
      "grad_norm": 0.055855777114629745,
      "learning_rate": 6.247020786417412e-06,
      "loss": 0.06108808517456055,
      "step": 5321
    },
    {
      "epoch": 0.7195173474388658,
      "grad_norm": 0.10095803439617157,
      "learning_rate": 6.241463404671516e-06,
      "loss": 0.06770330667495728,
      "step": 5322
    },
    {
      "epoch": 0.7196525442347016,
      "grad_norm": 0.061530306935310364,
      "learning_rate": 6.235907846499655e-06,
      "loss": 0.06612920761108398,
      "step": 5323
    },
    {
      "epoch": 0.7197877410305376,
      "grad_norm": 0.044040828943252563,
      "learning_rate": 6.230354113058505e-06,
      "loss": 0.04344582557678223,
      "step": 5324
    },
    {
      "epoch": 0.7199229378263735,
      "grad_norm": 0.06790877878665924,
      "learning_rate": 6.2248022055044e-06,
      "loss": 0.04950857162475586,
      "step": 5325
    },
    {
      "epoch": 0.7200581346222095,
      "grad_norm": 0.07887686043977737,
      "learning_rate": 6.219252124993271e-06,
      "loss": 0.05790352821350098,
      "step": 5326
    },
    {
      "epoch": 0.7201933314180454,
      "grad_norm": 0.05709953233599663,
      "learning_rate": 6.213703872680668e-06,
      "loss": 0.07248306274414062,
      "step": 5327
    },
    {
      "epoch": 0.7203285282138814,
      "grad_norm": 0.09998508542776108,
      "learning_rate": 6.208157449721785e-06,
      "loss": 0.07018446922302246,
      "step": 5328
    },
    {
      "epoch": 0.7204637250097172,
      "grad_norm": 0.14174021780490875,
      "learning_rate": 6.202612857271393e-06,
      "loss": 0.08308422565460205,
      "step": 5329
    },
    {
      "epoch": 0.7205989218055532,
      "grad_norm": 0.05510513857007027,
      "learning_rate": 6.197070096483923e-06,
      "loss": 0.05639457702636719,
      "step": 5330
    },
    {
      "epoch": 0.7207341186013891,
      "grad_norm": 0.06637461483478546,
      "learning_rate": 6.191529168513403e-06,
      "loss": 0.06119728088378906,
      "step": 5331
    },
    {
      "epoch": 0.7208693153972251,
      "grad_norm": 0.07741259783506393,
      "learning_rate": 6.1859900745134755e-06,
      "loss": 0.0739288330078125,
      "step": 5332
    },
    {
      "epoch": 0.721004512193061,
      "grad_norm": 0.0876217633485794,
      "learning_rate": 6.180452815637429e-06,
      "loss": 0.07354676723480225,
      "step": 5333
    },
    {
      "epoch": 0.721139708988897,
      "grad_norm": 0.1901162564754486,
      "learning_rate": 6.174917393038126e-06,
      "loss": 0.07827091217041016,
      "step": 5334
    },
    {
      "epoch": 0.7212749057847329,
      "grad_norm": 0.08351711928844452,
      "learning_rate": 6.169383807868088e-06,
      "loss": 0.06247431039810181,
      "step": 5335
    },
    {
      "epoch": 0.7214101025805688,
      "grad_norm": 0.10454121977090836,
      "learning_rate": 6.163852061279432e-06,
      "loss": 0.05945134162902832,
      "step": 5336
    },
    {
      "epoch": 0.7215452993764048,
      "grad_norm": 0.09074094146490097,
      "learning_rate": 6.158322154423897e-06,
      "loss": 0.06362414360046387,
      "step": 5337
    },
    {
      "epoch": 0.7216804961722407,
      "grad_norm": 0.041065674275159836,
      "learning_rate": 6.15279408845284e-06,
      "loss": 0.03328752517700195,
      "step": 5338
    },
    {
      "epoch": 0.7218156929680767,
      "grad_norm": 0.1519712507724762,
      "learning_rate": 6.147267864517226e-06,
      "loss": 0.05492144823074341,
      "step": 5339
    },
    {
      "epoch": 0.7219508897639126,
      "grad_norm": 0.06271402537822723,
      "learning_rate": 6.141743483767658e-06,
      "loss": 0.03983864188194275,
      "step": 5340
    },
    {
      "epoch": 0.7220860865597485,
      "grad_norm": 0.070847287774086,
      "learning_rate": 6.136220947354333e-06,
      "loss": 0.07993173599243164,
      "step": 5341
    },
    {
      "epoch": 0.7222212833555844,
      "grad_norm": 0.10266253352165222,
      "learning_rate": 6.130700256427075e-06,
      "loss": 0.0762782096862793,
      "step": 5342
    },
    {
      "epoch": 0.7223564801514204,
      "grad_norm": 0.0735422745347023,
      "learning_rate": 6.1251814121353204e-06,
      "loss": 0.06935369968414307,
      "step": 5343
    },
    {
      "epoch": 0.7224916769472564,
      "grad_norm": 0.05423269793391228,
      "learning_rate": 6.1196644156281175e-06,
      "loss": 0.05670738220214844,
      "step": 5344
    },
    {
      "epoch": 0.7226268737430923,
      "grad_norm": 0.04924255609512329,
      "learning_rate": 6.114149268054143e-06,
      "loss": 0.035392045974731445,
      "step": 5345
    },
    {
      "epoch": 0.7227620705389283,
      "grad_norm": 0.17496107518672943,
      "learning_rate": 6.108635970561679e-06,
      "loss": 0.09158384799957275,
      "step": 5346
    },
    {
      "epoch": 0.7228972673347641,
      "grad_norm": 0.04416352137923241,
      "learning_rate": 6.103124524298617e-06,
      "loss": 0.04716068506240845,
      "step": 5347
    },
    {
      "epoch": 0.7230324641306001,
      "grad_norm": 0.05858367681503296,
      "learning_rate": 6.097614930412475e-06,
      "loss": 0.050646618008613586,
      "step": 5348
    },
    {
      "epoch": 0.723167660926436,
      "grad_norm": 0.09099132567644119,
      "learning_rate": 6.092107190050371e-06,
      "loss": 0.06693851947784424,
      "step": 5349
    },
    {
      "epoch": 0.723302857722272,
      "grad_norm": 0.07344671338796616,
      "learning_rate": 6.086601304359059e-06,
      "loss": 0.05541038513183594,
      "step": 5350
    },
    {
      "epoch": 0.7234380545181079,
      "grad_norm": 0.04452231898903847,
      "learning_rate": 6.081097274484887e-06,
      "loss": 0.04966890811920166,
      "step": 5351
    },
    {
      "epoch": 0.7235732513139439,
      "grad_norm": 0.047814320772886276,
      "learning_rate": 6.075595101573825e-06,
      "loss": 0.04716920852661133,
      "step": 5352
    },
    {
      "epoch": 0.7237084481097797,
      "grad_norm": 0.07035131752490997,
      "learning_rate": 6.070094786771451e-06,
      "loss": 0.05995357036590576,
      "step": 5353
    },
    {
      "epoch": 0.7238436449056157,
      "grad_norm": 0.09810557216405869,
      "learning_rate": 6.06459633122296e-06,
      "loss": 0.07535481452941895,
      "step": 5354
    },
    {
      "epoch": 0.7239788417014517,
      "grad_norm": 0.05407467484474182,
      "learning_rate": 6.059099736073166e-06,
      "loss": 0.054631948471069336,
      "step": 5355
    },
    {
      "epoch": 0.7241140384972876,
      "grad_norm": 0.0503917932510376,
      "learning_rate": 6.0536050024664865e-06,
      "loss": 0.05037069320678711,
      "step": 5356
    },
    {
      "epoch": 0.7242492352931236,
      "grad_norm": 0.086210235953331,
      "learning_rate": 6.048112131546953e-06,
      "loss": 0.07773375511169434,
      "step": 5357
    },
    {
      "epoch": 0.7243844320889595,
      "grad_norm": 0.10230488330125809,
      "learning_rate": 6.0426211244582105e-06,
      "loss": 0.08366107940673828,
      "step": 5358
    },
    {
      "epoch": 0.7245196288847955,
      "grad_norm": 0.09285559505224228,
      "learning_rate": 6.03713198234351e-06,
      "loss": 0.08132076263427734,
      "step": 5359
    },
    {
      "epoch": 0.7246548256806313,
      "grad_norm": 0.06652617454528809,
      "learning_rate": 6.0316447063457395e-06,
      "loss": 0.05439925193786621,
      "step": 5360
    },
    {
      "epoch": 0.7247900224764673,
      "grad_norm": 0.09366998821496964,
      "learning_rate": 6.026159297607356e-06,
      "loss": 0.07218027114868164,
      "step": 5361
    },
    {
      "epoch": 0.7249252192723032,
      "grad_norm": 0.06269343197345734,
      "learning_rate": 6.020675757270466e-06,
      "loss": 0.05474060773849487,
      "step": 5362
    },
    {
      "epoch": 0.7250604160681392,
      "grad_norm": 0.05139464512467384,
      "learning_rate": 6.015194086476766e-06,
      "loss": 0.04461979866027832,
      "step": 5363
    },
    {
      "epoch": 0.7251956128639752,
      "grad_norm": 0.09796824306249619,
      "learning_rate": 6.009714286367565e-06,
      "loss": 0.06590306758880615,
      "step": 5364
    },
    {
      "epoch": 0.7253308096598111,
      "grad_norm": 0.062123388051986694,
      "learning_rate": 6.004236358083802e-06,
      "loss": 0.04872941970825195,
      "step": 5365
    },
    {
      "epoch": 0.725466006455647,
      "grad_norm": 0.10065522789955139,
      "learning_rate": 5.998760302765989e-06,
      "loss": 0.04776954650878906,
      "step": 5366
    },
    {
      "epoch": 0.7256012032514829,
      "grad_norm": 0.12137159705162048,
      "learning_rate": 5.993286121554289e-06,
      "loss": 0.09652739763259888,
      "step": 5367
    },
    {
      "epoch": 0.7257364000473189,
      "grad_norm": 0.10050668567419052,
      "learning_rate": 5.987813815588447e-06,
      "loss": 0.07247316837310791,
      "step": 5368
    },
    {
      "epoch": 0.7258715968431548,
      "grad_norm": 0.10011740028858185,
      "learning_rate": 5.982343386007827e-06,
      "loss": 0.06283712387084961,
      "step": 5369
    },
    {
      "epoch": 0.7260067936389908,
      "grad_norm": 0.06181425228714943,
      "learning_rate": 5.976874833951404e-06,
      "loss": 0.037781357765197754,
      "step": 5370
    },
    {
      "epoch": 0.7261419904348267,
      "grad_norm": 0.046439073979854584,
      "learning_rate": 5.971408160557751e-06,
      "loss": 0.043087005615234375,
      "step": 5371
    },
    {
      "epoch": 0.7262771872306626,
      "grad_norm": 0.12095876038074493,
      "learning_rate": 5.965943366965069e-06,
      "loss": 0.07113271951675415,
      "step": 5372
    },
    {
      "epoch": 0.7264123840264985,
      "grad_norm": 0.09663762897253036,
      "learning_rate": 5.960480454311155e-06,
      "loss": 0.08069181442260742,
      "step": 5373
    },
    {
      "epoch": 0.7265475808223345,
      "grad_norm": 0.11518411338329315,
      "learning_rate": 5.955019423733416e-06,
      "loss": 0.06019997596740723,
      "step": 5374
    },
    {
      "epoch": 0.7266827776181705,
      "grad_norm": 0.05858995020389557,
      "learning_rate": 5.949560276368866e-06,
      "loss": 0.06565618515014648,
      "step": 5375
    },
    {
      "epoch": 0.7268179744140064,
      "grad_norm": 0.1236773282289505,
      "learning_rate": 5.9441030133541235e-06,
      "loss": 0.050461649894714355,
      "step": 5376
    },
    {
      "epoch": 0.7269531712098424,
      "grad_norm": 0.13728676736354828,
      "learning_rate": 5.938647635825432e-06,
      "loss": 0.07996439933776855,
      "step": 5377
    },
    {
      "epoch": 0.7270883680056782,
      "grad_norm": 0.1156599372625351,
      "learning_rate": 5.933194144918623e-06,
      "loss": 0.1054983139038086,
      "step": 5378
    },
    {
      "epoch": 0.7272235648015142,
      "grad_norm": 0.0730336382985115,
      "learning_rate": 5.927742541769142e-06,
      "loss": 0.07008910179138184,
      "step": 5379
    },
    {
      "epoch": 0.7273587615973501,
      "grad_norm": 0.07192105799913406,
      "learning_rate": 5.9222928275120445e-06,
      "loss": 0.04196673631668091,
      "step": 5380
    },
    {
      "epoch": 0.7274939583931861,
      "grad_norm": 0.054765958338975906,
      "learning_rate": 5.916845003281983e-06,
      "loss": 0.061878979206085205,
      "step": 5381
    },
    {
      "epoch": 0.727629155189022,
      "grad_norm": 0.08369892090559006,
      "learning_rate": 5.911399070213234e-06,
      "loss": 0.053739070892333984,
      "step": 5382
    },
    {
      "epoch": 0.727764351984858,
      "grad_norm": 0.08738262206315994,
      "learning_rate": 5.905955029439665e-06,
      "loss": 0.06627261638641357,
      "step": 5383
    },
    {
      "epoch": 0.7278995487806939,
      "grad_norm": 0.05693286657333374,
      "learning_rate": 5.900512882094754e-06,
      "loss": 0.04614543914794922,
      "step": 5384
    },
    {
      "epoch": 0.7280347455765298,
      "grad_norm": 0.07425574213266373,
      "learning_rate": 5.8950726293115855e-06,
      "loss": 0.07261180877685547,
      "step": 5385
    },
    {
      "epoch": 0.7281699423723658,
      "grad_norm": 0.061569489538669586,
      "learning_rate": 5.889634272222844e-06,
      "loss": 0.06549835205078125,
      "step": 5386
    },
    {
      "epoch": 0.7283051391682017,
      "grad_norm": 0.08389081805944443,
      "learning_rate": 5.8841978119608345e-06,
      "loss": 0.05438733100891113,
      "step": 5387
    },
    {
      "epoch": 0.7284403359640377,
      "grad_norm": 0.04788355156779289,
      "learning_rate": 5.878763249657452e-06,
      "loss": 0.04955792427062988,
      "step": 5388
    },
    {
      "epoch": 0.7285755327598736,
      "grad_norm": 0.09368545562028885,
      "learning_rate": 5.873330586444202e-06,
      "loss": 0.06662511825561523,
      "step": 5389
    },
    {
      "epoch": 0.7287107295557095,
      "grad_norm": 0.08902400732040405,
      "learning_rate": 5.867899823452193e-06,
      "loss": 0.06015467643737793,
      "step": 5390
    },
    {
      "epoch": 0.7288459263515454,
      "grad_norm": 0.046715009957551956,
      "learning_rate": 5.862470961812133e-06,
      "loss": 0.03995227813720703,
      "step": 5391
    },
    {
      "epoch": 0.7289811231473814,
      "grad_norm": 0.17600621283054352,
      "learning_rate": 5.857044002654357e-06,
      "loss": 0.06038618087768555,
      "step": 5392
    },
    {
      "epoch": 0.7291163199432174,
      "grad_norm": 0.052368585020303726,
      "learning_rate": 5.851618947108764e-06,
      "loss": 0.04511260986328125,
      "step": 5393
    },
    {
      "epoch": 0.7292515167390533,
      "grad_norm": 0.0867835059762001,
      "learning_rate": 5.8461957963048984e-06,
      "loss": 0.046033382415771484,
      "step": 5394
    },
    {
      "epoch": 0.7293867135348893,
      "grad_norm": 0.09164764732122421,
      "learning_rate": 5.840774551371882e-06,
      "loss": 0.06015515327453613,
      "step": 5395
    },
    {
      "epoch": 0.7295219103307251,
      "grad_norm": 0.06385641545057297,
      "learning_rate": 5.8353552134384405e-06,
      "loss": 0.0616612434387207,
      "step": 5396
    },
    {
      "epoch": 0.7296571071265611,
      "grad_norm": 0.03947180137038231,
      "learning_rate": 5.829937783632926e-06,
      "loss": 0.04448610544204712,
      "step": 5397
    },
    {
      "epoch": 0.729792303922397,
      "grad_norm": 0.11268360167741776,
      "learning_rate": 5.824522263083256e-06,
      "loss": 0.05181407928466797,
      "step": 5398
    },
    {
      "epoch": 0.729927500718233,
      "grad_norm": 0.08491433411836624,
      "learning_rate": 5.8191086529169855e-06,
      "loss": 0.06472170352935791,
      "step": 5399
    },
    {
      "epoch": 0.7300626975140689,
      "grad_norm": 0.12871123850345612,
      "learning_rate": 5.813696954261253e-06,
      "loss": 0.05559062957763672,
      "step": 5400
    },
    {
      "epoch": 0.7301978943099049,
      "grad_norm": 0.08232863247394562,
      "learning_rate": 5.8082871682428e-06,
      "loss": 0.06604957580566406,
      "step": 5401
    },
    {
      "epoch": 0.7303330911057407,
      "grad_norm": 0.15112295746803284,
      "learning_rate": 5.802879295987975e-06,
      "loss": 0.06478571891784668,
      "step": 5402
    },
    {
      "epoch": 0.7304682879015767,
      "grad_norm": 0.10919829457998276,
      "learning_rate": 5.797473338622722e-06,
      "loss": 0.08009648323059082,
      "step": 5403
    },
    {
      "epoch": 0.7306034846974127,
      "grad_norm": 0.10116737335920334,
      "learning_rate": 5.792069297272599e-06,
      "loss": 0.06067085266113281,
      "step": 5404
    },
    {
      "epoch": 0.7307386814932486,
      "grad_norm": 0.07250949740409851,
      "learning_rate": 5.7866671730627485e-06,
      "loss": 0.04123795032501221,
      "step": 5405
    },
    {
      "epoch": 0.7308738782890846,
      "grad_norm": 0.06884024292230606,
      "learning_rate": 5.781266967117925e-06,
      "loss": 0.06620335578918457,
      "step": 5406
    },
    {
      "epoch": 0.7310090750849205,
      "grad_norm": 0.06492038071155548,
      "learning_rate": 5.7758686805624815e-06,
      "loss": 0.06308764219284058,
      "step": 5407
    },
    {
      "epoch": 0.7311442718807565,
      "grad_norm": 0.08034491539001465,
      "learning_rate": 5.7704723145203605e-06,
      "loss": 0.05358099937438965,
      "step": 5408
    },
    {
      "epoch": 0.7312794686765923,
      "grad_norm": 0.08367525041103363,
      "learning_rate": 5.765077870115126e-06,
      "loss": 0.05752992630004883,
      "step": 5409
    },
    {
      "epoch": 0.7314146654724283,
      "grad_norm": 0.07757303863763809,
      "learning_rate": 5.759685348469928e-06,
      "loss": 0.07065486907958984,
      "step": 5410
    },
    {
      "epoch": 0.7315498622682642,
      "grad_norm": 0.06057950481772423,
      "learning_rate": 5.754294750707514e-06,
      "loss": 0.04633328318595886,
      "step": 5411
    },
    {
      "epoch": 0.7316850590641002,
      "grad_norm": 0.07146965712308884,
      "learning_rate": 5.748906077950237e-06,
      "loss": 0.07584881782531738,
      "step": 5412
    },
    {
      "epoch": 0.7318202558599362,
      "grad_norm": 0.11002994328737259,
      "learning_rate": 5.743519331320042e-06,
      "loss": 0.06497859954833984,
      "step": 5413
    },
    {
      "epoch": 0.7319554526557721,
      "grad_norm": 0.037900302559137344,
      "learning_rate": 5.73813451193849e-06,
      "loss": 0.03794050216674805,
      "step": 5414
    },
    {
      "epoch": 0.732090649451608,
      "grad_norm": 0.05841132253408432,
      "learning_rate": 5.7327516209267225e-06,
      "loss": 0.052388131618499756,
      "step": 5415
    },
    {
      "epoch": 0.7322258462474439,
      "grad_norm": 0.06615713983774185,
      "learning_rate": 5.727370659405486e-06,
      "loss": 0.06970357894897461,
      "step": 5416
    },
    {
      "epoch": 0.7323610430432799,
      "grad_norm": 0.1121114194393158,
      "learning_rate": 5.7219916284951265e-06,
      "loss": 0.07740664482116699,
      "step": 5417
    },
    {
      "epoch": 0.7324962398391158,
      "grad_norm": 0.04503386840224266,
      "learning_rate": 5.716614529315582e-06,
      "loss": 0.043839067220687866,
      "step": 5418
    },
    {
      "epoch": 0.7326314366349518,
      "grad_norm": 0.08000829815864563,
      "learning_rate": 5.711239362986401e-06,
      "loss": 0.06727242469787598,
      "step": 5419
    },
    {
      "epoch": 0.7327666334307877,
      "grad_norm": 0.06890582293272018,
      "learning_rate": 5.705866130626719e-06,
      "loss": 0.05676770210266113,
      "step": 5420
    },
    {
      "epoch": 0.7329018302266236,
      "grad_norm": 0.11593379080295563,
      "learning_rate": 5.700494833355271e-06,
      "loss": 0.0728616714477539,
      "step": 5421
    },
    {
      "epoch": 0.7330370270224595,
      "grad_norm": 0.10198880732059479,
      "learning_rate": 5.69512547229039e-06,
      "loss": 0.07434606552124023,
      "step": 5422
    },
    {
      "epoch": 0.7331722238182955,
      "grad_norm": 0.03199901804327965,
      "learning_rate": 5.689758048550001e-06,
      "loss": 0.02387791872024536,
      "step": 5423
    },
    {
      "epoch": 0.7333074206141315,
      "grad_norm": 0.07188189029693604,
      "learning_rate": 5.684392563251644e-06,
      "loss": 0.07494354248046875,
      "step": 5424
    },
    {
      "epoch": 0.7334426174099674,
      "grad_norm": 0.08367427438497543,
      "learning_rate": 5.679029017512422e-06,
      "loss": 0.06900453567504883,
      "step": 5425
    },
    {
      "epoch": 0.7335778142058034,
      "grad_norm": 0.045845434069633484,
      "learning_rate": 5.6736674124490684e-06,
      "loss": 0.04075121879577637,
      "step": 5426
    },
    {
      "epoch": 0.7337130110016392,
      "grad_norm": 0.06971453130245209,
      "learning_rate": 5.6683077491778935e-06,
      "loss": 0.058294057846069336,
      "step": 5427
    },
    {
      "epoch": 0.7338482077974752,
      "grad_norm": 0.08261123299598694,
      "learning_rate": 5.6629500288148044e-06,
      "loss": 0.0534512996673584,
      "step": 5428
    },
    {
      "epoch": 0.7339834045933111,
      "grad_norm": 0.07943519204854965,
      "learning_rate": 5.657594252475319e-06,
      "loss": 0.06360459327697754,
      "step": 5429
    },
    {
      "epoch": 0.7341186013891471,
      "grad_norm": 0.058413002640008926,
      "learning_rate": 5.652240421274521e-06,
      "loss": 0.06244230270385742,
      "step": 5430
    },
    {
      "epoch": 0.734253798184983,
      "grad_norm": 0.1283576786518097,
      "learning_rate": 5.646888536327121e-06,
      "loss": 0.09134674072265625,
      "step": 5431
    },
    {
      "epoch": 0.734388994980819,
      "grad_norm": 0.0840093344449997,
      "learning_rate": 5.641538598747403e-06,
      "loss": 0.06548523902893066,
      "step": 5432
    },
    {
      "epoch": 0.7345241917766548,
      "grad_norm": 0.07205220311880112,
      "learning_rate": 5.6361906096492495e-06,
      "loss": 0.03668212890625,
      "step": 5433
    },
    {
      "epoch": 0.7346593885724908,
      "grad_norm": 0.06774645298719406,
      "learning_rate": 5.630844570146157e-06,
      "loss": 0.04929041862487793,
      "step": 5434
    },
    {
      "epoch": 0.7347945853683268,
      "grad_norm": 0.15659502148628235,
      "learning_rate": 5.625500481351176e-06,
      "loss": 0.06994763016700745,
      "step": 5435
    },
    {
      "epoch": 0.7349297821641627,
      "grad_norm": 0.09570881724357605,
      "learning_rate": 5.6201583443769895e-06,
      "loss": 0.06996774673461914,
      "step": 5436
    },
    {
      "epoch": 0.7350649789599987,
      "grad_norm": 0.07199430465698242,
      "learning_rate": 5.614818160335857e-06,
      "loss": 0.05830782651901245,
      "step": 5437
    },
    {
      "epoch": 0.7352001757558346,
      "grad_norm": 0.15612199902534485,
      "learning_rate": 5.6094799303396315e-06,
      "loss": 0.09366416931152344,
      "step": 5438
    },
    {
      "epoch": 0.7353353725516705,
      "grad_norm": 0.1137339323759079,
      "learning_rate": 5.6041436554997595e-06,
      "loss": 0.04423785209655762,
      "step": 5439
    },
    {
      "epoch": 0.7354705693475064,
      "grad_norm": 0.077876515686512,
      "learning_rate": 5.598809336927278e-06,
      "loss": 0.06204056739807129,
      "step": 5440
    },
    {
      "epoch": 0.7356057661433424,
      "grad_norm": 0.06212661787867546,
      "learning_rate": 5.5934769757328325e-06,
      "loss": 0.048281192779541016,
      "step": 5441
    },
    {
      "epoch": 0.7357409629391783,
      "grad_norm": 0.04554282873868942,
      "learning_rate": 5.588146573026642e-06,
      "loss": 0.044075578451156616,
      "step": 5442
    },
    {
      "epoch": 0.7358761597350143,
      "grad_norm": 0.10456150025129318,
      "learning_rate": 5.582818129918525e-06,
      "loss": 0.08704376220703125,
      "step": 5443
    },
    {
      "epoch": 0.7360113565308503,
      "grad_norm": 0.14092624187469482,
      "learning_rate": 5.5774916475178915e-06,
      "loss": 0.1082000732421875,
      "step": 5444
    },
    {
      "epoch": 0.7361465533266861,
      "grad_norm": 0.07932636141777039,
      "learning_rate": 5.572167126933738e-06,
      "loss": 0.04230976104736328,
      "step": 5445
    },
    {
      "epoch": 0.7362817501225221,
      "grad_norm": 0.07973039895296097,
      "learning_rate": 5.566844569274669e-06,
      "loss": 0.062274277210235596,
      "step": 5446
    },
    {
      "epoch": 0.736416946918358,
      "grad_norm": 0.06242867186665535,
      "learning_rate": 5.5615239756488665e-06,
      "loss": 0.04746294021606445,
      "step": 5447
    },
    {
      "epoch": 0.736552143714194,
      "grad_norm": 0.05520540103316307,
      "learning_rate": 5.556205347164104e-06,
      "loss": 0.0515783429145813,
      "step": 5448
    },
    {
      "epoch": 0.7366873405100299,
      "grad_norm": 0.2551308870315552,
      "learning_rate": 5.550888684927746e-06,
      "loss": 0.08831024169921875,
      "step": 5449
    },
    {
      "epoch": 0.7368225373058659,
      "grad_norm": 0.07695929706096649,
      "learning_rate": 5.545573990046752e-06,
      "loss": 0.048056602478027344,
      "step": 5450
    },
    {
      "epoch": 0.7369577341017018,
      "grad_norm": 0.05442466214299202,
      "learning_rate": 5.540261263627672e-06,
      "loss": 0.05111503601074219,
      "step": 5451
    },
    {
      "epoch": 0.7370929308975377,
      "grad_norm": 0.06580621004104614,
      "learning_rate": 5.534950506776644e-06,
      "loss": 0.06391990184783936,
      "step": 5452
    },
    {
      "epoch": 0.7372281276933736,
      "grad_norm": 0.08065614104270935,
      "learning_rate": 5.529641720599393e-06,
      "loss": 0.053240060806274414,
      "step": 5453
    },
    {
      "epoch": 0.7373633244892096,
      "grad_norm": 0.06250309199094772,
      "learning_rate": 5.52433490620124e-06,
      "loss": 0.05691862106323242,
      "step": 5454
    },
    {
      "epoch": 0.7374985212850456,
      "grad_norm": 0.09920760989189148,
      "learning_rate": 5.519030064687082e-06,
      "loss": 0.07890653610229492,
      "step": 5455
    },
    {
      "epoch": 0.7376337180808815,
      "grad_norm": 0.04598747938871384,
      "learning_rate": 5.51372719716143e-06,
      "loss": 0.04904282093048096,
      "step": 5456
    },
    {
      "epoch": 0.7377689148767175,
      "grad_norm": 0.08992001414299011,
      "learning_rate": 5.508426304728363e-06,
      "loss": 0.04861760139465332,
      "step": 5457
    },
    {
      "epoch": 0.7379041116725533,
      "grad_norm": 0.08290814608335495,
      "learning_rate": 5.503127388491552e-06,
      "loss": 0.07291316986083984,
      "step": 5458
    },
    {
      "epoch": 0.7380393084683893,
      "grad_norm": 0.06295759230852127,
      "learning_rate": 5.497830449554266e-06,
      "loss": 0.06519508361816406,
      "step": 5459
    },
    {
      "epoch": 0.7381745052642252,
      "grad_norm": 0.1149107962846756,
      "learning_rate": 5.492535489019344e-06,
      "loss": 0.09826409816741943,
      "step": 5460
    },
    {
      "epoch": 0.7383097020600612,
      "grad_norm": 0.13201308250427246,
      "learning_rate": 5.4872425079892454e-06,
      "loss": 0.07710552215576172,
      "step": 5461
    },
    {
      "epoch": 0.7384448988558971,
      "grad_norm": 0.08112886548042297,
      "learning_rate": 5.481951507565973e-06,
      "loss": 0.06615877151489258,
      "step": 5462
    },
    {
      "epoch": 0.7385800956517331,
      "grad_norm": 0.04119957610964775,
      "learning_rate": 5.476662488851159e-06,
      "loss": 0.040248751640319824,
      "step": 5463
    },
    {
      "epoch": 0.738715292447569,
      "grad_norm": 0.04800136387348175,
      "learning_rate": 5.471375452946e-06,
      "loss": 0.0457518994808197,
      "step": 5464
    },
    {
      "epoch": 0.7388504892434049,
      "grad_norm": 0.0868966206908226,
      "learning_rate": 5.466090400951279e-06,
      "loss": 0.07658815383911133,
      "step": 5465
    },
    {
      "epoch": 0.7389856860392409,
      "grad_norm": 0.05991123616695404,
      "learning_rate": 5.460807333967387e-06,
      "loss": 0.03929173946380615,
      "step": 5466
    },
    {
      "epoch": 0.7391208828350768,
      "grad_norm": 0.09085798263549805,
      "learning_rate": 5.455526253094267e-06,
      "loss": 0.05458378791809082,
      "step": 5467
    },
    {
      "epoch": 0.7392560796309128,
      "grad_norm": 0.08786561340093613,
      "learning_rate": 5.450247159431486e-06,
      "loss": 0.044530630111694336,
      "step": 5468
    },
    {
      "epoch": 0.7393912764267487,
      "grad_norm": 0.039743077009916306,
      "learning_rate": 5.44497005407817e-06,
      "loss": 0.0347599983215332,
      "step": 5469
    },
    {
      "epoch": 0.7395264732225846,
      "grad_norm": 0.13131289184093475,
      "learning_rate": 5.439694938133042e-06,
      "loss": 0.07515549659729004,
      "step": 5470
    },
    {
      "epoch": 0.7396616700184205,
      "grad_norm": 0.04205523803830147,
      "learning_rate": 5.434421812694409e-06,
      "loss": 0.04130291938781738,
      "step": 5471
    },
    {
      "epoch": 0.7397968668142565,
      "grad_norm": 0.1523616909980774,
      "learning_rate": 5.4291506788601624e-06,
      "loss": 0.07195305824279785,
      "step": 5472
    },
    {
      "epoch": 0.7399320636100924,
      "grad_norm": 0.049770746380090714,
      "learning_rate": 5.423881537727785e-06,
      "loss": 0.039507508277893066,
      "step": 5473
    },
    {
      "epoch": 0.7400672604059284,
      "grad_norm": 0.15474846959114075,
      "learning_rate": 5.418614390394338e-06,
      "loss": 0.07103705406188965,
      "step": 5474
    },
    {
      "epoch": 0.7402024572017644,
      "grad_norm": 0.09364216774702072,
      "learning_rate": 5.413349237956469e-06,
      "loss": 0.054895877838134766,
      "step": 5475
    },
    {
      "epoch": 0.7403376539976002,
      "grad_norm": 0.10518147796392441,
      "learning_rate": 5.4080860815104125e-06,
      "loss": 0.06545042991638184,
      "step": 5476
    },
    {
      "epoch": 0.7404728507934362,
      "grad_norm": 0.10757801681756973,
      "learning_rate": 5.402824922151977e-06,
      "loss": 0.07328128814697266,
      "step": 5477
    },
    {
      "epoch": 0.7406080475892721,
      "grad_norm": 0.045876652002334595,
      "learning_rate": 5.397565760976577e-06,
      "loss": 0.03791801631450653,
      "step": 5478
    },
    {
      "epoch": 0.7407432443851081,
      "grad_norm": 0.09045185893774033,
      "learning_rate": 5.392308599079193e-06,
      "loss": 0.053752899169921875,
      "step": 5479
    },
    {
      "epoch": 0.740878441180944,
      "grad_norm": 0.09355717152357101,
      "learning_rate": 5.3870534375543916e-06,
      "loss": 0.05192971229553223,
      "step": 5480
    },
    {
      "epoch": 0.74101363797678,
      "grad_norm": 0.056373097002506256,
      "learning_rate": 5.381800277496328e-06,
      "loss": 0.05078864097595215,
      "step": 5481
    },
    {
      "epoch": 0.7411488347726158,
      "grad_norm": 0.21935756504535675,
      "learning_rate": 5.376549119998731e-06,
      "loss": 0.08102989196777344,
      "step": 5482
    },
    {
      "epoch": 0.7412840315684518,
      "grad_norm": 0.07524493336677551,
      "learning_rate": 5.3712999661549314e-06,
      "loss": 0.06669723987579346,
      "step": 5483
    },
    {
      "epoch": 0.7414192283642878,
      "grad_norm": 0.057628776878118515,
      "learning_rate": 5.366052817057826e-06,
      "loss": 0.053884029388427734,
      "step": 5484
    },
    {
      "epoch": 0.7415544251601237,
      "grad_norm": 0.0970100685954094,
      "learning_rate": 5.360807673799899e-06,
      "loss": 0.07878828048706055,
      "step": 5485
    },
    {
      "epoch": 0.7416896219559597,
      "grad_norm": 0.09406261146068573,
      "learning_rate": 5.355564537473214e-06,
      "loss": 0.043989211320877075,
      "step": 5486
    },
    {
      "epoch": 0.7418248187517956,
      "grad_norm": 0.10842932015657425,
      "learning_rate": 5.35032340916942e-06,
      "loss": 0.07433564215898514,
      "step": 5487
    },
    {
      "epoch": 0.7419600155476315,
      "grad_norm": 0.060150325298309326,
      "learning_rate": 5.345084289979755e-06,
      "loss": 0.03793051838874817,
      "step": 5488
    },
    {
      "epoch": 0.7420952123434674,
      "grad_norm": 0.08368875086307526,
      "learning_rate": 5.339847180995026e-06,
      "loss": 0.07368636131286621,
      "step": 5489
    },
    {
      "epoch": 0.7422304091393034,
      "grad_norm": 0.07861552387475967,
      "learning_rate": 5.33461208330563e-06,
      "loss": 0.06901264190673828,
      "step": 5490
    },
    {
      "epoch": 0.7423656059351393,
      "grad_norm": 0.06586482375860214,
      "learning_rate": 5.32937899800154e-06,
      "loss": 0.062071025371551514,
      "step": 5491
    },
    {
      "epoch": 0.7425008027309753,
      "grad_norm": 0.060190703719854355,
      "learning_rate": 5.324147926172307e-06,
      "loss": 0.049607276916503906,
      "step": 5492
    },
    {
      "epoch": 0.7426359995268113,
      "grad_norm": 0.055766988545656204,
      "learning_rate": 5.318918868907084e-06,
      "loss": 0.043805062770843506,
      "step": 5493
    },
    {
      "epoch": 0.7427711963226471,
      "grad_norm": 0.09595475345849991,
      "learning_rate": 5.313691827294568e-06,
      "loss": 0.04592317342758179,
      "step": 5494
    },
    {
      "epoch": 0.742906393118483,
      "grad_norm": 0.04057767987251282,
      "learning_rate": 5.308466802423072e-06,
      "loss": 0.034522950649261475,
      "step": 5495
    },
    {
      "epoch": 0.743041589914319,
      "grad_norm": 0.09693080186843872,
      "learning_rate": 5.303243795380471e-06,
      "loss": 0.07604217529296875,
      "step": 5496
    },
    {
      "epoch": 0.743176786710155,
      "grad_norm": 0.12563791871070862,
      "learning_rate": 5.298022807254215e-06,
      "loss": 0.052136898040771484,
      "step": 5497
    },
    {
      "epoch": 0.7433119835059909,
      "grad_norm": 0.0906025841832161,
      "learning_rate": 5.292803839131358e-06,
      "loss": 0.0597529411315918,
      "step": 5498
    },
    {
      "epoch": 0.7434471803018269,
      "grad_norm": 0.045552872121334076,
      "learning_rate": 5.287586892098496e-06,
      "loss": 0.04845905303955078,
      "step": 5499
    },
    {
      "epoch": 0.7435823770976628,
      "grad_norm": 0.09599911421537399,
      "learning_rate": 5.282371967241842e-06,
      "loss": 0.0832982063293457,
      "step": 5500
    },
    {
      "epoch": 0.7437175738934987,
      "grad_norm": 0.0929669588804245,
      "learning_rate": 5.277159065647164e-06,
      "loss": 0.06981372833251953,
      "step": 5501
    },
    {
      "epoch": 0.7438527706893346,
      "grad_norm": 0.06944774091243744,
      "learning_rate": 5.271948188399814e-06,
      "loss": 0.03140115737915039,
      "step": 5502
    },
    {
      "epoch": 0.7439879674851706,
      "grad_norm": 0.06827398389577866,
      "learning_rate": 5.266739336584735e-06,
      "loss": 0.04185876250267029,
      "step": 5503
    },
    {
      "epoch": 0.7441231642810066,
      "grad_norm": 0.09269287437200546,
      "learning_rate": 5.261532511286422e-06,
      "loss": 0.07217788696289062,
      "step": 5504
    },
    {
      "epoch": 0.7442583610768425,
      "grad_norm": 0.15041735768318176,
      "learning_rate": 5.256327713588977e-06,
      "loss": 0.07729315757751465,
      "step": 5505
    },
    {
      "epoch": 0.7443935578726785,
      "grad_norm": 0.04381795600056648,
      "learning_rate": 5.25112494457606e-06,
      "loss": 0.03384733200073242,
      "step": 5506
    },
    {
      "epoch": 0.7445287546685143,
      "grad_norm": 0.09961852431297302,
      "learning_rate": 5.245924205330919e-06,
      "loss": 0.05482053756713867,
      "step": 5507
    },
    {
      "epoch": 0.7446639514643503,
      "grad_norm": 0.06379222869873047,
      "learning_rate": 5.240725496936373e-06,
      "loss": 0.047136589884757996,
      "step": 5508
    },
    {
      "epoch": 0.7447991482601862,
      "grad_norm": 0.04229654371738434,
      "learning_rate": 5.2355288204748145e-06,
      "loss": 0.03293609619140625,
      "step": 5509
    },
    {
      "epoch": 0.7449343450560222,
      "grad_norm": 0.06408032774925232,
      "learning_rate": 5.230334177028233e-06,
      "loss": 0.047388315200805664,
      "step": 5510
    },
    {
      "epoch": 0.7450695418518581,
      "grad_norm": 0.052778828889131546,
      "learning_rate": 5.2251415676781726e-06,
      "loss": 0.041927993297576904,
      "step": 5511
    },
    {
      "epoch": 0.7452047386476941,
      "grad_norm": 0.12281057238578796,
      "learning_rate": 5.2199509935057655e-06,
      "loss": 0.0704350471496582,
      "step": 5512
    },
    {
      "epoch": 0.7453399354435299,
      "grad_norm": 0.141086608171463,
      "learning_rate": 5.214762455591713e-06,
      "loss": 0.08241963386535645,
      "step": 5513
    },
    {
      "epoch": 0.7454751322393659,
      "grad_norm": 0.0525985062122345,
      "learning_rate": 5.209575955016295e-06,
      "loss": 0.05631065368652344,
      "step": 5514
    },
    {
      "epoch": 0.7456103290352019,
      "grad_norm": 0.06778184324502945,
      "learning_rate": 5.204391492859377e-06,
      "loss": 0.06846237182617188,
      "step": 5515
    },
    {
      "epoch": 0.7457455258310378,
      "grad_norm": 0.1181851178407669,
      "learning_rate": 5.199209070200388e-06,
      "loss": 0.06189322471618652,
      "step": 5516
    },
    {
      "epoch": 0.7458807226268738,
      "grad_norm": 0.06505578011274338,
      "learning_rate": 5.194028688118332e-06,
      "loss": 0.06946802139282227,
      "step": 5517
    },
    {
      "epoch": 0.7460159194227097,
      "grad_norm": 0.16072849929332733,
      "learning_rate": 5.188850347691797e-06,
      "loss": 0.09738695621490479,
      "step": 5518
    },
    {
      "epoch": 0.7461511162185456,
      "grad_norm": 0.049335628747940063,
      "learning_rate": 5.183674049998934e-06,
      "loss": 0.04713273048400879,
      "step": 5519
    },
    {
      "epoch": 0.7462863130143815,
      "grad_norm": 0.1295800805091858,
      "learning_rate": 5.178499796117485e-06,
      "loss": 0.06938743591308594,
      "step": 5520
    },
    {
      "epoch": 0.7464215098102175,
      "grad_norm": 0.10742707550525665,
      "learning_rate": 5.173327587124753e-06,
      "loss": 0.08169126510620117,
      "step": 5521
    },
    {
      "epoch": 0.7465567066060534,
      "grad_norm": 0.07736071199178696,
      "learning_rate": 5.16815742409762e-06,
      "loss": 0.04987215995788574,
      "step": 5522
    },
    {
      "epoch": 0.7466919034018894,
      "grad_norm": 0.047543223947286606,
      "learning_rate": 5.16298930811254e-06,
      "loss": 0.054995596408843994,
      "step": 5523
    },
    {
      "epoch": 0.7468271001977254,
      "grad_norm": 0.08714444935321808,
      "learning_rate": 5.15782324024554e-06,
      "loss": 0.057285845279693604,
      "step": 5524
    },
    {
      "epoch": 0.7469622969935612,
      "grad_norm": 0.07695301622152328,
      "learning_rate": 5.152659221572231e-06,
      "loss": 0.052234649658203125,
      "step": 5525
    },
    {
      "epoch": 0.7470974937893972,
      "grad_norm": 0.08860433846712112,
      "learning_rate": 5.147497253167784e-06,
      "loss": 0.07405328750610352,
      "step": 5526
    },
    {
      "epoch": 0.7472326905852331,
      "grad_norm": 0.0780363604426384,
      "learning_rate": 5.142337336106948e-06,
      "loss": 0.0707550048828125,
      "step": 5527
    },
    {
      "epoch": 0.7473678873810691,
      "grad_norm": 0.20865824818611145,
      "learning_rate": 5.137179471464047e-06,
      "loss": 0.07080549001693726,
      "step": 5528
    },
    {
      "epoch": 0.747503084176905,
      "grad_norm": 0.10193055868148804,
      "learning_rate": 5.13202366031297e-06,
      "loss": 0.0694284439086914,
      "step": 5529
    },
    {
      "epoch": 0.747638280972741,
      "grad_norm": 0.10590600967407227,
      "learning_rate": 5.1268699037272e-06,
      "loss": 0.05528116226196289,
      "step": 5530
    },
    {
      "epoch": 0.7477734777685768,
      "grad_norm": 0.07347662001848221,
      "learning_rate": 5.121718202779756e-06,
      "loss": 0.06485557556152344,
      "step": 5531
    },
    {
      "epoch": 0.7479086745644128,
      "grad_norm": 0.0719623938202858,
      "learning_rate": 5.116568558543264e-06,
      "loss": 0.05587053298950195,
      "step": 5532
    },
    {
      "epoch": 0.7480438713602487,
      "grad_norm": 0.09866522997617722,
      "learning_rate": 5.1114209720899025e-06,
      "loss": 0.054007112979888916,
      "step": 5533
    },
    {
      "epoch": 0.7481790681560847,
      "grad_norm": 0.06220245361328125,
      "learning_rate": 5.106275444491423e-06,
      "loss": 0.04083812236785889,
      "step": 5534
    },
    {
      "epoch": 0.7483142649519207,
      "grad_norm": 0.05143401399254799,
      "learning_rate": 5.101131976819165e-06,
      "loss": 0.05747842788696289,
      "step": 5535
    },
    {
      "epoch": 0.7484494617477566,
      "grad_norm": 0.0980193242430687,
      "learning_rate": 5.095990570144008e-06,
      "loss": 0.044384002685546875,
      "step": 5536
    },
    {
      "epoch": 0.7485846585435925,
      "grad_norm": 0.07927174866199493,
      "learning_rate": 5.090851225536432e-06,
      "loss": 0.07108783721923828,
      "step": 5537
    },
    {
      "epoch": 0.7487198553394284,
      "grad_norm": 0.09014839679002762,
      "learning_rate": 5.085713944066474e-06,
      "loss": 0.06176948547363281,
      "step": 5538
    },
    {
      "epoch": 0.7488550521352644,
      "grad_norm": 0.07132940739393234,
      "learning_rate": 5.080578726803741e-06,
      "loss": 0.05525040626525879,
      "step": 5539
    },
    {
      "epoch": 0.7489902489311003,
      "grad_norm": 0.04437018558382988,
      "learning_rate": 5.075445574817415e-06,
      "loss": 0.040592730045318604,
      "step": 5540
    },
    {
      "epoch": 0.7491254457269363,
      "grad_norm": 0.09692610800266266,
      "learning_rate": 5.07031448917624e-06,
      "loss": 0.05897831916809082,
      "step": 5541
    },
    {
      "epoch": 0.7492606425227722,
      "grad_norm": 0.08678141981363297,
      "learning_rate": 5.065185470948544e-06,
      "loss": 0.0537571907043457,
      "step": 5542
    },
    {
      "epoch": 0.7493958393186082,
      "grad_norm": 0.11594772338867188,
      "learning_rate": 5.060058521202211e-06,
      "loss": 0.05883944034576416,
      "step": 5543
    },
    {
      "epoch": 0.749531036114444,
      "grad_norm": 0.06255055218935013,
      "learning_rate": 5.054933641004703e-06,
      "loss": 0.059362053871154785,
      "step": 5544
    },
    {
      "epoch": 0.74966623291028,
      "grad_norm": 0.057039447128772736,
      "learning_rate": 5.0498108314230425e-06,
      "loss": 0.03361295908689499,
      "step": 5545
    },
    {
      "epoch": 0.749801429706116,
      "grad_norm": 0.07067342847585678,
      "learning_rate": 5.044690093523823e-06,
      "loss": 0.04484063386917114,
      "step": 5546
    },
    {
      "epoch": 0.7499366265019519,
      "grad_norm": 0.061600882560014725,
      "learning_rate": 5.039571428373219e-06,
      "loss": 0.06914234161376953,
      "step": 5547
    },
    {
      "epoch": 0.7500718232977879,
      "grad_norm": 0.08802131563425064,
      "learning_rate": 5.034454837036959e-06,
      "loss": 0.05873298645019531,
      "step": 5548
    },
    {
      "epoch": 0.7502070200936238,
      "grad_norm": 0.19037026166915894,
      "learning_rate": 5.0293403205803455e-06,
      "loss": 0.09025657176971436,
      "step": 5549
    },
    {
      "epoch": 0.7503422168894597,
      "grad_norm": 0.05234413966536522,
      "learning_rate": 5.024227880068247e-06,
      "loss": 0.04296457767486572,
      "step": 5550
    },
    {
      "epoch": 0.7504774136852956,
      "grad_norm": 0.061269741505384445,
      "learning_rate": 5.019117516565096e-06,
      "loss": 0.047121524810791016,
      "step": 5551
    },
    {
      "epoch": 0.7506126104811316,
      "grad_norm": 0.2578270733356476,
      "learning_rate": 5.014009231134908e-06,
      "loss": 0.06601262092590332,
      "step": 5552
    },
    {
      "epoch": 0.7507478072769675,
      "grad_norm": 0.0900583490729332,
      "learning_rate": 5.008903024841248e-06,
      "loss": 0.09714412689208984,
      "step": 5553
    },
    {
      "epoch": 0.7508830040728035,
      "grad_norm": 0.06084698066115379,
      "learning_rate": 5.0037988987472595e-06,
      "loss": 0.05105161666870117,
      "step": 5554
    },
    {
      "epoch": 0.7510182008686395,
      "grad_norm": 0.1035371646285057,
      "learning_rate": 4.998696853915646e-06,
      "loss": 0.05336737632751465,
      "step": 5555
    },
    {
      "epoch": 0.7511533976644753,
      "grad_norm": 0.04608193039894104,
      "learning_rate": 4.993596891408676e-06,
      "loss": 0.033961713314056396,
      "step": 5556
    },
    {
      "epoch": 0.7512885944603113,
      "grad_norm": 0.13474397361278534,
      "learning_rate": 4.988499012288198e-06,
      "loss": 0.07436126470565796,
      "step": 5557
    },
    {
      "epoch": 0.7514237912561472,
      "grad_norm": 0.09331358969211578,
      "learning_rate": 4.983403217615614e-06,
      "loss": 0.07677650451660156,
      "step": 5558
    },
    {
      "epoch": 0.7515589880519832,
      "grad_norm": 0.13320577144622803,
      "learning_rate": 4.978309508451896e-06,
      "loss": 0.06755125522613525,
      "step": 5559
    },
    {
      "epoch": 0.7516941848478191,
      "grad_norm": 0.16972249746322632,
      "learning_rate": 4.973217885857578e-06,
      "loss": 0.06012868881225586,
      "step": 5560
    },
    {
      "epoch": 0.7518293816436551,
      "grad_norm": 0.06380508840084076,
      "learning_rate": 4.968128350892763e-06,
      "loss": 0.07260751724243164,
      "step": 5561
    },
    {
      "epoch": 0.7519645784394909,
      "grad_norm": 0.09126947075128555,
      "learning_rate": 4.963040904617131e-06,
      "loss": 0.04449129104614258,
      "step": 5562
    },
    {
      "epoch": 0.7520997752353269,
      "grad_norm": 0.11580614745616913,
      "learning_rate": 4.9579555480898955e-06,
      "loss": 0.10945892333984375,
      "step": 5563
    },
    {
      "epoch": 0.7522349720311629,
      "grad_norm": 0.05059579759836197,
      "learning_rate": 4.952872282369873e-06,
      "loss": 0.037107110023498535,
      "step": 5564
    },
    {
      "epoch": 0.7523701688269988,
      "grad_norm": 0.06567502021789551,
      "learning_rate": 4.947791108515417e-06,
      "loss": 0.05544617772102356,
      "step": 5565
    },
    {
      "epoch": 0.7525053656228348,
      "grad_norm": 0.04168238863348961,
      "learning_rate": 4.942712027584453e-06,
      "loss": 0.036566972732543945,
      "step": 5566
    },
    {
      "epoch": 0.7526405624186707,
      "grad_norm": 0.054161179810762405,
      "learning_rate": 4.937635040634485e-06,
      "loss": 0.052449941635131836,
      "step": 5567
    },
    {
      "epoch": 0.7527757592145066,
      "grad_norm": 0.12100677192211151,
      "learning_rate": 4.9325601487225545e-06,
      "loss": 0.06344461441040039,
      "step": 5568
    },
    {
      "epoch": 0.7529109560103425,
      "grad_norm": 0.0906938910484314,
      "learning_rate": 4.927487352905289e-06,
      "loss": 0.040997982025146484,
      "step": 5569
    },
    {
      "epoch": 0.7530461528061785,
      "grad_norm": 0.06256604194641113,
      "learning_rate": 4.92241665423887e-06,
      "loss": 0.05134701728820801,
      "step": 5570
    },
    {
      "epoch": 0.7531813496020144,
      "grad_norm": 0.09261125326156616,
      "learning_rate": 4.917348053779039e-06,
      "loss": 0.08533334732055664,
      "step": 5571
    },
    {
      "epoch": 0.7533165463978504,
      "grad_norm": 0.052472133189439774,
      "learning_rate": 4.912281552581122e-06,
      "loss": 0.059172868728637695,
      "step": 5572
    },
    {
      "epoch": 0.7534517431936864,
      "grad_norm": 0.06928516924381256,
      "learning_rate": 4.907217151699969e-06,
      "loss": 0.06076943874359131,
      "step": 5573
    },
    {
      "epoch": 0.7535869399895222,
      "grad_norm": 0.09416227042675018,
      "learning_rate": 4.9021548521900305e-06,
      "loss": 0.08258247375488281,
      "step": 5574
    },
    {
      "epoch": 0.7537221367853582,
      "grad_norm": 0.0987280085682869,
      "learning_rate": 4.8970946551053005e-06,
      "loss": 0.0695432722568512,
      "step": 5575
    },
    {
      "epoch": 0.7538573335811941,
      "grad_norm": 0.11186367273330688,
      "learning_rate": 4.892036561499339e-06,
      "loss": 0.06056022644042969,
      "step": 5576
    },
    {
      "epoch": 0.7539925303770301,
      "grad_norm": 0.06027604639530182,
      "learning_rate": 4.8869805724252675e-06,
      "loss": 0.04197561740875244,
      "step": 5577
    },
    {
      "epoch": 0.754127727172866,
      "grad_norm": 0.1207454651594162,
      "learning_rate": 4.8819266889357665e-06,
      "loss": 0.06840872764587402,
      "step": 5578
    },
    {
      "epoch": 0.754262923968702,
      "grad_norm": 0.11013337224721909,
      "learning_rate": 4.876874912083088e-06,
      "loss": 0.05472058057785034,
      "step": 5579
    },
    {
      "epoch": 0.7543981207645378,
      "grad_norm": 0.12484157085418701,
      "learning_rate": 4.871825242919037e-06,
      "loss": 0.08231687545776367,
      "step": 5580
    },
    {
      "epoch": 0.7545333175603738,
      "grad_norm": 0.07946226000785828,
      "learning_rate": 4.866777682494978e-06,
      "loss": 0.07387256622314453,
      "step": 5581
    },
    {
      "epoch": 0.7546685143562097,
      "grad_norm": 0.054265640676021576,
      "learning_rate": 4.861732231861845e-06,
      "loss": 0.0603179931640625,
      "step": 5582
    },
    {
      "epoch": 0.7548037111520457,
      "grad_norm": 0.049201589077711105,
      "learning_rate": 4.85668889207012e-06,
      "loss": 0.04238533973693848,
      "step": 5583
    },
    {
      "epoch": 0.7549389079478817,
      "grad_norm": 0.08540099114179611,
      "learning_rate": 4.851647664169862e-06,
      "loss": 0.06142902374267578,
      "step": 5584
    },
    {
      "epoch": 0.7550741047437176,
      "grad_norm": 0.13426458835601807,
      "learning_rate": 4.846608549210679e-06,
      "loss": 0.11732864379882812,
      "step": 5585
    },
    {
      "epoch": 0.7552093015395536,
      "grad_norm": 0.13599397242069244,
      "learning_rate": 4.841571548241741e-06,
      "loss": 0.07292807102203369,
      "step": 5586
    },
    {
      "epoch": 0.7553444983353894,
      "grad_norm": 0.09144137799739838,
      "learning_rate": 4.836536662311777e-06,
      "loss": 0.07636666297912598,
      "step": 5587
    },
    {
      "epoch": 0.7554796951312254,
      "grad_norm": 0.12282189726829529,
      "learning_rate": 4.8315038924690745e-06,
      "loss": 0.07147359848022461,
      "step": 5588
    },
    {
      "epoch": 0.7556148919270613,
      "grad_norm": 0.04325420781970024,
      "learning_rate": 4.82647323976149e-06,
      "loss": 0.051477909088134766,
      "step": 5589
    },
    {
      "epoch": 0.7557500887228973,
      "grad_norm": 0.06746050715446472,
      "learning_rate": 4.821444705236429e-06,
      "loss": 0.05562758445739746,
      "step": 5590
    },
    {
      "epoch": 0.7558852855187332,
      "grad_norm": 0.05664766579866409,
      "learning_rate": 4.81641828994086e-06,
      "loss": 0.07956457138061523,
      "step": 5591
    },
    {
      "epoch": 0.7560204823145692,
      "grad_norm": 0.07822947204113007,
      "learning_rate": 4.811393994921308e-06,
      "loss": 0.06314229965209961,
      "step": 5592
    },
    {
      "epoch": 0.756155679110405,
      "grad_norm": 0.046915702521800995,
      "learning_rate": 4.806371821223854e-06,
      "loss": 0.04811561107635498,
      "step": 5593
    },
    {
      "epoch": 0.756290875906241,
      "grad_norm": 0.04373643547296524,
      "learning_rate": 4.801351769894151e-06,
      "loss": 0.0394287109375,
      "step": 5594
    },
    {
      "epoch": 0.756426072702077,
      "grad_norm": 0.07949721068143845,
      "learning_rate": 4.796333841977394e-06,
      "loss": 0.05352514982223511,
      "step": 5595
    },
    {
      "epoch": 0.7565612694979129,
      "grad_norm": 0.0772608295083046,
      "learning_rate": 4.791318038518345e-06,
      "loss": 0.053855180740356445,
      "step": 5596
    },
    {
      "epoch": 0.7566964662937489,
      "grad_norm": 0.053752753883600235,
      "learning_rate": 4.7863043605613185e-06,
      "loss": 0.04370427131652832,
      "step": 5597
    },
    {
      "epoch": 0.7568316630895848,
      "grad_norm": 0.05103258043527603,
      "learning_rate": 4.7812928091501865e-06,
      "loss": 0.04986906051635742,
      "step": 5598
    },
    {
      "epoch": 0.7569668598854207,
      "grad_norm": 0.05900655314326286,
      "learning_rate": 4.7762833853283935e-06,
      "loss": 0.04721975326538086,
      "step": 5599
    },
    {
      "epoch": 0.7571020566812566,
      "grad_norm": 0.09072911739349365,
      "learning_rate": 4.77127609013891e-06,
      "loss": 0.07277536392211914,
      "step": 5600
    },
    {
      "epoch": 0.7572372534770926,
      "grad_norm": 0.13973002135753632,
      "learning_rate": 4.766270924624295e-06,
      "loss": 0.04611790180206299,
      "step": 5601
    },
    {
      "epoch": 0.7573724502729285,
      "grad_norm": 0.05370303615927696,
      "learning_rate": 4.761267889826647e-06,
      "loss": 0.055280208587646484,
      "step": 5602
    },
    {
      "epoch": 0.7575076470687645,
      "grad_norm": 0.06365011632442474,
      "learning_rate": 4.756266986787619e-06,
      "loss": 0.057503461837768555,
      "step": 5603
    },
    {
      "epoch": 0.7576428438646005,
      "grad_norm": 0.0598599836230278,
      "learning_rate": 4.751268216548439e-06,
      "loss": 0.04886817932128906,
      "step": 5604
    },
    {
      "epoch": 0.7577780406604363,
      "grad_norm": 0.13633494079113007,
      "learning_rate": 4.746271580149861e-06,
      "loss": 0.0834503173828125,
      "step": 5605
    },
    {
      "epoch": 0.7579132374562723,
      "grad_norm": 0.14826372265815735,
      "learning_rate": 4.7412770786322244e-06,
      "loss": 0.06831932067871094,
      "step": 5606
    },
    {
      "epoch": 0.7580484342521082,
      "grad_norm": 0.13457536697387695,
      "learning_rate": 4.736284713035406e-06,
      "loss": 0.08717679977416992,
      "step": 5607
    },
    {
      "epoch": 0.7581836310479442,
      "grad_norm": 0.06402499973773956,
      "learning_rate": 4.731294484398843e-06,
      "loss": 0.06055939197540283,
      "step": 5608
    },
    {
      "epoch": 0.7583188278437801,
      "grad_norm": 0.057129230350255966,
      "learning_rate": 4.726306393761526e-06,
      "loss": 0.039793968200683594,
      "step": 5609
    },
    {
      "epoch": 0.7584540246396161,
      "grad_norm": 0.05173155292868614,
      "learning_rate": 4.721320442162001e-06,
      "loss": 0.03661537170410156,
      "step": 5610
    },
    {
      "epoch": 0.7585892214354519,
      "grad_norm": 0.059408728033304214,
      "learning_rate": 4.716336630638378e-06,
      "loss": 0.031701087951660156,
      "step": 5611
    },
    {
      "epoch": 0.7587244182312879,
      "grad_norm": 0.07806425541639328,
      "learning_rate": 4.711354960228306e-06,
      "loss": 0.04264932870864868,
      "step": 5612
    },
    {
      "epoch": 0.7588596150271238,
      "grad_norm": 0.048639487475156784,
      "learning_rate": 4.706375431968998e-06,
      "loss": 0.028697893023490906,
      "step": 5613
    },
    {
      "epoch": 0.7589948118229598,
      "grad_norm": 0.05570409819483757,
      "learning_rate": 4.701398046897218e-06,
      "loss": 0.05447578430175781,
      "step": 5614
    },
    {
      "epoch": 0.7591300086187958,
      "grad_norm": 0.06081291660666466,
      "learning_rate": 4.696422806049277e-06,
      "loss": 0.039641499519348145,
      "step": 5615
    },
    {
      "epoch": 0.7592652054146317,
      "grad_norm": 0.11350715905427933,
      "learning_rate": 4.69144971046106e-06,
      "loss": 0.05987238883972168,
      "step": 5616
    },
    {
      "epoch": 0.7594004022104676,
      "grad_norm": 0.06956437230110168,
      "learning_rate": 4.686478761167984e-06,
      "loss": 0.06370258331298828,
      "step": 5617
    },
    {
      "epoch": 0.7595355990063035,
      "grad_norm": 0.05556348338723183,
      "learning_rate": 4.681509959205028e-06,
      "loss": 0.046308040618896484,
      "step": 5618
    },
    {
      "epoch": 0.7596707958021395,
      "grad_norm": 0.05913687124848366,
      "learning_rate": 4.676543305606724e-06,
      "loss": 0.04073485732078552,
      "step": 5619
    },
    {
      "epoch": 0.7598059925979754,
      "grad_norm": 0.07346321642398834,
      "learning_rate": 4.67157880140715e-06,
      "loss": 0.06905460357666016,
      "step": 5620
    },
    {
      "epoch": 0.7599411893938114,
      "grad_norm": 0.07231054455041885,
      "learning_rate": 4.666616447639952e-06,
      "loss": 0.0376744270324707,
      "step": 5621
    },
    {
      "epoch": 0.7600763861896473,
      "grad_norm": 0.10149803012609482,
      "learning_rate": 4.661656245338314e-06,
      "loss": 0.07171380519866943,
      "step": 5622
    },
    {
      "epoch": 0.7602115829854832,
      "grad_norm": 0.04784693196415901,
      "learning_rate": 4.656698195534978e-06,
      "loss": 0.05109739303588867,
      "step": 5623
    },
    {
      "epoch": 0.7603467797813191,
      "grad_norm": 0.07004769891500473,
      "learning_rate": 4.651742299262233e-06,
      "loss": 0.07720184326171875,
      "step": 5624
    },
    {
      "epoch": 0.7604819765771551,
      "grad_norm": 0.10608218610286713,
      "learning_rate": 4.646788557551921e-06,
      "loss": 0.08008909225463867,
      "step": 5625
    },
    {
      "epoch": 0.7606171733729911,
      "grad_norm": 0.1579318791627884,
      "learning_rate": 4.641836971435445e-06,
      "loss": 0.07655477523803711,
      "step": 5626
    },
    {
      "epoch": 0.760752370168827,
      "grad_norm": 0.09159674495458603,
      "learning_rate": 4.63688754194375e-06,
      "loss": 0.0635063648223877,
      "step": 5627
    },
    {
      "epoch": 0.760887566964663,
      "grad_norm": 0.07548218220472336,
      "learning_rate": 4.6319402701073295e-06,
      "loss": 0.06471824645996094,
      "step": 5628
    },
    {
      "epoch": 0.7610227637604988,
      "grad_norm": 0.11820026487112045,
      "learning_rate": 4.6269951569562355e-06,
      "loss": 0.06072878837585449,
      "step": 5629
    },
    {
      "epoch": 0.7611579605563348,
      "grad_norm": 0.07385329902172089,
      "learning_rate": 4.622052203520061e-06,
      "loss": 0.04602551460266113,
      "step": 5630
    },
    {
      "epoch": 0.7612931573521707,
      "grad_norm": 0.14370004832744598,
      "learning_rate": 4.617111410827968e-06,
      "loss": 0.07100248336791992,
      "step": 5631
    },
    {
      "epoch": 0.7614283541480067,
      "grad_norm": 0.06431196630001068,
      "learning_rate": 4.612172779908639e-06,
      "loss": 0.06270313262939453,
      "step": 5632
    },
    {
      "epoch": 0.7615635509438426,
      "grad_norm": 0.1484082192182541,
      "learning_rate": 4.607236311790335e-06,
      "loss": 0.06161355972290039,
      "step": 5633
    },
    {
      "epoch": 0.7616987477396786,
      "grad_norm": 0.1067439615726471,
      "learning_rate": 4.602302007500854e-06,
      "loss": 0.07177305221557617,
      "step": 5634
    },
    {
      "epoch": 0.7618339445355146,
      "grad_norm": 0.0956447571516037,
      "learning_rate": 4.597369868067537e-06,
      "loss": 0.0683283805847168,
      "step": 5635
    },
    {
      "epoch": 0.7619691413313504,
      "grad_norm": 0.07363173365592957,
      "learning_rate": 4.592439894517296e-06,
      "loss": 0.05650901794433594,
      "step": 5636
    },
    {
      "epoch": 0.7621043381271864,
      "grad_norm": 0.11286943405866623,
      "learning_rate": 4.587512087876559e-06,
      "loss": 0.05732917785644531,
      "step": 5637
    },
    {
      "epoch": 0.7622395349230223,
      "grad_norm": 0.0913057029247284,
      "learning_rate": 4.582586449171336e-06,
      "loss": 0.06699621677398682,
      "step": 5638
    },
    {
      "epoch": 0.7623747317188583,
      "grad_norm": 0.0982796847820282,
      "learning_rate": 4.577662979427168e-06,
      "loss": 0.07278728485107422,
      "step": 5639
    },
    {
      "epoch": 0.7625099285146942,
      "grad_norm": 0.052754923701286316,
      "learning_rate": 4.572741679669147e-06,
      "loss": 0.04321455955505371,
      "step": 5640
    },
    {
      "epoch": 0.7626451253105302,
      "grad_norm": 0.1420690417289734,
      "learning_rate": 4.567822550921912e-06,
      "loss": 0.059764981269836426,
      "step": 5641
    },
    {
      "epoch": 0.762780322106366,
      "grad_norm": 0.07855550199747086,
      "learning_rate": 4.562905594209647e-06,
      "loss": 0.09063005447387695,
      "step": 5642
    },
    {
      "epoch": 0.762915518902202,
      "grad_norm": 0.07625294476747513,
      "learning_rate": 4.557990810556102e-06,
      "loss": 0.07136201858520508,
      "step": 5643
    },
    {
      "epoch": 0.763050715698038,
      "grad_norm": 0.13812348246574402,
      "learning_rate": 4.553078200984553e-06,
      "loss": 0.0558469295501709,
      "step": 5644
    },
    {
      "epoch": 0.7631859124938739,
      "grad_norm": 0.13974198698997498,
      "learning_rate": 4.548167766517832e-06,
      "loss": 0.07493537664413452,
      "step": 5645
    },
    {
      "epoch": 0.7633211092897099,
      "grad_norm": 0.06511951237916946,
      "learning_rate": 4.543259508178318e-06,
      "loss": 0.055921971797943115,
      "step": 5646
    },
    {
      "epoch": 0.7634563060855458,
      "grad_norm": 0.13474225997924805,
      "learning_rate": 4.538353426987931e-06,
      "loss": 0.05825614929199219,
      "step": 5647
    },
    {
      "epoch": 0.7635915028813817,
      "grad_norm": 0.11956040561199188,
      "learning_rate": 4.533449523968154e-06,
      "loss": 0.0710597038269043,
      "step": 5648
    },
    {
      "epoch": 0.7637266996772176,
      "grad_norm": 0.08503083139657974,
      "learning_rate": 4.528547800140001e-06,
      "loss": 0.06499481201171875,
      "step": 5649
    },
    {
      "epoch": 0.7638618964730536,
      "grad_norm": 0.06388535350561142,
      "learning_rate": 4.523648256524037e-06,
      "loss": 0.06755924224853516,
      "step": 5650
    },
    {
      "epoch": 0.7639970932688895,
      "grad_norm": 0.18131384253501892,
      "learning_rate": 4.518750894140372e-06,
      "loss": 0.07106781005859375,
      "step": 5651
    },
    {
      "epoch": 0.7641322900647255,
      "grad_norm": 0.06497541069984436,
      "learning_rate": 4.513855714008659e-06,
      "loss": 0.03839111328125,
      "step": 5652
    },
    {
      "epoch": 0.7642674868605615,
      "grad_norm": 0.14690041542053223,
      "learning_rate": 4.508962717148111e-06,
      "loss": 0.06930422782897949,
      "step": 5653
    },
    {
      "epoch": 0.7644026836563973,
      "grad_norm": 0.07699696719646454,
      "learning_rate": 4.504071904577469e-06,
      "loss": 0.07210493087768555,
      "step": 5654
    },
    {
      "epoch": 0.7645378804522333,
      "grad_norm": 0.19169102609157562,
      "learning_rate": 4.499183277315027e-06,
      "loss": 0.06988239288330078,
      "step": 5655
    },
    {
      "epoch": 0.7646730772480692,
      "grad_norm": 0.046422988176345825,
      "learning_rate": 4.494296836378625e-06,
      "loss": 0.03894376754760742,
      "step": 5656
    },
    {
      "epoch": 0.7648082740439052,
      "grad_norm": 0.09825270622968674,
      "learning_rate": 4.4894125827856415e-06,
      "loss": 0.08876895904541016,
      "step": 5657
    },
    {
      "epoch": 0.7649434708397411,
      "grad_norm": 0.05888034403324127,
      "learning_rate": 4.4845305175530105e-06,
      "loss": 0.05085289478302002,
      "step": 5658
    },
    {
      "epoch": 0.7650786676355771,
      "grad_norm": 0.14505313336849213,
      "learning_rate": 4.479650641697201e-06,
      "loss": 0.09961342811584473,
      "step": 5659
    },
    {
      "epoch": 0.7652138644314129,
      "grad_norm": 0.06909554451704025,
      "learning_rate": 4.4747729562342305e-06,
      "loss": 0.07560539245605469,
      "step": 5660
    },
    {
      "epoch": 0.7653490612272489,
      "grad_norm": 0.11348436027765274,
      "learning_rate": 4.469897462179656e-06,
      "loss": 0.05191612243652344,
      "step": 5661
    },
    {
      "epoch": 0.7654842580230848,
      "grad_norm": 0.0610688254237175,
      "learning_rate": 4.46502416054858e-06,
      "loss": 0.031334519386291504,
      "step": 5662
    },
    {
      "epoch": 0.7656194548189208,
      "grad_norm": 0.15756669640541077,
      "learning_rate": 4.460153052355663e-06,
      "loss": 0.05693340301513672,
      "step": 5663
    },
    {
      "epoch": 0.7657546516147568,
      "grad_norm": 0.06645017117261887,
      "learning_rate": 4.455284138615074e-06,
      "loss": 0.051447510719299316,
      "step": 5664
    },
    {
      "epoch": 0.7658898484105927,
      "grad_norm": 0.14653916656970978,
      "learning_rate": 4.4504174203405656e-06,
      "loss": 0.09816455841064453,
      "step": 5665
    },
    {
      "epoch": 0.7660250452064286,
      "grad_norm": 0.18517173826694489,
      "learning_rate": 4.445552898545407e-06,
      "loss": 0.07475399971008301,
      "step": 5666
    },
    {
      "epoch": 0.7661602420022645,
      "grad_norm": 0.06996867805719376,
      "learning_rate": 4.440690574242413e-06,
      "loss": 0.053424060344696045,
      "step": 5667
    },
    {
      "epoch": 0.7662954387981005,
      "grad_norm": 0.06849701702594757,
      "learning_rate": 4.435830448443961e-06,
      "loss": 0.03186267614364624,
      "step": 5668
    },
    {
      "epoch": 0.7664306355939364,
      "grad_norm": 0.039524782449007034,
      "learning_rate": 4.430972522161934e-06,
      "loss": 0.03809213638305664,
      "step": 5669
    },
    {
      "epoch": 0.7665658323897724,
      "grad_norm": 0.11341464519500732,
      "learning_rate": 4.426116796407794e-06,
      "loss": 0.08158540725708008,
      "step": 5670
    },
    {
      "epoch": 0.7667010291856083,
      "grad_norm": 0.14792455732822418,
      "learning_rate": 4.421263272192523e-06,
      "loss": 0.07742452621459961,
      "step": 5671
    },
    {
      "epoch": 0.7668362259814442,
      "grad_norm": 0.08348523825407028,
      "learning_rate": 4.416411950526648e-06,
      "loss": 0.04569399356842041,
      "step": 5672
    },
    {
      "epoch": 0.7669714227772801,
      "grad_norm": 0.07707400619983673,
      "learning_rate": 4.411562832420252e-06,
      "loss": 0.06798934936523438,
      "step": 5673
    },
    {
      "epoch": 0.7671066195731161,
      "grad_norm": 0.12139245122671127,
      "learning_rate": 4.406715918882929e-06,
      "loss": 0.05596733093261719,
      "step": 5674
    },
    {
      "epoch": 0.7672418163689521,
      "grad_norm": 0.07738848775625229,
      "learning_rate": 4.4018712109238475e-06,
      "loss": 0.08188116550445557,
      "step": 5675
    },
    {
      "epoch": 0.767377013164788,
      "grad_norm": 0.12631875276565552,
      "learning_rate": 4.3970287095516965e-06,
      "loss": 0.06354045867919922,
      "step": 5676
    },
    {
      "epoch": 0.767512209960624,
      "grad_norm": 0.056322433054447174,
      "learning_rate": 4.39218841577471e-06,
      "loss": 0.049623727798461914,
      "step": 5677
    },
    {
      "epoch": 0.7676474067564599,
      "grad_norm": 0.062164556235075,
      "learning_rate": 4.387350330600662e-06,
      "loss": 0.0751807689666748,
      "step": 5678
    },
    {
      "epoch": 0.7677826035522958,
      "grad_norm": 0.10083353519439697,
      "learning_rate": 4.382514455036864e-06,
      "loss": 0.036834120750427246,
      "step": 5679
    },
    {
      "epoch": 0.7679178003481317,
      "grad_norm": 0.16395795345306396,
      "learning_rate": 4.377680790090182e-06,
      "loss": 0.06953191757202148,
      "step": 5680
    },
    {
      "epoch": 0.7680529971439677,
      "grad_norm": 0.07130637764930725,
      "learning_rate": 4.372849336767004e-06,
      "loss": 0.044980525970458984,
      "step": 5681
    },
    {
      "epoch": 0.7681881939398036,
      "grad_norm": 0.08591113239526749,
      "learning_rate": 4.3680200960732645e-06,
      "loss": 0.0762636661529541,
      "step": 5682
    },
    {
      "epoch": 0.7683233907356396,
      "grad_norm": 0.09558533132076263,
      "learning_rate": 4.363193069014439e-06,
      "loss": 0.11683797836303711,
      "step": 5683
    },
    {
      "epoch": 0.7684585875314756,
      "grad_norm": 0.03490419313311577,
      "learning_rate": 4.3583682565955325e-06,
      "loss": 0.02942836284637451,
      "step": 5684
    },
    {
      "epoch": 0.7685937843273114,
      "grad_norm": 0.059083644300699234,
      "learning_rate": 4.3535456598211074e-06,
      "loss": 0.05377912521362305,
      "step": 5685
    },
    {
      "epoch": 0.7687289811231474,
      "grad_norm": 0.14577992260456085,
      "learning_rate": 4.348725279695251e-06,
      "loss": 0.06428909301757812,
      "step": 5686
    },
    {
      "epoch": 0.7688641779189833,
      "grad_norm": 0.04511864483356476,
      "learning_rate": 4.343907117221591e-06,
      "loss": 0.04201316833496094,
      "step": 5687
    },
    {
      "epoch": 0.7689993747148193,
      "grad_norm": 0.08045602589845657,
      "learning_rate": 4.339091173403294e-06,
      "loss": 0.06159019470214844,
      "step": 5688
    },
    {
      "epoch": 0.7691345715106552,
      "grad_norm": 0.08565692603588104,
      "learning_rate": 4.334277449243061e-06,
      "loss": 0.07035970687866211,
      "step": 5689
    },
    {
      "epoch": 0.7692697683064912,
      "grad_norm": 0.17514848709106445,
      "learning_rate": 4.329465945743144e-06,
      "loss": 0.07039403915405273,
      "step": 5690
    },
    {
      "epoch": 0.769404965102327,
      "grad_norm": 0.048192404210567474,
      "learning_rate": 4.32465666390532e-06,
      "loss": 0.043877601623535156,
      "step": 5691
    },
    {
      "epoch": 0.769540161898163,
      "grad_norm": 0.08604114502668381,
      "learning_rate": 4.319849604730905e-06,
      "loss": 0.054634809494018555,
      "step": 5692
    },
    {
      "epoch": 0.769675358693999,
      "grad_norm": 0.09068559110164642,
      "learning_rate": 4.315044769220758e-06,
      "loss": 0.039613962173461914,
      "step": 5693
    },
    {
      "epoch": 0.7698105554898349,
      "grad_norm": 0.08562308549880981,
      "learning_rate": 4.310242158375264e-06,
      "loss": 0.053876399993896484,
      "step": 5694
    },
    {
      "epoch": 0.7699457522856709,
      "grad_norm": 0.1034131795167923,
      "learning_rate": 4.30544177319436e-06,
      "loss": 0.07596015930175781,
      "step": 5695
    },
    {
      "epoch": 0.7700809490815068,
      "grad_norm": 0.04676998406648636,
      "learning_rate": 4.300643614677511e-06,
      "loss": 0.05100703239440918,
      "step": 5696
    },
    {
      "epoch": 0.7702161458773427,
      "grad_norm": 0.052511557936668396,
      "learning_rate": 4.2958476838237165e-06,
      "loss": 0.053226470947265625,
      "step": 5697
    },
    {
      "epoch": 0.7703513426731786,
      "grad_norm": 0.1213599443435669,
      "learning_rate": 4.2910539816315166e-06,
      "loss": 0.06881570816040039,
      "step": 5698
    },
    {
      "epoch": 0.7704865394690146,
      "grad_norm": 0.08902842551469803,
      "learning_rate": 4.286262509098979e-06,
      "loss": 0.07626307010650635,
      "step": 5699
    },
    {
      "epoch": 0.7706217362648505,
      "grad_norm": 0.12481430917978287,
      "learning_rate": 4.28147326722373e-06,
      "loss": 0.05967378616333008,
      "step": 5700
    },
    {
      "epoch": 0.7707569330606865,
      "grad_norm": 0.09361330419778824,
      "learning_rate": 4.2766862570028965e-06,
      "loss": 0.0902797281742096,
      "step": 5701
    },
    {
      "epoch": 0.7708921298565224,
      "grad_norm": 0.11980674415826797,
      "learning_rate": 4.2719014794331715e-06,
      "loss": 0.08904170989990234,
      "step": 5702
    },
    {
      "epoch": 0.7710273266523583,
      "grad_norm": 0.13160812854766846,
      "learning_rate": 4.267118935510767e-06,
      "loss": 0.08312225341796875,
      "step": 5703
    },
    {
      "epoch": 0.7711625234481942,
      "grad_norm": 0.11950559914112091,
      "learning_rate": 4.2623386262314306e-06,
      "loss": 0.06126868724822998,
      "step": 5704
    },
    {
      "epoch": 0.7712977202440302,
      "grad_norm": 0.12375812232494354,
      "learning_rate": 4.257560552590461e-06,
      "loss": 0.07834815979003906,
      "step": 5705
    },
    {
      "epoch": 0.7714329170398662,
      "grad_norm": 0.1631339192390442,
      "learning_rate": 4.252784715582661e-06,
      "loss": 0.08001017570495605,
      "step": 5706
    },
    {
      "epoch": 0.7715681138357021,
      "grad_norm": 0.09016553312540054,
      "learning_rate": 4.2480111162024e-06,
      "loss": 0.0696326494216919,
      "step": 5707
    },
    {
      "epoch": 0.7717033106315381,
      "grad_norm": 0.09198438376188278,
      "learning_rate": 4.243239755443561e-06,
      "loss": 0.05337882041931152,
      "step": 5708
    },
    {
      "epoch": 0.7718385074273739,
      "grad_norm": 0.06559540331363678,
      "learning_rate": 4.238470634299567e-06,
      "loss": 0.05505228042602539,
      "step": 5709
    },
    {
      "epoch": 0.7719737042232099,
      "grad_norm": 0.13903583586215973,
      "learning_rate": 4.233703753763375e-06,
      "loss": 0.0497514009475708,
      "step": 5710
    },
    {
      "epoch": 0.7721089010190458,
      "grad_norm": 0.06737703830003738,
      "learning_rate": 4.228939114827469e-06,
      "loss": 0.07395601272583008,
      "step": 5711
    },
    {
      "epoch": 0.7722440978148818,
      "grad_norm": 0.10658776015043259,
      "learning_rate": 4.224176718483881e-06,
      "loss": 0.06396293640136719,
      "step": 5712
    },
    {
      "epoch": 0.7723792946107177,
      "grad_norm": 0.03172273933887482,
      "learning_rate": 4.219416565724165e-06,
      "loss": 0.02954268455505371,
      "step": 5713
    },
    {
      "epoch": 0.7725144914065537,
      "grad_norm": 0.13991527259349823,
      "learning_rate": 4.21465865753941e-06,
      "loss": 0.09024268388748169,
      "step": 5714
    },
    {
      "epoch": 0.7726496882023896,
      "grad_norm": 0.08384308964014053,
      "learning_rate": 4.209902994920236e-06,
      "loss": 0.05019557476043701,
      "step": 5715
    },
    {
      "epoch": 0.7727848849982255,
      "grad_norm": 0.12114990502595901,
      "learning_rate": 4.205149578856794e-06,
      "loss": 0.04989200830459595,
      "step": 5716
    },
    {
      "epoch": 0.7729200817940615,
      "grad_norm": 0.04591487720608711,
      "learning_rate": 4.200398410338779e-06,
      "loss": 0.03940868377685547,
      "step": 5717
    },
    {
      "epoch": 0.7730552785898974,
      "grad_norm": 0.07248803973197937,
      "learning_rate": 4.1956494903554056e-06,
      "loss": 0.04620814323425293,
      "step": 5718
    },
    {
      "epoch": 0.7731904753857334,
      "grad_norm": 0.11163653433322906,
      "learning_rate": 4.190902819895425e-06,
      "loss": 0.05672287940979004,
      "step": 5719
    },
    {
      "epoch": 0.7733256721815693,
      "grad_norm": 0.06271634250879288,
      "learning_rate": 4.186158399947118e-06,
      "loss": 0.057398319244384766,
      "step": 5720
    },
    {
      "epoch": 0.7734608689774052,
      "grad_norm": 0.10149148106575012,
      "learning_rate": 4.181416231498292e-06,
      "loss": 0.07166898250579834,
      "step": 5721
    },
    {
      "epoch": 0.7735960657732411,
      "grad_norm": 0.0582280233502388,
      "learning_rate": 4.176676315536306e-06,
      "loss": 0.057489871978759766,
      "step": 5722
    },
    {
      "epoch": 0.7737312625690771,
      "grad_norm": 0.0627729594707489,
      "learning_rate": 4.171938653048027e-06,
      "loss": 0.05167412757873535,
      "step": 5723
    },
    {
      "epoch": 0.773866459364913,
      "grad_norm": 0.11052908003330231,
      "learning_rate": 4.1672032450198616e-06,
      "loss": 0.07015776634216309,
      "step": 5724
    },
    {
      "epoch": 0.774001656160749,
      "grad_norm": 0.06602715700864792,
      "learning_rate": 4.16247009243775e-06,
      "loss": 0.05773043632507324,
      "step": 5725
    },
    {
      "epoch": 0.774136852956585,
      "grad_norm": 0.08838232606649399,
      "learning_rate": 4.1577391962871504e-06,
      "loss": 0.07994794845581055,
      "step": 5726
    },
    {
      "epoch": 0.7742720497524209,
      "grad_norm": 0.08086957037448883,
      "learning_rate": 4.153010557553076e-06,
      "loss": 0.0540698766708374,
      "step": 5727
    },
    {
      "epoch": 0.7744072465482568,
      "grad_norm": 0.07766497880220413,
      "learning_rate": 4.148284177220045e-06,
      "loss": 0.04623603820800781,
      "step": 5728
    },
    {
      "epoch": 0.7745424433440927,
      "grad_norm": 0.18808479607105255,
      "learning_rate": 4.143560056272117e-06,
      "loss": 0.06416220963001251,
      "step": 5729
    },
    {
      "epoch": 0.7746776401399287,
      "grad_norm": 0.1262831836938858,
      "learning_rate": 4.1388381956928796e-06,
      "loss": 0.07453316450119019,
      "step": 5730
    },
    {
      "epoch": 0.7748128369357646,
      "grad_norm": 0.13727988302707672,
      "learning_rate": 4.134118596465443e-06,
      "loss": 0.0624697208404541,
      "step": 5731
    },
    {
      "epoch": 0.7749480337316006,
      "grad_norm": 0.038737379014492035,
      "learning_rate": 4.1294012595724675e-06,
      "loss": 0.04214668273925781,
      "step": 5732
    },
    {
      "epoch": 0.7750832305274366,
      "grad_norm": 0.07069750875234604,
      "learning_rate": 4.1246861859961114e-06,
      "loss": 0.04573345184326172,
      "step": 5733
    },
    {
      "epoch": 0.7752184273232724,
      "grad_norm": 0.10714850574731827,
      "learning_rate": 4.119973376718089e-06,
      "loss": 0.06184267997741699,
      "step": 5734
    },
    {
      "epoch": 0.7753536241191084,
      "grad_norm": 0.08067939430475235,
      "learning_rate": 4.115262832719628e-06,
      "loss": 0.06285619735717773,
      "step": 5735
    },
    {
      "epoch": 0.7754888209149443,
      "grad_norm": 0.07119523733854294,
      "learning_rate": 4.110554554981486e-06,
      "loss": 0.03945732116699219,
      "step": 5736
    },
    {
      "epoch": 0.7756240177107803,
      "grad_norm": 0.07371142506599426,
      "learning_rate": 4.1058485444839655e-06,
      "loss": 0.07352876663208008,
      "step": 5737
    },
    {
      "epoch": 0.7757592145066162,
      "grad_norm": 0.0705307349562645,
      "learning_rate": 4.101144802206862e-06,
      "loss": 0.05722975730895996,
      "step": 5738
    },
    {
      "epoch": 0.7758944113024522,
      "grad_norm": 0.04387529566884041,
      "learning_rate": 4.096443329129535e-06,
      "loss": 0.026026010513305664,
      "step": 5739
    },
    {
      "epoch": 0.776029608098288,
      "grad_norm": 0.0630331039428711,
      "learning_rate": 4.091744126230853e-06,
      "loss": 0.05737662315368652,
      "step": 5740
    },
    {
      "epoch": 0.776164804894124,
      "grad_norm": 0.05152873322367668,
      "learning_rate": 4.08704719448921e-06,
      "loss": 0.04090714454650879,
      "step": 5741
    },
    {
      "epoch": 0.7763000016899599,
      "grad_norm": 0.09099595993757248,
      "learning_rate": 4.082352534882543e-06,
      "loss": 0.047368526458740234,
      "step": 5742
    },
    {
      "epoch": 0.7764351984857959,
      "grad_norm": 0.047396838665008545,
      "learning_rate": 4.07766014838829e-06,
      "loss": 0.03992140293121338,
      "step": 5743
    },
    {
      "epoch": 0.7765703952816319,
      "grad_norm": 0.05859273299574852,
      "learning_rate": 4.072970035983443e-06,
      "loss": 0.054872751235961914,
      "step": 5744
    },
    {
      "epoch": 0.7767055920774678,
      "grad_norm": 0.058849774301052094,
      "learning_rate": 4.068282198644505e-06,
      "loss": 0.061443328857421875,
      "step": 5745
    },
    {
      "epoch": 0.7768407888733037,
      "grad_norm": 0.0690573900938034,
      "learning_rate": 4.06359663734751e-06,
      "loss": 0.06186199188232422,
      "step": 5746
    },
    {
      "epoch": 0.7769759856691396,
      "grad_norm": 0.10576310753822327,
      "learning_rate": 4.058913353068013e-06,
      "loss": 0.08347606658935547,
      "step": 5747
    },
    {
      "epoch": 0.7771111824649756,
      "grad_norm": 0.07534656673669815,
      "learning_rate": 4.0542323467810985e-06,
      "loss": 0.054612159729003906,
      "step": 5748
    },
    {
      "epoch": 0.7772463792608115,
      "grad_norm": 0.07083319872617722,
      "learning_rate": 4.049553619461381e-06,
      "loss": 0.04709815979003906,
      "step": 5749
    },
    {
      "epoch": 0.7773815760566475,
      "grad_norm": 0.0577218234539032,
      "learning_rate": 4.044877172082997e-06,
      "loss": 0.04214411973953247,
      "step": 5750
    },
    {
      "epoch": 0.7775167728524834,
      "grad_norm": 0.07540381699800491,
      "learning_rate": 4.040203005619604e-06,
      "loss": 0.06586265563964844,
      "step": 5751
    },
    {
      "epoch": 0.7776519696483193,
      "grad_norm": 0.1522982269525528,
      "learning_rate": 4.035531121044392e-06,
      "loss": 0.07368206977844238,
      "step": 5752
    },
    {
      "epoch": 0.7777871664441552,
      "grad_norm": 0.049702271819114685,
      "learning_rate": 4.030861519330065e-06,
      "loss": 0.04225802421569824,
      "step": 5753
    },
    {
      "epoch": 0.7779223632399912,
      "grad_norm": 0.14703910052776337,
      "learning_rate": 4.026194201448868e-06,
      "loss": 0.086081862449646,
      "step": 5754
    },
    {
      "epoch": 0.7780575600358272,
      "grad_norm": 0.09344411641359329,
      "learning_rate": 4.021529168372558e-06,
      "loss": 0.06851387023925781,
      "step": 5755
    },
    {
      "epoch": 0.7781927568316631,
      "grad_norm": 0.10253384709358215,
      "learning_rate": 4.01686642107242e-06,
      "loss": 0.05027914047241211,
      "step": 5756
    },
    {
      "epoch": 0.7783279536274991,
      "grad_norm": 0.0766432136297226,
      "learning_rate": 4.0122059605192624e-06,
      "loss": 0.06454163789749146,
      "step": 5757
    },
    {
      "epoch": 0.7784631504233349,
      "grad_norm": 0.11253654956817627,
      "learning_rate": 4.007547787683412e-06,
      "loss": 0.05473829805850983,
      "step": 5758
    },
    {
      "epoch": 0.7785983472191709,
      "grad_norm": 0.09446777403354645,
      "learning_rate": 4.002891903534736e-06,
      "loss": 0.06635713577270508,
      "step": 5759
    },
    {
      "epoch": 0.7787335440150068,
      "grad_norm": 0.0888349637389183,
      "learning_rate": 3.998238309042611e-06,
      "loss": 0.07001912593841553,
      "step": 5760
    },
    {
      "epoch": 0.7788687408108428,
      "grad_norm": 0.10076601058244705,
      "learning_rate": 3.993587005175937e-06,
      "loss": 0.08518075942993164,
      "step": 5761
    },
    {
      "epoch": 0.7790039376066787,
      "grad_norm": 0.06220710277557373,
      "learning_rate": 3.988937992903144e-06,
      "loss": 0.04938507080078125,
      "step": 5762
    },
    {
      "epoch": 0.7791391344025147,
      "grad_norm": 0.14608749747276306,
      "learning_rate": 3.9842912731921716e-06,
      "loss": 0.07331264019012451,
      "step": 5763
    },
    {
      "epoch": 0.7792743311983505,
      "grad_norm": 0.09500111639499664,
      "learning_rate": 3.979646847010506e-06,
      "loss": 0.07750976085662842,
      "step": 5764
    },
    {
      "epoch": 0.7794095279941865,
      "grad_norm": 0.10417024791240692,
      "learning_rate": 3.975004715325134e-06,
      "loss": 0.09924459457397461,
      "step": 5765
    },
    {
      "epoch": 0.7795447247900225,
      "grad_norm": 0.07900304347276688,
      "learning_rate": 3.970364879102572e-06,
      "loss": 0.053186893463134766,
      "step": 5766
    },
    {
      "epoch": 0.7796799215858584,
      "grad_norm": 0.06283389776945114,
      "learning_rate": 3.96572733930886e-06,
      "loss": 0.07232850790023804,
      "step": 5767
    },
    {
      "epoch": 0.7798151183816944,
      "grad_norm": 0.09084179997444153,
      "learning_rate": 3.961092096909552e-06,
      "loss": 0.042107582092285156,
      "step": 5768
    },
    {
      "epoch": 0.7799503151775303,
      "grad_norm": 0.09927485883235931,
      "learning_rate": 3.9564591528697455e-06,
      "loss": 0.056203603744506836,
      "step": 5769
    },
    {
      "epoch": 0.7800855119733663,
      "grad_norm": 0.0700564756989479,
      "learning_rate": 3.9518285081540275e-06,
      "loss": 0.052378058433532715,
      "step": 5770
    },
    {
      "epoch": 0.7802207087692021,
      "grad_norm": 0.056838974356651306,
      "learning_rate": 3.947200163726534e-06,
      "loss": 0.0572657585144043,
      "step": 5771
    },
    {
      "epoch": 0.7803559055650381,
      "grad_norm": 0.08624331653118134,
      "learning_rate": 3.9425741205509055e-06,
      "loss": 0.051546335220336914,
      "step": 5772
    },
    {
      "epoch": 0.780491102360874,
      "grad_norm": 0.06352335959672928,
      "learning_rate": 3.9379503795903065e-06,
      "loss": 0.0782475471496582,
      "step": 5773
    },
    {
      "epoch": 0.78062629915671,
      "grad_norm": 0.04690715670585632,
      "learning_rate": 3.933328941807439e-06,
      "loss": 0.04123187065124512,
      "step": 5774
    },
    {
      "epoch": 0.780761495952546,
      "grad_norm": 0.0718836709856987,
      "learning_rate": 3.928709808164491e-06,
      "loss": 0.05680561065673828,
      "step": 5775
    },
    {
      "epoch": 0.7808966927483819,
      "grad_norm": 0.12922799587249756,
      "learning_rate": 3.924092979623203e-06,
      "loss": 0.07539081573486328,
      "step": 5776
    },
    {
      "epoch": 0.7810318895442178,
      "grad_norm": 0.10052436590194702,
      "learning_rate": 3.919478457144824e-06,
      "loss": 0.07405269145965576,
      "step": 5777
    },
    {
      "epoch": 0.7811670863400537,
      "grad_norm": 0.12398896366357803,
      "learning_rate": 3.914866241690115e-06,
      "loss": 0.05944347381591797,
      "step": 5778
    },
    {
      "epoch": 0.7813022831358897,
      "grad_norm": 0.08246397227048874,
      "learning_rate": 3.9102563342193695e-06,
      "loss": 0.07052463293075562,
      "step": 5779
    },
    {
      "epoch": 0.7814374799317256,
      "grad_norm": 0.09794431924819946,
      "learning_rate": 3.905648735692389e-06,
      "loss": 0.07231402397155762,
      "step": 5780
    },
    {
      "epoch": 0.7815726767275616,
      "grad_norm": 0.1034655049443245,
      "learning_rate": 3.901043447068508e-06,
      "loss": 0.08643817901611328,
      "step": 5781
    },
    {
      "epoch": 0.7817078735233975,
      "grad_norm": 0.06430243700742722,
      "learning_rate": 3.896440469306567e-06,
      "loss": 0.05180072784423828,
      "step": 5782
    },
    {
      "epoch": 0.7818430703192334,
      "grad_norm": 0.10815218091011047,
      "learning_rate": 3.891839803364934e-06,
      "loss": 0.05089712142944336,
      "step": 5783
    },
    {
      "epoch": 0.7819782671150693,
      "grad_norm": 0.08369754254817963,
      "learning_rate": 3.887241450201487e-06,
      "loss": 0.04656982421875,
      "step": 5784
    },
    {
      "epoch": 0.7821134639109053,
      "grad_norm": 0.04828212782740593,
      "learning_rate": 3.882645410773629e-06,
      "loss": 0.04306960105895996,
      "step": 5785
    },
    {
      "epoch": 0.7822486607067413,
      "grad_norm": 0.1344897449016571,
      "learning_rate": 3.878051686038284e-06,
      "loss": 0.09119057655334473,
      "step": 5786
    },
    {
      "epoch": 0.7823838575025772,
      "grad_norm": 0.08893289417028427,
      "learning_rate": 3.873460276951889e-06,
      "loss": 0.06274032592773438,
      "step": 5787
    },
    {
      "epoch": 0.7825190542984132,
      "grad_norm": 0.0842614695429802,
      "learning_rate": 3.868871184470397e-06,
      "loss": 0.060044050216674805,
      "step": 5788
    },
    {
      "epoch": 0.782654251094249,
      "grad_norm": 0.11243007332086563,
      "learning_rate": 3.864284409549282e-06,
      "loss": 0.07468581199645996,
      "step": 5789
    },
    {
      "epoch": 0.782789447890085,
      "grad_norm": 0.06746405363082886,
      "learning_rate": 3.859699953143532e-06,
      "loss": 0.05449056625366211,
      "step": 5790
    },
    {
      "epoch": 0.7829246446859209,
      "grad_norm": 0.2180185467004776,
      "learning_rate": 3.855117816207665e-06,
      "loss": 0.07177698612213135,
      "step": 5791
    },
    {
      "epoch": 0.7830598414817569,
      "grad_norm": 0.07927478104829788,
      "learning_rate": 3.850537999695699e-06,
      "loss": 0.06865692138671875,
      "step": 5792
    },
    {
      "epoch": 0.7831950382775928,
      "grad_norm": 0.11424429714679718,
      "learning_rate": 3.845960504561179e-06,
      "loss": 0.06290602684020996,
      "step": 5793
    },
    {
      "epoch": 0.7833302350734288,
      "grad_norm": 0.0978156253695488,
      "learning_rate": 3.841385331757161e-06,
      "loss": 0.058560967445373535,
      "step": 5794
    },
    {
      "epoch": 0.7834654318692647,
      "grad_norm": 0.07030512392520905,
      "learning_rate": 3.8368124822362184e-06,
      "loss": 0.06477928161621094,
      "step": 5795
    },
    {
      "epoch": 0.7836006286651006,
      "grad_norm": 0.15952721238136292,
      "learning_rate": 3.832241956950449e-06,
      "loss": 0.07527303695678711,
      "step": 5796
    },
    {
      "epoch": 0.7837358254609366,
      "grad_norm": 0.08921771496534348,
      "learning_rate": 3.82767375685146e-06,
      "loss": 0.0729837417602539,
      "step": 5797
    },
    {
      "epoch": 0.7838710222567725,
      "grad_norm": 0.06520732492208481,
      "learning_rate": 3.823107882890373e-06,
      "loss": 0.06862860918045044,
      "step": 5798
    },
    {
      "epoch": 0.7840062190526085,
      "grad_norm": 0.04068349674344063,
      "learning_rate": 3.8185443360178265e-06,
      "loss": 0.030504941940307617,
      "step": 5799
    },
    {
      "epoch": 0.7841414158484444,
      "grad_norm": 0.1395060122013092,
      "learning_rate": 3.813983117183973e-06,
      "loss": 0.07504129409790039,
      "step": 5800
    },
    {
      "epoch": 0.7842766126442803,
      "grad_norm": 0.1359846144914627,
      "learning_rate": 3.8094242273384932e-06,
      "loss": 0.09117484092712402,
      "step": 5801
    },
    {
      "epoch": 0.7844118094401162,
      "grad_norm": 0.04848155751824379,
      "learning_rate": 3.804867667430555e-06,
      "loss": 0.05596351623535156,
      "step": 5802
    },
    {
      "epoch": 0.7845470062359522,
      "grad_norm": 0.09631361067295074,
      "learning_rate": 3.800313438408874e-06,
      "loss": 0.06917047500610352,
      "step": 5803
    },
    {
      "epoch": 0.7846822030317882,
      "grad_norm": 0.06552135944366455,
      "learning_rate": 3.7957615412216582e-06,
      "loss": 0.057166337966918945,
      "step": 5804
    },
    {
      "epoch": 0.7848173998276241,
      "grad_norm": 0.05808146297931671,
      "learning_rate": 3.791211976816634e-06,
      "loss": 0.05141258239746094,
      "step": 5805
    },
    {
      "epoch": 0.7849525966234601,
      "grad_norm": 0.06210657209157944,
      "learning_rate": 3.786664746141057e-06,
      "loss": 0.03990602493286133,
      "step": 5806
    },
    {
      "epoch": 0.7850877934192959,
      "grad_norm": 0.07927499711513519,
      "learning_rate": 3.782119850141669e-06,
      "loss": 0.0732259750366211,
      "step": 5807
    },
    {
      "epoch": 0.7852229902151319,
      "grad_norm": 0.07797801494598389,
      "learning_rate": 3.777577289764752e-06,
      "loss": 0.052412986755371094,
      "step": 5808
    },
    {
      "epoch": 0.7853581870109678,
      "grad_norm": 0.058375872671604156,
      "learning_rate": 3.7730370659560904e-06,
      "loss": 0.03686520457267761,
      "step": 5809
    },
    {
      "epoch": 0.7854933838068038,
      "grad_norm": 0.10792986303567886,
      "learning_rate": 3.7684991796609746e-06,
      "loss": 0.061506032943725586,
      "step": 5810
    },
    {
      "epoch": 0.7856285806026397,
      "grad_norm": 0.11434817314147949,
      "learning_rate": 3.7639636318242344e-06,
      "loss": 0.07185375690460205,
      "step": 5811
    },
    {
      "epoch": 0.7857637773984757,
      "grad_norm": 0.06701134890317917,
      "learning_rate": 3.7594304233901738e-06,
      "loss": 0.05864572525024414,
      "step": 5812
    },
    {
      "epoch": 0.7858989741943117,
      "grad_norm": 0.12902435660362244,
      "learning_rate": 3.754899555302645e-06,
      "loss": 0.05607867240905762,
      "step": 5813
    },
    {
      "epoch": 0.7860341709901475,
      "grad_norm": 0.05804478004574776,
      "learning_rate": 3.7503710285049964e-06,
      "loss": 0.04204225540161133,
      "step": 5814
    },
    {
      "epoch": 0.7861693677859835,
      "grad_norm": 0.0753183662891388,
      "learning_rate": 3.7458448439400888e-06,
      "loss": 0.06431722640991211,
      "step": 5815
    },
    {
      "epoch": 0.7863045645818194,
      "grad_norm": 0.11462990939617157,
      "learning_rate": 3.7413210025502985e-06,
      "loss": 0.08178341388702393,
      "step": 5816
    },
    {
      "epoch": 0.7864397613776554,
      "grad_norm": 0.07708786427974701,
      "learning_rate": 3.7367995052775123e-06,
      "loss": 0.05583548545837402,
      "step": 5817
    },
    {
      "epoch": 0.7865749581734913,
      "grad_norm": 0.07676519453525543,
      "learning_rate": 3.732280353063133e-06,
      "loss": 0.05231192708015442,
      "step": 5818
    },
    {
      "epoch": 0.7867101549693273,
      "grad_norm": 0.09140397608280182,
      "learning_rate": 3.727763546848074e-06,
      "loss": 0.06257450580596924,
      "step": 5819
    },
    {
      "epoch": 0.7868453517651631,
      "grad_norm": 0.09161291271448135,
      "learning_rate": 3.7232490875727544e-06,
      "loss": 0.06683731079101562,
      "step": 5820
    },
    {
      "epoch": 0.7869805485609991,
      "grad_norm": 0.1485453099012375,
      "learning_rate": 3.718736976177108e-06,
      "loss": 0.0742807388305664,
      "step": 5821
    },
    {
      "epoch": 0.787115745356835,
      "grad_norm": 0.06495513021945953,
      "learning_rate": 3.71422721360058e-06,
      "loss": 0.06864309310913086,
      "step": 5822
    },
    {
      "epoch": 0.787250942152671,
      "grad_norm": 0.047900985926389694,
      "learning_rate": 3.709719800782133e-06,
      "loss": 0.03637409210205078,
      "step": 5823
    },
    {
      "epoch": 0.787386138948507,
      "grad_norm": 0.07076506316661835,
      "learning_rate": 3.7052147386602304e-06,
      "loss": 0.03696131706237793,
      "step": 5824
    },
    {
      "epoch": 0.7875213357443429,
      "grad_norm": 0.1019882783293724,
      "learning_rate": 3.700712028172851e-06,
      "loss": 0.06559419631958008,
      "step": 5825
    },
    {
      "epoch": 0.7876565325401788,
      "grad_norm": 0.0746975913643837,
      "learning_rate": 3.696211670257481e-06,
      "loss": 0.061260342597961426,
      "step": 5826
    },
    {
      "epoch": 0.7877917293360147,
      "grad_norm": 0.05356961488723755,
      "learning_rate": 3.691713665851117e-06,
      "loss": 0.05988669395446777,
      "step": 5827
    },
    {
      "epoch": 0.7879269261318507,
      "grad_norm": 0.08384816348552704,
      "learning_rate": 3.6872180158902764e-06,
      "loss": 0.05983622372150421,
      "step": 5828
    },
    {
      "epoch": 0.7880621229276866,
      "grad_norm": 0.06372924149036407,
      "learning_rate": 3.6827247213109705e-06,
      "loss": 0.05476784706115723,
      "step": 5829
    },
    {
      "epoch": 0.7881973197235226,
      "grad_norm": 0.08513446152210236,
      "learning_rate": 3.6782337830487294e-06,
      "loss": 0.05332779884338379,
      "step": 5830
    },
    {
      "epoch": 0.7883325165193585,
      "grad_norm": 0.10654087364673615,
      "learning_rate": 3.6737452020385886e-06,
      "loss": 0.09494352340698242,
      "step": 5831
    },
    {
      "epoch": 0.7884677133151944,
      "grad_norm": 0.07302692532539368,
      "learning_rate": 3.6692589792150923e-06,
      "loss": 0.04778182506561279,
      "step": 5832
    },
    {
      "epoch": 0.7886029101110303,
      "grad_norm": 0.07393590360879898,
      "learning_rate": 3.6647751155123026e-06,
      "loss": 0.06770825386047363,
      "step": 5833
    },
    {
      "epoch": 0.7887381069068663,
      "grad_norm": 0.0764707699418068,
      "learning_rate": 3.660293611863782e-06,
      "loss": 0.040668368339538574,
      "step": 5834
    },
    {
      "epoch": 0.7888733037027023,
      "grad_norm": 0.06927548348903656,
      "learning_rate": 3.655814469202602e-06,
      "loss": 0.04694366455078125,
      "step": 5835
    },
    {
      "epoch": 0.7890085004985382,
      "grad_norm": 0.10794918239116669,
      "learning_rate": 3.6513376884613446e-06,
      "loss": 0.06515288352966309,
      "step": 5836
    },
    {
      "epoch": 0.7891436972943742,
      "grad_norm": 0.1458577811717987,
      "learning_rate": 3.6468632705720934e-06,
      "loss": 0.08035993576049805,
      "step": 5837
    },
    {
      "epoch": 0.78927889409021,
      "grad_norm": 0.07329337298870087,
      "learning_rate": 3.6423912164664606e-06,
      "loss": 0.048241376876831055,
      "step": 5838
    },
    {
      "epoch": 0.789414090886046,
      "grad_norm": 0.06068097800016403,
      "learning_rate": 3.637921527075534e-06,
      "loss": 0.052973270416259766,
      "step": 5839
    },
    {
      "epoch": 0.7895492876818819,
      "grad_norm": 0.04826470836997032,
      "learning_rate": 3.63345420332994e-06,
      "loss": 0.031844139099121094,
      "step": 5840
    },
    {
      "epoch": 0.7896844844777179,
      "grad_norm": 0.0977715328335762,
      "learning_rate": 3.628989246159795e-06,
      "loss": 0.08012515306472778,
      "step": 5841
    },
    {
      "epoch": 0.7898196812735538,
      "grad_norm": 0.11876091361045837,
      "learning_rate": 3.6245266564947205e-06,
      "loss": 0.057781100273132324,
      "step": 5842
    },
    {
      "epoch": 0.7899548780693898,
      "grad_norm": 0.044788289815187454,
      "learning_rate": 3.620066435263868e-06,
      "loss": 0.024888038635253906,
      "step": 5843
    },
    {
      "epoch": 0.7900900748652256,
      "grad_norm": 0.08737047016620636,
      "learning_rate": 3.6156085833958596e-06,
      "loss": 0.05821573734283447,
      "step": 5844
    },
    {
      "epoch": 0.7902252716610616,
      "grad_norm": 0.11403119564056396,
      "learning_rate": 3.6111531018188584e-06,
      "loss": 0.0505141019821167,
      "step": 5845
    },
    {
      "epoch": 0.7903604684568976,
      "grad_norm": 0.07073387503623962,
      "learning_rate": 3.606699991460513e-06,
      "loss": 0.061865806579589844,
      "step": 5846
    },
    {
      "epoch": 0.7904956652527335,
      "grad_norm": 0.11276034265756607,
      "learning_rate": 3.602249253247986e-06,
      "loss": 0.07218074798583984,
      "step": 5847
    },
    {
      "epoch": 0.7906308620485695,
      "grad_norm": 0.09160439670085907,
      "learning_rate": 3.5978008881079445e-06,
      "loss": 0.07404851913452148,
      "step": 5848
    },
    {
      "epoch": 0.7907660588444054,
      "grad_norm": 0.1586209386587143,
      "learning_rate": 3.5933548969665587e-06,
      "loss": 0.08043265342712402,
      "step": 5849
    },
    {
      "epoch": 0.7909012556402413,
      "grad_norm": 0.06926486641168594,
      "learning_rate": 3.5889112807495152e-06,
      "loss": 0.06469619274139404,
      "step": 5850
    },
    {
      "epoch": 0.7910364524360772,
      "grad_norm": 0.08617951720952988,
      "learning_rate": 3.5844700403819935e-06,
      "loss": 0.08688092231750488,
      "step": 5851
    },
    {
      "epoch": 0.7911716492319132,
      "grad_norm": 0.059393443167209625,
      "learning_rate": 3.5800311767886847e-06,
      "loss": 0.04292649030685425,
      "step": 5852
    },
    {
      "epoch": 0.7913068460277491,
      "grad_norm": 0.09488843381404877,
      "learning_rate": 3.575594690893784e-06,
      "loss": 0.06721115112304688,
      "step": 5853
    },
    {
      "epoch": 0.7914420428235851,
      "grad_norm": 0.11273609101772308,
      "learning_rate": 3.5711605836209853e-06,
      "loss": 0.07346487045288086,
      "step": 5854
    },
    {
      "epoch": 0.7915772396194211,
      "grad_norm": 0.09509149938821793,
      "learning_rate": 3.566728855893505e-06,
      "loss": 0.04930388927459717,
      "step": 5855
    },
    {
      "epoch": 0.7917124364152569,
      "grad_norm": 0.059821393340826035,
      "learning_rate": 3.5622995086340466e-06,
      "loss": 0.054166436195373535,
      "step": 5856
    },
    {
      "epoch": 0.7918476332110929,
      "grad_norm": 0.1228511705994606,
      "learning_rate": 3.5578725427648233e-06,
      "loss": 0.051804184913635254,
      "step": 5857
    },
    {
      "epoch": 0.7919828300069288,
      "grad_norm": 0.05270153284072876,
      "learning_rate": 3.553447959207553e-06,
      "loss": 0.03704023361206055,
      "step": 5858
    },
    {
      "epoch": 0.7921180268027648,
      "grad_norm": 0.0815044641494751,
      "learning_rate": 3.5490257588834552e-06,
      "loss": 0.04775381088256836,
      "step": 5859
    },
    {
      "epoch": 0.7922532235986007,
      "grad_norm": 0.10000798851251602,
      "learning_rate": 3.5446059427132615e-06,
      "loss": 0.09478378295898438,
      "step": 5860
    },
    {
      "epoch": 0.7923884203944367,
      "grad_norm": 0.08404769748449326,
      "learning_rate": 3.5401885116171977e-06,
      "loss": 0.05052304267883301,
      "step": 5861
    },
    {
      "epoch": 0.7925236171902726,
      "grad_norm": 0.04907684773206711,
      "learning_rate": 3.5357734665149983e-06,
      "loss": 0.040949106216430664,
      "step": 5862
    },
    {
      "epoch": 0.7926588139861085,
      "grad_norm": 0.07489892095327377,
      "learning_rate": 3.5313608083258975e-06,
      "loss": 0.06591272354125977,
      "step": 5863
    },
    {
      "epoch": 0.7927940107819444,
      "grad_norm": 0.07882266491651535,
      "learning_rate": 3.526950537968629e-06,
      "loss": 0.0416867733001709,
      "step": 5864
    },
    {
      "epoch": 0.7929292075777804,
      "grad_norm": 0.14749272167682648,
      "learning_rate": 3.5225426563614466e-06,
      "loss": 0.09202718734741211,
      "step": 5865
    },
    {
      "epoch": 0.7930644043736164,
      "grad_norm": 0.13158512115478516,
      "learning_rate": 3.518137164422088e-06,
      "loss": 0.07324403524398804,
      "step": 5866
    },
    {
      "epoch": 0.7931996011694523,
      "grad_norm": 0.09891493618488312,
      "learning_rate": 3.513734063067799e-06,
      "loss": 0.046997666358947754,
      "step": 5867
    },
    {
      "epoch": 0.7933347979652883,
      "grad_norm": 0.11289020627737045,
      "learning_rate": 3.5093333532153316e-06,
      "loss": 0.06782937049865723,
      "step": 5868
    },
    {
      "epoch": 0.7934699947611241,
      "grad_norm": 0.06969927251338959,
      "learning_rate": 3.504935035780931e-06,
      "loss": 0.06851321458816528,
      "step": 5869
    },
    {
      "epoch": 0.7936051915569601,
      "grad_norm": 0.05267447233200073,
      "learning_rate": 3.500539111680364e-06,
      "loss": 0.06720972061157227,
      "step": 5870
    },
    {
      "epoch": 0.793740388352796,
      "grad_norm": 0.13022063672542572,
      "learning_rate": 3.4961455818288683e-06,
      "loss": 0.048854708671569824,
      "step": 5871
    },
    {
      "epoch": 0.793875585148632,
      "grad_norm": 0.08528923243284225,
      "learning_rate": 3.491754447141212e-06,
      "loss": 0.0514446496963501,
      "step": 5872
    },
    {
      "epoch": 0.794010781944468,
      "grad_norm": 0.1936180740594864,
      "learning_rate": 3.4873657085316504e-06,
      "loss": 0.06984138488769531,
      "step": 5873
    },
    {
      "epoch": 0.7941459787403039,
      "grad_norm": 0.2576288878917694,
      "learning_rate": 3.482979366913935e-06,
      "loss": 0.12042355537414551,
      "step": 5874
    },
    {
      "epoch": 0.7942811755361397,
      "grad_norm": 0.0762060359120369,
      "learning_rate": 3.4785954232013423e-06,
      "loss": 0.0587085485458374,
      "step": 5875
    },
    {
      "epoch": 0.7944163723319757,
      "grad_norm": 0.07142416387796402,
      "learning_rate": 3.4742138783066122e-06,
      "loss": 0.054024696350097656,
      "step": 5876
    },
    {
      "epoch": 0.7945515691278117,
      "grad_norm": 0.0944850817322731,
      "learning_rate": 3.4698347331420206e-06,
      "loss": 0.05852842330932617,
      "step": 5877
    },
    {
      "epoch": 0.7946867659236476,
      "grad_norm": 0.16262933611869812,
      "learning_rate": 3.4654579886193223e-06,
      "loss": 0.07511472702026367,
      "step": 5878
    },
    {
      "epoch": 0.7948219627194836,
      "grad_norm": 0.06424234807491302,
      "learning_rate": 3.461083645649782e-06,
      "loss": 0.060793161392211914,
      "step": 5879
    },
    {
      "epoch": 0.7949571595153195,
      "grad_norm": 0.05788527801632881,
      "learning_rate": 3.4567117051441594e-06,
      "loss": 0.043811798095703125,
      "step": 5880
    },
    {
      "epoch": 0.7950923563111554,
      "grad_norm": 0.07045552134513855,
      "learning_rate": 3.4523421680127115e-06,
      "loss": 0.05317676067352295,
      "step": 5881
    },
    {
      "epoch": 0.7952275531069913,
      "grad_norm": 0.06306800246238708,
      "learning_rate": 3.447975035165209e-06,
      "loss": 0.052356839179992676,
      "step": 5882
    },
    {
      "epoch": 0.7953627499028273,
      "grad_norm": 0.17548410594463348,
      "learning_rate": 3.4436103075109076e-06,
      "loss": 0.07306051254272461,
      "step": 5883
    },
    {
      "epoch": 0.7954979466986632,
      "grad_norm": 0.09498739242553711,
      "learning_rate": 3.4392479859585642e-06,
      "loss": 0.05462980270385742,
      "step": 5884
    },
    {
      "epoch": 0.7956331434944992,
      "grad_norm": 0.137440025806427,
      "learning_rate": 3.4348880714164416e-06,
      "loss": 0.06501126289367676,
      "step": 5885
    },
    {
      "epoch": 0.7957683402903352,
      "grad_norm": 0.060025379061698914,
      "learning_rate": 3.430530564792289e-06,
      "loss": 0.060994505882263184,
      "step": 5886
    },
    {
      "epoch": 0.795903537086171,
      "grad_norm": 0.053567856550216675,
      "learning_rate": 3.426175466993374e-06,
      "loss": 0.0558621883392334,
      "step": 5887
    },
    {
      "epoch": 0.796038733882007,
      "grad_norm": 0.060433726757764816,
      "learning_rate": 3.4218227789264468e-06,
      "loss": 0.06391000747680664,
      "step": 5888
    },
    {
      "epoch": 0.7961739306778429,
      "grad_norm": 0.08559460192918777,
      "learning_rate": 3.417472501497758e-06,
      "loss": 0.07879400253295898,
      "step": 5889
    },
    {
      "epoch": 0.7963091274736789,
      "grad_norm": 0.11101758480072021,
      "learning_rate": 3.413124635613061e-06,
      "loss": 0.08178916573524475,
      "step": 5890
    },
    {
      "epoch": 0.7964443242695148,
      "grad_norm": 0.20513050258159637,
      "learning_rate": 3.4087791821775986e-06,
      "loss": 0.10080024600028992,
      "step": 5891
    },
    {
      "epoch": 0.7965795210653508,
      "grad_norm": 0.04593917727470398,
      "learning_rate": 3.4044361420961285e-06,
      "loss": 0.040120840072631836,
      "step": 5892
    },
    {
      "epoch": 0.7967147178611866,
      "grad_norm": 0.05298652499914169,
      "learning_rate": 3.4000955162728866e-06,
      "loss": 0.032567501068115234,
      "step": 5893
    },
    {
      "epoch": 0.7968499146570226,
      "grad_norm": 0.05706121027469635,
      "learning_rate": 3.3957573056116164e-06,
      "loss": 0.05450248718261719,
      "step": 5894
    },
    {
      "epoch": 0.7969851114528586,
      "grad_norm": 0.07949705421924591,
      "learning_rate": 3.391421511015558e-06,
      "loss": 0.06602859497070312,
      "step": 5895
    },
    {
      "epoch": 0.7971203082486945,
      "grad_norm": 0.07784634083509445,
      "learning_rate": 3.38708813338744e-06,
      "loss": 0.056526631116867065,
      "step": 5896
    },
    {
      "epoch": 0.7972555050445305,
      "grad_norm": 0.1227860227227211,
      "learning_rate": 3.382757173629506e-06,
      "loss": 0.07503247261047363,
      "step": 5897
    },
    {
      "epoch": 0.7973907018403664,
      "grad_norm": 0.0445844866335392,
      "learning_rate": 3.378428632643478e-06,
      "loss": 0.03827035427093506,
      "step": 5898
    },
    {
      "epoch": 0.7975258986362023,
      "grad_norm": 0.10407643020153046,
      "learning_rate": 3.3741025113305825e-06,
      "loss": 0.07637548446655273,
      "step": 5899
    },
    {
      "epoch": 0.7976610954320382,
      "grad_norm": 0.09840883314609528,
      "learning_rate": 3.369778810591541e-06,
      "loss": 0.053110599517822266,
      "step": 5900
    },
    {
      "epoch": 0.7977962922278742,
      "grad_norm": 0.06578446924686432,
      "learning_rate": 3.3654575313265664e-06,
      "loss": 0.04084807634353638,
      "step": 5901
    },
    {
      "epoch": 0.7979314890237101,
      "grad_norm": 0.05395715311169624,
      "learning_rate": 3.361138674435386e-06,
      "loss": 0.05305075645446777,
      "step": 5902
    },
    {
      "epoch": 0.7980666858195461,
      "grad_norm": 0.11053182184696198,
      "learning_rate": 3.35682224081719e-06,
      "loss": 0.06735897064208984,
      "step": 5903
    },
    {
      "epoch": 0.798201882615382,
      "grad_norm": 0.08346085250377655,
      "learning_rate": 3.352508231370699e-06,
      "loss": 0.059014320373535156,
      "step": 5904
    },
    {
      "epoch": 0.798337079411218,
      "grad_norm": 0.10122118145227432,
      "learning_rate": 3.3481966469941044e-06,
      "loss": 0.06972360610961914,
      "step": 5905
    },
    {
      "epoch": 0.7984722762070539,
      "grad_norm": 0.10055501013994217,
      "learning_rate": 3.3438874885850984e-06,
      "loss": 0.060906171798706055,
      "step": 5906
    },
    {
      "epoch": 0.7986074730028898,
      "grad_norm": 0.0458751805126667,
      "learning_rate": 3.3395807570408847e-06,
      "loss": 0.04475867748260498,
      "step": 5907
    },
    {
      "epoch": 0.7987426697987258,
      "grad_norm": 0.07080793380737305,
      "learning_rate": 3.33527645325813e-06,
      "loss": 0.05774796009063721,
      "step": 5908
    },
    {
      "epoch": 0.7988778665945617,
      "grad_norm": 0.07462490350008011,
      "learning_rate": 3.3309745781330247e-06,
      "loss": 0.05011320114135742,
      "step": 5909
    },
    {
      "epoch": 0.7990130633903977,
      "grad_norm": 0.06588814407587051,
      "learning_rate": 3.32667513256124e-06,
      "loss": 0.03573489189147949,
      "step": 5910
    },
    {
      "epoch": 0.7991482601862336,
      "grad_norm": 0.08016342669725418,
      "learning_rate": 3.3223781174379375e-06,
      "loss": 0.07845830917358398,
      "step": 5911
    },
    {
      "epoch": 0.7992834569820695,
      "grad_norm": 0.06420721858739853,
      "learning_rate": 3.3180835336577917e-06,
      "loss": 0.06180620193481445,
      "step": 5912
    },
    {
      "epoch": 0.7994186537779054,
      "grad_norm": 0.08457471430301666,
      "learning_rate": 3.313791382114943e-06,
      "loss": 0.07882165908813477,
      "step": 5913
    },
    {
      "epoch": 0.7995538505737414,
      "grad_norm": 0.19721674919128418,
      "learning_rate": 3.3095016637030505e-06,
      "loss": 0.06384015083312988,
      "step": 5914
    },
    {
      "epoch": 0.7996890473695774,
      "grad_norm": 0.11330549418926239,
      "learning_rate": 3.3052143793152524e-06,
      "loss": 0.0427393913269043,
      "step": 5915
    },
    {
      "epoch": 0.7998242441654133,
      "grad_norm": 0.042108796536922455,
      "learning_rate": 3.3009295298441855e-06,
      "loss": 0.03156137466430664,
      "step": 5916
    },
    {
      "epoch": 0.7999594409612493,
      "grad_norm": 0.13222014904022217,
      "learning_rate": 3.2966471161819767e-06,
      "loss": 0.07802689075469971,
      "step": 5917
    },
    {
      "epoch": 0.8000946377570851,
      "grad_norm": 0.05636896938085556,
      "learning_rate": 3.292367139220246e-06,
      "loss": 0.05374807119369507,
      "step": 5918
    },
    {
      "epoch": 0.8002298345529211,
      "grad_norm": 0.07512359321117401,
      "learning_rate": 3.288089599850112e-06,
      "loss": 0.04730796813964844,
      "step": 5919
    },
    {
      "epoch": 0.800365031348757,
      "grad_norm": 0.1327861249446869,
      "learning_rate": 3.2838144989621795e-06,
      "loss": 0.08629894256591797,
      "step": 5920
    },
    {
      "epoch": 0.800500228144593,
      "grad_norm": 0.05734610557556152,
      "learning_rate": 3.2795418374465458e-06,
      "loss": 0.049720048904418945,
      "step": 5921
    },
    {
      "epoch": 0.8006354249404289,
      "grad_norm": 0.04889432713389397,
      "learning_rate": 3.275271616192803e-06,
      "loss": 0.04104304313659668,
      "step": 5922
    },
    {
      "epoch": 0.8007706217362649,
      "grad_norm": 0.03819684684276581,
      "learning_rate": 3.2710038360900303e-06,
      "loss": 0.040823400020599365,
      "step": 5923
    },
    {
      "epoch": 0.8009058185321007,
      "grad_norm": 0.07306928187608719,
      "learning_rate": 3.266738498026808e-06,
      "loss": 0.05711174011230469,
      "step": 5924
    },
    {
      "epoch": 0.8010410153279367,
      "grad_norm": 0.09495928138494492,
      "learning_rate": 3.2624756028912005e-06,
      "loss": 0.08249831199645996,
      "step": 5925
    },
    {
      "epoch": 0.8011762121237727,
      "grad_norm": 0.10753341764211655,
      "learning_rate": 3.2582151515707655e-06,
      "loss": 0.06158852577209473,
      "step": 5926
    },
    {
      "epoch": 0.8013114089196086,
      "grad_norm": 0.09272406250238419,
      "learning_rate": 3.253957144952551e-06,
      "loss": 0.08705449104309082,
      "step": 5927
    },
    {
      "epoch": 0.8014466057154446,
      "grad_norm": 0.13662089407444,
      "learning_rate": 3.249701583923091e-06,
      "loss": 0.09015393257141113,
      "step": 5928
    },
    {
      "epoch": 0.8015818025112805,
      "grad_norm": 0.14579512178897858,
      "learning_rate": 3.2454484693684257e-06,
      "loss": 0.060903072357177734,
      "step": 5929
    },
    {
      "epoch": 0.8017169993071164,
      "grad_norm": 0.11067908257246017,
      "learning_rate": 3.2411978021740727e-06,
      "loss": 0.07764029502868652,
      "step": 5930
    },
    {
      "epoch": 0.8018521961029523,
      "grad_norm": 0.05757344886660576,
      "learning_rate": 3.2369495832250434e-06,
      "loss": 0.04710733890533447,
      "step": 5931
    },
    {
      "epoch": 0.8019873928987883,
      "grad_norm": 0.06984589993953705,
      "learning_rate": 3.2327038134058378e-06,
      "loss": 0.0804290771484375,
      "step": 5932
    },
    {
      "epoch": 0.8021225896946242,
      "grad_norm": 0.061518795788288116,
      "learning_rate": 3.228460493600446e-06,
      "loss": 0.05602109432220459,
      "step": 5933
    },
    {
      "epoch": 0.8022577864904602,
      "grad_norm": 0.04481853172183037,
      "learning_rate": 3.2242196246923554e-06,
      "loss": 0.04060816764831543,
      "step": 5934
    },
    {
      "epoch": 0.8023929832862962,
      "grad_norm": 0.04452382028102875,
      "learning_rate": 3.2199812075645375e-06,
      "loss": 0.04606294631958008,
      "step": 5935
    },
    {
      "epoch": 0.802528180082132,
      "grad_norm": 0.10836662352085114,
      "learning_rate": 3.2157452430994487e-06,
      "loss": 0.07996106147766113,
      "step": 5936
    },
    {
      "epoch": 0.802663376877968,
      "grad_norm": 0.08266950398683548,
      "learning_rate": 3.2115117321790427e-06,
      "loss": 0.05531609058380127,
      "step": 5937
    },
    {
      "epoch": 0.8027985736738039,
      "grad_norm": 0.09042800217866898,
      "learning_rate": 3.207280675684754e-06,
      "loss": 0.07632696628570557,
      "step": 5938
    },
    {
      "epoch": 0.8029337704696399,
      "grad_norm": 0.095298171043396,
      "learning_rate": 3.203052074497523e-06,
      "loss": 0.0833730697631836,
      "step": 5939
    },
    {
      "epoch": 0.8030689672654758,
      "grad_norm": 0.0649867132306099,
      "learning_rate": 3.198825929497752e-06,
      "loss": 0.0504755973815918,
      "step": 5940
    },
    {
      "epoch": 0.8032041640613118,
      "grad_norm": 0.08923760801553726,
      "learning_rate": 3.194602241565357e-06,
      "loss": 0.048790931701660156,
      "step": 5941
    },
    {
      "epoch": 0.8033393608571476,
      "grad_norm": 0.07723937183618546,
      "learning_rate": 3.1903810115797282e-06,
      "loss": 0.06285953521728516,
      "step": 5942
    },
    {
      "epoch": 0.8034745576529836,
      "grad_norm": 0.10952747613191605,
      "learning_rate": 3.1861622404197475e-06,
      "loss": 0.06120920181274414,
      "step": 5943
    },
    {
      "epoch": 0.8036097544488195,
      "grad_norm": 0.04562440514564514,
      "learning_rate": 3.181945928963794e-06,
      "loss": 0.05415225028991699,
      "step": 5944
    },
    {
      "epoch": 0.8037449512446555,
      "grad_norm": 0.07636386901140213,
      "learning_rate": 3.1777320780897124e-06,
      "loss": 0.06434714794158936,
      "step": 5945
    },
    {
      "epoch": 0.8038801480404915,
      "grad_norm": 0.040571264922618866,
      "learning_rate": 3.1735206886748602e-06,
      "loss": 0.04030346870422363,
      "step": 5946
    },
    {
      "epoch": 0.8040153448363274,
      "grad_norm": 0.10323947668075562,
      "learning_rate": 3.1693117615960665e-06,
      "loss": 0.08153200149536133,
      "step": 5947
    },
    {
      "epoch": 0.8041505416321633,
      "grad_norm": 0.07518929243087769,
      "learning_rate": 3.1651052977296537e-06,
      "loss": 0.0575709342956543,
      "step": 5948
    },
    {
      "epoch": 0.8042857384279992,
      "grad_norm": 0.08393549919128418,
      "learning_rate": 3.1609012979514273e-06,
      "loss": 0.07520294189453125,
      "step": 5949
    },
    {
      "epoch": 0.8044209352238352,
      "grad_norm": 0.05883118510246277,
      "learning_rate": 3.156699763136683e-06,
      "loss": 0.052559852600097656,
      "step": 5950
    },
    {
      "epoch": 0.8045561320196711,
      "grad_norm": 0.06830005347728729,
      "learning_rate": 3.152500694160207e-06,
      "loss": 0.056876182556152344,
      "step": 5951
    },
    {
      "epoch": 0.8046913288155071,
      "grad_norm": 0.06434983760118484,
      "learning_rate": 3.148304091896265e-06,
      "loss": 0.053955793380737305,
      "step": 5952
    },
    {
      "epoch": 0.804826525611343,
      "grad_norm": 0.09729620069265366,
      "learning_rate": 3.144109957218612e-06,
      "loss": 0.0799567699432373,
      "step": 5953
    },
    {
      "epoch": 0.804961722407179,
      "grad_norm": 0.05446004122495651,
      "learning_rate": 3.1399182910004893e-06,
      "loss": 0.04692339897155762,
      "step": 5954
    },
    {
      "epoch": 0.8050969192030148,
      "grad_norm": 0.0832701027393341,
      "learning_rate": 3.1357290941146215e-06,
      "loss": 0.07751703262329102,
      "step": 5955
    },
    {
      "epoch": 0.8052321159988508,
      "grad_norm": 0.0588974803686142,
      "learning_rate": 3.1315423674332265e-06,
      "loss": 0.04872506856918335,
      "step": 5956
    },
    {
      "epoch": 0.8053673127946868,
      "grad_norm": 0.0907578393816948,
      "learning_rate": 3.127358111828002e-06,
      "loss": 0.06001472473144531,
      "step": 5957
    },
    {
      "epoch": 0.8055025095905227,
      "grad_norm": 0.10768931359052658,
      "learning_rate": 3.123176328170131e-06,
      "loss": 0.0764617919921875,
      "step": 5958
    },
    {
      "epoch": 0.8056377063863587,
      "grad_norm": 0.1117824986577034,
      "learning_rate": 3.1189970173302816e-06,
      "loss": 0.0704648494720459,
      "step": 5959
    },
    {
      "epoch": 0.8057729031821946,
      "grad_norm": 0.12021458894014359,
      "learning_rate": 3.1148201801786085e-06,
      "loss": 0.050488829612731934,
      "step": 5960
    },
    {
      "epoch": 0.8059080999780305,
      "grad_norm": 0.08330684155225754,
      "learning_rate": 3.1106458175847572e-06,
      "loss": 0.07120692729949951,
      "step": 5961
    },
    {
      "epoch": 0.8060432967738664,
      "grad_norm": 0.07720736414194107,
      "learning_rate": 3.106473930417848e-06,
      "loss": 0.05233168601989746,
      "step": 5962
    },
    {
      "epoch": 0.8061784935697024,
      "grad_norm": 0.02804306335747242,
      "learning_rate": 3.1023045195464903e-06,
      "loss": 0.023141920566558838,
      "step": 5963
    },
    {
      "epoch": 0.8063136903655383,
      "grad_norm": 0.06017075479030609,
      "learning_rate": 3.098137585838779e-06,
      "loss": 0.05536526441574097,
      "step": 5964
    },
    {
      "epoch": 0.8064488871613743,
      "grad_norm": 0.07462954521179199,
      "learning_rate": 3.093973130162286e-06,
      "loss": 0.05035829544067383,
      "step": 5965
    },
    {
      "epoch": 0.8065840839572103,
      "grad_norm": 0.027970924973487854,
      "learning_rate": 3.089811153384083e-06,
      "loss": 0.02471745014190674,
      "step": 5966
    },
    {
      "epoch": 0.8067192807530461,
      "grad_norm": 0.09911976754665375,
      "learning_rate": 3.08565165637071e-06,
      "loss": 0.07069277763366699,
      "step": 5967
    },
    {
      "epoch": 0.8068544775488821,
      "grad_norm": 0.03710552304983139,
      "learning_rate": 3.081494639988196e-06,
      "loss": 0.03911447525024414,
      "step": 5968
    },
    {
      "epoch": 0.806989674344718,
      "grad_norm": 0.06144268065690994,
      "learning_rate": 3.077340105102057e-06,
      "loss": 0.06378602981567383,
      "step": 5969
    },
    {
      "epoch": 0.807124871140554,
      "grad_norm": 0.09934115409851074,
      "learning_rate": 3.0731880525772817e-06,
      "loss": 0.07390356063842773,
      "step": 5970
    },
    {
      "epoch": 0.8072600679363899,
      "grad_norm": 0.06829255819320679,
      "learning_rate": 3.069038483278364e-06,
      "loss": 0.05382704734802246,
      "step": 5971
    },
    {
      "epoch": 0.8073952647322259,
      "grad_norm": 0.06651218980550766,
      "learning_rate": 3.0648913980692505e-06,
      "loss": 0.05577385425567627,
      "step": 5972
    },
    {
      "epoch": 0.8075304615280617,
      "grad_norm": 0.07062459737062454,
      "learning_rate": 3.0607467978133985e-06,
      "loss": 0.04932081699371338,
      "step": 5973
    },
    {
      "epoch": 0.8076656583238977,
      "grad_norm": 0.07902832329273224,
      "learning_rate": 3.0566046833737294e-06,
      "loss": 0.05445671081542969,
      "step": 5974
    },
    {
      "epoch": 0.8078008551197337,
      "grad_norm": 0.07453961670398712,
      "learning_rate": 3.0524650556126517e-06,
      "loss": 0.042623937129974365,
      "step": 5975
    },
    {
      "epoch": 0.8079360519155696,
      "grad_norm": 0.04572852700948715,
      "learning_rate": 3.048327915392069e-06,
      "loss": 0.04386827349662781,
      "step": 5976
    },
    {
      "epoch": 0.8080712487114056,
      "grad_norm": 0.07879126071929932,
      "learning_rate": 3.044193263573341e-06,
      "loss": 0.07558250427246094,
      "step": 5977
    },
    {
      "epoch": 0.8082064455072415,
      "grad_norm": 0.12469901889562607,
      "learning_rate": 3.0400611010173355e-06,
      "loss": 0.07771492004394531,
      "step": 5978
    },
    {
      "epoch": 0.8083416423030774,
      "grad_norm": 0.058566123247146606,
      "learning_rate": 3.0359314285843863e-06,
      "loss": 0.04871082305908203,
      "step": 5979
    },
    {
      "epoch": 0.8084768390989133,
      "grad_norm": 0.09537943452596664,
      "learning_rate": 3.0318042471343104e-06,
      "loss": 0.07441234588623047,
      "step": 5980
    },
    {
      "epoch": 0.8086120358947493,
      "grad_norm": 0.09384571760892868,
      "learning_rate": 3.027679557526422e-06,
      "loss": 0.0855410099029541,
      "step": 5981
    },
    {
      "epoch": 0.8087472326905852,
      "grad_norm": 0.06223554164171219,
      "learning_rate": 3.0235573606194844e-06,
      "loss": 0.04994767904281616,
      "step": 5982
    },
    {
      "epoch": 0.8088824294864212,
      "grad_norm": 0.06187053024768829,
      "learning_rate": 3.0194376572717743e-06,
      "loss": 0.04442010819911957,
      "step": 5983
    },
    {
      "epoch": 0.8090176262822572,
      "grad_norm": 0.16885416209697723,
      "learning_rate": 3.0153204483410318e-06,
      "loss": 0.07637345790863037,
      "step": 5984
    },
    {
      "epoch": 0.809152823078093,
      "grad_norm": 0.07922068983316422,
      "learning_rate": 3.0112057346844834e-06,
      "loss": 0.06919288635253906,
      "step": 5985
    },
    {
      "epoch": 0.809288019873929,
      "grad_norm": 0.10322743654251099,
      "learning_rate": 3.007093517158832e-06,
      "loss": 0.06879043579101562,
      "step": 5986
    },
    {
      "epoch": 0.8094232166697649,
      "grad_norm": 0.17313022911548615,
      "learning_rate": 3.002983796620261e-06,
      "loss": 0.12035131454467773,
      "step": 5987
    },
    {
      "epoch": 0.8095584134656009,
      "grad_norm": 0.17304308712482452,
      "learning_rate": 2.9988765739244427e-06,
      "loss": 0.07915401458740234,
      "step": 5988
    },
    {
      "epoch": 0.8096936102614368,
      "grad_norm": 0.07025691866874695,
      "learning_rate": 2.9947718499265197e-06,
      "loss": 0.05477166175842285,
      "step": 5989
    },
    {
      "epoch": 0.8098288070572728,
      "grad_norm": 0.11004645377397537,
      "learning_rate": 2.9906696254811184e-06,
      "loss": 0.07807785272598267,
      "step": 5990
    },
    {
      "epoch": 0.8099640038531086,
      "grad_norm": 0.062981016933918,
      "learning_rate": 2.9865699014423404e-06,
      "loss": 0.056607723236083984,
      "step": 5991
    },
    {
      "epoch": 0.8100992006489446,
      "grad_norm": 0.13426287472248077,
      "learning_rate": 2.9824726786637698e-06,
      "loss": 0.06773710250854492,
      "step": 5992
    },
    {
      "epoch": 0.8102343974447805,
      "grad_norm": 0.06862668693065643,
      "learning_rate": 2.978377957998477e-06,
      "loss": 0.06562089920043945,
      "step": 5993
    },
    {
      "epoch": 0.8103695942406165,
      "grad_norm": 0.07917889952659607,
      "learning_rate": 2.974285740299001e-06,
      "loss": 0.08538436889648438,
      "step": 5994
    },
    {
      "epoch": 0.8105047910364525,
      "grad_norm": 0.07612813264131546,
      "learning_rate": 2.9701960264173612e-06,
      "loss": 0.05460333824157715,
      "step": 5995
    },
    {
      "epoch": 0.8106399878322884,
      "grad_norm": 0.07773082703351974,
      "learning_rate": 2.96610881720506e-06,
      "loss": 0.06387847661972046,
      "step": 5996
    },
    {
      "epoch": 0.8107751846281244,
      "grad_norm": 0.04482792317867279,
      "learning_rate": 2.9620241135130715e-06,
      "loss": 0.040294647216796875,
      "step": 5997
    },
    {
      "epoch": 0.8109103814239602,
      "grad_norm": 0.12301840633153915,
      "learning_rate": 2.9579419161918607e-06,
      "loss": 0.07533693313598633,
      "step": 5998
    },
    {
      "epoch": 0.8110455782197962,
      "grad_norm": 0.06225501745939255,
      "learning_rate": 2.9538622260913595e-06,
      "loss": 0.05575919151306152,
      "step": 5999
    },
    {
      "epoch": 0.8111807750156321,
      "grad_norm": 0.11108613014221191,
      "learning_rate": 2.9497850440609814e-06,
      "loss": 0.08737683296203613,
      "step": 6000
    },
    {
      "epoch": 0.8113159718114681,
      "grad_norm": 0.05581967905163765,
      "learning_rate": 2.945710370949616e-06,
      "loss": 0.05908411741256714,
      "step": 6001
    },
    {
      "epoch": 0.811451168607304,
      "grad_norm": 0.10604329407215118,
      "learning_rate": 2.941638207605629e-06,
      "loss": 0.07580089569091797,
      "step": 6002
    },
    {
      "epoch": 0.81158636540314,
      "grad_norm": 0.12798796594142914,
      "learning_rate": 2.937568554876873e-06,
      "loss": 0.07222437858581543,
      "step": 6003
    },
    {
      "epoch": 0.8117215621989758,
      "grad_norm": 0.06994862109422684,
      "learning_rate": 2.9335014136106704e-06,
      "loss": 0.053401947021484375,
      "step": 6004
    },
    {
      "epoch": 0.8118567589948118,
      "grad_norm": 0.04690518230199814,
      "learning_rate": 2.929436784653818e-06,
      "loss": 0.03975629806518555,
      "step": 6005
    },
    {
      "epoch": 0.8119919557906478,
      "grad_norm": 0.08309400081634521,
      "learning_rate": 2.925374668852597e-06,
      "loss": 0.08277511596679688,
      "step": 6006
    },
    {
      "epoch": 0.8121271525864837,
      "grad_norm": 0.07683993875980377,
      "learning_rate": 2.921315067052754e-06,
      "loss": 0.052703857421875,
      "step": 6007
    },
    {
      "epoch": 0.8122623493823197,
      "grad_norm": 0.07410429418087006,
      "learning_rate": 2.917257980099535e-06,
      "loss": 0.06193804740905762,
      "step": 6008
    },
    {
      "epoch": 0.8123975461781556,
      "grad_norm": 0.1447361558675766,
      "learning_rate": 2.913203408837629e-06,
      "loss": 0.07332003116607666,
      "step": 6009
    },
    {
      "epoch": 0.8125327429739915,
      "grad_norm": 0.093193918466568,
      "learning_rate": 2.909151354111232e-06,
      "loss": 0.056601524353027344,
      "step": 6010
    },
    {
      "epoch": 0.8126679397698274,
      "grad_norm": 0.08866371959447861,
      "learning_rate": 2.905101816763998e-06,
      "loss": 0.0764303207397461,
      "step": 6011
    },
    {
      "epoch": 0.8128031365656634,
      "grad_norm": 0.09341947734355927,
      "learning_rate": 2.9010547976390617e-06,
      "loss": 0.057110607624053955,
      "step": 6012
    },
    {
      "epoch": 0.8129383333614993,
      "grad_norm": 0.043275635689496994,
      "learning_rate": 2.897010297579042e-06,
      "loss": 0.04230165481567383,
      "step": 6013
    },
    {
      "epoch": 0.8130735301573353,
      "grad_norm": 0.0816507637500763,
      "learning_rate": 2.8929683174260133e-06,
      "loss": 0.06053817272186279,
      "step": 6014
    },
    {
      "epoch": 0.8132087269531713,
      "grad_norm": 0.04000457748770714,
      "learning_rate": 2.8889288580215467e-06,
      "loss": 0.04054450988769531,
      "step": 6015
    },
    {
      "epoch": 0.8133439237490071,
      "grad_norm": 0.08455421030521393,
      "learning_rate": 2.8848919202066752e-06,
      "loss": 0.07124114036560059,
      "step": 6016
    },
    {
      "epoch": 0.8134791205448431,
      "grad_norm": 0.11735472828149796,
      "learning_rate": 2.8808575048219123e-06,
      "loss": 0.08161544799804688,
      "step": 6017
    },
    {
      "epoch": 0.813614317340679,
      "grad_norm": 0.0470954105257988,
      "learning_rate": 2.8768256127072436e-06,
      "loss": 0.042212843894958496,
      "step": 6018
    },
    {
      "epoch": 0.813749514136515,
      "grad_norm": 0.08306185156106949,
      "learning_rate": 2.872796244702128e-06,
      "loss": 0.039395809173583984,
      "step": 6019
    },
    {
      "epoch": 0.8138847109323509,
      "grad_norm": 0.07860999554395676,
      "learning_rate": 2.8687694016455075e-06,
      "loss": 0.05400872230529785,
      "step": 6020
    },
    {
      "epoch": 0.8140199077281869,
      "grad_norm": 0.09967145323753357,
      "learning_rate": 2.86474508437579e-06,
      "loss": 0.06613636016845703,
      "step": 6021
    },
    {
      "epoch": 0.8141551045240227,
      "grad_norm": 0.13687734305858612,
      "learning_rate": 2.8607232937308587e-06,
      "loss": 0.08966779708862305,
      "step": 6022
    },
    {
      "epoch": 0.8142903013198587,
      "grad_norm": 0.09404417127370834,
      "learning_rate": 2.856704030548072e-06,
      "loss": 0.03868138790130615,
      "step": 6023
    },
    {
      "epoch": 0.8144254981156946,
      "grad_norm": 0.07717647403478622,
      "learning_rate": 2.8526872956642568e-06,
      "loss": 0.056162476539611816,
      "step": 6024
    },
    {
      "epoch": 0.8145606949115306,
      "grad_norm": 0.1365281641483307,
      "learning_rate": 2.84867308991573e-06,
      "loss": 0.043724775314331055,
      "step": 6025
    },
    {
      "epoch": 0.8146958917073666,
      "grad_norm": 0.08922185748815536,
      "learning_rate": 2.8446614141382638e-06,
      "loss": 0.05934906005859375,
      "step": 6026
    },
    {
      "epoch": 0.8148310885032025,
      "grad_norm": 0.062307439744472504,
      "learning_rate": 2.8406522691671104e-06,
      "loss": 0.04979896545410156,
      "step": 6027
    },
    {
      "epoch": 0.8149662852990384,
      "grad_norm": 0.06810615956783295,
      "learning_rate": 2.8366456558369975e-06,
      "loss": 0.057108163833618164,
      "step": 6028
    },
    {
      "epoch": 0.8151014820948743,
      "grad_norm": 0.08926712721586227,
      "learning_rate": 2.8326415749821186e-06,
      "loss": 0.044838547706604004,
      "step": 6029
    },
    {
      "epoch": 0.8152366788907103,
      "grad_norm": 0.0773458480834961,
      "learning_rate": 2.828640027436151e-06,
      "loss": 0.04747915267944336,
      "step": 6030
    },
    {
      "epoch": 0.8153718756865462,
      "grad_norm": 0.08920859545469284,
      "learning_rate": 2.824641014032235e-06,
      "loss": 0.05875211954116821,
      "step": 6031
    },
    {
      "epoch": 0.8155070724823822,
      "grad_norm": 0.11277290433645248,
      "learning_rate": 2.820644535602987e-06,
      "loss": 0.05080556869506836,
      "step": 6032
    },
    {
      "epoch": 0.8156422692782181,
      "grad_norm": 0.14920265972614288,
      "learning_rate": 2.8166505929804953e-06,
      "loss": 0.06355094909667969,
      "step": 6033
    },
    {
      "epoch": 0.815777466074054,
      "grad_norm": 0.06301344186067581,
      "learning_rate": 2.8126591869963163e-06,
      "loss": 0.04443478584289551,
      "step": 6034
    },
    {
      "epoch": 0.81591266286989,
      "grad_norm": 0.060130201280117035,
      "learning_rate": 2.8086703184814887e-06,
      "loss": 0.05647540092468262,
      "step": 6035
    },
    {
      "epoch": 0.8160478596657259,
      "grad_norm": 0.05849962681531906,
      "learning_rate": 2.8046839882665134e-06,
      "loss": 0.04903674125671387,
      "step": 6036
    },
    {
      "epoch": 0.8161830564615619,
      "grad_norm": 0.08409731835126877,
      "learning_rate": 2.800700197181364e-06,
      "loss": 0.06807661056518555,
      "step": 6037
    },
    {
      "epoch": 0.8163182532573978,
      "grad_norm": 0.07239754498004913,
      "learning_rate": 2.7967189460554876e-06,
      "loss": 0.05629897117614746,
      "step": 6038
    },
    {
      "epoch": 0.8164534500532338,
      "grad_norm": 0.10625198483467102,
      "learning_rate": 2.792740235717801e-06,
      "loss": 0.07801389694213867,
      "step": 6039
    },
    {
      "epoch": 0.8165886468490697,
      "grad_norm": 0.07800500839948654,
      "learning_rate": 2.7887640669967e-06,
      "loss": 0.06755685806274414,
      "step": 6040
    },
    {
      "epoch": 0.8167238436449056,
      "grad_norm": 0.049376923590898514,
      "learning_rate": 2.7847904407200327e-06,
      "loss": 0.05147850513458252,
      "step": 6041
    },
    {
      "epoch": 0.8168590404407415,
      "grad_norm": 0.14100128412246704,
      "learning_rate": 2.7808193577151363e-06,
      "loss": 0.08207482099533081,
      "step": 6042
    },
    {
      "epoch": 0.8169942372365775,
      "grad_norm": 0.08680377900600433,
      "learning_rate": 2.776850818808812e-06,
      "loss": 0.06275999546051025,
      "step": 6043
    },
    {
      "epoch": 0.8171294340324134,
      "grad_norm": 0.06518331915140152,
      "learning_rate": 2.772884824827325e-06,
      "loss": 0.05167031288146973,
      "step": 6044
    },
    {
      "epoch": 0.8172646308282494,
      "grad_norm": 0.08073441684246063,
      "learning_rate": 2.768921376596429e-06,
      "loss": 0.03640162944793701,
      "step": 6045
    },
    {
      "epoch": 0.8173998276240854,
      "grad_norm": 0.06530150026082993,
      "learning_rate": 2.7649604749413176e-06,
      "loss": 0.052735328674316406,
      "step": 6046
    },
    {
      "epoch": 0.8175350244199212,
      "grad_norm": 0.15498989820480347,
      "learning_rate": 2.7610021206866837e-06,
      "loss": 0.06781244277954102,
      "step": 6047
    },
    {
      "epoch": 0.8176702212157572,
      "grad_norm": 0.1281179040670395,
      "learning_rate": 2.757046314656676e-06,
      "loss": 0.05950215458869934,
      "step": 6048
    },
    {
      "epoch": 0.8178054180115931,
      "grad_norm": 0.05508748069405556,
      "learning_rate": 2.753093057674909e-06,
      "loss": 0.053448550403118134,
      "step": 6049
    },
    {
      "epoch": 0.8179406148074291,
      "grad_norm": 0.13670194149017334,
      "learning_rate": 2.749142350564483e-06,
      "loss": 0.09100627899169922,
      "step": 6050
    },
    {
      "epoch": 0.818075811603265,
      "grad_norm": 0.045564886182546616,
      "learning_rate": 2.7451941941479414e-06,
      "loss": 0.04336047172546387,
      "step": 6051
    },
    {
      "epoch": 0.818211008399101,
      "grad_norm": 0.07601013779640198,
      "learning_rate": 2.741248589247323e-06,
      "loss": 0.06634902954101562,
      "step": 6052
    },
    {
      "epoch": 0.8183462051949368,
      "grad_norm": 0.0758441761136055,
      "learning_rate": 2.73730553668412e-06,
      "loss": 0.05973625183105469,
      "step": 6053
    },
    {
      "epoch": 0.8184814019907728,
      "grad_norm": 0.05449908599257469,
      "learning_rate": 2.7333650372792978e-06,
      "loss": 0.054160118103027344,
      "step": 6054
    },
    {
      "epoch": 0.8186165987866088,
      "grad_norm": 0.08510931581258774,
      "learning_rate": 2.7294270918532876e-06,
      "loss": 0.05118298530578613,
      "step": 6055
    },
    {
      "epoch": 0.8187517955824447,
      "grad_norm": 0.13163307309150696,
      "learning_rate": 2.7254917012259882e-06,
      "loss": 0.11231327056884766,
      "step": 6056
    },
    {
      "epoch": 0.8188869923782807,
      "grad_norm": 0.07046449184417725,
      "learning_rate": 2.721558866216776e-06,
      "loss": 0.049848198890686035,
      "step": 6057
    },
    {
      "epoch": 0.8190221891741166,
      "grad_norm": 0.17723996937274933,
      "learning_rate": 2.7176285876444846e-06,
      "loss": 0.06799310445785522,
      "step": 6058
    },
    {
      "epoch": 0.8191573859699525,
      "grad_norm": 0.09139125794172287,
      "learning_rate": 2.713700866327417e-06,
      "loss": 0.05922532081604004,
      "step": 6059
    },
    {
      "epoch": 0.8192925827657884,
      "grad_norm": 0.07579890638589859,
      "learning_rate": 2.7097757030833497e-06,
      "loss": 0.05732536315917969,
      "step": 6060
    },
    {
      "epoch": 0.8194277795616244,
      "grad_norm": 0.194512277841568,
      "learning_rate": 2.705853098729517e-06,
      "loss": 0.07489776611328125,
      "step": 6061
    },
    {
      "epoch": 0.8195629763574603,
      "grad_norm": 0.16335757076740265,
      "learning_rate": 2.7019330540826325e-06,
      "loss": 0.0741119384765625,
      "step": 6062
    },
    {
      "epoch": 0.8196981731532963,
      "grad_norm": 0.047551579773426056,
      "learning_rate": 2.6980155699588666e-06,
      "loss": 0.03277790546417236,
      "step": 6063
    },
    {
      "epoch": 0.8198333699491323,
      "grad_norm": 0.1278848499059677,
      "learning_rate": 2.6941006471738633e-06,
      "loss": 0.055891990661621094,
      "step": 6064
    },
    {
      "epoch": 0.8199685667449681,
      "grad_norm": 0.07446646690368652,
      "learning_rate": 2.690188286542726e-06,
      "loss": 0.06618189811706543,
      "step": 6065
    },
    {
      "epoch": 0.820103763540804,
      "grad_norm": 0.13082048296928406,
      "learning_rate": 2.686278488880029e-06,
      "loss": 0.08866214752197266,
      "step": 6066
    },
    {
      "epoch": 0.82023896033664,
      "grad_norm": 0.0632932186126709,
      "learning_rate": 2.6823712549998187e-06,
      "loss": 0.057805657386779785,
      "step": 6067
    },
    {
      "epoch": 0.820374157132476,
      "grad_norm": 0.0960746705532074,
      "learning_rate": 2.678466585715599e-06,
      "loss": 0.05298125743865967,
      "step": 6068
    },
    {
      "epoch": 0.8205093539283119,
      "grad_norm": 0.11340481042861938,
      "learning_rate": 2.6745644818403426e-06,
      "loss": 0.07090479135513306,
      "step": 6069
    },
    {
      "epoch": 0.8206445507241479,
      "grad_norm": 0.20877237617969513,
      "learning_rate": 2.6706649441864883e-06,
      "loss": 0.06981277465820312,
      "step": 6070
    },
    {
      "epoch": 0.8207797475199837,
      "grad_norm": 0.06653616577386856,
      "learning_rate": 2.666767973565937e-06,
      "loss": 0.055539608001708984,
      "step": 6071
    },
    {
      "epoch": 0.8209149443158197,
      "grad_norm": 0.18026667833328247,
      "learning_rate": 2.6628735707900653e-06,
      "loss": 0.07924997806549072,
      "step": 6072
    },
    {
      "epoch": 0.8210501411116556,
      "grad_norm": 0.11764676868915558,
      "learning_rate": 2.658981736669707e-06,
      "loss": 0.045154869556427,
      "step": 6073
    },
    {
      "epoch": 0.8211853379074916,
      "grad_norm": 0.11506257951259613,
      "learning_rate": 2.655092472015161e-06,
      "loss": 0.04538559913635254,
      "step": 6074
    },
    {
      "epoch": 0.8213205347033276,
      "grad_norm": 0.08606728911399841,
      "learning_rate": 2.6512057776361935e-06,
      "loss": 0.06519508361816406,
      "step": 6075
    },
    {
      "epoch": 0.8214557314991635,
      "grad_norm": 0.10523980110883713,
      "learning_rate": 2.64732165434203e-06,
      "loss": 0.0547870397567749,
      "step": 6076
    },
    {
      "epoch": 0.8215909282949994,
      "grad_norm": 0.08119869977235794,
      "learning_rate": 2.6434401029413792e-06,
      "loss": 0.042222678661346436,
      "step": 6077
    },
    {
      "epoch": 0.8217261250908353,
      "grad_norm": 0.11899159848690033,
      "learning_rate": 2.639561124242385e-06,
      "loss": 0.06535935401916504,
      "step": 6078
    },
    {
      "epoch": 0.8218613218866713,
      "grad_norm": 0.052903711795806885,
      "learning_rate": 2.635684719052682e-06,
      "loss": 0.03343915939331055,
      "step": 6079
    },
    {
      "epoch": 0.8219965186825072,
      "grad_norm": 0.038406386971473694,
      "learning_rate": 2.631810888179355e-06,
      "loss": 0.030523747205734253,
      "step": 6080
    },
    {
      "epoch": 0.8221317154783432,
      "grad_norm": 0.040950872004032135,
      "learning_rate": 2.627939632428952e-06,
      "loss": 0.04197734594345093,
      "step": 6081
    },
    {
      "epoch": 0.8222669122741791,
      "grad_norm": 0.06389142572879791,
      "learning_rate": 2.624070952607502e-06,
      "loss": 0.04342985153198242,
      "step": 6082
    },
    {
      "epoch": 0.822402109070015,
      "grad_norm": 0.12661327421665192,
      "learning_rate": 2.620204849520468e-06,
      "loss": 0.07553958892822266,
      "step": 6083
    },
    {
      "epoch": 0.8225373058658509,
      "grad_norm": 0.06931939721107483,
      "learning_rate": 2.616341323972806e-06,
      "loss": 0.053159356117248535,
      "step": 6084
    },
    {
      "epoch": 0.8226725026616869,
      "grad_norm": 0.09741483628749847,
      "learning_rate": 2.612480376768917e-06,
      "loss": 0.0687553882598877,
      "step": 6085
    },
    {
      "epoch": 0.8228076994575229,
      "grad_norm": 0.06940137594938278,
      "learning_rate": 2.608622008712672e-06,
      "loss": 0.059928178787231445,
      "step": 6086
    },
    {
      "epoch": 0.8229428962533588,
      "grad_norm": 0.07559097558259964,
      "learning_rate": 2.6047662206074034e-06,
      "loss": 0.053426504135131836,
      "step": 6087
    },
    {
      "epoch": 0.8230780930491948,
      "grad_norm": 0.12496233731508255,
      "learning_rate": 2.600913013255904e-06,
      "loss": 0.0745849609375,
      "step": 6088
    },
    {
      "epoch": 0.8232132898450307,
      "grad_norm": 0.1043851375579834,
      "learning_rate": 2.59706238746044e-06,
      "loss": 0.06103801727294922,
      "step": 6089
    },
    {
      "epoch": 0.8233484866408666,
      "grad_norm": 0.11068032681941986,
      "learning_rate": 2.593214344022725e-06,
      "loss": 0.06898415088653564,
      "step": 6090
    },
    {
      "epoch": 0.8234836834367025,
      "grad_norm": 0.07048270106315613,
      "learning_rate": 2.5893688837439474e-06,
      "loss": 0.06960821151733398,
      "step": 6091
    },
    {
      "epoch": 0.8236188802325385,
      "grad_norm": 0.061021458357572556,
      "learning_rate": 2.5855260074247473e-06,
      "loss": 0.03547525405883789,
      "step": 6092
    },
    {
      "epoch": 0.8237540770283744,
      "grad_norm": 0.05903881415724754,
      "learning_rate": 2.581685715865232e-06,
      "loss": 0.0526430606842041,
      "step": 6093
    },
    {
      "epoch": 0.8238892738242104,
      "grad_norm": 0.05299198627471924,
      "learning_rate": 2.5778480098649766e-06,
      "loss": 0.05292820930480957,
      "step": 6094
    },
    {
      "epoch": 0.8240244706200464,
      "grad_norm": 0.06044149771332741,
      "learning_rate": 2.5740128902230087e-06,
      "loss": 0.05222296714782715,
      "step": 6095
    },
    {
      "epoch": 0.8241596674158822,
      "grad_norm": 0.07073352485895157,
      "learning_rate": 2.5701803577378214e-06,
      "loss": 0.06006669998168945,
      "step": 6096
    },
    {
      "epoch": 0.8242948642117182,
      "grad_norm": 0.08059313148260117,
      "learning_rate": 2.566350413207366e-06,
      "loss": 0.0693356990814209,
      "step": 6097
    },
    {
      "epoch": 0.8244300610075541,
      "grad_norm": 0.03012383170425892,
      "learning_rate": 2.5625230574290554e-06,
      "loss": 0.026161015033721924,
      "step": 6098
    },
    {
      "epoch": 0.8245652578033901,
      "grad_norm": 0.04499491676688194,
      "learning_rate": 2.558698291199773e-06,
      "loss": 0.03758955001831055,
      "step": 6099
    },
    {
      "epoch": 0.824700454599226,
      "grad_norm": 0.06370619684457779,
      "learning_rate": 2.5548761153158524e-06,
      "loss": 0.06589460372924805,
      "step": 6100
    },
    {
      "epoch": 0.824835651395062,
      "grad_norm": 0.0985814705491066,
      "learning_rate": 2.55105653057309e-06,
      "loss": 0.05526965856552124,
      "step": 6101
    },
    {
      "epoch": 0.8249708481908978,
      "grad_norm": 0.13217344880104065,
      "learning_rate": 2.547239537766743e-06,
      "loss": 0.0590825080871582,
      "step": 6102
    },
    {
      "epoch": 0.8251060449867338,
      "grad_norm": 0.1189265102148056,
      "learning_rate": 2.543425137691526e-06,
      "loss": 0.08219194412231445,
      "step": 6103
    },
    {
      "epoch": 0.8252412417825697,
      "grad_norm": 0.2378210425376892,
      "learning_rate": 2.5396133311416264e-06,
      "loss": 0.10416507720947266,
      "step": 6104
    },
    {
      "epoch": 0.8253764385784057,
      "grad_norm": 0.06367621570825577,
      "learning_rate": 2.5358041189106784e-06,
      "loss": 0.06879687309265137,
      "step": 6105
    },
    {
      "epoch": 0.8255116353742417,
      "grad_norm": 0.04631015658378601,
      "learning_rate": 2.531997501791779e-06,
      "loss": 0.05221867561340332,
      "step": 6106
    },
    {
      "epoch": 0.8256468321700776,
      "grad_norm": 0.055972620844841,
      "learning_rate": 2.528193480577489e-06,
      "loss": 0.05498814582824707,
      "step": 6107
    },
    {
      "epoch": 0.8257820289659135,
      "grad_norm": 0.11147332191467285,
      "learning_rate": 2.5243920560598186e-06,
      "loss": 0.04958057403564453,
      "step": 6108
    },
    {
      "epoch": 0.8259172257617494,
      "grad_norm": 0.16497240960597992,
      "learning_rate": 2.5205932290302598e-06,
      "loss": 0.06731081008911133,
      "step": 6109
    },
    {
      "epoch": 0.8260524225575854,
      "grad_norm": 0.08538535237312317,
      "learning_rate": 2.516797000279729e-06,
      "loss": 0.06237339973449707,
      "step": 6110
    },
    {
      "epoch": 0.8261876193534213,
      "grad_norm": 0.08002668619155884,
      "learning_rate": 2.513003370598637e-06,
      "loss": 0.04990202188491821,
      "step": 6111
    },
    {
      "epoch": 0.8263228161492573,
      "grad_norm": 0.18474170565605164,
      "learning_rate": 2.509212340776832e-06,
      "loss": 0.07783794403076172,
      "step": 6112
    },
    {
      "epoch": 0.8264580129450932,
      "grad_norm": 0.06811453402042389,
      "learning_rate": 2.505423911603622e-06,
      "loss": 0.05213165283203125,
      "step": 6113
    },
    {
      "epoch": 0.8265932097409291,
      "grad_norm": 0.10158013552427292,
      "learning_rate": 2.501638083867789e-06,
      "loss": 0.05647468566894531,
      "step": 6114
    },
    {
      "epoch": 0.826728406536765,
      "grad_norm": 0.1357419490814209,
      "learning_rate": 2.497854858357552e-06,
      "loss": 0.058693885803222656,
      "step": 6115
    },
    {
      "epoch": 0.826863603332601,
      "grad_norm": 0.08465646952390671,
      "learning_rate": 2.494074235860604e-06,
      "loss": 0.06606149673461914,
      "step": 6116
    },
    {
      "epoch": 0.826998800128437,
      "grad_norm": 0.04319896176457405,
      "learning_rate": 2.4902962171640913e-06,
      "loss": 0.02218031883239746,
      "step": 6117
    },
    {
      "epoch": 0.8271339969242729,
      "grad_norm": 0.09505002200603485,
      "learning_rate": 2.4865208030546167e-06,
      "loss": 0.04378557205200195,
      "step": 6118
    },
    {
      "epoch": 0.8272691937201089,
      "grad_norm": 0.05450539290904999,
      "learning_rate": 2.482747994318239e-06,
      "loss": 0.06740474700927734,
      "step": 6119
    },
    {
      "epoch": 0.8274043905159447,
      "grad_norm": 0.11607981473207474,
      "learning_rate": 2.478977791740477e-06,
      "loss": 0.08897721767425537,
      "step": 6120
    },
    {
      "epoch": 0.8275395873117807,
      "grad_norm": 0.11895623058080673,
      "learning_rate": 2.475210196106313e-06,
      "loss": 0.051215410232543945,
      "step": 6121
    },
    {
      "epoch": 0.8276747841076166,
      "grad_norm": 0.08639207482337952,
      "learning_rate": 2.4714452082001753e-06,
      "loss": 0.05059981346130371,
      "step": 6122
    },
    {
      "epoch": 0.8278099809034526,
      "grad_norm": 0.07842420041561127,
      "learning_rate": 2.467682828805956e-06,
      "loss": 0.05503344535827637,
      "step": 6123
    },
    {
      "epoch": 0.8279451776992885,
      "grad_norm": 0.07524443417787552,
      "learning_rate": 2.4639230587070017e-06,
      "loss": 0.08451175689697266,
      "step": 6124
    },
    {
      "epoch": 0.8280803744951245,
      "grad_norm": 0.06326291710138321,
      "learning_rate": 2.460165898686114e-06,
      "loss": 0.050694823265075684,
      "step": 6125
    },
    {
      "epoch": 0.8282155712909604,
      "grad_norm": 0.033216092735528946,
      "learning_rate": 2.4564113495255597e-06,
      "loss": 0.03252220153808594,
      "step": 6126
    },
    {
      "epoch": 0.8283507680867963,
      "grad_norm": 0.10565914958715439,
      "learning_rate": 2.4526594120070545e-06,
      "loss": 0.04973626136779785,
      "step": 6127
    },
    {
      "epoch": 0.8284859648826323,
      "grad_norm": 0.03650711104273796,
      "learning_rate": 2.4489100869117686e-06,
      "loss": 0.04173707962036133,
      "step": 6128
    },
    {
      "epoch": 0.8286211616784682,
      "grad_norm": 0.11496257036924362,
      "learning_rate": 2.4451633750203344e-06,
      "loss": 0.05859875679016113,
      "step": 6129
    },
    {
      "epoch": 0.8287563584743042,
      "grad_norm": 0.05917206034064293,
      "learning_rate": 2.441419277112831e-06,
      "loss": 0.047881126403808594,
      "step": 6130
    },
    {
      "epoch": 0.8288915552701401,
      "grad_norm": 0.08097994327545166,
      "learning_rate": 2.4376777939688107e-06,
      "loss": 0.07488274574279785,
      "step": 6131
    },
    {
      "epoch": 0.8290267520659761,
      "grad_norm": 0.08778148889541626,
      "learning_rate": 2.4339389263672625e-06,
      "loss": 0.052522242069244385,
      "step": 6132
    },
    {
      "epoch": 0.8291619488618119,
      "grad_norm": 0.06195569410920143,
      "learning_rate": 2.4302026750866406e-06,
      "loss": 0.05515170097351074,
      "step": 6133
    },
    {
      "epoch": 0.8292971456576479,
      "grad_norm": 0.07184108346700668,
      "learning_rate": 2.4264690409048517e-06,
      "loss": 0.06892275810241699,
      "step": 6134
    },
    {
      "epoch": 0.8294323424534839,
      "grad_norm": 0.1113007664680481,
      "learning_rate": 2.4227380245992555e-06,
      "loss": 0.06133723258972168,
      "step": 6135
    },
    {
      "epoch": 0.8295675392493198,
      "grad_norm": 0.13878749310970306,
      "learning_rate": 2.4190096269466767e-06,
      "loss": 0.09042072296142578,
      "step": 6136
    },
    {
      "epoch": 0.8297027360451558,
      "grad_norm": 0.1017458587884903,
      "learning_rate": 2.415283848723383e-06,
      "loss": 0.056163787841796875,
      "step": 6137
    },
    {
      "epoch": 0.8298379328409917,
      "grad_norm": 0.07465185970067978,
      "learning_rate": 2.411560690705101e-06,
      "loss": 0.06628525257110596,
      "step": 6138
    },
    {
      "epoch": 0.8299731296368276,
      "grad_norm": 0.10816925019025803,
      "learning_rate": 2.4078401536670146e-06,
      "loss": 0.0725405216217041,
      "step": 6139
    },
    {
      "epoch": 0.8301083264326635,
      "grad_norm": 0.07475408166646957,
      "learning_rate": 2.4041222383837538e-06,
      "loss": 0.04389333724975586,
      "step": 6140
    },
    {
      "epoch": 0.8302435232284995,
      "grad_norm": 0.052984319627285004,
      "learning_rate": 2.400406945629418e-06,
      "loss": 0.04824090003967285,
      "step": 6141
    },
    {
      "epoch": 0.8303787200243354,
      "grad_norm": 0.08021537214517593,
      "learning_rate": 2.3966942761775396e-06,
      "loss": 0.04873466491699219,
      "step": 6142
    },
    {
      "epoch": 0.8305139168201714,
      "grad_norm": 0.11845871806144714,
      "learning_rate": 2.3929842308011263e-06,
      "loss": 0.056719303131103516,
      "step": 6143
    },
    {
      "epoch": 0.8306491136160074,
      "grad_norm": 0.06073852255940437,
      "learning_rate": 2.3892768102726236e-06,
      "loss": 0.06802797317504883,
      "step": 6144
    },
    {
      "epoch": 0.8307843104118432,
      "grad_norm": 0.07287821173667908,
      "learning_rate": 2.3855720153639344e-06,
      "loss": 0.06714892387390137,
      "step": 6145
    },
    {
      "epoch": 0.8309195072076792,
      "grad_norm": 0.04477251321077347,
      "learning_rate": 2.381869846846428e-06,
      "loss": 0.04327821731567383,
      "step": 6146
    },
    {
      "epoch": 0.8310547040035151,
      "grad_norm": 0.059387944638729095,
      "learning_rate": 2.3781703054908993e-06,
      "loss": 0.05419564247131348,
      "step": 6147
    },
    {
      "epoch": 0.8311899007993511,
      "grad_norm": 0.11517562717199326,
      "learning_rate": 2.374473392067624e-06,
      "loss": 0.06175410747528076,
      "step": 6148
    },
    {
      "epoch": 0.831325097595187,
      "grad_norm": 0.11997115612030029,
      "learning_rate": 2.370779107346317e-06,
      "loss": 0.06570863723754883,
      "step": 6149
    },
    {
      "epoch": 0.831460294391023,
      "grad_norm": 0.04856313019990921,
      "learning_rate": 2.3670874520961437e-06,
      "loss": 0.0464855432510376,
      "step": 6150
    },
    {
      "epoch": 0.8315954911868588,
      "grad_norm": 0.04417033493518829,
      "learning_rate": 2.3633984270857367e-06,
      "loss": 0.04998207092285156,
      "step": 6151
    },
    {
      "epoch": 0.8317306879826948,
      "grad_norm": 0.07728380709886551,
      "learning_rate": 2.359712033083156e-06,
      "loss": 0.03751516342163086,
      "step": 6152
    },
    {
      "epoch": 0.8318658847785307,
      "grad_norm": 0.058518558740615845,
      "learning_rate": 2.35602827085594e-06,
      "loss": 0.05121469497680664,
      "step": 6153
    },
    {
      "epoch": 0.8320010815743667,
      "grad_norm": 0.10029186308383942,
      "learning_rate": 2.3523471411710644e-06,
      "loss": 0.05831766128540039,
      "step": 6154
    },
    {
      "epoch": 0.8321362783702027,
      "grad_norm": 0.15154419839382172,
      "learning_rate": 2.3486686447949585e-06,
      "loss": 0.07819652557373047,
      "step": 6155
    },
    {
      "epoch": 0.8322714751660386,
      "grad_norm": 0.14425918459892273,
      "learning_rate": 2.3449927824935075e-06,
      "loss": 0.04434826970100403,
      "step": 6156
    },
    {
      "epoch": 0.8324066719618745,
      "grad_norm": 0.044169846922159195,
      "learning_rate": 2.3413195550320393e-06,
      "loss": 0.048297882080078125,
      "step": 6157
    },
    {
      "epoch": 0.8325418687577104,
      "grad_norm": 0.08879756927490234,
      "learning_rate": 2.3376489631753474e-06,
      "loss": 0.0571284294128418,
      "step": 6158
    },
    {
      "epoch": 0.8326770655535464,
      "grad_norm": 0.1313953548669815,
      "learning_rate": 2.3339810076876665e-06,
      "loss": 0.06466960906982422,
      "step": 6159
    },
    {
      "epoch": 0.8328122623493823,
      "grad_norm": 0.04046588018536568,
      "learning_rate": 2.3303156893326815e-06,
      "loss": 0.025317251682281494,
      "step": 6160
    },
    {
      "epoch": 0.8329474591452183,
      "grad_norm": 0.05012265220284462,
      "learning_rate": 2.326653008873535e-06,
      "loss": 0.04676949977874756,
      "step": 6161
    },
    {
      "epoch": 0.8330826559410542,
      "grad_norm": 0.064298115670681,
      "learning_rate": 2.3229929670728085e-06,
      "loss": 0.049768269062042236,
      "step": 6162
    },
    {
      "epoch": 0.8332178527368901,
      "grad_norm": 0.12205102294683456,
      "learning_rate": 2.319335564692554e-06,
      "loss": 0.06771159172058105,
      "step": 6163
    },
    {
      "epoch": 0.833353049532726,
      "grad_norm": 0.08381735533475876,
      "learning_rate": 2.315680802494256e-06,
      "loss": 0.06295493245124817,
      "step": 6164
    },
    {
      "epoch": 0.833488246328562,
      "grad_norm": 0.08064660429954529,
      "learning_rate": 2.312028681238856e-06,
      "loss": 0.04182124137878418,
      "step": 6165
    },
    {
      "epoch": 0.833623443124398,
      "grad_norm": 0.1880762130022049,
      "learning_rate": 2.3083792016867434e-06,
      "loss": 0.09366559982299805,
      "step": 6166
    },
    {
      "epoch": 0.8337586399202339,
      "grad_norm": 0.05527402088046074,
      "learning_rate": 2.304732364597759e-06,
      "loss": 0.04879117012023926,
      "step": 6167
    },
    {
      "epoch": 0.8338938367160699,
      "grad_norm": 0.041861142963171005,
      "learning_rate": 2.3010881707311994e-06,
      "loss": 0.03336071968078613,
      "step": 6168
    },
    {
      "epoch": 0.8340290335119057,
      "grad_norm": 0.12642940878868103,
      "learning_rate": 2.2974466208458017e-06,
      "loss": 0.06938660144805908,
      "step": 6169
    },
    {
      "epoch": 0.8341642303077417,
      "grad_norm": 0.037987321615219116,
      "learning_rate": 2.293807715699755e-06,
      "loss": 0.03653538227081299,
      "step": 6170
    },
    {
      "epoch": 0.8342994271035776,
      "grad_norm": 0.11393680423498154,
      "learning_rate": 2.2901714560507e-06,
      "loss": 0.08897686004638672,
      "step": 6171
    },
    {
      "epoch": 0.8344346238994136,
      "grad_norm": 0.16335955262184143,
      "learning_rate": 2.286537842655722e-06,
      "loss": 0.07434940338134766,
      "step": 6172
    },
    {
      "epoch": 0.8345698206952495,
      "grad_norm": 0.06670046597719193,
      "learning_rate": 2.2829068762713633e-06,
      "loss": 0.0550236701965332,
      "step": 6173
    },
    {
      "epoch": 0.8347050174910855,
      "grad_norm": 0.05221809074282646,
      "learning_rate": 2.279278557653611e-06,
      "loss": 0.041936516761779785,
      "step": 6174
    },
    {
      "epoch": 0.8348402142869213,
      "grad_norm": 0.08448124676942825,
      "learning_rate": 2.2756528875578965e-06,
      "loss": 0.07308614253997803,
      "step": 6175
    },
    {
      "epoch": 0.8349754110827573,
      "grad_norm": 0.07125779986381531,
      "learning_rate": 2.2720298667391067e-06,
      "loss": 0.06439208984375,
      "step": 6176
    },
    {
      "epoch": 0.8351106078785933,
      "grad_norm": 0.08707048743963242,
      "learning_rate": 2.268409495951568e-06,
      "loss": 0.06947469711303711,
      "step": 6177
    },
    {
      "epoch": 0.8352458046744292,
      "grad_norm": 0.06138402596116066,
      "learning_rate": 2.2647917759490723e-06,
      "loss": 0.04711627960205078,
      "step": 6178
    },
    {
      "epoch": 0.8353810014702652,
      "grad_norm": 0.06254766136407852,
      "learning_rate": 2.261176707484834e-06,
      "loss": 0.060301780700683594,
      "step": 6179
    },
    {
      "epoch": 0.8355161982661011,
      "grad_norm": 0.0709356889128685,
      "learning_rate": 2.2575642913115408e-06,
      "loss": 0.05669832229614258,
      "step": 6180
    },
    {
      "epoch": 0.8356513950619371,
      "grad_norm": 0.0727134570479393,
      "learning_rate": 2.253954528181313e-06,
      "loss": 0.05334728956222534,
      "step": 6181
    },
    {
      "epoch": 0.8357865918577729,
      "grad_norm": 0.06224825978279114,
      "learning_rate": 2.2503474188457206e-06,
      "loss": 0.046790242195129395,
      "step": 6182
    },
    {
      "epoch": 0.8359217886536089,
      "grad_norm": 0.06544957309961319,
      "learning_rate": 2.2467429640557903e-06,
      "loss": 0.054503440856933594,
      "step": 6183
    },
    {
      "epoch": 0.8360569854494448,
      "grad_norm": 0.07928641885519028,
      "learning_rate": 2.2431411645619776e-06,
      "loss": 0.04765033721923828,
      "step": 6184
    },
    {
      "epoch": 0.8361921822452808,
      "grad_norm": 0.07874953746795654,
      "learning_rate": 2.239542021114205e-06,
      "loss": 0.05323934555053711,
      "step": 6185
    },
    {
      "epoch": 0.8363273790411168,
      "grad_norm": 0.07254187017679214,
      "learning_rate": 2.2359455344618306e-06,
      "loss": 0.06038475036621094,
      "step": 6186
    },
    {
      "epoch": 0.8364625758369527,
      "grad_norm": 0.04952237010002136,
      "learning_rate": 2.232351705353663e-06,
      "loss": 0.045972347259521484,
      "step": 6187
    },
    {
      "epoch": 0.8365977726327886,
      "grad_norm": 0.08560050278902054,
      "learning_rate": 2.228760534537955e-06,
      "loss": 0.08045077323913574,
      "step": 6188
    },
    {
      "epoch": 0.8367329694286245,
      "grad_norm": 0.051411427557468414,
      "learning_rate": 2.2251720227624044e-06,
      "loss": 0.04520010948181152,
      "step": 6189
    },
    {
      "epoch": 0.8368681662244605,
      "grad_norm": 0.07734411209821701,
      "learning_rate": 2.2215861707741666e-06,
      "loss": 0.049101829528808594,
      "step": 6190
    },
    {
      "epoch": 0.8370033630202964,
      "grad_norm": 0.08254475146532059,
      "learning_rate": 2.2180029793198313e-06,
      "loss": 0.053664207458496094,
      "step": 6191
    },
    {
      "epoch": 0.8371385598161324,
      "grad_norm": 0.06961311399936676,
      "learning_rate": 2.2144224491454363e-06,
      "loss": 0.05651545524597168,
      "step": 6192
    },
    {
      "epoch": 0.8372737566119683,
      "grad_norm": 0.06746058166027069,
      "learning_rate": 2.2108445809964695e-06,
      "loss": 0.05730557441711426,
      "step": 6193
    },
    {
      "epoch": 0.8374089534078042,
      "grad_norm": 0.04513616859912872,
      "learning_rate": 2.2072693756178567e-06,
      "loss": 0.0493779182434082,
      "step": 6194
    },
    {
      "epoch": 0.8375441502036401,
      "grad_norm": 0.16110411286354065,
      "learning_rate": 2.203696833753983e-06,
      "loss": 0.0833125114440918,
      "step": 6195
    },
    {
      "epoch": 0.8376793469994761,
      "grad_norm": 0.07493987679481506,
      "learning_rate": 2.200126956148668e-06,
      "loss": 0.07624554634094238,
      "step": 6196
    },
    {
      "epoch": 0.8378145437953121,
      "grad_norm": 0.08896756172180176,
      "learning_rate": 2.196559743545177e-06,
      "loss": 0.058403462171554565,
      "step": 6197
    },
    {
      "epoch": 0.837949740591148,
      "grad_norm": 0.09199562668800354,
      "learning_rate": 2.1929951966862233e-06,
      "loss": 0.10968255996704102,
      "step": 6198
    },
    {
      "epoch": 0.838084937386984,
      "grad_norm": 0.055772487074136734,
      "learning_rate": 2.1894333163139607e-06,
      "loss": 0.05327785015106201,
      "step": 6199
    },
    {
      "epoch": 0.8382201341828198,
      "grad_norm": 0.07642515748739243,
      "learning_rate": 2.1858741031700015e-06,
      "loss": 0.06313419342041016,
      "step": 6200
    },
    {
      "epoch": 0.8383553309786558,
      "grad_norm": 0.07300259917974472,
      "learning_rate": 2.1823175579953856e-06,
      "loss": 0.0383075475692749,
      "step": 6201
    },
    {
      "epoch": 0.8384905277744917,
      "grad_norm": 0.09861123561859131,
      "learning_rate": 2.1787636815306065e-06,
      "loss": 0.06454706192016602,
      "step": 6202
    },
    {
      "epoch": 0.8386257245703277,
      "grad_norm": 0.06772708147764206,
      "learning_rate": 2.1752124745156005e-06,
      "loss": 0.05844491720199585,
      "step": 6203
    },
    {
      "epoch": 0.8387609213661636,
      "grad_norm": 0.07884043455123901,
      "learning_rate": 2.171663937689744e-06,
      "loss": 0.0663137435913086,
      "step": 6204
    },
    {
      "epoch": 0.8388961181619996,
      "grad_norm": 0.09211120009422302,
      "learning_rate": 2.168118071791868e-06,
      "loss": 0.049735426902770996,
      "step": 6205
    },
    {
      "epoch": 0.8390313149578355,
      "grad_norm": 0.08327385038137436,
      "learning_rate": 2.164574877560237e-06,
      "loss": 0.06398111581802368,
      "step": 6206
    },
    {
      "epoch": 0.8391665117536714,
      "grad_norm": 0.06487874686717987,
      "learning_rate": 2.161034355732564e-06,
      "loss": 0.062338173389434814,
      "step": 6207
    },
    {
      "epoch": 0.8393017085495074,
      "grad_norm": 0.0806300938129425,
      "learning_rate": 2.1574965070460047e-06,
      "loss": 0.06329292058944702,
      "step": 6208
    },
    {
      "epoch": 0.8394369053453433,
      "grad_norm": 0.05296044051647186,
      "learning_rate": 2.1539613322371527e-06,
      "loss": 0.057851314544677734,
      "step": 6209
    },
    {
      "epoch": 0.8395721021411793,
      "grad_norm": 0.04556573927402496,
      "learning_rate": 2.1504288320420613e-06,
      "loss": 0.03304433822631836,
      "step": 6210
    },
    {
      "epoch": 0.8397072989370152,
      "grad_norm": 0.05280722677707672,
      "learning_rate": 2.1468990071962038e-06,
      "loss": 0.03946375846862793,
      "step": 6211
    },
    {
      "epoch": 0.8398424957328511,
      "grad_norm": 0.05114326253533363,
      "learning_rate": 2.143371858434515e-06,
      "loss": 0.03976273536682129,
      "step": 6212
    },
    {
      "epoch": 0.839977692528687,
      "grad_norm": 0.14617249369621277,
      "learning_rate": 2.139847386491367e-06,
      "loss": 0.07262492179870605,
      "step": 6213
    },
    {
      "epoch": 0.840112889324523,
      "grad_norm": 0.0926634669303894,
      "learning_rate": 2.1363255921005685e-06,
      "loss": 0.07397842407226562,
      "step": 6214
    },
    {
      "epoch": 0.840248086120359,
      "grad_norm": 0.08349807560443878,
      "learning_rate": 2.1328064759953853e-06,
      "loss": 0.054540157318115234,
      "step": 6215
    },
    {
      "epoch": 0.8403832829161949,
      "grad_norm": 0.08934856951236725,
      "learning_rate": 2.129290038908504e-06,
      "loss": 0.06272006034851074,
      "step": 6216
    },
    {
      "epoch": 0.8405184797120309,
      "grad_norm": 0.12437047809362411,
      "learning_rate": 2.1257762815720745e-06,
      "loss": 0.054378509521484375,
      "step": 6217
    },
    {
      "epoch": 0.8406536765078667,
      "grad_norm": 0.10967876762151718,
      "learning_rate": 2.122265204717678e-06,
      "loss": 0.07696533203125,
      "step": 6218
    },
    {
      "epoch": 0.8407888733037027,
      "grad_norm": 0.07253522425889969,
      "learning_rate": 2.1187568090763328e-06,
      "loss": 0.062096595764160156,
      "step": 6219
    },
    {
      "epoch": 0.8409240700995386,
      "grad_norm": 0.14087271690368652,
      "learning_rate": 2.1152510953785196e-06,
      "loss": 0.07767057418823242,
      "step": 6220
    },
    {
      "epoch": 0.8410592668953746,
      "grad_norm": 0.10102976113557816,
      "learning_rate": 2.1117480643541304e-06,
      "loss": 0.0737464427947998,
      "step": 6221
    },
    {
      "epoch": 0.8411944636912105,
      "grad_norm": 0.06434816122055054,
      "learning_rate": 2.1082477167325275e-06,
      "loss": 0.05290699005126953,
      "step": 6222
    },
    {
      "epoch": 0.8413296604870465,
      "grad_norm": 0.06449396163225174,
      "learning_rate": 2.1047500532424968e-06,
      "loss": 0.04849052429199219,
      "step": 6223
    },
    {
      "epoch": 0.8414648572828825,
      "grad_norm": 0.06327880918979645,
      "learning_rate": 2.1012550746122705e-06,
      "loss": 0.07477855682373047,
      "step": 6224
    },
    {
      "epoch": 0.8416000540787183,
      "grad_norm": 0.10428311675786972,
      "learning_rate": 2.0977627815695217e-06,
      "loss": 0.07810711860656738,
      "step": 6225
    },
    {
      "epoch": 0.8417352508745543,
      "grad_norm": 0.11010656505823135,
      "learning_rate": 2.094273174841362e-06,
      "loss": 0.07221651077270508,
      "step": 6226
    },
    {
      "epoch": 0.8418704476703902,
      "grad_norm": 0.05010322853922844,
      "learning_rate": 2.0907862551543516e-06,
      "loss": 0.03895598649978638,
      "step": 6227
    },
    {
      "epoch": 0.8420056444662262,
      "grad_norm": 0.09927906095981598,
      "learning_rate": 2.087302023234485e-06,
      "loss": 0.0524754524230957,
      "step": 6228
    },
    {
      "epoch": 0.8421408412620621,
      "grad_norm": 0.09309618920087814,
      "learning_rate": 2.083820479807194e-06,
      "loss": 0.071075439453125,
      "step": 6229
    },
    {
      "epoch": 0.8422760380578981,
      "grad_norm": 0.0662274956703186,
      "learning_rate": 2.0803416255973585e-06,
      "loss": 0.048351287841796875,
      "step": 6230
    },
    {
      "epoch": 0.8424112348537339,
      "grad_norm": 0.030056042596697807,
      "learning_rate": 2.0768654613292887e-06,
      "loss": 0.0256117582321167,
      "step": 6231
    },
    {
      "epoch": 0.8425464316495699,
      "grad_norm": 0.1632716804742813,
      "learning_rate": 2.0733919877267477e-06,
      "loss": 0.06250828504562378,
      "step": 6232
    },
    {
      "epoch": 0.8426816284454058,
      "grad_norm": 0.10840299725532532,
      "learning_rate": 2.0699212055129268e-06,
      "loss": 0.05613970756530762,
      "step": 6233
    },
    {
      "epoch": 0.8428168252412418,
      "grad_norm": 0.15623316168785095,
      "learning_rate": 2.066453115410463e-06,
      "loss": 0.08043575286865234,
      "step": 6234
    },
    {
      "epoch": 0.8429520220370778,
      "grad_norm": 0.07265260815620422,
      "learning_rate": 2.062987718141431e-06,
      "loss": 0.0694270133972168,
      "step": 6235
    },
    {
      "epoch": 0.8430872188329137,
      "grad_norm": 0.08338449150323868,
      "learning_rate": 2.0595250144273423e-06,
      "loss": 0.04454910755157471,
      "step": 6236
    },
    {
      "epoch": 0.8432224156287496,
      "grad_norm": 0.11968682706356049,
      "learning_rate": 2.056065004989155e-06,
      "loss": 0.06780791282653809,
      "step": 6237
    },
    {
      "epoch": 0.8433576124245855,
      "grad_norm": 0.10412868857383728,
      "learning_rate": 2.0526076905472585e-06,
      "loss": 0.045495063066482544,
      "step": 6238
    },
    {
      "epoch": 0.8434928092204215,
      "grad_norm": 0.18360310792922974,
      "learning_rate": 2.0491530718214855e-06,
      "loss": 0.06943023204803467,
      "step": 6239
    },
    {
      "epoch": 0.8436280060162574,
      "grad_norm": 0.10384724289178848,
      "learning_rate": 2.0457011495311045e-06,
      "loss": 0.05608558654785156,
      "step": 6240
    },
    {
      "epoch": 0.8437632028120934,
      "grad_norm": 0.12235202640295029,
      "learning_rate": 2.0422519243948232e-06,
      "loss": 0.04145188629627228,
      "step": 6241
    },
    {
      "epoch": 0.8438983996079293,
      "grad_norm": 0.08759697526693344,
      "learning_rate": 2.0388053971307927e-06,
      "loss": 0.08816301822662354,
      "step": 6242
    },
    {
      "epoch": 0.8440335964037652,
      "grad_norm": 0.07537052780389786,
      "learning_rate": 2.0353615684565956e-06,
      "loss": 0.06231236457824707,
      "step": 6243
    },
    {
      "epoch": 0.8441687931996011,
      "grad_norm": 0.1309131532907486,
      "learning_rate": 2.0319204390892566e-06,
      "loss": 0.09835195541381836,
      "step": 6244
    },
    {
      "epoch": 0.8443039899954371,
      "grad_norm": 0.14323389530181885,
      "learning_rate": 2.0284820097452374e-06,
      "loss": 0.08635735511779785,
      "step": 6245
    },
    {
      "epoch": 0.844439186791273,
      "grad_norm": 0.06707777827978134,
      "learning_rate": 2.02504628114043e-06,
      "loss": 0.06563258171081543,
      "step": 6246
    },
    {
      "epoch": 0.844574383587109,
      "grad_norm": 0.109011709690094,
      "learning_rate": 2.0216132539901865e-06,
      "loss": 0.06651854515075684,
      "step": 6247
    },
    {
      "epoch": 0.844709580382945,
      "grad_norm": 0.09222259372472763,
      "learning_rate": 2.0181829290092663e-06,
      "loss": 0.07601046562194824,
      "step": 6248
    },
    {
      "epoch": 0.8448447771787808,
      "grad_norm": 0.08063499629497528,
      "learning_rate": 2.014755306911891e-06,
      "loss": 0.054407358169555664,
      "step": 6249
    },
    {
      "epoch": 0.8449799739746168,
      "grad_norm": 0.11135166138410568,
      "learning_rate": 2.0113303884117057e-06,
      "loss": 0.06512022018432617,
      "step": 6250
    },
    {
      "epoch": 0.8451151707704527,
      "grad_norm": 0.07940854877233505,
      "learning_rate": 2.0079081742217957e-06,
      "loss": 0.08141922950744629,
      "step": 6251
    },
    {
      "epoch": 0.8452503675662887,
      "grad_norm": 0.06298718601465225,
      "learning_rate": 2.0044886650546915e-06,
      "loss": 0.06259512901306152,
      "step": 6252
    },
    {
      "epoch": 0.8453855643621246,
      "grad_norm": 0.15344993770122528,
      "learning_rate": 2.0010718616223406e-06,
      "loss": 0.061038196086883545,
      "step": 6253
    },
    {
      "epoch": 0.8455207611579606,
      "grad_norm": 0.07286316156387329,
      "learning_rate": 1.9976577646361514e-06,
      "loss": 0.06543761491775513,
      "step": 6254
    },
    {
      "epoch": 0.8456559579537964,
      "grad_norm": 0.09404855221509933,
      "learning_rate": 1.994246374806953e-06,
      "loss": 0.07211899757385254,
      "step": 6255
    },
    {
      "epoch": 0.8457911547496324,
      "grad_norm": 0.13926230370998383,
      "learning_rate": 1.9908376928450128e-06,
      "loss": 0.06549692153930664,
      "step": 6256
    },
    {
      "epoch": 0.8459263515454684,
      "grad_norm": 0.07288368046283722,
      "learning_rate": 1.987431719460039e-06,
      "loss": 0.056478023529052734,
      "step": 6257
    },
    {
      "epoch": 0.8460615483413043,
      "grad_norm": 0.10324602574110031,
      "learning_rate": 1.9840284553611706e-06,
      "loss": 0.04634881019592285,
      "step": 6258
    },
    {
      "epoch": 0.8461967451371403,
      "grad_norm": 0.1018214002251625,
      "learning_rate": 1.980627901256989e-06,
      "loss": 0.065582275390625,
      "step": 6259
    },
    {
      "epoch": 0.8463319419329762,
      "grad_norm": 0.1337098628282547,
      "learning_rate": 1.9772300578555062e-06,
      "loss": 0.09366035461425781,
      "step": 6260
    },
    {
      "epoch": 0.8464671387288121,
      "grad_norm": 0.05762092024087906,
      "learning_rate": 1.973834925864172e-06,
      "loss": 0.059086740016937256,
      "step": 6261
    },
    {
      "epoch": 0.846602335524648,
      "grad_norm": 0.06379427015781403,
      "learning_rate": 1.97044250598987e-06,
      "loss": 0.047879695892333984,
      "step": 6262
    },
    {
      "epoch": 0.846737532320484,
      "grad_norm": 0.1311216652393341,
      "learning_rate": 1.9670527989389177e-06,
      "loss": 0.0744023323059082,
      "step": 6263
    },
    {
      "epoch": 0.8468727291163199,
      "grad_norm": 0.06466415524482727,
      "learning_rate": 1.9636658054170747e-06,
      "loss": 0.05891752243041992,
      "step": 6264
    },
    {
      "epoch": 0.8470079259121559,
      "grad_norm": 0.09935913980007172,
      "learning_rate": 1.960281526129531e-06,
      "loss": 0.053803443908691406,
      "step": 6265
    },
    {
      "epoch": 0.8471431227079919,
      "grad_norm": 0.10352463275194168,
      "learning_rate": 1.9568999617809077e-06,
      "loss": 0.0675363540649414,
      "step": 6266
    },
    {
      "epoch": 0.8472783195038278,
      "grad_norm": 0.06425938755273819,
      "learning_rate": 1.9535211130752676e-06,
      "loss": 0.04772520065307617,
      "step": 6267
    },
    {
      "epoch": 0.8474135162996637,
      "grad_norm": 0.07881362736225128,
      "learning_rate": 1.950144980716101e-06,
      "loss": 0.049092769622802734,
      "step": 6268
    },
    {
      "epoch": 0.8475487130954996,
      "grad_norm": 0.09278977662324905,
      "learning_rate": 1.9467715654063444e-06,
      "loss": 0.0947115421295166,
      "step": 6269
    },
    {
      "epoch": 0.8476839098913356,
      "grad_norm": 0.1020590290427208,
      "learning_rate": 1.9434008678483532e-06,
      "loss": 0.08096301555633545,
      "step": 6270
    },
    {
      "epoch": 0.8478191066871715,
      "grad_norm": 0.07810697704553604,
      "learning_rate": 1.9400328887439295e-06,
      "loss": 0.06392109394073486,
      "step": 6271
    },
    {
      "epoch": 0.8479543034830075,
      "grad_norm": 0.10360649228096008,
      "learning_rate": 1.9366676287943038e-06,
      "loss": 0.06831574440002441,
      "step": 6272
    },
    {
      "epoch": 0.8480895002788434,
      "grad_norm": 0.07456852495670319,
      "learning_rate": 1.9333050887001337e-06,
      "loss": 0.0719747543334961,
      "step": 6273
    },
    {
      "epoch": 0.8482246970746793,
      "grad_norm": 0.08693388104438782,
      "learning_rate": 1.9299452691615293e-06,
      "loss": 0.09015655517578125,
      "step": 6274
    },
    {
      "epoch": 0.8483598938705152,
      "grad_norm": 0.08228273689746857,
      "learning_rate": 1.9265881708780182e-06,
      "loss": 0.07655012607574463,
      "step": 6275
    },
    {
      "epoch": 0.8484950906663512,
      "grad_norm": 0.12034351378679276,
      "learning_rate": 1.9232337945485657e-06,
      "loss": 0.07366728782653809,
      "step": 6276
    },
    {
      "epoch": 0.8486302874621872,
      "grad_norm": 0.07807251065969467,
      "learning_rate": 1.91988214087157e-06,
      "loss": 0.06461310386657715,
      "step": 6277
    },
    {
      "epoch": 0.8487654842580231,
      "grad_norm": 0.11119145900011063,
      "learning_rate": 1.9165332105448613e-06,
      "loss": 0.05874037742614746,
      "step": 6278
    },
    {
      "epoch": 0.8489006810538591,
      "grad_norm": 0.048088837414979935,
      "learning_rate": 1.913187004265715e-06,
      "loss": 0.041738033294677734,
      "step": 6279
    },
    {
      "epoch": 0.8490358778496949,
      "grad_norm": 0.06721097975969315,
      "learning_rate": 1.909843522730814e-06,
      "loss": 0.05828624963760376,
      "step": 6280
    },
    {
      "epoch": 0.8491710746455309,
      "grad_norm": 0.07519472390413284,
      "learning_rate": 1.9065027666363017e-06,
      "loss": 0.054174184799194336,
      "step": 6281
    },
    {
      "epoch": 0.8493062714413668,
      "grad_norm": 0.06168925389647484,
      "learning_rate": 1.903164736677736e-06,
      "loss": 0.045952796936035156,
      "step": 6282
    },
    {
      "epoch": 0.8494414682372028,
      "grad_norm": 0.09956198930740356,
      "learning_rate": 1.8998294335501082e-06,
      "loss": 0.060864925384521484,
      "step": 6283
    },
    {
      "epoch": 0.8495766650330387,
      "grad_norm": 0.08205647021532059,
      "learning_rate": 1.8964968579478592e-06,
      "loss": 0.07142376899719238,
      "step": 6284
    },
    {
      "epoch": 0.8497118618288747,
      "grad_norm": 0.07485213875770569,
      "learning_rate": 1.893167010564834e-06,
      "loss": 0.0645444393157959,
      "step": 6285
    },
    {
      "epoch": 0.8498470586247105,
      "grad_norm": 0.08847309648990631,
      "learning_rate": 1.8898398920943349e-06,
      "loss": 0.07096385955810547,
      "step": 6286
    },
    {
      "epoch": 0.8499822554205465,
      "grad_norm": 0.1267227977514267,
      "learning_rate": 1.886515503229081e-06,
      "loss": 0.06325006484985352,
      "step": 6287
    },
    {
      "epoch": 0.8501174522163825,
      "grad_norm": 0.14188162982463837,
      "learning_rate": 1.8831938446612269e-06,
      "loss": 0.06595194339752197,
      "step": 6288
    },
    {
      "epoch": 0.8502526490122184,
      "grad_norm": 0.07252778112888336,
      "learning_rate": 1.8798749170823676e-06,
      "loss": 0.05689281225204468,
      "step": 6289
    },
    {
      "epoch": 0.8503878458080544,
      "grad_norm": 0.07427436113357544,
      "learning_rate": 1.8765587211835089e-06,
      "loss": 0.05604982376098633,
      "step": 6290
    },
    {
      "epoch": 0.8505230426038903,
      "grad_norm": 0.03143913671374321,
      "learning_rate": 1.8732452576551102e-06,
      "loss": 0.03462386131286621,
      "step": 6291
    },
    {
      "epoch": 0.8506582393997262,
      "grad_norm": 0.12276497483253479,
      "learning_rate": 1.8699345271870493e-06,
      "loss": 0.05977284908294678,
      "step": 6292
    },
    {
      "epoch": 0.8507934361955621,
      "grad_norm": 0.14457136392593384,
      "learning_rate": 1.8666265304686387e-06,
      "loss": 0.0800027847290039,
      "step": 6293
    },
    {
      "epoch": 0.8509286329913981,
      "grad_norm": 0.13843873143196106,
      "learning_rate": 1.8633212681886203e-06,
      "loss": 0.0555652379989624,
      "step": 6294
    },
    {
      "epoch": 0.851063829787234,
      "grad_norm": 0.06576929986476898,
      "learning_rate": 1.8600187410351621e-06,
      "loss": 0.05843544006347656,
      "step": 6295
    },
    {
      "epoch": 0.85119902658307,
      "grad_norm": 0.07701294869184494,
      "learning_rate": 1.8567189496958776e-06,
      "loss": 0.0521770715713501,
      "step": 6296
    },
    {
      "epoch": 0.851334223378906,
      "grad_norm": 0.0984310656785965,
      "learning_rate": 1.853421894857797e-06,
      "loss": 0.06432175636291504,
      "step": 6297
    },
    {
      "epoch": 0.8514694201747418,
      "grad_norm": 0.11677698791027069,
      "learning_rate": 1.8501275772073827e-06,
      "loss": 0.06746864318847656,
      "step": 6298
    },
    {
      "epoch": 0.8516046169705778,
      "grad_norm": 0.04813215509057045,
      "learning_rate": 1.8468359974305315e-06,
      "loss": 0.04853618144989014,
      "step": 6299
    },
    {
      "epoch": 0.8517398137664137,
      "grad_norm": 0.09483543783426285,
      "learning_rate": 1.8435471562125633e-06,
      "loss": 0.07366073131561279,
      "step": 6300
    },
    {
      "epoch": 0.8518750105622497,
      "grad_norm": 0.08742405474185944,
      "learning_rate": 1.8402610542382386e-06,
      "loss": 0.055742740631103516,
      "step": 6301
    },
    {
      "epoch": 0.8520102073580856,
      "grad_norm": 0.09750664979219437,
      "learning_rate": 1.836977692191742e-06,
      "loss": 0.062215328216552734,
      "step": 6302
    },
    {
      "epoch": 0.8521454041539216,
      "grad_norm": 0.05253782495856285,
      "learning_rate": 1.8336970707566781e-06,
      "loss": 0.052700042724609375,
      "step": 6303
    },
    {
      "epoch": 0.8522806009497574,
      "grad_norm": 0.0742323100566864,
      "learning_rate": 1.8304191906160973e-06,
      "loss": 0.03688383102416992,
      "step": 6304
    },
    {
      "epoch": 0.8524157977455934,
      "grad_norm": 0.052266329526901245,
      "learning_rate": 1.8271440524524668e-06,
      "loss": 0.0567936897277832,
      "step": 6305
    },
    {
      "epoch": 0.8525509945414294,
      "grad_norm": 0.05437502637505531,
      "learning_rate": 1.8238716569476949e-06,
      "loss": 0.03851604461669922,
      "step": 6306
    },
    {
      "epoch": 0.8526861913372653,
      "grad_norm": 0.07193050533533096,
      "learning_rate": 1.8206020047831078e-06,
      "loss": 0.07477569580078125,
      "step": 6307
    },
    {
      "epoch": 0.8528213881331013,
      "grad_norm": 0.06978517770767212,
      "learning_rate": 1.8173350966394648e-06,
      "loss": 0.05048036575317383,
      "step": 6308
    },
    {
      "epoch": 0.8529565849289372,
      "grad_norm": 0.029680214822292328,
      "learning_rate": 1.8140709331969513e-06,
      "loss": 0.02732035517692566,
      "step": 6309
    },
    {
      "epoch": 0.8530917817247731,
      "grad_norm": 0.05021276697516441,
      "learning_rate": 1.810809515135184e-06,
      "loss": 0.04528331756591797,
      "step": 6310
    },
    {
      "epoch": 0.853226978520609,
      "grad_norm": 0.09833616763353348,
      "learning_rate": 1.8075508431332111e-06,
      "loss": 0.04064047336578369,
      "step": 6311
    },
    {
      "epoch": 0.853362175316445,
      "grad_norm": 0.07684347778558731,
      "learning_rate": 1.8042949178695034e-06,
      "loss": 0.0632781982421875,
      "step": 6312
    },
    {
      "epoch": 0.8534973721122809,
      "grad_norm": 0.1029210090637207,
      "learning_rate": 1.8010417400219636e-06,
      "loss": 0.06871858239173889,
      "step": 6313
    },
    {
      "epoch": 0.8536325689081169,
      "grad_norm": 0.10687116533517838,
      "learning_rate": 1.7977913102679167e-06,
      "loss": 0.09267044067382812,
      "step": 6314
    },
    {
      "epoch": 0.8537677657039529,
      "grad_norm": 0.09061265736818314,
      "learning_rate": 1.7945436292841193e-06,
      "loss": 0.06876087188720703,
      "step": 6315
    },
    {
      "epoch": 0.8539029624997888,
      "grad_norm": 0.06640708446502686,
      "learning_rate": 1.791298697746766e-06,
      "loss": 0.04834812879562378,
      "step": 6316
    },
    {
      "epoch": 0.8540381592956247,
      "grad_norm": 0.11356284469366074,
      "learning_rate": 1.7880565163314545e-06,
      "loss": 0.07348084449768066,
      "step": 6317
    },
    {
      "epoch": 0.8541733560914606,
      "grad_norm": 0.0762079507112503,
      "learning_rate": 1.784817085713233e-06,
      "loss": 0.0552525520324707,
      "step": 6318
    },
    {
      "epoch": 0.8543085528872966,
      "grad_norm": 0.09714668989181519,
      "learning_rate": 1.7815804065665669e-06,
      "loss": 0.05569267272949219,
      "step": 6319
    },
    {
      "epoch": 0.8544437496831325,
      "grad_norm": 0.0934680625796318,
      "learning_rate": 1.778346479565346e-06,
      "loss": 0.06911420822143555,
      "step": 6320
    },
    {
      "epoch": 0.8545789464789685,
      "grad_norm": 0.061427902430295944,
      "learning_rate": 1.7751153053829011e-06,
      "loss": 0.06157207489013672,
      "step": 6321
    },
    {
      "epoch": 0.8547141432748044,
      "grad_norm": 0.04562776908278465,
      "learning_rate": 1.7718868846919662e-06,
      "loss": 0.03926849365234375,
      "step": 6322
    },
    {
      "epoch": 0.8548493400706403,
      "grad_norm": 0.14218325912952423,
      "learning_rate": 1.7686612181647266e-06,
      "loss": 0.06679058074951172,
      "step": 6323
    },
    {
      "epoch": 0.8549845368664762,
      "grad_norm": 0.08945848792791367,
      "learning_rate": 1.7654383064727802e-06,
      "loss": 0.04896807670593262,
      "step": 6324
    },
    {
      "epoch": 0.8551197336623122,
      "grad_norm": 0.05994424223899841,
      "learning_rate": 1.762218150287152e-06,
      "loss": 0.0772860050201416,
      "step": 6325
    },
    {
      "epoch": 0.8552549304581482,
      "grad_norm": 0.09649288654327393,
      "learning_rate": 1.759000750278299e-06,
      "loss": 0.05411577224731445,
      "step": 6326
    },
    {
      "epoch": 0.8553901272539841,
      "grad_norm": 0.11805935204029083,
      "learning_rate": 1.7557861071160953e-06,
      "loss": 0.04920238256454468,
      "step": 6327
    },
    {
      "epoch": 0.8555253240498201,
      "grad_norm": 0.09620372205972672,
      "learning_rate": 1.7525742214698538e-06,
      "loss": 0.06359004974365234,
      "step": 6328
    },
    {
      "epoch": 0.8556605208456559,
      "grad_norm": 0.05717959627509117,
      "learning_rate": 1.7493650940083045e-06,
      "loss": 0.061995506286621094,
      "step": 6329
    },
    {
      "epoch": 0.8557957176414919,
      "grad_norm": 0.05568467453122139,
      "learning_rate": 1.746158725399603e-06,
      "loss": 0.05309653282165527,
      "step": 6330
    },
    {
      "epoch": 0.8559309144373278,
      "grad_norm": 0.0946793407201767,
      "learning_rate": 1.7429551163113322e-06,
      "loss": 0.042881131172180176,
      "step": 6331
    },
    {
      "epoch": 0.8560661112331638,
      "grad_norm": 0.11048036813735962,
      "learning_rate": 1.7397542674105e-06,
      "loss": 0.05759406089782715,
      "step": 6332
    },
    {
      "epoch": 0.8562013080289997,
      "grad_norm": 0.1003682091832161,
      "learning_rate": 1.7365561793635431e-06,
      "loss": 0.0689697265625,
      "step": 6333
    },
    {
      "epoch": 0.8563365048248357,
      "grad_norm": 0.09098163992166519,
      "learning_rate": 1.7333608528363227e-06,
      "loss": 0.07697200775146484,
      "step": 6334
    },
    {
      "epoch": 0.8564717016206715,
      "grad_norm": 0.08087503165006638,
      "learning_rate": 1.7301682884941128e-06,
      "loss": 0.06306719779968262,
      "step": 6335
    },
    {
      "epoch": 0.8566068984165075,
      "grad_norm": 0.0822829082608223,
      "learning_rate": 1.726978487001632e-06,
      "loss": 0.06019258499145508,
      "step": 6336
    },
    {
      "epoch": 0.8567420952123435,
      "grad_norm": 0.07111406326293945,
      "learning_rate": 1.7237914490230072e-06,
      "loss": 0.06962203979492188,
      "step": 6337
    },
    {
      "epoch": 0.8568772920081794,
      "grad_norm": 0.06223228573799133,
      "learning_rate": 1.7206071752218027e-06,
      "loss": 0.051291823387145996,
      "step": 6338
    },
    {
      "epoch": 0.8570124888040154,
      "grad_norm": 0.035153474658727646,
      "learning_rate": 1.7174256662610032e-06,
      "loss": 0.03708934783935547,
      "step": 6339
    },
    {
      "epoch": 0.8571476855998513,
      "grad_norm": 0.0890684425830841,
      "learning_rate": 1.714246922803004e-06,
      "loss": 0.06419539451599121,
      "step": 6340
    },
    {
      "epoch": 0.8572828823956872,
      "grad_norm": 0.04762372374534607,
      "learning_rate": 1.7110709455096468e-06,
      "loss": 0.038701534271240234,
      "step": 6341
    },
    {
      "epoch": 0.8574180791915231,
      "grad_norm": 0.12369276583194733,
      "learning_rate": 1.7078977350421815e-06,
      "loss": 0.0657840371131897,
      "step": 6342
    },
    {
      "epoch": 0.8575532759873591,
      "grad_norm": 0.026443932205438614,
      "learning_rate": 1.7047272920612926e-06,
      "loss": 0.020498961210250854,
      "step": 6343
    },
    {
      "epoch": 0.857688472783195,
      "grad_norm": 0.041917845606803894,
      "learning_rate": 1.7015596172270841e-06,
      "loss": 0.03945636749267578,
      "step": 6344
    },
    {
      "epoch": 0.857823669579031,
      "grad_norm": 0.041961077600717545,
      "learning_rate": 1.6983947111990717e-06,
      "loss": 0.03614044189453125,
      "step": 6345
    },
    {
      "epoch": 0.857958866374867,
      "grad_norm": 0.06169293448328972,
      "learning_rate": 1.695232574636218e-06,
      "loss": 0.05241173505783081,
      "step": 6346
    },
    {
      "epoch": 0.8580940631707028,
      "grad_norm": 0.1265294998884201,
      "learning_rate": 1.6920732081968882e-06,
      "loss": 0.0695028305053711,
      "step": 6347
    },
    {
      "epoch": 0.8582292599665388,
      "grad_norm": 0.08882235735654831,
      "learning_rate": 1.6889166125388878e-06,
      "loss": 0.05205345153808594,
      "step": 6348
    },
    {
      "epoch": 0.8583644567623747,
      "grad_norm": 0.07851480692625046,
      "learning_rate": 1.6857627883194277e-06,
      "loss": 0.041811466217041016,
      "step": 6349
    },
    {
      "epoch": 0.8584996535582107,
      "grad_norm": 0.10772155225276947,
      "learning_rate": 1.6826117361951577e-06,
      "loss": 0.08319354057312012,
      "step": 6350
    },
    {
      "epoch": 0.8586348503540466,
      "grad_norm": 0.0804390013217926,
      "learning_rate": 1.6794634568221412e-06,
      "loss": 0.06261491775512695,
      "step": 6351
    },
    {
      "epoch": 0.8587700471498826,
      "grad_norm": 0.04552323743700981,
      "learning_rate": 1.676317950855864e-06,
      "loss": 0.04698586463928223,
      "step": 6352
    },
    {
      "epoch": 0.8589052439457184,
      "grad_norm": 0.15698979794979095,
      "learning_rate": 1.6731752189512456e-06,
      "loss": 0.08278107643127441,
      "step": 6353
    },
    {
      "epoch": 0.8590404407415544,
      "grad_norm": 0.05993194878101349,
      "learning_rate": 1.6700352617626092e-06,
      "loss": 0.054193854331970215,
      "step": 6354
    },
    {
      "epoch": 0.8591756375373903,
      "grad_norm": 0.07977376878261566,
      "learning_rate": 1.6668980799437167e-06,
      "loss": 0.04984641075134277,
      "step": 6355
    },
    {
      "epoch": 0.8593108343332263,
      "grad_norm": 0.05026770383119583,
      "learning_rate": 1.6637636741477458e-06,
      "loss": 0.03841531276702881,
      "step": 6356
    },
    {
      "epoch": 0.8594460311290623,
      "grad_norm": 0.1296905130147934,
      "learning_rate": 1.6606320450272943e-06,
      "loss": 0.0711016058921814,
      "step": 6357
    },
    {
      "epoch": 0.8595812279248982,
      "grad_norm": 0.06973902136087418,
      "learning_rate": 1.657503193234386e-06,
      "loss": 0.05986309051513672,
      "step": 6358
    },
    {
      "epoch": 0.8597164247207342,
      "grad_norm": 0.06751244515180588,
      "learning_rate": 1.654377119420461e-06,
      "loss": 0.049954235553741455,
      "step": 6359
    },
    {
      "epoch": 0.85985162151657,
      "grad_norm": 0.17090043425559998,
      "learning_rate": 1.6512538242363889e-06,
      "loss": 0.08775246143341064,
      "step": 6360
    },
    {
      "epoch": 0.859986818312406,
      "grad_norm": 0.14085212349891663,
      "learning_rate": 1.6481333083324563e-06,
      "loss": 0.06416988372802734,
      "step": 6361
    },
    {
      "epoch": 0.8601220151082419,
      "grad_norm": 0.0892537534236908,
      "learning_rate": 1.6450155723583698e-06,
      "loss": 0.07040572166442871,
      "step": 6362
    },
    {
      "epoch": 0.8602572119040779,
      "grad_norm": 0.08346202969551086,
      "learning_rate": 1.6419006169632573e-06,
      "loss": 0.07629728317260742,
      "step": 6363
    },
    {
      "epoch": 0.8603924086999138,
      "grad_norm": 0.11694836616516113,
      "learning_rate": 1.638788442795668e-06,
      "loss": 0.0595853328704834,
      "step": 6364
    },
    {
      "epoch": 0.8605276054957498,
      "grad_norm": 0.07121625542640686,
      "learning_rate": 1.6356790505035785e-06,
      "loss": 0.06560654938220978,
      "step": 6365
    },
    {
      "epoch": 0.8606628022915856,
      "grad_norm": 0.042880576103925705,
      "learning_rate": 1.6325724407343795e-06,
      "loss": 0.03971362113952637,
      "step": 6366
    },
    {
      "epoch": 0.8607979990874216,
      "grad_norm": 0.0610007643699646,
      "learning_rate": 1.6294686141348801e-06,
      "loss": 0.04184424877166748,
      "step": 6367
    },
    {
      "epoch": 0.8609331958832576,
      "grad_norm": 0.1306169331073761,
      "learning_rate": 1.626367571351317e-06,
      "loss": 0.047617197036743164,
      "step": 6368
    },
    {
      "epoch": 0.8610683926790935,
      "grad_norm": 0.0728248730301857,
      "learning_rate": 1.6232693130293386e-06,
      "loss": 0.0598452091217041,
      "step": 6369
    },
    {
      "epoch": 0.8612035894749295,
      "grad_norm": 0.1365279257297516,
      "learning_rate": 1.6201738398140254e-06,
      "loss": 0.08390998840332031,
      "step": 6370
    },
    {
      "epoch": 0.8613387862707654,
      "grad_norm": 0.06906488537788391,
      "learning_rate": 1.6170811523498718e-06,
      "loss": 0.05446815490722656,
      "step": 6371
    },
    {
      "epoch": 0.8614739830666013,
      "grad_norm": 0.052067436277866364,
      "learning_rate": 1.613991251280783e-06,
      "loss": 0.04620945453643799,
      "step": 6372
    },
    {
      "epoch": 0.8616091798624372,
      "grad_norm": 0.05181831866502762,
      "learning_rate": 1.6109041372501028e-06,
      "loss": 0.045969247817993164,
      "step": 6373
    },
    {
      "epoch": 0.8617443766582732,
      "grad_norm": 0.1120200976729393,
      "learning_rate": 1.6078198109005766e-06,
      "loss": 0.06043815612792969,
      "step": 6374
    },
    {
      "epoch": 0.8618795734541091,
      "grad_norm": 0.09435294568538666,
      "learning_rate": 1.6047382728743843e-06,
      "loss": 0.06208467483520508,
      "step": 6375
    },
    {
      "epoch": 0.8620147702499451,
      "grad_norm": 0.0897468775510788,
      "learning_rate": 1.6016595238131176e-06,
      "loss": 0.06417274475097656,
      "step": 6376
    },
    {
      "epoch": 0.8621499670457811,
      "grad_norm": 0.17501172423362732,
      "learning_rate": 1.5985835643577824e-06,
      "loss": 0.08631551265716553,
      "step": 6377
    },
    {
      "epoch": 0.8622851638416169,
      "grad_norm": 0.0604604072868824,
      "learning_rate": 1.5955103951488177e-06,
      "loss": 0.05453217029571533,
      "step": 6378
    },
    {
      "epoch": 0.8624203606374529,
      "grad_norm": 0.07583752274513245,
      "learning_rate": 1.5924400168260666e-06,
      "loss": 0.07736921310424805,
      "step": 6379
    },
    {
      "epoch": 0.8625555574332888,
      "grad_norm": 0.23592795431613922,
      "learning_rate": 1.5893724300288064e-06,
      "loss": 0.07904243469238281,
      "step": 6380
    },
    {
      "epoch": 0.8626907542291248,
      "grad_norm": 0.051680997014045715,
      "learning_rate": 1.5863076353957196e-06,
      "loss": 0.03864121437072754,
      "step": 6381
    },
    {
      "epoch": 0.8628259510249607,
      "grad_norm": 0.07241277396678925,
      "learning_rate": 1.5832456335649104e-06,
      "loss": 0.05378365516662598,
      "step": 6382
    },
    {
      "epoch": 0.8629611478207967,
      "grad_norm": 0.05904010683298111,
      "learning_rate": 1.580186425173909e-06,
      "loss": 0.04015928506851196,
      "step": 6383
    },
    {
      "epoch": 0.8630963446166325,
      "grad_norm": 0.10654803365468979,
      "learning_rate": 1.5771300108596543e-06,
      "loss": 0.06052899360656738,
      "step": 6384
    },
    {
      "epoch": 0.8632315414124685,
      "grad_norm": 0.08163385838270187,
      "learning_rate": 1.5740763912585171e-06,
      "loss": 0.06293773651123047,
      "step": 6385
    },
    {
      "epoch": 0.8633667382083045,
      "grad_norm": 0.06008554995059967,
      "learning_rate": 1.5710255670062657e-06,
      "loss": 0.06254291534423828,
      "step": 6386
    },
    {
      "epoch": 0.8635019350041404,
      "grad_norm": 0.049865126609802246,
      "learning_rate": 1.567977538738105e-06,
      "loss": 0.049284160137176514,
      "step": 6387
    },
    {
      "epoch": 0.8636371317999764,
      "grad_norm": 0.16868528723716736,
      "learning_rate": 1.5649323070886494e-06,
      "loss": 0.05782385170459747,
      "step": 6388
    },
    {
      "epoch": 0.8637723285958123,
      "grad_norm": 0.04562008008360863,
      "learning_rate": 1.5618898726919284e-06,
      "loss": 0.036471009254455566,
      "step": 6389
    },
    {
      "epoch": 0.8639075253916482,
      "grad_norm": 0.041728463023900986,
      "learning_rate": 1.5588502361814032e-06,
      "loss": 0.033078670501708984,
      "step": 6390
    },
    {
      "epoch": 0.8640427221874841,
      "grad_norm": 0.06650735437870026,
      "learning_rate": 1.5558133981899314e-06,
      "loss": 0.05904310941696167,
      "step": 6391
    },
    {
      "epoch": 0.8641779189833201,
      "grad_norm": 0.12570209801197052,
      "learning_rate": 1.5527793593498053e-06,
      "loss": 0.08055615425109863,
      "step": 6392
    },
    {
      "epoch": 0.864313115779156,
      "grad_norm": 0.056440144777297974,
      "learning_rate": 1.5497481202927244e-06,
      "loss": 0.053415536880493164,
      "step": 6393
    },
    {
      "epoch": 0.864448312574992,
      "grad_norm": 0.06267638504505157,
      "learning_rate": 1.5467196816498107e-06,
      "loss": 0.052117109298706055,
      "step": 6394
    },
    {
      "epoch": 0.864583509370828,
      "grad_norm": 0.09015851467847824,
      "learning_rate": 1.5436940440516018e-06,
      "loss": 0.046578384935855865,
      "step": 6395
    },
    {
      "epoch": 0.8647187061666638,
      "grad_norm": 0.041786979883909225,
      "learning_rate": 1.5406712081280484e-06,
      "loss": 0.03513932228088379,
      "step": 6396
    },
    {
      "epoch": 0.8648539029624998,
      "grad_norm": 0.12346234917640686,
      "learning_rate": 1.5376511745085254e-06,
      "loss": 0.05023002624511719,
      "step": 6397
    },
    {
      "epoch": 0.8649890997583357,
      "grad_norm": 0.0752951055765152,
      "learning_rate": 1.5346339438218181e-06,
      "loss": 0.06355094909667969,
      "step": 6398
    },
    {
      "epoch": 0.8651242965541717,
      "grad_norm": 0.0592183843255043,
      "learning_rate": 1.5316195166961295e-06,
      "loss": 0.04720258712768555,
      "step": 6399
    },
    {
      "epoch": 0.8652594933500076,
      "grad_norm": 0.10795643925666809,
      "learning_rate": 1.5286078937590802e-06,
      "loss": 0.07311928272247314,
      "step": 6400
    },
    {
      "epoch": 0.8653946901458436,
      "grad_norm": 0.08860747516155243,
      "learning_rate": 1.5255990756377025e-06,
      "loss": 0.07887554168701172,
      "step": 6401
    },
    {
      "epoch": 0.8655298869416794,
      "grad_norm": 0.0760936439037323,
      "learning_rate": 1.5225930629584534e-06,
      "loss": 0.06229209899902344,
      "step": 6402
    },
    {
      "epoch": 0.8656650837375154,
      "grad_norm": 0.0736214891076088,
      "learning_rate": 1.5195898563472038e-06,
      "loss": 0.06769466400146484,
      "step": 6403
    },
    {
      "epoch": 0.8658002805333513,
      "grad_norm": 0.06015823781490326,
      "learning_rate": 1.5165894564292254e-06,
      "loss": 0.05713927745819092,
      "step": 6404
    },
    {
      "epoch": 0.8659354773291873,
      "grad_norm": 0.07352651655673981,
      "learning_rate": 1.5135918638292269e-06,
      "loss": 0.061180710792541504,
      "step": 6405
    },
    {
      "epoch": 0.8660706741250233,
      "grad_norm": 0.10576921701431274,
      "learning_rate": 1.5105970791713186e-06,
      "loss": 0.06670761108398438,
      "step": 6406
    },
    {
      "epoch": 0.8662058709208592,
      "grad_norm": 0.25882086157798767,
      "learning_rate": 1.5076051030790355e-06,
      "loss": 0.07758545875549316,
      "step": 6407
    },
    {
      "epoch": 0.8663410677166952,
      "grad_norm": 0.0488014779984951,
      "learning_rate": 1.5046159361753226e-06,
      "loss": 0.06250643730163574,
      "step": 6408
    },
    {
      "epoch": 0.866476264512531,
      "grad_norm": 0.07846884429454803,
      "learning_rate": 1.5016295790825336e-06,
      "loss": 0.04989957809448242,
      "step": 6409
    },
    {
      "epoch": 0.866611461308367,
      "grad_norm": 0.13717509806156158,
      "learning_rate": 1.4986460324224493e-06,
      "loss": 0.07029587030410767,
      "step": 6410
    },
    {
      "epoch": 0.8667466581042029,
      "grad_norm": 0.079444520175457,
      "learning_rate": 1.4956652968162582e-06,
      "loss": 0.08028125762939453,
      "step": 6411
    },
    {
      "epoch": 0.8668818549000389,
      "grad_norm": 0.08889906853437424,
      "learning_rate": 1.492687372884567e-06,
      "loss": 0.0611722469329834,
      "step": 6412
    },
    {
      "epoch": 0.8670170516958748,
      "grad_norm": 0.155953511595726,
      "learning_rate": 1.4897122612473978e-06,
      "loss": 0.07591962814331055,
      "step": 6413
    },
    {
      "epoch": 0.8671522484917108,
      "grad_norm": 0.17730297148227692,
      "learning_rate": 1.4867399625241772e-06,
      "loss": 0.06012523174285889,
      "step": 6414
    },
    {
      "epoch": 0.8672874452875466,
      "grad_norm": 0.06775970011949539,
      "learning_rate": 1.4837704773337602e-06,
      "loss": 0.05423814058303833,
      "step": 6415
    },
    {
      "epoch": 0.8674226420833826,
      "grad_norm": 0.0893474817276001,
      "learning_rate": 1.4808038062944036e-06,
      "loss": 0.07478094100952148,
      "step": 6416
    },
    {
      "epoch": 0.8675578388792186,
      "grad_norm": 0.07795003056526184,
      "learning_rate": 1.4778399500237933e-06,
      "loss": 0.040270090103149414,
      "step": 6417
    },
    {
      "epoch": 0.8676930356750545,
      "grad_norm": 0.09488378465175629,
      "learning_rate": 1.4748789091390124e-06,
      "loss": 0.07558703422546387,
      "step": 6418
    },
    {
      "epoch": 0.8678282324708905,
      "grad_norm": 0.07098179310560226,
      "learning_rate": 1.471920684256563e-06,
      "loss": 0.07665181159973145,
      "step": 6419
    },
    {
      "epoch": 0.8679634292667264,
      "grad_norm": 0.08094755560159683,
      "learning_rate": 1.4689652759923721e-06,
      "loss": 0.06972360610961914,
      "step": 6420
    },
    {
      "epoch": 0.8680986260625623,
      "grad_norm": 0.22221647202968597,
      "learning_rate": 1.4660126849617645e-06,
      "loss": 0.0952761173248291,
      "step": 6421
    },
    {
      "epoch": 0.8682338228583982,
      "grad_norm": 0.09138244390487671,
      "learning_rate": 1.4630629117794914e-06,
      "loss": 0.0722116231918335,
      "step": 6422
    },
    {
      "epoch": 0.8683690196542342,
      "grad_norm": 0.06013472378253937,
      "learning_rate": 1.4601159570597033e-06,
      "loss": 0.04598188400268555,
      "step": 6423
    },
    {
      "epoch": 0.8685042164500701,
      "grad_norm": 0.08180181682109833,
      "learning_rate": 1.4571718214159795e-06,
      "loss": 0.059500694274902344,
      "step": 6424
    },
    {
      "epoch": 0.8686394132459061,
      "grad_norm": 0.11993712931871414,
      "learning_rate": 1.454230505461303e-06,
      "loss": 0.0618743896484375,
      "step": 6425
    },
    {
      "epoch": 0.8687746100417421,
      "grad_norm": 0.06103155016899109,
      "learning_rate": 1.4512920098080672e-06,
      "loss": 0.03406292200088501,
      "step": 6426
    },
    {
      "epoch": 0.8689098068375779,
      "grad_norm": 0.05530266836285591,
      "learning_rate": 1.4483563350680878e-06,
      "loss": 0.05251955986022949,
      "step": 6427
    },
    {
      "epoch": 0.8690450036334139,
      "grad_norm": 0.07514005899429321,
      "learning_rate": 1.4454234818525824e-06,
      "loss": 0.05506563186645508,
      "step": 6428
    },
    {
      "epoch": 0.8691802004292498,
      "grad_norm": 0.04610864445567131,
      "learning_rate": 1.4424934507721926e-06,
      "loss": 0.031284451484680176,
      "step": 6429
    },
    {
      "epoch": 0.8693153972250858,
      "grad_norm": 0.1004101112484932,
      "learning_rate": 1.4395662424369622e-06,
      "loss": 0.057344913482666016,
      "step": 6430
    },
    {
      "epoch": 0.8694505940209217,
      "grad_norm": 0.0901050791144371,
      "learning_rate": 1.436641857456355e-06,
      "loss": 0.08454418182373047,
      "step": 6431
    },
    {
      "epoch": 0.8695857908167577,
      "grad_norm": 0.12006206065416336,
      "learning_rate": 1.4337202964392409e-06,
      "loss": 0.06084930896759033,
      "step": 6432
    },
    {
      "epoch": 0.8697209876125935,
      "grad_norm": 0.038087278604507446,
      "learning_rate": 1.4308015599939033e-06,
      "loss": 0.024875640869140625,
      "step": 6433
    },
    {
      "epoch": 0.8698561844084295,
      "grad_norm": 0.08178307116031647,
      "learning_rate": 1.4278856487280428e-06,
      "loss": 0.07812666893005371,
      "step": 6434
    },
    {
      "epoch": 0.8699913812042654,
      "grad_norm": 0.059284958988428116,
      "learning_rate": 1.4249725632487653e-06,
      "loss": 0.0552448034286499,
      "step": 6435
    },
    {
      "epoch": 0.8701265780001014,
      "grad_norm": 0.082722507417202,
      "learning_rate": 1.4220623041625924e-06,
      "loss": 0.05972862243652344,
      "step": 6436
    },
    {
      "epoch": 0.8702617747959374,
      "grad_norm": 0.11636929214000702,
      "learning_rate": 1.4191548720754527e-06,
      "loss": 0.0633622407913208,
      "step": 6437
    },
    {
      "epoch": 0.8703969715917733,
      "grad_norm": 0.060393013060092926,
      "learning_rate": 1.4162502675926887e-06,
      "loss": 0.05115801841020584,
      "step": 6438
    },
    {
      "epoch": 0.8705321683876092,
      "grad_norm": 0.06341095268726349,
      "learning_rate": 1.4133484913190596e-06,
      "loss": 0.05683088302612305,
      "step": 6439
    },
    {
      "epoch": 0.8706673651834451,
      "grad_norm": 0.059537675231695175,
      "learning_rate": 1.4104495438587295e-06,
      "loss": 0.035642027854919434,
      "step": 6440
    },
    {
      "epoch": 0.8708025619792811,
      "grad_norm": 0.08261965215206146,
      "learning_rate": 1.4075534258152667e-06,
      "loss": 0.05670452117919922,
      "step": 6441
    },
    {
      "epoch": 0.870937758775117,
      "grad_norm": 0.0936427116394043,
      "learning_rate": 1.4046601377916673e-06,
      "loss": 0.08454561233520508,
      "step": 6442
    },
    {
      "epoch": 0.871072955570953,
      "grad_norm": 0.18029776215553284,
      "learning_rate": 1.4017696803903246e-06,
      "loss": 0.07367730140686035,
      "step": 6443
    },
    {
      "epoch": 0.871208152366789,
      "grad_norm": 0.13922595977783203,
      "learning_rate": 1.3988820542130504e-06,
      "loss": 0.08683896064758301,
      "step": 6444
    },
    {
      "epoch": 0.8713433491626248,
      "grad_norm": 0.12686164677143097,
      "learning_rate": 1.395997259861067e-06,
      "loss": 0.07768511772155762,
      "step": 6445
    },
    {
      "epoch": 0.8714785459584607,
      "grad_norm": 0.14138749241828918,
      "learning_rate": 1.3931152979349926e-06,
      "loss": 0.07920169830322266,
      "step": 6446
    },
    {
      "epoch": 0.8716137427542967,
      "grad_norm": 0.13059298694133759,
      "learning_rate": 1.3902361690348769e-06,
      "loss": 0.06594228744506836,
      "step": 6447
    },
    {
      "epoch": 0.8717489395501327,
      "grad_norm": 0.05394219607114792,
      "learning_rate": 1.3873598737601639e-06,
      "loss": 0.042543888092041016,
      "step": 6448
    },
    {
      "epoch": 0.8718841363459686,
      "grad_norm": 0.055598847568035126,
      "learning_rate": 1.3844864127097229e-06,
      "loss": 0.05002450942993164,
      "step": 6449
    },
    {
      "epoch": 0.8720193331418046,
      "grad_norm": 0.12051721662282944,
      "learning_rate": 1.3816157864818151e-06,
      "loss": 0.06903493404388428,
      "step": 6450
    },
    {
      "epoch": 0.8721545299376405,
      "grad_norm": 0.040695756673812866,
      "learning_rate": 1.3787479956741194e-06,
      "loss": 0.049883127212524414,
      "step": 6451
    },
    {
      "epoch": 0.8722897267334764,
      "grad_norm": 0.07189293950796127,
      "learning_rate": 1.3758830408837314e-06,
      "loss": 0.06341159343719482,
      "step": 6452
    },
    {
      "epoch": 0.8724249235293123,
      "grad_norm": 0.0494973286986351,
      "learning_rate": 1.3730209227071439e-06,
      "loss": 0.04201161861419678,
      "step": 6453
    },
    {
      "epoch": 0.8725601203251483,
      "grad_norm": 0.13717636466026306,
      "learning_rate": 1.3701616417402734e-06,
      "loss": 0.09896183013916016,
      "step": 6454
    },
    {
      "epoch": 0.8726953171209842,
      "grad_norm": 0.06824973225593567,
      "learning_rate": 1.367305198578429e-06,
      "loss": 0.0618937611579895,
      "step": 6455
    },
    {
      "epoch": 0.8728305139168202,
      "grad_norm": 0.07186117023229599,
      "learning_rate": 1.36445159381634e-06,
      "loss": 0.05238616466522217,
      "step": 6456
    },
    {
      "epoch": 0.8729657107126562,
      "grad_norm": 0.16033010184764862,
      "learning_rate": 1.361600828048144e-06,
      "loss": 0.07703065872192383,
      "step": 6457
    },
    {
      "epoch": 0.873100907508492,
      "grad_norm": 0.06869401782751083,
      "learning_rate": 1.3587529018673816e-06,
      "loss": 0.05274546146392822,
      "step": 6458
    },
    {
      "epoch": 0.873236104304328,
      "grad_norm": 0.1351899653673172,
      "learning_rate": 1.3559078158670152e-06,
      "loss": 0.07019662857055664,
      "step": 6459
    },
    {
      "epoch": 0.8733713011001639,
      "grad_norm": 0.12785902619361877,
      "learning_rate": 1.353065570639394e-06,
      "loss": 0.06765604019165039,
      "step": 6460
    },
    {
      "epoch": 0.8735064978959999,
      "grad_norm": 0.094843789935112,
      "learning_rate": 1.3502261667763e-06,
      "loss": 0.04578077793121338,
      "step": 6461
    },
    {
      "epoch": 0.8736416946918358,
      "grad_norm": 0.0506066232919693,
      "learning_rate": 1.3473896048689067e-06,
      "loss": 0.04484367370605469,
      "step": 6462
    },
    {
      "epoch": 0.8737768914876718,
      "grad_norm": 0.05722508579492569,
      "learning_rate": 1.3445558855078017e-06,
      "loss": 0.04959213733673096,
      "step": 6463
    },
    {
      "epoch": 0.8739120882835076,
      "grad_norm": 0.07660440355539322,
      "learning_rate": 1.3417250092829814e-06,
      "loss": 0.0594487190246582,
      "step": 6464
    },
    {
      "epoch": 0.8740472850793436,
      "grad_norm": 0.09044113755226135,
      "learning_rate": 1.338896976783846e-06,
      "loss": 0.058452486991882324,
      "step": 6465
    },
    {
      "epoch": 0.8741824818751796,
      "grad_norm": 0.1256711184978485,
      "learning_rate": 1.336071788599213e-06,
      "loss": 0.06017589569091797,
      "step": 6466
    },
    {
      "epoch": 0.8743176786710155,
      "grad_norm": 0.09011678397655487,
      "learning_rate": 1.3332494453172982e-06,
      "loss": 0.08049631118774414,
      "step": 6467
    },
    {
      "epoch": 0.8744528754668515,
      "grad_norm": 0.07490069419145584,
      "learning_rate": 1.3304299475257287e-06,
      "loss": 0.06297492980957031,
      "step": 6468
    },
    {
      "epoch": 0.8745880722626874,
      "grad_norm": 0.09171540290117264,
      "learning_rate": 1.3276132958115394e-06,
      "loss": 0.058758556842803955,
      "step": 6469
    },
    {
      "epoch": 0.8747232690585233,
      "grad_norm": 0.05838421359658241,
      "learning_rate": 1.32479949076117e-06,
      "loss": 0.05766165256500244,
      "step": 6470
    },
    {
      "epoch": 0.8748584658543592,
      "grad_norm": 0.11151295155286789,
      "learning_rate": 1.3219885329604747e-06,
      "loss": 0.05652773380279541,
      "step": 6471
    },
    {
      "epoch": 0.8749936626501952,
      "grad_norm": 0.060704734176397324,
      "learning_rate": 1.319180422994709e-06,
      "loss": 0.03921103477478027,
      "step": 6472
    },
    {
      "epoch": 0.8751288594460311,
      "grad_norm": 0.07868202775716782,
      "learning_rate": 1.3163751614485287e-06,
      "loss": 0.07724672555923462,
      "step": 6473
    },
    {
      "epoch": 0.8752640562418671,
      "grad_norm": 0.07434491068124771,
      "learning_rate": 1.3135727489060113e-06,
      "loss": 0.058353424072265625,
      "step": 6474
    },
    {
      "epoch": 0.875399253037703,
      "grad_norm": 0.09557139873504639,
      "learning_rate": 1.3107731859506317e-06,
      "loss": 0.07402229309082031,
      "step": 6475
    },
    {
      "epoch": 0.8755344498335389,
      "grad_norm": 0.09994301199913025,
      "learning_rate": 1.3079764731652772e-06,
      "loss": 0.05762887001037598,
      "step": 6476
    },
    {
      "epoch": 0.8756696466293749,
      "grad_norm": 0.07793942093849182,
      "learning_rate": 1.3051826111322368e-06,
      "loss": 0.07263338565826416,
      "step": 6477
    },
    {
      "epoch": 0.8758048434252108,
      "grad_norm": 0.07906635850667953,
      "learning_rate": 1.3023916004332021e-06,
      "loss": 0.06643915176391602,
      "step": 6478
    },
    {
      "epoch": 0.8759400402210468,
      "grad_norm": 0.09959713369607925,
      "learning_rate": 1.2996034416492847e-06,
      "loss": 0.06354546546936035,
      "step": 6479
    },
    {
      "epoch": 0.8760752370168827,
      "grad_norm": 0.055056583136320114,
      "learning_rate": 1.2968181353609854e-06,
      "loss": 0.05894780158996582,
      "step": 6480
    },
    {
      "epoch": 0.8762104338127187,
      "grad_norm": 0.07187693566083908,
      "learning_rate": 1.2940356821482285e-06,
      "loss": 0.04204225540161133,
      "step": 6481
    },
    {
      "epoch": 0.8763456306085545,
      "grad_norm": 0.06411078572273254,
      "learning_rate": 1.291256082590334e-06,
      "loss": 0.05649673938751221,
      "step": 6482
    },
    {
      "epoch": 0.8764808274043905,
      "grad_norm": 0.14113785326480865,
      "learning_rate": 1.2884793372660208e-06,
      "loss": 0.07549786567687988,
      "step": 6483
    },
    {
      "epoch": 0.8766160242002264,
      "grad_norm": 0.05166800692677498,
      "learning_rate": 1.285705446753433e-06,
      "loss": 0.0476759672164917,
      "step": 6484
    },
    {
      "epoch": 0.8767512209960624,
      "grad_norm": 0.07620629668235779,
      "learning_rate": 1.2829344116301e-06,
      "loss": 0.06698322296142578,
      "step": 6485
    },
    {
      "epoch": 0.8768864177918984,
      "grad_norm": 0.11518441140651703,
      "learning_rate": 1.2801662324729774e-06,
      "loss": 0.06765782833099365,
      "step": 6486
    },
    {
      "epoch": 0.8770216145877343,
      "grad_norm": 0.06900651752948761,
      "learning_rate": 1.2774009098584055e-06,
      "loss": 0.05993431806564331,
      "step": 6487
    },
    {
      "epoch": 0.8771568113835702,
      "grad_norm": 0.09508582204580307,
      "learning_rate": 1.274638444362139e-06,
      "loss": 0.08182764053344727,
      "step": 6488
    },
    {
      "epoch": 0.8772920081794061,
      "grad_norm": 0.07269113510847092,
      "learning_rate": 1.2718788365593443e-06,
      "loss": 0.06250470876693726,
      "step": 6489
    },
    {
      "epoch": 0.8774272049752421,
      "grad_norm": 0.05016093701124191,
      "learning_rate": 1.26912208702458e-06,
      "loss": 0.04906296730041504,
      "step": 6490
    },
    {
      "epoch": 0.877562401771078,
      "grad_norm": 0.12695634365081787,
      "learning_rate": 1.2663681963318242e-06,
      "loss": 0.09316396713256836,
      "step": 6491
    },
    {
      "epoch": 0.877697598566914,
      "grad_norm": 0.06621631234884262,
      "learning_rate": 1.2636171650544443e-06,
      "loss": 0.0663766860961914,
      "step": 6492
    },
    {
      "epoch": 0.8778327953627499,
      "grad_norm": 0.11906833946704865,
      "learning_rate": 1.260868993765219e-06,
      "loss": 0.08048009872436523,
      "step": 6493
    },
    {
      "epoch": 0.8779679921585858,
      "grad_norm": 0.10896174609661102,
      "learning_rate": 1.258123683036339e-06,
      "loss": 0.06722664833068848,
      "step": 6494
    },
    {
      "epoch": 0.8781031889544217,
      "grad_norm": 0.0913810133934021,
      "learning_rate": 1.2553812334393872e-06,
      "loss": 0.0606541633605957,
      "step": 6495
    },
    {
      "epoch": 0.8782383857502577,
      "grad_norm": 0.0818769559264183,
      "learning_rate": 1.2526416455453582e-06,
      "loss": 0.058750152587890625,
      "step": 6496
    },
    {
      "epoch": 0.8783735825460937,
      "grad_norm": 0.1103515625,
      "learning_rate": 1.249904919924646e-06,
      "loss": 0.06089746952056885,
      "step": 6497
    },
    {
      "epoch": 0.8785087793419296,
      "grad_norm": 0.06198696419596672,
      "learning_rate": 1.2471710571470579e-06,
      "loss": 0.044019997119903564,
      "step": 6498
    },
    {
      "epoch": 0.8786439761377656,
      "grad_norm": 0.0762113630771637,
      "learning_rate": 1.2444400577817922e-06,
      "loss": 0.0538862943649292,
      "step": 6499
    },
    {
      "epoch": 0.8787791729336015,
      "grad_norm": 0.08310767263174057,
      "learning_rate": 1.2417119223974621e-06,
      "loss": 0.042129844427108765,
      "step": 6500
    },
    {
      "epoch": 0.8789143697294374,
      "grad_norm": 0.1790885329246521,
      "learning_rate": 1.2389866515620768e-06,
      "loss": 0.07005803287029266,
      "step": 6501
    },
    {
      "epoch": 0.8790495665252733,
      "grad_norm": 0.1959482580423355,
      "learning_rate": 1.2362642458430505e-06,
      "loss": 0.10361224412918091,
      "step": 6502
    },
    {
      "epoch": 0.8791847633211093,
      "grad_norm": 0.08510959893465042,
      "learning_rate": 1.2335447058072103e-06,
      "loss": 0.05635261535644531,
      "step": 6503
    },
    {
      "epoch": 0.8793199601169452,
      "grad_norm": 0.04887179285287857,
      "learning_rate": 1.230828032020771e-06,
      "loss": 0.03641188144683838,
      "step": 6504
    },
    {
      "epoch": 0.8794551569127812,
      "grad_norm": 0.07791148126125336,
      "learning_rate": 1.2281142250493638e-06,
      "loss": 0.05992138385772705,
      "step": 6505
    },
    {
      "epoch": 0.8795903537086172,
      "grad_norm": 0.07690058648586273,
      "learning_rate": 1.225403285458015e-06,
      "loss": 0.07168054580688477,
      "step": 6506
    },
    {
      "epoch": 0.879725550504453,
      "grad_norm": 0.1491979956626892,
      "learning_rate": 1.2226952138111546e-06,
      "loss": 0.07484722137451172,
      "step": 6507
    },
    {
      "epoch": 0.879860747300289,
      "grad_norm": 0.13239221274852753,
      "learning_rate": 1.219990010672622e-06,
      "loss": 0.06789350509643555,
      "step": 6508
    },
    {
      "epoch": 0.8799959440961249,
      "grad_norm": 0.13040296733379364,
      "learning_rate": 1.2172876766056562e-06,
      "loss": 0.05312696099281311,
      "step": 6509
    },
    {
      "epoch": 0.8801311408919609,
      "grad_norm": 0.12148870527744293,
      "learning_rate": 1.2145882121728906e-06,
      "loss": 0.059514760971069336,
      "step": 6510
    },
    {
      "epoch": 0.8802663376877968,
      "grad_norm": 0.1337987184524536,
      "learning_rate": 1.2118916179363727e-06,
      "loss": 0.08131563663482666,
      "step": 6511
    },
    {
      "epoch": 0.8804015344836328,
      "grad_norm": 0.0752175822854042,
      "learning_rate": 1.209197894457546e-06,
      "loss": 0.046537160873413086,
      "step": 6512
    },
    {
      "epoch": 0.8805367312794686,
      "grad_norm": 0.11400206387042999,
      "learning_rate": 1.2065070422972606e-06,
      "loss": 0.05353587865829468,
      "step": 6513
    },
    {
      "epoch": 0.8806719280753046,
      "grad_norm": 0.10037486255168915,
      "learning_rate": 1.2038190620157685e-06,
      "loss": 0.07340002059936523,
      "step": 6514
    },
    {
      "epoch": 0.8808071248711405,
      "grad_norm": 0.061386581510305405,
      "learning_rate": 1.2011339541727117e-06,
      "loss": 0.04145979881286621,
      "step": 6515
    },
    {
      "epoch": 0.8809423216669765,
      "grad_norm": 0.05132085084915161,
      "learning_rate": 1.198451719327155e-06,
      "loss": 0.049624860286712646,
      "step": 6516
    },
    {
      "epoch": 0.8810775184628125,
      "grad_norm": 0.06391947716474533,
      "learning_rate": 1.1957723580375447e-06,
      "loss": 0.047849178314208984,
      "step": 6517
    },
    {
      "epoch": 0.8812127152586484,
      "grad_norm": 0.04655515402555466,
      "learning_rate": 1.193095870861748e-06,
      "loss": 0.04912972450256348,
      "step": 6518
    },
    {
      "epoch": 0.8813479120544843,
      "grad_norm": 0.10650839656591415,
      "learning_rate": 1.1904222583570156e-06,
      "loss": 0.04141688346862793,
      "step": 6519
    },
    {
      "epoch": 0.8814831088503202,
      "grad_norm": 0.10529980808496475,
      "learning_rate": 1.1877515210800077e-06,
      "loss": 0.0585789680480957,
      "step": 6520
    },
    {
      "epoch": 0.8816183056461562,
      "grad_norm": 0.09063461422920227,
      "learning_rate": 1.1850836595867925e-06,
      "loss": 0.04475545883178711,
      "step": 6521
    },
    {
      "epoch": 0.8817535024419921,
      "grad_norm": 0.11571851372718811,
      "learning_rate": 1.1824186744328259e-06,
      "loss": 0.08102607727050781,
      "step": 6522
    },
    {
      "epoch": 0.8818886992378281,
      "grad_norm": 0.12886452674865723,
      "learning_rate": 1.179756566172982e-06,
      "loss": 0.07580780982971191,
      "step": 6523
    },
    {
      "epoch": 0.882023896033664,
      "grad_norm": 0.05057654529809952,
      "learning_rate": 1.177097335361516e-06,
      "loss": 0.03611111640930176,
      "step": 6524
    },
    {
      "epoch": 0.8821590928294999,
      "grad_norm": 0.08624257147312164,
      "learning_rate": 1.1744409825520969e-06,
      "loss": 0.07086181640625,
      "step": 6525
    },
    {
      "epoch": 0.8822942896253358,
      "grad_norm": 0.08988473564386368,
      "learning_rate": 1.171787508297792e-06,
      "loss": 0.07788324356079102,
      "step": 6526
    },
    {
      "epoch": 0.8824294864211718,
      "grad_norm": 0.07482783496379852,
      "learning_rate": 1.1691369131510676e-06,
      "loss": 0.039360642433166504,
      "step": 6527
    },
    {
      "epoch": 0.8825646832170078,
      "grad_norm": 0.03648466616868973,
      "learning_rate": 1.1664891976637992e-06,
      "loss": 0.02965986728668213,
      "step": 6528
    },
    {
      "epoch": 0.8826998800128437,
      "grad_norm": 0.0596202127635479,
      "learning_rate": 1.1638443623872442e-06,
      "loss": 0.05133655667304993,
      "step": 6529
    },
    {
      "epoch": 0.8828350768086797,
      "grad_norm": 0.11942271143198013,
      "learning_rate": 1.1612024078720752e-06,
      "loss": 0.054498091340065,
      "step": 6530
    },
    {
      "epoch": 0.8829702736045155,
      "grad_norm": 0.05281819775700569,
      "learning_rate": 1.1585633346683655e-06,
      "loss": 0.05314207077026367,
      "step": 6531
    },
    {
      "epoch": 0.8831054704003515,
      "grad_norm": 0.048382170498371124,
      "learning_rate": 1.155927143325579e-06,
      "loss": 0.04397740960121155,
      "step": 6532
    },
    {
      "epoch": 0.8832406671961874,
      "grad_norm": 0.05771726742386818,
      "learning_rate": 1.1532938343925887e-06,
      "loss": 0.049910902976989746,
      "step": 6533
    },
    {
      "epoch": 0.8833758639920234,
      "grad_norm": 0.04804328456521034,
      "learning_rate": 1.1506634084176587e-06,
      "loss": 0.04077553749084473,
      "step": 6534
    },
    {
      "epoch": 0.8835110607878593,
      "grad_norm": 0.0726543515920639,
      "learning_rate": 1.148035865948463e-06,
      "loss": 0.06297606229782104,
      "step": 6535
    },
    {
      "epoch": 0.8836462575836953,
      "grad_norm": 0.11173998564481735,
      "learning_rate": 1.1454112075320688e-06,
      "loss": 0.05508708953857422,
      "step": 6536
    },
    {
      "epoch": 0.8837814543795312,
      "grad_norm": 0.10670069605112076,
      "learning_rate": 1.1427894337149426e-06,
      "loss": 0.09070062637329102,
      "step": 6537
    },
    {
      "epoch": 0.8839166511753671,
      "grad_norm": 0.0660172700881958,
      "learning_rate": 1.1401705450429506e-06,
      "loss": 0.07214498519897461,
      "step": 6538
    },
    {
      "epoch": 0.8840518479712031,
      "grad_norm": 0.022525470703840256,
      "learning_rate": 1.1375545420613586e-06,
      "loss": 0.019761621952056885,
      "step": 6539
    },
    {
      "epoch": 0.884187044767039,
      "grad_norm": 0.07742978632450104,
      "learning_rate": 1.1349414253148377e-06,
      "loss": 0.051624417304992676,
      "step": 6540
    },
    {
      "epoch": 0.884322241562875,
      "grad_norm": 0.08196750283241272,
      "learning_rate": 1.1323311953474524e-06,
      "loss": 0.07137250900268555,
      "step": 6541
    },
    {
      "epoch": 0.8844574383587109,
      "grad_norm": 0.07884689420461655,
      "learning_rate": 1.1297238527026582e-06,
      "loss": 0.0605010986328125,
      "step": 6542
    },
    {
      "epoch": 0.8845926351545469,
      "grad_norm": 0.05428411066532135,
      "learning_rate": 1.1271193979233258e-06,
      "loss": 0.056774139404296875,
      "step": 6543
    },
    {
      "epoch": 0.8847278319503827,
      "grad_norm": 0.04940522834658623,
      "learning_rate": 1.1245178315517113e-06,
      "loss": 0.05476069450378418,
      "step": 6544
    },
    {
      "epoch": 0.8848630287462187,
      "grad_norm": 0.10771366953849792,
      "learning_rate": 1.1219191541294798e-06,
      "loss": 0.07926285266876221,
      "step": 6545
    },
    {
      "epoch": 0.8849982255420547,
      "grad_norm": 0.12758027017116547,
      "learning_rate": 1.1193233661976887e-06,
      "loss": 0.0677649974822998,
      "step": 6546
    },
    {
      "epoch": 0.8851334223378906,
      "grad_norm": 0.15208247303962708,
      "learning_rate": 1.1167304682967904e-06,
      "loss": 0.06079322099685669,
      "step": 6547
    },
    {
      "epoch": 0.8852686191337266,
      "grad_norm": 0.05892002582550049,
      "learning_rate": 1.114140460966645e-06,
      "loss": 0.05365419387817383,
      "step": 6548
    },
    {
      "epoch": 0.8854038159295625,
      "grad_norm": 0.07697014510631561,
      "learning_rate": 1.111553344746501e-06,
      "loss": 0.06839609146118164,
      "step": 6549
    },
    {
      "epoch": 0.8855390127253984,
      "grad_norm": 0.0700211301445961,
      "learning_rate": 1.1089691201750174e-06,
      "loss": 0.045278072357177734,
      "step": 6550
    },
    {
      "epoch": 0.8856742095212343,
      "grad_norm": 0.1417781412601471,
      "learning_rate": 1.106387787790239e-06,
      "loss": 0.08942627906799316,
      "step": 6551
    },
    {
      "epoch": 0.8858094063170703,
      "grad_norm": 0.1318359375,
      "learning_rate": 1.1038093481296091e-06,
      "loss": 0.09242439270019531,
      "step": 6552
    },
    {
      "epoch": 0.8859446031129062,
      "grad_norm": 0.055270396173000336,
      "learning_rate": 1.10123380172998e-06,
      "loss": 0.04508829116821289,
      "step": 6553
    },
    {
      "epoch": 0.8860797999087422,
      "grad_norm": 0.08247637003660202,
      "learning_rate": 1.098661149127586e-06,
      "loss": 0.05121803283691406,
      "step": 6554
    },
    {
      "epoch": 0.8862149967045782,
      "grad_norm": 0.05553044006228447,
      "learning_rate": 1.0960913908580788e-06,
      "loss": 0.050912559032440186,
      "step": 6555
    },
    {
      "epoch": 0.886350193500414,
      "grad_norm": 0.04795118793845177,
      "learning_rate": 1.0935245274564852e-06,
      "loss": 0.04163360595703125,
      "step": 6556
    },
    {
      "epoch": 0.88648539029625,
      "grad_norm": 0.12914790213108063,
      "learning_rate": 1.0909605594572413e-06,
      "loss": 0.050554513931274414,
      "step": 6557
    },
    {
      "epoch": 0.8866205870920859,
      "grad_norm": 0.1704612374305725,
      "learning_rate": 1.0883994873941816e-06,
      "loss": 0.09965455532073975,
      "step": 6558
    },
    {
      "epoch": 0.8867557838879219,
      "grad_norm": 0.06700973957777023,
      "learning_rate": 1.0858413118005345e-06,
      "loss": 0.0470738410949707,
      "step": 6559
    },
    {
      "epoch": 0.8868909806837578,
      "grad_norm": 0.19251349568367004,
      "learning_rate": 1.0832860332089288e-06,
      "loss": 0.06895112991333008,
      "step": 6560
    },
    {
      "epoch": 0.8870261774795938,
      "grad_norm": 0.10217111557722092,
      "learning_rate": 1.0807336521513828e-06,
      "loss": 0.043393611907958984,
      "step": 6561
    },
    {
      "epoch": 0.8871613742754296,
      "grad_norm": 0.07085638493299484,
      "learning_rate": 1.0781841691593142e-06,
      "loss": 0.047635555267333984,
      "step": 6562
    },
    {
      "epoch": 0.8872965710712656,
      "grad_norm": 0.11534271389245987,
      "learning_rate": 1.0756375847635435e-06,
      "loss": 0.08743667602539062,
      "step": 6563
    },
    {
      "epoch": 0.8874317678671015,
      "grad_norm": 0.042759209871292114,
      "learning_rate": 1.0730938994942818e-06,
      "loss": 0.04030489921569824,
      "step": 6564
    },
    {
      "epoch": 0.8875669646629375,
      "grad_norm": 0.07410503923892975,
      "learning_rate": 1.070553113881137e-06,
      "loss": 0.06687331199645996,
      "step": 6565
    },
    {
      "epoch": 0.8877021614587735,
      "grad_norm": 0.06275312602519989,
      "learning_rate": 1.0680152284531158e-06,
      "loss": 0.040627479553222656,
      "step": 6566
    },
    {
      "epoch": 0.8878373582546094,
      "grad_norm": 0.0754169449210167,
      "learning_rate": 1.0654802437386157e-06,
      "loss": 0.045983314514160156,
      "step": 6567
    },
    {
      "epoch": 0.8879725550504453,
      "grad_norm": 0.06962469220161438,
      "learning_rate": 1.062948160265438e-06,
      "loss": 0.057654619216918945,
      "step": 6568
    },
    {
      "epoch": 0.8881077518462812,
      "grad_norm": 0.0985807329416275,
      "learning_rate": 1.0604189785607772e-06,
      "loss": 0.06784486770629883,
      "step": 6569
    },
    {
      "epoch": 0.8882429486421172,
      "grad_norm": 0.13257592916488647,
      "learning_rate": 1.0578926991512171e-06,
      "loss": 0.07983052730560303,
      "step": 6570
    },
    {
      "epoch": 0.8883781454379531,
      "grad_norm": 0.08442649245262146,
      "learning_rate": 1.0553693225627458e-06,
      "loss": 0.05161643028259277,
      "step": 6571
    },
    {
      "epoch": 0.8885133422337891,
      "grad_norm": 0.09726759046316147,
      "learning_rate": 1.0528488493207444e-06,
      "loss": 0.06853289902210236,
      "step": 6572
    },
    {
      "epoch": 0.888648539029625,
      "grad_norm": 0.08419055491685867,
      "learning_rate": 1.0503312799499898e-06,
      "loss": 0.08247542381286621,
      "step": 6573
    },
    {
      "epoch": 0.8887837358254609,
      "grad_norm": 0.06069283187389374,
      "learning_rate": 1.0478166149746476e-06,
      "loss": 0.052207231521606445,
      "step": 6574
    },
    {
      "epoch": 0.8889189326212968,
      "grad_norm": 0.05220550671219826,
      "learning_rate": 1.0453048549182892e-06,
      "loss": 0.040143370628356934,
      "step": 6575
    },
    {
      "epoch": 0.8890541294171328,
      "grad_norm": 0.12548846006393433,
      "learning_rate": 1.0427960003038744e-06,
      "loss": 0.05970048904418945,
      "step": 6576
    },
    {
      "epoch": 0.8891893262129688,
      "grad_norm": 0.10632126778364182,
      "learning_rate": 1.040290051653764e-06,
      "loss": 0.06353044509887695,
      "step": 6577
    },
    {
      "epoch": 0.8893245230088047,
      "grad_norm": 0.05560656636953354,
      "learning_rate": 1.0377870094897085e-06,
      "loss": 0.04810154438018799,
      "step": 6578
    },
    {
      "epoch": 0.8894597198046407,
      "grad_norm": 0.07954412698745728,
      "learning_rate": 1.0352868743328497e-06,
      "loss": 0.048441171646118164,
      "step": 6579
    },
    {
      "epoch": 0.8895949166004765,
      "grad_norm": 0.06586475670337677,
      "learning_rate": 1.032789646703733e-06,
      "loss": 0.04835247993469238,
      "step": 6580
    },
    {
      "epoch": 0.8897301133963125,
      "grad_norm": 0.10117357224225998,
      "learning_rate": 1.0302953271222938e-06,
      "loss": 0.07866621017456055,
      "step": 6581
    },
    {
      "epoch": 0.8898653101921484,
      "grad_norm": 0.12333240360021591,
      "learning_rate": 1.0278039161078634e-06,
      "loss": 0.07080721855163574,
      "step": 6582
    },
    {
      "epoch": 0.8900005069879844,
      "grad_norm": 0.10532978177070618,
      "learning_rate": 1.0253154141791705e-06,
      "loss": 0.09575176239013672,
      "step": 6583
    },
    {
      "epoch": 0.8901357037838203,
      "grad_norm": 0.06972696632146835,
      "learning_rate": 1.0228298218543253e-06,
      "loss": 0.04833173751831055,
      "step": 6584
    },
    {
      "epoch": 0.8902709005796563,
      "grad_norm": 0.10652590543031693,
      "learning_rate": 1.020347139650849e-06,
      "loss": 0.0778203010559082,
      "step": 6585
    },
    {
      "epoch": 0.8904060973754923,
      "grad_norm": 0.08605887740850449,
      "learning_rate": 1.0178673680856448e-06,
      "loss": 0.05930304527282715,
      "step": 6586
    },
    {
      "epoch": 0.8905412941713281,
      "grad_norm": 0.07626832276582718,
      "learning_rate": 1.0153905076750196e-06,
      "loss": 0.07395672798156738,
      "step": 6587
    },
    {
      "epoch": 0.8906764909671641,
      "grad_norm": 0.06566493213176727,
      "learning_rate": 1.0129165589346644e-06,
      "loss": 0.05273330211639404,
      "step": 6588
    },
    {
      "epoch": 0.890811687763,
      "grad_norm": 0.06652422994375229,
      "learning_rate": 1.0104455223796688e-06,
      "loss": 0.060160696506500244,
      "step": 6589
    },
    {
      "epoch": 0.890946884558836,
      "grad_norm": 0.07430104911327362,
      "learning_rate": 1.0079773985245178e-06,
      "loss": 0.05275726318359375,
      "step": 6590
    },
    {
      "epoch": 0.8910820813546719,
      "grad_norm": 0.10063283890485764,
      "learning_rate": 1.0055121878830837e-06,
      "loss": 0.0899662971496582,
      "step": 6591
    },
    {
      "epoch": 0.8912172781505079,
      "grad_norm": 0.06059126555919647,
      "learning_rate": 1.0030498909686458e-06,
      "loss": 0.05206775665283203,
      "step": 6592
    },
    {
      "epoch": 0.8913524749463437,
      "grad_norm": 0.05247807502746582,
      "learning_rate": 1.0005905082938593e-06,
      "loss": 0.03754448890686035,
      "step": 6593
    },
    {
      "epoch": 0.8914876717421797,
      "grad_norm": 0.05807942524552345,
      "learning_rate": 9.981340403707794e-07,
      "loss": 0.03816002607345581,
      "step": 6594
    },
    {
      "epoch": 0.8916228685380156,
      "grad_norm": 0.061262134462594986,
      "learning_rate": 9.956804877108638e-07,
      "loss": 0.043181657791137695,
      "step": 6595
    },
    {
      "epoch": 0.8917580653338516,
      "grad_norm": 0.07644299417734146,
      "learning_rate": 9.932298508249488e-07,
      "loss": 0.06563282012939453,
      "step": 6596
    },
    {
      "epoch": 0.8918932621296876,
      "grad_norm": 0.07854171842336655,
      "learning_rate": 9.907821302232729e-07,
      "loss": 0.06135678291320801,
      "step": 6597
    },
    {
      "epoch": 0.8920284589255235,
      "grad_norm": 0.0681791827082634,
      "learning_rate": 9.883373264154633e-07,
      "loss": 0.046141624450683594,
      "step": 6598
    },
    {
      "epoch": 0.8921636557213594,
      "grad_norm": 0.055947039276361465,
      "learning_rate": 9.858954399105397e-07,
      "loss": 0.04207766056060791,
      "step": 6599
    },
    {
      "epoch": 0.8922988525171953,
      "grad_norm": 0.06317632645368576,
      "learning_rate": 9.834564712169202e-07,
      "loss": 0.05024838447570801,
      "step": 6600
    },
    {
      "epoch": 0.8924340493130313,
      "grad_norm": 0.09453734755516052,
      "learning_rate": 9.81020420842409e-07,
      "loss": 0.062131643295288086,
      "step": 6601
    },
    {
      "epoch": 0.8925692461088672,
      "grad_norm": 0.0783371552824974,
      "learning_rate": 9.785872892942033e-07,
      "loss": 0.055925726890563965,
      "step": 6602
    },
    {
      "epoch": 0.8927044429047032,
      "grad_norm": 0.09139455109834671,
      "learning_rate": 9.761570770788964e-07,
      "loss": 0.07199573516845703,
      "step": 6603
    },
    {
      "epoch": 0.8928396397005391,
      "grad_norm": 0.055934008210897446,
      "learning_rate": 9.737297847024685e-07,
      "loss": 0.05993682146072388,
      "step": 6604
    },
    {
      "epoch": 0.892974836496375,
      "grad_norm": 0.11721453815698624,
      "learning_rate": 9.713054126702968e-07,
      "loss": 0.04857802391052246,
      "step": 6605
    },
    {
      "epoch": 0.893110033292211,
      "grad_norm": 0.09238698333501816,
      "learning_rate": 9.688839614871497e-07,
      "loss": 0.08888101577758789,
      "step": 6606
    },
    {
      "epoch": 0.8932452300880469,
      "grad_norm": 0.10019996017217636,
      "learning_rate": 9.664654316571852e-07,
      "loss": 0.10535812377929688,
      "step": 6607
    },
    {
      "epoch": 0.8933804268838829,
      "grad_norm": 0.04973414167761803,
      "learning_rate": 9.640498236839507e-07,
      "loss": 0.05350446701049805,
      "step": 6608
    },
    {
      "epoch": 0.8935156236797188,
      "grad_norm": 0.09560219943523407,
      "learning_rate": 9.616371380703953e-07,
      "loss": 0.06395101547241211,
      "step": 6609
    },
    {
      "epoch": 0.8936508204755548,
      "grad_norm": 0.06275220215320587,
      "learning_rate": 9.592273753188507e-07,
      "loss": 0.06680488586425781,
      "step": 6610
    },
    {
      "epoch": 0.8937860172713906,
      "grad_norm": 0.065501369535923,
      "learning_rate": 9.568205359310372e-07,
      "loss": 0.04054737091064453,
      "step": 6611
    },
    {
      "epoch": 0.8939212140672266,
      "grad_norm": 0.12545578181743622,
      "learning_rate": 9.544166204080772e-07,
      "loss": 0.054447293281555176,
      "step": 6612
    },
    {
      "epoch": 0.8940564108630625,
      "grad_norm": 0.10772591829299927,
      "learning_rate": 9.520156292504739e-07,
      "loss": 0.0805930495262146,
      "step": 6613
    },
    {
      "epoch": 0.8941916076588985,
      "grad_norm": 0.07085982710123062,
      "learning_rate": 9.496175629581322e-07,
      "loss": 0.04218930006027222,
      "step": 6614
    },
    {
      "epoch": 0.8943268044547344,
      "grad_norm": 0.10710301250219345,
      "learning_rate": 9.472224220303427e-07,
      "loss": 0.059938594698905945,
      "step": 6615
    },
    {
      "epoch": 0.8944620012505704,
      "grad_norm": 0.06642382591962814,
      "learning_rate": 9.448302069657799e-07,
      "loss": 0.048637986183166504,
      "step": 6616
    },
    {
      "epoch": 0.8945971980464063,
      "grad_norm": 0.07268780469894409,
      "learning_rate": 9.424409182625205e-07,
      "loss": 0.06392055749893188,
      "step": 6617
    },
    {
      "epoch": 0.8947323948422422,
      "grad_norm": 0.1987529844045639,
      "learning_rate": 9.40054556418023e-07,
      "loss": 0.07520723342895508,
      "step": 6618
    },
    {
      "epoch": 0.8948675916380782,
      "grad_norm": 0.09070106595754623,
      "learning_rate": 9.376711219291483e-07,
      "loss": 0.05606275796890259,
      "step": 6619
    },
    {
      "epoch": 0.8950027884339141,
      "grad_norm": 0.0709957405924797,
      "learning_rate": 9.352906152921348e-07,
      "loss": 0.060788631439208984,
      "step": 6620
    },
    {
      "epoch": 0.8951379852297501,
      "grad_norm": 0.09190468490123749,
      "learning_rate": 9.32913037002614e-07,
      "loss": 0.04943346977233887,
      "step": 6621
    },
    {
      "epoch": 0.895273182025586,
      "grad_norm": 0.15285420417785645,
      "learning_rate": 9.30538387555615e-07,
      "loss": 0.049030959606170654,
      "step": 6622
    },
    {
      "epoch": 0.8954083788214219,
      "grad_norm": 0.0674080178141594,
      "learning_rate": 9.281666674455508e-07,
      "loss": 0.046922922134399414,
      "step": 6623
    },
    {
      "epoch": 0.8955435756172578,
      "grad_norm": 0.14944292604923248,
      "learning_rate": 9.257978771662295e-07,
      "loss": 0.06735897064208984,
      "step": 6624
    },
    {
      "epoch": 0.8956787724130938,
      "grad_norm": 0.07170566916465759,
      "learning_rate": 9.234320172108418e-07,
      "loss": 0.07200336456298828,
      "step": 6625
    },
    {
      "epoch": 0.8958139692089298,
      "grad_norm": 0.0958380252122879,
      "learning_rate": 9.210690880719719e-07,
      "loss": 0.0538938045501709,
      "step": 6626
    },
    {
      "epoch": 0.8959491660047657,
      "grad_norm": 0.13218814134597778,
      "learning_rate": 9.187090902415962e-07,
      "loss": 0.08578038215637207,
      "step": 6627
    },
    {
      "epoch": 0.8960843628006017,
      "grad_norm": 0.1322334110736847,
      "learning_rate": 9.163520242110784e-07,
      "loss": 0.06138420104980469,
      "step": 6628
    },
    {
      "epoch": 0.8962195595964375,
      "grad_norm": 0.041208699345588684,
      "learning_rate": 9.13997890471176e-07,
      "loss": 0.03258466720581055,
      "step": 6629
    },
    {
      "epoch": 0.8963547563922735,
      "grad_norm": 0.0809289962053299,
      "learning_rate": 9.116466895120251e-07,
      "loss": 0.06803727149963379,
      "step": 6630
    },
    {
      "epoch": 0.8964899531881094,
      "grad_norm": 0.04746970906853676,
      "learning_rate": 9.092984218231609e-07,
      "loss": 0.044019997119903564,
      "step": 6631
    },
    {
      "epoch": 0.8966251499839454,
      "grad_norm": 0.07196468114852905,
      "learning_rate": 9.069530878935072e-07,
      "loss": 0.059203267097473145,
      "step": 6632
    },
    {
      "epoch": 0.8967603467797813,
      "grad_norm": 0.18243137001991272,
      "learning_rate": 9.046106882113753e-07,
      "loss": 0.09470769762992859,
      "step": 6633
    },
    {
      "epoch": 0.8968955435756173,
      "grad_norm": 0.05943652614951134,
      "learning_rate": 9.022712232644631e-07,
      "loss": 0.03992493450641632,
      "step": 6634
    },
    {
      "epoch": 0.8970307403714533,
      "grad_norm": 0.04794257879257202,
      "learning_rate": 8.999346935398611e-07,
      "loss": 0.05938720703125,
      "step": 6635
    },
    {
      "epoch": 0.8971659371672891,
      "grad_norm": 0.07320133596658707,
      "learning_rate": 8.976010995240436e-07,
      "loss": 0.07886743545532227,
      "step": 6636
    },
    {
      "epoch": 0.897301133963125,
      "grad_norm": 0.0737534761428833,
      "learning_rate": 8.952704417028818e-07,
      "loss": 0.05039691925048828,
      "step": 6637
    },
    {
      "epoch": 0.897436330758961,
      "grad_norm": 0.07039651274681091,
      "learning_rate": 8.929427205616308e-07,
      "loss": 0.055694580078125,
      "step": 6638
    },
    {
      "epoch": 0.897571527554797,
      "grad_norm": 0.05775744095444679,
      "learning_rate": 8.906179365849332e-07,
      "loss": 0.04315638542175293,
      "step": 6639
    },
    {
      "epoch": 0.8977067243506329,
      "grad_norm": 0.06370531767606735,
      "learning_rate": 8.882960902568216e-07,
      "loss": 0.027716398239135742,
      "step": 6640
    },
    {
      "epoch": 0.8978419211464689,
      "grad_norm": 0.06135125458240509,
      "learning_rate": 8.85977182060716e-07,
      "loss": 0.05271148681640625,
      "step": 6641
    },
    {
      "epoch": 0.8979771179423047,
      "grad_norm": 0.0692668929696083,
      "learning_rate": 8.836612124794285e-07,
      "loss": 0.05684030055999756,
      "step": 6642
    },
    {
      "epoch": 0.8981123147381407,
      "grad_norm": 0.060235388576984406,
      "learning_rate": 8.813481819951502e-07,
      "loss": 0.051661014556884766,
      "step": 6643
    },
    {
      "epoch": 0.8982475115339766,
      "grad_norm": 0.16336274147033691,
      "learning_rate": 8.790380910894724e-07,
      "loss": 0.06594669818878174,
      "step": 6644
    },
    {
      "epoch": 0.8983827083298126,
      "grad_norm": 0.10090608149766922,
      "learning_rate": 8.767309402433671e-07,
      "loss": 0.07398748397827148,
      "step": 6645
    },
    {
      "epoch": 0.8985179051256486,
      "grad_norm": 0.09646004438400269,
      "learning_rate": 8.744267299371917e-07,
      "loss": 0.05701184272766113,
      "step": 6646
    },
    {
      "epoch": 0.8986531019214845,
      "grad_norm": 0.18584474921226501,
      "learning_rate": 8.721254606507023e-07,
      "loss": 0.09387397766113281,
      "step": 6647
    },
    {
      "epoch": 0.8987882987173204,
      "grad_norm": 0.060422707349061966,
      "learning_rate": 8.698271328630275e-07,
      "loss": 0.06333351135253906,
      "step": 6648
    },
    {
      "epoch": 0.8989234955131563,
      "grad_norm": 0.07320088893175125,
      "learning_rate": 8.675317470526961e-07,
      "loss": 0.06865859031677246,
      "step": 6649
    },
    {
      "epoch": 0.8990586923089923,
      "grad_norm": 0.09771262109279633,
      "learning_rate": 8.652393036976159e-07,
      "loss": 0.06720256805419922,
      "step": 6650
    },
    {
      "epoch": 0.8991938891048282,
      "grad_norm": 0.11702706664800644,
      "learning_rate": 8.629498032750916e-07,
      "loss": 0.05758178234100342,
      "step": 6651
    },
    {
      "epoch": 0.8993290859006642,
      "grad_norm": 0.08316123485565186,
      "learning_rate": 8.606632462618069e-07,
      "loss": 0.06345510482788086,
      "step": 6652
    },
    {
      "epoch": 0.8994642826965001,
      "grad_norm": 0.05326274037361145,
      "learning_rate": 8.583796331338311e-07,
      "loss": 0.04036378860473633,
      "step": 6653
    },
    {
      "epoch": 0.899599479492336,
      "grad_norm": 0.07759150117635727,
      "learning_rate": 8.560989643666306e-07,
      "loss": 0.06870555877685547,
      "step": 6654
    },
    {
      "epoch": 0.8997346762881719,
      "grad_norm": 0.07183264940977097,
      "learning_rate": 8.538212404350471e-07,
      "loss": 0.05275440216064453,
      "step": 6655
    },
    {
      "epoch": 0.8998698730840079,
      "grad_norm": 0.0721798911690712,
      "learning_rate": 8.515464618133228e-07,
      "loss": 0.07120466232299805,
      "step": 6656
    },
    {
      "epoch": 0.9000050698798439,
      "grad_norm": 0.12253816425800323,
      "learning_rate": 8.492746289750725e-07,
      "loss": 0.06766670942306519,
      "step": 6657
    },
    {
      "epoch": 0.9001402666756798,
      "grad_norm": 0.09368474781513214,
      "learning_rate": 8.470057423933026e-07,
      "loss": 0.05730867385864258,
      "step": 6658
    },
    {
      "epoch": 0.9002754634715158,
      "grad_norm": 0.04538887366652489,
      "learning_rate": 8.447398025404118e-07,
      "loss": 0.04840803146362305,
      "step": 6659
    },
    {
      "epoch": 0.9004106602673516,
      "grad_norm": 0.06308309733867645,
      "learning_rate": 8.42476809888178e-07,
      "loss": 0.046575188636779785,
      "step": 6660
    },
    {
      "epoch": 0.9005458570631876,
      "grad_norm": 0.05168364197015762,
      "learning_rate": 8.402167649077725e-07,
      "loss": 0.04607181251049042,
      "step": 6661
    },
    {
      "epoch": 0.9006810538590235,
      "grad_norm": 0.05424867570400238,
      "learning_rate": 8.379596680697454e-07,
      "loss": 0.04422760009765625,
      "step": 6662
    },
    {
      "epoch": 0.9008162506548595,
      "grad_norm": 0.11579970270395279,
      "learning_rate": 8.357055198440328e-07,
      "loss": 0.05030703544616699,
      "step": 6663
    },
    {
      "epoch": 0.9009514474506954,
      "grad_norm": 0.0502345971763134,
      "learning_rate": 8.334543206999673e-07,
      "loss": 0.04666423797607422,
      "step": 6664
    },
    {
      "epoch": 0.9010866442465314,
      "grad_norm": 0.1272699236869812,
      "learning_rate": 8.312060711062558e-07,
      "loss": 0.06950902938842773,
      "step": 6665
    },
    {
      "epoch": 0.9012218410423672,
      "grad_norm": 0.11976861953735352,
      "learning_rate": 8.289607715309988e-07,
      "loss": 0.06149446964263916,
      "step": 6666
    },
    {
      "epoch": 0.9013570378382032,
      "grad_norm": 0.054975979030132294,
      "learning_rate": 8.267184224416791e-07,
      "loss": 0.05180993676185608,
      "step": 6667
    },
    {
      "epoch": 0.9014922346340392,
      "grad_norm": 0.08684058487415314,
      "learning_rate": 8.244790243051614e-07,
      "loss": 0.07228899002075195,
      "step": 6668
    },
    {
      "epoch": 0.9016274314298751,
      "grad_norm": 0.05990104377269745,
      "learning_rate": 8.222425775877079e-07,
      "loss": 0.029060959815979004,
      "step": 6669
    },
    {
      "epoch": 0.9017626282257111,
      "grad_norm": 0.050528429448604584,
      "learning_rate": 8.200090827549527e-07,
      "loss": 0.052945494651794434,
      "step": 6670
    },
    {
      "epoch": 0.901897825021547,
      "grad_norm": 0.16809502243995667,
      "learning_rate": 8.17778540271924e-07,
      "loss": 0.0878763198852539,
      "step": 6671
    },
    {
      "epoch": 0.9020330218173829,
      "grad_norm": 0.11046073585748672,
      "learning_rate": 8.155509506030334e-07,
      "loss": 0.08097012341022491,
      "step": 6672
    },
    {
      "epoch": 0.9021682186132188,
      "grad_norm": 0.057846855372190475,
      "learning_rate": 8.133263142120717e-07,
      "loss": 0.0437619686126709,
      "step": 6673
    },
    {
      "epoch": 0.9023034154090548,
      "grad_norm": 0.09421905130147934,
      "learning_rate": 8.111046315622284e-07,
      "loss": 0.06803774833679199,
      "step": 6674
    },
    {
      "epoch": 0.9024386122048907,
      "grad_norm": 0.06651435047388077,
      "learning_rate": 8.088859031160633e-07,
      "loss": 0.06949138641357422,
      "step": 6675
    },
    {
      "epoch": 0.9025738090007267,
      "grad_norm": 0.11255306750535965,
      "learning_rate": 8.066701293355288e-07,
      "loss": 0.05358874797821045,
      "step": 6676
    },
    {
      "epoch": 0.9027090057965627,
      "grad_norm": 0.08829524368047714,
      "learning_rate": 8.044573106819625e-07,
      "loss": 0.08324098587036133,
      "step": 6677
    },
    {
      "epoch": 0.9028442025923986,
      "grad_norm": 0.04944612830877304,
      "learning_rate": 8.022474476160824e-07,
      "loss": 0.045194387435913086,
      "step": 6678
    },
    {
      "epoch": 0.9029793993882345,
      "grad_norm": 0.08622890710830688,
      "learning_rate": 8.000405405979988e-07,
      "loss": 0.07186532020568848,
      "step": 6679
    },
    {
      "epoch": 0.9031145961840704,
      "grad_norm": 0.057132501155138016,
      "learning_rate": 7.978365900871943e-07,
      "loss": 0.040940821170806885,
      "step": 6680
    },
    {
      "epoch": 0.9032497929799064,
      "grad_norm": 0.06894759088754654,
      "learning_rate": 7.956355965425482e-07,
      "loss": 0.0367276668548584,
      "step": 6681
    },
    {
      "epoch": 0.9033849897757423,
      "grad_norm": 0.044041674584150314,
      "learning_rate": 7.934375604223193e-07,
      "loss": 0.039381980895996094,
      "step": 6682
    },
    {
      "epoch": 0.9035201865715783,
      "grad_norm": 0.1334766298532486,
      "learning_rate": 7.912424821841463e-07,
      "loss": 0.06312200427055359,
      "step": 6683
    },
    {
      "epoch": 0.9036553833674142,
      "grad_norm": 0.1010698527097702,
      "learning_rate": 7.89050362285062e-07,
      "loss": 0.060697078704833984,
      "step": 6684
    },
    {
      "epoch": 0.9037905801632501,
      "grad_norm": 0.039662402123212814,
      "learning_rate": 7.868612011814713e-07,
      "loss": 0.038825154304504395,
      "step": 6685
    },
    {
      "epoch": 0.903925776959086,
      "grad_norm": 0.08480498194694519,
      "learning_rate": 7.846749993291746e-07,
      "loss": 0.06257915496826172,
      "step": 6686
    },
    {
      "epoch": 0.904060973754922,
      "grad_norm": 0.04940733313560486,
      "learning_rate": 7.824917571833445e-07,
      "loss": 0.04669523239135742,
      "step": 6687
    },
    {
      "epoch": 0.904196170550758,
      "grad_norm": 0.08223103731870651,
      "learning_rate": 7.80311475198554e-07,
      "loss": 0.05759322643280029,
      "step": 6688
    },
    {
      "epoch": 0.9043313673465939,
      "grad_norm": 0.07242028415203094,
      "learning_rate": 7.781341538287384e-07,
      "loss": 0.0670386552810669,
      "step": 6689
    },
    {
      "epoch": 0.9044665641424299,
      "grad_norm": 0.09505094587802887,
      "learning_rate": 7.759597935272316e-07,
      "loss": 0.07269632816314697,
      "step": 6690
    },
    {
      "epoch": 0.9046017609382657,
      "grad_norm": 0.06608331948518753,
      "learning_rate": 7.7378839474675e-07,
      "loss": 0.07195568084716797,
      "step": 6691
    },
    {
      "epoch": 0.9047369577341017,
      "grad_norm": 0.05378718301653862,
      "learning_rate": 7.716199579393851e-07,
      "loss": 0.03840494155883789,
      "step": 6692
    },
    {
      "epoch": 0.9048721545299376,
      "grad_norm": 0.07324188202619553,
      "learning_rate": 7.694544835566259e-07,
      "loss": 0.0505528450012207,
      "step": 6693
    },
    {
      "epoch": 0.9050073513257736,
      "grad_norm": 0.04650648683309555,
      "learning_rate": 7.672919720493249e-07,
      "loss": 0.041406214237213135,
      "step": 6694
    },
    {
      "epoch": 0.9051425481216095,
      "grad_norm": 0.11132923513650894,
      "learning_rate": 7.651324238677338e-07,
      "loss": 0.06971263885498047,
      "step": 6695
    },
    {
      "epoch": 0.9052777449174455,
      "grad_norm": 0.061683230102062225,
      "learning_rate": 7.629758394614828e-07,
      "loss": 0.055091023445129395,
      "step": 6696
    },
    {
      "epoch": 0.9054129417132813,
      "grad_norm": 0.11126754432916641,
      "learning_rate": 7.608222192795794e-07,
      "loss": 0.06579005718231201,
      "step": 6697
    },
    {
      "epoch": 0.9055481385091173,
      "grad_norm": 0.08219487965106964,
      "learning_rate": 7.586715637704284e-07,
      "loss": 0.08167386054992676,
      "step": 6698
    },
    {
      "epoch": 0.9056833353049533,
      "grad_norm": 0.058290623128414154,
      "learning_rate": 7.565238733817998e-07,
      "loss": 0.056861162185668945,
      "step": 6699
    },
    {
      "epoch": 0.9058185321007892,
      "grad_norm": 0.03432619944214821,
      "learning_rate": 7.543791485608542e-07,
      "loss": 0.03377723693847656,
      "step": 6700
    },
    {
      "epoch": 0.9059537288966252,
      "grad_norm": 0.0552808940410614,
      "learning_rate": 7.52237389754138e-07,
      "loss": 0.04391968250274658,
      "step": 6701
    },
    {
      "epoch": 0.9060889256924611,
      "grad_norm": 0.04706662893295288,
      "learning_rate": 7.500985974075758e-07,
      "loss": 0.03809022903442383,
      "step": 6702
    },
    {
      "epoch": 0.906224122488297,
      "grad_norm": 0.06526406854391098,
      "learning_rate": 7.479627719664767e-07,
      "loss": 0.04655647277832031,
      "step": 6703
    },
    {
      "epoch": 0.9063593192841329,
      "grad_norm": 0.06305108964443207,
      "learning_rate": 7.458299138755281e-07,
      "loss": 0.06414985656738281,
      "step": 6704
    },
    {
      "epoch": 0.9064945160799689,
      "grad_norm": 0.050837915390729904,
      "learning_rate": 7.437000235788033e-07,
      "loss": 0.04151761531829834,
      "step": 6705
    },
    {
      "epoch": 0.9066297128758048,
      "grad_norm": 0.139189675450325,
      "learning_rate": 7.415731015197575e-07,
      "loss": 0.06122446060180664,
      "step": 6706
    },
    {
      "epoch": 0.9067649096716408,
      "grad_norm": 0.11274229735136032,
      "learning_rate": 7.39449148141228e-07,
      "loss": 0.06790328025817871,
      "step": 6707
    },
    {
      "epoch": 0.9069001064674768,
      "grad_norm": 0.04873769357800484,
      "learning_rate": 7.373281638854329e-07,
      "loss": 0.04820966720581055,
      "step": 6708
    },
    {
      "epoch": 0.9070353032633126,
      "grad_norm": 0.08780638873577118,
      "learning_rate": 7.352101491939722e-07,
      "loss": 0.05083608627319336,
      "step": 6709
    },
    {
      "epoch": 0.9071705000591486,
      "grad_norm": 0.07825469970703125,
      "learning_rate": 7.330951045078249e-07,
      "loss": 0.0728602409362793,
      "step": 6710
    },
    {
      "epoch": 0.9073056968549845,
      "grad_norm": 0.096316859126091,
      "learning_rate": 7.309830302673621e-07,
      "loss": 0.04883098602294922,
      "step": 6711
    },
    {
      "epoch": 0.9074408936508205,
      "grad_norm": 0.20904502272605896,
      "learning_rate": 7.288739269123184e-07,
      "loss": 0.07873964309692383,
      "step": 6712
    },
    {
      "epoch": 0.9075760904466564,
      "grad_norm": 0.08129238337278366,
      "learning_rate": 7.267677948818296e-07,
      "loss": 0.062062978744506836,
      "step": 6713
    },
    {
      "epoch": 0.9077112872424924,
      "grad_norm": 0.14217080175876617,
      "learning_rate": 7.246646346143997e-07,
      "loss": 0.0724191665649414,
      "step": 6714
    },
    {
      "epoch": 0.9078464840383282,
      "grad_norm": 0.15996962785720825,
      "learning_rate": 7.225644465479153e-07,
      "loss": 0.07497239112854004,
      "step": 6715
    },
    {
      "epoch": 0.9079816808341642,
      "grad_norm": 0.045987971127033234,
      "learning_rate": 7.204672311196547e-07,
      "loss": 0.03400075435638428,
      "step": 6716
    },
    {
      "epoch": 0.9081168776300002,
      "grad_norm": 0.07036244124174118,
      "learning_rate": 7.183729887662604e-07,
      "loss": 0.04816460609436035,
      "step": 6717
    },
    {
      "epoch": 0.9082520744258361,
      "grad_norm": 0.0665932148694992,
      "learning_rate": 7.162817199237703e-07,
      "loss": 0.07290220260620117,
      "step": 6718
    },
    {
      "epoch": 0.9083872712216721,
      "grad_norm": 0.13764077425003052,
      "learning_rate": 7.141934250275978e-07,
      "loss": 0.08014583587646484,
      "step": 6719
    },
    {
      "epoch": 0.908522468017508,
      "grad_norm": 0.1037042960524559,
      "learning_rate": 7.121081045125316e-07,
      "loss": 0.06541967391967773,
      "step": 6720
    },
    {
      "epoch": 0.9086576648133439,
      "grad_norm": 0.060645900666713715,
      "learning_rate": 7.100257588127545e-07,
      "loss": 0.04391634464263916,
      "step": 6721
    },
    {
      "epoch": 0.9087928616091798,
      "grad_norm": 0.08745758980512619,
      "learning_rate": 7.079463883618148e-07,
      "loss": 0.07028853893280029,
      "step": 6722
    },
    {
      "epoch": 0.9089280584050158,
      "grad_norm": 0.09235255420207977,
      "learning_rate": 7.058699935926527e-07,
      "loss": 0.052414894104003906,
      "step": 6723
    },
    {
      "epoch": 0.9090632552008517,
      "grad_norm": 0.114885613322258,
      "learning_rate": 7.037965749375808e-07,
      "loss": 0.061815500259399414,
      "step": 6724
    },
    {
      "epoch": 0.9091984519966877,
      "grad_norm": 0.09272823482751846,
      "learning_rate": 7.017261328283037e-07,
      "loss": 0.08160018920898438,
      "step": 6725
    },
    {
      "epoch": 0.9093336487925237,
      "grad_norm": 0.15473969280719757,
      "learning_rate": 6.996586676958916e-07,
      "loss": 0.0699617862701416,
      "step": 6726
    },
    {
      "epoch": 0.9094688455883596,
      "grad_norm": 0.09846507012844086,
      "learning_rate": 6.975941799708019e-07,
      "loss": 0.07828950881958008,
      "step": 6727
    },
    {
      "epoch": 0.9096040423841955,
      "grad_norm": 0.12600812315940857,
      "learning_rate": 6.955326700828757e-07,
      "loss": 0.08341217041015625,
      "step": 6728
    },
    {
      "epoch": 0.9097392391800314,
      "grad_norm": 0.18116658926010132,
      "learning_rate": 6.934741384613246e-07,
      "loss": 0.06686651706695557,
      "step": 6729
    },
    {
      "epoch": 0.9098744359758674,
      "grad_norm": 0.060926783829927444,
      "learning_rate": 6.91418585534756e-07,
      "loss": 0.062319278717041016,
      "step": 6730
    },
    {
      "epoch": 0.9100096327717033,
      "grad_norm": 0.08466867357492447,
      "learning_rate": 6.893660117311373e-07,
      "loss": 0.0571293830871582,
      "step": 6731
    },
    {
      "epoch": 0.9101448295675393,
      "grad_norm": 0.1459810733795166,
      "learning_rate": 6.873164174778252e-07,
      "loss": 0.0804172158241272,
      "step": 6732
    },
    {
      "epoch": 0.9102800263633752,
      "grad_norm": 0.095429927110672,
      "learning_rate": 6.852698032015631e-07,
      "loss": 0.07612800598144531,
      "step": 6733
    },
    {
      "epoch": 0.9104152231592111,
      "grad_norm": 0.12787973880767822,
      "learning_rate": 6.832261693284636e-07,
      "loss": 0.07410478591918945,
      "step": 6734
    },
    {
      "epoch": 0.910550419955047,
      "grad_norm": 0.11708753556013107,
      "learning_rate": 6.811855162840214e-07,
      "loss": 0.0609288215637207,
      "step": 6735
    },
    {
      "epoch": 0.910685616750883,
      "grad_norm": 0.08623882383108139,
      "learning_rate": 6.791478444931132e-07,
      "loss": 0.06495475769042969,
      "step": 6736
    },
    {
      "epoch": 0.910820813546719,
      "grad_norm": 0.08381275087594986,
      "learning_rate": 6.77113154379988e-07,
      "loss": 0.04236018657684326,
      "step": 6737
    },
    {
      "epoch": 0.9109560103425549,
      "grad_norm": 0.05358347296714783,
      "learning_rate": 6.75081446368287e-07,
      "loss": 0.050638437271118164,
      "step": 6738
    },
    {
      "epoch": 0.9110912071383909,
      "grad_norm": 0.07180771231651306,
      "learning_rate": 6.730527208810166e-07,
      "loss": 0.04754924774169922,
      "step": 6739
    },
    {
      "epoch": 0.9112264039342267,
      "grad_norm": 0.07693973183631897,
      "learning_rate": 6.710269783405709e-07,
      "loss": 0.05288201570510864,
      "step": 6740
    },
    {
      "epoch": 0.9113616007300627,
      "grad_norm": 0.056572265923023224,
      "learning_rate": 6.690042191687206e-07,
      "loss": 0.04134058952331543,
      "step": 6741
    },
    {
      "epoch": 0.9114967975258986,
      "grad_norm": 0.11207717657089233,
      "learning_rate": 6.669844437866124e-07,
      "loss": 0.06454777717590332,
      "step": 6742
    },
    {
      "epoch": 0.9116319943217346,
      "grad_norm": 0.08345456421375275,
      "learning_rate": 6.649676526147764e-07,
      "loss": 0.05972456932067871,
      "step": 6743
    },
    {
      "epoch": 0.9117671911175705,
      "grad_norm": 0.03956246376037598,
      "learning_rate": 6.629538460731199e-07,
      "loss": 0.02511155605316162,
      "step": 6744
    },
    {
      "epoch": 0.9119023879134065,
      "grad_norm": 0.08143890649080276,
      "learning_rate": 6.609430245809261e-07,
      "loss": 0.06749105453491211,
      "step": 6745
    },
    {
      "epoch": 0.9120375847092423,
      "grad_norm": 0.09006833285093307,
      "learning_rate": 6.589351885568617e-07,
      "loss": 0.08936500549316406,
      "step": 6746
    },
    {
      "epoch": 0.9121727815050783,
      "grad_norm": 0.07447297126054764,
      "learning_rate": 6.569303384189624e-07,
      "loss": 0.07627391815185547,
      "step": 6747
    },
    {
      "epoch": 0.9123079783009143,
      "grad_norm": 0.108567014336586,
      "learning_rate": 6.54928474584659e-07,
      "loss": 0.0623016357421875,
      "step": 6748
    },
    {
      "epoch": 0.9124431750967502,
      "grad_norm": 0.08347965031862259,
      "learning_rate": 6.5292959747074e-07,
      "loss": 0.04192352294921875,
      "step": 6749
    },
    {
      "epoch": 0.9125783718925862,
      "grad_norm": 0.03402784839272499,
      "learning_rate": 6.509337074933891e-07,
      "loss": 0.037055015563964844,
      "step": 6750
    },
    {
      "epoch": 0.9127135686884221,
      "grad_norm": 0.04279724135994911,
      "learning_rate": 6.489408050681589e-07,
      "loss": 0.03884100914001465,
      "step": 6751
    },
    {
      "epoch": 0.912848765484258,
      "grad_norm": 0.06605985015630722,
      "learning_rate": 6.469508906099792e-07,
      "loss": 0.05292713642120361,
      "step": 6752
    },
    {
      "epoch": 0.9129839622800939,
      "grad_norm": 0.06558471918106079,
      "learning_rate": 6.449639645331684e-07,
      "loss": 0.06713080406188965,
      "step": 6753
    },
    {
      "epoch": 0.9131191590759299,
      "grad_norm": 0.07589532434940338,
      "learning_rate": 6.429800272514058e-07,
      "loss": 0.048555850982666016,
      "step": 6754
    },
    {
      "epoch": 0.9132543558717658,
      "grad_norm": 0.043322283774614334,
      "learning_rate": 6.409990791777659e-07,
      "loss": 0.04274141788482666,
      "step": 6755
    },
    {
      "epoch": 0.9133895526676018,
      "grad_norm": 0.08356264978647232,
      "learning_rate": 6.390211207246888e-07,
      "loss": 0.07425069808959961,
      "step": 6756
    },
    {
      "epoch": 0.9135247494634378,
      "grad_norm": 0.06733854860067368,
      "learning_rate": 6.370461523039967e-07,
      "loss": 0.05004429817199707,
      "step": 6757
    },
    {
      "epoch": 0.9136599462592736,
      "grad_norm": 0.06644568592309952,
      "learning_rate": 6.350741743268873e-07,
      "loss": 0.054903388023376465,
      "step": 6758
    },
    {
      "epoch": 0.9137951430551096,
      "grad_norm": 0.07641942799091339,
      "learning_rate": 6.331051872039373e-07,
      "loss": 0.044489383697509766,
      "step": 6759
    },
    {
      "epoch": 0.9139303398509455,
      "grad_norm": 0.07348938286304474,
      "learning_rate": 6.31139191345102e-07,
      "loss": 0.05042374134063721,
      "step": 6760
    },
    {
      "epoch": 0.9140655366467815,
      "grad_norm": 0.06905170530080795,
      "learning_rate": 6.291761871597091e-07,
      "loss": 0.047342449426651,
      "step": 6761
    },
    {
      "epoch": 0.9142007334426174,
      "grad_norm": 0.0952875167131424,
      "learning_rate": 6.272161750564731e-07,
      "loss": 0.06904077529907227,
      "step": 6762
    },
    {
      "epoch": 0.9143359302384534,
      "grad_norm": 0.059866953641176224,
      "learning_rate": 6.252591554434728e-07,
      "loss": 0.05153632164001465,
      "step": 6763
    },
    {
      "epoch": 0.9144711270342892,
      "grad_norm": 0.06745777279138565,
      "learning_rate": 6.233051287281688e-07,
      "loss": 0.062015533447265625,
      "step": 6764
    },
    {
      "epoch": 0.9146063238301252,
      "grad_norm": 0.062436845153570175,
      "learning_rate": 6.213540953174057e-07,
      "loss": 0.0528794527053833,
      "step": 6765
    },
    {
      "epoch": 0.9147415206259611,
      "grad_norm": 0.0822429433465004,
      "learning_rate": 6.194060556173953e-07,
      "loss": 0.07905292510986328,
      "step": 6766
    },
    {
      "epoch": 0.9148767174217971,
      "grad_norm": 0.044968072324991226,
      "learning_rate": 6.17461010033733e-07,
      "loss": 0.0505375862121582,
      "step": 6767
    },
    {
      "epoch": 0.9150119142176331,
      "grad_norm": 0.09746953845024109,
      "learning_rate": 6.155189589713833e-07,
      "loss": 0.05556917190551758,
      "step": 6768
    },
    {
      "epoch": 0.915147111013469,
      "grad_norm": 0.15062448382377625,
      "learning_rate": 6.135799028346928e-07,
      "loss": 0.06459987163543701,
      "step": 6769
    },
    {
      "epoch": 0.915282307809305,
      "grad_norm": 0.08286046236753464,
      "learning_rate": 6.116438420273868e-07,
      "loss": 0.06889796257019043,
      "step": 6770
    },
    {
      "epoch": 0.9154175046051408,
      "grad_norm": 0.09522316604852676,
      "learning_rate": 6.097107769525595e-07,
      "loss": 0.06338739395141602,
      "step": 6771
    },
    {
      "epoch": 0.9155527014009768,
      "grad_norm": 0.1612064093351364,
      "learning_rate": 6.077807080126873e-07,
      "loss": 0.06690263748168945,
      "step": 6772
    },
    {
      "epoch": 0.9156878981968127,
      "grad_norm": 0.07434094697237015,
      "learning_rate": 6.058536356096206e-07,
      "loss": 0.06382691860198975,
      "step": 6773
    },
    {
      "epoch": 0.9158230949926487,
      "grad_norm": 0.07352715730667114,
      "learning_rate": 6.039295601445833e-07,
      "loss": 0.06329339742660522,
      "step": 6774
    },
    {
      "epoch": 0.9159582917884846,
      "grad_norm": 0.155765563249588,
      "learning_rate": 6.020084820181831e-07,
      "loss": 0.07581257820129395,
      "step": 6775
    },
    {
      "epoch": 0.9160934885843206,
      "grad_norm": 0.04926585033535957,
      "learning_rate": 6.000904016303971e-07,
      "loss": 0.0404854416847229,
      "step": 6776
    },
    {
      "epoch": 0.9162286853801564,
      "grad_norm": 0.0913032665848732,
      "learning_rate": 5.981753193805789e-07,
      "loss": 0.08167791366577148,
      "step": 6777
    },
    {
      "epoch": 0.9163638821759924,
      "grad_norm": 0.0839015319943428,
      "learning_rate": 5.962632356674597e-07,
      "loss": 0.05323457717895508,
      "step": 6778
    },
    {
      "epoch": 0.9164990789718284,
      "grad_norm": 0.11051979660987854,
      "learning_rate": 5.94354150889141e-07,
      "loss": 0.07772564888000488,
      "step": 6779
    },
    {
      "epoch": 0.9166342757676643,
      "grad_norm": 0.0768633633852005,
      "learning_rate": 5.924480654431147e-07,
      "loss": 0.05799245834350586,
      "step": 6780
    },
    {
      "epoch": 0.9167694725635003,
      "grad_norm": 0.1127280592918396,
      "learning_rate": 5.905449797262252e-07,
      "loss": 0.06021428108215332,
      "step": 6781
    },
    {
      "epoch": 0.9169046693593362,
      "grad_norm": 0.07609863579273224,
      "learning_rate": 5.886448941347156e-07,
      "loss": 0.06479433178901672,
      "step": 6782
    },
    {
      "epoch": 0.9170398661551721,
      "grad_norm": 0.04923299327492714,
      "learning_rate": 5.867478090641892e-07,
      "loss": 0.040313392877578735,
      "step": 6783
    },
    {
      "epoch": 0.917175062951008,
      "grad_norm": 0.09888669848442078,
      "learning_rate": 5.848537249096269e-07,
      "loss": 0.05931282043457031,
      "step": 6784
    },
    {
      "epoch": 0.917310259746844,
      "grad_norm": 0.12415360659360886,
      "learning_rate": 5.829626420653949e-07,
      "loss": 0.05547034740447998,
      "step": 6785
    },
    {
      "epoch": 0.91744545654268,
      "grad_norm": 0.08682490885257721,
      "learning_rate": 5.810745609252166e-07,
      "loss": 0.0454595685005188,
      "step": 6786
    },
    {
      "epoch": 0.9175806533385159,
      "grad_norm": 0.11459515243768692,
      "learning_rate": 5.791894818822091e-07,
      "loss": 0.05810356140136719,
      "step": 6787
    },
    {
      "epoch": 0.9177158501343519,
      "grad_norm": 0.09491605311632156,
      "learning_rate": 5.773074053288519e-07,
      "loss": 0.06572043895721436,
      "step": 6788
    },
    {
      "epoch": 0.9178510469301877,
      "grad_norm": 0.09550150483846664,
      "learning_rate": 5.75428331657003e-07,
      "loss": 0.08484911918640137,
      "step": 6789
    },
    {
      "epoch": 0.9179862437260237,
      "grad_norm": 0.07265599817037582,
      "learning_rate": 5.735522612578998e-07,
      "loss": 0.07644820213317871,
      "step": 6790
    },
    {
      "epoch": 0.9181214405218596,
      "grad_norm": 0.0837293341755867,
      "learning_rate": 5.716791945221444e-07,
      "loss": 0.06304502487182617,
      "step": 6791
    },
    {
      "epoch": 0.9182566373176956,
      "grad_norm": 0.03523504361510277,
      "learning_rate": 5.698091318397219e-07,
      "loss": 0.028102397918701172,
      "step": 6792
    },
    {
      "epoch": 0.9183918341135315,
      "grad_norm": 0.05463401600718498,
      "learning_rate": 5.679420735999908e-07,
      "loss": 0.05178022384643555,
      "step": 6793
    },
    {
      "epoch": 0.9185270309093675,
      "grad_norm": 0.06810581684112549,
      "learning_rate": 5.660780201916799e-07,
      "loss": 0.030712023377418518,
      "step": 6794
    },
    {
      "epoch": 0.9186622277052033,
      "grad_norm": 0.052742671221494675,
      "learning_rate": 5.642169720028973e-07,
      "loss": 0.04141056537628174,
      "step": 6795
    },
    {
      "epoch": 0.9187974245010393,
      "grad_norm": 0.10572075843811035,
      "learning_rate": 5.623589294211196e-07,
      "loss": 0.06541228294372559,
      "step": 6796
    },
    {
      "epoch": 0.9189326212968753,
      "grad_norm": 0.05111096799373627,
      "learning_rate": 5.605038928332057e-07,
      "loss": 0.04832315444946289,
      "step": 6797
    },
    {
      "epoch": 0.9190678180927112,
      "grad_norm": 0.044135306030511856,
      "learning_rate": 5.586518626253817e-07,
      "loss": 0.037868618965148926,
      "step": 6798
    },
    {
      "epoch": 0.9192030148885472,
      "grad_norm": 0.06819167733192444,
      "learning_rate": 5.568028391832524e-07,
      "loss": 0.057631611824035645,
      "step": 6799
    },
    {
      "epoch": 0.9193382116843831,
      "grad_norm": 0.070487380027771,
      "learning_rate": 5.549568228917917e-07,
      "loss": 0.05653238296508789,
      "step": 6800
    },
    {
      "epoch": 0.919473408480219,
      "grad_norm": 0.1173807829618454,
      "learning_rate": 5.531138141353486e-07,
      "loss": 0.061377644538879395,
      "step": 6801
    },
    {
      "epoch": 0.9196086052760549,
      "grad_norm": 0.12492447346448898,
      "learning_rate": 5.512738132976514e-07,
      "loss": 0.05899500846862793,
      "step": 6802
    },
    {
      "epoch": 0.9197438020718909,
      "grad_norm": 0.14777909219264984,
      "learning_rate": 5.49436820761795e-07,
      "loss": 0.06536829471588135,
      "step": 6803
    },
    {
      "epoch": 0.9198789988677268,
      "grad_norm": 0.09018542617559433,
      "learning_rate": 5.476028369102537e-07,
      "loss": 0.06111180782318115,
      "step": 6804
    },
    {
      "epoch": 0.9200141956635628,
      "grad_norm": 0.07368019223213196,
      "learning_rate": 5.45771862124872e-07,
      "loss": 0.061571359634399414,
      "step": 6805
    },
    {
      "epoch": 0.9201493924593988,
      "grad_norm": 0.14074435830116272,
      "learning_rate": 5.439438967868649e-07,
      "loss": 0.08659911155700684,
      "step": 6806
    },
    {
      "epoch": 0.9202845892552346,
      "grad_norm": 0.08623439073562622,
      "learning_rate": 5.421189412768296e-07,
      "loss": 0.04534733295440674,
      "step": 6807
    },
    {
      "epoch": 0.9204197860510706,
      "grad_norm": 0.12841518223285675,
      "learning_rate": 5.402969959747306e-07,
      "loss": 0.062029361724853516,
      "step": 6808
    },
    {
      "epoch": 0.9205549828469065,
      "grad_norm": 0.11448809504508972,
      "learning_rate": 5.384780612599044e-07,
      "loss": 0.0379372239112854,
      "step": 6809
    },
    {
      "epoch": 0.9206901796427425,
      "grad_norm": 0.05587799474596977,
      "learning_rate": 5.366621375110647e-07,
      "loss": 0.03373980522155762,
      "step": 6810
    },
    {
      "epoch": 0.9208253764385784,
      "grad_norm": 0.06052003055810928,
      "learning_rate": 5.348492251062942e-07,
      "loss": 0.04782605171203613,
      "step": 6811
    },
    {
      "epoch": 0.9209605732344144,
      "grad_norm": 0.07805603742599487,
      "learning_rate": 5.330393244230558e-07,
      "loss": 0.053964436054229736,
      "step": 6812
    },
    {
      "epoch": 0.9210957700302503,
      "grad_norm": 0.10421637445688248,
      "learning_rate": 5.312324358381731e-07,
      "loss": 0.07691431045532227,
      "step": 6813
    },
    {
      "epoch": 0.9212309668260862,
      "grad_norm": 0.10782796889543533,
      "learning_rate": 5.29428559727857e-07,
      "loss": 0.08362793922424316,
      "step": 6814
    },
    {
      "epoch": 0.9213661636219221,
      "grad_norm": 0.09339733421802521,
      "learning_rate": 5.276276964676802e-07,
      "loss": 0.06957411766052246,
      "step": 6815
    },
    {
      "epoch": 0.9215013604177581,
      "grad_norm": 0.05293793976306915,
      "learning_rate": 5.258298464325928e-07,
      "loss": 0.03976625204086304,
      "step": 6816
    },
    {
      "epoch": 0.921636557213594,
      "grad_norm": 0.06883855164051056,
      "learning_rate": 5.240350099969204e-07,
      "loss": 0.05500668287277222,
      "step": 6817
    },
    {
      "epoch": 0.92177175400943,
      "grad_norm": 0.10639632493257523,
      "learning_rate": 5.222431875343492e-07,
      "loss": 0.05985760688781738,
      "step": 6818
    },
    {
      "epoch": 0.921906950805266,
      "grad_norm": 0.08619681745767593,
      "learning_rate": 5.204543794179539e-07,
      "loss": 0.0613057017326355,
      "step": 6819
    },
    {
      "epoch": 0.9220421476011018,
      "grad_norm": 0.06661764532327652,
      "learning_rate": 5.186685860201717e-07,
      "loss": 0.05211687088012695,
      "step": 6820
    },
    {
      "epoch": 0.9221773443969378,
      "grad_norm": 0.09220027923583984,
      "learning_rate": 5.16885807712812e-07,
      "loss": 0.04319643974304199,
      "step": 6821
    },
    {
      "epoch": 0.9223125411927737,
      "grad_norm": 0.09631132334470749,
      "learning_rate": 5.151060448670625e-07,
      "loss": 0.05782175064086914,
      "step": 6822
    },
    {
      "epoch": 0.9224477379886097,
      "grad_norm": 0.053621068596839905,
      "learning_rate": 5.133292978534754e-07,
      "loss": 0.04122114181518555,
      "step": 6823
    },
    {
      "epoch": 0.9225829347844456,
      "grad_norm": 0.07459413260221481,
      "learning_rate": 5.115555670419814e-07,
      "loss": 0.05349469184875488,
      "step": 6824
    },
    {
      "epoch": 0.9227181315802816,
      "grad_norm": 0.08251022547483444,
      "learning_rate": 5.097848528018817e-07,
      "loss": 0.046590209007263184,
      "step": 6825
    },
    {
      "epoch": 0.9228533283761174,
      "grad_norm": 0.10208684206008911,
      "learning_rate": 5.080171555018448e-07,
      "loss": 0.08421897888183594,
      "step": 6826
    },
    {
      "epoch": 0.9229885251719534,
      "grad_norm": 0.07252252101898193,
      "learning_rate": 5.06252475509918e-07,
      "loss": 0.05518531799316406,
      "step": 6827
    },
    {
      "epoch": 0.9231237219677894,
      "grad_norm": 0.07434801757335663,
      "learning_rate": 5.044908131935139e-07,
      "loss": 0.052265167236328125,
      "step": 6828
    },
    {
      "epoch": 0.9232589187636253,
      "grad_norm": 0.08358155936002731,
      "learning_rate": 5.027321689194242e-07,
      "loss": 0.053508758544921875,
      "step": 6829
    },
    {
      "epoch": 0.9233941155594613,
      "grad_norm": 0.06201353296637535,
      "learning_rate": 5.009765430538061e-07,
      "loss": 0.06040000915527344,
      "step": 6830
    },
    {
      "epoch": 0.9235293123552972,
      "grad_norm": 0.07067056000232697,
      "learning_rate": 4.992239359621886e-07,
      "loss": 0.05200427770614624,
      "step": 6831
    },
    {
      "epoch": 0.9236645091511331,
      "grad_norm": 0.07980097830295563,
      "learning_rate": 4.974743480094767e-07,
      "loss": 0.06742334365844727,
      "step": 6832
    },
    {
      "epoch": 0.923799705946969,
      "grad_norm": 0.10068795830011368,
      "learning_rate": 4.957277795599407e-07,
      "loss": 0.06462322175502777,
      "step": 6833
    },
    {
      "epoch": 0.923934902742805,
      "grad_norm": 0.056041549891233444,
      "learning_rate": 4.93984230977228e-07,
      "loss": 0.05373716354370117,
      "step": 6834
    },
    {
      "epoch": 0.9240700995386409,
      "grad_norm": 0.0804627314209938,
      "learning_rate": 4.922437026243531e-07,
      "loss": 0.06502079963684082,
      "step": 6835
    },
    {
      "epoch": 0.9242052963344769,
      "grad_norm": 0.12691618502140045,
      "learning_rate": 4.905061948637063e-07,
      "loss": 0.060770273208618164,
      "step": 6836
    },
    {
      "epoch": 0.9243404931303129,
      "grad_norm": 0.09311460703611374,
      "learning_rate": 4.887717080570431e-07,
      "loss": 0.06514084339141846,
      "step": 6837
    },
    {
      "epoch": 0.9244756899261487,
      "grad_norm": 0.08764898031949997,
      "learning_rate": 4.870402425654913e-07,
      "loss": 0.0783529281616211,
      "step": 6838
    },
    {
      "epoch": 0.9246108867219847,
      "grad_norm": 0.08409794420003891,
      "learning_rate": 4.853117987495542e-07,
      "loss": 0.06356537342071533,
      "step": 6839
    },
    {
      "epoch": 0.9247460835178206,
      "grad_norm": 0.0501297190785408,
      "learning_rate": 4.83586376969104e-07,
      "loss": 0.04085803031921387,
      "step": 6840
    },
    {
      "epoch": 0.9248812803136566,
      "grad_norm": 0.05269839987158775,
      "learning_rate": 4.818639775833816e-07,
      "loss": 0.04741021245718002,
      "step": 6841
    },
    {
      "epoch": 0.9250164771094925,
      "grad_norm": 0.05799153074622154,
      "learning_rate": 4.801446009509969e-07,
      "loss": 0.03713059425354004,
      "step": 6842
    },
    {
      "epoch": 0.9251516739053285,
      "grad_norm": 0.07219300419092178,
      "learning_rate": 4.784282474299367e-07,
      "loss": 0.06656515598297119,
      "step": 6843
    },
    {
      "epoch": 0.9252868707011643,
      "grad_norm": 0.09508062154054642,
      "learning_rate": 4.767149173775537e-07,
      "loss": 0.042938947677612305,
      "step": 6844
    },
    {
      "epoch": 0.9254220674970003,
      "grad_norm": 0.09607056528329849,
      "learning_rate": 4.750046111505724e-07,
      "loss": 0.06508302688598633,
      "step": 6845
    },
    {
      "epoch": 0.9255572642928362,
      "grad_norm": 0.061726950109004974,
      "learning_rate": 4.732973291050896e-07,
      "loss": 0.04613304138183594,
      "step": 6846
    },
    {
      "epoch": 0.9256924610886722,
      "grad_norm": 0.06372092664241791,
      "learning_rate": 4.7159307159656607e-07,
      "loss": 0.04676032066345215,
      "step": 6847
    },
    {
      "epoch": 0.9258276578845082,
      "grad_norm": 0.17753368616104126,
      "learning_rate": 4.6989183897983954e-07,
      "loss": 0.05688464641571045,
      "step": 6848
    },
    {
      "epoch": 0.9259628546803441,
      "grad_norm": 0.08127326518297195,
      "learning_rate": 4.681936316091201e-07,
      "loss": 0.05523126572370529,
      "step": 6849
    },
    {
      "epoch": 0.92609805147618,
      "grad_norm": 0.07534569501876831,
      "learning_rate": 4.664984498379765e-07,
      "loss": 0.05183809995651245,
      "step": 6850
    },
    {
      "epoch": 0.9262332482720159,
      "grad_norm": 0.043646253645420074,
      "learning_rate": 4.6480629401935814e-07,
      "loss": 0.03409457206726074,
      "step": 6851
    },
    {
      "epoch": 0.9263684450678519,
      "grad_norm": 0.06272473931312561,
      "learning_rate": 4.631171645055815e-07,
      "loss": 0.04832053184509277,
      "step": 6852
    },
    {
      "epoch": 0.9265036418636878,
      "grad_norm": 0.08324024826288223,
      "learning_rate": 4.614310616483286e-07,
      "loss": 0.05918073654174805,
      "step": 6853
    },
    {
      "epoch": 0.9266388386595238,
      "grad_norm": 0.07095614820718765,
      "learning_rate": 4.5974798579866193e-07,
      "loss": 0.04324913024902344,
      "step": 6854
    },
    {
      "epoch": 0.9267740354553597,
      "grad_norm": 0.0709993913769722,
      "learning_rate": 4.580679373069996e-07,
      "loss": 0.0503043532371521,
      "step": 6855
    },
    {
      "epoch": 0.9269092322511956,
      "grad_norm": 0.10518515110015869,
      "learning_rate": 4.5639091652314e-07,
      "loss": 0.06734371185302734,
      "step": 6856
    },
    {
      "epoch": 0.9270444290470315,
      "grad_norm": 0.10777423530817032,
      "learning_rate": 4.54716923796249e-07,
      "loss": 0.05843234062194824,
      "step": 6857
    },
    {
      "epoch": 0.9271796258428675,
      "grad_norm": 0.09842325001955032,
      "learning_rate": 4.5304595947485927e-07,
      "loss": 0.06998658180236816,
      "step": 6858
    },
    {
      "epoch": 0.9273148226387035,
      "grad_norm": 0.13332384824752808,
      "learning_rate": 4.5137802390687433e-07,
      "loss": 0.06135225296020508,
      "step": 6859
    },
    {
      "epoch": 0.9274500194345394,
      "grad_norm": 0.12472131103277206,
      "learning_rate": 4.497131174395663e-07,
      "loss": 0.06375408172607422,
      "step": 6860
    },
    {
      "epoch": 0.9275852162303754,
      "grad_norm": 0.08886054158210754,
      "learning_rate": 4.4805124041957967e-07,
      "loss": 0.057554006576538086,
      "step": 6861
    },
    {
      "epoch": 0.9277204130262113,
      "grad_norm": 0.15123821794986725,
      "learning_rate": 4.463923931929259e-07,
      "loss": 0.06347322463989258,
      "step": 6862
    },
    {
      "epoch": 0.9278556098220472,
      "grad_norm": 0.050435133278369904,
      "learning_rate": 4.4473657610498377e-07,
      "loss": 0.035222649574279785,
      "step": 6863
    },
    {
      "epoch": 0.9279908066178831,
      "grad_norm": 0.12263201177120209,
      "learning_rate": 4.430837895005058e-07,
      "loss": 0.07786369323730469,
      "step": 6864
    },
    {
      "epoch": 0.9281260034137191,
      "grad_norm": 0.07072308659553528,
      "learning_rate": 4.4143403372360836e-07,
      "loss": 0.04534858465194702,
      "step": 6865
    },
    {
      "epoch": 0.928261200209555,
      "grad_norm": 0.08886685967445374,
      "learning_rate": 4.3978730911778176e-07,
      "loss": 0.05535316467285156,
      "step": 6866
    },
    {
      "epoch": 0.928396397005391,
      "grad_norm": 0.06547877192497253,
      "learning_rate": 4.381436160258834e-07,
      "loss": 0.059301525354385376,
      "step": 6867
    },
    {
      "epoch": 0.928531593801227,
      "grad_norm": 0.055500734597444534,
      "learning_rate": 4.3650295479013615e-07,
      "loss": 0.04136532545089722,
      "step": 6868
    },
    {
      "epoch": 0.9286667905970628,
      "grad_norm": 0.22133472561836243,
      "learning_rate": 4.348653257521351e-07,
      "loss": 0.07832002639770508,
      "step": 6869
    },
    {
      "epoch": 0.9288019873928988,
      "grad_norm": 0.0687526986002922,
      "learning_rate": 4.332307292528442e-07,
      "loss": 0.057385802268981934,
      "step": 6870
    },
    {
      "epoch": 0.9289371841887347,
      "grad_norm": 0.1347644031047821,
      "learning_rate": 4.315991656325946e-07,
      "loss": 0.08509588241577148,
      "step": 6871
    },
    {
      "epoch": 0.9290723809845707,
      "grad_norm": 0.10149525851011276,
      "learning_rate": 4.299706352310895e-07,
      "loss": 0.08035612106323242,
      "step": 6872
    },
    {
      "epoch": 0.9292075777804066,
      "grad_norm": 0.11910875886678696,
      "learning_rate": 4.283451383873926e-07,
      "loss": 0.07304835319519043,
      "step": 6873
    },
    {
      "epoch": 0.9293427745762426,
      "grad_norm": 0.10757680982351303,
      "learning_rate": 4.26722675439945e-07,
      "loss": 0.0601048469543457,
      "step": 6874
    },
    {
      "epoch": 0.9294779713720784,
      "grad_norm": 0.130685493350029,
      "learning_rate": 4.251032467265481e-07,
      "loss": 0.08267974853515625,
      "step": 6875
    },
    {
      "epoch": 0.9296131681679144,
      "grad_norm": 0.07935380935668945,
      "learning_rate": 4.234868525843805e-07,
      "loss": 0.05322980880737305,
      "step": 6876
    },
    {
      "epoch": 0.9297483649637504,
      "grad_norm": 0.12427707016468048,
      "learning_rate": 4.218734933499796e-07,
      "loss": 0.06458759307861328,
      "step": 6877
    },
    {
      "epoch": 0.9298835617595863,
      "grad_norm": 0.08099223673343658,
      "learning_rate": 4.202631693592601e-07,
      "loss": 0.04376697540283203,
      "step": 6878
    },
    {
      "epoch": 0.9300187585554223,
      "grad_norm": 0.051493365317583084,
      "learning_rate": 4.186558809474955e-07,
      "loss": 0.055303096771240234,
      "step": 6879
    },
    {
      "epoch": 0.9301539553512582,
      "grad_norm": 0.0659814402461052,
      "learning_rate": 4.170516284493331e-07,
      "loss": 0.06276798248291016,
      "step": 6880
    },
    {
      "epoch": 0.9302891521470941,
      "grad_norm": 0.13775216042995453,
      "learning_rate": 4.1545041219879063e-07,
      "loss": 0.044480085372924805,
      "step": 6881
    },
    {
      "epoch": 0.93042434894293,
      "grad_norm": 0.055290453135967255,
      "learning_rate": 4.138522325292432e-07,
      "loss": 0.03249400854110718,
      "step": 6882
    },
    {
      "epoch": 0.930559545738766,
      "grad_norm": 0.039650361984968185,
      "learning_rate": 4.1225708977344457e-07,
      "loss": 0.027518153190612793,
      "step": 6883
    },
    {
      "epoch": 0.9306947425346019,
      "grad_norm": 0.12958823144435883,
      "learning_rate": 4.106649842635124e-07,
      "loss": 0.06265556812286377,
      "step": 6884
    },
    {
      "epoch": 0.9308299393304379,
      "grad_norm": 0.043656568974256516,
      "learning_rate": 4.090759163309282e-07,
      "loss": 0.043683528900146484,
      "step": 6885
    },
    {
      "epoch": 0.9309651361262739,
      "grad_norm": 0.07034581899642944,
      "learning_rate": 4.07489886306549e-07,
      "loss": 0.05775117874145508,
      "step": 6886
    },
    {
      "epoch": 0.9311003329221097,
      "grad_norm": 0.06452146917581558,
      "learning_rate": 4.059068945205907e-07,
      "loss": 0.034664273262023926,
      "step": 6887
    },
    {
      "epoch": 0.9312355297179457,
      "grad_norm": 0.05761135369539261,
      "learning_rate": 4.043269413026429e-07,
      "loss": 0.03112006187438965,
      "step": 6888
    },
    {
      "epoch": 0.9313707265137816,
      "grad_norm": 0.1013440415263176,
      "learning_rate": 4.027500269816592e-07,
      "loss": 0.05026751756668091,
      "step": 6889
    },
    {
      "epoch": 0.9315059233096176,
      "grad_norm": 0.07188749313354492,
      "learning_rate": 4.011761518859619e-07,
      "loss": 0.05354803800582886,
      "step": 6890
    },
    {
      "epoch": 0.9316411201054535,
      "grad_norm": 0.10590389370918274,
      "learning_rate": 3.996053163432406e-07,
      "loss": 0.05647069215774536,
      "step": 6891
    },
    {
      "epoch": 0.9317763169012895,
      "grad_norm": 0.060343142598867416,
      "learning_rate": 3.980375206805503e-07,
      "loss": 0.04913735389709473,
      "step": 6892
    },
    {
      "epoch": 0.9319115136971253,
      "grad_norm": 0.06492726504802704,
      "learning_rate": 3.9647276522431664e-07,
      "loss": 0.05971336364746094,
      "step": 6893
    },
    {
      "epoch": 0.9320467104929613,
      "grad_norm": 0.11738261580467224,
      "learning_rate": 3.949110503003289e-07,
      "loss": 0.05980730056762695,
      "step": 6894
    },
    {
      "epoch": 0.9321819072887972,
      "grad_norm": 0.06924591213464737,
      "learning_rate": 3.9335237623374377e-07,
      "loss": 0.0476231575012207,
      "step": 6895
    },
    {
      "epoch": 0.9323171040846332,
      "grad_norm": 0.08711355179548264,
      "learning_rate": 3.917967433490849e-07,
      "loss": 0.08286428451538086,
      "step": 6896
    },
    {
      "epoch": 0.9324523008804692,
      "grad_norm": 0.04757753759622574,
      "learning_rate": 3.902441519702449e-07,
      "loss": 0.050853729248046875,
      "step": 6897
    },
    {
      "epoch": 0.9325874976763051,
      "grad_norm": 0.059648267924785614,
      "learning_rate": 3.886946024204818e-07,
      "loss": 0.052712440490722656,
      "step": 6898
    },
    {
      "epoch": 0.932722694472141,
      "grad_norm": 0.07008446007966995,
      "learning_rate": 3.871480950224193e-07,
      "loss": 0.058917999267578125,
      "step": 6899
    },
    {
      "epoch": 0.9328578912679769,
      "grad_norm": 0.028207402676343918,
      "learning_rate": 3.856046300980498e-07,
      "loss": 0.032725900411605835,
      "step": 6900
    },
    {
      "epoch": 0.9329930880638129,
      "grad_norm": 0.11810152232646942,
      "learning_rate": 3.8406420796872953e-07,
      "loss": 0.0660848617553711,
      "step": 6901
    },
    {
      "epoch": 0.9331282848596488,
      "grad_norm": 0.07528826594352722,
      "learning_rate": 3.825268289551803e-07,
      "loss": 0.04598945379257202,
      "step": 6902
    },
    {
      "epoch": 0.9332634816554848,
      "grad_norm": 0.06875009089708328,
      "learning_rate": 3.8099249337749777e-07,
      "loss": 0.050786495208740234,
      "step": 6903
    },
    {
      "epoch": 0.9333986784513207,
      "grad_norm": 0.029056215658783913,
      "learning_rate": 3.7946120155513465e-07,
      "loss": 0.0377044677734375,
      "step": 6904
    },
    {
      "epoch": 0.9335338752471567,
      "grad_norm": 0.08003580570220947,
      "learning_rate": 3.7793295380691595e-07,
      "loss": 0.05276012420654297,
      "step": 6905
    },
    {
      "epoch": 0.9336690720429925,
      "grad_norm": 0.05763528496026993,
      "learning_rate": 3.7640775045103214e-07,
      "loss": 0.052684903144836426,
      "step": 6906
    },
    {
      "epoch": 0.9338042688388285,
      "grad_norm": 0.08067308366298676,
      "learning_rate": 3.7488559180503423e-07,
      "loss": 0.039466023445129395,
      "step": 6907
    },
    {
      "epoch": 0.9339394656346645,
      "grad_norm": 0.11801900714635849,
      "learning_rate": 3.7336647818584866e-07,
      "loss": 0.07549607753753662,
      "step": 6908
    },
    {
      "epoch": 0.9340746624305004,
      "grad_norm": 0.1064031645655632,
      "learning_rate": 3.718504099097625e-07,
      "loss": 0.057752132415771484,
      "step": 6909
    },
    {
      "epoch": 0.9342098592263364,
      "grad_norm": 0.071253702044487,
      "learning_rate": 3.703373872924265e-07,
      "loss": 0.04210996627807617,
      "step": 6910
    },
    {
      "epoch": 0.9343450560221723,
      "grad_norm": 0.05311444774270058,
      "learning_rate": 3.688274106488604e-07,
      "loss": 0.04536402225494385,
      "step": 6911
    },
    {
      "epoch": 0.9344802528180082,
      "grad_norm": 0.11050664633512497,
      "learning_rate": 3.67320480293451e-07,
      "loss": 0.07140803337097168,
      "step": 6912
    },
    {
      "epoch": 0.9346154496138441,
      "grad_norm": 0.12018724530935287,
      "learning_rate": 3.6581659653994736e-07,
      "loss": 0.08607006072998047,
      "step": 6913
    },
    {
      "epoch": 0.9347506464096801,
      "grad_norm": 0.050158578902482986,
      "learning_rate": 3.64315759701469e-07,
      "loss": 0.037581801414489746,
      "step": 6914
    },
    {
      "epoch": 0.934885843205516,
      "grad_norm": 0.04933745041489601,
      "learning_rate": 3.6281797009049765e-07,
      "loss": 0.03782397508621216,
      "step": 6915
    },
    {
      "epoch": 0.935021040001352,
      "grad_norm": 0.04649464040994644,
      "learning_rate": 3.613232280188772e-07,
      "loss": 0.034948378801345825,
      "step": 6916
    },
    {
      "epoch": 0.935156236797188,
      "grad_norm": 0.08013904094696045,
      "learning_rate": 3.5983153379782363e-07,
      "loss": 0.06444597244262695,
      "step": 6917
    },
    {
      "epoch": 0.9352914335930238,
      "grad_norm": 0.08652915805578232,
      "learning_rate": 3.5834288773791854e-07,
      "loss": 0.05016756057739258,
      "step": 6918
    },
    {
      "epoch": 0.9354266303888598,
      "grad_norm": 0.11209945380687714,
      "learning_rate": 3.568572901491007e-07,
      "loss": 0.06284534931182861,
      "step": 6919
    },
    {
      "epoch": 0.9355618271846957,
      "grad_norm": 0.08809870481491089,
      "learning_rate": 3.553747413406827e-07,
      "loss": 0.062105655670166016,
      "step": 6920
    },
    {
      "epoch": 0.9356970239805317,
      "grad_norm": 0.10317903012037277,
      "learning_rate": 3.538952416213376e-07,
      "loss": 0.0722661018371582,
      "step": 6921
    },
    {
      "epoch": 0.9358322207763676,
      "grad_norm": 0.09842576831579208,
      "learning_rate": 3.524187912991056e-07,
      "loss": 0.046496033668518066,
      "step": 6922
    },
    {
      "epoch": 0.9359674175722036,
      "grad_norm": 0.07721572369337082,
      "learning_rate": 3.5094539068139254e-07,
      "loss": 0.06457924842834473,
      "step": 6923
    },
    {
      "epoch": 0.9361026143680394,
      "grad_norm": 0.10088503360748291,
      "learning_rate": 3.494750400749663e-07,
      "loss": 0.06267070770263672,
      "step": 6924
    },
    {
      "epoch": 0.9362378111638754,
      "grad_norm": 0.09097261726856232,
      "learning_rate": 3.480077397859638e-07,
      "loss": 0.07631969451904297,
      "step": 6925
    },
    {
      "epoch": 0.9363730079597113,
      "grad_norm": 0.07021842896938324,
      "learning_rate": 3.4654349011988384e-07,
      "loss": 0.05778908729553223,
      "step": 6926
    },
    {
      "epoch": 0.9365082047555473,
      "grad_norm": 0.14088305830955505,
      "learning_rate": 3.4508229138159095e-07,
      "loss": 0.07998228073120117,
      "step": 6927
    },
    {
      "epoch": 0.9366434015513833,
      "grad_norm": 0.061945971101522446,
      "learning_rate": 3.4362414387531516e-07,
      "loss": 0.05021810531616211,
      "step": 6928
    },
    {
      "epoch": 0.9367785983472192,
      "grad_norm": 0.1050914004445076,
      "learning_rate": 3.4216904790464854e-07,
      "loss": 0.053131103515625,
      "step": 6929
    },
    {
      "epoch": 0.9369137951430551,
      "grad_norm": 0.05285362899303436,
      "learning_rate": 3.407170037725521e-07,
      "loss": 0.049791574478149414,
      "step": 6930
    },
    {
      "epoch": 0.937048991938891,
      "grad_norm": 0.07455403357744217,
      "learning_rate": 3.3926801178134737e-07,
      "loss": 0.058953285217285156,
      "step": 6931
    },
    {
      "epoch": 0.937184188734727,
      "grad_norm": 0.16067960858345032,
      "learning_rate": 3.3782207223272467e-07,
      "loss": 0.06723785400390625,
      "step": 6932
    },
    {
      "epoch": 0.9373193855305629,
      "grad_norm": 0.11488193273544312,
      "learning_rate": 3.363791854277348e-07,
      "loss": 0.07103681564331055,
      "step": 6933
    },
    {
      "epoch": 0.9374545823263989,
      "grad_norm": 0.10198970884084702,
      "learning_rate": 3.349393516667926e-07,
      "loss": 0.07474517822265625,
      "step": 6934
    },
    {
      "epoch": 0.9375897791222348,
      "grad_norm": 0.06418319046497345,
      "learning_rate": 3.335025712496814e-07,
      "loss": 0.05176424980163574,
      "step": 6935
    },
    {
      "epoch": 0.9377249759180707,
      "grad_norm": 0.07827525585889816,
      "learning_rate": 3.320688444755471e-07,
      "loss": 0.05374348163604736,
      "step": 6936
    },
    {
      "epoch": 0.9378601727139066,
      "grad_norm": 0.12284183502197266,
      "learning_rate": 3.306381716428991e-07,
      "loss": 0.08468794822692871,
      "step": 6937
    },
    {
      "epoch": 0.9379953695097426,
      "grad_norm": 0.09308009594678879,
      "learning_rate": 3.2921055304960925e-07,
      "loss": 0.05273151397705078,
      "step": 6938
    },
    {
      "epoch": 0.9381305663055786,
      "grad_norm": 0.11022285372018814,
      "learning_rate": 3.277859889929147e-07,
      "loss": 0.053481101989746094,
      "step": 6939
    },
    {
      "epoch": 0.9382657631014145,
      "grad_norm": 0.1451026350259781,
      "learning_rate": 3.263644797694215e-07,
      "loss": 0.08204197883605957,
      "step": 6940
    },
    {
      "epoch": 0.9384009598972505,
      "grad_norm": 0.057752303779125214,
      "learning_rate": 3.2494602567509303e-07,
      "loss": 0.031233549118041992,
      "step": 6941
    },
    {
      "epoch": 0.9385361566930863,
      "grad_norm": 0.07729972153902054,
      "learning_rate": 3.2353062700525794e-07,
      "loss": 0.060292959213256836,
      "step": 6942
    },
    {
      "epoch": 0.9386713534889223,
      "grad_norm": 0.0540657602250576,
      "learning_rate": 3.221182840546122e-07,
      "loss": 0.04839235544204712,
      "step": 6943
    },
    {
      "epoch": 0.9388065502847582,
      "grad_norm": 0.08703777194023132,
      "learning_rate": 3.207089971172089e-07,
      "loss": 0.06039083003997803,
      "step": 6944
    },
    {
      "epoch": 0.9389417470805942,
      "grad_norm": 0.0702759325504303,
      "learning_rate": 3.1930276648647504e-07,
      "loss": 0.05310368537902832,
      "step": 6945
    },
    {
      "epoch": 0.9390769438764301,
      "grad_norm": 0.05793435126543045,
      "learning_rate": 3.178995924551914e-07,
      "loss": 0.0605621337890625,
      "step": 6946
    },
    {
      "epoch": 0.9392121406722661,
      "grad_norm": 0.18370160460472107,
      "learning_rate": 3.164994753155059e-07,
      "loss": 0.051070213317871094,
      "step": 6947
    },
    {
      "epoch": 0.939347337468102,
      "grad_norm": 0.1674375832080841,
      "learning_rate": 3.1510241535893215e-07,
      "loss": 0.07889413833618164,
      "step": 6948
    },
    {
      "epoch": 0.9394825342639379,
      "grad_norm": 0.10024214535951614,
      "learning_rate": 3.1370841287634567e-07,
      "loss": 0.04334592819213867,
      "step": 6949
    },
    {
      "epoch": 0.9396177310597739,
      "grad_norm": 0.06585827469825745,
      "learning_rate": 3.1231746815798436e-07,
      "loss": 0.05802345275878906,
      "step": 6950
    },
    {
      "epoch": 0.9397529278556098,
      "grad_norm": 0.08680598437786102,
      "learning_rate": 3.1092958149344985e-07,
      "loss": 0.0465472936630249,
      "step": 6951
    },
    {
      "epoch": 0.9398881246514458,
      "grad_norm": 0.12279269844293594,
      "learning_rate": 3.095447531717077e-07,
      "loss": 0.07993555068969727,
      "step": 6952
    },
    {
      "epoch": 0.9400233214472817,
      "grad_norm": 0.11014531552791595,
      "learning_rate": 3.08162983481089e-07,
      "loss": 0.08346843719482422,
      "step": 6953
    },
    {
      "epoch": 0.9401585182431177,
      "grad_norm": 0.04653956741094589,
      "learning_rate": 3.067842727092801e-07,
      "loss": 0.043128013610839844,
      "step": 6954
    },
    {
      "epoch": 0.9402937150389535,
      "grad_norm": 0.07728352397680283,
      "learning_rate": 3.0540862114334323e-07,
      "loss": 0.0655207633972168,
      "step": 6955
    },
    {
      "epoch": 0.9404289118347895,
      "grad_norm": 0.07413399964570999,
      "learning_rate": 3.0403602906969086e-07,
      "loss": 0.07045590877532959,
      "step": 6956
    },
    {
      "epoch": 0.9405641086306255,
      "grad_norm": 0.07928022742271423,
      "learning_rate": 3.0266649677410605e-07,
      "loss": 0.06980705261230469,
      "step": 6957
    },
    {
      "epoch": 0.9406993054264614,
      "grad_norm": 0.09311087429523468,
      "learning_rate": 3.0130002454173243e-07,
      "loss": 0.06402349472045898,
      "step": 6958
    },
    {
      "epoch": 0.9408345022222974,
      "grad_norm": 0.10275434702634811,
      "learning_rate": 2.9993661265707407e-07,
      "loss": 0.07254600524902344,
      "step": 6959
    },
    {
      "epoch": 0.9409696990181333,
      "grad_norm": 0.04765182361006737,
      "learning_rate": 2.985762614040072e-07,
      "loss": 0.034528255462646484,
      "step": 6960
    },
    {
      "epoch": 0.9411048958139692,
      "grad_norm": 0.08414024859666824,
      "learning_rate": 2.972189710657586e-07,
      "loss": 0.054224252700805664,
      "step": 6961
    },
    {
      "epoch": 0.9412400926098051,
      "grad_norm": 0.05579359829425812,
      "learning_rate": 2.958647419249255e-07,
      "loss": 0.0525822639465332,
      "step": 6962
    },
    {
      "epoch": 0.9413752894056411,
      "grad_norm": 0.06046384572982788,
      "learning_rate": 2.9451357426346415e-07,
      "loss": 0.05457305908203125,
      "step": 6963
    },
    {
      "epoch": 0.941510486201477,
      "grad_norm": 0.04738999530673027,
      "learning_rate": 2.9316546836269776e-07,
      "loss": 0.030657470226287842,
      "step": 6964
    },
    {
      "epoch": 0.941645682997313,
      "grad_norm": 0.08934573084115982,
      "learning_rate": 2.9182042450330516e-07,
      "loss": 0.0614393949508667,
      "step": 6965
    },
    {
      "epoch": 0.941780879793149,
      "grad_norm": 0.06653844565153122,
      "learning_rate": 2.9047844296533397e-07,
      "loss": 0.049520254135131836,
      "step": 6966
    },
    {
      "epoch": 0.9419160765889848,
      "grad_norm": 0.06392962485551834,
      "learning_rate": 2.8913952402819246e-07,
      "loss": 0.06250190734863281,
      "step": 6967
    },
    {
      "epoch": 0.9420512733848208,
      "grad_norm": 0.06784596294164658,
      "learning_rate": 2.878036679706492e-07,
      "loss": 0.0459897518157959,
      "step": 6968
    },
    {
      "epoch": 0.9421864701806567,
      "grad_norm": 0.061562154442071915,
      "learning_rate": 2.8647087507083837e-07,
      "loss": 0.05291271209716797,
      "step": 6969
    },
    {
      "epoch": 0.9423216669764927,
      "grad_norm": 0.058297812938690186,
      "learning_rate": 2.8514114560625303e-07,
      "loss": 0.05229926109313965,
      "step": 6970
    },
    {
      "epoch": 0.9424568637723286,
      "grad_norm": 0.0412035770714283,
      "learning_rate": 2.8381447985375007e-07,
      "loss": 0.037505149841308594,
      "step": 6971
    },
    {
      "epoch": 0.9425920605681646,
      "grad_norm": 0.050766970962285995,
      "learning_rate": 2.8249087808954853e-07,
      "loss": 0.03810882568359375,
      "step": 6972
    },
    {
      "epoch": 0.9427272573640004,
      "grad_norm": 0.07791668176651001,
      "learning_rate": 2.811703405892296e-07,
      "loss": 0.07019102573394775,
      "step": 6973
    },
    {
      "epoch": 0.9428624541598364,
      "grad_norm": 0.054396599531173706,
      "learning_rate": 2.798528676277368e-07,
      "loss": 0.04919242858886719,
      "step": 6974
    },
    {
      "epoch": 0.9429976509556723,
      "grad_norm": 0.07416316121816635,
      "learning_rate": 2.785384594793738e-07,
      "loss": 0.05912590026855469,
      "step": 6975
    },
    {
      "epoch": 0.9431328477515083,
      "grad_norm": 0.08746141940355301,
      "learning_rate": 2.772271164178086e-07,
      "loss": 0.044594764709472656,
      "step": 6976
    },
    {
      "epoch": 0.9432680445473443,
      "grad_norm": 0.07813379168510437,
      "learning_rate": 2.759188387160677e-07,
      "loss": 0.05049169063568115,
      "step": 6977
    },
    {
      "epoch": 0.9434032413431802,
      "grad_norm": 0.04530896618962288,
      "learning_rate": 2.746136266465449e-07,
      "loss": 0.03863561153411865,
      "step": 6978
    },
    {
      "epoch": 0.9435384381390161,
      "grad_norm": 0.06993469595909119,
      "learning_rate": 2.7331148048098943e-07,
      "loss": 0.054784297943115234,
      "step": 6979
    },
    {
      "epoch": 0.943673634934852,
      "grad_norm": 0.06468288600444794,
      "learning_rate": 2.7201240049051613e-07,
      "loss": 0.05142354965209961,
      "step": 6980
    },
    {
      "epoch": 0.943808831730688,
      "grad_norm": 0.043773747980594635,
      "learning_rate": 2.707163869455986e-07,
      "loss": 0.03817009925842285,
      "step": 6981
    },
    {
      "epoch": 0.9439440285265239,
      "grad_norm": 0.11098359525203705,
      "learning_rate": 2.694234401160778e-07,
      "loss": 0.07822735607624054,
      "step": 6982
    },
    {
      "epoch": 0.9440792253223599,
      "grad_norm": 0.12570399045944214,
      "learning_rate": 2.6813356027114986e-07,
      "loss": 0.08304531872272491,
      "step": 6983
    },
    {
      "epoch": 0.9442144221181958,
      "grad_norm": 0.0534651055932045,
      "learning_rate": 2.6684674767937346e-07,
      "loss": 0.04551506042480469,
      "step": 6984
    },
    {
      "epoch": 0.9443496189140317,
      "grad_norm": 0.13213485479354858,
      "learning_rate": 2.655630026086708e-07,
      "loss": 0.09651803970336914,
      "step": 6985
    },
    {
      "epoch": 0.9444848157098676,
      "grad_norm": 0.06385689973831177,
      "learning_rate": 2.642823253263249e-07,
      "loss": 0.054630815982818604,
      "step": 6986
    },
    {
      "epoch": 0.9446200125057036,
      "grad_norm": 0.07354943454265594,
      "learning_rate": 2.630047160989807e-07,
      "loss": 0.0584794282913208,
      "step": 6987
    },
    {
      "epoch": 0.9447552093015396,
      "grad_norm": 0.04315639287233353,
      "learning_rate": 2.6173017519263875e-07,
      "loss": 0.04550975561141968,
      "step": 6988
    },
    {
      "epoch": 0.9448904060973755,
      "grad_norm": 0.0740494504570961,
      "learning_rate": 2.6045870287267014e-07,
      "loss": 0.053692519664764404,
      "step": 6989
    },
    {
      "epoch": 0.9450256028932115,
      "grad_norm": 0.05050640553236008,
      "learning_rate": 2.5919029940380147e-07,
      "loss": 0.03628945350646973,
      "step": 6990
    },
    {
      "epoch": 0.9451607996890473,
      "grad_norm": 0.09610525518655777,
      "learning_rate": 2.5792496505011807e-07,
      "loss": 0.08017885684967041,
      "step": 6991
    },
    {
      "epoch": 0.9452959964848833,
      "grad_norm": 0.14941200613975525,
      "learning_rate": 2.5666270007507266e-07,
      "loss": 0.07337033748626709,
      "step": 6992
    },
    {
      "epoch": 0.9454311932807192,
      "grad_norm": 0.057521216571331024,
      "learning_rate": 2.5540350474147324e-07,
      "loss": 0.04524034261703491,
      "step": 6993
    },
    {
      "epoch": 0.9455663900765552,
      "grad_norm": 0.05158543586730957,
      "learning_rate": 2.5414737931149346e-07,
      "loss": 0.03907042741775513,
      "step": 6994
    },
    {
      "epoch": 0.9457015868723911,
      "grad_norm": 0.06468634307384491,
      "learning_rate": 2.5289432404666246e-07,
      "loss": 0.05111527442932129,
      "step": 6995
    },
    {
      "epoch": 0.9458367836682271,
      "grad_norm": 0.07391882687807083,
      "learning_rate": 2.5164433920787487e-07,
      "loss": 0.07205140590667725,
      "step": 6996
    },
    {
      "epoch": 0.9459719804640631,
      "grad_norm": 0.03957660496234894,
      "learning_rate": 2.503974250553842e-07,
      "loss": 0.04009866714477539,
      "step": 6997
    },
    {
      "epoch": 0.9461071772598989,
      "grad_norm": 0.06391185522079468,
      "learning_rate": 2.491535818488011e-07,
      "loss": 0.04989814758300781,
      "step": 6998
    },
    {
      "epoch": 0.9462423740557349,
      "grad_norm": 0.16935226321220398,
      "learning_rate": 2.479128098471067e-07,
      "loss": 0.060219258069992065,
      "step": 6999
    },
    {
      "epoch": 0.9463775708515708,
      "grad_norm": 0.08015145361423492,
      "learning_rate": 2.466751093086328e-07,
      "loss": 0.05099630355834961,
      "step": 7000
    },
    {
      "epoch": 0.9465127676474068,
      "grad_norm": 0.07341624796390533,
      "learning_rate": 2.454404804910748e-07,
      "loss": 0.050408005714416504,
      "step": 7001
    },
    {
      "epoch": 0.9466479644432427,
      "grad_norm": 0.045486945658922195,
      "learning_rate": 2.442089236514888e-07,
      "loss": 0.03636839985847473,
      "step": 7002
    },
    {
      "epoch": 0.9467831612390787,
      "grad_norm": 0.06731517612934113,
      "learning_rate": 2.429804390462931e-07,
      "loss": 0.06496882438659668,
      "step": 7003
    },
    {
      "epoch": 0.9469183580349145,
      "grad_norm": 0.057919710874557495,
      "learning_rate": 2.4175502693126293e-07,
      "loss": 0.04881696403026581,
      "step": 7004
    },
    {
      "epoch": 0.9470535548307505,
      "grad_norm": 0.10030993819236755,
      "learning_rate": 2.4053268756153933e-07,
      "loss": 0.049046725034713745,
      "step": 7005
    },
    {
      "epoch": 0.9471887516265864,
      "grad_norm": 0.11372458189725876,
      "learning_rate": 2.393134211916154e-07,
      "loss": 0.09297668933868408,
      "step": 7006
    },
    {
      "epoch": 0.9473239484224224,
      "grad_norm": 0.13003067672252655,
      "learning_rate": 2.3809722807535128e-07,
      "loss": 0.05327713489532471,
      "step": 7007
    },
    {
      "epoch": 0.9474591452182584,
      "grad_norm": 0.13583219051361084,
      "learning_rate": 2.3688410846596287e-07,
      "loss": 0.06853389739990234,
      "step": 7008
    },
    {
      "epoch": 0.9475943420140943,
      "grad_norm": 0.08847765624523163,
      "learning_rate": 2.3567406261603143e-07,
      "loss": 0.05989217758178711,
      "step": 7009
    },
    {
      "epoch": 0.9477295388099302,
      "grad_norm": 0.05565788596868515,
      "learning_rate": 2.3446709077749206e-07,
      "loss": 0.0533231645822525,
      "step": 7010
    },
    {
      "epoch": 0.9478647356057661,
      "grad_norm": 0.07357075065374374,
      "learning_rate": 2.3326319320164546e-07,
      "loss": 0.08249294757843018,
      "step": 7011
    },
    {
      "epoch": 0.9479999324016021,
      "grad_norm": 0.11841505765914917,
      "learning_rate": 2.320623701391461e-07,
      "loss": 0.05325794219970703,
      "step": 7012
    },
    {
      "epoch": 0.948135129197438,
      "grad_norm": 0.04237016290426254,
      "learning_rate": 2.30864621840014e-07,
      "loss": 0.04903078079223633,
      "step": 7013
    },
    {
      "epoch": 0.948270325993274,
      "grad_norm": 0.04357681795954704,
      "learning_rate": 2.2966994855362633e-07,
      "loss": 0.043715476989746094,
      "step": 7014
    },
    {
      "epoch": 0.94840552278911,
      "grad_norm": 0.07724309712648392,
      "learning_rate": 2.2847835052872079e-07,
      "loss": 0.053927481174468994,
      "step": 7015
    },
    {
      "epoch": 0.9485407195849458,
      "grad_norm": 0.05171968787908554,
      "learning_rate": 2.2728982801339392e-07,
      "loss": 0.04001206159591675,
      "step": 7016
    },
    {
      "epoch": 0.9486759163807817,
      "grad_norm": 0.05697864294052124,
      "learning_rate": 2.261043812551028e-07,
      "loss": 0.04075813293457031,
      "step": 7017
    },
    {
      "epoch": 0.9488111131766177,
      "grad_norm": 0.08187760412693024,
      "learning_rate": 2.249220105006633e-07,
      "loss": 0.05548381805419922,
      "step": 7018
    },
    {
      "epoch": 0.9489463099724537,
      "grad_norm": 0.07538948953151703,
      "learning_rate": 2.2374271599625185e-07,
      "loss": 0.0672069787979126,
      "step": 7019
    },
    {
      "epoch": 0.9490815067682896,
      "grad_norm": 0.08061212301254272,
      "learning_rate": 2.2256649798740204e-07,
      "loss": 0.04513835906982422,
      "step": 7020
    },
    {
      "epoch": 0.9492167035641256,
      "grad_norm": 0.16609831154346466,
      "learning_rate": 2.2139335671901294e-07,
      "loss": 0.09791260957717896,
      "step": 7021
    },
    {
      "epoch": 0.9493519003599614,
      "grad_norm": 0.03898888826370239,
      "learning_rate": 2.2022329243533422e-07,
      "loss": 0.033918142318725586,
      "step": 7022
    },
    {
      "epoch": 0.9494870971557974,
      "grad_norm": 0.07824122160673141,
      "learning_rate": 2.19056305379981e-07,
      "loss": 0.049491167068481445,
      "step": 7023
    },
    {
      "epoch": 0.9496222939516333,
      "grad_norm": 0.09263777732849121,
      "learning_rate": 2.178923957959289e-07,
      "loss": 0.05721890926361084,
      "step": 7024
    },
    {
      "epoch": 0.9497574907474693,
      "grad_norm": 0.05878986418247223,
      "learning_rate": 2.1673156392550408e-07,
      "loss": 0.05939221382141113,
      "step": 7025
    },
    {
      "epoch": 0.9498926875433052,
      "grad_norm": 0.16350974142551422,
      "learning_rate": 2.155738100104049e-07,
      "loss": 0.08903127908706665,
      "step": 7026
    },
    {
      "epoch": 0.9500278843391412,
      "grad_norm": 0.05151699110865593,
      "learning_rate": 2.1441913429167682e-07,
      "loss": 0.02555510401725769,
      "step": 7027
    },
    {
      "epoch": 0.950163081134977,
      "grad_norm": 0.1628103256225586,
      "learning_rate": 2.1326753700973256e-07,
      "loss": 0.0980062484741211,
      "step": 7028
    },
    {
      "epoch": 0.950298277930813,
      "grad_norm": 0.15239852666854858,
      "learning_rate": 2.1211901840434034e-07,
      "loss": 0.0710453987121582,
      "step": 7029
    },
    {
      "epoch": 0.950433474726649,
      "grad_norm": 0.06128983572125435,
      "learning_rate": 2.1097357871462386e-07,
      "loss": 0.037694454193115234,
      "step": 7030
    },
    {
      "epoch": 0.9505686715224849,
      "grad_norm": 0.08963966369628906,
      "learning_rate": 2.098312181790757e-07,
      "loss": 0.06391239166259766,
      "step": 7031
    },
    {
      "epoch": 0.9507038683183209,
      "grad_norm": 0.0813160240650177,
      "learning_rate": 2.086919370355389e-07,
      "loss": 0.06886947154998779,
      "step": 7032
    },
    {
      "epoch": 0.9508390651141568,
      "grad_norm": 0.08002682775259018,
      "learning_rate": 2.075557355212171e-07,
      "loss": 0.06606996059417725,
      "step": 7033
    },
    {
      "epoch": 0.9509742619099927,
      "grad_norm": 0.1222519651055336,
      "learning_rate": 2.0642261387267268e-07,
      "loss": 0.06932210922241211,
      "step": 7034
    },
    {
      "epoch": 0.9511094587058286,
      "grad_norm": 0.03813016042113304,
      "learning_rate": 2.0529257232583033e-07,
      "loss": 0.03692770004272461,
      "step": 7035
    },
    {
      "epoch": 0.9512446555016646,
      "grad_norm": 0.05937309190630913,
      "learning_rate": 2.0416561111596844e-07,
      "loss": 0.044332027435302734,
      "step": 7036
    },
    {
      "epoch": 0.9513798522975006,
      "grad_norm": 0.07218378037214279,
      "learning_rate": 2.0304173047772933e-07,
      "loss": 0.05002550780773163,
      "step": 7037
    },
    {
      "epoch": 0.9515150490933365,
      "grad_norm": 0.06665070354938507,
      "learning_rate": 2.0192093064510753e-07,
      "loss": 0.04206490516662598,
      "step": 7038
    },
    {
      "epoch": 0.9516502458891725,
      "grad_norm": 0.038208600133657455,
      "learning_rate": 2.0080321185146134e-07,
      "loss": 0.027156829833984375,
      "step": 7039
    },
    {
      "epoch": 0.9517854426850084,
      "grad_norm": 0.07616526633501053,
      "learning_rate": 1.996885743295046e-07,
      "loss": 0.06332600116729736,
      "step": 7040
    },
    {
      "epoch": 0.9519206394808443,
      "grad_norm": 0.08870632946491241,
      "learning_rate": 1.985770183113117e-07,
      "loss": 0.060047268867492676,
      "step": 7041
    },
    {
      "epoch": 0.9520558362766802,
      "grad_norm": 0.05216154828667641,
      "learning_rate": 1.9746854402831583e-07,
      "loss": 0.0407564640045166,
      "step": 7042
    },
    {
      "epoch": 0.9521910330725162,
      "grad_norm": 0.09871078282594681,
      "learning_rate": 1.963631517113057e-07,
      "loss": 0.08051782846450806,
      "step": 7043
    },
    {
      "epoch": 0.9523262298683521,
      "grad_norm": 0.0778815969824791,
      "learning_rate": 1.952608415904289e-07,
      "loss": 0.053549885749816895,
      "step": 7044
    },
    {
      "epoch": 0.9524614266641881,
      "grad_norm": 0.05686617270112038,
      "learning_rate": 1.9416161389519348e-07,
      "loss": 0.04177820682525635,
      "step": 7045
    },
    {
      "epoch": 0.952596623460024,
      "grad_norm": 0.05548636242747307,
      "learning_rate": 1.9306546885446475e-07,
      "loss": 0.04991555213928223,
      "step": 7046
    },
    {
      "epoch": 0.9527318202558599,
      "grad_norm": 0.09127151221036911,
      "learning_rate": 1.919724066964651e-07,
      "loss": 0.04975247383117676,
      "step": 7047
    },
    {
      "epoch": 0.9528670170516959,
      "grad_norm": 0.0802159234881401,
      "learning_rate": 1.908824276487775e-07,
      "loss": 0.05361223220825195,
      "step": 7048
    },
    {
      "epoch": 0.9530022138475318,
      "grad_norm": 0.0906638652086258,
      "learning_rate": 1.8979553193833876e-07,
      "loss": 0.0714186429977417,
      "step": 7049
    },
    {
      "epoch": 0.9531374106433678,
      "grad_norm": 0.10248968005180359,
      "learning_rate": 1.8871171979144786e-07,
      "loss": 0.04005098342895508,
      "step": 7050
    },
    {
      "epoch": 0.9532726074392037,
      "grad_norm": 0.06315653771162033,
      "learning_rate": 1.8763099143376262e-07,
      "loss": 0.04148685932159424,
      "step": 7051
    },
    {
      "epoch": 0.9534078042350397,
      "grad_norm": 0.11445050686597824,
      "learning_rate": 1.8655334709029303e-07,
      "loss": 0.0642225444316864,
      "step": 7052
    },
    {
      "epoch": 0.9535430010308755,
      "grad_norm": 0.11468757688999176,
      "learning_rate": 1.8547878698541132e-07,
      "loss": 0.07507681846618652,
      "step": 7053
    },
    {
      "epoch": 0.9536781978267115,
      "grad_norm": 0.08256664872169495,
      "learning_rate": 1.8440731134284684e-07,
      "loss": 0.05866289138793945,
      "step": 7054
    },
    {
      "epoch": 0.9538133946225474,
      "grad_norm": 0.07681503891944885,
      "learning_rate": 1.833389203856861e-07,
      "loss": 0.06564569473266602,
      "step": 7055
    },
    {
      "epoch": 0.9539485914183834,
      "grad_norm": 0.04535171762108803,
      "learning_rate": 1.8227361433637625e-07,
      "loss": 0.04410219192504883,
      "step": 7056
    },
    {
      "epoch": 0.9540837882142194,
      "grad_norm": 0.07874813675880432,
      "learning_rate": 1.812113934167148e-07,
      "loss": 0.05748295783996582,
      "step": 7057
    },
    {
      "epoch": 0.9542189850100553,
      "grad_norm": 0.06298123300075531,
      "learning_rate": 1.8015225784786483e-07,
      "loss": 0.05262655019760132,
      "step": 7058
    },
    {
      "epoch": 0.9543541818058912,
      "grad_norm": 0.1091485545039177,
      "learning_rate": 1.7909620785034663e-07,
      "loss": 0.0746760368347168,
      "step": 7059
    },
    {
      "epoch": 0.9544893786017271,
      "grad_norm": 0.10086274147033691,
      "learning_rate": 1.7804324364402936e-07,
      "loss": 0.04511404037475586,
      "step": 7060
    },
    {
      "epoch": 0.9546245753975631,
      "grad_norm": 0.0677308589220047,
      "learning_rate": 1.769933654481526e-07,
      "loss": 0.05417579412460327,
      "step": 7061
    },
    {
      "epoch": 0.954759772193399,
      "grad_norm": 0.18758665025234222,
      "learning_rate": 1.7594657348129984e-07,
      "loss": 0.07589542865753174,
      "step": 7062
    },
    {
      "epoch": 0.954894968989235,
      "grad_norm": 0.07013189792633057,
      "learning_rate": 1.749028679614234e-07,
      "loss": 0.0611720085144043,
      "step": 7063
    },
    {
      "epoch": 0.9550301657850709,
      "grad_norm": 0.1402924507856369,
      "learning_rate": 1.7386224910582615e-07,
      "loss": 0.06402111053466797,
      "step": 7064
    },
    {
      "epoch": 0.9551653625809068,
      "grad_norm": 0.11010725051164627,
      "learning_rate": 1.728247171311731e-07,
      "loss": 0.10113859176635742,
      "step": 7065
    },
    {
      "epoch": 0.9553005593767427,
      "grad_norm": 0.06461682170629501,
      "learning_rate": 1.7179027225348142e-07,
      "loss": 0.056375741958618164,
      "step": 7066
    },
    {
      "epoch": 0.9554357561725787,
      "grad_norm": 0.11112015694379807,
      "learning_rate": 1.7075891468812722e-07,
      "loss": 0.08144569396972656,
      "step": 7067
    },
    {
      "epoch": 0.9555709529684147,
      "grad_norm": 0.08525556325912476,
      "learning_rate": 1.69730644649847e-07,
      "loss": 0.05287277698516846,
      "step": 7068
    },
    {
      "epoch": 0.9557061497642506,
      "grad_norm": 0.1605376899242401,
      "learning_rate": 1.687054623527312e-07,
      "loss": 0.06701302528381348,
      "step": 7069
    },
    {
      "epoch": 0.9558413465600866,
      "grad_norm": 0.11374951153993607,
      "learning_rate": 1.676833680102291e-07,
      "loss": 0.07854437828063965,
      "step": 7070
    },
    {
      "epoch": 0.9559765433559224,
      "grad_norm": 0.10113842040300369,
      "learning_rate": 1.6666436183514378e-07,
      "loss": 0.049597740173339844,
      "step": 7071
    },
    {
      "epoch": 0.9561117401517584,
      "grad_norm": 0.06007857620716095,
      "learning_rate": 1.6564844403964053e-07,
      "loss": 0.048806190490722656,
      "step": 7072
    },
    {
      "epoch": 0.9562469369475943,
      "grad_norm": 0.04048439860343933,
      "learning_rate": 1.6463561483523682e-07,
      "loss": 0.0379909873008728,
      "step": 7073
    },
    {
      "epoch": 0.9563821337434303,
      "grad_norm": 0.06618840247392654,
      "learning_rate": 1.6362587443281063e-07,
      "loss": 0.05173051357269287,
      "step": 7074
    },
    {
      "epoch": 0.9565173305392662,
      "grad_norm": 0.09910482168197632,
      "learning_rate": 1.626192230425938e-07,
      "loss": 0.05704766511917114,
      "step": 7075
    },
    {
      "epoch": 0.9566525273351022,
      "grad_norm": 0.050393156707286835,
      "learning_rate": 1.6161566087417868e-07,
      "loss": 0.04679848253726959,
      "step": 7076
    },
    {
      "epoch": 0.956787724130938,
      "grad_norm": 0.05030618980526924,
      "learning_rate": 1.6061518813650977e-07,
      "loss": 0.05124664306640625,
      "step": 7077
    },
    {
      "epoch": 0.956922920926774,
      "grad_norm": 0.04689155891537666,
      "learning_rate": 1.5961780503789215e-07,
      "loss": 0.04890674352645874,
      "step": 7078
    },
    {
      "epoch": 0.95705811772261,
      "grad_norm": 0.13164012134075165,
      "learning_rate": 1.5862351178598633e-07,
      "loss": 0.06755131483078003,
      "step": 7079
    },
    {
      "epoch": 0.9571933145184459,
      "grad_norm": 0.06548308581113815,
      "learning_rate": 1.5763230858781008e-07,
      "loss": 0.04771256446838379,
      "step": 7080
    },
    {
      "epoch": 0.9573285113142819,
      "grad_norm": 0.05163513869047165,
      "learning_rate": 1.5664419564973497e-07,
      "loss": 0.0486447811126709,
      "step": 7081
    },
    {
      "epoch": 0.9574637081101178,
      "grad_norm": 0.08825210481882095,
      "learning_rate": 1.5565917317749146e-07,
      "loss": 0.05896306037902832,
      "step": 7082
    },
    {
      "epoch": 0.9575989049059537,
      "grad_norm": 0.053509511053562164,
      "learning_rate": 1.5467724137617046e-07,
      "loss": 0.0438838005065918,
      "step": 7083
    },
    {
      "epoch": 0.9577341017017896,
      "grad_norm": 0.08421257138252258,
      "learning_rate": 1.5369840045021178e-07,
      "loss": 0.08101487159729004,
      "step": 7084
    },
    {
      "epoch": 0.9578692984976256,
      "grad_norm": 0.08983268588781357,
      "learning_rate": 1.5272265060341572e-07,
      "loss": 0.06287050247192383,
      "step": 7085
    },
    {
      "epoch": 0.9580044952934615,
      "grad_norm": 0.06738763302564621,
      "learning_rate": 1.517499920389398e-07,
      "loss": 0.06264352798461914,
      "step": 7086
    },
    {
      "epoch": 0.9581396920892975,
      "grad_norm": 0.0920373946428299,
      "learning_rate": 1.5078042495929534e-07,
      "loss": 0.05201005935668945,
      "step": 7087
    },
    {
      "epoch": 0.9582748888851335,
      "grad_norm": 0.07492781430482864,
      "learning_rate": 1.498139495663542e-07,
      "loss": 0.04308652877807617,
      "step": 7088
    },
    {
      "epoch": 0.9584100856809694,
      "grad_norm": 0.10955853015184402,
      "learning_rate": 1.4885056606133707e-07,
      "loss": 0.06994962692260742,
      "step": 7089
    },
    {
      "epoch": 0.9585452824768053,
      "grad_norm": 0.06498569995164871,
      "learning_rate": 1.478902746448302e-07,
      "loss": 0.0503995418548584,
      "step": 7090
    },
    {
      "epoch": 0.9586804792726412,
      "grad_norm": 0.09128726273775101,
      "learning_rate": 1.469330755167686e-07,
      "loss": 0.06974124908447266,
      "step": 7091
    },
    {
      "epoch": 0.9588156760684772,
      "grad_norm": 0.13307435810565948,
      "learning_rate": 1.4597896887644458e-07,
      "loss": 0.0832834243774414,
      "step": 7092
    },
    {
      "epoch": 0.9589508728643131,
      "grad_norm": 0.07883147150278091,
      "learning_rate": 1.4502795492251418e-07,
      "loss": 0.04476213455200195,
      "step": 7093
    },
    {
      "epoch": 0.9590860696601491,
      "grad_norm": 0.0650823637843132,
      "learning_rate": 1.4408003385297742e-07,
      "loss": 0.029877454042434692,
      "step": 7094
    },
    {
      "epoch": 0.959221266455985,
      "grad_norm": 0.11174848675727844,
      "learning_rate": 1.4313520586519968e-07,
      "loss": 0.0674666166305542,
      "step": 7095
    },
    {
      "epoch": 0.9593564632518209,
      "grad_norm": 0.057634565979242325,
      "learning_rate": 1.4219347115589863e-07,
      "loss": 0.04873943328857422,
      "step": 7096
    },
    {
      "epoch": 0.9594916600476568,
      "grad_norm": 0.09915723651647568,
      "learning_rate": 1.4125482992114914e-07,
      "loss": 0.07328510284423828,
      "step": 7097
    },
    {
      "epoch": 0.9596268568434928,
      "grad_norm": 0.09499924629926682,
      "learning_rate": 1.403192823563798e-07,
      "loss": 0.05839717388153076,
      "step": 7098
    },
    {
      "epoch": 0.9597620536393288,
      "grad_norm": 0.13537491858005524,
      "learning_rate": 1.3938682865637654e-07,
      "loss": 0.08152484893798828,
      "step": 7099
    },
    {
      "epoch": 0.9598972504351647,
      "grad_norm": 0.10001429915428162,
      "learning_rate": 1.38457469015284e-07,
      "loss": 0.06018853187561035,
      "step": 7100
    },
    {
      "epoch": 0.9600324472310007,
      "grad_norm": 0.06562701612710953,
      "learning_rate": 1.3753120362659576e-07,
      "loss": 0.0664825439453125,
      "step": 7101
    },
    {
      "epoch": 0.9601676440268365,
      "grad_norm": 0.03771786764264107,
      "learning_rate": 1.3660803268316925e-07,
      "loss": 0.03746604919433594,
      "step": 7102
    },
    {
      "epoch": 0.9603028408226725,
      "grad_norm": 0.044692859053611755,
      "learning_rate": 1.3568795637721065e-07,
      "loss": 0.0363612174987793,
      "step": 7103
    },
    {
      "epoch": 0.9604380376185084,
      "grad_norm": 0.04847042262554169,
      "learning_rate": 1.347709749002851e-07,
      "loss": 0.040572166442871094,
      "step": 7104
    },
    {
      "epoch": 0.9605732344143444,
      "grad_norm": 0.16118532419204712,
      "learning_rate": 1.338570884433149e-07,
      "loss": 0.07378178834915161,
      "step": 7105
    },
    {
      "epoch": 0.9607084312101803,
      "grad_norm": 0.048179082572460175,
      "learning_rate": 1.3294629719657448e-07,
      "loss": 0.05052661895751953,
      "step": 7106
    },
    {
      "epoch": 0.9608436280060163,
      "grad_norm": 0.06481017917394638,
      "learning_rate": 1.3203860134969548e-07,
      "loss": 0.034099578857421875,
      "step": 7107
    },
    {
      "epoch": 0.9609788248018521,
      "grad_norm": 0.09291870146989822,
      "learning_rate": 1.3113400109166508e-07,
      "loss": 0.04570817947387695,
      "step": 7108
    },
    {
      "epoch": 0.9611140215976881,
      "grad_norm": 0.07057782262563705,
      "learning_rate": 1.3023249661082592e-07,
      "loss": 0.06121492385864258,
      "step": 7109
    },
    {
      "epoch": 0.9612492183935241,
      "grad_norm": 0.07935141772031784,
      "learning_rate": 1.2933408809487623e-07,
      "loss": 0.0702776312828064,
      "step": 7110
    },
    {
      "epoch": 0.96138441518936,
      "grad_norm": 0.08906565606594086,
      "learning_rate": 1.2843877573086972e-07,
      "loss": 0.08137989044189453,
      "step": 7111
    },
    {
      "epoch": 0.961519611985196,
      "grad_norm": 0.1408509612083435,
      "learning_rate": 1.2754655970521556e-07,
      "loss": 0.08360910415649414,
      "step": 7112
    },
    {
      "epoch": 0.9616548087810319,
      "grad_norm": 0.12053748965263367,
      "learning_rate": 1.2665744020367686e-07,
      "loss": 0.06547069549560547,
      "step": 7113
    },
    {
      "epoch": 0.9617900055768678,
      "grad_norm": 0.0661163479089737,
      "learning_rate": 1.2577141741137388e-07,
      "loss": 0.06647157669067383,
      "step": 7114
    },
    {
      "epoch": 0.9619252023727037,
      "grad_norm": 0.11322446167469025,
      "learning_rate": 1.248884915127807e-07,
      "loss": 0.06463050842285156,
      "step": 7115
    },
    {
      "epoch": 0.9620603991685397,
      "grad_norm": 0.08344893902540207,
      "learning_rate": 1.2400866269172694e-07,
      "loss": 0.055196285247802734,
      "step": 7116
    },
    {
      "epoch": 0.9621955959643756,
      "grad_norm": 0.165816068649292,
      "learning_rate": 1.2313193113139777e-07,
      "loss": 0.06582069396972656,
      "step": 7117
    },
    {
      "epoch": 0.9623307927602116,
      "grad_norm": 0.047542229294776917,
      "learning_rate": 1.2225829701433545e-07,
      "loss": 0.038216590881347656,
      "step": 7118
    },
    {
      "epoch": 0.9624659895560476,
      "grad_norm": 0.14484909176826477,
      "learning_rate": 1.2138776052243116e-07,
      "loss": 0.055254220962524414,
      "step": 7119
    },
    {
      "epoch": 0.9626011863518834,
      "grad_norm": 0.0934133380651474,
      "learning_rate": 1.2052032183693996e-07,
      "loss": 0.06642961502075195,
      "step": 7120
    },
    {
      "epoch": 0.9627363831477194,
      "grad_norm": 0.10766567289829254,
      "learning_rate": 1.196559811384623e-07,
      "loss": 0.06594440340995789,
      "step": 7121
    },
    {
      "epoch": 0.9628715799435553,
      "grad_norm": 0.0728660449385643,
      "learning_rate": 1.1879473860696266e-07,
      "loss": 0.04430907964706421,
      "step": 7122
    },
    {
      "epoch": 0.9630067767393913,
      "grad_norm": 0.10184944421052933,
      "learning_rate": 1.179365944217542e-07,
      "loss": 0.053729116916656494,
      "step": 7123
    },
    {
      "epoch": 0.9631419735352272,
      "grad_norm": 0.13168060779571533,
      "learning_rate": 1.1708154876150735e-07,
      "loss": 0.08052492141723633,
      "step": 7124
    },
    {
      "epoch": 0.9632771703310632,
      "grad_norm": 0.09283110499382019,
      "learning_rate": 1.1622960180424801e-07,
      "loss": 0.054796457290649414,
      "step": 7125
    },
    {
      "epoch": 0.963412367126899,
      "grad_norm": 0.09164842963218689,
      "learning_rate": 1.1538075372735435e-07,
      "loss": 0.051995038986206055,
      "step": 7126
    },
    {
      "epoch": 0.963547563922735,
      "grad_norm": 0.11548691987991333,
      "learning_rate": 1.1453500470756328e-07,
      "loss": 0.06072866916656494,
      "step": 7127
    },
    {
      "epoch": 0.963682760718571,
      "grad_norm": 0.171317458152771,
      "learning_rate": 1.1369235492096397e-07,
      "loss": 0.08099818229675293,
      "step": 7128
    },
    {
      "epoch": 0.9638179575144069,
      "grad_norm": 0.0761416107416153,
      "learning_rate": 1.1285280454299774e-07,
      "loss": 0.05479097366333008,
      "step": 7129
    },
    {
      "epoch": 0.9639531543102429,
      "grad_norm": 0.05584999918937683,
      "learning_rate": 1.1201635374846808e-07,
      "loss": 0.05953693389892578,
      "step": 7130
    },
    {
      "epoch": 0.9640883511060788,
      "grad_norm": 0.06278429925441742,
      "learning_rate": 1.1118300271152404e-07,
      "loss": 0.04982280731201172,
      "step": 7131
    },
    {
      "epoch": 0.9642235479019148,
      "grad_norm": 0.12395094335079193,
      "learning_rate": 1.1035275160567682e-07,
      "loss": 0.06191062927246094,
      "step": 7132
    },
    {
      "epoch": 0.9643587446977506,
      "grad_norm": 0.07014553248882294,
      "learning_rate": 1.0952560060378813e-07,
      "loss": 0.05559396743774414,
      "step": 7133
    },
    {
      "epoch": 0.9644939414935866,
      "grad_norm": 0.15083228051662445,
      "learning_rate": 1.0870154987807523e-07,
      "loss": 0.04974627494812012,
      "step": 7134
    },
    {
      "epoch": 0.9646291382894225,
      "grad_norm": 0.09887823462486267,
      "learning_rate": 1.0788059960010921e-07,
      "loss": 0.07526826858520508,
      "step": 7135
    },
    {
      "epoch": 0.9647643350852585,
      "grad_norm": 0.08582594245672226,
      "learning_rate": 1.0706274994081499e-07,
      "loss": 0.05914449691772461,
      "step": 7136
    },
    {
      "epoch": 0.9648995318810945,
      "grad_norm": 0.103627510368824,
      "learning_rate": 1.0624800107047805e-07,
      "loss": 0.06508827209472656,
      "step": 7137
    },
    {
      "epoch": 0.9650347286769304,
      "grad_norm": 0.08308898657560349,
      "learning_rate": 1.0543635315872934e-07,
      "loss": 0.05675673484802246,
      "step": 7138
    },
    {
      "epoch": 0.9651699254727663,
      "grad_norm": 0.12995609641075134,
      "learning_rate": 1.0462780637455871e-07,
      "loss": 0.04163527488708496,
      "step": 7139
    },
    {
      "epoch": 0.9653051222686022,
      "grad_norm": 0.04397675767540932,
      "learning_rate": 1.0382236088631148e-07,
      "loss": 0.041150569915771484,
      "step": 7140
    },
    {
      "epoch": 0.9654403190644382,
      "grad_norm": 0.05513301119208336,
      "learning_rate": 1.0302001686168349e-07,
      "loss": 0.05726146697998047,
      "step": 7141
    },
    {
      "epoch": 0.9655755158602741,
      "grad_norm": 0.06386985629796982,
      "learning_rate": 1.0222077446772949e-07,
      "loss": 0.060245513916015625,
      "step": 7142
    },
    {
      "epoch": 0.9657107126561101,
      "grad_norm": 0.11923275887966156,
      "learning_rate": 1.0142463387085465e-07,
      "loss": 0.0731656551361084,
      "step": 7143
    },
    {
      "epoch": 0.965845909451946,
      "grad_norm": 0.09917903691530228,
      "learning_rate": 1.0063159523682142e-07,
      "loss": 0.0786084234714508,
      "step": 7144
    },
    {
      "epoch": 0.9659811062477819,
      "grad_norm": 0.13281050324440002,
      "learning_rate": 9.984165873074102e-08,
      "loss": 0.0727696418762207,
      "step": 7145
    },
    {
      "epoch": 0.9661163030436178,
      "grad_norm": 0.1297747641801834,
      "learning_rate": 9.905482451708526e-08,
      "loss": 0.07542753219604492,
      "step": 7146
    },
    {
      "epoch": 0.9662514998394538,
      "grad_norm": 0.07161790132522583,
      "learning_rate": 9.827109275967638e-08,
      "loss": 0.05749034881591797,
      "step": 7147
    },
    {
      "epoch": 0.9663866966352898,
      "grad_norm": 0.09546342492103577,
      "learning_rate": 9.749046362169223e-08,
      "loss": 0.06731534004211426,
      "step": 7148
    },
    {
      "epoch": 0.9665218934311257,
      "grad_norm": 0.055850736796855927,
      "learning_rate": 9.671293726566443e-08,
      "loss": 0.045067548751831055,
      "step": 7149
    },
    {
      "epoch": 0.9666570902269617,
      "grad_norm": 0.11278649419546127,
      "learning_rate": 9.593851385347518e-08,
      "loss": 0.07315880060195923,
      "step": 7150
    },
    {
      "epoch": 0.9667922870227975,
      "grad_norm": 0.05858650803565979,
      "learning_rate": 9.516719354636716e-08,
      "loss": 0.032311201095581055,
      "step": 7151
    },
    {
      "epoch": 0.9669274838186335,
      "grad_norm": 0.05344470962882042,
      "learning_rate": 9.439897650493024e-08,
      "loss": 0.04370051622390747,
      "step": 7152
    },
    {
      "epoch": 0.9670626806144694,
      "grad_norm": 0.053297895938158035,
      "learning_rate": 9.363386288911313e-08,
      "loss": 0.0451812744140625,
      "step": 7153
    },
    {
      "epoch": 0.9671978774103054,
      "grad_norm": 0.08843020349740982,
      "learning_rate": 9.287185285821675e-08,
      "loss": 0.07363772392272949,
      "step": 7154
    },
    {
      "epoch": 0.9673330742061413,
      "grad_norm": 0.10848035663366318,
      "learning_rate": 9.211294657089587e-08,
      "loss": 0.0557703971862793,
      "step": 7155
    },
    {
      "epoch": 0.9674682710019773,
      "grad_norm": 0.08803831040859222,
      "learning_rate": 9.135714418515573e-08,
      "loss": 0.04780101776123047,
      "step": 7156
    },
    {
      "epoch": 0.9676034677978131,
      "grad_norm": 0.08944506198167801,
      "learning_rate": 9.060444585836381e-08,
      "loss": 0.06505727767944336,
      "step": 7157
    },
    {
      "epoch": 0.9677386645936491,
      "grad_norm": 0.10320509970188141,
      "learning_rate": 8.985485174722974e-08,
      "loss": 0.05461621284484863,
      "step": 7158
    },
    {
      "epoch": 0.9678738613894851,
      "grad_norm": 0.08219880610704422,
      "learning_rate": 8.910836200782868e-08,
      "loss": 0.060118675231933594,
      "step": 7159
    },
    {
      "epoch": 0.968009058185321,
      "grad_norm": 0.09920164197683334,
      "learning_rate": 8.836497679557964e-08,
      "loss": 0.06039631366729736,
      "step": 7160
    },
    {
      "epoch": 0.968144254981157,
      "grad_norm": 0.11348406970500946,
      "learning_rate": 8.762469626526048e-08,
      "loss": 0.07171249389648438,
      "step": 7161
    },
    {
      "epoch": 0.9682794517769929,
      "grad_norm": 0.10537512600421906,
      "learning_rate": 8.688752057100457e-08,
      "loss": 0.09157943725585938,
      "step": 7162
    },
    {
      "epoch": 0.9684146485728288,
      "grad_norm": 0.05739142373204231,
      "learning_rate": 8.615344986629082e-08,
      "loss": 0.043684959411621094,
      "step": 7163
    },
    {
      "epoch": 0.9685498453686647,
      "grad_norm": 0.12338952720165253,
      "learning_rate": 8.542248430396027e-08,
      "loss": 0.056375980377197266,
      "step": 7164
    },
    {
      "epoch": 0.9686850421645007,
      "grad_norm": 0.07579125463962555,
      "learning_rate": 8.469462403620287e-08,
      "loss": 0.041066765785217285,
      "step": 7165
    },
    {
      "epoch": 0.9688202389603366,
      "grad_norm": 0.13266021013259888,
      "learning_rate": 8.39698692145624e-08,
      "loss": 0.04218494892120361,
      "step": 7166
    },
    {
      "epoch": 0.9689554357561726,
      "grad_norm": 0.04894666746258736,
      "learning_rate": 8.324821998993648e-08,
      "loss": 0.03298592567443848,
      "step": 7167
    },
    {
      "epoch": 0.9690906325520086,
      "grad_norm": 0.059127770364284515,
      "learning_rate": 8.252967651257826e-08,
      "loss": 0.044711530208587646,
      "step": 7168
    },
    {
      "epoch": 0.9692258293478444,
      "grad_norm": 0.07320144027471542,
      "learning_rate": 8.181423893208973e-08,
      "loss": 0.06138050556182861,
      "step": 7169
    },
    {
      "epoch": 0.9693610261436804,
      "grad_norm": 0.08178849518299103,
      "learning_rate": 8.110190739743172e-08,
      "loss": 0.04873943328857422,
      "step": 7170
    },
    {
      "epoch": 0.9694962229395163,
      "grad_norm": 0.06717972457408905,
      "learning_rate": 8.03926820569123e-08,
      "loss": 0.04816484451293945,
      "step": 7171
    },
    {
      "epoch": 0.9696314197353523,
      "grad_norm": 0.15655025839805603,
      "learning_rate": 7.968656305819833e-08,
      "loss": 0.05584096908569336,
      "step": 7172
    },
    {
      "epoch": 0.9697666165311882,
      "grad_norm": 0.06611785292625427,
      "learning_rate": 7.898355054830719e-08,
      "loss": 0.05482077598571777,
      "step": 7173
    },
    {
      "epoch": 0.9699018133270242,
      "grad_norm": 0.10130004584789276,
      "learning_rate": 7.828364467360849e-08,
      "loss": 0.06627640128135681,
      "step": 7174
    },
    {
      "epoch": 0.97003701012286,
      "grad_norm": 0.07988695800304413,
      "learning_rate": 7.758684557982731e-08,
      "loss": 0.056477874517440796,
      "step": 7175
    },
    {
      "epoch": 0.970172206918696,
      "grad_norm": 0.07316162437200546,
      "learning_rate": 7.689315341204262e-08,
      "loss": 0.06038975715637207,
      "step": 7176
    },
    {
      "epoch": 0.970307403714532,
      "grad_norm": 0.02777114138007164,
      "learning_rate": 7.62025683146822e-08,
      "loss": 0.02889883518218994,
      "step": 7177
    },
    {
      "epoch": 0.9704426005103679,
      "grad_norm": 0.05920135974884033,
      "learning_rate": 7.551509043152937e-08,
      "loss": 0.0374600887298584,
      "step": 7178
    },
    {
      "epoch": 0.9705777973062039,
      "grad_norm": 0.08617579936981201,
      "learning_rate": 7.483071990572132e-08,
      "loss": 0.05197763442993164,
      "step": 7179
    },
    {
      "epoch": 0.9707129941020398,
      "grad_norm": 0.07711437344551086,
      "learning_rate": 7.414945687975072e-08,
      "loss": 0.06940889358520508,
      "step": 7180
    },
    {
      "epoch": 0.9708481908978758,
      "grad_norm": 0.15267902612686157,
      "learning_rate": 7.347130149545578e-08,
      "loss": 0.09007930755615234,
      "step": 7181
    },
    {
      "epoch": 0.9709833876937116,
      "grad_norm": 0.11504143476486206,
      "learning_rate": 7.279625389403355e-08,
      "loss": 0.07199335098266602,
      "step": 7182
    },
    {
      "epoch": 0.9711185844895476,
      "grad_norm": 0.09618354588747025,
      "learning_rate": 7.212431421603327e-08,
      "loss": 0.0663142204284668,
      "step": 7183
    },
    {
      "epoch": 0.9712537812853835,
      "grad_norm": 0.06710486859083176,
      "learning_rate": 7.145548260135638e-08,
      "loss": 0.056787967681884766,
      "step": 7184
    },
    {
      "epoch": 0.9713889780812195,
      "grad_norm": 0.10158078372478485,
      "learning_rate": 7.078975918925645e-08,
      "loss": 0.0723709836602211,
      "step": 7185
    },
    {
      "epoch": 0.9715241748770554,
      "grad_norm": 0.05166007578372955,
      "learning_rate": 7.012714411834098e-08,
      "loss": 0.04442107677459717,
      "step": 7186
    },
    {
      "epoch": 0.9716593716728914,
      "grad_norm": 0.04055127128958702,
      "learning_rate": 6.946763752656959e-08,
      "loss": 0.04286599159240723,
      "step": 7187
    },
    {
      "epoch": 0.9717945684687272,
      "grad_norm": 0.1107621118426323,
      "learning_rate": 6.881123955125579e-08,
      "loss": 0.08385276794433594,
      "step": 7188
    },
    {
      "epoch": 0.9719297652645632,
      "grad_norm": 0.06057797744870186,
      "learning_rate": 6.815795032906524e-08,
      "loss": 0.05335426330566406,
      "step": 7189
    },
    {
      "epoch": 0.9720649620603992,
      "grad_norm": 0.07267526537179947,
      "learning_rate": 6.750776999601415e-08,
      "loss": 0.05660605430603027,
      "step": 7190
    },
    {
      "epoch": 0.9722001588562351,
      "grad_norm": 0.1337149441242218,
      "learning_rate": 6.68606986874759e-08,
      "loss": 0.05895906686782837,
      "step": 7191
    },
    {
      "epoch": 0.9723353556520711,
      "grad_norm": 0.08872835338115692,
      "learning_rate": 6.62167365381744e-08,
      "loss": 0.05801975727081299,
      "step": 7192
    },
    {
      "epoch": 0.972470552447907,
      "grad_norm": 0.0661938264966011,
      "learning_rate": 6.557588368218237e-08,
      "loss": 0.05499553680419922,
      "step": 7193
    },
    {
      "epoch": 0.9726057492437429,
      "grad_norm": 0.12666253745555878,
      "learning_rate": 6.493814025293476e-08,
      "loss": 0.06466269493103027,
      "step": 7194
    },
    {
      "epoch": 0.9727409460395788,
      "grad_norm": 0.07209885120391846,
      "learning_rate": 6.430350638320704e-08,
      "loss": 0.05181649327278137,
      "step": 7195
    },
    {
      "epoch": 0.9728761428354148,
      "grad_norm": 0.08253423869609833,
      "learning_rate": 6.367198220513848e-08,
      "loss": 0.05649292469024658,
      "step": 7196
    },
    {
      "epoch": 0.9730113396312507,
      "grad_norm": 0.07808000594377518,
      "learning_rate": 6.304356785021226e-08,
      "loss": 0.07116007804870605,
      "step": 7197
    },
    {
      "epoch": 0.9731465364270867,
      "grad_norm": 0.05046825483441353,
      "learning_rate": 6.241826344926704e-08,
      "loss": 0.0363239049911499,
      "step": 7198
    },
    {
      "epoch": 0.9732817332229227,
      "grad_norm": 0.08713020384311676,
      "learning_rate": 6.17960691324987e-08,
      "loss": 0.054728031158447266,
      "step": 7199
    },
    {
      "epoch": 0.9734169300187585,
      "grad_norm": 0.06168987601995468,
      "learning_rate": 6.117698502944857e-08,
      "loss": 0.03146451711654663,
      "step": 7200
    },
    {
      "epoch": 0.9735521268145945,
      "grad_norm": 0.13072951138019562,
      "learning_rate": 6.056101126901358e-08,
      "loss": 0.07768011093139648,
      "step": 7201
    },
    {
      "epoch": 0.9736873236104304,
      "grad_norm": 0.056231435388326645,
      "learning_rate": 5.994814797944281e-08,
      "loss": 0.055208683013916016,
      "step": 7202
    },
    {
      "epoch": 0.9738225204062664,
      "grad_norm": 0.15364541113376617,
      "learning_rate": 5.933839528833751e-08,
      "loss": 0.06423664093017578,
      "step": 7203
    },
    {
      "epoch": 0.9739577172021023,
      "grad_norm": 0.09420917928218842,
      "learning_rate": 5.873175332265279e-08,
      "loss": 0.0685586929321289,
      "step": 7204
    },
    {
      "epoch": 0.9740929139979383,
      "grad_norm": 0.1318870633840561,
      "learning_rate": 5.812822220869096e-08,
      "loss": 0.07795560359954834,
      "step": 7205
    },
    {
      "epoch": 0.9742281107937741,
      "grad_norm": 0.1434108316898346,
      "learning_rate": 5.752780207211483e-08,
      "loss": 0.08306550979614258,
      "step": 7206
    },
    {
      "epoch": 0.9743633075896101,
      "grad_norm": 0.07491429895162582,
      "learning_rate": 5.693049303793274e-08,
      "loss": 0.04940158128738403,
      "step": 7207
    },
    {
      "epoch": 0.974498504385446,
      "grad_norm": 0.048310086131095886,
      "learning_rate": 5.6336295230508536e-08,
      "loss": 0.036936283111572266,
      "step": 7208
    },
    {
      "epoch": 0.974633701181282,
      "grad_norm": 0.12229932099580765,
      "learning_rate": 5.5745208773558266e-08,
      "loss": 0.045446574687957764,
      "step": 7209
    },
    {
      "epoch": 0.974768897977118,
      "grad_norm": 0.13875699043273926,
      "learning_rate": 5.515723379014681e-08,
      "loss": 0.05692625045776367,
      "step": 7210
    },
    {
      "epoch": 0.9749040947729539,
      "grad_norm": 0.08782761543989182,
      "learning_rate": 5.4572370402694583e-08,
      "loss": 0.07181024551391602,
      "step": 7211
    },
    {
      "epoch": 0.9750392915687898,
      "grad_norm": 0.07965756952762604,
      "learning_rate": 5.399061873297417e-08,
      "loss": 0.049305230379104614,
      "step": 7212
    },
    {
      "epoch": 0.9751744883646257,
      "grad_norm": 0.05086463689804077,
      "learning_rate": 5.341197890210869e-08,
      "loss": 0.03241884708404541,
      "step": 7213
    },
    {
      "epoch": 0.9753096851604617,
      "grad_norm": 0.06518615037202835,
      "learning_rate": 5.283645103057344e-08,
      "loss": 0.04001426696777344,
      "step": 7214
    },
    {
      "epoch": 0.9754448819562976,
      "grad_norm": 0.08768096566200256,
      "learning_rate": 5.226403523819756e-08,
      "loss": 0.07132196426391602,
      "step": 7215
    },
    {
      "epoch": 0.9755800787521336,
      "grad_norm": 0.13590385019779205,
      "learning_rate": 5.169473164416072e-08,
      "loss": 0.0567096471786499,
      "step": 7216
    },
    {
      "epoch": 0.9757152755479696,
      "grad_norm": 0.04079766944050789,
      "learning_rate": 5.112854036699477e-08,
      "loss": 0.03612542152404785,
      "step": 7217
    },
    {
      "epoch": 0.9758504723438054,
      "grad_norm": 0.1411398947238922,
      "learning_rate": 5.0565461524583745e-08,
      "loss": 0.09985828399658203,
      "step": 7218
    },
    {
      "epoch": 0.9759856691396414,
      "grad_norm": 0.06892736256122589,
      "learning_rate": 5.0005495234163865e-08,
      "loss": 0.04550325870513916,
      "step": 7219
    },
    {
      "epoch": 0.9761208659354773,
      "grad_norm": 0.058781880885362625,
      "learning_rate": 4.9448641612321874e-08,
      "loss": 0.044287681579589844,
      "step": 7220
    },
    {
      "epoch": 0.9762560627313133,
      "grad_norm": 0.14731939136981964,
      "learning_rate": 4.889490077500003e-08,
      "loss": 0.058156371116638184,
      "step": 7221
    },
    {
      "epoch": 0.9763912595271492,
      "grad_norm": 0.12105114758014679,
      "learning_rate": 4.8344272837489434e-08,
      "loss": 0.08028936386108398,
      "step": 7222
    },
    {
      "epoch": 0.9765264563229852,
      "grad_norm": 0.09475019574165344,
      "learning_rate": 4.779675791443172e-08,
      "loss": 0.07539916038513184,
      "step": 7223
    },
    {
      "epoch": 0.9766616531188211,
      "grad_norm": 0.06494379788637161,
      "learning_rate": 4.72523561198257e-08,
      "loss": 0.05502772331237793,
      "step": 7224
    },
    {
      "epoch": 0.976796849914657,
      "grad_norm": 0.051857080310583115,
      "learning_rate": 4.6711067567014044e-08,
      "loss": 0.0483931303024292,
      "step": 7225
    },
    {
      "epoch": 0.9769320467104929,
      "grad_norm": 0.04452521353960037,
      "learning_rate": 4.6172892368701595e-08,
      "loss": 0.03325977921485901,
      "step": 7226
    },
    {
      "epoch": 0.9770672435063289,
      "grad_norm": 0.06195031851530075,
      "learning_rate": 4.5637830636935385e-08,
      "loss": 0.055794715881347656,
      "step": 7227
    },
    {
      "epoch": 0.9772024403021649,
      "grad_norm": 0.05458105355501175,
      "learning_rate": 4.5105882483119643e-08,
      "loss": 0.050904035568237305,
      "step": 7228
    },
    {
      "epoch": 0.9773376370980008,
      "grad_norm": 0.07816591113805771,
      "learning_rate": 4.4577048018007436e-08,
      "loss": 0.07050323486328125,
      "step": 7229
    },
    {
      "epoch": 0.9774728338938368,
      "grad_norm": 0.05761956423521042,
      "learning_rate": 4.405132735170569e-08,
      "loss": 0.05326282978057861,
      "step": 7230
    },
    {
      "epoch": 0.9776080306896726,
      "grad_norm": 0.0911867767572403,
      "learning_rate": 4.3528720593675184e-08,
      "loss": 0.0633094310760498,
      "step": 7231
    },
    {
      "epoch": 0.9777432274855086,
      "grad_norm": 0.13230380415916443,
      "learning_rate": 4.300922785271888e-08,
      "loss": 0.07651877403259277,
      "step": 7232
    },
    {
      "epoch": 0.9778784242813445,
      "grad_norm": 0.14072942733764648,
      "learning_rate": 4.249284923700358e-08,
      "loss": 0.048138707876205444,
      "step": 7233
    },
    {
      "epoch": 0.9780136210771805,
      "grad_norm": 0.11088144779205322,
      "learning_rate": 4.197958485404163e-08,
      "loss": 0.0700385570526123,
      "step": 7234
    },
    {
      "epoch": 0.9781488178730164,
      "grad_norm": 0.06729786098003387,
      "learning_rate": 4.1469434810694206e-08,
      "loss": 0.054015517234802246,
      "step": 7235
    },
    {
      "epoch": 0.9782840146688524,
      "grad_norm": 0.05143892765045166,
      "learning_rate": 4.096239921317968e-08,
      "loss": 0.04444146156311035,
      "step": 7236
    },
    {
      "epoch": 0.9784192114646882,
      "grad_norm": 0.16051821410655975,
      "learning_rate": 4.045847816706361e-08,
      "loss": 0.06583833694458008,
      "step": 7237
    },
    {
      "epoch": 0.9785544082605242,
      "grad_norm": 0.0900055542588234,
      "learning_rate": 3.9957671777268724e-08,
      "loss": 0.04785585403442383,
      "step": 7238
    },
    {
      "epoch": 0.9786896050563602,
      "grad_norm": 0.08714839816093445,
      "learning_rate": 3.945998014806163e-08,
      "loss": 0.04713749885559082,
      "step": 7239
    },
    {
      "epoch": 0.9788248018521961,
      "grad_norm": 0.06403005123138428,
      "learning_rate": 3.896540338306609e-08,
      "loss": 0.0555572509765625,
      "step": 7240
    },
    {
      "epoch": 0.9789599986480321,
      "grad_norm": 0.04254044219851494,
      "learning_rate": 3.847394158525641e-08,
      "loss": 0.02935624122619629,
      "step": 7241
    },
    {
      "epoch": 0.979095195443868,
      "grad_norm": 0.09209254384040833,
      "learning_rate": 3.798559485695574e-08,
      "loss": 0.07053327560424805,
      "step": 7242
    },
    {
      "epoch": 0.9792303922397039,
      "grad_norm": 0.10998386144638062,
      "learning_rate": 3.7500363299842746e-08,
      "loss": 0.07876110076904297,
      "step": 7243
    },
    {
      "epoch": 0.9793655890355398,
      "grad_norm": 0.04358939826488495,
      "learning_rate": 3.701824701494327e-08,
      "loss": 0.0397646427154541,
      "step": 7244
    },
    {
      "epoch": 0.9795007858313758,
      "grad_norm": 0.08134984970092773,
      "learning_rate": 3.653924610263703e-08,
      "loss": 0.051084041595458984,
      "step": 7245
    },
    {
      "epoch": 0.9796359826272117,
      "grad_norm": 0.16544319689273834,
      "learning_rate": 3.6063360662654255e-08,
      "loss": 0.06447172164916992,
      "step": 7246
    },
    {
      "epoch": 0.9797711794230477,
      "grad_norm": 0.060077276080846786,
      "learning_rate": 3.559059079407734e-08,
      "loss": 0.05467081069946289,
      "step": 7247
    },
    {
      "epoch": 0.9799063762188837,
      "grad_norm": 0.06446938961744308,
      "learning_rate": 3.512093659533922e-08,
      "loss": 0.05666780471801758,
      "step": 7248
    },
    {
      "epoch": 0.9800415730147195,
      "grad_norm": 0.06144725903868675,
      "learning_rate": 3.4654398164225e-08,
      "loss": 0.04402303695678711,
      "step": 7249
    },
    {
      "epoch": 0.9801767698105555,
      "grad_norm": 0.06942480802536011,
      "learning_rate": 3.4190975597870325e-08,
      "loss": 0.06115114688873291,
      "step": 7250
    },
    {
      "epoch": 0.9803119666063914,
      "grad_norm": 0.10614340752363205,
      "learning_rate": 3.373066899276134e-08,
      "loss": 0.0576479434967041,
      "step": 7251
    },
    {
      "epoch": 0.9804471634022274,
      "grad_norm": 0.10466573387384415,
      "learning_rate": 3.3273478444736386e-08,
      "loss": 0.04949367046356201,
      "step": 7252
    },
    {
      "epoch": 0.9805823601980633,
      "grad_norm": 0.08140348643064499,
      "learning_rate": 3.281940404898764e-08,
      "loss": 0.065673828125,
      "step": 7253
    },
    {
      "epoch": 0.9807175569938993,
      "grad_norm": 0.12110307067632675,
      "learning_rate": 3.236844590005117e-08,
      "loss": 0.06766033172607422,
      "step": 7254
    },
    {
      "epoch": 0.9808527537897351,
      "grad_norm": 0.06678981333971024,
      "learning_rate": 3.192060409182351e-08,
      "loss": 0.050559282302856445,
      "step": 7255
    },
    {
      "epoch": 0.9809879505855711,
      "grad_norm": 0.09222646057605743,
      "learning_rate": 3.147587871754509e-08,
      "loss": 0.07135415077209473,
      "step": 7256
    },
    {
      "epoch": 0.981123147381407,
      "grad_norm": 0.0893658846616745,
      "learning_rate": 3.1034269869810174e-08,
      "loss": 0.034998536109924316,
      "step": 7257
    },
    {
      "epoch": 0.981258344177243,
      "grad_norm": 0.13266150653362274,
      "learning_rate": 3.05957776405652e-08,
      "loss": 0.046663522720336914,
      "step": 7258
    },
    {
      "epoch": 0.981393540973079,
      "grad_norm": 0.07934768497943878,
      "learning_rate": 3.016040212110549e-08,
      "loss": 0.0372769832611084,
      "step": 7259
    },
    {
      "epoch": 0.9815287377689149,
      "grad_norm": 0.07053276896476746,
      "learning_rate": 2.9728143402078522e-08,
      "loss": 0.052707672119140625,
      "step": 7260
    },
    {
      "epoch": 0.9816639345647508,
      "grad_norm": 0.06684913486242294,
      "learning_rate": 2.9299001573483975e-08,
      "loss": 0.0597834587097168,
      "step": 7261
    },
    {
      "epoch": 0.9817991313605867,
      "grad_norm": 0.06725136935710907,
      "learning_rate": 2.8872976724670375e-08,
      "loss": 0.050567626953125,
      "step": 7262
    },
    {
      "epoch": 0.9819343281564227,
      "grad_norm": 0.06243680417537689,
      "learning_rate": 2.8450068944338436e-08,
      "loss": 0.05402660369873047,
      "step": 7263
    },
    {
      "epoch": 0.9820695249522586,
      "grad_norm": 0.06280216574668884,
      "learning_rate": 2.803027832054106e-08,
      "loss": 0.04726457595825195,
      "step": 7264
    },
    {
      "epoch": 0.9822047217480946,
      "grad_norm": 0.12313113361597061,
      "learning_rate": 2.7613604940679995e-08,
      "loss": 0.06346559524536133,
      "step": 7265
    },
    {
      "epoch": 0.9823399185439305,
      "grad_norm": 0.2555335462093353,
      "learning_rate": 2.7200048891509176e-08,
      "loss": 0.10686159133911133,
      "step": 7266
    },
    {
      "epoch": 0.9824751153397665,
      "grad_norm": 0.05700648948550224,
      "learning_rate": 2.67896102591314e-08,
      "loss": 0.052342891693115234,
      "step": 7267
    },
    {
      "epoch": 0.9826103121356023,
      "grad_norm": 0.053727101534605026,
      "learning_rate": 2.6382289129004978e-08,
      "loss": 0.04517078399658203,
      "step": 7268
    },
    {
      "epoch": 0.9827455089314383,
      "grad_norm": 0.0781005397439003,
      "learning_rate": 2.5978085585935395e-08,
      "loss": 0.07580232620239258,
      "step": 7269
    },
    {
      "epoch": 0.9828807057272743,
      "grad_norm": 0.034053657203912735,
      "learning_rate": 2.5576999714078676e-08,
      "loss": 0.0352177619934082,
      "step": 7270
    },
    {
      "epoch": 0.9830159025231102,
      "grad_norm": 0.07882603257894516,
      "learning_rate": 2.517903159694468e-08,
      "loss": 0.047789573669433594,
      "step": 7271
    },
    {
      "epoch": 0.9831510993189462,
      "grad_norm": 0.04295378923416138,
      "learning_rate": 2.4784181317390465e-08,
      "loss": 0.027587890625,
      "step": 7272
    },
    {
      "epoch": 0.9832862961147821,
      "grad_norm": 0.1099075973033905,
      "learning_rate": 2.4392448957628598e-08,
      "loss": 0.0524439811706543,
      "step": 7273
    },
    {
      "epoch": 0.983421492910618,
      "grad_norm": 0.15199990570545197,
      "learning_rate": 2.4003834599217177e-08,
      "loss": 0.08845829963684082,
      "step": 7274
    },
    {
      "epoch": 0.9835566897064539,
      "grad_norm": 0.13113896548748016,
      "learning_rate": 2.3618338323071474e-08,
      "loss": 0.08320808410644531,
      "step": 7275
    },
    {
      "epoch": 0.9836918865022899,
      "grad_norm": 0.09326287358999252,
      "learning_rate": 2.3235960209448958e-08,
      "loss": 0.05952954292297363,
      "step": 7276
    },
    {
      "epoch": 0.9838270832981258,
      "grad_norm": 0.07457846403121948,
      "learning_rate": 2.2856700337967606e-08,
      "loss": 0.045603036880493164,
      "step": 7277
    },
    {
      "epoch": 0.9839622800939618,
      "grad_norm": 0.09696845710277557,
      "learning_rate": 2.2480558787587592e-08,
      "loss": 0.06147873401641846,
      "step": 7278
    },
    {
      "epoch": 0.9840974768897978,
      "grad_norm": 0.036029696464538574,
      "learning_rate": 2.2107535636626263e-08,
      "loss": 0.02816241979598999,
      "step": 7279
    },
    {
      "epoch": 0.9842326736856336,
      "grad_norm": 0.10190917551517487,
      "learning_rate": 2.1737630962746502e-08,
      "loss": 0.07305526733398438,
      "step": 7280
    },
    {
      "epoch": 0.9843678704814696,
      "grad_norm": 0.07059134542942047,
      "learning_rate": 2.1370844842966696e-08,
      "loss": 0.0533447265625,
      "step": 7281
    },
    {
      "epoch": 0.9845030672773055,
      "grad_norm": 0.08832582086324692,
      "learning_rate": 2.100717735365243e-08,
      "loss": 0.051555335521698,
      "step": 7282
    },
    {
      "epoch": 0.9846382640731415,
      "grad_norm": 0.1104244813323021,
      "learning_rate": 2.0646628570521464e-08,
      "loss": 0.08354133367538452,
      "step": 7283
    },
    {
      "epoch": 0.9847734608689774,
      "grad_norm": 0.0623931810259819,
      "learning_rate": 2.028919856864375e-08,
      "loss": 0.04960620403289795,
      "step": 7284
    },
    {
      "epoch": 0.9849086576648134,
      "grad_norm": 0.09466513246297836,
      "learning_rate": 1.9934887422434766e-08,
      "loss": 0.06949377059936523,
      "step": 7285
    },
    {
      "epoch": 0.9850438544606492,
      "grad_norm": 0.03106377087533474,
      "learning_rate": 1.9583695205665496e-08,
      "loss": 0.02700960636138916,
      "step": 7286
    },
    {
      "epoch": 0.9851790512564852,
      "grad_norm": 0.05648316070437431,
      "learning_rate": 1.9235621991457454e-08,
      "loss": 0.03772091865539551,
      "step": 7287
    },
    {
      "epoch": 0.9853142480523212,
      "grad_norm": 0.07968435436487198,
      "learning_rate": 1.889066785227933e-08,
      "loss": 0.04428505897521973,
      "step": 7288
    },
    {
      "epoch": 0.9854494448481571,
      "grad_norm": 0.07778004556894302,
      "learning_rate": 1.854883285995368e-08,
      "loss": 0.06160151958465576,
      "step": 7289
    },
    {
      "epoch": 0.9855846416439931,
      "grad_norm": 0.05391272157430649,
      "learning_rate": 1.8210117085651902e-08,
      "loss": 0.03964868187904358,
      "step": 7290
    },
    {
      "epoch": 0.985719838439829,
      "grad_norm": 0.08631722629070282,
      "learning_rate": 1.7874520599894252e-08,
      "loss": 0.0598507821559906,
      "step": 7291
    },
    {
      "epoch": 0.9858550352356649,
      "grad_norm": 0.08468089252710342,
      "learning_rate": 1.7542043472558166e-08,
      "loss": 0.07365989685058594,
      "step": 7292
    },
    {
      "epoch": 0.9859902320315008,
      "grad_norm": 0.12107174098491669,
      "learning_rate": 1.7212685772864945e-08,
      "loss": 0.07382214069366455,
      "step": 7293
    },
    {
      "epoch": 0.9861254288273368,
      "grad_norm": 0.07340415567159653,
      "learning_rate": 1.68864475693864e-08,
      "loss": 0.04373335838317871,
      "step": 7294
    },
    {
      "epoch": 0.9862606256231727,
      "grad_norm": 0.11149155348539352,
      "learning_rate": 1.6563328930051526e-08,
      "loss": 0.0391998291015625,
      "step": 7295
    },
    {
      "epoch": 0.9863958224190087,
      "grad_norm": 0.12067710608243942,
      "learning_rate": 1.624332992213151e-08,
      "loss": 0.07052433490753174,
      "step": 7296
    },
    {
      "epoch": 0.9865310192148447,
      "grad_norm": 0.06118866801261902,
      "learning_rate": 1.5926450612254728e-08,
      "loss": 0.03546231985092163,
      "step": 7297
    },
    {
      "epoch": 0.9866662160106805,
      "grad_norm": 0.06422483921051025,
      "learning_rate": 1.5612691066395068e-08,
      "loss": 0.049187541007995605,
      "step": 7298
    },
    {
      "epoch": 0.9868014128065165,
      "grad_norm": 0.08040238171815872,
      "learning_rate": 1.530205134987861e-08,
      "loss": 0.07235908508300781,
      "step": 7299
    },
    {
      "epoch": 0.9869366096023524,
      "grad_norm": 0.048238355666399,
      "learning_rate": 1.499453152738528e-08,
      "loss": 0.04775857925415039,
      "step": 7300
    },
    {
      "epoch": 0.9870718063981884,
      "grad_norm": 0.11366164684295654,
      "learning_rate": 1.4690131662938866e-08,
      "loss": 0.09452676773071289,
      "step": 7301
    },
    {
      "epoch": 0.9872070031940243,
      "grad_norm": 0.07032579183578491,
      "learning_rate": 1.438885181991867e-08,
      "loss": 0.05037975311279297,
      "step": 7302
    },
    {
      "epoch": 0.9873421999898603,
      "grad_norm": 0.052110254764556885,
      "learning_rate": 1.4090692061052846e-08,
      "loss": 0.05149391293525696,
      "step": 7303
    },
    {
      "epoch": 0.9874773967856961,
      "grad_norm": 0.11958946287631989,
      "learning_rate": 1.3795652448420071e-08,
      "loss": 0.08651185035705566,
      "step": 7304
    },
    {
      "epoch": 0.9876125935815321,
      "grad_norm": 0.03916904330253601,
      "learning_rate": 1.3503733043447874e-08,
      "loss": 0.029096364974975586,
      "step": 7305
    },
    {
      "epoch": 0.987747790377368,
      "grad_norm": 0.07844559848308563,
      "learning_rate": 1.3214933906915971e-08,
      "loss": 0.0634145438671112,
      "step": 7306
    },
    {
      "epoch": 0.987882987173204,
      "grad_norm": 0.1720583587884903,
      "learning_rate": 1.2929255098954596e-08,
      "loss": 0.060106754302978516,
      "step": 7307
    },
    {
      "epoch": 0.98801818396904,
      "grad_norm": 0.05431899055838585,
      "learning_rate": 1.2646696679042835e-08,
      "loss": 0.041239380836486816,
      "step": 7308
    },
    {
      "epoch": 0.9881533807648759,
      "grad_norm": 0.05242586508393288,
      "learning_rate": 1.2367258706010298e-08,
      "loss": 0.041638731956481934,
      "step": 7309
    },
    {
      "epoch": 0.9882885775607118,
      "grad_norm": 0.0778360441327095,
      "learning_rate": 1.2090941238040443e-08,
      "loss": 0.05116933584213257,
      "step": 7310
    },
    {
      "epoch": 0.9884237743565477,
      "grad_norm": 0.16519494354724884,
      "learning_rate": 1.1817744332660584e-08,
      "loss": 0.05723142623901367,
      "step": 7311
    },
    {
      "epoch": 0.9885589711523837,
      "grad_norm": 0.024900518357753754,
      "learning_rate": 1.1547668046751891e-08,
      "loss": 0.022734999656677246,
      "step": 7312
    },
    {
      "epoch": 0.9886941679482196,
      "grad_norm": 0.120774045586586,
      "learning_rate": 1.1280712436549379e-08,
      "loss": 0.0833730697631836,
      "step": 7313
    },
    {
      "epoch": 0.9888293647440556,
      "grad_norm": 0.08230368047952652,
      "learning_rate": 1.1016877557630257e-08,
      "loss": 0.04598128795623779,
      "step": 7314
    },
    {
      "epoch": 0.9889645615398915,
      "grad_norm": 0.06282849609851837,
      "learning_rate": 1.0756163464928915e-08,
      "loss": 0.05870342254638672,
      "step": 7315
    },
    {
      "epoch": 0.9890997583357275,
      "grad_norm": 0.0756852999329567,
      "learning_rate": 1.0498570212726932e-08,
      "loss": 0.07147502899169922,
      "step": 7316
    },
    {
      "epoch": 0.9892349551315633,
      "grad_norm": 0.04308709874749184,
      "learning_rate": 1.024409785465641e-08,
      "loss": 0.042134761810302734,
      "step": 7317
    },
    {
      "epoch": 0.9893701519273993,
      "grad_norm": 0.11743681877851486,
      "learning_rate": 9.992746443699962e-09,
      "loss": 0.06059384346008301,
      "step": 7318
    },
    {
      "epoch": 0.9895053487232353,
      "grad_norm": 0.05181999132037163,
      "learning_rate": 9.744516032190731e-09,
      "loss": 0.043380022048950195,
      "step": 7319
    },
    {
      "epoch": 0.9896405455190712,
      "grad_norm": 0.07927172631025314,
      "learning_rate": 9.499406671809041e-09,
      "loss": 0.06056690216064453,
      "step": 7320
    },
    {
      "epoch": 0.9897757423149072,
      "grad_norm": 0.09364833682775497,
      "learning_rate": 9.2574184135924e-09,
      "loss": 0.05971479415893555,
      "step": 7321
    },
    {
      "epoch": 0.9899109391107431,
      "grad_norm": 0.05973111838102341,
      "learning_rate": 9.018551307920508e-09,
      "loss": 0.035719871520996094,
      "step": 7322
    },
    {
      "epoch": 0.990046135906579,
      "grad_norm": 0.06893175095319748,
      "learning_rate": 8.782805404526917e-09,
      "loss": 0.04376935958862305,
      "step": 7323
    },
    {
      "epoch": 0.9901813327024149,
      "grad_norm": 0.07715661078691483,
      "learning_rate": 8.55018075249736e-09,
      "loss": 0.044313669204711914,
      "step": 7324
    },
    {
      "epoch": 0.9903165294982509,
      "grad_norm": 0.08267417550086975,
      "learning_rate": 8.320677400264764e-09,
      "loss": 0.0799022912979126,
      "step": 7325
    },
    {
      "epoch": 0.9904517262940868,
      "grad_norm": 0.08510856330394745,
      "learning_rate": 8.094295395610906e-09,
      "loss": 0.047110795974731445,
      "step": 7326
    },
    {
      "epoch": 0.9905869230899228,
      "grad_norm": 0.10011366754770279,
      "learning_rate": 7.87103478567308e-09,
      "loss": 0.06752943992614746,
      "step": 7327
    },
    {
      "epoch": 0.9907221198857588,
      "grad_norm": 0.11906718462705612,
      "learning_rate": 7.65089561693244e-09,
      "loss": 0.09389448165893555,
      "step": 7328
    },
    {
      "epoch": 0.9908573166815946,
      "grad_norm": 0.0777163952589035,
      "learning_rate": 7.433877935225652e-09,
      "loss": 0.05770304799079895,
      "step": 7329
    },
    {
      "epoch": 0.9909925134774306,
      "grad_norm": 0.0878908783197403,
      "learning_rate": 7.219981785733243e-09,
      "loss": 0.06152153015136719,
      "step": 7330
    },
    {
      "epoch": 0.9911277102732665,
      "grad_norm": 0.13056887686252594,
      "learning_rate": 7.009207212992919e-09,
      "loss": 0.08936762809753418,
      "step": 7331
    },
    {
      "epoch": 0.9912629070691025,
      "grad_norm": 0.07207076251506805,
      "learning_rate": 6.801554260889575e-09,
      "loss": 0.057196855545043945,
      "step": 7332
    },
    {
      "epoch": 0.9913981038649384,
      "grad_norm": 0.09439665824174881,
      "learning_rate": 6.5970229726552976e-09,
      "loss": 0.07920801639556885,
      "step": 7333
    },
    {
      "epoch": 0.9915333006607744,
      "grad_norm": 0.08448266237974167,
      "learning_rate": 6.3956133908743556e-09,
      "loss": 0.0589594841003418,
      "step": 7334
    },
    {
      "epoch": 0.9916684974566102,
      "grad_norm": 0.07885945588350296,
      "learning_rate": 6.197325557483202e-09,
      "loss": 0.060319721698760986,
      "step": 7335
    },
    {
      "epoch": 0.9918036942524462,
      "grad_norm": 0.07197452336549759,
      "learning_rate": 6.002159513765482e-09,
      "loss": 0.04823002219200134,
      "step": 7336
    },
    {
      "epoch": 0.9919388910482821,
      "grad_norm": 0.04160919785499573,
      "learning_rate": 5.810115300355357e-09,
      "loss": 0.03542327880859375,
      "step": 7337
    },
    {
      "epoch": 0.9920740878441181,
      "grad_norm": 0.08904135972261429,
      "learning_rate": 5.621192957239174e-09,
      "loss": 0.07356023788452148,
      "step": 7338
    },
    {
      "epoch": 0.9922092846399541,
      "grad_norm": 0.06244876980781555,
      "learning_rate": 5.435392523748806e-09,
      "loss": 0.05868411064147949,
      "step": 7339
    },
    {
      "epoch": 0.99234448143579,
      "grad_norm": 0.0832950696349144,
      "learning_rate": 5.252714038571638e-09,
      "loss": 0.07088655233383179,
      "step": 7340
    },
    {
      "epoch": 0.9924796782316259,
      "grad_norm": 0.10306744277477264,
      "learning_rate": 5.073157539742246e-09,
      "loss": 0.06284856796264648,
      "step": 7341
    },
    {
      "epoch": 0.9926148750274618,
      "grad_norm": 0.06996965408325195,
      "learning_rate": 4.896723064642394e-09,
      "loss": 0.0495152473449707,
      "step": 7342
    },
    {
      "epoch": 0.9927500718232978,
      "grad_norm": 0.09587699919939041,
      "learning_rate": 4.723410650009363e-09,
      "loss": 0.05898404121398926,
      "step": 7343
    },
    {
      "epoch": 0.9928852686191337,
      "grad_norm": 0.05208943411707878,
      "learning_rate": 4.553220331925956e-09,
      "loss": 0.05413079261779785,
      "step": 7344
    },
    {
      "epoch": 0.9930204654149697,
      "grad_norm": 0.05242899805307388,
      "learning_rate": 4.38615214582716e-09,
      "loss": 0.03993582725524902,
      "step": 7345
    },
    {
      "epoch": 0.9931556622108056,
      "grad_norm": 0.07335025072097778,
      "learning_rate": 4.2222061265001496e-09,
      "loss": 0.05894613265991211,
      "step": 7346
    },
    {
      "epoch": 0.9932908590066415,
      "grad_norm": 0.09117382764816284,
      "learning_rate": 4.0613823080742905e-09,
      "loss": 0.06613492965698242,
      "step": 7347
    },
    {
      "epoch": 0.9934260558024774,
      "grad_norm": 0.042938247323036194,
      "learning_rate": 3.903680724037795e-09,
      "loss": 0.04477345943450928,
      "step": 7348
    },
    {
      "epoch": 0.9935612525983134,
      "grad_norm": 0.058746881783008575,
      "learning_rate": 3.749101407224398e-09,
      "loss": 0.06394344568252563,
      "step": 7349
    },
    {
      "epoch": 0.9936964493941494,
      "grad_norm": 0.13351649045944214,
      "learning_rate": 3.597644389818355e-09,
      "loss": 0.09245014190673828,
      "step": 7350
    },
    {
      "epoch": 0.9938316461899853,
      "grad_norm": 0.034742143005132675,
      "learning_rate": 3.4493097033527767e-09,
      "loss": 0.03508591651916504,
      "step": 7351
    },
    {
      "epoch": 0.9939668429858213,
      "grad_norm": 0.08260580152273178,
      "learning_rate": 3.3040973787112904e-09,
      "loss": 0.08246707916259766,
      "step": 7352
    },
    {
      "epoch": 0.9941020397816571,
      "grad_norm": 0.085362009704113,
      "learning_rate": 3.162007446129711e-09,
      "loss": 0.05268669128417969,
      "step": 7353
    },
    {
      "epoch": 0.9942372365774931,
      "grad_norm": 0.04928789660334587,
      "learning_rate": 3.023039935191041e-09,
      "loss": 0.04291635751724243,
      "step": 7354
    },
    {
      "epoch": 0.994372433373329,
      "grad_norm": 0.11875756829977036,
      "learning_rate": 2.887194874830468e-09,
      "loss": 0.064117431640625,
      "step": 7355
    },
    {
      "epoch": 0.994507630169165,
      "grad_norm": 0.060207534581422806,
      "learning_rate": 2.7544722933287026e-09,
      "loss": 0.05113565921783447,
      "step": 7356
    },
    {
      "epoch": 0.994642826965001,
      "grad_norm": 0.08969463407993317,
      "learning_rate": 2.6248722183203066e-09,
      "loss": 0.07146716117858887,
      "step": 7357
    },
    {
      "epoch": 0.9947780237608369,
      "grad_norm": 0.08677563071250916,
      "learning_rate": 2.498394676790361e-09,
      "loss": 0.07406949996948242,
      "step": 7358
    },
    {
      "epoch": 0.9949132205566729,
      "grad_norm": 0.15845677256584167,
      "learning_rate": 2.375039695071135e-09,
      "loss": 0.05764305591583252,
      "step": 7359
    },
    {
      "epoch": 0.9950484173525087,
      "grad_norm": 0.10074788331985474,
      "learning_rate": 2.2548072988454184e-09,
      "loss": 0.06560428440570831,
      "step": 7360
    },
    {
      "epoch": 0.9951836141483447,
      "grad_norm": 0.08231482654809952,
      "learning_rate": 2.1376975131465194e-09,
      "loss": 0.05889749526977539,
      "step": 7361
    },
    {
      "epoch": 0.9953188109441806,
      "grad_norm": 0.12215745449066162,
      "learning_rate": 2.023710362356601e-09,
      "loss": 0.07542967796325684,
      "step": 7362
    },
    {
      "epoch": 0.9954540077400166,
      "grad_norm": 0.07525450736284256,
      "learning_rate": 1.9128458702100117e-09,
      "loss": 0.05368232727050781,
      "step": 7363
    },
    {
      "epoch": 0.9955892045358525,
      "grad_norm": 0.07393708825111389,
      "learning_rate": 1.8051040597882873e-09,
      "loss": 0.03982853889465332,
      "step": 7364
    },
    {
      "epoch": 0.9957244013316885,
      "grad_norm": 0.06943808495998383,
      "learning_rate": 1.70048495352515e-09,
      "loss": 0.05326399207115173,
      "step": 7365
    },
    {
      "epoch": 0.9958595981275243,
      "grad_norm": 0.11843971163034439,
      "learning_rate": 1.5989885731998443e-09,
      "loss": 0.0825052261352539,
      "step": 7366
    },
    {
      "epoch": 0.9959947949233603,
      "grad_norm": 0.0996103286743164,
      "learning_rate": 1.5006149399487966e-09,
      "loss": 0.0654592514038086,
      "step": 7367
    },
    {
      "epoch": 0.9961299917191963,
      "grad_norm": 0.05979285389184952,
      "learning_rate": 1.4053640742489604e-09,
      "loss": 0.04976773262023926,
      "step": 7368
    },
    {
      "epoch": 0.9962651885150322,
      "grad_norm": 0.13997714221477509,
      "learning_rate": 1.3132359959361351e-09,
      "loss": 0.08460056781768799,
      "step": 7369
    },
    {
      "epoch": 0.9964003853108682,
      "grad_norm": 0.06448747217655182,
      "learning_rate": 1.2242307241899787e-09,
      "loss": 0.052953124046325684,
      "step": 7370
    },
    {
      "epoch": 0.9965355821067041,
      "grad_norm": 0.041787318885326385,
      "learning_rate": 1.1383482775406685e-09,
      "loss": 0.03504984825849533,
      "step": 7371
    },
    {
      "epoch": 0.99667077890254,
      "grad_norm": 0.12908028066158295,
      "learning_rate": 1.0555886738738973e-09,
      "loss": 0.0784159004688263,
      "step": 7372
    },
    {
      "epoch": 0.9968059756983759,
      "grad_norm": 0.08244071900844574,
      "learning_rate": 9.75951930415886e-10,
      "loss": 0.06368780136108398,
      "step": 7373
    },
    {
      "epoch": 0.9969411724942119,
      "grad_norm": 0.10129138082265854,
      "learning_rate": 8.994380637483701e-10,
      "loss": 0.052173614501953125,
      "step": 7374
    },
    {
      "epoch": 0.9970763692900478,
      "grad_norm": 0.05334750562906265,
      "learning_rate": 8.260470898036054e-10,
      "loss": 0.045610785484313965,
      "step": 7375
    },
    {
      "epoch": 0.9972115660858838,
      "grad_norm": 0.10708335041999817,
      "learning_rate": 7.557790238627016e-10,
      "loss": 0.054087936878204346,
      "step": 7376
    },
    {
      "epoch": 0.9973467628817198,
      "grad_norm": 0.07844416797161102,
      "learning_rate": 6.886338805522918e-10,
      "loss": 0.04304170608520508,
      "step": 7377
    },
    {
      "epoch": 0.9974819596775556,
      "grad_norm": 0.0769147202372551,
      "learning_rate": 6.246116738561903e-10,
      "loss": 0.061806440353393555,
      "step": 7378
    },
    {
      "epoch": 0.9976171564733916,
      "grad_norm": 0.07373236864805222,
      "learning_rate": 5.637124171004038e-10,
      "loss": 0.04380142688751221,
      "step": 7379
    },
    {
      "epoch": 0.9977523532692275,
      "grad_norm": 0.11770644038915634,
      "learning_rate": 5.059361229681203e-10,
      "loss": 0.08809971809387207,
      "step": 7380
    },
    {
      "epoch": 0.9978875500650635,
      "grad_norm": 0.09737664461135864,
      "learning_rate": 4.5128280348638583e-10,
      "loss": 0.0631260871887207,
      "step": 7381
    },
    {
      "epoch": 0.9980227468608994,
      "grad_norm": 0.04032571613788605,
      "learning_rate": 3.9975247003443127e-10,
      "loss": 0.03821921348571777,
      "step": 7382
    },
    {
      "epoch": 0.9981579436567354,
      "grad_norm": 0.06047150120139122,
      "learning_rate": 3.51345133342007e-10,
      "loss": 0.03913068771362305,
      "step": 7383
    },
    {
      "epoch": 0.9982931404525712,
      "grad_norm": 0.10665332525968552,
      "learning_rate": 3.060608034877177e-10,
      "loss": 0.06272637844085693,
      "step": 7384
    },
    {
      "epoch": 0.9984283372484072,
      "grad_norm": 0.0563480518758297,
      "learning_rate": 2.638994898990221e-10,
      "loss": 0.041876569390296936,
      "step": 7385
    },
    {
      "epoch": 0.9985635340442431,
      "grad_norm": 0.05935134366154671,
      "learning_rate": 2.2486120135556398e-10,
      "loss": 0.05381488800048828,
      "step": 7386
    },
    {
      "epoch": 0.9986987308400791,
      "grad_norm": 0.08404790610074997,
      "learning_rate": 1.889459459841758e-10,
      "loss": 0.06423234939575195,
      "step": 7387
    },
    {
      "epoch": 0.998833927635915,
      "grad_norm": 0.0969076156616211,
      "learning_rate": 1.56153731263875e-10,
      "loss": 0.05673551559448242,
      "step": 7388
    },
    {
      "epoch": 0.998969124431751,
      "grad_norm": 0.05608352646231651,
      "learning_rate": 1.2648456402086784e-10,
      "loss": 0.050678133964538574,
      "step": 7389
    },
    {
      "epoch": 0.9991043212275869,
      "grad_norm": 0.03873895853757858,
      "learning_rate": 9.99384504318801e-11,
      "loss": 0.03996634483337402,
      "step": 7390
    },
    {
      "epoch": 0.9992395180234228,
      "grad_norm": 0.07401720434427261,
      "learning_rate": 7.651539602582247e-11,
      "loss": 0.04620383679866791,
      "step": 7391
    },
    {
      "epoch": 0.9993747148192588,
      "grad_norm": 0.15651960670948029,
      "learning_rate": 5.6215405678794464e-11,
      "loss": 0.07623988389968872,
      "step": 7392
    },
    {
      "epoch": 0.9995099116150947,
      "grad_norm": 0.10791472345590591,
      "learning_rate": 3.9038483615749795e-11,
      "loss": 0.051717519760131836,
      "step": 7393
    },
    {
      "epoch": 0.9996451084109307,
      "grad_norm": 0.06786425411701202,
      "learning_rate": 2.4984633415492398e-11,
      "loss": 0.06241750717163086,
      "step": 7394
    },
    {
      "epoch": 0.9997803052067666,
      "grad_norm": 0.033510491251945496,
      "learning_rate": 1.4053858004015041e-11,
      "loss": 0.03623461723327637,
      "step": 7395
    },
    {
      "epoch": 0.9999155020026025,
      "grad_norm": 0.08831936866044998,
      "learning_rate": 6.246159654499373e-12,
      "loss": 0.06389164924621582,
      "step": 7396
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.08253154158592224,
      "learning_rate": 1.561539995642569e-12,
      "loss": 0.07079315185546875,
      "step": 7397
    },
    {
      "epoch": 1.0,
      "step": 7397,
      "total_flos": 2.893062630172682e+19,
      "train_loss": 0.011039156519602775,
      "train_runtime": 12905.1847,
      "train_samples_per_second": 146.725,
      "train_steps_per_second": 0.573
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 7397,
  "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": 2.893062630172682e+19,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}