{
  "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": 0.00013519336876526207,
      "grad_norm": 88.03363800048828,
      "learning_rate": 0.0,
      "loss": 0.767974853515625,
      "step": 1
    },
    {
      "epoch": 0.00027038673753052413,
      "grad_norm": 89.86097717285156,
      "learning_rate": 5.859375e-08,
      "loss": 0.769378662109375,
      "step": 2
    },
    {
      "epoch": 0.0004055801062957862,
      "grad_norm": 91.50601196289062,
      "learning_rate": 1.171875e-07,
      "loss": 0.768035888671875,
      "step": 3
    },
    {
      "epoch": 0.0005407734750610483,
      "grad_norm": 84.51274871826172,
      "learning_rate": 1.7578125e-07,
      "loss": 0.7620849609375,
      "step": 4
    },
    {
      "epoch": 0.0006759668438263103,
      "grad_norm": 91.75151824951172,
      "learning_rate": 2.34375e-07,
      "loss": 0.767547607421875,
      "step": 5
    },
    {
      "epoch": 0.0008111602125915724,
      "grad_norm": 88.40542602539062,
      "learning_rate": 2.9296875000000003e-07,
      "loss": 0.76806640625,
      "step": 6
    },
    {
      "epoch": 0.0009463535813568345,
      "grad_norm": 88.53487396240234,
      "learning_rate": 3.515625e-07,
      "loss": 0.767822265625,
      "step": 7
    },
    {
      "epoch": 0.0010815469501220965,
      "grad_norm": 89.52659606933594,
      "learning_rate": 4.1015625e-07,
      "loss": 0.765960693359375,
      "step": 8
    },
    {
      "epoch": 0.0012167403188873586,
      "grad_norm": 91.50320434570312,
      "learning_rate": 4.6875e-07,
      "loss": 0.76519775390625,
      "step": 9
    },
    {
      "epoch": 0.0013519336876526207,
      "grad_norm": 88.8525619506836,
      "learning_rate": 5.2734375e-07,
      "loss": 0.76153564453125,
      "step": 10
    },
    {
      "epoch": 0.0014871270564178827,
      "grad_norm": 89.35160827636719,
      "learning_rate": 5.859375000000001e-07,
      "loss": 0.76251220703125,
      "step": 11
    },
    {
      "epoch": 0.0016223204251831448,
      "grad_norm": 89.51719665527344,
      "learning_rate": 6.4453125e-07,
      "loss": 0.76220703125,
      "step": 12
    },
    {
      "epoch": 0.0017575137939484068,
      "grad_norm": 91.03227996826172,
      "learning_rate": 7.03125e-07,
      "loss": 0.75811767578125,
      "step": 13
    },
    {
      "epoch": 0.001892707162713669,
      "grad_norm": 91.70550537109375,
      "learning_rate": 7.6171875e-07,
      "loss": 0.75830078125,
      "step": 14
    },
    {
      "epoch": 0.0020279005314789308,
      "grad_norm": 91.33809661865234,
      "learning_rate": 8.203125e-07,
      "loss": 0.75732421875,
      "step": 15
    },
    {
      "epoch": 0.002163093900244193,
      "grad_norm": 89.25840759277344,
      "learning_rate": 8.7890625e-07,
      "loss": 0.749755859375,
      "step": 16
    },
    {
      "epoch": 0.002298287269009455,
      "grad_norm": 90.64176177978516,
      "learning_rate": 9.375e-07,
      "loss": 0.748870849609375,
      "step": 17
    },
    {
      "epoch": 0.002433480637774717,
      "grad_norm": 85.99710083007812,
      "learning_rate": 9.9609375e-07,
      "loss": 0.738128662109375,
      "step": 18
    },
    {
      "epoch": 0.002568674006539979,
      "grad_norm": 90.21903991699219,
      "learning_rate": 1.0546875e-06,
      "loss": 0.736785888671875,
      "step": 19
    },
    {
      "epoch": 0.0027038673753052413,
      "grad_norm": 85.76763916015625,
      "learning_rate": 1.11328125e-06,
      "loss": 0.7325439453125,
      "step": 20
    },
    {
      "epoch": 0.002839060744070503,
      "grad_norm": 86.10233306884766,
      "learning_rate": 1.1718750000000001e-06,
      "loss": 0.731903076171875,
      "step": 21
    },
    {
      "epoch": 0.0029742541128357654,
      "grad_norm": 82.16548919677734,
      "learning_rate": 1.23046875e-06,
      "loss": 0.70294189453125,
      "step": 22
    },
    {
      "epoch": 0.0031094474816010273,
      "grad_norm": 84.72628021240234,
      "learning_rate": 1.2890625e-06,
      "loss": 0.70013427734375,
      "step": 23
    },
    {
      "epoch": 0.0032446408503662896,
      "grad_norm": 83.80533599853516,
      "learning_rate": 1.34765625e-06,
      "loss": 0.698394775390625,
      "step": 24
    },
    {
      "epoch": 0.0033798342191315514,
      "grad_norm": 81.4764175415039,
      "learning_rate": 1.40625e-06,
      "loss": 0.670745849609375,
      "step": 25
    },
    {
      "epoch": 0.0035150275878968137,
      "grad_norm": 79.31185913085938,
      "learning_rate": 1.46484375e-06,
      "loss": 0.666015625,
      "step": 26
    },
    {
      "epoch": 0.0036502209566620755,
      "grad_norm": 79.19160461425781,
      "learning_rate": 1.5234375e-06,
      "loss": 0.660888671875,
      "step": 27
    },
    {
      "epoch": 0.003785414325427338,
      "grad_norm": 76.50727844238281,
      "learning_rate": 1.5820312500000001e-06,
      "loss": 0.660675048828125,
      "step": 28
    },
    {
      "epoch": 0.0039206076941926,
      "grad_norm": 79.90402221679688,
      "learning_rate": 1.640625e-06,
      "loss": 0.648712158203125,
      "step": 29
    },
    {
      "epoch": 0.0040558010629578615,
      "grad_norm": 67.9672622680664,
      "learning_rate": 1.69921875e-06,
      "loss": 0.5775146484375,
      "step": 30
    },
    {
      "epoch": 0.004190994431723124,
      "grad_norm": 68.07891082763672,
      "learning_rate": 1.7578125e-06,
      "loss": 0.565826416015625,
      "step": 31
    },
    {
      "epoch": 0.004326187800488386,
      "grad_norm": 69.5920639038086,
      "learning_rate": 1.81640625e-06,
      "loss": 0.560821533203125,
      "step": 32
    },
    {
      "epoch": 0.004461381169253648,
      "grad_norm": 70.55146789550781,
      "learning_rate": 1.875e-06,
      "loss": 0.5455322265625,
      "step": 33
    },
    {
      "epoch": 0.00459657453801891,
      "grad_norm": 62.89144515991211,
      "learning_rate": 1.93359375e-06,
      "loss": 0.5310516357421875,
      "step": 34
    },
    {
      "epoch": 0.0047317679067841725,
      "grad_norm": 62.215370178222656,
      "learning_rate": 1.9921875e-06,
      "loss": 0.5177001953125,
      "step": 35
    },
    {
      "epoch": 0.004866961275549434,
      "grad_norm": 60.91170883178711,
      "learning_rate": 2.05078125e-06,
      "loss": 0.5041046142578125,
      "step": 36
    },
    {
      "epoch": 0.005002154644314696,
      "grad_norm": 62.39719772338867,
      "learning_rate": 2.109375e-06,
      "loss": 0.4893798828125,
      "step": 37
    },
    {
      "epoch": 0.005137348013079958,
      "grad_norm": 60.21321487426758,
      "learning_rate": 2.16796875e-06,
      "loss": 0.4813232421875,
      "step": 38
    },
    {
      "epoch": 0.005272541381845221,
      "grad_norm": 52.77524185180664,
      "learning_rate": 2.2265625e-06,
      "loss": 0.470611572265625,
      "step": 39
    },
    {
      "epoch": 0.005407734750610483,
      "grad_norm": 45.11795425415039,
      "learning_rate": 2.28515625e-06,
      "loss": 0.432891845703125,
      "step": 40
    },
    {
      "epoch": 0.0055429281193757445,
      "grad_norm": 40.704158782958984,
      "learning_rate": 2.3437500000000002e-06,
      "loss": 0.3948516845703125,
      "step": 41
    },
    {
      "epoch": 0.005678121488141006,
      "grad_norm": 36.75060272216797,
      "learning_rate": 2.40234375e-06,
      "loss": 0.3881683349609375,
      "step": 42
    },
    {
      "epoch": 0.005813314856906269,
      "grad_norm": 32.00703811645508,
      "learning_rate": 2.4609375e-06,
      "loss": 0.402679443359375,
      "step": 43
    },
    {
      "epoch": 0.005948508225671531,
      "grad_norm": 35.68880844116211,
      "learning_rate": 2.5195312500000003e-06,
      "loss": 0.3690643310546875,
      "step": 44
    },
    {
      "epoch": 0.006083701594436793,
      "grad_norm": 35.64313507080078,
      "learning_rate": 2.578125e-06,
      "loss": 0.356475830078125,
      "step": 45
    },
    {
      "epoch": 0.0062188949632020546,
      "grad_norm": 36.784759521484375,
      "learning_rate": 2.63671875e-06,
      "loss": 0.3323516845703125,
      "step": 46
    },
    {
      "epoch": 0.006354088331967317,
      "grad_norm": 30.713638305664062,
      "learning_rate": 2.6953125e-06,
      "loss": 0.3551788330078125,
      "step": 47
    },
    {
      "epoch": 0.006489281700732579,
      "grad_norm": 34.66780471801758,
      "learning_rate": 2.75390625e-06,
      "loss": 0.3091888427734375,
      "step": 48
    },
    {
      "epoch": 0.006624475069497841,
      "grad_norm": 30.918582916259766,
      "learning_rate": 2.8125e-06,
      "loss": 0.31719970703125,
      "step": 49
    },
    {
      "epoch": 0.006759668438263103,
      "grad_norm": 28.330442428588867,
      "learning_rate": 2.87109375e-06,
      "loss": 0.32224273681640625,
      "step": 50
    },
    {
      "epoch": 0.0068948618070283655,
      "grad_norm": 25.909151077270508,
      "learning_rate": 2.9296875e-06,
      "loss": 0.32096099853515625,
      "step": 51
    },
    {
      "epoch": 0.007030055175793627,
      "grad_norm": 24.142841339111328,
      "learning_rate": 2.9882812500000002e-06,
      "loss": 0.31163787841796875,
      "step": 52
    },
    {
      "epoch": 0.007165248544558889,
      "grad_norm": 23.957773208618164,
      "learning_rate": 3.046875e-06,
      "loss": 0.29045867919921875,
      "step": 53
    },
    {
      "epoch": 0.007300441913324151,
      "grad_norm": 22.4742488861084,
      "learning_rate": 3.10546875e-06,
      "loss": 0.281585693359375,
      "step": 54
    },
    {
      "epoch": 0.007435635282089414,
      "grad_norm": 18.76061248779297,
      "learning_rate": 3.1640625000000003e-06,
      "loss": 0.294525146484375,
      "step": 55
    },
    {
      "epoch": 0.007570828650854676,
      "grad_norm": 17.834041595458984,
      "learning_rate": 3.22265625e-06,
      "loss": 0.2843017578125,
      "step": 56
    },
    {
      "epoch": 0.0077060220196199375,
      "grad_norm": 18.969289779663086,
      "learning_rate": 3.28125e-06,
      "loss": 0.25803375244140625,
      "step": 57
    },
    {
      "epoch": 0.0078412153883852,
      "grad_norm": 16.735214233398438,
      "learning_rate": 3.3398437500000003e-06,
      "loss": 0.26306915283203125,
      "step": 58
    },
    {
      "epoch": 0.007976408757150461,
      "grad_norm": 14.971871376037598,
      "learning_rate": 3.3984375e-06,
      "loss": 0.26770782470703125,
      "step": 59
    },
    {
      "epoch": 0.008111602125915723,
      "grad_norm": 14.197041511535645,
      "learning_rate": 3.45703125e-06,
      "loss": 0.26336669921875,
      "step": 60
    },
    {
      "epoch": 0.008246795494680987,
      "grad_norm": 14.076885223388672,
      "learning_rate": 3.515625e-06,
      "loss": 0.2498626708984375,
      "step": 61
    },
    {
      "epoch": 0.008381988863446248,
      "grad_norm": 13.335953712463379,
      "learning_rate": 3.57421875e-06,
      "loss": 0.243988037109375,
      "step": 62
    },
    {
      "epoch": 0.00851718223221151,
      "grad_norm": 11.799309730529785,
      "learning_rate": 3.6328125e-06,
      "loss": 0.24451446533203125,
      "step": 63
    },
    {
      "epoch": 0.008652375600976772,
      "grad_norm": 12.843095779418945,
      "learning_rate": 3.69140625e-06,
      "loss": 0.21314239501953125,
      "step": 64
    },
    {
      "epoch": 0.008787568969742034,
      "grad_norm": 10.60861587524414,
      "learning_rate": 3.75e-06,
      "loss": 0.22778701782226562,
      "step": 65
    },
    {
      "epoch": 0.008922762338507296,
      "grad_norm": 4.623114585876465,
      "learning_rate": 3.8085937500000002e-06,
      "loss": 0.2890357971191406,
      "step": 66
    },
    {
      "epoch": 0.009057955707272558,
      "grad_norm": 7.7785725593566895,
      "learning_rate": 3.8671875e-06,
      "loss": 0.23848342895507812,
      "step": 67
    },
    {
      "epoch": 0.00919314907603782,
      "grad_norm": 7.840065956115723,
      "learning_rate": 3.92578125e-06,
      "loss": 0.22467803955078125,
      "step": 68
    },
    {
      "epoch": 0.009328342444803083,
      "grad_norm": 8.62641716003418,
      "learning_rate": 3.984375e-06,
      "loss": 0.20023727416992188,
      "step": 69
    },
    {
      "epoch": 0.009463535813568345,
      "grad_norm": 8.374451637268066,
      "learning_rate": 4.0429687500000004e-06,
      "loss": 0.19073486328125,
      "step": 70
    },
    {
      "epoch": 0.009598729182333607,
      "grad_norm": 7.400808811187744,
      "learning_rate": 4.1015625e-06,
      "loss": 0.190460205078125,
      "step": 71
    },
    {
      "epoch": 0.009733922551098869,
      "grad_norm": 6.118727207183838,
      "learning_rate": 4.16015625e-06,
      "loss": 0.1943359375,
      "step": 72
    },
    {
      "epoch": 0.00986911591986413,
      "grad_norm": 4.306054592132568,
      "learning_rate": 4.21875e-06,
      "loss": 0.21149826049804688,
      "step": 73
    },
    {
      "epoch": 0.010004309288629392,
      "grad_norm": 2.8444812297821045,
      "learning_rate": 4.27734375e-06,
      "loss": 0.22626876831054688,
      "step": 74
    },
    {
      "epoch": 0.010139502657394654,
      "grad_norm": 2.178194046020508,
      "learning_rate": 4.3359375e-06,
      "loss": 0.2353515625,
      "step": 75
    },
    {
      "epoch": 0.010274696026159916,
      "grad_norm": 6.895608425140381,
      "learning_rate": 4.3945312500000005e-06,
      "loss": 0.1833648681640625,
      "step": 76
    },
    {
      "epoch": 0.01040988939492518,
      "grad_norm": 3.906135082244873,
      "learning_rate": 4.453125e-06,
      "loss": 0.19829177856445312,
      "step": 77
    },
    {
      "epoch": 0.010545082763690442,
      "grad_norm": 3.4365999698638916,
      "learning_rate": 4.51171875e-06,
      "loss": 0.1984233856201172,
      "step": 78
    },
    {
      "epoch": 0.010680276132455703,
      "grad_norm": 8.071659088134766,
      "learning_rate": 4.5703125e-06,
      "loss": 0.14834213256835938,
      "step": 79
    },
    {
      "epoch": 0.010815469501220965,
      "grad_norm": 5.206031322479248,
      "learning_rate": 4.62890625e-06,
      "loss": 0.16190338134765625,
      "step": 80
    },
    {
      "epoch": 0.010950662869986227,
      "grad_norm": 9.365129470825195,
      "learning_rate": 4.6875000000000004e-06,
      "loss": 0.16129302978515625,
      "step": 81
    },
    {
      "epoch": 0.011085856238751489,
      "grad_norm": 15.743746757507324,
      "learning_rate": 4.74609375e-06,
      "loss": 0.18073081970214844,
      "step": 82
    },
    {
      "epoch": 0.01122104960751675,
      "grad_norm": 6.636305332183838,
      "learning_rate": 4.8046875e-06,
      "loss": 0.16434097290039062,
      "step": 83
    },
    {
      "epoch": 0.011356242976282013,
      "grad_norm": 26.415525436401367,
      "learning_rate": 4.86328125e-06,
      "loss": 0.21837615966796875,
      "step": 84
    },
    {
      "epoch": 0.011491436345047276,
      "grad_norm": 26.364681243896484,
      "learning_rate": 4.921875e-06,
      "loss": 0.2114429473876953,
      "step": 85
    },
    {
      "epoch": 0.011626629713812538,
      "grad_norm": 29.980968475341797,
      "learning_rate": 4.98046875e-06,
      "loss": 0.18030548095703125,
      "step": 86
    },
    {
      "epoch": 0.0117618230825778,
      "grad_norm": 33.83200454711914,
      "learning_rate": 5.0390625000000005e-06,
      "loss": 0.19058799743652344,
      "step": 87
    },
    {
      "epoch": 0.011897016451343062,
      "grad_norm": 8.606954574584961,
      "learning_rate": 5.09765625e-06,
      "loss": 0.17997169494628906,
      "step": 88
    },
    {
      "epoch": 0.012032209820108324,
      "grad_norm": 4.509511947631836,
      "learning_rate": 5.15625e-06,
      "loss": 0.1287860870361328,
      "step": 89
    },
    {
      "epoch": 0.012167403188873585,
      "grad_norm": 5.7349162101745605,
      "learning_rate": 5.21484375e-06,
      "loss": 0.1718158721923828,
      "step": 90
    },
    {
      "epoch": 0.012302596557638847,
      "grad_norm": 23.79114532470703,
      "learning_rate": 5.2734375e-06,
      "loss": 0.1282062530517578,
      "step": 91
    },
    {
      "epoch": 0.012437789926404109,
      "grad_norm": 6.960205554962158,
      "learning_rate": 5.3320312500000004e-06,
      "loss": 0.1727924346923828,
      "step": 92
    },
    {
      "epoch": 0.012572983295169373,
      "grad_norm": 5.630927085876465,
      "learning_rate": 5.390625e-06,
      "loss": 0.1682109832763672,
      "step": 93
    },
    {
      "epoch": 0.012708176663934635,
      "grad_norm": 33.38479232788086,
      "learning_rate": 5.44921875e-06,
      "loss": 0.1725749969482422,
      "step": 94
    },
    {
      "epoch": 0.012843370032699896,
      "grad_norm": 25.859128952026367,
      "learning_rate": 5.5078125e-06,
      "loss": 0.14342308044433594,
      "step": 95
    },
    {
      "epoch": 0.012978563401465158,
      "grad_norm": 17.265249252319336,
      "learning_rate": 5.56640625e-06,
      "loss": 0.1385517120361328,
      "step": 96
    },
    {
      "epoch": 0.01311375677023042,
      "grad_norm": 13.762451171875,
      "learning_rate": 5.625e-06,
      "loss": 0.19484519958496094,
      "step": 97
    },
    {
      "epoch": 0.013248950138995682,
      "grad_norm": 11.274735450744629,
      "learning_rate": 5.6835937500000005e-06,
      "loss": 0.13864517211914062,
      "step": 98
    },
    {
      "epoch": 0.013384143507760944,
      "grad_norm": 30.696157455444336,
      "learning_rate": 5.7421875e-06,
      "loss": 0.1842355728149414,
      "step": 99
    },
    {
      "epoch": 0.013519336876526206,
      "grad_norm": 42.534507751464844,
      "learning_rate": 5.80078125e-06,
      "loss": 0.15462398529052734,
      "step": 100
    },
    {
      "epoch": 0.01365453024529147,
      "grad_norm": 10.378792762756348,
      "learning_rate": 5.859375e-06,
      "loss": 0.11968231201171875,
      "step": 101
    },
    {
      "epoch": 0.013789723614056731,
      "grad_norm": 5.789700031280518,
      "learning_rate": 5.91796875e-06,
      "loss": 0.18852615356445312,
      "step": 102
    },
    {
      "epoch": 0.013924916982821993,
      "grad_norm": 35.43119430541992,
      "learning_rate": 5.9765625000000004e-06,
      "loss": 0.1937999725341797,
      "step": 103
    },
    {
      "epoch": 0.014060110351587255,
      "grad_norm": 14.727746963500977,
      "learning_rate": 6.03515625e-06,
      "loss": 0.16605758666992188,
      "step": 104
    },
    {
      "epoch": 0.014195303720352517,
      "grad_norm": 32.22308349609375,
      "learning_rate": 6.09375e-06,
      "loss": 0.14065933227539062,
      "step": 105
    },
    {
      "epoch": 0.014330497089117778,
      "grad_norm": 27.742570877075195,
      "learning_rate": 6.15234375e-06,
      "loss": 0.1688098907470703,
      "step": 106
    },
    {
      "epoch": 0.01446569045788304,
      "grad_norm": 6.195815563201904,
      "learning_rate": 6.2109375e-06,
      "loss": 0.11390304565429688,
      "step": 107
    },
    {
      "epoch": 0.014600883826648302,
      "grad_norm": 6.322120666503906,
      "learning_rate": 6.26953125e-06,
      "loss": 0.147125244140625,
      "step": 108
    },
    {
      "epoch": 0.014736077195413566,
      "grad_norm": 24.588651657104492,
      "learning_rate": 6.3281250000000005e-06,
      "loss": 0.20888996124267578,
      "step": 109
    },
    {
      "epoch": 0.014871270564178828,
      "grad_norm": 21.922161102294922,
      "learning_rate": 6.38671875e-06,
      "loss": 0.1313915252685547,
      "step": 110
    },
    {
      "epoch": 0.01500646393294409,
      "grad_norm": 14.735121726989746,
      "learning_rate": 6.4453125e-06,
      "loss": 0.12705707550048828,
      "step": 111
    },
    {
      "epoch": 0.015141657301709351,
      "grad_norm": 17.854413986206055,
      "learning_rate": 6.50390625e-06,
      "loss": 0.12715816497802734,
      "step": 112
    },
    {
      "epoch": 0.015276850670474613,
      "grad_norm": 15.228596687316895,
      "learning_rate": 6.5625e-06,
      "loss": 0.12988948822021484,
      "step": 113
    },
    {
      "epoch": 0.015412044039239875,
      "grad_norm": 11.644367218017578,
      "learning_rate": 6.6210937500000004e-06,
      "loss": 0.160980224609375,
      "step": 114
    },
    {
      "epoch": 0.015547237408005137,
      "grad_norm": 38.006412506103516,
      "learning_rate": 6.679687500000001e-06,
      "loss": 0.16860294342041016,
      "step": 115
    },
    {
      "epoch": 0.0156824307767704,
      "grad_norm": 18.42274284362793,
      "learning_rate": 6.73828125e-06,
      "loss": 0.10754013061523438,
      "step": 116
    },
    {
      "epoch": 0.015817624145535662,
      "grad_norm": 5.729445934295654,
      "learning_rate": 6.796875e-06,
      "loss": 0.13247394561767578,
      "step": 117
    },
    {
      "epoch": 0.015952817514300922,
      "grad_norm": 8.216533660888672,
      "learning_rate": 6.85546875e-06,
      "loss": 0.13490581512451172,
      "step": 118
    },
    {
      "epoch": 0.016088010883066186,
      "grad_norm": 28.392295837402344,
      "learning_rate": 6.9140625e-06,
      "loss": 0.13926315307617188,
      "step": 119
    },
    {
      "epoch": 0.016223204251831446,
      "grad_norm": 13.207259178161621,
      "learning_rate": 6.9726562500000005e-06,
      "loss": 0.12112712860107422,
      "step": 120
    },
    {
      "epoch": 0.01635839762059671,
      "grad_norm": 14.374978065490723,
      "learning_rate": 7.03125e-06,
      "loss": 0.1497325897216797,
      "step": 121
    },
    {
      "epoch": 0.016493590989361973,
      "grad_norm": 14.391389846801758,
      "learning_rate": 7.08984375e-06,
      "loss": 0.10625457763671875,
      "step": 122
    },
    {
      "epoch": 0.016628784358127233,
      "grad_norm": 9.666500091552734,
      "learning_rate": 7.1484375e-06,
      "loss": 0.12035751342773438,
      "step": 123
    },
    {
      "epoch": 0.016763977726892497,
      "grad_norm": 9.767655372619629,
      "learning_rate": 7.20703125e-06,
      "loss": 0.11335468292236328,
      "step": 124
    },
    {
      "epoch": 0.016899171095657757,
      "grad_norm": 19.927940368652344,
      "learning_rate": 7.265625e-06,
      "loss": 0.16068744659423828,
      "step": 125
    },
    {
      "epoch": 0.01703436446442302,
      "grad_norm": 24.16889190673828,
      "learning_rate": 7.3242187500000006e-06,
      "loss": 0.13986587524414062,
      "step": 126
    },
    {
      "epoch": 0.01716955783318828,
      "grad_norm": 15.296323776245117,
      "learning_rate": 7.3828125e-06,
      "loss": 0.1767120361328125,
      "step": 127
    },
    {
      "epoch": 0.017304751201953544,
      "grad_norm": 9.244256973266602,
      "learning_rate": 7.44140625e-06,
      "loss": 0.1714944839477539,
      "step": 128
    },
    {
      "epoch": 0.017439944570718808,
      "grad_norm": 18.481584548950195,
      "learning_rate": 7.5e-06,
      "loss": 0.13541030883789062,
      "step": 129
    },
    {
      "epoch": 0.017575137939484068,
      "grad_norm": 20.11408233642578,
      "learning_rate": 7.55859375e-06,
      "loss": 0.16762447357177734,
      "step": 130
    },
    {
      "epoch": 0.01771033130824933,
      "grad_norm": 12.163348197937012,
      "learning_rate": 7.6171875000000005e-06,
      "loss": 0.13053226470947266,
      "step": 131
    },
    {
      "epoch": 0.01784552467701459,
      "grad_norm": 10.763473510742188,
      "learning_rate": 7.67578125e-06,
      "loss": 0.14937305450439453,
      "step": 132
    },
    {
      "epoch": 0.017980718045779855,
      "grad_norm": 18.54998207092285,
      "learning_rate": 7.734375e-06,
      "loss": 0.18165111541748047,
      "step": 133
    },
    {
      "epoch": 0.018115911414545115,
      "grad_norm": 14.987088203430176,
      "learning_rate": 7.792968750000001e-06,
      "loss": 0.11704730987548828,
      "step": 134
    },
    {
      "epoch": 0.01825110478331038,
      "grad_norm": 24.72374153137207,
      "learning_rate": 7.8515625e-06,
      "loss": 0.15272188186645508,
      "step": 135
    },
    {
      "epoch": 0.01838629815207564,
      "grad_norm": 15.230940818786621,
      "learning_rate": 7.91015625e-06,
      "loss": 0.12484550476074219,
      "step": 136
    },
    {
      "epoch": 0.018521491520840903,
      "grad_norm": 10.244939804077148,
      "learning_rate": 7.96875e-06,
      "loss": 0.1355447769165039,
      "step": 137
    },
    {
      "epoch": 0.018656684889606166,
      "grad_norm": 5.154263973236084,
      "learning_rate": 8.02734375e-06,
      "loss": 0.12298297882080078,
      "step": 138
    },
    {
      "epoch": 0.018791878258371426,
      "grad_norm": 12.942228317260742,
      "learning_rate": 8.085937500000001e-06,
      "loss": 0.126708984375,
      "step": 139
    },
    {
      "epoch": 0.01892707162713669,
      "grad_norm": 9.724465370178223,
      "learning_rate": 8.14453125e-06,
      "loss": 0.10453987121582031,
      "step": 140
    },
    {
      "epoch": 0.01906226499590195,
      "grad_norm": 19.185394287109375,
      "learning_rate": 8.203125e-06,
      "loss": 0.15420866012573242,
      "step": 141
    },
    {
      "epoch": 0.019197458364667214,
      "grad_norm": 21.297924041748047,
      "learning_rate": 8.26171875e-06,
      "loss": 0.1211385726928711,
      "step": 142
    },
    {
      "epoch": 0.019332651733432474,
      "grad_norm": 5.746158123016357,
      "learning_rate": 8.3203125e-06,
      "loss": 0.08761024475097656,
      "step": 143
    },
    {
      "epoch": 0.019467845102197737,
      "grad_norm": 40.37514114379883,
      "learning_rate": 8.37890625e-06,
      "loss": 0.19170904159545898,
      "step": 144
    },
    {
      "epoch": 0.019603038470963,
      "grad_norm": 41.8349609375,
      "learning_rate": 8.4375e-06,
      "loss": 0.17560100555419922,
      "step": 145
    },
    {
      "epoch": 0.01973823183972826,
      "grad_norm": 15.578014373779297,
      "learning_rate": 8.49609375e-06,
      "loss": 0.14916515350341797,
      "step": 146
    },
    {
      "epoch": 0.019873425208493525,
      "grad_norm": 13.305092811584473,
      "learning_rate": 8.5546875e-06,
      "loss": 0.12070083618164062,
      "step": 147
    },
    {
      "epoch": 0.020008618577258785,
      "grad_norm": 16.86214256286621,
      "learning_rate": 8.61328125e-06,
      "loss": 0.1086874008178711,
      "step": 148
    },
    {
      "epoch": 0.02014381194602405,
      "grad_norm": 7.134828090667725,
      "learning_rate": 8.671875e-06,
      "loss": 0.13480234146118164,
      "step": 149
    },
    {
      "epoch": 0.02027900531478931,
      "grad_norm": 21.424076080322266,
      "learning_rate": 8.73046875e-06,
      "loss": 0.18094730377197266,
      "step": 150
    },
    {
      "epoch": 0.020414198683554572,
      "grad_norm": 13.057845115661621,
      "learning_rate": 8.789062500000001e-06,
      "loss": 0.12444591522216797,
      "step": 151
    },
    {
      "epoch": 0.020549392052319832,
      "grad_norm": 4.724929332733154,
      "learning_rate": 8.84765625e-06,
      "loss": 0.1180415153503418,
      "step": 152
    },
    {
      "epoch": 0.020684585421085096,
      "grad_norm": 20.824792861938477,
      "learning_rate": 8.90625e-06,
      "loss": 0.17496871948242188,
      "step": 153
    },
    {
      "epoch": 0.02081977878985036,
      "grad_norm": 25.18788719177246,
      "learning_rate": 8.96484375e-06,
      "loss": 0.1593494415283203,
      "step": 154
    },
    {
      "epoch": 0.02095497215861562,
      "grad_norm": 17.29711151123047,
      "learning_rate": 9.0234375e-06,
      "loss": 0.13325977325439453,
      "step": 155
    },
    {
      "epoch": 0.021090165527380883,
      "grad_norm": 31.869096755981445,
      "learning_rate": 9.082031250000001e-06,
      "loss": 0.1534104347229004,
      "step": 156
    },
    {
      "epoch": 0.021225358896146143,
      "grad_norm": 35.00883865356445,
      "learning_rate": 9.140625e-06,
      "loss": 0.1826152801513672,
      "step": 157
    },
    {
      "epoch": 0.021360552264911407,
      "grad_norm": 19.997268676757812,
      "learning_rate": 9.19921875e-06,
      "loss": 0.14528989791870117,
      "step": 158
    },
    {
      "epoch": 0.021495745633676667,
      "grad_norm": 8.106069564819336,
      "learning_rate": 9.2578125e-06,
      "loss": 0.13573265075683594,
      "step": 159
    },
    {
      "epoch": 0.02163093900244193,
      "grad_norm": 11.43966293334961,
      "learning_rate": 9.31640625e-06,
      "loss": 0.13237857818603516,
      "step": 160
    },
    {
      "epoch": 0.021766132371207194,
      "grad_norm": 2.361283302307129,
      "learning_rate": 9.375000000000001e-06,
      "loss": 0.1256718635559082,
      "step": 161
    },
    {
      "epoch": 0.021901325739972454,
      "grad_norm": 13.342789649963379,
      "learning_rate": 9.43359375e-06,
      "loss": 0.14096593856811523,
      "step": 162
    },
    {
      "epoch": 0.022036519108737718,
      "grad_norm": 8.534233093261719,
      "learning_rate": 9.4921875e-06,
      "loss": 0.12235212326049805,
      "step": 163
    },
    {
      "epoch": 0.022171712477502978,
      "grad_norm": 3.0962111949920654,
      "learning_rate": 9.55078125e-06,
      "loss": 0.13095426559448242,
      "step": 164
    },
    {
      "epoch": 0.02230690584626824,
      "grad_norm": 9.89217758178711,
      "learning_rate": 9.609375e-06,
      "loss": 0.08809947967529297,
      "step": 165
    },
    {
      "epoch": 0.0224420992150335,
      "grad_norm": 5.463018417358398,
      "learning_rate": 9.66796875e-06,
      "loss": 0.10288095474243164,
      "step": 166
    },
    {
      "epoch": 0.022577292583798765,
      "grad_norm": 8.819249153137207,
      "learning_rate": 9.7265625e-06,
      "loss": 0.13170766830444336,
      "step": 167
    },
    {
      "epoch": 0.022712485952564025,
      "grad_norm": 8.91738224029541,
      "learning_rate": 9.78515625e-06,
      "loss": 0.11728525161743164,
      "step": 168
    },
    {
      "epoch": 0.02284767932132929,
      "grad_norm": 3.5539848804473877,
      "learning_rate": 9.84375e-06,
      "loss": 0.11752843856811523,
      "step": 169
    },
    {
      "epoch": 0.022982872690094552,
      "grad_norm": 5.445803642272949,
      "learning_rate": 9.90234375e-06,
      "loss": 0.14904022216796875,
      "step": 170
    },
    {
      "epoch": 0.023118066058859812,
      "grad_norm": 5.846347332000732,
      "learning_rate": 9.9609375e-06,
      "loss": 0.09321165084838867,
      "step": 171
    },
    {
      "epoch": 0.023253259427625076,
      "grad_norm": 2.9143776893615723,
      "learning_rate": 1.001953125e-05,
      "loss": 0.11193323135375977,
      "step": 172
    },
    {
      "epoch": 0.023388452796390336,
      "grad_norm": 1.7001529932022095,
      "learning_rate": 1.0078125000000001e-05,
      "loss": 0.08716917037963867,
      "step": 173
    },
    {
      "epoch": 0.0235236461651556,
      "grad_norm": 4.321895122528076,
      "learning_rate": 1.013671875e-05,
      "loss": 0.1195530891418457,
      "step": 174
    },
    {
      "epoch": 0.02365883953392086,
      "grad_norm": 6.5774970054626465,
      "learning_rate": 1.01953125e-05,
      "loss": 0.13956212997436523,
      "step": 175
    },
    {
      "epoch": 0.023794032902686123,
      "grad_norm": 4.023632049560547,
      "learning_rate": 1.025390625e-05,
      "loss": 0.10661029815673828,
      "step": 176
    },
    {
      "epoch": 0.023929226271451387,
      "grad_norm": 4.219208240509033,
      "learning_rate": 1.03125e-05,
      "loss": 0.12139606475830078,
      "step": 177
    },
    {
      "epoch": 0.024064419640216647,
      "grad_norm": 2.6929924488067627,
      "learning_rate": 1.0371093750000001e-05,
      "loss": 0.11550712585449219,
      "step": 178
    },
    {
      "epoch": 0.02419961300898191,
      "grad_norm": 7.381962299346924,
      "learning_rate": 1.04296875e-05,
      "loss": 0.12038326263427734,
      "step": 179
    },
    {
      "epoch": 0.02433480637774717,
      "grad_norm": 2.777801752090454,
      "learning_rate": 1.048828125e-05,
      "loss": 0.10917186737060547,
      "step": 180
    },
    {
      "epoch": 0.024469999746512434,
      "grad_norm": 12.518394470214844,
      "learning_rate": 1.0546875e-05,
      "loss": 0.09705829620361328,
      "step": 181
    },
    {
      "epoch": 0.024605193115277695,
      "grad_norm": 11.730567932128906,
      "learning_rate": 1.060546875e-05,
      "loss": 0.11596536636352539,
      "step": 182
    },
    {
      "epoch": 0.024740386484042958,
      "grad_norm": 9.598226547241211,
      "learning_rate": 1.0664062500000001e-05,
      "loss": 0.15691423416137695,
      "step": 183
    },
    {
      "epoch": 0.024875579852808218,
      "grad_norm": 8.342377662658691,
      "learning_rate": 1.072265625e-05,
      "loss": 0.1196894645690918,
      "step": 184
    },
    {
      "epoch": 0.025010773221573482,
      "grad_norm": 8.918068885803223,
      "learning_rate": 1.078125e-05,
      "loss": 0.12574005126953125,
      "step": 185
    },
    {
      "epoch": 0.025145966590338745,
      "grad_norm": 3.005324363708496,
      "learning_rate": 1.083984375e-05,
      "loss": 0.12017583847045898,
      "step": 186
    },
    {
      "epoch": 0.025281159959104006,
      "grad_norm": 10.396531105041504,
      "learning_rate": 1.08984375e-05,
      "loss": 0.12627720832824707,
      "step": 187
    },
    {
      "epoch": 0.02541635332786927,
      "grad_norm": 23.966936111450195,
      "learning_rate": 1.095703125e-05,
      "loss": 0.169081449508667,
      "step": 188
    },
    {
      "epoch": 0.02555154669663453,
      "grad_norm": 14.574640274047852,
      "learning_rate": 1.1015625e-05,
      "loss": 0.12852835655212402,
      "step": 189
    },
    {
      "epoch": 0.025686740065399793,
      "grad_norm": 1.5561856031417847,
      "learning_rate": 1.1074218750000001e-05,
      "loss": 0.1138162612915039,
      "step": 190
    },
    {
      "epoch": 0.025821933434165053,
      "grad_norm": 12.832440376281738,
      "learning_rate": 1.11328125e-05,
      "loss": 0.11830949783325195,
      "step": 191
    },
    {
      "epoch": 0.025957126802930316,
      "grad_norm": 8.318037033081055,
      "learning_rate": 1.119140625e-05,
      "loss": 0.08885073661804199,
      "step": 192
    },
    {
      "epoch": 0.026092320171695577,
      "grad_norm": 2.1923887729644775,
      "learning_rate": 1.125e-05,
      "loss": 0.11017823219299316,
      "step": 193
    },
    {
      "epoch": 0.02622751354046084,
      "grad_norm": 19.12151336669922,
      "learning_rate": 1.130859375e-05,
      "loss": 0.17975258827209473,
      "step": 194
    },
    {
      "epoch": 0.026362706909226104,
      "grad_norm": 13.106063842773438,
      "learning_rate": 1.1367187500000001e-05,
      "loss": 0.12197279930114746,
      "step": 195
    },
    {
      "epoch": 0.026497900277991364,
      "grad_norm": 11.476738929748535,
      "learning_rate": 1.142578125e-05,
      "loss": 0.1254715919494629,
      "step": 196
    },
    {
      "epoch": 0.026633093646756627,
      "grad_norm": 2.7868235111236572,
      "learning_rate": 1.1484375e-05,
      "loss": 0.0923452377319336,
      "step": 197
    },
    {
      "epoch": 0.026768287015521888,
      "grad_norm": 4.950589179992676,
      "learning_rate": 1.154296875e-05,
      "loss": 0.1008601188659668,
      "step": 198
    },
    {
      "epoch": 0.02690348038428715,
      "grad_norm": 3.5299816131591797,
      "learning_rate": 1.16015625e-05,
      "loss": 0.1331310272216797,
      "step": 199
    },
    {
      "epoch": 0.02703867375305241,
      "grad_norm": 4.6229166984558105,
      "learning_rate": 1.1660156250000001e-05,
      "loss": 0.11776399612426758,
      "step": 200
    },
    {
      "epoch": 0.027173867121817675,
      "grad_norm": 5.123754978179932,
      "learning_rate": 1.171875e-05,
      "loss": 0.11093664169311523,
      "step": 201
    },
    {
      "epoch": 0.02730906049058294,
      "grad_norm": 1.3746099472045898,
      "learning_rate": 1.177734375e-05,
      "loss": 0.10559463500976562,
      "step": 202
    },
    {
      "epoch": 0.0274442538593482,
      "grad_norm": 5.895517349243164,
      "learning_rate": 1.18359375e-05,
      "loss": 0.1263742446899414,
      "step": 203
    },
    {
      "epoch": 0.027579447228113462,
      "grad_norm": 6.1869378089904785,
      "learning_rate": 1.189453125e-05,
      "loss": 0.1250779628753662,
      "step": 204
    },
    {
      "epoch": 0.027714640596878722,
      "grad_norm": 1.8732315301895142,
      "learning_rate": 1.1953125000000001e-05,
      "loss": 0.09824371337890625,
      "step": 205
    },
    {
      "epoch": 0.027849833965643986,
      "grad_norm": 3.040196657180786,
      "learning_rate": 1.201171875e-05,
      "loss": 0.13049840927124023,
      "step": 206
    },
    {
      "epoch": 0.027985027334409246,
      "grad_norm": 5.667483329772949,
      "learning_rate": 1.20703125e-05,
      "loss": 0.1139378547668457,
      "step": 207
    },
    {
      "epoch": 0.02812022070317451,
      "grad_norm": 1.674835205078125,
      "learning_rate": 1.212890625e-05,
      "loss": 0.08015131950378418,
      "step": 208
    },
    {
      "epoch": 0.02825541407193977,
      "grad_norm": 3.9440863132476807,
      "learning_rate": 1.21875e-05,
      "loss": 0.09508085250854492,
      "step": 209
    },
    {
      "epoch": 0.028390607440705033,
      "grad_norm": 2.1189498901367188,
      "learning_rate": 1.224609375e-05,
      "loss": 0.11002635955810547,
      "step": 210
    },
    {
      "epoch": 0.028525800809470297,
      "grad_norm": 2.2205352783203125,
      "learning_rate": 1.23046875e-05,
      "loss": 0.09447610378265381,
      "step": 211
    },
    {
      "epoch": 0.028660994178235557,
      "grad_norm": 9.254800796508789,
      "learning_rate": 1.2363281250000001e-05,
      "loss": 0.1143198013305664,
      "step": 212
    },
    {
      "epoch": 0.02879618754700082,
      "grad_norm": 6.819960117340088,
      "learning_rate": 1.2421875e-05,
      "loss": 0.10503458976745605,
      "step": 213
    },
    {
      "epoch": 0.02893138091576608,
      "grad_norm": 12.390182495117188,
      "learning_rate": 1.248046875e-05,
      "loss": 0.13003826141357422,
      "step": 214
    },
    {
      "epoch": 0.029066574284531344,
      "grad_norm": 2.660463333129883,
      "learning_rate": 1.25390625e-05,
      "loss": 0.0995473861694336,
      "step": 215
    },
    {
      "epoch": 0.029201767653296604,
      "grad_norm": 2.5199267864227295,
      "learning_rate": 1.259765625e-05,
      "loss": 0.09882307052612305,
      "step": 216
    },
    {
      "epoch": 0.029336961022061868,
      "grad_norm": 3.099677085876465,
      "learning_rate": 1.2656250000000001e-05,
      "loss": 0.1080394983291626,
      "step": 217
    },
    {
      "epoch": 0.02947215439082713,
      "grad_norm": 10.921390533447266,
      "learning_rate": 1.271484375e-05,
      "loss": 0.1387948989868164,
      "step": 218
    },
    {
      "epoch": 0.02960734775959239,
      "grad_norm": 1.8139811754226685,
      "learning_rate": 1.27734375e-05,
      "loss": 0.08462834358215332,
      "step": 219
    },
    {
      "epoch": 0.029742541128357655,
      "grad_norm": 2.6657485961914062,
      "learning_rate": 1.283203125e-05,
      "loss": 0.11264526844024658,
      "step": 220
    },
    {
      "epoch": 0.029877734497122915,
      "grad_norm": 8.381332397460938,
      "learning_rate": 1.2890625e-05,
      "loss": 0.12684082984924316,
      "step": 221
    },
    {
      "epoch": 0.03001292786588818,
      "grad_norm": 3.141138792037964,
      "learning_rate": 1.2949218750000001e-05,
      "loss": 0.1339268684387207,
      "step": 222
    },
    {
      "epoch": 0.03014812123465344,
      "grad_norm": 8.974042892456055,
      "learning_rate": 1.30078125e-05,
      "loss": 0.12345361709594727,
      "step": 223
    },
    {
      "epoch": 0.030283314603418703,
      "grad_norm": 5.3078436851501465,
      "learning_rate": 1.306640625e-05,
      "loss": 0.09230971336364746,
      "step": 224
    },
    {
      "epoch": 0.030418507972183963,
      "grad_norm": 1.307445764541626,
      "learning_rate": 1.3125e-05,
      "loss": 0.08745694160461426,
      "step": 225
    },
    {
      "epoch": 0.030553701340949226,
      "grad_norm": 4.963911056518555,
      "learning_rate": 1.318359375e-05,
      "loss": 0.09627199172973633,
      "step": 226
    },
    {
      "epoch": 0.03068889470971449,
      "grad_norm": 3.65700364112854,
      "learning_rate": 1.3242187500000001e-05,
      "loss": 0.10228323936462402,
      "step": 227
    },
    {
      "epoch": 0.03082408807847975,
      "grad_norm": 3.8001017570495605,
      "learning_rate": 1.330078125e-05,
      "loss": 0.12071537971496582,
      "step": 228
    },
    {
      "epoch": 0.030959281447245014,
      "grad_norm": 10.60185718536377,
      "learning_rate": 1.3359375000000001e-05,
      "loss": 0.10210299491882324,
      "step": 229
    },
    {
      "epoch": 0.031094474816010274,
      "grad_norm": 16.104564666748047,
      "learning_rate": 1.341796875e-05,
      "loss": 0.11900615692138672,
      "step": 230
    },
    {
      "epoch": 0.031229668184775537,
      "grad_norm": 8.709383010864258,
      "learning_rate": 1.34765625e-05,
      "loss": 0.15693950653076172,
      "step": 231
    },
    {
      "epoch": 0.0313648615535408,
      "grad_norm": 15.20444107055664,
      "learning_rate": 1.353515625e-05,
      "loss": 0.12135660648345947,
      "step": 232
    },
    {
      "epoch": 0.03150005492230606,
      "grad_norm": 14.495709419250488,
      "learning_rate": 1.359375e-05,
      "loss": 0.12809765338897705,
      "step": 233
    },
    {
      "epoch": 0.031635248291071325,
      "grad_norm": 1.579339861869812,
      "learning_rate": 1.3652343750000001e-05,
      "loss": 0.06402897834777832,
      "step": 234
    },
    {
      "epoch": 0.031770441659836585,
      "grad_norm": 4.174615383148193,
      "learning_rate": 1.37109375e-05,
      "loss": 0.09828853607177734,
      "step": 235
    },
    {
      "epoch": 0.031905635028601845,
      "grad_norm": 4.023648262023926,
      "learning_rate": 1.376953125e-05,
      "loss": 0.12398958206176758,
      "step": 236
    },
    {
      "epoch": 0.03204082839736711,
      "grad_norm": 2.566143035888672,
      "learning_rate": 1.3828125e-05,
      "loss": 0.09736800193786621,
      "step": 237
    },
    {
      "epoch": 0.03217602176613237,
      "grad_norm": 3.476241111755371,
      "learning_rate": 1.388671875e-05,
      "loss": 0.11614418029785156,
      "step": 238
    },
    {
      "epoch": 0.03231121513489763,
      "grad_norm": 2.4547955989837646,
      "learning_rate": 1.3945312500000001e-05,
      "loss": 0.10267853736877441,
      "step": 239
    },
    {
      "epoch": 0.03244640850366289,
      "grad_norm": 2.6375529766082764,
      "learning_rate": 1.400390625e-05,
      "loss": 0.08126914501190186,
      "step": 240
    },
    {
      "epoch": 0.03258160187242816,
      "grad_norm": 2.8025100231170654,
      "learning_rate": 1.40625e-05,
      "loss": 0.07041752338409424,
      "step": 241
    },
    {
      "epoch": 0.03271679524119342,
      "grad_norm": 3.087529182434082,
      "learning_rate": 1.412109375e-05,
      "loss": 0.08452367782592773,
      "step": 242
    },
    {
      "epoch": 0.03285198860995868,
      "grad_norm": 2.6189351081848145,
      "learning_rate": 1.41796875e-05,
      "loss": 0.10260534286499023,
      "step": 243
    },
    {
      "epoch": 0.032987181978723946,
      "grad_norm": 5.7942399978637695,
      "learning_rate": 1.4238281250000001e-05,
      "loss": 0.1002422571182251,
      "step": 244
    },
    {
      "epoch": 0.03312237534748921,
      "grad_norm": 2.233701705932617,
      "learning_rate": 1.4296875e-05,
      "loss": 0.1160893440246582,
      "step": 245
    },
    {
      "epoch": 0.03325756871625447,
      "grad_norm": 1.7791078090667725,
      "learning_rate": 1.435546875e-05,
      "loss": 0.14070630073547363,
      "step": 246
    },
    {
      "epoch": 0.03339276208501973,
      "grad_norm": 4.403859615325928,
      "learning_rate": 1.44140625e-05,
      "loss": 0.1036067008972168,
      "step": 247
    },
    {
      "epoch": 0.033527955453784994,
      "grad_norm": 4.750607967376709,
      "learning_rate": 1.447265625e-05,
      "loss": 0.10971546173095703,
      "step": 248
    },
    {
      "epoch": 0.033663148822550254,
      "grad_norm": 2.128018856048584,
      "learning_rate": 1.453125e-05,
      "loss": 0.0850905179977417,
      "step": 249
    },
    {
      "epoch": 0.033798342191315514,
      "grad_norm": 3.6158080101013184,
      "learning_rate": 1.458984375e-05,
      "loss": 0.0932382345199585,
      "step": 250
    },
    {
      "epoch": 0.03393353556008078,
      "grad_norm": 2.7658867835998535,
      "learning_rate": 1.4648437500000001e-05,
      "loss": 0.07552975416183472,
      "step": 251
    },
    {
      "epoch": 0.03406872892884604,
      "grad_norm": 4.574974060058594,
      "learning_rate": 1.470703125e-05,
      "loss": 0.0757284164428711,
      "step": 252
    },
    {
      "epoch": 0.0342039222976113,
      "grad_norm": 5.151226997375488,
      "learning_rate": 1.4765625e-05,
      "loss": 0.0752873420715332,
      "step": 253
    },
    {
      "epoch": 0.03433911566637656,
      "grad_norm": 2.2354824542999268,
      "learning_rate": 1.482421875e-05,
      "loss": 0.10274958610534668,
      "step": 254
    },
    {
      "epoch": 0.03447430903514183,
      "grad_norm": 2.9898624420166016,
      "learning_rate": 1.48828125e-05,
      "loss": 0.09230160713195801,
      "step": 255
    },
    {
      "epoch": 0.03460950240390709,
      "grad_norm": 4.158736228942871,
      "learning_rate": 1.4941406250000001e-05,
      "loss": 0.08805322647094727,
      "step": 256
    },
    {
      "epoch": 0.03474469577267235,
      "grad_norm": 1.8160712718963623,
      "learning_rate": 1.5e-05,
      "loss": 0.08407831192016602,
      "step": 257
    },
    {
      "epoch": 0.034879889141437616,
      "grad_norm": 11.49789047241211,
      "learning_rate": 1.505859375e-05,
      "loss": 0.11480516195297241,
      "step": 258
    },
    {
      "epoch": 0.035015082510202876,
      "grad_norm": 6.841657638549805,
      "learning_rate": 1.51171875e-05,
      "loss": 0.09477484226226807,
      "step": 259
    },
    {
      "epoch": 0.035150275878968136,
      "grad_norm": 2.060495615005493,
      "learning_rate": 1.517578125e-05,
      "loss": 0.07318687438964844,
      "step": 260
    },
    {
      "epoch": 0.035285469247733396,
      "grad_norm": 5.91054105758667,
      "learning_rate": 1.5234375000000001e-05,
      "loss": 0.12297433614730835,
      "step": 261
    },
    {
      "epoch": 0.03542066261649866,
      "grad_norm": 9.994206428527832,
      "learning_rate": 1.529296875e-05,
      "loss": 0.10230755805969238,
      "step": 262
    },
    {
      "epoch": 0.03555585598526392,
      "grad_norm": 9.000950813293457,
      "learning_rate": 1.53515625e-05,
      "loss": 0.10456538200378418,
      "step": 263
    },
    {
      "epoch": 0.03569104935402918,
      "grad_norm": 3.6928207874298096,
      "learning_rate": 1.541015625e-05,
      "loss": 0.09900754690170288,
      "step": 264
    },
    {
      "epoch": 0.035826242722794444,
      "grad_norm": 2.467583179473877,
      "learning_rate": 1.546875e-05,
      "loss": 0.08031642436981201,
      "step": 265
    },
    {
      "epoch": 0.03596143609155971,
      "grad_norm": 15.056185722351074,
      "learning_rate": 1.552734375e-05,
      "loss": 0.15174436569213867,
      "step": 266
    },
    {
      "epoch": 0.03609662946032497,
      "grad_norm": 8.540653228759766,
      "learning_rate": 1.5585937500000002e-05,
      "loss": 0.10565590858459473,
      "step": 267
    },
    {
      "epoch": 0.03623182282909023,
      "grad_norm": 4.716104507446289,
      "learning_rate": 1.564453125e-05,
      "loss": 0.09699249267578125,
      "step": 268
    },
    {
      "epoch": 0.0363670161978555,
      "grad_norm": 2.3965976238250732,
      "learning_rate": 1.5703125e-05,
      "loss": 0.11889266967773438,
      "step": 269
    },
    {
      "epoch": 0.03650220956662076,
      "grad_norm": 1.9796611070632935,
      "learning_rate": 1.576171875e-05,
      "loss": 0.09106159210205078,
      "step": 270
    },
    {
      "epoch": 0.03663740293538602,
      "grad_norm": 2.821047306060791,
      "learning_rate": 1.58203125e-05,
      "loss": 0.061782002449035645,
      "step": 271
    },
    {
      "epoch": 0.03677259630415128,
      "grad_norm": 4.7325310707092285,
      "learning_rate": 1.587890625e-05,
      "loss": 0.09113740921020508,
      "step": 272
    },
    {
      "epoch": 0.036907789672916545,
      "grad_norm": 2.668755054473877,
      "learning_rate": 1.59375e-05,
      "loss": 0.09154939651489258,
      "step": 273
    },
    {
      "epoch": 0.037042983041681805,
      "grad_norm": 5.259016513824463,
      "learning_rate": 1.599609375e-05,
      "loss": 0.11018514633178711,
      "step": 274
    },
    {
      "epoch": 0.037178176410447066,
      "grad_norm": 4.589200496673584,
      "learning_rate": 1.60546875e-05,
      "loss": 0.10451722145080566,
      "step": 275
    },
    {
      "epoch": 0.03731336977921233,
      "grad_norm": 2.119976758956909,
      "learning_rate": 1.611328125e-05,
      "loss": 0.07092821598052979,
      "step": 276
    },
    {
      "epoch": 0.03744856314797759,
      "grad_norm": 2.3516979217529297,
      "learning_rate": 1.6171875000000002e-05,
      "loss": 0.09616303443908691,
      "step": 277
    },
    {
      "epoch": 0.03758375651674285,
      "grad_norm": 1.3791124820709229,
      "learning_rate": 1.623046875e-05,
      "loss": 0.09056389331817627,
      "step": 278
    },
    {
      "epoch": 0.03771894988550811,
      "grad_norm": 8.076896667480469,
      "learning_rate": 1.62890625e-05,
      "loss": 0.11744379997253418,
      "step": 279
    },
    {
      "epoch": 0.03785414325427338,
      "grad_norm": 3.684563636779785,
      "learning_rate": 1.634765625e-05,
      "loss": 0.08807528018951416,
      "step": 280
    },
    {
      "epoch": 0.03798933662303864,
      "grad_norm": 4.626955509185791,
      "learning_rate": 1.640625e-05,
      "loss": 0.10694241523742676,
      "step": 281
    },
    {
      "epoch": 0.0381245299918039,
      "grad_norm": 7.342104911804199,
      "learning_rate": 1.646484375e-05,
      "loss": 0.09790802001953125,
      "step": 282
    },
    {
      "epoch": 0.03825972336056917,
      "grad_norm": 7.423117637634277,
      "learning_rate": 1.65234375e-05,
      "loss": 0.12955760955810547,
      "step": 283
    },
    {
      "epoch": 0.03839491672933443,
      "grad_norm": 2.925480604171753,
      "learning_rate": 1.6582031250000002e-05,
      "loss": 0.09709501266479492,
      "step": 284
    },
    {
      "epoch": 0.03853011009809969,
      "grad_norm": 9.6500825881958,
      "learning_rate": 1.6640625e-05,
      "loss": 0.09623748064041138,
      "step": 285
    },
    {
      "epoch": 0.03866530346686495,
      "grad_norm": 10.83087158203125,
      "learning_rate": 1.669921875e-05,
      "loss": 0.10638195276260376,
      "step": 286
    },
    {
      "epoch": 0.038800496835630215,
      "grad_norm": 5.118061065673828,
      "learning_rate": 1.67578125e-05,
      "loss": 0.0744776725769043,
      "step": 287
    },
    {
      "epoch": 0.038935690204395475,
      "grad_norm": 1.7427037954330444,
      "learning_rate": 1.681640625e-05,
      "loss": 0.09165191650390625,
      "step": 288
    },
    {
      "epoch": 0.039070883573160735,
      "grad_norm": 4.346134662628174,
      "learning_rate": 1.6875e-05,
      "loss": 0.08267402648925781,
      "step": 289
    },
    {
      "epoch": 0.039206076941926,
      "grad_norm": 2.6110188961029053,
      "learning_rate": 1.693359375e-05,
      "loss": 0.08368420600891113,
      "step": 290
    },
    {
      "epoch": 0.03934127031069126,
      "grad_norm": 3.2085206508636475,
      "learning_rate": 1.69921875e-05,
      "loss": 0.0966956615447998,
      "step": 291
    },
    {
      "epoch": 0.03947646367945652,
      "grad_norm": 2.1107547283172607,
      "learning_rate": 1.705078125e-05,
      "loss": 0.10750734806060791,
      "step": 292
    },
    {
      "epoch": 0.03961165704822178,
      "grad_norm": 1.9825631380081177,
      "learning_rate": 1.7109375e-05,
      "loss": 0.1081845760345459,
      "step": 293
    },
    {
      "epoch": 0.03974685041698705,
      "grad_norm": 2.3374953269958496,
      "learning_rate": 1.7167968750000002e-05,
      "loss": 0.10037648677825928,
      "step": 294
    },
    {
      "epoch": 0.03988204378575231,
      "grad_norm": 1.1485666036605835,
      "learning_rate": 1.72265625e-05,
      "loss": 0.06504380702972412,
      "step": 295
    },
    {
      "epoch": 0.04001723715451757,
      "grad_norm": 1.4012730121612549,
      "learning_rate": 1.728515625e-05,
      "loss": 0.10542726516723633,
      "step": 296
    },
    {
      "epoch": 0.04015243052328283,
      "grad_norm": 3.4263930320739746,
      "learning_rate": 1.734375e-05,
      "loss": 0.08764410018920898,
      "step": 297
    },
    {
      "epoch": 0.0402876238920481,
      "grad_norm": 2.7934041023254395,
      "learning_rate": 1.740234375e-05,
      "loss": 0.07379543781280518,
      "step": 298
    },
    {
      "epoch": 0.04042281726081336,
      "grad_norm": 2.6881332397460938,
      "learning_rate": 1.74609375e-05,
      "loss": 0.06783699989318848,
      "step": 299
    },
    {
      "epoch": 0.04055801062957862,
      "grad_norm": 6.708810806274414,
      "learning_rate": 1.751953125e-05,
      "loss": 0.10776853561401367,
      "step": 300
    },
    {
      "epoch": 0.040693203998343884,
      "grad_norm": 5.752043724060059,
      "learning_rate": 1.7578125000000002e-05,
      "loss": 0.09417986869812012,
      "step": 301
    },
    {
      "epoch": 0.040828397367109144,
      "grad_norm": 4.546014308929443,
      "learning_rate": 1.763671875e-05,
      "loss": 0.09645891189575195,
      "step": 302
    },
    {
      "epoch": 0.040963590735874404,
      "grad_norm": 7.75122594833374,
      "learning_rate": 1.76953125e-05,
      "loss": 0.1010664701461792,
      "step": 303
    },
    {
      "epoch": 0.041098784104639664,
      "grad_norm": 5.185743808746338,
      "learning_rate": 1.775390625e-05,
      "loss": 0.09725743532180786,
      "step": 304
    },
    {
      "epoch": 0.04123397747340493,
      "grad_norm": 1.7176100015640259,
      "learning_rate": 1.78125e-05,
      "loss": 0.08327817916870117,
      "step": 305
    },
    {
      "epoch": 0.04136917084217019,
      "grad_norm": 5.91613245010376,
      "learning_rate": 1.787109375e-05,
      "loss": 0.09066081047058105,
      "step": 306
    },
    {
      "epoch": 0.04150436421093545,
      "grad_norm": 5.451717376708984,
      "learning_rate": 1.79296875e-05,
      "loss": 0.10851454734802246,
      "step": 307
    },
    {
      "epoch": 0.04163955757970072,
      "grad_norm": 1.243614673614502,
      "learning_rate": 1.798828125e-05,
      "loss": 0.07808101177215576,
      "step": 308
    },
    {
      "epoch": 0.04177475094846598,
      "grad_norm": 3.067702293395996,
      "learning_rate": 1.8046875e-05,
      "loss": 0.06620454788208008,
      "step": 309
    },
    {
      "epoch": 0.04190994431723124,
      "grad_norm": 2.0741145610809326,
      "learning_rate": 1.810546875e-05,
      "loss": 0.0906221866607666,
      "step": 310
    },
    {
      "epoch": 0.0420451376859965,
      "grad_norm": 1.381949782371521,
      "learning_rate": 1.8164062500000002e-05,
      "loss": 0.09767007827758789,
      "step": 311
    },
    {
      "epoch": 0.042180331054761766,
      "grad_norm": 3.3403677940368652,
      "learning_rate": 1.822265625e-05,
      "loss": 0.08916366100311279,
      "step": 312
    },
    {
      "epoch": 0.042315524423527026,
      "grad_norm": 6.893871307373047,
      "learning_rate": 1.828125e-05,
      "loss": 0.10216140747070312,
      "step": 313
    },
    {
      "epoch": 0.042450717792292286,
      "grad_norm": 7.702648162841797,
      "learning_rate": 1.833984375e-05,
      "loss": 0.09231257438659668,
      "step": 314
    },
    {
      "epoch": 0.04258591116105755,
      "grad_norm": 1.366671085357666,
      "learning_rate": 1.83984375e-05,
      "loss": 0.0905611515045166,
      "step": 315
    },
    {
      "epoch": 0.04272110452982281,
      "grad_norm": 3.0089540481567383,
      "learning_rate": 1.845703125e-05,
      "loss": 0.07928508520126343,
      "step": 316
    },
    {
      "epoch": 0.042856297898588074,
      "grad_norm": 1.875510811805725,
      "learning_rate": 1.8515625e-05,
      "loss": 0.09736001491546631,
      "step": 317
    },
    {
      "epoch": 0.042991491267353334,
      "grad_norm": 1.8423699140548706,
      "learning_rate": 1.8574218750000002e-05,
      "loss": 0.1001434326171875,
      "step": 318
    },
    {
      "epoch": 0.0431266846361186,
      "grad_norm": 1.1062639951705933,
      "learning_rate": 1.86328125e-05,
      "loss": 0.04751598834991455,
      "step": 319
    },
    {
      "epoch": 0.04326187800488386,
      "grad_norm": 7.962644100189209,
      "learning_rate": 1.869140625e-05,
      "loss": 0.09258556365966797,
      "step": 320
    },
    {
      "epoch": 0.04339707137364912,
      "grad_norm": 9.39422607421875,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 0.11903655529022217,
      "step": 321
    },
    {
      "epoch": 0.04353226474241439,
      "grad_norm": 1.427226185798645,
      "learning_rate": 1.880859375e-05,
      "loss": 0.10064029693603516,
      "step": 322
    },
    {
      "epoch": 0.04366745811117965,
      "grad_norm": 3.133540630340576,
      "learning_rate": 1.88671875e-05,
      "loss": 0.06096315383911133,
      "step": 323
    },
    {
      "epoch": 0.04380265147994491,
      "grad_norm": 3.97255277633667,
      "learning_rate": 1.892578125e-05,
      "loss": 0.09675478935241699,
      "step": 324
    },
    {
      "epoch": 0.04393784484871017,
      "grad_norm": 1.0997313261032104,
      "learning_rate": 1.8984375e-05,
      "loss": 0.0714145302772522,
      "step": 325
    },
    {
      "epoch": 0.044073038217475435,
      "grad_norm": 5.258992671966553,
      "learning_rate": 1.904296875e-05,
      "loss": 0.1031460165977478,
      "step": 326
    },
    {
      "epoch": 0.044208231586240695,
      "grad_norm": 2.8044447898864746,
      "learning_rate": 1.91015625e-05,
      "loss": 0.07067525386810303,
      "step": 327
    },
    {
      "epoch": 0.044343424955005956,
      "grad_norm": 8.673918724060059,
      "learning_rate": 1.9160156250000002e-05,
      "loss": 0.1001860499382019,
      "step": 328
    },
    {
      "epoch": 0.044478618323771216,
      "grad_norm": 9.760953903198242,
      "learning_rate": 1.921875e-05,
      "loss": 0.10470497608184814,
      "step": 329
    },
    {
      "epoch": 0.04461381169253648,
      "grad_norm": 2.4386723041534424,
      "learning_rate": 1.927734375e-05,
      "loss": 0.07603228092193604,
      "step": 330
    },
    {
      "epoch": 0.04474900506130174,
      "grad_norm": 14.846535682678223,
      "learning_rate": 1.93359375e-05,
      "loss": 0.1105351597070694,
      "step": 331
    },
    {
      "epoch": 0.044884198430067,
      "grad_norm": 8.958602905273438,
      "learning_rate": 1.939453125e-05,
      "loss": 0.10117067396640778,
      "step": 332
    },
    {
      "epoch": 0.04501939179883227,
      "grad_norm": 8.11074161529541,
      "learning_rate": 1.9453125e-05,
      "loss": 0.10616093873977661,
      "step": 333
    },
    {
      "epoch": 0.04515458516759753,
      "grad_norm": 6.426239013671875,
      "learning_rate": 1.951171875e-05,
      "loss": 0.09173119068145752,
      "step": 334
    },
    {
      "epoch": 0.04528977853636279,
      "grad_norm": 11.673319816589355,
      "learning_rate": 1.95703125e-05,
      "loss": 0.10270881652832031,
      "step": 335
    },
    {
      "epoch": 0.04542497190512805,
      "grad_norm": 16.70077896118164,
      "learning_rate": 1.962890625e-05,
      "loss": 0.13658809661865234,
      "step": 336
    },
    {
      "epoch": 0.04556016527389332,
      "grad_norm": 16.626657485961914,
      "learning_rate": 1.96875e-05,
      "loss": 0.11444759368896484,
      "step": 337
    },
    {
      "epoch": 0.04569535864265858,
      "grad_norm": 5.551499843597412,
      "learning_rate": 1.9746093750000002e-05,
      "loss": 0.08840703964233398,
      "step": 338
    },
    {
      "epoch": 0.04583055201142384,
      "grad_norm": 1.8800626993179321,
      "learning_rate": 1.98046875e-05,
      "loss": 0.06582427024841309,
      "step": 339
    },
    {
      "epoch": 0.045965745380189105,
      "grad_norm": 6.9070000648498535,
      "learning_rate": 1.986328125e-05,
      "loss": 0.10113584995269775,
      "step": 340
    },
    {
      "epoch": 0.046100938748954365,
      "grad_norm": 6.177674770355225,
      "learning_rate": 1.9921875e-05,
      "loss": 0.10582369565963745,
      "step": 341
    },
    {
      "epoch": 0.046236132117719625,
      "grad_norm": 7.566307544708252,
      "learning_rate": 1.998046875e-05,
      "loss": 0.11455971002578735,
      "step": 342
    },
    {
      "epoch": 0.046371325486484885,
      "grad_norm": 1.430694580078125,
      "learning_rate": 2.00390625e-05,
      "loss": 0.07259982824325562,
      "step": 343
    },
    {
      "epoch": 0.04650651885525015,
      "grad_norm": 4.00285005569458,
      "learning_rate": 2.009765625e-05,
      "loss": 0.0822591781616211,
      "step": 344
    },
    {
      "epoch": 0.04664171222401541,
      "grad_norm": 3.8489129543304443,
      "learning_rate": 2.0156250000000002e-05,
      "loss": 0.10387992858886719,
      "step": 345
    },
    {
      "epoch": 0.04677690559278067,
      "grad_norm": 1.9880205392837524,
      "learning_rate": 2.021484375e-05,
      "loss": 0.09433865547180176,
      "step": 346
    },
    {
      "epoch": 0.04691209896154594,
      "grad_norm": 2.200664758682251,
      "learning_rate": 2.02734375e-05,
      "loss": 0.08370310068130493,
      "step": 347
    },
    {
      "epoch": 0.0470472923303112,
      "grad_norm": 1.9416441917419434,
      "learning_rate": 2.033203125e-05,
      "loss": 0.09031820297241211,
      "step": 348
    },
    {
      "epoch": 0.04718248569907646,
      "grad_norm": 2.0605087280273438,
      "learning_rate": 2.0390625e-05,
      "loss": 0.06338119506835938,
      "step": 349
    },
    {
      "epoch": 0.04731767906784172,
      "grad_norm": 3.013564348220825,
      "learning_rate": 2.044921875e-05,
      "loss": 0.10893821716308594,
      "step": 350
    },
    {
      "epoch": 0.04745287243660699,
      "grad_norm": 5.238158226013184,
      "learning_rate": 2.05078125e-05,
      "loss": 0.07785534858703613,
      "step": 351
    },
    {
      "epoch": 0.04758806580537225,
      "grad_norm": 6.897862911224365,
      "learning_rate": 2.056640625e-05,
      "loss": 0.08934783935546875,
      "step": 352
    },
    {
      "epoch": 0.04772325917413751,
      "grad_norm": 4.818323612213135,
      "learning_rate": 2.0625e-05,
      "loss": 0.07204675674438477,
      "step": 353
    },
    {
      "epoch": 0.047858452542902774,
      "grad_norm": 7.2365031242370605,
      "learning_rate": 2.068359375e-05,
      "loss": 0.12442159652709961,
      "step": 354
    },
    {
      "epoch": 0.047993645911668034,
      "grad_norm": 3.4385592937469482,
      "learning_rate": 2.0742187500000002e-05,
      "loss": 0.0787164568901062,
      "step": 355
    },
    {
      "epoch": 0.048128839280433294,
      "grad_norm": 5.228093147277832,
      "learning_rate": 2.080078125e-05,
      "loss": 0.1132049560546875,
      "step": 356
    },
    {
      "epoch": 0.048264032649198554,
      "grad_norm": 1.9497636556625366,
      "learning_rate": 2.0859375e-05,
      "loss": 0.09700345993041992,
      "step": 357
    },
    {
      "epoch": 0.04839922601796382,
      "grad_norm": 5.747951030731201,
      "learning_rate": 2.091796875e-05,
      "loss": 0.09501749277114868,
      "step": 358
    },
    {
      "epoch": 0.04853441938672908,
      "grad_norm": 1.3901766538619995,
      "learning_rate": 2.09765625e-05,
      "loss": 0.1029747724533081,
      "step": 359
    },
    {
      "epoch": 0.04866961275549434,
      "grad_norm": 1.3685859441757202,
      "learning_rate": 2.103515625e-05,
      "loss": 0.061025023460388184,
      "step": 360
    },
    {
      "epoch": 0.0488048061242596,
      "grad_norm": 1.9651250839233398,
      "learning_rate": 2.109375e-05,
      "loss": 0.08706998825073242,
      "step": 361
    },
    {
      "epoch": 0.04893999949302487,
      "grad_norm": 1.6208951473236084,
      "learning_rate": 2.1152343750000002e-05,
      "loss": 0.08523666858673096,
      "step": 362
    },
    {
      "epoch": 0.04907519286179013,
      "grad_norm": 4.106655120849609,
      "learning_rate": 2.12109375e-05,
      "loss": 0.11466550827026367,
      "step": 363
    },
    {
      "epoch": 0.04921038623055539,
      "grad_norm": 2.599555015563965,
      "learning_rate": 2.126953125e-05,
      "loss": 0.1063699722290039,
      "step": 364
    },
    {
      "epoch": 0.049345579599320656,
      "grad_norm": 2.2370595932006836,
      "learning_rate": 2.1328125000000002e-05,
      "loss": 0.07172304391860962,
      "step": 365
    },
    {
      "epoch": 0.049480772968085916,
      "grad_norm": 3.7827062606811523,
      "learning_rate": 2.138671875e-05,
      "loss": 0.0763203501701355,
      "step": 366
    },
    {
      "epoch": 0.049615966336851176,
      "grad_norm": 1.5155869722366333,
      "learning_rate": 2.14453125e-05,
      "loss": 0.09769988059997559,
      "step": 367
    },
    {
      "epoch": 0.049751159705616436,
      "grad_norm": 5.802121162414551,
      "learning_rate": 2.150390625e-05,
      "loss": 0.10133606195449829,
      "step": 368
    },
    {
      "epoch": 0.049886353074381704,
      "grad_norm": 4.722776412963867,
      "learning_rate": 2.15625e-05,
      "loss": 0.06864500045776367,
      "step": 369
    },
    {
      "epoch": 0.050021546443146964,
      "grad_norm": 4.014957904815674,
      "learning_rate": 2.162109375e-05,
      "loss": 0.06608808040618896,
      "step": 370
    },
    {
      "epoch": 0.050156739811912224,
      "grad_norm": 3.3784916400909424,
      "learning_rate": 2.16796875e-05,
      "loss": 0.08661270141601562,
      "step": 371
    },
    {
      "epoch": 0.05029193318067749,
      "grad_norm": 0.9976571798324585,
      "learning_rate": 2.1738281250000002e-05,
      "loss": 0.06785595417022705,
      "step": 372
    },
    {
      "epoch": 0.05042712654944275,
      "grad_norm": 3.677727222442627,
      "learning_rate": 2.1796875e-05,
      "loss": 0.08674323558807373,
      "step": 373
    },
    {
      "epoch": 0.05056231991820801,
      "grad_norm": 2.9944887161254883,
      "learning_rate": 2.185546875e-05,
      "loss": 0.07923847436904907,
      "step": 374
    },
    {
      "epoch": 0.05069751328697327,
      "grad_norm": 1.8594979047775269,
      "learning_rate": 2.19140625e-05,
      "loss": 0.0975416898727417,
      "step": 375
    },
    {
      "epoch": 0.05083270665573854,
      "grad_norm": 7.510520935058594,
      "learning_rate": 2.197265625e-05,
      "loss": 0.0970761775970459,
      "step": 376
    },
    {
      "epoch": 0.0509679000245038,
      "grad_norm": 7.723931312561035,
      "learning_rate": 2.203125e-05,
      "loss": 0.09255567193031311,
      "step": 377
    },
    {
      "epoch": 0.05110309339326906,
      "grad_norm": 3.676070213317871,
      "learning_rate": 2.208984375e-05,
      "loss": 0.08611512184143066,
      "step": 378
    },
    {
      "epoch": 0.051238286762034325,
      "grad_norm": 1.5313795804977417,
      "learning_rate": 2.2148437500000002e-05,
      "loss": 0.08518537878990173,
      "step": 379
    },
    {
      "epoch": 0.051373480130799586,
      "grad_norm": 2.037703037261963,
      "learning_rate": 2.220703125e-05,
      "loss": 0.0772629976272583,
      "step": 380
    },
    {
      "epoch": 0.051508673499564846,
      "grad_norm": 2.8023152351379395,
      "learning_rate": 2.2265625e-05,
      "loss": 0.0715177059173584,
      "step": 381
    },
    {
      "epoch": 0.051643866868330106,
      "grad_norm": 2.998567581176758,
      "learning_rate": 2.2324218750000002e-05,
      "loss": 0.10310161113739014,
      "step": 382
    },
    {
      "epoch": 0.05177906023709537,
      "grad_norm": 2.0281074047088623,
      "learning_rate": 2.23828125e-05,
      "loss": 0.05637192726135254,
      "step": 383
    },
    {
      "epoch": 0.05191425360586063,
      "grad_norm": 2.003948450088501,
      "learning_rate": 2.244140625e-05,
      "loss": 0.08940792083740234,
      "step": 384
    },
    {
      "epoch": 0.05204944697462589,
      "grad_norm": 3.8771634101867676,
      "learning_rate": 2.25e-05,
      "loss": 0.07818818092346191,
      "step": 385
    },
    {
      "epoch": 0.05218464034339115,
      "grad_norm": 2.335020065307617,
      "learning_rate": 2.255859375e-05,
      "loss": 0.07242804765701294,
      "step": 386
    },
    {
      "epoch": 0.05231983371215642,
      "grad_norm": 2.4745888710021973,
      "learning_rate": 2.26171875e-05,
      "loss": 0.0739285945892334,
      "step": 387
    },
    {
      "epoch": 0.05245502708092168,
      "grad_norm": 6.174681186676025,
      "learning_rate": 2.267578125e-05,
      "loss": 0.1017613410949707,
      "step": 388
    },
    {
      "epoch": 0.05259022044968694,
      "grad_norm": 4.108165264129639,
      "learning_rate": 2.2734375000000002e-05,
      "loss": 0.08620131015777588,
      "step": 389
    },
    {
      "epoch": 0.05272541381845221,
      "grad_norm": 2.4019179344177246,
      "learning_rate": 2.279296875e-05,
      "loss": 0.0928642749786377,
      "step": 390
    },
    {
      "epoch": 0.05286060718721747,
      "grad_norm": 0.8760350346565247,
      "learning_rate": 2.28515625e-05,
      "loss": 0.08316290378570557,
      "step": 391
    },
    {
      "epoch": 0.05299580055598273,
      "grad_norm": 5.666714191436768,
      "learning_rate": 2.291015625e-05,
      "loss": 0.09477740526199341,
      "step": 392
    },
    {
      "epoch": 0.05313099392474799,
      "grad_norm": 5.060141086578369,
      "learning_rate": 2.296875e-05,
      "loss": 0.09440922737121582,
      "step": 393
    },
    {
      "epoch": 0.053266187293513255,
      "grad_norm": 7.304379940032959,
      "learning_rate": 2.302734375e-05,
      "loss": 0.11314010620117188,
      "step": 394
    },
    {
      "epoch": 0.053401380662278515,
      "grad_norm": 1.990586757659912,
      "learning_rate": 2.30859375e-05,
      "loss": 0.09145021438598633,
      "step": 395
    },
    {
      "epoch": 0.053536574031043775,
      "grad_norm": 3.8750035762786865,
      "learning_rate": 2.3144531250000002e-05,
      "loss": 0.06574606895446777,
      "step": 396
    },
    {
      "epoch": 0.05367176739980904,
      "grad_norm": 3.6363365650177,
      "learning_rate": 2.3203125e-05,
      "loss": 0.07064366340637207,
      "step": 397
    },
    {
      "epoch": 0.0538069607685743,
      "grad_norm": 4.949976921081543,
      "learning_rate": 2.326171875e-05,
      "loss": 0.08974480628967285,
      "step": 398
    },
    {
      "epoch": 0.05394215413733956,
      "grad_norm": 2.357069730758667,
      "learning_rate": 2.3320312500000002e-05,
      "loss": 0.09490275382995605,
      "step": 399
    },
    {
      "epoch": 0.05407734750610482,
      "grad_norm": 2.3310916423797607,
      "learning_rate": 2.337890625e-05,
      "loss": 0.07440567016601562,
      "step": 400
    },
    {
      "epoch": 0.05421254087487009,
      "grad_norm": 1.8446704149246216,
      "learning_rate": 2.34375e-05,
      "loss": 0.06704896688461304,
      "step": 401
    },
    {
      "epoch": 0.05434773424363535,
      "grad_norm": 2.1257400512695312,
      "learning_rate": 2.349609375e-05,
      "loss": 0.07876241207122803,
      "step": 402
    },
    {
      "epoch": 0.05448292761240061,
      "grad_norm": 0.6201032996177673,
      "learning_rate": 2.35546875e-05,
      "loss": 0.06959986686706543,
      "step": 403
    },
    {
      "epoch": 0.05461812098116588,
      "grad_norm": 1.6564972400665283,
      "learning_rate": 2.361328125e-05,
      "loss": 0.05341958999633789,
      "step": 404
    },
    {
      "epoch": 0.05475331434993114,
      "grad_norm": 3.3766963481903076,
      "learning_rate": 2.3671875e-05,
      "loss": 0.09309649467468262,
      "step": 405
    },
    {
      "epoch": 0.0548885077186964,
      "grad_norm": 1.0515787601470947,
      "learning_rate": 2.3730468750000002e-05,
      "loss": 0.09221798181533813,
      "step": 406
    },
    {
      "epoch": 0.05502370108746166,
      "grad_norm": 4.562459945678711,
      "learning_rate": 2.37890625e-05,
      "loss": 0.11785227060317993,
      "step": 407
    },
    {
      "epoch": 0.055158894456226924,
      "grad_norm": 3.8840770721435547,
      "learning_rate": 2.384765625e-05,
      "loss": 0.08860963582992554,
      "step": 408
    },
    {
      "epoch": 0.055294087824992184,
      "grad_norm": 2.603105306625366,
      "learning_rate": 2.3906250000000002e-05,
      "loss": 0.06699860095977783,
      "step": 409
    },
    {
      "epoch": 0.055429281193757445,
      "grad_norm": 2.319427967071533,
      "learning_rate": 2.396484375e-05,
      "loss": 0.08991217613220215,
      "step": 410
    },
    {
      "epoch": 0.05556447456252271,
      "grad_norm": 2.4066224098205566,
      "learning_rate": 2.40234375e-05,
      "loss": 0.0811762809753418,
      "step": 411
    },
    {
      "epoch": 0.05569966793128797,
      "grad_norm": 1.0495777130126953,
      "learning_rate": 2.408203125e-05,
      "loss": 0.0670466423034668,
      "step": 412
    },
    {
      "epoch": 0.05583486130005323,
      "grad_norm": 2.1325764656066895,
      "learning_rate": 2.4140625e-05,
      "loss": 0.0867302417755127,
      "step": 413
    },
    {
      "epoch": 0.05597005466881849,
      "grad_norm": 0.9265480637550354,
      "learning_rate": 2.419921875e-05,
      "loss": 0.08390694856643677,
      "step": 414
    },
    {
      "epoch": 0.05610524803758376,
      "grad_norm": 2.2873709201812744,
      "learning_rate": 2.42578125e-05,
      "loss": 0.10381579399108887,
      "step": 415
    },
    {
      "epoch": 0.05624044140634902,
      "grad_norm": 2.1423025131225586,
      "learning_rate": 2.4316406250000002e-05,
      "loss": 0.0864713191986084,
      "step": 416
    },
    {
      "epoch": 0.05637563477511428,
      "grad_norm": 2.467217206954956,
      "learning_rate": 2.4375e-05,
      "loss": 0.09174138307571411,
      "step": 417
    },
    {
      "epoch": 0.05651082814387954,
      "grad_norm": 2.6019210815429688,
      "learning_rate": 2.443359375e-05,
      "loss": 0.08538818359375,
      "step": 418
    },
    {
      "epoch": 0.056646021512644806,
      "grad_norm": 1.3806113004684448,
      "learning_rate": 2.44921875e-05,
      "loss": 0.07542300224304199,
      "step": 419
    },
    {
      "epoch": 0.056781214881410066,
      "grad_norm": 1.9980984926223755,
      "learning_rate": 2.455078125e-05,
      "loss": 0.08560431003570557,
      "step": 420
    },
    {
      "epoch": 0.05691640825017533,
      "grad_norm": 2.729419708251953,
      "learning_rate": 2.4609375e-05,
      "loss": 0.08438265323638916,
      "step": 421
    },
    {
      "epoch": 0.057051601618940594,
      "grad_norm": 1.1821060180664062,
      "learning_rate": 2.466796875e-05,
      "loss": 0.050698280334472656,
      "step": 422
    },
    {
      "epoch": 0.057186794987705854,
      "grad_norm": 1.6192868947982788,
      "learning_rate": 2.4726562500000002e-05,
      "loss": 0.0817415714263916,
      "step": 423
    },
    {
      "epoch": 0.057321988356471114,
      "grad_norm": 1.4687600135803223,
      "learning_rate": 2.478515625e-05,
      "loss": 0.09992074966430664,
      "step": 424
    },
    {
      "epoch": 0.057457181725236374,
      "grad_norm": 4.619570255279541,
      "learning_rate": 2.484375e-05,
      "loss": 0.11551809310913086,
      "step": 425
    },
    {
      "epoch": 0.05759237509400164,
      "grad_norm": 3.8509414196014404,
      "learning_rate": 2.4902343750000002e-05,
      "loss": 0.07223761081695557,
      "step": 426
    },
    {
      "epoch": 0.0577275684627669,
      "grad_norm": 2.7280728816986084,
      "learning_rate": 2.49609375e-05,
      "loss": 0.07535338401794434,
      "step": 427
    },
    {
      "epoch": 0.05786276183153216,
      "grad_norm": 2.048595905303955,
      "learning_rate": 2.501953125e-05,
      "loss": 0.08168274164199829,
      "step": 428
    },
    {
      "epoch": 0.05799795520029743,
      "grad_norm": 2.0069172382354736,
      "learning_rate": 2.5078125e-05,
      "loss": 0.0855332612991333,
      "step": 429
    },
    {
      "epoch": 0.05813314856906269,
      "grad_norm": 2.6839518547058105,
      "learning_rate": 2.513671875e-05,
      "loss": 0.07594740390777588,
      "step": 430
    },
    {
      "epoch": 0.05826834193782795,
      "grad_norm": 1.9031440019607544,
      "learning_rate": 2.51953125e-05,
      "loss": 0.08525919914245605,
      "step": 431
    },
    {
      "epoch": 0.05840353530659321,
      "grad_norm": 2.022561550140381,
      "learning_rate": 2.525390625e-05,
      "loss": 0.0855109691619873,
      "step": 432
    },
    {
      "epoch": 0.058538728675358476,
      "grad_norm": 3.3286380767822266,
      "learning_rate": 2.5312500000000002e-05,
      "loss": 0.08324706554412842,
      "step": 433
    },
    {
      "epoch": 0.058673922044123736,
      "grad_norm": 2.32784104347229,
      "learning_rate": 2.537109375e-05,
      "loss": 0.09593796730041504,
      "step": 434
    },
    {
      "epoch": 0.058809115412888996,
      "grad_norm": 1.7362346649169922,
      "learning_rate": 2.54296875e-05,
      "loss": 0.08457082509994507,
      "step": 435
    },
    {
      "epoch": 0.05894430878165426,
      "grad_norm": 1.3542895317077637,
      "learning_rate": 2.548828125e-05,
      "loss": 0.08672499656677246,
      "step": 436
    },
    {
      "epoch": 0.05907950215041952,
      "grad_norm": 1.7253117561340332,
      "learning_rate": 2.5546875e-05,
      "loss": 0.09296071529388428,
      "step": 437
    },
    {
      "epoch": 0.05921469551918478,
      "grad_norm": 2.2812447547912598,
      "learning_rate": 2.560546875e-05,
      "loss": 0.06116044521331787,
      "step": 438
    },
    {
      "epoch": 0.05934988888795004,
      "grad_norm": 2.4810948371887207,
      "learning_rate": 2.56640625e-05,
      "loss": 0.07155156135559082,
      "step": 439
    },
    {
      "epoch": 0.05948508225671531,
      "grad_norm": 1.031551718711853,
      "learning_rate": 2.5722656250000002e-05,
      "loss": 0.08875316381454468,
      "step": 440
    },
    {
      "epoch": 0.05962027562548057,
      "grad_norm": 1.3491324186325073,
      "learning_rate": 2.578125e-05,
      "loss": 0.0728384256362915,
      "step": 441
    },
    {
      "epoch": 0.05975546899424583,
      "grad_norm": 4.78794527053833,
      "learning_rate": 2.583984375e-05,
      "loss": 0.07623755931854248,
      "step": 442
    },
    {
      "epoch": 0.0598906623630111,
      "grad_norm": 3.6588025093078613,
      "learning_rate": 2.5898437500000002e-05,
      "loss": 0.06744241714477539,
      "step": 443
    },
    {
      "epoch": 0.06002585573177636,
      "grad_norm": 1.644033670425415,
      "learning_rate": 2.595703125e-05,
      "loss": 0.08454227447509766,
      "step": 444
    },
    {
      "epoch": 0.06016104910054162,
      "grad_norm": 2.69010853767395,
      "learning_rate": 2.6015625e-05,
      "loss": 0.07021188735961914,
      "step": 445
    },
    {
      "epoch": 0.06029624246930688,
      "grad_norm": 2.966491222381592,
      "learning_rate": 2.607421875e-05,
      "loss": 0.10312604904174805,
      "step": 446
    },
    {
      "epoch": 0.060431435838072145,
      "grad_norm": 2.192582130432129,
      "learning_rate": 2.61328125e-05,
      "loss": 0.08733201026916504,
      "step": 447
    },
    {
      "epoch": 0.060566629206837405,
      "grad_norm": 3.28017520904541,
      "learning_rate": 2.619140625e-05,
      "loss": 0.11056116223335266,
      "step": 448
    },
    {
      "epoch": 0.060701822575602665,
      "grad_norm": 1.294121265411377,
      "learning_rate": 2.625e-05,
      "loss": 0.08379864692687988,
      "step": 449
    },
    {
      "epoch": 0.060837015944367925,
      "grad_norm": 2.3864622116088867,
      "learning_rate": 2.6308593750000002e-05,
      "loss": 0.06312590837478638,
      "step": 450
    },
    {
      "epoch": 0.06097220931313319,
      "grad_norm": 3.0067105293273926,
      "learning_rate": 2.63671875e-05,
      "loss": 0.07582861185073853,
      "step": 451
    },
    {
      "epoch": 0.06110740268189845,
      "grad_norm": 2.875403881072998,
      "learning_rate": 2.642578125e-05,
      "loss": 0.08906316757202148,
      "step": 452
    },
    {
      "epoch": 0.06124259605066371,
      "grad_norm": 3.5831849575042725,
      "learning_rate": 2.6484375000000002e-05,
      "loss": 0.10461187362670898,
      "step": 453
    },
    {
      "epoch": 0.06137778941942898,
      "grad_norm": 5.713657855987549,
      "learning_rate": 2.654296875e-05,
      "loss": 0.08087611198425293,
      "step": 454
    },
    {
      "epoch": 0.06151298278819424,
      "grad_norm": 4.763205051422119,
      "learning_rate": 2.66015625e-05,
      "loss": 0.09847617149353027,
      "step": 455
    },
    {
      "epoch": 0.0616481761569595,
      "grad_norm": 2.2618470191955566,
      "learning_rate": 2.666015625e-05,
      "loss": 0.07884633541107178,
      "step": 456
    },
    {
      "epoch": 0.06178336952572476,
      "grad_norm": 4.422484397888184,
      "learning_rate": 2.6718750000000002e-05,
      "loss": 0.09517449140548706,
      "step": 457
    },
    {
      "epoch": 0.06191856289449003,
      "grad_norm": 6.361186504364014,
      "learning_rate": 2.677734375e-05,
      "loss": 0.11206620931625366,
      "step": 458
    },
    {
      "epoch": 0.06205375626325529,
      "grad_norm": 1.5987541675567627,
      "learning_rate": 2.68359375e-05,
      "loss": 0.04566866159439087,
      "step": 459
    },
    {
      "epoch": 0.06218894963202055,
      "grad_norm": 3.8181939125061035,
      "learning_rate": 2.6894531250000002e-05,
      "loss": 0.09405183792114258,
      "step": 460
    },
    {
      "epoch": 0.062324143000785814,
      "grad_norm": 0.971908450126648,
      "learning_rate": 2.6953125e-05,
      "loss": 0.0977165699005127,
      "step": 461
    },
    {
      "epoch": 0.062459336369551074,
      "grad_norm": 4.934218883514404,
      "learning_rate": 2.701171875e-05,
      "loss": 0.07342004776000977,
      "step": 462
    },
    {
      "epoch": 0.06259452973831633,
      "grad_norm": 5.575514316558838,
      "learning_rate": 2.70703125e-05,
      "loss": 0.11276412010192871,
      "step": 463
    },
    {
      "epoch": 0.0627297231070816,
      "grad_norm": 6.003695964813232,
      "learning_rate": 2.712890625e-05,
      "loss": 0.09714007377624512,
      "step": 464
    },
    {
      "epoch": 0.06286491647584685,
      "grad_norm": 3.977492332458496,
      "learning_rate": 2.71875e-05,
      "loss": 0.12041759490966797,
      "step": 465
    },
    {
      "epoch": 0.06300010984461211,
      "grad_norm": 3.8524303436279297,
      "learning_rate": 2.724609375e-05,
      "loss": 0.0952996015548706,
      "step": 466
    },
    {
      "epoch": 0.06313530321337739,
      "grad_norm": 0.717673659324646,
      "learning_rate": 2.7304687500000002e-05,
      "loss": 0.0744776725769043,
      "step": 467
    },
    {
      "epoch": 0.06327049658214265,
      "grad_norm": 3.0212314128875732,
      "learning_rate": 2.736328125e-05,
      "loss": 0.09562563896179199,
      "step": 468
    },
    {
      "epoch": 0.06340568995090791,
      "grad_norm": 1.3355679512023926,
      "learning_rate": 2.7421875e-05,
      "loss": 0.057322025299072266,
      "step": 469
    },
    {
      "epoch": 0.06354088331967317,
      "grad_norm": 1.1125469207763672,
      "learning_rate": 2.7480468750000002e-05,
      "loss": 0.07634854316711426,
      "step": 470
    },
    {
      "epoch": 0.06367607668843843,
      "grad_norm": 0.9093754887580872,
      "learning_rate": 2.75390625e-05,
      "loss": 0.06976306438446045,
      "step": 471
    },
    {
      "epoch": 0.06381127005720369,
      "grad_norm": 0.6387351751327515,
      "learning_rate": 2.759765625e-05,
      "loss": 0.07417264580726624,
      "step": 472
    },
    {
      "epoch": 0.06394646342596895,
      "grad_norm": 0.9516288042068481,
      "learning_rate": 2.765625e-05,
      "loss": 0.0605187714099884,
      "step": 473
    },
    {
      "epoch": 0.06408165679473422,
      "grad_norm": 2.21431040763855,
      "learning_rate": 2.7714843750000002e-05,
      "loss": 0.0928950309753418,
      "step": 474
    },
    {
      "epoch": 0.06421685016349948,
      "grad_norm": 0.692080557346344,
      "learning_rate": 2.77734375e-05,
      "loss": 0.06328105926513672,
      "step": 475
    },
    {
      "epoch": 0.06435204353226474,
      "grad_norm": 1.8471779823303223,
      "learning_rate": 2.783203125e-05,
      "loss": 0.06584012508392334,
      "step": 476
    },
    {
      "epoch": 0.06448723690103,
      "grad_norm": 0.7896122932434082,
      "learning_rate": 2.7890625000000002e-05,
      "loss": 0.07595396041870117,
      "step": 477
    },
    {
      "epoch": 0.06462243026979526,
      "grad_norm": 0.8789713382720947,
      "learning_rate": 2.794921875e-05,
      "loss": 0.052197396755218506,
      "step": 478
    },
    {
      "epoch": 0.06475762363856052,
      "grad_norm": 1.795386552810669,
      "learning_rate": 2.80078125e-05,
      "loss": 0.06410270929336548,
      "step": 479
    },
    {
      "epoch": 0.06489281700732578,
      "grad_norm": 2.4986648559570312,
      "learning_rate": 2.806640625e-05,
      "loss": 0.09851151704788208,
      "step": 480
    },
    {
      "epoch": 0.06502801037609106,
      "grad_norm": 2.338993549346924,
      "learning_rate": 2.8125e-05,
      "loss": 0.08768641948699951,
      "step": 481
    },
    {
      "epoch": 0.06516320374485632,
      "grad_norm": 1.754858136177063,
      "learning_rate": 2.818359375e-05,
      "loss": 0.08150345087051392,
      "step": 482
    },
    {
      "epoch": 0.06529839711362158,
      "grad_norm": 1.4037175178527832,
      "learning_rate": 2.82421875e-05,
      "loss": 0.07955771684646606,
      "step": 483
    },
    {
      "epoch": 0.06543359048238684,
      "grad_norm": 2.536918878555298,
      "learning_rate": 2.8300781250000002e-05,
      "loss": 0.06898238509893417,
      "step": 484
    },
    {
      "epoch": 0.0655687838511521,
      "grad_norm": 2.3428120613098145,
      "learning_rate": 2.8359375e-05,
      "loss": 0.09016305208206177,
      "step": 485
    },
    {
      "epoch": 0.06570397721991736,
      "grad_norm": 1.6057106256484985,
      "learning_rate": 2.841796875e-05,
      "loss": 0.07038795948028564,
      "step": 486
    },
    {
      "epoch": 0.06583917058868262,
      "grad_norm": 1.4538984298706055,
      "learning_rate": 2.8476562500000002e-05,
      "loss": 0.07903861999511719,
      "step": 487
    },
    {
      "epoch": 0.06597436395744789,
      "grad_norm": 4.265200138092041,
      "learning_rate": 2.853515625e-05,
      "loss": 0.0682828426361084,
      "step": 488
    },
    {
      "epoch": 0.06610955732621315,
      "grad_norm": 1.1278741359710693,
      "learning_rate": 2.859375e-05,
      "loss": 0.07071077823638916,
      "step": 489
    },
    {
      "epoch": 0.06624475069497841,
      "grad_norm": 1.5449624061584473,
      "learning_rate": 2.865234375e-05,
      "loss": 0.07739472389221191,
      "step": 490
    },
    {
      "epoch": 0.06637994406374367,
      "grad_norm": 3.274684190750122,
      "learning_rate": 2.87109375e-05,
      "loss": 0.0890035629272461,
      "step": 491
    },
    {
      "epoch": 0.06651513743250893,
      "grad_norm": 1.5417492389678955,
      "learning_rate": 2.876953125e-05,
      "loss": 0.04077804088592529,
      "step": 492
    },
    {
      "epoch": 0.0666503308012742,
      "grad_norm": 0.8454017043113708,
      "learning_rate": 2.8828125e-05,
      "loss": 0.06175266206264496,
      "step": 493
    },
    {
      "epoch": 0.06678552417003945,
      "grad_norm": 2.7841005325317383,
      "learning_rate": 2.8886718750000002e-05,
      "loss": 0.08558833599090576,
      "step": 494
    },
    {
      "epoch": 0.06692071753880473,
      "grad_norm": 4.5589494705200195,
      "learning_rate": 2.89453125e-05,
      "loss": 0.07117438316345215,
      "step": 495
    },
    {
      "epoch": 0.06705591090756999,
      "grad_norm": 4.542187213897705,
      "learning_rate": 2.900390625e-05,
      "loss": 0.10571330785751343,
      "step": 496
    },
    {
      "epoch": 0.06719110427633525,
      "grad_norm": 1.6217830181121826,
      "learning_rate": 2.90625e-05,
      "loss": 0.10186421871185303,
      "step": 497
    },
    {
      "epoch": 0.06732629764510051,
      "grad_norm": 3.4604969024658203,
      "learning_rate": 2.912109375e-05,
      "loss": 0.08647912740707397,
      "step": 498
    },
    {
      "epoch": 0.06746149101386577,
      "grad_norm": 5.712713241577148,
      "learning_rate": 2.91796875e-05,
      "loss": 0.10363689064979553,
      "step": 499
    },
    {
      "epoch": 0.06759668438263103,
      "grad_norm": 4.759398937225342,
      "learning_rate": 2.923828125e-05,
      "loss": 0.09876960515975952,
      "step": 500
    },
    {
      "epoch": 0.06773187775139629,
      "grad_norm": 3.2786519527435303,
      "learning_rate": 2.9296875000000002e-05,
      "loss": 0.08078700304031372,
      "step": 501
    },
    {
      "epoch": 0.06786707112016156,
      "grad_norm": 1.162019968032837,
      "learning_rate": 2.935546875e-05,
      "loss": 0.06313735246658325,
      "step": 502
    },
    {
      "epoch": 0.06800226448892682,
      "grad_norm": 1.308668851852417,
      "learning_rate": 2.94140625e-05,
      "loss": 0.07171192765235901,
      "step": 503
    },
    {
      "epoch": 0.06813745785769208,
      "grad_norm": 2.578092575073242,
      "learning_rate": 2.9472656250000002e-05,
      "loss": 0.08997377753257751,
      "step": 504
    },
    {
      "epoch": 0.06827265122645734,
      "grad_norm": 1.9555453062057495,
      "learning_rate": 2.953125e-05,
      "loss": 0.07397922873497009,
      "step": 505
    },
    {
      "epoch": 0.0684078445952226,
      "grad_norm": 0.8753219246864319,
      "learning_rate": 2.958984375e-05,
      "loss": 0.07477635145187378,
      "step": 506
    },
    {
      "epoch": 0.06854303796398786,
      "grad_norm": 2.393610715866089,
      "learning_rate": 2.96484375e-05,
      "loss": 0.08859503269195557,
      "step": 507
    },
    {
      "epoch": 0.06867823133275312,
      "grad_norm": 1.1795601844787598,
      "learning_rate": 2.970703125e-05,
      "loss": 0.07753872871398926,
      "step": 508
    },
    {
      "epoch": 0.0688134247015184,
      "grad_norm": 0.7777423858642578,
      "learning_rate": 2.9765625e-05,
      "loss": 0.05384558439254761,
      "step": 509
    },
    {
      "epoch": 0.06894861807028366,
      "grad_norm": 1.0782616138458252,
      "learning_rate": 2.982421875e-05,
      "loss": 0.0829768031835556,
      "step": 510
    },
    {
      "epoch": 0.06908381143904892,
      "grad_norm": 2.6904473304748535,
      "learning_rate": 2.9882812500000002e-05,
      "loss": 0.07865709066390991,
      "step": 511
    },
    {
      "epoch": 0.06921900480781418,
      "grad_norm": 0.9617339372634888,
      "learning_rate": 2.994140625e-05,
      "loss": 0.08347833156585693,
      "step": 512
    },
    {
      "epoch": 0.06935419817657944,
      "grad_norm": 4.226638317108154,
      "learning_rate": 3e-05,
      "loss": 0.06993937492370605,
      "step": 513
    },
    {
      "epoch": 0.0694893915453447,
      "grad_norm": 5.760720729827881,
      "learning_rate": 2.9999998438460004e-05,
      "loss": 0.08640438318252563,
      "step": 514
    },
    {
      "epoch": 0.06962458491410996,
      "grad_norm": 7.5532145500183105,
      "learning_rate": 2.9999993753840344e-05,
      "loss": 0.11704134941101074,
      "step": 515
    },
    {
      "epoch": 0.06975977828287523,
      "grad_norm": 2.440594434738159,
      "learning_rate": 2.9999985946141995e-05,
      "loss": 0.08039286732673645,
      "step": 516
    },
    {
      "epoch": 0.06989497165164049,
      "grad_norm": 1.566990852355957,
      "learning_rate": 2.9999975015366586e-05,
      "loss": 0.05487659573554993,
      "step": 517
    },
    {
      "epoch": 0.07003016502040575,
      "grad_norm": 3.4617419242858887,
      "learning_rate": 2.9999960961516384e-05,
      "loss": 0.07434442639350891,
      "step": 518
    },
    {
      "epoch": 0.07016535838917101,
      "grad_norm": 4.425759315490723,
      "learning_rate": 2.9999943784594325e-05,
      "loss": 0.08561226725578308,
      "step": 519
    },
    {
      "epoch": 0.07030055175793627,
      "grad_norm": 0.42576825618743896,
      "learning_rate": 2.9999923484603975e-05,
      "loss": 0.0590876042842865,
      "step": 520
    },
    {
      "epoch": 0.07043574512670153,
      "grad_norm": 1.4820572137832642,
      "learning_rate": 2.999990006154957e-05,
      "loss": 0.05504500865936279,
      "step": 521
    },
    {
      "epoch": 0.07057093849546679,
      "grad_norm": 2.1246414184570312,
      "learning_rate": 2.9999873515435977e-05,
      "loss": 0.08048522472381592,
      "step": 522
    },
    {
      "epoch": 0.07070613186423207,
      "grad_norm": 0.5896093845367432,
      "learning_rate": 2.9999843846268735e-05,
      "loss": 0.07550835609436035,
      "step": 523
    },
    {
      "epoch": 0.07084132523299733,
      "grad_norm": 0.4509059190750122,
      "learning_rate": 2.9999811054054018e-05,
      "loss": 0.06059771776199341,
      "step": 524
    },
    {
      "epoch": 0.07097651860176259,
      "grad_norm": 2.119579553604126,
      "learning_rate": 2.9999775138798646e-05,
      "loss": 0.07610955834388733,
      "step": 525
    },
    {
      "epoch": 0.07111171197052785,
      "grad_norm": 0.6424253582954407,
      "learning_rate": 2.99997361005101e-05,
      "loss": 0.05062618851661682,
      "step": 526
    },
    {
      "epoch": 0.0712469053392931,
      "grad_norm": 0.4129358232021332,
      "learning_rate": 2.9999693939196513e-05,
      "loss": 0.04428476095199585,
      "step": 527
    },
    {
      "epoch": 0.07138209870805837,
      "grad_norm": 0.9434506893157959,
      "learning_rate": 2.999964865486666e-05,
      "loss": 0.06673705577850342,
      "step": 528
    },
    {
      "epoch": 0.07151729207682363,
      "grad_norm": 1.987471103668213,
      "learning_rate": 2.999960024752997e-05,
      "loss": 0.07437312602996826,
      "step": 529
    },
    {
      "epoch": 0.07165248544558889,
      "grad_norm": 1.8667017221450806,
      "learning_rate": 2.9999548717196516e-05,
      "loss": 0.08354781568050385,
      "step": 530
    },
    {
      "epoch": 0.07178767881435416,
      "grad_norm": 1.7856428623199463,
      "learning_rate": 2.999949406387703e-05,
      "loss": 0.09004330635070801,
      "step": 531
    },
    {
      "epoch": 0.07192287218311942,
      "grad_norm": 1.2929048538208008,
      "learning_rate": 2.9999436287582903e-05,
      "loss": 0.07697892189025879,
      "step": 532
    },
    {
      "epoch": 0.07205806555188468,
      "grad_norm": 1.8377665281295776,
      "learning_rate": 2.9999375388326145e-05,
      "loss": 0.0900411605834961,
      "step": 533
    },
    {
      "epoch": 0.07219325892064994,
      "grad_norm": 0.629837155342102,
      "learning_rate": 2.9999311366119447e-05,
      "loss": 0.07256340980529785,
      "step": 534
    },
    {
      "epoch": 0.0723284522894152,
      "grad_norm": 1.0979360342025757,
      "learning_rate": 2.9999244220976137e-05,
      "loss": 0.059804320335388184,
      "step": 535
    },
    {
      "epoch": 0.07246364565818046,
      "grad_norm": 1.495607614517212,
      "learning_rate": 2.9999173952910197e-05,
      "loss": 0.059341222047805786,
      "step": 536
    },
    {
      "epoch": 0.07259883902694572,
      "grad_norm": 2.1448705196380615,
      "learning_rate": 2.9999100561936252e-05,
      "loss": 0.06675183773040771,
      "step": 537
    },
    {
      "epoch": 0.072734032395711,
      "grad_norm": 2.3952038288116455,
      "learning_rate": 2.9999024048069585e-05,
      "loss": 0.06725847721099854,
      "step": 538
    },
    {
      "epoch": 0.07286922576447626,
      "grad_norm": 2.405796766281128,
      "learning_rate": 2.9998944411326127e-05,
      "loss": 0.05848759412765503,
      "step": 539
    },
    {
      "epoch": 0.07300441913324152,
      "grad_norm": 2.3532772064208984,
      "learning_rate": 2.999886165172246e-05,
      "loss": 0.08737562596797943,
      "step": 540
    },
    {
      "epoch": 0.07313961250200678,
      "grad_norm": 1.9639641046524048,
      "learning_rate": 2.9998775769275814e-05,
      "loss": 0.08353215456008911,
      "step": 541
    },
    {
      "epoch": 0.07327480587077204,
      "grad_norm": 1.4823710918426514,
      "learning_rate": 2.9998686764004067e-05,
      "loss": 0.08125150203704834,
      "step": 542
    },
    {
      "epoch": 0.0734099992395373,
      "grad_norm": 4.220778942108154,
      "learning_rate": 2.9998594635925755e-05,
      "loss": 0.10129022598266602,
      "step": 543
    },
    {
      "epoch": 0.07354519260830256,
      "grad_norm": 6.046679973602295,
      "learning_rate": 2.999849938506005e-05,
      "loss": 0.09592664241790771,
      "step": 544
    },
    {
      "epoch": 0.07368038597706783,
      "grad_norm": 3.6834492683410645,
      "learning_rate": 2.99984010114268e-05,
      "loss": 0.062161028385162354,
      "step": 545
    },
    {
      "epoch": 0.07381557934583309,
      "grad_norm": 3.0782973766326904,
      "learning_rate": 2.9998299515046475e-05,
      "loss": 0.08035112917423248,
      "step": 546
    },
    {
      "epoch": 0.07395077271459835,
      "grad_norm": 1.19057035446167,
      "learning_rate": 2.9998194895940213e-05,
      "loss": 0.06646406650543213,
      "step": 547
    },
    {
      "epoch": 0.07408596608336361,
      "grad_norm": 0.42260244488716125,
      "learning_rate": 2.9998087154129792e-05,
      "loss": 0.06359273195266724,
      "step": 548
    },
    {
      "epoch": 0.07422115945212887,
      "grad_norm": 1.7558311223983765,
      "learning_rate": 2.9997976289637645e-05,
      "loss": 0.0758281946182251,
      "step": 549
    },
    {
      "epoch": 0.07435635282089413,
      "grad_norm": 0.8219915628433228,
      "learning_rate": 2.9997862302486855e-05,
      "loss": 0.06392264366149902,
      "step": 550
    },
    {
      "epoch": 0.07449154618965939,
      "grad_norm": 1.645617127418518,
      "learning_rate": 2.9997745192701153e-05,
      "loss": 0.07526493072509766,
      "step": 551
    },
    {
      "epoch": 0.07462673955842467,
      "grad_norm": 0.48630088567733765,
      "learning_rate": 2.9997624960304926e-05,
      "loss": 0.06985509395599365,
      "step": 552
    },
    {
      "epoch": 0.07476193292718993,
      "grad_norm": 0.5417790412902832,
      "learning_rate": 2.9997501605323214e-05,
      "loss": 0.06212824583053589,
      "step": 553
    },
    {
      "epoch": 0.07489712629595519,
      "grad_norm": 0.9031376242637634,
      "learning_rate": 2.999737512778168e-05,
      "loss": 0.059658557176589966,
      "step": 554
    },
    {
      "epoch": 0.07503231966472045,
      "grad_norm": 1.3054426908493042,
      "learning_rate": 2.9997245527706674e-05,
      "loss": 0.054506897926330566,
      "step": 555
    },
    {
      "epoch": 0.0751675130334857,
      "grad_norm": 2.3331143856048584,
      "learning_rate": 2.999711280512517e-05,
      "loss": 0.07523204386234283,
      "step": 556
    },
    {
      "epoch": 0.07530270640225097,
      "grad_norm": 1.8983051776885986,
      "learning_rate": 2.9996976960064807e-05,
      "loss": 0.08041766285896301,
      "step": 557
    },
    {
      "epoch": 0.07543789977101623,
      "grad_norm": 2.1832733154296875,
      "learning_rate": 2.999683799255387e-05,
      "loss": 0.07708972692489624,
      "step": 558
    },
    {
      "epoch": 0.0755730931397815,
      "grad_norm": 1.53640615940094,
      "learning_rate": 2.999669590262129e-05,
      "loss": 0.07007499039173126,
      "step": 559
    },
    {
      "epoch": 0.07570828650854676,
      "grad_norm": 2.4458889961242676,
      "learning_rate": 2.999655069029665e-05,
      "loss": 0.0823965072631836,
      "step": 560
    },
    {
      "epoch": 0.07584347987731202,
      "grad_norm": 0.9590111970901489,
      "learning_rate": 2.9996402355610183e-05,
      "loss": 0.06776762008666992,
      "step": 561
    },
    {
      "epoch": 0.07597867324607728,
      "grad_norm": 0.502263069152832,
      "learning_rate": 2.9996250898592777e-05,
      "loss": 0.06980323791503906,
      "step": 562
    },
    {
      "epoch": 0.07611386661484254,
      "grad_norm": 1.8956334590911865,
      "learning_rate": 2.9996096319275962e-05,
      "loss": 0.06285884976387024,
      "step": 563
    },
    {
      "epoch": 0.0762490599836078,
      "grad_norm": 3.2829816341400146,
      "learning_rate": 2.9995938617691925e-05,
      "loss": 0.09168515354394913,
      "step": 564
    },
    {
      "epoch": 0.07638425335237306,
      "grad_norm": 1.0252958536148071,
      "learning_rate": 2.9995777793873504e-05,
      "loss": 0.0881321132183075,
      "step": 565
    },
    {
      "epoch": 0.07651944672113833,
      "grad_norm": 1.8914507627487183,
      "learning_rate": 2.9995613847854176e-05,
      "loss": 0.07276776432991028,
      "step": 566
    },
    {
      "epoch": 0.0766546400899036,
      "grad_norm": 1.0187723636627197,
      "learning_rate": 2.9995446779668078e-05,
      "loss": 0.05905801057815552,
      "step": 567
    },
    {
      "epoch": 0.07678983345866885,
      "grad_norm": 2.9869515895843506,
      "learning_rate": 2.9995276589349992e-05,
      "loss": 0.06865787506103516,
      "step": 568
    },
    {
      "epoch": 0.07692502682743411,
      "grad_norm": 2.001732349395752,
      "learning_rate": 2.9995103276935357e-05,
      "loss": 0.09824866056442261,
      "step": 569
    },
    {
      "epoch": 0.07706022019619937,
      "grad_norm": 1.145727276802063,
      "learning_rate": 2.9994926842460258e-05,
      "loss": 0.06862831115722656,
      "step": 570
    },
    {
      "epoch": 0.07719541356496464,
      "grad_norm": 2.2794854640960693,
      "learning_rate": 2.9994747285961428e-05,
      "loss": 0.07974135875701904,
      "step": 571
    },
    {
      "epoch": 0.0773306069337299,
      "grad_norm": 2.717662811279297,
      "learning_rate": 2.9994564607476255e-05,
      "loss": 0.07998615503311157,
      "step": 572
    },
    {
      "epoch": 0.07746580030249517,
      "grad_norm": 1.9366658926010132,
      "learning_rate": 2.9994378807042762e-05,
      "loss": 0.08855938911437988,
      "step": 573
    },
    {
      "epoch": 0.07760099367126043,
      "grad_norm": 1.265599012374878,
      "learning_rate": 2.9994189884699647e-05,
      "loss": 0.08413612842559814,
      "step": 574
    },
    {
      "epoch": 0.07773618704002569,
      "grad_norm": 0.8053746223449707,
      "learning_rate": 2.9993997840486233e-05,
      "loss": 0.08846956491470337,
      "step": 575
    },
    {
      "epoch": 0.07787138040879095,
      "grad_norm": 1.0635226964950562,
      "learning_rate": 2.9993802674442516e-05,
      "loss": 0.07659375667572021,
      "step": 576
    },
    {
      "epoch": 0.07800657377755621,
      "grad_norm": 0.9337055087089539,
      "learning_rate": 2.999360438660913e-05,
      "loss": 0.08028608560562134,
      "step": 577
    },
    {
      "epoch": 0.07814176714632147,
      "grad_norm": 1.3205060958862305,
      "learning_rate": 2.9993402977027346e-05,
      "loss": 0.05538719892501831,
      "step": 578
    },
    {
      "epoch": 0.07827696051508673,
      "grad_norm": 1.600622534751892,
      "learning_rate": 2.999319844573911e-05,
      "loss": 0.05767008662223816,
      "step": 579
    },
    {
      "epoch": 0.078412153883852,
      "grad_norm": 0.590765118598938,
      "learning_rate": 2.9992990792787007e-05,
      "loss": 0.03631296753883362,
      "step": 580
    },
    {
      "epoch": 0.07854734725261726,
      "grad_norm": 1.2146821022033691,
      "learning_rate": 2.999278001821427e-05,
      "loss": 0.05524358153343201,
      "step": 581
    },
    {
      "epoch": 0.07868254062138252,
      "grad_norm": 1.7299154996871948,
      "learning_rate": 2.9992566122064775e-05,
      "loss": 0.054759085178375244,
      "step": 582
    },
    {
      "epoch": 0.07881773399014778,
      "grad_norm": 1.8429731130599976,
      "learning_rate": 2.999234910438307e-05,
      "loss": 0.06211972236633301,
      "step": 583
    },
    {
      "epoch": 0.07895292735891304,
      "grad_norm": 2.8555471897125244,
      "learning_rate": 2.999212896521433e-05,
      "loss": 0.07378911972045898,
      "step": 584
    },
    {
      "epoch": 0.0790881207276783,
      "grad_norm": 0.525478720664978,
      "learning_rate": 2.999190570460439e-05,
      "loss": 0.0759100466966629,
      "step": 585
    },
    {
      "epoch": 0.07922331409644356,
      "grad_norm": 0.7448561191558838,
      "learning_rate": 2.9991679322599734e-05,
      "loss": 0.07554101943969727,
      "step": 586
    },
    {
      "epoch": 0.07935850746520882,
      "grad_norm": 2.3114960193634033,
      "learning_rate": 2.9991449819247505e-05,
      "loss": 0.06755508482456207,
      "step": 587
    },
    {
      "epoch": 0.0794937008339741,
      "grad_norm": 1.1557061672210693,
      "learning_rate": 2.9991217194595474e-05,
      "loss": 0.06702160835266113,
      "step": 588
    },
    {
      "epoch": 0.07962889420273936,
      "grad_norm": 2.0428478717803955,
      "learning_rate": 2.9990981448692078e-05,
      "loss": 0.07164239883422852,
      "step": 589
    },
    {
      "epoch": 0.07976408757150462,
      "grad_norm": 1.5015708208084106,
      "learning_rate": 2.999074258158641e-05,
      "loss": 0.08351626992225647,
      "step": 590
    },
    {
      "epoch": 0.07989928094026988,
      "grad_norm": 1.9153437614440918,
      "learning_rate": 2.9990500593328192e-05,
      "loss": 0.06143990159034729,
      "step": 591
    },
    {
      "epoch": 0.08003447430903514,
      "grad_norm": 0.9049278497695923,
      "learning_rate": 2.999025548396781e-05,
      "loss": 0.0896698534488678,
      "step": 592
    },
    {
      "epoch": 0.0801696676778004,
      "grad_norm": 0.7840317487716675,
      "learning_rate": 2.9990007253556302e-05,
      "loss": 0.06274673342704773,
      "step": 593
    },
    {
      "epoch": 0.08030486104656566,
      "grad_norm": 2.514416456222534,
      "learning_rate": 2.9989755902145345e-05,
      "loss": 0.08844804763793945,
      "step": 594
    },
    {
      "epoch": 0.08044005441533093,
      "grad_norm": 0.6164042353630066,
      "learning_rate": 2.9989501429787273e-05,
      "loss": 0.08358362317085266,
      "step": 595
    },
    {
      "epoch": 0.0805752477840962,
      "grad_norm": 0.4986203908920288,
      "learning_rate": 2.9989243836535073e-05,
      "loss": 0.0751338005065918,
      "step": 596
    },
    {
      "epoch": 0.08071044115286145,
      "grad_norm": 1.5278631448745728,
      "learning_rate": 2.998898312244237e-05,
      "loss": 0.07638660073280334,
      "step": 597
    },
    {
      "epoch": 0.08084563452162671,
      "grad_norm": 3.135820150375366,
      "learning_rate": 2.9988719287563452e-05,
      "loss": 0.08369582891464233,
      "step": 598
    },
    {
      "epoch": 0.08098082789039197,
      "grad_norm": 0.5640255212783813,
      "learning_rate": 2.998845233195325e-05,
      "loss": 0.05716729164123535,
      "step": 599
    },
    {
      "epoch": 0.08111602125915723,
      "grad_norm": 0.6250023245811462,
      "learning_rate": 2.998818225566734e-05,
      "loss": 0.05709981918334961,
      "step": 600
    },
    {
      "epoch": 0.0812512146279225,
      "grad_norm": 1.8859413862228394,
      "learning_rate": 2.998790905876196e-05,
      "loss": 0.0794154703617096,
      "step": 601
    },
    {
      "epoch": 0.08138640799668777,
      "grad_norm": 2.544265031814575,
      "learning_rate": 2.9987632741293987e-05,
      "loss": 0.08674255013465881,
      "step": 602
    },
    {
      "epoch": 0.08152160136545303,
      "grad_norm": 0.7365748882293701,
      "learning_rate": 2.998735330332096e-05,
      "loss": 0.061058998107910156,
      "step": 603
    },
    {
      "epoch": 0.08165679473421829,
      "grad_norm": 1.6129382848739624,
      "learning_rate": 2.9987070744901046e-05,
      "loss": 0.07812857627868652,
      "step": 604
    },
    {
      "epoch": 0.08179198810298355,
      "grad_norm": 1.4834262132644653,
      "learning_rate": 2.9986785066093084e-05,
      "loss": 0.05322533845901489,
      "step": 605
    },
    {
      "epoch": 0.08192718147174881,
      "grad_norm": 1.7839930057525635,
      "learning_rate": 2.9986496266956556e-05,
      "loss": 0.06453073024749756,
      "step": 606
    },
    {
      "epoch": 0.08206237484051407,
      "grad_norm": 2.200597047805786,
      "learning_rate": 2.9986204347551583e-05,
      "loss": 0.0850030779838562,
      "step": 607
    },
    {
      "epoch": 0.08219756820927933,
      "grad_norm": 1.8930333852767944,
      "learning_rate": 2.9985909307938948e-05,
      "loss": 0.05780499428510666,
      "step": 608
    },
    {
      "epoch": 0.0823327615780446,
      "grad_norm": 0.8153626918792725,
      "learning_rate": 2.9985611148180082e-05,
      "loss": 0.07434721291065216,
      "step": 609
    },
    {
      "epoch": 0.08246795494680986,
      "grad_norm": 2.432506561279297,
      "learning_rate": 2.9985309868337063e-05,
      "loss": 0.054645657539367676,
      "step": 610
    },
    {
      "epoch": 0.08260314831557512,
      "grad_norm": 0.8336548805236816,
      "learning_rate": 2.9985005468472617e-05,
      "loss": 0.07022076845169067,
      "step": 611
    },
    {
      "epoch": 0.08273834168434038,
      "grad_norm": 0.8005855083465576,
      "learning_rate": 2.9984697948650124e-05,
      "loss": 0.04986768960952759,
      "step": 612
    },
    {
      "epoch": 0.08287353505310564,
      "grad_norm": 0.8254513144493103,
      "learning_rate": 2.998438730893361e-05,
      "loss": 0.07963043451309204,
      "step": 613
    },
    {
      "epoch": 0.0830087284218709,
      "grad_norm": 2.1858198642730713,
      "learning_rate": 2.9984073549387747e-05,
      "loss": 0.07175400853157043,
      "step": 614
    },
    {
      "epoch": 0.08314392179063616,
      "grad_norm": 2.355132818222046,
      "learning_rate": 2.998375667007787e-05,
      "loss": 0.07016070187091827,
      "step": 615
    },
    {
      "epoch": 0.08327911515940144,
      "grad_norm": 0.599641740322113,
      "learning_rate": 2.998343667106995e-05,
      "loss": 0.058913350105285645,
      "step": 616
    },
    {
      "epoch": 0.0834143085281667,
      "grad_norm": 0.9613671898841858,
      "learning_rate": 2.9983113552430616e-05,
      "loss": 0.07916253805160522,
      "step": 617
    },
    {
      "epoch": 0.08354950189693196,
      "grad_norm": 3.1716086864471436,
      "learning_rate": 2.9982787314227134e-05,
      "loss": 0.07137918472290039,
      "step": 618
    },
    {
      "epoch": 0.08368469526569722,
      "grad_norm": 1.6789171695709229,
      "learning_rate": 2.998245795652744e-05,
      "loss": 0.05489552021026611,
      "step": 619
    },
    {
      "epoch": 0.08381988863446248,
      "grad_norm": 2.5331814289093018,
      "learning_rate": 2.9982125479400106e-05,
      "loss": 0.07533860206604004,
      "step": 620
    },
    {
      "epoch": 0.08395508200322774,
      "grad_norm": 1.989997148513794,
      "learning_rate": 2.9981789882914352e-05,
      "loss": 0.08089403808116913,
      "step": 621
    },
    {
      "epoch": 0.084090275371993,
      "grad_norm": 0.8958866596221924,
      "learning_rate": 2.9981451167140048e-05,
      "loss": 0.05067843198776245,
      "step": 622
    },
    {
      "epoch": 0.08422546874075827,
      "grad_norm": 3.3056395053863525,
      "learning_rate": 2.9981109332147722e-05,
      "loss": 0.0888616293668747,
      "step": 623
    },
    {
      "epoch": 0.08436066210952353,
      "grad_norm": 0.6550520658493042,
      "learning_rate": 2.9980764378008545e-05,
      "loss": 0.04205785691738129,
      "step": 624
    },
    {
      "epoch": 0.08449585547828879,
      "grad_norm": 1.6792329549789429,
      "learning_rate": 2.9980416304794332e-05,
      "loss": 0.05595056712627411,
      "step": 625
    },
    {
      "epoch": 0.08463104884705405,
      "grad_norm": 1.2456620931625366,
      "learning_rate": 2.9980065112577565e-05,
      "loss": 0.06043350696563721,
      "step": 626
    },
    {
      "epoch": 0.08476624221581931,
      "grad_norm": 1.6745126247406006,
      "learning_rate": 2.9979710801431357e-05,
      "loss": 0.06566761434078217,
      "step": 627
    },
    {
      "epoch": 0.08490143558458457,
      "grad_norm": 2.0548148155212402,
      "learning_rate": 2.997935337142948e-05,
      "loss": 0.06921306252479553,
      "step": 628
    },
    {
      "epoch": 0.08503662895334983,
      "grad_norm": 1.9589701890945435,
      "learning_rate": 2.9978992822646347e-05,
      "loss": 0.08202598243951797,
      "step": 629
    },
    {
      "epoch": 0.0851718223221151,
      "grad_norm": 0.798828125,
      "learning_rate": 2.9978629155157036e-05,
      "loss": 0.06456416845321655,
      "step": 630
    },
    {
      "epoch": 0.08530701569088037,
      "grad_norm": 1.6530522108078003,
      "learning_rate": 2.9978262369037252e-05,
      "loss": 0.05158402770757675,
      "step": 631
    },
    {
      "epoch": 0.08544220905964563,
      "grad_norm": 1.7948981523513794,
      "learning_rate": 2.9977892464363375e-05,
      "loss": 0.09780353307723999,
      "step": 632
    },
    {
      "epoch": 0.08557740242841089,
      "grad_norm": 0.8527246117591858,
      "learning_rate": 2.9977519441212412e-05,
      "loss": 0.054360270500183105,
      "step": 633
    },
    {
      "epoch": 0.08571259579717615,
      "grad_norm": 2.158431053161621,
      "learning_rate": 2.9977143299662034e-05,
      "loss": 0.09504848718643188,
      "step": 634
    },
    {
      "epoch": 0.08584778916594141,
      "grad_norm": 0.8429413437843323,
      "learning_rate": 2.997676403979055e-05,
      "loss": 0.0699925422668457,
      "step": 635
    },
    {
      "epoch": 0.08598298253470667,
      "grad_norm": 1.0217684507369995,
      "learning_rate": 2.997638166167693e-05,
      "loss": 0.06281322240829468,
      "step": 636
    },
    {
      "epoch": 0.08611817590347194,
      "grad_norm": 1.3085033893585205,
      "learning_rate": 2.9975996165400786e-05,
      "loss": 0.07656960189342499,
      "step": 637
    },
    {
      "epoch": 0.0862533692722372,
      "grad_norm": 0.8830587267875671,
      "learning_rate": 2.9975607551042373e-05,
      "loss": 0.06641332060098648,
      "step": 638
    },
    {
      "epoch": 0.08638856264100246,
      "grad_norm": 0.7558915615081787,
      "learning_rate": 2.9975215818682607e-05,
      "loss": 0.08561281859874725,
      "step": 639
    },
    {
      "epoch": 0.08652375600976772,
      "grad_norm": 0.6917787194252014,
      "learning_rate": 2.9974820968403056e-05,
      "loss": 0.06669548153877258,
      "step": 640
    },
    {
      "epoch": 0.08665894937853298,
      "grad_norm": 2.1940808296203613,
      "learning_rate": 2.9974423000285923e-05,
      "loss": 0.07260659337043762,
      "step": 641
    },
    {
      "epoch": 0.08679414274729824,
      "grad_norm": 0.9054264426231384,
      "learning_rate": 2.9974021914414068e-05,
      "loss": 0.07992389798164368,
      "step": 642
    },
    {
      "epoch": 0.0869293361160635,
      "grad_norm": 1.1337133646011353,
      "learning_rate": 2.9973617710871e-05,
      "loss": 0.07288509607315063,
      "step": 643
    },
    {
      "epoch": 0.08706452948482878,
      "grad_norm": 2.689429759979248,
      "learning_rate": 2.997321038974087e-05,
      "loss": 0.07699355483055115,
      "step": 644
    },
    {
      "epoch": 0.08719972285359404,
      "grad_norm": 1.0502344369888306,
      "learning_rate": 2.997279995110849e-05,
      "loss": 0.082648366689682,
      "step": 645
    },
    {
      "epoch": 0.0873349162223593,
      "grad_norm": 2.049832820892334,
      "learning_rate": 2.997238639505932e-05,
      "loss": 0.08546966314315796,
      "step": 646
    },
    {
      "epoch": 0.08747010959112456,
      "grad_norm": 2.7138779163360596,
      "learning_rate": 2.997196972167946e-05,
      "loss": 0.0652693510055542,
      "step": 647
    },
    {
      "epoch": 0.08760530295988982,
      "grad_norm": 1.1469850540161133,
      "learning_rate": 2.9971549931055665e-05,
      "loss": 0.07901674509048462,
      "step": 648
    },
    {
      "epoch": 0.08774049632865508,
      "grad_norm": 0.536758303642273,
      "learning_rate": 2.997112702327533e-05,
      "loss": 0.06264278292655945,
      "step": 649
    },
    {
      "epoch": 0.08787568969742034,
      "grad_norm": 0.7747340202331543,
      "learning_rate": 2.9970700998426518e-05,
      "loss": 0.07239297032356262,
      "step": 650
    },
    {
      "epoch": 0.0880108830661856,
      "grad_norm": 2.103119373321533,
      "learning_rate": 2.9970271856597925e-05,
      "loss": 0.06559926271438599,
      "step": 651
    },
    {
      "epoch": 0.08814607643495087,
      "grad_norm": 2.075385332107544,
      "learning_rate": 2.9969839597878896e-05,
      "loss": 0.06409762799739838,
      "step": 652
    },
    {
      "epoch": 0.08828126980371613,
      "grad_norm": 2.3380393981933594,
      "learning_rate": 2.9969404222359436e-05,
      "loss": 0.09425145387649536,
      "step": 653
    },
    {
      "epoch": 0.08841646317248139,
      "grad_norm": 3.265293598175049,
      "learning_rate": 2.9968965730130188e-05,
      "loss": 0.06211280822753906,
      "step": 654
    },
    {
      "epoch": 0.08855165654124665,
      "grad_norm": 3.7281758785247803,
      "learning_rate": 2.9968524121282455e-05,
      "loss": 0.07705247402191162,
      "step": 655
    },
    {
      "epoch": 0.08868684991001191,
      "grad_norm": 4.030885696411133,
      "learning_rate": 2.9968079395908178e-05,
      "loss": 0.09976205229759216,
      "step": 656
    },
    {
      "epoch": 0.08882204327877717,
      "grad_norm": 1.7111213207244873,
      "learning_rate": 2.9967631554099947e-05,
      "loss": 0.05883723497390747,
      "step": 657
    },
    {
      "epoch": 0.08895723664754243,
      "grad_norm": 1.616750955581665,
      "learning_rate": 2.996718059595101e-05,
      "loss": 0.07166612148284912,
      "step": 658
    },
    {
      "epoch": 0.0890924300163077,
      "grad_norm": 1.1883772611618042,
      "learning_rate": 2.9966726521555265e-05,
      "loss": 0.07008233666419983,
      "step": 659
    },
    {
      "epoch": 0.08922762338507297,
      "grad_norm": 1.9944844245910645,
      "learning_rate": 2.996626933100724e-05,
      "loss": 0.06989702582359314,
      "step": 660
    },
    {
      "epoch": 0.08936281675383823,
      "grad_norm": 1.8264142274856567,
      "learning_rate": 2.996580902440213e-05,
      "loss": 0.06547290086746216,
      "step": 661
    },
    {
      "epoch": 0.08949801012260349,
      "grad_norm": 2.1894795894622803,
      "learning_rate": 2.9965345601835773e-05,
      "loss": 0.06390075385570526,
      "step": 662
    },
    {
      "epoch": 0.08963320349136875,
      "grad_norm": 1.2526521682739258,
      "learning_rate": 2.996487906340466e-05,
      "loss": 0.06138528883457184,
      "step": 663
    },
    {
      "epoch": 0.089768396860134,
      "grad_norm": 1.6418873071670532,
      "learning_rate": 2.996440940920592e-05,
      "loss": 0.0744391679763794,
      "step": 664
    },
    {
      "epoch": 0.08990359022889927,
      "grad_norm": 0.9773853421211243,
      "learning_rate": 2.996393663933735e-05,
      "loss": 0.07288479804992676,
      "step": 665
    },
    {
      "epoch": 0.09003878359766454,
      "grad_norm": 1.7122795581817627,
      "learning_rate": 2.9963460753897364e-05,
      "loss": 0.06747186183929443,
      "step": 666
    },
    {
      "epoch": 0.0901739769664298,
      "grad_norm": 1.5426825284957886,
      "learning_rate": 2.996298175298506e-05,
      "loss": 0.08441933989524841,
      "step": 667
    },
    {
      "epoch": 0.09030917033519506,
      "grad_norm": 0.7839665412902832,
      "learning_rate": 2.996249963670016e-05,
      "loss": 0.07313214242458344,
      "step": 668
    },
    {
      "epoch": 0.09044436370396032,
      "grad_norm": 0.9040767550468445,
      "learning_rate": 2.9962014405143042e-05,
      "loss": 0.08949136734008789,
      "step": 669
    },
    {
      "epoch": 0.09057955707272558,
      "grad_norm": 0.5139188766479492,
      "learning_rate": 2.9961526058414745e-05,
      "loss": 0.06375241279602051,
      "step": 670
    },
    {
      "epoch": 0.09071475044149084,
      "grad_norm": 2.4885027408599854,
      "learning_rate": 2.9961034596616936e-05,
      "loss": 0.06757906079292297,
      "step": 671
    },
    {
      "epoch": 0.0908499438102561,
      "grad_norm": 2.298752546310425,
      "learning_rate": 2.996054001985194e-05,
      "loss": 0.061887532472610474,
      "step": 672
    },
    {
      "epoch": 0.09098513717902137,
      "grad_norm": 1.8645328283309937,
      "learning_rate": 2.9960042328222732e-05,
      "loss": 0.0736398696899414,
      "step": 673
    },
    {
      "epoch": 0.09112033054778663,
      "grad_norm": 0.5860607624053955,
      "learning_rate": 2.995954152183294e-05,
      "loss": 0.06288467347621918,
      "step": 674
    },
    {
      "epoch": 0.0912555239165519,
      "grad_norm": 1.5632336139678955,
      "learning_rate": 2.9959037600786822e-05,
      "loss": 0.07192747294902802,
      "step": 675
    },
    {
      "epoch": 0.09139071728531716,
      "grad_norm": 2.513230323791504,
      "learning_rate": 2.9958530565189307e-05,
      "loss": 0.0633854866027832,
      "step": 676
    },
    {
      "epoch": 0.09152591065408242,
      "grad_norm": 1.0967401266098022,
      "learning_rate": 2.995802041514596e-05,
      "loss": 0.08458459377288818,
      "step": 677
    },
    {
      "epoch": 0.09166110402284768,
      "grad_norm": 0.4788946807384491,
      "learning_rate": 2.9957507150762996e-05,
      "loss": 0.061201997101306915,
      "step": 678
    },
    {
      "epoch": 0.09179629739161294,
      "grad_norm": 0.6017131805419922,
      "learning_rate": 2.9956990772147283e-05,
      "loss": 0.053073760122060776,
      "step": 679
    },
    {
      "epoch": 0.09193149076037821,
      "grad_norm": 1.4322870969772339,
      "learning_rate": 2.9956471279406324e-05,
      "loss": 0.049558788537979126,
      "step": 680
    },
    {
      "epoch": 0.09206668412914347,
      "grad_norm": 0.6398486495018005,
      "learning_rate": 2.9955948672648298e-05,
      "loss": 0.062316685914993286,
      "step": 681
    },
    {
      "epoch": 0.09220187749790873,
      "grad_norm": 0.848783016204834,
      "learning_rate": 2.9955422951981994e-05,
      "loss": 0.0711323618888855,
      "step": 682
    },
    {
      "epoch": 0.09233707086667399,
      "grad_norm": 1.5682917833328247,
      "learning_rate": 2.995489411751688e-05,
      "loss": 0.08610308170318604,
      "step": 683
    },
    {
      "epoch": 0.09247226423543925,
      "grad_norm": 0.8479408621788025,
      "learning_rate": 2.9954362169363064e-05,
      "loss": 0.06241118907928467,
      "step": 684
    },
    {
      "epoch": 0.09260745760420451,
      "grad_norm": 0.9955445528030396,
      "learning_rate": 2.99538271076313e-05,
      "loss": 0.05744418501853943,
      "step": 685
    },
    {
      "epoch": 0.09274265097296977,
      "grad_norm": 0.5702171921730042,
      "learning_rate": 2.9953288932432985e-05,
      "loss": 0.051501452922821045,
      "step": 686
    },
    {
      "epoch": 0.09287784434173504,
      "grad_norm": 1.3401936292648315,
      "learning_rate": 2.995274764388018e-05,
      "loss": 0.068433478474617,
      "step": 687
    },
    {
      "epoch": 0.0930130377105003,
      "grad_norm": 2.055107593536377,
      "learning_rate": 2.9952203242085566e-05,
      "loss": 0.1082228571176529,
      "step": 688
    },
    {
      "epoch": 0.09314823107926556,
      "grad_norm": 1.7486412525177002,
      "learning_rate": 2.995165572716251e-05,
      "loss": 0.07478171586990356,
      "step": 689
    },
    {
      "epoch": 0.09328342444803082,
      "grad_norm": 0.4199610948562622,
      "learning_rate": 2.9951105099225003e-05,
      "loss": 0.06392104923725128,
      "step": 690
    },
    {
      "epoch": 0.09341861781679608,
      "grad_norm": 2.0826423168182373,
      "learning_rate": 2.995055135838768e-05,
      "loss": 0.08926263451576233,
      "step": 691
    },
    {
      "epoch": 0.09355381118556134,
      "grad_norm": 0.5368213057518005,
      "learning_rate": 2.994999450476584e-05,
      "loss": 0.06264454126358032,
      "step": 692
    },
    {
      "epoch": 0.0936890045543266,
      "grad_norm": 0.5854647159576416,
      "learning_rate": 2.9949434538475414e-05,
      "loss": 0.07068228721618652,
      "step": 693
    },
    {
      "epoch": 0.09382419792309188,
      "grad_norm": 1.64569890499115,
      "learning_rate": 2.9948871459633008e-05,
      "loss": 0.08626413345336914,
      "step": 694
    },
    {
      "epoch": 0.09395939129185714,
      "grad_norm": 1.3701481819152832,
      "learning_rate": 2.994830526835584e-05,
      "loss": 0.06382119655609131,
      "step": 695
    },
    {
      "epoch": 0.0940945846606224,
      "grad_norm": 1.5206036567687988,
      "learning_rate": 2.9947735964761803e-05,
      "loss": 0.08059477806091309,
      "step": 696
    },
    {
      "epoch": 0.09422977802938766,
      "grad_norm": 1.9635858535766602,
      "learning_rate": 2.9947163548969428e-05,
      "loss": 0.08320224285125732,
      "step": 697
    },
    {
      "epoch": 0.09436497139815292,
      "grad_norm": 3.18457293510437,
      "learning_rate": 2.9946588021097893e-05,
      "loss": 0.07568666338920593,
      "step": 698
    },
    {
      "epoch": 0.09450016476691818,
      "grad_norm": 0.7685016989707947,
      "learning_rate": 2.9946009381267028e-05,
      "loss": 0.06975710391998291,
      "step": 699
    },
    {
      "epoch": 0.09463535813568344,
      "grad_norm": 1.169830083847046,
      "learning_rate": 2.9945427629597306e-05,
      "loss": 0.05975687503814697,
      "step": 700
    },
    {
      "epoch": 0.09477055150444871,
      "grad_norm": 2.346580743789673,
      "learning_rate": 2.9944842766209853e-05,
      "loss": 0.06218397617340088,
      "step": 701
    },
    {
      "epoch": 0.09490574487321397,
      "grad_norm": 2.587913751602173,
      "learning_rate": 2.9944254791226444e-05,
      "loss": 0.0864153504371643,
      "step": 702
    },
    {
      "epoch": 0.09504093824197923,
      "grad_norm": 0.5739260315895081,
      "learning_rate": 2.994366370476949e-05,
      "loss": 0.06660217046737671,
      "step": 703
    },
    {
      "epoch": 0.0951761316107445,
      "grad_norm": 2.2969987392425537,
      "learning_rate": 2.9943069506962067e-05,
      "loss": 0.07569676637649536,
      "step": 704
    },
    {
      "epoch": 0.09531132497950975,
      "grad_norm": 0.560544490814209,
      "learning_rate": 2.9942472197927886e-05,
      "loss": 0.055384933948516846,
      "step": 705
    },
    {
      "epoch": 0.09544651834827501,
      "grad_norm": 1.0866575241088867,
      "learning_rate": 2.994187177779131e-05,
      "loss": 0.061683833599090576,
      "step": 706
    },
    {
      "epoch": 0.09558171171704027,
      "grad_norm": 0.6969910264015198,
      "learning_rate": 2.9941268246677353e-05,
      "loss": 0.06661128997802734,
      "step": 707
    },
    {
      "epoch": 0.09571690508580555,
      "grad_norm": 2.0652143955230713,
      "learning_rate": 2.9940661604711664e-05,
      "loss": 0.07562792301177979,
      "step": 708
    },
    {
      "epoch": 0.09585209845457081,
      "grad_norm": 2.230104923248291,
      "learning_rate": 2.994005185202056e-05,
      "loss": 0.06700295209884644,
      "step": 709
    },
    {
      "epoch": 0.09598729182333607,
      "grad_norm": 1.7517598867416382,
      "learning_rate": 2.9939438988730986e-05,
      "loss": 0.05676276981830597,
      "step": 710
    },
    {
      "epoch": 0.09612248519210133,
      "grad_norm": 1.8654752969741821,
      "learning_rate": 2.9938823014970553e-05,
      "loss": 0.057524263858795166,
      "step": 711
    },
    {
      "epoch": 0.09625767856086659,
      "grad_norm": 2.1614320278167725,
      "learning_rate": 2.99382039308675e-05,
      "loss": 0.06328561902046204,
      "step": 712
    },
    {
      "epoch": 0.09639287192963185,
      "grad_norm": 1.671129822731018,
      "learning_rate": 2.993758173655073e-05,
      "loss": 0.07212221622467041,
      "step": 713
    },
    {
      "epoch": 0.09652806529839711,
      "grad_norm": 0.696434497833252,
      "learning_rate": 2.993695643214979e-05,
      "loss": 0.07994964718818665,
      "step": 714
    },
    {
      "epoch": 0.09666325866716237,
      "grad_norm": 1.1885391473770142,
      "learning_rate": 2.9936328017794864e-05,
      "loss": 0.07391977310180664,
      "step": 715
    },
    {
      "epoch": 0.09679845203592764,
      "grad_norm": 0.6399041414260864,
      "learning_rate": 2.9935696493616796e-05,
      "loss": 0.060434311628341675,
      "step": 716
    },
    {
      "epoch": 0.0969336454046929,
      "grad_norm": 2.2488210201263428,
      "learning_rate": 2.9935061859747065e-05,
      "loss": 0.0648459643125534,
      "step": 717
    },
    {
      "epoch": 0.09706883877345816,
      "grad_norm": 2.4420716762542725,
      "learning_rate": 2.993442411631782e-05,
      "loss": 0.09676936268806458,
      "step": 718
    },
    {
      "epoch": 0.09720403214222342,
      "grad_norm": 0.48561614751815796,
      "learning_rate": 2.9933783263461827e-05,
      "loss": 0.06036248430609703,
      "step": 719
    },
    {
      "epoch": 0.09733922551098868,
      "grad_norm": 0.5393803119659424,
      "learning_rate": 2.9933139301312526e-05,
      "loss": 0.04506488889455795,
      "step": 720
    },
    {
      "epoch": 0.09747441887975394,
      "grad_norm": 1.6485989093780518,
      "learning_rate": 2.9932492230003984e-05,
      "loss": 0.068889319896698,
      "step": 721
    },
    {
      "epoch": 0.0976096122485192,
      "grad_norm": 0.4607279300689697,
      "learning_rate": 2.993184204967094e-05,
      "loss": 0.04657787084579468,
      "step": 722
    },
    {
      "epoch": 0.09774480561728448,
      "grad_norm": 2.5502140522003174,
      "learning_rate": 2.9931188760448748e-05,
      "loss": 0.09501942247152328,
      "step": 723
    },
    {
      "epoch": 0.09787999898604974,
      "grad_norm": 1.5143232345581055,
      "learning_rate": 2.9930532362473433e-05,
      "loss": 0.07087540626525879,
      "step": 724
    },
    {
      "epoch": 0.098015192354815,
      "grad_norm": 0.9561923146247864,
      "learning_rate": 2.9929872855881663e-05,
      "loss": 0.07919055223464966,
      "step": 725
    },
    {
      "epoch": 0.09815038572358026,
      "grad_norm": 2.1918020248413086,
      "learning_rate": 2.9929210240810744e-05,
      "loss": 0.06227082014083862,
      "step": 726
    },
    {
      "epoch": 0.09828557909234552,
      "grad_norm": 0.3666427731513977,
      "learning_rate": 2.9928544517398644e-05,
      "loss": 0.053845345973968506,
      "step": 727
    },
    {
      "epoch": 0.09842077246111078,
      "grad_norm": 1.5570653676986694,
      "learning_rate": 2.9927875685783966e-05,
      "loss": 0.07275102287530899,
      "step": 728
    },
    {
      "epoch": 0.09855596582987604,
      "grad_norm": 1.464547038078308,
      "learning_rate": 2.9927203746105968e-05,
      "loss": 0.06417824327945709,
      "step": 729
    },
    {
      "epoch": 0.09869115919864131,
      "grad_norm": 0.7170355319976807,
      "learning_rate": 2.9926528698504546e-05,
      "loss": 0.08210603147745132,
      "step": 730
    },
    {
      "epoch": 0.09882635256740657,
      "grad_norm": 0.7994945049285889,
      "learning_rate": 2.992585054312025e-05,
      "loss": 0.06015549600124359,
      "step": 731
    },
    {
      "epoch": 0.09896154593617183,
      "grad_norm": 1.777430772781372,
      "learning_rate": 2.9925169280094278e-05,
      "loss": 0.08092288672924042,
      "step": 732
    },
    {
      "epoch": 0.09909673930493709,
      "grad_norm": 2.4413414001464844,
      "learning_rate": 2.9924484909568472e-05,
      "loss": 0.0873079001903534,
      "step": 733
    },
    {
      "epoch": 0.09923193267370235,
      "grad_norm": 0.8687604665756226,
      "learning_rate": 2.9923797431685322e-05,
      "loss": 0.0524190291762352,
      "step": 734
    },
    {
      "epoch": 0.09936712604246761,
      "grad_norm": 0.5589920878410339,
      "learning_rate": 2.992310684658796e-05,
      "loss": 0.053833842277526855,
      "step": 735
    },
    {
      "epoch": 0.09950231941123287,
      "grad_norm": 1.160480260848999,
      "learning_rate": 2.9922413154420173e-05,
      "loss": 0.07847365736961365,
      "step": 736
    },
    {
      "epoch": 0.09963751277999815,
      "grad_norm": 0.40571966767311096,
      "learning_rate": 2.9921716355326393e-05,
      "loss": 0.0431639701128006,
      "step": 737
    },
    {
      "epoch": 0.09977270614876341,
      "grad_norm": 0.6975798606872559,
      "learning_rate": 2.9921016449451695e-05,
      "loss": 0.06810298562049866,
      "step": 738
    },
    {
      "epoch": 0.09990789951752867,
      "grad_norm": 0.7199623584747314,
      "learning_rate": 2.9920313436941805e-05,
      "loss": 0.05668321251869202,
      "step": 739
    },
    {
      "epoch": 0.10004309288629393,
      "grad_norm": 1.9567862749099731,
      "learning_rate": 2.991960731794309e-05,
      "loss": 0.07285964488983154,
      "step": 740
    },
    {
      "epoch": 0.10017828625505919,
      "grad_norm": 0.6199095845222473,
      "learning_rate": 2.991889809260257e-05,
      "loss": 0.08010756969451904,
      "step": 741
    },
    {
      "epoch": 0.10031347962382445,
      "grad_norm": 2.2614054679870605,
      "learning_rate": 2.9918185761067912e-05,
      "loss": 0.06480199098587036,
      "step": 742
    },
    {
      "epoch": 0.10044867299258971,
      "grad_norm": 2.2115559577941895,
      "learning_rate": 2.9917470323487423e-05,
      "loss": 0.0716850757598877,
      "step": 743
    },
    {
      "epoch": 0.10058386636135498,
      "grad_norm": 1.1710504293441772,
      "learning_rate": 2.9916751780010063e-05,
      "loss": 0.06521201133728027,
      "step": 744
    },
    {
      "epoch": 0.10071905973012024,
      "grad_norm": 1.2195080518722534,
      "learning_rate": 2.9916030130785436e-05,
      "loss": 0.08190056681632996,
      "step": 745
    },
    {
      "epoch": 0.1008542530988855,
      "grad_norm": 2.0479962825775146,
      "learning_rate": 2.99153053759638e-05,
      "loss": 0.09446015954017639,
      "step": 746
    },
    {
      "epoch": 0.10098944646765076,
      "grad_norm": 1.6798561811447144,
      "learning_rate": 2.991457751569604e-05,
      "loss": 0.06819860637187958,
      "step": 747
    },
    {
      "epoch": 0.10112463983641602,
      "grad_norm": 2.639376401901245,
      "learning_rate": 2.991384655013371e-05,
      "loss": 0.0689087063074112,
      "step": 748
    },
    {
      "epoch": 0.10125983320518128,
      "grad_norm": 0.7151151895523071,
      "learning_rate": 2.9913112479429e-05,
      "loss": 0.05495786666870117,
      "step": 749
    },
    {
      "epoch": 0.10139502657394654,
      "grad_norm": 0.5547966361045837,
      "learning_rate": 2.991237530373474e-05,
      "loss": 0.05543568730354309,
      "step": 750
    },
    {
      "epoch": 0.10153021994271182,
      "grad_norm": 0.76650470495224,
      "learning_rate": 2.9911635023204423e-05,
      "loss": 0.0550074577331543,
      "step": 751
    },
    {
      "epoch": 0.10166541331147708,
      "grad_norm": 1.9718046188354492,
      "learning_rate": 2.9910891637992172e-05,
      "loss": 0.07963669300079346,
      "step": 752
    },
    {
      "epoch": 0.10180060668024234,
      "grad_norm": 1.059044361114502,
      "learning_rate": 2.991014514825277e-05,
      "loss": 0.06547009944915771,
      "step": 753
    },
    {
      "epoch": 0.1019358000490076,
      "grad_norm": 0.5173660516738892,
      "learning_rate": 2.9909395554141638e-05,
      "loss": 0.06538379192352295,
      "step": 754
    },
    {
      "epoch": 0.10207099341777286,
      "grad_norm": 0.7970746159553528,
      "learning_rate": 2.9908642855814844e-05,
      "loss": 0.0796213150024414,
      "step": 755
    },
    {
      "epoch": 0.10220618678653812,
      "grad_norm": 0.509709358215332,
      "learning_rate": 2.9907887053429107e-05,
      "loss": 0.07489272952079773,
      "step": 756
    },
    {
      "epoch": 0.10234138015530338,
      "grad_norm": 0.43984255194664,
      "learning_rate": 2.9907128147141783e-05,
      "loss": 0.052528321743011475,
      "step": 757
    },
    {
      "epoch": 0.10247657352406865,
      "grad_norm": 2.5199241638183594,
      "learning_rate": 2.990636613711089e-05,
      "loss": 0.07140716910362244,
      "step": 758
    },
    {
      "epoch": 0.10261176689283391,
      "grad_norm": 0.7608581185340881,
      "learning_rate": 2.990560102349507e-05,
      "loss": 0.08876919746398926,
      "step": 759
    },
    {
      "epoch": 0.10274696026159917,
      "grad_norm": 0.3698713779449463,
      "learning_rate": 2.9904832806453635e-05,
      "loss": 0.0644555538892746,
      "step": 760
    },
    {
      "epoch": 0.10288215363036443,
      "grad_norm": 0.6989520788192749,
      "learning_rate": 2.9904061486146524e-05,
      "loss": 0.04861116409301758,
      "step": 761
    },
    {
      "epoch": 0.10301734699912969,
      "grad_norm": 0.44074511528015137,
      "learning_rate": 2.9903287062734333e-05,
      "loss": 0.054500073194503784,
      "step": 762
    },
    {
      "epoch": 0.10315254036789495,
      "grad_norm": 0.6014143824577332,
      "learning_rate": 2.990250953637831e-05,
      "loss": 0.07462161779403687,
      "step": 763
    },
    {
      "epoch": 0.10328773373666021,
      "grad_norm": 1.3084285259246826,
      "learning_rate": 2.9901728907240326e-05,
      "loss": 0.05892127752304077,
      "step": 764
    },
    {
      "epoch": 0.10342292710542549,
      "grad_norm": 1.170800805091858,
      "learning_rate": 2.9900945175482916e-05,
      "loss": 0.07677042484283447,
      "step": 765
    },
    {
      "epoch": 0.10355812047419075,
      "grad_norm": 0.8300243616104126,
      "learning_rate": 2.990015834126926e-05,
      "loss": 0.053655147552490234,
      "step": 766
    },
    {
      "epoch": 0.103693313842956,
      "grad_norm": 1.9866926670074463,
      "learning_rate": 2.989936840476318e-05,
      "loss": 0.08869680762290955,
      "step": 767
    },
    {
      "epoch": 0.10382850721172127,
      "grad_norm": 1.2555097341537476,
      "learning_rate": 2.9898575366129145e-05,
      "loss": 0.08476987481117249,
      "step": 768
    },
    {
      "epoch": 0.10396370058048653,
      "grad_norm": 1.3849974870681763,
      "learning_rate": 2.9897779225532273e-05,
      "loss": 0.07022476196289062,
      "step": 769
    },
    {
      "epoch": 0.10409889394925179,
      "grad_norm": 1.173958659172058,
      "learning_rate": 2.989697998313832e-05,
      "loss": 0.0702330470085144,
      "step": 770
    },
    {
      "epoch": 0.10423408731801705,
      "grad_norm": 1.936564326286316,
      "learning_rate": 2.989617763911369e-05,
      "loss": 0.07958516478538513,
      "step": 771
    },
    {
      "epoch": 0.1043692806867823,
      "grad_norm": 1.0194870233535767,
      "learning_rate": 2.9895372193625442e-05,
      "loss": 0.045984745025634766,
      "step": 772
    },
    {
      "epoch": 0.10450447405554758,
      "grad_norm": 1.7459651231765747,
      "learning_rate": 2.9894563646841273e-05,
      "loss": 0.09469527006149292,
      "step": 773
    },
    {
      "epoch": 0.10463966742431284,
      "grad_norm": 0.6957114338874817,
      "learning_rate": 2.9893751998929523e-05,
      "loss": 0.08088427782058716,
      "step": 774
    },
    {
      "epoch": 0.1047748607930781,
      "grad_norm": 0.6176539063453674,
      "learning_rate": 2.9892937250059187e-05,
      "loss": 0.06235021352767944,
      "step": 775
    },
    {
      "epoch": 0.10491005416184336,
      "grad_norm": 1.5887959003448486,
      "learning_rate": 2.9892119400399894e-05,
      "loss": 0.06383442878723145,
      "step": 776
    },
    {
      "epoch": 0.10504524753060862,
      "grad_norm": 1.5024386644363403,
      "learning_rate": 2.989129845012193e-05,
      "loss": 0.06038999557495117,
      "step": 777
    },
    {
      "epoch": 0.10518044089937388,
      "grad_norm": 2.5025908946990967,
      "learning_rate": 2.989047439939621e-05,
      "loss": 0.08759409189224243,
      "step": 778
    },
    {
      "epoch": 0.10531563426813914,
      "grad_norm": 0.47842252254486084,
      "learning_rate": 2.9889647248394324e-05,
      "loss": 0.07418322563171387,
      "step": 779
    },
    {
      "epoch": 0.10545082763690442,
      "grad_norm": 1.1762672662734985,
      "learning_rate": 2.9888816997288475e-05,
      "loss": 0.06556662172079086,
      "step": 780
    },
    {
      "epoch": 0.10558602100566968,
      "grad_norm": 1.706656813621521,
      "learning_rate": 2.988798364625153e-05,
      "loss": 0.07296490669250488,
      "step": 781
    },
    {
      "epoch": 0.10572121437443494,
      "grad_norm": 0.6584717035293579,
      "learning_rate": 2.9887147195457002e-05,
      "loss": 0.059939682483673096,
      "step": 782
    },
    {
      "epoch": 0.1058564077432002,
      "grad_norm": 1.2951629161834717,
      "learning_rate": 2.9886307645079037e-05,
      "loss": 0.07447600364685059,
      "step": 783
    },
    {
      "epoch": 0.10599160111196546,
      "grad_norm": 0.5977193713188171,
      "learning_rate": 2.9885464995292436e-05,
      "loss": 0.054437726736068726,
      "step": 784
    },
    {
      "epoch": 0.10612679448073072,
      "grad_norm": 1.9214591979980469,
      "learning_rate": 2.9884619246272648e-05,
      "loss": 0.07319247722625732,
      "step": 785
    },
    {
      "epoch": 0.10626198784949598,
      "grad_norm": 0.9901397824287415,
      "learning_rate": 2.988377039819575e-05,
      "loss": 0.08041355013847351,
      "step": 786
    },
    {
      "epoch": 0.10639718121826125,
      "grad_norm": 1.2070285081863403,
      "learning_rate": 2.9882918451238494e-05,
      "loss": 0.06483381986618042,
      "step": 787
    },
    {
      "epoch": 0.10653237458702651,
      "grad_norm": 1.4205820560455322,
      "learning_rate": 2.988206340557825e-05,
      "loss": 0.06248599290847778,
      "step": 788
    },
    {
      "epoch": 0.10666756795579177,
      "grad_norm": 1.5785423517227173,
      "learning_rate": 2.9881205261393037e-05,
      "loss": 0.06281136721372604,
      "step": 789
    },
    {
      "epoch": 0.10680276132455703,
      "grad_norm": 0.6278877258300781,
      "learning_rate": 2.988034401886154e-05,
      "loss": 0.07736283540725708,
      "step": 790
    },
    {
      "epoch": 0.10693795469332229,
      "grad_norm": 3.044987201690674,
      "learning_rate": 2.9879479678163065e-05,
      "loss": 0.07018107175827026,
      "step": 791
    },
    {
      "epoch": 0.10707314806208755,
      "grad_norm": 0.5835976004600525,
      "learning_rate": 2.9878612239477568e-05,
      "loss": 0.06766742467880249,
      "step": 792
    },
    {
      "epoch": 0.10720834143085281,
      "grad_norm": 0.8784189224243164,
      "learning_rate": 2.9877741702985666e-05,
      "loss": 0.08805882930755615,
      "step": 793
    },
    {
      "epoch": 0.10734353479961808,
      "grad_norm": 0.8258385062217712,
      "learning_rate": 2.98768680688686e-05,
      "loss": 0.05716429650783539,
      "step": 794
    },
    {
      "epoch": 0.10747872816838334,
      "grad_norm": 0.5803067088127136,
      "learning_rate": 2.9875991337308274e-05,
      "loss": 0.0653550773859024,
      "step": 795
    },
    {
      "epoch": 0.1076139215371486,
      "grad_norm": 0.9804613590240479,
      "learning_rate": 2.987511150848722e-05,
      "loss": 0.07762131094932556,
      "step": 796
    },
    {
      "epoch": 0.10774911490591386,
      "grad_norm": 1.0947750806808472,
      "learning_rate": 2.9874228582588627e-05,
      "loss": 0.07066251337528229,
      "step": 797
    },
    {
      "epoch": 0.10788430827467912,
      "grad_norm": 0.5301551818847656,
      "learning_rate": 2.9873342559796325e-05,
      "loss": 0.06156620383262634,
      "step": 798
    },
    {
      "epoch": 0.10801950164344438,
      "grad_norm": 0.9586540460586548,
      "learning_rate": 2.9872453440294787e-05,
      "loss": 0.07479715347290039,
      "step": 799
    },
    {
      "epoch": 0.10815469501220965,
      "grad_norm": 0.683612585067749,
      "learning_rate": 2.9871561224269134e-05,
      "loss": 0.05333220958709717,
      "step": 800
    },
    {
      "epoch": 0.10828988838097492,
      "grad_norm": 0.7130185961723328,
      "learning_rate": 2.9870665911905127e-05,
      "loss": 0.05451911687850952,
      "step": 801
    },
    {
      "epoch": 0.10842508174974018,
      "grad_norm": 0.5970738530158997,
      "learning_rate": 2.9869767503389176e-05,
      "loss": 0.0738338828086853,
      "step": 802
    },
    {
      "epoch": 0.10856027511850544,
      "grad_norm": 2.720468282699585,
      "learning_rate": 2.986886599890834e-05,
      "loss": 0.08367794752120972,
      "step": 803
    },
    {
      "epoch": 0.1086954684872707,
      "grad_norm": 0.5943778157234192,
      "learning_rate": 2.9867961398650306e-05,
      "loss": 0.06560193002223969,
      "step": 804
    },
    {
      "epoch": 0.10883066185603596,
      "grad_norm": 0.3316863775253296,
      "learning_rate": 2.9867053702803425e-05,
      "loss": 0.06915531307458878,
      "step": 805
    },
    {
      "epoch": 0.10896585522480122,
      "grad_norm": 1.3966037034988403,
      "learning_rate": 2.9866142911556685e-05,
      "loss": 0.055244386196136475,
      "step": 806
    },
    {
      "epoch": 0.10910104859356648,
      "grad_norm": 0.6676267981529236,
      "learning_rate": 2.9865229025099713e-05,
      "loss": 0.05885940045118332,
      "step": 807
    },
    {
      "epoch": 0.10923624196233175,
      "grad_norm": 1.6562120914459229,
      "learning_rate": 2.986431204362279e-05,
      "loss": 0.07612279802560806,
      "step": 808
    },
    {
      "epoch": 0.10937143533109701,
      "grad_norm": 1.0587570667266846,
      "learning_rate": 2.9863391967316835e-05,
      "loss": 0.07631005346775055,
      "step": 809
    },
    {
      "epoch": 0.10950662869986227,
      "grad_norm": 1.4671709537506104,
      "learning_rate": 2.9862468796373404e-05,
      "loss": 0.06622779369354248,
      "step": 810
    },
    {
      "epoch": 0.10964182206862753,
      "grad_norm": 1.1340380907058716,
      "learning_rate": 2.9861542530984718e-05,
      "loss": 0.07871812582015991,
      "step": 811
    },
    {
      "epoch": 0.1097770154373928,
      "grad_norm": 3.6486597061157227,
      "learning_rate": 2.9860613171343624e-05,
      "loss": 0.0973467230796814,
      "step": 812
    },
    {
      "epoch": 0.10991220880615805,
      "grad_norm": 3.2831902503967285,
      "learning_rate": 2.9859680717643623e-05,
      "loss": 0.07520408928394318,
      "step": 813
    },
    {
      "epoch": 0.11004740217492331,
      "grad_norm": 1.5942511558532715,
      "learning_rate": 2.985874517007885e-05,
      "loss": 0.05213823914527893,
      "step": 814
    },
    {
      "epoch": 0.11018259554368859,
      "grad_norm": 2.048729181289673,
      "learning_rate": 2.98578065288441e-05,
      "loss": 0.052881523966789246,
      "step": 815
    },
    {
      "epoch": 0.11031778891245385,
      "grad_norm": 0.43020763993263245,
      "learning_rate": 2.9856864794134798e-05,
      "loss": 0.04438048601150513,
      "step": 816
    },
    {
      "epoch": 0.11045298228121911,
      "grad_norm": 2.1527047157287598,
      "learning_rate": 2.9855919966147025e-05,
      "loss": 0.06952786445617676,
      "step": 817
    },
    {
      "epoch": 0.11058817564998437,
      "grad_norm": 1.5365175008773804,
      "learning_rate": 2.9854972045077485e-05,
      "loss": 0.05727046728134155,
      "step": 818
    },
    {
      "epoch": 0.11072336901874963,
      "grad_norm": 0.9647189974784851,
      "learning_rate": 2.9854021031123555e-05,
      "loss": 0.05073575675487518,
      "step": 819
    },
    {
      "epoch": 0.11085856238751489,
      "grad_norm": 1.5040180683135986,
      "learning_rate": 2.9853066924483232e-05,
      "loss": 0.0715726763010025,
      "step": 820
    },
    {
      "epoch": 0.11099375575628015,
      "grad_norm": 1.0900242328643799,
      "learning_rate": 2.9852109725355173e-05,
      "loss": 0.06380832195281982,
      "step": 821
    },
    {
      "epoch": 0.11112894912504542,
      "grad_norm": 1.058659315109253,
      "learning_rate": 2.9851149433938662e-05,
      "loss": 0.07913881540298462,
      "step": 822
    },
    {
      "epoch": 0.11126414249381068,
      "grad_norm": 1.3789092302322388,
      "learning_rate": 2.9850186050433645e-05,
      "loss": 0.05524668097496033,
      "step": 823
    },
    {
      "epoch": 0.11139933586257594,
      "grad_norm": 4.019479751586914,
      "learning_rate": 2.9849219575040708e-05,
      "loss": 0.07588204741477966,
      "step": 824
    },
    {
      "epoch": 0.1115345292313412,
      "grad_norm": 2.0158936977386475,
      "learning_rate": 2.984825000796106e-05,
      "loss": 0.07111486792564392,
      "step": 825
    },
    {
      "epoch": 0.11166972260010646,
      "grad_norm": 1.8207803964614868,
      "learning_rate": 2.9847277349396586e-05,
      "loss": 0.07971811294555664,
      "step": 826
    },
    {
      "epoch": 0.11180491596887172,
      "grad_norm": 0.3318125307559967,
      "learning_rate": 2.984630159954979e-05,
      "loss": 0.06541299819946289,
      "step": 827
    },
    {
      "epoch": 0.11194010933763698,
      "grad_norm": 0.5564690232276917,
      "learning_rate": 2.9845322758623833e-05,
      "loss": 0.06687232851982117,
      "step": 828
    },
    {
      "epoch": 0.11207530270640226,
      "grad_norm": 4.113993167877197,
      "learning_rate": 2.984434082682251e-05,
      "loss": 0.09146535396575928,
      "step": 829
    },
    {
      "epoch": 0.11221049607516752,
      "grad_norm": 1.7399711608886719,
      "learning_rate": 2.984335580435027e-05,
      "loss": 0.06328044831752777,
      "step": 830
    },
    {
      "epoch": 0.11234568944393278,
      "grad_norm": 0.9513656497001648,
      "learning_rate": 2.9842367691412192e-05,
      "loss": 0.07036569714546204,
      "step": 831
    },
    {
      "epoch": 0.11248088281269804,
      "grad_norm": 0.7836118340492249,
      "learning_rate": 2.9841376488214015e-05,
      "loss": 0.06594693660736084,
      "step": 832
    },
    {
      "epoch": 0.1126160761814633,
      "grad_norm": 1.620522141456604,
      "learning_rate": 2.984038219496211e-05,
      "loss": 0.049950480461120605,
      "step": 833
    },
    {
      "epoch": 0.11275126955022856,
      "grad_norm": 0.8438585996627808,
      "learning_rate": 2.9839384811863493e-05,
      "loss": 0.05754745006561279,
      "step": 834
    },
    {
      "epoch": 0.11288646291899382,
      "grad_norm": 1.3088704347610474,
      "learning_rate": 2.9838384339125824e-05,
      "loss": 0.04670900106430054,
      "step": 835
    },
    {
      "epoch": 0.11302165628775908,
      "grad_norm": 2.862694025039673,
      "learning_rate": 2.9837380776957405e-05,
      "loss": 0.07493305206298828,
      "step": 836
    },
    {
      "epoch": 0.11315684965652435,
      "grad_norm": 0.6323216557502747,
      "learning_rate": 2.9836374125567193e-05,
      "loss": 0.0698775202035904,
      "step": 837
    },
    {
      "epoch": 0.11329204302528961,
      "grad_norm": 0.9724240899085999,
      "learning_rate": 2.9835364385164764e-05,
      "loss": 0.07646346092224121,
      "step": 838
    },
    {
      "epoch": 0.11342723639405487,
      "grad_norm": 0.9290037155151367,
      "learning_rate": 2.983435155596036e-05,
      "loss": 0.06411190330982208,
      "step": 839
    },
    {
      "epoch": 0.11356242976282013,
      "grad_norm": 1.4180476665496826,
      "learning_rate": 2.9833335638164858e-05,
      "loss": 0.06833842396736145,
      "step": 840
    },
    {
      "epoch": 0.11369762313158539,
      "grad_norm": 1.3837205171585083,
      "learning_rate": 2.9832316631989774e-05,
      "loss": 0.06893129646778107,
      "step": 841
    },
    {
      "epoch": 0.11383281650035065,
      "grad_norm": 0.9323716163635254,
      "learning_rate": 2.9831294537647272e-05,
      "loss": 0.07906448096036911,
      "step": 842
    },
    {
      "epoch": 0.11396800986911591,
      "grad_norm": 0.43834733963012695,
      "learning_rate": 2.9830269355350155e-05,
      "loss": 0.06279188394546509,
      "step": 843
    },
    {
      "epoch": 0.11410320323788119,
      "grad_norm": 1.4275168180465698,
      "learning_rate": 2.9829241085311872e-05,
      "loss": 0.04370114952325821,
      "step": 844
    },
    {
      "epoch": 0.11423839660664645,
      "grad_norm": 0.7436087727546692,
      "learning_rate": 2.9828209727746522e-05,
      "loss": 0.07141464948654175,
      "step": 845
    },
    {
      "epoch": 0.11437358997541171,
      "grad_norm": 1.1526052951812744,
      "learning_rate": 2.982717528286883e-05,
      "loss": 0.06148737668991089,
      "step": 846
    },
    {
      "epoch": 0.11450878334417697,
      "grad_norm": 0.6929847002029419,
      "learning_rate": 2.9826137750894176e-05,
      "loss": 0.047165051102638245,
      "step": 847
    },
    {
      "epoch": 0.11464397671294223,
      "grad_norm": 1.6904442310333252,
      "learning_rate": 2.9825097132038578e-05,
      "loss": 0.07676011323928833,
      "step": 848
    },
    {
      "epoch": 0.11477917008170749,
      "grad_norm": 1.3478825092315674,
      "learning_rate": 2.9824053426518703e-05,
      "loss": 0.07952776551246643,
      "step": 849
    },
    {
      "epoch": 0.11491436345047275,
      "grad_norm": 0.4725055992603302,
      "learning_rate": 2.9823006634551848e-05,
      "loss": 0.06164819002151489,
      "step": 850
    },
    {
      "epoch": 0.11504955681923802,
      "grad_norm": 0.39076611399650574,
      "learning_rate": 2.9821956756355973e-05,
      "loss": 0.05239543318748474,
      "step": 851
    },
    {
      "epoch": 0.11518475018800328,
      "grad_norm": 0.5654911994934082,
      "learning_rate": 2.9820903792149653e-05,
      "loss": 0.0677807629108429,
      "step": 852
    },
    {
      "epoch": 0.11531994355676854,
      "grad_norm": 0.42723843455314636,
      "learning_rate": 2.981984774215214e-05,
      "loss": 0.04578013718128204,
      "step": 853
    },
    {
      "epoch": 0.1154551369255338,
      "grad_norm": 1.8594659566879272,
      "learning_rate": 2.9818788606583286e-05,
      "loss": 0.08013622462749481,
      "step": 854
    },
    {
      "epoch": 0.11559033029429906,
      "grad_norm": 1.2543073892593384,
      "learning_rate": 2.9817726385663627e-05,
      "loss": 0.05468093603849411,
      "step": 855
    },
    {
      "epoch": 0.11572552366306432,
      "grad_norm": 0.7090474963188171,
      "learning_rate": 2.9816661079614316e-05,
      "loss": 0.05808126926422119,
      "step": 856
    },
    {
      "epoch": 0.11586071703182958,
      "grad_norm": 0.5124869346618652,
      "learning_rate": 2.9815592688657154e-05,
      "loss": 0.06463829427957535,
      "step": 857
    },
    {
      "epoch": 0.11599591040059486,
      "grad_norm": 0.6906826496124268,
      "learning_rate": 2.9814521213014588e-05,
      "loss": 0.07837387919425964,
      "step": 858
    },
    {
      "epoch": 0.11613110376936012,
      "grad_norm": 0.575003981590271,
      "learning_rate": 2.9813446652909707e-05,
      "loss": 0.04965294897556305,
      "step": 859
    },
    {
      "epoch": 0.11626629713812538,
      "grad_norm": 1.7726686000823975,
      "learning_rate": 2.981236900856624e-05,
      "loss": 0.06196914613246918,
      "step": 860
    },
    {
      "epoch": 0.11640149050689064,
      "grad_norm": 4.158353328704834,
      "learning_rate": 2.9811288280208552e-05,
      "loss": 0.1057438999414444,
      "step": 861
    },
    {
      "epoch": 0.1165366838756559,
      "grad_norm": 2.241472005844116,
      "learning_rate": 2.9810204468061664e-05,
      "loss": 0.09526422619819641,
      "step": 862
    },
    {
      "epoch": 0.11667187724442116,
      "grad_norm": 1.0776751041412354,
      "learning_rate": 2.9809117572351223e-05,
      "loss": 0.09615468978881836,
      "step": 863
    },
    {
      "epoch": 0.11680707061318642,
      "grad_norm": 1.199429988861084,
      "learning_rate": 2.9808027593303537e-05,
      "loss": 0.07904808223247528,
      "step": 864
    },
    {
      "epoch": 0.11694226398195169,
      "grad_norm": 1.8700757026672363,
      "learning_rate": 2.980693453114554e-05,
      "loss": 0.07645630836486816,
      "step": 865
    },
    {
      "epoch": 0.11707745735071695,
      "grad_norm": 0.6234368085861206,
      "learning_rate": 2.980583838610481e-05,
      "loss": 0.058155059814453125,
      "step": 866
    },
    {
      "epoch": 0.11721265071948221,
      "grad_norm": 1.198128342628479,
      "learning_rate": 2.980473915840957e-05,
      "loss": 0.0551808625459671,
      "step": 867
    },
    {
      "epoch": 0.11734784408824747,
      "grad_norm": 0.831923246383667,
      "learning_rate": 2.9803636848288696e-05,
      "loss": 0.06490929424762726,
      "step": 868
    },
    {
      "epoch": 0.11748303745701273,
      "grad_norm": 2.5640063285827637,
      "learning_rate": 2.9802531455971686e-05,
      "loss": 0.06057201325893402,
      "step": 869
    },
    {
      "epoch": 0.11761823082577799,
      "grad_norm": 0.9984391927719116,
      "learning_rate": 2.980142298168869e-05,
      "loss": 0.04081990197300911,
      "step": 870
    },
    {
      "epoch": 0.11775342419454325,
      "grad_norm": 1.7455575466156006,
      "learning_rate": 2.9800311425670495e-05,
      "loss": 0.05051445960998535,
      "step": 871
    },
    {
      "epoch": 0.11788861756330853,
      "grad_norm": 1.3665772676467896,
      "learning_rate": 2.9799196788148538e-05,
      "loss": 0.06193774938583374,
      "step": 872
    },
    {
      "epoch": 0.11802381093207379,
      "grad_norm": 1.3060983419418335,
      "learning_rate": 2.9798079069354893e-05,
      "loss": 0.04442111402750015,
      "step": 873
    },
    {
      "epoch": 0.11815900430083905,
      "grad_norm": 1.7216105461120605,
      "learning_rate": 2.9796958269522273e-05,
      "loss": 0.06150418519973755,
      "step": 874
    },
    {
      "epoch": 0.1182941976696043,
      "grad_norm": 1.261020302772522,
      "learning_rate": 2.9795834388884034e-05,
      "loss": 0.06411415338516235,
      "step": 875
    },
    {
      "epoch": 0.11842939103836957,
      "grad_norm": 1.3385952711105347,
      "learning_rate": 2.979470742767417e-05,
      "loss": 0.040393806993961334,
      "step": 876
    },
    {
      "epoch": 0.11856458440713483,
      "grad_norm": 1.189489483833313,
      "learning_rate": 2.9793577386127327e-05,
      "loss": 0.059862058609724045,
      "step": 877
    },
    {
      "epoch": 0.11869977777590009,
      "grad_norm": 0.8483715057373047,
      "learning_rate": 2.9792444264478784e-05,
      "loss": 0.09011414647102356,
      "step": 878
    },
    {
      "epoch": 0.11883497114466536,
      "grad_norm": 0.8031742572784424,
      "learning_rate": 2.979130806296446e-05,
      "loss": 0.0652703046798706,
      "step": 879
    },
    {
      "epoch": 0.11897016451343062,
      "grad_norm": 1.6406009197235107,
      "learning_rate": 2.9790168781820925e-05,
      "loss": 0.05732802674174309,
      "step": 880
    },
    {
      "epoch": 0.11910535788219588,
      "grad_norm": 2.006895065307617,
      "learning_rate": 2.9789026421285375e-05,
      "loss": 0.06943021714687347,
      "step": 881
    },
    {
      "epoch": 0.11924055125096114,
      "grad_norm": 0.8554799556732178,
      "learning_rate": 2.9787880981595663e-05,
      "loss": 0.06801632046699524,
      "step": 882
    },
    {
      "epoch": 0.1193757446197264,
      "grad_norm": 0.3612128794193268,
      "learning_rate": 2.9786732462990267e-05,
      "loss": 0.05761076509952545,
      "step": 883
    },
    {
      "epoch": 0.11951093798849166,
      "grad_norm": 0.5959079265594482,
      "learning_rate": 2.9785580865708323e-05,
      "loss": 0.05626345053315163,
      "step": 884
    },
    {
      "epoch": 0.11964613135725692,
      "grad_norm": 1.26160728931427,
      "learning_rate": 2.97844261899896e-05,
      "loss": 0.06064796447753906,
      "step": 885
    },
    {
      "epoch": 0.1197813247260222,
      "grad_norm": 0.37730908393859863,
      "learning_rate": 2.9783268436074495e-05,
      "loss": 0.03973902761936188,
      "step": 886
    },
    {
      "epoch": 0.11991651809478746,
      "grad_norm": 1.4231590032577515,
      "learning_rate": 2.978210760420407e-05,
      "loss": 0.05634953826665878,
      "step": 887
    },
    {
      "epoch": 0.12005171146355272,
      "grad_norm": 1.3626042604446411,
      "learning_rate": 2.978094369462002e-05,
      "loss": 0.06598908454179764,
      "step": 888
    },
    {
      "epoch": 0.12018690483231798,
      "grad_norm": 1.5572553873062134,
      "learning_rate": 2.977977670756467e-05,
      "loss": 0.05168672651052475,
      "step": 889
    },
    {
      "epoch": 0.12032209820108324,
      "grad_norm": 0.9045180082321167,
      "learning_rate": 2.9778606643280987e-05,
      "loss": 0.05312810838222504,
      "step": 890
    },
    {
      "epoch": 0.1204572915698485,
      "grad_norm": 0.4083821475505829,
      "learning_rate": 2.97774335020126e-05,
      "loss": 0.06432473659515381,
      "step": 891
    },
    {
      "epoch": 0.12059248493861376,
      "grad_norm": 1.6102540493011475,
      "learning_rate": 2.9776257284003748e-05,
      "loss": 0.06559291481971741,
      "step": 892
    },
    {
      "epoch": 0.12072767830737903,
      "grad_norm": 0.7310768365859985,
      "learning_rate": 2.9775077989499338e-05,
      "loss": 0.0623704195022583,
      "step": 893
    },
    {
      "epoch": 0.12086287167614429,
      "grad_norm": 1.1899971961975098,
      "learning_rate": 2.97738956187449e-05,
      "loss": 0.06307214498519897,
      "step": 894
    },
    {
      "epoch": 0.12099806504490955,
      "grad_norm": 0.5913503766059875,
      "learning_rate": 2.9772710171986605e-05,
      "loss": 0.05548405647277832,
      "step": 895
    },
    {
      "epoch": 0.12113325841367481,
      "grad_norm": 0.3939407467842102,
      "learning_rate": 2.977152164947128e-05,
      "loss": 0.05418830364942551,
      "step": 896
    },
    {
      "epoch": 0.12126845178244007,
      "grad_norm": 0.5984494686126709,
      "learning_rate": 2.9770330051446373e-05,
      "loss": 0.06009635329246521,
      "step": 897
    },
    {
      "epoch": 0.12140364515120533,
      "grad_norm": 0.4753250777721405,
      "learning_rate": 2.976913537815999e-05,
      "loss": 0.05103091523051262,
      "step": 898
    },
    {
      "epoch": 0.12153883851997059,
      "grad_norm": 1.1283420324325562,
      "learning_rate": 2.9767937629860853e-05,
      "loss": 0.061839379370212555,
      "step": 899
    },
    {
      "epoch": 0.12167403188873585,
      "grad_norm": 0.39484310150146484,
      "learning_rate": 2.9766736806798353e-05,
      "loss": 0.03699915111064911,
      "step": 900
    },
    {
      "epoch": 0.12180922525750112,
      "grad_norm": 1.2396390438079834,
      "learning_rate": 2.9765532909222512e-05,
      "loss": 0.09471902251243591,
      "step": 901
    },
    {
      "epoch": 0.12194441862626638,
      "grad_norm": 0.6397807002067566,
      "learning_rate": 2.976432593738397e-05,
      "loss": 0.06966215372085571,
      "step": 902
    },
    {
      "epoch": 0.12207961199503164,
      "grad_norm": 2.4892289638519287,
      "learning_rate": 2.9763115891534036e-05,
      "loss": 0.06544773280620575,
      "step": 903
    },
    {
      "epoch": 0.1222148053637969,
      "grad_norm": 1.672637939453125,
      "learning_rate": 2.9761902771924648e-05,
      "loss": 0.06459483504295349,
      "step": 904
    },
    {
      "epoch": 0.12234999873256217,
      "grad_norm": 1.1361255645751953,
      "learning_rate": 2.9760686578808387e-05,
      "loss": 0.098862424492836,
      "step": 905
    },
    {
      "epoch": 0.12248519210132743,
      "grad_norm": 0.5333594083786011,
      "learning_rate": 2.9759467312438462e-05,
      "loss": 0.055858880281448364,
      "step": 906
    },
    {
      "epoch": 0.12262038547009269,
      "grad_norm": 4.857579708099365,
      "learning_rate": 2.975824497306874e-05,
      "loss": 0.08147549629211426,
      "step": 907
    },
    {
      "epoch": 0.12275557883885796,
      "grad_norm": 2.879769802093506,
      "learning_rate": 2.9757019560953707e-05,
      "loss": 0.06890100240707397,
      "step": 908
    },
    {
      "epoch": 0.12289077220762322,
      "grad_norm": 1.9118424654006958,
      "learning_rate": 2.9755791076348517e-05,
      "loss": 0.07549156248569489,
      "step": 909
    },
    {
      "epoch": 0.12302596557638848,
      "grad_norm": 1.9207574129104614,
      "learning_rate": 2.9754559519508924e-05,
      "loss": 0.05638545751571655,
      "step": 910
    },
    {
      "epoch": 0.12316115894515374,
      "grad_norm": 1.3111354112625122,
      "learning_rate": 2.975332489069137e-05,
      "loss": 0.06364589929580688,
      "step": 911
    },
    {
      "epoch": 0.123296352313919,
      "grad_norm": 1.1506239175796509,
      "learning_rate": 2.9752087190152893e-05,
      "loss": 0.05111977458000183,
      "step": 912
    },
    {
      "epoch": 0.12343154568268426,
      "grad_norm": 2.007946491241455,
      "learning_rate": 2.97508464181512e-05,
      "loss": 0.06762352585792542,
      "step": 913
    },
    {
      "epoch": 0.12356673905144952,
      "grad_norm": 3.3811326026916504,
      "learning_rate": 2.9749602574944615e-05,
      "loss": 0.06516307592391968,
      "step": 914
    },
    {
      "epoch": 0.1237019324202148,
      "grad_norm": 3.5728046894073486,
      "learning_rate": 2.9748355660792125e-05,
      "loss": 0.07153135538101196,
      "step": 915
    },
    {
      "epoch": 0.12383712578898005,
      "grad_norm": 5.441051006317139,
      "learning_rate": 2.9747105675953338e-05,
      "loss": 0.10422030091285706,
      "step": 916
    },
    {
      "epoch": 0.12397231915774531,
      "grad_norm": 1.6395776271820068,
      "learning_rate": 2.9745852620688506e-05,
      "loss": 0.05035269260406494,
      "step": 917
    },
    {
      "epoch": 0.12410751252651057,
      "grad_norm": 2.6748416423797607,
      "learning_rate": 2.974459649525853e-05,
      "loss": 0.0686805248260498,
      "step": 918
    },
    {
      "epoch": 0.12424270589527583,
      "grad_norm": 0.5252537727355957,
      "learning_rate": 2.9743337299924925e-05,
      "loss": 0.07046585530042648,
      "step": 919
    },
    {
      "epoch": 0.1243778992640411,
      "grad_norm": 0.5221901535987854,
      "learning_rate": 2.9742075034949883e-05,
      "loss": 0.05802202224731445,
      "step": 920
    },
    {
      "epoch": 0.12451309263280635,
      "grad_norm": 1.5236924886703491,
      "learning_rate": 2.97408097005962e-05,
      "loss": 0.049811482429504395,
      "step": 921
    },
    {
      "epoch": 0.12464828600157163,
      "grad_norm": 0.9998456835746765,
      "learning_rate": 2.973954129712733e-05,
      "loss": 0.0606459379196167,
      "step": 922
    },
    {
      "epoch": 0.12478347937033689,
      "grad_norm": 1.7388442754745483,
      "learning_rate": 2.973826982480736e-05,
      "loss": 0.04614129662513733,
      "step": 923
    },
    {
      "epoch": 0.12491867273910215,
      "grad_norm": 0.809036374092102,
      "learning_rate": 2.9736995283901022e-05,
      "loss": 0.07797861099243164,
      "step": 924
    },
    {
      "epoch": 0.12505386610786742,
      "grad_norm": 0.873992919921875,
      "learning_rate": 2.9735717674673676e-05,
      "loss": 0.048703208565711975,
      "step": 925
    },
    {
      "epoch": 0.12518905947663267,
      "grad_norm": 0.6025969386100769,
      "learning_rate": 2.973443699739133e-05,
      "loss": 0.05472603440284729,
      "step": 926
    },
    {
      "epoch": 0.12532425284539794,
      "grad_norm": 1.1863207817077637,
      "learning_rate": 2.973315325232063e-05,
      "loss": 0.06854510307312012,
      "step": 927
    },
    {
      "epoch": 0.1254594462141632,
      "grad_norm": 0.7763834595680237,
      "learning_rate": 2.9731866439728853e-05,
      "loss": 0.07515951991081238,
      "step": 928
    },
    {
      "epoch": 0.12559463958292846,
      "grad_norm": 1.0926889181137085,
      "learning_rate": 2.9730576559883924e-05,
      "loss": 0.0751802921295166,
      "step": 929
    },
    {
      "epoch": 0.1257298329516937,
      "grad_norm": 0.715045154094696,
      "learning_rate": 2.97292836130544e-05,
      "loss": 0.06445688009262085,
      "step": 930
    },
    {
      "epoch": 0.12586502632045898,
      "grad_norm": 0.6126769185066223,
      "learning_rate": 2.9727987599509485e-05,
      "loss": 0.07244141399860382,
      "step": 931
    },
    {
      "epoch": 0.12600021968922423,
      "grad_norm": 2.9130282402038574,
      "learning_rate": 2.972668851951901e-05,
      "loss": 0.08251702785491943,
      "step": 932
    },
    {
      "epoch": 0.1261354130579895,
      "grad_norm": 3.5020670890808105,
      "learning_rate": 2.9725386373353455e-05,
      "loss": 0.0746297836303711,
      "step": 933
    },
    {
      "epoch": 0.12627060642675478,
      "grad_norm": 2.260486364364624,
      "learning_rate": 2.972408116128393e-05,
      "loss": 0.055924415588378906,
      "step": 934
    },
    {
      "epoch": 0.12640579979552002,
      "grad_norm": 0.6031903624534607,
      "learning_rate": 2.972277288358219e-05,
      "loss": 0.07023245096206665,
      "step": 935
    },
    {
      "epoch": 0.1265409931642853,
      "grad_norm": 1.5049775838851929,
      "learning_rate": 2.9721461540520628e-05,
      "loss": 0.07419735193252563,
      "step": 936
    },
    {
      "epoch": 0.12667618653305054,
      "grad_norm": 0.8745632171630859,
      "learning_rate": 2.9720147132372265e-05,
      "loss": 0.047914132475852966,
      "step": 937
    },
    {
      "epoch": 0.12681137990181582,
      "grad_norm": 1.3793575763702393,
      "learning_rate": 2.9718829659410772e-05,
      "loss": 0.07423095405101776,
      "step": 938
    },
    {
      "epoch": 0.12694657327058106,
      "grad_norm": 1.5917911529541016,
      "learning_rate": 2.9717509121910453e-05,
      "loss": 0.08392691612243652,
      "step": 939
    },
    {
      "epoch": 0.12708176663934634,
      "grad_norm": 1.1619272232055664,
      "learning_rate": 2.971618552014625e-05,
      "loss": 0.0750933438539505,
      "step": 940
    },
    {
      "epoch": 0.1272169600081116,
      "grad_norm": 1.3756927251815796,
      "learning_rate": 2.971485885439375e-05,
      "loss": 0.05483636260032654,
      "step": 941
    },
    {
      "epoch": 0.12735215337687686,
      "grad_norm": 1.454074501991272,
      "learning_rate": 2.9713529124929163e-05,
      "loss": 0.06254559755325317,
      "step": 942
    },
    {
      "epoch": 0.12748734674564213,
      "grad_norm": 0.8426021933555603,
      "learning_rate": 2.9712196332029352e-05,
      "loss": 0.06235361099243164,
      "step": 943
    },
    {
      "epoch": 0.12762254011440738,
      "grad_norm": 0.5614035129547119,
      "learning_rate": 2.971086047597181e-05,
      "loss": 0.0639098584651947,
      "step": 944
    },
    {
      "epoch": 0.12775773348317265,
      "grad_norm": 0.4427438974380493,
      "learning_rate": 2.9709521557034668e-05,
      "loss": 0.05317005515098572,
      "step": 945
    },
    {
      "epoch": 0.1278929268519379,
      "grad_norm": 1.5867512226104736,
      "learning_rate": 2.9708179575496696e-05,
      "loss": 0.06115603446960449,
      "step": 946
    },
    {
      "epoch": 0.12802812022070317,
      "grad_norm": 1.1928867101669312,
      "learning_rate": 2.9706834531637303e-05,
      "loss": 0.06077621877193451,
      "step": 947
    },
    {
      "epoch": 0.12816331358946845,
      "grad_norm": 1.291746973991394,
      "learning_rate": 2.9705486425736537e-05,
      "loss": 0.06889256089925766,
      "step": 948
    },
    {
      "epoch": 0.1282985069582337,
      "grad_norm": 0.37629184126853943,
      "learning_rate": 2.9704135258075077e-05,
      "loss": 0.056202471256256104,
      "step": 949
    },
    {
      "epoch": 0.12843370032699897,
      "grad_norm": 0.49766677618026733,
      "learning_rate": 2.970278102893424e-05,
      "loss": 0.06517137587070465,
      "step": 950
    },
    {
      "epoch": 0.1285688936957642,
      "grad_norm": 3.0935404300689697,
      "learning_rate": 2.9701423738595992e-05,
      "loss": 0.06942284107208252,
      "step": 951
    },
    {
      "epoch": 0.1287040870645295,
      "grad_norm": 0.8044374585151672,
      "learning_rate": 2.9700063387342925e-05,
      "loss": 0.05415910482406616,
      "step": 952
    },
    {
      "epoch": 0.12883928043329473,
      "grad_norm": 1.8038911819458008,
      "learning_rate": 2.969869997545827e-05,
      "loss": 0.05029727518558502,
      "step": 953
    },
    {
      "epoch": 0.12897447380206,
      "grad_norm": 1.792656660079956,
      "learning_rate": 2.9697333503225897e-05,
      "loss": 0.08134263753890991,
      "step": 954
    },
    {
      "epoch": 0.12910966717082528,
      "grad_norm": 1.6244159936904907,
      "learning_rate": 2.969596397093031e-05,
      "loss": 0.08842146396636963,
      "step": 955
    },
    {
      "epoch": 0.12924486053959053,
      "grad_norm": 1.7177331447601318,
      "learning_rate": 2.969459137885666e-05,
      "loss": 0.06281697750091553,
      "step": 956
    },
    {
      "epoch": 0.1293800539083558,
      "grad_norm": 0.923691987991333,
      "learning_rate": 2.969321572729072e-05,
      "loss": 0.06180095672607422,
      "step": 957
    },
    {
      "epoch": 0.12951524727712105,
      "grad_norm": 1.5431050062179565,
      "learning_rate": 2.9691837016518915e-05,
      "loss": 0.05874697118997574,
      "step": 958
    },
    {
      "epoch": 0.12965044064588632,
      "grad_norm": 1.1677707433700562,
      "learning_rate": 2.9690455246828294e-05,
      "loss": 0.07713724672794342,
      "step": 959
    },
    {
      "epoch": 0.12978563401465157,
      "grad_norm": 0.8348269462585449,
      "learning_rate": 2.968907041850655e-05,
      "loss": 0.09317457675933838,
      "step": 960
    },
    {
      "epoch": 0.12992082738341684,
      "grad_norm": 1.7617661952972412,
      "learning_rate": 2.968768253184202e-05,
      "loss": 0.08088464289903641,
      "step": 961
    },
    {
      "epoch": 0.13005602075218212,
      "grad_norm": 0.8855709433555603,
      "learning_rate": 2.9686291587123655e-05,
      "loss": 0.05380135774612427,
      "step": 962
    },
    {
      "epoch": 0.13019121412094736,
      "grad_norm": 1.5261743068695068,
      "learning_rate": 2.968489758464107e-05,
      "loss": 0.08649598062038422,
      "step": 963
    },
    {
      "epoch": 0.13032640748971264,
      "grad_norm": 0.648188591003418,
      "learning_rate": 2.9683500524684494e-05,
      "loss": 0.07602918148040771,
      "step": 964
    },
    {
      "epoch": 0.13046160085847788,
      "grad_norm": 0.34662652015686035,
      "learning_rate": 2.9682100407544812e-05,
      "loss": 0.057284027338027954,
      "step": 965
    },
    {
      "epoch": 0.13059679422724316,
      "grad_norm": 0.922051727771759,
      "learning_rate": 2.9680697233513526e-05,
      "loss": 0.06950879096984863,
      "step": 966
    },
    {
      "epoch": 0.1307319875960084,
      "grad_norm": 0.43236103653907776,
      "learning_rate": 2.9679291002882793e-05,
      "loss": 0.058712005615234375,
      "step": 967
    },
    {
      "epoch": 0.13086718096477368,
      "grad_norm": 0.8399235010147095,
      "learning_rate": 2.967788171594539e-05,
      "loss": 0.0585600882768631,
      "step": 968
    },
    {
      "epoch": 0.13100237433353895,
      "grad_norm": 1.3014203310012817,
      "learning_rate": 2.967646937299474e-05,
      "loss": 0.03898021578788757,
      "step": 969
    },
    {
      "epoch": 0.1311375677023042,
      "grad_norm": 0.47541749477386475,
      "learning_rate": 2.9675053974324907e-05,
      "loss": 0.07133780419826508,
      "step": 970
    },
    {
      "epoch": 0.13127276107106947,
      "grad_norm": 0.48997896909713745,
      "learning_rate": 2.9673635520230576e-05,
      "loss": 0.06527069211006165,
      "step": 971
    },
    {
      "epoch": 0.13140795443983472,
      "grad_norm": 0.5041906237602234,
      "learning_rate": 2.9672214011007087e-05,
      "loss": 0.07057970762252808,
      "step": 972
    },
    {
      "epoch": 0.1315431478086,
      "grad_norm": 0.5192206501960754,
      "learning_rate": 2.9670789446950396e-05,
      "loss": 0.07677845656871796,
      "step": 973
    },
    {
      "epoch": 0.13167834117736524,
      "grad_norm": 0.5961480736732483,
      "learning_rate": 2.9669361828357105e-05,
      "loss": 0.06424421072006226,
      "step": 974
    },
    {
      "epoch": 0.1318135345461305,
      "grad_norm": 0.3776947259902954,
      "learning_rate": 2.9667931155524454e-05,
      "loss": 0.05686786770820618,
      "step": 975
    },
    {
      "epoch": 0.13194872791489579,
      "grad_norm": 0.7766925096511841,
      "learning_rate": 2.966649742875032e-05,
      "loss": 0.03621548414230347,
      "step": 976
    },
    {
      "epoch": 0.13208392128366103,
      "grad_norm": 1.2372055053710938,
      "learning_rate": 2.9665060648333206e-05,
      "loss": 0.08590242266654968,
      "step": 977
    },
    {
      "epoch": 0.1322191146524263,
      "grad_norm": 0.538113534450531,
      "learning_rate": 2.9663620814572266e-05,
      "loss": 0.049870461225509644,
      "step": 978
    },
    {
      "epoch": 0.13235430802119155,
      "grad_norm": 0.5654088854789734,
      "learning_rate": 2.966217792776728e-05,
      "loss": 0.04519476741552353,
      "step": 979
    },
    {
      "epoch": 0.13248950138995683,
      "grad_norm": 1.511000633239746,
      "learning_rate": 2.9660731988218652e-05,
      "loss": 0.05653522536158562,
      "step": 980
    },
    {
      "epoch": 0.13262469475872207,
      "grad_norm": 1.6901277303695679,
      "learning_rate": 2.965928299622745e-05,
      "loss": 0.07455629110336304,
      "step": 981
    },
    {
      "epoch": 0.13275988812748735,
      "grad_norm": 1.12834632396698,
      "learning_rate": 2.965783095209535e-05,
      "loss": 0.058841437101364136,
      "step": 982
    },
    {
      "epoch": 0.13289508149625262,
      "grad_norm": 0.83742755651474,
      "learning_rate": 2.965637585612469e-05,
      "loss": 0.06962665915489197,
      "step": 983
    },
    {
      "epoch": 0.13303027486501787,
      "grad_norm": 0.4493842124938965,
      "learning_rate": 2.965491770861841e-05,
      "loss": 0.05994352698326111,
      "step": 984
    },
    {
      "epoch": 0.13316546823378314,
      "grad_norm": 0.37051379680633545,
      "learning_rate": 2.965345650988012e-05,
      "loss": 0.050818562507629395,
      "step": 985
    },
    {
      "epoch": 0.1333006616025484,
      "grad_norm": 1.1987723112106323,
      "learning_rate": 2.9651992260214035e-05,
      "loss": 0.08498618006706238,
      "step": 986
    },
    {
      "epoch": 0.13343585497131366,
      "grad_norm": 0.7185893058776855,
      "learning_rate": 2.9650524959925037e-05,
      "loss": 0.04328185319900513,
      "step": 987
    },
    {
      "epoch": 0.1335710483400789,
      "grad_norm": 1.0075762271881104,
      "learning_rate": 2.9649054609318607e-05,
      "loss": 0.060393065214157104,
      "step": 988
    },
    {
      "epoch": 0.13370624170884418,
      "grad_norm": 1.6080729961395264,
      "learning_rate": 2.9647581208700894e-05,
      "loss": 0.06977052986621857,
      "step": 989
    },
    {
      "epoch": 0.13384143507760946,
      "grad_norm": 0.7243795394897461,
      "learning_rate": 2.9646104758378666e-05,
      "loss": 0.06446462869644165,
      "step": 990
    },
    {
      "epoch": 0.1339766284463747,
      "grad_norm": 0.9294887185096741,
      "learning_rate": 2.964462525865932e-05,
      "loss": 0.0632893443107605,
      "step": 991
    },
    {
      "epoch": 0.13411182181513998,
      "grad_norm": 0.4036575257778168,
      "learning_rate": 2.96431427098509e-05,
      "loss": 0.06646198034286499,
      "step": 992
    },
    {
      "epoch": 0.13424701518390522,
      "grad_norm": 0.7816623449325562,
      "learning_rate": 2.9641657112262084e-05,
      "loss": 0.06584139168262482,
      "step": 993
    },
    {
      "epoch": 0.1343822085526705,
      "grad_norm": 1.8476940393447876,
      "learning_rate": 2.9640168466202174e-05,
      "loss": 0.06878417730331421,
      "step": 994
    },
    {
      "epoch": 0.13451740192143574,
      "grad_norm": 1.5821086168289185,
      "learning_rate": 2.9638676771981124e-05,
      "loss": 0.0807846188545227,
      "step": 995
    },
    {
      "epoch": 0.13465259529020102,
      "grad_norm": 0.7830606698989868,
      "learning_rate": 2.9637182029909508e-05,
      "loss": 0.08169925212860107,
      "step": 996
    },
    {
      "epoch": 0.1347877886589663,
      "grad_norm": 0.4132815897464752,
      "learning_rate": 2.9635684240298532e-05,
      "loss": 0.07634192705154419,
      "step": 997
    },
    {
      "epoch": 0.13492298202773154,
      "grad_norm": 2.0433011054992676,
      "learning_rate": 2.9634183403460053e-05,
      "loss": 0.06839564442634583,
      "step": 998
    },
    {
      "epoch": 0.1350581753964968,
      "grad_norm": 1.4567112922668457,
      "learning_rate": 2.9632679519706553e-05,
      "loss": 0.0772167444229126,
      "step": 999
    },
    {
      "epoch": 0.13519336876526206,
      "grad_norm": 1.1899508237838745,
      "learning_rate": 2.9631172589351137e-05,
      "loss": 0.07662512362003326,
      "step": 1000
    },
    {
      "epoch": 0.13532856213402733,
      "grad_norm": 0.44104450941085815,
      "learning_rate": 2.962966261270758e-05,
      "loss": 0.0652366429567337,
      "step": 1001
    },
    {
      "epoch": 0.13546375550279258,
      "grad_norm": 1.766348123550415,
      "learning_rate": 2.962814959009024e-05,
      "loss": 0.06794068217277527,
      "step": 1002
    },
    {
      "epoch": 0.13559894887155785,
      "grad_norm": 1.7373936176300049,
      "learning_rate": 2.962663352181415e-05,
      "loss": 0.0638173520565033,
      "step": 1003
    },
    {
      "epoch": 0.13573414224032312,
      "grad_norm": 0.5086437463760376,
      "learning_rate": 2.9625114408194966e-05,
      "loss": 0.06538169085979462,
      "step": 1004
    },
    {
      "epoch": 0.13586933560908837,
      "grad_norm": 1.0753984451293945,
      "learning_rate": 2.962359224954897e-05,
      "loss": 0.059414565563201904,
      "step": 1005
    },
    {
      "epoch": 0.13600452897785364,
      "grad_norm": 0.728958010673523,
      "learning_rate": 2.9622067046193086e-05,
      "loss": 0.07270858436822891,
      "step": 1006
    },
    {
      "epoch": 0.1361397223466189,
      "grad_norm": 1.2851983308792114,
      "learning_rate": 2.9620538798444867e-05,
      "loss": 0.05009317398071289,
      "step": 1007
    },
    {
      "epoch": 0.13627491571538417,
      "grad_norm": 0.3558655083179474,
      "learning_rate": 2.9619007506622506e-05,
      "loss": 0.040700286626815796,
      "step": 1008
    },
    {
      "epoch": 0.1364101090841494,
      "grad_norm": 0.8928632736206055,
      "learning_rate": 2.961747317104482e-05,
      "loss": 0.06870028376579285,
      "step": 1009
    },
    {
      "epoch": 0.13654530245291469,
      "grad_norm": 1.4920967817306519,
      "learning_rate": 2.9615935792031274e-05,
      "loss": 0.06532464176416397,
      "step": 1010
    },
    {
      "epoch": 0.13668049582167996,
      "grad_norm": 0.9910970330238342,
      "learning_rate": 2.9614395369901953e-05,
      "loss": 0.05240216851234436,
      "step": 1011
    },
    {
      "epoch": 0.1368156891904452,
      "grad_norm": 0.6473343372344971,
      "learning_rate": 2.9612851904977582e-05,
      "loss": 0.059423238039016724,
      "step": 1012
    },
    {
      "epoch": 0.13695088255921048,
      "grad_norm": 0.6781181693077087,
      "learning_rate": 2.9611305397579518e-05,
      "loss": 0.0826173722743988,
      "step": 1013
    },
    {
      "epoch": 0.13708607592797573,
      "grad_norm": 2.144813299179077,
      "learning_rate": 2.9609755848029755e-05,
      "loss": 0.06403718888759613,
      "step": 1014
    },
    {
      "epoch": 0.137221269296741,
      "grad_norm": 1.1143856048583984,
      "learning_rate": 2.9608203256650916e-05,
      "loss": 0.07878640294075012,
      "step": 1015
    },
    {
      "epoch": 0.13735646266550625,
      "grad_norm": 1.0002622604370117,
      "learning_rate": 2.9606647623766257e-05,
      "loss": 0.07778096944093704,
      "step": 1016
    },
    {
      "epoch": 0.13749165603427152,
      "grad_norm": 2.941098690032959,
      "learning_rate": 2.9605088949699672e-05,
      "loss": 0.08105441927909851,
      "step": 1017
    },
    {
      "epoch": 0.1376268494030368,
      "grad_norm": 0.9422494769096375,
      "learning_rate": 2.9603527234775682e-05,
      "loss": 0.040878817439079285,
      "step": 1018
    },
    {
      "epoch": 0.13776204277180204,
      "grad_norm": 1.3126517534255981,
      "learning_rate": 2.960196247931945e-05,
      "loss": 0.05973470211029053,
      "step": 1019
    },
    {
      "epoch": 0.13789723614056731,
      "grad_norm": 1.0239661931991577,
      "learning_rate": 2.960039468365676e-05,
      "loss": 0.0669485330581665,
      "step": 1020
    },
    {
      "epoch": 0.13803242950933256,
      "grad_norm": 1.0040719509124756,
      "learning_rate": 2.959882384811404e-05,
      "loss": 0.057542502880096436,
      "step": 1021
    },
    {
      "epoch": 0.13816762287809783,
      "grad_norm": 1.0834072828292847,
      "learning_rate": 2.9597249973018343e-05,
      "loss": 0.0507512241601944,
      "step": 1022
    },
    {
      "epoch": 0.13830281624686308,
      "grad_norm": 1.3461211919784546,
      "learning_rate": 2.959567305869736e-05,
      "loss": 0.06819938868284225,
      "step": 1023
    },
    {
      "epoch": 0.13843800961562835,
      "grad_norm": 1.9950796365737915,
      "learning_rate": 2.9594093105479413e-05,
      "loss": 0.05097164213657379,
      "step": 1024
    },
    {
      "epoch": 0.13857320298439363,
      "grad_norm": 0.788266658782959,
      "learning_rate": 2.959251011369345e-05,
      "loss": 0.056308358907699585,
      "step": 1025
    },
    {
      "epoch": 0.13870839635315887,
      "grad_norm": 0.7522055506706238,
      "learning_rate": 2.959092408366907e-05,
      "loss": 0.06282210350036621,
      "step": 1026
    },
    {
      "epoch": 0.13884358972192415,
      "grad_norm": 1.9201449155807495,
      "learning_rate": 2.958933501573649e-05,
      "loss": 0.07393568754196167,
      "step": 1027
    },
    {
      "epoch": 0.1389787830906894,
      "grad_norm": 0.47177568078041077,
      "learning_rate": 2.9587742910226555e-05,
      "loss": 0.06012946367263794,
      "step": 1028
    },
    {
      "epoch": 0.13911397645945467,
      "grad_norm": 0.7972776293754578,
      "learning_rate": 2.958614776747076e-05,
      "loss": 0.06980639696121216,
      "step": 1029
    },
    {
      "epoch": 0.13924916982821992,
      "grad_norm": 1.3969213962554932,
      "learning_rate": 2.9584549587801213e-05,
      "loss": 0.08450901508331299,
      "step": 1030
    },
    {
      "epoch": 0.1393843631969852,
      "grad_norm": 0.9194837808609009,
      "learning_rate": 2.958294837155067e-05,
      "loss": 0.06217719614505768,
      "step": 1031
    },
    {
      "epoch": 0.13951955656575046,
      "grad_norm": 0.4585019052028656,
      "learning_rate": 2.9581344119052508e-05,
      "loss": 0.059274494647979736,
      "step": 1032
    },
    {
      "epoch": 0.1396547499345157,
      "grad_norm": 0.32113614678382874,
      "learning_rate": 2.957973683064074e-05,
      "loss": 0.06717707961797714,
      "step": 1033
    },
    {
      "epoch": 0.13978994330328098,
      "grad_norm": 1.8591957092285156,
      "learning_rate": 2.957812650665002e-05,
      "loss": 0.05622287094593048,
      "step": 1034
    },
    {
      "epoch": 0.13992513667204623,
      "grad_norm": 2.2799623012542725,
      "learning_rate": 2.957651314741562e-05,
      "loss": 0.05717763304710388,
      "step": 1035
    },
    {
      "epoch": 0.1400603300408115,
      "grad_norm": 1.5666700601577759,
      "learning_rate": 2.9574896753273454e-05,
      "loss": 0.08140039443969727,
      "step": 1036
    },
    {
      "epoch": 0.14019552340957675,
      "grad_norm": 0.5222755074501038,
      "learning_rate": 2.9573277324560058e-05,
      "loss": 0.054591115564107895,
      "step": 1037
    },
    {
      "epoch": 0.14033071677834202,
      "grad_norm": 1.7303684949874878,
      "learning_rate": 2.9571654861612608e-05,
      "loss": 0.06008630990982056,
      "step": 1038
    },
    {
      "epoch": 0.1404659101471073,
      "grad_norm": 1.889082431793213,
      "learning_rate": 2.957002936476891e-05,
      "loss": 0.06525285542011261,
      "step": 1039
    },
    {
      "epoch": 0.14060110351587254,
      "grad_norm": 1.5210939645767212,
      "learning_rate": 2.9568400834367406e-05,
      "loss": 0.06174790859222412,
      "step": 1040
    },
    {
      "epoch": 0.14073629688463782,
      "grad_norm": 1.0190696716308594,
      "learning_rate": 2.9566769270747158e-05,
      "loss": 0.06712540984153748,
      "step": 1041
    },
    {
      "epoch": 0.14087149025340306,
      "grad_norm": 0.42528876662254333,
      "learning_rate": 2.9565134674247864e-05,
      "loss": 0.06162145733833313,
      "step": 1042
    },
    {
      "epoch": 0.14100668362216834,
      "grad_norm": 0.9857556223869324,
      "learning_rate": 2.9563497045209866e-05,
      "loss": 0.07064189016819,
      "step": 1043
    },
    {
      "epoch": 0.14114187699093358,
      "grad_norm": 2.3487162590026855,
      "learning_rate": 2.9561856383974118e-05,
      "loss": 0.0643070861697197,
      "step": 1044
    },
    {
      "epoch": 0.14127707035969886,
      "grad_norm": 0.4388207495212555,
      "learning_rate": 2.9560212690882218e-05,
      "loss": 0.056035563349723816,
      "step": 1045
    },
    {
      "epoch": 0.14141226372846413,
      "grad_norm": 1.9741652011871338,
      "learning_rate": 2.9558565966276395e-05,
      "loss": 0.06551049649715424,
      "step": 1046
    },
    {
      "epoch": 0.14154745709722938,
      "grad_norm": 2.2496907711029053,
      "learning_rate": 2.9556916210499497e-05,
      "loss": 0.0794958546757698,
      "step": 1047
    },
    {
      "epoch": 0.14168265046599465,
      "grad_norm": 0.6142613887786865,
      "learning_rate": 2.9555263423895016e-05,
      "loss": 0.07496144622564316,
      "step": 1048
    },
    {
      "epoch": 0.1418178438347599,
      "grad_norm": 0.809754490852356,
      "learning_rate": 2.955360760680708e-05,
      "loss": 0.0774577409029007,
      "step": 1049
    },
    {
      "epoch": 0.14195303720352517,
      "grad_norm": 1.5004494190216064,
      "learning_rate": 2.9551948759580423e-05,
      "loss": 0.06169368326663971,
      "step": 1050
    },
    {
      "epoch": 0.14208823057229042,
      "grad_norm": 1.9727504253387451,
      "learning_rate": 2.9550286882560435e-05,
      "loss": 0.07013940811157227,
      "step": 1051
    },
    {
      "epoch": 0.1422234239410557,
      "grad_norm": 1.5525301694869995,
      "learning_rate": 2.9548621976093126e-05,
      "loss": 0.057036034762859344,
      "step": 1052
    },
    {
      "epoch": 0.14235861730982094,
      "grad_norm": 0.6294586062431335,
      "learning_rate": 2.9546954040525144e-05,
      "loss": 0.07674640417098999,
      "step": 1053
    },
    {
      "epoch": 0.1424938106785862,
      "grad_norm": 0.6791669130325317,
      "learning_rate": 2.9545283076203753e-05,
      "loss": 0.05670631676912308,
      "step": 1054
    },
    {
      "epoch": 0.1426290040473515,
      "grad_norm": 2.116180658340454,
      "learning_rate": 2.954360908347686e-05,
      "loss": 0.09419101476669312,
      "step": 1055
    },
    {
      "epoch": 0.14276419741611673,
      "grad_norm": 0.5166284441947937,
      "learning_rate": 2.9541932062693e-05,
      "loss": 0.06673908233642578,
      "step": 1056
    },
    {
      "epoch": 0.142899390784882,
      "grad_norm": 0.5847052931785583,
      "learning_rate": 2.954025201420134e-05,
      "loss": 0.06297552585601807,
      "step": 1057
    },
    {
      "epoch": 0.14303458415364725,
      "grad_norm": 1.0459882020950317,
      "learning_rate": 2.9538568938351672e-05,
      "loss": 0.061056531965732574,
      "step": 1058
    },
    {
      "epoch": 0.14316977752241253,
      "grad_norm": 0.5410304069519043,
      "learning_rate": 2.953688283549442e-05,
      "loss": 0.0664343312382698,
      "step": 1059
    },
    {
      "epoch": 0.14330497089117777,
      "grad_norm": 1.6860905885696411,
      "learning_rate": 2.9535193705980642e-05,
      "loss": 0.06649842858314514,
      "step": 1060
    },
    {
      "epoch": 0.14344016425994305,
      "grad_norm": 1.0052019357681274,
      "learning_rate": 2.9533501550162028e-05,
      "loss": 0.07814857363700867,
      "step": 1061
    },
    {
      "epoch": 0.14357535762870832,
      "grad_norm": 1.191643476486206,
      "learning_rate": 2.9531806368390882e-05,
      "loss": 0.08305811882019043,
      "step": 1062
    },
    {
      "epoch": 0.14371055099747357,
      "grad_norm": 0.4967811107635498,
      "learning_rate": 2.953010816102016e-05,
      "loss": 0.04764849692583084,
      "step": 1063
    },
    {
      "epoch": 0.14384574436623884,
      "grad_norm": 1.2741310596466064,
      "learning_rate": 2.952840692840343e-05,
      "loss": 0.0677926242351532,
      "step": 1064
    },
    {
      "epoch": 0.1439809377350041,
      "grad_norm": 0.5323997735977173,
      "learning_rate": 2.9526702670894914e-05,
      "loss": 0.06263640522956848,
      "step": 1065
    },
    {
      "epoch": 0.14411613110376936,
      "grad_norm": 0.45657527446746826,
      "learning_rate": 2.952499538884943e-05,
      "loss": 0.06672221422195435,
      "step": 1066
    },
    {
      "epoch": 0.1442513244725346,
      "grad_norm": 1.4525842666625977,
      "learning_rate": 2.9523285082622448e-05,
      "loss": 0.07497847080230713,
      "step": 1067
    },
    {
      "epoch": 0.14438651784129988,
      "grad_norm": 0.9516348838806152,
      "learning_rate": 2.9521571752570064e-05,
      "loss": 0.060328468680381775,
      "step": 1068
    },
    {
      "epoch": 0.14452171121006516,
      "grad_norm": 0.7086668014526367,
      "learning_rate": 2.9519855399049004e-05,
      "loss": 0.06583045423030853,
      "step": 1069
    },
    {
      "epoch": 0.1446569045788304,
      "grad_norm": 0.5770422220230103,
      "learning_rate": 2.951813602241662e-05,
      "loss": 0.06792578101158142,
      "step": 1070
    },
    {
      "epoch": 0.14479209794759568,
      "grad_norm": 2.717829704284668,
      "learning_rate": 2.9516413623030896e-05,
      "loss": 0.09775790572166443,
      "step": 1071
    },
    {
      "epoch": 0.14492729131636092,
      "grad_norm": 0.45390549302101135,
      "learning_rate": 2.951468820125045e-05,
      "loss": 0.06244179606437683,
      "step": 1072
    },
    {
      "epoch": 0.1450624846851262,
      "grad_norm": 0.7627196311950684,
      "learning_rate": 2.9512959757434508e-05,
      "loss": 0.056396666914224625,
      "step": 1073
    },
    {
      "epoch": 0.14519767805389144,
      "grad_norm": 0.975857138633728,
      "learning_rate": 2.951122829194296e-05,
      "loss": 0.07269716262817383,
      "step": 1074
    },
    {
      "epoch": 0.14533287142265672,
      "grad_norm": 0.3419383466243744,
      "learning_rate": 2.9509493805136296e-05,
      "loss": 0.05531579256057739,
      "step": 1075
    },
    {
      "epoch": 0.145468064791422,
      "grad_norm": 1.8316603899002075,
      "learning_rate": 2.9507756297375648e-05,
      "loss": 0.06720045953989029,
      "step": 1076
    },
    {
      "epoch": 0.14560325816018724,
      "grad_norm": 2.4128952026367188,
      "learning_rate": 2.9506015769022778e-05,
      "loss": 0.07311362773180008,
      "step": 1077
    },
    {
      "epoch": 0.1457384515289525,
      "grad_norm": 1.209806203842163,
      "learning_rate": 2.950427222044006e-05,
      "loss": 0.08212907612323761,
      "step": 1078
    },
    {
      "epoch": 0.14587364489771776,
      "grad_norm": 0.38941672444343567,
      "learning_rate": 2.9502525651990525e-05,
      "loss": 0.061763226985931396,
      "step": 1079
    },
    {
      "epoch": 0.14600883826648303,
      "grad_norm": 0.8612173795700073,
      "learning_rate": 2.9500776064037813e-05,
      "loss": 0.05280739814043045,
      "step": 1080
    },
    {
      "epoch": 0.14614403163524828,
      "grad_norm": 0.653526782989502,
      "learning_rate": 2.9499023456946194e-05,
      "loss": 0.05213388428092003,
      "step": 1081
    },
    {
      "epoch": 0.14627922500401355,
      "grad_norm": 0.7387879490852356,
      "learning_rate": 2.9497267831080575e-05,
      "loss": 0.06898682564496994,
      "step": 1082
    },
    {
      "epoch": 0.14641441837277883,
      "grad_norm": 0.5965262651443481,
      "learning_rate": 2.949550918680649e-05,
      "loss": 0.06924512982368469,
      "step": 1083
    },
    {
      "epoch": 0.14654961174154407,
      "grad_norm": 2.2757368087768555,
      "learning_rate": 2.9493747524490086e-05,
      "loss": 0.08496251702308655,
      "step": 1084
    },
    {
      "epoch": 0.14668480511030935,
      "grad_norm": 0.6187565326690674,
      "learning_rate": 2.9491982844498156e-05,
      "loss": 0.05621740221977234,
      "step": 1085
    },
    {
      "epoch": 0.1468199984790746,
      "grad_norm": 0.42957448959350586,
      "learning_rate": 2.949021514719812e-05,
      "loss": 0.04010415077209473,
      "step": 1086
    },
    {
      "epoch": 0.14695519184783987,
      "grad_norm": 1.1361998319625854,
      "learning_rate": 2.948844443295802e-05,
      "loss": 0.08248896896839142,
      "step": 1087
    },
    {
      "epoch": 0.1470903852166051,
      "grad_norm": 0.5923175811767578,
      "learning_rate": 2.9486670702146526e-05,
      "loss": 0.0988461971282959,
      "step": 1088
    },
    {
      "epoch": 0.1472255785853704,
      "grad_norm": 0.3613094985485077,
      "learning_rate": 2.948489395513294e-05,
      "loss": 0.05367446690797806,
      "step": 1089
    },
    {
      "epoch": 0.14736077195413566,
      "grad_norm": 0.9075596332550049,
      "learning_rate": 2.948311419228719e-05,
      "loss": 0.0606498122215271,
      "step": 1090
    },
    {
      "epoch": 0.1474959653229009,
      "grad_norm": 1.122167706489563,
      "learning_rate": 2.948133141397983e-05,
      "loss": 0.05617364123463631,
      "step": 1091
    },
    {
      "epoch": 0.14763115869166618,
      "grad_norm": 0.9702511429786682,
      "learning_rate": 2.9479545620582047e-05,
      "loss": 0.06455537676811218,
      "step": 1092
    },
    {
      "epoch": 0.14776635206043143,
      "grad_norm": 0.6575385928153992,
      "learning_rate": 2.9477756812465652e-05,
      "loss": 0.07123291492462158,
      "step": 1093
    },
    {
      "epoch": 0.1479015454291967,
      "grad_norm": 1.3090535402297974,
      "learning_rate": 2.9475964990003085e-05,
      "loss": 0.05015579238533974,
      "step": 1094
    },
    {
      "epoch": 0.14803673879796195,
      "grad_norm": 0.7073326706886292,
      "learning_rate": 2.9474170153567406e-05,
      "loss": 0.06102776527404785,
      "step": 1095
    },
    {
      "epoch": 0.14817193216672722,
      "grad_norm": 0.5905848741531372,
      "learning_rate": 2.947237230353232e-05,
      "loss": 0.07047636061906815,
      "step": 1096
    },
    {
      "epoch": 0.1483071255354925,
      "grad_norm": 1.980143666267395,
      "learning_rate": 2.9470571440272147e-05,
      "loss": 0.0791713297367096,
      "step": 1097
    },
    {
      "epoch": 0.14844231890425774,
      "grad_norm": 0.4122215807437897,
      "learning_rate": 2.946876756416183e-05,
      "loss": 0.04994921386241913,
      "step": 1098
    },
    {
      "epoch": 0.14857751227302302,
      "grad_norm": 0.716984212398529,
      "learning_rate": 2.946696067557695e-05,
      "loss": 0.06360213458538055,
      "step": 1099
    },
    {
      "epoch": 0.14871270564178826,
      "grad_norm": 1.2254761457443237,
      "learning_rate": 2.9465150774893706e-05,
      "loss": 0.06261074542999268,
      "step": 1100
    },
    {
      "epoch": 0.14884789901055354,
      "grad_norm": 0.25284162163734436,
      "learning_rate": 2.9463337862488938e-05,
      "loss": 0.055991947650909424,
      "step": 1101
    },
    {
      "epoch": 0.14898309237931878,
      "grad_norm": 0.4695529639720917,
      "learning_rate": 2.9461521938740096e-05,
      "loss": 0.06830859184265137,
      "step": 1102
    },
    {
      "epoch": 0.14911828574808406,
      "grad_norm": 1.6265945434570312,
      "learning_rate": 2.9459703004025273e-05,
      "loss": 0.06502760201692581,
      "step": 1103
    },
    {
      "epoch": 0.14925347911684933,
      "grad_norm": 1.1338313817977905,
      "learning_rate": 2.9457881058723174e-05,
      "loss": 0.06526631116867065,
      "step": 1104
    },
    {
      "epoch": 0.14938867248561458,
      "grad_norm": 0.8361053466796875,
      "learning_rate": 2.9456056103213137e-05,
      "loss": 0.04766133427619934,
      "step": 1105
    },
    {
      "epoch": 0.14952386585437985,
      "grad_norm": 1.4736934900283813,
      "learning_rate": 2.945422813787513e-05,
      "loss": 0.06487038731575012,
      "step": 1106
    },
    {
      "epoch": 0.1496590592231451,
      "grad_norm": 0.996687650680542,
      "learning_rate": 2.9452397163089748e-05,
      "loss": 0.0539180152118206,
      "step": 1107
    },
    {
      "epoch": 0.14979425259191037,
      "grad_norm": 0.9717349410057068,
      "learning_rate": 2.9450563179238207e-05,
      "loss": 0.05086350440979004,
      "step": 1108
    },
    {
      "epoch": 0.14992944596067562,
      "grad_norm": 1.2033283710479736,
      "learning_rate": 2.9448726186702354e-05,
      "loss": 0.05426774546504021,
      "step": 1109
    },
    {
      "epoch": 0.1500646393294409,
      "grad_norm": 0.32348763942718506,
      "learning_rate": 2.9446886185864652e-05,
      "loss": 0.05802512168884277,
      "step": 1110
    },
    {
      "epoch": 0.15019983269820616,
      "grad_norm": 0.8118970394134521,
      "learning_rate": 2.944504317710821e-05,
      "loss": 0.06699618697166443,
      "step": 1111
    },
    {
      "epoch": 0.1503350260669714,
      "grad_norm": 0.8924899697303772,
      "learning_rate": 2.944319716081675e-05,
      "loss": 0.06311993300914764,
      "step": 1112
    },
    {
      "epoch": 0.15047021943573669,
      "grad_norm": 0.5372419357299805,
      "learning_rate": 2.944134813737462e-05,
      "loss": 0.06188049167394638,
      "step": 1113
    },
    {
      "epoch": 0.15060541280450193,
      "grad_norm": 1.255583643913269,
      "learning_rate": 2.9439496107166796e-05,
      "loss": 0.09018206596374512,
      "step": 1114
    },
    {
      "epoch": 0.1507406061732672,
      "grad_norm": 2.200819492340088,
      "learning_rate": 2.943764107057888e-05,
      "loss": 0.06547680497169495,
      "step": 1115
    },
    {
      "epoch": 0.15087579954203245,
      "grad_norm": 1.4588055610656738,
      "learning_rate": 2.9435783027997106e-05,
      "loss": 0.07638072222471237,
      "step": 1116
    },
    {
      "epoch": 0.15101099291079773,
      "grad_norm": 1.3274680376052856,
      "learning_rate": 2.9433921979808323e-05,
      "loss": 0.06149342656135559,
      "step": 1117
    },
    {
      "epoch": 0.151146186279563,
      "grad_norm": 1.705327033996582,
      "learning_rate": 2.9432057926400014e-05,
      "loss": 0.08070620894432068,
      "step": 1118
    },
    {
      "epoch": 0.15128137964832825,
      "grad_norm": 0.4748374819755554,
      "learning_rate": 2.943019086816028e-05,
      "loss": 0.05764502286911011,
      "step": 1119
    },
    {
      "epoch": 0.15141657301709352,
      "grad_norm": 0.8783624768257141,
      "learning_rate": 2.9428320805477855e-05,
      "loss": 0.06809592247009277,
      "step": 1120
    },
    {
      "epoch": 0.15155176638585877,
      "grad_norm": 0.36441802978515625,
      "learning_rate": 2.9426447738742104e-05,
      "loss": 0.06841164827346802,
      "step": 1121
    },
    {
      "epoch": 0.15168695975462404,
      "grad_norm": 1.1301729679107666,
      "learning_rate": 2.9424571668343e-05,
      "loss": 0.04396027326583862,
      "step": 1122
    },
    {
      "epoch": 0.1518221531233893,
      "grad_norm": 2.087982416152954,
      "learning_rate": 2.942269259467115e-05,
      "loss": 0.0730208158493042,
      "step": 1123
    },
    {
      "epoch": 0.15195734649215456,
      "grad_norm": 0.44096827507019043,
      "learning_rate": 2.9420810518117794e-05,
      "loss": 0.05848512053489685,
      "step": 1124
    },
    {
      "epoch": 0.15209253986091983,
      "grad_norm": 1.1126229763031006,
      "learning_rate": 2.9418925439074784e-05,
      "loss": 0.0867539644241333,
      "step": 1125
    },
    {
      "epoch": 0.15222773322968508,
      "grad_norm": 2.0272955894470215,
      "learning_rate": 2.9417037357934606e-05,
      "loss": 0.07255285978317261,
      "step": 1126
    },
    {
      "epoch": 0.15236292659845035,
      "grad_norm": 1.182340383529663,
      "learning_rate": 2.9415146275090373e-05,
      "loss": 0.06449049711227417,
      "step": 1127
    },
    {
      "epoch": 0.1524981199672156,
      "grad_norm": 2.9131686687469482,
      "learning_rate": 2.9413252190935813e-05,
      "loss": 0.05471482872962952,
      "step": 1128
    },
    {
      "epoch": 0.15263331333598087,
      "grad_norm": 1.4113824367523193,
      "learning_rate": 2.9411355105865286e-05,
      "loss": 0.06527480483055115,
      "step": 1129
    },
    {
      "epoch": 0.15276850670474612,
      "grad_norm": 1.8772748708724976,
      "learning_rate": 2.9409455020273775e-05,
      "loss": 0.06518785655498505,
      "step": 1130
    },
    {
      "epoch": 0.1529037000735114,
      "grad_norm": 0.6438419818878174,
      "learning_rate": 2.940755193455689e-05,
      "loss": 0.05799649655818939,
      "step": 1131
    },
    {
      "epoch": 0.15303889344227667,
      "grad_norm": 0.5866788029670715,
      "learning_rate": 2.940564584911086e-05,
      "loss": 0.04870527982711792,
      "step": 1132
    },
    {
      "epoch": 0.15317408681104192,
      "grad_norm": 2.1813197135925293,
      "learning_rate": 2.9403736764332543e-05,
      "loss": 0.0656803548336029,
      "step": 1133
    },
    {
      "epoch": 0.1533092801798072,
      "grad_norm": 1.3716038465499878,
      "learning_rate": 2.9401824680619423e-05,
      "loss": 0.08586659282445908,
      "step": 1134
    },
    {
      "epoch": 0.15344447354857244,
      "grad_norm": 1.49837327003479,
      "learning_rate": 2.9399909598369604e-05,
      "loss": 0.07058757543563843,
      "step": 1135
    },
    {
      "epoch": 0.1535796669173377,
      "grad_norm": 0.5299723148345947,
      "learning_rate": 2.939799151798182e-05,
      "loss": 0.05751442909240723,
      "step": 1136
    },
    {
      "epoch": 0.15371486028610296,
      "grad_norm": 1.1080832481384277,
      "learning_rate": 2.9396070439855417e-05,
      "loss": 0.0840616226196289,
      "step": 1137
    },
    {
      "epoch": 0.15385005365486823,
      "grad_norm": 0.9604690670967102,
      "learning_rate": 2.9394146364390382e-05,
      "loss": 0.05214349925518036,
      "step": 1138
    },
    {
      "epoch": 0.1539852470236335,
      "grad_norm": 1.7251310348510742,
      "learning_rate": 2.9392219291987315e-05,
      "loss": 0.08107581734657288,
      "step": 1139
    },
    {
      "epoch": 0.15412044039239875,
      "grad_norm": 0.4901770353317261,
      "learning_rate": 2.939028922304744e-05,
      "loss": 0.0697246789932251,
      "step": 1140
    },
    {
      "epoch": 0.15425563376116402,
      "grad_norm": 0.36542144417762756,
      "learning_rate": 2.9388356157972615e-05,
      "loss": 0.0458989292383194,
      "step": 1141
    },
    {
      "epoch": 0.15439082712992927,
      "grad_norm": 1.7223154306411743,
      "learning_rate": 2.938642009716531e-05,
      "loss": 0.07267695665359497,
      "step": 1142
    },
    {
      "epoch": 0.15452602049869454,
      "grad_norm": 1.3905787467956543,
      "learning_rate": 2.938448104102862e-05,
      "loss": 0.08286727219820023,
      "step": 1143
    },
    {
      "epoch": 0.1546612138674598,
      "grad_norm": 0.7059480547904968,
      "learning_rate": 2.9382538989966267e-05,
      "loss": 0.05648154020309448,
      "step": 1144
    },
    {
      "epoch": 0.15479640723622506,
      "grad_norm": 1.1272727251052856,
      "learning_rate": 2.9380593944382605e-05,
      "loss": 0.0527062863111496,
      "step": 1145
    },
    {
      "epoch": 0.15493160060499034,
      "grad_norm": 0.5262485146522522,
      "learning_rate": 2.9378645904682596e-05,
      "loss": 0.05934404581785202,
      "step": 1146
    },
    {
      "epoch": 0.15506679397375558,
      "grad_norm": 0.7232634425163269,
      "learning_rate": 2.937669487127183e-05,
      "loss": 0.0660967230796814,
      "step": 1147
    },
    {
      "epoch": 0.15520198734252086,
      "grad_norm": 1.1460777521133423,
      "learning_rate": 2.9374740844556532e-05,
      "loss": 0.07225543260574341,
      "step": 1148
    },
    {
      "epoch": 0.1553371807112861,
      "grad_norm": 0.42113828659057617,
      "learning_rate": 2.937278382494353e-05,
      "loss": 0.04665200412273407,
      "step": 1149
    },
    {
      "epoch": 0.15547237408005138,
      "grad_norm": 0.37123903632164,
      "learning_rate": 2.9370823812840287e-05,
      "loss": 0.06203605234622955,
      "step": 1150
    },
    {
      "epoch": 0.15560756744881662,
      "grad_norm": 0.6488979458808899,
      "learning_rate": 2.93688608086549e-05,
      "loss": 0.059460125863552094,
      "step": 1151
    },
    {
      "epoch": 0.1557427608175819,
      "grad_norm": 1.1914242506027222,
      "learning_rate": 2.9366894812796064e-05,
      "loss": 0.057233214378356934,
      "step": 1152
    },
    {
      "epoch": 0.15587795418634717,
      "grad_norm": 0.7370941638946533,
      "learning_rate": 2.9364925825673117e-05,
      "loss": 0.06552474200725555,
      "step": 1153
    },
    {
      "epoch": 0.15601314755511242,
      "grad_norm": 0.6630440950393677,
      "learning_rate": 2.9362953847696006e-05,
      "loss": 0.07680156081914902,
      "step": 1154
    },
    {
      "epoch": 0.1561483409238777,
      "grad_norm": 0.8082736134529114,
      "learning_rate": 2.9360978879275313e-05,
      "loss": 0.06067699193954468,
      "step": 1155
    },
    {
      "epoch": 0.15628353429264294,
      "grad_norm": 1.5821971893310547,
      "learning_rate": 2.9359000920822237e-05,
      "loss": 0.05686017870903015,
      "step": 1156
    },
    {
      "epoch": 0.1564187276614082,
      "grad_norm": 1.0934772491455078,
      "learning_rate": 2.9357019972748594e-05,
      "loss": 0.08420317620038986,
      "step": 1157
    },
    {
      "epoch": 0.15655392103017346,
      "grad_norm": 0.9861279129981995,
      "learning_rate": 2.9355036035466836e-05,
      "loss": 0.07504907250404358,
      "step": 1158
    },
    {
      "epoch": 0.15668911439893873,
      "grad_norm": 0.45539483428001404,
      "learning_rate": 2.935304910939002e-05,
      "loss": 0.04558587074279785,
      "step": 1159
    },
    {
      "epoch": 0.156824307767704,
      "grad_norm": 0.658811092376709,
      "learning_rate": 2.935105919493184e-05,
      "loss": 0.055750906467437744,
      "step": 1160
    },
    {
      "epoch": 0.15695950113646925,
      "grad_norm": 0.8782719373703003,
      "learning_rate": 2.9349066292506613e-05,
      "loss": 0.047747790813446045,
      "step": 1161
    },
    {
      "epoch": 0.15709469450523453,
      "grad_norm": 1.0695005655288696,
      "learning_rate": 2.934707040252926e-05,
      "loss": 0.07084289193153381,
      "step": 1162
    },
    {
      "epoch": 0.15722988787399977,
      "grad_norm": 0.6481457352638245,
      "learning_rate": 2.9345071525415342e-05,
      "loss": 0.0865902528166771,
      "step": 1163
    },
    {
      "epoch": 0.15736508124276505,
      "grad_norm": 1.2633981704711914,
      "learning_rate": 2.9343069661581035e-05,
      "loss": 0.06998138874769211,
      "step": 1164
    },
    {
      "epoch": 0.1575002746115303,
      "grad_norm": 0.6916375756263733,
      "learning_rate": 2.9341064811443138e-05,
      "loss": 0.06468892097473145,
      "step": 1165
    },
    {
      "epoch": 0.15763546798029557,
      "grad_norm": 1.4526642560958862,
      "learning_rate": 2.9339056975419078e-05,
      "loss": 0.06961843371391296,
      "step": 1166
    },
    {
      "epoch": 0.15777066134906084,
      "grad_norm": 0.7845077514648438,
      "learning_rate": 2.9337046153926882e-05,
      "loss": 0.07602876424789429,
      "step": 1167
    },
    {
      "epoch": 0.1579058547178261,
      "grad_norm": 0.8499192595481873,
      "learning_rate": 2.9335032347385224e-05,
      "loss": 0.04190849885344505,
      "step": 1168
    },
    {
      "epoch": 0.15804104808659136,
      "grad_norm": 1.3594536781311035,
      "learning_rate": 2.933301555621339e-05,
      "loss": 0.04575201869010925,
      "step": 1169
    },
    {
      "epoch": 0.1581762414553566,
      "grad_norm": 0.6063716411590576,
      "learning_rate": 2.933099578083128e-05,
      "loss": 0.05104599893093109,
      "step": 1170
    },
    {
      "epoch": 0.15831143482412188,
      "grad_norm": 0.5349283814430237,
      "learning_rate": 2.932897302165943e-05,
      "loss": 0.05895107239484787,
      "step": 1171
    },
    {
      "epoch": 0.15844662819288713,
      "grad_norm": 0.3548339903354645,
      "learning_rate": 2.9326947279118983e-05,
      "loss": 0.051456570625305176,
      "step": 1172
    },
    {
      "epoch": 0.1585818215616524,
      "grad_norm": 0.6213728785514832,
      "learning_rate": 2.9324918553631716e-05,
      "loss": 0.061855852603912354,
      "step": 1173
    },
    {
      "epoch": 0.15871701493041765,
      "grad_norm": 0.5467653274536133,
      "learning_rate": 2.9322886845620013e-05,
      "loss": 0.03827083855867386,
      "step": 1174
    },
    {
      "epoch": 0.15885220829918292,
      "grad_norm": 0.5223851799964905,
      "learning_rate": 2.932085215550689e-05,
      "loss": 0.07667255401611328,
      "step": 1175
    },
    {
      "epoch": 0.1589874016679482,
      "grad_norm": 0.601033091545105,
      "learning_rate": 2.9318814483715982e-05,
      "loss": 0.05599546432495117,
      "step": 1176
    },
    {
      "epoch": 0.15912259503671344,
      "grad_norm": 1.7533951997756958,
      "learning_rate": 2.9316773830671537e-05,
      "loss": 0.05947062373161316,
      "step": 1177
    },
    {
      "epoch": 0.15925778840547872,
      "grad_norm": 0.3397280275821686,
      "learning_rate": 2.9314730196798437e-05,
      "loss": 0.05962613224983215,
      "step": 1178
    },
    {
      "epoch": 0.15939298177424396,
      "grad_norm": 0.8302063345909119,
      "learning_rate": 2.9312683582522178e-05,
      "loss": 0.07750225067138672,
      "step": 1179
    },
    {
      "epoch": 0.15952817514300924,
      "grad_norm": 0.6219926476478577,
      "learning_rate": 2.9310633988268868e-05,
      "loss": 0.06360742449760437,
      "step": 1180
    },
    {
      "epoch": 0.15966336851177448,
      "grad_norm": 1.2431893348693848,
      "learning_rate": 2.9308581414465246e-05,
      "loss": 0.047799646854400635,
      "step": 1181
    },
    {
      "epoch": 0.15979856188053976,
      "grad_norm": 0.7167524695396423,
      "learning_rate": 2.9306525861538674e-05,
      "loss": 0.07473218441009521,
      "step": 1182
    },
    {
      "epoch": 0.15993375524930503,
      "grad_norm": 1.344822645187378,
      "learning_rate": 2.9304467329917127e-05,
      "loss": 0.06404919177293777,
      "step": 1183
    },
    {
      "epoch": 0.16006894861807028,
      "grad_norm": 1.3459992408752441,
      "learning_rate": 2.9302405820029198e-05,
      "loss": 0.06698673218488693,
      "step": 1184
    },
    {
      "epoch": 0.16020414198683555,
      "grad_norm": 0.4024495482444763,
      "learning_rate": 2.9300341332304114e-05,
      "loss": 0.049604594707489014,
      "step": 1185
    },
    {
      "epoch": 0.1603393353556008,
      "grad_norm": 0.6731967329978943,
      "learning_rate": 2.9298273867171697e-05,
      "loss": 0.06139141321182251,
      "step": 1186
    },
    {
      "epoch": 0.16047452872436607,
      "grad_norm": 1.0775129795074463,
      "learning_rate": 2.929620342506242e-05,
      "loss": 0.07561543583869934,
      "step": 1187
    },
    {
      "epoch": 0.16060972209313132,
      "grad_norm": 1.8730332851409912,
      "learning_rate": 2.929413000640735e-05,
      "loss": 0.0670536607503891,
      "step": 1188
    },
    {
      "epoch": 0.1607449154618966,
      "grad_norm": 1.2282514572143555,
      "learning_rate": 2.9292053611638187e-05,
      "loss": 0.05790209770202637,
      "step": 1189
    },
    {
      "epoch": 0.16088010883066187,
      "grad_norm": 1.6782622337341309,
      "learning_rate": 2.928997424118725e-05,
      "loss": 0.0759919285774231,
      "step": 1190
    },
    {
      "epoch": 0.1610153021994271,
      "grad_norm": 1.298192024230957,
      "learning_rate": 2.928789189548747e-05,
      "loss": 0.0586373507976532,
      "step": 1191
    },
    {
      "epoch": 0.1611504955681924,
      "grad_norm": 2.9043021202087402,
      "learning_rate": 2.9285806574972405e-05,
      "loss": 0.08343648910522461,
      "step": 1192
    },
    {
      "epoch": 0.16128568893695763,
      "grad_norm": 0.8272789120674133,
      "learning_rate": 2.928371828007623e-05,
      "loss": 0.05761042982339859,
      "step": 1193
    },
    {
      "epoch": 0.1614208823057229,
      "grad_norm": 4.33075475692749,
      "learning_rate": 2.928162701123374e-05,
      "loss": 0.09251654148101807,
      "step": 1194
    },
    {
      "epoch": 0.16155607567448815,
      "grad_norm": 0.8157700896263123,
      "learning_rate": 2.9279532768880345e-05,
      "loss": 0.0797078013420105,
      "step": 1195
    },
    {
      "epoch": 0.16169126904325343,
      "grad_norm": 1.2659385204315186,
      "learning_rate": 2.9277435553452084e-05,
      "loss": 0.08196496963500977,
      "step": 1196
    },
    {
      "epoch": 0.1618264624120187,
      "grad_norm": 1.4740269184112549,
      "learning_rate": 2.9275335365385602e-05,
      "loss": 0.06738023459911346,
      "step": 1197
    },
    {
      "epoch": 0.16196165578078395,
      "grad_norm": 0.792144775390625,
      "learning_rate": 2.927323220511817e-05,
      "loss": 0.0665399506688118,
      "step": 1198
    },
    {
      "epoch": 0.16209684914954922,
      "grad_norm": 0.5530847907066345,
      "learning_rate": 2.9271126073087684e-05,
      "loss": 0.05725696682929993,
      "step": 1199
    },
    {
      "epoch": 0.16223204251831447,
      "grad_norm": 0.64042067527771,
      "learning_rate": 2.926901696973264e-05,
      "loss": 0.06065577268600464,
      "step": 1200
    },
    {
      "epoch": 0.16236723588707974,
      "grad_norm": 1.6298916339874268,
      "learning_rate": 2.9266904895492177e-05,
      "loss": 0.05311492830514908,
      "step": 1201
    },
    {
      "epoch": 0.162502429255845,
      "grad_norm": 1.9529551267623901,
      "learning_rate": 2.926478985080603e-05,
      "loss": 0.06378045678138733,
      "step": 1202
    },
    {
      "epoch": 0.16263762262461026,
      "grad_norm": 1.2222492694854736,
      "learning_rate": 2.9262671836114568e-05,
      "loss": 0.05665773153305054,
      "step": 1203
    },
    {
      "epoch": 0.16277281599337554,
      "grad_norm": 0.7065329551696777,
      "learning_rate": 2.9260550851858774e-05,
      "loss": 0.05856727808713913,
      "step": 1204
    },
    {
      "epoch": 0.16290800936214078,
      "grad_norm": 0.5530824661254883,
      "learning_rate": 2.9258426898480243e-05,
      "loss": 0.06612825393676758,
      "step": 1205
    },
    {
      "epoch": 0.16304320273090606,
      "grad_norm": 1.5942386388778687,
      "learning_rate": 2.9256299976421198e-05,
      "loss": 0.06233048439025879,
      "step": 1206
    },
    {
      "epoch": 0.1631783960996713,
      "grad_norm": 0.7886584997177124,
      "learning_rate": 2.9254170086124474e-05,
      "loss": 0.08418315649032593,
      "step": 1207
    },
    {
      "epoch": 0.16331358946843658,
      "grad_norm": 0.8411967754364014,
      "learning_rate": 2.9252037228033526e-05,
      "loss": 0.07825121283531189,
      "step": 1208
    },
    {
      "epoch": 0.16344878283720182,
      "grad_norm": 0.752139687538147,
      "learning_rate": 2.9249901402592424e-05,
      "loss": 0.04684019088745117,
      "step": 1209
    },
    {
      "epoch": 0.1635839762059671,
      "grad_norm": 0.592678964138031,
      "learning_rate": 2.9247762610245863e-05,
      "loss": 0.0651087760925293,
      "step": 1210
    },
    {
      "epoch": 0.16371916957473237,
      "grad_norm": 1.1372289657592773,
      "learning_rate": 2.9245620851439146e-05,
      "loss": 0.06731021404266357,
      "step": 1211
    },
    {
      "epoch": 0.16385436294349762,
      "grad_norm": 0.7190410494804382,
      "learning_rate": 2.92434761266182e-05,
      "loss": 0.06134688854217529,
      "step": 1212
    },
    {
      "epoch": 0.1639895563122629,
      "grad_norm": 0.6636428236961365,
      "learning_rate": 2.924132843622957e-05,
      "loss": 0.04543939232826233,
      "step": 1213
    },
    {
      "epoch": 0.16412474968102814,
      "grad_norm": 0.8182775974273682,
      "learning_rate": 2.9239177780720418e-05,
      "loss": 0.056656286120414734,
      "step": 1214
    },
    {
      "epoch": 0.1642599430497934,
      "grad_norm": 0.709373414516449,
      "learning_rate": 2.923702416053852e-05,
      "loss": 0.07238680124282837,
      "step": 1215
    },
    {
      "epoch": 0.16439513641855866,
      "grad_norm": 0.6388227939605713,
      "learning_rate": 2.9234867576132268e-05,
      "loss": 0.0633966475725174,
      "step": 1216
    },
    {
      "epoch": 0.16453032978732393,
      "grad_norm": 0.5293400883674622,
      "learning_rate": 2.923270802795068e-05,
      "loss": 0.06284171342849731,
      "step": 1217
    },
    {
      "epoch": 0.1646655231560892,
      "grad_norm": 1.3790321350097656,
      "learning_rate": 2.9230545516443378e-05,
      "loss": 0.07694780826568604,
      "step": 1218
    },
    {
      "epoch": 0.16480071652485445,
      "grad_norm": 0.5367887020111084,
      "learning_rate": 2.9228380042060615e-05,
      "loss": 0.05496096611022949,
      "step": 1219
    },
    {
      "epoch": 0.16493590989361973,
      "grad_norm": 0.3577832281589508,
      "learning_rate": 2.9226211605253252e-05,
      "loss": 0.06105245277285576,
      "step": 1220
    },
    {
      "epoch": 0.16507110326238497,
      "grad_norm": 0.7726598978042603,
      "learning_rate": 2.922404020647277e-05,
      "loss": 0.05503867194056511,
      "step": 1221
    },
    {
      "epoch": 0.16520629663115025,
      "grad_norm": 0.6038638353347778,
      "learning_rate": 2.9221865846171264e-05,
      "loss": 0.05856103450059891,
      "step": 1222
    },
    {
      "epoch": 0.1653414899999155,
      "grad_norm": 1.0350080728530884,
      "learning_rate": 2.9219688524801446e-05,
      "loss": 0.05294673144817352,
      "step": 1223
    },
    {
      "epoch": 0.16547668336868077,
      "grad_norm": 1.02299964427948,
      "learning_rate": 2.9217508242816653e-05,
      "loss": 0.06431066989898682,
      "step": 1224
    },
    {
      "epoch": 0.16561187673744604,
      "grad_norm": 0.2556920051574707,
      "learning_rate": 2.921532500067083e-05,
      "loss": 0.04862751066684723,
      "step": 1225
    },
    {
      "epoch": 0.16574707010621129,
      "grad_norm": 0.6257906556129456,
      "learning_rate": 2.9213138798818528e-05,
      "loss": 0.04743689298629761,
      "step": 1226
    },
    {
      "epoch": 0.16588226347497656,
      "grad_norm": 0.8982285857200623,
      "learning_rate": 2.921094963771494e-05,
      "loss": 0.045286282896995544,
      "step": 1227
    },
    {
      "epoch": 0.1660174568437418,
      "grad_norm": 0.5756014585494995,
      "learning_rate": 2.9208757517815855e-05,
      "loss": 0.08197182416915894,
      "step": 1228
    },
    {
      "epoch": 0.16615265021250708,
      "grad_norm": 1.370456576347351,
      "learning_rate": 2.9206562439577684e-05,
      "loss": 0.05535238981246948,
      "step": 1229
    },
    {
      "epoch": 0.16628784358127233,
      "grad_norm": 2.5601351261138916,
      "learning_rate": 2.9204364403457452e-05,
      "loss": 0.09890396147966385,
      "step": 1230
    },
    {
      "epoch": 0.1664230369500376,
      "grad_norm": 0.9468607306480408,
      "learning_rate": 2.9202163409912808e-05,
      "loss": 0.07507592439651489,
      "step": 1231
    },
    {
      "epoch": 0.16655823031880287,
      "grad_norm": 0.7571756839752197,
      "learning_rate": 2.9199959459402003e-05,
      "loss": 0.04222369194030762,
      "step": 1232
    },
    {
      "epoch": 0.16669342368756812,
      "grad_norm": 2.2120721340179443,
      "learning_rate": 2.919775255238392e-05,
      "loss": 0.07921065390110016,
      "step": 1233
    },
    {
      "epoch": 0.1668286170563334,
      "grad_norm": 1.0212641954421997,
      "learning_rate": 2.919554268931804e-05,
      "loss": 0.05951683595776558,
      "step": 1234
    },
    {
      "epoch": 0.16696381042509864,
      "grad_norm": 0.497661828994751,
      "learning_rate": 2.9193329870664475e-05,
      "loss": 0.06232532858848572,
      "step": 1235
    },
    {
      "epoch": 0.16709900379386392,
      "grad_norm": 0.5672320127487183,
      "learning_rate": 2.9191114096883938e-05,
      "loss": 0.05264327675104141,
      "step": 1236
    },
    {
      "epoch": 0.16723419716262916,
      "grad_norm": 0.7996569871902466,
      "learning_rate": 2.9188895368437774e-05,
      "loss": 0.07075543701648712,
      "step": 1237
    },
    {
      "epoch": 0.16736939053139444,
      "grad_norm": 1.8040475845336914,
      "learning_rate": 2.9186673685787926e-05,
      "loss": 0.05779653787612915,
      "step": 1238
    },
    {
      "epoch": 0.1675045839001597,
      "grad_norm": 0.94858318567276,
      "learning_rate": 2.918444904939697e-05,
      "loss": 0.07351627945899963,
      "step": 1239
    },
    {
      "epoch": 0.16763977726892496,
      "grad_norm": 0.5381576418876648,
      "learning_rate": 2.9182221459728078e-05,
      "loss": 0.053827933967113495,
      "step": 1240
    },
    {
      "epoch": 0.16777497063769023,
      "grad_norm": 1.689529299736023,
      "learning_rate": 2.917999091724505e-05,
      "loss": 0.07044818997383118,
      "step": 1241
    },
    {
      "epoch": 0.16791016400645548,
      "grad_norm": 1.0230121612548828,
      "learning_rate": 2.9177757422412294e-05,
      "loss": 0.07499171793460846,
      "step": 1242
    },
    {
      "epoch": 0.16804535737522075,
      "grad_norm": 1.2331953048706055,
      "learning_rate": 2.917552097569484e-05,
      "loss": 0.05908238887786865,
      "step": 1243
    },
    {
      "epoch": 0.168180550743986,
      "grad_norm": 2.16436767578125,
      "learning_rate": 2.917328157755832e-05,
      "loss": 0.08111009001731873,
      "step": 1244
    },
    {
      "epoch": 0.16831574411275127,
      "grad_norm": 0.577461302280426,
      "learning_rate": 2.9171039228469003e-05,
      "loss": 0.05244028568267822,
      "step": 1245
    },
    {
      "epoch": 0.16845093748151654,
      "grad_norm": 0.6482664942741394,
      "learning_rate": 2.9168793928893747e-05,
      "loss": 0.06110365688800812,
      "step": 1246
    },
    {
      "epoch": 0.1685861308502818,
      "grad_norm": 0.887102484703064,
      "learning_rate": 2.9166545679300036e-05,
      "loss": 0.07000339031219482,
      "step": 1247
    },
    {
      "epoch": 0.16872132421904706,
      "grad_norm": 1.3720703125,
      "learning_rate": 2.9164294480155966e-05,
      "loss": 0.07628023624420166,
      "step": 1248
    },
    {
      "epoch": 0.1688565175878123,
      "grad_norm": 0.4543033242225647,
      "learning_rate": 2.9162040331930256e-05,
      "loss": 0.05563601851463318,
      "step": 1249
    },
    {
      "epoch": 0.16899171095657758,
      "grad_norm": 0.9618221521377563,
      "learning_rate": 2.915978323509223e-05,
      "loss": 0.0719805359840393,
      "step": 1250
    },
    {
      "epoch": 0.16912690432534283,
      "grad_norm": 0.8158279061317444,
      "learning_rate": 2.915752319011182e-05,
      "loss": 0.07083217054605484,
      "step": 1251
    },
    {
      "epoch": 0.1692620976941081,
      "grad_norm": 0.6591756939888,
      "learning_rate": 2.9155260197459588e-05,
      "loss": 0.07081496715545654,
      "step": 1252
    },
    {
      "epoch": 0.16939729106287338,
      "grad_norm": 1.1029133796691895,
      "learning_rate": 2.91529942576067e-05,
      "loss": 0.06827642023563385,
      "step": 1253
    },
    {
      "epoch": 0.16953248443163862,
      "grad_norm": 0.8691407442092896,
      "learning_rate": 2.915072537102493e-05,
      "loss": 0.05868637561798096,
      "step": 1254
    },
    {
      "epoch": 0.1696676778004039,
      "grad_norm": 0.8354164958000183,
      "learning_rate": 2.914845353818668e-05,
      "loss": 0.05098971724510193,
      "step": 1255
    },
    {
      "epoch": 0.16980287116916915,
      "grad_norm": 0.6224926710128784,
      "learning_rate": 2.9146178759564953e-05,
      "loss": 0.07958734035491943,
      "step": 1256
    },
    {
      "epoch": 0.16993806453793442,
      "grad_norm": 0.8625409007072449,
      "learning_rate": 2.914390103563337e-05,
      "loss": 0.07424217462539673,
      "step": 1257
    },
    {
      "epoch": 0.17007325790669967,
      "grad_norm": 1.2023978233337402,
      "learning_rate": 2.914162036686617e-05,
      "loss": 0.06286744773387909,
      "step": 1258
    },
    {
      "epoch": 0.17020845127546494,
      "grad_norm": 0.9498358964920044,
      "learning_rate": 2.9139336753738196e-05,
      "loss": 0.05720394849777222,
      "step": 1259
    },
    {
      "epoch": 0.1703436446442302,
      "grad_norm": 0.5057894587516785,
      "learning_rate": 2.913705019672491e-05,
      "loss": 0.06330971419811249,
      "step": 1260
    },
    {
      "epoch": 0.17047883801299546,
      "grad_norm": 2.2030510902404785,
      "learning_rate": 2.9134760696302386e-05,
      "loss": 0.05698040500283241,
      "step": 1261
    },
    {
      "epoch": 0.17061403138176073,
      "grad_norm": 0.359602153301239,
      "learning_rate": 2.9132468252947306e-05,
      "loss": 0.05025045573711395,
      "step": 1262
    },
    {
      "epoch": 0.17074922475052598,
      "grad_norm": 1.3271559476852417,
      "learning_rate": 2.9130172867136974e-05,
      "loss": 0.06531503796577454,
      "step": 1263
    },
    {
      "epoch": 0.17088441811929125,
      "grad_norm": 0.3179835081100464,
      "learning_rate": 2.91278745393493e-05,
      "loss": 0.03917551040649414,
      "step": 1264
    },
    {
      "epoch": 0.1710196114880565,
      "grad_norm": 0.5240478515625,
      "learning_rate": 2.9125573270062812e-05,
      "loss": 0.07011768221855164,
      "step": 1265
    },
    {
      "epoch": 0.17115480485682177,
      "grad_norm": 0.5413030385971069,
      "learning_rate": 2.9123269059756634e-05,
      "loss": 0.09455788135528564,
      "step": 1266
    },
    {
      "epoch": 0.17128999822558705,
      "grad_norm": 0.47013336420059204,
      "learning_rate": 2.9120961908910528e-05,
      "loss": 0.0679527074098587,
      "step": 1267
    },
    {
      "epoch": 0.1714251915943523,
      "grad_norm": 0.9408896565437317,
      "learning_rate": 2.911865181800485e-05,
      "loss": 0.06111685186624527,
      "step": 1268
    },
    {
      "epoch": 0.17156038496311757,
      "grad_norm": 0.47430142760276794,
      "learning_rate": 2.9116338787520577e-05,
      "loss": 0.07369215786457062,
      "step": 1269
    },
    {
      "epoch": 0.17169557833188281,
      "grad_norm": 1.098947286605835,
      "learning_rate": 2.9114022817939283e-05,
      "loss": 0.05156645178794861,
      "step": 1270
    },
    {
      "epoch": 0.1718307717006481,
      "grad_norm": 1.2353764772415161,
      "learning_rate": 2.911170390974318e-05,
      "loss": 0.07163624465465546,
      "step": 1271
    },
    {
      "epoch": 0.17196596506941333,
      "grad_norm": 1.6305971145629883,
      "learning_rate": 2.9109382063415067e-05,
      "loss": 0.06071142852306366,
      "step": 1272
    },
    {
      "epoch": 0.1721011584381786,
      "grad_norm": 1.8492131233215332,
      "learning_rate": 2.9107057279438372e-05,
      "loss": 0.06312686949968338,
      "step": 1273
    },
    {
      "epoch": 0.17223635180694388,
      "grad_norm": 0.71107417345047,
      "learning_rate": 2.910472955829712e-05,
      "loss": 0.06514546275138855,
      "step": 1274
    },
    {
      "epoch": 0.17237154517570913,
      "grad_norm": 0.898524284362793,
      "learning_rate": 2.9102398900475958e-05,
      "loss": 0.06705622375011444,
      "step": 1275
    },
    {
      "epoch": 0.1725067385444744,
      "grad_norm": 0.2860540449619293,
      "learning_rate": 2.910006530646014e-05,
      "loss": 0.051441192626953125,
      "step": 1276
    },
    {
      "epoch": 0.17264193191323965,
      "grad_norm": 0.9001814723014832,
      "learning_rate": 2.909772877673554e-05,
      "loss": 0.07040204852819443,
      "step": 1277
    },
    {
      "epoch": 0.17277712528200492,
      "grad_norm": 0.3261363208293915,
      "learning_rate": 2.9095389311788626e-05,
      "loss": 0.06222736835479736,
      "step": 1278
    },
    {
      "epoch": 0.17291231865077017,
      "grad_norm": 1.1822023391723633,
      "learning_rate": 2.9093046912106494e-05,
      "loss": 0.08206778764724731,
      "step": 1279
    },
    {
      "epoch": 0.17304751201953544,
      "grad_norm": 0.6466618776321411,
      "learning_rate": 2.909070157817684e-05,
      "loss": 0.0743969976902008,
      "step": 1280
    },
    {
      "epoch": 0.17318270538830072,
      "grad_norm": 0.6141491532325745,
      "learning_rate": 2.9088353310487976e-05,
      "loss": 0.04727473109960556,
      "step": 1281
    },
    {
      "epoch": 0.17331789875706596,
      "grad_norm": 1.0952966213226318,
      "learning_rate": 2.9086002109528825e-05,
      "loss": 0.04447062313556671,
      "step": 1282
    },
    {
      "epoch": 0.17345309212583124,
      "grad_norm": 0.9889422655105591,
      "learning_rate": 2.908364797578892e-05,
      "loss": 0.056224673986434937,
      "step": 1283
    },
    {
      "epoch": 0.17358828549459648,
      "grad_norm": 0.2930924594402313,
      "learning_rate": 2.9081290909758405e-05,
      "loss": 0.04904172569513321,
      "step": 1284
    },
    {
      "epoch": 0.17372347886336176,
      "grad_norm": 1.41915762424469,
      "learning_rate": 2.9078930911928033e-05,
      "loss": 0.08801735937595367,
      "step": 1285
    },
    {
      "epoch": 0.173858672232127,
      "grad_norm": 1.8939464092254639,
      "learning_rate": 2.907656798278916e-05,
      "loss": 0.07156491279602051,
      "step": 1286
    },
    {
      "epoch": 0.17399386560089228,
      "grad_norm": 0.5577313303947449,
      "learning_rate": 2.9074202122833773e-05,
      "loss": 0.07089945673942566,
      "step": 1287
    },
    {
      "epoch": 0.17412905896965755,
      "grad_norm": 1.4324188232421875,
      "learning_rate": 2.907183333255445e-05,
      "loss": 0.06999829411506653,
      "step": 1288
    },
    {
      "epoch": 0.1742642523384228,
      "grad_norm": 0.33485084772109985,
      "learning_rate": 2.9069461612444384e-05,
      "loss": 0.0712369829416275,
      "step": 1289
    },
    {
      "epoch": 0.17439944570718807,
      "grad_norm": 0.3693605065345764,
      "learning_rate": 2.9067086962997385e-05,
      "loss": 0.07062338292598724,
      "step": 1290
    },
    {
      "epoch": 0.17453463907595332,
      "grad_norm": 0.7931060791015625,
      "learning_rate": 2.9064709384707868e-05,
      "loss": 0.0733291506767273,
      "step": 1291
    },
    {
      "epoch": 0.1746698324447186,
      "grad_norm": 0.7573596239089966,
      "learning_rate": 2.9062328878070855e-05,
      "loss": 0.06712360680103302,
      "step": 1292
    },
    {
      "epoch": 0.17480502581348384,
      "grad_norm": 1.6741069555282593,
      "learning_rate": 2.905994544358198e-05,
      "loss": 0.057468071579933167,
      "step": 1293
    },
    {
      "epoch": 0.1749402191822491,
      "grad_norm": 0.9005102515220642,
      "learning_rate": 2.9057559081737482e-05,
      "loss": 0.06759119033813477,
      "step": 1294
    },
    {
      "epoch": 0.1750754125510144,
      "grad_norm": 0.9607921838760376,
      "learning_rate": 2.9055169793034225e-05,
      "loss": 0.05060403048992157,
      "step": 1295
    },
    {
      "epoch": 0.17521060591977963,
      "grad_norm": 0.49658510088920593,
      "learning_rate": 2.9052777577969656e-05,
      "loss": 0.0803871750831604,
      "step": 1296
    },
    {
      "epoch": 0.1753457992885449,
      "grad_norm": 0.31053662300109863,
      "learning_rate": 2.9050382437041868e-05,
      "loss": 0.046495724469423294,
      "step": 1297
    },
    {
      "epoch": 0.17548099265731015,
      "grad_norm": 0.6438810229301453,
      "learning_rate": 2.9047984370749526e-05,
      "loss": 0.0692463219165802,
      "step": 1298
    },
    {
      "epoch": 0.17561618602607543,
      "grad_norm": 0.2605363130569458,
      "learning_rate": 2.9045583379591925e-05,
      "loss": 0.055267587304115295,
      "step": 1299
    },
    {
      "epoch": 0.17575137939484067,
      "grad_norm": 0.3073757588863373,
      "learning_rate": 2.9043179464068965e-05,
      "loss": 0.057579156011343,
      "step": 1300
    },
    {
      "epoch": 0.17588657276360595,
      "grad_norm": 1.2076513767242432,
      "learning_rate": 2.9040772624681152e-05,
      "loss": 0.08651989698410034,
      "step": 1301
    },
    {
      "epoch": 0.1760217661323712,
      "grad_norm": 0.597019612789154,
      "learning_rate": 2.9038362861929603e-05,
      "loss": 0.06141549348831177,
      "step": 1302
    },
    {
      "epoch": 0.17615695950113647,
      "grad_norm": 0.6154230237007141,
      "learning_rate": 2.903595017631605e-05,
      "loss": 0.0814509391784668,
      "step": 1303
    },
    {
      "epoch": 0.17629215286990174,
      "grad_norm": 0.9492348432540894,
      "learning_rate": 2.903353456834282e-05,
      "loss": 0.07166945934295654,
      "step": 1304
    },
    {
      "epoch": 0.176427346238667,
      "grad_norm": 1.2229962348937988,
      "learning_rate": 2.903111603851285e-05,
      "loss": 0.0598774254322052,
      "step": 1305
    },
    {
      "epoch": 0.17656253960743226,
      "grad_norm": 0.944716215133667,
      "learning_rate": 2.9028694587329704e-05,
      "loss": 0.061225906014442444,
      "step": 1306
    },
    {
      "epoch": 0.1766977329761975,
      "grad_norm": 0.7197076678276062,
      "learning_rate": 2.902627021529753e-05,
      "loss": 0.05019553005695343,
      "step": 1307
    },
    {
      "epoch": 0.17683292634496278,
      "grad_norm": 0.9098347425460815,
      "learning_rate": 2.9023842922921105e-05,
      "loss": 0.07456699013710022,
      "step": 1308
    },
    {
      "epoch": 0.17696811971372803,
      "grad_norm": 2.485454559326172,
      "learning_rate": 2.90214127107058e-05,
      "loss": 0.07402080297470093,
      "step": 1309
    },
    {
      "epoch": 0.1771033130824933,
      "grad_norm": 1.4531241655349731,
      "learning_rate": 2.9018979579157592e-05,
      "loss": 0.07091991603374481,
      "step": 1310
    },
    {
      "epoch": 0.17723850645125858,
      "grad_norm": 0.4783823788166046,
      "learning_rate": 2.901654352878308e-05,
      "loss": 0.0479360967874527,
      "step": 1311
    },
    {
      "epoch": 0.17737369982002382,
      "grad_norm": 0.5426583886146545,
      "learning_rate": 2.9014104560089462e-05,
      "loss": 0.043479904532432556,
      "step": 1312
    },
    {
      "epoch": 0.1775088931887891,
      "grad_norm": 0.6178447604179382,
      "learning_rate": 2.9011662673584538e-05,
      "loss": 0.08286845684051514,
      "step": 1313
    },
    {
      "epoch": 0.17764408655755434,
      "grad_norm": 1.829768180847168,
      "learning_rate": 2.900921786977673e-05,
      "loss": 0.049149930477142334,
      "step": 1314
    },
    {
      "epoch": 0.17777927992631962,
      "grad_norm": 0.8166171312332153,
      "learning_rate": 2.900677014917505e-05,
      "loss": 0.07144433259963989,
      "step": 1315
    },
    {
      "epoch": 0.17791447329508486,
      "grad_norm": 0.9345900416374207,
      "learning_rate": 2.9004319512289136e-05,
      "loss": 0.05137139558792114,
      "step": 1316
    },
    {
      "epoch": 0.17804966666385014,
      "grad_norm": 0.4796876013278961,
      "learning_rate": 2.9001865959629222e-05,
      "loss": 0.04606900364160538,
      "step": 1317
    },
    {
      "epoch": 0.1781848600326154,
      "grad_norm": 0.7798415422439575,
      "learning_rate": 2.8999409491706143e-05,
      "loss": 0.05695191025733948,
      "step": 1318
    },
    {
      "epoch": 0.17832005340138066,
      "grad_norm": 0.9366718530654907,
      "learning_rate": 2.8996950109031355e-05,
      "loss": 0.07336622476577759,
      "step": 1319
    },
    {
      "epoch": 0.17845524677014593,
      "grad_norm": 1.6822129487991333,
      "learning_rate": 2.8994487812116917e-05,
      "loss": 0.08025699853897095,
      "step": 1320
    },
    {
      "epoch": 0.17859044013891118,
      "grad_norm": 0.8868669271469116,
      "learning_rate": 2.8992022601475483e-05,
      "loss": 0.09152978658676147,
      "step": 1321
    },
    {
      "epoch": 0.17872563350767645,
      "grad_norm": 0.9628618359565735,
      "learning_rate": 2.8989554477620332e-05,
      "loss": 0.0634050965309143,
      "step": 1322
    },
    {
      "epoch": 0.1788608268764417,
      "grad_norm": 0.36020126938819885,
      "learning_rate": 2.8987083441065335e-05,
      "loss": 0.05945882201194763,
      "step": 1323
    },
    {
      "epoch": 0.17899602024520697,
      "grad_norm": 1.1709458827972412,
      "learning_rate": 2.8984609492324983e-05,
      "loss": 0.07281702756881714,
      "step": 1324
    },
    {
      "epoch": 0.17913121361397225,
      "grad_norm": 0.5962613821029663,
      "learning_rate": 2.8982132631914357e-05,
      "loss": 0.06544731557369232,
      "step": 1325
    },
    {
      "epoch": 0.1792664069827375,
      "grad_norm": 1.8113411664962769,
      "learning_rate": 2.8979652860349154e-05,
      "loss": 0.06583040952682495,
      "step": 1326
    },
    {
      "epoch": 0.17940160035150277,
      "grad_norm": 1.1780592203140259,
      "learning_rate": 2.8977170178145675e-05,
      "loss": 0.06304009258747101,
      "step": 1327
    },
    {
      "epoch": 0.179536793720268,
      "grad_norm": 1.8488304615020752,
      "learning_rate": 2.8974684585820833e-05,
      "loss": 0.0766972005367279,
      "step": 1328
    },
    {
      "epoch": 0.17967198708903329,
      "grad_norm": 0.4457748234272003,
      "learning_rate": 2.8972196083892138e-05,
      "loss": 0.05921986699104309,
      "step": 1329
    },
    {
      "epoch": 0.17980718045779853,
      "grad_norm": 0.4643270671367645,
      "learning_rate": 2.8969704672877707e-05,
      "loss": 0.07630157470703125,
      "step": 1330
    },
    {
      "epoch": 0.1799423738265638,
      "grad_norm": 0.3910136818885803,
      "learning_rate": 2.896721035329627e-05,
      "loss": 0.05909073352813721,
      "step": 1331
    },
    {
      "epoch": 0.18007756719532908,
      "grad_norm": 1.9842510223388672,
      "learning_rate": 2.8964713125667153e-05,
      "loss": 0.06459778547286987,
      "step": 1332
    },
    {
      "epoch": 0.18021276056409433,
      "grad_norm": 1.379435658454895,
      "learning_rate": 2.8962212990510294e-05,
      "loss": 0.07560950517654419,
      "step": 1333
    },
    {
      "epoch": 0.1803479539328596,
      "grad_norm": 1.6055450439453125,
      "learning_rate": 2.8959709948346237e-05,
      "loss": 0.05632704496383667,
      "step": 1334
    },
    {
      "epoch": 0.18048314730162485,
      "grad_norm": 0.4815007448196411,
      "learning_rate": 2.8957203999696124e-05,
      "loss": 0.06901296973228455,
      "step": 1335
    },
    {
      "epoch": 0.18061834067039012,
      "grad_norm": 0.3896736800670624,
      "learning_rate": 2.8954695145081713e-05,
      "loss": 0.053732067346572876,
      "step": 1336
    },
    {
      "epoch": 0.18075353403915537,
      "grad_norm": 1.5841916799545288,
      "learning_rate": 2.8952183385025356e-05,
      "loss": 0.06032158434391022,
      "step": 1337
    },
    {
      "epoch": 0.18088872740792064,
      "grad_norm": 0.43961912393569946,
      "learning_rate": 2.8949668720050014e-05,
      "loss": 0.058547914028167725,
      "step": 1338
    },
    {
      "epoch": 0.18102392077668591,
      "grad_norm": 0.8505260348320007,
      "learning_rate": 2.8947151150679256e-05,
      "loss": 0.059689223766326904,
      "step": 1339
    },
    {
      "epoch": 0.18115911414545116,
      "grad_norm": 0.6629796624183655,
      "learning_rate": 2.8944630677437255e-05,
      "loss": 0.07312893867492676,
      "step": 1340
    },
    {
      "epoch": 0.18129430751421644,
      "grad_norm": 0.802157998085022,
      "learning_rate": 2.8942107300848784e-05,
      "loss": 0.05700739473104477,
      "step": 1341
    },
    {
      "epoch": 0.18142950088298168,
      "grad_norm": 0.6724606156349182,
      "learning_rate": 2.8939581021439225e-05,
      "loss": 0.0663527324795723,
      "step": 1342
    },
    {
      "epoch": 0.18156469425174696,
      "grad_norm": 0.5345813632011414,
      "learning_rate": 2.8937051839734563e-05,
      "loss": 0.043694838881492615,
      "step": 1343
    },
    {
      "epoch": 0.1816998876205122,
      "grad_norm": 0.34863507747650146,
      "learning_rate": 2.8934519756261384e-05,
      "loss": 0.05110260099172592,
      "step": 1344
    },
    {
      "epoch": 0.18183508098927748,
      "grad_norm": 1.1135504245758057,
      "learning_rate": 2.8931984771546885e-05,
      "loss": 0.058423250913619995,
      "step": 1345
    },
    {
      "epoch": 0.18197027435804275,
      "grad_norm": 1.122864007949829,
      "learning_rate": 2.8929446886118866e-05,
      "loss": 0.06855830550193787,
      "step": 1346
    },
    {
      "epoch": 0.182105467726808,
      "grad_norm": 1.548689603805542,
      "learning_rate": 2.892690610050572e-05,
      "loss": 0.06757400184869766,
      "step": 1347
    },
    {
      "epoch": 0.18224066109557327,
      "grad_norm": 0.3105394244194031,
      "learning_rate": 2.892436241523646e-05,
      "loss": 0.047551169991493225,
      "step": 1348
    },
    {
      "epoch": 0.18237585446433852,
      "grad_norm": 1.830718755722046,
      "learning_rate": 2.8921815830840685e-05,
      "loss": 0.08233155310153961,
      "step": 1349
    },
    {
      "epoch": 0.1825110478331038,
      "grad_norm": 0.4397461712360382,
      "learning_rate": 2.891926634784862e-05,
      "loss": 0.06138932704925537,
      "step": 1350
    },
    {
      "epoch": 0.18264624120186904,
      "grad_norm": 0.7208978533744812,
      "learning_rate": 2.8916713966791076e-05,
      "loss": 0.05991073325276375,
      "step": 1351
    },
    {
      "epoch": 0.1827814345706343,
      "grad_norm": 0.7274355292320251,
      "learning_rate": 2.8914158688199464e-05,
      "loss": 0.060209520161151886,
      "step": 1352
    },
    {
      "epoch": 0.18291662793939958,
      "grad_norm": 1.2630335092544556,
      "learning_rate": 2.891160051260582e-05,
      "loss": 0.06155611574649811,
      "step": 1353
    },
    {
      "epoch": 0.18305182130816483,
      "grad_norm": 1.1451889276504517,
      "learning_rate": 2.8909039440542758e-05,
      "loss": 0.062114208936691284,
      "step": 1354
    },
    {
      "epoch": 0.1831870146769301,
      "grad_norm": 1.212422251701355,
      "learning_rate": 2.890647547254352e-05,
      "loss": 0.060322076082229614,
      "step": 1355
    },
    {
      "epoch": 0.18332220804569535,
      "grad_norm": 0.9648664593696594,
      "learning_rate": 2.8903908609141923e-05,
      "loss": 0.08123242855072021,
      "step": 1356
    },
    {
      "epoch": 0.18345740141446062,
      "grad_norm": 1.5143882036209106,
      "learning_rate": 2.8901338850872413e-05,
      "loss": 0.06759816408157349,
      "step": 1357
    },
    {
      "epoch": 0.18359259478322587,
      "grad_norm": 0.5402228832244873,
      "learning_rate": 2.8898766198270022e-05,
      "loss": 0.07894018292427063,
      "step": 1358
    },
    {
      "epoch": 0.18372778815199114,
      "grad_norm": 0.6195774674415588,
      "learning_rate": 2.8896190651870392e-05,
      "loss": 0.09232117235660553,
      "step": 1359
    },
    {
      "epoch": 0.18386298152075642,
      "grad_norm": 1.2298245429992676,
      "learning_rate": 2.8893612212209763e-05,
      "loss": 0.05993261933326721,
      "step": 1360
    },
    {
      "epoch": 0.18399817488952167,
      "grad_norm": 2.5731022357940674,
      "learning_rate": 2.8891030879824985e-05,
      "loss": 0.07145705819129944,
      "step": 1361
    },
    {
      "epoch": 0.18413336825828694,
      "grad_norm": 0.5495437979698181,
      "learning_rate": 2.88884466552535e-05,
      "loss": 0.06904041767120361,
      "step": 1362
    },
    {
      "epoch": 0.18426856162705219,
      "grad_norm": 0.8204841613769531,
      "learning_rate": 2.888585953903336e-05,
      "loss": 0.051793456077575684,
      "step": 1363
    },
    {
      "epoch": 0.18440375499581746,
      "grad_norm": 1.3093513250350952,
      "learning_rate": 2.888326953170321e-05,
      "loss": 0.06429917365312576,
      "step": 1364
    },
    {
      "epoch": 0.1845389483645827,
      "grad_norm": 0.4332553446292877,
      "learning_rate": 2.8880676633802314e-05,
      "loss": 0.059311643242836,
      "step": 1365
    },
    {
      "epoch": 0.18467414173334798,
      "grad_norm": 0.9532291293144226,
      "learning_rate": 2.8878080845870522e-05,
      "loss": 0.07912647724151611,
      "step": 1366
    },
    {
      "epoch": 0.18480933510211325,
      "grad_norm": 0.7556528449058533,
      "learning_rate": 2.887548216844829e-05,
      "loss": 0.04770098999142647,
      "step": 1367
    },
    {
      "epoch": 0.1849445284708785,
      "grad_norm": 1.159386157989502,
      "learning_rate": 2.8872880602076675e-05,
      "loss": 0.053607672452926636,
      "step": 1368
    },
    {
      "epoch": 0.18507972183964377,
      "grad_norm": 0.6745975017547607,
      "learning_rate": 2.8870276147297344e-05,
      "loss": 0.08379173278808594,
      "step": 1369
    },
    {
      "epoch": 0.18521491520840902,
      "grad_norm": 1.985194206237793,
      "learning_rate": 2.8867668804652552e-05,
      "loss": 0.06111246347427368,
      "step": 1370
    },
    {
      "epoch": 0.1853501085771743,
      "grad_norm": 0.5915805697441101,
      "learning_rate": 2.886505857468516e-05,
      "loss": 0.06341639161109924,
      "step": 1371
    },
    {
      "epoch": 0.18548530194593954,
      "grad_norm": 0.8633459210395813,
      "learning_rate": 2.8862445457938642e-05,
      "loss": 0.0885532796382904,
      "step": 1372
    },
    {
      "epoch": 0.18562049531470481,
      "grad_norm": 0.5439592599868774,
      "learning_rate": 2.8859829454957053e-05,
      "loss": 0.05765097588300705,
      "step": 1373
    },
    {
      "epoch": 0.1857556886834701,
      "grad_norm": 0.846956193447113,
      "learning_rate": 2.8857210566285062e-05,
      "loss": 0.06486758589744568,
      "step": 1374
    },
    {
      "epoch": 0.18589088205223533,
      "grad_norm": 1.9822721481323242,
      "learning_rate": 2.8854588792467932e-05,
      "loss": 0.07421214878559113,
      "step": 1375
    },
    {
      "epoch": 0.1860260754210006,
      "grad_norm": 2.105954647064209,
      "learning_rate": 2.8851964134051535e-05,
      "loss": 0.07359392940998077,
      "step": 1376
    },
    {
      "epoch": 0.18616126878976585,
      "grad_norm": 1.0604692697525024,
      "learning_rate": 2.884933659158234e-05,
      "loss": 0.07753920555114746,
      "step": 1377
    },
    {
      "epoch": 0.18629646215853113,
      "grad_norm": 1.4054888486862183,
      "learning_rate": 2.8846706165607415e-05,
      "loss": 0.06569766998291016,
      "step": 1378
    },
    {
      "epoch": 0.18643165552729637,
      "grad_norm": 0.9416623711585999,
      "learning_rate": 2.8844072856674422e-05,
      "loss": 0.0600811243057251,
      "step": 1379
    },
    {
      "epoch": 0.18656684889606165,
      "grad_norm": 0.5811194777488708,
      "learning_rate": 2.8841436665331634e-05,
      "loss": 0.05377200245857239,
      "step": 1380
    },
    {
      "epoch": 0.18670204226482692,
      "grad_norm": 1.1855747699737549,
      "learning_rate": 2.8838797592127927e-05,
      "loss": 0.04610684514045715,
      "step": 1381
    },
    {
      "epoch": 0.18683723563359217,
      "grad_norm": 0.3341962993144989,
      "learning_rate": 2.883615563761276e-05,
      "loss": 0.04806177318096161,
      "step": 1382
    },
    {
      "epoch": 0.18697242900235744,
      "grad_norm": 0.5546738505363464,
      "learning_rate": 2.8833510802336206e-05,
      "loss": 0.03987765312194824,
      "step": 1383
    },
    {
      "epoch": 0.1871076223711227,
      "grad_norm": 0.3393515944480896,
      "learning_rate": 2.883086308684893e-05,
      "loss": 0.05964048206806183,
      "step": 1384
    },
    {
      "epoch": 0.18724281573988796,
      "grad_norm": 0.5718085169792175,
      "learning_rate": 2.882821249170221e-05,
      "loss": 0.04978965222835541,
      "step": 1385
    },
    {
      "epoch": 0.1873780091086532,
      "grad_norm": 2.1170740127563477,
      "learning_rate": 2.8825559017447905e-05,
      "loss": 0.05634533613920212,
      "step": 1386
    },
    {
      "epoch": 0.18751320247741848,
      "grad_norm": 1.212235689163208,
      "learning_rate": 2.8822902664638487e-05,
      "loss": 0.05493786931037903,
      "step": 1387
    },
    {
      "epoch": 0.18764839584618376,
      "grad_norm": 0.3020192086696625,
      "learning_rate": 2.882024343382702e-05,
      "loss": 0.056016504764556885,
      "step": 1388
    },
    {
      "epoch": 0.187783589214949,
      "grad_norm": 1.2162086963653564,
      "learning_rate": 2.8817581325567174e-05,
      "loss": 0.06100273132324219,
      "step": 1389
    },
    {
      "epoch": 0.18791878258371428,
      "grad_norm": 1.113671898841858,
      "learning_rate": 2.8814916340413205e-05,
      "loss": 0.052695900201797485,
      "step": 1390
    },
    {
      "epoch": 0.18805397595247952,
      "grad_norm": 1.1613482236862183,
      "learning_rate": 2.881224847891999e-05,
      "loss": 0.050453707575798035,
      "step": 1391
    },
    {
      "epoch": 0.1881891693212448,
      "grad_norm": 1.153808355331421,
      "learning_rate": 2.8809577741642987e-05,
      "loss": 0.061996977776288986,
      "step": 1392
    },
    {
      "epoch": 0.18832436269001004,
      "grad_norm": 0.34761378169059753,
      "learning_rate": 2.8806904129138255e-05,
      "loss": 0.050693221390247345,
      "step": 1393
    },
    {
      "epoch": 0.18845955605877532,
      "grad_norm": 0.4242549240589142,
      "learning_rate": 2.8804227641962457e-05,
      "loss": 0.04989054054021835,
      "step": 1394
    },
    {
      "epoch": 0.1885947494275406,
      "grad_norm": 1.1975650787353516,
      "learning_rate": 2.8801548280672847e-05,
      "loss": 0.07356453686952591,
      "step": 1395
    },
    {
      "epoch": 0.18872994279630584,
      "grad_norm": 0.8233917355537415,
      "learning_rate": 2.8798866045827288e-05,
      "loss": 0.04728338122367859,
      "step": 1396
    },
    {
      "epoch": 0.1888651361650711,
      "grad_norm": 0.7185053825378418,
      "learning_rate": 2.8796180937984234e-05,
      "loss": 0.06343841552734375,
      "step": 1397
    },
    {
      "epoch": 0.18900032953383636,
      "grad_norm": 0.8097310066223145,
      "learning_rate": 2.8793492957702738e-05,
      "loss": 0.0484083853662014,
      "step": 1398
    },
    {
      "epoch": 0.18913552290260163,
      "grad_norm": 1.0301105976104736,
      "learning_rate": 2.8790802105542454e-05,
      "loss": 0.03863731026649475,
      "step": 1399
    },
    {
      "epoch": 0.18927071627136688,
      "grad_norm": 0.9395861625671387,
      "learning_rate": 2.8788108382063628e-05,
      "loss": 0.0369383729994297,
      "step": 1400
    },
    {
      "epoch": 0.18940590964013215,
      "grad_norm": 0.5939541459083557,
      "learning_rate": 2.878541178782711e-05,
      "loss": 0.04944443702697754,
      "step": 1401
    },
    {
      "epoch": 0.18954110300889743,
      "grad_norm": 0.3824097514152527,
      "learning_rate": 2.8782712323394344e-05,
      "loss": 0.06165945529937744,
      "step": 1402
    },
    {
      "epoch": 0.18967629637766267,
      "grad_norm": 0.6497465372085571,
      "learning_rate": 2.878000998932738e-05,
      "loss": 0.061732374131679535,
      "step": 1403
    },
    {
      "epoch": 0.18981148974642795,
      "grad_norm": 0.8493558168411255,
      "learning_rate": 2.8777304786188847e-05,
      "loss": 0.06216737627983093,
      "step": 1404
    },
    {
      "epoch": 0.1899466831151932,
      "grad_norm": 0.565345287322998,
      "learning_rate": 2.8774596714541988e-05,
      "loss": 0.058844342827796936,
      "step": 1405
    },
    {
      "epoch": 0.19008187648395847,
      "grad_norm": 0.26175230741500854,
      "learning_rate": 2.8771885774950637e-05,
      "loss": 0.0428578220307827,
      "step": 1406
    },
    {
      "epoch": 0.1902170698527237,
      "grad_norm": 1.515351414680481,
      "learning_rate": 2.876917196797923e-05,
      "loss": 0.06139880791306496,
      "step": 1407
    },
    {
      "epoch": 0.190352263221489,
      "grad_norm": 0.29945918917655945,
      "learning_rate": 2.876645529419279e-05,
      "loss": 0.04580672085285187,
      "step": 1408
    },
    {
      "epoch": 0.19048745659025426,
      "grad_norm": 0.5630703568458557,
      "learning_rate": 2.876373575415695e-05,
      "loss": 0.06493443250656128,
      "step": 1409
    },
    {
      "epoch": 0.1906226499590195,
      "grad_norm": 0.6920896172523499,
      "learning_rate": 2.8761013348437926e-05,
      "loss": 0.04265984147787094,
      "step": 1410
    },
    {
      "epoch": 0.19075784332778478,
      "grad_norm": 0.6778731942176819,
      "learning_rate": 2.875828807760254e-05,
      "loss": 0.06861774623394012,
      "step": 1411
    },
    {
      "epoch": 0.19089303669655003,
      "grad_norm": 0.705251932144165,
      "learning_rate": 2.875555994221821e-05,
      "loss": 0.049150049686431885,
      "step": 1412
    },
    {
      "epoch": 0.1910282300653153,
      "grad_norm": 0.42983245849609375,
      "learning_rate": 2.8752828942852943e-05,
      "loss": 0.06673504412174225,
      "step": 1413
    },
    {
      "epoch": 0.19116342343408055,
      "grad_norm": 0.5979132056236267,
      "learning_rate": 2.875009508007535e-05,
      "loss": 0.05231538414955139,
      "step": 1414
    },
    {
      "epoch": 0.19129861680284582,
      "grad_norm": 0.5539986491203308,
      "learning_rate": 2.8747358354454642e-05,
      "loss": 0.05697956681251526,
      "step": 1415
    },
    {
      "epoch": 0.1914338101716111,
      "grad_norm": 1.201626181602478,
      "learning_rate": 2.8744618766560614e-05,
      "loss": 0.062018804252147675,
      "step": 1416
    },
    {
      "epoch": 0.19156900354037634,
      "grad_norm": 0.4408376216888428,
      "learning_rate": 2.8741876316963664e-05,
      "loss": 0.07073703408241272,
      "step": 1417
    },
    {
      "epoch": 0.19170419690914162,
      "grad_norm": 0.5461744666099548,
      "learning_rate": 2.873913100623478e-05,
      "loss": 0.06411121785640717,
      "step": 1418
    },
    {
      "epoch": 0.19183939027790686,
      "grad_norm": 0.5853143334388733,
      "learning_rate": 2.873638283494556e-05,
      "loss": 0.06544029712677002,
      "step": 1419
    },
    {
      "epoch": 0.19197458364667214,
      "grad_norm": 0.3352160155773163,
      "learning_rate": 2.8733631803668178e-05,
      "loss": 0.052479326725006104,
      "step": 1420
    },
    {
      "epoch": 0.19210977701543738,
      "grad_norm": 0.587458074092865,
      "learning_rate": 2.8730877912975418e-05,
      "loss": 0.058047693222761154,
      "step": 1421
    },
    {
      "epoch": 0.19224497038420266,
      "grad_norm": 0.5267054438591003,
      "learning_rate": 2.8728121163440656e-05,
      "loss": 0.06486907601356506,
      "step": 1422
    },
    {
      "epoch": 0.1923801637529679,
      "grad_norm": 0.9581182599067688,
      "learning_rate": 2.8725361555637863e-05,
      "loss": 0.065681591629982,
      "step": 1423
    },
    {
      "epoch": 0.19251535712173318,
      "grad_norm": 0.5375241041183472,
      "learning_rate": 2.8722599090141598e-05,
      "loss": 0.04519478231668472,
      "step": 1424
    },
    {
      "epoch": 0.19265055049049845,
      "grad_norm": 0.4517899751663208,
      "learning_rate": 2.8719833767527026e-05,
      "loss": 0.058487050235271454,
      "step": 1425
    },
    {
      "epoch": 0.1927857438592637,
      "grad_norm": 1.9817936420440674,
      "learning_rate": 2.8717065588369896e-05,
      "loss": 0.0784255713224411,
      "step": 1426
    },
    {
      "epoch": 0.19292093722802897,
      "grad_norm": 1.4670841693878174,
      "learning_rate": 2.871429455324657e-05,
      "loss": 0.06524771451950073,
      "step": 1427
    },
    {
      "epoch": 0.19305613059679422,
      "grad_norm": 0.8886244297027588,
      "learning_rate": 2.871152066273398e-05,
      "loss": 0.059401385486125946,
      "step": 1428
    },
    {
      "epoch": 0.1931913239655595,
      "grad_norm": 0.4757377803325653,
      "learning_rate": 2.870874391740967e-05,
      "loss": 0.05879536271095276,
      "step": 1429
    },
    {
      "epoch": 0.19332651733432474,
      "grad_norm": 0.8385919332504272,
      "learning_rate": 2.8705964317851774e-05,
      "loss": 0.06977000832557678,
      "step": 1430
    },
    {
      "epoch": 0.19346171070309,
      "grad_norm": 1.4974596500396729,
      "learning_rate": 2.8703181864639013e-05,
      "loss": 0.06021403521299362,
      "step": 1431
    },
    {
      "epoch": 0.19359690407185529,
      "grad_norm": 1.6825629472732544,
      "learning_rate": 2.870039655835072e-05,
      "loss": 0.07012343406677246,
      "step": 1432
    },
    {
      "epoch": 0.19373209744062053,
      "grad_norm": 0.38851383328437805,
      "learning_rate": 2.8697608399566796e-05,
      "loss": 0.050712019205093384,
      "step": 1433
    },
    {
      "epoch": 0.1938672908093858,
      "grad_norm": 0.5127747058868408,
      "learning_rate": 2.869481738886777e-05,
      "loss": 0.07452350854873657,
      "step": 1434
    },
    {
      "epoch": 0.19400248417815105,
      "grad_norm": 0.3593316674232483,
      "learning_rate": 2.8692023526834725e-05,
      "loss": 0.07108698785305023,
      "step": 1435
    },
    {
      "epoch": 0.19413767754691633,
      "grad_norm": 2.6567790508270264,
      "learning_rate": 2.8689226814049367e-05,
      "loss": 0.08526412397623062,
      "step": 1436
    },
    {
      "epoch": 0.19427287091568157,
      "grad_norm": 1.179962158203125,
      "learning_rate": 2.868642725109399e-05,
      "loss": 0.06299266219139099,
      "step": 1437
    },
    {
      "epoch": 0.19440806428444685,
      "grad_norm": 1.5013912916183472,
      "learning_rate": 2.868362483855147e-05,
      "loss": 0.06303150206804276,
      "step": 1438
    },
    {
      "epoch": 0.19454325765321212,
      "grad_norm": 0.7080300450325012,
      "learning_rate": 2.8680819577005295e-05,
      "loss": 0.04272356629371643,
      "step": 1439
    },
    {
      "epoch": 0.19467845102197737,
      "grad_norm": 0.361674964427948,
      "learning_rate": 2.8678011467039526e-05,
      "loss": 0.06747883558273315,
      "step": 1440
    },
    {
      "epoch": 0.19481364439074264,
      "grad_norm": 0.835003674030304,
      "learning_rate": 2.867520050923883e-05,
      "loss": 0.0589282363653183,
      "step": 1441
    },
    {
      "epoch": 0.1949488377595079,
      "grad_norm": 0.40837588906288147,
      "learning_rate": 2.8672386704188466e-05,
      "loss": 0.06220751255750656,
      "step": 1442
    },
    {
      "epoch": 0.19508403112827316,
      "grad_norm": 0.6291946172714233,
      "learning_rate": 2.8669570052474273e-05,
      "loss": 0.0549418106675148,
      "step": 1443
    },
    {
      "epoch": 0.1952192244970384,
      "grad_norm": 0.40878450870513916,
      "learning_rate": 2.86667505546827e-05,
      "loss": 0.04268855229020119,
      "step": 1444
    },
    {
      "epoch": 0.19535441786580368,
      "grad_norm": 0.4511062800884247,
      "learning_rate": 2.866392821140079e-05,
      "loss": 0.07379084825515747,
      "step": 1445
    },
    {
      "epoch": 0.19548961123456896,
      "grad_norm": 0.9196026921272278,
      "learning_rate": 2.8661103023216154e-05,
      "loss": 0.06482299417257309,
      "step": 1446
    },
    {
      "epoch": 0.1956248046033342,
      "grad_norm": 0.7536078691482544,
      "learning_rate": 2.8658274990717018e-05,
      "loss": 0.06228230148553848,
      "step": 1447
    },
    {
      "epoch": 0.19575999797209948,
      "grad_norm": 1.1135989427566528,
      "learning_rate": 2.86554441144922e-05,
      "loss": 0.06930319964885712,
      "step": 1448
    },
    {
      "epoch": 0.19589519134086472,
      "grad_norm": 0.5206980109214783,
      "learning_rate": 2.8652610395131097e-05,
      "loss": 0.07230779528617859,
      "step": 1449
    },
    {
      "epoch": 0.19603038470963,
      "grad_norm": 0.5373529195785522,
      "learning_rate": 2.8649773833223702e-05,
      "loss": 0.06830896437168121,
      "step": 1450
    },
    {
      "epoch": 0.19616557807839524,
      "grad_norm": 0.3036535084247589,
      "learning_rate": 2.8646934429360606e-05,
      "loss": 0.049472615122795105,
      "step": 1451
    },
    {
      "epoch": 0.19630077144716052,
      "grad_norm": 0.5945667624473572,
      "learning_rate": 2.8644092184132986e-05,
      "loss": 0.06595683097839355,
      "step": 1452
    },
    {
      "epoch": 0.1964359648159258,
      "grad_norm": 0.8892114758491516,
      "learning_rate": 2.864124709813262e-05,
      "loss": 0.05675327777862549,
      "step": 1453
    },
    {
      "epoch": 0.19657115818469104,
      "grad_norm": 0.727644681930542,
      "learning_rate": 2.8638399171951856e-05,
      "loss": 0.06490229070186615,
      "step": 1454
    },
    {
      "epoch": 0.1967063515534563,
      "grad_norm": 0.6210764050483704,
      "learning_rate": 2.8635548406183664e-05,
      "loss": 0.04059150815010071,
      "step": 1455
    },
    {
      "epoch": 0.19684154492222156,
      "grad_norm": 0.6249570250511169,
      "learning_rate": 2.8632694801421576e-05,
      "loss": 0.07593834400177002,
      "step": 1456
    },
    {
      "epoch": 0.19697673829098683,
      "grad_norm": 1.0537331104278564,
      "learning_rate": 2.862983835825973e-05,
      "loss": 0.06349700689315796,
      "step": 1457
    },
    {
      "epoch": 0.19711193165975208,
      "grad_norm": 0.6601712107658386,
      "learning_rate": 2.8626979077292856e-05,
      "loss": 0.042082592844963074,
      "step": 1458
    },
    {
      "epoch": 0.19724712502851735,
      "grad_norm": 0.5785070061683655,
      "learning_rate": 2.862411695911627e-05,
      "loss": 0.06087595224380493,
      "step": 1459
    },
    {
      "epoch": 0.19738231839728262,
      "grad_norm": 0.5136145353317261,
      "learning_rate": 2.862125200432588e-05,
      "loss": 0.06294526904821396,
      "step": 1460
    },
    {
      "epoch": 0.19751751176604787,
      "grad_norm": 0.856765866279602,
      "learning_rate": 2.8618384213518188e-05,
      "loss": 0.04286312311887741,
      "step": 1461
    },
    {
      "epoch": 0.19765270513481314,
      "grad_norm": 0.7110387682914734,
      "learning_rate": 2.861551358729028e-05,
      "loss": 0.07258517295122147,
      "step": 1462
    },
    {
      "epoch": 0.1977878985035784,
      "grad_norm": 1.8345304727554321,
      "learning_rate": 2.8612640126239836e-05,
      "loss": 0.07168304920196533,
      "step": 1463
    },
    {
      "epoch": 0.19792309187234366,
      "grad_norm": 0.32499468326568604,
      "learning_rate": 2.8609763830965126e-05,
      "loss": 0.032810524106025696,
      "step": 1464
    },
    {
      "epoch": 0.1980582852411089,
      "grad_norm": 0.4453463554382324,
      "learning_rate": 2.860688470206501e-05,
      "loss": 0.07522565126419067,
      "step": 1465
    },
    {
      "epoch": 0.19819347860987419,
      "grad_norm": 0.577467143535614,
      "learning_rate": 2.8604002740138936e-05,
      "loss": 0.06885667890310287,
      "step": 1466
    },
    {
      "epoch": 0.19832867197863946,
      "grad_norm": 0.6764350533485413,
      "learning_rate": 2.860111794578695e-05,
      "loss": 0.05699273943901062,
      "step": 1467
    },
    {
      "epoch": 0.1984638653474047,
      "grad_norm": 1.195464849472046,
      "learning_rate": 2.8598230319609677e-05,
      "loss": 0.055881090462207794,
      "step": 1468
    },
    {
      "epoch": 0.19859905871616998,
      "grad_norm": 0.4497874975204468,
      "learning_rate": 2.8595339862208336e-05,
      "loss": 0.06361427903175354,
      "step": 1469
    },
    {
      "epoch": 0.19873425208493523,
      "grad_norm": 0.5617062449455261,
      "learning_rate": 2.8592446574184733e-05,
      "loss": 0.08015744388103485,
      "step": 1470
    },
    {
      "epoch": 0.1988694454537005,
      "grad_norm": 0.571229100227356,
      "learning_rate": 2.8589550456141274e-05,
      "loss": 0.08003470301628113,
      "step": 1471
    },
    {
      "epoch": 0.19900463882246575,
      "grad_norm": 0.84321129322052,
      "learning_rate": 2.8586651508680942e-05,
      "loss": 0.04576558619737625,
      "step": 1472
    },
    {
      "epoch": 0.19913983219123102,
      "grad_norm": 0.3449549674987793,
      "learning_rate": 2.8583749732407312e-05,
      "loss": 0.038829684257507324,
      "step": 1473
    },
    {
      "epoch": 0.1992750255599963,
      "grad_norm": 0.6807087659835815,
      "learning_rate": 2.8580845127924546e-05,
      "loss": 0.04984404146671295,
      "step": 1474
    },
    {
      "epoch": 0.19941021892876154,
      "grad_norm": 0.3755718171596527,
      "learning_rate": 2.8577937695837408e-05,
      "loss": 0.048488616943359375,
      "step": 1475
    },
    {
      "epoch": 0.19954541229752681,
      "grad_norm": 1.0324113368988037,
      "learning_rate": 2.8575027436751235e-05,
      "loss": 0.05285117030143738,
      "step": 1476
    },
    {
      "epoch": 0.19968060566629206,
      "grad_norm": 1.789212703704834,
      "learning_rate": 2.8572114351271955e-05,
      "loss": 0.0629136934876442,
      "step": 1477
    },
    {
      "epoch": 0.19981579903505733,
      "grad_norm": 0.4355745315551758,
      "learning_rate": 2.85691984400061e-05,
      "loss": 0.06396521627902985,
      "step": 1478
    },
    {
      "epoch": 0.19995099240382258,
      "grad_norm": 1.519565463066101,
      "learning_rate": 2.8566279703560762e-05,
      "loss": 0.05704353377223015,
      "step": 1479
    },
    {
      "epoch": 0.20008618577258785,
      "grad_norm": 0.5814024806022644,
      "learning_rate": 2.8563358142543648e-05,
      "loss": 0.0474286787211895,
      "step": 1480
    },
    {
      "epoch": 0.20022137914135313,
      "grad_norm": 3.2591962814331055,
      "learning_rate": 2.856043375756304e-05,
      "loss": 0.07081346213817596,
      "step": 1481
    },
    {
      "epoch": 0.20035657251011837,
      "grad_norm": 1.7282776832580566,
      "learning_rate": 2.855750654922781e-05,
      "loss": 0.05421200394630432,
      "step": 1482
    },
    {
      "epoch": 0.20049176587888365,
      "grad_norm": 2.3187155723571777,
      "learning_rate": 2.855457651814742e-05,
      "loss": 0.04969785362482071,
      "step": 1483
    },
    {
      "epoch": 0.2006269592476489,
      "grad_norm": 1.7427756786346436,
      "learning_rate": 2.8551643664931916e-05,
      "loss": 0.052194394171237946,
      "step": 1484
    },
    {
      "epoch": 0.20076215261641417,
      "grad_norm": 0.4092828631401062,
      "learning_rate": 2.8548707990191933e-05,
      "loss": 0.05346745252609253,
      "step": 1485
    },
    {
      "epoch": 0.20089734598517942,
      "grad_norm": 0.6095959544181824,
      "learning_rate": 2.8545769494538698e-05,
      "loss": 0.060761407017707825,
      "step": 1486
    },
    {
      "epoch": 0.2010325393539447,
      "grad_norm": 2.5191256999969482,
      "learning_rate": 2.854282817858402e-05,
      "loss": 0.06619636714458466,
      "step": 1487
    },
    {
      "epoch": 0.20116773272270996,
      "grad_norm": 1.464105248451233,
      "learning_rate": 2.85398840429403e-05,
      "loss": 0.05017085000872612,
      "step": 1488
    },
    {
      "epoch": 0.2013029260914752,
      "grad_norm": 1.7387162446975708,
      "learning_rate": 2.853693708822051e-05,
      "loss": 0.0615549311041832,
      "step": 1489
    },
    {
      "epoch": 0.20143811946024048,
      "grad_norm": 1.7491748332977295,
      "learning_rate": 2.8533987315038234e-05,
      "loss": 0.08808696269989014,
      "step": 1490
    },
    {
      "epoch": 0.20157331282900573,
      "grad_norm": 0.8185397386550903,
      "learning_rate": 2.8531034724007627e-05,
      "loss": 0.05078073590993881,
      "step": 1491
    },
    {
      "epoch": 0.201708506197771,
      "grad_norm": 1.4671510457992554,
      "learning_rate": 2.8528079315743435e-05,
      "loss": 0.054104551672935486,
      "step": 1492
    },
    {
      "epoch": 0.20184369956653625,
      "grad_norm": 0.38106274604797363,
      "learning_rate": 2.852512109086099e-05,
      "loss": 0.05543220043182373,
      "step": 1493
    },
    {
      "epoch": 0.20197889293530152,
      "grad_norm": 1.7341113090515137,
      "learning_rate": 2.8522160049976208e-05,
      "loss": 0.074765145778656,
      "step": 1494
    },
    {
      "epoch": 0.2021140863040668,
      "grad_norm": 1.8391804695129395,
      "learning_rate": 2.8519196193705595e-05,
      "loss": 0.05571398138999939,
      "step": 1495
    },
    {
      "epoch": 0.20224927967283204,
      "grad_norm": 3.4821393489837646,
      "learning_rate": 2.8516229522666243e-05,
      "loss": 0.07515992224216461,
      "step": 1496
    },
    {
      "epoch": 0.20238447304159732,
      "grad_norm": 3.051548480987549,
      "learning_rate": 2.8513260037475825e-05,
      "loss": 0.08510702848434448,
      "step": 1497
    },
    {
      "epoch": 0.20251966641036256,
      "grad_norm": 1.314646601676941,
      "learning_rate": 2.8510287738752604e-05,
      "loss": 0.0674319863319397,
      "step": 1498
    },
    {
      "epoch": 0.20265485977912784,
      "grad_norm": 1.2149888277053833,
      "learning_rate": 2.8507312627115435e-05,
      "loss": 0.06909724324941635,
      "step": 1499
    },
    {
      "epoch": 0.20279005314789308,
      "grad_norm": 1.0091185569763184,
      "learning_rate": 2.850433470318374e-05,
      "loss": 0.059906914830207825,
      "step": 1500
    },
    {
      "epoch": 0.20292524651665836,
      "grad_norm": 0.4593426287174225,
      "learning_rate": 2.8501353967577556e-05,
      "loss": 0.05634528398513794,
      "step": 1501
    },
    {
      "epoch": 0.20306043988542363,
      "grad_norm": 0.7114489674568176,
      "learning_rate": 2.8498370420917468e-05,
      "loss": 0.05177636444568634,
      "step": 1502
    },
    {
      "epoch": 0.20319563325418888,
      "grad_norm": 0.8836167454719543,
      "learning_rate": 2.8495384063824683e-05,
      "loss": 0.04526149481534958,
      "step": 1503
    },
    {
      "epoch": 0.20333082662295415,
      "grad_norm": 0.6767558455467224,
      "learning_rate": 2.8492394896920964e-05,
      "loss": 0.04810865968465805,
      "step": 1504
    },
    {
      "epoch": 0.2034660199917194,
      "grad_norm": 0.8853635787963867,
      "learning_rate": 2.848940292082868e-05,
      "loss": 0.05030532553792,
      "step": 1505
    },
    {
      "epoch": 0.20360121336048467,
      "grad_norm": 1.0097651481628418,
      "learning_rate": 2.8486408136170772e-05,
      "loss": 0.05914625525474548,
      "step": 1506
    },
    {
      "epoch": 0.20373640672924992,
      "grad_norm": 0.9485298991203308,
      "learning_rate": 2.8483410543570776e-05,
      "loss": 0.053879305720329285,
      "step": 1507
    },
    {
      "epoch": 0.2038716000980152,
      "grad_norm": 0.3891317546367645,
      "learning_rate": 2.8480410143652803e-05,
      "loss": 0.03825295343995094,
      "step": 1508
    },
    {
      "epoch": 0.20400679346678047,
      "grad_norm": 0.304548978805542,
      "learning_rate": 2.8477406937041547e-05,
      "loss": 0.058615073561668396,
      "step": 1509
    },
    {
      "epoch": 0.2041419868355457,
      "grad_norm": 0.4876135289669037,
      "learning_rate": 2.8474400924362298e-05,
      "loss": 0.0701942965388298,
      "step": 1510
    },
    {
      "epoch": 0.204277180204311,
      "grad_norm": 0.617000162601471,
      "learning_rate": 2.847139210624092e-05,
      "loss": 0.07015332579612732,
      "step": 1511
    },
    {
      "epoch": 0.20441237357307623,
      "grad_norm": 0.3961528539657593,
      "learning_rate": 2.8468380483303873e-05,
      "loss": 0.06465774774551392,
      "step": 1512
    },
    {
      "epoch": 0.2045475669418415,
      "grad_norm": 1.253760814666748,
      "learning_rate": 2.8465366056178183e-05,
      "loss": 0.049995020031929016,
      "step": 1513
    },
    {
      "epoch": 0.20468276031060675,
      "grad_norm": 0.6975782513618469,
      "learning_rate": 2.8462348825491475e-05,
      "loss": 0.06051470339298248,
      "step": 1514
    },
    {
      "epoch": 0.20481795367937203,
      "grad_norm": 1.233405590057373,
      "learning_rate": 2.8459328791871953e-05,
      "loss": 0.06607866287231445,
      "step": 1515
    },
    {
      "epoch": 0.2049531470481373,
      "grad_norm": 2.070460557937622,
      "learning_rate": 2.8456305955948402e-05,
      "loss": 0.07863160967826843,
      "step": 1516
    },
    {
      "epoch": 0.20508834041690255,
      "grad_norm": 0.9268529415130615,
      "learning_rate": 2.845328031835019e-05,
      "loss": 0.07612798362970352,
      "step": 1517
    },
    {
      "epoch": 0.20522353378566782,
      "grad_norm": 0.5493766665458679,
      "learning_rate": 2.8450251879707277e-05,
      "loss": 0.05937439203262329,
      "step": 1518
    },
    {
      "epoch": 0.20535872715443307,
      "grad_norm": 0.817802906036377,
      "learning_rate": 2.8447220640650194e-05,
      "loss": 0.043556079268455505,
      "step": 1519
    },
    {
      "epoch": 0.20549392052319834,
      "grad_norm": 0.5602957010269165,
      "learning_rate": 2.8444186601810068e-05,
      "loss": 0.05191570520401001,
      "step": 1520
    },
    {
      "epoch": 0.2056291138919636,
      "grad_norm": 1.5219169855117798,
      "learning_rate": 2.84411497638186e-05,
      "loss": 0.062377363443374634,
      "step": 1521
    },
    {
      "epoch": 0.20576430726072886,
      "grad_norm": 0.41533032059669495,
      "learning_rate": 2.843811012730807e-05,
      "loss": 0.06018230319023132,
      "step": 1522
    },
    {
      "epoch": 0.20589950062949414,
      "grad_norm": 0.42811867594718933,
      "learning_rate": 2.8435067692911353e-05,
      "loss": 0.06564238667488098,
      "step": 1523
    },
    {
      "epoch": 0.20603469399825938,
      "grad_norm": 0.4506815969944,
      "learning_rate": 2.8432022461261897e-05,
      "loss": 0.05740030109882355,
      "step": 1524
    },
    {
      "epoch": 0.20616988736702466,
      "grad_norm": 0.7717320919036865,
      "learning_rate": 2.8428974432993736e-05,
      "loss": 0.07080622762441635,
      "step": 1525
    },
    {
      "epoch": 0.2063050807357899,
      "grad_norm": 0.30265215039253235,
      "learning_rate": 2.8425923608741486e-05,
      "loss": 0.06177237629890442,
      "step": 1526
    },
    {
      "epoch": 0.20644027410455518,
      "grad_norm": 0.9989373683929443,
      "learning_rate": 2.8422869989140343e-05,
      "loss": 0.08423186093568802,
      "step": 1527
    },
    {
      "epoch": 0.20657546747332042,
      "grad_norm": 0.8502436876296997,
      "learning_rate": 2.8419813574826093e-05,
      "loss": 0.06335252523422241,
      "step": 1528
    },
    {
      "epoch": 0.2067106608420857,
      "grad_norm": 1.1274170875549316,
      "learning_rate": 2.8416754366435092e-05,
      "loss": 0.05677357316017151,
      "step": 1529
    },
    {
      "epoch": 0.20684585421085097,
      "grad_norm": 0.5897433161735535,
      "learning_rate": 2.8413692364604285e-05,
      "loss": 0.05167311057448387,
      "step": 1530
    },
    {
      "epoch": 0.20698104757961622,
      "grad_norm": 0.7777832746505737,
      "learning_rate": 2.8410627569971197e-05,
      "loss": 0.054501548409461975,
      "step": 1531
    },
    {
      "epoch": 0.2071162409483815,
      "grad_norm": 0.8203860521316528,
      "learning_rate": 2.8407559983173934e-05,
      "loss": 0.0593453012406826,
      "step": 1532
    },
    {
      "epoch": 0.20725143431714674,
      "grad_norm": 0.7851532101631165,
      "learning_rate": 2.8404489604851186e-05,
      "loss": 0.07835626602172852,
      "step": 1533
    },
    {
      "epoch": 0.207386627685912,
      "grad_norm": 0.6273208856582642,
      "learning_rate": 2.840141643564222e-05,
      "loss": 0.06668758392333984,
      "step": 1534
    },
    {
      "epoch": 0.20752182105467726,
      "grad_norm": 0.8988673090934753,
      "learning_rate": 2.8398340476186885e-05,
      "loss": 0.0568954274058342,
      "step": 1535
    },
    {
      "epoch": 0.20765701442344253,
      "grad_norm": 0.3905712962150574,
      "learning_rate": 2.8395261727125617e-05,
      "loss": 0.04055122286081314,
      "step": 1536
    },
    {
      "epoch": 0.2077922077922078,
      "grad_norm": 1.662516474723816,
      "learning_rate": 2.8392180189099425e-05,
      "loss": 0.04748539626598358,
      "step": 1537
    },
    {
      "epoch": 0.20792740116097305,
      "grad_norm": 1.5907397270202637,
      "learning_rate": 2.83890958627499e-05,
      "loss": 0.06497156620025635,
      "step": 1538
    },
    {
      "epoch": 0.20806259452973833,
      "grad_norm": 1.8041553497314453,
      "learning_rate": 2.8386008748719216e-05,
      "loss": 0.06606616824865341,
      "step": 1539
    },
    {
      "epoch": 0.20819778789850357,
      "grad_norm": 0.3390914499759674,
      "learning_rate": 2.838291884765013e-05,
      "loss": 0.056449294090270996,
      "step": 1540
    },
    {
      "epoch": 0.20833298126726885,
      "grad_norm": 0.8439186215400696,
      "learning_rate": 2.8379826160185975e-05,
      "loss": 0.053850337862968445,
      "step": 1541
    },
    {
      "epoch": 0.2084681746360341,
      "grad_norm": 0.4911138713359833,
      "learning_rate": 2.8376730686970664e-05,
      "loss": 0.0673052966594696,
      "step": 1542
    },
    {
      "epoch": 0.20860336800479937,
      "grad_norm": 1.1656914949417114,
      "learning_rate": 2.8373632428648683e-05,
      "loss": 0.0522618293762207,
      "step": 1543
    },
    {
      "epoch": 0.2087385613735646,
      "grad_norm": 0.8624969720840454,
      "learning_rate": 2.8370531385865124e-05,
      "loss": 0.07638722658157349,
      "step": 1544
    },
    {
      "epoch": 0.2088737547423299,
      "grad_norm": 0.8319230675697327,
      "learning_rate": 2.8367427559265622e-05,
      "loss": 0.0550997257232666,
      "step": 1545
    },
    {
      "epoch": 0.20900894811109516,
      "grad_norm": 0.6036023497581482,
      "learning_rate": 2.836432094949642e-05,
      "loss": 0.051479071378707886,
      "step": 1546
    },
    {
      "epoch": 0.2091441414798604,
      "grad_norm": 0.4117427468299866,
      "learning_rate": 2.836121155720433e-05,
      "loss": 0.0465034544467926,
      "step": 1547
    },
    {
      "epoch": 0.20927933484862568,
      "grad_norm": 0.3497418761253357,
      "learning_rate": 2.8358099383036745e-05,
      "loss": 0.052825331687927246,
      "step": 1548
    },
    {
      "epoch": 0.20941452821739093,
      "grad_norm": 0.4250861704349518,
      "learning_rate": 2.8354984427641634e-05,
      "loss": 0.04520291090011597,
      "step": 1549
    },
    {
      "epoch": 0.2095497215861562,
      "grad_norm": 0.9436564445495605,
      "learning_rate": 2.8351866691667544e-05,
      "loss": 0.061111681163311005,
      "step": 1550
    },
    {
      "epoch": 0.20968491495492145,
      "grad_norm": 0.8618022799491882,
      "learning_rate": 2.8348746175763613e-05,
      "loss": 0.0505509227514267,
      "step": 1551
    },
    {
      "epoch": 0.20982010832368672,
      "grad_norm": 0.30750373005867004,
      "learning_rate": 2.8345622880579537e-05,
      "loss": 0.048603713512420654,
      "step": 1552
    },
    {
      "epoch": 0.209955301692452,
      "grad_norm": 0.40893998742103577,
      "learning_rate": 2.8342496806765615e-05,
      "loss": 0.049013957381248474,
      "step": 1553
    },
    {
      "epoch": 0.21009049506121724,
      "grad_norm": 0.5988844633102417,
      "learning_rate": 2.833936795497271e-05,
      "loss": 0.045951008796691895,
      "step": 1554
    },
    {
      "epoch": 0.21022568842998252,
      "grad_norm": 0.9709116220474243,
      "learning_rate": 2.8336236325852257e-05,
      "loss": 0.05929470434784889,
      "step": 1555
    },
    {
      "epoch": 0.21036088179874776,
      "grad_norm": 0.5662141442298889,
      "learning_rate": 2.8333101920056285e-05,
      "loss": 0.07690741866827011,
      "step": 1556
    },
    {
      "epoch": 0.21049607516751304,
      "grad_norm": 0.5925481915473938,
      "learning_rate": 2.8329964738237392e-05,
      "loss": 0.04631420969963074,
      "step": 1557
    },
    {
      "epoch": 0.21063126853627828,
      "grad_norm": 0.6251190900802612,
      "learning_rate": 2.8326824781048756e-05,
      "loss": 0.052150797098875046,
      "step": 1558
    },
    {
      "epoch": 0.21076646190504356,
      "grad_norm": 0.700613796710968,
      "learning_rate": 2.8323682049144135e-05,
      "loss": 0.0730980932712555,
      "step": 1559
    },
    {
      "epoch": 0.21090165527380883,
      "grad_norm": 0.8474962115287781,
      "learning_rate": 2.832053654317786e-05,
      "loss": 0.05908976495265961,
      "step": 1560
    },
    {
      "epoch": 0.21103684864257408,
      "grad_norm": 1.0694419145584106,
      "learning_rate": 2.8317388263804842e-05,
      "loss": 0.050022710114717484,
      "step": 1561
    },
    {
      "epoch": 0.21117204201133935,
      "grad_norm": 1.1018961668014526,
      "learning_rate": 2.8314237211680573e-05,
      "loss": 0.06538151204586029,
      "step": 1562
    },
    {
      "epoch": 0.2113072353801046,
      "grad_norm": 1.3340040445327759,
      "learning_rate": 2.8311083387461118e-05,
      "loss": 0.06075182557106018,
      "step": 1563
    },
    {
      "epoch": 0.21144242874886987,
      "grad_norm": 0.39237844944000244,
      "learning_rate": 2.8307926791803114e-05,
      "loss": 0.04414108395576477,
      "step": 1564
    },
    {
      "epoch": 0.21157762211763512,
      "grad_norm": 0.8417506217956543,
      "learning_rate": 2.8304767425363785e-05,
      "loss": 0.07103466987609863,
      "step": 1565
    },
    {
      "epoch": 0.2117128154864004,
      "grad_norm": 0.7777599692344666,
      "learning_rate": 2.830160528880093e-05,
      "loss": 0.05266956984996796,
      "step": 1566
    },
    {
      "epoch": 0.21184800885516566,
      "grad_norm": 1.6419627666473389,
      "learning_rate": 2.829844038277292e-05,
      "loss": 0.055458784103393555,
      "step": 1567
    },
    {
      "epoch": 0.2119832022239309,
      "grad_norm": 1.389513611793518,
      "learning_rate": 2.8295272707938706e-05,
      "loss": 0.07247611880302429,
      "step": 1568
    },
    {
      "epoch": 0.21211839559269619,
      "grad_norm": 0.29261651635169983,
      "learning_rate": 2.8292102264957817e-05,
      "loss": 0.051503024995326996,
      "step": 1569
    },
    {
      "epoch": 0.21225358896146143,
      "grad_norm": 1.0491795539855957,
      "learning_rate": 2.8288929054490357e-05,
      "loss": 0.055749163031578064,
      "step": 1570
    },
    {
      "epoch": 0.2123887823302267,
      "grad_norm": 0.7268002033233643,
      "learning_rate": 2.8285753077196998e-05,
      "loss": 0.05362197756767273,
      "step": 1571
    },
    {
      "epoch": 0.21252397569899195,
      "grad_norm": 0.41287580132484436,
      "learning_rate": 2.8282574333739006e-05,
      "loss": 0.05210597813129425,
      "step": 1572
    },
    {
      "epoch": 0.21265916906775723,
      "grad_norm": 0.5221613645553589,
      "learning_rate": 2.8279392824778197e-05,
      "loss": 0.06534935534000397,
      "step": 1573
    },
    {
      "epoch": 0.2127943624365225,
      "grad_norm": 0.9850326180458069,
      "learning_rate": 2.8276208550976993e-05,
      "loss": 0.04884040355682373,
      "step": 1574
    },
    {
      "epoch": 0.21292955580528775,
      "grad_norm": 0.43634629249572754,
      "learning_rate": 2.8273021512998372e-05,
      "loss": 0.047863513231277466,
      "step": 1575
    },
    {
      "epoch": 0.21306474917405302,
      "grad_norm": 0.9789410829544067,
      "learning_rate": 2.826983171150589e-05,
      "loss": 0.08013477921485901,
      "step": 1576
    },
    {
      "epoch": 0.21319994254281827,
      "grad_norm": 0.37938928604125977,
      "learning_rate": 2.826663914716368e-05,
      "loss": 0.06604525446891785,
      "step": 1577
    },
    {
      "epoch": 0.21333513591158354,
      "grad_norm": 0.5171474814414978,
      "learning_rate": 2.826344382063646e-05,
      "loss": 0.053354568779468536,
      "step": 1578
    },
    {
      "epoch": 0.2134703292803488,
      "grad_norm": 0.9365546703338623,
      "learning_rate": 2.8260245732589503e-05,
      "loss": 0.06443160772323608,
      "step": 1579
    },
    {
      "epoch": 0.21360552264911406,
      "grad_norm": 0.5557537078857422,
      "learning_rate": 2.8257044883688672e-05,
      "loss": 0.06397908180952072,
      "step": 1580
    },
    {
      "epoch": 0.21374071601787933,
      "grad_norm": 0.6537729501724243,
      "learning_rate": 2.82538412746004e-05,
      "loss": 0.07449346035718918,
      "step": 1581
    },
    {
      "epoch": 0.21387590938664458,
      "grad_norm": 1.3514331579208374,
      "learning_rate": 2.8250634905991695e-05,
      "loss": 0.06255114078521729,
      "step": 1582
    },
    {
      "epoch": 0.21401110275540985,
      "grad_norm": 0.7896385192871094,
      "learning_rate": 2.824742577853015e-05,
      "loss": 0.05204898118972778,
      "step": 1583
    },
    {
      "epoch": 0.2141462961241751,
      "grad_norm": 0.33208876848220825,
      "learning_rate": 2.8244213892883907e-05,
      "loss": 0.04208874702453613,
      "step": 1584
    },
    {
      "epoch": 0.21428148949294037,
      "grad_norm": 1.2684580087661743,
      "learning_rate": 2.82409992497217e-05,
      "loss": 0.06291976571083069,
      "step": 1585
    },
    {
      "epoch": 0.21441668286170562,
      "grad_norm": 0.4156642556190491,
      "learning_rate": 2.8237781849712852e-05,
      "loss": 0.043654412031173706,
      "step": 1586
    },
    {
      "epoch": 0.2145518762304709,
      "grad_norm": 1.0025794506072998,
      "learning_rate": 2.8234561693527222e-05,
      "loss": 0.06891020387411118,
      "step": 1587
    },
    {
      "epoch": 0.21468706959923617,
      "grad_norm": 0.40402752161026,
      "learning_rate": 2.8231338781835275e-05,
      "loss": 0.05821501463651657,
      "step": 1588
    },
    {
      "epoch": 0.21482226296800142,
      "grad_norm": 1.179532766342163,
      "learning_rate": 2.8228113115308032e-05,
      "loss": 0.04632169008255005,
      "step": 1589
    },
    {
      "epoch": 0.2149574563367667,
      "grad_norm": 0.5531177520751953,
      "learning_rate": 2.82248846946171e-05,
      "loss": 0.050778210163116455,
      "step": 1590
    },
    {
      "epoch": 0.21509264970553194,
      "grad_norm": 0.7919912338256836,
      "learning_rate": 2.822165352043465e-05,
      "loss": 0.051515668630599976,
      "step": 1591
    },
    {
      "epoch": 0.2152278430742972,
      "grad_norm": 0.30874982476234436,
      "learning_rate": 2.8218419593433437e-05,
      "loss": 0.05695490539073944,
      "step": 1592
    },
    {
      "epoch": 0.21536303644306246,
      "grad_norm": 0.47324231266975403,
      "learning_rate": 2.8215182914286768e-05,
      "loss": 0.051553875207901,
      "step": 1593
    },
    {
      "epoch": 0.21549822981182773,
      "grad_norm": 0.4900946617126465,
      "learning_rate": 2.8211943483668546e-05,
      "loss": 0.07251714169979095,
      "step": 1594
    },
    {
      "epoch": 0.215633423180593,
      "grad_norm": 1.1118718385696411,
      "learning_rate": 2.8208701302253237e-05,
      "loss": 0.07170483469963074,
      "step": 1595
    },
    {
      "epoch": 0.21576861654935825,
      "grad_norm": 1.024431824684143,
      "learning_rate": 2.820545637071588e-05,
      "loss": 0.040563106536865234,
      "step": 1596
    },
    {
      "epoch": 0.21590380991812352,
      "grad_norm": 1.1180799007415771,
      "learning_rate": 2.8202208689732083e-05,
      "loss": 0.05583508312702179,
      "step": 1597
    },
    {
      "epoch": 0.21603900328688877,
      "grad_norm": 1.9988240003585815,
      "learning_rate": 2.819895825997804e-05,
      "loss": 0.05353221297264099,
      "step": 1598
    },
    {
      "epoch": 0.21617419665565404,
      "grad_norm": 1.0622143745422363,
      "learning_rate": 2.81957050821305e-05,
      "loss": 0.05039858818054199,
      "step": 1599
    },
    {
      "epoch": 0.2163093900244193,
      "grad_norm": 1.1034172773361206,
      "learning_rate": 2.8192449156866787e-05,
      "loss": 0.04590676724910736,
      "step": 1600
    },
    {
      "epoch": 0.21644458339318456,
      "grad_norm": 0.7669665217399597,
      "learning_rate": 2.8189190484864814e-05,
      "loss": 0.05868140608072281,
      "step": 1601
    },
    {
      "epoch": 0.21657977676194984,
      "grad_norm": 0.396586537361145,
      "learning_rate": 2.8185929066803052e-05,
      "loss": 0.0575755313038826,
      "step": 1602
    },
    {
      "epoch": 0.21671497013071508,
      "grad_norm": 1.191956877708435,
      "learning_rate": 2.818266490336054e-05,
      "loss": 0.036856234073638916,
      "step": 1603
    },
    {
      "epoch": 0.21685016349948036,
      "grad_norm": 1.4901243448257446,
      "learning_rate": 2.817939799521689e-05,
      "loss": 0.0770600438117981,
      "step": 1604
    },
    {
      "epoch": 0.2169853568682456,
      "grad_norm": 2.492192268371582,
      "learning_rate": 2.8176128343052304e-05,
      "loss": 0.06851261854171753,
      "step": 1605
    },
    {
      "epoch": 0.21712055023701088,
      "grad_norm": 0.568828284740448,
      "learning_rate": 2.817285594754753e-05,
      "loss": 0.04781460762023926,
      "step": 1606
    },
    {
      "epoch": 0.21725574360577612,
      "grad_norm": 0.9603391885757446,
      "learning_rate": 2.8169580809383902e-05,
      "loss": 0.05101378262042999,
      "step": 1607
    },
    {
      "epoch": 0.2173909369745414,
      "grad_norm": 0.4263308644294739,
      "learning_rate": 2.8166302929243326e-05,
      "loss": 0.052577219903469086,
      "step": 1608
    },
    {
      "epoch": 0.21752613034330667,
      "grad_norm": 1.386176347732544,
      "learning_rate": 2.8163022307808264e-05,
      "loss": 0.06691841036081314,
      "step": 1609
    },
    {
      "epoch": 0.21766132371207192,
      "grad_norm": 0.5585028529167175,
      "learning_rate": 2.8159738945761764e-05,
      "loss": 0.04208102077245712,
      "step": 1610
    },
    {
      "epoch": 0.2177965170808372,
      "grad_norm": 0.48720434308052063,
      "learning_rate": 2.8156452843787438e-05,
      "loss": 0.05007272958755493,
      "step": 1611
    },
    {
      "epoch": 0.21793171044960244,
      "grad_norm": 0.8018759489059448,
      "learning_rate": 2.815316400256947e-05,
      "loss": 0.07517276704311371,
      "step": 1612
    },
    {
      "epoch": 0.2180669038183677,
      "grad_norm": 0.680761992931366,
      "learning_rate": 2.814987242279262e-05,
      "loss": 0.06988190114498138,
      "step": 1613
    },
    {
      "epoch": 0.21820209718713296,
      "grad_norm": 0.3617783188819885,
      "learning_rate": 2.8146578105142202e-05,
      "loss": 0.0472244918346405,
      "step": 1614
    },
    {
      "epoch": 0.21833729055589823,
      "grad_norm": 1.3154038190841675,
      "learning_rate": 2.814328105030412e-05,
      "loss": 0.05513004958629608,
      "step": 1615
    },
    {
      "epoch": 0.2184724839246635,
      "grad_norm": 1.6053988933563232,
      "learning_rate": 2.8139981258964836e-05,
      "loss": 0.07096618413925171,
      "step": 1616
    },
    {
      "epoch": 0.21860767729342875,
      "grad_norm": 0.7854763269424438,
      "learning_rate": 2.8136678731811385e-05,
      "loss": 0.06544134020805359,
      "step": 1617
    },
    {
      "epoch": 0.21874287066219403,
      "grad_norm": 1.5256870985031128,
      "learning_rate": 2.8133373469531362e-05,
      "loss": 0.0759010910987854,
      "step": 1618
    },
    {
      "epoch": 0.21887806403095927,
      "grad_norm": 1.4773973226547241,
      "learning_rate": 2.8130065472812952e-05,
      "loss": 0.05293330177664757,
      "step": 1619
    },
    {
      "epoch": 0.21901325739972455,
      "grad_norm": 0.6383475065231323,
      "learning_rate": 2.812675474234489e-05,
      "loss": 0.06383223831653595,
      "step": 1620
    },
    {
      "epoch": 0.2191484507684898,
      "grad_norm": 0.7026002407073975,
      "learning_rate": 2.812344127881649e-05,
      "loss": 0.055020660161972046,
      "step": 1621
    },
    {
      "epoch": 0.21928364413725507,
      "grad_norm": 1.5413810014724731,
      "learning_rate": 2.8120125082917638e-05,
      "loss": 0.051977306604385376,
      "step": 1622
    },
    {
      "epoch": 0.21941883750602034,
      "grad_norm": 0.6406027674674988,
      "learning_rate": 2.8116806155338773e-05,
      "loss": 0.07189971208572388,
      "step": 1623
    },
    {
      "epoch": 0.2195540308747856,
      "grad_norm": 1.2294435501098633,
      "learning_rate": 2.8113484496770923e-05,
      "loss": 0.06025470793247223,
      "step": 1624
    },
    {
      "epoch": 0.21968922424355086,
      "grad_norm": 0.6029508709907532,
      "learning_rate": 2.811016010790567e-05,
      "loss": 0.05814117193222046,
      "step": 1625
    },
    {
      "epoch": 0.2198244176123161,
      "grad_norm": 1.443228006362915,
      "learning_rate": 2.8106832989435165e-05,
      "loss": 0.07699942588806152,
      "step": 1626
    },
    {
      "epoch": 0.21995961098108138,
      "grad_norm": 0.37957221269607544,
      "learning_rate": 2.8103503142052146e-05,
      "loss": 0.05255141854286194,
      "step": 1627
    },
    {
      "epoch": 0.22009480434984663,
      "grad_norm": 0.9441279768943787,
      "learning_rate": 2.8100170566449892e-05,
      "loss": 0.04794152081012726,
      "step": 1628
    },
    {
      "epoch": 0.2202299977186119,
      "grad_norm": 0.3265136182308197,
      "learning_rate": 2.8096835263322266e-05,
      "loss": 0.0478270947933197,
      "step": 1629
    },
    {
      "epoch": 0.22036519108737718,
      "grad_norm": 0.4508991241455078,
      "learning_rate": 2.8093497233363702e-05,
      "loss": 0.046484291553497314,
      "step": 1630
    },
    {
      "epoch": 0.22050038445614242,
      "grad_norm": 0.9232129454612732,
      "learning_rate": 2.8090156477269185e-05,
      "loss": 0.054828643798828125,
      "step": 1631
    },
    {
      "epoch": 0.2206355778249077,
      "grad_norm": 0.8288602828979492,
      "learning_rate": 2.808681299573429e-05,
      "loss": 0.04206643998622894,
      "step": 1632
    },
    {
      "epoch": 0.22077077119367294,
      "grad_norm": 1.2797791957855225,
      "learning_rate": 2.8083466789455137e-05,
      "loss": 0.08406156301498413,
      "step": 1633
    },
    {
      "epoch": 0.22090596456243822,
      "grad_norm": 0.3598003685474396,
      "learning_rate": 2.808011785912843e-05,
      "loss": 0.04149273782968521,
      "step": 1634
    },
    {
      "epoch": 0.22104115793120346,
      "grad_norm": 0.6481890678405762,
      "learning_rate": 2.8076766205451435e-05,
      "loss": 0.05434077978134155,
      "step": 1635
    },
    {
      "epoch": 0.22117635129996874,
      "grad_norm": 1.588051676750183,
      "learning_rate": 2.8073411829121983e-05,
      "loss": 0.05897385999560356,
      "step": 1636
    },
    {
      "epoch": 0.221311544668734,
      "grad_norm": 0.5865188837051392,
      "learning_rate": 2.8070054730838467e-05,
      "loss": 0.05248776078224182,
      "step": 1637
    },
    {
      "epoch": 0.22144673803749926,
      "grad_norm": 1.468625545501709,
      "learning_rate": 2.8066694911299865e-05,
      "loss": 0.07838153839111328,
      "step": 1638
    },
    {
      "epoch": 0.22158193140626453,
      "grad_norm": 1.1525522470474243,
      "learning_rate": 2.8063332371205698e-05,
      "loss": 0.05041515827178955,
      "step": 1639
    },
    {
      "epoch": 0.22171712477502978,
      "grad_norm": 1.5627365112304688,
      "learning_rate": 2.8059967111256072e-05,
      "loss": 0.06059056520462036,
      "step": 1640
    },
    {
      "epoch": 0.22185231814379505,
      "grad_norm": 0.6790928244590759,
      "learning_rate": 2.8056599132151647e-05,
      "loss": 0.06592552363872528,
      "step": 1641
    },
    {
      "epoch": 0.2219875115125603,
      "grad_norm": 0.33519986271858215,
      "learning_rate": 2.8053228434593656e-05,
      "loss": 0.06335578113794327,
      "step": 1642
    },
    {
      "epoch": 0.22212270488132557,
      "grad_norm": 0.39631593227386475,
      "learning_rate": 2.8049855019283895e-05,
      "loss": 0.052461713552474976,
      "step": 1643
    },
    {
      "epoch": 0.22225789825009085,
      "grad_norm": 0.6668994426727295,
      "learning_rate": 2.8046478886924736e-05,
      "loss": 0.07242193818092346,
      "step": 1644
    },
    {
      "epoch": 0.2223930916188561,
      "grad_norm": 2.2576355934143066,
      "learning_rate": 2.804310003821909e-05,
      "loss": 0.0729055404663086,
      "step": 1645
    },
    {
      "epoch": 0.22252828498762137,
      "grad_norm": 0.3832198679447174,
      "learning_rate": 2.8039718473870473e-05,
      "loss": 0.05689685046672821,
      "step": 1646
    },
    {
      "epoch": 0.2226634783563866,
      "grad_norm": 0.8421013951301575,
      "learning_rate": 2.8036334194582924e-05,
      "loss": 0.06981846690177917,
      "step": 1647
    },
    {
      "epoch": 0.2227986717251519,
      "grad_norm": 0.9113370180130005,
      "learning_rate": 2.8032947201061084e-05,
      "loss": 0.05107051134109497,
      "step": 1648
    },
    {
      "epoch": 0.22293386509391713,
      "grad_norm": 0.7018494009971619,
      "learning_rate": 2.8029557494010132e-05,
      "loss": 0.07083918154239655,
      "step": 1649
    },
    {
      "epoch": 0.2230690584626824,
      "grad_norm": 1.2353463172912598,
      "learning_rate": 2.802616507413583e-05,
      "loss": 0.057971615344285965,
      "step": 1650
    },
    {
      "epoch": 0.22320425183144768,
      "grad_norm": 0.6947340369224548,
      "learning_rate": 2.8022769942144492e-05,
      "loss": 0.059219732880592346,
      "step": 1651
    },
    {
      "epoch": 0.22333944520021293,
      "grad_norm": 0.6718868613243103,
      "learning_rate": 2.801937209874301e-05,
      "loss": 0.053752824664115906,
      "step": 1652
    },
    {
      "epoch": 0.2234746385689782,
      "grad_norm": 1.9309371709823608,
      "learning_rate": 2.8015971544638832e-05,
      "loss": 0.08076757192611694,
      "step": 1653
    },
    {
      "epoch": 0.22360983193774345,
      "grad_norm": 1.4071406126022339,
      "learning_rate": 2.8012568280539964e-05,
      "loss": 0.06257554888725281,
      "step": 1654
    },
    {
      "epoch": 0.22374502530650872,
      "grad_norm": 0.9768237471580505,
      "learning_rate": 2.800916230715499e-05,
      "loss": 0.06816723942756653,
      "step": 1655
    },
    {
      "epoch": 0.22388021867527397,
      "grad_norm": 1.7064036130905151,
      "learning_rate": 2.800575362519305e-05,
      "loss": 0.06085750460624695,
      "step": 1656
    },
    {
      "epoch": 0.22401541204403924,
      "grad_norm": 1.0066249370574951,
      "learning_rate": 2.800234223536385e-05,
      "loss": 0.06367231905460358,
      "step": 1657
    },
    {
      "epoch": 0.22415060541280452,
      "grad_norm": 0.6331930160522461,
      "learning_rate": 2.799892813837766e-05,
      "loss": 0.052688345313072205,
      "step": 1658
    },
    {
      "epoch": 0.22428579878156976,
      "grad_norm": 0.5558263063430786,
      "learning_rate": 2.7995511334945315e-05,
      "loss": 0.040113601833581924,
      "step": 1659
    },
    {
      "epoch": 0.22442099215033504,
      "grad_norm": 0.5435409545898438,
      "learning_rate": 2.7992091825778202e-05,
      "loss": 0.05569447576999664,
      "step": 1660
    },
    {
      "epoch": 0.22455618551910028,
      "grad_norm": 1.3237699270248413,
      "learning_rate": 2.7988669611588295e-05,
      "loss": 0.06360754370689392,
      "step": 1661
    },
    {
      "epoch": 0.22469137888786556,
      "grad_norm": 0.49552586674690247,
      "learning_rate": 2.7985244693088112e-05,
      "loss": 0.06739620119333267,
      "step": 1662
    },
    {
      "epoch": 0.2248265722566308,
      "grad_norm": 0.3336208164691925,
      "learning_rate": 2.7981817070990736e-05,
      "loss": 0.057138100266456604,
      "step": 1663
    },
    {
      "epoch": 0.22496176562539608,
      "grad_norm": 0.6160723567008972,
      "learning_rate": 2.7978386746009813e-05,
      "loss": 0.06904512643814087,
      "step": 1664
    },
    {
      "epoch": 0.22509695899416135,
      "grad_norm": 0.9872597455978394,
      "learning_rate": 2.797495371885957e-05,
      "loss": 0.06188105046749115,
      "step": 1665
    },
    {
      "epoch": 0.2252321523629266,
      "grad_norm": 0.9342311024665833,
      "learning_rate": 2.7971517990254768e-05,
      "loss": 0.07091188430786133,
      "step": 1666
    },
    {
      "epoch": 0.22536734573169187,
      "grad_norm": 0.7085817456245422,
      "learning_rate": 2.7968079560910744e-05,
      "loss": 0.05051851272583008,
      "step": 1667
    },
    {
      "epoch": 0.22550253910045712,
      "grad_norm": 0.6815599203109741,
      "learning_rate": 2.7964638431543402e-05,
      "loss": 0.030333321541547775,
      "step": 1668
    },
    {
      "epoch": 0.2256377324692224,
      "grad_norm": 1.7448046207427979,
      "learning_rate": 2.7961194602869208e-05,
      "loss": 0.05529448390007019,
      "step": 1669
    },
    {
      "epoch": 0.22577292583798764,
      "grad_norm": 1.0418426990509033,
      "learning_rate": 2.7957748075605178e-05,
      "loss": 0.09649249911308289,
      "step": 1670
    },
    {
      "epoch": 0.2259081192067529,
      "grad_norm": 1.0943551063537598,
      "learning_rate": 2.7954298850468898e-05,
      "loss": 0.046424269676208496,
      "step": 1671
    },
    {
      "epoch": 0.22604331257551816,
      "grad_norm": 1.1015396118164062,
      "learning_rate": 2.7950846928178517e-05,
      "loss": 0.05632650852203369,
      "step": 1672
    },
    {
      "epoch": 0.22617850594428343,
      "grad_norm": 1.5949186086654663,
      "learning_rate": 2.7947392309452744e-05,
      "loss": 0.06467847526073456,
      "step": 1673
    },
    {
      "epoch": 0.2263136993130487,
      "grad_norm": 0.7427306771278381,
      "learning_rate": 2.7943934995010845e-05,
      "loss": 0.06725446879863739,
      "step": 1674
    },
    {
      "epoch": 0.22644889268181395,
      "grad_norm": 0.7063242793083191,
      "learning_rate": 2.7940474985572657e-05,
      "loss": 0.06955593824386597,
      "step": 1675
    },
    {
      "epoch": 0.22658408605057923,
      "grad_norm": 1.926409363746643,
      "learning_rate": 2.793701228185857e-05,
      "loss": 0.07470649480819702,
      "step": 1676
    },
    {
      "epoch": 0.22671927941934447,
      "grad_norm": 1.3241316080093384,
      "learning_rate": 2.7933546884589536e-05,
      "loss": 0.067235067486763,
      "step": 1677
    },
    {
      "epoch": 0.22685447278810975,
      "grad_norm": 1.6580857038497925,
      "learning_rate": 2.7930078794487077e-05,
      "loss": 0.05165666341781616,
      "step": 1678
    },
    {
      "epoch": 0.226989666156875,
      "grad_norm": 0.7732198238372803,
      "learning_rate": 2.7926608012273253e-05,
      "loss": 0.07548081874847412,
      "step": 1679
    },
    {
      "epoch": 0.22712485952564027,
      "grad_norm": 1.19050133228302,
      "learning_rate": 2.7923134538670715e-05,
      "loss": 0.09109258651733398,
      "step": 1680
    },
    {
      "epoch": 0.22726005289440554,
      "grad_norm": 0.34316375851631165,
      "learning_rate": 2.7919658374402645e-05,
      "loss": 0.06333279609680176,
      "step": 1681
    },
    {
      "epoch": 0.22739524626317079,
      "grad_norm": 0.8293101191520691,
      "learning_rate": 2.7916179520192807e-05,
      "loss": 0.06527724862098694,
      "step": 1682
    },
    {
      "epoch": 0.22753043963193606,
      "grad_norm": 1.1587144136428833,
      "learning_rate": 2.7912697976765516e-05,
      "loss": 0.07280510663986206,
      "step": 1683
    },
    {
      "epoch": 0.2276656330007013,
      "grad_norm": 0.510677695274353,
      "learning_rate": 2.790921374484565e-05,
      "loss": 0.06580553203821182,
      "step": 1684
    },
    {
      "epoch": 0.22780082636946658,
      "grad_norm": 0.9677261114120483,
      "learning_rate": 2.7905726825158637e-05,
      "loss": 0.05581563711166382,
      "step": 1685
    },
    {
      "epoch": 0.22793601973823183,
      "grad_norm": 0.2613687515258789,
      "learning_rate": 2.7902237218430485e-05,
      "loss": 0.04570366442203522,
      "step": 1686
    },
    {
      "epoch": 0.2280712131069971,
      "grad_norm": 0.4171440899372101,
      "learning_rate": 2.7898744925387735e-05,
      "loss": 0.04238733649253845,
      "step": 1687
    },
    {
      "epoch": 0.22820640647576237,
      "grad_norm": 0.946252167224884,
      "learning_rate": 2.7895249946757505e-05,
      "loss": 0.0751807689666748,
      "step": 1688
    },
    {
      "epoch": 0.22834159984452762,
      "grad_norm": 0.36497074365615845,
      "learning_rate": 2.7891752283267474e-05,
      "loss": 0.052467137575149536,
      "step": 1689
    },
    {
      "epoch": 0.2284767932132929,
      "grad_norm": 0.43665435910224915,
      "learning_rate": 2.788825193564587e-05,
      "loss": 0.05565465986728668,
      "step": 1690
    },
    {
      "epoch": 0.22861198658205814,
      "grad_norm": 0.5362083315849304,
      "learning_rate": 2.7884748904621483e-05,
      "loss": 0.058127254247665405,
      "step": 1691
    },
    {
      "epoch": 0.22874717995082341,
      "grad_norm": 1.1373850107192993,
      "learning_rate": 2.7881243190923667e-05,
      "loss": 0.0649629458785057,
      "step": 1692
    },
    {
      "epoch": 0.22888237331958866,
      "grad_norm": 0.5474332571029663,
      "learning_rate": 2.7877734795282326e-05,
      "loss": 0.04449063912034035,
      "step": 1693
    },
    {
      "epoch": 0.22901756668835394,
      "grad_norm": 0.5453364849090576,
      "learning_rate": 2.7874223718427926e-05,
      "loss": 0.07171127945184708,
      "step": 1694
    },
    {
      "epoch": 0.2291527600571192,
      "grad_norm": 1.0143977403640747,
      "learning_rate": 2.78707099610915e-05,
      "loss": 0.056407347321510315,
      "step": 1695
    },
    {
      "epoch": 0.22928795342588446,
      "grad_norm": 0.5908769369125366,
      "learning_rate": 2.7867193524004618e-05,
      "loss": 0.06064420938491821,
      "step": 1696
    },
    {
      "epoch": 0.22942314679464973,
      "grad_norm": 0.8280333280563354,
      "learning_rate": 2.786367440789943e-05,
      "loss": 0.059423357248306274,
      "step": 1697
    },
    {
      "epoch": 0.22955834016341498,
      "grad_norm": 0.5103542804718018,
      "learning_rate": 2.7860152613508634e-05,
      "loss": 0.07177874445915222,
      "step": 1698
    },
    {
      "epoch": 0.22969353353218025,
      "grad_norm": 0.542658805847168,
      "learning_rate": 2.7856628141565484e-05,
      "loss": 0.0574946254491806,
      "step": 1699
    },
    {
      "epoch": 0.2298287269009455,
      "grad_norm": 0.5609861612319946,
      "learning_rate": 2.7853100992803797e-05,
      "loss": 0.05437430739402771,
      "step": 1700
    },
    {
      "epoch": 0.22996392026971077,
      "grad_norm": 1.6992119550704956,
      "learning_rate": 2.7849571167957942e-05,
      "loss": 0.07060274481773376,
      "step": 1701
    },
    {
      "epoch": 0.23009911363847604,
      "grad_norm": 0.4380677044391632,
      "learning_rate": 2.784603866776285e-05,
      "loss": 0.06999701261520386,
      "step": 1702
    },
    {
      "epoch": 0.2302343070072413,
      "grad_norm": 1.9180126190185547,
      "learning_rate": 2.7842503492953996e-05,
      "loss": 0.06108155846595764,
      "step": 1703
    },
    {
      "epoch": 0.23036950037600656,
      "grad_norm": 0.9410768151283264,
      "learning_rate": 2.7838965644267435e-05,
      "loss": 0.06441255658864975,
      "step": 1704
    },
    {
      "epoch": 0.2305046937447718,
      "grad_norm": 0.7353812456130981,
      "learning_rate": 2.7835425122439764e-05,
      "loss": 0.05866764485836029,
      "step": 1705
    },
    {
      "epoch": 0.23063988711353708,
      "grad_norm": 0.787632405757904,
      "learning_rate": 2.7831881928208128e-05,
      "loss": 0.05790655314922333,
      "step": 1706
    },
    {
      "epoch": 0.23077508048230233,
      "grad_norm": 0.43534642457962036,
      "learning_rate": 2.7828336062310252e-05,
      "loss": 0.056324705481529236,
      "step": 1707
    },
    {
      "epoch": 0.2309102738510676,
      "grad_norm": 1.1952860355377197,
      "learning_rate": 2.7824787525484403e-05,
      "loss": 0.05890044569969177,
      "step": 1708
    },
    {
      "epoch": 0.23104546721983288,
      "grad_norm": 2.0317039489746094,
      "learning_rate": 2.7821236318469395e-05,
      "loss": 0.0760083720088005,
      "step": 1709
    },
    {
      "epoch": 0.23118066058859812,
      "grad_norm": 1.2457159757614136,
      "learning_rate": 2.7817682442004615e-05,
      "loss": 0.07557854056358337,
      "step": 1710
    },
    {
      "epoch": 0.2313158539573634,
      "grad_norm": 0.46890324354171753,
      "learning_rate": 2.781412589683e-05,
      "loss": 0.06290195882320404,
      "step": 1711
    },
    {
      "epoch": 0.23145104732612864,
      "grad_norm": 0.507620632648468,
      "learning_rate": 2.781056668368604e-05,
      "loss": 0.0546550378203392,
      "step": 1712
    },
    {
      "epoch": 0.23158624069489392,
      "grad_norm": 0.7180529236793518,
      "learning_rate": 2.780700480331378e-05,
      "loss": 0.04927615821361542,
      "step": 1713
    },
    {
      "epoch": 0.23172143406365917,
      "grad_norm": 0.559109091758728,
      "learning_rate": 2.7803440256454825e-05,
      "loss": 0.08365112543106079,
      "step": 1714
    },
    {
      "epoch": 0.23185662743242444,
      "grad_norm": 0.7803539037704468,
      "learning_rate": 2.7799873043851337e-05,
      "loss": 0.04947352409362793,
      "step": 1715
    },
    {
      "epoch": 0.2319918208011897,
      "grad_norm": 1.0731490850448608,
      "learning_rate": 2.7796303166246016e-05,
      "loss": 0.06417693197727203,
      "step": 1716
    },
    {
      "epoch": 0.23212701416995496,
      "grad_norm": 0.3587137460708618,
      "learning_rate": 2.7792730624382142e-05,
      "loss": 0.051572009921073914,
      "step": 1717
    },
    {
      "epoch": 0.23226220753872023,
      "grad_norm": 0.7012569308280945,
      "learning_rate": 2.778915541900353e-05,
      "loss": 0.052128903567790985,
      "step": 1718
    },
    {
      "epoch": 0.23239740090748548,
      "grad_norm": 1.057073712348938,
      "learning_rate": 2.7785577550854566e-05,
      "loss": 0.053997982293367386,
      "step": 1719
    },
    {
      "epoch": 0.23253259427625075,
      "grad_norm": 0.7684621810913086,
      "learning_rate": 2.778199702068017e-05,
      "loss": 0.05164429545402527,
      "step": 1720
    },
    {
      "epoch": 0.232667787645016,
      "grad_norm": 0.3256686329841614,
      "learning_rate": 2.777841382922583e-05,
      "loss": 0.060562632977962494,
      "step": 1721
    },
    {
      "epoch": 0.23280298101378127,
      "grad_norm": 0.444046288728714,
      "learning_rate": 2.7774827977237596e-05,
      "loss": 0.06021483242511749,
      "step": 1722
    },
    {
      "epoch": 0.23293817438254655,
      "grad_norm": 0.7290605902671814,
      "learning_rate": 2.777123946546205e-05,
      "loss": 0.06646603345870972,
      "step": 1723
    },
    {
      "epoch": 0.2330733677513118,
      "grad_norm": 0.8953688740730286,
      "learning_rate": 2.776764829464634e-05,
      "loss": 0.05031698942184448,
      "step": 1724
    },
    {
      "epoch": 0.23320856112007707,
      "grad_norm": 1.12669038772583,
      "learning_rate": 2.7764054465538173e-05,
      "loss": 0.06480012089014053,
      "step": 1725
    },
    {
      "epoch": 0.23334375448884231,
      "grad_norm": 0.8605860471725464,
      "learning_rate": 2.7760457978885794e-05,
      "loss": 0.06884333491325378,
      "step": 1726
    },
    {
      "epoch": 0.2334789478576076,
      "grad_norm": 1.0614045858383179,
      "learning_rate": 2.7756858835438022e-05,
      "loss": 0.06185644865036011,
      "step": 1727
    },
    {
      "epoch": 0.23361414122637283,
      "grad_norm": 0.6377485990524292,
      "learning_rate": 2.7753257035944216e-05,
      "loss": 0.0591130405664444,
      "step": 1728
    },
    {
      "epoch": 0.2337493345951381,
      "grad_norm": 0.7270711064338684,
      "learning_rate": 2.7749652581154277e-05,
      "loss": 0.07110196352005005,
      "step": 1729
    },
    {
      "epoch": 0.23388452796390338,
      "grad_norm": 0.30580636858940125,
      "learning_rate": 2.7746045471818685e-05,
      "loss": 0.03449159860610962,
      "step": 1730
    },
    {
      "epoch": 0.23401972133266863,
      "grad_norm": 0.9647356271743774,
      "learning_rate": 2.7742435708688458e-05,
      "loss": 0.06302178651094437,
      "step": 1731
    },
    {
      "epoch": 0.2341549147014339,
      "grad_norm": 0.6172770857810974,
      "learning_rate": 2.7738823292515167e-05,
      "loss": 0.05716082453727722,
      "step": 1732
    },
    {
      "epoch": 0.23429010807019915,
      "grad_norm": 1.1884669065475464,
      "learning_rate": 2.773520822405093e-05,
      "loss": 0.08634865283966064,
      "step": 1733
    },
    {
      "epoch": 0.23442530143896442,
      "grad_norm": 1.4400255680084229,
      "learning_rate": 2.7731590504048433e-05,
      "loss": 0.06523825228214264,
      "step": 1734
    },
    {
      "epoch": 0.23456049480772967,
      "grad_norm": 0.8287622332572937,
      "learning_rate": 2.7727970133260896e-05,
      "loss": 0.0670129656791687,
      "step": 1735
    },
    {
      "epoch": 0.23469568817649494,
      "grad_norm": 0.34829503297805786,
      "learning_rate": 2.7724347112442106e-05,
      "loss": 0.0478961318731308,
      "step": 1736
    },
    {
      "epoch": 0.23483088154526022,
      "grad_norm": 0.8971961140632629,
      "learning_rate": 2.772072144234639e-05,
      "loss": 0.05987761914730072,
      "step": 1737
    },
    {
      "epoch": 0.23496607491402546,
      "grad_norm": 0.5579344034194946,
      "learning_rate": 2.7717093123728634e-05,
      "loss": 0.045401155948638916,
      "step": 1738
    },
    {
      "epoch": 0.23510126828279074,
      "grad_norm": 0.5325412750244141,
      "learning_rate": 2.771346215734428e-05,
      "loss": 0.055544640868902206,
      "step": 1739
    },
    {
      "epoch": 0.23523646165155598,
      "grad_norm": 0.9315246343612671,
      "learning_rate": 2.7709828543949302e-05,
      "loss": 0.07714492082595825,
      "step": 1740
    },
    {
      "epoch": 0.23537165502032126,
      "grad_norm": 1.5582971572875977,
      "learning_rate": 2.770619228430025e-05,
      "loss": 0.06546777486801147,
      "step": 1741
    },
    {
      "epoch": 0.2355068483890865,
      "grad_norm": 0.9863983392715454,
      "learning_rate": 2.77025533791542e-05,
      "loss": 0.06010594964027405,
      "step": 1742
    },
    {
      "epoch": 0.23564204175785178,
      "grad_norm": 1.0297991037368774,
      "learning_rate": 2.76989118292688e-05,
      "loss": 0.06947432458400726,
      "step": 1743
    },
    {
      "epoch": 0.23577723512661705,
      "grad_norm": 0.45426660776138306,
      "learning_rate": 2.7695267635402242e-05,
      "loss": 0.053979575634002686,
      "step": 1744
    },
    {
      "epoch": 0.2359124284953823,
      "grad_norm": 0.641617476940155,
      "learning_rate": 2.7691620798313258e-05,
      "loss": 0.056682419031858444,
      "step": 1745
    },
    {
      "epoch": 0.23604762186414757,
      "grad_norm": 1.4111003875732422,
      "learning_rate": 2.7687971318761145e-05,
      "loss": 0.060999996960163116,
      "step": 1746
    },
    {
      "epoch": 0.23618281523291282,
      "grad_norm": 1.1374071836471558,
      "learning_rate": 2.7684319197505746e-05,
      "loss": 0.051502518355846405,
      "step": 1747
    },
    {
      "epoch": 0.2363180086016781,
      "grad_norm": 1.2918708324432373,
      "learning_rate": 2.7680664435307446e-05,
      "loss": 0.04531306400895119,
      "step": 1748
    },
    {
      "epoch": 0.23645320197044334,
      "grad_norm": 0.8245479464530945,
      "learning_rate": 2.767700703292719e-05,
      "loss": 0.06491582095623016,
      "step": 1749
    },
    {
      "epoch": 0.2365883953392086,
      "grad_norm": 0.6467477083206177,
      "learning_rate": 2.767334699112647e-05,
      "loss": 0.0583115816116333,
      "step": 1750
    },
    {
      "epoch": 0.2367235887079739,
      "grad_norm": 1.161240577697754,
      "learning_rate": 2.7669684310667318e-05,
      "loss": 0.06556473672389984,
      "step": 1751
    },
    {
      "epoch": 0.23685878207673913,
      "grad_norm": 0.6168792843818665,
      "learning_rate": 2.7666018992312333e-05,
      "loss": 0.07531628012657166,
      "step": 1752
    },
    {
      "epoch": 0.2369939754455044,
      "grad_norm": 0.29516834020614624,
      "learning_rate": 2.7662351036824653e-05,
      "loss": 0.04231659322977066,
      "step": 1753
    },
    {
      "epoch": 0.23712916881426965,
      "grad_norm": 0.4645411968231201,
      "learning_rate": 2.7658680444967964e-05,
      "loss": 0.0712827742099762,
      "step": 1754
    },
    {
      "epoch": 0.23726436218303493,
      "grad_norm": 0.38169869780540466,
      "learning_rate": 2.76550072175065e-05,
      "loss": 0.0872441977262497,
      "step": 1755
    },
    {
      "epoch": 0.23739955555180017,
      "grad_norm": 0.6510182619094849,
      "learning_rate": 2.7651331355205044e-05,
      "loss": 0.04930286854505539,
      "step": 1756
    },
    {
      "epoch": 0.23753474892056545,
      "grad_norm": 1.2145999670028687,
      "learning_rate": 2.7647652858828936e-05,
      "loss": 0.06656190752983093,
      "step": 1757
    },
    {
      "epoch": 0.23766994228933072,
      "grad_norm": 0.5234197974205017,
      "learning_rate": 2.764397172914406e-05,
      "loss": 0.05475449562072754,
      "step": 1758
    },
    {
      "epoch": 0.23780513565809597,
      "grad_norm": 0.9165642857551575,
      "learning_rate": 2.7640287966916845e-05,
      "loss": 0.07122278213500977,
      "step": 1759
    },
    {
      "epoch": 0.23794032902686124,
      "grad_norm": 0.2774665653705597,
      "learning_rate": 2.7636601572914266e-05,
      "loss": 0.0409126877784729,
      "step": 1760
    },
    {
      "epoch": 0.2380755223956265,
      "grad_norm": 1.0802595615386963,
      "learning_rate": 2.7632912547903855e-05,
      "loss": 0.06233122944831848,
      "step": 1761
    },
    {
      "epoch": 0.23821071576439176,
      "grad_norm": 0.2581363022327423,
      "learning_rate": 2.7629220892653685e-05,
      "loss": 0.05251719802618027,
      "step": 1762
    },
    {
      "epoch": 0.238345909133157,
      "grad_norm": 1.029417872428894,
      "learning_rate": 2.7625526607932378e-05,
      "loss": 0.06709900498390198,
      "step": 1763
    },
    {
      "epoch": 0.23848110250192228,
      "grad_norm": 0.8297461271286011,
      "learning_rate": 2.76218296945091e-05,
      "loss": 0.05600571632385254,
      "step": 1764
    },
    {
      "epoch": 0.23861629587068756,
      "grad_norm": 0.8343843817710876,
      "learning_rate": 2.7618130153153577e-05,
      "loss": 0.07020968198776245,
      "step": 1765
    },
    {
      "epoch": 0.2387514892394528,
      "grad_norm": 0.6480276584625244,
      "learning_rate": 2.7614427984636063e-05,
      "loss": 0.0596952810883522,
      "step": 1766
    },
    {
      "epoch": 0.23888668260821808,
      "grad_norm": 0.33661749958992004,
      "learning_rate": 2.7610723189727377e-05,
      "loss": 0.07578754425048828,
      "step": 1767
    },
    {
      "epoch": 0.23902187597698332,
      "grad_norm": 0.4645842909812927,
      "learning_rate": 2.760701576919888e-05,
      "loss": 0.058799296617507935,
      "step": 1768
    },
    {
      "epoch": 0.2391570693457486,
      "grad_norm": 0.24671271443367004,
      "learning_rate": 2.760330572382246e-05,
      "loss": 0.03767775744199753,
      "step": 1769
    },
    {
      "epoch": 0.23929226271451384,
      "grad_norm": 0.6260137557983398,
      "learning_rate": 2.7599593054370584e-05,
      "loss": 0.06268168985843658,
      "step": 1770
    },
    {
      "epoch": 0.23942745608327912,
      "grad_norm": 1.1759059429168701,
      "learning_rate": 2.7595877761616246e-05,
      "loss": 0.0525277704000473,
      "step": 1771
    },
    {
      "epoch": 0.2395626494520444,
      "grad_norm": 0.5533081889152527,
      "learning_rate": 2.759215984633299e-05,
      "loss": 0.060588620603084564,
      "step": 1772
    },
    {
      "epoch": 0.23969784282080964,
      "grad_norm": 2.2983760833740234,
      "learning_rate": 2.7588439309294902e-05,
      "loss": 0.08851748704910278,
      "step": 1773
    },
    {
      "epoch": 0.2398330361895749,
      "grad_norm": 0.4648050367832184,
      "learning_rate": 2.7584716151276623e-05,
      "loss": 0.05141003802418709,
      "step": 1774
    },
    {
      "epoch": 0.23996822955834016,
      "grad_norm": 1.3984687328338623,
      "learning_rate": 2.7580990373053325e-05,
      "loss": 0.05576743185520172,
      "step": 1775
    },
    {
      "epoch": 0.24010342292710543,
      "grad_norm": 0.6347246170043945,
      "learning_rate": 2.7577261975400747e-05,
      "loss": 0.06411650776863098,
      "step": 1776
    },
    {
      "epoch": 0.24023861629587068,
      "grad_norm": 0.9491327404975891,
      "learning_rate": 2.7573530959095154e-05,
      "loss": 0.048143547028303146,
      "step": 1777
    },
    {
      "epoch": 0.24037380966463595,
      "grad_norm": 0.46420589089393616,
      "learning_rate": 2.756979732491336e-05,
      "loss": 0.04543163627386093,
      "step": 1778
    },
    {
      "epoch": 0.24050900303340123,
      "grad_norm": 0.5693197250366211,
      "learning_rate": 2.756606107363274e-05,
      "loss": 0.07411319017410278,
      "step": 1779
    },
    {
      "epoch": 0.24064419640216647,
      "grad_norm": 1.1672110557556152,
      "learning_rate": 2.7562322206031192e-05,
      "loss": 0.0833682268857956,
      "step": 1780
    },
    {
      "epoch": 0.24077938977093175,
      "grad_norm": 1.0420798063278198,
      "learning_rate": 2.7558580722887166e-05,
      "loss": 0.0583530068397522,
      "step": 1781
    },
    {
      "epoch": 0.240914583139697,
      "grad_norm": 0.573514997959137,
      "learning_rate": 2.7554836624979666e-05,
      "loss": 0.055614590644836426,
      "step": 1782
    },
    {
      "epoch": 0.24104977650846227,
      "grad_norm": 0.5239211320877075,
      "learning_rate": 2.7551089913088233e-05,
      "loss": 0.06242452189326286,
      "step": 1783
    },
    {
      "epoch": 0.2411849698772275,
      "grad_norm": 0.6482314467430115,
      "learning_rate": 2.7547340587992948e-05,
      "loss": 0.05109933018684387,
      "step": 1784
    },
    {
      "epoch": 0.24132016324599279,
      "grad_norm": 1.2163139581680298,
      "learning_rate": 2.754358865047444e-05,
      "loss": 0.07107076793909073,
      "step": 1785
    },
    {
      "epoch": 0.24145535661475806,
      "grad_norm": 0.4805852472782135,
      "learning_rate": 2.7539834101313885e-05,
      "loss": 0.06997501850128174,
      "step": 1786
    },
    {
      "epoch": 0.2415905499835233,
      "grad_norm": 2.0008926391601562,
      "learning_rate": 2.7536076941293003e-05,
      "loss": 0.06454098224639893,
      "step": 1787
    },
    {
      "epoch": 0.24172574335228858,
      "grad_norm": 0.32264941930770874,
      "learning_rate": 2.753231717119405e-05,
      "loss": 0.038575127720832825,
      "step": 1788
    },
    {
      "epoch": 0.24186093672105383,
      "grad_norm": 0.38871124386787415,
      "learning_rate": 2.7528554791799826e-05,
      "loss": 0.04571658372879028,
      "step": 1789
    },
    {
      "epoch": 0.2419961300898191,
      "grad_norm": 0.9657143354415894,
      "learning_rate": 2.7524789803893686e-05,
      "loss": 0.07892800867557526,
      "step": 1790
    },
    {
      "epoch": 0.24213132345858435,
      "grad_norm": 0.28928887844085693,
      "learning_rate": 2.7521022208259526e-05,
      "loss": 0.05117356777191162,
      "step": 1791
    },
    {
      "epoch": 0.24226651682734962,
      "grad_norm": 0.5177487134933472,
      "learning_rate": 2.7517252005681762e-05,
      "loss": 0.06379137188196182,
      "step": 1792
    },
    {
      "epoch": 0.24240171019611487,
      "grad_norm": 1.0492885112762451,
      "learning_rate": 2.7513479196945385e-05,
      "loss": 0.07026708126068115,
      "step": 1793
    },
    {
      "epoch": 0.24253690356488014,
      "grad_norm": 1.247847318649292,
      "learning_rate": 2.750970378283591e-05,
      "loss": 0.06689602136611938,
      "step": 1794
    },
    {
      "epoch": 0.24267209693364541,
      "grad_norm": 2.1468605995178223,
      "learning_rate": 2.7505925764139398e-05,
      "loss": 0.06966928392648697,
      "step": 1795
    },
    {
      "epoch": 0.24280729030241066,
      "grad_norm": 1.622549295425415,
      "learning_rate": 2.7502145141642447e-05,
      "loss": 0.04319116473197937,
      "step": 1796
    },
    {
      "epoch": 0.24294248367117594,
      "grad_norm": 0.9661479592323303,
      "learning_rate": 2.7498361916132212e-05,
      "loss": 0.046201109886169434,
      "step": 1797
    },
    {
      "epoch": 0.24307767703994118,
      "grad_norm": 1.2449190616607666,
      "learning_rate": 2.7494576088396376e-05,
      "loss": 0.0622846782207489,
      "step": 1798
    },
    {
      "epoch": 0.24321287040870646,
      "grad_norm": 0.784413754940033,
      "learning_rate": 2.749078765922317e-05,
      "loss": 0.05469036102294922,
      "step": 1799
    },
    {
      "epoch": 0.2433480637774717,
      "grad_norm": 0.9067596793174744,
      "learning_rate": 2.7486996629401366e-05,
      "loss": 0.05928552895784378,
      "step": 1800
    },
    {
      "epoch": 0.24348325714623698,
      "grad_norm": 1.948528528213501,
      "learning_rate": 2.7483202999720272e-05,
      "loss": 0.06363476812839508,
      "step": 1801
    },
    {
      "epoch": 0.24361845051500225,
      "grad_norm": 1.1192318201065063,
      "learning_rate": 2.7479406770969747e-05,
      "loss": 0.04887492209672928,
      "step": 1802
    },
    {
      "epoch": 0.2437536438837675,
      "grad_norm": 0.898763120174408,
      "learning_rate": 2.7475607943940182e-05,
      "loss": 0.04625377804040909,
      "step": 1803
    },
    {
      "epoch": 0.24388883725253277,
      "grad_norm": 1.0470030307769775,
      "learning_rate": 2.7471806519422514e-05,
      "loss": 0.06388384103775024,
      "step": 1804
    },
    {
      "epoch": 0.24402403062129802,
      "grad_norm": 0.7634301781654358,
      "learning_rate": 2.746800249820822e-05,
      "loss": 0.05902114510536194,
      "step": 1805
    },
    {
      "epoch": 0.2441592239900633,
      "grad_norm": 1.5930242538452148,
      "learning_rate": 2.7464195881089323e-05,
      "loss": 0.06449858844280243,
      "step": 1806
    },
    {
      "epoch": 0.24429441735882854,
      "grad_norm": 1.7291353940963745,
      "learning_rate": 2.746038666885837e-05,
      "loss": 0.0698220357298851,
      "step": 1807
    },
    {
      "epoch": 0.2444296107275938,
      "grad_norm": 2.733517646789551,
      "learning_rate": 2.7456574862308474e-05,
      "loss": 0.06998045742511749,
      "step": 1808
    },
    {
      "epoch": 0.24456480409635908,
      "grad_norm": 2.2248353958129883,
      "learning_rate": 2.745276046223326e-05,
      "loss": 0.055635951459407806,
      "step": 1809
    },
    {
      "epoch": 0.24469999746512433,
      "grad_norm": 2.745321035385132,
      "learning_rate": 2.744894346942691e-05,
      "loss": 0.06990677118301392,
      "step": 1810
    },
    {
      "epoch": 0.2448351908338896,
      "grad_norm": 1.1119916439056396,
      "learning_rate": 2.744512388468415e-05,
      "loss": 0.05870842933654785,
      "step": 1811
    },
    {
      "epoch": 0.24497038420265485,
      "grad_norm": 0.6725683808326721,
      "learning_rate": 2.7441301708800227e-05,
      "loss": 0.07774585485458374,
      "step": 1812
    },
    {
      "epoch": 0.24510557757142012,
      "grad_norm": 0.804542601108551,
      "learning_rate": 2.7437476942570942e-05,
      "loss": 0.07373358309268951,
      "step": 1813
    },
    {
      "epoch": 0.24524077094018537,
      "grad_norm": 0.2727304995059967,
      "learning_rate": 2.7433649586792637e-05,
      "loss": 0.05927370488643646,
      "step": 1814
    },
    {
      "epoch": 0.24537596430895064,
      "grad_norm": 2.648794174194336,
      "learning_rate": 2.7429819642262178e-05,
      "loss": 0.08700032532215118,
      "step": 1815
    },
    {
      "epoch": 0.24551115767771592,
      "grad_norm": 1.0932246446609497,
      "learning_rate": 2.7425987109776994e-05,
      "loss": 0.0529962033033371,
      "step": 1816
    },
    {
      "epoch": 0.24564635104648117,
      "grad_norm": 1.0143872499465942,
      "learning_rate": 2.7422151990135022e-05,
      "loss": 0.0611632764339447,
      "step": 1817
    },
    {
      "epoch": 0.24578154441524644,
      "grad_norm": 1.2205675840377808,
      "learning_rate": 2.741831428413477e-05,
      "loss": 0.06258591264486313,
      "step": 1818
    },
    {
      "epoch": 0.24591673778401169,
      "grad_norm": 1.5245801210403442,
      "learning_rate": 2.7414473992575257e-05,
      "loss": 0.06870941817760468,
      "step": 1819
    },
    {
      "epoch": 0.24605193115277696,
      "grad_norm": 0.6271198987960815,
      "learning_rate": 2.7410631116256054e-05,
      "loss": 0.048978958278894424,
      "step": 1820
    },
    {
      "epoch": 0.2461871245215422,
      "grad_norm": 1.5812677145004272,
      "learning_rate": 2.7406785655977275e-05,
      "loss": 0.0609498955309391,
      "step": 1821
    },
    {
      "epoch": 0.24632231789030748,
      "grad_norm": 1.5800942182540894,
      "learning_rate": 2.7402937612539563e-05,
      "loss": 0.058327898383140564,
      "step": 1822
    },
    {
      "epoch": 0.24645751125907275,
      "grad_norm": 1.137834072113037,
      "learning_rate": 2.7399086986744095e-05,
      "loss": 0.05759924650192261,
      "step": 1823
    },
    {
      "epoch": 0.246592704627838,
      "grad_norm": 1.3274519443511963,
      "learning_rate": 2.7395233779392598e-05,
      "loss": 0.06996330618858337,
      "step": 1824
    },
    {
      "epoch": 0.24672789799660327,
      "grad_norm": 0.9218474626541138,
      "learning_rate": 2.739137799128733e-05,
      "loss": 0.03890904784202576,
      "step": 1825
    },
    {
      "epoch": 0.24686309136536852,
      "grad_norm": 1.3220356702804565,
      "learning_rate": 2.7387519623231085e-05,
      "loss": 0.0814785361289978,
      "step": 1826
    },
    {
      "epoch": 0.2469982847341338,
      "grad_norm": 0.50826495885849,
      "learning_rate": 2.7383658676027195e-05,
      "loss": 0.06760821491479874,
      "step": 1827
    },
    {
      "epoch": 0.24713347810289904,
      "grad_norm": 0.7867356538772583,
      "learning_rate": 2.7379795150479535e-05,
      "loss": 0.06059293448925018,
      "step": 1828
    },
    {
      "epoch": 0.24726867147166431,
      "grad_norm": 1.4877345561981201,
      "learning_rate": 2.73759290473925e-05,
      "loss": 0.06261137127876282,
      "step": 1829
    },
    {
      "epoch": 0.2474038648404296,
      "grad_norm": 0.9563901424407959,
      "learning_rate": 2.7372060367571044e-05,
      "loss": 0.060074642300605774,
      "step": 1830
    },
    {
      "epoch": 0.24753905820919483,
      "grad_norm": 0.351698637008667,
      "learning_rate": 2.7368189111820648e-05,
      "loss": 0.05316463112831116,
      "step": 1831
    },
    {
      "epoch": 0.2476742515779601,
      "grad_norm": 0.4171631932258606,
      "learning_rate": 2.736431528094732e-05,
      "loss": 0.045943550765514374,
      "step": 1832
    },
    {
      "epoch": 0.24780944494672535,
      "grad_norm": 0.359910786151886,
      "learning_rate": 2.7360438875757614e-05,
      "loss": 0.054830729961395264,
      "step": 1833
    },
    {
      "epoch": 0.24794463831549063,
      "grad_norm": 0.3965355157852173,
      "learning_rate": 2.7356559897058624e-05,
      "loss": 0.056278809905052185,
      "step": 1834
    },
    {
      "epoch": 0.24807983168425587,
      "grad_norm": 1.5511910915374756,
      "learning_rate": 2.735267834565797e-05,
      "loss": 0.06495694816112518,
      "step": 1835
    },
    {
      "epoch": 0.24821502505302115,
      "grad_norm": 0.47934770584106445,
      "learning_rate": 2.734879422236381e-05,
      "loss": 0.04668864607810974,
      "step": 1836
    },
    {
      "epoch": 0.24835021842178642,
      "grad_norm": 0.8820375204086304,
      "learning_rate": 2.734490752798484e-05,
      "loss": 0.06065814197063446,
      "step": 1837
    },
    {
      "epoch": 0.24848541179055167,
      "grad_norm": 0.6350238919258118,
      "learning_rate": 2.7341018263330296e-05,
      "loss": 0.042666926980018616,
      "step": 1838
    },
    {
      "epoch": 0.24862060515931694,
      "grad_norm": 0.6458765268325806,
      "learning_rate": 2.7337126429209935e-05,
      "loss": 0.06931717693805695,
      "step": 1839
    },
    {
      "epoch": 0.2487557985280822,
      "grad_norm": 0.721137285232544,
      "learning_rate": 2.7333232026434064e-05,
      "loss": 0.06862339377403259,
      "step": 1840
    },
    {
      "epoch": 0.24889099189684746,
      "grad_norm": 0.2562505006790161,
      "learning_rate": 2.7329335055813517e-05,
      "loss": 0.047194480895996094,
      "step": 1841
    },
    {
      "epoch": 0.2490261852656127,
      "grad_norm": 0.30537474155426025,
      "learning_rate": 2.732543551815966e-05,
      "loss": 0.05128122493624687,
      "step": 1842
    },
    {
      "epoch": 0.24916137863437798,
      "grad_norm": 0.4329995810985565,
      "learning_rate": 2.7321533414284404e-05,
      "loss": 0.057897549122571945,
      "step": 1843
    },
    {
      "epoch": 0.24929657200314326,
      "grad_norm": 1.046766996383667,
      "learning_rate": 2.731762874500018e-05,
      "loss": 0.07247406989336014,
      "step": 1844
    },
    {
      "epoch": 0.2494317653719085,
      "grad_norm": 0.4092596173286438,
      "learning_rate": 2.7313721511119972e-05,
      "loss": 0.051524870097637177,
      "step": 1845
    },
    {
      "epoch": 0.24956695874067378,
      "grad_norm": 0.36781468987464905,
      "learning_rate": 2.7309811713457275e-05,
      "loss": 0.05429118871688843,
      "step": 1846
    },
    {
      "epoch": 0.24970215210943902,
      "grad_norm": 0.8583148717880249,
      "learning_rate": 2.730589935282614e-05,
      "loss": 0.053462326526641846,
      "step": 1847
    },
    {
      "epoch": 0.2498373454782043,
      "grad_norm": 0.3260382115840912,
      "learning_rate": 2.7301984430041135e-05,
      "loss": 0.0619199275970459,
      "step": 1848
    },
    {
      "epoch": 0.24997253884696954,
      "grad_norm": 1.4224365949630737,
      "learning_rate": 2.7298066945917368e-05,
      "loss": 0.05573566257953644,
      "step": 1849
    },
    {
      "epoch": 0.25010773221573485,
      "grad_norm": 0.350322425365448,
      "learning_rate": 2.7294146901270482e-05,
      "loss": 0.0641157478094101,
      "step": 1850
    },
    {
      "epoch": 0.25024292558450006,
      "grad_norm": 0.47027844190597534,
      "learning_rate": 2.7290224296916653e-05,
      "loss": 0.05068720504641533,
      "step": 1851
    },
    {
      "epoch": 0.25037811895326534,
      "grad_norm": 0.9299583435058594,
      "learning_rate": 2.7286299133672584e-05,
      "loss": 0.0563206672668457,
      "step": 1852
    },
    {
      "epoch": 0.2505133123220306,
      "grad_norm": 0.9145474433898926,
      "learning_rate": 2.728237141235552e-05,
      "loss": 0.04449716955423355,
      "step": 1853
    },
    {
      "epoch": 0.2506485056907959,
      "grad_norm": 1.6957886219024658,
      "learning_rate": 2.727844113378322e-05,
      "loss": 0.06835083663463593,
      "step": 1854
    },
    {
      "epoch": 0.2507836990595611,
      "grad_norm": 0.529062807559967,
      "learning_rate": 2.7274508298774013e-05,
      "loss": 0.062294989824295044,
      "step": 1855
    },
    {
      "epoch": 0.2509188924283264,
      "grad_norm": 1.1401989459991455,
      "learning_rate": 2.727057290814672e-05,
      "loss": 0.06933862715959549,
      "step": 1856
    },
    {
      "epoch": 0.25105408579709165,
      "grad_norm": 0.7565990090370178,
      "learning_rate": 2.7266634962720704e-05,
      "loss": 0.053271643817424774,
      "step": 1857
    },
    {
      "epoch": 0.2511892791658569,
      "grad_norm": 0.3381154537200928,
      "learning_rate": 2.726269446331588e-05,
      "loss": 0.048572588711977005,
      "step": 1858
    },
    {
      "epoch": 0.2513244725346222,
      "grad_norm": 0.40603747963905334,
      "learning_rate": 2.7258751410752676e-05,
      "loss": 0.06588006764650345,
      "step": 1859
    },
    {
      "epoch": 0.2514596659033874,
      "grad_norm": 0.6585817337036133,
      "learning_rate": 2.725480580585206e-05,
      "loss": 0.04722384363412857,
      "step": 1860
    },
    {
      "epoch": 0.2515948592721527,
      "grad_norm": 1.5531675815582275,
      "learning_rate": 2.7250857649435522e-05,
      "loss": 0.062035009264945984,
      "step": 1861
    },
    {
      "epoch": 0.25173005264091797,
      "grad_norm": 0.7977484464645386,
      "learning_rate": 2.724690694232509e-05,
      "loss": 0.07116580009460449,
      "step": 1862
    },
    {
      "epoch": 0.25186524600968324,
      "grad_norm": 0.7185953855514526,
      "learning_rate": 2.7242953685343327e-05,
      "loss": 0.0728900358080864,
      "step": 1863
    },
    {
      "epoch": 0.25200043937844846,
      "grad_norm": 0.4016585648059845,
      "learning_rate": 2.723899787931332e-05,
      "loss": 0.039566099643707275,
      "step": 1864
    },
    {
      "epoch": 0.25213563274721373,
      "grad_norm": 1.5395886898040771,
      "learning_rate": 2.7235039525058684e-05,
      "loss": 0.056715503334999084,
      "step": 1865
    },
    {
      "epoch": 0.252270826115979,
      "grad_norm": 0.8606144189834595,
      "learning_rate": 2.7231078623403575e-05,
      "loss": 0.046288251876831055,
      "step": 1866
    },
    {
      "epoch": 0.2524060194847443,
      "grad_norm": 2.1993372440338135,
      "learning_rate": 2.722711517517267e-05,
      "loss": 0.07265862822532654,
      "step": 1867
    },
    {
      "epoch": 0.25254121285350956,
      "grad_norm": 1.4779186248779297,
      "learning_rate": 2.7223149181191187e-05,
      "loss": 0.06568378210067749,
      "step": 1868
    },
    {
      "epoch": 0.2526764062222748,
      "grad_norm": 0.7808976769447327,
      "learning_rate": 2.7219180642284864e-05,
      "loss": 0.08012668043375015,
      "step": 1869
    },
    {
      "epoch": 0.25281159959104005,
      "grad_norm": 0.465035617351532,
      "learning_rate": 2.721520955927997e-05,
      "loss": 0.054050326347351074,
      "step": 1870
    },
    {
      "epoch": 0.2529467929598053,
      "grad_norm": 0.5058714747428894,
      "learning_rate": 2.7211235933003302e-05,
      "loss": 0.050615519285202026,
      "step": 1871
    },
    {
      "epoch": 0.2530819863285706,
      "grad_norm": 1.8749537467956543,
      "learning_rate": 2.72072597642822e-05,
      "loss": 0.08371332287788391,
      "step": 1872
    },
    {
      "epoch": 0.25321717969733587,
      "grad_norm": 0.6472415328025818,
      "learning_rate": 2.7203281053944512e-05,
      "loss": 0.053905874490737915,
      "step": 1873
    },
    {
      "epoch": 0.2533523730661011,
      "grad_norm": 1.0982719659805298,
      "learning_rate": 2.719929980281864e-05,
      "loss": 0.05468970537185669,
      "step": 1874
    },
    {
      "epoch": 0.25348756643486636,
      "grad_norm": 0.7528514862060547,
      "learning_rate": 2.719531601173349e-05,
      "loss": 0.06013808399438858,
      "step": 1875
    },
    {
      "epoch": 0.25362275980363164,
      "grad_norm": 0.4620010554790497,
      "learning_rate": 2.7191329681518512e-05,
      "loss": 0.06881189346313477,
      "step": 1876
    },
    {
      "epoch": 0.2537579531723969,
      "grad_norm": 0.6234038472175598,
      "learning_rate": 2.7187340813003682e-05,
      "loss": 0.04425733536481857,
      "step": 1877
    },
    {
      "epoch": 0.25389314654116213,
      "grad_norm": 1.4850654602050781,
      "learning_rate": 2.718334940701951e-05,
      "loss": 0.0729677677154541,
      "step": 1878
    },
    {
      "epoch": 0.2540283399099274,
      "grad_norm": 0.9001442193984985,
      "learning_rate": 2.7179355464397014e-05,
      "loss": 0.07417845726013184,
      "step": 1879
    },
    {
      "epoch": 0.2541635332786927,
      "grad_norm": 0.9688825011253357,
      "learning_rate": 2.7175358985967763e-05,
      "loss": 0.07931607961654663,
      "step": 1880
    },
    {
      "epoch": 0.25429872664745795,
      "grad_norm": 0.5000962615013123,
      "learning_rate": 2.717135997256385e-05,
      "loss": 0.0718928873538971,
      "step": 1881
    },
    {
      "epoch": 0.2544339200162232,
      "grad_norm": 0.28846657276153564,
      "learning_rate": 2.7167358425017882e-05,
      "loss": 0.04686398804187775,
      "step": 1882
    },
    {
      "epoch": 0.25456911338498844,
      "grad_norm": 1.3982936143875122,
      "learning_rate": 2.7163354344163004e-05,
      "loss": 0.052206553518772125,
      "step": 1883
    },
    {
      "epoch": 0.2547043067537537,
      "grad_norm": 0.30318522453308105,
      "learning_rate": 2.715934773083289e-05,
      "loss": 0.053683459758758545,
      "step": 1884
    },
    {
      "epoch": 0.254839500122519,
      "grad_norm": 1.3615477085113525,
      "learning_rate": 2.715533858586174e-05,
      "loss": 0.05026477575302124,
      "step": 1885
    },
    {
      "epoch": 0.25497469349128427,
      "grad_norm": 1.1193678379058838,
      "learning_rate": 2.715132691008427e-05,
      "loss": 0.0637890100479126,
      "step": 1886
    },
    {
      "epoch": 0.25510988686004954,
      "grad_norm": 0.7913280725479126,
      "learning_rate": 2.714731270433574e-05,
      "loss": 0.05701271444559097,
      "step": 1887
    },
    {
      "epoch": 0.25524508022881476,
      "grad_norm": 0.6793217062950134,
      "learning_rate": 2.7143295969451933e-05,
      "loss": 0.05757066607475281,
      "step": 1888
    },
    {
      "epoch": 0.25538027359758003,
      "grad_norm": 0.9453518390655518,
      "learning_rate": 2.7139276706269147e-05,
      "loss": 0.06984785199165344,
      "step": 1889
    },
    {
      "epoch": 0.2555154669663453,
      "grad_norm": 0.595427930355072,
      "learning_rate": 2.7135254915624213e-05,
      "loss": 0.05994880199432373,
      "step": 1890
    },
    {
      "epoch": 0.2556506603351106,
      "grad_norm": 1.0444555282592773,
      "learning_rate": 2.7131230598354497e-05,
      "loss": 0.05637727677822113,
      "step": 1891
    },
    {
      "epoch": 0.2557858537038758,
      "grad_norm": 0.9351038932800293,
      "learning_rate": 2.712720375529787e-05,
      "loss": 0.0469779372215271,
      "step": 1892
    },
    {
      "epoch": 0.2559210470726411,
      "grad_norm": 0.8016331195831299,
      "learning_rate": 2.7123174387292758e-05,
      "loss": 0.0498119592666626,
      "step": 1893
    },
    {
      "epoch": 0.25605624044140635,
      "grad_norm": 0.31011295318603516,
      "learning_rate": 2.7119142495178088e-05,
      "loss": 0.043321095407009125,
      "step": 1894
    },
    {
      "epoch": 0.2561914338101716,
      "grad_norm": 1.458289623260498,
      "learning_rate": 2.711510807979333e-05,
      "loss": 0.05448979139328003,
      "step": 1895
    },
    {
      "epoch": 0.2563266271789369,
      "grad_norm": 1.7407903671264648,
      "learning_rate": 2.7111071141978452e-05,
      "loss": 0.05992579460144043,
      "step": 1896
    },
    {
      "epoch": 0.2564618205477021,
      "grad_norm": 0.3617306649684906,
      "learning_rate": 2.7107031682573987e-05,
      "loss": 0.054641127586364746,
      "step": 1897
    },
    {
      "epoch": 0.2565970139164674,
      "grad_norm": 0.6111330986022949,
      "learning_rate": 2.710298970242096e-05,
      "loss": 0.053890250623226166,
      "step": 1898
    },
    {
      "epoch": 0.25673220728523266,
      "grad_norm": 0.42177915573120117,
      "learning_rate": 2.7098945202360937e-05,
      "loss": 0.05688124895095825,
      "step": 1899
    },
    {
      "epoch": 0.25686740065399793,
      "grad_norm": 0.7408020496368408,
      "learning_rate": 2.7094898183236e-05,
      "loss": 0.06984029710292816,
      "step": 1900
    },
    {
      "epoch": 0.2570025940227632,
      "grad_norm": 0.5317678451538086,
      "learning_rate": 2.709084864588877e-05,
      "loss": 0.06329578906297684,
      "step": 1901
    },
    {
      "epoch": 0.2571377873915284,
      "grad_norm": 0.6379442811012268,
      "learning_rate": 2.708679659116237e-05,
      "loss": 0.06518420577049255,
      "step": 1902
    },
    {
      "epoch": 0.2572729807602937,
      "grad_norm": 1.0077797174453735,
      "learning_rate": 2.708274201990047e-05,
      "loss": 0.04854530096054077,
      "step": 1903
    },
    {
      "epoch": 0.257408174129059,
      "grad_norm": 0.6798588633537292,
      "learning_rate": 2.7078684932947247e-05,
      "loss": 0.06895408034324646,
      "step": 1904
    },
    {
      "epoch": 0.25754336749782425,
      "grad_norm": 0.48604434728622437,
      "learning_rate": 2.7074625331147407e-05,
      "loss": 0.047987326979637146,
      "step": 1905
    },
    {
      "epoch": 0.25767856086658947,
      "grad_norm": 0.48254266381263733,
      "learning_rate": 2.7070563215346184e-05,
      "loss": 0.05127117037773132,
      "step": 1906
    },
    {
      "epoch": 0.25781375423535474,
      "grad_norm": 0.5055396556854248,
      "learning_rate": 2.7066498586389332e-05,
      "loss": 0.05714119225740433,
      "step": 1907
    },
    {
      "epoch": 0.25794894760412,
      "grad_norm": 1.2692639827728271,
      "learning_rate": 2.7062431445123127e-05,
      "loss": 0.06041686609387398,
      "step": 1908
    },
    {
      "epoch": 0.2580841409728853,
      "grad_norm": 0.5003594160079956,
      "learning_rate": 2.705836179239437e-05,
      "loss": 0.05806645005941391,
      "step": 1909
    },
    {
      "epoch": 0.25821933434165056,
      "grad_norm": 0.9051533341407776,
      "learning_rate": 2.705428962905039e-05,
      "loss": 0.07491414248943329,
      "step": 1910
    },
    {
      "epoch": 0.2583545277104158,
      "grad_norm": 0.5739898085594177,
      "learning_rate": 2.705021495593902e-05,
      "loss": 0.05252230167388916,
      "step": 1911
    },
    {
      "epoch": 0.25848972107918106,
      "grad_norm": 1.1121389865875244,
      "learning_rate": 2.704613777390864e-05,
      "loss": 0.072174072265625,
      "step": 1912
    },
    {
      "epoch": 0.25862491444794633,
      "grad_norm": 0.8885291814804077,
      "learning_rate": 2.7042058083808135e-05,
      "loss": 0.06262989342212677,
      "step": 1913
    },
    {
      "epoch": 0.2587601078167116,
      "grad_norm": 0.9516357779502869,
      "learning_rate": 2.7037975886486928e-05,
      "loss": 0.04772152006626129,
      "step": 1914
    },
    {
      "epoch": 0.2588953011854769,
      "grad_norm": 0.3271455764770508,
      "learning_rate": 2.7033891182794942e-05,
      "loss": 0.030513525009155273,
      "step": 1915
    },
    {
      "epoch": 0.2590304945542421,
      "grad_norm": 0.7102453112602234,
      "learning_rate": 2.7029803973582642e-05,
      "loss": 0.052742987871170044,
      "step": 1916
    },
    {
      "epoch": 0.25916568792300737,
      "grad_norm": 0.6446483135223389,
      "learning_rate": 2.7025714259701e-05,
      "loss": 0.054237931966781616,
      "step": 1917
    },
    {
      "epoch": 0.25930088129177264,
      "grad_norm": 2.5130269527435303,
      "learning_rate": 2.7021622042001524e-05,
      "loss": 0.07978424429893494,
      "step": 1918
    },
    {
      "epoch": 0.2594360746605379,
      "grad_norm": 1.4490081071853638,
      "learning_rate": 2.701752732133623e-05,
      "loss": 0.06512665748596191,
      "step": 1919
    },
    {
      "epoch": 0.25957126802930314,
      "grad_norm": 1.5117402076721191,
      "learning_rate": 2.7013430098557664e-05,
      "loss": 0.06800676882266998,
      "step": 1920
    },
    {
      "epoch": 0.2597064613980684,
      "grad_norm": 0.4595567286014557,
      "learning_rate": 2.7009330374518885e-05,
      "loss": 0.07122290134429932,
      "step": 1921
    },
    {
      "epoch": 0.2598416547668337,
      "grad_norm": 1.3470333814620972,
      "learning_rate": 2.7005228150073483e-05,
      "loss": 0.06333541870117188,
      "step": 1922
    },
    {
      "epoch": 0.25997684813559896,
      "grad_norm": 0.5032757520675659,
      "learning_rate": 2.7001123426075558e-05,
      "loss": 0.05834554135799408,
      "step": 1923
    },
    {
      "epoch": 0.26011204150436423,
      "grad_norm": 0.28954771161079407,
      "learning_rate": 2.699701620337974e-05,
      "loss": 0.04165119305253029,
      "step": 1924
    },
    {
      "epoch": 0.26024723487312945,
      "grad_norm": 0.20491264760494232,
      "learning_rate": 2.699290648284117e-05,
      "loss": 0.03780645132064819,
      "step": 1925
    },
    {
      "epoch": 0.2603824282418947,
      "grad_norm": 0.7633419632911682,
      "learning_rate": 2.6988794265315522e-05,
      "loss": 0.04897985979914665,
      "step": 1926
    },
    {
      "epoch": 0.26051762161066,
      "grad_norm": 0.5443441271781921,
      "learning_rate": 2.698467955165897e-05,
      "loss": 0.046710193157196045,
      "step": 1927
    },
    {
      "epoch": 0.2606528149794253,
      "grad_norm": 1.3065940141677856,
      "learning_rate": 2.6980562342728226e-05,
      "loss": 0.0639866441488266,
      "step": 1928
    },
    {
      "epoch": 0.26078800834819055,
      "grad_norm": 2.1279795169830322,
      "learning_rate": 2.6976442639380516e-05,
      "loss": 0.09672144055366516,
      "step": 1929
    },
    {
      "epoch": 0.26092320171695577,
      "grad_norm": 0.3977057635784149,
      "learning_rate": 2.6972320442473583e-05,
      "loss": 0.05983440577983856,
      "step": 1930
    },
    {
      "epoch": 0.26105839508572104,
      "grad_norm": 0.3161788880825043,
      "learning_rate": 2.6968195752865686e-05,
      "loss": 0.0703844353556633,
      "step": 1931
    },
    {
      "epoch": 0.2611935884544863,
      "grad_norm": 0.7519835233688354,
      "learning_rate": 2.6964068571415613e-05,
      "loss": 0.039884477853775024,
      "step": 1932
    },
    {
      "epoch": 0.2613287818232516,
      "grad_norm": 0.949356198310852,
      "learning_rate": 2.6959938898982667e-05,
      "loss": 0.06464026123285294,
      "step": 1933
    },
    {
      "epoch": 0.2614639751920168,
      "grad_norm": 0.3803577125072479,
      "learning_rate": 2.6955806736426657e-05,
      "loss": 0.04969945549964905,
      "step": 1934
    },
    {
      "epoch": 0.2615991685607821,
      "grad_norm": 0.5054728984832764,
      "learning_rate": 2.6951672084607937e-05,
      "loss": 0.05244843661785126,
      "step": 1935
    },
    {
      "epoch": 0.26173436192954735,
      "grad_norm": 0.9093344807624817,
      "learning_rate": 2.694753494438735e-05,
      "loss": 0.05855073779821396,
      "step": 1936
    },
    {
      "epoch": 0.26186955529831263,
      "grad_norm": 0.2236987054347992,
      "learning_rate": 2.6943395316626272e-05,
      "loss": 0.0456208810210228,
      "step": 1937
    },
    {
      "epoch": 0.2620047486670779,
      "grad_norm": 0.29676297307014465,
      "learning_rate": 2.69392532021866e-05,
      "loss": 0.055048078298568726,
      "step": 1938
    },
    {
      "epoch": 0.2621399420358431,
      "grad_norm": 0.4697316586971283,
      "learning_rate": 2.693510860193075e-05,
      "loss": 0.046270325779914856,
      "step": 1939
    },
    {
      "epoch": 0.2622751354046084,
      "grad_norm": 1.1346640586853027,
      "learning_rate": 2.6930961516721638e-05,
      "loss": 0.07410261034965515,
      "step": 1940
    },
    {
      "epoch": 0.26241032877337367,
      "grad_norm": 1.0093399286270142,
      "learning_rate": 2.6926811947422717e-05,
      "loss": 0.061102449893951416,
      "step": 1941
    },
    {
      "epoch": 0.26254552214213894,
      "grad_norm": 0.32292136549949646,
      "learning_rate": 2.6922659894897946e-05,
      "loss": 0.03949351608753204,
      "step": 1942
    },
    {
      "epoch": 0.2626807155109042,
      "grad_norm": 0.4040752947330475,
      "learning_rate": 2.6918505360011805e-05,
      "loss": 0.058416783809661865,
      "step": 1943
    },
    {
      "epoch": 0.26281590887966944,
      "grad_norm": 0.6055988669395447,
      "learning_rate": 2.6914348343629292e-05,
      "loss": 0.0581560879945755,
      "step": 1944
    },
    {
      "epoch": 0.2629511022484347,
      "grad_norm": 0.7155444622039795,
      "learning_rate": 2.6910188846615918e-05,
      "loss": 0.07350419461727142,
      "step": 1945
    },
    {
      "epoch": 0.2630862956172,
      "grad_norm": 0.5275724530220032,
      "learning_rate": 2.6906026869837714e-05,
      "loss": 0.06944310665130615,
      "step": 1946
    },
    {
      "epoch": 0.26322148898596526,
      "grad_norm": 0.7116784453392029,
      "learning_rate": 2.6901862414161222e-05,
      "loss": 0.06892804801464081,
      "step": 1947
    },
    {
      "epoch": 0.2633566823547305,
      "grad_norm": 0.4181476831436157,
      "learning_rate": 2.689769548045351e-05,
      "loss": 0.0550856739282608,
      "step": 1948
    },
    {
      "epoch": 0.26349187572349575,
      "grad_norm": 0.33882415294647217,
      "learning_rate": 2.6893526069582154e-05,
      "loss": 0.05200064182281494,
      "step": 1949
    },
    {
      "epoch": 0.263627069092261,
      "grad_norm": 0.6489690542221069,
      "learning_rate": 2.6889354182415245e-05,
      "loss": 0.05584654211997986,
      "step": 1950
    },
    {
      "epoch": 0.2637622624610263,
      "grad_norm": 0.6772111654281616,
      "learning_rate": 2.688517981982139e-05,
      "loss": 0.06969714164733887,
      "step": 1951
    },
    {
      "epoch": 0.26389745582979157,
      "grad_norm": 1.2031954526901245,
      "learning_rate": 2.6881002982669723e-05,
      "loss": 0.06634595990180969,
      "step": 1952
    },
    {
      "epoch": 0.2640326491985568,
      "grad_norm": 0.5072511434555054,
      "learning_rate": 2.6876823671829874e-05,
      "loss": 0.061920978128910065,
      "step": 1953
    },
    {
      "epoch": 0.26416784256732206,
      "grad_norm": 1.2806464433670044,
      "learning_rate": 2.6872641888172e-05,
      "loss": 0.05857972800731659,
      "step": 1954
    },
    {
      "epoch": 0.26430303593608734,
      "grad_norm": 0.6853794455528259,
      "learning_rate": 2.6868457632566774e-05,
      "loss": 0.06263160705566406,
      "step": 1955
    },
    {
      "epoch": 0.2644382293048526,
      "grad_norm": 0.5920037627220154,
      "learning_rate": 2.6864270905885377e-05,
      "loss": 0.054866619408130646,
      "step": 1956
    },
    {
      "epoch": 0.2645734226736179,
      "grad_norm": 1.4363727569580078,
      "learning_rate": 2.6860081708999515e-05,
      "loss": 0.0517156682908535,
      "step": 1957
    },
    {
      "epoch": 0.2647086160423831,
      "grad_norm": 0.7071147561073303,
      "learning_rate": 2.685589004278139e-05,
      "loss": 0.06194252520799637,
      "step": 1958
    },
    {
      "epoch": 0.2648438094111484,
      "grad_norm": 1.235511302947998,
      "learning_rate": 2.6851695908103737e-05,
      "loss": 0.04652991518378258,
      "step": 1959
    },
    {
      "epoch": 0.26497900277991365,
      "grad_norm": 0.5754021406173706,
      "learning_rate": 2.6847499305839796e-05,
      "loss": 0.04518058896064758,
      "step": 1960
    },
    {
      "epoch": 0.2651141961486789,
      "grad_norm": 0.875260055065155,
      "learning_rate": 2.684330023686332e-05,
      "loss": 0.06999662518501282,
      "step": 1961
    },
    {
      "epoch": 0.26524938951744415,
      "grad_norm": 1.3374043703079224,
      "learning_rate": 2.6839098702048577e-05,
      "loss": 0.07331164926290512,
      "step": 1962
    },
    {
      "epoch": 0.2653845828862094,
      "grad_norm": 0.9756178259849548,
      "learning_rate": 2.683489470227035e-05,
      "loss": 0.04712153598666191,
      "step": 1963
    },
    {
      "epoch": 0.2655197762549747,
      "grad_norm": 1.1614489555358887,
      "learning_rate": 2.6830688238403936e-05,
      "loss": 0.05995406210422516,
      "step": 1964
    },
    {
      "epoch": 0.26565496962373997,
      "grad_norm": 0.9477787017822266,
      "learning_rate": 2.682647931132514e-05,
      "loss": 0.04590131342411041,
      "step": 1965
    },
    {
      "epoch": 0.26579016299250524,
      "grad_norm": 1.38800847530365,
      "learning_rate": 2.682226792191029e-05,
      "loss": 0.056134700775146484,
      "step": 1966
    },
    {
      "epoch": 0.26592535636127046,
      "grad_norm": 2.509615182876587,
      "learning_rate": 2.681805407103621e-05,
      "loss": 0.08689026534557343,
      "step": 1967
    },
    {
      "epoch": 0.26606054973003573,
      "grad_norm": 0.551263153553009,
      "learning_rate": 2.6813837759580253e-05,
      "loss": 0.05793978273868561,
      "step": 1968
    },
    {
      "epoch": 0.266195743098801,
      "grad_norm": 0.6268625855445862,
      "learning_rate": 2.6809618988420274e-05,
      "loss": 0.06955307722091675,
      "step": 1969
    },
    {
      "epoch": 0.2663309364675663,
      "grad_norm": 0.3579012453556061,
      "learning_rate": 2.6805397758434647e-05,
      "loss": 0.05733288824558258,
      "step": 1970
    },
    {
      "epoch": 0.26646612983633156,
      "grad_norm": 0.539228618144989,
      "learning_rate": 2.6801174070502248e-05,
      "loss": 0.05921357870101929,
      "step": 1971
    },
    {
      "epoch": 0.2666013232050968,
      "grad_norm": 1.1919562816619873,
      "learning_rate": 2.679694792550248e-05,
      "loss": 0.05999784544110298,
      "step": 1972
    },
    {
      "epoch": 0.26673651657386205,
      "grad_norm": 1.372760534286499,
      "learning_rate": 2.6792719324315248e-05,
      "loss": 0.049474261701107025,
      "step": 1973
    },
    {
      "epoch": 0.2668717099426273,
      "grad_norm": 0.9844462275505066,
      "learning_rate": 2.678848826782096e-05,
      "loss": 0.07781863212585449,
      "step": 1974
    },
    {
      "epoch": 0.2670069033113926,
      "grad_norm": 0.37535110116004944,
      "learning_rate": 2.678425475690055e-05,
      "loss": 0.04978233575820923,
      "step": 1975
    },
    {
      "epoch": 0.2671420966801578,
      "grad_norm": 0.5681155323982239,
      "learning_rate": 2.6780018792435464e-05,
      "loss": 0.05559802055358887,
      "step": 1976
    },
    {
      "epoch": 0.2672772900489231,
      "grad_norm": 1.2379475831985474,
      "learning_rate": 2.6775780375307645e-05,
      "loss": 0.05596546828746796,
      "step": 1977
    },
    {
      "epoch": 0.26741248341768836,
      "grad_norm": 0.3967132270336151,
      "learning_rate": 2.6771539506399555e-05,
      "loss": 0.04318176209926605,
      "step": 1978
    },
    {
      "epoch": 0.26754767678645364,
      "grad_norm": 0.40085673332214355,
      "learning_rate": 2.6767296186594165e-05,
      "loss": 0.05271802097558975,
      "step": 1979
    },
    {
      "epoch": 0.2676828701552189,
      "grad_norm": 0.7064202427864075,
      "learning_rate": 2.676305041677496e-05,
      "loss": 0.04608451575040817,
      "step": 1980
    },
    {
      "epoch": 0.26781806352398413,
      "grad_norm": 0.3746558427810669,
      "learning_rate": 2.675880219782593e-05,
      "loss": 0.06824222207069397,
      "step": 1981
    },
    {
      "epoch": 0.2679532568927494,
      "grad_norm": 0.364446759223938,
      "learning_rate": 2.6754551530631575e-05,
      "loss": 0.037284694612026215,
      "step": 1982
    },
    {
      "epoch": 0.2680884502615147,
      "grad_norm": 0.26884880661964417,
      "learning_rate": 2.6750298416076907e-05,
      "loss": 0.03772097826004028,
      "step": 1983
    },
    {
      "epoch": 0.26822364363027995,
      "grad_norm": 1.5052671432495117,
      "learning_rate": 2.674604285504745e-05,
      "loss": 0.07238410413265228,
      "step": 1984
    },
    {
      "epoch": 0.26835883699904517,
      "grad_norm": 0.869867742061615,
      "learning_rate": 2.6741784848429235e-05,
      "loss": 0.05454730987548828,
      "step": 1985
    },
    {
      "epoch": 0.26849403036781044,
      "grad_norm": 1.318238377571106,
      "learning_rate": 2.67375243971088e-05,
      "loss": 0.06269331276416779,
      "step": 1986
    },
    {
      "epoch": 0.2686292237365757,
      "grad_norm": 0.535857081413269,
      "learning_rate": 2.6733261501973192e-05,
      "loss": 0.060758545994758606,
      "step": 1987
    },
    {
      "epoch": 0.268764417105341,
      "grad_norm": 1.0873723030090332,
      "learning_rate": 2.672899616390997e-05,
      "loss": 0.06212398409843445,
      "step": 1988
    },
    {
      "epoch": 0.26889961047410627,
      "grad_norm": 0.9297671914100647,
      "learning_rate": 2.67247283838072e-05,
      "loss": 0.07655810564756393,
      "step": 1989
    },
    {
      "epoch": 0.2690348038428715,
      "grad_norm": 0.6478985548019409,
      "learning_rate": 2.6720458162553457e-05,
      "loss": 0.04838288575410843,
      "step": 1990
    },
    {
      "epoch": 0.26916999721163676,
      "grad_norm": 1.0387420654296875,
      "learning_rate": 2.6716185501037822e-05,
      "loss": 0.06883057951927185,
      "step": 1991
    },
    {
      "epoch": 0.26930519058040203,
      "grad_norm": 0.5580405592918396,
      "learning_rate": 2.671191040014989e-05,
      "loss": 0.046358756721019745,
      "step": 1992
    },
    {
      "epoch": 0.2694403839491673,
      "grad_norm": 1.4209740161895752,
      "learning_rate": 2.6707632860779756e-05,
      "loss": 0.09473806619644165,
      "step": 1993
    },
    {
      "epoch": 0.2695755773179326,
      "grad_norm": 0.5205309391021729,
      "learning_rate": 2.6703352883818024e-05,
      "loss": 0.043007511645555496,
      "step": 1994
    },
    {
      "epoch": 0.2697107706866978,
      "grad_norm": 0.38811278343200684,
      "learning_rate": 2.6699070470155816e-05,
      "loss": 0.04337712749838829,
      "step": 1995
    },
    {
      "epoch": 0.2698459640554631,
      "grad_norm": 0.45077791810035706,
      "learning_rate": 2.669478562068475e-05,
      "loss": 0.07259446382522583,
      "step": 1996
    },
    {
      "epoch": 0.26998115742422835,
      "grad_norm": 0.7099560499191284,
      "learning_rate": 2.6690498336296955e-05,
      "loss": 0.06714433431625366,
      "step": 1997
    },
    {
      "epoch": 0.2701163507929936,
      "grad_norm": 1.017167568206787,
      "learning_rate": 2.6686208617885057e-05,
      "loss": 0.03857170045375824,
      "step": 1998
    },
    {
      "epoch": 0.27025154416175884,
      "grad_norm": 1.6143343448638916,
      "learning_rate": 2.668191646634221e-05,
      "loss": 0.06279931217432022,
      "step": 1999
    },
    {
      "epoch": 0.2703867375305241,
      "grad_norm": 0.24972137808799744,
      "learning_rate": 2.667762188256206e-05,
      "loss": 0.040912751108407974,
      "step": 2000
    },
    {
      "epoch": 0.2705219308992894,
      "grad_norm": 0.7155330777168274,
      "learning_rate": 2.6673324867438764e-05,
      "loss": 0.05724817514419556,
      "step": 2001
    },
    {
      "epoch": 0.27065712426805466,
      "grad_norm": 1.0139793157577515,
      "learning_rate": 2.666902542186698e-05,
      "loss": 0.07184246182441711,
      "step": 2002
    },
    {
      "epoch": 0.27079231763681993,
      "grad_norm": 0.6095170378684998,
      "learning_rate": 2.666472354674187e-05,
      "loss": 0.0645485520362854,
      "step": 2003
    },
    {
      "epoch": 0.27092751100558515,
      "grad_norm": 1.3737459182739258,
      "learning_rate": 2.666041924295912e-05,
      "loss": 0.05256572365760803,
      "step": 2004
    },
    {
      "epoch": 0.2710627043743504,
      "grad_norm": 0.966224730014801,
      "learning_rate": 2.6656112511414902e-05,
      "loss": 0.06350427865982056,
      "step": 2005
    },
    {
      "epoch": 0.2711978977431157,
      "grad_norm": 2.5293614864349365,
      "learning_rate": 2.6651803353005896e-05,
      "loss": 0.06370405107736588,
      "step": 2006
    },
    {
      "epoch": 0.271333091111881,
      "grad_norm": 0.6755800247192383,
      "learning_rate": 2.66474917686293e-05,
      "loss": 0.06628291308879852,
      "step": 2007
    },
    {
      "epoch": 0.27146828448064625,
      "grad_norm": 0.6063771843910217,
      "learning_rate": 2.664317775918281e-05,
      "loss": 0.05570520460605621,
      "step": 2008
    },
    {
      "epoch": 0.27160347784941147,
      "grad_norm": 0.4344748854637146,
      "learning_rate": 2.6638861325564615e-05,
      "loss": 0.04128864407539368,
      "step": 2009
    },
    {
      "epoch": 0.27173867121817674,
      "grad_norm": 0.7044264078140259,
      "learning_rate": 2.6634542468673432e-05,
      "loss": 0.0394858792424202,
      "step": 2010
    },
    {
      "epoch": 0.271873864586942,
      "grad_norm": 0.4111612141132355,
      "learning_rate": 2.663022118940846e-05,
      "loss": 0.053647249937057495,
      "step": 2011
    },
    {
      "epoch": 0.2720090579557073,
      "grad_norm": 1.2196134328842163,
      "learning_rate": 2.662589748866942e-05,
      "loss": 0.05058395862579346,
      "step": 2012
    },
    {
      "epoch": 0.2721442513244725,
      "grad_norm": 0.36669281125068665,
      "learning_rate": 2.6621571367356522e-05,
      "loss": 0.048945896327495575,
      "step": 2013
    },
    {
      "epoch": 0.2722794446932378,
      "grad_norm": 0.4089919328689575,
      "learning_rate": 2.6617242826370495e-05,
      "loss": 0.05728524923324585,
      "step": 2014
    },
    {
      "epoch": 0.27241463806200306,
      "grad_norm": 0.5235752463340759,
      "learning_rate": 2.661291186661256e-05,
      "loss": 0.0560784712433815,
      "step": 2015
    },
    {
      "epoch": 0.27254983143076833,
      "grad_norm": 0.6754398941993713,
      "learning_rate": 2.6608578488984444e-05,
      "loss": 0.0594220831990242,
      "step": 2016
    },
    {
      "epoch": 0.2726850247995336,
      "grad_norm": 0.25663843750953674,
      "learning_rate": 2.6604242694388388e-05,
      "loss": 0.042109452188014984,
      "step": 2017
    },
    {
      "epoch": 0.2728202181682988,
      "grad_norm": 0.6239408254623413,
      "learning_rate": 2.6599904483727116e-05,
      "loss": 0.05000513046979904,
      "step": 2018
    },
    {
      "epoch": 0.2729554115370641,
      "grad_norm": 0.9525699615478516,
      "learning_rate": 2.6595563857903872e-05,
      "loss": 0.04286642372608185,
      "step": 2019
    },
    {
      "epoch": 0.27309060490582937,
      "grad_norm": 0.6130484342575073,
      "learning_rate": 2.6591220817822405e-05,
      "loss": 0.07682472467422485,
      "step": 2020
    },
    {
      "epoch": 0.27322579827459464,
      "grad_norm": 0.7622012495994568,
      "learning_rate": 2.658687536438694e-05,
      "loss": 0.07658661901950836,
      "step": 2021
    },
    {
      "epoch": 0.2733609916433599,
      "grad_norm": 1.308983325958252,
      "learning_rate": 2.6582527498502243e-05,
      "loss": 0.05244206637144089,
      "step": 2022
    },
    {
      "epoch": 0.27349618501212514,
      "grad_norm": 0.9608461260795593,
      "learning_rate": 2.6578177221073556e-05,
      "loss": 0.06910043954849243,
      "step": 2023
    },
    {
      "epoch": 0.2736313783808904,
      "grad_norm": 1.6253361701965332,
      "learning_rate": 2.6573824533006628e-05,
      "loss": 0.06038973480463028,
      "step": 2024
    },
    {
      "epoch": 0.2737665717496557,
      "grad_norm": 0.36416172981262207,
      "learning_rate": 2.6569469435207712e-05,
      "loss": 0.06410835683345795,
      "step": 2025
    },
    {
      "epoch": 0.27390176511842096,
      "grad_norm": 1.387352705001831,
      "learning_rate": 2.656511192858356e-05,
      "loss": 0.06703147292137146,
      "step": 2026
    },
    {
      "epoch": 0.2740369584871862,
      "grad_norm": 1.038543939590454,
      "learning_rate": 2.6560752014041438e-05,
      "loss": 0.063834547996521,
      "step": 2027
    },
    {
      "epoch": 0.27417215185595145,
      "grad_norm": 1.0874626636505127,
      "learning_rate": 2.6556389692489098e-05,
      "loss": 0.05736267566680908,
      "step": 2028
    },
    {
      "epoch": 0.2743073452247167,
      "grad_norm": 0.49104976654052734,
      "learning_rate": 2.6552024964834795e-05,
      "loss": 0.06279715895652771,
      "step": 2029
    },
    {
      "epoch": 0.274442538593482,
      "grad_norm": 0.8350416421890259,
      "learning_rate": 2.6547657831987286e-05,
      "loss": 0.07277926057577133,
      "step": 2030
    },
    {
      "epoch": 0.2745777319622473,
      "grad_norm": 1.1978554725646973,
      "learning_rate": 2.6543288294855843e-05,
      "loss": 0.06596343219280243,
      "step": 2031
    },
    {
      "epoch": 0.2747129253310125,
      "grad_norm": 0.3906417787075043,
      "learning_rate": 2.653891635435022e-05,
      "loss": 0.0609915629029274,
      "step": 2032
    },
    {
      "epoch": 0.27484811869977777,
      "grad_norm": 1.3309073448181152,
      "learning_rate": 2.653454201138068e-05,
      "loss": 0.06403705477714539,
      "step": 2033
    },
    {
      "epoch": 0.27498331206854304,
      "grad_norm": 0.4766184985637665,
      "learning_rate": 2.653016526685798e-05,
      "loss": 0.06566610932350159,
      "step": 2034
    },
    {
      "epoch": 0.2751185054373083,
      "grad_norm": 0.593104362487793,
      "learning_rate": 2.6525786121693387e-05,
      "loss": 0.05452541261911392,
      "step": 2035
    },
    {
      "epoch": 0.2752536988060736,
      "grad_norm": 1.4176524877548218,
      "learning_rate": 2.652140457679866e-05,
      "loss": 0.08625222742557526,
      "step": 2036
    },
    {
      "epoch": 0.2753888921748388,
      "grad_norm": 0.2285621613264084,
      "learning_rate": 2.6517020633086064e-05,
      "loss": 0.044230446219444275,
      "step": 2037
    },
    {
      "epoch": 0.2755240855436041,
      "grad_norm": 1.102249264717102,
      "learning_rate": 2.6512634291468354e-05,
      "loss": 0.05788937211036682,
      "step": 2038
    },
    {
      "epoch": 0.27565927891236935,
      "grad_norm": 1.042309045791626,
      "learning_rate": 2.6508245552858792e-05,
      "loss": 0.07709316909313202,
      "step": 2039
    },
    {
      "epoch": 0.27579447228113463,
      "grad_norm": 0.4332238733768463,
      "learning_rate": 2.6503854418171133e-05,
      "loss": 0.053490445017814636,
      "step": 2040
    },
    {
      "epoch": 0.27592966564989985,
      "grad_norm": 0.5355224013328552,
      "learning_rate": 2.6499460888319644e-05,
      "loss": 0.06303292512893677,
      "step": 2041
    },
    {
      "epoch": 0.2760648590186651,
      "grad_norm": 0.6453908681869507,
      "learning_rate": 2.6495064964219073e-05,
      "loss": 0.06252860277891159,
      "step": 2042
    },
    {
      "epoch": 0.2762000523874304,
      "grad_norm": 0.6886420845985413,
      "learning_rate": 2.649066664678467e-05,
      "loss": 0.06643763929605484,
      "step": 2043
    },
    {
      "epoch": 0.27633524575619567,
      "grad_norm": 0.5869872570037842,
      "learning_rate": 2.6486265936932205e-05,
      "loss": 0.07961654663085938,
      "step": 2044
    },
    {
      "epoch": 0.27647043912496094,
      "grad_norm": 0.6214112043380737,
      "learning_rate": 2.6481862835577915e-05,
      "loss": 0.05826053023338318,
      "step": 2045
    },
    {
      "epoch": 0.27660563249372616,
      "grad_norm": 0.9288861155509949,
      "learning_rate": 2.6477457343638557e-05,
      "loss": 0.05841217562556267,
      "step": 2046
    },
    {
      "epoch": 0.27674082586249144,
      "grad_norm": 1.3822966814041138,
      "learning_rate": 2.647304946203137e-05,
      "loss": 0.06501396745443344,
      "step": 2047
    },
    {
      "epoch": 0.2768760192312567,
      "grad_norm": 0.4394715428352356,
      "learning_rate": 2.6468639191674106e-05,
      "loss": 0.0452602282166481,
      "step": 2048
    },
    {
      "epoch": 0.277011212600022,
      "grad_norm": 0.38890260457992554,
      "learning_rate": 2.6464226533485007e-05,
      "loss": 0.06317457556724548,
      "step": 2049
    },
    {
      "epoch": 0.27714640596878726,
      "grad_norm": 0.9562539458274841,
      "learning_rate": 2.6459811488382806e-05,
      "loss": 0.0622381754219532,
      "step": 2050
    },
    {
      "epoch": 0.2772815993375525,
      "grad_norm": 0.7285279631614685,
      "learning_rate": 2.645539405728674e-05,
      "loss": 0.06485778093338013,
      "step": 2051
    },
    {
      "epoch": 0.27741679270631775,
      "grad_norm": 0.5501190423965454,
      "learning_rate": 2.6450974241116545e-05,
      "loss": 0.05617007613182068,
      "step": 2052
    },
    {
      "epoch": 0.277551986075083,
      "grad_norm": 0.7587544322013855,
      "learning_rate": 2.644655204079245e-05,
      "loss": 0.06515318900346756,
      "step": 2053
    },
    {
      "epoch": 0.2776871794438483,
      "grad_norm": 0.7682645916938782,
      "learning_rate": 2.6442127457235177e-05,
      "loss": 0.054640933871269226,
      "step": 2054
    },
    {
      "epoch": 0.2778223728126135,
      "grad_norm": 1.141964316368103,
      "learning_rate": 2.6437700491365957e-05,
      "loss": 0.06025565043091774,
      "step": 2055
    },
    {
      "epoch": 0.2779575661813788,
      "grad_norm": 0.38962647318840027,
      "learning_rate": 2.6433271144106495e-05,
      "loss": 0.05071568489074707,
      "step": 2056
    },
    {
      "epoch": 0.27809275955014406,
      "grad_norm": 0.4076961576938629,
      "learning_rate": 2.6428839416379015e-05,
      "loss": 0.06274399906396866,
      "step": 2057
    },
    {
      "epoch": 0.27822795291890934,
      "grad_norm": 0.4470677375793457,
      "learning_rate": 2.642440530910622e-05,
      "loss": 0.055875495076179504,
      "step": 2058
    },
    {
      "epoch": 0.2783631462876746,
      "grad_norm": 1.213831901550293,
      "learning_rate": 2.6419968823211318e-05,
      "loss": 0.07357648015022278,
      "step": 2059
    },
    {
      "epoch": 0.27849833965643983,
      "grad_norm": 0.6747694611549377,
      "learning_rate": 2.641552995961801e-05,
      "loss": 0.06557597219944,
      "step": 2060
    },
    {
      "epoch": 0.2786335330252051,
      "grad_norm": 0.438777357339859,
      "learning_rate": 2.6411088719250484e-05,
      "loss": 0.07125673443078995,
      "step": 2061
    },
    {
      "epoch": 0.2787687263939704,
      "grad_norm": 0.6402414441108704,
      "learning_rate": 2.6406645103033442e-05,
      "loss": 0.05008098483085632,
      "step": 2062
    },
    {
      "epoch": 0.27890391976273565,
      "grad_norm": 0.47974082827568054,
      "learning_rate": 2.640219911189206e-05,
      "loss": 0.052809394896030426,
      "step": 2063
    },
    {
      "epoch": 0.2790391131315009,
      "grad_norm": 1.3230500221252441,
      "learning_rate": 2.6397750746752015e-05,
      "loss": 0.06357279419898987,
      "step": 2064
    },
    {
      "epoch": 0.27917430650026615,
      "grad_norm": 0.6687652468681335,
      "learning_rate": 2.6393300008539488e-05,
      "loss": 0.07994140684604645,
      "step": 2065
    },
    {
      "epoch": 0.2793094998690314,
      "grad_norm": 1.1518137454986572,
      "learning_rate": 2.6388846898181143e-05,
      "loss": 0.06600034236907959,
      "step": 2066
    },
    {
      "epoch": 0.2794446932377967,
      "grad_norm": 0.45129120349884033,
      "learning_rate": 2.6384391416604142e-05,
      "loss": 0.07186335325241089,
      "step": 2067
    },
    {
      "epoch": 0.27957988660656197,
      "grad_norm": 0.560322105884552,
      "learning_rate": 2.6379933564736136e-05,
      "loss": 0.06591766327619553,
      "step": 2068
    },
    {
      "epoch": 0.2797150799753272,
      "grad_norm": 0.24667029082775116,
      "learning_rate": 2.637547334350528e-05,
      "loss": 0.03618912398815155,
      "step": 2069
    },
    {
      "epoch": 0.27985027334409246,
      "grad_norm": 0.8697183728218079,
      "learning_rate": 2.637101075384021e-05,
      "loss": 0.05519033968448639,
      "step": 2070
    },
    {
      "epoch": 0.27998546671285773,
      "grad_norm": 0.7805359363555908,
      "learning_rate": 2.636654579667006e-05,
      "loss": 0.06149636209011078,
      "step": 2071
    },
    {
      "epoch": 0.280120660081623,
      "grad_norm": 1.6960641145706177,
      "learning_rate": 2.6362078472924467e-05,
      "loss": 0.08044280111789703,
      "step": 2072
    },
    {
      "epoch": 0.2802558534503883,
      "grad_norm": 0.3118131458759308,
      "learning_rate": 2.6357608783533545e-05,
      "loss": 0.04548192769289017,
      "step": 2073
    },
    {
      "epoch": 0.2803910468191535,
      "grad_norm": 0.6053363680839539,
      "learning_rate": 2.6353136729427907e-05,
      "loss": 0.0536673367023468,
      "step": 2074
    },
    {
      "epoch": 0.2805262401879188,
      "grad_norm": 0.5909741520881653,
      "learning_rate": 2.6348662311538657e-05,
      "loss": 0.06355074048042297,
      "step": 2075
    },
    {
      "epoch": 0.28066143355668405,
      "grad_norm": 0.47632163763046265,
      "learning_rate": 2.6344185530797398e-05,
      "loss": 0.06311307847499847,
      "step": 2076
    },
    {
      "epoch": 0.2807966269254493,
      "grad_norm": 1.1889612674713135,
      "learning_rate": 2.633970638813622e-05,
      "loss": 0.05792977288365364,
      "step": 2077
    },
    {
      "epoch": 0.2809318202942146,
      "grad_norm": 1.0011950731277466,
      "learning_rate": 2.6335224884487698e-05,
      "loss": 0.04652509093284607,
      "step": 2078
    },
    {
      "epoch": 0.2810670136629798,
      "grad_norm": 0.3565572202205658,
      "learning_rate": 2.6330741020784905e-05,
      "loss": 0.06233729422092438,
      "step": 2079
    },
    {
      "epoch": 0.2812022070317451,
      "grad_norm": 1.7147787809371948,
      "learning_rate": 2.6326254797961415e-05,
      "loss": 0.06864406913518906,
      "step": 2080
    },
    {
      "epoch": 0.28133740040051036,
      "grad_norm": 1.502392053604126,
      "learning_rate": 2.6321766216951273e-05,
      "loss": 0.06973086297512054,
      "step": 2081
    },
    {
      "epoch": 0.28147259376927564,
      "grad_norm": 0.4112927317619324,
      "learning_rate": 2.631727527868903e-05,
      "loss": 0.05096091330051422,
      "step": 2082
    },
    {
      "epoch": 0.28160778713804085,
      "grad_norm": 0.5707592368125916,
      "learning_rate": 2.6312781984109727e-05,
      "loss": 0.05502316355705261,
      "step": 2083
    },
    {
      "epoch": 0.28174298050680613,
      "grad_norm": 0.5201780200004578,
      "learning_rate": 2.6308286334148882e-05,
      "loss": 0.06917780637741089,
      "step": 2084
    },
    {
      "epoch": 0.2818781738755714,
      "grad_norm": 1.4294310808181763,
      "learning_rate": 2.630378832974252e-05,
      "loss": 0.06621050834655762,
      "step": 2085
    },
    {
      "epoch": 0.2820133672443367,
      "grad_norm": 0.5480486750602722,
      "learning_rate": 2.6299287971827154e-05,
      "loss": 0.046482332050800323,
      "step": 2086
    },
    {
      "epoch": 0.28214856061310195,
      "grad_norm": 0.7295203804969788,
      "learning_rate": 2.629478526133977e-05,
      "loss": 0.046879373490810394,
      "step": 2087
    },
    {
      "epoch": 0.28228375398186717,
      "grad_norm": 1.3096091747283936,
      "learning_rate": 2.6290280199217867e-05,
      "loss": 0.07276512682437897,
      "step": 2088
    },
    {
      "epoch": 0.28241894735063244,
      "grad_norm": 0.5544393062591553,
      "learning_rate": 2.6285772786399424e-05,
      "loss": 0.07403305172920227,
      "step": 2089
    },
    {
      "epoch": 0.2825541407193977,
      "grad_norm": 1.1611231565475464,
      "learning_rate": 2.6281263023822894e-05,
      "loss": 0.05479201301932335,
      "step": 2090
    },
    {
      "epoch": 0.282689334088163,
      "grad_norm": 0.589711606502533,
      "learning_rate": 2.627675091242725e-05,
      "loss": 0.07190374284982681,
      "step": 2091
    },
    {
      "epoch": 0.28282452745692827,
      "grad_norm": 0.37469837069511414,
      "learning_rate": 2.627223645315193e-05,
      "loss": 0.04291430488228798,
      "step": 2092
    },
    {
      "epoch": 0.2829597208256935,
      "grad_norm": 0.579419732093811,
      "learning_rate": 2.6267719646936868e-05,
      "loss": 0.05952748656272888,
      "step": 2093
    },
    {
      "epoch": 0.28309491419445876,
      "grad_norm": 1.2522783279418945,
      "learning_rate": 2.626320049472249e-05,
      "loss": 0.055046483874320984,
      "step": 2094
    },
    {
      "epoch": 0.28323010756322403,
      "grad_norm": 0.848613440990448,
      "learning_rate": 2.6258678997449705e-05,
      "loss": 0.0651765838265419,
      "step": 2095
    },
    {
      "epoch": 0.2833653009319893,
      "grad_norm": 1.2239712476730347,
      "learning_rate": 2.6254155156059912e-05,
      "loss": 0.05742935836315155,
      "step": 2096
    },
    {
      "epoch": 0.2835004943007545,
      "grad_norm": 0.4563262462615967,
      "learning_rate": 2.6249628971495006e-05,
      "loss": 0.061477482318878174,
      "step": 2097
    },
    {
      "epoch": 0.2836356876695198,
      "grad_norm": 0.484327107667923,
      "learning_rate": 2.6245100444697353e-05,
      "loss": 0.05643342435359955,
      "step": 2098
    },
    {
      "epoch": 0.28377088103828507,
      "grad_norm": 0.6670452356338501,
      "learning_rate": 2.6240569576609824e-05,
      "loss": 0.07622776925563812,
      "step": 2099
    },
    {
      "epoch": 0.28390607440705035,
      "grad_norm": 0.6406234502792358,
      "learning_rate": 2.623603636817577e-05,
      "loss": 0.08094936609268188,
      "step": 2100
    },
    {
      "epoch": 0.2840412677758156,
      "grad_norm": 0.917285144329071,
      "learning_rate": 2.6231500820339024e-05,
      "loss": 0.030194712802767754,
      "step": 2101
    },
    {
      "epoch": 0.28417646114458084,
      "grad_norm": 0.886455237865448,
      "learning_rate": 2.6226962934043913e-05,
      "loss": 0.04499289393424988,
      "step": 2102
    },
    {
      "epoch": 0.2843116545133461,
      "grad_norm": 0.5797407627105713,
      "learning_rate": 2.622242271023525e-05,
      "loss": 0.08102905750274658,
      "step": 2103
    },
    {
      "epoch": 0.2844468478821114,
      "grad_norm": 1.250813364982605,
      "learning_rate": 2.6217880149858333e-05,
      "loss": 0.05921363830566406,
      "step": 2104
    },
    {
      "epoch": 0.28458204125087666,
      "grad_norm": 0.9572173953056335,
      "learning_rate": 2.621333525385895e-05,
      "loss": 0.0761459469795227,
      "step": 2105
    },
    {
      "epoch": 0.2847172346196419,
      "grad_norm": 0.5949355363845825,
      "learning_rate": 2.6208788023183366e-05,
      "loss": 0.0530424565076828,
      "step": 2106
    },
    {
      "epoch": 0.28485242798840715,
      "grad_norm": 1.1584101915359497,
      "learning_rate": 2.6204238458778346e-05,
      "loss": 0.055029988288879395,
      "step": 2107
    },
    {
      "epoch": 0.2849876213571724,
      "grad_norm": 0.647292971611023,
      "learning_rate": 2.619968656159113e-05,
      "loss": 0.05487114191055298,
      "step": 2108
    },
    {
      "epoch": 0.2851228147259377,
      "grad_norm": 0.7586051821708679,
      "learning_rate": 2.6195132332569445e-05,
      "loss": 0.053023576736450195,
      "step": 2109
    },
    {
      "epoch": 0.285258008094703,
      "grad_norm": 1.4620863199234009,
      "learning_rate": 2.619057577266151e-05,
      "loss": 0.0534789115190506,
      "step": 2110
    },
    {
      "epoch": 0.2853932014634682,
      "grad_norm": 1.3903439044952393,
      "learning_rate": 2.6186016882816027e-05,
      "loss": 0.05233687534928322,
      "step": 2111
    },
    {
      "epoch": 0.28552839483223347,
      "grad_norm": 1.9320210218429565,
      "learning_rate": 2.6181455663982175e-05,
      "loss": 0.058652929961681366,
      "step": 2112
    },
    {
      "epoch": 0.28566358820099874,
      "grad_norm": 1.721946120262146,
      "learning_rate": 2.6176892117109628e-05,
      "loss": 0.06058088690042496,
      "step": 2113
    },
    {
      "epoch": 0.285798781569764,
      "grad_norm": 1.8031491041183472,
      "learning_rate": 2.617232624314854e-05,
      "loss": 0.061432018876075745,
      "step": 2114
    },
    {
      "epoch": 0.2859339749385293,
      "grad_norm": 1.1871031522750854,
      "learning_rate": 2.616775804304955e-05,
      "loss": 0.06383770704269409,
      "step": 2115
    },
    {
      "epoch": 0.2860691683072945,
      "grad_norm": 1.7513153553009033,
      "learning_rate": 2.616318751776378e-05,
      "loss": 0.05388442799448967,
      "step": 2116
    },
    {
      "epoch": 0.2862043616760598,
      "grad_norm": 0.41766881942749023,
      "learning_rate": 2.615861466824284e-05,
      "loss": 0.06770746409893036,
      "step": 2117
    },
    {
      "epoch": 0.28633955504482506,
      "grad_norm": 0.5413923859596252,
      "learning_rate": 2.6154039495438825e-05,
      "loss": 0.06088829040527344,
      "step": 2118
    },
    {
      "epoch": 0.28647474841359033,
      "grad_norm": 1.5705360174179077,
      "learning_rate": 2.6149462000304302e-05,
      "loss": 0.07825294882059097,
      "step": 2119
    },
    {
      "epoch": 0.28660994178235555,
      "grad_norm": 1.2637450695037842,
      "learning_rate": 2.6144882183792335e-05,
      "loss": 0.04790281131863594,
      "step": 2120
    },
    {
      "epoch": 0.2867451351511208,
      "grad_norm": 1.96914803981781,
      "learning_rate": 2.6140300046856468e-05,
      "loss": 0.06698126345872879,
      "step": 2121
    },
    {
      "epoch": 0.2868803285198861,
      "grad_norm": 2.1521503925323486,
      "learning_rate": 2.6135715590450722e-05,
      "loss": 0.06972119212150574,
      "step": 2122
    },
    {
      "epoch": 0.28701552188865137,
      "grad_norm": 2.6057887077331543,
      "learning_rate": 2.6131128815529608e-05,
      "loss": 0.07070231437683105,
      "step": 2123
    },
    {
      "epoch": 0.28715071525741664,
      "grad_norm": 0.8588773608207703,
      "learning_rate": 2.6126539723048115e-05,
      "loss": 0.0656328797340393,
      "step": 2124
    },
    {
      "epoch": 0.28728590862618186,
      "grad_norm": 0.9165468811988831,
      "learning_rate": 2.612194831396172e-05,
      "loss": 0.04477657750248909,
      "step": 2125
    },
    {
      "epoch": 0.28742110199494714,
      "grad_norm": 1.0606473684310913,
      "learning_rate": 2.611735458922637e-05,
      "loss": 0.047616153955459595,
      "step": 2126
    },
    {
      "epoch": 0.2875562953637124,
      "grad_norm": 1.8390511274337769,
      "learning_rate": 2.6112758549798515e-05,
      "loss": 0.0676884651184082,
      "step": 2127
    },
    {
      "epoch": 0.2876914887324777,
      "grad_norm": 1.0135793685913086,
      "learning_rate": 2.610816019663507e-05,
      "loss": 0.07285168766975403,
      "step": 2128
    },
    {
      "epoch": 0.28782668210124296,
      "grad_norm": 0.8273748159408569,
      "learning_rate": 2.6103559530693436e-05,
      "loss": 0.0653875470161438,
      "step": 2129
    },
    {
      "epoch": 0.2879618754700082,
      "grad_norm": 1.3856375217437744,
      "learning_rate": 2.6098956552931495e-05,
      "loss": 0.07490475475788116,
      "step": 2130
    },
    {
      "epoch": 0.28809706883877345,
      "grad_norm": 0.3101572096347809,
      "learning_rate": 2.6094351264307613e-05,
      "loss": 0.0505615696310997,
      "step": 2131
    },
    {
      "epoch": 0.2882322622075387,
      "grad_norm": 0.6453413367271423,
      "learning_rate": 2.6089743665780635e-05,
      "loss": 0.04547475278377533,
      "step": 2132
    },
    {
      "epoch": 0.288367455576304,
      "grad_norm": 0.412611186504364,
      "learning_rate": 2.6085133758309887e-05,
      "loss": 0.0630529522895813,
      "step": 2133
    },
    {
      "epoch": 0.2885026489450692,
      "grad_norm": 0.6633384823799133,
      "learning_rate": 2.6080521542855182e-05,
      "loss": 0.06623699516057968,
      "step": 2134
    },
    {
      "epoch": 0.2886378423138345,
      "grad_norm": 0.5063520073890686,
      "learning_rate": 2.60759070203768e-05,
      "loss": 0.03371000289916992,
      "step": 2135
    },
    {
      "epoch": 0.28877303568259977,
      "grad_norm": 0.6748740077018738,
      "learning_rate": 2.607129019183551e-05,
      "loss": 0.059692561626434326,
      "step": 2136
    },
    {
      "epoch": 0.28890822905136504,
      "grad_norm": 0.8152740001678467,
      "learning_rate": 2.6066671058192566e-05,
      "loss": 0.07377050817012787,
      "step": 2137
    },
    {
      "epoch": 0.2890434224201303,
      "grad_norm": 0.45241445302963257,
      "learning_rate": 2.606204962040969e-05,
      "loss": 0.05911744385957718,
      "step": 2138
    },
    {
      "epoch": 0.28917861578889553,
      "grad_norm": 0.9087333679199219,
      "learning_rate": 2.6057425879449095e-05,
      "loss": 0.06385991722345352,
      "step": 2139
    },
    {
      "epoch": 0.2893138091576608,
      "grad_norm": 1.0540632009506226,
      "learning_rate": 2.605279983627347e-05,
      "loss": 0.060062795877456665,
      "step": 2140
    },
    {
      "epoch": 0.2894490025264261,
      "grad_norm": 0.36895206570625305,
      "learning_rate": 2.6048171491845974e-05,
      "loss": 0.05678999423980713,
      "step": 2141
    },
    {
      "epoch": 0.28958419589519135,
      "grad_norm": 1.6579879522323608,
      "learning_rate": 2.604354084713026e-05,
      "loss": 0.055744290351867676,
      "step": 2142
    },
    {
      "epoch": 0.28971938926395663,
      "grad_norm": 0.5126224756240845,
      "learning_rate": 2.6038907903090446e-05,
      "loss": 0.05784796550869942,
      "step": 2143
    },
    {
      "epoch": 0.28985458263272185,
      "grad_norm": 0.527445912361145,
      "learning_rate": 2.6034272660691143e-05,
      "loss": 0.07844924926757812,
      "step": 2144
    },
    {
      "epoch": 0.2899897760014871,
      "grad_norm": 0.43016624450683594,
      "learning_rate": 2.6029635120897434e-05,
      "loss": 0.06831195950508118,
      "step": 2145
    },
    {
      "epoch": 0.2901249693702524,
      "grad_norm": 1.5274966955184937,
      "learning_rate": 2.6024995284674867e-05,
      "loss": 0.06620031595230103,
      "step": 2146
    },
    {
      "epoch": 0.29026016273901767,
      "grad_norm": 0.7344223856925964,
      "learning_rate": 2.6020353152989496e-05,
      "loss": 0.0732392966747284,
      "step": 2147
    },
    {
      "epoch": 0.2903953561077829,
      "grad_norm": 0.8761911988258362,
      "learning_rate": 2.601570872680783e-05,
      "loss": 0.052635855972766876,
      "step": 2148
    },
    {
      "epoch": 0.29053054947654816,
      "grad_norm": 0.7074698805809021,
      "learning_rate": 2.6011062007096857e-05,
      "loss": 0.05327111482620239,
      "step": 2149
    },
    {
      "epoch": 0.29066574284531344,
      "grad_norm": 0.50385981798172,
      "learning_rate": 2.6006412994824067e-05,
      "loss": 0.05209881067276001,
      "step": 2150
    },
    {
      "epoch": 0.2908009362140787,
      "grad_norm": 0.3549734652042389,
      "learning_rate": 2.6001761690957388e-05,
      "loss": 0.042777977883815765,
      "step": 2151
    },
    {
      "epoch": 0.290936129582844,
      "grad_norm": 0.6660646200180054,
      "learning_rate": 2.5997108096465263e-05,
      "loss": 0.03807990252971649,
      "step": 2152
    },
    {
      "epoch": 0.2910713229516092,
      "grad_norm": 0.4637339413166046,
      "learning_rate": 2.599245221231659e-05,
      "loss": 0.04703313857316971,
      "step": 2153
    },
    {
      "epoch": 0.2912065163203745,
      "grad_norm": 0.6196834444999695,
      "learning_rate": 2.5987794039480743e-05,
      "loss": 0.06262943148612976,
      "step": 2154
    },
    {
      "epoch": 0.29134170968913975,
      "grad_norm": 0.9010622501373291,
      "learning_rate": 2.5983133578927584e-05,
      "loss": 0.05753566324710846,
      "step": 2155
    },
    {
      "epoch": 0.291476903057905,
      "grad_norm": 0.5884128212928772,
      "learning_rate": 2.5978470831627444e-05,
      "loss": 0.07157447934150696,
      "step": 2156
    },
    {
      "epoch": 0.2916120964266703,
      "grad_norm": 0.505321741104126,
      "learning_rate": 2.597380579855113e-05,
      "loss": 0.06195785105228424,
      "step": 2157
    },
    {
      "epoch": 0.2917472897954355,
      "grad_norm": 0.5729936957359314,
      "learning_rate": 2.5969138480669936e-05,
      "loss": 0.07561683654785156,
      "step": 2158
    },
    {
      "epoch": 0.2918824831642008,
      "grad_norm": 0.7477554082870483,
      "learning_rate": 2.5964468878955614e-05,
      "loss": 0.06464653462171555,
      "step": 2159
    },
    {
      "epoch": 0.29201767653296606,
      "grad_norm": 0.8106328248977661,
      "learning_rate": 2.5959796994380397e-05,
      "loss": 0.061203569173812866,
      "step": 2160
    },
    {
      "epoch": 0.29215286990173134,
      "grad_norm": 0.34237033128738403,
      "learning_rate": 2.5955122827917004e-05,
      "loss": 0.05926303565502167,
      "step": 2161
    },
    {
      "epoch": 0.29228806327049656,
      "grad_norm": 0.291145920753479,
      "learning_rate": 2.595044638053862e-05,
      "loss": 0.042369719594717026,
      "step": 2162
    },
    {
      "epoch": 0.29242325663926183,
      "grad_norm": 0.35732415318489075,
      "learning_rate": 2.59457676532189e-05,
      "loss": 0.04472386837005615,
      "step": 2163
    },
    {
      "epoch": 0.2925584500080271,
      "grad_norm": 0.943020761013031,
      "learning_rate": 2.594108664693199e-05,
      "loss": 0.06752137839794159,
      "step": 2164
    },
    {
      "epoch": 0.2926936433767924,
      "grad_norm": 1.8312469720840454,
      "learning_rate": 2.5936403362652494e-05,
      "loss": 0.07746871560811996,
      "step": 2165
    },
    {
      "epoch": 0.29282883674555765,
      "grad_norm": 0.3293423056602478,
      "learning_rate": 2.5931717801355497e-05,
      "loss": 0.07278227061033249,
      "step": 2166
    },
    {
      "epoch": 0.29296403011432287,
      "grad_norm": 0.3762620985507965,
      "learning_rate": 2.5927029964016556e-05,
      "loss": 0.0636572614312172,
      "step": 2167
    },
    {
      "epoch": 0.29309922348308814,
      "grad_norm": 0.7130229473114014,
      "learning_rate": 2.592233985161171e-05,
      "loss": 0.037568364292383194,
      "step": 2168
    },
    {
      "epoch": 0.2932344168518534,
      "grad_norm": 0.6726861000061035,
      "learning_rate": 2.5917647465117463e-05,
      "loss": 0.06736618280410767,
      "step": 2169
    },
    {
      "epoch": 0.2933696102206187,
      "grad_norm": 0.47206631302833557,
      "learning_rate": 2.591295280551079e-05,
      "loss": 0.049211252480745316,
      "step": 2170
    },
    {
      "epoch": 0.29350480358938397,
      "grad_norm": 0.41759103536605835,
      "learning_rate": 2.590825587376915e-05,
      "loss": 0.059932708740234375,
      "step": 2171
    },
    {
      "epoch": 0.2936399969581492,
      "grad_norm": 0.6596844792366028,
      "learning_rate": 2.5903556670870464e-05,
      "loss": 0.0686102956533432,
      "step": 2172
    },
    {
      "epoch": 0.29377519032691446,
      "grad_norm": 0.7449108362197876,
      "learning_rate": 2.589885519779314e-05,
      "loss": 0.06023481488227844,
      "step": 2173
    },
    {
      "epoch": 0.29391038369567973,
      "grad_norm": 0.3134186565876007,
      "learning_rate": 2.5894151455516043e-05,
      "loss": 0.04394517093896866,
      "step": 2174
    },
    {
      "epoch": 0.294045577064445,
      "grad_norm": 0.9985197186470032,
      "learning_rate": 2.5889445445018513e-05,
      "loss": 0.07373246550559998,
      "step": 2175
    },
    {
      "epoch": 0.2941807704332102,
      "grad_norm": 0.6612058281898499,
      "learning_rate": 2.5884737167280375e-05,
      "loss": 0.062397778034210205,
      "step": 2176
    },
    {
      "epoch": 0.2943159638019755,
      "grad_norm": 0.4066429138183594,
      "learning_rate": 2.5880026623281914e-05,
      "loss": 0.07649281620979309,
      "step": 2177
    },
    {
      "epoch": 0.2944511571707408,
      "grad_norm": 0.7098007202148438,
      "learning_rate": 2.5875313814003892e-05,
      "loss": 0.07545536756515503,
      "step": 2178
    },
    {
      "epoch": 0.29458635053950605,
      "grad_norm": 0.5521203875541687,
      "learning_rate": 2.587059874042754e-05,
      "loss": 0.07319119572639465,
      "step": 2179
    },
    {
      "epoch": 0.2947215439082713,
      "grad_norm": 0.5925273299217224,
      "learning_rate": 2.5865881403534557e-05,
      "loss": 0.05126646161079407,
      "step": 2180
    },
    {
      "epoch": 0.29485673727703654,
      "grad_norm": 0.7179078459739685,
      "learning_rate": 2.5861161804307124e-05,
      "loss": 0.07141532003879547,
      "step": 2181
    },
    {
      "epoch": 0.2949919306458018,
      "grad_norm": 0.39854222536087036,
      "learning_rate": 2.5856439943727886e-05,
      "loss": 0.06480929255485535,
      "step": 2182
    },
    {
      "epoch": 0.2951271240145671,
      "grad_norm": 0.45212194323539734,
      "learning_rate": 2.5851715822779954e-05,
      "loss": 0.05433202162384987,
      "step": 2183
    },
    {
      "epoch": 0.29526231738333236,
      "grad_norm": 1.2842271327972412,
      "learning_rate": 2.5846989442446926e-05,
      "loss": 0.06126861274242401,
      "step": 2184
    },
    {
      "epoch": 0.29539751075209764,
      "grad_norm": 0.5720062255859375,
      "learning_rate": 2.584226080371285e-05,
      "loss": 0.07492798566818237,
      "step": 2185
    },
    {
      "epoch": 0.29553270412086285,
      "grad_norm": 0.706356942653656,
      "learning_rate": 2.5837529907562258e-05,
      "loss": 0.0765361487865448,
      "step": 2186
    },
    {
      "epoch": 0.29566789748962813,
      "grad_norm": 0.5648061037063599,
      "learning_rate": 2.5832796754980138e-05,
      "loss": 0.08554080128669739,
      "step": 2187
    },
    {
      "epoch": 0.2958030908583934,
      "grad_norm": 0.3338322639465332,
      "learning_rate": 2.5828061346951974e-05,
      "loss": 0.05611315369606018,
      "step": 2188
    },
    {
      "epoch": 0.2959382842271587,
      "grad_norm": 0.47631970047950745,
      "learning_rate": 2.5823323684463693e-05,
      "loss": 0.06342458724975586,
      "step": 2189
    },
    {
      "epoch": 0.2960734775959239,
      "grad_norm": 0.44482526183128357,
      "learning_rate": 2.5818583768501708e-05,
      "loss": 0.050715237855911255,
      "step": 2190
    },
    {
      "epoch": 0.29620867096468917,
      "grad_norm": 1.0387810468673706,
      "learning_rate": 2.5813841600052887e-05,
      "loss": 0.0709274411201477,
      "step": 2191
    },
    {
      "epoch": 0.29634386433345444,
      "grad_norm": 0.5544099807739258,
      "learning_rate": 2.580909718010458e-05,
      "loss": 0.06838691234588623,
      "step": 2192
    },
    {
      "epoch": 0.2964790577022197,
      "grad_norm": 0.7458712458610535,
      "learning_rate": 2.58043505096446e-05,
      "loss": 0.08356204628944397,
      "step": 2193
    },
    {
      "epoch": 0.296614251070985,
      "grad_norm": 0.35629406571388245,
      "learning_rate": 2.5799601589661223e-05,
      "loss": 0.04793331027030945,
      "step": 2194
    },
    {
      "epoch": 0.2967494444397502,
      "grad_norm": 1.2872802019119263,
      "learning_rate": 2.579485042114321e-05,
      "loss": 0.05886256694793701,
      "step": 2195
    },
    {
      "epoch": 0.2968846378085155,
      "grad_norm": 0.4057595431804657,
      "learning_rate": 2.5790097005079766e-05,
      "loss": 0.05406687408685684,
      "step": 2196
    },
    {
      "epoch": 0.29701983117728076,
      "grad_norm": 0.6697300672531128,
      "learning_rate": 2.5785341342460595e-05,
      "loss": 0.07221139967441559,
      "step": 2197
    },
    {
      "epoch": 0.29715502454604603,
      "grad_norm": 0.6518707275390625,
      "learning_rate": 2.5780583434275837e-05,
      "loss": 0.058696746826171875,
      "step": 2198
    },
    {
      "epoch": 0.2972902179148113,
      "grad_norm": 0.5879749655723572,
      "learning_rate": 2.577582328151612e-05,
      "loss": 0.05843890458345413,
      "step": 2199
    },
    {
      "epoch": 0.2974254112835765,
      "grad_norm": 0.8899679780006409,
      "learning_rate": 2.5771060885172532e-05,
      "loss": 0.06020255386829376,
      "step": 2200
    },
    {
      "epoch": 0.2975606046523418,
      "grad_norm": 1.8119174242019653,
      "learning_rate": 2.5766296246236628e-05,
      "loss": 0.07687734812498093,
      "step": 2201
    },
    {
      "epoch": 0.29769579802110707,
      "grad_norm": 0.8819155693054199,
      "learning_rate": 2.5761529365700437e-05,
      "loss": 0.04933363199234009,
      "step": 2202
    },
    {
      "epoch": 0.29783099138987235,
      "grad_norm": 1.332566261291504,
      "learning_rate": 2.5756760244556445e-05,
      "loss": 0.0746423602104187,
      "step": 2203
    },
    {
      "epoch": 0.29796618475863756,
      "grad_norm": 0.31044360995292664,
      "learning_rate": 2.5751988883797603e-05,
      "loss": 0.05552416667342186,
      "step": 2204
    },
    {
      "epoch": 0.29810137812740284,
      "grad_norm": 1.0255037546157837,
      "learning_rate": 2.574721528441734e-05,
      "loss": 0.05071808397769928,
      "step": 2205
    },
    {
      "epoch": 0.2982365714961681,
      "grad_norm": 0.8949851393699646,
      "learning_rate": 2.5742439447409545e-05,
      "loss": 0.05579739063978195,
      "step": 2206
    },
    {
      "epoch": 0.2983717648649334,
      "grad_norm": 0.3716931641101837,
      "learning_rate": 2.5737661373768568e-05,
      "loss": 0.06341147422790527,
      "step": 2207
    },
    {
      "epoch": 0.29850695823369866,
      "grad_norm": 0.523248553276062,
      "learning_rate": 2.5732881064489237e-05,
      "loss": 0.04970260709524155,
      "step": 2208
    },
    {
      "epoch": 0.2986421516024639,
      "grad_norm": 0.4543715715408325,
      "learning_rate": 2.572809852056683e-05,
      "loss": 0.04435265064239502,
      "step": 2209
    },
    {
      "epoch": 0.29877734497122915,
      "grad_norm": 0.36390364170074463,
      "learning_rate": 2.572331374299711e-05,
      "loss": 0.060278236865997314,
      "step": 2210
    },
    {
      "epoch": 0.2989125383399944,
      "grad_norm": 2.006216049194336,
      "learning_rate": 2.5718526732776276e-05,
      "loss": 0.06773050129413605,
      "step": 2211
    },
    {
      "epoch": 0.2990477317087597,
      "grad_norm": 0.7467211484909058,
      "learning_rate": 2.5713737490901023e-05,
      "loss": 0.05065278708934784,
      "step": 2212
    },
    {
      "epoch": 0.299182925077525,
      "grad_norm": 1.6297553777694702,
      "learning_rate": 2.570894601836849e-05,
      "loss": 0.07146915793418884,
      "step": 2213
    },
    {
      "epoch": 0.2993181184462902,
      "grad_norm": 0.7554977536201477,
      "learning_rate": 2.5704152316176287e-05,
      "loss": 0.06410757452249527,
      "step": 2214
    },
    {
      "epoch": 0.29945331181505547,
      "grad_norm": 1.449791669845581,
      "learning_rate": 2.5699356385322487e-05,
      "loss": 0.06757006049156189,
      "step": 2215
    },
    {
      "epoch": 0.29958850518382074,
      "grad_norm": 0.5825949311256409,
      "learning_rate": 2.5694558226805643e-05,
      "loss": 0.06181889772415161,
      "step": 2216
    },
    {
      "epoch": 0.299723698552586,
      "grad_norm": 0.8457855582237244,
      "learning_rate": 2.568975784162474e-05,
      "loss": 0.04547801613807678,
      "step": 2217
    },
    {
      "epoch": 0.29985889192135123,
      "grad_norm": 0.6279097199440002,
      "learning_rate": 2.5684955230779245e-05,
      "loss": 0.05461207777261734,
      "step": 2218
    },
    {
      "epoch": 0.2999940852901165,
      "grad_norm": 0.3246513605117798,
      "learning_rate": 2.5680150395269096e-05,
      "loss": 0.04975739121437073,
      "step": 2219
    },
    {
      "epoch": 0.3001292786588818,
      "grad_norm": 0.2790122330188751,
      "learning_rate": 2.5675343336094683e-05,
      "loss": 0.04424440115690231,
      "step": 2220
    },
    {
      "epoch": 0.30026447202764706,
      "grad_norm": 0.6023923754692078,
      "learning_rate": 2.5670534054256855e-05,
      "loss": 0.06825404614210129,
      "step": 2221
    },
    {
      "epoch": 0.30039966539641233,
      "grad_norm": 0.4277121424674988,
      "learning_rate": 2.5665722550756937e-05,
      "loss": 0.05392634496092796,
      "step": 2222
    },
    {
      "epoch": 0.30053485876517755,
      "grad_norm": 0.534571647644043,
      "learning_rate": 2.5660908826596707e-05,
      "loss": 0.0506836473941803,
      "step": 2223
    },
    {
      "epoch": 0.3006700521339428,
      "grad_norm": 0.8310690522193909,
      "learning_rate": 2.5656092882778413e-05,
      "loss": 0.05335317552089691,
      "step": 2224
    },
    {
      "epoch": 0.3008052455027081,
      "grad_norm": 1.2586736679077148,
      "learning_rate": 2.565127472030475e-05,
      "loss": 0.06451964378356934,
      "step": 2225
    },
    {
      "epoch": 0.30094043887147337,
      "grad_norm": 0.854870080947876,
      "learning_rate": 2.5646454340178894e-05,
      "loss": 0.06391061097383499,
      "step": 2226
    },
    {
      "epoch": 0.3010756322402386,
      "grad_norm": 0.41664719581604004,
      "learning_rate": 2.564163174340447e-05,
      "loss": 0.0637710690498352,
      "step": 2227
    },
    {
      "epoch": 0.30121082560900386,
      "grad_norm": 0.9564629197120667,
      "learning_rate": 2.5636806930985565e-05,
      "loss": 0.06166139245033264,
      "step": 2228
    },
    {
      "epoch": 0.30134601897776914,
      "grad_norm": 1.286916971206665,
      "learning_rate": 2.5631979903926738e-05,
      "loss": 0.05765240266919136,
      "step": 2229
    },
    {
      "epoch": 0.3014812123465344,
      "grad_norm": 0.39217379689216614,
      "learning_rate": 2.5627150663233e-05,
      "loss": 0.07160265743732452,
      "step": 2230
    },
    {
      "epoch": 0.3016164057152997,
      "grad_norm": 0.5052042603492737,
      "learning_rate": 2.5622319209909817e-05,
      "loss": 0.060192883014678955,
      "step": 2231
    },
    {
      "epoch": 0.3017515990840649,
      "grad_norm": 1.1523587703704834,
      "learning_rate": 2.5617485544963135e-05,
      "loss": 0.0801660567522049,
      "step": 2232
    },
    {
      "epoch": 0.3018867924528302,
      "grad_norm": 0.3566419184207916,
      "learning_rate": 2.561264966939934e-05,
      "loss": 0.05440396070480347,
      "step": 2233
    },
    {
      "epoch": 0.30202198582159545,
      "grad_norm": 0.657753586769104,
      "learning_rate": 2.5607811584225294e-05,
      "loss": 0.06699851155281067,
      "step": 2234
    },
    {
      "epoch": 0.3021571791903607,
      "grad_norm": 0.5132169723510742,
      "learning_rate": 2.5602971290448305e-05,
      "loss": 0.0679093524813652,
      "step": 2235
    },
    {
      "epoch": 0.302292372559126,
      "grad_norm": 0.614510715007782,
      "learning_rate": 2.5598128789076152e-05,
      "loss": 0.0653022825717926,
      "step": 2236
    },
    {
      "epoch": 0.3024275659278912,
      "grad_norm": 1.667647361755371,
      "learning_rate": 2.559328408111707e-05,
      "loss": 0.06635583937168121,
      "step": 2237
    },
    {
      "epoch": 0.3025627592966565,
      "grad_norm": 0.348704993724823,
      "learning_rate": 2.5588437167579755e-05,
      "loss": 0.04212060943245888,
      "step": 2238
    },
    {
      "epoch": 0.30269795266542177,
      "grad_norm": 0.5667931437492371,
      "learning_rate": 2.558358804947335e-05,
      "loss": 0.054676353931427,
      "step": 2239
    },
    {
      "epoch": 0.30283314603418704,
      "grad_norm": 2.80304217338562,
      "learning_rate": 2.557873672780748e-05,
      "loss": 0.07425614446401596,
      "step": 2240
    },
    {
      "epoch": 0.30296833940295226,
      "grad_norm": 0.4265073835849762,
      "learning_rate": 2.557388320359221e-05,
      "loss": 0.04567320644855499,
      "step": 2241
    },
    {
      "epoch": 0.30310353277171753,
      "grad_norm": 0.47821781039237976,
      "learning_rate": 2.5569027477838068e-05,
      "loss": 0.05417218804359436,
      "step": 2242
    },
    {
      "epoch": 0.3032387261404828,
      "grad_norm": 1.1858329772949219,
      "learning_rate": 2.5564169551556044e-05,
      "loss": 0.05075885355472565,
      "step": 2243
    },
    {
      "epoch": 0.3033739195092481,
      "grad_norm": 0.6808141469955444,
      "learning_rate": 2.5559309425757586e-05,
      "loss": 0.06217425316572189,
      "step": 2244
    },
    {
      "epoch": 0.30350911287801335,
      "grad_norm": 1.2015057802200317,
      "learning_rate": 2.5554447101454597e-05,
      "loss": 0.05612587183713913,
      "step": 2245
    },
    {
      "epoch": 0.3036443062467786,
      "grad_norm": 1.3352810144424438,
      "learning_rate": 2.554958257965944e-05,
      "loss": 0.07060706615447998,
      "step": 2246
    },
    {
      "epoch": 0.30377949961554385,
      "grad_norm": 1.2676361799240112,
      "learning_rate": 2.554471586138493e-05,
      "loss": 0.053166020661592484,
      "step": 2247
    },
    {
      "epoch": 0.3039146929843091,
      "grad_norm": 1.0703836679458618,
      "learning_rate": 2.5539846947644342e-05,
      "loss": 0.059078484773635864,
      "step": 2248
    },
    {
      "epoch": 0.3040498863530744,
      "grad_norm": 0.7100827097892761,
      "learning_rate": 2.5534975839451416e-05,
      "loss": 0.059625498950481415,
      "step": 2249
    },
    {
      "epoch": 0.30418507972183967,
      "grad_norm": 1.0836572647094727,
      "learning_rate": 2.5530102537820348e-05,
      "loss": 0.06885454803705215,
      "step": 2250
    },
    {
      "epoch": 0.3043202730906049,
      "grad_norm": 0.6359640955924988,
      "learning_rate": 2.5525227043765774e-05,
      "loss": 0.04971761628985405,
      "step": 2251
    },
    {
      "epoch": 0.30445546645937016,
      "grad_norm": 0.6875571012496948,
      "learning_rate": 2.55203493583028e-05,
      "loss": 0.0724475234746933,
      "step": 2252
    },
    {
      "epoch": 0.30459065982813543,
      "grad_norm": 0.9050090312957764,
      "learning_rate": 2.551546948244699e-05,
      "loss": 0.046121858060359955,
      "step": 2253
    },
    {
      "epoch": 0.3047258531969007,
      "grad_norm": 0.32779523730278015,
      "learning_rate": 2.551058741721436e-05,
      "loss": 0.05801308900117874,
      "step": 2254
    },
    {
      "epoch": 0.3048610465656659,
      "grad_norm": 0.30548548698425293,
      "learning_rate": 2.550570316362138e-05,
      "loss": 0.047217901796102524,
      "step": 2255
    },
    {
      "epoch": 0.3049962399344312,
      "grad_norm": 1.3277095556259155,
      "learning_rate": 2.5500816722684975e-05,
      "loss": 0.06256397813558578,
      "step": 2256
    },
    {
      "epoch": 0.3051314333031965,
      "grad_norm": 0.7082242369651794,
      "learning_rate": 2.549592809542253e-05,
      "loss": 0.06021872162818909,
      "step": 2257
    },
    {
      "epoch": 0.30526662667196175,
      "grad_norm": 1.5869684219360352,
      "learning_rate": 2.549103728285189e-05,
      "loss": 0.055726319551467896,
      "step": 2258
    },
    {
      "epoch": 0.305401820040727,
      "grad_norm": 0.4927757680416107,
      "learning_rate": 2.548614428599134e-05,
      "loss": 0.048781394958496094,
      "step": 2259
    },
    {
      "epoch": 0.30553701340949224,
      "grad_norm": 0.48715919256210327,
      "learning_rate": 2.5481249105859633e-05,
      "loss": 0.039811015129089355,
      "step": 2260
    },
    {
      "epoch": 0.3056722067782575,
      "grad_norm": 0.5090923309326172,
      "learning_rate": 2.5476351743475964e-05,
      "loss": 0.06070142984390259,
      "step": 2261
    },
    {
      "epoch": 0.3058074001470228,
      "grad_norm": 1.5543603897094727,
      "learning_rate": 2.547145219986e-05,
      "loss": 0.054766714572906494,
      "step": 2262
    },
    {
      "epoch": 0.30594259351578806,
      "grad_norm": 0.726177990436554,
      "learning_rate": 2.5466550476031846e-05,
      "loss": 0.037304095923900604,
      "step": 2263
    },
    {
      "epoch": 0.30607778688455334,
      "grad_norm": 1.7814775705337524,
      "learning_rate": 2.5461646573012072e-05,
      "loss": 0.06137313321232796,
      "step": 2264
    },
    {
      "epoch": 0.30621298025331856,
      "grad_norm": 0.7154324054718018,
      "learning_rate": 2.5456740491821687e-05,
      "loss": 0.06463988125324249,
      "step": 2265
    },
    {
      "epoch": 0.30634817362208383,
      "grad_norm": 0.40536412596702576,
      "learning_rate": 2.5451832233482172e-05,
      "loss": 0.05746598541736603,
      "step": 2266
    },
    {
      "epoch": 0.3064833669908491,
      "grad_norm": 0.451678067445755,
      "learning_rate": 2.544692179901545e-05,
      "loss": 0.044037047773599625,
      "step": 2267
    },
    {
      "epoch": 0.3066185603596144,
      "grad_norm": 0.40946945548057556,
      "learning_rate": 2.5442009189443902e-05,
      "loss": 0.052533067762851715,
      "step": 2268
    },
    {
      "epoch": 0.3067537537283796,
      "grad_norm": 0.37241262197494507,
      "learning_rate": 2.5437094405790355e-05,
      "loss": 0.06470764428377151,
      "step": 2269
    },
    {
      "epoch": 0.30688894709714487,
      "grad_norm": 0.559974730014801,
      "learning_rate": 2.5432177449078096e-05,
      "loss": 0.05968666449189186,
      "step": 2270
    },
    {
      "epoch": 0.30702414046591014,
      "grad_norm": 0.6734687685966492,
      "learning_rate": 2.5427258320330857e-05,
      "loss": 0.056102193892002106,
      "step": 2271
    },
    {
      "epoch": 0.3071593338346754,
      "grad_norm": 0.4163251519203186,
      "learning_rate": 2.5422337020572835e-05,
      "loss": 0.05599336698651314,
      "step": 2272
    },
    {
      "epoch": 0.3072945272034407,
      "grad_norm": 1.0217441320419312,
      "learning_rate": 2.5417413550828667e-05,
      "loss": 0.06663127988576889,
      "step": 2273
    },
    {
      "epoch": 0.3074297205722059,
      "grad_norm": 0.23586855828762054,
      "learning_rate": 2.5412487912123444e-05,
      "loss": 0.03475649654865265,
      "step": 2274
    },
    {
      "epoch": 0.3075649139409712,
      "grad_norm": 2.002098321914673,
      "learning_rate": 2.5407560105482708e-05,
      "loss": 0.06683699041604996,
      "step": 2275
    },
    {
      "epoch": 0.30770010730973646,
      "grad_norm": 0.3159419894218445,
      "learning_rate": 2.540263013193246e-05,
      "loss": 0.038786984980106354,
      "step": 2276
    },
    {
      "epoch": 0.30783530067850173,
      "grad_norm": 0.8217936754226685,
      "learning_rate": 2.539769799249915e-05,
      "loss": 0.07454089820384979,
      "step": 2277
    },
    {
      "epoch": 0.307970494047267,
      "grad_norm": 0.9815629720687866,
      "learning_rate": 2.5392763688209666e-05,
      "loss": 0.06661137193441391,
      "step": 2278
    },
    {
      "epoch": 0.3081056874160322,
      "grad_norm": 1.1893088817596436,
      "learning_rate": 2.5387827220091362e-05,
      "loss": 0.05191679298877716,
      "step": 2279
    },
    {
      "epoch": 0.3082408807847975,
      "grad_norm": 0.6995902061462402,
      "learning_rate": 2.538288858917204e-05,
      "loss": 0.05519968271255493,
      "step": 2280
    },
    {
      "epoch": 0.3083760741535628,
      "grad_norm": 1.0510425567626953,
      "learning_rate": 2.5377947796479936e-05,
      "loss": 0.05884440243244171,
      "step": 2281
    },
    {
      "epoch": 0.30851126752232805,
      "grad_norm": 0.5182132124900818,
      "learning_rate": 2.537300484304377e-05,
      "loss": 0.05221012979745865,
      "step": 2282
    },
    {
      "epoch": 0.30864646089109327,
      "grad_norm": 0.3225332498550415,
      "learning_rate": 2.536805972989267e-05,
      "loss": 0.03802039474248886,
      "step": 2283
    },
    {
      "epoch": 0.30878165425985854,
      "grad_norm": 0.7955539226531982,
      "learning_rate": 2.5363112458056252e-05,
      "loss": 0.051208097487688065,
      "step": 2284
    },
    {
      "epoch": 0.3089168476286238,
      "grad_norm": 1.0788735151290894,
      "learning_rate": 2.5358163028564552e-05,
      "loss": 0.07112982869148254,
      "step": 2285
    },
    {
      "epoch": 0.3090520409973891,
      "grad_norm": 0.375968873500824,
      "learning_rate": 2.535321144244808e-05,
      "loss": 0.06154714897274971,
      "step": 2286
    },
    {
      "epoch": 0.30918723436615436,
      "grad_norm": 1.073516607284546,
      "learning_rate": 2.534825770073777e-05,
      "loss": 0.04923577234148979,
      "step": 2287
    },
    {
      "epoch": 0.3093224277349196,
      "grad_norm": 0.5616546869277954,
      "learning_rate": 2.5343301804465026e-05,
      "loss": 0.058992281556129456,
      "step": 2288
    },
    {
      "epoch": 0.30945762110368485,
      "grad_norm": 0.6748343706130981,
      "learning_rate": 2.533834375466169e-05,
      "loss": 0.042425546795129776,
      "step": 2289
    },
    {
      "epoch": 0.30959281447245013,
      "grad_norm": 0.910538911819458,
      "learning_rate": 2.533338355236005e-05,
      "loss": 0.043878909200429916,
      "step": 2290
    },
    {
      "epoch": 0.3097280078412154,
      "grad_norm": 1.8196405172348022,
      "learning_rate": 2.532842119859285e-05,
      "loss": 0.05300411581993103,
      "step": 2291
    },
    {
      "epoch": 0.3098632012099807,
      "grad_norm": 0.3632139265537262,
      "learning_rate": 2.532345669439328e-05,
      "loss": 0.04527851939201355,
      "step": 2292
    },
    {
      "epoch": 0.3099983945787459,
      "grad_norm": 0.6463128328323364,
      "learning_rate": 2.5318490040794975e-05,
      "loss": 0.06771989166736603,
      "step": 2293
    },
    {
      "epoch": 0.31013358794751117,
      "grad_norm": 1.31852126121521,
      "learning_rate": 2.531352123883202e-05,
      "loss": 0.06382325291633606,
      "step": 2294
    },
    {
      "epoch": 0.31026878131627644,
      "grad_norm": 0.793941080570221,
      "learning_rate": 2.530855028953894e-05,
      "loss": 0.06026911735534668,
      "step": 2295
    },
    {
      "epoch": 0.3104039746850417,
      "grad_norm": 0.7083117365837097,
      "learning_rate": 2.5303577193950724e-05,
      "loss": 0.0671968162059784,
      "step": 2296
    },
    {
      "epoch": 0.31053916805380694,
      "grad_norm": 0.7987262606620789,
      "learning_rate": 2.5298601953102785e-05,
      "loss": 0.06650563329458237,
      "step": 2297
    },
    {
      "epoch": 0.3106743614225722,
      "grad_norm": 0.36385875940322876,
      "learning_rate": 2.5293624568031008e-05,
      "loss": 0.057412274181842804,
      "step": 2298
    },
    {
      "epoch": 0.3108095547913375,
      "grad_norm": 0.8501790761947632,
      "learning_rate": 2.5288645039771697e-05,
      "loss": 0.04437267780303955,
      "step": 2299
    },
    {
      "epoch": 0.31094474816010276,
      "grad_norm": 1.2511833906173706,
      "learning_rate": 2.5283663369361624e-05,
      "loss": 0.07001717388629913,
      "step": 2300
    },
    {
      "epoch": 0.31107994152886803,
      "grad_norm": 0.4312867522239685,
      "learning_rate": 2.5278679557837998e-05,
      "loss": 0.04859645664691925,
      "step": 2301
    },
    {
      "epoch": 0.31121513489763325,
      "grad_norm": 0.6450917720794678,
      "learning_rate": 2.5273693606238474e-05,
      "loss": 0.04278920590877533,
      "step": 2302
    },
    {
      "epoch": 0.3113503282663985,
      "grad_norm": 0.2746388912200928,
      "learning_rate": 2.5268705515601164e-05,
      "loss": 0.05155005306005478,
      "step": 2303
    },
    {
      "epoch": 0.3114855216351638,
      "grad_norm": 0.3895500898361206,
      "learning_rate": 2.5263715286964596e-05,
      "loss": 0.059623122215270996,
      "step": 2304
    },
    {
      "epoch": 0.31162071500392907,
      "grad_norm": 0.6174415946006775,
      "learning_rate": 2.525872292136778e-05,
      "loss": 0.06191803514957428,
      "step": 2305
    },
    {
      "epoch": 0.31175590837269435,
      "grad_norm": 0.6154640316963196,
      "learning_rate": 2.525372841985014e-05,
      "loss": 0.03563769906759262,
      "step": 2306
    },
    {
      "epoch": 0.31189110174145956,
      "grad_norm": 0.699211835861206,
      "learning_rate": 2.5248731783451567e-05,
      "loss": 0.05020824447274208,
      "step": 2307
    },
    {
      "epoch": 0.31202629511022484,
      "grad_norm": 0.3884332478046417,
      "learning_rate": 2.524373301321238e-05,
      "loss": 0.046756379306316376,
      "step": 2308
    },
    {
      "epoch": 0.3121614884789901,
      "grad_norm": 0.5133849382400513,
      "learning_rate": 2.5238732110173356e-05,
      "loss": 0.07218366861343384,
      "step": 2309
    },
    {
      "epoch": 0.3122966818477554,
      "grad_norm": 0.4838017225265503,
      "learning_rate": 2.5233729075375708e-05,
      "loss": 0.0739641785621643,
      "step": 2310
    },
    {
      "epoch": 0.3124318752165206,
      "grad_norm": 0.41110891103744507,
      "learning_rate": 2.522872390986109e-05,
      "loss": 0.04120361804962158,
      "step": 2311
    },
    {
      "epoch": 0.3125670685852859,
      "grad_norm": 0.7661836743354797,
      "learning_rate": 2.522371661467161e-05,
      "loss": 0.04032690078020096,
      "step": 2312
    },
    {
      "epoch": 0.31270226195405115,
      "grad_norm": 0.5305105447769165,
      "learning_rate": 2.521870719084981e-05,
      "loss": 0.05806836485862732,
      "step": 2313
    },
    {
      "epoch": 0.3128374553228164,
      "grad_norm": 0.47661060094833374,
      "learning_rate": 2.5213695639438686e-05,
      "loss": 0.07388916611671448,
      "step": 2314
    },
    {
      "epoch": 0.3129726486915817,
      "grad_norm": 0.8824151158332825,
      "learning_rate": 2.5208681961481657e-05,
      "loss": 0.07833479344844818,
      "step": 2315
    },
    {
      "epoch": 0.3131078420603469,
      "grad_norm": 1.5235333442687988,
      "learning_rate": 2.5203666158022607e-05,
      "loss": 0.05425835773348808,
      "step": 2316
    },
    {
      "epoch": 0.3132430354291122,
      "grad_norm": 1.2147819995880127,
      "learning_rate": 2.519864823010585e-05,
      "loss": 0.05932708457112312,
      "step": 2317
    },
    {
      "epoch": 0.31337822879787747,
      "grad_norm": 0.73869389295578,
      "learning_rate": 2.5193628178776148e-05,
      "loss": 0.06545159220695496,
      "step": 2318
    },
    {
      "epoch": 0.31351342216664274,
      "grad_norm": 0.3748724162578583,
      "learning_rate": 2.5188606005078695e-05,
      "loss": 0.05377781391143799,
      "step": 2319
    },
    {
      "epoch": 0.313648615535408,
      "grad_norm": 1.026969313621521,
      "learning_rate": 2.518358171005914e-05,
      "loss": 0.07322761416435242,
      "step": 2320
    },
    {
      "epoch": 0.31378380890417323,
      "grad_norm": 0.8221989274024963,
      "learning_rate": 2.517855529476357e-05,
      "loss": 0.0666864737868309,
      "step": 2321
    },
    {
      "epoch": 0.3139190022729385,
      "grad_norm": 0.8605358004570007,
      "learning_rate": 2.517352676023851e-05,
      "loss": 0.04999266564846039,
      "step": 2322
    },
    {
      "epoch": 0.3140541956417038,
      "grad_norm": 0.8627109527587891,
      "learning_rate": 2.5168496107530925e-05,
      "loss": 0.08818463981151581,
      "step": 2323
    },
    {
      "epoch": 0.31418938901046906,
      "grad_norm": 0.5400346517562866,
      "learning_rate": 2.5163463337688224e-05,
      "loss": 0.05868957191705704,
      "step": 2324
    },
    {
      "epoch": 0.3143245823792343,
      "grad_norm": 1.3682141304016113,
      "learning_rate": 2.515842845175826e-05,
      "loss": 0.05405328795313835,
      "step": 2325
    },
    {
      "epoch": 0.31445977574799955,
      "grad_norm": 0.6122546792030334,
      "learning_rate": 2.5153391450789326e-05,
      "loss": 0.06808657944202423,
      "step": 2326
    },
    {
      "epoch": 0.3145949691167648,
      "grad_norm": 0.6714510321617126,
      "learning_rate": 2.514835233583014e-05,
      "loss": 0.05392294377088547,
      "step": 2327
    },
    {
      "epoch": 0.3147301624855301,
      "grad_norm": 0.5753044486045837,
      "learning_rate": 2.514331110792988e-05,
      "loss": 0.0617678165435791,
      "step": 2328
    },
    {
      "epoch": 0.31486535585429537,
      "grad_norm": 0.8453408479690552,
      "learning_rate": 2.513826776813816e-05,
      "loss": 0.0437600314617157,
      "step": 2329
    },
    {
      "epoch": 0.3150005492230606,
      "grad_norm": 1.5375425815582275,
      "learning_rate": 2.5133222317505024e-05,
      "loss": 0.058279894292354584,
      "step": 2330
    },
    {
      "epoch": 0.31513574259182586,
      "grad_norm": 0.22465813159942627,
      "learning_rate": 2.5128174757080965e-05,
      "loss": 0.0558398962020874,
      "step": 2331
    },
    {
      "epoch": 0.31527093596059114,
      "grad_norm": 0.7125957608222961,
      "learning_rate": 2.5123125087916916e-05,
      "loss": 0.06689736247062683,
      "step": 2332
    },
    {
      "epoch": 0.3154061293293564,
      "grad_norm": 0.7507221698760986,
      "learning_rate": 2.5118073311064236e-05,
      "loss": 0.06189854443073273,
      "step": 2333
    },
    {
      "epoch": 0.3155413226981217,
      "grad_norm": 0.7014270424842834,
      "learning_rate": 2.5113019427574734e-05,
      "loss": 0.04525640979409218,
      "step": 2334
    },
    {
      "epoch": 0.3156765160668869,
      "grad_norm": 0.692531943321228,
      "learning_rate": 2.5107963438500666e-05,
      "loss": 0.04493245482444763,
      "step": 2335
    },
    {
      "epoch": 0.3158117094356522,
      "grad_norm": 0.8417144417762756,
      "learning_rate": 2.51029053448947e-05,
      "loss": 0.055364251136779785,
      "step": 2336
    },
    {
      "epoch": 0.31594690280441745,
      "grad_norm": 1.539441466331482,
      "learning_rate": 2.509784514780997e-05,
      "loss": 0.06964397430419922,
      "step": 2337
    },
    {
      "epoch": 0.3160820961731827,
      "grad_norm": 0.40144404768943787,
      "learning_rate": 2.5092782848300033e-05,
      "loss": 0.05018016695976257,
      "step": 2338
    },
    {
      "epoch": 0.31621728954194794,
      "grad_norm": 0.47542232275009155,
      "learning_rate": 2.5087718447418886e-05,
      "loss": 0.05702904611825943,
      "step": 2339
    },
    {
      "epoch": 0.3163524829107132,
      "grad_norm": 0.4485418200492859,
      "learning_rate": 2.5082651946220958e-05,
      "loss": 0.06316009163856506,
      "step": 2340
    },
    {
      "epoch": 0.3164876762794785,
      "grad_norm": 1.029420018196106,
      "learning_rate": 2.507758334576113e-05,
      "loss": 0.06851062178611755,
      "step": 2341
    },
    {
      "epoch": 0.31662286964824377,
      "grad_norm": 1.2221734523773193,
      "learning_rate": 2.5072512647094713e-05,
      "loss": 0.05392313003540039,
      "step": 2342
    },
    {
      "epoch": 0.31675806301700904,
      "grad_norm": 0.9792153239250183,
      "learning_rate": 2.506743985127745e-05,
      "loss": 0.07204374670982361,
      "step": 2343
    },
    {
      "epoch": 0.31689325638577426,
      "grad_norm": 0.5060032606124878,
      "learning_rate": 2.506236495936552e-05,
      "loss": 0.055379390716552734,
      "step": 2344
    },
    {
      "epoch": 0.31702844975453953,
      "grad_norm": 0.3366352319717407,
      "learning_rate": 2.5057287972415547e-05,
      "loss": 0.0541258230805397,
      "step": 2345
    },
    {
      "epoch": 0.3171636431233048,
      "grad_norm": 0.3888050317764282,
      "learning_rate": 2.5052208891484588e-05,
      "loss": 0.05308638513088226,
      "step": 2346
    },
    {
      "epoch": 0.3172988364920701,
      "grad_norm": 1.1438707113265991,
      "learning_rate": 2.504712771763013e-05,
      "loss": 0.06358611583709717,
      "step": 2347
    },
    {
      "epoch": 0.3174340298608353,
      "grad_norm": 0.8864173293113708,
      "learning_rate": 2.5042044451910108e-05,
      "loss": 0.06125408038496971,
      "step": 2348
    },
    {
      "epoch": 0.3175692232296006,
      "grad_norm": 0.3055857717990875,
      "learning_rate": 2.5036959095382875e-05,
      "loss": 0.06520285457372665,
      "step": 2349
    },
    {
      "epoch": 0.31770441659836585,
      "grad_norm": 1.0140036344528198,
      "learning_rate": 2.5031871649107233e-05,
      "loss": 0.03953090310096741,
      "step": 2350
    },
    {
      "epoch": 0.3178396099671311,
      "grad_norm": 1.01449716091156,
      "learning_rate": 2.5026782114142426e-05,
      "loss": 0.05470992624759674,
      "step": 2351
    },
    {
      "epoch": 0.3179748033358964,
      "grad_norm": 0.47396790981292725,
      "learning_rate": 2.5021690491548107e-05,
      "loss": 0.05226321518421173,
      "step": 2352
    },
    {
      "epoch": 0.3181099967046616,
      "grad_norm": 1.720629334449768,
      "learning_rate": 2.5016596782384387e-05,
      "loss": 0.06110893189907074,
      "step": 2353
    },
    {
      "epoch": 0.3182451900734269,
      "grad_norm": 0.32993414998054504,
      "learning_rate": 2.5011500987711804e-05,
      "loss": 0.041723668575286865,
      "step": 2354
    },
    {
      "epoch": 0.31838038344219216,
      "grad_norm": 0.8112986087799072,
      "learning_rate": 2.5006403108591325e-05,
      "loss": 0.061730362474918365,
      "step": 2355
    },
    {
      "epoch": 0.31851557681095743,
      "grad_norm": 0.878517210483551,
      "learning_rate": 2.500130314608436e-05,
      "loss": 0.059821128845214844,
      "step": 2356
    },
    {
      "epoch": 0.3186507701797227,
      "grad_norm": 0.5326772332191467,
      "learning_rate": 2.4996201101252742e-05,
      "loss": 0.043070752173662186,
      "step": 2357
    },
    {
      "epoch": 0.3187859635484879,
      "grad_norm": 0.2386907935142517,
      "learning_rate": 2.4991096975158757e-05,
      "loss": 0.050444453954696655,
      "step": 2358
    },
    {
      "epoch": 0.3189211569172532,
      "grad_norm": 0.3969920873641968,
      "learning_rate": 2.4985990768865095e-05,
      "loss": 0.060389913618564606,
      "step": 2359
    },
    {
      "epoch": 0.3190563502860185,
      "grad_norm": 0.7786385416984558,
      "learning_rate": 2.4980882483434904e-05,
      "loss": 0.048563480377197266,
      "step": 2360
    },
    {
      "epoch": 0.31919154365478375,
      "grad_norm": 0.7249764800071716,
      "learning_rate": 2.497577211993176e-05,
      "loss": 0.07420365512371063,
      "step": 2361
    },
    {
      "epoch": 0.31932673702354897,
      "grad_norm": 2.392258644104004,
      "learning_rate": 2.4970659679419658e-05,
      "loss": 0.07418099045753479,
      "step": 2362
    },
    {
      "epoch": 0.31946193039231424,
      "grad_norm": 0.2727169692516327,
      "learning_rate": 2.496554516296304e-05,
      "loss": 0.055803120136260986,
      "step": 2363
    },
    {
      "epoch": 0.3195971237610795,
      "grad_norm": 0.8510539531707764,
      "learning_rate": 2.4960428571626784e-05,
      "loss": 0.06812772154808044,
      "step": 2364
    },
    {
      "epoch": 0.3197323171298448,
      "grad_norm": 0.3870421350002289,
      "learning_rate": 2.4955309906476177e-05,
      "loss": 0.06500202417373657,
      "step": 2365
    },
    {
      "epoch": 0.31986751049861006,
      "grad_norm": 0.33770906925201416,
      "learning_rate": 2.495018916857696e-05,
      "loss": 0.05120396241545677,
      "step": 2366
    },
    {
      "epoch": 0.3200027038673753,
      "grad_norm": 0.5311005711555481,
      "learning_rate": 2.4945066358995304e-05,
      "loss": 0.04961118474602699,
      "step": 2367
    },
    {
      "epoch": 0.32013789723614056,
      "grad_norm": 0.5402520895004272,
      "learning_rate": 2.493994147879779e-05,
      "loss": 0.0555969774723053,
      "step": 2368
    },
    {
      "epoch": 0.32027309060490583,
      "grad_norm": 0.8443195223808289,
      "learning_rate": 2.4934814529051458e-05,
      "loss": 0.05740940570831299,
      "step": 2369
    },
    {
      "epoch": 0.3204082839736711,
      "grad_norm": 0.2561611831188202,
      "learning_rate": 2.4929685510823763e-05,
      "loss": 0.05358386039733887,
      "step": 2370
    },
    {
      "epoch": 0.3205434773424364,
      "grad_norm": 0.8193274736404419,
      "learning_rate": 2.492455442518259e-05,
      "loss": 0.05188105255365372,
      "step": 2371
    },
    {
      "epoch": 0.3206786707112016,
      "grad_norm": 0.6803062558174133,
      "learning_rate": 2.4919421273196262e-05,
      "loss": 0.05494639277458191,
      "step": 2372
    },
    {
      "epoch": 0.32081386407996687,
      "grad_norm": 0.5245711207389832,
      "learning_rate": 2.4914286055933527e-05,
      "loss": 0.05685272812843323,
      "step": 2373
    },
    {
      "epoch": 0.32094905744873214,
      "grad_norm": 1.1571366786956787,
      "learning_rate": 2.4909148774463572e-05,
      "loss": 0.04307565093040466,
      "step": 2374
    },
    {
      "epoch": 0.3210842508174974,
      "grad_norm": 0.43709444999694824,
      "learning_rate": 2.4904009429855992e-05,
      "loss": 0.06487181782722473,
      "step": 2375
    },
    {
      "epoch": 0.32121944418626264,
      "grad_norm": 0.8173585534095764,
      "learning_rate": 2.4898868023180844e-05,
      "loss": 0.048045456409454346,
      "step": 2376
    },
    {
      "epoch": 0.3213546375550279,
      "grad_norm": 0.45800986886024475,
      "learning_rate": 2.4893724555508575e-05,
      "loss": 0.06433533132076263,
      "step": 2377
    },
    {
      "epoch": 0.3214898309237932,
      "grad_norm": 0.3575770854949951,
      "learning_rate": 2.4888579027910105e-05,
      "loss": 0.05344129353761673,
      "step": 2378
    },
    {
      "epoch": 0.32162502429255846,
      "grad_norm": 0.4732600748538971,
      "learning_rate": 2.4883431441456738e-05,
      "loss": 0.07580220699310303,
      "step": 2379
    },
    {
      "epoch": 0.32176021766132373,
      "grad_norm": 0.508769154548645,
      "learning_rate": 2.4878281797220244e-05,
      "loss": 0.04080207645893097,
      "step": 2380
    },
    {
      "epoch": 0.32189541103008895,
      "grad_norm": 0.6944779753684998,
      "learning_rate": 2.4873130096272805e-05,
      "loss": 0.07719346880912781,
      "step": 2381
    },
    {
      "epoch": 0.3220306043988542,
      "grad_norm": 0.3551993668079376,
      "learning_rate": 2.4867976339687026e-05,
      "loss": 0.043351590633392334,
      "step": 2382
    },
    {
      "epoch": 0.3221657977676195,
      "grad_norm": 0.4997990131378174,
      "learning_rate": 2.4862820528535955e-05,
      "loss": 0.04125227406620979,
      "step": 2383
    },
    {
      "epoch": 0.3223009911363848,
      "grad_norm": 0.27644866704940796,
      "learning_rate": 2.4857662663893054e-05,
      "loss": 0.03502889722585678,
      "step": 2384
    },
    {
      "epoch": 0.32243618450515005,
      "grad_norm": 0.8085530400276184,
      "learning_rate": 2.485250274683222e-05,
      "loss": 0.05728539824485779,
      "step": 2385
    },
    {
      "epoch": 0.32257137787391527,
      "grad_norm": 1.5582599639892578,
      "learning_rate": 2.4847340778427772e-05,
      "loss": 0.05420401692390442,
      "step": 2386
    },
    {
      "epoch": 0.32270657124268054,
      "grad_norm": 0.4201025366783142,
      "learning_rate": 2.484217675975446e-05,
      "loss": 0.06086397171020508,
      "step": 2387
    },
    {
      "epoch": 0.3228417646114458,
      "grad_norm": 0.5157440304756165,
      "learning_rate": 2.4837010691887466e-05,
      "loss": 0.0627305805683136,
      "step": 2388
    },
    {
      "epoch": 0.3229769579802111,
      "grad_norm": 0.7646204233169556,
      "learning_rate": 2.4831842575902383e-05,
      "loss": 0.046876609325408936,
      "step": 2389
    },
    {
      "epoch": 0.3231121513489763,
      "grad_norm": 0.8501638770103455,
      "learning_rate": 2.482667241287525e-05,
      "loss": 0.06470996141433716,
      "step": 2390
    },
    {
      "epoch": 0.3232473447177416,
      "grad_norm": 1.5645791292190552,
      "learning_rate": 2.4821500203882517e-05,
      "loss": 0.06021870672702789,
      "step": 2391
    },
    {
      "epoch": 0.32338253808650685,
      "grad_norm": 1.2629863023757935,
      "learning_rate": 2.4816325950001067e-05,
      "loss": 0.04962673783302307,
      "step": 2392
    },
    {
      "epoch": 0.32351773145527213,
      "grad_norm": 0.4654386341571808,
      "learning_rate": 2.4811149652308205e-05,
      "loss": 0.04819222912192345,
      "step": 2393
    },
    {
      "epoch": 0.3236529248240374,
      "grad_norm": 0.5356674790382385,
      "learning_rate": 2.480597131188167e-05,
      "loss": 0.06122288107872009,
      "step": 2394
    },
    {
      "epoch": 0.3237881181928026,
      "grad_norm": 0.6159305572509766,
      "learning_rate": 2.4800790929799614e-05,
      "loss": 0.06149455904960632,
      "step": 2395
    },
    {
      "epoch": 0.3239233115615679,
      "grad_norm": 0.7850677371025085,
      "learning_rate": 2.4795608507140623e-05,
      "loss": 0.0695309042930603,
      "step": 2396
    },
    {
      "epoch": 0.32405850493033317,
      "grad_norm": 1.8402515649795532,
      "learning_rate": 2.4790424044983705e-05,
      "loss": 0.07553311437368393,
      "step": 2397
    },
    {
      "epoch": 0.32419369829909844,
      "grad_norm": 0.8507335186004639,
      "learning_rate": 2.4785237544408288e-05,
      "loss": 0.041149843484163284,
      "step": 2398
    },
    {
      "epoch": 0.3243288916678637,
      "grad_norm": 1.0684906244277954,
      "learning_rate": 2.478004900649424e-05,
      "loss": 0.05163618177175522,
      "step": 2399
    },
    {
      "epoch": 0.32446408503662894,
      "grad_norm": 0.7404788136482239,
      "learning_rate": 2.477485843232183e-05,
      "loss": 0.058647915720939636,
      "step": 2400
    },
    {
      "epoch": 0.3245992784053942,
      "grad_norm": 0.1775343120098114,
      "learning_rate": 2.476966582297177e-05,
      "loss": 0.04130779951810837,
      "step": 2401
    },
    {
      "epoch": 0.3247344717741595,
      "grad_norm": 0.3759896159172058,
      "learning_rate": 2.4764471179525188e-05,
      "loss": 0.04953281581401825,
      "step": 2402
    },
    {
      "epoch": 0.32486966514292476,
      "grad_norm": 1.2053723335266113,
      "learning_rate": 2.4759274503063632e-05,
      "loss": 0.05366501957178116,
      "step": 2403
    },
    {
      "epoch": 0.32500485851169,
      "grad_norm": 0.45025211572647095,
      "learning_rate": 2.4754075794669088e-05,
      "loss": 0.05164482444524765,
      "step": 2404
    },
    {
      "epoch": 0.32514005188045525,
      "grad_norm": 0.6710517406463623,
      "learning_rate": 2.4748875055423942e-05,
      "loss": 0.056668512523174286,
      "step": 2405
    },
    {
      "epoch": 0.3252752452492205,
      "grad_norm": 0.8575512766838074,
      "learning_rate": 2.4743672286411027e-05,
      "loss": 0.057905301451683044,
      "step": 2406
    },
    {
      "epoch": 0.3254104386179858,
      "grad_norm": 0.5478147864341736,
      "learning_rate": 2.4738467488713582e-05,
      "loss": 0.07564461976289749,
      "step": 2407
    },
    {
      "epoch": 0.32554563198675107,
      "grad_norm": 0.3981148302555084,
      "learning_rate": 2.473326066341527e-05,
      "loss": 0.06725983321666718,
      "step": 2408
    },
    {
      "epoch": 0.3256808253555163,
      "grad_norm": 0.6456949710845947,
      "learning_rate": 2.4728051811600184e-05,
      "loss": 0.07121357321739197,
      "step": 2409
    },
    {
      "epoch": 0.32581601872428156,
      "grad_norm": 1.2709277868270874,
      "learning_rate": 2.4722840934352838e-05,
      "loss": 0.08546609431505203,
      "step": 2410
    },
    {
      "epoch": 0.32595121209304684,
      "grad_norm": 0.30143308639526367,
      "learning_rate": 2.471762803275816e-05,
      "loss": 0.043651144951581955,
      "step": 2411
    },
    {
      "epoch": 0.3260864054618121,
      "grad_norm": 1.11068594455719,
      "learning_rate": 2.4712413107901504e-05,
      "loss": 0.050117939710617065,
      "step": 2412
    },
    {
      "epoch": 0.3262215988305774,
      "grad_norm": 0.5859435200691223,
      "learning_rate": 2.470719616086865e-05,
      "loss": 0.04886026680469513,
      "step": 2413
    },
    {
      "epoch": 0.3263567921993426,
      "grad_norm": 0.6050263047218323,
      "learning_rate": 2.4701977192745785e-05,
      "loss": 0.04854444041848183,
      "step": 2414
    },
    {
      "epoch": 0.3264919855681079,
      "grad_norm": 0.542130708694458,
      "learning_rate": 2.4696756204619535e-05,
      "loss": 0.07222840189933777,
      "step": 2415
    },
    {
      "epoch": 0.32662717893687315,
      "grad_norm": 0.4350861608982086,
      "learning_rate": 2.469153319757693e-05,
      "loss": 0.04858303815126419,
      "step": 2416
    },
    {
      "epoch": 0.3267623723056384,
      "grad_norm": 0.4467565417289734,
      "learning_rate": 2.4686308172705433e-05,
      "loss": 0.05715404450893402,
      "step": 2417
    },
    {
      "epoch": 0.32689756567440365,
      "grad_norm": 0.5131983757019043,
      "learning_rate": 2.4681081131092926e-05,
      "loss": 0.07129460573196411,
      "step": 2418
    },
    {
      "epoch": 0.3270327590431689,
      "grad_norm": 0.9890090823173523,
      "learning_rate": 2.467585207382769e-05,
      "loss": 0.08803065121173859,
      "step": 2419
    },
    {
      "epoch": 0.3271679524119342,
      "grad_norm": 1.0912023782730103,
      "learning_rate": 2.4670621001998467e-05,
      "loss": 0.049446359276771545,
      "step": 2420
    },
    {
      "epoch": 0.32730314578069947,
      "grad_norm": 0.37686675786972046,
      "learning_rate": 2.466538791669437e-05,
      "loss": 0.05590030550956726,
      "step": 2421
    },
    {
      "epoch": 0.32743833914946474,
      "grad_norm": 0.7751613855361938,
      "learning_rate": 2.4660152819004973e-05,
      "loss": 0.04032851755619049,
      "step": 2422
    },
    {
      "epoch": 0.32757353251822996,
      "grad_norm": 0.6678184866905212,
      "learning_rate": 2.4654915710020246e-05,
      "loss": 0.06188032776117325,
      "step": 2423
    },
    {
      "epoch": 0.32770872588699523,
      "grad_norm": 1.0484058856964111,
      "learning_rate": 2.464967659083058e-05,
      "loss": 0.06437119841575623,
      "step": 2424
    },
    {
      "epoch": 0.3278439192557605,
      "grad_norm": 0.4111539125442505,
      "learning_rate": 2.464443546252679e-05,
      "loss": 0.05515044182538986,
      "step": 2425
    },
    {
      "epoch": 0.3279791126245258,
      "grad_norm": 0.36039870977401733,
      "learning_rate": 2.4639192326200104e-05,
      "loss": 0.06106668710708618,
      "step": 2426
    },
    {
      "epoch": 0.32811430599329106,
      "grad_norm": 0.332709401845932,
      "learning_rate": 2.463394718294218e-05,
      "loss": 0.044600386172533035,
      "step": 2427
    },
    {
      "epoch": 0.3282494993620563,
      "grad_norm": 0.2628311514854431,
      "learning_rate": 2.4628700033845072e-05,
      "loss": 0.04399694502353668,
      "step": 2428
    },
    {
      "epoch": 0.32838469273082155,
      "grad_norm": 0.7415270209312439,
      "learning_rate": 2.4623450880001268e-05,
      "loss": 0.057974159717559814,
      "step": 2429
    },
    {
      "epoch": 0.3285198860995868,
      "grad_norm": 0.8735581636428833,
      "learning_rate": 2.4618199722503676e-05,
      "loss": 0.07110685110092163,
      "step": 2430
    },
    {
      "epoch": 0.3286550794683521,
      "grad_norm": 0.38375768065452576,
      "learning_rate": 2.4612946562445613e-05,
      "loss": 0.04996195808053017,
      "step": 2431
    },
    {
      "epoch": 0.3287902728371173,
      "grad_norm": 0.39115580916404724,
      "learning_rate": 2.460769140092081e-05,
      "loss": 0.06056630611419678,
      "step": 2432
    },
    {
      "epoch": 0.3289254662058826,
      "grad_norm": 1.1033483743667603,
      "learning_rate": 2.460243423902342e-05,
      "loss": 0.0673246830701828,
      "step": 2433
    },
    {
      "epoch": 0.32906065957464786,
      "grad_norm": 0.6083263158798218,
      "learning_rate": 2.459717507784802e-05,
      "loss": 0.0530889630317688,
      "step": 2434
    },
    {
      "epoch": 0.32919585294341314,
      "grad_norm": 0.5067211985588074,
      "learning_rate": 2.459191391848959e-05,
      "loss": 0.04698304086923599,
      "step": 2435
    },
    {
      "epoch": 0.3293310463121784,
      "grad_norm": 0.6329554915428162,
      "learning_rate": 2.4586650762043538e-05,
      "loss": 0.0533183254301548,
      "step": 2436
    },
    {
      "epoch": 0.32946623968094363,
      "grad_norm": 0.9279079437255859,
      "learning_rate": 2.4581385609605665e-05,
      "loss": 0.05001988261938095,
      "step": 2437
    },
    {
      "epoch": 0.3296014330497089,
      "grad_norm": 0.842490017414093,
      "learning_rate": 2.4576118462272218e-05,
      "loss": 0.05696358531713486,
      "step": 2438
    },
    {
      "epoch": 0.3297366264184742,
      "grad_norm": 1.5504297018051147,
      "learning_rate": 2.4570849321139836e-05,
      "loss": 0.06988736987113953,
      "step": 2439
    },
    {
      "epoch": 0.32987181978723945,
      "grad_norm": 0.7859694957733154,
      "learning_rate": 2.4565578187305596e-05,
      "loss": 0.048929743468761444,
      "step": 2440
    },
    {
      "epoch": 0.3300070131560047,
      "grad_norm": 0.8285436630249023,
      "learning_rate": 2.456030506186696e-05,
      "loss": 0.05581335723400116,
      "step": 2441
    },
    {
      "epoch": 0.33014220652476994,
      "grad_norm": 0.9999210238456726,
      "learning_rate": 2.4555029945921832e-05,
      "loss": 0.07747752964496613,
      "step": 2442
    },
    {
      "epoch": 0.3302773998935352,
      "grad_norm": 0.7843573689460754,
      "learning_rate": 2.4549752840568516e-05,
      "loss": 0.07872691750526428,
      "step": 2443
    },
    {
      "epoch": 0.3304125932623005,
      "grad_norm": 0.3269481956958771,
      "learning_rate": 2.4544473746905733e-05,
      "loss": 0.04337403178215027,
      "step": 2444
    },
    {
      "epoch": 0.33054778663106577,
      "grad_norm": 0.678648054599762,
      "learning_rate": 2.4539192666032617e-05,
      "loss": 0.05248695984482765,
      "step": 2445
    },
    {
      "epoch": 0.330682979999831,
      "grad_norm": 0.27597638964653015,
      "learning_rate": 2.4533909599048718e-05,
      "loss": 0.04707809537649155,
      "step": 2446
    },
    {
      "epoch": 0.33081817336859626,
      "grad_norm": 0.7942515015602112,
      "learning_rate": 2.4528624547054003e-05,
      "loss": 0.05427852272987366,
      "step": 2447
    },
    {
      "epoch": 0.33095336673736153,
      "grad_norm": 0.8145894408226013,
      "learning_rate": 2.4523337511148843e-05,
      "loss": 0.08071470260620117,
      "step": 2448
    },
    {
      "epoch": 0.3310885601061268,
      "grad_norm": 0.579809844493866,
      "learning_rate": 2.4518048492434028e-05,
      "loss": 0.051956817507743835,
      "step": 2449
    },
    {
      "epoch": 0.3312237534748921,
      "grad_norm": 1.0467652082443237,
      "learning_rate": 2.4512757492010762e-05,
      "loss": 0.0688856691122055,
      "step": 2450
    },
    {
      "epoch": 0.3313589468436573,
      "grad_norm": 0.26254987716674805,
      "learning_rate": 2.4507464510980652e-05,
      "loss": 0.03904147073626518,
      "step": 2451
    },
    {
      "epoch": 0.33149414021242257,
      "grad_norm": 1.1274890899658203,
      "learning_rate": 2.450216955044574e-05,
      "loss": 0.06311538070440292,
      "step": 2452
    },
    {
      "epoch": 0.33162933358118785,
      "grad_norm": 0.32946228981018066,
      "learning_rate": 2.449687261150845e-05,
      "loss": 0.05260097607970238,
      "step": 2453
    },
    {
      "epoch": 0.3317645269499531,
      "grad_norm": 0.4412408471107483,
      "learning_rate": 2.449157369527164e-05,
      "loss": 0.05872887000441551,
      "step": 2454
    },
    {
      "epoch": 0.3318997203187184,
      "grad_norm": 0.3664141297340393,
      "learning_rate": 2.448627280283857e-05,
      "loss": 0.058825358748435974,
      "step": 2455
    },
    {
      "epoch": 0.3320349136874836,
      "grad_norm": 0.5627807378768921,
      "learning_rate": 2.4480969935312917e-05,
      "loss": 0.06658937782049179,
      "step": 2456
    },
    {
      "epoch": 0.3321701070562489,
      "grad_norm": 0.6522341966629028,
      "learning_rate": 2.4475665093798766e-05,
      "loss": 0.04885073006153107,
      "step": 2457
    },
    {
      "epoch": 0.33230530042501416,
      "grad_norm": 0.6704056262969971,
      "learning_rate": 2.447035827940061e-05,
      "loss": 0.051568403840065,
      "step": 2458
    },
    {
      "epoch": 0.33244049379377943,
      "grad_norm": 0.5867211818695068,
      "learning_rate": 2.4465049493223356e-05,
      "loss": 0.05047309398651123,
      "step": 2459
    },
    {
      "epoch": 0.33257568716254465,
      "grad_norm": 0.6995148062705994,
      "learning_rate": 2.4459738736372327e-05,
      "loss": 0.05371546745300293,
      "step": 2460
    },
    {
      "epoch": 0.3327108805313099,
      "grad_norm": 0.3484623432159424,
      "learning_rate": 2.4454426009953252e-05,
      "loss": 0.04776644706726074,
      "step": 2461
    },
    {
      "epoch": 0.3328460739000752,
      "grad_norm": 0.8619898557662964,
      "learning_rate": 2.4449111315072254e-05,
      "loss": 0.07687251269817352,
      "step": 2462
    },
    {
      "epoch": 0.3329812672688405,
      "grad_norm": 0.7768975496292114,
      "learning_rate": 2.44437946528359e-05,
      "loss": 0.07029340416193008,
      "step": 2463
    },
    {
      "epoch": 0.33311646063760575,
      "grad_norm": 0.30667319893836975,
      "learning_rate": 2.4438476024351138e-05,
      "loss": 0.0642610639333725,
      "step": 2464
    },
    {
      "epoch": 0.33325165400637097,
      "grad_norm": 0.37118950486183167,
      "learning_rate": 2.4433155430725333e-05,
      "loss": 0.04756537824869156,
      "step": 2465
    },
    {
      "epoch": 0.33338684737513624,
      "grad_norm": 1.2667423486709595,
      "learning_rate": 2.4427832873066262e-05,
      "loss": 0.06080685555934906,
      "step": 2466
    },
    {
      "epoch": 0.3335220407439015,
      "grad_norm": 0.5492689609527588,
      "learning_rate": 2.4422508352482113e-05,
      "loss": 0.07088005542755127,
      "step": 2467
    },
    {
      "epoch": 0.3336572341126668,
      "grad_norm": 0.3840407431125641,
      "learning_rate": 2.441718187008148e-05,
      "loss": 0.06379091739654541,
      "step": 2468
    },
    {
      "epoch": 0.333792427481432,
      "grad_norm": 0.5432056784629822,
      "learning_rate": 2.441185342697336e-05,
      "loss": 0.03955751657485962,
      "step": 2469
    },
    {
      "epoch": 0.3339276208501973,
      "grad_norm": 1.3248522281646729,
      "learning_rate": 2.440652302426717e-05,
      "loss": 0.06881004571914673,
      "step": 2470
    },
    {
      "epoch": 0.33406281421896256,
      "grad_norm": 0.5760420560836792,
      "learning_rate": 2.440119066307272e-05,
      "loss": 0.07075256109237671,
      "step": 2471
    },
    {
      "epoch": 0.33419800758772783,
      "grad_norm": 1.4234977960586548,
      "learning_rate": 2.4395856344500244e-05,
      "loss": 0.06021004915237427,
      "step": 2472
    },
    {
      "epoch": 0.3343332009564931,
      "grad_norm": 0.3846748173236847,
      "learning_rate": 2.4390520069660377e-05,
      "loss": 0.04977214336395264,
      "step": 2473
    },
    {
      "epoch": 0.3344683943252583,
      "grad_norm": 0.5696228742599487,
      "learning_rate": 2.4385181839664146e-05,
      "loss": 0.0500163808465004,
      "step": 2474
    },
    {
      "epoch": 0.3346035876940236,
      "grad_norm": 0.8239949941635132,
      "learning_rate": 2.437984165562301e-05,
      "loss": 0.05204898118972778,
      "step": 2475
    },
    {
      "epoch": 0.33473878106278887,
      "grad_norm": 0.8784533739089966,
      "learning_rate": 2.4374499518648827e-05,
      "loss": 0.07036906480789185,
      "step": 2476
    },
    {
      "epoch": 0.33487397443155414,
      "grad_norm": 0.3877400755882263,
      "learning_rate": 2.436915542985385e-05,
      "loss": 0.047933436930179596,
      "step": 2477
    },
    {
      "epoch": 0.3350091678003194,
      "grad_norm": 1.5199875831604004,
      "learning_rate": 2.436380939035075e-05,
      "loss": 0.06770068407058716,
      "step": 2478
    },
    {
      "epoch": 0.33514436116908464,
      "grad_norm": 0.9085196852684021,
      "learning_rate": 2.43584614012526e-05,
      "loss": 0.06982289254665375,
      "step": 2479
    },
    {
      "epoch": 0.3352795545378499,
      "grad_norm": 0.47763633728027344,
      "learning_rate": 2.4353111463672882e-05,
      "loss": 0.059069663286209106,
      "step": 2480
    },
    {
      "epoch": 0.3354147479066152,
      "grad_norm": 0.803889274597168,
      "learning_rate": 2.4347759578725482e-05,
      "loss": 0.062149446457624435,
      "step": 2481
    },
    {
      "epoch": 0.33554994127538046,
      "grad_norm": 0.48583748936653137,
      "learning_rate": 2.4342405747524685e-05,
      "loss": 0.05786242336034775,
      "step": 2482
    },
    {
      "epoch": 0.3356851346441457,
      "grad_norm": 0.26858198642730713,
      "learning_rate": 2.4337049971185194e-05,
      "loss": 0.05728241801261902,
      "step": 2483
    },
    {
      "epoch": 0.33582032801291095,
      "grad_norm": 0.26471465826034546,
      "learning_rate": 2.433169225082211e-05,
      "loss": 0.057354867458343506,
      "step": 2484
    },
    {
      "epoch": 0.3359555213816762,
      "grad_norm": 0.5075029730796814,
      "learning_rate": 2.432633258755093e-05,
      "loss": 0.06345418840646744,
      "step": 2485
    },
    {
      "epoch": 0.3360907147504415,
      "grad_norm": 0.6484138369560242,
      "learning_rate": 2.432097098248758e-05,
      "loss": 0.06858913600444794,
      "step": 2486
    },
    {
      "epoch": 0.3362259081192068,
      "grad_norm": 1.5889604091644287,
      "learning_rate": 2.4315607436748362e-05,
      "loss": 0.05519643425941467,
      "step": 2487
    },
    {
      "epoch": 0.336361101487972,
      "grad_norm": 0.627388060092926,
      "learning_rate": 2.4310241951449997e-05,
      "loss": 0.05793236941099167,
      "step": 2488
    },
    {
      "epoch": 0.33649629485673727,
      "grad_norm": 0.4585854411125183,
      "learning_rate": 2.4304874527709614e-05,
      "loss": 0.039617158472537994,
      "step": 2489
    },
    {
      "epoch": 0.33663148822550254,
      "grad_norm": 0.5693581700325012,
      "learning_rate": 2.429950516664473e-05,
      "loss": 0.05294934660196304,
      "step": 2490
    },
    {
      "epoch": 0.3367666815942678,
      "grad_norm": 1.1733219623565674,
      "learning_rate": 2.4294133869373284e-05,
      "loss": 0.07497140765190125,
      "step": 2491
    },
    {
      "epoch": 0.3369018749630331,
      "grad_norm": 0.5426054000854492,
      "learning_rate": 2.42887606370136e-05,
      "loss": 0.05442477762699127,
      "step": 2492
    },
    {
      "epoch": 0.3370370683317983,
      "grad_norm": 0.41151532530784607,
      "learning_rate": 2.428338547068442e-05,
      "loss": 0.047723107039928436,
      "step": 2493
    },
    {
      "epoch": 0.3371722617005636,
      "grad_norm": 0.699028491973877,
      "learning_rate": 2.427800837150488e-05,
      "loss": 0.054842229932546616,
      "step": 2494
    },
    {
      "epoch": 0.33730745506932885,
      "grad_norm": 0.600603461265564,
      "learning_rate": 2.4272629340594518e-05,
      "loss": 0.06443243473768234,
      "step": 2495
    },
    {
      "epoch": 0.33744264843809413,
      "grad_norm": 0.587218701839447,
      "learning_rate": 2.426724837907328e-05,
      "loss": 0.05162160098552704,
      "step": 2496
    },
    {
      "epoch": 0.33757784180685935,
      "grad_norm": 0.3580301105976105,
      "learning_rate": 2.4261865488061512e-05,
      "loss": 0.041045911610126495,
      "step": 2497
    },
    {
      "epoch": 0.3377130351756246,
      "grad_norm": 0.7507015466690063,
      "learning_rate": 2.4256480668679958e-05,
      "loss": 0.04999418556690216,
      "step": 2498
    },
    {
      "epoch": 0.3378482285443899,
      "grad_norm": 1.4667502641677856,
      "learning_rate": 2.4251093922049766e-05,
      "loss": 0.0518585704267025,
      "step": 2499
    },
    {
      "epoch": 0.33798342191315517,
      "grad_norm": 2.0040831565856934,
      "learning_rate": 2.4245705249292494e-05,
      "loss": 0.05939832329750061,
      "step": 2500
    },
    {
      "epoch": 0.33811861528192044,
      "grad_norm": 0.571308434009552,
      "learning_rate": 2.4240314651530073e-05,
      "loss": 0.049097444862127304,
      "step": 2501
    },
    {
      "epoch": 0.33825380865068566,
      "grad_norm": 0.3827081322669983,
      "learning_rate": 2.4234922129884873e-05,
      "loss": 0.04473471641540527,
      "step": 2502
    },
    {
      "epoch": 0.33838900201945094,
      "grad_norm": 0.7014095187187195,
      "learning_rate": 2.4229527685479644e-05,
      "loss": 0.08057114481925964,
      "step": 2503
    },
    {
      "epoch": 0.3385241953882162,
      "grad_norm": 1.4268348217010498,
      "learning_rate": 2.4224131319437523e-05,
      "loss": 0.05230224132537842,
      "step": 2504
    },
    {
      "epoch": 0.3386593887569815,
      "grad_norm": 1.3716235160827637,
      "learning_rate": 2.421873303288208e-05,
      "loss": 0.0688956081867218,
      "step": 2505
    },
    {
      "epoch": 0.33879458212574676,
      "grad_norm": 0.8329153060913086,
      "learning_rate": 2.4213332826937255e-05,
      "loss": 0.04252082109451294,
      "step": 2506
    },
    {
      "epoch": 0.338929775494512,
      "grad_norm": 0.7675012946128845,
      "learning_rate": 2.4207930702727404e-05,
      "loss": 0.06650793552398682,
      "step": 2507
    },
    {
      "epoch": 0.33906496886327725,
      "grad_norm": 0.7769269943237305,
      "learning_rate": 2.420252666137728e-05,
      "loss": 0.0697876513004303,
      "step": 2508
    },
    {
      "epoch": 0.3392001622320425,
      "grad_norm": 0.3359142541885376,
      "learning_rate": 2.419712070401203e-05,
      "loss": 0.047617167234420776,
      "step": 2509
    },
    {
      "epoch": 0.3393353556008078,
      "grad_norm": 0.5980057716369629,
      "learning_rate": 2.4191712831757203e-05,
      "loss": 0.05942881107330322,
      "step": 2510
    },
    {
      "epoch": 0.339470548969573,
      "grad_norm": 0.4190271496772766,
      "learning_rate": 2.418630304573875e-05,
      "loss": 0.0539388507604599,
      "step": 2511
    },
    {
      "epoch": 0.3396057423383383,
      "grad_norm": 0.43068206310272217,
      "learning_rate": 2.418089134708302e-05,
      "loss": 0.061523064970970154,
      "step": 2512
    },
    {
      "epoch": 0.33974093570710356,
      "grad_norm": 0.5115730166435242,
      "learning_rate": 2.4175477736916743e-05,
      "loss": 0.06761537492275238,
      "step": 2513
    },
    {
      "epoch": 0.33987612907586884,
      "grad_norm": 1.7118775844573975,
      "learning_rate": 2.4170062216367082e-05,
      "loss": 0.053287506103515625,
      "step": 2514
    },
    {
      "epoch": 0.3400113224446341,
      "grad_norm": 0.6027407050132751,
      "learning_rate": 2.416464478656156e-05,
      "loss": 0.06524059176445007,
      "step": 2515
    },
    {
      "epoch": 0.34014651581339933,
      "grad_norm": 1.1736122369766235,
      "learning_rate": 2.4159225448628123e-05,
      "loss": 0.06616820394992828,
      "step": 2516
    },
    {
      "epoch": 0.3402817091821646,
      "grad_norm": 1.180332064628601,
      "learning_rate": 2.4153804203695103e-05,
      "loss": 0.05459309369325638,
      "step": 2517
    },
    {
      "epoch": 0.3404169025509299,
      "grad_norm": 0.42003291845321655,
      "learning_rate": 2.4148381052891236e-05,
      "loss": 0.04738977551460266,
      "step": 2518
    },
    {
      "epoch": 0.34055209591969515,
      "grad_norm": 0.3957957923412323,
      "learning_rate": 2.4142955997345648e-05,
      "loss": 0.053268320858478546,
      "step": 2519
    },
    {
      "epoch": 0.3406872892884604,
      "grad_norm": 0.49949464201927185,
      "learning_rate": 2.4137529038187864e-05,
      "loss": 0.06918475776910782,
      "step": 2520
    },
    {
      "epoch": 0.34082248265722564,
      "grad_norm": 1.229874849319458,
      "learning_rate": 2.413210017654781e-05,
      "loss": 0.07177431881427765,
      "step": 2521
    },
    {
      "epoch": 0.3409576760259909,
      "grad_norm": 0.9811297655105591,
      "learning_rate": 2.4126669413555802e-05,
      "loss": 0.05535629391670227,
      "step": 2522
    },
    {
      "epoch": 0.3410928693947562,
      "grad_norm": 0.46806463599205017,
      "learning_rate": 2.4121236750342548e-05,
      "loss": 0.045290857553482056,
      "step": 2523
    },
    {
      "epoch": 0.34122806276352147,
      "grad_norm": 0.5662136673927307,
      "learning_rate": 2.4115802188039165e-05,
      "loss": 0.07392966747283936,
      "step": 2524
    },
    {
      "epoch": 0.3413632561322867,
      "grad_norm": 1.0050934553146362,
      "learning_rate": 2.4110365727777156e-05,
      "loss": 0.07055419683456421,
      "step": 2525
    },
    {
      "epoch": 0.34149844950105196,
      "grad_norm": 0.765142023563385,
      "learning_rate": 2.410492737068842e-05,
      "loss": 0.0515645295381546,
      "step": 2526
    },
    {
      "epoch": 0.34163364286981723,
      "grad_norm": 0.5425080060958862,
      "learning_rate": 2.409948711790525e-05,
      "loss": 0.057616740465164185,
      "step": 2527
    },
    {
      "epoch": 0.3417688362385825,
      "grad_norm": 0.6883987784385681,
      "learning_rate": 2.4094044970560336e-05,
      "loss": 0.066932812333107,
      "step": 2528
    },
    {
      "epoch": 0.3419040296073478,
      "grad_norm": 0.5013260245323181,
      "learning_rate": 2.4088600929786767e-05,
      "loss": 0.06805744767189026,
      "step": 2529
    },
    {
      "epoch": 0.342039222976113,
      "grad_norm": 0.5903222560882568,
      "learning_rate": 2.408315499671802e-05,
      "loss": 0.07115764915943146,
      "step": 2530
    },
    {
      "epoch": 0.3421744163448783,
      "grad_norm": 0.34074294567108154,
      "learning_rate": 2.407770717248796e-05,
      "loss": 0.06422656774520874,
      "step": 2531
    },
    {
      "epoch": 0.34230960971364355,
      "grad_norm": 0.6268665194511414,
      "learning_rate": 2.407225745823086e-05,
      "loss": 0.05897696316242218,
      "step": 2532
    },
    {
      "epoch": 0.3424448030824088,
      "grad_norm": 0.8784014582633972,
      "learning_rate": 2.4066805855081378e-05,
      "loss": 0.059277549386024475,
      "step": 2533
    },
    {
      "epoch": 0.3425799964511741,
      "grad_norm": 0.8250194787979126,
      "learning_rate": 2.406135236417457e-05,
      "loss": 0.04818663001060486,
      "step": 2534
    },
    {
      "epoch": 0.3427151898199393,
      "grad_norm": 0.39451804757118225,
      "learning_rate": 2.4055896986645875e-05,
      "loss": 0.06961485743522644,
      "step": 2535
    },
    {
      "epoch": 0.3428503831887046,
      "grad_norm": 0.3706168830394745,
      "learning_rate": 2.4050439723631136e-05,
      "loss": 0.039464518427848816,
      "step": 2536
    },
    {
      "epoch": 0.34298557655746986,
      "grad_norm": 0.27246546745300293,
      "learning_rate": 2.404498057626659e-05,
      "loss": 0.053841859102249146,
      "step": 2537
    },
    {
      "epoch": 0.34312076992623514,
      "grad_norm": 0.5579825043678284,
      "learning_rate": 2.4039519545688848e-05,
      "loss": 0.06240345537662506,
      "step": 2538
    },
    {
      "epoch": 0.34325596329500035,
      "grad_norm": 0.2667763829231262,
      "learning_rate": 2.4034056633034932e-05,
      "loss": 0.055660054087638855,
      "step": 2539
    },
    {
      "epoch": 0.34339115666376563,
      "grad_norm": 0.4128856360912323,
      "learning_rate": 2.402859183944225e-05,
      "loss": 0.045962586998939514,
      "step": 2540
    },
    {
      "epoch": 0.3435263500325309,
      "grad_norm": 0.7272021770477295,
      "learning_rate": 2.4023125166048597e-05,
      "loss": 0.06458261609077454,
      "step": 2541
    },
    {
      "epoch": 0.3436615434012962,
      "grad_norm": 0.6898627877235413,
      "learning_rate": 2.401765661399218e-05,
      "loss": 0.0651286393404007,
      "step": 2542
    },
    {
      "epoch": 0.34379673677006145,
      "grad_norm": 0.4911874830722809,
      "learning_rate": 2.4012186184411556e-05,
      "loss": 0.0685330182313919,
      "step": 2543
    },
    {
      "epoch": 0.34393193013882667,
      "grad_norm": 0.7272791862487793,
      "learning_rate": 2.400671387844571e-05,
      "loss": 0.050691794604063034,
      "step": 2544
    },
    {
      "epoch": 0.34406712350759194,
      "grad_norm": 1.5425516366958618,
      "learning_rate": 2.4001239697234008e-05,
      "loss": 0.06000924110412598,
      "step": 2545
    },
    {
      "epoch": 0.3442023168763572,
      "grad_norm": 0.47782620787620544,
      "learning_rate": 2.3995763641916205e-05,
      "loss": 0.05121324956417084,
      "step": 2546
    },
    {
      "epoch": 0.3443375102451225,
      "grad_norm": 1.0811127424240112,
      "learning_rate": 2.3990285713632436e-05,
      "loss": 0.054869458079338074,
      "step": 2547
    },
    {
      "epoch": 0.34447270361388777,
      "grad_norm": 0.7305948734283447,
      "learning_rate": 2.398480591352324e-05,
      "loss": 0.05475684627890587,
      "step": 2548
    },
    {
      "epoch": 0.344607896982653,
      "grad_norm": 0.2501955032348633,
      "learning_rate": 2.3979324242729537e-05,
      "loss": 0.05399274826049805,
      "step": 2549
    },
    {
      "epoch": 0.34474309035141826,
      "grad_norm": 0.5100407600402832,
      "learning_rate": 2.3973840702392646e-05,
      "loss": 0.04482141137123108,
      "step": 2550
    },
    {
      "epoch": 0.34487828372018353,
      "grad_norm": 0.7808458209037781,
      "learning_rate": 2.3968355293654267e-05,
      "loss": 0.060536690056324005,
      "step": 2551
    },
    {
      "epoch": 0.3450134770889488,
      "grad_norm": 0.45987311005592346,
      "learning_rate": 2.396286801765649e-05,
      "loss": 0.05622217059135437,
      "step": 2552
    },
    {
      "epoch": 0.345148670457714,
      "grad_norm": 0.44538676738739014,
      "learning_rate": 2.3957378875541795e-05,
      "loss": 0.048855721950531006,
      "step": 2553
    },
    {
      "epoch": 0.3452838638264793,
      "grad_norm": 0.47922489047050476,
      "learning_rate": 2.395188786845305e-05,
      "loss": 0.05611765757203102,
      "step": 2554
    },
    {
      "epoch": 0.34541905719524457,
      "grad_norm": 0.7451590895652771,
      "learning_rate": 2.3946394997533516e-05,
      "loss": 0.0712137222290039,
      "step": 2555
    },
    {
      "epoch": 0.34555425056400985,
      "grad_norm": 1.1291475296020508,
      "learning_rate": 2.3940900263926833e-05,
      "loss": 0.04754869267344475,
      "step": 2556
    },
    {
      "epoch": 0.3456894439327751,
      "grad_norm": 0.8264445662498474,
      "learning_rate": 2.393540366877704e-05,
      "loss": 0.05984792858362198,
      "step": 2557
    },
    {
      "epoch": 0.34582463730154034,
      "grad_norm": 0.7591851353645325,
      "learning_rate": 2.392990521322855e-05,
      "loss": 0.058432627469301224,
      "step": 2558
    },
    {
      "epoch": 0.3459598306703056,
      "grad_norm": 0.4125320613384247,
      "learning_rate": 2.392440489842618e-05,
      "loss": 0.03875409811735153,
      "step": 2559
    },
    {
      "epoch": 0.3460950240390709,
      "grad_norm": 0.3128974139690399,
      "learning_rate": 2.3918902725515118e-05,
      "loss": 0.038277775049209595,
      "step": 2560
    },
    {
      "epoch": 0.34623021740783616,
      "grad_norm": 0.5362597703933716,
      "learning_rate": 2.391339869564094e-05,
      "loss": 0.07293155789375305,
      "step": 2561
    },
    {
      "epoch": 0.34636541077660143,
      "grad_norm": 0.6536089777946472,
      "learning_rate": 2.3907892809949628e-05,
      "loss": 0.06180323660373688,
      "step": 2562
    },
    {
      "epoch": 0.34650060414536665,
      "grad_norm": 0.8963726162910461,
      "learning_rate": 2.390238506958753e-05,
      "loss": 0.0651576817035675,
      "step": 2563
    },
    {
      "epoch": 0.3466357975141319,
      "grad_norm": 0.300552099943161,
      "learning_rate": 2.3896875475701387e-05,
      "loss": 0.04789632558822632,
      "step": 2564
    },
    {
      "epoch": 0.3467709908828972,
      "grad_norm": 0.8877284526824951,
      "learning_rate": 2.3891364029438323e-05,
      "loss": 0.06549087911844254,
      "step": 2565
    },
    {
      "epoch": 0.3469061842516625,
      "grad_norm": 0.5803410410881042,
      "learning_rate": 2.3885850731945857e-05,
      "loss": 0.05586116015911102,
      "step": 2566
    },
    {
      "epoch": 0.3470413776204277,
      "grad_norm": 1.0514369010925293,
      "learning_rate": 2.3880335584371884e-05,
      "loss": 0.06165571138262749,
      "step": 2567
    },
    {
      "epoch": 0.34717657098919297,
      "grad_norm": 0.5227709412574768,
      "learning_rate": 2.387481858786468e-05,
      "loss": 0.04741154611110687,
      "step": 2568
    },
    {
      "epoch": 0.34731176435795824,
      "grad_norm": 1.0869237184524536,
      "learning_rate": 2.386929974357293e-05,
      "loss": 0.0511382594704628,
      "step": 2569
    },
    {
      "epoch": 0.3474469577267235,
      "grad_norm": 0.79176265001297,
      "learning_rate": 2.386377905264567e-05,
      "loss": 0.06429392099380493,
      "step": 2570
    },
    {
      "epoch": 0.3475821510954888,
      "grad_norm": 0.6106847524642944,
      "learning_rate": 2.3858256516232346e-05,
      "loss": 0.06466436386108398,
      "step": 2571
    },
    {
      "epoch": 0.347717344464254,
      "grad_norm": 0.4373892545700073,
      "learning_rate": 2.3852732135482775e-05,
      "loss": 0.04785441234707832,
      "step": 2572
    },
    {
      "epoch": 0.3478525378330193,
      "grad_norm": 0.7229081988334656,
      "learning_rate": 2.3847205911547166e-05,
      "loss": 0.05422341823577881,
      "step": 2573
    },
    {
      "epoch": 0.34798773120178456,
      "grad_norm": 1.1516376733779907,
      "learning_rate": 2.3841677845576108e-05,
      "loss": 0.062233179807662964,
      "step": 2574
    },
    {
      "epoch": 0.34812292457054983,
      "grad_norm": 0.7101799249649048,
      "learning_rate": 2.383614793872057e-05,
      "loss": 0.061386242508888245,
      "step": 2575
    },
    {
      "epoch": 0.3482581179393151,
      "grad_norm": 0.2678720951080322,
      "learning_rate": 2.3830616192131913e-05,
      "loss": 0.03412139043211937,
      "step": 2576
    },
    {
      "epoch": 0.3483933113080803,
      "grad_norm": 0.4181729555130005,
      "learning_rate": 2.3825082606961876e-05,
      "loss": 0.05492118000984192,
      "step": 2577
    },
    {
      "epoch": 0.3485285046768456,
      "grad_norm": 0.582635760307312,
      "learning_rate": 2.3819547184362575e-05,
      "loss": 0.060674965381622314,
      "step": 2578
    },
    {
      "epoch": 0.34866369804561087,
      "grad_norm": 0.420721173286438,
      "learning_rate": 2.3814009925486522e-05,
      "loss": 0.05094034969806671,
      "step": 2579
    },
    {
      "epoch": 0.34879889141437614,
      "grad_norm": 0.9418121576309204,
      "learning_rate": 2.38084708314866e-05,
      "loss": 0.04535314440727234,
      "step": 2580
    },
    {
      "epoch": 0.34893408478314136,
      "grad_norm": 0.3692968487739563,
      "learning_rate": 2.380292990351608e-05,
      "loss": 0.04885189235210419,
      "step": 2581
    },
    {
      "epoch": 0.34906927815190664,
      "grad_norm": 0.7663249373435974,
      "learning_rate": 2.3797387142728607e-05,
      "loss": 0.046023085713386536,
      "step": 2582
    },
    {
      "epoch": 0.3492044715206719,
      "grad_norm": 0.5300248265266418,
      "learning_rate": 2.379184255027822e-05,
      "loss": 0.054738618433475494,
      "step": 2583
    },
    {
      "epoch": 0.3493396648894372,
      "grad_norm": 0.5208472013473511,
      "learning_rate": 2.378629612731933e-05,
      "loss": 0.06298881024122238,
      "step": 2584
    },
    {
      "epoch": 0.34947485825820246,
      "grad_norm": 0.559243381023407,
      "learning_rate": 2.3780747875006735e-05,
      "loss": 0.04366272687911987,
      "step": 2585
    },
    {
      "epoch": 0.3496100516269677,
      "grad_norm": 0.29652270674705505,
      "learning_rate": 2.37751977944956e-05,
      "loss": 0.05863502621650696,
      "step": 2586
    },
    {
      "epoch": 0.34974524499573295,
      "grad_norm": 0.5760659575462341,
      "learning_rate": 2.3769645886941497e-05,
      "loss": 0.0788021981716156,
      "step": 2587
    },
    {
      "epoch": 0.3498804383644982,
      "grad_norm": 0.27831214666366577,
      "learning_rate": 2.376409215350035e-05,
      "loss": 0.043708428740501404,
      "step": 2588
    },
    {
      "epoch": 0.3500156317332635,
      "grad_norm": 0.7583713531494141,
      "learning_rate": 2.3758536595328486e-05,
      "loss": 0.06890417635440826,
      "step": 2589
    },
    {
      "epoch": 0.3501508251020288,
      "grad_norm": 0.3065214157104492,
      "learning_rate": 2.375297921358259e-05,
      "loss": 0.05722136050462723,
      "step": 2590
    },
    {
      "epoch": 0.350286018470794,
      "grad_norm": 0.6269283890724182,
      "learning_rate": 2.3747420009419745e-05,
      "loss": 0.05212374031543732,
      "step": 2591
    },
    {
      "epoch": 0.35042121183955927,
      "grad_norm": 0.3478796184062958,
      "learning_rate": 2.3741858983997415e-05,
      "loss": 0.0655936747789383,
      "step": 2592
    },
    {
      "epoch": 0.35055640520832454,
      "grad_norm": 0.2571185231208801,
      "learning_rate": 2.373629613847342e-05,
      "loss": 0.046925801783800125,
      "step": 2593
    },
    {
      "epoch": 0.3506915985770898,
      "grad_norm": 0.4292133152484894,
      "learning_rate": 2.3730731474005988e-05,
      "loss": 0.03944733738899231,
      "step": 2594
    },
    {
      "epoch": 0.35082679194585503,
      "grad_norm": 0.4198874831199646,
      "learning_rate": 2.37251649917537e-05,
      "loss": 0.04771158844232559,
      "step": 2595
    },
    {
      "epoch": 0.3509619853146203,
      "grad_norm": 0.4447716176509857,
      "learning_rate": 2.3719596692875534e-05,
      "loss": 0.057852745056152344,
      "step": 2596
    },
    {
      "epoch": 0.3510971786833856,
      "grad_norm": 0.7505528330802917,
      "learning_rate": 2.3714026578530836e-05,
      "loss": 0.06255149841308594,
      "step": 2597
    },
    {
      "epoch": 0.35123237205215085,
      "grad_norm": 0.31214281916618347,
      "learning_rate": 2.370845464987934e-05,
      "loss": 0.04574630409479141,
      "step": 2598
    },
    {
      "epoch": 0.35136756542091613,
      "grad_norm": 0.3981100618839264,
      "learning_rate": 2.370288090808114e-05,
      "loss": 0.0443689338862896,
      "step": 2599
    },
    {
      "epoch": 0.35150275878968135,
      "grad_norm": 0.48687514662742615,
      "learning_rate": 2.369730535429673e-05,
      "loss": 0.05019274353981018,
      "step": 2600
    },
    {
      "epoch": 0.3516379521584466,
      "grad_norm": 0.38240477442741394,
      "learning_rate": 2.369172798968697e-05,
      "loss": 0.06334640830755234,
      "step": 2601
    },
    {
      "epoch": 0.3517731455272119,
      "grad_norm": 0.3267514109611511,
      "learning_rate": 2.3686148815413083e-05,
      "loss": 0.050984565168619156,
      "step": 2602
    },
    {
      "epoch": 0.35190833889597717,
      "grad_norm": 0.557716429233551,
      "learning_rate": 2.3680567832636695e-05,
      "loss": 0.05017661675810814,
      "step": 2603
    },
    {
      "epoch": 0.3520435322647424,
      "grad_norm": 0.29590439796447754,
      "learning_rate": 2.3674985042519795e-05,
      "loss": 0.06020501255989075,
      "step": 2604
    },
    {
      "epoch": 0.35217872563350766,
      "grad_norm": 0.2776332199573517,
      "learning_rate": 2.366940044622475e-05,
      "loss": 0.04737381637096405,
      "step": 2605
    },
    {
      "epoch": 0.35231391900227294,
      "grad_norm": 1.3460701704025269,
      "learning_rate": 2.3663814044914302e-05,
      "loss": 0.050212420523166656,
      "step": 2606
    },
    {
      "epoch": 0.3524491123710382,
      "grad_norm": 0.32826700806617737,
      "learning_rate": 2.3658225839751566e-05,
      "loss": 0.05376420542597771,
      "step": 2607
    },
    {
      "epoch": 0.3525843057398035,
      "grad_norm": 0.604899525642395,
      "learning_rate": 2.3652635831900043e-05,
      "loss": 0.07410729676485062,
      "step": 2608
    },
    {
      "epoch": 0.3527194991085687,
      "grad_norm": 1.2091988325119019,
      "learning_rate": 2.3647044022523595e-05,
      "loss": 0.0536273717880249,
      "step": 2609
    },
    {
      "epoch": 0.352854692477334,
      "grad_norm": 0.8003138303756714,
      "learning_rate": 2.364145041278647e-05,
      "loss": 0.0515962690114975,
      "step": 2610
    },
    {
      "epoch": 0.35298988584609925,
      "grad_norm": 0.2733721137046814,
      "learning_rate": 2.3635855003853287e-05,
      "loss": 0.03086373209953308,
      "step": 2611
    },
    {
      "epoch": 0.3531250792148645,
      "grad_norm": 1.9662023782730103,
      "learning_rate": 2.363025779688904e-05,
      "loss": 0.07728084921836853,
      "step": 2612
    },
    {
      "epoch": 0.3532602725836298,
      "grad_norm": 0.4124818444252014,
      "learning_rate": 2.3624658793059103e-05,
      "loss": 0.07044723629951477,
      "step": 2613
    },
    {
      "epoch": 0.353395465952395,
      "grad_norm": 1.3796569108963013,
      "learning_rate": 2.3619057993529204e-05,
      "loss": 0.07363824546337128,
      "step": 2614
    },
    {
      "epoch": 0.3535306593211603,
      "grad_norm": 1.6578117609024048,
      "learning_rate": 2.3613455399465475e-05,
      "loss": 0.048297226428985596,
      "step": 2615
    },
    {
      "epoch": 0.35366585268992556,
      "grad_norm": 0.39314690232276917,
      "learning_rate": 2.3607851012034394e-05,
      "loss": 0.057796917855739594,
      "step": 2616
    },
    {
      "epoch": 0.35380104605869084,
      "grad_norm": 0.8022534251213074,
      "learning_rate": 2.3602244832402838e-05,
      "loss": 0.043085239827632904,
      "step": 2617
    },
    {
      "epoch": 0.35393623942745606,
      "grad_norm": 2.165907382965088,
      "learning_rate": 2.3596636861738024e-05,
      "loss": 0.05915340781211853,
      "step": 2618
    },
    {
      "epoch": 0.35407143279622133,
      "grad_norm": 0.5089679956436157,
      "learning_rate": 2.3591027101207578e-05,
      "loss": 0.05461547523736954,
      "step": 2619
    },
    {
      "epoch": 0.3542066261649866,
      "grad_norm": 0.3693690001964569,
      "learning_rate": 2.3585415551979476e-05,
      "loss": 0.041728127747774124,
      "step": 2620
    },
    {
      "epoch": 0.3543418195337519,
      "grad_norm": 0.3994734287261963,
      "learning_rate": 2.3579802215222076e-05,
      "loss": 0.06538271903991699,
      "step": 2621
    },
    {
      "epoch": 0.35447701290251715,
      "grad_norm": 0.8325362801551819,
      "learning_rate": 2.35741870921041e-05,
      "loss": 0.08045431226491928,
      "step": 2622
    },
    {
      "epoch": 0.35461220627128237,
      "grad_norm": 0.28094246983528137,
      "learning_rate": 2.3568570183794645e-05,
      "loss": 0.05441633239388466,
      "step": 2623
    },
    {
      "epoch": 0.35474739964004764,
      "grad_norm": 1.7707818746566772,
      "learning_rate": 2.356295149146319e-05,
      "loss": 0.048590898513793945,
      "step": 2624
    },
    {
      "epoch": 0.3548825930088129,
      "grad_norm": 1.2356181144714355,
      "learning_rate": 2.3557331016279567e-05,
      "loss": 0.05864347517490387,
      "step": 2625
    },
    {
      "epoch": 0.3550177863775782,
      "grad_norm": 0.47317442297935486,
      "learning_rate": 2.3551708759413998e-05,
      "loss": 0.05378357693552971,
      "step": 2626
    },
    {
      "epoch": 0.35515297974634347,
      "grad_norm": 0.9529665112495422,
      "learning_rate": 2.354608472203706e-05,
      "loss": 0.0599437952041626,
      "step": 2627
    },
    {
      "epoch": 0.3552881731151087,
      "grad_norm": 0.6597905158996582,
      "learning_rate": 2.3540458905319705e-05,
      "loss": 0.05852166563272476,
      "step": 2628
    },
    {
      "epoch": 0.35542336648387396,
      "grad_norm": 0.6731898784637451,
      "learning_rate": 2.3534831310433264e-05,
      "loss": 0.061750784516334534,
      "step": 2629
    },
    {
      "epoch": 0.35555855985263923,
      "grad_norm": 0.6791136264801025,
      "learning_rate": 2.3529201938549434e-05,
      "loss": 0.06681104004383087,
      "step": 2630
    },
    {
      "epoch": 0.3556937532214045,
      "grad_norm": 0.5720198750495911,
      "learning_rate": 2.3523570790840274e-05,
      "loss": 0.06063258647918701,
      "step": 2631
    },
    {
      "epoch": 0.3558289465901697,
      "grad_norm": 0.73337322473526,
      "learning_rate": 2.3517937868478228e-05,
      "loss": 0.04149096831679344,
      "step": 2632
    },
    {
      "epoch": 0.355964139958935,
      "grad_norm": 0.27330920100212097,
      "learning_rate": 2.3512303172636092e-05,
      "loss": 0.04757782071828842,
      "step": 2633
    },
    {
      "epoch": 0.3560993333277003,
      "grad_norm": 0.28042715787887573,
      "learning_rate": 2.3506666704487033e-05,
      "loss": 0.060640573501586914,
      "step": 2634
    },
    {
      "epoch": 0.35623452669646555,
      "grad_norm": 0.6427616477012634,
      "learning_rate": 2.3501028465204614e-05,
      "loss": 0.06296513974666595,
      "step": 2635
    },
    {
      "epoch": 0.3563697200652308,
      "grad_norm": 0.47043800354003906,
      "learning_rate": 2.3495388455962734e-05,
      "loss": 0.06619837880134583,
      "step": 2636
    },
    {
      "epoch": 0.35650491343399604,
      "grad_norm": 1.122891902923584,
      "learning_rate": 2.3489746677935673e-05,
      "loss": 0.07286417484283447,
      "step": 2637
    },
    {
      "epoch": 0.3566401068027613,
      "grad_norm": 0.5454763174057007,
      "learning_rate": 2.3484103132298082e-05,
      "loss": 0.06258266419172287,
      "step": 2638
    },
    {
      "epoch": 0.3567753001715266,
      "grad_norm": 1.1158266067504883,
      "learning_rate": 2.347845782022497e-05,
      "loss": 0.05772816389799118,
      "step": 2639
    },
    {
      "epoch": 0.35691049354029186,
      "grad_norm": 0.47275757789611816,
      "learning_rate": 2.3472810742891734e-05,
      "loss": 0.05300324410200119,
      "step": 2640
    },
    {
      "epoch": 0.35704568690905714,
      "grad_norm": 2.019820213317871,
      "learning_rate": 2.3467161901474118e-05,
      "loss": 0.06003296375274658,
      "step": 2641
    },
    {
      "epoch": 0.35718088027782235,
      "grad_norm": 1.201438307762146,
      "learning_rate": 2.346151129714824e-05,
      "loss": 0.0818963423371315,
      "step": 2642
    },
    {
      "epoch": 0.35731607364658763,
      "grad_norm": 0.5656431317329407,
      "learning_rate": 2.3455858931090588e-05,
      "loss": 0.05859818309545517,
      "step": 2643
    },
    {
      "epoch": 0.3574512670153529,
      "grad_norm": 0.9253957867622375,
      "learning_rate": 2.3450204804478014e-05,
      "loss": 0.0758921205997467,
      "step": 2644
    },
    {
      "epoch": 0.3575864603841182,
      "grad_norm": 0.4945479929447174,
      "learning_rate": 2.344454891848774e-05,
      "loss": 0.06418024003505707,
      "step": 2645
    },
    {
      "epoch": 0.3577216537528834,
      "grad_norm": 0.21417228877544403,
      "learning_rate": 2.3438891274297348e-05,
      "loss": 0.045081257820129395,
      "step": 2646
    },
    {
      "epoch": 0.35785684712164867,
      "grad_norm": 1.4353468418121338,
      "learning_rate": 2.343323187308479e-05,
      "loss": 0.0547470897436142,
      "step": 2647
    },
    {
      "epoch": 0.35799204049041394,
      "grad_norm": 0.4400724470615387,
      "learning_rate": 2.342757071602839e-05,
      "loss": 0.05882125720381737,
      "step": 2648
    },
    {
      "epoch": 0.3581272338591792,
      "grad_norm": 1.221472144126892,
      "learning_rate": 2.3421907804306816e-05,
      "loss": 0.058724045753479004,
      "step": 2649
    },
    {
      "epoch": 0.3582624272279445,
      "grad_norm": 0.5090343952178955,
      "learning_rate": 2.341624313909913e-05,
      "loss": 0.04208928346633911,
      "step": 2650
    },
    {
      "epoch": 0.3583976205967097,
      "grad_norm": 0.3714367151260376,
      "learning_rate": 2.3410576721584742e-05,
      "loss": 0.04616578668355942,
      "step": 2651
    },
    {
      "epoch": 0.358532813965475,
      "grad_norm": 0.8840522766113281,
      "learning_rate": 2.3404908552943435e-05,
      "loss": 0.04824555665254593,
      "step": 2652
    },
    {
      "epoch": 0.35866800733424026,
      "grad_norm": 0.5322878956794739,
      "learning_rate": 2.339923863435534e-05,
      "loss": 0.048618778586387634,
      "step": 2653
    },
    {
      "epoch": 0.35880320070300553,
      "grad_norm": 0.8479301929473877,
      "learning_rate": 2.3393566967000974e-05,
      "loss": 0.06648284196853638,
      "step": 2654
    },
    {
      "epoch": 0.3589383940717708,
      "grad_norm": 1.0377086400985718,
      "learning_rate": 2.3387893552061202e-05,
      "loss": 0.06648188829421997,
      "step": 2655
    },
    {
      "epoch": 0.359073587440536,
      "grad_norm": 0.5264196395874023,
      "learning_rate": 2.3382218390717268e-05,
      "loss": 0.06007471680641174,
      "step": 2656
    },
    {
      "epoch": 0.3592087808093013,
      "grad_norm": 1.1664226055145264,
      "learning_rate": 2.3376541484150762e-05,
      "loss": 0.05334685742855072,
      "step": 2657
    },
    {
      "epoch": 0.35934397417806657,
      "grad_norm": 0.4176519513130188,
      "learning_rate": 2.3370862833543652e-05,
      "loss": 0.05965322256088257,
      "step": 2658
    },
    {
      "epoch": 0.35947916754683185,
      "grad_norm": 0.46354198455810547,
      "learning_rate": 2.336518244007826e-05,
      "loss": 0.05492527782917023,
      "step": 2659
    },
    {
      "epoch": 0.35961436091559706,
      "grad_norm": 0.6516409516334534,
      "learning_rate": 2.3359500304937274e-05,
      "loss": 0.05586463212966919,
      "step": 2660
    },
    {
      "epoch": 0.35974955428436234,
      "grad_norm": 0.6066004633903503,
      "learning_rate": 2.335381642930375e-05,
      "loss": 0.07232370972633362,
      "step": 2661
    },
    {
      "epoch": 0.3598847476531276,
      "grad_norm": 0.9156617522239685,
      "learning_rate": 2.3348130814361094e-05,
      "loss": 0.03705447167158127,
      "step": 2662
    },
    {
      "epoch": 0.3600199410218929,
      "grad_norm": 0.6057007908821106,
      "learning_rate": 2.334244346129309e-05,
      "loss": 0.04913758113980293,
      "step": 2663
    },
    {
      "epoch": 0.36015513439065816,
      "grad_norm": 0.31930455565452576,
      "learning_rate": 2.3336754371283862e-05,
      "loss": 0.0419168621301651,
      "step": 2664
    },
    {
      "epoch": 0.3602903277594234,
      "grad_norm": 0.30003830790519714,
      "learning_rate": 2.333106354551792e-05,
      "loss": 0.05088382586836815,
      "step": 2665
    },
    {
      "epoch": 0.36042552112818865,
      "grad_norm": 0.9144757390022278,
      "learning_rate": 2.332537098518012e-05,
      "loss": 0.07023753225803375,
      "step": 2666
    },
    {
      "epoch": 0.3605607144969539,
      "grad_norm": 0.21278853714466095,
      "learning_rate": 2.3319676691455686e-05,
      "loss": 0.03908340632915497,
      "step": 2667
    },
    {
      "epoch": 0.3606959078657192,
      "grad_norm": 0.425176203250885,
      "learning_rate": 2.3313980665530205e-05,
      "loss": 0.052151650190353394,
      "step": 2668
    },
    {
      "epoch": 0.3608311012344845,
      "grad_norm": 0.7476831078529358,
      "learning_rate": 2.3308282908589606e-05,
      "loss": 0.04653248190879822,
      "step": 2669
    },
    {
      "epoch": 0.3609662946032497,
      "grad_norm": 1.47437584400177,
      "learning_rate": 2.330258342182021e-05,
      "loss": 0.05309110879898071,
      "step": 2670
    },
    {
      "epoch": 0.36110148797201497,
      "grad_norm": 1.4785507917404175,
      "learning_rate": 2.329688220640866e-05,
      "loss": 0.05830750614404678,
      "step": 2671
    },
    {
      "epoch": 0.36123668134078024,
      "grad_norm": 1.1409943103790283,
      "learning_rate": 2.329117926354199e-05,
      "loss": 0.0646219328045845,
      "step": 2672
    },
    {
      "epoch": 0.3613718747095455,
      "grad_norm": 0.2731613516807556,
      "learning_rate": 2.3285474594407588e-05,
      "loss": 0.03335670754313469,
      "step": 2673
    },
    {
      "epoch": 0.36150706807831073,
      "grad_norm": 1.2858467102050781,
      "learning_rate": 2.327976820019319e-05,
      "loss": 0.06651624292135239,
      "step": 2674
    },
    {
      "epoch": 0.361642261447076,
      "grad_norm": 0.5883856415748596,
      "learning_rate": 2.32740600820869e-05,
      "loss": 0.0514206625521183,
      "step": 2675
    },
    {
      "epoch": 0.3617774548158413,
      "grad_norm": 1.06332266330719,
      "learning_rate": 2.326835024127718e-05,
      "loss": 0.07948241382837296,
      "step": 2676
    },
    {
      "epoch": 0.36191264818460656,
      "grad_norm": 0.47783055901527405,
      "learning_rate": 2.326263867895285e-05,
      "loss": 0.07001806795597076,
      "step": 2677
    },
    {
      "epoch": 0.36204784155337183,
      "grad_norm": 0.7671249508857727,
      "learning_rate": 2.3256925396303076e-05,
      "loss": 0.04274945706129074,
      "step": 2678
    },
    {
      "epoch": 0.36218303492213705,
      "grad_norm": 0.8156877160072327,
      "learning_rate": 2.3251210394517412e-05,
      "loss": 0.05127374827861786,
      "step": 2679
    },
    {
      "epoch": 0.3623182282909023,
      "grad_norm": 0.48961812257766724,
      "learning_rate": 2.3245493674785742e-05,
      "loss": 0.06561869382858276,
      "step": 2680
    },
    {
      "epoch": 0.3624534216596676,
      "grad_norm": 0.4243928790092468,
      "learning_rate": 2.3239775238298316e-05,
      "loss": 0.06535175442695618,
      "step": 2681
    },
    {
      "epoch": 0.36258861502843287,
      "grad_norm": 1.0964494943618774,
      "learning_rate": 2.3234055086245744e-05,
      "loss": 0.06075618043541908,
      "step": 2682
    },
    {
      "epoch": 0.36272380839719814,
      "grad_norm": 0.7016053795814514,
      "learning_rate": 2.3228333219818998e-05,
      "loss": 0.051271483302116394,
      "step": 2683
    },
    {
      "epoch": 0.36285900176596336,
      "grad_norm": 0.358658105134964,
      "learning_rate": 2.3222609640209397e-05,
      "loss": 0.05346973240375519,
      "step": 2684
    },
    {
      "epoch": 0.36299419513472864,
      "grad_norm": 0.361572265625,
      "learning_rate": 2.3216884348608614e-05,
      "loss": 0.06042298674583435,
      "step": 2685
    },
    {
      "epoch": 0.3631293885034939,
      "grad_norm": 0.8352283835411072,
      "learning_rate": 2.32111573462087e-05,
      "loss": 0.06254858523607254,
      "step": 2686
    },
    {
      "epoch": 0.3632645818722592,
      "grad_norm": 0.563700258731842,
      "learning_rate": 2.3205428634202028e-05,
      "loss": 0.043987635523080826,
      "step": 2687
    },
    {
      "epoch": 0.3633997752410244,
      "grad_norm": 0.49695372581481934,
      "learning_rate": 2.3199698213781367e-05,
      "loss": 0.053213298320770264,
      "step": 2688
    },
    {
      "epoch": 0.3635349686097897,
      "grad_norm": 0.6971237659454346,
      "learning_rate": 2.319396608613981e-05,
      "loss": 0.06444203853607178,
      "step": 2689
    },
    {
      "epoch": 0.36367016197855495,
      "grad_norm": 0.3077828288078308,
      "learning_rate": 2.318823225247082e-05,
      "loss": 0.0511898472905159,
      "step": 2690
    },
    {
      "epoch": 0.3638053553473202,
      "grad_norm": 0.2585951089859009,
      "learning_rate": 2.3182496713968208e-05,
      "loss": 0.050716765224933624,
      "step": 2691
    },
    {
      "epoch": 0.3639405487160855,
      "grad_norm": 0.4897538721561432,
      "learning_rate": 2.3176759471826143e-05,
      "loss": 0.0444374606013298,
      "step": 2692
    },
    {
      "epoch": 0.3640757420848507,
      "grad_norm": 0.469706654548645,
      "learning_rate": 2.3171020527239155e-05,
      "loss": 0.04763749986886978,
      "step": 2693
    },
    {
      "epoch": 0.364210935453616,
      "grad_norm": 0.27218949794769287,
      "learning_rate": 2.316527988140212e-05,
      "loss": 0.031900662928819656,
      "step": 2694
    },
    {
      "epoch": 0.36434612882238127,
      "grad_norm": 0.5049083232879639,
      "learning_rate": 2.315953753551027e-05,
      "loss": 0.04299360513687134,
      "step": 2695
    },
    {
      "epoch": 0.36448132219114654,
      "grad_norm": 0.3708631992340088,
      "learning_rate": 2.3153793490759197e-05,
      "loss": 0.048834264278411865,
      "step": 2696
    },
    {
      "epoch": 0.3646165155599118,
      "grad_norm": 1.4542580842971802,
      "learning_rate": 2.3148047748344835e-05,
      "loss": 0.05752076208591461,
      "step": 2697
    },
    {
      "epoch": 0.36475170892867703,
      "grad_norm": 0.33774030208587646,
      "learning_rate": 2.314230030946348e-05,
      "loss": 0.04336026310920715,
      "step": 2698
    },
    {
      "epoch": 0.3648869022974423,
      "grad_norm": 1.5721356868743896,
      "learning_rate": 2.3136551175311782e-05,
      "loss": 0.06467732042074203,
      "step": 2699
    },
    {
      "epoch": 0.3650220956662076,
      "grad_norm": 0.2294367402791977,
      "learning_rate": 2.313080034708674e-05,
      "loss": 0.04604820907115936,
      "step": 2700
    },
    {
      "epoch": 0.36515728903497285,
      "grad_norm": 0.3811178505420685,
      "learning_rate": 2.312504782598571e-05,
      "loss": 0.05043311044573784,
      "step": 2701
    },
    {
      "epoch": 0.3652924824037381,
      "grad_norm": 0.3909724950790405,
      "learning_rate": 2.311929361320639e-05,
      "loss": 0.061378300189971924,
      "step": 2702
    },
    {
      "epoch": 0.36542767577250335,
      "grad_norm": 0.7049826979637146,
      "learning_rate": 2.311353770994684e-05,
      "loss": 0.04515687748789787,
      "step": 2703
    },
    {
      "epoch": 0.3655628691412686,
      "grad_norm": 0.2656460106372833,
      "learning_rate": 2.310778011740548e-05,
      "loss": 0.048185400664806366,
      "step": 2704
    },
    {
      "epoch": 0.3656980625100339,
      "grad_norm": 1.106172800064087,
      "learning_rate": 2.310202083678106e-05,
      "loss": 0.06623083353042603,
      "step": 2705
    },
    {
      "epoch": 0.36583325587879917,
      "grad_norm": 1.0797213315963745,
      "learning_rate": 2.3096259869272694e-05,
      "loss": 0.05998314544558525,
      "step": 2706
    },
    {
      "epoch": 0.3659684492475644,
      "grad_norm": 1.055147647857666,
      "learning_rate": 2.309049721607985e-05,
      "loss": 0.07395990192890167,
      "step": 2707
    },
    {
      "epoch": 0.36610364261632966,
      "grad_norm": 0.5194681286811829,
      "learning_rate": 2.3084732878402342e-05,
      "loss": 0.05055910348892212,
      "step": 2708
    },
    {
      "epoch": 0.36623883598509493,
      "grad_norm": 0.5933238863945007,
      "learning_rate": 2.307896685744034e-05,
      "loss": 0.06610037386417389,
      "step": 2709
    },
    {
      "epoch": 0.3663740293538602,
      "grad_norm": 0.25912216305732727,
      "learning_rate": 2.3073199154394352e-05,
      "loss": 0.037207845598459244,
      "step": 2710
    },
    {
      "epoch": 0.3665092227226255,
      "grad_norm": 0.8808997273445129,
      "learning_rate": 2.3067429770465246e-05,
      "loss": 0.065489262342453,
      "step": 2711
    },
    {
      "epoch": 0.3666444160913907,
      "grad_norm": 0.8706968426704407,
      "learning_rate": 2.3061658706854244e-05,
      "loss": 0.06867433339357376,
      "step": 2712
    },
    {
      "epoch": 0.366779609460156,
      "grad_norm": 0.5530458688735962,
      "learning_rate": 2.3055885964762907e-05,
      "loss": 0.059696100652217865,
      "step": 2713
    },
    {
      "epoch": 0.36691480282892125,
      "grad_norm": 0.8945398330688477,
      "learning_rate": 2.3050111545393156e-05,
      "loss": 0.05344860255718231,
      "step": 2714
    },
    {
      "epoch": 0.3670499961976865,
      "grad_norm": 1.4480206966400146,
      "learning_rate": 2.304433544994725e-05,
      "loss": 0.057323891669511795,
      "step": 2715
    },
    {
      "epoch": 0.36718518956645174,
      "grad_norm": 0.6804830431938171,
      "learning_rate": 2.303855767962781e-05,
      "loss": 0.05345511436462402,
      "step": 2716
    },
    {
      "epoch": 0.367320382935217,
      "grad_norm": 0.5865732431411743,
      "learning_rate": 2.303277823563779e-05,
      "loss": 0.060651302337646484,
      "step": 2717
    },
    {
      "epoch": 0.3674555763039823,
      "grad_norm": 0.46498337388038635,
      "learning_rate": 2.3026997119180507e-05,
      "loss": 0.04442206025123596,
      "step": 2718
    },
    {
      "epoch": 0.36759076967274756,
      "grad_norm": 0.7256549596786499,
      "learning_rate": 2.3021214331459616e-05,
      "loss": 0.04756008833646774,
      "step": 2719
    },
    {
      "epoch": 0.36772596304151284,
      "grad_norm": 1.356463074684143,
      "learning_rate": 2.301542987367913e-05,
      "loss": 0.0645064115524292,
      "step": 2720
    },
    {
      "epoch": 0.36786115641027806,
      "grad_norm": 0.3430469036102295,
      "learning_rate": 2.3009643747043403e-05,
      "loss": 0.03879815340042114,
      "step": 2721
    },
    {
      "epoch": 0.36799634977904333,
      "grad_norm": 0.5895841717720032,
      "learning_rate": 2.3003855952757132e-05,
      "loss": 0.0735158622264862,
      "step": 2722
    },
    {
      "epoch": 0.3681315431478086,
      "grad_norm": 1.2479759454727173,
      "learning_rate": 2.2998066492025372e-05,
      "loss": 0.05335110425949097,
      "step": 2723
    },
    {
      "epoch": 0.3682667365165739,
      "grad_norm": 0.5266788601875305,
      "learning_rate": 2.2992275366053513e-05,
      "loss": 0.06457316875457764,
      "step": 2724
    },
    {
      "epoch": 0.3684019298853391,
      "grad_norm": 0.6879621148109436,
      "learning_rate": 2.2986482576047305e-05,
      "loss": 0.07312683761119843,
      "step": 2725
    },
    {
      "epoch": 0.36853712325410437,
      "grad_norm": 0.4298947751522064,
      "learning_rate": 2.298068812321284e-05,
      "loss": 0.03696560487151146,
      "step": 2726
    },
    {
      "epoch": 0.36867231662286964,
      "grad_norm": 0.4071572422981262,
      "learning_rate": 2.297489200875654e-05,
      "loss": 0.036071836948394775,
      "step": 2727
    },
    {
      "epoch": 0.3688075099916349,
      "grad_norm": 0.8441872000694275,
      "learning_rate": 2.2969094233885204e-05,
      "loss": 0.048501357436180115,
      "step": 2728
    },
    {
      "epoch": 0.3689427033604002,
      "grad_norm": 0.9283362627029419,
      "learning_rate": 2.296329479980595e-05,
      "loss": 0.06408321857452393,
      "step": 2729
    },
    {
      "epoch": 0.3690778967291654,
      "grad_norm": 0.7360424995422363,
      "learning_rate": 2.2957493707726252e-05,
      "loss": 0.049999088048934937,
      "step": 2730
    },
    {
      "epoch": 0.3692130900979307,
      "grad_norm": 0.39487510919570923,
      "learning_rate": 2.2951690958853932e-05,
      "loss": 0.05361481010913849,
      "step": 2731
    },
    {
      "epoch": 0.36934828346669596,
      "grad_norm": 0.5209821462631226,
      "learning_rate": 2.2945886554397154e-05,
      "loss": 0.05601753294467926,
      "step": 2732
    },
    {
      "epoch": 0.36948347683546123,
      "grad_norm": 0.42719781398773193,
      "learning_rate": 2.294008049556441e-05,
      "loss": 0.056699272245168686,
      "step": 2733
    },
    {
      "epoch": 0.3696186702042265,
      "grad_norm": 0.38892847299575806,
      "learning_rate": 2.2934272783564577e-05,
      "loss": 0.06433428823947906,
      "step": 2734
    },
    {
      "epoch": 0.3697538635729917,
      "grad_norm": 0.29517319798469543,
      "learning_rate": 2.2928463419606835e-05,
      "loss": 0.03910673037171364,
      "step": 2735
    },
    {
      "epoch": 0.369889056941757,
      "grad_norm": 1.056670904159546,
      "learning_rate": 2.292265240490073e-05,
      "loss": 0.05807200074195862,
      "step": 2736
    },
    {
      "epoch": 0.3700242503105223,
      "grad_norm": 0.22617408633232117,
      "learning_rate": 2.2916839740656154e-05,
      "loss": 0.03555326163768768,
      "step": 2737
    },
    {
      "epoch": 0.37015944367928755,
      "grad_norm": 1.2022522687911987,
      "learning_rate": 2.2911025428083316e-05,
      "loss": 0.0521751344203949,
      "step": 2738
    },
    {
      "epoch": 0.37029463704805277,
      "grad_norm": 0.2654719948768616,
      "learning_rate": 2.2905209468392798e-05,
      "loss": 0.051092736423015594,
      "step": 2739
    },
    {
      "epoch": 0.37042983041681804,
      "grad_norm": 0.5073826909065247,
      "learning_rate": 2.2899391862795514e-05,
      "loss": 0.04168633744120598,
      "step": 2740
    },
    {
      "epoch": 0.3705650237855833,
      "grad_norm": 0.35489004850387573,
      "learning_rate": 2.2893572612502718e-05,
      "loss": 0.06316696852445602,
      "step": 2741
    },
    {
      "epoch": 0.3707002171543486,
      "grad_norm": 0.6103214621543884,
      "learning_rate": 2.2887751718726013e-05,
      "loss": 0.04942135512828827,
      "step": 2742
    },
    {
      "epoch": 0.37083541052311386,
      "grad_norm": 0.3600391447544098,
      "learning_rate": 2.288192918267734e-05,
      "loss": 0.05429067462682724,
      "step": 2743
    },
    {
      "epoch": 0.3709706038918791,
      "grad_norm": 0.295093297958374,
      "learning_rate": 2.2876105005568974e-05,
      "loss": 0.047676101326942444,
      "step": 2744
    },
    {
      "epoch": 0.37110579726064435,
      "grad_norm": 1.0416713953018188,
      "learning_rate": 2.287027918861355e-05,
      "loss": 0.05760412663221359,
      "step": 2745
    },
    {
      "epoch": 0.37124099062940963,
      "grad_norm": 0.42710763216018677,
      "learning_rate": 2.2864451733024024e-05,
      "loss": 0.06148383766412735,
      "step": 2746
    },
    {
      "epoch": 0.3713761839981749,
      "grad_norm": 0.5819581747055054,
      "learning_rate": 2.2858622640013716e-05,
      "loss": 0.06356078386306763,
      "step": 2747
    },
    {
      "epoch": 0.3715113773669402,
      "grad_norm": 1.125916838645935,
      "learning_rate": 2.285279191079626e-05,
      "loss": 0.05261853337287903,
      "step": 2748
    },
    {
      "epoch": 0.3716465707357054,
      "grad_norm": 1.5052939653396606,
      "learning_rate": 2.2846959546585656e-05,
      "loss": 0.04465719312429428,
      "step": 2749
    },
    {
      "epoch": 0.37178176410447067,
      "grad_norm": 1.099485993385315,
      "learning_rate": 2.2841125548596225e-05,
      "loss": 0.06966456770896912,
      "step": 2750
    },
    {
      "epoch": 0.37191695747323594,
      "grad_norm": 0.36587652564048767,
      "learning_rate": 2.2835289918042648e-05,
      "loss": 0.06398959457874298,
      "step": 2751
    },
    {
      "epoch": 0.3720521508420012,
      "grad_norm": 0.3534739315509796,
      "learning_rate": 2.282945265613992e-05,
      "loss": 0.05443921685218811,
      "step": 2752
    },
    {
      "epoch": 0.37218734421076644,
      "grad_norm": 0.21372392773628235,
      "learning_rate": 2.2823613764103406e-05,
      "loss": 0.044846728444099426,
      "step": 2753
    },
    {
      "epoch": 0.3723225375795317,
      "grad_norm": 0.2837085723876953,
      "learning_rate": 2.2817773243148776e-05,
      "loss": 0.03718673065304756,
      "step": 2754
    },
    {
      "epoch": 0.372457730948297,
      "grad_norm": 0.5193917155265808,
      "learning_rate": 2.2811931094492074e-05,
      "loss": 0.07798799872398376,
      "step": 2755
    },
    {
      "epoch": 0.37259292431706226,
      "grad_norm": 0.48120924830436707,
      "learning_rate": 2.280608731934966e-05,
      "loss": 0.054799821227788925,
      "step": 2756
    },
    {
      "epoch": 0.37272811768582753,
      "grad_norm": 0.4023662507534027,
      "learning_rate": 2.280024191893823e-05,
      "loss": 0.05645638704299927,
      "step": 2757
    },
    {
      "epoch": 0.37286331105459275,
      "grad_norm": 0.6345804333686829,
      "learning_rate": 2.279439489447485e-05,
      "loss": 0.06636759638786316,
      "step": 2758
    },
    {
      "epoch": 0.372998504423358,
      "grad_norm": 0.7730368971824646,
      "learning_rate": 2.278854624717688e-05,
      "loss": 0.04965287446975708,
      "step": 2759
    },
    {
      "epoch": 0.3731336977921233,
      "grad_norm": 0.39420077204704285,
      "learning_rate": 2.2782695978262045e-05,
      "loss": 0.06774768233299255,
      "step": 2760
    },
    {
      "epoch": 0.37326889116088857,
      "grad_norm": 0.9592914581298828,
      "learning_rate": 2.2776844088948406e-05,
      "loss": 0.06119055300951004,
      "step": 2761
    },
    {
      "epoch": 0.37340408452965385,
      "grad_norm": 1.4865642786026,
      "learning_rate": 2.2770990580454364e-05,
      "loss": 0.06855255365371704,
      "step": 2762
    },
    {
      "epoch": 0.37353927789841906,
      "grad_norm": 1.741552472114563,
      "learning_rate": 2.276513545399864e-05,
      "loss": 0.059029772877693176,
      "step": 2763
    },
    {
      "epoch": 0.37367447126718434,
      "grad_norm": 0.21837197244167328,
      "learning_rate": 2.2759278710800306e-05,
      "loss": 0.03799697756767273,
      "step": 2764
    },
    {
      "epoch": 0.3738096646359496,
      "grad_norm": 0.784138560295105,
      "learning_rate": 2.275342035207876e-05,
      "loss": 0.05388563871383667,
      "step": 2765
    },
    {
      "epoch": 0.3739448580047149,
      "grad_norm": 1.3761948347091675,
      "learning_rate": 2.2747560379053752e-05,
      "loss": 0.0802559107542038,
      "step": 2766
    },
    {
      "epoch": 0.3740800513734801,
      "grad_norm": 0.3484854996204376,
      "learning_rate": 2.2741698792945364e-05,
      "loss": 0.0629880428314209,
      "step": 2767
    },
    {
      "epoch": 0.3742152447422454,
      "grad_norm": 0.3504173457622528,
      "learning_rate": 2.2735835594974003e-05,
      "loss": 0.05580052733421326,
      "step": 2768
    },
    {
      "epoch": 0.37435043811101065,
      "grad_norm": 0.7178674340248108,
      "learning_rate": 2.272997078636042e-05,
      "loss": 0.04396190121769905,
      "step": 2769
    },
    {
      "epoch": 0.3744856314797759,
      "grad_norm": 0.7006298303604126,
      "learning_rate": 2.272410436832569e-05,
      "loss": 0.06192396208643913,
      "step": 2770
    },
    {
      "epoch": 0.3746208248485412,
      "grad_norm": 0.6319068074226379,
      "learning_rate": 2.2718236342091248e-05,
      "loss": 0.04144385829567909,
      "step": 2771
    },
    {
      "epoch": 0.3747560182173064,
      "grad_norm": 1.289096713066101,
      "learning_rate": 2.2712366708878838e-05,
      "loss": 0.06428465247154236,
      "step": 2772
    },
    {
      "epoch": 0.3748912115860717,
      "grad_norm": 0.6738072633743286,
      "learning_rate": 2.2706495469910552e-05,
      "loss": 0.05183836817741394,
      "step": 2773
    },
    {
      "epoch": 0.37502640495483697,
      "grad_norm": 0.5354626178741455,
      "learning_rate": 2.2700622626408814e-05,
      "loss": 0.046249933540821075,
      "step": 2774
    },
    {
      "epoch": 0.37516159832360224,
      "grad_norm": 0.5030537843704224,
      "learning_rate": 2.2694748179596375e-05,
      "loss": 0.05862545967102051,
      "step": 2775
    },
    {
      "epoch": 0.3752967916923675,
      "grad_norm": 0.4092203676700592,
      "learning_rate": 2.2688872130696342e-05,
      "loss": 0.05227592587471008,
      "step": 2776
    },
    {
      "epoch": 0.37543198506113273,
      "grad_norm": 0.3558029532432556,
      "learning_rate": 2.268299448093212e-05,
      "loss": 0.05536097288131714,
      "step": 2777
    },
    {
      "epoch": 0.375567178429898,
      "grad_norm": 0.6395118832588196,
      "learning_rate": 2.2677115231527482e-05,
      "loss": 0.049967266619205475,
      "step": 2778
    },
    {
      "epoch": 0.3757023717986633,
      "grad_norm": 0.6145211458206177,
      "learning_rate": 2.267123438370651e-05,
      "loss": 0.07054963707923889,
      "step": 2779
    },
    {
      "epoch": 0.37583756516742856,
      "grad_norm": 0.6234350800514221,
      "learning_rate": 2.266535193869363e-05,
      "loss": 0.056785374879837036,
      "step": 2780
    },
    {
      "epoch": 0.3759727585361938,
      "grad_norm": 0.3293749988079071,
      "learning_rate": 2.2659467897713604e-05,
      "loss": 0.053153231739997864,
      "step": 2781
    },
    {
      "epoch": 0.37610795190495905,
      "grad_norm": 0.6051040291786194,
      "learning_rate": 2.2653582261991516e-05,
      "loss": 0.05777176097035408,
      "step": 2782
    },
    {
      "epoch": 0.3762431452737243,
      "grad_norm": 0.5731443166732788,
      "learning_rate": 2.2647695032752785e-05,
      "loss": 0.05151727795600891,
      "step": 2783
    },
    {
      "epoch": 0.3763783386424896,
      "grad_norm": 0.5319263935089111,
      "learning_rate": 2.264180621122317e-05,
      "loss": 0.04741548374295235,
      "step": 2784
    },
    {
      "epoch": 0.37651353201125487,
      "grad_norm": 0.3491629958152771,
      "learning_rate": 2.2635915798628747e-05,
      "loss": 0.05430998653173447,
      "step": 2785
    },
    {
      "epoch": 0.3766487253800201,
      "grad_norm": 0.5214946269989014,
      "learning_rate": 2.2630023796195932e-05,
      "loss": 0.05097288638353348,
      "step": 2786
    },
    {
      "epoch": 0.37678391874878536,
      "grad_norm": 0.4239116609096527,
      "learning_rate": 2.262413020515148e-05,
      "loss": 0.047918349504470825,
      "step": 2787
    },
    {
      "epoch": 0.37691911211755064,
      "grad_norm": 0.7723445296287537,
      "learning_rate": 2.261823502672246e-05,
      "loss": 0.06741046905517578,
      "step": 2788
    },
    {
      "epoch": 0.3770543054863159,
      "grad_norm": 0.6508707404136658,
      "learning_rate": 2.261233826213628e-05,
      "loss": 0.05089424550533295,
      "step": 2789
    },
    {
      "epoch": 0.3771894988550812,
      "grad_norm": 0.9169923663139343,
      "learning_rate": 2.2606439912620688e-05,
      "loss": 0.05232696235179901,
      "step": 2790
    },
    {
      "epoch": 0.3773246922238464,
      "grad_norm": 0.7651610970497131,
      "learning_rate": 2.2600539979403734e-05,
      "loss": 0.0608372688293457,
      "step": 2791
    },
    {
      "epoch": 0.3774598855926117,
      "grad_norm": 0.9743182063102722,
      "learning_rate": 2.259463846371383e-05,
      "loss": 0.0558069609105587,
      "step": 2792
    },
    {
      "epoch": 0.37759507896137695,
      "grad_norm": 0.3944699764251709,
      "learning_rate": 2.2588735366779698e-05,
      "loss": 0.0599839985370636,
      "step": 2793
    },
    {
      "epoch": 0.3777302723301422,
      "grad_norm": 0.4863327741622925,
      "learning_rate": 2.2582830689830394e-05,
      "loss": 0.059852585196495056,
      "step": 2794
    },
    {
      "epoch": 0.37786546569890744,
      "grad_norm": 0.30877846479415894,
      "learning_rate": 2.2576924434095305e-05,
      "loss": 0.03760230168700218,
      "step": 2795
    },
    {
      "epoch": 0.3780006590676727,
      "grad_norm": 0.4678315222263336,
      "learning_rate": 2.257101660080414e-05,
      "loss": 0.047534674406051636,
      "step": 2796
    },
    {
      "epoch": 0.378135852436438,
      "grad_norm": 0.38959020376205444,
      "learning_rate": 2.256510719118695e-05,
      "loss": 0.03578207641839981,
      "step": 2797
    },
    {
      "epoch": 0.37827104580520327,
      "grad_norm": 0.9588425159454346,
      "learning_rate": 2.2559196206474094e-05,
      "loss": 0.04739844799041748,
      "step": 2798
    },
    {
      "epoch": 0.37840623917396854,
      "grad_norm": 0.807678759098053,
      "learning_rate": 2.2553283647896287e-05,
      "loss": 0.0760999321937561,
      "step": 2799
    },
    {
      "epoch": 0.37854143254273376,
      "grad_norm": 1.932219386100769,
      "learning_rate": 2.254736951668454e-05,
      "loss": 0.048617780208587646,
      "step": 2800
    },
    {
      "epoch": 0.37867662591149903,
      "grad_norm": 0.5309522747993469,
      "learning_rate": 2.2541453814070212e-05,
      "loss": 0.06753760576248169,
      "step": 2801
    },
    {
      "epoch": 0.3788118192802643,
      "grad_norm": 0.8778548240661621,
      "learning_rate": 2.2535536541284983e-05,
      "loss": 0.06830979883670807,
      "step": 2802
    },
    {
      "epoch": 0.3789470126490296,
      "grad_norm": 0.6769487857818604,
      "learning_rate": 2.2529617699560857e-05,
      "loss": 0.05845741555094719,
      "step": 2803
    },
    {
      "epoch": 0.37908220601779485,
      "grad_norm": 0.326483815908432,
      "learning_rate": 2.2523697290130185e-05,
      "loss": 0.061376750469207764,
      "step": 2804
    },
    {
      "epoch": 0.3792173993865601,
      "grad_norm": 0.8158133029937744,
      "learning_rate": 2.251777531422561e-05,
      "loss": 0.04257579892873764,
      "step": 2805
    },
    {
      "epoch": 0.37935259275532535,
      "grad_norm": 1.3088444471359253,
      "learning_rate": 2.2511851773080127e-05,
      "loss": 0.05869834125041962,
      "step": 2806
    },
    {
      "epoch": 0.3794877861240906,
      "grad_norm": 0.33457887172698975,
      "learning_rate": 2.2505926667927043e-05,
      "loss": 0.05326327681541443,
      "step": 2807
    },
    {
      "epoch": 0.3796229794928559,
      "grad_norm": 0.21494072675704956,
      "learning_rate": 2.25e-05,
      "loss": 0.035292450338602066,
      "step": 2808
    },
    {
      "epoch": 0.3797581728616211,
      "grad_norm": 0.6951077580451965,
      "learning_rate": 2.2494071770532966e-05,
      "loss": 0.06634253263473511,
      "step": 2809
    },
    {
      "epoch": 0.3798933662303864,
      "grad_norm": 0.7285453081130981,
      "learning_rate": 2.2488141980760223e-05,
      "loss": 0.07704959809780121,
      "step": 2810
    },
    {
      "epoch": 0.38002855959915166,
      "grad_norm": 0.21081511676311493,
      "learning_rate": 2.248221063191639e-05,
      "loss": 0.05316847562789917,
      "step": 2811
    },
    {
      "epoch": 0.38016375296791693,
      "grad_norm": 0.9092246890068054,
      "learning_rate": 2.24762777252364e-05,
      "loss": 0.06442821025848389,
      "step": 2812
    },
    {
      "epoch": 0.3802989463366822,
      "grad_norm": 0.2412085086107254,
      "learning_rate": 2.2470343261955525e-05,
      "loss": 0.060990944504737854,
      "step": 2813
    },
    {
      "epoch": 0.3804341397054474,
      "grad_norm": 0.4725414514541626,
      "learning_rate": 2.246440724330934e-05,
      "loss": 0.06503316015005112,
      "step": 2814
    },
    {
      "epoch": 0.3805693330742127,
      "grad_norm": 0.8974441289901733,
      "learning_rate": 2.2458469670533765e-05,
      "loss": 0.06016567349433899,
      "step": 2815
    },
    {
      "epoch": 0.380704526442978,
      "grad_norm": 0.2822750210762024,
      "learning_rate": 2.2452530544865034e-05,
      "loss": 0.04852447658777237,
      "step": 2816
    },
    {
      "epoch": 0.38083971981174325,
      "grad_norm": 0.8395393490791321,
      "learning_rate": 2.24465898675397e-05,
      "loss": 0.0711175799369812,
      "step": 2817
    },
    {
      "epoch": 0.3809749131805085,
      "grad_norm": 0.6369546055793762,
      "learning_rate": 2.244064763979464e-05,
      "loss": 0.061800286173820496,
      "step": 2818
    },
    {
      "epoch": 0.38111010654927374,
      "grad_norm": 0.6756783723831177,
      "learning_rate": 2.2434703862867068e-05,
      "loss": 0.05368724465370178,
      "step": 2819
    },
    {
      "epoch": 0.381245299918039,
      "grad_norm": 0.4739999771118164,
      "learning_rate": 2.2428758537994504e-05,
      "loss": 0.061122551560401917,
      "step": 2820
    },
    {
      "epoch": 0.3813804932868043,
      "grad_norm": 0.3748827278614044,
      "learning_rate": 2.24228116664148e-05,
      "loss": 0.03840610384941101,
      "step": 2821
    },
    {
      "epoch": 0.38151568665556956,
      "grad_norm": 0.4113057255744934,
      "learning_rate": 2.2416863249366125e-05,
      "loss": 0.058167681097984314,
      "step": 2822
    },
    {
      "epoch": 0.3816508800243348,
      "grad_norm": 0.5908141732215881,
      "learning_rate": 2.241091328808696e-05,
      "loss": 0.06661194562911987,
      "step": 2823
    },
    {
      "epoch": 0.38178607339310006,
      "grad_norm": 1.5720164775848389,
      "learning_rate": 2.240496178381614e-05,
      "loss": 0.06552646309137344,
      "step": 2824
    },
    {
      "epoch": 0.38192126676186533,
      "grad_norm": 0.783643364906311,
      "learning_rate": 2.239900873779278e-05,
      "loss": 0.06343169510364532,
      "step": 2825
    },
    {
      "epoch": 0.3820564601306306,
      "grad_norm": 0.40840062499046326,
      "learning_rate": 2.2393054151256352e-05,
      "loss": 0.0626457929611206,
      "step": 2826
    },
    {
      "epoch": 0.3821916534993959,
      "grad_norm": 0.290694922208786,
      "learning_rate": 2.238709802544662e-05,
      "loss": 0.04983705282211304,
      "step": 2827
    },
    {
      "epoch": 0.3823268468681611,
      "grad_norm": 0.38798582553863525,
      "learning_rate": 2.2381140361603686e-05,
      "loss": 0.0732077807188034,
      "step": 2828
    },
    {
      "epoch": 0.38246204023692637,
      "grad_norm": 0.5534157156944275,
      "learning_rate": 2.237518116096797e-05,
      "loss": 0.053943514823913574,
      "step": 2829
    },
    {
      "epoch": 0.38259723360569164,
      "grad_norm": 0.34634479880332947,
      "learning_rate": 2.2369220424780203e-05,
      "loss": 0.060630276799201965,
      "step": 2830
    },
    {
      "epoch": 0.3827324269744569,
      "grad_norm": 0.561322808265686,
      "learning_rate": 2.2363258154281452e-05,
      "loss": 0.046656541526317596,
      "step": 2831
    },
    {
      "epoch": 0.3828676203432222,
      "grad_norm": 0.3773760199546814,
      "learning_rate": 2.2357294350713088e-05,
      "loss": 0.057856976985931396,
      "step": 2832
    },
    {
      "epoch": 0.3830028137119874,
      "grad_norm": 0.31006112694740295,
      "learning_rate": 2.2351329015316802e-05,
      "loss": 0.05619370937347412,
      "step": 2833
    },
    {
      "epoch": 0.3831380070807527,
      "grad_norm": 0.4072780907154083,
      "learning_rate": 2.2345362149334613e-05,
      "loss": 0.05969315767288208,
      "step": 2834
    },
    {
      "epoch": 0.38327320044951796,
      "grad_norm": 0.6256647706031799,
      "learning_rate": 2.2339393754008854e-05,
      "loss": 0.06885421276092529,
      "step": 2835
    },
    {
      "epoch": 0.38340839381828323,
      "grad_norm": 0.5506196618080139,
      "learning_rate": 2.233342383058218e-05,
      "loss": 0.058861277997493744,
      "step": 2836
    },
    {
      "epoch": 0.38354358718704845,
      "grad_norm": 0.5706689953804016,
      "learning_rate": 2.2327452380297554e-05,
      "loss": 0.05928713083267212,
      "step": 2837
    },
    {
      "epoch": 0.3836787805558137,
      "grad_norm": 0.6916266083717346,
      "learning_rate": 2.232147940439827e-05,
      "loss": 0.054866909980773926,
      "step": 2838
    },
    {
      "epoch": 0.383813973924579,
      "grad_norm": 1.3635517358779907,
      "learning_rate": 2.2315504904127936e-05,
      "loss": 0.06967507302761078,
      "step": 2839
    },
    {
      "epoch": 0.3839491672933443,
      "grad_norm": 0.40929463505744934,
      "learning_rate": 2.2309528880730463e-05,
      "loss": 0.057257890701293945,
      "step": 2840
    },
    {
      "epoch": 0.38408436066210955,
      "grad_norm": 1.0062419176101685,
      "learning_rate": 2.2303551335450096e-05,
      "loss": 0.04601931944489479,
      "step": 2841
    },
    {
      "epoch": 0.38421955403087477,
      "grad_norm": 0.3769949674606323,
      "learning_rate": 2.2297572269531398e-05,
      "loss": 0.05190584063529968,
      "step": 2842
    },
    {
      "epoch": 0.38435474739964004,
      "grad_norm": 0.32087311148643494,
      "learning_rate": 2.2291591684219243e-05,
      "loss": 0.041286490857601166,
      "step": 2843
    },
    {
      "epoch": 0.3844899407684053,
      "grad_norm": 0.292425274848938,
      "learning_rate": 2.2285609580758806e-05,
      "loss": 0.05107489973306656,
      "step": 2844
    },
    {
      "epoch": 0.3846251341371706,
      "grad_norm": 0.5732579827308655,
      "learning_rate": 2.227962596039561e-05,
      "loss": 0.07750353217124939,
      "step": 2845
    },
    {
      "epoch": 0.3847603275059358,
      "grad_norm": 0.4695257842540741,
      "learning_rate": 2.2273640824375462e-05,
      "loss": 0.0394764244556427,
      "step": 2846
    },
    {
      "epoch": 0.3848955208747011,
      "grad_norm": 0.25707632303237915,
      "learning_rate": 2.2267654173944515e-05,
      "loss": 0.04923805594444275,
      "step": 2847
    },
    {
      "epoch": 0.38503071424346635,
      "grad_norm": 0.4969066381454468,
      "learning_rate": 2.2261666010349212e-05,
      "loss": 0.05975627154111862,
      "step": 2848
    },
    {
      "epoch": 0.38516590761223163,
      "grad_norm": 0.5696791410446167,
      "learning_rate": 2.2255676334836317e-05,
      "loss": 0.06848347187042236,
      "step": 2849
    },
    {
      "epoch": 0.3853011009809969,
      "grad_norm": 0.2937922179698944,
      "learning_rate": 2.2249685148652917e-05,
      "loss": 0.05625009536743164,
      "step": 2850
    },
    {
      "epoch": 0.3854362943497621,
      "grad_norm": 0.5420044660568237,
      "learning_rate": 2.224369245304641e-05,
      "loss": 0.056445397436618805,
      "step": 2851
    },
    {
      "epoch": 0.3855714877185274,
      "grad_norm": 0.3149201571941376,
      "learning_rate": 2.2237698249264507e-05,
      "loss": 0.0482441782951355,
      "step": 2852
    },
    {
      "epoch": 0.38570668108729267,
      "grad_norm": 0.9132434129714966,
      "learning_rate": 2.2231702538555235e-05,
      "loss": 0.05407146364450455,
      "step": 2853
    },
    {
      "epoch": 0.38584187445605794,
      "grad_norm": 0.4237323999404907,
      "learning_rate": 2.2225705322166928e-05,
      "loss": 0.04787974804639816,
      "step": 2854
    },
    {
      "epoch": 0.3859770678248232,
      "grad_norm": 0.4087650179862976,
      "learning_rate": 2.2219706601348242e-05,
      "loss": 0.041061609983444214,
      "step": 2855
    },
    {
      "epoch": 0.38611226119358844,
      "grad_norm": 1.379231572151184,
      "learning_rate": 2.221370637734814e-05,
      "loss": 0.07291083037853241,
      "step": 2856
    },
    {
      "epoch": 0.3862474545623537,
      "grad_norm": 0.4396313726902008,
      "learning_rate": 2.22077046514159e-05,
      "loss": 0.05511271953582764,
      "step": 2857
    },
    {
      "epoch": 0.386382647931119,
      "grad_norm": 0.3383207321166992,
      "learning_rate": 2.220170142480112e-05,
      "loss": 0.055924803018569946,
      "step": 2858
    },
    {
      "epoch": 0.38651784129988426,
      "grad_norm": 0.22424940764904022,
      "learning_rate": 2.2195696698753695e-05,
      "loss": 0.036877959966659546,
      "step": 2859
    },
    {
      "epoch": 0.3866530346686495,
      "grad_norm": 0.6279323101043701,
      "learning_rate": 2.2189690474523844e-05,
      "loss": 0.06415797024965286,
      "step": 2860
    },
    {
      "epoch": 0.38678822803741475,
      "grad_norm": 0.4282463490962982,
      "learning_rate": 2.21836827533621e-05,
      "loss": 0.05457483232021332,
      "step": 2861
    },
    {
      "epoch": 0.38692342140618,
      "grad_norm": 0.3078089654445648,
      "learning_rate": 2.2177673536519297e-05,
      "loss": 0.04797707498073578,
      "step": 2862
    },
    {
      "epoch": 0.3870586147749453,
      "grad_norm": 0.2630147635936737,
      "learning_rate": 2.217166282524659e-05,
      "loss": 0.05074311047792435,
      "step": 2863
    },
    {
      "epoch": 0.38719380814371057,
      "grad_norm": 0.6083945631980896,
      "learning_rate": 2.216565062079544e-05,
      "loss": 0.04444461315870285,
      "step": 2864
    },
    {
      "epoch": 0.3873290015124758,
      "grad_norm": 0.9929003119468689,
      "learning_rate": 2.2159636924417612e-05,
      "loss": 0.06157923489809036,
      "step": 2865
    },
    {
      "epoch": 0.38746419488124106,
      "grad_norm": 0.6618122458457947,
      "learning_rate": 2.2153621737365205e-05,
      "loss": 0.04268050193786621,
      "step": 2866
    },
    {
      "epoch": 0.38759938825000634,
      "grad_norm": 0.424345999956131,
      "learning_rate": 2.2147605060890598e-05,
      "loss": 0.040484413504600525,
      "step": 2867
    },
    {
      "epoch": 0.3877345816187716,
      "grad_norm": 1.0123282670974731,
      "learning_rate": 2.2141586896246503e-05,
      "loss": 0.061379581689834595,
      "step": 2868
    },
    {
      "epoch": 0.3878697749875369,
      "grad_norm": 0.46783512830734253,
      "learning_rate": 2.2135567244685933e-05,
      "loss": 0.047980695962905884,
      "step": 2869
    },
    {
      "epoch": 0.3880049683563021,
      "grad_norm": 0.5564154982566833,
      "learning_rate": 2.2129546107462214e-05,
      "loss": 0.05251924321055412,
      "step": 2870
    },
    {
      "epoch": 0.3881401617250674,
      "grad_norm": 0.6100112795829773,
      "learning_rate": 2.212352348582897e-05,
      "loss": 0.06808388233184814,
      "step": 2871
    },
    {
      "epoch": 0.38827535509383265,
      "grad_norm": 0.32874441146850586,
      "learning_rate": 2.2117499381040157e-05,
      "loss": 0.06659463047981262,
      "step": 2872
    },
    {
      "epoch": 0.3884105484625979,
      "grad_norm": 0.5706319212913513,
      "learning_rate": 2.211147379435001e-05,
      "loss": 0.04702724516391754,
      "step": 2873
    },
    {
      "epoch": 0.38854574183136315,
      "grad_norm": 1.2208558320999146,
      "learning_rate": 2.2105446727013098e-05,
      "loss": 0.06388106197118759,
      "step": 2874
    },
    {
      "epoch": 0.3886809352001284,
      "grad_norm": 0.40987250208854675,
      "learning_rate": 2.209941818028429e-05,
      "loss": 0.059205979108810425,
      "step": 2875
    },
    {
      "epoch": 0.3888161285688937,
      "grad_norm": 0.4029499590396881,
      "learning_rate": 2.2093388155418757e-05,
      "loss": 0.05426451563835144,
      "step": 2876
    },
    {
      "epoch": 0.38895132193765897,
      "grad_norm": 0.621417224407196,
      "learning_rate": 2.2087356653671982e-05,
      "loss": 0.05353108048439026,
      "step": 2877
    },
    {
      "epoch": 0.38908651530642424,
      "grad_norm": 0.24934527277946472,
      "learning_rate": 2.2081323676299756e-05,
      "loss": 0.046502117067575455,
      "step": 2878
    },
    {
      "epoch": 0.38922170867518946,
      "grad_norm": 0.34832093119621277,
      "learning_rate": 2.207528922455818e-05,
      "loss": 0.05424058437347412,
      "step": 2879
    },
    {
      "epoch": 0.38935690204395473,
      "grad_norm": 0.46701887249946594,
      "learning_rate": 2.206925329970366e-05,
      "loss": 0.047336872667074203,
      "step": 2880
    },
    {
      "epoch": 0.38949209541272,
      "grad_norm": 1.636969804763794,
      "learning_rate": 2.20632159029929e-05,
      "loss": 0.08789052069187164,
      "step": 2881
    },
    {
      "epoch": 0.3896272887814853,
      "grad_norm": 1.1907824277877808,
      "learning_rate": 2.2057177035682926e-05,
      "loss": 0.07766807079315186,
      "step": 2882
    },
    {
      "epoch": 0.38976248215025056,
      "grad_norm": 0.5773687362670898,
      "learning_rate": 2.2051136699031058e-05,
      "loss": 0.07759416103363037,
      "step": 2883
    },
    {
      "epoch": 0.3898976755190158,
      "grad_norm": 0.2676785886287689,
      "learning_rate": 2.2045094894294933e-05,
      "loss": 0.046090394258499146,
      "step": 2884
    },
    {
      "epoch": 0.39003286888778105,
      "grad_norm": 0.4202217161655426,
      "learning_rate": 2.203905162273248e-05,
      "loss": 0.04131387919187546,
      "step": 2885
    },
    {
      "epoch": 0.3901680622565463,
      "grad_norm": 1.154231309890747,
      "learning_rate": 2.203300688560194e-05,
      "loss": 0.06303060054779053,
      "step": 2886
    },
    {
      "epoch": 0.3903032556253116,
      "grad_norm": 0.3597363531589508,
      "learning_rate": 2.2026960684161862e-05,
      "loss": 0.05045539140701294,
      "step": 2887
    },
    {
      "epoch": 0.3904384489940768,
      "grad_norm": 0.9783898591995239,
      "learning_rate": 2.2020913019671097e-05,
      "loss": 0.06813019514083862,
      "step": 2888
    },
    {
      "epoch": 0.3905736423628421,
      "grad_norm": 0.726236879825592,
      "learning_rate": 2.20148638933888e-05,
      "loss": 0.06245800107717514,
      "step": 2889
    },
    {
      "epoch": 0.39070883573160736,
      "grad_norm": 1.0685704946517944,
      "learning_rate": 2.2008813306574438e-05,
      "loss": 0.05793141573667526,
      "step": 2890
    },
    {
      "epoch": 0.39084402910037264,
      "grad_norm": 0.4038330316543579,
      "learning_rate": 2.200276126048777e-05,
      "loss": 0.05636018514633179,
      "step": 2891
    },
    {
      "epoch": 0.3909792224691379,
      "grad_norm": 1.0796375274658203,
      "learning_rate": 2.199670775638886e-05,
      "loss": 0.07029972970485687,
      "step": 2892
    },
    {
      "epoch": 0.39111441583790313,
      "grad_norm": 0.5935426354408264,
      "learning_rate": 2.1990652795538085e-05,
      "loss": 0.08655223995447159,
      "step": 2893
    },
    {
      "epoch": 0.3912496092066684,
      "grad_norm": 0.42973801493644714,
      "learning_rate": 2.1984596379196117e-05,
      "loss": 0.05218249931931496,
      "step": 2894
    },
    {
      "epoch": 0.3913848025754337,
      "grad_norm": 0.36252933740615845,
      "learning_rate": 2.1978538508623942e-05,
      "loss": 0.04683312773704529,
      "step": 2895
    },
    {
      "epoch": 0.39151999594419895,
      "grad_norm": 0.8237360119819641,
      "learning_rate": 2.197247918508283e-05,
      "loss": 0.04931851103901863,
      "step": 2896
    },
    {
      "epoch": 0.3916551893129642,
      "grad_norm": 0.5000747442245483,
      "learning_rate": 2.1966418409834374e-05,
      "loss": 0.046158745884895325,
      "step": 2897
    },
    {
      "epoch": 0.39179038268172944,
      "grad_norm": 0.4377847909927368,
      "learning_rate": 2.1960356184140453e-05,
      "loss": 0.0648607611656189,
      "step": 2898
    },
    {
      "epoch": 0.3919255760504947,
      "grad_norm": 1.9298532009124756,
      "learning_rate": 2.1954292509263258e-05,
      "loss": 0.06234771013259888,
      "step": 2899
    },
    {
      "epoch": 0.39206076941926,
      "grad_norm": 0.48706138134002686,
      "learning_rate": 2.194822738646528e-05,
      "loss": 0.06517994403839111,
      "step": 2900
    },
    {
      "epoch": 0.39219596278802527,
      "grad_norm": 0.3350776731967926,
      "learning_rate": 2.1942160817009304e-05,
      "loss": 0.05278658866882324,
      "step": 2901
    },
    {
      "epoch": 0.3923311561567905,
      "grad_norm": 0.8154242634773254,
      "learning_rate": 2.193609280215843e-05,
      "loss": 0.03802967816591263,
      "step": 2902
    },
    {
      "epoch": 0.39246634952555576,
      "grad_norm": 0.45581865310668945,
      "learning_rate": 2.1930023343176044e-05,
      "loss": 0.05451982468366623,
      "step": 2903
    },
    {
      "epoch": 0.39260154289432103,
      "grad_norm": 0.533525288105011,
      "learning_rate": 2.1923952441325837e-05,
      "loss": 0.05818824842572212,
      "step": 2904
    },
    {
      "epoch": 0.3927367362630863,
      "grad_norm": 0.4630977213382721,
      "learning_rate": 2.191788009787182e-05,
      "loss": 0.04634997993707657,
      "step": 2905
    },
    {
      "epoch": 0.3928719296318516,
      "grad_norm": 0.8678225874900818,
      "learning_rate": 2.1911806314078267e-05,
      "loss": 0.08176819980144501,
      "step": 2906
    },
    {
      "epoch": 0.3930071230006168,
      "grad_norm": 0.7259595990180969,
      "learning_rate": 2.1905731091209786e-05,
      "loss": 0.06727480888366699,
      "step": 2907
    },
    {
      "epoch": 0.39314231636938207,
      "grad_norm": 0.3742668926715851,
      "learning_rate": 2.1899654430531262e-05,
      "loss": 0.049308210611343384,
      "step": 2908
    },
    {
      "epoch": 0.39327750973814735,
      "grad_norm": 0.699688732624054,
      "learning_rate": 2.18935763333079e-05,
      "loss": 0.051163263618946075,
      "step": 2909
    },
    {
      "epoch": 0.3934127031069126,
      "grad_norm": 1.3867462873458862,
      "learning_rate": 2.1887496800805175e-05,
      "loss": 0.06994909048080444,
      "step": 2910
    },
    {
      "epoch": 0.3935478964756779,
      "grad_norm": 0.29912394285202026,
      "learning_rate": 2.188141583428889e-05,
      "loss": 0.050768643617630005,
      "step": 2911
    },
    {
      "epoch": 0.3936830898444431,
      "grad_norm": 0.4691031873226166,
      "learning_rate": 2.1875333435025138e-05,
      "loss": 0.04402311146259308,
      "step": 2912
    },
    {
      "epoch": 0.3938182832132084,
      "grad_norm": 0.3649757504463196,
      "learning_rate": 2.1869249604280296e-05,
      "loss": 0.06400299072265625,
      "step": 2913
    },
    {
      "epoch": 0.39395347658197366,
      "grad_norm": 0.43007636070251465,
      "learning_rate": 2.1863164343321057e-05,
      "loss": 0.04669588804244995,
      "step": 2914
    },
    {
      "epoch": 0.39408866995073893,
      "grad_norm": 0.6839202642440796,
      "learning_rate": 2.1857077653414397e-05,
      "loss": 0.07176864147186279,
      "step": 2915
    },
    {
      "epoch": 0.39422386331950415,
      "grad_norm": 1.8159350156784058,
      "learning_rate": 2.185098953582761e-05,
      "loss": 0.07163586467504501,
      "step": 2916
    },
    {
      "epoch": 0.3943590566882694,
      "grad_norm": 0.7029412984848022,
      "learning_rate": 2.1844899991828265e-05,
      "loss": 0.05822014808654785,
      "step": 2917
    },
    {
      "epoch": 0.3944942500570347,
      "grad_norm": 0.7634820342063904,
      "learning_rate": 2.1838809022684247e-05,
      "loss": 0.04471708834171295,
      "step": 2918
    },
    {
      "epoch": 0.3946294434258,
      "grad_norm": 0.8820807933807373,
      "learning_rate": 2.1832716629663712e-05,
      "loss": 0.052506059408187866,
      "step": 2919
    },
    {
      "epoch": 0.39476463679456525,
      "grad_norm": 0.4695444107055664,
      "learning_rate": 2.1826622814035138e-05,
      "loss": 0.050924718379974365,
      "step": 2920
    },
    {
      "epoch": 0.39489983016333047,
      "grad_norm": 0.5966514945030212,
      "learning_rate": 2.1820527577067293e-05,
      "loss": 0.06248673424124718,
      "step": 2921
    },
    {
      "epoch": 0.39503502353209574,
      "grad_norm": 0.2945012152194977,
      "learning_rate": 2.1814430920029238e-05,
      "loss": 0.04624895006418228,
      "step": 2922
    },
    {
      "epoch": 0.395170216900861,
      "grad_norm": 0.681483805179596,
      "learning_rate": 2.1808332844190325e-05,
      "loss": 0.07906293869018555,
      "step": 2923
    },
    {
      "epoch": 0.3953054102696263,
      "grad_norm": 0.32935208082199097,
      "learning_rate": 2.1802233350820203e-05,
      "loss": 0.04863521456718445,
      "step": 2924
    },
    {
      "epoch": 0.39544060363839156,
      "grad_norm": 0.662350058555603,
      "learning_rate": 2.179613244118883e-05,
      "loss": 0.058676302433013916,
      "step": 2925
    },
    {
      "epoch": 0.3955757970071568,
      "grad_norm": 0.4267386198043823,
      "learning_rate": 2.1790030116566436e-05,
      "loss": 0.07477055490016937,
      "step": 2926
    },
    {
      "epoch": 0.39571099037592206,
      "grad_norm": 0.48754751682281494,
      "learning_rate": 2.1783926378223563e-05,
      "loss": 0.06251576542854309,
      "step": 2927
    },
    {
      "epoch": 0.39584618374468733,
      "grad_norm": 0.9842320680618286,
      "learning_rate": 2.1777821227431048e-05,
      "loss": 0.0857323408126831,
      "step": 2928
    },
    {
      "epoch": 0.3959813771134526,
      "grad_norm": 0.7058376669883728,
      "learning_rate": 2.1771714665460005e-05,
      "loss": 0.04794486612081528,
      "step": 2929
    },
    {
      "epoch": 0.3961165704822178,
      "grad_norm": 0.7239179015159607,
      "learning_rate": 2.1765606693581857e-05,
      "loss": 0.06123724579811096,
      "step": 2930
    },
    {
      "epoch": 0.3962517638509831,
      "grad_norm": 0.5751566290855408,
      "learning_rate": 2.1759497313068316e-05,
      "loss": 0.06578932702541351,
      "step": 2931
    },
    {
      "epoch": 0.39638695721974837,
      "grad_norm": 0.36836645007133484,
      "learning_rate": 2.175338652519139e-05,
      "loss": 0.054364919662475586,
      "step": 2932
    },
    {
      "epoch": 0.39652215058851364,
      "grad_norm": 0.8371866345405579,
      "learning_rate": 2.1747274331223377e-05,
      "loss": 0.041383594274520874,
      "step": 2933
    },
    {
      "epoch": 0.3966573439572789,
      "grad_norm": 0.8707664608955383,
      "learning_rate": 2.1741160732436865e-05,
      "loss": 0.051559921354055405,
      "step": 2934
    },
    {
      "epoch": 0.39679253732604414,
      "grad_norm": 0.6024278998374939,
      "learning_rate": 2.1735045730104746e-05,
      "loss": 0.049465976655483246,
      "step": 2935
    },
    {
      "epoch": 0.3969277306948094,
      "grad_norm": 0.39770883321762085,
      "learning_rate": 2.1728929325500183e-05,
      "loss": 0.04017620533704758,
      "step": 2936
    },
    {
      "epoch": 0.3970629240635747,
      "grad_norm": 0.33352670073509216,
      "learning_rate": 2.1722811519896654e-05,
      "loss": 0.04881981760263443,
      "step": 2937
    },
    {
      "epoch": 0.39719811743233996,
      "grad_norm": 1.5417475700378418,
      "learning_rate": 2.171669231456792e-05,
      "loss": 0.07164902985095978,
      "step": 2938
    },
    {
      "epoch": 0.39733331080110523,
      "grad_norm": 1.3256629705429077,
      "learning_rate": 2.1710571710788025e-05,
      "loss": 0.06817503273487091,
      "step": 2939
    },
    {
      "epoch": 0.39746850416987045,
      "grad_norm": 0.73344886302948,
      "learning_rate": 2.1704449709831312e-05,
      "loss": 0.061049096286296844,
      "step": 2940
    },
    {
      "epoch": 0.3976036975386357,
      "grad_norm": 1.999131679534912,
      "learning_rate": 2.1698326312972423e-05,
      "loss": 0.05661129951477051,
      "step": 2941
    },
    {
      "epoch": 0.397738890907401,
      "grad_norm": 0.4522429406642914,
      "learning_rate": 2.1692201521486268e-05,
      "loss": 0.04034745693206787,
      "step": 2942
    },
    {
      "epoch": 0.3978740842761663,
      "grad_norm": 1.2400321960449219,
      "learning_rate": 2.1686075336648075e-05,
      "loss": 0.05283085256814957,
      "step": 2943
    },
    {
      "epoch": 0.3980092776449315,
      "grad_norm": 0.31111636757850647,
      "learning_rate": 2.167994775973334e-05,
      "loss": 0.04952539503574371,
      "step": 2944
    },
    {
      "epoch": 0.39814447101369677,
      "grad_norm": 1.8355605602264404,
      "learning_rate": 2.167381879201786e-05,
      "loss": 0.06255283951759338,
      "step": 2945
    },
    {
      "epoch": 0.39827966438246204,
      "grad_norm": 0.5530511736869812,
      "learning_rate": 2.166768843477772e-05,
      "loss": 0.05090506374835968,
      "step": 2946
    },
    {
      "epoch": 0.3984148577512273,
      "grad_norm": 0.7286508679389954,
      "learning_rate": 2.166155668928929e-05,
      "loss": 0.06268691271543503,
      "step": 2947
    },
    {
      "epoch": 0.3985500511199926,
      "grad_norm": 0.24765899777412415,
      "learning_rate": 2.1655423556829233e-05,
      "loss": 0.050964921712875366,
      "step": 2948
    },
    {
      "epoch": 0.3986852444887578,
      "grad_norm": 0.25484296679496765,
      "learning_rate": 2.1649289038674504e-05,
      "loss": 0.050675757229328156,
      "step": 2949
    },
    {
      "epoch": 0.3988204378575231,
      "grad_norm": 1.1110823154449463,
      "learning_rate": 2.1643153136102333e-05,
      "loss": 0.0936761274933815,
      "step": 2950
    },
    {
      "epoch": 0.39895563122628835,
      "grad_norm": 2.2881293296813965,
      "learning_rate": 2.1637015850390255e-05,
      "loss": 0.0666525661945343,
      "step": 2951
    },
    {
      "epoch": 0.39909082459505363,
      "grad_norm": 1.5375266075134277,
      "learning_rate": 2.1630877182816087e-05,
      "loss": 0.05764663964509964,
      "step": 2952
    },
    {
      "epoch": 0.3992260179638189,
      "grad_norm": 0.5055779218673706,
      "learning_rate": 2.162473713465793e-05,
      "loss": 0.050413694232702255,
      "step": 2953
    },
    {
      "epoch": 0.3993612113325841,
      "grad_norm": 0.6004561185836792,
      "learning_rate": 2.161859570719417e-05,
      "loss": 0.04441293329000473,
      "step": 2954
    },
    {
      "epoch": 0.3994964047013494,
      "grad_norm": 0.9383441209793091,
      "learning_rate": 2.161245290170349e-05,
      "loss": 0.05407094955444336,
      "step": 2955
    },
    {
      "epoch": 0.39963159807011467,
      "grad_norm": 1.520416021347046,
      "learning_rate": 2.1606308719464858e-05,
      "loss": 0.05162256956100464,
      "step": 2956
    },
    {
      "epoch": 0.39976679143887994,
      "grad_norm": 0.6215782761573792,
      "learning_rate": 2.160016316175752e-05,
      "loss": 0.06491442024707794,
      "step": 2957
    },
    {
      "epoch": 0.39990198480764516,
      "grad_norm": 0.4168660044670105,
      "learning_rate": 2.159401622986101e-05,
      "loss": 0.039890021085739136,
      "step": 2958
    },
    {
      "epoch": 0.40003717817641044,
      "grad_norm": 1.2815821170806885,
      "learning_rate": 2.1587867925055165e-05,
      "loss": 0.054726384580135345,
      "step": 2959
    },
    {
      "epoch": 0.4001723715451757,
      "grad_norm": 0.6798976063728333,
      "learning_rate": 2.158171824862008e-05,
      "loss": 0.04645705223083496,
      "step": 2960
    },
    {
      "epoch": 0.400307564913941,
      "grad_norm": 1.2501940727233887,
      "learning_rate": 2.157556720183616e-05,
      "loss": 0.04481580853462219,
      "step": 2961
    },
    {
      "epoch": 0.40044275828270626,
      "grad_norm": 0.9489818215370178,
      "learning_rate": 2.156941478598409e-05,
      "loss": 0.08229565620422363,
      "step": 2962
    },
    {
      "epoch": 0.4005779516514715,
      "grad_norm": 0.469913512468338,
      "learning_rate": 2.156326100234482e-05,
      "loss": 0.05371943116188049,
      "step": 2963
    },
    {
      "epoch": 0.40071314502023675,
      "grad_norm": 0.4225693941116333,
      "learning_rate": 2.1557105852199612e-05,
      "loss": 0.03603549301624298,
      "step": 2964
    },
    {
      "epoch": 0.400848338389002,
      "grad_norm": 0.4249083697795868,
      "learning_rate": 2.155094933683e-05,
      "loss": 0.05573997646570206,
      "step": 2965
    },
    {
      "epoch": 0.4009835317577673,
      "grad_norm": 1.0302414894104004,
      "learning_rate": 2.1544791457517802e-05,
      "loss": 0.07400310784578323,
      "step": 2966
    },
    {
      "epoch": 0.4011187251265325,
      "grad_norm": 0.5317590236663818,
      "learning_rate": 2.1538632215545126e-05,
      "loss": 0.04713648557662964,
      "step": 2967
    },
    {
      "epoch": 0.4012539184952978,
      "grad_norm": 0.7051037549972534,
      "learning_rate": 2.153247161219435e-05,
      "loss": 0.04982832074165344,
      "step": 2968
    },
    {
      "epoch": 0.40138911186406306,
      "grad_norm": 1.0028332471847534,
      "learning_rate": 2.1526309648748147e-05,
      "loss": 0.07157133519649506,
      "step": 2969
    },
    {
      "epoch": 0.40152430523282834,
      "grad_norm": 0.7827900052070618,
      "learning_rate": 2.1520146326489476e-05,
      "loss": 0.06770962476730347,
      "step": 2970
    },
    {
      "epoch": 0.4016594986015936,
      "grad_norm": 0.4194885194301605,
      "learning_rate": 2.151398164670157e-05,
      "loss": 0.06479065865278244,
      "step": 2971
    },
    {
      "epoch": 0.40179469197035883,
      "grad_norm": 0.6523520350456238,
      "learning_rate": 2.1507815610667948e-05,
      "loss": 0.06123092770576477,
      "step": 2972
    },
    {
      "epoch": 0.4019298853391241,
      "grad_norm": 0.4864334762096405,
      "learning_rate": 2.1501648219672407e-05,
      "loss": 0.0620855987071991,
      "step": 2973
    },
    {
      "epoch": 0.4020650787078894,
      "grad_norm": 1.055147409439087,
      "learning_rate": 2.149547947499904e-05,
      "loss": 0.06187228113412857,
      "step": 2974
    },
    {
      "epoch": 0.40220027207665465,
      "grad_norm": 0.6128971576690674,
      "learning_rate": 2.1489309377932212e-05,
      "loss": 0.05860280245542526,
      "step": 2975
    },
    {
      "epoch": 0.4023354654454199,
      "grad_norm": 0.544592022895813,
      "learning_rate": 2.1483137929756562e-05,
      "loss": 0.03803078830242157,
      "step": 2976
    },
    {
      "epoch": 0.40247065881418514,
      "grad_norm": 1.7454712390899658,
      "learning_rate": 2.147696513175702e-05,
      "loss": 0.06334815919399261,
      "step": 2977
    },
    {
      "epoch": 0.4026058521829504,
      "grad_norm": 0.2908405065536499,
      "learning_rate": 2.1470790985218804e-05,
      "loss": 0.047992438077926636,
      "step": 2978
    },
    {
      "epoch": 0.4027410455517157,
      "grad_norm": 0.2873890697956085,
      "learning_rate": 2.1464615491427393e-05,
      "loss": 0.042032837867736816,
      "step": 2979
    },
    {
      "epoch": 0.40287623892048097,
      "grad_norm": 0.26165837049484253,
      "learning_rate": 2.1458438651668567e-05,
      "loss": 0.04348848760128021,
      "step": 2980
    },
    {
      "epoch": 0.4030114322892462,
      "grad_norm": 0.549848198890686,
      "learning_rate": 2.1452260467228376e-05,
      "loss": 0.06827035546302795,
      "step": 2981
    },
    {
      "epoch": 0.40314662565801146,
      "grad_norm": 0.2638908624649048,
      "learning_rate": 2.144608093939314e-05,
      "loss": 0.06703531742095947,
      "step": 2982
    },
    {
      "epoch": 0.40328181902677673,
      "grad_norm": 0.34968826174736023,
      "learning_rate": 2.1439900069449483e-05,
      "loss": 0.042341575026512146,
      "step": 2983
    },
    {
      "epoch": 0.403417012395542,
      "grad_norm": 0.4925195872783661,
      "learning_rate": 2.1433717858684286e-05,
      "loss": 0.04880249500274658,
      "step": 2984
    },
    {
      "epoch": 0.4035522057643073,
      "grad_norm": 0.4777568280696869,
      "learning_rate": 2.1427534308384724e-05,
      "loss": 0.061606962233781815,
      "step": 2985
    },
    {
      "epoch": 0.4036873991330725,
      "grad_norm": 0.45032647252082825,
      "learning_rate": 2.1421349419838245e-05,
      "loss": 0.069297194480896,
      "step": 2986
    },
    {
      "epoch": 0.4038225925018378,
      "grad_norm": 0.6713958382606506,
      "learning_rate": 2.1415163194332574e-05,
      "loss": 0.059538375586271286,
      "step": 2987
    },
    {
      "epoch": 0.40395778587060305,
      "grad_norm": 0.37850481271743774,
      "learning_rate": 2.1408975633155715e-05,
      "loss": 0.03341542184352875,
      "step": 2988
    },
    {
      "epoch": 0.4040929792393683,
      "grad_norm": 0.8188773989677429,
      "learning_rate": 2.140278673759595e-05,
      "loss": 0.04894550144672394,
      "step": 2989
    },
    {
      "epoch": 0.4042281726081336,
      "grad_norm": 0.5429801344871521,
      "learning_rate": 2.1396596508941847e-05,
      "loss": 0.05429689586162567,
      "step": 2990
    },
    {
      "epoch": 0.4043633659768988,
      "grad_norm": 0.42324841022491455,
      "learning_rate": 2.1390404948482238e-05,
      "loss": 0.054476361721754074,
      "step": 2991
    },
    {
      "epoch": 0.4044985593456641,
      "grad_norm": 0.3851200342178345,
      "learning_rate": 2.1384212057506243e-05,
      "loss": 0.05033630132675171,
      "step": 2992
    },
    {
      "epoch": 0.40463375271442936,
      "grad_norm": 0.35794851183891296,
      "learning_rate": 2.137801783730325e-05,
      "loss": 0.04697723686695099,
      "step": 2993
    },
    {
      "epoch": 0.40476894608319464,
      "grad_norm": 0.4880930483341217,
      "learning_rate": 2.137182228916293e-05,
      "loss": 0.030466772615909576,
      "step": 2994
    },
    {
      "epoch": 0.40490413945195985,
      "grad_norm": 0.24870483577251434,
      "learning_rate": 2.136562541437523e-05,
      "loss": 0.04872870445251465,
      "step": 2995
    },
    {
      "epoch": 0.40503933282072513,
      "grad_norm": 0.3435058891773224,
      "learning_rate": 2.135942721423038e-05,
      "loss": 0.04835168272256851,
      "step": 2996
    },
    {
      "epoch": 0.4051745261894904,
      "grad_norm": 0.9559407234191895,
      "learning_rate": 2.1353227690018865e-05,
      "loss": 0.05477004870772362,
      "step": 2997
    },
    {
      "epoch": 0.4053097195582557,
      "grad_norm": 0.6487329006195068,
      "learning_rate": 2.1347026843031467e-05,
      "loss": 0.07681387662887573,
      "step": 2998
    },
    {
      "epoch": 0.40544491292702095,
      "grad_norm": 0.9528166055679321,
      "learning_rate": 2.1340824674559238e-05,
      "loss": 0.059322237968444824,
      "step": 2999
    },
    {
      "epoch": 0.40558010629578617,
      "grad_norm": 0.4483122229576111,
      "learning_rate": 2.133462118589349e-05,
      "loss": 0.05703192576766014,
      "step": 3000
    },
    {
      "epoch": 0.40571529966455144,
      "grad_norm": 0.6086049675941467,
      "learning_rate": 2.1328416378325837e-05,
      "loss": 0.05557991564273834,
      "step": 3001
    },
    {
      "epoch": 0.4058504930333167,
      "grad_norm": 0.39140862226486206,
      "learning_rate": 2.1322210253148144e-05,
      "loss": 0.0404512844979763,
      "step": 3002
    },
    {
      "epoch": 0.405985686402082,
      "grad_norm": 0.3951510787010193,
      "learning_rate": 2.131600281165257e-05,
      "loss": 0.05150478705763817,
      "step": 3003
    },
    {
      "epoch": 0.40612087977084727,
      "grad_norm": 0.48007631301879883,
      "learning_rate": 2.130979405513152e-05,
      "loss": 0.056524619460105896,
      "step": 3004
    },
    {
      "epoch": 0.4062560731396125,
      "grad_norm": 0.39255964756011963,
      "learning_rate": 2.1303583984877697e-05,
      "loss": 0.060637444257736206,
      "step": 3005
    },
    {
      "epoch": 0.40639126650837776,
      "grad_norm": 0.49237433075904846,
      "learning_rate": 2.1297372602184085e-05,
      "loss": 0.057009633630514145,
      "step": 3006
    },
    {
      "epoch": 0.40652645987714303,
      "grad_norm": 1.7834043502807617,
      "learning_rate": 2.1291159908343907e-05,
      "loss": 0.051647309213876724,
      "step": 3007
    },
    {
      "epoch": 0.4066616532459083,
      "grad_norm": 1.6294023990631104,
      "learning_rate": 2.1284945904650693e-05,
      "loss": 0.0586942583322525,
      "step": 3008
    },
    {
      "epoch": 0.4067968466146735,
      "grad_norm": 0.6993623971939087,
      "learning_rate": 2.127873059239822e-05,
      "loss": 0.07575511932373047,
      "step": 3009
    },
    {
      "epoch": 0.4069320399834388,
      "grad_norm": 0.3398081362247467,
      "learning_rate": 2.127251397288056e-05,
      "loss": 0.05911538004875183,
      "step": 3010
    },
    {
      "epoch": 0.40706723335220407,
      "grad_norm": 0.9419555068016052,
      "learning_rate": 2.126629604739204e-05,
      "loss": 0.06722565740346909,
      "step": 3011
    },
    {
      "epoch": 0.40720242672096935,
      "grad_norm": 0.7504845261573792,
      "learning_rate": 2.1260076817227268e-05,
      "loss": 0.048842914402484894,
      "step": 3012
    },
    {
      "epoch": 0.4073376200897346,
      "grad_norm": 0.9056908488273621,
      "learning_rate": 2.1253856283681122e-05,
      "loss": 0.05707564949989319,
      "step": 3013
    },
    {
      "epoch": 0.40747281345849984,
      "grad_norm": 0.3084135949611664,
      "learning_rate": 2.1247634448048743e-05,
      "loss": 0.05317723751068115,
      "step": 3014
    },
    {
      "epoch": 0.4076080068272651,
      "grad_norm": 0.7882542014122009,
      "learning_rate": 2.1241411311625562e-05,
      "loss": 0.05224326625466347,
      "step": 3015
    },
    {
      "epoch": 0.4077432001960304,
      "grad_norm": 0.3396093547344208,
      "learning_rate": 2.1235186875707257e-05,
      "loss": 0.06457221508026123,
      "step": 3016
    },
    {
      "epoch": 0.40787839356479566,
      "grad_norm": 0.35858070850372314,
      "learning_rate": 2.1228961141589797e-05,
      "loss": 0.050357699394226074,
      "step": 3017
    },
    {
      "epoch": 0.40801358693356093,
      "grad_norm": 0.35762953758239746,
      "learning_rate": 2.122273411056941e-05,
      "loss": 0.07766478508710861,
      "step": 3018
    },
    {
      "epoch": 0.40814878030232615,
      "grad_norm": 1.071114182472229,
      "learning_rate": 2.1216505783942592e-05,
      "loss": 0.06400462985038757,
      "step": 3019
    },
    {
      "epoch": 0.4082839736710914,
      "grad_norm": 1.0491416454315186,
      "learning_rate": 2.121027616300613e-05,
      "loss": 0.05831335112452507,
      "step": 3020
    },
    {
      "epoch": 0.4084191670398567,
      "grad_norm": 0.5437660217285156,
      "learning_rate": 2.1204045249057043e-05,
      "loss": 0.06762094795703888,
      "step": 3021
    },
    {
      "epoch": 0.408554360408622,
      "grad_norm": 1.0614601373672485,
      "learning_rate": 2.119781304339266e-05,
      "loss": 0.05835258960723877,
      "step": 3022
    },
    {
      "epoch": 0.4086895537773872,
      "grad_norm": 0.7614403367042542,
      "learning_rate": 2.1191579547310547e-05,
      "loss": 0.04494254291057587,
      "step": 3023
    },
    {
      "epoch": 0.40882474714615247,
      "grad_norm": 0.3509546220302582,
      "learning_rate": 2.1185344762108556e-05,
      "loss": 0.06634199619293213,
      "step": 3024
    },
    {
      "epoch": 0.40895994051491774,
      "grad_norm": 0.24866631627082825,
      "learning_rate": 2.11791086890848e-05,
      "loss": 0.03005295991897583,
      "step": 3025
    },
    {
      "epoch": 0.409095133883683,
      "grad_norm": 1.4082615375518799,
      "learning_rate": 2.1172871329537662e-05,
      "loss": 0.056225575506687164,
      "step": 3026
    },
    {
      "epoch": 0.4092303272524483,
      "grad_norm": 0.5511521100997925,
      "learning_rate": 2.1166632684765794e-05,
      "loss": 0.055938757956027985,
      "step": 3027
    },
    {
      "epoch": 0.4093655206212135,
      "grad_norm": 0.7599920034408569,
      "learning_rate": 2.1160392756068124e-05,
      "loss": 0.05547342449426651,
      "step": 3028
    },
    {
      "epoch": 0.4095007139899788,
      "grad_norm": 0.8268606662750244,
      "learning_rate": 2.1154151544743826e-05,
      "loss": 0.05882567539811134,
      "step": 3029
    },
    {
      "epoch": 0.40963590735874406,
      "grad_norm": 0.23169812560081482,
      "learning_rate": 2.114790905209236e-05,
      "loss": 0.04227090999484062,
      "step": 3030
    },
    {
      "epoch": 0.40977110072750933,
      "grad_norm": 0.6046125888824463,
      "learning_rate": 2.1141665279413444e-05,
      "loss": 0.05067253112792969,
      "step": 3031
    },
    {
      "epoch": 0.4099062940962746,
      "grad_norm": 0.6065956950187683,
      "learning_rate": 2.1135420228007062e-05,
      "loss": 0.052970755845308304,
      "step": 3032
    },
    {
      "epoch": 0.4100414874650398,
      "grad_norm": 0.35070839524269104,
      "learning_rate": 2.1129173899173474e-05,
      "loss": 0.054492343217134476,
      "step": 3033
    },
    {
      "epoch": 0.4101766808338051,
      "grad_norm": 0.6365925073623657,
      "learning_rate": 2.11229262942132e-05,
      "loss": 0.05697426199913025,
      "step": 3034
    },
    {
      "epoch": 0.41031187420257037,
      "grad_norm": 1.2109260559082031,
      "learning_rate": 2.1116677414427008e-05,
      "loss": 0.06884992122650146,
      "step": 3035
    },
    {
      "epoch": 0.41044706757133564,
      "grad_norm": 0.396894633769989,
      "learning_rate": 2.1110427261115972e-05,
      "loss": 0.0466761589050293,
      "step": 3036
    },
    {
      "epoch": 0.41058226094010086,
      "grad_norm": 1.1930204629898071,
      "learning_rate": 2.1104175835581386e-05,
      "loss": 0.07297670841217041,
      "step": 3037
    },
    {
      "epoch": 0.41071745430886614,
      "grad_norm": 0.46178895235061646,
      "learning_rate": 2.1097923139124846e-05,
      "loss": 0.05735020712018013,
      "step": 3038
    },
    {
      "epoch": 0.4108526476776314,
      "grad_norm": 0.7437040209770203,
      "learning_rate": 2.109166917304819e-05,
      "loss": 0.04401330649852753,
      "step": 3039
    },
    {
      "epoch": 0.4109878410463967,
      "grad_norm": 2.015864849090576,
      "learning_rate": 2.1085413938653532e-05,
      "loss": 0.07502947747707367,
      "step": 3040
    },
    {
      "epoch": 0.41112303441516196,
      "grad_norm": 1.0889643430709839,
      "learning_rate": 2.107915743724323e-05,
      "loss": 0.04603596031665802,
      "step": 3041
    },
    {
      "epoch": 0.4112582277839272,
      "grad_norm": 1.5796411037445068,
      "learning_rate": 2.1072899670119935e-05,
      "loss": 0.04325581341981888,
      "step": 3042
    },
    {
      "epoch": 0.41139342115269245,
      "grad_norm": 0.6258715987205505,
      "learning_rate": 2.1066640638586543e-05,
      "loss": 0.06860720366239548,
      "step": 3043
    },
    {
      "epoch": 0.4115286145214577,
      "grad_norm": 0.32061874866485596,
      "learning_rate": 2.1060380343946223e-05,
      "loss": 0.06643766164779663,
      "step": 3044
    },
    {
      "epoch": 0.411663807890223,
      "grad_norm": 0.42691174149513245,
      "learning_rate": 2.10541187875024e-05,
      "loss": 0.05105191469192505,
      "step": 3045
    },
    {
      "epoch": 0.4117990012589883,
      "grad_norm": 1.1686564683914185,
      "learning_rate": 2.1047855970558753e-05,
      "loss": 0.06568668782711029,
      "step": 3046
    },
    {
      "epoch": 0.4119341946277535,
      "grad_norm": 0.44537875056266785,
      "learning_rate": 2.1041591894419244e-05,
      "loss": 0.05183150991797447,
      "step": 3047
    },
    {
      "epoch": 0.41206938799651877,
      "grad_norm": 0.47970667481422424,
      "learning_rate": 2.1035326560388087e-05,
      "loss": 0.0559847354888916,
      "step": 3048
    },
    {
      "epoch": 0.41220458136528404,
      "grad_norm": 0.3690253794193268,
      "learning_rate": 2.1029059969769756e-05,
      "loss": 0.044215962290763855,
      "step": 3049
    },
    {
      "epoch": 0.4123397747340493,
      "grad_norm": 0.47140994668006897,
      "learning_rate": 2.1022792123868986e-05,
      "loss": 0.04593484103679657,
      "step": 3050
    },
    {
      "epoch": 0.41247496810281453,
      "grad_norm": 1.8404030799865723,
      "learning_rate": 2.1016523023990783e-05,
      "loss": 0.05107954517006874,
      "step": 3051
    },
    {
      "epoch": 0.4126101614715798,
      "grad_norm": 0.5912984013557434,
      "learning_rate": 2.1010252671440398e-05,
      "loss": 0.04349742829799652,
      "step": 3052
    },
    {
      "epoch": 0.4127453548403451,
      "grad_norm": 0.6136721968650818,
      "learning_rate": 2.1003981067523358e-05,
      "loss": 0.06135822832584381,
      "step": 3053
    },
    {
      "epoch": 0.41288054820911035,
      "grad_norm": 0.936673641204834,
      "learning_rate": 2.099770821354544e-05,
      "loss": 0.05388724058866501,
      "step": 3054
    },
    {
      "epoch": 0.41301574157787563,
      "grad_norm": 0.3524787425994873,
      "learning_rate": 2.0991434110812692e-05,
      "loss": 0.06820368766784668,
      "step": 3055
    },
    {
      "epoch": 0.41315093494664085,
      "grad_norm": 0.29848381876945496,
      "learning_rate": 2.0985158760631415e-05,
      "loss": 0.05465012788772583,
      "step": 3056
    },
    {
      "epoch": 0.4132861283154061,
      "grad_norm": 0.4333903193473816,
      "learning_rate": 2.0978882164308157e-05,
      "loss": 0.05514828860759735,
      "step": 3057
    },
    {
      "epoch": 0.4134213216841714,
      "grad_norm": 0.49294960498809814,
      "learning_rate": 2.0972604323149755e-05,
      "loss": 0.052114322781562805,
      "step": 3058
    },
    {
      "epoch": 0.41355651505293667,
      "grad_norm": 0.3174901604652405,
      "learning_rate": 2.0966325238463283e-05,
      "loss": 0.04909542202949524,
      "step": 3059
    },
    {
      "epoch": 0.41369170842170194,
      "grad_norm": 0.814630925655365,
      "learning_rate": 2.096004491155608e-05,
      "loss": 0.06041857972741127,
      "step": 3060
    },
    {
      "epoch": 0.41382690179046716,
      "grad_norm": 0.7371432781219482,
      "learning_rate": 2.0953763343735746e-05,
      "loss": 0.05977824330329895,
      "step": 3061
    },
    {
      "epoch": 0.41396209515923243,
      "grad_norm": 0.521529495716095,
      "learning_rate": 2.0947480536310133e-05,
      "loss": 0.06525170803070068,
      "step": 3062
    },
    {
      "epoch": 0.4140972885279977,
      "grad_norm": 0.4722524881362915,
      "learning_rate": 2.0941196490587352e-05,
      "loss": 0.05617760121822357,
      "step": 3063
    },
    {
      "epoch": 0.414232481896763,
      "grad_norm": 0.3377145826816559,
      "learning_rate": 2.0934911207875782e-05,
      "loss": 0.0448685958981514,
      "step": 3064
    },
    {
      "epoch": 0.4143676752655282,
      "grad_norm": 0.5349330306053162,
      "learning_rate": 2.092862468948405e-05,
      "loss": 0.04958082735538483,
      "step": 3065
    },
    {
      "epoch": 0.4145028686342935,
      "grad_norm": 1.1507644653320312,
      "learning_rate": 2.0922336936721044e-05,
      "loss": 0.06242275983095169,
      "step": 3066
    },
    {
      "epoch": 0.41463806200305875,
      "grad_norm": 0.3838142454624176,
      "learning_rate": 2.0916047950895907e-05,
      "loss": 0.05453038215637207,
      "step": 3067
    },
    {
      "epoch": 0.414773255371824,
      "grad_norm": 0.3850749433040619,
      "learning_rate": 2.0909757733318035e-05,
      "loss": 0.06268344819545746,
      "step": 3068
    },
    {
      "epoch": 0.4149084487405893,
      "grad_norm": 0.4310440719127655,
      "learning_rate": 2.090346628529709e-05,
      "loss": 0.05703897774219513,
      "step": 3069
    },
    {
      "epoch": 0.4150436421093545,
      "grad_norm": 0.41503193974494934,
      "learning_rate": 2.089717360814298e-05,
      "loss": 0.03965657949447632,
      "step": 3070
    },
    {
      "epoch": 0.4151788354781198,
      "grad_norm": 0.5820544362068176,
      "learning_rate": 2.0890879703165885e-05,
      "loss": 0.04829868674278259,
      "step": 3071
    },
    {
      "epoch": 0.41531402884688506,
      "grad_norm": 0.5983740091323853,
      "learning_rate": 2.0884584571676217e-05,
      "loss": 0.061264362186193466,
      "step": 3072
    },
    {
      "epoch": 0.41544922221565034,
      "grad_norm": 0.8227573037147522,
      "learning_rate": 2.0878288214984657e-05,
      "loss": 0.06077104061841965,
      "step": 3073
    },
    {
      "epoch": 0.4155844155844156,
      "grad_norm": 0.8223791122436523,
      "learning_rate": 2.0871990634402147e-05,
      "loss": 0.05666433274745941,
      "step": 3074
    },
    {
      "epoch": 0.41571960895318083,
      "grad_norm": 0.6710504293441772,
      "learning_rate": 2.0865691831239877e-05,
      "loss": 0.07451266050338745,
      "step": 3075
    },
    {
      "epoch": 0.4158548023219461,
      "grad_norm": 0.6195180416107178,
      "learning_rate": 2.0859391806809285e-05,
      "loss": 0.07262474298477173,
      "step": 3076
    },
    {
      "epoch": 0.4159899956907114,
      "grad_norm": 0.4847974181175232,
      "learning_rate": 2.0853090562422072e-05,
      "loss": 0.058167219161987305,
      "step": 3077
    },
    {
      "epoch": 0.41612518905947665,
      "grad_norm": 1.5001903772354126,
      "learning_rate": 2.084678809939019e-05,
      "loss": 0.05016686022281647,
      "step": 3078
    },
    {
      "epoch": 0.41626038242824187,
      "grad_norm": 2.3016607761383057,
      "learning_rate": 2.084048441902585e-05,
      "loss": 0.069181427359581,
      "step": 3079
    },
    {
      "epoch": 0.41639557579700714,
      "grad_norm": 1.4211180210113525,
      "learning_rate": 2.0834179522641508e-05,
      "loss": 0.06011167913675308,
      "step": 3080
    },
    {
      "epoch": 0.4165307691657724,
      "grad_norm": 0.2641412615776062,
      "learning_rate": 2.0827873411549877e-05,
      "loss": 0.06048768758773804,
      "step": 3081
    },
    {
      "epoch": 0.4166659625345377,
      "grad_norm": 1.3323535919189453,
      "learning_rate": 2.0821566087063926e-05,
      "loss": 0.07238896191120148,
      "step": 3082
    },
    {
      "epoch": 0.41680115590330297,
      "grad_norm": 0.356070339679718,
      "learning_rate": 2.081525755049687e-05,
      "loss": 0.059626445174217224,
      "step": 3083
    },
    {
      "epoch": 0.4169363492720682,
      "grad_norm": 0.48678216338157654,
      "learning_rate": 2.0808947803162182e-05,
      "loss": 0.06464888155460358,
      "step": 3084
    },
    {
      "epoch": 0.41707154264083346,
      "grad_norm": 0.3271723985671997,
      "learning_rate": 2.0802636846373578e-05,
      "loss": 0.04913259297609329,
      "step": 3085
    },
    {
      "epoch": 0.41720673600959873,
      "grad_norm": 0.8940332531929016,
      "learning_rate": 2.0796324681445045e-05,
      "loss": 0.054951827973127365,
      "step": 3086
    },
    {
      "epoch": 0.417341929378364,
      "grad_norm": 0.6586296558380127,
      "learning_rate": 2.0790011309690806e-05,
      "loss": 0.04194846749305725,
      "step": 3087
    },
    {
      "epoch": 0.4174771227471292,
      "grad_norm": 0.40827396512031555,
      "learning_rate": 2.0783696732425332e-05,
      "loss": 0.055915653705596924,
      "step": 3088
    },
    {
      "epoch": 0.4176123161158945,
      "grad_norm": 0.5980594754219055,
      "learning_rate": 2.0777380950963355e-05,
      "loss": 0.06726858019828796,
      "step": 3089
    },
    {
      "epoch": 0.4177475094846598,
      "grad_norm": 0.42285361886024475,
      "learning_rate": 2.0771063966619854e-05,
      "loss": 0.026932690292596817,
      "step": 3090
    },
    {
      "epoch": 0.41788270285342505,
      "grad_norm": 0.3072812259197235,
      "learning_rate": 2.0764745780710065e-05,
      "loss": 0.04799140244722366,
      "step": 3091
    },
    {
      "epoch": 0.4180178962221903,
      "grad_norm": 0.7011281251907349,
      "learning_rate": 2.075842639454946e-05,
      "loss": 0.044330716133117676,
      "step": 3092
    },
    {
      "epoch": 0.41815308959095554,
      "grad_norm": 0.9490211009979248,
      "learning_rate": 2.075210580945378e-05,
      "loss": 0.05724972486495972,
      "step": 3093
    },
    {
      "epoch": 0.4182882829597208,
      "grad_norm": 0.5081564784049988,
      "learning_rate": 2.0745784026738984e-05,
      "loss": 0.05207130312919617,
      "step": 3094
    },
    {
      "epoch": 0.4184234763284861,
      "grad_norm": 1.4715677499771118,
      "learning_rate": 2.073946104772132e-05,
      "loss": 0.06876707077026367,
      "step": 3095
    },
    {
      "epoch": 0.41855866969725136,
      "grad_norm": 1.1546006202697754,
      "learning_rate": 2.0733136873717258e-05,
      "loss": 0.0642404556274414,
      "step": 3096
    },
    {
      "epoch": 0.41869386306601664,
      "grad_norm": 0.49467188119888306,
      "learning_rate": 2.0726811506043527e-05,
      "loss": 0.04943768307566643,
      "step": 3097
    },
    {
      "epoch": 0.41882905643478185,
      "grad_norm": 0.4919363260269165,
      "learning_rate": 2.0720484946017104e-05,
      "loss": 0.03366868942975998,
      "step": 3098
    },
    {
      "epoch": 0.41896424980354713,
      "grad_norm": 0.9652361273765564,
      "learning_rate": 2.0714157194955202e-05,
      "loss": 0.07263242453336716,
      "step": 3099
    },
    {
      "epoch": 0.4190994431723124,
      "grad_norm": 0.3756640553474426,
      "learning_rate": 2.070782825417531e-05,
      "loss": 0.045490097254514694,
      "step": 3100
    },
    {
      "epoch": 0.4192346365410777,
      "grad_norm": 0.5862973928451538,
      "learning_rate": 2.0701498124995127e-05,
      "loss": 0.0809047669172287,
      "step": 3101
    },
    {
      "epoch": 0.4193698299098429,
      "grad_norm": 0.4543468654155731,
      "learning_rate": 2.069516680873264e-05,
      "loss": 0.0488109290599823,
      "step": 3102
    },
    {
      "epoch": 0.41950502327860817,
      "grad_norm": 0.47270965576171875,
      "learning_rate": 2.0688834306706047e-05,
      "loss": 0.06073812395334244,
      "step": 3103
    },
    {
      "epoch": 0.41964021664737344,
      "grad_norm": 1.1174681186676025,
      "learning_rate": 2.0682500620233815e-05,
      "loss": 0.05444493889808655,
      "step": 3104
    },
    {
      "epoch": 0.4197754100161387,
      "grad_norm": 0.4161030054092407,
      "learning_rate": 2.0676165750634656e-05,
      "loss": 0.061250392347574234,
      "step": 3105
    },
    {
      "epoch": 0.419910603384904,
      "grad_norm": 0.8984009027481079,
      "learning_rate": 2.0669829699227513e-05,
      "loss": 0.037275247275829315,
      "step": 3106
    },
    {
      "epoch": 0.4200457967536692,
      "grad_norm": 0.8720649480819702,
      "learning_rate": 2.06634924673316e-05,
      "loss": 0.05299574136734009,
      "step": 3107
    },
    {
      "epoch": 0.4201809901224345,
      "grad_norm": 0.44392624497413635,
      "learning_rate": 2.0657154056266346e-05,
      "loss": 0.04778685420751572,
      "step": 3108
    },
    {
      "epoch": 0.42031618349119976,
      "grad_norm": 0.4944213628768921,
      "learning_rate": 2.0650814467351452e-05,
      "loss": 0.05455660820007324,
      "step": 3109
    },
    {
      "epoch": 0.42045137685996503,
      "grad_norm": 1.4108939170837402,
      "learning_rate": 2.064447370190685e-05,
      "loss": 0.06568697094917297,
      "step": 3110
    },
    {
      "epoch": 0.4205865702287303,
      "grad_norm": 1.3381396532058716,
      "learning_rate": 2.0638131761252724e-05,
      "loss": 0.07705916464328766,
      "step": 3111
    },
    {
      "epoch": 0.4207217635974955,
      "grad_norm": 0.5389488339424133,
      "learning_rate": 2.06317886467095e-05,
      "loss": 0.05240902304649353,
      "step": 3112
    },
    {
      "epoch": 0.4208569569662608,
      "grad_norm": 0.9664044380187988,
      "learning_rate": 2.0625444359597847e-05,
      "loss": 0.07163377106189728,
      "step": 3113
    },
    {
      "epoch": 0.42099215033502607,
      "grad_norm": 0.28362175822257996,
      "learning_rate": 2.0619098901238684e-05,
      "loss": 0.05485422909259796,
      "step": 3114
    },
    {
      "epoch": 0.42112734370379135,
      "grad_norm": 0.7811862826347351,
      "learning_rate": 2.0612752272953158e-05,
      "loss": 0.06677161902189255,
      "step": 3115
    },
    {
      "epoch": 0.42126253707255656,
      "grad_norm": 0.8218448758125305,
      "learning_rate": 2.060640447606268e-05,
      "loss": 0.037672460079193115,
      "step": 3116
    },
    {
      "epoch": 0.42139773044132184,
      "grad_norm": 1.2673053741455078,
      "learning_rate": 2.0600055511888895e-05,
      "loss": 0.04295028746128082,
      "step": 3117
    },
    {
      "epoch": 0.4215329238100871,
      "grad_norm": 1.0854992866516113,
      "learning_rate": 2.059370538175369e-05,
      "loss": 0.06312769651412964,
      "step": 3118
    },
    {
      "epoch": 0.4216681171788524,
      "grad_norm": 0.5508562326431274,
      "learning_rate": 2.0587354086979194e-05,
      "loss": 0.040297526866197586,
      "step": 3119
    },
    {
      "epoch": 0.42180331054761766,
      "grad_norm": 0.3192689120769501,
      "learning_rate": 2.0581001628887785e-05,
      "loss": 0.05371555685997009,
      "step": 3120
    },
    {
      "epoch": 0.4219385039163829,
      "grad_norm": 1.5956919193267822,
      "learning_rate": 2.057464800880207e-05,
      "loss": 0.07366383075714111,
      "step": 3121
    },
    {
      "epoch": 0.42207369728514815,
      "grad_norm": 0.721843421459198,
      "learning_rate": 2.0568293228044914e-05,
      "loss": 0.05905497819185257,
      "step": 3122
    },
    {
      "epoch": 0.4222088906539134,
      "grad_norm": 0.3545355796813965,
      "learning_rate": 2.0561937287939413e-05,
      "loss": 0.03873894736170769,
      "step": 3123
    },
    {
      "epoch": 0.4223440840226787,
      "grad_norm": 0.4149668514728546,
      "learning_rate": 2.055558018980891e-05,
      "loss": 0.054845765233039856,
      "step": 3124
    },
    {
      "epoch": 0.422479277391444,
      "grad_norm": 0.7542999386787415,
      "learning_rate": 2.0549221934976987e-05,
      "loss": 0.05623306334018707,
      "step": 3125
    },
    {
      "epoch": 0.4226144707602092,
      "grad_norm": 0.5587112903594971,
      "learning_rate": 2.054286252476746e-05,
      "loss": 0.05437682569026947,
      "step": 3126
    },
    {
      "epoch": 0.42274966412897447,
      "grad_norm": 0.6315963268280029,
      "learning_rate": 2.05365019605044e-05,
      "loss": 0.046305060386657715,
      "step": 3127
    },
    {
      "epoch": 0.42288485749773974,
      "grad_norm": 0.4447070360183716,
      "learning_rate": 2.053014024351211e-05,
      "loss": 0.05291140079498291,
      "step": 3128
    },
    {
      "epoch": 0.423020050866505,
      "grad_norm": 1.1890922784805298,
      "learning_rate": 2.0523777375115133e-05,
      "loss": 0.054853226989507675,
      "step": 3129
    },
    {
      "epoch": 0.42315524423527023,
      "grad_norm": 0.19163751602172852,
      "learning_rate": 2.0517413356638245e-05,
      "loss": 0.03453580290079117,
      "step": 3130
    },
    {
      "epoch": 0.4232904376040355,
      "grad_norm": 0.860350489616394,
      "learning_rate": 2.0511048189406472e-05,
      "loss": 0.07074567675590515,
      "step": 3131
    },
    {
      "epoch": 0.4234256309728008,
      "grad_norm": 0.3687298595905304,
      "learning_rate": 2.0504681874745082e-05,
      "loss": 0.05705578625202179,
      "step": 3132
    },
    {
      "epoch": 0.42356082434156606,
      "grad_norm": 0.5984498858451843,
      "learning_rate": 2.049831441397957e-05,
      "loss": 0.0578913688659668,
      "step": 3133
    },
    {
      "epoch": 0.42369601771033133,
      "grad_norm": 0.5607897639274597,
      "learning_rate": 2.0491945808435674e-05,
      "loss": 0.049040380865335464,
      "step": 3134
    },
    {
      "epoch": 0.42383121107909655,
      "grad_norm": 0.4735546410083771,
      "learning_rate": 2.048557605943938e-05,
      "loss": 0.07487677037715912,
      "step": 3135
    },
    {
      "epoch": 0.4239664044478618,
      "grad_norm": 0.4827166497707367,
      "learning_rate": 2.047920516831689e-05,
      "loss": 0.05880272388458252,
      "step": 3136
    },
    {
      "epoch": 0.4241015978166271,
      "grad_norm": 0.29004523158073425,
      "learning_rate": 2.047283313639467e-05,
      "loss": 0.040888816118240356,
      "step": 3137
    },
    {
      "epoch": 0.42423679118539237,
      "grad_norm": 0.87261563539505,
      "learning_rate": 2.0466459964999408e-05,
      "loss": 0.049823611974716187,
      "step": 3138
    },
    {
      "epoch": 0.42437198455415764,
      "grad_norm": 0.9923464059829712,
      "learning_rate": 2.0460085655458025e-05,
      "loss": 0.05934444069862366,
      "step": 3139
    },
    {
      "epoch": 0.42450717792292286,
      "grad_norm": 0.6945972442626953,
      "learning_rate": 2.0453710209097697e-05,
      "loss": 0.04255807772278786,
      "step": 3140
    },
    {
      "epoch": 0.42464237129168814,
      "grad_norm": 0.22872036695480347,
      "learning_rate": 2.044733362724582e-05,
      "loss": 0.050856903195381165,
      "step": 3141
    },
    {
      "epoch": 0.4247775646604534,
      "grad_norm": 1.0634028911590576,
      "learning_rate": 2.0440955911230028e-05,
      "loss": 0.08041512966156006,
      "step": 3142
    },
    {
      "epoch": 0.4249127580292187,
      "grad_norm": 0.3631167411804199,
      "learning_rate": 2.0434577062378203e-05,
      "loss": 0.049963533878326416,
      "step": 3143
    },
    {
      "epoch": 0.4250479513979839,
      "grad_norm": 0.7671000361442566,
      "learning_rate": 2.0428197082018458e-05,
      "loss": 0.06532436609268188,
      "step": 3144
    },
    {
      "epoch": 0.4251831447667492,
      "grad_norm": 0.37645432353019714,
      "learning_rate": 2.042181597147913e-05,
      "loss": 0.05664876103401184,
      "step": 3145
    },
    {
      "epoch": 0.42531833813551445,
      "grad_norm": 0.38391217589378357,
      "learning_rate": 2.0415433732088806e-05,
      "loss": 0.05824176222085953,
      "step": 3146
    },
    {
      "epoch": 0.4254535315042797,
      "grad_norm": 1.1771982908248901,
      "learning_rate": 2.0409050365176294e-05,
      "loss": 0.05723366141319275,
      "step": 3147
    },
    {
      "epoch": 0.425588724873045,
      "grad_norm": 0.8841889500617981,
      "learning_rate": 2.0402665872070656e-05,
      "loss": 0.059717755764722824,
      "step": 3148
    },
    {
      "epoch": 0.4257239182418102,
      "grad_norm": 0.2481062412261963,
      "learning_rate": 2.0396280254101172e-05,
      "loss": 0.04214269667863846,
      "step": 3149
    },
    {
      "epoch": 0.4258591116105755,
      "grad_norm": 0.5287272930145264,
      "learning_rate": 2.0389893512597364e-05,
      "loss": 0.05356164276599884,
      "step": 3150
    },
    {
      "epoch": 0.42599430497934077,
      "grad_norm": 0.8665952682495117,
      "learning_rate": 2.0383505648888986e-05,
      "loss": 0.04244658350944519,
      "step": 3151
    },
    {
      "epoch": 0.42612949834810604,
      "grad_norm": 0.9140850305557251,
      "learning_rate": 2.037711666430602e-05,
      "loss": 0.07658889889717102,
      "step": 3152
    },
    {
      "epoch": 0.4262646917168713,
      "grad_norm": 1.0958154201507568,
      "learning_rate": 2.0370726560178693e-05,
      "loss": 0.05771195888519287,
      "step": 3153
    },
    {
      "epoch": 0.42639988508563653,
      "grad_norm": 0.3769657611846924,
      "learning_rate": 2.036433533783745e-05,
      "loss": 0.039417192339897156,
      "step": 3154
    },
    {
      "epoch": 0.4265350784544018,
      "grad_norm": 0.9372339248657227,
      "learning_rate": 2.0357942998612988e-05,
      "loss": 0.05889472737908363,
      "step": 3155
    },
    {
      "epoch": 0.4266702718231671,
      "grad_norm": 0.41475072503089905,
      "learning_rate": 2.0351549543836224e-05,
      "loss": 0.05053633823990822,
      "step": 3156
    },
    {
      "epoch": 0.42680546519193235,
      "grad_norm": 0.9089570045471191,
      "learning_rate": 2.0345154974838307e-05,
      "loss": 0.04572737216949463,
      "step": 3157
    },
    {
      "epoch": 0.4269406585606976,
      "grad_norm": 0.34928643703460693,
      "learning_rate": 2.0338759292950618e-05,
      "loss": 0.044844917953014374,
      "step": 3158
    },
    {
      "epoch": 0.42707585192946285,
      "grad_norm": 0.38216763734817505,
      "learning_rate": 2.033236249950477e-05,
      "loss": 0.05039975047111511,
      "step": 3159
    },
    {
      "epoch": 0.4272110452982281,
      "grad_norm": 1.0492303371429443,
      "learning_rate": 2.0325964595832618e-05,
      "loss": 0.050439879298210144,
      "step": 3160
    },
    {
      "epoch": 0.4273462386669934,
      "grad_norm": 0.29913753271102905,
      "learning_rate": 2.031956558326624e-05,
      "loss": 0.06860119104385376,
      "step": 3161
    },
    {
      "epoch": 0.42748143203575867,
      "grad_norm": 1.319534182548523,
      "learning_rate": 2.0313165463137935e-05,
      "loss": 0.06233935058116913,
      "step": 3162
    },
    {
      "epoch": 0.4276166254045239,
      "grad_norm": 0.9906255602836609,
      "learning_rate": 2.030676423678025e-05,
      "loss": 0.06712990254163742,
      "step": 3163
    },
    {
      "epoch": 0.42775181877328916,
      "grad_norm": 0.25455281138420105,
      "learning_rate": 2.030036190552595e-05,
      "loss": 0.05154913663864136,
      "step": 3164
    },
    {
      "epoch": 0.42788701214205443,
      "grad_norm": 0.5683528184890747,
      "learning_rate": 2.029395847070803e-05,
      "loss": 0.0626387894153595,
      "step": 3165
    },
    {
      "epoch": 0.4280222055108197,
      "grad_norm": 0.8501094579696655,
      "learning_rate": 2.0287553933659735e-05,
      "loss": 0.05703309178352356,
      "step": 3166
    },
    {
      "epoch": 0.428157398879585,
      "grad_norm": 0.44154056906700134,
      "learning_rate": 2.0281148295714512e-05,
      "loss": 0.04259533807635307,
      "step": 3167
    },
    {
      "epoch": 0.4282925922483502,
      "grad_norm": 1.5962021350860596,
      "learning_rate": 2.027474155820605e-05,
      "loss": 0.05909880995750427,
      "step": 3168
    },
    {
      "epoch": 0.4284277856171155,
      "grad_norm": 0.6748774647712708,
      "learning_rate": 2.026833372246827e-05,
      "loss": 0.057523906230926514,
      "step": 3169
    },
    {
      "epoch": 0.42856297898588075,
      "grad_norm": 0.5485327839851379,
      "learning_rate": 2.026192478983531e-05,
      "loss": 0.055931925773620605,
      "step": 3170
    },
    {
      "epoch": 0.428698172354646,
      "grad_norm": 1.0570600032806396,
      "learning_rate": 2.0255514761641555e-05,
      "loss": 0.06017902493476868,
      "step": 3171
    },
    {
      "epoch": 0.42883336572341124,
      "grad_norm": 0.7914089560508728,
      "learning_rate": 2.0249103639221597e-05,
      "loss": 0.05615999549627304,
      "step": 3172
    },
    {
      "epoch": 0.4289685590921765,
      "grad_norm": 0.26349112391471863,
      "learning_rate": 2.024269142391027e-05,
      "loss": 0.04027664661407471,
      "step": 3173
    },
    {
      "epoch": 0.4291037524609418,
      "grad_norm": 0.3956128656864166,
      "learning_rate": 2.023627811704263e-05,
      "loss": 0.061393264681100845,
      "step": 3174
    },
    {
      "epoch": 0.42923894582970706,
      "grad_norm": 0.5028055310249329,
      "learning_rate": 2.0229863719953963e-05,
      "loss": 0.05934426188468933,
      "step": 3175
    },
    {
      "epoch": 0.42937413919847234,
      "grad_norm": 0.6211058497428894,
      "learning_rate": 2.0223448233979785e-05,
      "loss": 0.05135301500558853,
      "step": 3176
    },
    {
      "epoch": 0.42950933256723756,
      "grad_norm": 0.7907568216323853,
      "learning_rate": 2.0217031660455825e-05,
      "loss": 0.0515744686126709,
      "step": 3177
    },
    {
      "epoch": 0.42964452593600283,
      "grad_norm": 0.3090481162071228,
      "learning_rate": 2.0210614000718054e-05,
      "loss": 0.047113820910453796,
      "step": 3178
    },
    {
      "epoch": 0.4297797193047681,
      "grad_norm": 0.4285459518432617,
      "learning_rate": 2.020419525610266e-05,
      "loss": 0.05762678012251854,
      "step": 3179
    },
    {
      "epoch": 0.4299149126735334,
      "grad_norm": 0.7162730693817139,
      "learning_rate": 2.0197775427946066e-05,
      "loss": 0.061169058084487915,
      "step": 3180
    },
    {
      "epoch": 0.43005010604229865,
      "grad_norm": 0.3998583257198334,
      "learning_rate": 2.0191354517584902e-05,
      "loss": 0.061767011880874634,
      "step": 3181
    },
    {
      "epoch": 0.43018529941106387,
      "grad_norm": 0.34154802560806274,
      "learning_rate": 2.018493252635605e-05,
      "loss": 0.05337631702423096,
      "step": 3182
    },
    {
      "epoch": 0.43032049277982914,
      "grad_norm": 0.34869688749313354,
      "learning_rate": 2.0178509455596598e-05,
      "loss": 0.05695056915283203,
      "step": 3183
    },
    {
      "epoch": 0.4304556861485944,
      "grad_norm": 0.5041409730911255,
      "learning_rate": 2.017208530664386e-05,
      "loss": 0.047867417335510254,
      "step": 3184
    },
    {
      "epoch": 0.4305908795173597,
      "grad_norm": 1.3873205184936523,
      "learning_rate": 2.016566008083538e-05,
      "loss": 0.06679412722587585,
      "step": 3185
    },
    {
      "epoch": 0.4307260728861249,
      "grad_norm": 0.38294702768325806,
      "learning_rate": 2.0159233779508923e-05,
      "loss": 0.060565900057554245,
      "step": 3186
    },
    {
      "epoch": 0.4308612662548902,
      "grad_norm": 0.6644356846809387,
      "learning_rate": 2.0152806404002482e-05,
      "loss": 0.06483253836631775,
      "step": 3187
    },
    {
      "epoch": 0.43099645962365546,
      "grad_norm": 0.7891088724136353,
      "learning_rate": 2.014637795565427e-05,
      "loss": 0.053404733538627625,
      "step": 3188
    },
    {
      "epoch": 0.43113165299242073,
      "grad_norm": 0.4857129156589508,
      "learning_rate": 2.0139948435802722e-05,
      "loss": 0.05069521814584732,
      "step": 3189
    },
    {
      "epoch": 0.431266846361186,
      "grad_norm": 0.40805938839912415,
      "learning_rate": 2.0133517845786504e-05,
      "loss": 0.05410213768482208,
      "step": 3190
    },
    {
      "epoch": 0.4314020397299512,
      "grad_norm": 0.8711510300636292,
      "learning_rate": 2.012708618694449e-05,
      "loss": 0.04502767324447632,
      "step": 3191
    },
    {
      "epoch": 0.4315372330987165,
      "grad_norm": 0.9506723880767822,
      "learning_rate": 2.0120653460615795e-05,
      "loss": 0.06844263523817062,
      "step": 3192
    },
    {
      "epoch": 0.4316724264674818,
      "grad_norm": 0.6595968008041382,
      "learning_rate": 2.011421966813974e-05,
      "loss": 0.05935592204332352,
      "step": 3193
    },
    {
      "epoch": 0.43180761983624705,
      "grad_norm": 0.60198575258255,
      "learning_rate": 2.0107784810855882e-05,
      "loss": 0.04870133101940155,
      "step": 3194
    },
    {
      "epoch": 0.4319428132050123,
      "grad_norm": 0.7255297303199768,
      "learning_rate": 2.0101348890103985e-05,
      "loss": 0.05051903426647186,
      "step": 3195
    },
    {
      "epoch": 0.43207800657377754,
      "grad_norm": 0.3370662033557892,
      "learning_rate": 2.0094911907224043e-05,
      "loss": 0.05318271368741989,
      "step": 3196
    },
    {
      "epoch": 0.4322131999425428,
      "grad_norm": 0.7351927757263184,
      "learning_rate": 2.008847386355628e-05,
      "loss": 0.0428081750869751,
      "step": 3197
    },
    {
      "epoch": 0.4323483933113081,
      "grad_norm": 0.4200049340724945,
      "learning_rate": 2.008203476044112e-05,
      "loss": 0.052372708916664124,
      "step": 3198
    },
    {
      "epoch": 0.43248358668007336,
      "grad_norm": 2.193608283996582,
      "learning_rate": 2.007559459921922e-05,
      "loss": 0.0705721452832222,
      "step": 3199
    },
    {
      "epoch": 0.4326187800488386,
      "grad_norm": 1.3866478204727173,
      "learning_rate": 2.0069153381231456e-05,
      "loss": 0.05817686766386032,
      "step": 3200
    },
    {
      "epoch": 0.43275397341760385,
      "grad_norm": 1.2520829439163208,
      "learning_rate": 2.0062711107818933e-05,
      "loss": 0.06678328663110733,
      "step": 3201
    },
    {
      "epoch": 0.43288916678636913,
      "grad_norm": 0.2560953199863434,
      "learning_rate": 2.0056267780322953e-05,
      "loss": 0.04996124655008316,
      "step": 3202
    },
    {
      "epoch": 0.4330243601551344,
      "grad_norm": 0.5135100483894348,
      "learning_rate": 2.004982340008506e-05,
      "loss": 0.053074657917022705,
      "step": 3203
    },
    {
      "epoch": 0.4331595535238997,
      "grad_norm": 0.4020565450191498,
      "learning_rate": 2.004337796844701e-05,
      "loss": 0.0586613267660141,
      "step": 3204
    },
    {
      "epoch": 0.4332947468926649,
      "grad_norm": 0.45359864830970764,
      "learning_rate": 2.003693148675077e-05,
      "loss": 0.04980979859828949,
      "step": 3205
    },
    {
      "epoch": 0.43342994026143017,
      "grad_norm": 0.36846843361854553,
      "learning_rate": 2.003048395633853e-05,
      "loss": 0.06583218276500702,
      "step": 3206
    },
    {
      "epoch": 0.43356513363019544,
      "grad_norm": 0.3749634921550751,
      "learning_rate": 2.0024035378552708e-05,
      "loss": 0.061269551515579224,
      "step": 3207
    },
    {
      "epoch": 0.4337003269989607,
      "grad_norm": 0.431803435087204,
      "learning_rate": 2.001758575473593e-05,
      "loss": 0.06340453028678894,
      "step": 3208
    },
    {
      "epoch": 0.43383552036772594,
      "grad_norm": 0.3598921000957489,
      "learning_rate": 2.0011135086231042e-05,
      "loss": 0.06528577208518982,
      "step": 3209
    },
    {
      "epoch": 0.4339707137364912,
      "grad_norm": 0.4522221088409424,
      "learning_rate": 2.0004683374381104e-05,
      "loss": 0.0628328025341034,
      "step": 3210
    },
    {
      "epoch": 0.4341059071052565,
      "grad_norm": 0.5399999618530273,
      "learning_rate": 1.9998230620529395e-05,
      "loss": 0.06149455904960632,
      "step": 3211
    },
    {
      "epoch": 0.43424110047402176,
      "grad_norm": 0.7971583008766174,
      "learning_rate": 1.999177682601942e-05,
      "loss": 0.06106923520565033,
      "step": 3212
    },
    {
      "epoch": 0.43437629384278703,
      "grad_norm": 0.40637439489364624,
      "learning_rate": 1.9985321992194896e-05,
      "loss": 0.04652516171336174,
      "step": 3213
    },
    {
      "epoch": 0.43451148721155225,
      "grad_norm": 0.4229534864425659,
      "learning_rate": 1.9978866120399746e-05,
      "loss": 0.058541879057884216,
      "step": 3214
    },
    {
      "epoch": 0.4346466805803175,
      "grad_norm": 0.38915035128593445,
      "learning_rate": 1.9972409211978116e-05,
      "loss": 0.06971043348312378,
      "step": 3215
    },
    {
      "epoch": 0.4347818739490828,
      "grad_norm": 0.7337201833724976,
      "learning_rate": 1.9965951268274373e-05,
      "loss": 0.06125558912754059,
      "step": 3216
    },
    {
      "epoch": 0.43491706731784807,
      "grad_norm": 0.4810763895511627,
      "learning_rate": 1.9959492290633093e-05,
      "loss": 0.061926454305648804,
      "step": 3217
    },
    {
      "epoch": 0.43505226068661335,
      "grad_norm": 0.8450936675071716,
      "learning_rate": 1.995303228039907e-05,
      "loss": 0.09613863378763199,
      "step": 3218
    },
    {
      "epoch": 0.43518745405537856,
      "grad_norm": 0.5844223499298096,
      "learning_rate": 1.994657123891732e-05,
      "loss": 0.07898298650979996,
      "step": 3219
    },
    {
      "epoch": 0.43532264742414384,
      "grad_norm": 0.9888449311256409,
      "learning_rate": 1.9940109167533055e-05,
      "loss": 0.0579981803894043,
      "step": 3220
    },
    {
      "epoch": 0.4354578407929091,
      "grad_norm": 0.44750744104385376,
      "learning_rate": 1.9933646067591716e-05,
      "loss": 0.054648369550704956,
      "step": 3221
    },
    {
      "epoch": 0.4355930341616744,
      "grad_norm": 0.5578551292419434,
      "learning_rate": 1.992718194043896e-05,
      "loss": 0.0631774291396141,
      "step": 3222
    },
    {
      "epoch": 0.4357282275304396,
      "grad_norm": 0.556036651134491,
      "learning_rate": 1.9920716787420643e-05,
      "loss": 0.057975560426712036,
      "step": 3223
    },
    {
      "epoch": 0.4358634208992049,
      "grad_norm": 0.520154595375061,
      "learning_rate": 1.9914250609882858e-05,
      "loss": 0.062328506261110306,
      "step": 3224
    },
    {
      "epoch": 0.43599861426797015,
      "grad_norm": 0.33388620615005493,
      "learning_rate": 1.9907783409171885e-05,
      "loss": 0.06294068694114685,
      "step": 3225
    },
    {
      "epoch": 0.4361338076367354,
      "grad_norm": 0.8506790995597839,
      "learning_rate": 1.990131518663424e-05,
      "loss": 0.04879724979400635,
      "step": 3226
    },
    {
      "epoch": 0.4362690010055007,
      "grad_norm": 0.34767946600914,
      "learning_rate": 1.9894845943616632e-05,
      "loss": 0.0620148591697216,
      "step": 3227
    },
    {
      "epoch": 0.4364041943742659,
      "grad_norm": 0.36129623651504517,
      "learning_rate": 1.988837568146599e-05,
      "loss": 0.047696761786937714,
      "step": 3228
    },
    {
      "epoch": 0.4365393877430312,
      "grad_norm": 0.3131442666053772,
      "learning_rate": 1.988190440152947e-05,
      "loss": 0.05027863383293152,
      "step": 3229
    },
    {
      "epoch": 0.43667458111179647,
      "grad_norm": 0.28045254945755005,
      "learning_rate": 1.9875432105154424e-05,
      "loss": 0.048961855471134186,
      "step": 3230
    },
    {
      "epoch": 0.43680977448056174,
      "grad_norm": 0.33952921628952026,
      "learning_rate": 1.9868958793688412e-05,
      "loss": 0.0586174875497818,
      "step": 3231
    },
    {
      "epoch": 0.436944967849327,
      "grad_norm": 0.4964987635612488,
      "learning_rate": 1.9862484468479213e-05,
      "loss": 0.04754757881164551,
      "step": 3232
    },
    {
      "epoch": 0.43708016121809223,
      "grad_norm": 1.9296250343322754,
      "learning_rate": 1.985600913087482e-05,
      "loss": 0.07404591143131256,
      "step": 3233
    },
    {
      "epoch": 0.4372153545868575,
      "grad_norm": 0.638374388217926,
      "learning_rate": 1.9849532782223425e-05,
      "loss": 0.048216745257377625,
      "step": 3234
    },
    {
      "epoch": 0.4373505479556228,
      "grad_norm": 0.4848732054233551,
      "learning_rate": 1.9843055423873447e-05,
      "loss": 0.05438791215419769,
      "step": 3235
    },
    {
      "epoch": 0.43748574132438806,
      "grad_norm": 0.3670552670955658,
      "learning_rate": 1.9836577057173507e-05,
      "loss": 0.04548480361700058,
      "step": 3236
    },
    {
      "epoch": 0.4376209346931533,
      "grad_norm": 0.6398751139640808,
      "learning_rate": 1.9830097683472427e-05,
      "loss": 0.06153556704521179,
      "step": 3237
    },
    {
      "epoch": 0.43775612806191855,
      "grad_norm": 0.31612691283226013,
      "learning_rate": 1.9823617304119252e-05,
      "loss": 0.05666649341583252,
      "step": 3238
    },
    {
      "epoch": 0.4378913214306838,
      "grad_norm": 0.41127637028694153,
      "learning_rate": 1.9817135920463232e-05,
      "loss": 0.05100924149155617,
      "step": 3239
    },
    {
      "epoch": 0.4380265147994491,
      "grad_norm": 0.3253238797187805,
      "learning_rate": 1.9810653533853826e-05,
      "loss": 0.04596482589840889,
      "step": 3240
    },
    {
      "epoch": 0.43816170816821437,
      "grad_norm": 0.3820272982120514,
      "learning_rate": 1.9804170145640706e-05,
      "loss": 0.04634726047515869,
      "step": 3241
    },
    {
      "epoch": 0.4382969015369796,
      "grad_norm": 0.5377208590507507,
      "learning_rate": 1.9797685757173737e-05,
      "loss": 0.04338394105434418,
      "step": 3242
    },
    {
      "epoch": 0.43843209490574486,
      "grad_norm": 0.7749507427215576,
      "learning_rate": 1.979120036980301e-05,
      "loss": 0.05895250290632248,
      "step": 3243
    },
    {
      "epoch": 0.43856728827451014,
      "grad_norm": 0.7766172885894775,
      "learning_rate": 1.9784713984878814e-05,
      "loss": 0.06698055565357208,
      "step": 3244
    },
    {
      "epoch": 0.4387024816432754,
      "grad_norm": 0.652805507183075,
      "learning_rate": 1.9778226603751652e-05,
      "loss": 0.06592923402786255,
      "step": 3245
    },
    {
      "epoch": 0.4388376750120407,
      "grad_norm": 0.512524425983429,
      "learning_rate": 1.9771738227772235e-05,
      "loss": 0.047188758850097656,
      "step": 3246
    },
    {
      "epoch": 0.4389728683808059,
      "grad_norm": 0.41331592202186584,
      "learning_rate": 1.976524885829147e-05,
      "loss": 0.050759319216012955,
      "step": 3247
    },
    {
      "epoch": 0.4391080617495712,
      "grad_norm": 0.17818917334079742,
      "learning_rate": 1.975875849666048e-05,
      "loss": 0.03525908291339874,
      "step": 3248
    },
    {
      "epoch": 0.43924325511833645,
      "grad_norm": 1.3401192426681519,
      "learning_rate": 1.9752267144230595e-05,
      "loss": 0.09731544554233551,
      "step": 3249
    },
    {
      "epoch": 0.4393784484871017,
      "grad_norm": 0.4023383855819702,
      "learning_rate": 1.9745774802353347e-05,
      "loss": 0.06433025747537613,
      "step": 3250
    },
    {
      "epoch": 0.43951364185586694,
      "grad_norm": 1.0541208982467651,
      "learning_rate": 1.973928147238048e-05,
      "loss": 0.05375806987285614,
      "step": 3251
    },
    {
      "epoch": 0.4396488352246322,
      "grad_norm": 0.3709910809993744,
      "learning_rate": 1.973278715566394e-05,
      "loss": 0.047917380928993225,
      "step": 3252
    },
    {
      "epoch": 0.4397840285933975,
      "grad_norm": 0.4579212963581085,
      "learning_rate": 1.972629185355587e-05,
      "loss": 0.05296555161476135,
      "step": 3253
    },
    {
      "epoch": 0.43991922196216277,
      "grad_norm": 0.38086408376693726,
      "learning_rate": 1.971979556740864e-05,
      "loss": 0.04802680015563965,
      "step": 3254
    },
    {
      "epoch": 0.44005441533092804,
      "grad_norm": 0.7378490567207336,
      "learning_rate": 1.971329829857479e-05,
      "loss": 0.09295317530632019,
      "step": 3255
    },
    {
      "epoch": 0.44018960869969326,
      "grad_norm": 0.3409130871295929,
      "learning_rate": 1.9706800048407112e-05,
      "loss": 0.06915012001991272,
      "step": 3256
    },
    {
      "epoch": 0.44032480206845853,
      "grad_norm": 0.5004827380180359,
      "learning_rate": 1.9700300818258566e-05,
      "loss": 0.05587489530444145,
      "step": 3257
    },
    {
      "epoch": 0.4404599954372238,
      "grad_norm": 0.4734703004360199,
      "learning_rate": 1.9693800609482318e-05,
      "loss": 0.08199352025985718,
      "step": 3258
    },
    {
      "epoch": 0.4405951888059891,
      "grad_norm": 1.2099796533584595,
      "learning_rate": 1.9687299423431754e-05,
      "loss": 0.047711823135614395,
      "step": 3259
    },
    {
      "epoch": 0.44073038217475435,
      "grad_norm": 0.4385868310928345,
      "learning_rate": 1.968079726146045e-05,
      "loss": 0.07175090909004211,
      "step": 3260
    },
    {
      "epoch": 0.44086557554351957,
      "grad_norm": 0.9640140533447266,
      "learning_rate": 1.9674294124922204e-05,
      "loss": 0.05886758118867874,
      "step": 3261
    },
    {
      "epoch": 0.44100076891228485,
      "grad_norm": 0.8408608436584473,
      "learning_rate": 1.966779001517099e-05,
      "loss": 0.056329671293497086,
      "step": 3262
    },
    {
      "epoch": 0.4411359622810501,
      "grad_norm": 0.9086351990699768,
      "learning_rate": 1.9661284933561006e-05,
      "loss": 0.06688427925109863,
      "step": 3263
    },
    {
      "epoch": 0.4412711556498154,
      "grad_norm": 0.3593735098838806,
      "learning_rate": 1.9654778881446636e-05,
      "loss": 0.06216779351234436,
      "step": 3264
    },
    {
      "epoch": 0.4414063490185806,
      "grad_norm": 0.5820820331573486,
      "learning_rate": 1.9648271860182487e-05,
      "loss": 0.05620887503027916,
      "step": 3265
    },
    {
      "epoch": 0.4415415423873459,
      "grad_norm": 0.3773776590824127,
      "learning_rate": 1.9641763871123345e-05,
      "loss": 0.05723300576210022,
      "step": 3266
    },
    {
      "epoch": 0.44167673575611116,
      "grad_norm": 0.6403797268867493,
      "learning_rate": 1.963525491562421e-05,
      "loss": 0.06503838300704956,
      "step": 3267
    },
    {
      "epoch": 0.44181192912487643,
      "grad_norm": 0.6258330941200256,
      "learning_rate": 1.9628744995040287e-05,
      "loss": 0.04666862636804581,
      "step": 3268
    },
    {
      "epoch": 0.4419471224936417,
      "grad_norm": 0.5811514258384705,
      "learning_rate": 1.9622234110726976e-05,
      "loss": 0.06009425222873688,
      "step": 3269
    },
    {
      "epoch": 0.4420823158624069,
      "grad_norm": 0.42902350425720215,
      "learning_rate": 1.9615722264039868e-05,
      "loss": 0.05902581661939621,
      "step": 3270
    },
    {
      "epoch": 0.4422175092311722,
      "grad_norm": 0.24103787541389465,
      "learning_rate": 1.9609209456334772e-05,
      "loss": 0.03415320813655853,
      "step": 3271
    },
    {
      "epoch": 0.4423527025999375,
      "grad_norm": 0.40127962827682495,
      "learning_rate": 1.960269568896769e-05,
      "loss": 0.03563021123409271,
      "step": 3272
    },
    {
      "epoch": 0.44248789596870275,
      "grad_norm": 0.558793306350708,
      "learning_rate": 1.9596180963294822e-05,
      "loss": 0.02755964919924736,
      "step": 3273
    },
    {
      "epoch": 0.442623089337468,
      "grad_norm": 0.7912986278533936,
      "learning_rate": 1.9589665280672564e-05,
      "loss": 0.04704117029905319,
      "step": 3274
    },
    {
      "epoch": 0.44275828270623324,
      "grad_norm": 0.46691492199897766,
      "learning_rate": 1.958314864245752e-05,
      "loss": 0.060979537665843964,
      "step": 3275
    },
    {
      "epoch": 0.4428934760749985,
      "grad_norm": 0.45566725730895996,
      "learning_rate": 1.957663105000649e-05,
      "loss": 0.06779834628105164,
      "step": 3276
    },
    {
      "epoch": 0.4430286694437638,
      "grad_norm": 0.38593584299087524,
      "learning_rate": 1.957011250467647e-05,
      "loss": 0.05130264535546303,
      "step": 3277
    },
    {
      "epoch": 0.44316386281252906,
      "grad_norm": 0.41180941462516785,
      "learning_rate": 1.9563593007824658e-05,
      "loss": 0.04429095238447189,
      "step": 3278
    },
    {
      "epoch": 0.4432990561812943,
      "grad_norm": 0.36301419138908386,
      "learning_rate": 1.9557072560808442e-05,
      "loss": 0.059429511427879333,
      "step": 3279
    },
    {
      "epoch": 0.44343424955005956,
      "grad_norm": 0.5429235696792603,
      "learning_rate": 1.9550551164985418e-05,
      "loss": 0.057512715458869934,
      "step": 3280
    },
    {
      "epoch": 0.44356944291882483,
      "grad_norm": 0.41305485367774963,
      "learning_rate": 1.9544028821713372e-05,
      "loss": 0.05373512953519821,
      "step": 3281
    },
    {
      "epoch": 0.4437046362875901,
      "grad_norm": 0.8366455435752869,
      "learning_rate": 1.9537505532350298e-05,
      "loss": 0.06713327765464783,
      "step": 3282
    },
    {
      "epoch": 0.4438398296563554,
      "grad_norm": 0.5665419697761536,
      "learning_rate": 1.9530981298254376e-05,
      "loss": 0.048474207520484924,
      "step": 3283
    },
    {
      "epoch": 0.4439750230251206,
      "grad_norm": 0.8575229644775391,
      "learning_rate": 1.9524456120783983e-05,
      "loss": 0.04002584517002106,
      "step": 3284
    },
    {
      "epoch": 0.44411021639388587,
      "grad_norm": 1.1227996349334717,
      "learning_rate": 1.95179300012977e-05,
      "loss": 0.06277889758348465,
      "step": 3285
    },
    {
      "epoch": 0.44424540976265114,
      "grad_norm": 0.4908751845359802,
      "learning_rate": 1.9511402941154296e-05,
      "loss": 0.03839091211557388,
      "step": 3286
    },
    {
      "epoch": 0.4443806031314164,
      "grad_norm": 0.701174259185791,
      "learning_rate": 1.950487494171274e-05,
      "loss": 0.057075630873441696,
      "step": 3287
    },
    {
      "epoch": 0.4445157965001817,
      "grad_norm": 0.2505267858505249,
      "learning_rate": 1.9498346004332203e-05,
      "loss": 0.04473879933357239,
      "step": 3288
    },
    {
      "epoch": 0.4446509898689469,
      "grad_norm": 0.5283285975456238,
      "learning_rate": 1.949181613037204e-05,
      "loss": 0.06413161754608154,
      "step": 3289
    },
    {
      "epoch": 0.4447861832377122,
      "grad_norm": 0.29598984122276306,
      "learning_rate": 1.9485285321191804e-05,
      "loss": 0.03635593503713608,
      "step": 3290
    },
    {
      "epoch": 0.44492137660647746,
      "grad_norm": 0.8156299591064453,
      "learning_rate": 1.9478753578151244e-05,
      "loss": 0.06624409556388855,
      "step": 3291
    },
    {
      "epoch": 0.44505656997524273,
      "grad_norm": 0.5123705267906189,
      "learning_rate": 1.9472220902610304e-05,
      "loss": 0.05679801106452942,
      "step": 3292
    },
    {
      "epoch": 0.44519176334400795,
      "grad_norm": 0.5711596012115479,
      "learning_rate": 1.9465687295929127e-05,
      "loss": 0.04937441647052765,
      "step": 3293
    },
    {
      "epoch": 0.4453269567127732,
      "grad_norm": 0.7444480061531067,
      "learning_rate": 1.945915275946804e-05,
      "loss": 0.05021470785140991,
      "step": 3294
    },
    {
      "epoch": 0.4454621500815385,
      "grad_norm": 0.6331806182861328,
      "learning_rate": 1.9452617294587573e-05,
      "loss": 0.042013660073280334,
      "step": 3295
    },
    {
      "epoch": 0.4455973434503038,
      "grad_norm": 0.29099100828170776,
      "learning_rate": 1.9446080902648435e-05,
      "loss": 0.055853720754384995,
      "step": 3296
    },
    {
      "epoch": 0.44573253681906905,
      "grad_norm": 0.688414990901947,
      "learning_rate": 1.943954358501154e-05,
      "loss": 0.05329141765832901,
      "step": 3297
    },
    {
      "epoch": 0.44586773018783427,
      "grad_norm": 0.30019357800483704,
      "learning_rate": 1.943300534303801e-05,
      "loss": 0.050568610429763794,
      "step": 3298
    },
    {
      "epoch": 0.44600292355659954,
      "grad_norm": 0.2570483982563019,
      "learning_rate": 1.9426466178089116e-05,
      "loss": 0.05790006369352341,
      "step": 3299
    },
    {
      "epoch": 0.4461381169253648,
      "grad_norm": 0.5182380676269531,
      "learning_rate": 1.9419926091526367e-05,
      "loss": 0.0813378095626831,
      "step": 3300
    },
    {
      "epoch": 0.4462733102941301,
      "grad_norm": 0.3804183900356293,
      "learning_rate": 1.9413385084711425e-05,
      "loss": 0.06054868921637535,
      "step": 3301
    },
    {
      "epoch": 0.44640850366289536,
      "grad_norm": 0.34261995553970337,
      "learning_rate": 1.9406843159006183e-05,
      "loss": 0.044449031352996826,
      "step": 3302
    },
    {
      "epoch": 0.4465436970316606,
      "grad_norm": 0.5972133278846741,
      "learning_rate": 1.940030031577269e-05,
      "loss": 0.05956093966960907,
      "step": 3303
    },
    {
      "epoch": 0.44667889040042585,
      "grad_norm": 0.425552636384964,
      "learning_rate": 1.9393756556373212e-05,
      "loss": 0.061608828604221344,
      "step": 3304
    },
    {
      "epoch": 0.44681408376919113,
      "grad_norm": 0.4237003028392792,
      "learning_rate": 1.9387211882170184e-05,
      "loss": 0.0635639876127243,
      "step": 3305
    },
    {
      "epoch": 0.4469492771379564,
      "grad_norm": 0.6458073258399963,
      "learning_rate": 1.9380666294526243e-05,
      "loss": 0.0525660365819931,
      "step": 3306
    },
    {
      "epoch": 0.4470844705067216,
      "grad_norm": 0.5430435538291931,
      "learning_rate": 1.9374119794804228e-05,
      "loss": 0.05350537225604057,
      "step": 3307
    },
    {
      "epoch": 0.4472196638754869,
      "grad_norm": 0.28671368956565857,
      "learning_rate": 1.936757238436714e-05,
      "loss": 0.045153431594371796,
      "step": 3308
    },
    {
      "epoch": 0.44735485724425217,
      "grad_norm": 0.41515183448791504,
      "learning_rate": 1.936102406457819e-05,
      "loss": 0.06470730900764465,
      "step": 3309
    },
    {
      "epoch": 0.44749005061301744,
      "grad_norm": 0.24309509992599487,
      "learning_rate": 1.935447483680078e-05,
      "loss": 0.04342038184404373,
      "step": 3310
    },
    {
      "epoch": 0.4476252439817827,
      "grad_norm": 0.32473987340927124,
      "learning_rate": 1.9347924702398484e-05,
      "loss": 0.05396459996700287,
      "step": 3311
    },
    {
      "epoch": 0.44776043735054794,
      "grad_norm": 2.080249786376953,
      "learning_rate": 1.9341373662735075e-05,
      "loss": 0.07097083330154419,
      "step": 3312
    },
    {
      "epoch": 0.4478956307193132,
      "grad_norm": 0.371449738740921,
      "learning_rate": 1.9334821719174524e-05,
      "loss": 0.056131958961486816,
      "step": 3313
    },
    {
      "epoch": 0.4480308240880785,
      "grad_norm": 0.6857618689537048,
      "learning_rate": 1.9328268873080974e-05,
      "loss": 0.05432959273457527,
      "step": 3314
    },
    {
      "epoch": 0.44816601745684376,
      "grad_norm": 0.7514235377311707,
      "learning_rate": 1.9321715125818765e-05,
      "loss": 0.04985123872756958,
      "step": 3315
    },
    {
      "epoch": 0.44830121082560903,
      "grad_norm": 0.7892392873764038,
      "learning_rate": 1.931516047875242e-05,
      "loss": 0.07212784886360168,
      "step": 3316
    },
    {
      "epoch": 0.44843640419437425,
      "grad_norm": 0.42287296056747437,
      "learning_rate": 1.930860493324665e-05,
      "loss": 0.044905733317136765,
      "step": 3317
    },
    {
      "epoch": 0.4485715975631395,
      "grad_norm": 0.762205183506012,
      "learning_rate": 1.9302048490666356e-05,
      "loss": 0.04661571606993675,
      "step": 3318
    },
    {
      "epoch": 0.4487067909319048,
      "grad_norm": 1.1957151889801025,
      "learning_rate": 1.9295491152376623e-05,
      "loss": 0.054930999875068665,
      "step": 3319
    },
    {
      "epoch": 0.44884198430067007,
      "grad_norm": 0.6091325879096985,
      "learning_rate": 1.928893291974273e-05,
      "loss": 0.0738677978515625,
      "step": 3320
    },
    {
      "epoch": 0.4489771776694353,
      "grad_norm": 0.5154362916946411,
      "learning_rate": 1.9282373794130132e-05,
      "loss": 0.05926453322172165,
      "step": 3321
    },
    {
      "epoch": 0.44911237103820056,
      "grad_norm": 0.8428158760070801,
      "learning_rate": 1.9275813776904472e-05,
      "loss": 0.0578303337097168,
      "step": 3322
    },
    {
      "epoch": 0.44924756440696584,
      "grad_norm": 1.1031607389450073,
      "learning_rate": 1.9269252869431582e-05,
      "loss": 0.05805609002709389,
      "step": 3323
    },
    {
      "epoch": 0.4493827577757311,
      "grad_norm": 1.2092585563659668,
      "learning_rate": 1.9262691073077476e-05,
      "loss": 0.07075157761573792,
      "step": 3324
    },
    {
      "epoch": 0.4495179511444964,
      "grad_norm": 0.5066566467285156,
      "learning_rate": 1.9256128389208363e-05,
      "loss": 0.039241723716259,
      "step": 3325
    },
    {
      "epoch": 0.4496531445132616,
      "grad_norm": 1.1367995738983154,
      "learning_rate": 1.924956481919062e-05,
      "loss": 0.08321493864059448,
      "step": 3326
    },
    {
      "epoch": 0.4497883378820269,
      "grad_norm": 0.36701610684394836,
      "learning_rate": 1.9243000364390825e-05,
      "loss": 0.04881875216960907,
      "step": 3327
    },
    {
      "epoch": 0.44992353125079215,
      "grad_norm": 0.3730778396129608,
      "learning_rate": 1.9236435026175717e-05,
      "loss": 0.05519121512770653,
      "step": 3328
    },
    {
      "epoch": 0.4500587246195574,
      "grad_norm": 0.877098798751831,
      "learning_rate": 1.9229868805912248e-05,
      "loss": 0.04307505488395691,
      "step": 3329
    },
    {
      "epoch": 0.4501939179883227,
      "grad_norm": 0.5113497376441956,
      "learning_rate": 1.9223301704967543e-05,
      "loss": 0.06186281889677048,
      "step": 3330
    },
    {
      "epoch": 0.4503291113570879,
      "grad_norm": 0.5405557751655579,
      "learning_rate": 1.92167337247089e-05,
      "loss": 0.0581301748752594,
      "step": 3331
    },
    {
      "epoch": 0.4504643047258532,
      "grad_norm": 0.661689281463623,
      "learning_rate": 1.9210164866503808e-05,
      "loss": 0.050081897526979446,
      "step": 3332
    },
    {
      "epoch": 0.45059949809461847,
      "grad_norm": 0.6406334638595581,
      "learning_rate": 1.9203595131719935e-05,
      "loss": 0.04544217884540558,
      "step": 3333
    },
    {
      "epoch": 0.45073469146338374,
      "grad_norm": 0.35986214876174927,
      "learning_rate": 1.9197024521725148e-05,
      "loss": 0.031505897641181946,
      "step": 3334
    },
    {
      "epoch": 0.45086988483214896,
      "grad_norm": 0.8860594034194946,
      "learning_rate": 1.9190453037887465e-05,
      "loss": 0.0622507706284523,
      "step": 3335
    },
    {
      "epoch": 0.45100507820091423,
      "grad_norm": 0.25882935523986816,
      "learning_rate": 1.918388068157512e-05,
      "loss": 0.04336508363485336,
      "step": 3336
    },
    {
      "epoch": 0.4511402715696795,
      "grad_norm": 0.9596101641654968,
      "learning_rate": 1.9177307454156507e-05,
      "loss": 0.03922261297702789,
      "step": 3337
    },
    {
      "epoch": 0.4512754649384448,
      "grad_norm": 0.47697144746780396,
      "learning_rate": 1.9170733357000202e-05,
      "loss": 0.06904802471399307,
      "step": 3338
    },
    {
      "epoch": 0.45141065830721006,
      "grad_norm": 0.4825569987297058,
      "learning_rate": 1.916415839147497e-05,
      "loss": 0.06335628032684326,
      "step": 3339
    },
    {
      "epoch": 0.4515458516759753,
      "grad_norm": 0.6721900701522827,
      "learning_rate": 1.9157582558949756e-05,
      "loss": 0.061501145362854004,
      "step": 3340
    },
    {
      "epoch": 0.45168104504474055,
      "grad_norm": 0.43628573417663574,
      "learning_rate": 1.9151005860793682e-05,
      "loss": 0.06130243092775345,
      "step": 3341
    },
    {
      "epoch": 0.4518162384135058,
      "grad_norm": 0.37040936946868896,
      "learning_rate": 1.9144428298376056e-05,
      "loss": 0.05750388652086258,
      "step": 3342
    },
    {
      "epoch": 0.4519514317822711,
      "grad_norm": 0.185185506939888,
      "learning_rate": 1.9137849873066355e-05,
      "loss": 0.030655065551400185,
      "step": 3343
    },
    {
      "epoch": 0.4520866251510363,
      "grad_norm": 0.9033924341201782,
      "learning_rate": 1.9131270586234243e-05,
      "loss": 0.06248645484447479,
      "step": 3344
    },
    {
      "epoch": 0.4522218185198016,
      "grad_norm": 0.9251096844673157,
      "learning_rate": 1.9124690439249564e-05,
      "loss": 0.07921051979064941,
      "step": 3345
    },
    {
      "epoch": 0.45235701188856686,
      "grad_norm": 0.5152464509010315,
      "learning_rate": 1.9118109433482342e-05,
      "loss": 0.06715615093708038,
      "step": 3346
    },
    {
      "epoch": 0.45249220525733214,
      "grad_norm": 0.5808513760566711,
      "learning_rate": 1.911152757030278e-05,
      "loss": 0.06280335783958435,
      "step": 3347
    },
    {
      "epoch": 0.4526273986260974,
      "grad_norm": 0.2645490765571594,
      "learning_rate": 1.9104944851081247e-05,
      "loss": 0.03735634684562683,
      "step": 3348
    },
    {
      "epoch": 0.45276259199486263,
      "grad_norm": 1.25602126121521,
      "learning_rate": 1.9098361277188303e-05,
      "loss": 0.058656878769397736,
      "step": 3349
    },
    {
      "epoch": 0.4528977853636279,
      "grad_norm": 1.070167064666748,
      "learning_rate": 1.909177684999469e-05,
      "loss": 0.06288847327232361,
      "step": 3350
    },
    {
      "epoch": 0.4530329787323932,
      "grad_norm": 1.1462658643722534,
      "learning_rate": 1.9085191570871316e-05,
      "loss": 0.05812841281294823,
      "step": 3351
    },
    {
      "epoch": 0.45316817210115845,
      "grad_norm": 1.4589474201202393,
      "learning_rate": 1.9078605441189275e-05,
      "loss": 0.07801459729671478,
      "step": 3352
    },
    {
      "epoch": 0.4533033654699237,
      "grad_norm": 0.8323929905891418,
      "learning_rate": 1.9072018462319828e-05,
      "loss": 0.04907388612627983,
      "step": 3353
    },
    {
      "epoch": 0.45343855883868894,
      "grad_norm": 0.56675785779953,
      "learning_rate": 1.9065430635634422e-05,
      "loss": 0.05784274637699127,
      "step": 3354
    },
    {
      "epoch": 0.4535737522074542,
      "grad_norm": 0.314178466796875,
      "learning_rate": 1.9058841962504677e-05,
      "loss": 0.05712277814745903,
      "step": 3355
    },
    {
      "epoch": 0.4537089455762195,
      "grad_norm": 0.542799711227417,
      "learning_rate": 1.9052252444302394e-05,
      "loss": 0.06617625057697296,
      "step": 3356
    },
    {
      "epoch": 0.45384413894498477,
      "grad_norm": 0.19473232328891754,
      "learning_rate": 1.904566208239954e-05,
      "loss": 0.04961702227592468,
      "step": 3357
    },
    {
      "epoch": 0.45397933231375,
      "grad_norm": 0.7159522175788879,
      "learning_rate": 1.903907087816827e-05,
      "loss": 0.03728868067264557,
      "step": 3358
    },
    {
      "epoch": 0.45411452568251526,
      "grad_norm": 1.2334436178207397,
      "learning_rate": 1.9032478832980902e-05,
      "loss": 0.05597636103630066,
      "step": 3359
    },
    {
      "epoch": 0.45424971905128053,
      "grad_norm": 0.3319777250289917,
      "learning_rate": 1.9025885948209938e-05,
      "loss": 0.0323946475982666,
      "step": 3360
    },
    {
      "epoch": 0.4543849124200458,
      "grad_norm": 0.8528616428375244,
      "learning_rate": 1.901929222522805e-05,
      "loss": 0.03945460915565491,
      "step": 3361
    },
    {
      "epoch": 0.4545201057888111,
      "grad_norm": 0.7063406705856323,
      "learning_rate": 1.901269766540809e-05,
      "loss": 0.05806471407413483,
      "step": 3362
    },
    {
      "epoch": 0.4546552991575763,
      "grad_norm": 0.9343569278717041,
      "learning_rate": 1.9006102270123076e-05,
      "loss": 0.05490513890981674,
      "step": 3363
    },
    {
      "epoch": 0.45479049252634157,
      "grad_norm": 0.6323087811470032,
      "learning_rate": 1.8999506040746208e-05,
      "loss": 0.06549021601676941,
      "step": 3364
    },
    {
      "epoch": 0.45492568589510685,
      "grad_norm": 0.35148748755455017,
      "learning_rate": 1.899290897865085e-05,
      "loss": 0.051027994602918625,
      "step": 3365
    },
    {
      "epoch": 0.4550608792638721,
      "grad_norm": 0.3146888315677643,
      "learning_rate": 1.898631108521055e-05,
      "loss": 0.05037091672420502,
      "step": 3366
    },
    {
      "epoch": 0.4551960726326374,
      "grad_norm": 0.8882355093955994,
      "learning_rate": 1.8979712361799027e-05,
      "loss": 0.05013510584831238,
      "step": 3367
    },
    {
      "epoch": 0.4553312660014026,
      "grad_norm": 0.9671123027801514,
      "learning_rate": 1.8973112809790168e-05,
      "loss": 0.044209957122802734,
      "step": 3368
    },
    {
      "epoch": 0.4554664593701679,
      "grad_norm": 0.2821582555770874,
      "learning_rate": 1.8966512430558036e-05,
      "loss": 0.05418318510055542,
      "step": 3369
    },
    {
      "epoch": 0.45560165273893316,
      "grad_norm": 1.2645623683929443,
      "learning_rate": 1.8959911225476858e-05,
      "loss": 0.0573011189699173,
      "step": 3370
    },
    {
      "epoch": 0.45573684610769843,
      "grad_norm": 0.3427494466304779,
      "learning_rate": 1.895330919592105e-05,
      "loss": 0.03825434297323227,
      "step": 3371
    },
    {
      "epoch": 0.45587203947646365,
      "grad_norm": 0.6923047304153442,
      "learning_rate": 1.8946706343265187e-05,
      "loss": 0.05962452292442322,
      "step": 3372
    },
    {
      "epoch": 0.4560072328452289,
      "grad_norm": 0.4523015022277832,
      "learning_rate": 1.8940102668884016e-05,
      "loss": 0.060275617986917496,
      "step": 3373
    },
    {
      "epoch": 0.4561424262139942,
      "grad_norm": 0.3502708673477173,
      "learning_rate": 1.893349817415246e-05,
      "loss": 0.04480568319559097,
      "step": 3374
    },
    {
      "epoch": 0.4562776195827595,
      "grad_norm": 0.31605008244514465,
      "learning_rate": 1.8926892860445607e-05,
      "loss": 0.038295455276966095,
      "step": 3375
    },
    {
      "epoch": 0.45641281295152475,
      "grad_norm": 0.8085044622421265,
      "learning_rate": 1.8920286729138718e-05,
      "loss": 0.056893039494752884,
      "step": 3376
    },
    {
      "epoch": 0.45654800632028997,
      "grad_norm": 0.748875081539154,
      "learning_rate": 1.8913679781607225e-05,
      "loss": 0.0810333639383316,
      "step": 3377
    },
    {
      "epoch": 0.45668319968905524,
      "grad_norm": 0.48029693961143494,
      "learning_rate": 1.8907072019226734e-05,
      "loss": 0.06152554228901863,
      "step": 3378
    },
    {
      "epoch": 0.4568183930578205,
      "grad_norm": 0.6049940586090088,
      "learning_rate": 1.8900463443373015e-05,
      "loss": 0.05309459567070007,
      "step": 3379
    },
    {
      "epoch": 0.4569535864265858,
      "grad_norm": 1.2409980297088623,
      "learning_rate": 1.889385405542201e-05,
      "loss": 0.06103135645389557,
      "step": 3380
    },
    {
      "epoch": 0.45708877979535106,
      "grad_norm": 1.6766926050186157,
      "learning_rate": 1.8887243856749816e-05,
      "loss": 0.08241996169090271,
      "step": 3381
    },
    {
      "epoch": 0.4572239731641163,
      "grad_norm": 0.535280704498291,
      "learning_rate": 1.8880632848732723e-05,
      "loss": 0.05949222296476364,
      "step": 3382
    },
    {
      "epoch": 0.45735916653288156,
      "grad_norm": 0.7128165364265442,
      "learning_rate": 1.8874021032747185e-05,
      "loss": 0.05191230773925781,
      "step": 3383
    },
    {
      "epoch": 0.45749435990164683,
      "grad_norm": 0.6367013454437256,
      "learning_rate": 1.8867408410169803e-05,
      "loss": 0.07156538963317871,
      "step": 3384
    },
    {
      "epoch": 0.4576295532704121,
      "grad_norm": 0.6274704933166504,
      "learning_rate": 1.886079498237737e-05,
      "loss": 0.06594263017177582,
      "step": 3385
    },
    {
      "epoch": 0.4577647466391773,
      "grad_norm": 0.8115509748458862,
      "learning_rate": 1.885418075074683e-05,
      "loss": 0.04770982265472412,
      "step": 3386
    },
    {
      "epoch": 0.4578999400079426,
      "grad_norm": 0.8404709696769714,
      "learning_rate": 1.884756571665531e-05,
      "loss": 0.05110614746809006,
      "step": 3387
    },
    {
      "epoch": 0.45803513337670787,
      "grad_norm": 0.7161521315574646,
      "learning_rate": 1.8840949881480085e-05,
      "loss": 0.052481502294540405,
      "step": 3388
    },
    {
      "epoch": 0.45817032674547314,
      "grad_norm": 1.0244470834732056,
      "learning_rate": 1.8834333246598613e-05,
      "loss": 0.05016744136810303,
      "step": 3389
    },
    {
      "epoch": 0.4583055201142384,
      "grad_norm": 0.2579854130744934,
      "learning_rate": 1.8827715813388514e-05,
      "loss": 0.04288007318973541,
      "step": 3390
    },
    {
      "epoch": 0.45844071348300364,
      "grad_norm": 0.25869378447532654,
      "learning_rate": 1.8821097583227572e-05,
      "loss": 0.04056933522224426,
      "step": 3391
    },
    {
      "epoch": 0.4585759068517689,
      "grad_norm": 0.2815791368484497,
      "learning_rate": 1.8814478557493732e-05,
      "loss": 0.04701721668243408,
      "step": 3392
    },
    {
      "epoch": 0.4587111002205342,
      "grad_norm": 0.7656209468841553,
      "learning_rate": 1.8807858737565118e-05,
      "loss": 0.05907118320465088,
      "step": 3393
    },
    {
      "epoch": 0.45884629358929946,
      "grad_norm": 0.3615525960922241,
      "learning_rate": 1.880123812482001e-05,
      "loss": 0.051198653876781464,
      "step": 3394
    },
    {
      "epoch": 0.45898148695806473,
      "grad_norm": 0.7177746295928955,
      "learning_rate": 1.8794616720636853e-05,
      "loss": 0.05915292724967003,
      "step": 3395
    },
    {
      "epoch": 0.45911668032682995,
      "grad_norm": 0.37491458654403687,
      "learning_rate": 1.8787994526394257e-05,
      "loss": 0.06178012490272522,
      "step": 3396
    },
    {
      "epoch": 0.4592518736955952,
      "grad_norm": 0.3707510232925415,
      "learning_rate": 1.8781371543471e-05,
      "loss": 0.0526086688041687,
      "step": 3397
    },
    {
      "epoch": 0.4593870670643605,
      "grad_norm": 1.5047873258590698,
      "learning_rate": 1.8774747773246024e-05,
      "loss": 0.06546938419342041,
      "step": 3398
    },
    {
      "epoch": 0.4595222604331258,
      "grad_norm": 0.6238464117050171,
      "learning_rate": 1.8768123217098438e-05,
      "loss": 0.06101621687412262,
      "step": 3399
    },
    {
      "epoch": 0.459657453801891,
      "grad_norm": 0.5374392867088318,
      "learning_rate": 1.8761497876407496e-05,
      "loss": 0.0675344169139862,
      "step": 3400
    },
    {
      "epoch": 0.45979264717065627,
      "grad_norm": 0.621504008769989,
      "learning_rate": 1.8754871752552646e-05,
      "loss": 0.07805784791707993,
      "step": 3401
    },
    {
      "epoch": 0.45992784053942154,
      "grad_norm": 0.28153881430625916,
      "learning_rate": 1.8748244846913463e-05,
      "loss": 0.05409560352563858,
      "step": 3402
    },
    {
      "epoch": 0.4600630339081868,
      "grad_norm": 0.4041364789009094,
      "learning_rate": 1.874161716086972e-05,
      "loss": 0.04168283939361572,
      "step": 3403
    },
    {
      "epoch": 0.4601982272769521,
      "grad_norm": 0.3518540561199188,
      "learning_rate": 1.8734988695801333e-05,
      "loss": 0.0461282804608345,
      "step": 3404
    },
    {
      "epoch": 0.4603334206457173,
      "grad_norm": 0.9236600399017334,
      "learning_rate": 1.8728359453088382e-05,
      "loss": 0.055802371352910995,
      "step": 3405
    },
    {
      "epoch": 0.4604686140144826,
      "grad_norm": 0.4125099182128906,
      "learning_rate": 1.8721729434111108e-05,
      "loss": 0.042859822511672974,
      "step": 3406
    },
    {
      "epoch": 0.46060380738324785,
      "grad_norm": 0.5350891351699829,
      "learning_rate": 1.871509864024992e-05,
      "loss": 0.05139743164181709,
      "step": 3407
    },
    {
      "epoch": 0.46073900075201313,
      "grad_norm": 0.7655153274536133,
      "learning_rate": 1.8708467072885385e-05,
      "loss": 0.06916612386703491,
      "step": 3408
    },
    {
      "epoch": 0.4608741941207784,
      "grad_norm": 0.8144391179084778,
      "learning_rate": 1.8701834733398227e-05,
      "loss": 0.04741179570555687,
      "step": 3409
    },
    {
      "epoch": 0.4610093874895436,
      "grad_norm": 0.24424514174461365,
      "learning_rate": 1.8695201623169335e-05,
      "loss": 0.05104641616344452,
      "step": 3410
    },
    {
      "epoch": 0.4611445808583089,
      "grad_norm": 0.5939989686012268,
      "learning_rate": 1.868856774357977e-05,
      "loss": 0.06348701566457748,
      "step": 3411
    },
    {
      "epoch": 0.46127977422707417,
      "grad_norm": 0.4160653352737427,
      "learning_rate": 1.868193309601072e-05,
      "loss": 0.06438948959112167,
      "step": 3412
    },
    {
      "epoch": 0.46141496759583944,
      "grad_norm": 0.3901027739048004,
      "learning_rate": 1.867529768184357e-05,
      "loss": 0.053491801023483276,
      "step": 3413
    },
    {
      "epoch": 0.46155016096460466,
      "grad_norm": 0.37588658928871155,
      "learning_rate": 1.8668661502459842e-05,
      "loss": 0.060191426426172256,
      "step": 3414
    },
    {
      "epoch": 0.46168535433336994,
      "grad_norm": 0.662421703338623,
      "learning_rate": 1.866202455924123e-05,
      "loss": 0.05157659202814102,
      "step": 3415
    },
    {
      "epoch": 0.4618205477021352,
      "grad_norm": 0.366858571767807,
      "learning_rate": 1.865538685356957e-05,
      "loss": 0.06058875471353531,
      "step": 3416
    },
    {
      "epoch": 0.4619557410709005,
      "grad_norm": 0.48786288499832153,
      "learning_rate": 1.864874838682688e-05,
      "loss": 0.061392560601234436,
      "step": 3417
    },
    {
      "epoch": 0.46209093443966576,
      "grad_norm": 0.2137085497379303,
      "learning_rate": 1.864210916039531e-05,
      "loss": 0.04010477662086487,
      "step": 3418
    },
    {
      "epoch": 0.462226127808431,
      "grad_norm": 0.27187997102737427,
      "learning_rate": 1.86354691756572e-05,
      "loss": 0.05676757171750069,
      "step": 3419
    },
    {
      "epoch": 0.46236132117719625,
      "grad_norm": 0.31700649857521057,
      "learning_rate": 1.8628828433995013e-05,
      "loss": 0.04720379412174225,
      "step": 3420
    },
    {
      "epoch": 0.4624965145459615,
      "grad_norm": 0.3173675835132599,
      "learning_rate": 1.86221869367914e-05,
      "loss": 0.04724174737930298,
      "step": 3421
    },
    {
      "epoch": 0.4626317079147268,
      "grad_norm": 1.117053747177124,
      "learning_rate": 1.8615544685429153e-05,
      "loss": 0.05501580983400345,
      "step": 3422
    },
    {
      "epoch": 0.46276690128349207,
      "grad_norm": 1.2657760381698608,
      "learning_rate": 1.860890168129122e-05,
      "loss": 0.06483148038387299,
      "step": 3423
    },
    {
      "epoch": 0.4629020946522573,
      "grad_norm": 0.3202558159828186,
      "learning_rate": 1.8602257925760708e-05,
      "loss": 0.04469441622495651,
      "step": 3424
    },
    {
      "epoch": 0.46303728802102256,
      "grad_norm": 0.38990518450737,
      "learning_rate": 1.859561342022089e-05,
      "loss": 0.07769818603992462,
      "step": 3425
    },
    {
      "epoch": 0.46317248138978784,
      "grad_norm": 0.9692981243133545,
      "learning_rate": 1.8588968166055185e-05,
      "loss": 0.04871663823723793,
      "step": 3426
    },
    {
      "epoch": 0.4633076747585531,
      "grad_norm": 1.47981858253479,
      "learning_rate": 1.8582322164647166e-05,
      "loss": 0.05820249021053314,
      "step": 3427
    },
    {
      "epoch": 0.46344286812731833,
      "grad_norm": 0.3757861852645874,
      "learning_rate": 1.8575675417380568e-05,
      "loss": 0.054184168577194214,
      "step": 3428
    },
    {
      "epoch": 0.4635780614960836,
      "grad_norm": 0.4767962098121643,
      "learning_rate": 1.856902792563928e-05,
      "loss": 0.051788631826639175,
      "step": 3429
    },
    {
      "epoch": 0.4637132548648489,
      "grad_norm": 0.4673423767089844,
      "learning_rate": 1.856237969080735e-05,
      "loss": 0.044467851519584656,
      "step": 3430
    },
    {
      "epoch": 0.46384844823361415,
      "grad_norm": 0.42419612407684326,
      "learning_rate": 1.8555730714268967e-05,
      "loss": 0.034988969564437866,
      "step": 3431
    },
    {
      "epoch": 0.4639836416023794,
      "grad_norm": 0.6036272644996643,
      "learning_rate": 1.8549080997408492e-05,
      "loss": 0.07301966845989227,
      "step": 3432
    },
    {
      "epoch": 0.46411883497114464,
      "grad_norm": 0.43555545806884766,
      "learning_rate": 1.8542430541610426e-05,
      "loss": 0.06862740218639374,
      "step": 3433
    },
    {
      "epoch": 0.4642540283399099,
      "grad_norm": 0.2437300980091095,
      "learning_rate": 1.8535779348259428e-05,
      "loss": 0.05106465518474579,
      "step": 3434
    },
    {
      "epoch": 0.4643892217086752,
      "grad_norm": 0.8447793126106262,
      "learning_rate": 1.852912741874032e-05,
      "loss": 0.06574514508247375,
      "step": 3435
    },
    {
      "epoch": 0.46452441507744047,
      "grad_norm": 0.44374769926071167,
      "learning_rate": 1.8522474754438056e-05,
      "loss": 0.06850980222225189,
      "step": 3436
    },
    {
      "epoch": 0.46465960844620574,
      "grad_norm": 0.6663258075714111,
      "learning_rate": 1.851582135673777e-05,
      "loss": 0.0613822340965271,
      "step": 3437
    },
    {
      "epoch": 0.46479480181497096,
      "grad_norm": 0.6282140612602234,
      "learning_rate": 1.850916722702473e-05,
      "loss": 0.06273164600133896,
      "step": 3438
    },
    {
      "epoch": 0.46492999518373623,
      "grad_norm": 0.633892834186554,
      "learning_rate": 1.8502512366684355e-05,
      "loss": 0.050341397523880005,
      "step": 3439
    },
    {
      "epoch": 0.4650651885525015,
      "grad_norm": 0.7119957804679871,
      "learning_rate": 1.8495856777102232e-05,
      "loss": 0.05206059291958809,
      "step": 3440
    },
    {
      "epoch": 0.4652003819212668,
      "grad_norm": 0.3833909332752228,
      "learning_rate": 1.848920045966408e-05,
      "loss": 0.06406398117542267,
      "step": 3441
    },
    {
      "epoch": 0.465335575290032,
      "grad_norm": 0.18850788474082947,
      "learning_rate": 1.8482543415755797e-05,
      "loss": 0.026808317750692368,
      "step": 3442
    },
    {
      "epoch": 0.4654707686587973,
      "grad_norm": 0.5003129839897156,
      "learning_rate": 1.8475885646763394e-05,
      "loss": 0.053905852138996124,
      "step": 3443
    },
    {
      "epoch": 0.46560596202756255,
      "grad_norm": 0.9879121780395508,
      "learning_rate": 1.8469227154073064e-05,
      "loss": 0.05546775460243225,
      "step": 3444
    },
    {
      "epoch": 0.4657411553963278,
      "grad_norm": 1.4978803396224976,
      "learning_rate": 1.8462567939071142e-05,
      "loss": 0.08119261264801025,
      "step": 3445
    },
    {
      "epoch": 0.4658763487650931,
      "grad_norm": 0.65383380651474,
      "learning_rate": 1.845590800314411e-05,
      "loss": 0.06052582710981369,
      "step": 3446
    },
    {
      "epoch": 0.4660115421338583,
      "grad_norm": 0.48527494072914124,
      "learning_rate": 1.8449247347678607e-05,
      "loss": 0.04265797138214111,
      "step": 3447
    },
    {
      "epoch": 0.4661467355026236,
      "grad_norm": 0.36451455950737,
      "learning_rate": 1.8442585974061405e-05,
      "loss": 0.04752948880195618,
      "step": 3448
    },
    {
      "epoch": 0.46628192887138886,
      "grad_norm": 0.7075680494308472,
      "learning_rate": 1.8435923883679452e-05,
      "loss": 0.05228123068809509,
      "step": 3449
    },
    {
      "epoch": 0.46641712224015414,
      "grad_norm": 0.5869306921958923,
      "learning_rate": 1.8429261077919818e-05,
      "loss": 0.055602557957172394,
      "step": 3450
    },
    {
      "epoch": 0.4665523156089194,
      "grad_norm": 0.3610169589519501,
      "learning_rate": 1.8422597558169742e-05,
      "loss": 0.05988682806491852,
      "step": 3451
    },
    {
      "epoch": 0.46668750897768463,
      "grad_norm": 0.5981849431991577,
      "learning_rate": 1.84159333258166e-05,
      "loss": 0.053812094032764435,
      "step": 3452
    },
    {
      "epoch": 0.4668227023464499,
      "grad_norm": 0.27451688051223755,
      "learning_rate": 1.8409268382247933e-05,
      "loss": 0.0516011118888855,
      "step": 3453
    },
    {
      "epoch": 0.4669578957152152,
      "grad_norm": 1.1405771970748901,
      "learning_rate": 1.8402602728851405e-05,
      "loss": 0.0609176903963089,
      "step": 3454
    },
    {
      "epoch": 0.46709308908398045,
      "grad_norm": 1.826614260673523,
      "learning_rate": 1.839593636701484e-05,
      "loss": 0.058491140604019165,
      "step": 3455
    },
    {
      "epoch": 0.46722828245274567,
      "grad_norm": 1.3256646394729614,
      "learning_rate": 1.8389269298126214e-05,
      "loss": 0.04908771812915802,
      "step": 3456
    },
    {
      "epoch": 0.46736347582151094,
      "grad_norm": 0.5058795809745789,
      "learning_rate": 1.838260152357365e-05,
      "loss": 0.03418835625052452,
      "step": 3457
    },
    {
      "epoch": 0.4674986691902762,
      "grad_norm": 0.7562401294708252,
      "learning_rate": 1.837593304474541e-05,
      "loss": 0.04584929347038269,
      "step": 3458
    },
    {
      "epoch": 0.4676338625590415,
      "grad_norm": 0.37185177206993103,
      "learning_rate": 1.836926386302991e-05,
      "loss": 0.06661425530910492,
      "step": 3459
    },
    {
      "epoch": 0.46776905592780677,
      "grad_norm": 0.3748548626899719,
      "learning_rate": 1.8362593979815696e-05,
      "loss": 0.05052608251571655,
      "step": 3460
    },
    {
      "epoch": 0.467904249296572,
      "grad_norm": 0.7036848068237305,
      "learning_rate": 1.8355923396491496e-05,
      "loss": 0.0682167336344719,
      "step": 3461
    },
    {
      "epoch": 0.46803944266533726,
      "grad_norm": 1.4316294193267822,
      "learning_rate": 1.8349252114446138e-05,
      "loss": 0.048164620995521545,
      "step": 3462
    },
    {
      "epoch": 0.46817463603410253,
      "grad_norm": 0.3539901673793793,
      "learning_rate": 1.834258013506864e-05,
      "loss": 0.05239434540271759,
      "step": 3463
    },
    {
      "epoch": 0.4683098294028678,
      "grad_norm": 1.9415796995162964,
      "learning_rate": 1.833590745974813e-05,
      "loss": 0.062813401222229,
      "step": 3464
    },
    {
      "epoch": 0.468445022771633,
      "grad_norm": 0.7124485373497009,
      "learning_rate": 1.8329234089873892e-05,
      "loss": 0.06759695708751678,
      "step": 3465
    },
    {
      "epoch": 0.4685802161403983,
      "grad_norm": 1.0345145463943481,
      "learning_rate": 1.8322560026835366e-05,
      "loss": 0.051732808351516724,
      "step": 3466
    },
    {
      "epoch": 0.46871540950916357,
      "grad_norm": 0.7185250520706177,
      "learning_rate": 1.8315885272022125e-05,
      "loss": 0.07194608449935913,
      "step": 3467
    },
    {
      "epoch": 0.46885060287792885,
      "grad_norm": 0.6272517442703247,
      "learning_rate": 1.830920982682389e-05,
      "loss": 0.05070233345031738,
      "step": 3468
    },
    {
      "epoch": 0.4689857962466941,
      "grad_norm": 0.955436110496521,
      "learning_rate": 1.830253369263052e-05,
      "loss": 0.04889228194952011,
      "step": 3469
    },
    {
      "epoch": 0.46912098961545934,
      "grad_norm": 0.5173132419586182,
      "learning_rate": 1.8295856870832024e-05,
      "loss": 0.04823637008666992,
      "step": 3470
    },
    {
      "epoch": 0.4692561829842246,
      "grad_norm": 0.44537919759750366,
      "learning_rate": 1.828917936281855e-05,
      "loss": 0.06603503227233887,
      "step": 3471
    },
    {
      "epoch": 0.4693913763529899,
      "grad_norm": 0.633919894695282,
      "learning_rate": 1.8282501169980396e-05,
      "loss": 0.06749454140663147,
      "step": 3472
    },
    {
      "epoch": 0.46952656972175516,
      "grad_norm": 1.38040292263031,
      "learning_rate": 1.8275822293707992e-05,
      "loss": 0.05902627110481262,
      "step": 3473
    },
    {
      "epoch": 0.46966176309052043,
      "grad_norm": 1.7549225091934204,
      "learning_rate": 1.8269142735391917e-05,
      "loss": 0.07592406868934631,
      "step": 3474
    },
    {
      "epoch": 0.46979695645928565,
      "grad_norm": 1.3159812688827515,
      "learning_rate": 1.8262462496422893e-05,
      "loss": 0.04475869983434677,
      "step": 3475
    },
    {
      "epoch": 0.4699321498280509,
      "grad_norm": 0.3364255130290985,
      "learning_rate": 1.8255781578191778e-05,
      "loss": 0.048334427177906036,
      "step": 3476
    },
    {
      "epoch": 0.4700673431968162,
      "grad_norm": 0.8258396983146667,
      "learning_rate": 1.824909998208958e-05,
      "loss": 0.05500677973031998,
      "step": 3477
    },
    {
      "epoch": 0.4702025365655815,
      "grad_norm": 1.0199753046035767,
      "learning_rate": 1.8242417709507434e-05,
      "loss": 0.056221529841423035,
      "step": 3478
    },
    {
      "epoch": 0.4703377299343467,
      "grad_norm": 0.5179654955863953,
      "learning_rate": 1.8235734761836637e-05,
      "loss": 0.06112375110387802,
      "step": 3479
    },
    {
      "epoch": 0.47047292330311197,
      "grad_norm": 0.5609266757965088,
      "learning_rate": 1.82290511404686e-05,
      "loss": 0.04634537547826767,
      "step": 3480
    },
    {
      "epoch": 0.47060811667187724,
      "grad_norm": 0.456564724445343,
      "learning_rate": 1.8222366846794904e-05,
      "loss": 0.044562939554452896,
      "step": 3481
    },
    {
      "epoch": 0.4707433100406425,
      "grad_norm": 0.6344621181488037,
      "learning_rate": 1.8215681882207238e-05,
      "loss": 0.04998070001602173,
      "step": 3482
    },
    {
      "epoch": 0.4708785034094078,
      "grad_norm": 1.3882066011428833,
      "learning_rate": 1.8208996248097462e-05,
      "loss": 0.06087478995323181,
      "step": 3483
    },
    {
      "epoch": 0.471013696778173,
      "grad_norm": 0.4571479856967926,
      "learning_rate": 1.8202309945857557e-05,
      "loss": 0.05453101545572281,
      "step": 3484
    },
    {
      "epoch": 0.4711488901469383,
      "grad_norm": 0.543473482131958,
      "learning_rate": 1.8195622976879638e-05,
      "loss": 0.06498540937900543,
      "step": 3485
    },
    {
      "epoch": 0.47128408351570356,
      "grad_norm": 0.596011757850647,
      "learning_rate": 1.8188935342555977e-05,
      "loss": 0.048884838819503784,
      "step": 3486
    },
    {
      "epoch": 0.47141927688446883,
      "grad_norm": 0.769769012928009,
      "learning_rate": 1.818224704427897e-05,
      "loss": 0.06901255995035172,
      "step": 3487
    },
    {
      "epoch": 0.4715544702532341,
      "grad_norm": 0.3739570081233978,
      "learning_rate": 1.8175558083441162e-05,
      "loss": 0.06934164464473724,
      "step": 3488
    },
    {
      "epoch": 0.4716896636219993,
      "grad_norm": 0.36289092898368835,
      "learning_rate": 1.8168868461435225e-05,
      "loss": 0.03892380744218826,
      "step": 3489
    },
    {
      "epoch": 0.4718248569907646,
      "grad_norm": 0.5343566536903381,
      "learning_rate": 1.8162178179653977e-05,
      "loss": 0.05886805057525635,
      "step": 3490
    },
    {
      "epoch": 0.47196005035952987,
      "grad_norm": 0.5023666024208069,
      "learning_rate": 1.815548723949037e-05,
      "loss": 0.05915558338165283,
      "step": 3491
    },
    {
      "epoch": 0.47209524372829514,
      "grad_norm": 0.46567168831825256,
      "learning_rate": 1.814879564233749e-05,
      "loss": 0.07154445350170135,
      "step": 3492
    },
    {
      "epoch": 0.47223043709706036,
      "grad_norm": 0.45786207914352417,
      "learning_rate": 1.8142103389588567e-05,
      "loss": 0.05651085823774338,
      "step": 3493
    },
    {
      "epoch": 0.47236563046582564,
      "grad_norm": 1.2955033779144287,
      "learning_rate": 1.813541048263696e-05,
      "loss": 0.047192446887493134,
      "step": 3494
    },
    {
      "epoch": 0.4725008238345909,
      "grad_norm": 0.8994665741920471,
      "learning_rate": 1.8128716922876178e-05,
      "loss": 0.04672951251268387,
      "step": 3495
    },
    {
      "epoch": 0.4726360172033562,
      "grad_norm": 0.39284980297088623,
      "learning_rate": 1.812202271169984e-05,
      "loss": 0.047931570559740067,
      "step": 3496
    },
    {
      "epoch": 0.47277121057212146,
      "grad_norm": 1.5437575578689575,
      "learning_rate": 1.8115327850501726e-05,
      "loss": 0.06407858431339264,
      "step": 3497
    },
    {
      "epoch": 0.4729064039408867,
      "grad_norm": 0.5672667622566223,
      "learning_rate": 1.8108632340675746e-05,
      "loss": 0.04306498169898987,
      "step": 3498
    },
    {
      "epoch": 0.47304159730965195,
      "grad_norm": 1.1058542728424072,
      "learning_rate": 1.810193618361593e-05,
      "loss": 0.059718698263168335,
      "step": 3499
    },
    {
      "epoch": 0.4731767906784172,
      "grad_norm": 0.32095372676849365,
      "learning_rate": 1.8095239380716464e-05,
      "loss": 0.043613508343696594,
      "step": 3500
    },
    {
      "epoch": 0.4733119840471825,
      "grad_norm": 0.5730457305908203,
      "learning_rate": 1.808854193337165e-05,
      "loss": 0.0419449508190155,
      "step": 3501
    },
    {
      "epoch": 0.4734471774159478,
      "grad_norm": 1.6305323839187622,
      "learning_rate": 1.8081843842975935e-05,
      "loss": 0.06303280591964722,
      "step": 3502
    },
    {
      "epoch": 0.473582370784713,
      "grad_norm": 1.2520970106124878,
      "learning_rate": 1.80751451109239e-05,
      "loss": 0.0593019537627697,
      "step": 3503
    },
    {
      "epoch": 0.47371756415347827,
      "grad_norm": 1.0650655031204224,
      "learning_rate": 1.806844573861025e-05,
      "loss": 0.04477345570921898,
      "step": 3504
    },
    {
      "epoch": 0.47385275752224354,
      "grad_norm": 0.3829644024372101,
      "learning_rate": 1.806174572742984e-05,
      "loss": 0.06156153231859207,
      "step": 3505
    },
    {
      "epoch": 0.4739879508910088,
      "grad_norm": 0.6122543811798096,
      "learning_rate": 1.8055045078777634e-05,
      "loss": 0.03903737664222717,
      "step": 3506
    },
    {
      "epoch": 0.47412314425977403,
      "grad_norm": 0.40458688139915466,
      "learning_rate": 1.8048343794048762e-05,
      "loss": 0.0539078563451767,
      "step": 3507
    },
    {
      "epoch": 0.4742583376285393,
      "grad_norm": 0.933499276638031,
      "learning_rate": 1.8041641874638445e-05,
      "loss": 0.07325289398431778,
      "step": 3508
    },
    {
      "epoch": 0.4743935309973046,
      "grad_norm": 0.2896197736263275,
      "learning_rate": 1.8034939321942077e-05,
      "loss": 0.047521188855171204,
      "step": 3509
    },
    {
      "epoch": 0.47452872436606985,
      "grad_norm": 0.46743831038475037,
      "learning_rate": 1.8028236137355154e-05,
      "loss": 0.06093984842300415,
      "step": 3510
    },
    {
      "epoch": 0.47466391773483513,
      "grad_norm": 0.4030322730541229,
      "learning_rate": 1.8021532322273327e-05,
      "loss": 0.04286172240972519,
      "step": 3511
    },
    {
      "epoch": 0.47479911110360035,
      "grad_norm": 1.35879647731781,
      "learning_rate": 1.8014827878092347e-05,
      "loss": 0.05102885514497757,
      "step": 3512
    },
    {
      "epoch": 0.4749343044723656,
      "grad_norm": 0.5118727684020996,
      "learning_rate": 1.800812280620813e-05,
      "loss": 0.0731704980134964,
      "step": 3513
    },
    {
      "epoch": 0.4750694978411309,
      "grad_norm": 0.4939217269420624,
      "learning_rate": 1.80014171080167e-05,
      "loss": 0.0499170646071434,
      "step": 3514
    },
    {
      "epoch": 0.47520469120989617,
      "grad_norm": 0.29561617970466614,
      "learning_rate": 1.7994710784914227e-05,
      "loss": 0.051967136561870575,
      "step": 3515
    },
    {
      "epoch": 0.47533988457866144,
      "grad_norm": 0.5234993696212769,
      "learning_rate": 1.7988003838297e-05,
      "loss": 0.05290350317955017,
      "step": 3516
    },
    {
      "epoch": 0.47547507794742666,
      "grad_norm": 0.415535032749176,
      "learning_rate": 1.7981296269561438e-05,
      "loss": 0.05252787843346596,
      "step": 3517
    },
    {
      "epoch": 0.47561027131619193,
      "grad_norm": 0.31221282482147217,
      "learning_rate": 1.7974588080104095e-05,
      "loss": 0.055691197514534,
      "step": 3518
    },
    {
      "epoch": 0.4757454646849572,
      "grad_norm": 0.2609007954597473,
      "learning_rate": 1.7967879271321648e-05,
      "loss": 0.05249927192926407,
      "step": 3519
    },
    {
      "epoch": 0.4758806580537225,
      "grad_norm": 0.5697064995765686,
      "learning_rate": 1.7961169844610913e-05,
      "loss": 0.03712989389896393,
      "step": 3520
    },
    {
      "epoch": 0.4760158514224877,
      "grad_norm": 0.9957741498947144,
      "learning_rate": 1.795445980136883e-05,
      "loss": 0.061709798872470856,
      "step": 3521
    },
    {
      "epoch": 0.476151044791253,
      "grad_norm": 1.1795940399169922,
      "learning_rate": 1.794774914299245e-05,
      "loss": 0.058040887117385864,
      "step": 3522
    },
    {
      "epoch": 0.47628623816001825,
      "grad_norm": 0.8981585502624512,
      "learning_rate": 1.794103787087899e-05,
      "loss": 0.062032561749219894,
      "step": 3523
    },
    {
      "epoch": 0.4764214315287835,
      "grad_norm": 0.3013242781162262,
      "learning_rate": 1.7934325986425755e-05,
      "loss": 0.04807254672050476,
      "step": 3524
    },
    {
      "epoch": 0.4765566248975488,
      "grad_norm": 0.4229622185230255,
      "learning_rate": 1.7927613491030204e-05,
      "loss": 0.05180754512548447,
      "step": 3525
    },
    {
      "epoch": 0.476691818266314,
      "grad_norm": 0.39684975147247314,
      "learning_rate": 1.7920900386089913e-05,
      "loss": 0.042893849313259125,
      "step": 3526
    },
    {
      "epoch": 0.4768270116350793,
      "grad_norm": 0.374948114156723,
      "learning_rate": 1.7914186673002588e-05,
      "loss": 0.04052875190973282,
      "step": 3527
    },
    {
      "epoch": 0.47696220500384456,
      "grad_norm": 1.1256228685379028,
      "learning_rate": 1.790747235316605e-05,
      "loss": 0.06336495280265808,
      "step": 3528
    },
    {
      "epoch": 0.47709739837260984,
      "grad_norm": 1.4504151344299316,
      "learning_rate": 1.790075742797827e-05,
      "loss": 0.04924369975924492,
      "step": 3529
    },
    {
      "epoch": 0.4772325917413751,
      "grad_norm": 0.3167463541030884,
      "learning_rate": 1.789404189883732e-05,
      "loss": 0.05668431520462036,
      "step": 3530
    },
    {
      "epoch": 0.47736778511014033,
      "grad_norm": 0.6608914136886597,
      "learning_rate": 1.7887325767141415e-05,
      "loss": 0.05914923548698425,
      "step": 3531
    },
    {
      "epoch": 0.4775029784789056,
      "grad_norm": 0.16262947022914886,
      "learning_rate": 1.7880609034288894e-05,
      "loss": 0.03452926129102707,
      "step": 3532
    },
    {
      "epoch": 0.4776381718476709,
      "grad_norm": 0.38108453154563904,
      "learning_rate": 1.7873891701678208e-05,
      "loss": 0.0512351468205452,
      "step": 3533
    },
    {
      "epoch": 0.47777336521643615,
      "grad_norm": 0.43436264991760254,
      "learning_rate": 1.786717377070794e-05,
      "loss": 0.05606218799948692,
      "step": 3534
    },
    {
      "epoch": 0.47790855858520137,
      "grad_norm": 0.5354858040809631,
      "learning_rate": 1.7860455242776803e-05,
      "loss": 0.05875754728913307,
      "step": 3535
    },
    {
      "epoch": 0.47804375195396664,
      "grad_norm": 0.36154821515083313,
      "learning_rate": 1.7853736119283635e-05,
      "loss": 0.058581605553627014,
      "step": 3536
    },
    {
      "epoch": 0.4781789453227319,
      "grad_norm": 0.8720853924751282,
      "learning_rate": 1.7847016401627388e-05,
      "loss": 0.037343598902225494,
      "step": 3537
    },
    {
      "epoch": 0.4783141386914972,
      "grad_norm": 0.288687527179718,
      "learning_rate": 1.7840296091207144e-05,
      "loss": 0.04128693416714668,
      "step": 3538
    },
    {
      "epoch": 0.47844933206026247,
      "grad_norm": 0.23239724338054657,
      "learning_rate": 1.7833575189422107e-05,
      "loss": 0.04333505406975746,
      "step": 3539
    },
    {
      "epoch": 0.4785845254290277,
      "grad_norm": 0.40633338689804077,
      "learning_rate": 1.7826853697671604e-05,
      "loss": 0.05566149204969406,
      "step": 3540
    },
    {
      "epoch": 0.47871971879779296,
      "grad_norm": 0.8720612525939941,
      "learning_rate": 1.782013161735509e-05,
      "loss": 0.06611773371696472,
      "step": 3541
    },
    {
      "epoch": 0.47885491216655823,
      "grad_norm": 0.40848857164382935,
      "learning_rate": 1.781340894987213e-05,
      "loss": 0.05184454098343849,
      "step": 3542
    },
    {
      "epoch": 0.4789901055353235,
      "grad_norm": 0.29652848839759827,
      "learning_rate": 1.7806685696622426e-05,
      "loss": 0.04452505707740784,
      "step": 3543
    },
    {
      "epoch": 0.4791252989040888,
      "grad_norm": 0.7281109690666199,
      "learning_rate": 1.779996185900579e-05,
      "loss": 0.05375983566045761,
      "step": 3544
    },
    {
      "epoch": 0.479260492272854,
      "grad_norm": 0.790040135383606,
      "learning_rate": 1.7793237438422165e-05,
      "loss": 0.05115380883216858,
      "step": 3545
    },
    {
      "epoch": 0.4793956856416193,
      "grad_norm": 0.6497330665588379,
      "learning_rate": 1.7786512436271617e-05,
      "loss": 0.04705348610877991,
      "step": 3546
    },
    {
      "epoch": 0.47953087901038455,
      "grad_norm": 0.842193603515625,
      "learning_rate": 1.777978685395431e-05,
      "loss": 0.06490755081176758,
      "step": 3547
    },
    {
      "epoch": 0.4796660723791498,
      "grad_norm": 0.4795830547809601,
      "learning_rate": 1.7773060692870564e-05,
      "loss": 0.05699475109577179,
      "step": 3548
    },
    {
      "epoch": 0.47980126574791504,
      "grad_norm": 0.6608070135116577,
      "learning_rate": 1.7766333954420794e-05,
      "loss": 0.06021735072135925,
      "step": 3549
    },
    {
      "epoch": 0.4799364591166803,
      "grad_norm": 0.3167383670806885,
      "learning_rate": 1.775960664000554e-05,
      "loss": 0.061611637473106384,
      "step": 3550
    },
    {
      "epoch": 0.4800716524854456,
      "grad_norm": 2.2271599769592285,
      "learning_rate": 1.7752878751025463e-05,
      "loss": 0.07767148315906525,
      "step": 3551
    },
    {
      "epoch": 0.48020684585421086,
      "grad_norm": 0.7745353579521179,
      "learning_rate": 1.7746150288881352e-05,
      "loss": 0.057197004556655884,
      "step": 3552
    },
    {
      "epoch": 0.48034203922297614,
      "grad_norm": 0.6138405203819275,
      "learning_rate": 1.7739421254974114e-05,
      "loss": 0.06160023435950279,
      "step": 3553
    },
    {
      "epoch": 0.48047723259174135,
      "grad_norm": 0.50901859998703,
      "learning_rate": 1.7732691650704756e-05,
      "loss": 0.06081277132034302,
      "step": 3554
    },
    {
      "epoch": 0.48061242596050663,
      "grad_norm": 0.5683940649032593,
      "learning_rate": 1.7725961477474423e-05,
      "loss": 0.05153261125087738,
      "step": 3555
    },
    {
      "epoch": 0.4807476193292719,
      "grad_norm": 0.5863044857978821,
      "learning_rate": 1.7719230736684375e-05,
      "loss": 0.05348905920982361,
      "step": 3556
    },
    {
      "epoch": 0.4808828126980372,
      "grad_norm": 0.4014899730682373,
      "learning_rate": 1.771249942973599e-05,
      "loss": 0.03674754127860069,
      "step": 3557
    },
    {
      "epoch": 0.48101800606680245,
      "grad_norm": 0.3008640706539154,
      "learning_rate": 1.7705767558030756e-05,
      "loss": 0.05308730900287628,
      "step": 3558
    },
    {
      "epoch": 0.48115319943556767,
      "grad_norm": 0.6073523759841919,
      "learning_rate": 1.769903512297029e-05,
      "loss": 0.04392029717564583,
      "step": 3559
    },
    {
      "epoch": 0.48128839280433294,
      "grad_norm": 0.5493504405021667,
      "learning_rate": 1.7692302125956315e-05,
      "loss": 0.04186969995498657,
      "step": 3560
    },
    {
      "epoch": 0.4814235861730982,
      "grad_norm": 0.5778635740280151,
      "learning_rate": 1.768556856839068e-05,
      "loss": 0.06150808930397034,
      "step": 3561
    },
    {
      "epoch": 0.4815587795418635,
      "grad_norm": 0.7606193423271179,
      "learning_rate": 1.767883445167535e-05,
      "loss": 0.059654057025909424,
      "step": 3562
    },
    {
      "epoch": 0.4816939729106287,
      "grad_norm": 0.7920559048652649,
      "learning_rate": 1.7672099777212398e-05,
      "loss": 0.0523810088634491,
      "step": 3563
    },
    {
      "epoch": 0.481829166279394,
      "grad_norm": 1.0121209621429443,
      "learning_rate": 1.7665364546404034e-05,
      "loss": 0.061217568814754486,
      "step": 3564
    },
    {
      "epoch": 0.48196435964815926,
      "grad_norm": 0.7576692700386047,
      "learning_rate": 1.7658628760652548e-05,
      "loss": 0.040264010429382324,
      "step": 3565
    },
    {
      "epoch": 0.48209955301692453,
      "grad_norm": 0.7712067365646362,
      "learning_rate": 1.765189242136038e-05,
      "loss": 0.048045091331005096,
      "step": 3566
    },
    {
      "epoch": 0.4822347463856898,
      "grad_norm": 0.3825013339519501,
      "learning_rate": 1.7645155529930065e-05,
      "loss": 0.059248507022857666,
      "step": 3567
    },
    {
      "epoch": 0.482369939754455,
      "grad_norm": 0.4226016104221344,
      "learning_rate": 1.763841808776426e-05,
      "loss": 0.03228508681058884,
      "step": 3568
    },
    {
      "epoch": 0.4825051331232203,
      "grad_norm": 1.3603301048278809,
      "learning_rate": 1.763168009626575e-05,
      "loss": 0.057688385248184204,
      "step": 3569
    },
    {
      "epoch": 0.48264032649198557,
      "grad_norm": 1.6420178413391113,
      "learning_rate": 1.7624941556837406e-05,
      "loss": 0.05993642657995224,
      "step": 3570
    },
    {
      "epoch": 0.48277551986075085,
      "grad_norm": 0.24214904010295868,
      "learning_rate": 1.7618202470882233e-05,
      "loss": 0.054488539695739746,
      "step": 3571
    },
    {
      "epoch": 0.4829107132295161,
      "grad_norm": 0.23728974163532257,
      "learning_rate": 1.7611462839803336e-05,
      "loss": 0.0469437837600708,
      "step": 3572
    },
    {
      "epoch": 0.48304590659828134,
      "grad_norm": 0.31635594367980957,
      "learning_rate": 1.760472266500396e-05,
      "loss": 0.05480599403381348,
      "step": 3573
    },
    {
      "epoch": 0.4831810999670466,
      "grad_norm": 0.43468227982521057,
      "learning_rate": 1.759798194788743e-05,
      "loss": 0.04596111923456192,
      "step": 3574
    },
    {
      "epoch": 0.4833162933358119,
      "grad_norm": 0.3930162191390991,
      "learning_rate": 1.75912406898572e-05,
      "loss": 0.0596642941236496,
      "step": 3575
    },
    {
      "epoch": 0.48345148670457716,
      "grad_norm": 0.34544721245765686,
      "learning_rate": 1.758449889231685e-05,
      "loss": 0.03439716994762421,
      "step": 3576
    },
    {
      "epoch": 0.4835866800733424,
      "grad_norm": 0.4070488512516022,
      "learning_rate": 1.757775655667004e-05,
      "loss": 0.04105016589164734,
      "step": 3577
    },
    {
      "epoch": 0.48372187344210765,
      "grad_norm": 0.8999466300010681,
      "learning_rate": 1.757101368432057e-05,
      "loss": 0.06334449350833893,
      "step": 3578
    },
    {
      "epoch": 0.4838570668108729,
      "grad_norm": 0.6146352291107178,
      "learning_rate": 1.7564270276672343e-05,
      "loss": 0.06370649486780167,
      "step": 3579
    },
    {
      "epoch": 0.4839922601796382,
      "grad_norm": 0.3868604898452759,
      "learning_rate": 1.7557526335129372e-05,
      "loss": 0.05511938035488129,
      "step": 3580
    },
    {
      "epoch": 0.4841274535484035,
      "grad_norm": 0.4834255278110504,
      "learning_rate": 1.7550781861095774e-05,
      "loss": 0.04977622628211975,
      "step": 3581
    },
    {
      "epoch": 0.4842626469171687,
      "grad_norm": 0.8302196860313416,
      "learning_rate": 1.754403685597579e-05,
      "loss": 0.05525088310241699,
      "step": 3582
    },
    {
      "epoch": 0.48439784028593397,
      "grad_norm": 0.9061380624771118,
      "learning_rate": 1.7537291321173773e-05,
      "loss": 0.05217995494604111,
      "step": 3583
    },
    {
      "epoch": 0.48453303365469924,
      "grad_norm": 0.27826252579689026,
      "learning_rate": 1.7530545258094165e-05,
      "loss": 0.044787466526031494,
      "step": 3584
    },
    {
      "epoch": 0.4846682270234645,
      "grad_norm": 1.808152437210083,
      "learning_rate": 1.7523798668141548e-05,
      "loss": 0.0737103819847107,
      "step": 3585
    },
    {
      "epoch": 0.48480342039222973,
      "grad_norm": 0.47455063462257385,
      "learning_rate": 1.7517051552720584e-05,
      "loss": 0.058237701654434204,
      "step": 3586
    },
    {
      "epoch": 0.484938613760995,
      "grad_norm": 0.47527530789375305,
      "learning_rate": 1.7510303913236066e-05,
      "loss": 0.06245797872543335,
      "step": 3587
    },
    {
      "epoch": 0.4850738071297603,
      "grad_norm": 0.37395185232162476,
      "learning_rate": 1.7503555751092883e-05,
      "loss": 0.05553579330444336,
      "step": 3588
    },
    {
      "epoch": 0.48520900049852556,
      "grad_norm": 0.4764285683631897,
      "learning_rate": 1.7496807067696046e-05,
      "loss": 0.0616372674703598,
      "step": 3589
    },
    {
      "epoch": 0.48534419386729083,
      "grad_norm": 0.3947109282016754,
      "learning_rate": 1.7490057864450665e-05,
      "loss": 0.0552312433719635,
      "step": 3590
    },
    {
      "epoch": 0.48547938723605605,
      "grad_norm": 0.6810263395309448,
      "learning_rate": 1.748330814276195e-05,
      "loss": 0.06639324128627777,
      "step": 3591
    },
    {
      "epoch": 0.4856145806048213,
      "grad_norm": 0.37617674469947815,
      "learning_rate": 1.7476557904035243e-05,
      "loss": 0.054632678627967834,
      "step": 3592
    },
    {
      "epoch": 0.4857497739735866,
      "grad_norm": 0.5433797240257263,
      "learning_rate": 1.7469807149675973e-05,
      "loss": 0.060484692454338074,
      "step": 3593
    },
    {
      "epoch": 0.48588496734235187,
      "grad_norm": 1.147133469581604,
      "learning_rate": 1.7463055881089685e-05,
      "loss": 0.061211928725242615,
      "step": 3594
    },
    {
      "epoch": 0.48602016071111714,
      "grad_norm": 0.4189985692501068,
      "learning_rate": 1.7456304099682024e-05,
      "loss": 0.06052279472351074,
      "step": 3595
    },
    {
      "epoch": 0.48615535407988236,
      "grad_norm": 0.24788349866867065,
      "learning_rate": 1.7449551806858756e-05,
      "loss": 0.03750964254140854,
      "step": 3596
    },
    {
      "epoch": 0.48629054744864764,
      "grad_norm": 0.41173654794692993,
      "learning_rate": 1.7442799004025733e-05,
      "loss": 0.05764399468898773,
      "step": 3597
    },
    {
      "epoch": 0.4864257408174129,
      "grad_norm": 0.2908349633216858,
      "learning_rate": 1.7436045692588934e-05,
      "loss": 0.041406065225601196,
      "step": 3598
    },
    {
      "epoch": 0.4865609341861782,
      "grad_norm": 1.3618172407150269,
      "learning_rate": 1.742929187395443e-05,
      "loss": 0.06943368911743164,
      "step": 3599
    },
    {
      "epoch": 0.4866961275549434,
      "grad_norm": 0.31823453307151794,
      "learning_rate": 1.7422537549528402e-05,
      "loss": 0.04396023601293564,
      "step": 3600
    },
    {
      "epoch": 0.4868313209237087,
      "grad_norm": 0.3217940628528595,
      "learning_rate": 1.7415782720717147e-05,
      "loss": 0.05544228479266167,
      "step": 3601
    },
    {
      "epoch": 0.48696651429247395,
      "grad_norm": 0.4987863600254059,
      "learning_rate": 1.740902738892704e-05,
      "loss": 0.04933873564004898,
      "step": 3602
    },
    {
      "epoch": 0.4871017076612392,
      "grad_norm": 0.4408535361289978,
      "learning_rate": 1.7402271555564585e-05,
      "loss": 0.06837975978851318,
      "step": 3603
    },
    {
      "epoch": 0.4872369010300045,
      "grad_norm": 0.6852837204933167,
      "learning_rate": 1.739551522203638e-05,
      "loss": 0.05049797147512436,
      "step": 3604
    },
    {
      "epoch": 0.4873720943987697,
      "grad_norm": 0.8003811240196228,
      "learning_rate": 1.738875838974913e-05,
      "loss": 0.07135294377803802,
      "step": 3605
    },
    {
      "epoch": 0.487507287767535,
      "grad_norm": 0.27592015266418457,
      "learning_rate": 1.7382001060109652e-05,
      "loss": 0.028982965275645256,
      "step": 3606
    },
    {
      "epoch": 0.48764248113630027,
      "grad_norm": 0.64210444688797,
      "learning_rate": 1.7375243234524843e-05,
      "loss": 0.0634416788816452,
      "step": 3607
    },
    {
      "epoch": 0.48777767450506554,
      "grad_norm": 0.31281939148902893,
      "learning_rate": 1.736848491440173e-05,
      "loss": 0.046354249119758606,
      "step": 3608
    },
    {
      "epoch": 0.4879128678738308,
      "grad_norm": 0.71690833568573,
      "learning_rate": 1.7361726101147424e-05,
      "loss": 0.056982751935720444,
      "step": 3609
    },
    {
      "epoch": 0.48804806124259603,
      "grad_norm": 0.783579409122467,
      "learning_rate": 1.7354966796169157e-05,
      "loss": 0.05837477743625641,
      "step": 3610
    },
    {
      "epoch": 0.4881832546113613,
      "grad_norm": 0.5557472109794617,
      "learning_rate": 1.734820700087424e-05,
      "loss": 0.06476366519927979,
      "step": 3611
    },
    {
      "epoch": 0.4883184479801266,
      "grad_norm": 0.46474185585975647,
      "learning_rate": 1.7341446716670103e-05,
      "loss": 0.0403604730963707,
      "step": 3612
    },
    {
      "epoch": 0.48845364134889185,
      "grad_norm": 0.9985072612762451,
      "learning_rate": 1.7334685944964272e-05,
      "loss": 0.05111198127269745,
      "step": 3613
    },
    {
      "epoch": 0.4885888347176571,
      "grad_norm": 0.4394057095050812,
      "learning_rate": 1.732792468716438e-05,
      "loss": 0.058096371591091156,
      "step": 3614
    },
    {
      "epoch": 0.48872402808642235,
      "grad_norm": 0.5287598371505737,
      "learning_rate": 1.7321162944678155e-05,
      "loss": 0.0497131384909153,
      "step": 3615
    },
    {
      "epoch": 0.4888592214551876,
      "grad_norm": 1.235806941986084,
      "learning_rate": 1.7314400718913425e-05,
      "loss": 0.06044672429561615,
      "step": 3616
    },
    {
      "epoch": 0.4889944148239529,
      "grad_norm": 0.39756840467453003,
      "learning_rate": 1.7307638011278126e-05,
      "loss": 0.051515936851501465,
      "step": 3617
    },
    {
      "epoch": 0.48912960819271817,
      "grad_norm": 0.3148517906665802,
      "learning_rate": 1.7300874823180284e-05,
      "loss": 0.048752591013908386,
      "step": 3618
    },
    {
      "epoch": 0.4892648015614834,
      "grad_norm": 0.7498884797096252,
      "learning_rate": 1.7294111156028034e-05,
      "loss": 0.06937260180711746,
      "step": 3619
    },
    {
      "epoch": 0.48939999493024866,
      "grad_norm": 0.5027851462364197,
      "learning_rate": 1.7287347011229605e-05,
      "loss": 0.04623818397521973,
      "step": 3620
    },
    {
      "epoch": 0.48953518829901393,
      "grad_norm": 0.8526375889778137,
      "learning_rate": 1.7280582390193333e-05,
      "loss": 0.05940094590187073,
      "step": 3621
    },
    {
      "epoch": 0.4896703816677792,
      "grad_norm": 0.5256524085998535,
      "learning_rate": 1.7273817294327653e-05,
      "loss": 0.037637438625097275,
      "step": 3622
    },
    {
      "epoch": 0.4898055750365445,
      "grad_norm": 0.5717564225196838,
      "learning_rate": 1.726705172504108e-05,
      "loss": 0.0569169819355011,
      "step": 3623
    },
    {
      "epoch": 0.4899407684053097,
      "grad_norm": 0.7404683232307434,
      "learning_rate": 1.7260285683742248e-05,
      "loss": 0.03827935457229614,
      "step": 3624
    },
    {
      "epoch": 0.490075961774075,
      "grad_norm": 0.5484645366668701,
      "learning_rate": 1.7253519171839883e-05,
      "loss": 0.03865198791027069,
      "step": 3625
    },
    {
      "epoch": 0.49021115514284025,
      "grad_norm": 1.9374899864196777,
      "learning_rate": 1.724675219074281e-05,
      "loss": 0.05823603272438049,
      "step": 3626
    },
    {
      "epoch": 0.4903463485116055,
      "grad_norm": 0.561532199382782,
      "learning_rate": 1.7239984741859957e-05,
      "loss": 0.048579275608062744,
      "step": 3627
    },
    {
      "epoch": 0.49048154188037074,
      "grad_norm": 0.9890323281288147,
      "learning_rate": 1.7233216826600324e-05,
      "loss": 0.05866961181163788,
      "step": 3628
    },
    {
      "epoch": 0.490616735249136,
      "grad_norm": 0.6588320136070251,
      "learning_rate": 1.7226448446373047e-05,
      "loss": 0.06463548541069031,
      "step": 3629
    },
    {
      "epoch": 0.4907519286179013,
      "grad_norm": 0.3036702573299408,
      "learning_rate": 1.7219679602587326e-05,
      "loss": 0.04972761869430542,
      "step": 3630
    },
    {
      "epoch": 0.49088712198666656,
      "grad_norm": 0.507706880569458,
      "learning_rate": 1.7212910296652476e-05,
      "loss": 0.0412788987159729,
      "step": 3631
    },
    {
      "epoch": 0.49102231535543184,
      "grad_norm": 1.1250947713851929,
      "learning_rate": 1.7206140529977904e-05,
      "loss": 0.05426212400197983,
      "step": 3632
    },
    {
      "epoch": 0.49115750872419706,
      "grad_norm": 0.5611497759819031,
      "learning_rate": 1.719937030397311e-05,
      "loss": 0.06120358407497406,
      "step": 3633
    },
    {
      "epoch": 0.49129270209296233,
      "grad_norm": 1.639595627784729,
      "learning_rate": 1.7192599620047683e-05,
      "loss": 0.07729215919971466,
      "step": 3634
    },
    {
      "epoch": 0.4914278954617276,
      "grad_norm": 1.4780056476593018,
      "learning_rate": 1.718582847961133e-05,
      "loss": 0.06893113255500793,
      "step": 3635
    },
    {
      "epoch": 0.4915630888304929,
      "grad_norm": 0.35439521074295044,
      "learning_rate": 1.7179056884073826e-05,
      "loss": 0.0451018288731575,
      "step": 3636
    },
    {
      "epoch": 0.49169828219925815,
      "grad_norm": 0.9456911683082581,
      "learning_rate": 1.717228483484506e-05,
      "loss": 0.06447529792785645,
      "step": 3637
    },
    {
      "epoch": 0.49183347556802337,
      "grad_norm": 0.49770087003707886,
      "learning_rate": 1.7165512333335013e-05,
      "loss": 0.04016174376010895,
      "step": 3638
    },
    {
      "epoch": 0.49196866893678864,
      "grad_norm": 0.4902314841747284,
      "learning_rate": 1.715873938095374e-05,
      "loss": 0.05016827583312988,
      "step": 3639
    },
    {
      "epoch": 0.4921038623055539,
      "grad_norm": 0.7114960551261902,
      "learning_rate": 1.7151965979111427e-05,
      "loss": 0.05821014940738678,
      "step": 3640
    },
    {
      "epoch": 0.4922390556743192,
      "grad_norm": 0.5455580353736877,
      "learning_rate": 1.7145192129218313e-05,
      "loss": 0.0648580938577652,
      "step": 3641
    },
    {
      "epoch": 0.4923742490430844,
      "grad_norm": 0.5809600949287415,
      "learning_rate": 1.7138417832684763e-05,
      "loss": 0.06623226404190063,
      "step": 3642
    },
    {
      "epoch": 0.4925094424118497,
      "grad_norm": 0.43233659863471985,
      "learning_rate": 1.7131643090921216e-05,
      "loss": 0.07102841138839722,
      "step": 3643
    },
    {
      "epoch": 0.49264463578061496,
      "grad_norm": 0.4675983786582947,
      "learning_rate": 1.712486790533821e-05,
      "loss": 0.058873116970062256,
      "step": 3644
    },
    {
      "epoch": 0.49277982914938023,
      "grad_norm": 0.7671048641204834,
      "learning_rate": 1.7118092277346372e-05,
      "loss": 0.06054091453552246,
      "step": 3645
    },
    {
      "epoch": 0.4929150225181455,
      "grad_norm": 1.271550178527832,
      "learning_rate": 1.7111316208356428e-05,
      "loss": 0.05534249544143677,
      "step": 3646
    },
    {
      "epoch": 0.4930502158869107,
      "grad_norm": 0.302577942609787,
      "learning_rate": 1.7104539699779192e-05,
      "loss": 0.04478336125612259,
      "step": 3647
    },
    {
      "epoch": 0.493185409255676,
      "grad_norm": 0.3643266558647156,
      "learning_rate": 1.709776275302557e-05,
      "loss": 0.054196715354919434,
      "step": 3648
    },
    {
      "epoch": 0.4933206026244413,
      "grad_norm": 0.8454395532608032,
      "learning_rate": 1.7090985369506555e-05,
      "loss": 0.06105788052082062,
      "step": 3649
    },
    {
      "epoch": 0.49345579599320655,
      "grad_norm": 0.4101276695728302,
      "learning_rate": 1.708420755063323e-05,
      "loss": 0.03762774169445038,
      "step": 3650
    },
    {
      "epoch": 0.4935909893619718,
      "grad_norm": 0.3875649869441986,
      "learning_rate": 1.707742929781678e-05,
      "loss": 0.05809156596660614,
      "step": 3651
    },
    {
      "epoch": 0.49372618273073704,
      "grad_norm": 0.40271130204200745,
      "learning_rate": 1.707065061246848e-05,
      "loss": 0.05698806047439575,
      "step": 3652
    },
    {
      "epoch": 0.4938613760995023,
      "grad_norm": 0.8993547558784485,
      "learning_rate": 1.7063871495999677e-05,
      "loss": 0.07150521874427795,
      "step": 3653
    },
    {
      "epoch": 0.4939965694682676,
      "grad_norm": 1.4272431135177612,
      "learning_rate": 1.705709194982182e-05,
      "loss": 0.045105963945388794,
      "step": 3654
    },
    {
      "epoch": 0.49413176283703286,
      "grad_norm": 1.2940682172775269,
      "learning_rate": 1.7050311975346447e-05,
      "loss": 0.07005545496940613,
      "step": 3655
    },
    {
      "epoch": 0.4942669562057981,
      "grad_norm": 1.3660322427749634,
      "learning_rate": 1.704353157398519e-05,
      "loss": 0.06419873237609863,
      "step": 3656
    },
    {
      "epoch": 0.49440214957456335,
      "grad_norm": 0.4984577000141144,
      "learning_rate": 1.7036750747149764e-05,
      "loss": 0.06574180722236633,
      "step": 3657
    },
    {
      "epoch": 0.49453734294332863,
      "grad_norm": 0.4779943525791168,
      "learning_rate": 1.702996949625197e-05,
      "loss": 0.06674687564373016,
      "step": 3658
    },
    {
      "epoch": 0.4946725363120939,
      "grad_norm": 0.3778025805950165,
      "learning_rate": 1.7023187822703702e-05,
      "loss": 0.04428689926862717,
      "step": 3659
    },
    {
      "epoch": 0.4948077296808592,
      "grad_norm": 0.43517693877220154,
      "learning_rate": 1.7016405727916936e-05,
      "loss": 0.05122418701648712,
      "step": 3660
    },
    {
      "epoch": 0.4949429230496244,
      "grad_norm": 0.8824998736381531,
      "learning_rate": 1.700962321330375e-05,
      "loss": 0.05420784652233124,
      "step": 3661
    },
    {
      "epoch": 0.49507811641838967,
      "grad_norm": 0.4531822204589844,
      "learning_rate": 1.700284028027629e-05,
      "loss": 0.055697277188301086,
      "step": 3662
    },
    {
      "epoch": 0.49521330978715494,
      "grad_norm": 0.9987421035766602,
      "learning_rate": 1.6996056930246807e-05,
      "loss": 0.06256911158561707,
      "step": 3663
    },
    {
      "epoch": 0.4953485031559202,
      "grad_norm": 1.1916788816452026,
      "learning_rate": 1.6989273164627626e-05,
      "loss": 0.04673158377408981,
      "step": 3664
    },
    {
      "epoch": 0.4954836965246855,
      "grad_norm": 0.9413378238677979,
      "learning_rate": 1.6982488984831163e-05,
      "loss": 0.04367947578430176,
      "step": 3665
    },
    {
      "epoch": 0.4956188898934507,
      "grad_norm": 0.5563457608222961,
      "learning_rate": 1.697570439226992e-05,
      "loss": 0.053668826818466187,
      "step": 3666
    },
    {
      "epoch": 0.495754083262216,
      "grad_norm": 0.5396509766578674,
      "learning_rate": 1.6968919388356486e-05,
      "loss": 0.060034189373254776,
      "step": 3667
    },
    {
      "epoch": 0.49588927663098126,
      "grad_norm": 0.4821469485759735,
      "learning_rate": 1.696213397450354e-05,
      "loss": 0.05835293233394623,
      "step": 3668
    },
    {
      "epoch": 0.49602446999974653,
      "grad_norm": 0.758277177810669,
      "learning_rate": 1.695534815212384e-05,
      "loss": 0.06622427701950073,
      "step": 3669
    },
    {
      "epoch": 0.49615966336851175,
      "grad_norm": 0.3500867486000061,
      "learning_rate": 1.6948561922630223e-05,
      "loss": 0.030244916677474976,
      "step": 3670
    },
    {
      "epoch": 0.496294856737277,
      "grad_norm": 0.42022255063056946,
      "learning_rate": 1.694177528743562e-05,
      "loss": 0.056450068950653076,
      "step": 3671
    },
    {
      "epoch": 0.4964300501060423,
      "grad_norm": 0.3428938686847687,
      "learning_rate": 1.6934988247953053e-05,
      "loss": 0.034330885857343674,
      "step": 3672
    },
    {
      "epoch": 0.49656524347480757,
      "grad_norm": 0.9088848829269409,
      "learning_rate": 1.6928200805595606e-05,
      "loss": 0.049365729093551636,
      "step": 3673
    },
    {
      "epoch": 0.49670043684357285,
      "grad_norm": 1.6058392524719238,
      "learning_rate": 1.6921412961776475e-05,
      "loss": 0.06689521670341492,
      "step": 3674
    },
    {
      "epoch": 0.49683563021233806,
      "grad_norm": 0.7300386428833008,
      "learning_rate": 1.6914624717908922e-05,
      "loss": 0.061847418546676636,
      "step": 3675
    },
    {
      "epoch": 0.49697082358110334,
      "grad_norm": 1.1861059665679932,
      "learning_rate": 1.6907836075406288e-05,
      "loss": 0.04366448521614075,
      "step": 3676
    },
    {
      "epoch": 0.4971060169498686,
      "grad_norm": 0.352228045463562,
      "learning_rate": 1.690104703568201e-05,
      "loss": 0.06551776826381683,
      "step": 3677
    },
    {
      "epoch": 0.4972412103186339,
      "grad_norm": 0.4220818877220154,
      "learning_rate": 1.68942576001496e-05,
      "loss": 0.06559351086616516,
      "step": 3678
    },
    {
      "epoch": 0.49737640368739916,
      "grad_norm": 0.4009663462638855,
      "learning_rate": 1.6887467770222658e-05,
      "loss": 0.04567134380340576,
      "step": 3679
    },
    {
      "epoch": 0.4975115970561644,
      "grad_norm": 1.063888430595398,
      "learning_rate": 1.6880677547314865e-05,
      "loss": 0.05458955094218254,
      "step": 3680
    },
    {
      "epoch": 0.49764679042492965,
      "grad_norm": 0.7614601254463196,
      "learning_rate": 1.6873886932839973e-05,
      "loss": 0.0724869966506958,
      "step": 3681
    },
    {
      "epoch": 0.4977819837936949,
      "grad_norm": 0.31583043932914734,
      "learning_rate": 1.686709592821183e-05,
      "loss": 0.04564934968948364,
      "step": 3682
    },
    {
      "epoch": 0.4979171771624602,
      "grad_norm": 1.0900447368621826,
      "learning_rate": 1.6860304534844355e-05,
      "loss": 0.060673344880342484,
      "step": 3683
    },
    {
      "epoch": 0.4980523705312254,
      "grad_norm": 0.40941131114959717,
      "learning_rate": 1.6853512754151556e-05,
      "loss": 0.04743904620409012,
      "step": 3684
    },
    {
      "epoch": 0.4981875638999907,
      "grad_norm": 0.26381832361221313,
      "learning_rate": 1.684672058754752e-05,
      "loss": 0.05201912671327591,
      "step": 3685
    },
    {
      "epoch": 0.49832275726875597,
      "grad_norm": 1.4613826274871826,
      "learning_rate": 1.6839928036446416e-05,
      "loss": 0.05368952080607414,
      "step": 3686
    },
    {
      "epoch": 0.49845795063752124,
      "grad_norm": 1.019508957862854,
      "learning_rate": 1.6833135102262473e-05,
      "loss": 0.05993691086769104,
      "step": 3687
    },
    {
      "epoch": 0.4985931440062865,
      "grad_norm": 0.7055081129074097,
      "learning_rate": 1.682634178641003e-05,
      "loss": 0.07027196884155273,
      "step": 3688
    },
    {
      "epoch": 0.49872833737505173,
      "grad_norm": 1.317682147026062,
      "learning_rate": 1.6819548090303485e-05,
      "loss": 0.055247001349925995,
      "step": 3689
    },
    {
      "epoch": 0.498863530743817,
      "grad_norm": 1.383270263671875,
      "learning_rate": 1.6812754015357328e-05,
      "loss": 0.059474170207977295,
      "step": 3690
    },
    {
      "epoch": 0.4989987241125823,
      "grad_norm": 0.5482416152954102,
      "learning_rate": 1.680595956298612e-05,
      "loss": 0.056713249534368515,
      "step": 3691
    },
    {
      "epoch": 0.49913391748134756,
      "grad_norm": 0.8580617904663086,
      "learning_rate": 1.6799164734604497e-05,
      "loss": 0.052371665835380554,
      "step": 3692
    },
    {
      "epoch": 0.49926911085011283,
      "grad_norm": 0.7360714077949524,
      "learning_rate": 1.6792369531627186e-05,
      "loss": 0.060470789670944214,
      "step": 3693
    },
    {
      "epoch": 0.49940430421887805,
      "grad_norm": 0.5085428953170776,
      "learning_rate": 1.6785573955468974e-05,
      "loss": 0.035801105201244354,
      "step": 3694
    },
    {
      "epoch": 0.4995394975876433,
      "grad_norm": 0.5546193718910217,
      "learning_rate": 1.6778778007544745e-05,
      "loss": 0.058577146381139755,
      "step": 3695
    },
    {
      "epoch": 0.4996746909564086,
      "grad_norm": 0.5918977856636047,
      "learning_rate": 1.6771981689269452e-05,
      "loss": 0.047676801681518555,
      "step": 3696
    },
    {
      "epoch": 0.49980988432517387,
      "grad_norm": 0.5465928316116333,
      "learning_rate": 1.6765185002058123e-05,
      "loss": 0.05425170063972473,
      "step": 3697
    },
    {
      "epoch": 0.4999450776939391,
      "grad_norm": 0.28634053468704224,
      "learning_rate": 1.6758387947325856e-05,
      "loss": 0.04169359430670738,
      "step": 3698
    },
    {
      "epoch": 0.5000802710627044,
      "grad_norm": 1.013948917388916,
      "learning_rate": 1.6751590526487843e-05,
      "loss": 0.058259516954422,
      "step": 3699
    },
    {
      "epoch": 0.5002154644314697,
      "grad_norm": 0.4457032084465027,
      "learning_rate": 1.6744792740959347e-05,
      "loss": 0.06493023037910461,
      "step": 3700
    },
    {
      "epoch": 0.5003506578002349,
      "grad_norm": 0.762586772441864,
      "learning_rate": 1.6737994592155697e-05,
      "loss": 0.06665882468223572,
      "step": 3701
    },
    {
      "epoch": 0.5004858511690001,
      "grad_norm": 0.6518793702125549,
      "learning_rate": 1.6731196081492307e-05,
      "loss": 0.0613759383559227,
      "step": 3702
    },
    {
      "epoch": 0.5006210445377655,
      "grad_norm": 0.31194958090782166,
      "learning_rate": 1.6724397210384655e-05,
      "loss": 0.044153835624456406,
      "step": 3703
    },
    {
      "epoch": 0.5007562379065307,
      "grad_norm": 1.887346863746643,
      "learning_rate": 1.6717597980248308e-05,
      "loss": 0.06944090127944946,
      "step": 3704
    },
    {
      "epoch": 0.5008914312752959,
      "grad_norm": 1.396126389503479,
      "learning_rate": 1.6710798392498904e-05,
      "loss": 0.06309475004673004,
      "step": 3705
    },
    {
      "epoch": 0.5010266246440612,
      "grad_norm": 0.3908338248729706,
      "learning_rate": 1.6703998448552154e-05,
      "loss": 0.05575186014175415,
      "step": 3706
    },
    {
      "epoch": 0.5011618180128264,
      "grad_norm": 0.3825208842754364,
      "learning_rate": 1.669719814982384e-05,
      "loss": 0.055212318897247314,
      "step": 3707
    },
    {
      "epoch": 0.5012970113815918,
      "grad_norm": 0.24458405375480652,
      "learning_rate": 1.6690397497729818e-05,
      "loss": 0.042881518602371216,
      "step": 3708
    },
    {
      "epoch": 0.501432204750357,
      "grad_norm": 0.3589680790901184,
      "learning_rate": 1.6683596493686028e-05,
      "loss": 0.039042823016643524,
      "step": 3709
    },
    {
      "epoch": 0.5015673981191222,
      "grad_norm": 0.7032709121704102,
      "learning_rate": 1.667679513910846e-05,
      "loss": 0.05372938513755798,
      "step": 3710
    },
    {
      "epoch": 0.5017025914878875,
      "grad_norm": 0.5207394957542419,
      "learning_rate": 1.666999343541321e-05,
      "loss": 0.06177643686532974,
      "step": 3711
    },
    {
      "epoch": 0.5018377848566528,
      "grad_norm": 1.1613737344741821,
      "learning_rate": 1.6663191384016422e-05,
      "loss": 0.05667185038328171,
      "step": 3712
    },
    {
      "epoch": 0.5019729782254181,
      "grad_norm": 0.5407153964042664,
      "learning_rate": 1.6656388986334315e-05,
      "loss": 0.04644760861992836,
      "step": 3713
    },
    {
      "epoch": 0.5021081715941833,
      "grad_norm": 0.8264970183372498,
      "learning_rate": 1.6649586243783186e-05,
      "loss": 0.0530182421207428,
      "step": 3714
    },
    {
      "epoch": 0.5022433649629485,
      "grad_norm": 0.7608914971351624,
      "learning_rate": 1.6642783157779405e-05,
      "loss": 0.05709357559680939,
      "step": 3715
    },
    {
      "epoch": 0.5023785583317139,
      "grad_norm": 0.48501965403556824,
      "learning_rate": 1.6635979729739417e-05,
      "loss": 0.08360063284635544,
      "step": 3716
    },
    {
      "epoch": 0.5025137517004791,
      "grad_norm": 0.4754328727722168,
      "learning_rate": 1.662917596107972e-05,
      "loss": 0.05376851558685303,
      "step": 3717
    },
    {
      "epoch": 0.5026489450692444,
      "grad_norm": 0.9103453755378723,
      "learning_rate": 1.6622371853216904e-05,
      "loss": 0.050147056579589844,
      "step": 3718
    },
    {
      "epoch": 0.5027841384380096,
      "grad_norm": 1.55764901638031,
      "learning_rate": 1.661556740756761e-05,
      "loss": 0.06033648923039436,
      "step": 3719
    },
    {
      "epoch": 0.5029193318067748,
      "grad_norm": 0.27574923634529114,
      "learning_rate": 1.6608762625548572e-05,
      "loss": 0.06433028727769852,
      "step": 3720
    },
    {
      "epoch": 0.5030545251755402,
      "grad_norm": 0.7348741292953491,
      "learning_rate": 1.6601957508576573e-05,
      "loss": 0.04384501278400421,
      "step": 3721
    },
    {
      "epoch": 0.5031897185443054,
      "grad_norm": 1.7975655794143677,
      "learning_rate": 1.659515205806848e-05,
      "loss": 0.04705321788787842,
      "step": 3722
    },
    {
      "epoch": 0.5033249119130707,
      "grad_norm": 0.2675785422325134,
      "learning_rate": 1.6588346275441224e-05,
      "loss": 0.04868397116661072,
      "step": 3723
    },
    {
      "epoch": 0.5034601052818359,
      "grad_norm": 1.0152156352996826,
      "learning_rate": 1.65815401621118e-05,
      "loss": 0.052325956523418427,
      "step": 3724
    },
    {
      "epoch": 0.5035952986506012,
      "grad_norm": 0.5642540454864502,
      "learning_rate": 1.657473371949728e-05,
      "loss": 0.05865258723497391,
      "step": 3725
    },
    {
      "epoch": 0.5037304920193665,
      "grad_norm": 0.4270670413970947,
      "learning_rate": 1.6567926949014805e-05,
      "loss": 0.04334026575088501,
      "step": 3726
    },
    {
      "epoch": 0.5038656853881317,
      "grad_norm": 0.9907171130180359,
      "learning_rate": 1.6561119852081574e-05,
      "loss": 0.056989990174770355,
      "step": 3727
    },
    {
      "epoch": 0.5040008787568969,
      "grad_norm": 0.619630753993988,
      "learning_rate": 1.6554312430114868e-05,
      "loss": 0.06310467422008514,
      "step": 3728
    },
    {
      "epoch": 0.5041360721256622,
      "grad_norm": 0.35960811376571655,
      "learning_rate": 1.6547504684532026e-05,
      "loss": 0.03643840551376343,
      "step": 3729
    },
    {
      "epoch": 0.5042712654944275,
      "grad_norm": 0.6705995202064514,
      "learning_rate": 1.6540696616750454e-05,
      "loss": 0.06269891560077667,
      "step": 3730
    },
    {
      "epoch": 0.5044064588631928,
      "grad_norm": 0.42261171340942383,
      "learning_rate": 1.6533888228187628e-05,
      "loss": 0.04221126437187195,
      "step": 3731
    },
    {
      "epoch": 0.504541652231958,
      "grad_norm": 0.36732491850852966,
      "learning_rate": 1.6527079520261103e-05,
      "loss": 0.05267505347728729,
      "step": 3732
    },
    {
      "epoch": 0.5046768456007232,
      "grad_norm": 0.43220579624176025,
      "learning_rate": 1.6520270494388472e-05,
      "loss": 0.047395966947078705,
      "step": 3733
    },
    {
      "epoch": 0.5048120389694886,
      "grad_norm": 1.1709681749343872,
      "learning_rate": 1.6513461151987418e-05,
      "loss": 0.04951769858598709,
      "step": 3734
    },
    {
      "epoch": 0.5049472323382538,
      "grad_norm": 0.6413924098014832,
      "learning_rate": 1.6506651494475678e-05,
      "loss": 0.06443445384502411,
      "step": 3735
    },
    {
      "epoch": 0.5050824257070191,
      "grad_norm": 0.7721502780914307,
      "learning_rate": 1.6499841523271062e-05,
      "loss": 0.07294841855764389,
      "step": 3736
    },
    {
      "epoch": 0.5052176190757843,
      "grad_norm": 0.47612032294273376,
      "learning_rate": 1.649303123979145e-05,
      "loss": 0.053695209324359894,
      "step": 3737
    },
    {
      "epoch": 0.5053528124445495,
      "grad_norm": 0.4241267144680023,
      "learning_rate": 1.648622064545477e-05,
      "loss": 0.07089121639728546,
      "step": 3738
    },
    {
      "epoch": 0.5054880058133149,
      "grad_norm": 0.41727402806282043,
      "learning_rate": 1.6479409741679025e-05,
      "loss": 0.04553259164094925,
      "step": 3739
    },
    {
      "epoch": 0.5056231991820801,
      "grad_norm": 0.7261723279953003,
      "learning_rate": 1.6472598529882277e-05,
      "loss": 0.04759795963764191,
      "step": 3740
    },
    {
      "epoch": 0.5057583925508454,
      "grad_norm": 0.4641232490539551,
      "learning_rate": 1.646578701148267e-05,
      "loss": 0.05426902323961258,
      "step": 3741
    },
    {
      "epoch": 0.5058935859196106,
      "grad_norm": 0.331417441368103,
      "learning_rate": 1.6458975187898384e-05,
      "loss": 0.04891033470630646,
      "step": 3742
    },
    {
      "epoch": 0.5060287792883759,
      "grad_norm": 0.8553628325462341,
      "learning_rate": 1.6452163060547687e-05,
      "loss": 0.060906924307346344,
      "step": 3743
    },
    {
      "epoch": 0.5061639726571412,
      "grad_norm": 0.38397231698036194,
      "learning_rate": 1.64453506308489e-05,
      "loss": 0.06193387508392334,
      "step": 3744
    },
    {
      "epoch": 0.5062991660259064,
      "grad_norm": 0.5434215068817139,
      "learning_rate": 1.64385379002204e-05,
      "loss": 0.04837833344936371,
      "step": 3745
    },
    {
      "epoch": 0.5064343593946717,
      "grad_norm": 0.427910715341568,
      "learning_rate": 1.643172487008064e-05,
      "loss": 0.05249765142798424,
      "step": 3746
    },
    {
      "epoch": 0.506569552763437,
      "grad_norm": 0.6468247175216675,
      "learning_rate": 1.6424911541848124e-05,
      "loss": 0.06334252655506134,
      "step": 3747
    },
    {
      "epoch": 0.5067047461322022,
      "grad_norm": 0.33395493030548096,
      "learning_rate": 1.641809791694143e-05,
      "loss": 0.039932504296302795,
      "step": 3748
    },
    {
      "epoch": 0.5068399395009675,
      "grad_norm": 0.9206994771957397,
      "learning_rate": 1.6411283996779184e-05,
      "loss": 0.06204662472009659,
      "step": 3749
    },
    {
      "epoch": 0.5069751328697327,
      "grad_norm": 0.6091291308403015,
      "learning_rate": 1.6404469782780088e-05,
      "loss": 0.06372246891260147,
      "step": 3750
    },
    {
      "epoch": 0.507110326238498,
      "grad_norm": 0.9094211459159851,
      "learning_rate": 1.639765527636289e-05,
      "loss": 0.03974538296461105,
      "step": 3751
    },
    {
      "epoch": 0.5072455196072633,
      "grad_norm": 0.4258108139038086,
      "learning_rate": 1.639084047894641e-05,
      "loss": 0.05518403649330139,
      "step": 3752
    },
    {
      "epoch": 0.5073807129760285,
      "grad_norm": 0.5820168256759644,
      "learning_rate": 1.638402539194953e-05,
      "loss": 0.07280158996582031,
      "step": 3753
    },
    {
      "epoch": 0.5075159063447938,
      "grad_norm": 0.6401300430297852,
      "learning_rate": 1.6377210016791182e-05,
      "loss": 0.05840405076742172,
      "step": 3754
    },
    {
      "epoch": 0.507651099713559,
      "grad_norm": 0.7480988502502441,
      "learning_rate": 1.6370394354890364e-05,
      "loss": 0.044561177492141724,
      "step": 3755
    },
    {
      "epoch": 0.5077862930823243,
      "grad_norm": 0.6101779341697693,
      "learning_rate": 1.636357840766613e-05,
      "loss": 0.06543612480163574,
      "step": 3756
    },
    {
      "epoch": 0.5079214864510896,
      "grad_norm": 0.8819478154182434,
      "learning_rate": 1.6356762176537606e-05,
      "loss": 0.05290394276380539,
      "step": 3757
    },
    {
      "epoch": 0.5080566798198548,
      "grad_norm": 0.39467740058898926,
      "learning_rate": 1.6349945662923953e-05,
      "loss": 0.07516656816005707,
      "step": 3758
    },
    {
      "epoch": 0.5081918731886201,
      "grad_norm": 0.8476990461349487,
      "learning_rate": 1.634312886824442e-05,
      "loss": 0.0673803985118866,
      "step": 3759
    },
    {
      "epoch": 0.5083270665573854,
      "grad_norm": 0.29710081219673157,
      "learning_rate": 1.6336311793918298e-05,
      "loss": 0.05188039690256119,
      "step": 3760
    },
    {
      "epoch": 0.5084622599261506,
      "grad_norm": 0.7095276117324829,
      "learning_rate": 1.6329494441364925e-05,
      "loss": 0.06461555510759354,
      "step": 3761
    },
    {
      "epoch": 0.5085974532949159,
      "grad_norm": 0.3347511291503906,
      "learning_rate": 1.6322676812003727e-05,
      "loss": 0.04996323958039284,
      "step": 3762
    },
    {
      "epoch": 0.5087326466636811,
      "grad_norm": 0.8257862329483032,
      "learning_rate": 1.631585890725416e-05,
      "loss": 0.04910095036029816,
      "step": 3763
    },
    {
      "epoch": 0.5088678400324465,
      "grad_norm": 0.9673036336898804,
      "learning_rate": 1.630904072853575e-05,
      "loss": 0.06738243997097015,
      "step": 3764
    },
    {
      "epoch": 0.5090030334012117,
      "grad_norm": 0.49601346254348755,
      "learning_rate": 1.6302222277268085e-05,
      "loss": 0.053567975759506226,
      "step": 3765
    },
    {
      "epoch": 0.5091382267699769,
      "grad_norm": 0.21795786917209625,
      "learning_rate": 1.6295403554870794e-05,
      "loss": 0.051491305232048035,
      "step": 3766
    },
    {
      "epoch": 0.5092734201387422,
      "grad_norm": 1.084364652633667,
      "learning_rate": 1.6288584562763572e-05,
      "loss": 0.05329851806163788,
      "step": 3767
    },
    {
      "epoch": 0.5094086135075074,
      "grad_norm": 1.3986179828643799,
      "learning_rate": 1.6281765302366176e-05,
      "loss": 0.05432478338479996,
      "step": 3768
    },
    {
      "epoch": 0.5095438068762728,
      "grad_norm": 0.2347116768360138,
      "learning_rate": 1.6274945775098412e-05,
      "loss": 0.04109840840101242,
      "step": 3769
    },
    {
      "epoch": 0.509679000245038,
      "grad_norm": 0.36350691318511963,
      "learning_rate": 1.6268125982380135e-05,
      "loss": 0.05259581282734871,
      "step": 3770
    },
    {
      "epoch": 0.5098141936138032,
      "grad_norm": 0.36298471689224243,
      "learning_rate": 1.626130592563127e-05,
      "loss": 0.047557294368743896,
      "step": 3771
    },
    {
      "epoch": 0.5099493869825685,
      "grad_norm": 0.5269676446914673,
      "learning_rate": 1.6254485606271778e-05,
      "loss": 0.04008561372756958,
      "step": 3772
    },
    {
      "epoch": 0.5100845803513337,
      "grad_norm": 0.4229883551597595,
      "learning_rate": 1.6247665025721698e-05,
      "loss": 0.0676676407456398,
      "step": 3773
    },
    {
      "epoch": 0.5102197737200991,
      "grad_norm": 0.5472242832183838,
      "learning_rate": 1.62408441854011e-05,
      "loss": 0.042094528675079346,
      "step": 3774
    },
    {
      "epoch": 0.5103549670888643,
      "grad_norm": 0.22410422563552856,
      "learning_rate": 1.6234023086730136e-05,
      "loss": 0.021529115736484528,
      "step": 3775
    },
    {
      "epoch": 0.5104901604576295,
      "grad_norm": 0.2958018183708191,
      "learning_rate": 1.622720173112898e-05,
      "loss": 0.043539129197597504,
      "step": 3776
    },
    {
      "epoch": 0.5106253538263948,
      "grad_norm": 0.8274920582771301,
      "learning_rate": 1.6220380120017874e-05,
      "loss": 0.03790076822042465,
      "step": 3777
    },
    {
      "epoch": 0.5107605471951601,
      "grad_norm": 0.2734115421772003,
      "learning_rate": 1.6213558254817128e-05,
      "loss": 0.049877118319272995,
      "step": 3778
    },
    {
      "epoch": 0.5108957405639254,
      "grad_norm": 1.2204430103302002,
      "learning_rate": 1.6206736136947074e-05,
      "loss": 0.055194538086652756,
      "step": 3779
    },
    {
      "epoch": 0.5110309339326906,
      "grad_norm": 0.574268102645874,
      "learning_rate": 1.6199913767828126e-05,
      "loss": 0.047274813055992126,
      "step": 3780
    },
    {
      "epoch": 0.5111661273014558,
      "grad_norm": 1.3793399333953857,
      "learning_rate": 1.6193091148880733e-05,
      "loss": 0.06472516059875488,
      "step": 3781
    },
    {
      "epoch": 0.5113013206702212,
      "grad_norm": 0.5116593241691589,
      "learning_rate": 1.61862682815254e-05,
      "loss": 0.048178352415561676,
      "step": 3782
    },
    {
      "epoch": 0.5114365140389864,
      "grad_norm": 0.45036429166793823,
      "learning_rate": 1.617944516718268e-05,
      "loss": 0.05155980587005615,
      "step": 3783
    },
    {
      "epoch": 0.5115717074077516,
      "grad_norm": 0.8632551431655884,
      "learning_rate": 1.617262180727319e-05,
      "loss": 0.061946380883455276,
      "step": 3784
    },
    {
      "epoch": 0.5117069007765169,
      "grad_norm": 1.622459053993225,
      "learning_rate": 1.6165798203217588e-05,
      "loss": 0.04066108167171478,
      "step": 3785
    },
    {
      "epoch": 0.5118420941452821,
      "grad_norm": 0.623325765132904,
      "learning_rate": 1.6158974356436585e-05,
      "loss": 0.04205261170864105,
      "step": 3786
    },
    {
      "epoch": 0.5119772875140475,
      "grad_norm": 0.44530412554740906,
      "learning_rate": 1.6152150268350938e-05,
      "loss": 0.0872809886932373,
      "step": 3787
    },
    {
      "epoch": 0.5121124808828127,
      "grad_norm": 0.31950002908706665,
      "learning_rate": 1.6145325940381458e-05,
      "loss": 0.03133980557322502,
      "step": 3788
    },
    {
      "epoch": 0.5122476742515779,
      "grad_norm": 0.6281394362449646,
      "learning_rate": 1.6138501373949018e-05,
      "loss": 0.05071195960044861,
      "step": 3789
    },
    {
      "epoch": 0.5123828676203432,
      "grad_norm": 0.9504885077476501,
      "learning_rate": 1.613167657047451e-05,
      "loss": 0.044804222881793976,
      "step": 3790
    },
    {
      "epoch": 0.5125180609891085,
      "grad_norm": 0.7927709817886353,
      "learning_rate": 1.612485153137891e-05,
      "loss": 0.0602584145963192,
      "step": 3791
    },
    {
      "epoch": 0.5126532543578738,
      "grad_norm": 0.20674657821655273,
      "learning_rate": 1.611802625808323e-05,
      "loss": 0.03130554407835007,
      "step": 3792
    },
    {
      "epoch": 0.512788447726639,
      "grad_norm": 0.256592720746994,
      "learning_rate": 1.611120075200851e-05,
      "loss": 0.04670822620391846,
      "step": 3793
    },
    {
      "epoch": 0.5129236410954042,
      "grad_norm": 0.42805352807044983,
      "learning_rate": 1.610437501457587e-05,
      "loss": 0.04130428284406662,
      "step": 3794
    },
    {
      "epoch": 0.5130588344641696,
      "grad_norm": 0.4622546434402466,
      "learning_rate": 1.6097549047206464e-05,
      "loss": 0.0534285306930542,
      "step": 3795
    },
    {
      "epoch": 0.5131940278329348,
      "grad_norm": 0.2508544623851776,
      "learning_rate": 1.6090722851321497e-05,
      "loss": 0.05359993875026703,
      "step": 3796
    },
    {
      "epoch": 0.5133292212017001,
      "grad_norm": 0.7881059050559998,
      "learning_rate": 1.6083896428342213e-05,
      "loss": 0.05133604630827904,
      "step": 3797
    },
    {
      "epoch": 0.5134644145704653,
      "grad_norm": 0.5246431231498718,
      "learning_rate": 1.6077069779689915e-05,
      "loss": 0.07454442977905273,
      "step": 3798
    },
    {
      "epoch": 0.5135996079392305,
      "grad_norm": 0.42925789952278137,
      "learning_rate": 1.607024290678594e-05,
      "loss": 0.06013303995132446,
      "step": 3799
    },
    {
      "epoch": 0.5137348013079959,
      "grad_norm": 0.5546578168869019,
      "learning_rate": 1.6063415811051686e-05,
      "loss": 0.056167393922805786,
      "step": 3800
    },
    {
      "epoch": 0.5138699946767611,
      "grad_norm": 1.0981993675231934,
      "learning_rate": 1.6056588493908596e-05,
      "loss": 0.07775872945785522,
      "step": 3801
    },
    {
      "epoch": 0.5140051880455264,
      "grad_norm": 0.5263516902923584,
      "learning_rate": 1.604976095677814e-05,
      "loss": 0.05087617412209511,
      "step": 3802
    },
    {
      "epoch": 0.5141403814142916,
      "grad_norm": 0.4181288182735443,
      "learning_rate": 1.604293320108186e-05,
      "loss": 0.04873044788837433,
      "step": 3803
    },
    {
      "epoch": 0.5142755747830569,
      "grad_norm": 0.686302900314331,
      "learning_rate": 1.603610522824132e-05,
      "loss": 0.05752221494913101,
      "step": 3804
    },
    {
      "epoch": 0.5144107681518222,
      "grad_norm": 0.828981876373291,
      "learning_rate": 1.6029277039678153e-05,
      "loss": 0.05069413781166077,
      "step": 3805
    },
    {
      "epoch": 0.5145459615205874,
      "grad_norm": 0.5596721172332764,
      "learning_rate": 1.602244863681401e-05,
      "loss": 0.056078072637319565,
      "step": 3806
    },
    {
      "epoch": 0.5146811548893527,
      "grad_norm": 0.8351901769638062,
      "learning_rate": 1.6015620021070613e-05,
      "loss": 0.06332506239414215,
      "step": 3807
    },
    {
      "epoch": 0.514816348258118,
      "grad_norm": 0.535958468914032,
      "learning_rate": 1.6008791193869714e-05,
      "loss": 0.0573420524597168,
      "step": 3808
    },
    {
      "epoch": 0.5149515416268832,
      "grad_norm": 0.3675961196422577,
      "learning_rate": 1.6001962156633102e-05,
      "loss": 0.05918343365192413,
      "step": 3809
    },
    {
      "epoch": 0.5150867349956485,
      "grad_norm": 0.3773360848426819,
      "learning_rate": 1.5995132910782632e-05,
      "loss": 0.07070313394069672,
      "step": 3810
    },
    {
      "epoch": 0.5152219283644137,
      "grad_norm": 0.836632251739502,
      "learning_rate": 1.5988303457740178e-05,
      "loss": 0.06749022752046585,
      "step": 3811
    },
    {
      "epoch": 0.5153571217331789,
      "grad_norm": 0.5288264155387878,
      "learning_rate": 1.598147379892768e-05,
      "loss": 0.05578360706567764,
      "step": 3812
    },
    {
      "epoch": 0.5154923151019443,
      "grad_norm": 0.7572844624519348,
      "learning_rate": 1.5974643935767098e-05,
      "loss": 0.060728468000888824,
      "step": 3813
    },
    {
      "epoch": 0.5156275084707095,
      "grad_norm": 0.4169822037220001,
      "learning_rate": 1.5967813869680452e-05,
      "loss": 0.04759075492620468,
      "step": 3814
    },
    {
      "epoch": 0.5157627018394748,
      "grad_norm": 0.6279019713401794,
      "learning_rate": 1.59609836020898e-05,
      "loss": 0.05336674302816391,
      "step": 3815
    },
    {
      "epoch": 0.51589789520824,
      "grad_norm": 0.5039679408073425,
      "learning_rate": 1.5954153134417236e-05,
      "loss": 0.04822143539786339,
      "step": 3816
    },
    {
      "epoch": 0.5160330885770053,
      "grad_norm": 0.7431056499481201,
      "learning_rate": 1.59473224680849e-05,
      "loss": 0.04274154081940651,
      "step": 3817
    },
    {
      "epoch": 0.5161682819457706,
      "grad_norm": 0.43867871165275574,
      "learning_rate": 1.5940491604514976e-05,
      "loss": 0.05871080607175827,
      "step": 3818
    },
    {
      "epoch": 0.5163034753145358,
      "grad_norm": 0.29136285185813904,
      "learning_rate": 1.5933660545129683e-05,
      "loss": 0.04610811918973923,
      "step": 3819
    },
    {
      "epoch": 0.5164386686833011,
      "grad_norm": 0.37538301944732666,
      "learning_rate": 1.5926829291351288e-05,
      "loss": 0.05637273192405701,
      "step": 3820
    },
    {
      "epoch": 0.5165738620520663,
      "grad_norm": 0.4648245871067047,
      "learning_rate": 1.591999784460209e-05,
      "loss": 0.052748456597328186,
      "step": 3821
    },
    {
      "epoch": 0.5167090554208316,
      "grad_norm": 1.408056616783142,
      "learning_rate": 1.5913166206304435e-05,
      "loss": 0.056761935353279114,
      "step": 3822
    },
    {
      "epoch": 0.5168442487895969,
      "grad_norm": 1.2758344411849976,
      "learning_rate": 1.5906334377880707e-05,
      "loss": 0.0536779910326004,
      "step": 3823
    },
    {
      "epoch": 0.5169794421583621,
      "grad_norm": 1.120057463645935,
      "learning_rate": 1.589950236075333e-05,
      "loss": 0.05391278117895126,
      "step": 3824
    },
    {
      "epoch": 0.5171146355271274,
      "grad_norm": 0.9656091928482056,
      "learning_rate": 1.5892670156344764e-05,
      "loss": 0.055062390863895416,
      "step": 3825
    },
    {
      "epoch": 0.5172498288958927,
      "grad_norm": 0.8462112545967102,
      "learning_rate": 1.588583776607751e-05,
      "loss": 0.05114181339740753,
      "step": 3826
    },
    {
      "epoch": 0.5173850222646579,
      "grad_norm": 0.33182936906814575,
      "learning_rate": 1.5879005191374106e-05,
      "loss": 0.06080438196659088,
      "step": 3827
    },
    {
      "epoch": 0.5175202156334232,
      "grad_norm": 0.7704989314079285,
      "learning_rate": 1.587217243365714e-05,
      "loss": 0.0652606412768364,
      "step": 3828
    },
    {
      "epoch": 0.5176554090021884,
      "grad_norm": 0.9421502947807312,
      "learning_rate": 1.586533949434922e-05,
      "loss": 0.048460062593221664,
      "step": 3829
    },
    {
      "epoch": 0.5177906023709538,
      "grad_norm": 0.8190940022468567,
      "learning_rate": 1.5858506374872998e-05,
      "loss": 0.046723030507564545,
      "step": 3830
    },
    {
      "epoch": 0.517925795739719,
      "grad_norm": 0.9917197227478027,
      "learning_rate": 1.5851673076651178e-05,
      "loss": 0.057548344135284424,
      "step": 3831
    },
    {
      "epoch": 0.5180609891084842,
      "grad_norm": 0.9356064200401306,
      "learning_rate": 1.5844839601106477e-05,
      "loss": 0.06234029680490494,
      "step": 3832
    },
    {
      "epoch": 0.5181961824772495,
      "grad_norm": 0.3192106783390045,
      "learning_rate": 1.583800594966167e-05,
      "loss": 0.053430721163749695,
      "step": 3833
    },
    {
      "epoch": 0.5183313758460147,
      "grad_norm": 0.3324984312057495,
      "learning_rate": 1.583117212373955e-05,
      "loss": 0.06126798689365387,
      "step": 3834
    },
    {
      "epoch": 0.51846656921478,
      "grad_norm": 0.5636252164840698,
      "learning_rate": 1.5824338124762967e-05,
      "loss": 0.06817282736301422,
      "step": 3835
    },
    {
      "epoch": 0.5186017625835453,
      "grad_norm": 0.1898660510778427,
      "learning_rate": 1.581750395415479e-05,
      "loss": 0.0354049950838089,
      "step": 3836
    },
    {
      "epoch": 0.5187369559523105,
      "grad_norm": 0.7402545809745789,
      "learning_rate": 1.5810669613337922e-05,
      "loss": 0.053197041153907776,
      "step": 3837
    },
    {
      "epoch": 0.5188721493210758,
      "grad_norm": 0.2631089985370636,
      "learning_rate": 1.5803835103735327e-05,
      "loss": 0.04453294724225998,
      "step": 3838
    },
    {
      "epoch": 0.5190073426898411,
      "grad_norm": 0.5345631837844849,
      "learning_rate": 1.5797000426769973e-05,
      "loss": 0.04821440204977989,
      "step": 3839
    },
    {
      "epoch": 0.5191425360586063,
      "grad_norm": 0.5558984875679016,
      "learning_rate": 1.579016558386488e-05,
      "loss": 0.04820968210697174,
      "step": 3840
    },
    {
      "epoch": 0.5192777294273716,
      "grad_norm": 0.2662298381328583,
      "learning_rate": 1.5783330576443096e-05,
      "loss": 0.041163913905620575,
      "step": 3841
    },
    {
      "epoch": 0.5194129227961368,
      "grad_norm": 0.2704199552536011,
      "learning_rate": 1.5776495405927716e-05,
      "loss": 0.0519847609102726,
      "step": 3842
    },
    {
      "epoch": 0.5195481161649022,
      "grad_norm": 0.30000296235084534,
      "learning_rate": 1.5769660073741844e-05,
      "loss": 0.04474365711212158,
      "step": 3843
    },
    {
      "epoch": 0.5196833095336674,
      "grad_norm": 0.5938866138458252,
      "learning_rate": 1.5762824581308645e-05,
      "loss": 0.06992365419864655,
      "step": 3844
    },
    {
      "epoch": 0.5198185029024326,
      "grad_norm": 1.4845619201660156,
      "learning_rate": 1.5755988930051304e-05,
      "loss": 0.06222233176231384,
      "step": 3845
    },
    {
      "epoch": 0.5199536962711979,
      "grad_norm": 0.8205815553665161,
      "learning_rate": 1.5749153121393025e-05,
      "loss": 0.044046223163604736,
      "step": 3846
    },
    {
      "epoch": 0.5200888896399631,
      "grad_norm": 1.048462152481079,
      "learning_rate": 1.574231715675708e-05,
      "loss": 0.045677050948143005,
      "step": 3847
    },
    {
      "epoch": 0.5202240830087285,
      "grad_norm": 1.2512325048446655,
      "learning_rate": 1.573548103756674e-05,
      "loss": 0.056922078132629395,
      "step": 3848
    },
    {
      "epoch": 0.5203592763774937,
      "grad_norm": 0.6226654052734375,
      "learning_rate": 1.572864476524533e-05,
      "loss": 0.05940178036689758,
      "step": 3849
    },
    {
      "epoch": 0.5204944697462589,
      "grad_norm": 0.350815087556839,
      "learning_rate": 1.5721808341216195e-05,
      "loss": 0.06943237781524658,
      "step": 3850
    },
    {
      "epoch": 0.5206296631150242,
      "grad_norm": 0.8176279664039612,
      "learning_rate": 1.571497176690271e-05,
      "loss": 0.07380060851573944,
      "step": 3851
    },
    {
      "epoch": 0.5207648564837895,
      "grad_norm": 0.22101128101348877,
      "learning_rate": 1.570813504372829e-05,
      "loss": 0.03988974541425705,
      "step": 3852
    },
    {
      "epoch": 0.5209000498525548,
      "grad_norm": 0.41456660628318787,
      "learning_rate": 1.570129817311638e-05,
      "loss": 0.0563424788415432,
      "step": 3853
    },
    {
      "epoch": 0.52103524322132,
      "grad_norm": 0.328954815864563,
      "learning_rate": 1.5694461156490452e-05,
      "loss": 0.04352203756570816,
      "step": 3854
    },
    {
      "epoch": 0.5211704365900852,
      "grad_norm": 0.8576094508171082,
      "learning_rate": 1.5687623995274008e-05,
      "loss": 0.06292326003313065,
      "step": 3855
    },
    {
      "epoch": 0.5213056299588505,
      "grad_norm": 0.9740639925003052,
      "learning_rate": 1.568078669089058e-05,
      "loss": 0.07779178023338318,
      "step": 3856
    },
    {
      "epoch": 0.5214408233276158,
      "grad_norm": 0.6014882922172546,
      "learning_rate": 1.567394924476373e-05,
      "loss": 0.05543476343154907,
      "step": 3857
    },
    {
      "epoch": 0.5215760166963811,
      "grad_norm": 0.3653249442577362,
      "learning_rate": 1.5667111658317057e-05,
      "loss": 0.056862927973270416,
      "step": 3858
    },
    {
      "epoch": 0.5217112100651463,
      "grad_norm": 0.8777360320091248,
      "learning_rate": 1.5660273932974177e-05,
      "loss": 0.06048957258462906,
      "step": 3859
    },
    {
      "epoch": 0.5218464034339115,
      "grad_norm": 0.21385769546031952,
      "learning_rate": 1.5653436070158743e-05,
      "loss": 0.043873440474271774,
      "step": 3860
    },
    {
      "epoch": 0.5219815968026769,
      "grad_norm": 0.5791447758674622,
      "learning_rate": 1.564659807129444e-05,
      "loss": 0.05659090355038643,
      "step": 3861
    },
    {
      "epoch": 0.5221167901714421,
      "grad_norm": 0.532931387424469,
      "learning_rate": 1.5639759937804962e-05,
      "loss": 0.05180169641971588,
      "step": 3862
    },
    {
      "epoch": 0.5222519835402073,
      "grad_norm": 0.3157457411289215,
      "learning_rate": 1.5632921671114055e-05,
      "loss": 0.05484512448310852,
      "step": 3863
    },
    {
      "epoch": 0.5223871769089726,
      "grad_norm": 1.268995761871338,
      "learning_rate": 1.5626083272645485e-05,
      "loss": 0.06786813586950302,
      "step": 3864
    },
    {
      "epoch": 0.5225223702777378,
      "grad_norm": 0.676856279373169,
      "learning_rate": 1.5619244743823038e-05,
      "loss": 0.04725271090865135,
      "step": 3865
    },
    {
      "epoch": 0.5226575636465032,
      "grad_norm": 0.4009271264076233,
      "learning_rate": 1.5612406086070534e-05,
      "loss": 0.07125125080347061,
      "step": 3866
    },
    {
      "epoch": 0.5227927570152684,
      "grad_norm": 0.46584638953208923,
      "learning_rate": 1.560556730081181e-05,
      "loss": 0.06834045052528381,
      "step": 3867
    },
    {
      "epoch": 0.5229279503840336,
      "grad_norm": 0.38255366683006287,
      "learning_rate": 1.5598728389470754e-05,
      "loss": 0.047905176877975464,
      "step": 3868
    },
    {
      "epoch": 0.5230631437527989,
      "grad_norm": 0.824782133102417,
      "learning_rate": 1.5591889353471245e-05,
      "loss": 0.05604368448257446,
      "step": 3869
    },
    {
      "epoch": 0.5231983371215642,
      "grad_norm": 0.5316340327262878,
      "learning_rate": 1.5585050194237226e-05,
      "loss": 0.06252117455005646,
      "step": 3870
    },
    {
      "epoch": 0.5233335304903295,
      "grad_norm": 0.32712358236312866,
      "learning_rate": 1.557821091319263e-05,
      "loss": 0.0662146508693695,
      "step": 3871
    },
    {
      "epoch": 0.5234687238590947,
      "grad_norm": 0.4667937457561493,
      "learning_rate": 1.5571371511761446e-05,
      "loss": 0.053338099271059036,
      "step": 3872
    },
    {
      "epoch": 0.5236039172278599,
      "grad_norm": 0.6650628447532654,
      "learning_rate": 1.5564531991367658e-05,
      "loss": 0.06061176583170891,
      "step": 3873
    },
    {
      "epoch": 0.5237391105966253,
      "grad_norm": 1.3317973613739014,
      "learning_rate": 1.5557692353435302e-05,
      "loss": 0.0495610386133194,
      "step": 3874
    },
    {
      "epoch": 0.5238743039653905,
      "grad_norm": 0.6013226509094238,
      "learning_rate": 1.5550852599388424e-05,
      "loss": 0.0501331090927124,
      "step": 3875
    },
    {
      "epoch": 0.5240094973341558,
      "grad_norm": 2.2273976802825928,
      "learning_rate": 1.5544012730651096e-05,
      "loss": 0.06653538346290588,
      "step": 3876
    },
    {
      "epoch": 0.524144690702921,
      "grad_norm": 0.46187737584114075,
      "learning_rate": 1.5537172748647422e-05,
      "loss": 0.05617896467447281,
      "step": 3877
    },
    {
      "epoch": 0.5242798840716862,
      "grad_norm": 1.363012433052063,
      "learning_rate": 1.553033265480151e-05,
      "loss": 0.06546523422002792,
      "step": 3878
    },
    {
      "epoch": 0.5244150774404516,
      "grad_norm": 0.5969788432121277,
      "learning_rate": 1.552349245053752e-05,
      "loss": 0.05155107378959656,
      "step": 3879
    },
    {
      "epoch": 0.5245502708092168,
      "grad_norm": 0.8172316551208496,
      "learning_rate": 1.5516652137279597e-05,
      "loss": 0.06449680030345917,
      "step": 3880
    },
    {
      "epoch": 0.5246854641779821,
      "grad_norm": 0.7563415765762329,
      "learning_rate": 1.5509811716451955e-05,
      "loss": 0.050470590591430664,
      "step": 3881
    },
    {
      "epoch": 0.5248206575467473,
      "grad_norm": 0.46673232316970825,
      "learning_rate": 1.550297118947879e-05,
      "loss": 0.047976646572351456,
      "step": 3882
    },
    {
      "epoch": 0.5249558509155126,
      "grad_norm": 0.5018225908279419,
      "learning_rate": 1.5496130557784343e-05,
      "loss": 0.047616541385650635,
      "step": 3883
    },
    {
      "epoch": 0.5250910442842779,
      "grad_norm": 0.4791119396686554,
      "learning_rate": 1.5489289822792868e-05,
      "loss": 0.06036540865898132,
      "step": 3884
    },
    {
      "epoch": 0.5252262376530431,
      "grad_norm": 0.1800708770751953,
      "learning_rate": 1.5482448985928645e-05,
      "loss": 0.03271171823143959,
      "step": 3885
    },
    {
      "epoch": 0.5253614310218084,
      "grad_norm": 0.9041771292686462,
      "learning_rate": 1.5475608048615964e-05,
      "loss": 0.05772767215967178,
      "step": 3886
    },
    {
      "epoch": 0.5254966243905737,
      "grad_norm": 0.7105149626731873,
      "learning_rate": 1.546876701227916e-05,
      "loss": 0.05694102868437767,
      "step": 3887
    },
    {
      "epoch": 0.5256318177593389,
      "grad_norm": 0.34959179162979126,
      "learning_rate": 1.5461925878342558e-05,
      "loss": 0.037605538964271545,
      "step": 3888
    },
    {
      "epoch": 0.5257670111281042,
      "grad_norm": 1.0307186841964722,
      "learning_rate": 1.5455084648230527e-05,
      "loss": 0.062323808670043945,
      "step": 3889
    },
    {
      "epoch": 0.5259022044968694,
      "grad_norm": 0.23543967306613922,
      "learning_rate": 1.5448243323367438e-05,
      "loss": 0.044301412999629974,
      "step": 3890
    },
    {
      "epoch": 0.5260373978656346,
      "grad_norm": 0.3705684244632721,
      "learning_rate": 1.544140190517771e-05,
      "loss": 0.03656876087188721,
      "step": 3891
    },
    {
      "epoch": 0.5261725912344,
      "grad_norm": 0.8742043375968933,
      "learning_rate": 1.5434560395085745e-05,
      "loss": 0.05393455922603607,
      "step": 3892
    },
    {
      "epoch": 0.5263077846031652,
      "grad_norm": 0.8725761771202087,
      "learning_rate": 1.542771879451599e-05,
      "loss": 0.04833026975393295,
      "step": 3893
    },
    {
      "epoch": 0.5264429779719305,
      "grad_norm": 0.4318239092826843,
      "learning_rate": 1.54208771048929e-05,
      "loss": 0.04526226222515106,
      "step": 3894
    },
    {
      "epoch": 0.5265781713406957,
      "grad_norm": 1.3377554416656494,
      "learning_rate": 1.5414035327640958e-05,
      "loss": 0.053397081792354584,
      "step": 3895
    },
    {
      "epoch": 0.526713364709461,
      "grad_norm": 0.7858668565750122,
      "learning_rate": 1.5407193464184644e-05,
      "loss": 0.0437445342540741,
      "step": 3896
    },
    {
      "epoch": 0.5268485580782263,
      "grad_norm": 0.47253498435020447,
      "learning_rate": 1.5400351515948485e-05,
      "loss": 0.04661727696657181,
      "step": 3897
    },
    {
      "epoch": 0.5269837514469915,
      "grad_norm": 0.7696034908294678,
      "learning_rate": 1.5393509484357006e-05,
      "loss": 0.04863014817237854,
      "step": 3898
    },
    {
      "epoch": 0.5271189448157568,
      "grad_norm": 0.43488770723342896,
      "learning_rate": 1.538666737083475e-05,
      "loss": 0.0425231009721756,
      "step": 3899
    },
    {
      "epoch": 0.527254138184522,
      "grad_norm": 0.5269473195075989,
      "learning_rate": 1.537982517680629e-05,
      "loss": 0.04516787454485893,
      "step": 3900
    },
    {
      "epoch": 0.5273893315532873,
      "grad_norm": 0.7998220920562744,
      "learning_rate": 1.5372982903696196e-05,
      "loss": 0.05101688206195831,
      "step": 3901
    },
    {
      "epoch": 0.5275245249220526,
      "grad_norm": 0.9070618152618408,
      "learning_rate": 1.536614055292908e-05,
      "loss": 0.06188303232192993,
      "step": 3902
    },
    {
      "epoch": 0.5276597182908178,
      "grad_norm": 0.7143345475196838,
      "learning_rate": 1.535929812592955e-05,
      "loss": 0.06838368624448776,
      "step": 3903
    },
    {
      "epoch": 0.5277949116595831,
      "grad_norm": 1.5905269384384155,
      "learning_rate": 1.5352455624122227e-05,
      "loss": 0.07274457812309265,
      "step": 3904
    },
    {
      "epoch": 0.5279301050283484,
      "grad_norm": 0.5470003485679626,
      "learning_rate": 1.5345613048931765e-05,
      "loss": 0.04788441210985184,
      "step": 3905
    },
    {
      "epoch": 0.5280652983971136,
      "grad_norm": 0.8070629835128784,
      "learning_rate": 1.5338770401782822e-05,
      "loss": 0.05735981464385986,
      "step": 3906
    },
    {
      "epoch": 0.5282004917658789,
      "grad_norm": 0.3525625467300415,
      "learning_rate": 1.5331927684100077e-05,
      "loss": 0.07135386765003204,
      "step": 3907
    },
    {
      "epoch": 0.5283356851346441,
      "grad_norm": 0.46959879994392395,
      "learning_rate": 1.5325084897308218e-05,
      "loss": 0.05447271093726158,
      "step": 3908
    },
    {
      "epoch": 0.5284708785034095,
      "grad_norm": 0.5100918412208557,
      "learning_rate": 1.5318242042831952e-05,
      "loss": 0.05446285381913185,
      "step": 3909
    },
    {
      "epoch": 0.5286060718721747,
      "grad_norm": 1.1584619283676147,
      "learning_rate": 1.5311399122095992e-05,
      "loss": 0.04878786951303482,
      "step": 3910
    },
    {
      "epoch": 0.5287412652409399,
      "grad_norm": 1.0253782272338867,
      "learning_rate": 1.5304556136525074e-05,
      "loss": 0.05952572822570801,
      "step": 3911
    },
    {
      "epoch": 0.5288764586097052,
      "grad_norm": 0.5966041088104248,
      "learning_rate": 1.5297713087543948e-05,
      "loss": 0.03753407299518585,
      "step": 3912
    },
    {
      "epoch": 0.5290116519784704,
      "grad_norm": 0.8773438930511475,
      "learning_rate": 1.5290869976577365e-05,
      "loss": 0.04544057324528694,
      "step": 3913
    },
    {
      "epoch": 0.5291468453472358,
      "grad_norm": 0.44492125511169434,
      "learning_rate": 1.5284026805050107e-05,
      "loss": 0.05192258581519127,
      "step": 3914
    },
    {
      "epoch": 0.529282038716001,
      "grad_norm": 0.36962461471557617,
      "learning_rate": 1.5277183574386947e-05,
      "loss": 0.05602163076400757,
      "step": 3915
    },
    {
      "epoch": 0.5294172320847662,
      "grad_norm": 0.6350026726722717,
      "learning_rate": 1.5270340286012694e-05,
      "loss": 0.04901233687996864,
      "step": 3916
    },
    {
      "epoch": 0.5295524254535315,
      "grad_norm": 0.38578498363494873,
      "learning_rate": 1.526349694135215e-05,
      "loss": 0.054621048271656036,
      "step": 3917
    },
    {
      "epoch": 0.5296876188222968,
      "grad_norm": 0.3361757695674896,
      "learning_rate": 1.525665354183014e-05,
      "loss": 0.05293537676334381,
      "step": 3918
    },
    {
      "epoch": 0.529822812191062,
      "grad_norm": 0.39618435502052307,
      "learning_rate": 1.5249810088871493e-05,
      "loss": 0.06588000804185867,
      "step": 3919
    },
    {
      "epoch": 0.5299580055598273,
      "grad_norm": 0.5319632887840271,
      "learning_rate": 1.5242966583901052e-05,
      "loss": 0.03069375641644001,
      "step": 3920
    },
    {
      "epoch": 0.5300931989285925,
      "grad_norm": 0.2962198257446289,
      "learning_rate": 1.523612302834367e-05,
      "loss": 0.06402204930782318,
      "step": 3921
    },
    {
      "epoch": 0.5302283922973579,
      "grad_norm": 0.954925537109375,
      "learning_rate": 1.5229279423624217e-05,
      "loss": 0.06296830624341965,
      "step": 3922
    },
    {
      "epoch": 0.5303635856661231,
      "grad_norm": 0.34819495677948,
      "learning_rate": 1.5222435771167566e-05,
      "loss": 0.05464529991149902,
      "step": 3923
    },
    {
      "epoch": 0.5304987790348883,
      "grad_norm": 0.9567537307739258,
      "learning_rate": 1.5215592072398602e-05,
      "loss": 0.05670039355754852,
      "step": 3924
    },
    {
      "epoch": 0.5306339724036536,
      "grad_norm": 0.297484815120697,
      "learning_rate": 1.520874832874222e-05,
      "loss": 0.033783864229917526,
      "step": 3925
    },
    {
      "epoch": 0.5307691657724188,
      "grad_norm": 0.29705309867858887,
      "learning_rate": 1.5201904541623318e-05,
      "loss": 0.04822036623954773,
      "step": 3926
    },
    {
      "epoch": 0.5309043591411842,
      "grad_norm": 0.8676610589027405,
      "learning_rate": 1.5195060712466817e-05,
      "loss": 0.07909068465232849,
      "step": 3927
    },
    {
      "epoch": 0.5310395525099494,
      "grad_norm": 0.4062187373638153,
      "learning_rate": 1.5188216842697635e-05,
      "loss": 0.06048871576786041,
      "step": 3928
    },
    {
      "epoch": 0.5311747458787146,
      "grad_norm": 0.6600865125656128,
      "learning_rate": 1.5181372933740703e-05,
      "loss": 0.05761503800749779,
      "step": 3929
    },
    {
      "epoch": 0.5313099392474799,
      "grad_norm": 0.4390881061553955,
      "learning_rate": 1.5174528987020958e-05,
      "loss": 0.058613844215869904,
      "step": 3930
    },
    {
      "epoch": 0.5314451326162452,
      "grad_norm": 0.9420432448387146,
      "learning_rate": 1.5167685003963345e-05,
      "loss": 0.04771383851766586,
      "step": 3931
    },
    {
      "epoch": 0.5315803259850105,
      "grad_norm": 0.6272133588790894,
      "learning_rate": 1.5160840985992824e-05,
      "loss": 0.059007227420806885,
      "step": 3932
    },
    {
      "epoch": 0.5317155193537757,
      "grad_norm": 0.8019857406616211,
      "learning_rate": 1.515399693453435e-05,
      "loss": 0.05623268336057663,
      "step": 3933
    },
    {
      "epoch": 0.5318507127225409,
      "grad_norm": 0.4517199397087097,
      "learning_rate": 1.5147152851012894e-05,
      "loss": 0.04688534513115883,
      "step": 3934
    },
    {
      "epoch": 0.5319859060913062,
      "grad_norm": 0.5166458487510681,
      "learning_rate": 1.514030873685343e-05,
      "loss": 0.05230247601866722,
      "step": 3935
    },
    {
      "epoch": 0.5321210994600715,
      "grad_norm": 0.9699565172195435,
      "learning_rate": 1.513346459348094e-05,
      "loss": 0.06472494453191757,
      "step": 3936
    },
    {
      "epoch": 0.5322562928288368,
      "grad_norm": 0.8696311712265015,
      "learning_rate": 1.5126620422320405e-05,
      "loss": 0.058809250593185425,
      "step": 3937
    },
    {
      "epoch": 0.532391486197602,
      "grad_norm": 0.8052372336387634,
      "learning_rate": 1.5119776224796823e-05,
      "loss": 0.056269899010658264,
      "step": 3938
    },
    {
      "epoch": 0.5325266795663672,
      "grad_norm": 0.33290237188339233,
      "learning_rate": 1.5112932002335195e-05,
      "loss": 0.04272053390741348,
      "step": 3939
    },
    {
      "epoch": 0.5326618729351326,
      "grad_norm": 0.36302196979522705,
      "learning_rate": 1.5106087756360524e-05,
      "loss": 0.046406738460063934,
      "step": 3940
    },
    {
      "epoch": 0.5327970663038978,
      "grad_norm": 1.064314365386963,
      "learning_rate": 1.5099243488297816e-05,
      "loss": 0.05604291334748268,
      "step": 3941
    },
    {
      "epoch": 0.5329322596726631,
      "grad_norm": 0.49924832582473755,
      "learning_rate": 1.5092399199572083e-05,
      "loss": 0.05543386936187744,
      "step": 3942
    },
    {
      "epoch": 0.5330674530414283,
      "grad_norm": 0.39729470014572144,
      "learning_rate": 1.5085554891608343e-05,
      "loss": 0.05348895490169525,
      "step": 3943
    },
    {
      "epoch": 0.5332026464101935,
      "grad_norm": 0.6028935313224792,
      "learning_rate": 1.5078710565831616e-05,
      "loss": 0.05534949526190758,
      "step": 3944
    },
    {
      "epoch": 0.5333378397789589,
      "grad_norm": 0.5769929885864258,
      "learning_rate": 1.5071866223666935e-05,
      "loss": 0.04693952202796936,
      "step": 3945
    },
    {
      "epoch": 0.5334730331477241,
      "grad_norm": 0.7933577299118042,
      "learning_rate": 1.5065021866539323e-05,
      "loss": 0.05774712562561035,
      "step": 3946
    },
    {
      "epoch": 0.5336082265164893,
      "grad_norm": 0.5810902714729309,
      "learning_rate": 1.5058177495873805e-05,
      "loss": 0.04253597557544708,
      "step": 3947
    },
    {
      "epoch": 0.5337434198852546,
      "grad_norm": 0.3926798403263092,
      "learning_rate": 1.5051333113095429e-05,
      "loss": 0.05249135196208954,
      "step": 3948
    },
    {
      "epoch": 0.5338786132540199,
      "grad_norm": 0.4805140197277069,
      "learning_rate": 1.5044488719629218e-05,
      "loss": 0.047764845192432404,
      "step": 3949
    },
    {
      "epoch": 0.5340138066227852,
      "grad_norm": 0.3548891842365265,
      "learning_rate": 1.5037644316900227e-05,
      "loss": 0.051260799169540405,
      "step": 3950
    },
    {
      "epoch": 0.5341489999915504,
      "grad_norm": 0.37100163102149963,
      "learning_rate": 1.5030799906333484e-05,
      "loss": 0.04891455918550491,
      "step": 3951
    },
    {
      "epoch": 0.5342841933603156,
      "grad_norm": 0.7228818535804749,
      "learning_rate": 1.5023955489354031e-05,
      "loss": 0.06493876129388809,
      "step": 3952
    },
    {
      "epoch": 0.534419386729081,
      "grad_norm": 0.30967745184898376,
      "learning_rate": 1.5017111067386927e-05,
      "loss": 0.05843615531921387,
      "step": 3953
    },
    {
      "epoch": 0.5345545800978462,
      "grad_norm": 0.8512394428253174,
      "learning_rate": 1.50102666418572e-05,
      "loss": 0.06520982831716537,
      "step": 3954
    },
    {
      "epoch": 0.5346897734666115,
      "grad_norm": 0.8132785558700562,
      "learning_rate": 1.500342221418991e-05,
      "loss": 0.062483467161655426,
      "step": 3955
    },
    {
      "epoch": 0.5348249668353767,
      "grad_norm": 0.30524805188179016,
      "learning_rate": 1.4996577785810094e-05,
      "loss": 0.047003597021102905,
      "step": 3956
    },
    {
      "epoch": 0.5349601602041419,
      "grad_norm": 0.39114096760749817,
      "learning_rate": 1.4989733358142798e-05,
      "loss": 0.05080056190490723,
      "step": 3957
    },
    {
      "epoch": 0.5350953535729073,
      "grad_norm": 1.2146002054214478,
      "learning_rate": 1.498288893261308e-05,
      "loss": 0.05616268515586853,
      "step": 3958
    },
    {
      "epoch": 0.5352305469416725,
      "grad_norm": 0.5430983304977417,
      "learning_rate": 1.497604451064597e-05,
      "loss": 0.04470429569482803,
      "step": 3959
    },
    {
      "epoch": 0.5353657403104378,
      "grad_norm": 0.7658754587173462,
      "learning_rate": 1.496920009366652e-05,
      "loss": 0.06324173510074615,
      "step": 3960
    },
    {
      "epoch": 0.535500933679203,
      "grad_norm": 0.657473623752594,
      "learning_rate": 1.4962355683099777e-05,
      "loss": 0.05880952626466751,
      "step": 3961
    },
    {
      "epoch": 0.5356361270479683,
      "grad_norm": 0.4107910692691803,
      "learning_rate": 1.4955511280370782e-05,
      "loss": 0.04643833637237549,
      "step": 3962
    },
    {
      "epoch": 0.5357713204167336,
      "grad_norm": 0.25113463401794434,
      "learning_rate": 1.4948666886904579e-05,
      "loss": 0.05110061168670654,
      "step": 3963
    },
    {
      "epoch": 0.5359065137854988,
      "grad_norm": 0.2991109788417816,
      "learning_rate": 1.4941822504126199e-05,
      "loss": 0.04806319624185562,
      "step": 3964
    },
    {
      "epoch": 0.5360417071542641,
      "grad_norm": 0.47618919610977173,
      "learning_rate": 1.4934978133460681e-05,
      "loss": 0.050580233335494995,
      "step": 3965
    },
    {
      "epoch": 0.5361769005230294,
      "grad_norm": 0.7184830904006958,
      "learning_rate": 1.4928133776333068e-05,
      "loss": 0.049489960074424744,
      "step": 3966
    },
    {
      "epoch": 0.5363120938917946,
      "grad_norm": 0.3012215793132782,
      "learning_rate": 1.4921289434168386e-05,
      "loss": 0.043583422899246216,
      "step": 3967
    },
    {
      "epoch": 0.5364472872605599,
      "grad_norm": 0.8425238728523254,
      "learning_rate": 1.4914445108391663e-05,
      "loss": 0.06295725703239441,
      "step": 3968
    },
    {
      "epoch": 0.5365824806293251,
      "grad_norm": 0.412889301776886,
      "learning_rate": 1.4907600800427922e-05,
      "loss": 0.04477260261774063,
      "step": 3969
    },
    {
      "epoch": 0.5367176739980903,
      "grad_norm": 0.4890596270561218,
      "learning_rate": 1.4900756511702188e-05,
      "loss": 0.06844127178192139,
      "step": 3970
    },
    {
      "epoch": 0.5368528673668557,
      "grad_norm": 0.42125043272972107,
      "learning_rate": 1.4893912243639479e-05,
      "loss": 0.05592568218708038,
      "step": 3971
    },
    {
      "epoch": 0.5369880607356209,
      "grad_norm": 0.7336945533752441,
      "learning_rate": 1.4887067997664807e-05,
      "loss": 0.07025749981403351,
      "step": 3972
    },
    {
      "epoch": 0.5371232541043862,
      "grad_norm": 0.2617323100566864,
      "learning_rate": 1.488022377520318e-05,
      "loss": 0.052128762006759644,
      "step": 3973
    },
    {
      "epoch": 0.5372584474731514,
      "grad_norm": 0.6597862839698792,
      "learning_rate": 1.4873379577679599e-05,
      "loss": 0.04164543002843857,
      "step": 3974
    },
    {
      "epoch": 0.5373936408419167,
      "grad_norm": 0.927537739276886,
      "learning_rate": 1.4866535406519063e-05,
      "loss": 0.051501937210559845,
      "step": 3975
    },
    {
      "epoch": 0.537528834210682,
      "grad_norm": 0.31450849771499634,
      "learning_rate": 1.4859691263146574e-05,
      "loss": 0.04818488657474518,
      "step": 3976
    },
    {
      "epoch": 0.5376640275794472,
      "grad_norm": 0.447421669960022,
      "learning_rate": 1.485284714898711e-05,
      "loss": 0.05986437201499939,
      "step": 3977
    },
    {
      "epoch": 0.5377992209482125,
      "grad_norm": 1.162524700164795,
      "learning_rate": 1.4846003065465653e-05,
      "loss": 0.05748840421438217,
      "step": 3978
    },
    {
      "epoch": 0.5379344143169777,
      "grad_norm": 0.34466612339019775,
      "learning_rate": 1.4839159014007179e-05,
      "loss": 0.06284038722515106,
      "step": 3979
    },
    {
      "epoch": 0.538069607685743,
      "grad_norm": 0.7355208396911621,
      "learning_rate": 1.4832314996036653e-05,
      "loss": 0.04395366832613945,
      "step": 3980
    },
    {
      "epoch": 0.5382048010545083,
      "grad_norm": 0.4821547865867615,
      "learning_rate": 1.4825471012979048e-05,
      "loss": 0.05633664131164551,
      "step": 3981
    },
    {
      "epoch": 0.5383399944232735,
      "grad_norm": 0.6286754608154297,
      "learning_rate": 1.4818627066259301e-05,
      "loss": 0.0601787269115448,
      "step": 3982
    },
    {
      "epoch": 0.5384751877920388,
      "grad_norm": 0.37904462218284607,
      "learning_rate": 1.481178315730237e-05,
      "loss": 0.051263660192489624,
      "step": 3983
    },
    {
      "epoch": 0.5386103811608041,
      "grad_norm": 0.39313796162605286,
      "learning_rate": 1.4804939287533184e-05,
      "loss": 0.061426207423210144,
      "step": 3984
    },
    {
      "epoch": 0.5387455745295693,
      "grad_norm": 0.9317488074302673,
      "learning_rate": 1.4798095458376682e-05,
      "loss": 0.04274756461381912,
      "step": 3985
    },
    {
      "epoch": 0.5388807678983346,
      "grad_norm": 0.48566317558288574,
      "learning_rate": 1.4791251671257788e-05,
      "loss": 0.052697841078042984,
      "step": 3986
    },
    {
      "epoch": 0.5390159612670998,
      "grad_norm": 0.3606877326965332,
      "learning_rate": 1.4784407927601401e-05,
      "loss": 0.05537360906600952,
      "step": 3987
    },
    {
      "epoch": 0.5391511546358652,
      "grad_norm": 0.5394943952560425,
      "learning_rate": 1.4777564228832436e-05,
      "loss": 0.05346938967704773,
      "step": 3988
    },
    {
      "epoch": 0.5392863480046304,
      "grad_norm": 0.35650384426116943,
      "learning_rate": 1.4770720576375782e-05,
      "loss": 0.03862364962697029,
      "step": 3989
    },
    {
      "epoch": 0.5394215413733956,
      "grad_norm": 0.6105693578720093,
      "learning_rate": 1.4763876971656334e-05,
      "loss": 0.03420288488268852,
      "step": 3990
    },
    {
      "epoch": 0.5395567347421609,
      "grad_norm": 0.8025833964347839,
      "learning_rate": 1.4757033416098953e-05,
      "loss": 0.06326080858707428,
      "step": 3991
    },
    {
      "epoch": 0.5396919281109261,
      "grad_norm": 1.293094515800476,
      "learning_rate": 1.4750189911128511e-05,
      "loss": 0.06910311430692673,
      "step": 3992
    },
    {
      "epoch": 0.5398271214796915,
      "grad_norm": 0.5423516035079956,
      "learning_rate": 1.4743346458169863e-05,
      "loss": 0.05523611977696419,
      "step": 3993
    },
    {
      "epoch": 0.5399623148484567,
      "grad_norm": 0.5392700433731079,
      "learning_rate": 1.473650305864785e-05,
      "loss": 0.06953607499599457,
      "step": 3994
    },
    {
      "epoch": 0.5400975082172219,
      "grad_norm": 0.9406944513320923,
      "learning_rate": 1.472965971398731e-05,
      "loss": 0.0425386056303978,
      "step": 3995
    },
    {
      "epoch": 0.5402327015859872,
      "grad_norm": 0.35616856813430786,
      "learning_rate": 1.4722816425613054e-05,
      "loss": 0.05764736235141754,
      "step": 3996
    },
    {
      "epoch": 0.5403678949547525,
      "grad_norm": 0.3001463711261749,
      "learning_rate": 1.4715973194949895e-05,
      "loss": 0.03597217798233032,
      "step": 3997
    },
    {
      "epoch": 0.5405030883235177,
      "grad_norm": 0.31074583530426025,
      "learning_rate": 1.4709130023422636e-05,
      "loss": 0.04479756951332092,
      "step": 3998
    },
    {
      "epoch": 0.540638281692283,
      "grad_norm": 0.8563055396080017,
      "learning_rate": 1.4702286912456052e-05,
      "loss": 0.06081032752990723,
      "step": 3999
    },
    {
      "epoch": 0.5407734750610482,
      "grad_norm": 0.49117857217788696,
      "learning_rate": 1.4695443863474928e-05,
      "loss": 0.05906405299901962,
      "step": 4000
    },
    {
      "epoch": 0.5409086684298136,
      "grad_norm": 1.1349149942398071,
      "learning_rate": 1.4688600877904012e-05,
      "loss": 0.05307549238204956,
      "step": 4001
    },
    {
      "epoch": 0.5410438617985788,
      "grad_norm": 0.9400823712348938,
      "learning_rate": 1.468175795716805e-05,
      "loss": 0.06698894500732422,
      "step": 4002
    },
    {
      "epoch": 0.541179055167344,
      "grad_norm": 0.47587287425994873,
      "learning_rate": 1.4674915102691783e-05,
      "loss": 0.043459322303533554,
      "step": 4003
    },
    {
      "epoch": 0.5413142485361093,
      "grad_norm": 0.4782063663005829,
      "learning_rate": 1.4668072315899926e-05,
      "loss": 0.03177499771118164,
      "step": 4004
    },
    {
      "epoch": 0.5414494419048745,
      "grad_norm": 0.324441522359848,
      "learning_rate": 1.466122959821718e-05,
      "loss": 0.05222472548484802,
      "step": 4005
    },
    {
      "epoch": 0.5415846352736399,
      "grad_norm": 0.3802015781402588,
      "learning_rate": 1.4654386951068239e-05,
      "loss": 0.04731915518641472,
      "step": 4006
    },
    {
      "epoch": 0.5417198286424051,
      "grad_norm": 0.49423736333847046,
      "learning_rate": 1.4647544375877776e-05,
      "loss": 0.059997059404850006,
      "step": 4007
    },
    {
      "epoch": 0.5418550220111703,
      "grad_norm": 0.6311784386634827,
      "learning_rate": 1.4640701874070457e-05,
      "loss": 0.04727725684642792,
      "step": 4008
    },
    {
      "epoch": 0.5419902153799356,
      "grad_norm": 0.487170934677124,
      "learning_rate": 1.4633859447070922e-05,
      "loss": 0.04510696232318878,
      "step": 4009
    },
    {
      "epoch": 0.5421254087487009,
      "grad_norm": 0.6538010239601135,
      "learning_rate": 1.4627017096303805e-05,
      "loss": 0.048280373215675354,
      "step": 4010
    },
    {
      "epoch": 0.5422606021174662,
      "grad_norm": 0.7923305630683899,
      "learning_rate": 1.4620174823193711e-05,
      "loss": 0.04995451495051384,
      "step": 4011
    },
    {
      "epoch": 0.5423957954862314,
      "grad_norm": 0.6805353164672852,
      "learning_rate": 1.4613332629165249e-05,
      "loss": 0.05478479713201523,
      "step": 4012
    },
    {
      "epoch": 0.5425309888549966,
      "grad_norm": 1.2362356185913086,
      "learning_rate": 1.4606490515642998e-05,
      "loss": 0.06356494873762131,
      "step": 4013
    },
    {
      "epoch": 0.542666182223762,
      "grad_norm": 1.2735474109649658,
      "learning_rate": 1.4599648484051516e-05,
      "loss": 0.050921857357025146,
      "step": 4014
    },
    {
      "epoch": 0.5428013755925272,
      "grad_norm": 0.62109375,
      "learning_rate": 1.4592806535815358e-05,
      "loss": 0.05046012997627258,
      "step": 4015
    },
    {
      "epoch": 0.5429365689612925,
      "grad_norm": 0.3577922582626343,
      "learning_rate": 1.4585964672359045e-05,
      "loss": 0.06369826942682266,
      "step": 4016
    },
    {
      "epoch": 0.5430717623300577,
      "grad_norm": 0.5299736261367798,
      "learning_rate": 1.4579122895107098e-05,
      "loss": 0.05683884024620056,
      "step": 4017
    },
    {
      "epoch": 0.5432069556988229,
      "grad_norm": 1.1091710329055786,
      "learning_rate": 1.4572281205484012e-05,
      "loss": 0.053758978843688965,
      "step": 4018
    },
    {
      "epoch": 0.5433421490675883,
      "grad_norm": 1.0474162101745605,
      "learning_rate": 1.4565439604914256e-05,
      "loss": 0.05532175302505493,
      "step": 4019
    },
    {
      "epoch": 0.5434773424363535,
      "grad_norm": 0.9889752864837646,
      "learning_rate": 1.4558598094822294e-05,
      "loss": 0.0699634999036789,
      "step": 4020
    },
    {
      "epoch": 0.5436125358051188,
      "grad_norm": 0.7981903553009033,
      "learning_rate": 1.455175667663256e-05,
      "loss": 0.05378991737961769,
      "step": 4021
    },
    {
      "epoch": 0.543747729173884,
      "grad_norm": 0.4481160342693329,
      "learning_rate": 1.4544915351769476e-05,
      "loss": 0.04631450027227402,
      "step": 4022
    },
    {
      "epoch": 0.5438829225426492,
      "grad_norm": 0.574632465839386,
      "learning_rate": 1.4538074121657448e-05,
      "loss": 0.06112110614776611,
      "step": 4023
    },
    {
      "epoch": 0.5440181159114146,
      "grad_norm": 0.33643460273742676,
      "learning_rate": 1.4531232987720846e-05,
      "loss": 0.05754278600215912,
      "step": 4024
    },
    {
      "epoch": 0.5441533092801798,
      "grad_norm": 0.4023594260215759,
      "learning_rate": 1.4524391951384037e-05,
      "loss": 0.050510138273239136,
      "step": 4025
    },
    {
      "epoch": 0.544288502648945,
      "grad_norm": 0.47239968180656433,
      "learning_rate": 1.4517551014071358e-05,
      "loss": 0.053018130362033844,
      "step": 4026
    },
    {
      "epoch": 0.5444236960177103,
      "grad_norm": 1.0239540338516235,
      "learning_rate": 1.4510710177207137e-05,
      "loss": 0.050805479288101196,
      "step": 4027
    },
    {
      "epoch": 0.5445588893864756,
      "grad_norm": 1.0416115522384644,
      "learning_rate": 1.450386944221566e-05,
      "loss": 0.05809459090232849,
      "step": 4028
    },
    {
      "epoch": 0.5446940827552409,
      "grad_norm": 0.36053913831710815,
      "learning_rate": 1.449702881052121e-05,
      "loss": 0.055181846022605896,
      "step": 4029
    },
    {
      "epoch": 0.5448292761240061,
      "grad_norm": 0.7667855620384216,
      "learning_rate": 1.4490188283548048e-05,
      "loss": 0.060374438762664795,
      "step": 4030
    },
    {
      "epoch": 0.5449644694927713,
      "grad_norm": 0.495286226272583,
      "learning_rate": 1.44833478627204e-05,
      "loss": 0.0642457827925682,
      "step": 4031
    },
    {
      "epoch": 0.5450996628615367,
      "grad_norm": 0.4716796278953552,
      "learning_rate": 1.447650754946249e-05,
      "loss": 0.04884020984172821,
      "step": 4032
    },
    {
      "epoch": 0.5452348562303019,
      "grad_norm": 0.6766912937164307,
      "learning_rate": 1.4469667345198492e-05,
      "loss": 0.05094345286488533,
      "step": 4033
    },
    {
      "epoch": 0.5453700495990672,
      "grad_norm": 0.8531404733657837,
      "learning_rate": 1.446282725135258e-05,
      "loss": 0.03966781497001648,
      "step": 4034
    },
    {
      "epoch": 0.5455052429678324,
      "grad_norm": 0.5359408855438232,
      "learning_rate": 1.4455987269348904e-05,
      "loss": 0.056910835206508636,
      "step": 4035
    },
    {
      "epoch": 0.5456404363365976,
      "grad_norm": 0.49176695942878723,
      "learning_rate": 1.4449147400611578e-05,
      "loss": 0.053131937980651855,
      "step": 4036
    },
    {
      "epoch": 0.545775629705363,
      "grad_norm": 0.8822224140167236,
      "learning_rate": 1.4442307646564702e-05,
      "loss": 0.07119302451610565,
      "step": 4037
    },
    {
      "epoch": 0.5459108230741282,
      "grad_norm": 0.3242127299308777,
      "learning_rate": 1.4435468008632345e-05,
      "loss": 0.04253949224948883,
      "step": 4038
    },
    {
      "epoch": 0.5460460164428935,
      "grad_norm": 0.9133155941963196,
      "learning_rate": 1.4428628488238557e-05,
      "loss": 0.057325027883052826,
      "step": 4039
    },
    {
      "epoch": 0.5461812098116587,
      "grad_norm": 0.8766014575958252,
      "learning_rate": 1.442178908680737e-05,
      "loss": 0.06670939922332764,
      "step": 4040
    },
    {
      "epoch": 0.546316403180424,
      "grad_norm": 0.5208431482315063,
      "learning_rate": 1.4414949805762779e-05,
      "loss": 0.0591048002243042,
      "step": 4041
    },
    {
      "epoch": 0.5464515965491893,
      "grad_norm": 0.5918677449226379,
      "learning_rate": 1.4408110646528757e-05,
      "loss": 0.0653669536113739,
      "step": 4042
    },
    {
      "epoch": 0.5465867899179545,
      "grad_norm": 1.5518032312393188,
      "learning_rate": 1.440127161052925e-05,
      "loss": 0.05806749686598778,
      "step": 4043
    },
    {
      "epoch": 0.5467219832867198,
      "grad_norm": 0.4439472258090973,
      "learning_rate": 1.4394432699188188e-05,
      "loss": 0.05808722972869873,
      "step": 4044
    },
    {
      "epoch": 0.546857176655485,
      "grad_norm": 0.40956586599349976,
      "learning_rate": 1.4387593913929472e-05,
      "loss": 0.03346915543079376,
      "step": 4045
    },
    {
      "epoch": 0.5469923700242503,
      "grad_norm": 0.6225788593292236,
      "learning_rate": 1.4380755256176968e-05,
      "loss": 0.045644327998161316,
      "step": 4046
    },
    {
      "epoch": 0.5471275633930156,
      "grad_norm": 0.4095410108566284,
      "learning_rate": 1.437391672735452e-05,
      "loss": 0.05311070382595062,
      "step": 4047
    },
    {
      "epoch": 0.5472627567617808,
      "grad_norm": 0.4994106888771057,
      "learning_rate": 1.4367078328885946e-05,
      "loss": 0.06590622663497925,
      "step": 4048
    },
    {
      "epoch": 0.5473979501305462,
      "grad_norm": 0.4182433784008026,
      "learning_rate": 1.4360240062195039e-05,
      "loss": 0.07823541760444641,
      "step": 4049
    },
    {
      "epoch": 0.5475331434993114,
      "grad_norm": 0.36532631516456604,
      "learning_rate": 1.435340192870557e-05,
      "loss": 0.04127059876918793,
      "step": 4050
    },
    {
      "epoch": 0.5476683368680766,
      "grad_norm": 0.3086000978946686,
      "learning_rate": 1.434656392984126e-05,
      "loss": 0.03841794654726982,
      "step": 4051
    },
    {
      "epoch": 0.5478035302368419,
      "grad_norm": 0.5577488541603088,
      "learning_rate": 1.4339726067025828e-05,
      "loss": 0.05866488441824913,
      "step": 4052
    },
    {
      "epoch": 0.5479387236056071,
      "grad_norm": 0.4661047160625458,
      "learning_rate": 1.4332888341682947e-05,
      "loss": 0.04983644187450409,
      "step": 4053
    },
    {
      "epoch": 0.5480739169743724,
      "grad_norm": 0.40975701808929443,
      "learning_rate": 1.432605075523627e-05,
      "loss": 0.04896017909049988,
      "step": 4054
    },
    {
      "epoch": 0.5482091103431377,
      "grad_norm": 0.3015366792678833,
      "learning_rate": 1.4319213309109426e-05,
      "loss": 0.03304576501250267,
      "step": 4055
    },
    {
      "epoch": 0.5483443037119029,
      "grad_norm": 0.3199126124382019,
      "learning_rate": 1.4312376004725996e-05,
      "loss": 0.06221427768468857,
      "step": 4056
    },
    {
      "epoch": 0.5484794970806682,
      "grad_norm": 0.2792492210865021,
      "learning_rate": 1.430553884350955e-05,
      "loss": 0.05618011951446533,
      "step": 4057
    },
    {
      "epoch": 0.5486146904494335,
      "grad_norm": 0.23022061586380005,
      "learning_rate": 1.429870182688362e-05,
      "loss": 0.03981427103281021,
      "step": 4058
    },
    {
      "epoch": 0.5487498838181987,
      "grad_norm": 0.49077630043029785,
      "learning_rate": 1.4291864956271713e-05,
      "loss": 0.06821377575397491,
      "step": 4059
    },
    {
      "epoch": 0.548885077186964,
      "grad_norm": 0.6557405591011047,
      "learning_rate": 1.4285028233097293e-05,
      "loss": 0.06908242404460907,
      "step": 4060
    },
    {
      "epoch": 0.5490202705557292,
      "grad_norm": 0.960667073726654,
      "learning_rate": 1.4278191658783809e-05,
      "loss": 0.04825650155544281,
      "step": 4061
    },
    {
      "epoch": 0.5491554639244945,
      "grad_norm": 0.7569752335548401,
      "learning_rate": 1.427135523475467e-05,
      "loss": 0.0507570281624794,
      "step": 4062
    },
    {
      "epoch": 0.5492906572932598,
      "grad_norm": 0.3151237368583679,
      "learning_rate": 1.4264518962433258e-05,
      "loss": 0.052069079130887985,
      "step": 4063
    },
    {
      "epoch": 0.549425850662025,
      "grad_norm": 0.6898719668388367,
      "learning_rate": 1.4257682843242925e-05,
      "loss": 0.04433870315551758,
      "step": 4064
    },
    {
      "epoch": 0.5495610440307903,
      "grad_norm": 0.6530208587646484,
      "learning_rate": 1.4250846878606974e-05,
      "loss": 0.06546321511268616,
      "step": 4065
    },
    {
      "epoch": 0.5496962373995555,
      "grad_norm": 1.4969708919525146,
      "learning_rate": 1.4244011069948702e-05,
      "loss": 0.0656256452202797,
      "step": 4066
    },
    {
      "epoch": 0.5498314307683209,
      "grad_norm": 0.31266719102859497,
      "learning_rate": 1.4237175418691357e-05,
      "loss": 0.03966697305440903,
      "step": 4067
    },
    {
      "epoch": 0.5499666241370861,
      "grad_norm": 0.9247994422912598,
      "learning_rate": 1.4230339926258153e-05,
      "loss": 0.04583561420440674,
      "step": 4068
    },
    {
      "epoch": 0.5501018175058513,
      "grad_norm": 1.2100938558578491,
      "learning_rate": 1.422350459407229e-05,
      "loss": 0.04688480496406555,
      "step": 4069
    },
    {
      "epoch": 0.5502370108746166,
      "grad_norm": 0.440973699092865,
      "learning_rate": 1.4216669423556903e-05,
      "loss": 0.04519513621926308,
      "step": 4070
    },
    {
      "epoch": 0.5503722042433818,
      "grad_norm": 0.34939002990722656,
      "learning_rate": 1.420983441613512e-05,
      "loss": 0.042144447565078735,
      "step": 4071
    },
    {
      "epoch": 0.5505073976121472,
      "grad_norm": 0.7764574885368347,
      "learning_rate": 1.420299957323003e-05,
      "loss": 0.04697326570749283,
      "step": 4072
    },
    {
      "epoch": 0.5506425909809124,
      "grad_norm": 0.5128805041313171,
      "learning_rate": 1.4196164896264679e-05,
      "loss": 0.04348839819431305,
      "step": 4073
    },
    {
      "epoch": 0.5507777843496776,
      "grad_norm": 0.5639934539794922,
      "learning_rate": 1.418933038666208e-05,
      "loss": 0.06003371626138687,
      "step": 4074
    },
    {
      "epoch": 0.5509129777184429,
      "grad_norm": 0.3731740415096283,
      "learning_rate": 1.4182496045845217e-05,
      "loss": 0.04616701602935791,
      "step": 4075
    },
    {
      "epoch": 0.5510481710872082,
      "grad_norm": 0.5516098141670227,
      "learning_rate": 1.4175661875237036e-05,
      "loss": 0.05356122553348541,
      "step": 4076
    },
    {
      "epoch": 0.5511833644559734,
      "grad_norm": 0.7479949593544006,
      "learning_rate": 1.416882787626045e-05,
      "loss": 0.054915063083171844,
      "step": 4077
    },
    {
      "epoch": 0.5513185578247387,
      "grad_norm": 0.2697214186191559,
      "learning_rate": 1.4161994050338334e-05,
      "loss": 0.042996957898139954,
      "step": 4078
    },
    {
      "epoch": 0.5514537511935039,
      "grad_norm": 0.9078285098075867,
      "learning_rate": 1.4155160398893528e-05,
      "loss": 0.04638843983411789,
      "step": 4079
    },
    {
      "epoch": 0.5515889445622693,
      "grad_norm": 0.2647183835506439,
      "learning_rate": 1.4148326923348824e-05,
      "loss": 0.05164635181427002,
      "step": 4080
    },
    {
      "epoch": 0.5517241379310345,
      "grad_norm": 0.4467652440071106,
      "learning_rate": 1.4141493625127e-05,
      "loss": 0.06272059679031372,
      "step": 4081
    },
    {
      "epoch": 0.5518593312997997,
      "grad_norm": 0.24048303067684174,
      "learning_rate": 1.4134660505650786e-05,
      "loss": 0.03398372232913971,
      "step": 4082
    },
    {
      "epoch": 0.551994524668565,
      "grad_norm": 0.6048710942268372,
      "learning_rate": 1.4127827566342864e-05,
      "loss": 0.05136989802122116,
      "step": 4083
    },
    {
      "epoch": 0.5521297180373302,
      "grad_norm": 0.8332582712173462,
      "learning_rate": 1.4120994808625896e-05,
      "loss": 0.0650918185710907,
      "step": 4084
    },
    {
      "epoch": 0.5522649114060956,
      "grad_norm": 0.635187029838562,
      "learning_rate": 1.4114162233922494e-05,
      "loss": 0.053848572075366974,
      "step": 4085
    },
    {
      "epoch": 0.5524001047748608,
      "grad_norm": 0.9322073459625244,
      "learning_rate": 1.4107329843655238e-05,
      "loss": 0.07745838165283203,
      "step": 4086
    },
    {
      "epoch": 0.552535298143626,
      "grad_norm": 0.5019484758377075,
      "learning_rate": 1.4100497639246675e-05,
      "loss": 0.04726524278521538,
      "step": 4087
    },
    {
      "epoch": 0.5526704915123913,
      "grad_norm": 0.7686334252357483,
      "learning_rate": 1.4093665622119294e-05,
      "loss": 0.043291181325912476,
      "step": 4088
    },
    {
      "epoch": 0.5528056848811566,
      "grad_norm": 0.7255486249923706,
      "learning_rate": 1.4086833793695566e-05,
      "loss": 0.050574421882629395,
      "step": 4089
    },
    {
      "epoch": 0.5529408782499219,
      "grad_norm": 1.2916117906570435,
      "learning_rate": 1.408000215539791e-05,
      "loss": 0.05484726279973984,
      "step": 4090
    },
    {
      "epoch": 0.5530760716186871,
      "grad_norm": 0.3538680970668793,
      "learning_rate": 1.4073170708648711e-05,
      "loss": 0.06455840170383453,
      "step": 4091
    },
    {
      "epoch": 0.5532112649874523,
      "grad_norm": 0.8253899216651917,
      "learning_rate": 1.406633945487032e-05,
      "loss": 0.049087703227996826,
      "step": 4092
    },
    {
      "epoch": 0.5533464583562177,
      "grad_norm": 0.4992747902870178,
      "learning_rate": 1.4059508395485026e-05,
      "loss": 0.056317687034606934,
      "step": 4093
    },
    {
      "epoch": 0.5534816517249829,
      "grad_norm": 0.8482896685600281,
      "learning_rate": 1.4052677531915102e-05,
      "loss": 0.05254112184047699,
      "step": 4094
    },
    {
      "epoch": 0.5536168450937482,
      "grad_norm": 0.37596696615219116,
      "learning_rate": 1.4045846865582765e-05,
      "loss": 0.0493338480591774,
      "step": 4095
    },
    {
      "epoch": 0.5537520384625134,
      "grad_norm": 0.41362515091896057,
      "learning_rate": 1.4039016397910206e-05,
      "loss": 0.05972599983215332,
      "step": 4096
    },
    {
      "epoch": 0.5538872318312786,
      "grad_norm": 0.7611149549484253,
      "learning_rate": 1.403218613031955e-05,
      "loss": 0.05557525157928467,
      "step": 4097
    },
    {
      "epoch": 0.554022425200044,
      "grad_norm": 0.324356347322464,
      "learning_rate": 1.4025356064232903e-05,
      "loss": 0.036259621381759644,
      "step": 4098
    },
    {
      "epoch": 0.5541576185688092,
      "grad_norm": 0.5125876665115356,
      "learning_rate": 1.4018526201072324e-05,
      "loss": 0.05409299582242966,
      "step": 4099
    },
    {
      "epoch": 0.5542928119375745,
      "grad_norm": 0.420535147190094,
      "learning_rate": 1.4011696542259821e-05,
      "loss": 0.05342576652765274,
      "step": 4100
    },
    {
      "epoch": 0.5544280053063397,
      "grad_norm": 1.3484312295913696,
      "learning_rate": 1.4004867089217376e-05,
      "loss": 0.07324740290641785,
      "step": 4101
    },
    {
      "epoch": 0.554563198675105,
      "grad_norm": 1.4609699249267578,
      "learning_rate": 1.39980378433669e-05,
      "loss": 0.05767257511615753,
      "step": 4102
    },
    {
      "epoch": 0.5546983920438703,
      "grad_norm": 0.4036857783794403,
      "learning_rate": 1.399120880613029e-05,
      "loss": 0.050574153661727905,
      "step": 4103
    },
    {
      "epoch": 0.5548335854126355,
      "grad_norm": 1.92477548122406,
      "learning_rate": 1.3984379978929388e-05,
      "loss": 0.06461501121520996,
      "step": 4104
    },
    {
      "epoch": 0.5549687787814007,
      "grad_norm": 0.9905716776847839,
      "learning_rate": 1.3977551363185995e-05,
      "loss": 0.04796414077281952,
      "step": 4105
    },
    {
      "epoch": 0.555103972150166,
      "grad_norm": 1.1557170152664185,
      "learning_rate": 1.3970722960321854e-05,
      "loss": 0.05784928798675537,
      "step": 4106
    },
    {
      "epoch": 0.5552391655189313,
      "grad_norm": 0.7333866953849792,
      "learning_rate": 1.3963894771758682e-05,
      "loss": 0.0469755083322525,
      "step": 4107
    },
    {
      "epoch": 0.5553743588876966,
      "grad_norm": 0.7916989922523499,
      "learning_rate": 1.3957066798918143e-05,
      "loss": 0.04061264544725418,
      "step": 4108
    },
    {
      "epoch": 0.5555095522564618,
      "grad_norm": 0.9357433319091797,
      "learning_rate": 1.3950239043221861e-05,
      "loss": 0.050264425575733185,
      "step": 4109
    },
    {
      "epoch": 0.555644745625227,
      "grad_norm": 0.5857399106025696,
      "learning_rate": 1.3943411506091408e-05,
      "loss": 0.047240953892469406,
      "step": 4110
    },
    {
      "epoch": 0.5557799389939924,
      "grad_norm": 1.3813725709915161,
      "learning_rate": 1.3936584188948313e-05,
      "loss": 0.0697888657450676,
      "step": 4111
    },
    {
      "epoch": 0.5559151323627576,
      "grad_norm": 0.35123321413993835,
      "learning_rate": 1.3929757093214059e-05,
      "loss": 0.05244212597608566,
      "step": 4112
    },
    {
      "epoch": 0.5560503257315229,
      "grad_norm": 0.3247286081314087,
      "learning_rate": 1.3922930220310085e-05,
      "loss": 0.05216085910797119,
      "step": 4113
    },
    {
      "epoch": 0.5561855191002881,
      "grad_norm": 0.2647315561771393,
      "learning_rate": 1.3916103571657786e-05,
      "loss": 0.050299856811761856,
      "step": 4114
    },
    {
      "epoch": 0.5563207124690533,
      "grad_norm": 0.8535521626472473,
      "learning_rate": 1.3909277148678504e-05,
      "loss": 0.07653629034757614,
      "step": 4115
    },
    {
      "epoch": 0.5564559058378187,
      "grad_norm": 0.38206785917282104,
      "learning_rate": 1.3902450952793536e-05,
      "loss": 0.0732155591249466,
      "step": 4116
    },
    {
      "epoch": 0.5565910992065839,
      "grad_norm": 0.32354500889778137,
      "learning_rate": 1.389562498542413e-05,
      "loss": 0.049967825412750244,
      "step": 4117
    },
    {
      "epoch": 0.5567262925753492,
      "grad_norm": 0.5827561616897583,
      "learning_rate": 1.388879924799149e-05,
      "loss": 0.06330039352178574,
      "step": 4118
    },
    {
      "epoch": 0.5568614859441144,
      "grad_norm": 0.39095014333724976,
      "learning_rate": 1.388197374191678e-05,
      "loss": 0.03535168245434761,
      "step": 4119
    },
    {
      "epoch": 0.5569966793128797,
      "grad_norm": 0.3703196942806244,
      "learning_rate": 1.387514846862109e-05,
      "loss": 0.0664469376206398,
      "step": 4120
    },
    {
      "epoch": 0.557131872681645,
      "grad_norm": 0.6275187134742737,
      "learning_rate": 1.3868323429525492e-05,
      "loss": 0.04316431283950806,
      "step": 4121
    },
    {
      "epoch": 0.5572670660504102,
      "grad_norm": 0.3375455141067505,
      "learning_rate": 1.3861498626050986e-05,
      "loss": 0.05484852194786072,
      "step": 4122
    },
    {
      "epoch": 0.5574022594191755,
      "grad_norm": 0.6091442704200745,
      "learning_rate": 1.385467405961854e-05,
      "loss": 0.06739459931850433,
      "step": 4123
    },
    {
      "epoch": 0.5575374527879408,
      "grad_norm": 0.5915157794952393,
      "learning_rate": 1.3847849731649066e-05,
      "loss": 0.04694291576743126,
      "step": 4124
    },
    {
      "epoch": 0.557672646156706,
      "grad_norm": 0.2952704131603241,
      "learning_rate": 1.3841025643563418e-05,
      "loss": 0.04867354780435562,
      "step": 4125
    },
    {
      "epoch": 0.5578078395254713,
      "grad_norm": 0.3743056058883667,
      "learning_rate": 1.3834201796782413e-05,
      "loss": 0.04590238630771637,
      "step": 4126
    },
    {
      "epoch": 0.5579430328942365,
      "grad_norm": 0.2780226171016693,
      "learning_rate": 1.3827378192726808e-05,
      "loss": 0.04500555992126465,
      "step": 4127
    },
    {
      "epoch": 0.5580782262630019,
      "grad_norm": 0.626263439655304,
      "learning_rate": 1.3820554832817324e-05,
      "loss": 0.064442940056324,
      "step": 4128
    },
    {
      "epoch": 0.5582134196317671,
      "grad_norm": 0.7143221497535706,
      "learning_rate": 1.3813731718474606e-05,
      "loss": 0.05304088816046715,
      "step": 4129
    },
    {
      "epoch": 0.5583486130005323,
      "grad_norm": 0.44622063636779785,
      "learning_rate": 1.380690885111927e-05,
      "loss": 0.03524869680404663,
      "step": 4130
    },
    {
      "epoch": 0.5584838063692976,
      "grad_norm": 0.9775217771530151,
      "learning_rate": 1.3800086232171877e-05,
      "loss": 0.06411316990852356,
      "step": 4131
    },
    {
      "epoch": 0.5586189997380628,
      "grad_norm": 0.3309904634952545,
      "learning_rate": 1.3793263863052926e-05,
      "loss": 0.04349002242088318,
      "step": 4132
    },
    {
      "epoch": 0.5587541931068281,
      "grad_norm": 0.9526672959327698,
      "learning_rate": 1.3786441745182881e-05,
      "loss": 0.060160089284181595,
      "step": 4133
    },
    {
      "epoch": 0.5588893864755934,
      "grad_norm": 0.3098883628845215,
      "learning_rate": 1.3779619879982127e-05,
      "loss": 0.03340590000152588,
      "step": 4134
    },
    {
      "epoch": 0.5590245798443586,
      "grad_norm": 0.34839949011802673,
      "learning_rate": 1.3772798268871025e-05,
      "loss": 0.057577282190322876,
      "step": 4135
    },
    {
      "epoch": 0.5591597732131239,
      "grad_norm": 0.5307559967041016,
      "learning_rate": 1.376597691326987e-05,
      "loss": 0.04966269060969353,
      "step": 4136
    },
    {
      "epoch": 0.5592949665818892,
      "grad_norm": 0.6379033923149109,
      "learning_rate": 1.3759155814598898e-05,
      "loss": 0.06234271824359894,
      "step": 4137
    },
    {
      "epoch": 0.5594301599506544,
      "grad_norm": 0.7114551067352295,
      "learning_rate": 1.3752334974278308e-05,
      "loss": 0.05135638266801834,
      "step": 4138
    },
    {
      "epoch": 0.5595653533194197,
      "grad_norm": 0.7957108020782471,
      "learning_rate": 1.3745514393728225e-05,
      "loss": 0.050775811076164246,
      "step": 4139
    },
    {
      "epoch": 0.5597005466881849,
      "grad_norm": 0.6730691194534302,
      "learning_rate": 1.3738694074368735e-05,
      "loss": 0.05369863659143448,
      "step": 4140
    },
    {
      "epoch": 0.5598357400569502,
      "grad_norm": 0.772809624671936,
      "learning_rate": 1.3731874017619868e-05,
      "loss": 0.06065630167722702,
      "step": 4141
    },
    {
      "epoch": 0.5599709334257155,
      "grad_norm": 0.5261940360069275,
      "learning_rate": 1.3725054224901597e-05,
      "loss": 0.055207014083862305,
      "step": 4142
    },
    {
      "epoch": 0.5601061267944807,
      "grad_norm": 0.7934703826904297,
      "learning_rate": 1.3718234697633826e-05,
      "loss": 0.04808718338608742,
      "step": 4143
    },
    {
      "epoch": 0.560241320163246,
      "grad_norm": 0.7563589811325073,
      "learning_rate": 1.3711415437236427e-05,
      "loss": 0.059947334229946136,
      "step": 4144
    },
    {
      "epoch": 0.5603765135320112,
      "grad_norm": 0.7628953456878662,
      "learning_rate": 1.3704596445129207e-05,
      "loss": 0.057378850877285004,
      "step": 4145
    },
    {
      "epoch": 0.5605117069007766,
      "grad_norm": 0.3508414924144745,
      "learning_rate": 1.369777772273192e-05,
      "loss": 0.03536544740200043,
      "step": 4146
    },
    {
      "epoch": 0.5606469002695418,
      "grad_norm": 0.35734426975250244,
      "learning_rate": 1.369095927146425e-05,
      "loss": 0.0586828775703907,
      "step": 4147
    },
    {
      "epoch": 0.560782093638307,
      "grad_norm": 0.4771082103252411,
      "learning_rate": 1.3684141092745846e-05,
      "loss": 0.05103515088558197,
      "step": 4148
    },
    {
      "epoch": 0.5609172870070723,
      "grad_norm": 0.5046601295471191,
      "learning_rate": 1.3677323187996276e-05,
      "loss": 0.05139821767807007,
      "step": 4149
    },
    {
      "epoch": 0.5610524803758375,
      "grad_norm": 0.7221264243125916,
      "learning_rate": 1.3670505558635074e-05,
      "loss": 0.04857698827981949,
      "step": 4150
    },
    {
      "epoch": 0.5611876737446029,
      "grad_norm": 1.0628623962402344,
      "learning_rate": 1.366368820608171e-05,
      "loss": 0.06827078759670258,
      "step": 4151
    },
    {
      "epoch": 0.5613228671133681,
      "grad_norm": 0.27733466029167175,
      "learning_rate": 1.365687113175558e-05,
      "loss": 0.04822686314582825,
      "step": 4152
    },
    {
      "epoch": 0.5614580604821333,
      "grad_norm": 0.5919431447982788,
      "learning_rate": 1.3650054337076049e-05,
      "loss": 0.05639851093292236,
      "step": 4153
    },
    {
      "epoch": 0.5615932538508986,
      "grad_norm": 0.9658674001693726,
      "learning_rate": 1.3643237823462398e-05,
      "loss": 0.042504988610744476,
      "step": 4154
    },
    {
      "epoch": 0.5617284472196639,
      "grad_norm": 1.4385346174240112,
      "learning_rate": 1.363642159233387e-05,
      "loss": 0.06431318074464798,
      "step": 4155
    },
    {
      "epoch": 0.5618636405884292,
      "grad_norm": 0.37629932165145874,
      "learning_rate": 1.3629605645109642e-05,
      "loss": 0.042373064905405045,
      "step": 4156
    },
    {
      "epoch": 0.5619988339571944,
      "grad_norm": 0.8330971002578735,
      "learning_rate": 1.362278998320882e-05,
      "loss": 0.053013384342193604,
      "step": 4157
    },
    {
      "epoch": 0.5621340273259596,
      "grad_norm": 0.5067549347877502,
      "learning_rate": 1.3615974608050472e-05,
      "loss": 0.049661993980407715,
      "step": 4158
    },
    {
      "epoch": 0.562269220694725,
      "grad_norm": 1.273038387298584,
      "learning_rate": 1.3609159521053588e-05,
      "loss": 0.06212637573480606,
      "step": 4159
    },
    {
      "epoch": 0.5624044140634902,
      "grad_norm": 0.7922593355178833,
      "learning_rate": 1.3602344723637107e-05,
      "loss": 0.06535595655441284,
      "step": 4160
    },
    {
      "epoch": 0.5625396074322554,
      "grad_norm": 0.3038403391838074,
      "learning_rate": 1.3595530217219916e-05,
      "loss": 0.045795708894729614,
      "step": 4161
    },
    {
      "epoch": 0.5626748008010207,
      "grad_norm": 0.6399083137512207,
      "learning_rate": 1.3588716003220815e-05,
      "loss": 0.07586899399757385,
      "step": 4162
    },
    {
      "epoch": 0.5628099941697859,
      "grad_norm": 0.4581215977668762,
      "learning_rate": 1.3581902083058574e-05,
      "loss": 0.05741545557975769,
      "step": 4163
    },
    {
      "epoch": 0.5629451875385513,
      "grad_norm": 0.9531058669090271,
      "learning_rate": 1.3575088458151877e-05,
      "loss": 0.06747080385684967,
      "step": 4164
    },
    {
      "epoch": 0.5630803809073165,
      "grad_norm": 0.30632415413856506,
      "learning_rate": 1.3568275129919367e-05,
      "loss": 0.048171043395996094,
      "step": 4165
    },
    {
      "epoch": 0.5632155742760817,
      "grad_norm": 0.6023296117782593,
      "learning_rate": 1.3561462099779604e-05,
      "loss": 0.04519319534301758,
      "step": 4166
    },
    {
      "epoch": 0.563350767644847,
      "grad_norm": 0.4950801730155945,
      "learning_rate": 1.3554649369151104e-05,
      "loss": 0.05491878092288971,
      "step": 4167
    },
    {
      "epoch": 0.5634859610136123,
      "grad_norm": 0.6327624320983887,
      "learning_rate": 1.3547836939452315e-05,
      "loss": 0.04467090964317322,
      "step": 4168
    },
    {
      "epoch": 0.5636211543823776,
      "grad_norm": 0.959592342376709,
      "learning_rate": 1.3541024812101615e-05,
      "loss": 0.04802456498146057,
      "step": 4169
    },
    {
      "epoch": 0.5637563477511428,
      "grad_norm": 0.4934079349040985,
      "learning_rate": 1.3534212988517339e-05,
      "loss": 0.04421760141849518,
      "step": 4170
    },
    {
      "epoch": 0.563891541119908,
      "grad_norm": 1.5135420560836792,
      "learning_rate": 1.3527401470117726e-05,
      "loss": 0.06407340615987778,
      "step": 4171
    },
    {
      "epoch": 0.5640267344886734,
      "grad_norm": 0.2907327711582184,
      "learning_rate": 1.3520590258320981e-05,
      "loss": 0.04562245309352875,
      "step": 4172
    },
    {
      "epoch": 0.5641619278574386,
      "grad_norm": 1.0602513551712036,
      "learning_rate": 1.3513779354545235e-05,
      "loss": 0.06544183194637299,
      "step": 4173
    },
    {
      "epoch": 0.5642971212262039,
      "grad_norm": 0.6522163152694702,
      "learning_rate": 1.3506968760208557e-05,
      "loss": 0.04684177786111832,
      "step": 4174
    },
    {
      "epoch": 0.5644323145949691,
      "grad_norm": 0.6463688611984253,
      "learning_rate": 1.3500158476728938e-05,
      "loss": 0.07147551327943802,
      "step": 4175
    },
    {
      "epoch": 0.5645675079637343,
      "grad_norm": 0.585493803024292,
      "learning_rate": 1.3493348505524325e-05,
      "loss": 0.05106959864497185,
      "step": 4176
    },
    {
      "epoch": 0.5647027013324997,
      "grad_norm": 0.34694379568099976,
      "learning_rate": 1.3486538848012586e-05,
      "loss": 0.051603272557258606,
      "step": 4177
    },
    {
      "epoch": 0.5648378947012649,
      "grad_norm": 0.47639647126197815,
      "learning_rate": 1.3479729505611532e-05,
      "loss": 0.046169739216566086,
      "step": 4178
    },
    {
      "epoch": 0.5649730880700302,
      "grad_norm": 1.6975427865982056,
      "learning_rate": 1.3472920479738906e-05,
      "loss": 0.05696733295917511,
      "step": 4179
    },
    {
      "epoch": 0.5651082814387954,
      "grad_norm": 0.2580471932888031,
      "learning_rate": 1.346611177181237e-05,
      "loss": 0.050407782196998596,
      "step": 4180
    },
    {
      "epoch": 0.5652434748075607,
      "grad_norm": 0.3292336165904999,
      "learning_rate": 1.3459303383249547e-05,
      "loss": 0.053844138979911804,
      "step": 4181
    },
    {
      "epoch": 0.565378668176326,
      "grad_norm": 0.8269696831703186,
      "learning_rate": 1.3452495315467975e-05,
      "loss": 0.0390135832130909,
      "step": 4182
    },
    {
      "epoch": 0.5655138615450912,
      "grad_norm": 0.5612631440162659,
      "learning_rate": 1.3445687569885132e-05,
      "loss": 0.054399967193603516,
      "step": 4183
    },
    {
      "epoch": 0.5656490549138565,
      "grad_norm": 0.4437982738018036,
      "learning_rate": 1.3438880147918429e-05,
      "loss": 0.04522149637341499,
      "step": 4184
    },
    {
      "epoch": 0.5657842482826217,
      "grad_norm": 0.661548376083374,
      "learning_rate": 1.3432073050985201e-05,
      "loss": 0.05173173546791077,
      "step": 4185
    },
    {
      "epoch": 0.565919441651387,
      "grad_norm": 1.1240564584732056,
      "learning_rate": 1.3425266280502721e-05,
      "loss": 0.06077219173312187,
      "step": 4186
    },
    {
      "epoch": 0.5660546350201523,
      "grad_norm": 1.235144853591919,
      "learning_rate": 1.3418459837888202e-05,
      "loss": 0.06470440328121185,
      "step": 4187
    },
    {
      "epoch": 0.5661898283889175,
      "grad_norm": 0.2538681626319885,
      "learning_rate": 1.3411653724558784e-05,
      "loss": 0.05350903421640396,
      "step": 4188
    },
    {
      "epoch": 0.5663250217576827,
      "grad_norm": 0.5879151225090027,
      "learning_rate": 1.3404847941931523e-05,
      "loss": 0.0470278300344944,
      "step": 4189
    },
    {
      "epoch": 0.5664602151264481,
      "grad_norm": 1.0408427715301514,
      "learning_rate": 1.339804249142343e-05,
      "loss": 0.0490865521132946,
      "step": 4190
    },
    {
      "epoch": 0.5665954084952133,
      "grad_norm": 0.8385155200958252,
      "learning_rate": 1.3391237374451429e-05,
      "loss": 0.045292384922504425,
      "step": 4191
    },
    {
      "epoch": 0.5667306018639786,
      "grad_norm": 0.19643595814704895,
      "learning_rate": 1.3384432592432388e-05,
      "loss": 0.03735601156949997,
      "step": 4192
    },
    {
      "epoch": 0.5668657952327438,
      "grad_norm": 0.4666518270969391,
      "learning_rate": 1.3377628146783102e-05,
      "loss": 0.054945189505815506,
      "step": 4193
    },
    {
      "epoch": 0.567000988601509,
      "grad_norm": 0.9669063091278076,
      "learning_rate": 1.3370824038920281e-05,
      "loss": 0.06678605824708939,
      "step": 4194
    },
    {
      "epoch": 0.5671361819702744,
      "grad_norm": 0.9080616235733032,
      "learning_rate": 1.3364020270260586e-05,
      "loss": 0.06193765252828598,
      "step": 4195
    },
    {
      "epoch": 0.5672713753390396,
      "grad_norm": 0.6309118270874023,
      "learning_rate": 1.335721684222059e-05,
      "loss": 0.04887032508850098,
      "step": 4196
    },
    {
      "epoch": 0.5674065687078049,
      "grad_norm": 0.5722773671150208,
      "learning_rate": 1.3350413756216816e-05,
      "loss": 0.043982818722724915,
      "step": 4197
    },
    {
      "epoch": 0.5675417620765701,
      "grad_norm": 1.1884140968322754,
      "learning_rate": 1.334361101366569e-05,
      "loss": 0.06711280345916748,
      "step": 4198
    },
    {
      "epoch": 0.5676769554453354,
      "grad_norm": 0.22038215398788452,
      "learning_rate": 1.3336808615983582e-05,
      "loss": 0.03942631930112839,
      "step": 4199
    },
    {
      "epoch": 0.5678121488141007,
      "grad_norm": 1.034661054611206,
      "learning_rate": 1.3330006564586791e-05,
      "loss": 0.058673225343227386,
      "step": 4200
    },
    {
      "epoch": 0.5679473421828659,
      "grad_norm": 0.8482556939125061,
      "learning_rate": 1.3323204860891539e-05,
      "loss": 0.04450428485870361,
      "step": 4201
    },
    {
      "epoch": 0.5680825355516312,
      "grad_norm": 0.31858378648757935,
      "learning_rate": 1.3316403506313981e-05,
      "loss": 0.04843887314200401,
      "step": 4202
    },
    {
      "epoch": 0.5682177289203965,
      "grad_norm": 1.3350555896759033,
      "learning_rate": 1.3309602502270184e-05,
      "loss": 0.07182306051254272,
      "step": 4203
    },
    {
      "epoch": 0.5683529222891617,
      "grad_norm": 0.6230509281158447,
      "learning_rate": 1.3302801850176161e-05,
      "loss": 0.060498617589473724,
      "step": 4204
    },
    {
      "epoch": 0.568488115657927,
      "grad_norm": 0.33955833315849304,
      "learning_rate": 1.3296001551447848e-05,
      "loss": 0.060212284326553345,
      "step": 4205
    },
    {
      "epoch": 0.5686233090266922,
      "grad_norm": 0.33457639813423157,
      "learning_rate": 1.32892016075011e-05,
      "loss": 0.05924059823155403,
      "step": 4206
    },
    {
      "epoch": 0.5687585023954576,
      "grad_norm": 0.43424177169799805,
      "learning_rate": 1.3282402019751694e-05,
      "loss": 0.05278228223323822,
      "step": 4207
    },
    {
      "epoch": 0.5688936957642228,
      "grad_norm": 0.4012652039527893,
      "learning_rate": 1.327560278961535e-05,
      "loss": 0.05098462477326393,
      "step": 4208
    },
    {
      "epoch": 0.569028889132988,
      "grad_norm": 0.365847647190094,
      "learning_rate": 1.3268803918507699e-05,
      "loss": 0.06091932952404022,
      "step": 4209
    },
    {
      "epoch": 0.5691640825017533,
      "grad_norm": 1.8767354488372803,
      "learning_rate": 1.3262005407844306e-05,
      "loss": 0.06962120532989502,
      "step": 4210
    },
    {
      "epoch": 0.5692992758705185,
      "grad_norm": 1.3309123516082764,
      "learning_rate": 1.325520725904066e-05,
      "loss": 0.05918598920106888,
      "step": 4211
    },
    {
      "epoch": 0.5694344692392838,
      "grad_norm": 1.093928337097168,
      "learning_rate": 1.3248409473512158e-05,
      "loss": 0.043022722005844116,
      "step": 4212
    },
    {
      "epoch": 0.5695696626080491,
      "grad_norm": 0.6291282773017883,
      "learning_rate": 1.3241612052674146e-05,
      "loss": 0.04735366255044937,
      "step": 4213
    },
    {
      "epoch": 0.5697048559768143,
      "grad_norm": 0.28831589221954346,
      "learning_rate": 1.3234814997941883e-05,
      "loss": 0.04577290266752243,
      "step": 4214
    },
    {
      "epoch": 0.5698400493455796,
      "grad_norm": 0.2873865067958832,
      "learning_rate": 1.322801831073055e-05,
      "loss": 0.05296456813812256,
      "step": 4215
    },
    {
      "epoch": 0.5699752427143449,
      "grad_norm": 0.3886588215827942,
      "learning_rate": 1.322122199245526e-05,
      "loss": 0.05226030945777893,
      "step": 4216
    },
    {
      "epoch": 0.5701104360831101,
      "grad_norm": 0.3163350522518158,
      "learning_rate": 1.321442604453103e-05,
      "loss": 0.04614682495594025,
      "step": 4217
    },
    {
      "epoch": 0.5702456294518754,
      "grad_norm": 0.6476740837097168,
      "learning_rate": 1.320763046837282e-05,
      "loss": 0.0599798858165741,
      "step": 4218
    },
    {
      "epoch": 0.5703808228206406,
      "grad_norm": 0.6392112374305725,
      "learning_rate": 1.3200835265395504e-05,
      "loss": 0.04629442095756531,
      "step": 4219
    },
    {
      "epoch": 0.570516016189406,
      "grad_norm": 0.8076706528663635,
      "learning_rate": 1.3194040437013885e-05,
      "loss": 0.036763809621334076,
      "step": 4220
    },
    {
      "epoch": 0.5706512095581712,
      "grad_norm": 0.6790736317634583,
      "learning_rate": 1.3187245984642673e-05,
      "loss": 0.0466952919960022,
      "step": 4221
    },
    {
      "epoch": 0.5707864029269364,
      "grad_norm": 0.4466095566749573,
      "learning_rate": 1.3180451909696517e-05,
      "loss": 0.0447138249874115,
      "step": 4222
    },
    {
      "epoch": 0.5709215962957017,
      "grad_norm": 0.4621639847755432,
      "learning_rate": 1.3173658213589972e-05,
      "loss": 0.04115643352270126,
      "step": 4223
    },
    {
      "epoch": 0.5710567896644669,
      "grad_norm": 0.8061466217041016,
      "learning_rate": 1.3166864897737526e-05,
      "loss": 0.04671631008386612,
      "step": 4224
    },
    {
      "epoch": 0.5711919830332323,
      "grad_norm": 0.28725436329841614,
      "learning_rate": 1.3160071963553593e-05,
      "loss": 0.043019846081733704,
      "step": 4225
    },
    {
      "epoch": 0.5713271764019975,
      "grad_norm": 0.4862815737724304,
      "learning_rate": 1.315327941245248e-05,
      "loss": 0.0696071982383728,
      "step": 4226
    },
    {
      "epoch": 0.5714623697707627,
      "grad_norm": 0.5043781995773315,
      "learning_rate": 1.3146487245848445e-05,
      "loss": 0.07956242561340332,
      "step": 4227
    },
    {
      "epoch": 0.571597563139528,
      "grad_norm": 0.4982438087463379,
      "learning_rate": 1.3139695465155645e-05,
      "loss": 0.05159628391265869,
      "step": 4228
    },
    {
      "epoch": 0.5717327565082932,
      "grad_norm": 0.36959385871887207,
      "learning_rate": 1.3132904071788177e-05,
      "loss": 0.052197109907865524,
      "step": 4229
    },
    {
      "epoch": 0.5718679498770586,
      "grad_norm": 0.7914020419120789,
      "learning_rate": 1.3126113067160031e-05,
      "loss": 0.052085041999816895,
      "step": 4230
    },
    {
      "epoch": 0.5720031432458238,
      "grad_norm": 1.8427563905715942,
      "learning_rate": 1.3119322452685139e-05,
      "loss": 0.06533059477806091,
      "step": 4231
    },
    {
      "epoch": 0.572138336614589,
      "grad_norm": 0.3809410631656647,
      "learning_rate": 1.3112532229777344e-05,
      "loss": 0.04417727142572403,
      "step": 4232
    },
    {
      "epoch": 0.5722735299833543,
      "grad_norm": 0.38008368015289307,
      "learning_rate": 1.3105742399850399e-05,
      "loss": 0.03816206008195877,
      "step": 4233
    },
    {
      "epoch": 0.5724087233521196,
      "grad_norm": 0.3189476728439331,
      "learning_rate": 1.3098952964317996e-05,
      "loss": 0.06020701676607132,
      "step": 4234
    },
    {
      "epoch": 0.5725439167208849,
      "grad_norm": 0.47066813707351685,
      "learning_rate": 1.3092163924593717e-05,
      "loss": 0.058689720928668976,
      "step": 4235
    },
    {
      "epoch": 0.5726791100896501,
      "grad_norm": 0.6496806144714355,
      "learning_rate": 1.308537528209108e-05,
      "loss": 0.05438286066055298,
      "step": 4236
    },
    {
      "epoch": 0.5728143034584153,
      "grad_norm": 0.8448660373687744,
      "learning_rate": 1.3078587038223525e-05,
      "loss": 0.06135724484920502,
      "step": 4237
    },
    {
      "epoch": 0.5729494968271807,
      "grad_norm": 0.312496542930603,
      "learning_rate": 1.3071799194404392e-05,
      "loss": 0.039947301149368286,
      "step": 4238
    },
    {
      "epoch": 0.5730846901959459,
      "grad_norm": 0.8599078059196472,
      "learning_rate": 1.3065011752046955e-05,
      "loss": 0.0575120747089386,
      "step": 4239
    },
    {
      "epoch": 0.5732198835647111,
      "grad_norm": 0.3608524203300476,
      "learning_rate": 1.3058224712564382e-05,
      "loss": 0.05470991134643555,
      "step": 4240
    },
    {
      "epoch": 0.5733550769334764,
      "grad_norm": 0.3681836426258087,
      "learning_rate": 1.305143807736978e-05,
      "loss": 0.060732871294021606,
      "step": 4241
    },
    {
      "epoch": 0.5734902703022416,
      "grad_norm": 0.9190942645072937,
      "learning_rate": 1.3044651847876163e-05,
      "loss": 0.07144888490438461,
      "step": 4242
    },
    {
      "epoch": 0.573625463671007,
      "grad_norm": 0.9063126444816589,
      "learning_rate": 1.3037866025496466e-05,
      "loss": 0.05000302195549011,
      "step": 4243
    },
    {
      "epoch": 0.5737606570397722,
      "grad_norm": 0.7755569219589233,
      "learning_rate": 1.3031080611643514e-05,
      "loss": 0.05154004693031311,
      "step": 4244
    },
    {
      "epoch": 0.5738958504085374,
      "grad_norm": 0.7370505332946777,
      "learning_rate": 1.3024295607730083e-05,
      "loss": 0.0579124391078949,
      "step": 4245
    },
    {
      "epoch": 0.5740310437773027,
      "grad_norm": 0.7617559432983398,
      "learning_rate": 1.301751101516884e-05,
      "loss": 0.06045614555478096,
      "step": 4246
    },
    {
      "epoch": 0.574166237146068,
      "grad_norm": 0.9118973612785339,
      "learning_rate": 1.3010726835372377e-05,
      "loss": 0.05753954499959946,
      "step": 4247
    },
    {
      "epoch": 0.5743014305148333,
      "grad_norm": 0.6673192381858826,
      "learning_rate": 1.30039430697532e-05,
      "loss": 0.060206253081560135,
      "step": 4248
    },
    {
      "epoch": 0.5744366238835985,
      "grad_norm": 0.9776385426521301,
      "learning_rate": 1.2997159719723713e-05,
      "loss": 0.04279157519340515,
      "step": 4249
    },
    {
      "epoch": 0.5745718172523637,
      "grad_norm": 0.44115784764289856,
      "learning_rate": 1.2990376786696254e-05,
      "loss": 0.06441664695739746,
      "step": 4250
    },
    {
      "epoch": 0.574707010621129,
      "grad_norm": 0.43690550327301025,
      "learning_rate": 1.2983594272083063e-05,
      "loss": 0.04498184099793434,
      "step": 4251
    },
    {
      "epoch": 0.5748422039898943,
      "grad_norm": 0.834084689617157,
      "learning_rate": 1.2976812177296307e-05,
      "loss": 0.03776583820581436,
      "step": 4252
    },
    {
      "epoch": 0.5749773973586596,
      "grad_norm": 0.5029201507568359,
      "learning_rate": 1.2970030503748039e-05,
      "loss": 0.057890862226486206,
      "step": 4253
    },
    {
      "epoch": 0.5751125907274248,
      "grad_norm": 1.0927622318267822,
      "learning_rate": 1.2963249252850242e-05,
      "loss": 0.05455537885427475,
      "step": 4254
    },
    {
      "epoch": 0.57524778409619,
      "grad_norm": 0.5903801918029785,
      "learning_rate": 1.295646842601481e-05,
      "loss": 0.047474633902311325,
      "step": 4255
    },
    {
      "epoch": 0.5753829774649554,
      "grad_norm": 0.2989392876625061,
      "learning_rate": 1.294968802465355e-05,
      "loss": 0.056687116622924805,
      "step": 4256
    },
    {
      "epoch": 0.5755181708337206,
      "grad_norm": 0.5718480944633484,
      "learning_rate": 1.2942908050178187e-05,
      "loss": 0.05198892578482628,
      "step": 4257
    },
    {
      "epoch": 0.5756533642024859,
      "grad_norm": 0.2312406599521637,
      "learning_rate": 1.293612850400033e-05,
      "loss": 0.03864598274230957,
      "step": 4258
    },
    {
      "epoch": 0.5757885575712511,
      "grad_norm": 0.34641847014427185,
      "learning_rate": 1.2929349387531525e-05,
      "loss": 0.04378978908061981,
      "step": 4259
    },
    {
      "epoch": 0.5759237509400164,
      "grad_norm": 0.42612916231155396,
      "learning_rate": 1.2922570702183217e-05,
      "loss": 0.0561370924115181,
      "step": 4260
    },
    {
      "epoch": 0.5760589443087817,
      "grad_norm": 0.23974616825580597,
      "learning_rate": 1.2915792449366768e-05,
      "loss": 0.044177234172821045,
      "step": 4261
    },
    {
      "epoch": 0.5761941376775469,
      "grad_norm": 0.6883369088172913,
      "learning_rate": 1.2909014630493451e-05,
      "loss": 0.06474051624536514,
      "step": 4262
    },
    {
      "epoch": 0.5763293310463122,
      "grad_norm": 0.3299371302127838,
      "learning_rate": 1.2902237246974432e-05,
      "loss": 0.05071800947189331,
      "step": 4263
    },
    {
      "epoch": 0.5764645244150775,
      "grad_norm": 0.7747350931167603,
      "learning_rate": 1.289546030022081e-05,
      "loss": 0.06015794724225998,
      "step": 4264
    },
    {
      "epoch": 0.5765997177838427,
      "grad_norm": 0.4959127604961395,
      "learning_rate": 1.2888683791643572e-05,
      "loss": 0.05921699106693268,
      "step": 4265
    },
    {
      "epoch": 0.576734911152608,
      "grad_norm": 0.8913729786872864,
      "learning_rate": 1.2881907722653633e-05,
      "loss": 0.04792968928813934,
      "step": 4266
    },
    {
      "epoch": 0.5768701045213732,
      "grad_norm": 0.25339075922966003,
      "learning_rate": 1.2875132094661796e-05,
      "loss": 0.03906603902578354,
      "step": 4267
    },
    {
      "epoch": 0.5770052978901384,
      "grad_norm": 0.5254693031311035,
      "learning_rate": 1.2868356909078787e-05,
      "loss": 0.0628429651260376,
      "step": 4268
    },
    {
      "epoch": 0.5771404912589038,
      "grad_norm": 0.46943530440330505,
      "learning_rate": 1.286158216731524e-05,
      "loss": 0.07296016812324524,
      "step": 4269
    },
    {
      "epoch": 0.577275684627669,
      "grad_norm": 0.4199482798576355,
      "learning_rate": 1.2854807870781686e-05,
      "loss": 0.05589793622493744,
      "step": 4270
    },
    {
      "epoch": 0.5774108779964343,
      "grad_norm": 0.41195833683013916,
      "learning_rate": 1.284803402088858e-05,
      "loss": 0.054120901972055435,
      "step": 4271
    },
    {
      "epoch": 0.5775460713651995,
      "grad_norm": 0.8266782760620117,
      "learning_rate": 1.284126061904626e-05,
      "loss": 0.06698405742645264,
      "step": 4272
    },
    {
      "epoch": 0.5776812647339647,
      "grad_norm": 1.3787145614624023,
      "learning_rate": 1.283448766666499e-05,
      "loss": 0.07092546671628952,
      "step": 4273
    },
    {
      "epoch": 0.5778164581027301,
      "grad_norm": 0.4263189733028412,
      "learning_rate": 1.282771516515494e-05,
      "loss": 0.06073032319545746,
      "step": 4274
    },
    {
      "epoch": 0.5779516514714953,
      "grad_norm": 0.6744718551635742,
      "learning_rate": 1.282094311592618e-05,
      "loss": 0.07127252221107483,
      "step": 4275
    },
    {
      "epoch": 0.5780868448402606,
      "grad_norm": 1.0482310056686401,
      "learning_rate": 1.2814171520388676e-05,
      "loss": 0.059532083570957184,
      "step": 4276
    },
    {
      "epoch": 0.5782220382090258,
      "grad_norm": 0.3588486313819885,
      "learning_rate": 1.2807400379952318e-05,
      "loss": 0.04554516077041626,
      "step": 4277
    },
    {
      "epoch": 0.5783572315777911,
      "grad_norm": 0.25432875752449036,
      "learning_rate": 1.2800629696026895e-05,
      "loss": 0.041674643754959106,
      "step": 4278
    },
    {
      "epoch": 0.5784924249465564,
      "grad_norm": 0.5969200730323792,
      "learning_rate": 1.2793859470022098e-05,
      "loss": 0.05866185203194618,
      "step": 4279
    },
    {
      "epoch": 0.5786276183153216,
      "grad_norm": 0.9372228384017944,
      "learning_rate": 1.278708970334753e-05,
      "loss": 0.057602398097515106,
      "step": 4280
    },
    {
      "epoch": 0.5787628116840869,
      "grad_norm": 0.4330744445323944,
      "learning_rate": 1.2780320397412678e-05,
      "loss": 0.05368056893348694,
      "step": 4281
    },
    {
      "epoch": 0.5788980050528522,
      "grad_norm": 0.3051561415195465,
      "learning_rate": 1.2773551553626957e-05,
      "loss": 0.05303791165351868,
      "step": 4282
    },
    {
      "epoch": 0.5790331984216174,
      "grad_norm": 0.46041205525398254,
      "learning_rate": 1.2766783173399675e-05,
      "loss": 0.05813620612025261,
      "step": 4283
    },
    {
      "epoch": 0.5791683917903827,
      "grad_norm": 0.7252134680747986,
      "learning_rate": 1.276001525814005e-05,
      "loss": 0.04161641001701355,
      "step": 4284
    },
    {
      "epoch": 0.5793035851591479,
      "grad_norm": 0.5711507797241211,
      "learning_rate": 1.2753247809257192e-05,
      "loss": 0.05217631161212921,
      "step": 4285
    },
    {
      "epoch": 0.5794387785279133,
      "grad_norm": 0.3248074948787689,
      "learning_rate": 1.2746480828160119e-05,
      "loss": 0.05739147961139679,
      "step": 4286
    },
    {
      "epoch": 0.5795739718966785,
      "grad_norm": 0.20690594613552094,
      "learning_rate": 1.2739714316257753e-05,
      "loss": 0.04190623760223389,
      "step": 4287
    },
    {
      "epoch": 0.5797091652654437,
      "grad_norm": 0.7761908769607544,
      "learning_rate": 1.273294827495892e-05,
      "loss": 0.07080733776092529,
      "step": 4288
    },
    {
      "epoch": 0.579844358634209,
      "grad_norm": 0.23968324065208435,
      "learning_rate": 1.2726182705672352e-05,
      "loss": 0.03853154182434082,
      "step": 4289
    },
    {
      "epoch": 0.5799795520029742,
      "grad_norm": 0.25534650683403015,
      "learning_rate": 1.271941760980667e-05,
      "loss": 0.041600268334150314,
      "step": 4290
    },
    {
      "epoch": 0.5801147453717396,
      "grad_norm": 0.8996881246566772,
      "learning_rate": 1.2712652988770396e-05,
      "loss": 0.04725450277328491,
      "step": 4291
    },
    {
      "epoch": 0.5802499387405048,
      "grad_norm": 0.3842504620552063,
      "learning_rate": 1.2705888843971967e-05,
      "loss": 0.04753348231315613,
      "step": 4292
    },
    {
      "epoch": 0.58038513210927,
      "grad_norm": 0.4953189194202423,
      "learning_rate": 1.2699125176819717e-05,
      "loss": 0.05507143586874008,
      "step": 4293
    },
    {
      "epoch": 0.5805203254780353,
      "grad_norm": 0.7106717824935913,
      "learning_rate": 1.269236198872188e-05,
      "loss": 0.04401867091655731,
      "step": 4294
    },
    {
      "epoch": 0.5806555188468006,
      "grad_norm": 1.290099024772644,
      "learning_rate": 1.2685599281086577e-05,
      "loss": 0.06402313709259033,
      "step": 4295
    },
    {
      "epoch": 0.5807907122155658,
      "grad_norm": 0.4258081018924713,
      "learning_rate": 1.2678837055321849e-05,
      "loss": 0.059646766632795334,
      "step": 4296
    },
    {
      "epoch": 0.5809259055843311,
      "grad_norm": 0.3928106129169464,
      "learning_rate": 1.267207531283562e-05,
      "loss": 0.05644255876541138,
      "step": 4297
    },
    {
      "epoch": 0.5810610989530963,
      "grad_norm": 0.9362512826919556,
      "learning_rate": 1.266531405503573e-05,
      "loss": 0.06461828947067261,
      "step": 4298
    },
    {
      "epoch": 0.5811962923218617,
      "grad_norm": 0.5836531519889832,
      "learning_rate": 1.26585532833299e-05,
      "loss": 0.051962971687316895,
      "step": 4299
    },
    {
      "epoch": 0.5813314856906269,
      "grad_norm": 0.3997187316417694,
      "learning_rate": 1.2651792999125763e-05,
      "loss": 0.05108805000782013,
      "step": 4300
    },
    {
      "epoch": 0.5814666790593921,
      "grad_norm": 1.0038704872131348,
      "learning_rate": 1.2645033203830846e-05,
      "loss": 0.04372485354542732,
      "step": 4301
    },
    {
      "epoch": 0.5816018724281574,
      "grad_norm": 0.6462306976318359,
      "learning_rate": 1.2638273898852573e-05,
      "loss": 0.07015995681285858,
      "step": 4302
    },
    {
      "epoch": 0.5817370657969226,
      "grad_norm": 0.4095308184623718,
      "learning_rate": 1.2631515085598275e-05,
      "loss": 0.05098499357700348,
      "step": 4303
    },
    {
      "epoch": 0.581872259165688,
      "grad_norm": 0.5673921704292297,
      "learning_rate": 1.262475676547516e-05,
      "loss": 0.06684210151433945,
      "step": 4304
    },
    {
      "epoch": 0.5820074525344532,
      "grad_norm": 0.8474652767181396,
      "learning_rate": 1.2617998939890352e-05,
      "loss": 0.06170887127518654,
      "step": 4305
    },
    {
      "epoch": 0.5821426459032184,
      "grad_norm": 0.4729473888874054,
      "learning_rate": 1.261124161025087e-05,
      "loss": 0.04176437854766846,
      "step": 4306
    },
    {
      "epoch": 0.5822778392719837,
      "grad_norm": 0.36480385065078735,
      "learning_rate": 1.260448477796362e-05,
      "loss": 0.05516377091407776,
      "step": 4307
    },
    {
      "epoch": 0.582413032640749,
      "grad_norm": 0.3644324839115143,
      "learning_rate": 1.259772844443542e-05,
      "loss": 0.06253552436828613,
      "step": 4308
    },
    {
      "epoch": 0.5825482260095143,
      "grad_norm": 0.7161415815353394,
      "learning_rate": 1.2590972611072964e-05,
      "loss": 0.060322076082229614,
      "step": 4309
    },
    {
      "epoch": 0.5826834193782795,
      "grad_norm": 0.5066630840301514,
      "learning_rate": 1.2584217279282855e-05,
      "loss": 0.0561978816986084,
      "step": 4310
    },
    {
      "epoch": 0.5828186127470447,
      "grad_norm": 0.6844581961631775,
      "learning_rate": 1.2577462450471593e-05,
      "loss": 0.05494430661201477,
      "step": 4311
    },
    {
      "epoch": 0.58295380611581,
      "grad_norm": 1.0546153783798218,
      "learning_rate": 1.2570708126045574e-05,
      "loss": 0.052705734968185425,
      "step": 4312
    },
    {
      "epoch": 0.5830889994845753,
      "grad_norm": 0.45255061984062195,
      "learning_rate": 1.256395430741107e-05,
      "loss": 0.06387509405612946,
      "step": 4313
    },
    {
      "epoch": 0.5832241928533406,
      "grad_norm": 0.6577525734901428,
      "learning_rate": 1.2557200995974268e-05,
      "loss": 0.038536787033081055,
      "step": 4314
    },
    {
      "epoch": 0.5833593862221058,
      "grad_norm": 0.42378073930740356,
      "learning_rate": 1.2550448193141248e-05,
      "loss": 0.05845659226179123,
      "step": 4315
    },
    {
      "epoch": 0.583494579590871,
      "grad_norm": 0.9057872295379639,
      "learning_rate": 1.2543695900317977e-05,
      "loss": 0.06074342131614685,
      "step": 4316
    },
    {
      "epoch": 0.5836297729596364,
      "grad_norm": 1.0088768005371094,
      "learning_rate": 1.2536944118910323e-05,
      "loss": 0.04993507266044617,
      "step": 4317
    },
    {
      "epoch": 0.5837649663284016,
      "grad_norm": 1.4090949296951294,
      "learning_rate": 1.2530192850324032e-05,
      "loss": 0.0724797248840332,
      "step": 4318
    },
    {
      "epoch": 0.5839001596971669,
      "grad_norm": 0.49211621284484863,
      "learning_rate": 1.252344209596476e-05,
      "loss": 0.055135756731033325,
      "step": 4319
    },
    {
      "epoch": 0.5840353530659321,
      "grad_norm": 1.1818344593048096,
      "learning_rate": 1.251669185723805e-05,
      "loss": 0.08254462480545044,
      "step": 4320
    },
    {
      "epoch": 0.5841705464346973,
      "grad_norm": 0.421589732170105,
      "learning_rate": 1.2509942135549344e-05,
      "loss": 0.06350669264793396,
      "step": 4321
    },
    {
      "epoch": 0.5843057398034627,
      "grad_norm": 0.26521849632263184,
      "learning_rate": 1.250319293230396e-05,
      "loss": 0.04598189890384674,
      "step": 4322
    },
    {
      "epoch": 0.5844409331722279,
      "grad_norm": 0.8640342354774475,
      "learning_rate": 1.2496444248907121e-05,
      "loss": 0.06404251605272293,
      "step": 4323
    },
    {
      "epoch": 0.5845761265409931,
      "grad_norm": 0.8883600234985352,
      "learning_rate": 1.2489696086763939e-05,
      "loss": 0.06144168600440025,
      "step": 4324
    },
    {
      "epoch": 0.5847113199097584,
      "grad_norm": 0.33867132663726807,
      "learning_rate": 1.2482948447279417e-05,
      "loss": 0.041415244340896606,
      "step": 4325
    },
    {
      "epoch": 0.5848465132785237,
      "grad_norm": 0.3557130992412567,
      "learning_rate": 1.2476201331858458e-05,
      "loss": 0.054164670407772064,
      "step": 4326
    },
    {
      "epoch": 0.584981706647289,
      "grad_norm": 0.28703773021698,
      "learning_rate": 1.2469454741905839e-05,
      "loss": 0.04187518358230591,
      "step": 4327
    },
    {
      "epoch": 0.5851169000160542,
      "grad_norm": 0.45612937211990356,
      "learning_rate": 1.2462708678826233e-05,
      "loss": 0.045684874057769775,
      "step": 4328
    },
    {
      "epoch": 0.5852520933848194,
      "grad_norm": 0.5603744983673096,
      "learning_rate": 1.245596314402421e-05,
      "loss": 0.058832503855228424,
      "step": 4329
    },
    {
      "epoch": 0.5853872867535848,
      "grad_norm": 0.2811681926250458,
      "learning_rate": 1.2449218138904225e-05,
      "loss": 0.03117161989212036,
      "step": 4330
    },
    {
      "epoch": 0.58552248012235,
      "grad_norm": 0.5589156150817871,
      "learning_rate": 1.2442473664870636e-05,
      "loss": 0.07267424464225769,
      "step": 4331
    },
    {
      "epoch": 0.5856576734911153,
      "grad_norm": 0.5839823484420776,
      "learning_rate": 1.2435729723327661e-05,
      "loss": 0.051913969218730927,
      "step": 4332
    },
    {
      "epoch": 0.5857928668598805,
      "grad_norm": 0.42361190915107727,
      "learning_rate": 1.2428986315679433e-05,
      "loss": 0.05698402225971222,
      "step": 4333
    },
    {
      "epoch": 0.5859280602286457,
      "grad_norm": 0.521894097328186,
      "learning_rate": 1.2422243443329962e-05,
      "loss": 0.06634476780891418,
      "step": 4334
    },
    {
      "epoch": 0.5860632535974111,
      "grad_norm": 0.29309552907943726,
      "learning_rate": 1.241550110768316e-05,
      "loss": 0.04298561066389084,
      "step": 4335
    },
    {
      "epoch": 0.5861984469661763,
      "grad_norm": 0.42206260561943054,
      "learning_rate": 1.2408759310142803e-05,
      "loss": 0.0429934561252594,
      "step": 4336
    },
    {
      "epoch": 0.5863336403349416,
      "grad_norm": 0.341665655374527,
      "learning_rate": 1.2402018052112576e-05,
      "loss": 0.05137814208865166,
      "step": 4337
    },
    {
      "epoch": 0.5864688337037068,
      "grad_norm": 0.3751700818538666,
      "learning_rate": 1.2395277334996045e-05,
      "loss": 0.05620073154568672,
      "step": 4338
    },
    {
      "epoch": 0.5866040270724721,
      "grad_norm": 0.37570586800575256,
      "learning_rate": 1.2388537160196663e-05,
      "loss": 0.0541437491774559,
      "step": 4339
    },
    {
      "epoch": 0.5867392204412374,
      "grad_norm": 0.3547845184803009,
      "learning_rate": 1.2381797529117776e-05,
      "loss": 0.04071830213069916,
      "step": 4340
    },
    {
      "epoch": 0.5868744138100026,
      "grad_norm": 0.30910295248031616,
      "learning_rate": 1.23750584431626e-05,
      "loss": 0.05100833252072334,
      "step": 4341
    },
    {
      "epoch": 0.5870096071787679,
      "grad_norm": 0.4811614155769348,
      "learning_rate": 1.236831990373425e-05,
      "loss": 0.0544014535844326,
      "step": 4342
    },
    {
      "epoch": 0.5871448005475332,
      "grad_norm": 0.26779910922050476,
      "learning_rate": 1.2361581912235736e-05,
      "loss": 0.041720613837242126,
      "step": 4343
    },
    {
      "epoch": 0.5872799939162984,
      "grad_norm": 0.4530389904975891,
      "learning_rate": 1.235484447006994e-05,
      "loss": 0.05202691629528999,
      "step": 4344
    },
    {
      "epoch": 0.5874151872850637,
      "grad_norm": 0.8156394362449646,
      "learning_rate": 1.2348107578639627e-05,
      "loss": 0.04596954584121704,
      "step": 4345
    },
    {
      "epoch": 0.5875503806538289,
      "grad_norm": 0.44907045364379883,
      "learning_rate": 1.2341371239347454e-05,
      "loss": 0.05828942358493805,
      "step": 4346
    },
    {
      "epoch": 0.5876855740225941,
      "grad_norm": 1.5142748355865479,
      "learning_rate": 1.233463545359597e-05,
      "loss": 0.06395776569843292,
      "step": 4347
    },
    {
      "epoch": 0.5878207673913595,
      "grad_norm": 0.5775921940803528,
      "learning_rate": 1.23279002227876e-05,
      "loss": 0.04298216104507446,
      "step": 4348
    },
    {
      "epoch": 0.5879559607601247,
      "grad_norm": 0.6789389252662659,
      "learning_rate": 1.2321165548324655e-05,
      "loss": 0.05139341950416565,
      "step": 4349
    },
    {
      "epoch": 0.58809115412889,
      "grad_norm": 0.4374094009399414,
      "learning_rate": 1.2314431431609323e-05,
      "loss": 0.03571010380983353,
      "step": 4350
    },
    {
      "epoch": 0.5882263474976552,
      "grad_norm": 0.5122615098953247,
      "learning_rate": 1.2307697874043687e-05,
      "loss": 0.06129905581474304,
      "step": 4351
    },
    {
      "epoch": 0.5883615408664205,
      "grad_norm": 0.6931524276733398,
      "learning_rate": 1.2300964877029712e-05,
      "loss": 0.07287464290857315,
      "step": 4352
    },
    {
      "epoch": 0.5884967342351858,
      "grad_norm": 0.5996394157409668,
      "learning_rate": 1.2294232441969246e-05,
      "loss": 0.0418534129858017,
      "step": 4353
    },
    {
      "epoch": 0.588631927603951,
      "grad_norm": 0.6493504643440247,
      "learning_rate": 1.2287500570264017e-05,
      "loss": 0.062101706862449646,
      "step": 4354
    },
    {
      "epoch": 0.5887671209727163,
      "grad_norm": 0.6454962491989136,
      "learning_rate": 1.2280769263315628e-05,
      "loss": 0.04887927323579788,
      "step": 4355
    },
    {
      "epoch": 0.5889023143414815,
      "grad_norm": 0.5784348845481873,
      "learning_rate": 1.2274038522525577e-05,
      "loss": 0.05716042220592499,
      "step": 4356
    },
    {
      "epoch": 0.5890375077102468,
      "grad_norm": 0.5934639573097229,
      "learning_rate": 1.2267308349295246e-05,
      "loss": 0.04778507351875305,
      "step": 4357
    },
    {
      "epoch": 0.5891727010790121,
      "grad_norm": 0.338418573141098,
      "learning_rate": 1.2260578745025892e-05,
      "loss": 0.06151972711086273,
      "step": 4358
    },
    {
      "epoch": 0.5893078944477773,
      "grad_norm": 0.3730992078781128,
      "learning_rate": 1.225384971111865e-05,
      "loss": 0.03685858100652695,
      "step": 4359
    },
    {
      "epoch": 0.5894430878165426,
      "grad_norm": 1.1936970949172974,
      "learning_rate": 1.224712124897454e-05,
      "loss": 0.05756092816591263,
      "step": 4360
    },
    {
      "epoch": 0.5895782811853079,
      "grad_norm": 0.39170098304748535,
      "learning_rate": 1.2240393359994466e-05,
      "loss": 0.04370509833097458,
      "step": 4361
    },
    {
      "epoch": 0.5897134745540731,
      "grad_norm": 0.38018980622291565,
      "learning_rate": 1.2233666045579209e-05,
      "loss": 0.0440501868724823,
      "step": 4362
    },
    {
      "epoch": 0.5898486679228384,
      "grad_norm": 0.31766247749328613,
      "learning_rate": 1.222693930712944e-05,
      "loss": 0.03685295581817627,
      "step": 4363
    },
    {
      "epoch": 0.5899838612916036,
      "grad_norm": 1.0282713174819946,
      "learning_rate": 1.2220213146045691e-05,
      "loss": 0.059676796197891235,
      "step": 4364
    },
    {
      "epoch": 0.590119054660369,
      "grad_norm": 0.8460642695426941,
      "learning_rate": 1.2213487563728389e-05,
      "loss": 0.05196624994277954,
      "step": 4365
    },
    {
      "epoch": 0.5902542480291342,
      "grad_norm": 0.591384768486023,
      "learning_rate": 1.220676256157783e-05,
      "loss": 0.04675830900669098,
      "step": 4366
    },
    {
      "epoch": 0.5903894413978994,
      "grad_norm": 0.6578323841094971,
      "learning_rate": 1.2200038140994212e-05,
      "loss": 0.04303811490535736,
      "step": 4367
    },
    {
      "epoch": 0.5905246347666647,
      "grad_norm": 0.27735573053359985,
      "learning_rate": 1.2193314303377578e-05,
      "loss": 0.0528721883893013,
      "step": 4368
    },
    {
      "epoch": 0.5906598281354299,
      "grad_norm": 0.3221501111984253,
      "learning_rate": 1.2186591050127874e-05,
      "loss": 0.05126020312309265,
      "step": 4369
    },
    {
      "epoch": 0.5907950215041953,
      "grad_norm": 1.4605958461761475,
      "learning_rate": 1.2179868382644916e-05,
      "loss": 0.056049223989248276,
      "step": 4370
    },
    {
      "epoch": 0.5909302148729605,
      "grad_norm": 0.7284490466117859,
      "learning_rate": 1.2173146302328396e-05,
      "loss": 0.04205405339598656,
      "step": 4371
    },
    {
      "epoch": 0.5910654082417257,
      "grad_norm": 0.4030453860759735,
      "learning_rate": 1.21664248105779e-05,
      "loss": 0.060487955808639526,
      "step": 4372
    },
    {
      "epoch": 0.591200601610491,
      "grad_norm": 0.36900702118873596,
      "learning_rate": 1.2159703908792858e-05,
      "loss": 0.07081760466098785,
      "step": 4373
    },
    {
      "epoch": 0.5913357949792563,
      "grad_norm": 0.217972993850708,
      "learning_rate": 1.2152983598372613e-05,
      "loss": 0.03986116871237755,
      "step": 4374
    },
    {
      "epoch": 0.5914709883480215,
      "grad_norm": 0.5936857461929321,
      "learning_rate": 1.2146263880716366e-05,
      "loss": 0.06678003072738647,
      "step": 4375
    },
    {
      "epoch": 0.5916061817167868,
      "grad_norm": 0.3019724190235138,
      "learning_rate": 1.2139544757223194e-05,
      "loss": 0.050807081162929535,
      "step": 4376
    },
    {
      "epoch": 0.591741375085552,
      "grad_norm": 0.3493005335330963,
      "learning_rate": 1.2132826229292066e-05,
      "loss": 0.057315923273563385,
      "step": 4377
    },
    {
      "epoch": 0.5918765684543174,
      "grad_norm": 0.5496053099632263,
      "learning_rate": 1.2126108298321798e-05,
      "loss": 0.03934304043650627,
      "step": 4378
    },
    {
      "epoch": 0.5920117618230826,
      "grad_norm": 1.3614407777786255,
      "learning_rate": 1.2119390965711107e-05,
      "loss": 0.07513861358165741,
      "step": 4379
    },
    {
      "epoch": 0.5921469551918478,
      "grad_norm": 0.347663015127182,
      "learning_rate": 1.2112674232858582e-05,
      "loss": 0.048264436423778534,
      "step": 4380
    },
    {
      "epoch": 0.5922821485606131,
      "grad_norm": 0.30053818225860596,
      "learning_rate": 1.2105958101162684e-05,
      "loss": 0.061656564474105835,
      "step": 4381
    },
    {
      "epoch": 0.5924173419293783,
      "grad_norm": 0.4433741569519043,
      "learning_rate": 1.2099242572021735e-05,
      "loss": 0.06932751834392548,
      "step": 4382
    },
    {
      "epoch": 0.5925525352981437,
      "grad_norm": 0.6770027279853821,
      "learning_rate": 1.209252764683395e-05,
      "loss": 0.05713781714439392,
      "step": 4383
    },
    {
      "epoch": 0.5926877286669089,
      "grad_norm": 1.0216777324676514,
      "learning_rate": 1.2085813326997414e-05,
      "loss": 0.06954903155565262,
      "step": 4384
    },
    {
      "epoch": 0.5928229220356741,
      "grad_norm": 0.31264278292655945,
      "learning_rate": 1.2079099613910088e-05,
      "loss": 0.06308547407388687,
      "step": 4385
    },
    {
      "epoch": 0.5929581154044394,
      "grad_norm": 0.28358975052833557,
      "learning_rate": 1.20723865089698e-05,
      "loss": 0.046769797801971436,
      "step": 4386
    },
    {
      "epoch": 0.5930933087732047,
      "grad_norm": 0.4193544089794159,
      "learning_rate": 1.2065674013574248e-05,
      "loss": 0.05737566947937012,
      "step": 4387
    },
    {
      "epoch": 0.59322850214197,
      "grad_norm": 1.1413214206695557,
      "learning_rate": 1.2058962129121013e-05,
      "loss": 0.049747034907341,
      "step": 4388
    },
    {
      "epoch": 0.5933636955107352,
      "grad_norm": 0.47744250297546387,
      "learning_rate": 1.2052250857007548e-05,
      "loss": 0.04318264499306679,
      "step": 4389
    },
    {
      "epoch": 0.5934988888795004,
      "grad_norm": 1.706655502319336,
      "learning_rate": 1.2045540198631177e-05,
      "loss": 0.06314040720462799,
      "step": 4390
    },
    {
      "epoch": 0.5936340822482657,
      "grad_norm": 0.5557811260223389,
      "learning_rate": 1.2038830155389091e-05,
      "loss": 0.05621114373207092,
      "step": 4391
    },
    {
      "epoch": 0.593769275617031,
      "grad_norm": 0.46371743083000183,
      "learning_rate": 1.2032120728678354e-05,
      "loss": 0.05500394105911255,
      "step": 4392
    },
    {
      "epoch": 0.5939044689857963,
      "grad_norm": 1.143871545791626,
      "learning_rate": 1.2025411919895907e-05,
      "loss": 0.05228354036808014,
      "step": 4393
    },
    {
      "epoch": 0.5940396623545615,
      "grad_norm": 0.31671059131622314,
      "learning_rate": 1.2018703730438561e-05,
      "loss": 0.04256322234869003,
      "step": 4394
    },
    {
      "epoch": 0.5941748557233267,
      "grad_norm": 0.3683294951915741,
      "learning_rate": 1.2011996161703003e-05,
      "loss": 0.05954208970069885,
      "step": 4395
    },
    {
      "epoch": 0.5943100490920921,
      "grad_norm": 0.36772671341896057,
      "learning_rate": 1.2005289215085775e-05,
      "loss": 0.039159759879112244,
      "step": 4396
    },
    {
      "epoch": 0.5944452424608573,
      "grad_norm": 0.6880024075508118,
      "learning_rate": 1.19985828919833e-05,
      "loss": 0.05449862405657768,
      "step": 4397
    },
    {
      "epoch": 0.5945804358296226,
      "grad_norm": 0.4047585725784302,
      "learning_rate": 1.1991877193791872e-05,
      "loss": 0.05063389241695404,
      "step": 4398
    },
    {
      "epoch": 0.5947156291983878,
      "grad_norm": 0.40419384837150574,
      "learning_rate": 1.1985172121907653e-05,
      "loss": 0.05270978808403015,
      "step": 4399
    },
    {
      "epoch": 0.594850822567153,
      "grad_norm": 1.3940887451171875,
      "learning_rate": 1.1978467677726682e-05,
      "loss": 0.04564329981803894,
      "step": 4400
    },
    {
      "epoch": 0.5949860159359184,
      "grad_norm": 0.3156428933143616,
      "learning_rate": 1.197176386264485e-05,
      "loss": 0.04023021459579468,
      "step": 4401
    },
    {
      "epoch": 0.5951212093046836,
      "grad_norm": 0.3367822766304016,
      "learning_rate": 1.1965060678057927e-05,
      "loss": 0.047062501311302185,
      "step": 4402
    },
    {
      "epoch": 0.5952564026734488,
      "grad_norm": 1.0438036918640137,
      "learning_rate": 1.1958358125361554e-05,
      "loss": 0.04885775223374367,
      "step": 4403
    },
    {
      "epoch": 0.5953915960422141,
      "grad_norm": 0.30638495087623596,
      "learning_rate": 1.1951656205951247e-05,
      "loss": 0.037989918142557144,
      "step": 4404
    },
    {
      "epoch": 0.5955267894109794,
      "grad_norm": 0.2807495594024658,
      "learning_rate": 1.1944954921222367e-05,
      "loss": 0.030818786472082138,
      "step": 4405
    },
    {
      "epoch": 0.5956619827797447,
      "grad_norm": 0.876422643661499,
      "learning_rate": 1.1938254272570167e-05,
      "loss": 0.0718328207731247,
      "step": 4406
    },
    {
      "epoch": 0.5957971761485099,
      "grad_norm": 0.37925800681114197,
      "learning_rate": 1.1931554261389751e-05,
      "loss": 0.046253807842731476,
      "step": 4407
    },
    {
      "epoch": 0.5959323695172751,
      "grad_norm": 0.5791126489639282,
      "learning_rate": 1.1924854889076103e-05,
      "loss": 0.07300731539726257,
      "step": 4408
    },
    {
      "epoch": 0.5960675628860405,
      "grad_norm": 0.45043909549713135,
      "learning_rate": 1.191815615702407e-05,
      "loss": 0.05591295287013054,
      "step": 4409
    },
    {
      "epoch": 0.5962027562548057,
      "grad_norm": 0.34145453572273254,
      "learning_rate": 1.1911458066628353e-05,
      "loss": 0.05854158103466034,
      "step": 4410
    },
    {
      "epoch": 0.596337949623571,
      "grad_norm": 1.750978708267212,
      "learning_rate": 1.1904760619283537e-05,
      "loss": 0.079463891685009,
      "step": 4411
    },
    {
      "epoch": 0.5964731429923362,
      "grad_norm": 0.8478667736053467,
      "learning_rate": 1.1898063816384069e-05,
      "loss": 0.04810492694377899,
      "step": 4412
    },
    {
      "epoch": 0.5966083363611014,
      "grad_norm": 0.2722764313220978,
      "learning_rate": 1.189136765932426e-05,
      "loss": 0.04031342267990112,
      "step": 4413
    },
    {
      "epoch": 0.5967435297298668,
      "grad_norm": 0.6616509556770325,
      "learning_rate": 1.1884672149498276e-05,
      "loss": 0.050533972680568695,
      "step": 4414
    },
    {
      "epoch": 0.596878723098632,
      "grad_norm": 0.6429870128631592,
      "learning_rate": 1.187797728830016e-05,
      "loss": 0.05223057419061661,
      "step": 4415
    },
    {
      "epoch": 0.5970139164673973,
      "grad_norm": 0.9347872734069824,
      "learning_rate": 1.1871283077123823e-05,
      "loss": 0.04920932278037071,
      "step": 4416
    },
    {
      "epoch": 0.5971491098361625,
      "grad_norm": 0.5087645053863525,
      "learning_rate": 1.1864589517363038e-05,
      "loss": 0.05826232209801674,
      "step": 4417
    },
    {
      "epoch": 0.5972843032049278,
      "grad_norm": 1.1827343702316284,
      "learning_rate": 1.185789661041144e-05,
      "loss": 0.05395862087607384,
      "step": 4418
    },
    {
      "epoch": 0.5974194965736931,
      "grad_norm": 0.6172618269920349,
      "learning_rate": 1.1851204357662513e-05,
      "loss": 0.06921996176242828,
      "step": 4419
    },
    {
      "epoch": 0.5975546899424583,
      "grad_norm": 0.6347182989120483,
      "learning_rate": 1.1844512760509634e-05,
      "loss": 0.05589497089385986,
      "step": 4420
    },
    {
      "epoch": 0.5976898833112236,
      "grad_norm": 0.5600841641426086,
      "learning_rate": 1.1837821820346022e-05,
      "loss": 0.056553810834884644,
      "step": 4421
    },
    {
      "epoch": 0.5978250766799889,
      "grad_norm": 0.7159196734428406,
      "learning_rate": 1.1831131538564775e-05,
      "loss": 0.06223991513252258,
      "step": 4422
    },
    {
      "epoch": 0.5979602700487541,
      "grad_norm": 0.4582863748073578,
      "learning_rate": 1.1824441916558843e-05,
      "loss": 0.05742642283439636,
      "step": 4423
    },
    {
      "epoch": 0.5980954634175194,
      "grad_norm": 0.9712393879890442,
      "learning_rate": 1.1817752955721031e-05,
      "loss": 0.06445480138063431,
      "step": 4424
    },
    {
      "epoch": 0.5982306567862846,
      "grad_norm": 1.0242362022399902,
      "learning_rate": 1.1811064657444023e-05,
      "loss": 0.059829238802194595,
      "step": 4425
    },
    {
      "epoch": 0.59836585015505,
      "grad_norm": 0.3005829155445099,
      "learning_rate": 1.1804377023120361e-05,
      "loss": 0.04465368390083313,
      "step": 4426
    },
    {
      "epoch": 0.5985010435238152,
      "grad_norm": 0.5643232464790344,
      "learning_rate": 1.1797690054142451e-05,
      "loss": 0.05843439698219299,
      "step": 4427
    },
    {
      "epoch": 0.5986362368925804,
      "grad_norm": 1.0243879556655884,
      "learning_rate": 1.1791003751902542e-05,
      "loss": 0.05740996450185776,
      "step": 4428
    },
    {
      "epoch": 0.5987714302613457,
      "grad_norm": 0.41483309864997864,
      "learning_rate": 1.1784318117792763e-05,
      "loss": 0.06152985617518425,
      "step": 4429
    },
    {
      "epoch": 0.5989066236301109,
      "grad_norm": 0.3840685188770294,
      "learning_rate": 1.17776331532051e-05,
      "loss": 0.040231138467788696,
      "step": 4430
    },
    {
      "epoch": 0.5990418169988762,
      "grad_norm": 0.9017181396484375,
      "learning_rate": 1.1770948859531397e-05,
      "loss": 0.05257067084312439,
      "step": 4431
    },
    {
      "epoch": 0.5991770103676415,
      "grad_norm": 0.6576009392738342,
      "learning_rate": 1.1764265238163369e-05,
      "loss": 0.06276470422744751,
      "step": 4432
    },
    {
      "epoch": 0.5993122037364067,
      "grad_norm": 0.4700859785079956,
      "learning_rate": 1.1757582290492568e-05,
      "loss": 0.055005013942718506,
      "step": 4433
    },
    {
      "epoch": 0.599447397105172,
      "grad_norm": 0.9550535082817078,
      "learning_rate": 1.1750900017910425e-05,
      "loss": 0.07956385612487793,
      "step": 4434
    },
    {
      "epoch": 0.5995825904739372,
      "grad_norm": 0.7578852772712708,
      "learning_rate": 1.1744218421808221e-05,
      "loss": 0.05361468344926834,
      "step": 4435
    },
    {
      "epoch": 0.5997177838427025,
      "grad_norm": 0.6122483611106873,
      "learning_rate": 1.1737537503577112e-05,
      "loss": 0.06411196291446686,
      "step": 4436
    },
    {
      "epoch": 0.5998529772114678,
      "grad_norm": 0.7074890732765198,
      "learning_rate": 1.1730857264608086e-05,
      "loss": 0.05618126690387726,
      "step": 4437
    },
    {
      "epoch": 0.599988170580233,
      "grad_norm": 0.6040074229240417,
      "learning_rate": 1.1724177706292013e-05,
      "loss": 0.05763174593448639,
      "step": 4438
    },
    {
      "epoch": 0.6001233639489983,
      "grad_norm": 0.3741949796676636,
      "learning_rate": 1.1717498830019607e-05,
      "loss": 0.054901402443647385,
      "step": 4439
    },
    {
      "epoch": 0.6002585573177636,
      "grad_norm": 0.30857011675834656,
      "learning_rate": 1.1710820637181449e-05,
      "loss": 0.041115377098321915,
      "step": 4440
    },
    {
      "epoch": 0.6003937506865288,
      "grad_norm": 0.5500670671463013,
      "learning_rate": 1.170414312916798e-05,
      "loss": 0.0641489326953888,
      "step": 4441
    },
    {
      "epoch": 0.6005289440552941,
      "grad_norm": 0.3509334325790405,
      "learning_rate": 1.1697466307369484e-05,
      "loss": 0.04598594456911087,
      "step": 4442
    },
    {
      "epoch": 0.6006641374240593,
      "grad_norm": 0.2850635349750519,
      "learning_rate": 1.1690790173176116e-05,
      "loss": 0.055777426809072495,
      "step": 4443
    },
    {
      "epoch": 0.6007993307928247,
      "grad_norm": 0.2754749357700348,
      "learning_rate": 1.1684114727977876e-05,
      "loss": 0.039429061114788055,
      "step": 4444
    },
    {
      "epoch": 0.6009345241615899,
      "grad_norm": 1.281582236289978,
      "learning_rate": 1.167743997316464e-05,
      "loss": 0.0517854169011116,
      "step": 4445
    },
    {
      "epoch": 0.6010697175303551,
      "grad_norm": 0.7395382523536682,
      "learning_rate": 1.1670765910126112e-05,
      "loss": 0.05040641129016876,
      "step": 4446
    },
    {
      "epoch": 0.6012049108991204,
      "grad_norm": 0.5542140603065491,
      "learning_rate": 1.1664092540251877e-05,
      "loss": 0.053301576524972916,
      "step": 4447
    },
    {
      "epoch": 0.6013401042678856,
      "grad_norm": 0.7280396819114685,
      "learning_rate": 1.1657419864931361e-05,
      "loss": 0.04925832897424698,
      "step": 4448
    },
    {
      "epoch": 0.601475297636651,
      "grad_norm": 0.5647919774055481,
      "learning_rate": 1.165074788555386e-05,
      "loss": 0.07171936333179474,
      "step": 4449
    },
    {
      "epoch": 0.6016104910054162,
      "grad_norm": 0.4669032394886017,
      "learning_rate": 1.1644076603508514e-05,
      "loss": 0.06894838809967041,
      "step": 4450
    },
    {
      "epoch": 0.6017456843741814,
      "grad_norm": 0.525761604309082,
      "learning_rate": 1.1637406020184305e-05,
      "loss": 0.049926526844501495,
      "step": 4451
    },
    {
      "epoch": 0.6018808777429467,
      "grad_norm": 0.4559738337993622,
      "learning_rate": 1.1630736136970097e-05,
      "loss": 0.04470636323094368,
      "step": 4452
    },
    {
      "epoch": 0.602016071111712,
      "grad_norm": 0.9113357663154602,
      "learning_rate": 1.162406695525459e-05,
      "loss": 0.041763052344322205,
      "step": 4453
    },
    {
      "epoch": 0.6021512644804772,
      "grad_norm": 0.30894824862480164,
      "learning_rate": 1.161739847642635e-05,
      "loss": 0.02997482195496559,
      "step": 4454
    },
    {
      "epoch": 0.6022864578492425,
      "grad_norm": 0.4740419387817383,
      "learning_rate": 1.1610730701873788e-05,
      "loss": 0.0627528578042984,
      "step": 4455
    },
    {
      "epoch": 0.6024216512180077,
      "grad_norm": 0.5100468993186951,
      "learning_rate": 1.1604063632985163e-05,
      "loss": 0.04404717683792114,
      "step": 4456
    },
    {
      "epoch": 0.602556844586773,
      "grad_norm": 0.37073689699172974,
      "learning_rate": 1.1597397271148598e-05,
      "loss": 0.06574007123708725,
      "step": 4457
    },
    {
      "epoch": 0.6026920379555383,
      "grad_norm": 0.5196016430854797,
      "learning_rate": 1.1590731617752067e-05,
      "loss": 0.06108853220939636,
      "step": 4458
    },
    {
      "epoch": 0.6028272313243035,
      "grad_norm": 0.4110202193260193,
      "learning_rate": 1.1584066674183398e-05,
      "loss": 0.06466826796531677,
      "step": 4459
    },
    {
      "epoch": 0.6029624246930688,
      "grad_norm": 0.2575041651725769,
      "learning_rate": 1.1577402441830262e-05,
      "loss": 0.045179583132267,
      "step": 4460
    },
    {
      "epoch": 0.603097618061834,
      "grad_norm": 0.56150883436203,
      "learning_rate": 1.1570738922080185e-05,
      "loss": 0.06690779328346252,
      "step": 4461
    },
    {
      "epoch": 0.6032328114305994,
      "grad_norm": 0.5520632863044739,
      "learning_rate": 1.1564076116320552e-05,
      "loss": 0.06335223466157913,
      "step": 4462
    },
    {
      "epoch": 0.6033680047993646,
      "grad_norm": 0.44771501421928406,
      "learning_rate": 1.1557414025938592e-05,
      "loss": 0.04493759945034981,
      "step": 4463
    },
    {
      "epoch": 0.6035031981681298,
      "grad_norm": 0.4248267412185669,
      "learning_rate": 1.15507526523214e-05,
      "loss": 0.04206670820713043,
      "step": 4464
    },
    {
      "epoch": 0.6036383915368951,
      "grad_norm": 0.3631657660007477,
      "learning_rate": 1.1544091996855895e-05,
      "loss": 0.05404015630483627,
      "step": 4465
    },
    {
      "epoch": 0.6037735849056604,
      "grad_norm": 0.7033305764198303,
      "learning_rate": 1.153743206092886e-05,
      "loss": 0.05844687670469284,
      "step": 4466
    },
    {
      "epoch": 0.6039087782744257,
      "grad_norm": 0.5848333239555359,
      "learning_rate": 1.1530772845926936e-05,
      "loss": 0.05540715903043747,
      "step": 4467
    },
    {
      "epoch": 0.6040439716431909,
      "grad_norm": 0.4362988770008087,
      "learning_rate": 1.1524114353236614e-05,
      "loss": 0.04233485460281372,
      "step": 4468
    },
    {
      "epoch": 0.6041791650119561,
      "grad_norm": 0.4982118606567383,
      "learning_rate": 1.151745658424421e-05,
      "loss": 0.05209067091345787,
      "step": 4469
    },
    {
      "epoch": 0.6043143583807215,
      "grad_norm": 1.0047283172607422,
      "learning_rate": 1.151079954033592e-05,
      "loss": 0.05934375524520874,
      "step": 4470
    },
    {
      "epoch": 0.6044495517494867,
      "grad_norm": 0.4419027268886566,
      "learning_rate": 1.150414322289777e-05,
      "loss": 0.036733437329530716,
      "step": 4471
    },
    {
      "epoch": 0.604584745118252,
      "grad_norm": 0.6926829814910889,
      "learning_rate": 1.1497487633315643e-05,
      "loss": 0.03674779087305069,
      "step": 4472
    },
    {
      "epoch": 0.6047199384870172,
      "grad_norm": 0.4897218346595764,
      "learning_rate": 1.1490832772975275e-05,
      "loss": 0.06402263790369034,
      "step": 4473
    },
    {
      "epoch": 0.6048551318557824,
      "grad_norm": 0.7173218727111816,
      "learning_rate": 1.148417864326223e-05,
      "loss": 0.05959732085466385,
      "step": 4474
    },
    {
      "epoch": 0.6049903252245478,
      "grad_norm": 0.6146566271781921,
      "learning_rate": 1.1477525245561944e-05,
      "loss": 0.04254075512290001,
      "step": 4475
    },
    {
      "epoch": 0.605125518593313,
      "grad_norm": 0.5662301182746887,
      "learning_rate": 1.1470872581259684e-05,
      "loss": 0.05857444554567337,
      "step": 4476
    },
    {
      "epoch": 0.6052607119620783,
      "grad_norm": 0.2399030327796936,
      "learning_rate": 1.146422065174057e-05,
      "loss": 0.04444399103522301,
      "step": 4477
    },
    {
      "epoch": 0.6053959053308435,
      "grad_norm": 0.8195293545722961,
      "learning_rate": 1.1457569458389578e-05,
      "loss": 0.048936132341623306,
      "step": 4478
    },
    {
      "epoch": 0.6055310986996087,
      "grad_norm": 0.5751801133155823,
      "learning_rate": 1.145091900259151e-05,
      "loss": 0.059193696826696396,
      "step": 4479
    },
    {
      "epoch": 0.6056662920683741,
      "grad_norm": 0.4658619463443756,
      "learning_rate": 1.1444269285731032e-05,
      "loss": 0.038571201264858246,
      "step": 4480
    },
    {
      "epoch": 0.6058014854371393,
      "grad_norm": 0.4931628108024597,
      "learning_rate": 1.1437620309192652e-05,
      "loss": 0.05222680792212486,
      "step": 4481
    },
    {
      "epoch": 0.6059366788059045,
      "grad_norm": 0.582899808883667,
      "learning_rate": 1.1430972074360722e-05,
      "loss": 0.05361529067158699,
      "step": 4482
    },
    {
      "epoch": 0.6060718721746698,
      "grad_norm": 0.890861988067627,
      "learning_rate": 1.1424324582619435e-05,
      "loss": 0.05030763894319534,
      "step": 4483
    },
    {
      "epoch": 0.6062070655434351,
      "grad_norm": 0.26525622606277466,
      "learning_rate": 1.1417677835352837e-05,
      "loss": 0.049131568521261215,
      "step": 4484
    },
    {
      "epoch": 0.6063422589122004,
      "grad_norm": 0.3394700288772583,
      "learning_rate": 1.1411031833944816e-05,
      "loss": 0.061053432524204254,
      "step": 4485
    },
    {
      "epoch": 0.6064774522809656,
      "grad_norm": 0.47498729825019836,
      "learning_rate": 1.1404386579779111e-05,
      "loss": 0.05526217818260193,
      "step": 4486
    },
    {
      "epoch": 0.6066126456497308,
      "grad_norm": 1.155960202217102,
      "learning_rate": 1.1397742074239296e-05,
      "loss": 0.05119175463914871,
      "step": 4487
    },
    {
      "epoch": 0.6067478390184962,
      "grad_norm": 0.3275589048862457,
      "learning_rate": 1.1391098318708785e-05,
      "loss": 0.04987310245633125,
      "step": 4488
    },
    {
      "epoch": 0.6068830323872614,
      "grad_norm": 0.6260046362876892,
      "learning_rate": 1.1384455314570848e-05,
      "loss": 0.05643220245838165,
      "step": 4489
    },
    {
      "epoch": 0.6070182257560267,
      "grad_norm": 0.28886738419532776,
      "learning_rate": 1.1377813063208596e-05,
      "loss": 0.061378877609968185,
      "step": 4490
    },
    {
      "epoch": 0.6071534191247919,
      "grad_norm": 0.25748902559280396,
      "learning_rate": 1.1371171566004986e-05,
      "loss": 0.03589729964733124,
      "step": 4491
    },
    {
      "epoch": 0.6072886124935571,
      "grad_norm": 0.29101109504699707,
      "learning_rate": 1.1364530824342806e-05,
      "loss": 0.049356117844581604,
      "step": 4492
    },
    {
      "epoch": 0.6074238058623225,
      "grad_norm": 0.7595440745353699,
      "learning_rate": 1.1357890839604688e-05,
      "loss": 0.061377085745334625,
      "step": 4493
    },
    {
      "epoch": 0.6075589992310877,
      "grad_norm": 0.7708401679992676,
      "learning_rate": 1.1351251613173122e-05,
      "loss": 0.06804318726062775,
      "step": 4494
    },
    {
      "epoch": 0.607694192599853,
      "grad_norm": 0.5198320150375366,
      "learning_rate": 1.1344613146430428e-05,
      "loss": 0.05746417120099068,
      "step": 4495
    },
    {
      "epoch": 0.6078293859686182,
      "grad_norm": 1.0845645666122437,
      "learning_rate": 1.1337975440758775e-05,
      "loss": 0.06742439419031143,
      "step": 4496
    },
    {
      "epoch": 0.6079645793373835,
      "grad_norm": 0.19988715648651123,
      "learning_rate": 1.133133849754016e-05,
      "loss": 0.044403448700904846,
      "step": 4497
    },
    {
      "epoch": 0.6080997727061488,
      "grad_norm": 0.2857939600944519,
      "learning_rate": 1.1324702318156431e-05,
      "loss": 0.0392131507396698,
      "step": 4498
    },
    {
      "epoch": 0.608234966074914,
      "grad_norm": 0.458284854888916,
      "learning_rate": 1.1318066903989279e-05,
      "loss": 0.04939714074134827,
      "step": 4499
    },
    {
      "epoch": 0.6083701594436793,
      "grad_norm": 0.7546911239624023,
      "learning_rate": 1.1311432256420232e-05,
      "loss": 0.06240024045109749,
      "step": 4500
    },
    {
      "epoch": 0.6085053528124446,
      "grad_norm": 0.40432146191596985,
      "learning_rate": 1.1304798376830664e-05,
      "loss": 0.044800613075494766,
      "step": 4501
    },
    {
      "epoch": 0.6086405461812098,
      "grad_norm": 0.22065140306949615,
      "learning_rate": 1.1298165266601778e-05,
      "loss": 0.03623470291495323,
      "step": 4502
    },
    {
      "epoch": 0.6087757395499751,
      "grad_norm": 1.1977784633636475,
      "learning_rate": 1.129153292711462e-05,
      "loss": 0.07958650588989258,
      "step": 4503
    },
    {
      "epoch": 0.6089109329187403,
      "grad_norm": 0.6214346289634705,
      "learning_rate": 1.1284901359750082e-05,
      "loss": 0.039936840534210205,
      "step": 4504
    },
    {
      "epoch": 0.6090461262875057,
      "grad_norm": 0.5940772294998169,
      "learning_rate": 1.1278270565888897e-05,
      "loss": 0.06382440030574799,
      "step": 4505
    },
    {
      "epoch": 0.6091813196562709,
      "grad_norm": 0.6148561239242554,
      "learning_rate": 1.1271640546911624e-05,
      "loss": 0.05178374797105789,
      "step": 4506
    },
    {
      "epoch": 0.6093165130250361,
      "grad_norm": 0.5272179245948792,
      "learning_rate": 1.1265011304198672e-05,
      "loss": 0.05109657719731331,
      "step": 4507
    },
    {
      "epoch": 0.6094517063938014,
      "grad_norm": 0.37512943148612976,
      "learning_rate": 1.1258382839130282e-05,
      "loss": 0.051076628267765045,
      "step": 4508
    },
    {
      "epoch": 0.6095868997625666,
      "grad_norm": 0.4755030572414398,
      "learning_rate": 1.1251755153086536e-05,
      "loss": 0.0556274950504303,
      "step": 4509
    },
    {
      "epoch": 0.6097220931313319,
      "grad_norm": 0.4107944369316101,
      "learning_rate": 1.1245128247447362e-05,
      "loss": 0.05108289420604706,
      "step": 4510
    },
    {
      "epoch": 0.6098572865000972,
      "grad_norm": 0.5602015852928162,
      "learning_rate": 1.1238502123592507e-05,
      "loss": 0.04840603470802307,
      "step": 4511
    },
    {
      "epoch": 0.6099924798688624,
      "grad_norm": 0.4493732750415802,
      "learning_rate": 1.1231876782901568e-05,
      "loss": 0.06225484609603882,
      "step": 4512
    },
    {
      "epoch": 0.6101276732376277,
      "grad_norm": 0.6337834000587463,
      "learning_rate": 1.1225252226753975e-05,
      "loss": 0.06003597378730774,
      "step": 4513
    },
    {
      "epoch": 0.610262866606393,
      "grad_norm": 0.6227352023124695,
      "learning_rate": 1.1218628456529005e-05,
      "loss": 0.04350108653306961,
      "step": 4514
    },
    {
      "epoch": 0.6103980599751582,
      "grad_norm": 0.49585258960723877,
      "learning_rate": 1.1212005473605746e-05,
      "loss": 0.05694460868835449,
      "step": 4515
    },
    {
      "epoch": 0.6105332533439235,
      "grad_norm": 1.3771907091140747,
      "learning_rate": 1.120538327936315e-05,
      "loss": 0.06392525136470795,
      "step": 4516
    },
    {
      "epoch": 0.6106684467126887,
      "grad_norm": 0.7170501351356506,
      "learning_rate": 1.1198761875179993e-05,
      "loss": 0.066231869161129,
      "step": 4517
    },
    {
      "epoch": 0.610803640081454,
      "grad_norm": 0.6148501038551331,
      "learning_rate": 1.1192141262434883e-05,
      "loss": 0.07428601384162903,
      "step": 4518
    },
    {
      "epoch": 0.6109388334502193,
      "grad_norm": 0.3670237958431244,
      "learning_rate": 1.1185521442506272e-05,
      "loss": 0.04119445011019707,
      "step": 4519
    },
    {
      "epoch": 0.6110740268189845,
      "grad_norm": 0.42440328001976013,
      "learning_rate": 1.1178902416772432e-05,
      "loss": 0.04954603314399719,
      "step": 4520
    },
    {
      "epoch": 0.6112092201877498,
      "grad_norm": 0.334435373544693,
      "learning_rate": 1.1172284186611485e-05,
      "loss": 0.04701607674360275,
      "step": 4521
    },
    {
      "epoch": 0.611344413556515,
      "grad_norm": 0.30091506242752075,
      "learning_rate": 1.1165666753401384e-05,
      "loss": 0.051545873284339905,
      "step": 4522
    },
    {
      "epoch": 0.6114796069252804,
      "grad_norm": 0.32049858570098877,
      "learning_rate": 1.1159050118519914e-05,
      "loss": 0.05188825726509094,
      "step": 4523
    },
    {
      "epoch": 0.6116148002940456,
      "grad_norm": 0.5259817242622375,
      "learning_rate": 1.1152434283344696e-05,
      "loss": 0.061009228229522705,
      "step": 4524
    },
    {
      "epoch": 0.6117499936628108,
      "grad_norm": 0.3324960470199585,
      "learning_rate": 1.114581924925317e-05,
      "loss": 0.041055694222450256,
      "step": 4525
    },
    {
      "epoch": 0.6118851870315761,
      "grad_norm": 0.6279718279838562,
      "learning_rate": 1.113920501762263e-05,
      "loss": 0.05985286831855774,
      "step": 4526
    },
    {
      "epoch": 0.6120203804003413,
      "grad_norm": 1.092221736907959,
      "learning_rate": 1.1132591589830193e-05,
      "loss": 0.04456346482038498,
      "step": 4527
    },
    {
      "epoch": 0.6121555737691067,
      "grad_norm": 0.4922019839286804,
      "learning_rate": 1.1125978967252818e-05,
      "loss": 0.049274761229753494,
      "step": 4528
    },
    {
      "epoch": 0.6122907671378719,
      "grad_norm": 0.5381514430046082,
      "learning_rate": 1.1119367151267278e-05,
      "loss": 0.044018328189849854,
      "step": 4529
    },
    {
      "epoch": 0.6124259605066371,
      "grad_norm": 0.5391473770141602,
      "learning_rate": 1.1112756143250186e-05,
      "loss": 0.045425914227962494,
      "step": 4530
    },
    {
      "epoch": 0.6125611538754024,
      "grad_norm": 0.2799208462238312,
      "learning_rate": 1.1106145944577995e-05,
      "loss": 0.03928492218255997,
      "step": 4531
    },
    {
      "epoch": 0.6126963472441677,
      "grad_norm": 0.17868204414844513,
      "learning_rate": 1.1099536556626984e-05,
      "loss": 0.031209055334329605,
      "step": 4532
    },
    {
      "epoch": 0.612831540612933,
      "grad_norm": 0.346330851316452,
      "learning_rate": 1.1092927980773269e-05,
      "loss": 0.05431076139211655,
      "step": 4533
    },
    {
      "epoch": 0.6129667339816982,
      "grad_norm": 0.4241599440574646,
      "learning_rate": 1.1086320218392777e-05,
      "loss": 0.056727904826402664,
      "step": 4534
    },
    {
      "epoch": 0.6131019273504634,
      "grad_norm": 0.3166915774345398,
      "learning_rate": 1.1079713270861286e-05,
      "loss": 0.043045420199632645,
      "step": 4535
    },
    {
      "epoch": 0.6132371207192288,
      "grad_norm": 0.3482891321182251,
      "learning_rate": 1.1073107139554395e-05,
      "loss": 0.06476882845163345,
      "step": 4536
    },
    {
      "epoch": 0.613372314087994,
      "grad_norm": 0.8104740381240845,
      "learning_rate": 1.1066501825847545e-05,
      "loss": 0.062356606125831604,
      "step": 4537
    },
    {
      "epoch": 0.6135075074567592,
      "grad_norm": 0.42610758543014526,
      "learning_rate": 1.1059897331115985e-05,
      "loss": 0.04907872527837753,
      "step": 4538
    },
    {
      "epoch": 0.6136427008255245,
      "grad_norm": 0.42339351773262024,
      "learning_rate": 1.1053293656734816e-05,
      "loss": 0.0518743060529232,
      "step": 4539
    },
    {
      "epoch": 0.6137778941942897,
      "grad_norm": 0.4038066864013672,
      "learning_rate": 1.1046690804078949e-05,
      "loss": 0.04603969305753708,
      "step": 4540
    },
    {
      "epoch": 0.6139130875630551,
      "grad_norm": 0.32849767804145813,
      "learning_rate": 1.1040088774523139e-05,
      "loss": 0.04492694139480591,
      "step": 4541
    },
    {
      "epoch": 0.6140482809318203,
      "grad_norm": 0.6885644793510437,
      "learning_rate": 1.1033487569441971e-05,
      "loss": 0.04000355303287506,
      "step": 4542
    },
    {
      "epoch": 0.6141834743005855,
      "grad_norm": 0.9018462896347046,
      "learning_rate": 1.1026887190209834e-05,
      "loss": 0.05526731163263321,
      "step": 4543
    },
    {
      "epoch": 0.6143186676693508,
      "grad_norm": 0.27910539507865906,
      "learning_rate": 1.1020287638200977e-05,
      "loss": 0.034677326679229736,
      "step": 4544
    },
    {
      "epoch": 0.6144538610381161,
      "grad_norm": 1.0978777408599854,
      "learning_rate": 1.1013688914789452e-05,
      "loss": 0.06008952856063843,
      "step": 4545
    },
    {
      "epoch": 0.6145890544068814,
      "grad_norm": 0.7051708698272705,
      "learning_rate": 1.100709102134915e-05,
      "loss": 0.046445004642009735,
      "step": 4546
    },
    {
      "epoch": 0.6147242477756466,
      "grad_norm": 0.6244603395462036,
      "learning_rate": 1.10004939592538e-05,
      "loss": 0.06161254644393921,
      "step": 4547
    },
    {
      "epoch": 0.6148594411444118,
      "grad_norm": 0.3483380973339081,
      "learning_rate": 1.0993897729876927e-05,
      "loss": 0.05491615831851959,
      "step": 4548
    },
    {
      "epoch": 0.6149946345131772,
      "grad_norm": 1.7858338356018066,
      "learning_rate": 1.0987302334591915e-05,
      "loss": 0.06269332766532898,
      "step": 4549
    },
    {
      "epoch": 0.6151298278819424,
      "grad_norm": 0.3047940731048584,
      "learning_rate": 1.098070777477195e-05,
      "loss": 0.03702007979154587,
      "step": 4550
    },
    {
      "epoch": 0.6152650212507077,
      "grad_norm": 0.6988071799278259,
      "learning_rate": 1.0974114051790067e-05,
      "loss": 0.06821054220199585,
      "step": 4551
    },
    {
      "epoch": 0.6154002146194729,
      "grad_norm": 0.4086858928203583,
      "learning_rate": 1.09675211670191e-05,
      "loss": 0.0581369623541832,
      "step": 4552
    },
    {
      "epoch": 0.6155354079882381,
      "grad_norm": 0.43806758522987366,
      "learning_rate": 1.0960929121831732e-05,
      "loss": 0.040879987180233,
      "step": 4553
    },
    {
      "epoch": 0.6156706013570035,
      "grad_norm": 1.7018482685089111,
      "learning_rate": 1.095433791760046e-05,
      "loss": 0.056650206446647644,
      "step": 4554
    },
    {
      "epoch": 0.6158057947257687,
      "grad_norm": 0.4267836809158325,
      "learning_rate": 1.0947747555697609e-05,
      "loss": 0.06319084763526917,
      "step": 4555
    },
    {
      "epoch": 0.615940988094534,
      "grad_norm": 0.3949653208255768,
      "learning_rate": 1.0941158037495328e-05,
      "loss": 0.0571795254945755,
      "step": 4556
    },
    {
      "epoch": 0.6160761814632992,
      "grad_norm": 0.3105303943157196,
      "learning_rate": 1.0934569364365583e-05,
      "loss": 0.05321167781949043,
      "step": 4557
    },
    {
      "epoch": 0.6162113748320645,
      "grad_norm": 0.4802440106868744,
      "learning_rate": 1.0927981537680176e-05,
      "loss": 0.053725797683000565,
      "step": 4558
    },
    {
      "epoch": 0.6163465682008298,
      "grad_norm": 0.2842666506767273,
      "learning_rate": 1.0921394558810726e-05,
      "loss": 0.03103308379650116,
      "step": 4559
    },
    {
      "epoch": 0.616481761569595,
      "grad_norm": 0.347064346075058,
      "learning_rate": 1.0914808429128688e-05,
      "loss": 0.07509362697601318,
      "step": 4560
    },
    {
      "epoch": 0.6166169549383603,
      "grad_norm": 0.31892895698547363,
      "learning_rate": 1.0908223150005315e-05,
      "loss": 0.04672209173440933,
      "step": 4561
    },
    {
      "epoch": 0.6167521483071255,
      "grad_norm": 0.6527944803237915,
      "learning_rate": 1.09016387228117e-05,
      "loss": 0.0335516631603241,
      "step": 4562
    },
    {
      "epoch": 0.6168873416758908,
      "grad_norm": 0.7031928300857544,
      "learning_rate": 1.0895055148918758e-05,
      "loss": 0.056803472340106964,
      "step": 4563
    },
    {
      "epoch": 0.6170225350446561,
      "grad_norm": 1.5055878162384033,
      "learning_rate": 1.0888472429697223e-05,
      "loss": 0.05277127027511597,
      "step": 4564
    },
    {
      "epoch": 0.6171577284134213,
      "grad_norm": 0.40353551506996155,
      "learning_rate": 1.088189056651766e-05,
      "loss": 0.05125678330659866,
      "step": 4565
    },
    {
      "epoch": 0.6172929217821865,
      "grad_norm": 0.5008995532989502,
      "learning_rate": 1.0875309560750438e-05,
      "loss": 0.05714167654514313,
      "step": 4566
    },
    {
      "epoch": 0.6174281151509519,
      "grad_norm": 1.4121614694595337,
      "learning_rate": 1.086872941376576e-05,
      "loss": 0.07152986526489258,
      "step": 4567
    },
    {
      "epoch": 0.6175633085197171,
      "grad_norm": 0.2644006907939911,
      "learning_rate": 1.0862150126933648e-05,
      "loss": 0.035382434725761414,
      "step": 4568
    },
    {
      "epoch": 0.6176985018884824,
      "grad_norm": 0.9554007053375244,
      "learning_rate": 1.0855571701623942e-05,
      "loss": 0.07176284492015839,
      "step": 4569
    },
    {
      "epoch": 0.6178336952572476,
      "grad_norm": 0.32135018706321716,
      "learning_rate": 1.0848994139206317e-05,
      "loss": 0.041753847151994705,
      "step": 4570
    },
    {
      "epoch": 0.6179688886260128,
      "grad_norm": 0.46053165197372437,
      "learning_rate": 1.0842417441050247e-05,
      "loss": 0.05809721350669861,
      "step": 4571
    },
    {
      "epoch": 0.6181040819947782,
      "grad_norm": 0.418971449136734,
      "learning_rate": 1.0835841608525031e-05,
      "loss": 0.06268781423568726,
      "step": 4572
    },
    {
      "epoch": 0.6182392753635434,
      "grad_norm": 0.5150498747825623,
      "learning_rate": 1.08292666429998e-05,
      "loss": 0.04394363611936569,
      "step": 4573
    },
    {
      "epoch": 0.6183744687323087,
      "grad_norm": 0.4957038462162018,
      "learning_rate": 1.08226925458435e-05,
      "loss": 0.05378865450620651,
      "step": 4574
    },
    {
      "epoch": 0.6185096621010739,
      "grad_norm": 0.520465612411499,
      "learning_rate": 1.0816119318424882e-05,
      "loss": 0.04268661141395569,
      "step": 4575
    },
    {
      "epoch": 0.6186448554698392,
      "grad_norm": 0.8219638466835022,
      "learning_rate": 1.0809546962112535e-05,
      "loss": 0.04037794470787048,
      "step": 4576
    },
    {
      "epoch": 0.6187800488386045,
      "grad_norm": 0.437211275100708,
      "learning_rate": 1.0802975478274856e-05,
      "loss": 0.04212292656302452,
      "step": 4577
    },
    {
      "epoch": 0.6189152422073697,
      "grad_norm": 0.6094338893890381,
      "learning_rate": 1.0796404868280062e-05,
      "loss": 0.046925559639930725,
      "step": 4578
    },
    {
      "epoch": 0.619050435576135,
      "grad_norm": 0.4211161732673645,
      "learning_rate": 1.07898351334962e-05,
      "loss": 0.04569794237613678,
      "step": 4579
    },
    {
      "epoch": 0.6191856289449003,
      "grad_norm": 0.5433511734008789,
      "learning_rate": 1.0783266275291103e-05,
      "loss": 0.04710765182971954,
      "step": 4580
    },
    {
      "epoch": 0.6193208223136655,
      "grad_norm": 0.2898409962654114,
      "learning_rate": 1.077669829503246e-05,
      "loss": 0.03926851600408554,
      "step": 4581
    },
    {
      "epoch": 0.6194560156824308,
      "grad_norm": 0.29439783096313477,
      "learning_rate": 1.077013119408775e-05,
      "loss": 0.047526340931653976,
      "step": 4582
    },
    {
      "epoch": 0.619591209051196,
      "grad_norm": 0.5454329252243042,
      "learning_rate": 1.0763564973824289e-05,
      "loss": 0.05025076866149902,
      "step": 4583
    },
    {
      "epoch": 0.6197264024199614,
      "grad_norm": 0.3629743158817291,
      "learning_rate": 1.0756999635609185e-05,
      "loss": 0.04767164587974548,
      "step": 4584
    },
    {
      "epoch": 0.6198615957887266,
      "grad_norm": 0.21775631606578827,
      "learning_rate": 1.0750435180809381e-05,
      "loss": 0.03224128857254982,
      "step": 4585
    },
    {
      "epoch": 0.6199967891574918,
      "grad_norm": 0.27913036942481995,
      "learning_rate": 1.074387161079164e-05,
      "loss": 0.04899683594703674,
      "step": 4586
    },
    {
      "epoch": 0.6201319825262571,
      "grad_norm": 0.753542423248291,
      "learning_rate": 1.0737308926922521e-05,
      "loss": 0.05364812910556793,
      "step": 4587
    },
    {
      "epoch": 0.6202671758950223,
      "grad_norm": 0.5278282165527344,
      "learning_rate": 1.0730747130568424e-05,
      "loss": 0.03552785515785217,
      "step": 4588
    },
    {
      "epoch": 0.6204023692637876,
      "grad_norm": 0.6113235354423523,
      "learning_rate": 1.0724186223095532e-05,
      "loss": 0.05290553718805313,
      "step": 4589
    },
    {
      "epoch": 0.6205375626325529,
      "grad_norm": 0.4825994372367859,
      "learning_rate": 1.071762620586987e-05,
      "loss": 0.05363726615905762,
      "step": 4590
    },
    {
      "epoch": 0.6206727560013181,
      "grad_norm": 0.7129722833633423,
      "learning_rate": 1.0711067080257273e-05,
      "loss": 0.04131750017404556,
      "step": 4591
    },
    {
      "epoch": 0.6208079493700834,
      "grad_norm": 0.7018037438392639,
      "learning_rate": 1.0704508847623374e-05,
      "loss": 0.04199056699872017,
      "step": 4592
    },
    {
      "epoch": 0.6209431427388487,
      "grad_norm": 0.36936163902282715,
      "learning_rate": 1.069795150933365e-05,
      "loss": 0.04437442123889923,
      "step": 4593
    },
    {
      "epoch": 0.6210783361076139,
      "grad_norm": 0.39472657442092896,
      "learning_rate": 1.0691395066753357e-05,
      "loss": 0.05951548367738724,
      "step": 4594
    },
    {
      "epoch": 0.6212135294763792,
      "grad_norm": 0.3580392897129059,
      "learning_rate": 1.0684839521247584e-05,
      "loss": 0.041038259863853455,
      "step": 4595
    },
    {
      "epoch": 0.6213487228451444,
      "grad_norm": 0.32261937856674194,
      "learning_rate": 1.0678284874181234e-05,
      "loss": 0.05707205832004547,
      "step": 4596
    },
    {
      "epoch": 0.6214839162139097,
      "grad_norm": 1.4168047904968262,
      "learning_rate": 1.0671731126919028e-05,
      "loss": 0.06272955983877182,
      "step": 4597
    },
    {
      "epoch": 0.621619109582675,
      "grad_norm": 0.8052384853363037,
      "learning_rate": 1.066517828082548e-05,
      "loss": 0.06737357378005981,
      "step": 4598
    },
    {
      "epoch": 0.6217543029514402,
      "grad_norm": 1.3140414953231812,
      "learning_rate": 1.0658626337264926e-05,
      "loss": 0.06402401626110077,
      "step": 4599
    },
    {
      "epoch": 0.6218894963202055,
      "grad_norm": 0.26091107726097107,
      "learning_rate": 1.0652075297601518e-05,
      "loss": 0.043483246117830276,
      "step": 4600
    },
    {
      "epoch": 0.6220246896889707,
      "grad_norm": 0.31169095635414124,
      "learning_rate": 1.0645525163199222e-05,
      "loss": 0.037219852209091187,
      "step": 4601
    },
    {
      "epoch": 0.6221598830577361,
      "grad_norm": 0.46487393975257874,
      "learning_rate": 1.063897593542181e-05,
      "loss": 0.03937599062919617,
      "step": 4602
    },
    {
      "epoch": 0.6222950764265013,
      "grad_norm": 1.5859410762786865,
      "learning_rate": 1.0632427615632864e-05,
      "loss": 0.05201828479766846,
      "step": 4603
    },
    {
      "epoch": 0.6224302697952665,
      "grad_norm": 0.5500233173370361,
      "learning_rate": 1.0625880205195776e-05,
      "loss": 0.06203152611851692,
      "step": 4604
    },
    {
      "epoch": 0.6225654631640318,
      "grad_norm": 0.8580653667449951,
      "learning_rate": 1.0619333705473754e-05,
      "loss": 0.05133706331253052,
      "step": 4605
    },
    {
      "epoch": 0.622700656532797,
      "grad_norm": 0.8754221200942993,
      "learning_rate": 1.0612788117829821e-05,
      "loss": 0.057396914809942245,
      "step": 4606
    },
    {
      "epoch": 0.6228358499015624,
      "grad_norm": 0.23293782770633698,
      "learning_rate": 1.0606243443626792e-05,
      "loss": 0.050726406276226044,
      "step": 4607
    },
    {
      "epoch": 0.6229710432703276,
      "grad_norm": 0.46297305822372437,
      "learning_rate": 1.0599699684227313e-05,
      "loss": 0.05203748494386673,
      "step": 4608
    },
    {
      "epoch": 0.6231062366390928,
      "grad_norm": 0.6241655945777893,
      "learning_rate": 1.0593156840993818e-05,
      "loss": 0.04434943199157715,
      "step": 4609
    },
    {
      "epoch": 0.6232414300078581,
      "grad_norm": 0.5743916630744934,
      "learning_rate": 1.0586614915288571e-05,
      "loss": 0.04693271219730377,
      "step": 4610
    },
    {
      "epoch": 0.6233766233766234,
      "grad_norm": 0.2689257264137268,
      "learning_rate": 1.0580073908473641e-05,
      "loss": 0.04017655551433563,
      "step": 4611
    },
    {
      "epoch": 0.6235118167453887,
      "grad_norm": 0.5774421095848083,
      "learning_rate": 1.0573533821910885e-05,
      "loss": 0.06909120082855225,
      "step": 4612
    },
    {
      "epoch": 0.6236470101141539,
      "grad_norm": 1.0421887636184692,
      "learning_rate": 1.0566994656961997e-05,
      "loss": 0.0610082745552063,
      "step": 4613
    },
    {
      "epoch": 0.6237822034829191,
      "grad_norm": 0.5665892958641052,
      "learning_rate": 1.0560456414988456e-05,
      "loss": 0.06724916398525238,
      "step": 4614
    },
    {
      "epoch": 0.6239173968516845,
      "grad_norm": 0.5606773495674133,
      "learning_rate": 1.0553919097351564e-05,
      "loss": 0.0555778369307518,
      "step": 4615
    },
    {
      "epoch": 0.6240525902204497,
      "grad_norm": 0.5149750709533691,
      "learning_rate": 1.0547382705412434e-05,
      "loss": 0.057661134749650955,
      "step": 4616
    },
    {
      "epoch": 0.6241877835892149,
      "grad_norm": 0.5562970042228699,
      "learning_rate": 1.054084724053196e-05,
      "loss": 0.05339311063289642,
      "step": 4617
    },
    {
      "epoch": 0.6243229769579802,
      "grad_norm": 0.6296172738075256,
      "learning_rate": 1.0534312704070875e-05,
      "loss": 0.050513237714767456,
      "step": 4618
    },
    {
      "epoch": 0.6244581703267454,
      "grad_norm": 0.5522633194923401,
      "learning_rate": 1.0527779097389695e-05,
      "loss": 0.04391420632600784,
      "step": 4619
    },
    {
      "epoch": 0.6245933636955108,
      "grad_norm": 0.5698334574699402,
      "learning_rate": 1.0521246421848762e-05,
      "loss": 0.04198046028614044,
      "step": 4620
    },
    {
      "epoch": 0.624728557064276,
      "grad_norm": 0.6919045448303223,
      "learning_rate": 1.0514714678808202e-05,
      "loss": 0.047321055084466934,
      "step": 4621
    },
    {
      "epoch": 0.6248637504330412,
      "grad_norm": 1.213399887084961,
      "learning_rate": 1.0508183869627962e-05,
      "loss": 0.06009151041507721,
      "step": 4622
    },
    {
      "epoch": 0.6249989438018065,
      "grad_norm": 0.48268258571624756,
      "learning_rate": 1.0501653995667798e-05,
      "loss": 0.06529982388019562,
      "step": 4623
    },
    {
      "epoch": 0.6251341371705718,
      "grad_norm": 0.59559166431427,
      "learning_rate": 1.0495125058287258e-05,
      "loss": 0.06405478715896606,
      "step": 4624
    },
    {
      "epoch": 0.6252693305393371,
      "grad_norm": 0.9471227526664734,
      "learning_rate": 1.0488597058845708e-05,
      "loss": 0.058127835392951965,
      "step": 4625
    },
    {
      "epoch": 0.6254045239081023,
      "grad_norm": 0.40131238102912903,
      "learning_rate": 1.0482069998702304e-05,
      "loss": 0.040883246809244156,
      "step": 4626
    },
    {
      "epoch": 0.6255397172768675,
      "grad_norm": 0.4903443157672882,
      "learning_rate": 1.0475543879216017e-05,
      "loss": 0.05748749524354935,
      "step": 4627
    },
    {
      "epoch": 0.6256749106456329,
      "grad_norm": 0.664860188961029,
      "learning_rate": 1.0469018701745626e-05,
      "loss": 0.04879584163427353,
      "step": 4628
    },
    {
      "epoch": 0.6258101040143981,
      "grad_norm": 0.9136778116226196,
      "learning_rate": 1.0462494467649704e-05,
      "loss": 0.04561778903007507,
      "step": 4629
    },
    {
      "epoch": 0.6259452973831634,
      "grad_norm": 1.187482476234436,
      "learning_rate": 1.045597117828663e-05,
      "loss": 0.04310613498091698,
      "step": 4630
    },
    {
      "epoch": 0.6260804907519286,
      "grad_norm": 0.5049716234207153,
      "learning_rate": 1.0449448835014586e-05,
      "loss": 0.06675244867801666,
      "step": 4631
    },
    {
      "epoch": 0.6262156841206938,
      "grad_norm": 0.41891464591026306,
      "learning_rate": 1.044292743919156e-05,
      "loss": 0.046847864985466,
      "step": 4632
    },
    {
      "epoch": 0.6263508774894592,
      "grad_norm": 0.43414586782455444,
      "learning_rate": 1.0436406992175343e-05,
      "loss": 0.06464985758066177,
      "step": 4633
    },
    {
      "epoch": 0.6264860708582244,
      "grad_norm": 0.5278609991073608,
      "learning_rate": 1.0429887495323532e-05,
      "loss": 0.051476627588272095,
      "step": 4634
    },
    {
      "epoch": 0.6266212642269897,
      "grad_norm": 0.6428268551826477,
      "learning_rate": 1.0423368949993512e-05,
      "loss": 0.06466898322105408,
      "step": 4635
    },
    {
      "epoch": 0.6267564575957549,
      "grad_norm": 0.34842076897621155,
      "learning_rate": 1.041685135754248e-05,
      "loss": 0.05078856647014618,
      "step": 4636
    },
    {
      "epoch": 0.6268916509645202,
      "grad_norm": 0.5631418228149414,
      "learning_rate": 1.0410334719327435e-05,
      "loss": 0.060221489518880844,
      "step": 4637
    },
    {
      "epoch": 0.6270268443332855,
      "grad_norm": 0.8179173469543457,
      "learning_rate": 1.0403819036705177e-05,
      "loss": 0.06180504709482193,
      "step": 4638
    },
    {
      "epoch": 0.6271620377020507,
      "grad_norm": 1.1750370264053345,
      "learning_rate": 1.0397304311032311e-05,
      "loss": 0.06148827075958252,
      "step": 4639
    },
    {
      "epoch": 0.627297231070816,
      "grad_norm": 0.24618911743164062,
      "learning_rate": 1.039079054366523e-05,
      "loss": 0.039470549672842026,
      "step": 4640
    },
    {
      "epoch": 0.6274324244395812,
      "grad_norm": 0.5818641185760498,
      "learning_rate": 1.0384277735960133e-05,
      "loss": 0.060404010117053986,
      "step": 4641
    },
    {
      "epoch": 0.6275676178083465,
      "grad_norm": 0.4214550256729126,
      "learning_rate": 1.0377765889273025e-05,
      "loss": 0.04228523373603821,
      "step": 4642
    },
    {
      "epoch": 0.6277028111771118,
      "grad_norm": 0.984075129032135,
      "learning_rate": 1.0371255004959715e-05,
      "loss": 0.07434651255607605,
      "step": 4643
    },
    {
      "epoch": 0.627838004545877,
      "grad_norm": 0.32798850536346436,
      "learning_rate": 1.036474508437579e-05,
      "loss": 0.03830292075872421,
      "step": 4644
    },
    {
      "epoch": 0.6279731979146422,
      "grad_norm": 0.4325157403945923,
      "learning_rate": 1.035823612887666e-05,
      "loss": 0.05093963444232941,
      "step": 4645
    },
    {
      "epoch": 0.6281083912834076,
      "grad_norm": 0.8938845992088318,
      "learning_rate": 1.0351728139817517e-05,
      "loss": 0.06327836215496063,
      "step": 4646
    },
    {
      "epoch": 0.6282435846521728,
      "grad_norm": 0.7308549880981445,
      "learning_rate": 1.0345221118553362e-05,
      "loss": 0.052430376410484314,
      "step": 4647
    },
    {
      "epoch": 0.6283787780209381,
      "grad_norm": 0.568248450756073,
      "learning_rate": 1.0338715066439002e-05,
      "loss": 0.050008200109004974,
      "step": 4648
    },
    {
      "epoch": 0.6285139713897033,
      "grad_norm": 0.7952597737312317,
      "learning_rate": 1.0332209984829013e-05,
      "loss": 0.06555724143981934,
      "step": 4649
    },
    {
      "epoch": 0.6286491647584685,
      "grad_norm": 0.36300814151763916,
      "learning_rate": 1.03257058750778e-05,
      "loss": 0.04265543073415756,
      "step": 4650
    },
    {
      "epoch": 0.6287843581272339,
      "grad_norm": 0.405932754278183,
      "learning_rate": 1.0319202738539548e-05,
      "loss": 0.04134620726108551,
      "step": 4651
    },
    {
      "epoch": 0.6289195514959991,
      "grad_norm": 0.49853989481925964,
      "learning_rate": 1.0312700576568253e-05,
      "loss": 0.05231449007987976,
      "step": 4652
    },
    {
      "epoch": 0.6290547448647644,
      "grad_norm": 1.1169902086257935,
      "learning_rate": 1.0306199390517688e-05,
      "loss": 0.056012049317359924,
      "step": 4653
    },
    {
      "epoch": 0.6291899382335296,
      "grad_norm": 0.5426188111305237,
      "learning_rate": 1.0299699181741439e-05,
      "loss": 0.07195337116718292,
      "step": 4654
    },
    {
      "epoch": 0.6293251316022949,
      "grad_norm": 1.414876937866211,
      "learning_rate": 1.0293199951592889e-05,
      "loss": 0.06186673045158386,
      "step": 4655
    },
    {
      "epoch": 0.6294603249710602,
      "grad_norm": 0.3539164960384369,
      "learning_rate": 1.0286701701425206e-05,
      "loss": 0.05881623178720474,
      "step": 4656
    },
    {
      "epoch": 0.6295955183398254,
      "grad_norm": 0.7021041512489319,
      "learning_rate": 1.0280204432591369e-05,
      "loss": 0.07057327032089233,
      "step": 4657
    },
    {
      "epoch": 0.6297307117085907,
      "grad_norm": 1.7062206268310547,
      "learning_rate": 1.0273708146444133e-05,
      "loss": 0.06410694867372513,
      "step": 4658
    },
    {
      "epoch": 0.629865905077356,
      "grad_norm": 0.8552284240722656,
      "learning_rate": 1.0267212844336062e-05,
      "loss": 0.0669512152671814,
      "step": 4659
    },
    {
      "epoch": 0.6300010984461212,
      "grad_norm": 0.22374765574932098,
      "learning_rate": 1.026071852761952e-05,
      "loss": 0.03679882735013962,
      "step": 4660
    },
    {
      "epoch": 0.6301362918148865,
      "grad_norm": 0.8955230712890625,
      "learning_rate": 1.025422519764665e-05,
      "loss": 0.04569307714700699,
      "step": 4661
    },
    {
      "epoch": 0.6302714851836517,
      "grad_norm": 1.1031315326690674,
      "learning_rate": 1.024773285576941e-05,
      "loss": 0.04860438406467438,
      "step": 4662
    },
    {
      "epoch": 0.630406678552417,
      "grad_norm": 0.34250757098197937,
      "learning_rate": 1.0241241503339524e-05,
      "loss": 0.04318247362971306,
      "step": 4663
    },
    {
      "epoch": 0.6305418719211823,
      "grad_norm": 0.6776455640792847,
      "learning_rate": 1.023475114170853e-05,
      "loss": 0.05851234123110771,
      "step": 4664
    },
    {
      "epoch": 0.6306770652899475,
      "grad_norm": 0.6967629790306091,
      "learning_rate": 1.0228261772227768e-05,
      "loss": 0.0459999218583107,
      "step": 4665
    },
    {
      "epoch": 0.6308122586587128,
      "grad_norm": 0.7477776408195496,
      "learning_rate": 1.0221773396248349e-05,
      "loss": 0.06580895185470581,
      "step": 4666
    },
    {
      "epoch": 0.630947452027478,
      "grad_norm": 1.4138846397399902,
      "learning_rate": 1.021528601512119e-05,
      "loss": 0.06900068372488022,
      "step": 4667
    },
    {
      "epoch": 0.6310826453962434,
      "grad_norm": 0.9229210615158081,
      "learning_rate": 1.0208799630196994e-05,
      "loss": 0.0526098906993866,
      "step": 4668
    },
    {
      "epoch": 0.6312178387650086,
      "grad_norm": 0.5856477618217468,
      "learning_rate": 1.0202314242826264e-05,
      "loss": 0.05838462710380554,
      "step": 4669
    },
    {
      "epoch": 0.6313530321337738,
      "grad_norm": 0.31988847255706787,
      "learning_rate": 1.0195829854359299e-05,
      "loss": 0.0435689240694046,
      "step": 4670
    },
    {
      "epoch": 0.6314882255025391,
      "grad_norm": 0.37234628200531006,
      "learning_rate": 1.0189346466146175e-05,
      "loss": 0.043655455112457275,
      "step": 4671
    },
    {
      "epoch": 0.6316234188713044,
      "grad_norm": 0.4791053533554077,
      "learning_rate": 1.018286407953677e-05,
      "loss": 0.056932225823402405,
      "step": 4672
    },
    {
      "epoch": 0.6317586122400696,
      "grad_norm": 0.25158461928367615,
      "learning_rate": 1.017638269588075e-05,
      "loss": 0.05398217588663101,
      "step": 4673
    },
    {
      "epoch": 0.6318938056088349,
      "grad_norm": 0.9145091772079468,
      "learning_rate": 1.0169902316527575e-05,
      "loss": 0.07113750278949738,
      "step": 4674
    },
    {
      "epoch": 0.6320289989776001,
      "grad_norm": 0.7961694002151489,
      "learning_rate": 1.0163422942826502e-05,
      "loss": 0.06181144714355469,
      "step": 4675
    },
    {
      "epoch": 0.6321641923463654,
      "grad_norm": 0.5894361734390259,
      "learning_rate": 1.0156944576126555e-05,
      "loss": 0.04805958271026611,
      "step": 4676
    },
    {
      "epoch": 0.6322993857151307,
      "grad_norm": 0.2800850570201874,
      "learning_rate": 1.0150467217776579e-05,
      "loss": 0.049652036279439926,
      "step": 4677
    },
    {
      "epoch": 0.6324345790838959,
      "grad_norm": 0.4588675796985626,
      "learning_rate": 1.0143990869125185e-05,
      "loss": 0.053740665316581726,
      "step": 4678
    },
    {
      "epoch": 0.6325697724526612,
      "grad_norm": 0.4305167496204376,
      "learning_rate": 1.013751553152079e-05,
      "loss": 0.06193351745605469,
      "step": 4679
    },
    {
      "epoch": 0.6327049658214264,
      "grad_norm": 0.4192427694797516,
      "learning_rate": 1.0131041206311594e-05,
      "loss": 0.0593588650226593,
      "step": 4680
    },
    {
      "epoch": 0.6328401591901918,
      "grad_norm": 0.48952728509902954,
      "learning_rate": 1.0124567894845578e-05,
      "loss": 0.06378231942653656,
      "step": 4681
    },
    {
      "epoch": 0.632975352558957,
      "grad_norm": 0.7123900055885315,
      "learning_rate": 1.0118095598470528e-05,
      "loss": 0.04663345590233803,
      "step": 4682
    },
    {
      "epoch": 0.6331105459277222,
      "grad_norm": 0.3992023169994354,
      "learning_rate": 1.0111624318534006e-05,
      "loss": 0.04774162545800209,
      "step": 4683
    },
    {
      "epoch": 0.6332457392964875,
      "grad_norm": 1.074270486831665,
      "learning_rate": 1.0105154056383377e-05,
      "loss": 0.04648825526237488,
      "step": 4684
    },
    {
      "epoch": 0.6333809326652527,
      "grad_norm": 1.688263177871704,
      "learning_rate": 1.0098684813365764e-05,
      "loss": 0.058587878942489624,
      "step": 4685
    },
    {
      "epoch": 0.6335161260340181,
      "grad_norm": 0.4962066113948822,
      "learning_rate": 1.0092216590828115e-05,
      "loss": 0.03800930455327034,
      "step": 4686
    },
    {
      "epoch": 0.6336513194027833,
      "grad_norm": 0.9367076754570007,
      "learning_rate": 1.0085749390117146e-05,
      "loss": 0.05098221078515053,
      "step": 4687
    },
    {
      "epoch": 0.6337865127715485,
      "grad_norm": 0.3751489818096161,
      "learning_rate": 1.0079283212579354e-05,
      "loss": 0.04532918334007263,
      "step": 4688
    },
    {
      "epoch": 0.6339217061403138,
      "grad_norm": 0.5485007166862488,
      "learning_rate": 1.0072818059561045e-05,
      "loss": 0.041458070278167725,
      "step": 4689
    },
    {
      "epoch": 0.6340568995090791,
      "grad_norm": 0.595403254032135,
      "learning_rate": 1.0066353932408285e-05,
      "loss": 0.047010332345962524,
      "step": 4690
    },
    {
      "epoch": 0.6341920928778444,
      "grad_norm": 0.21888503432273865,
      "learning_rate": 1.0059890832466948e-05,
      "loss": 0.05118374526500702,
      "step": 4691
    },
    {
      "epoch": 0.6343272862466096,
      "grad_norm": 0.32095620036125183,
      "learning_rate": 1.0053428761082684e-05,
      "loss": 0.04708552360534668,
      "step": 4692
    },
    {
      "epoch": 0.6344624796153748,
      "grad_norm": 0.23541928827762604,
      "learning_rate": 1.0046967719600927e-05,
      "loss": 0.05972860008478165,
      "step": 4693
    },
    {
      "epoch": 0.6345976729841402,
      "grad_norm": 0.3196291923522949,
      "learning_rate": 1.0040507709366912e-05,
      "loss": 0.04170640558004379,
      "step": 4694
    },
    {
      "epoch": 0.6347328663529054,
      "grad_norm": 1.1211464405059814,
      "learning_rate": 1.0034048731725631e-05,
      "loss": 0.08250119537115097,
      "step": 4695
    },
    {
      "epoch": 0.6348680597216706,
      "grad_norm": 0.5909939408302307,
      "learning_rate": 1.0027590788021886e-05,
      "loss": 0.04364566504955292,
      "step": 4696
    },
    {
      "epoch": 0.6350032530904359,
      "grad_norm": 1.281913161277771,
      "learning_rate": 1.0021133879600258e-05,
      "loss": 0.05728508532047272,
      "step": 4697
    },
    {
      "epoch": 0.6351384464592011,
      "grad_norm": 0.33976858854293823,
      "learning_rate": 1.0014678007805108e-05,
      "loss": 0.05651407688856125,
      "step": 4698
    },
    {
      "epoch": 0.6352736398279665,
      "grad_norm": 0.3187199831008911,
      "learning_rate": 1.0008223173980579e-05,
      "loss": 0.05691501498222351,
      "step": 4699
    },
    {
      "epoch": 0.6354088331967317,
      "grad_norm": 0.29178905487060547,
      "learning_rate": 1.0001769379470604e-05,
      "loss": 0.03464853763580322,
      "step": 4700
    },
    {
      "epoch": 0.6355440265654969,
      "grad_norm": 0.48736095428466797,
      "learning_rate": 9.995316625618898e-06,
      "loss": 0.056324444711208344,
      "step": 4701
    },
    {
      "epoch": 0.6356792199342622,
      "grad_norm": 0.4598604440689087,
      "learning_rate": 9.988864913768962e-06,
      "loss": 0.06390278786420822,
      "step": 4702
    },
    {
      "epoch": 0.6358144133030275,
      "grad_norm": 0.3368128538131714,
      "learning_rate": 9.982414245264071e-06,
      "loss": 0.03932724520564079,
      "step": 4703
    },
    {
      "epoch": 0.6359496066717928,
      "grad_norm": 0.8145586848258972,
      "learning_rate": 9.975964621447293e-06,
      "loss": 0.05152442306280136,
      "step": 4704
    },
    {
      "epoch": 0.636084800040558,
      "grad_norm": 1.0392258167266846,
      "learning_rate": 9.96951604366147e-06,
      "loss": 0.04404053837060928,
      "step": 4705
    },
    {
      "epoch": 0.6362199934093232,
      "grad_norm": 0.2877473533153534,
      "learning_rate": 9.963068513249233e-06,
      "loss": 0.055663272738456726,
      "step": 4706
    },
    {
      "epoch": 0.6363551867780886,
      "grad_norm": 0.18260252475738525,
      "learning_rate": 9.956622031552996e-06,
      "loss": 0.03911535441875458,
      "step": 4707
    },
    {
      "epoch": 0.6364903801468538,
      "grad_norm": 0.6719172596931458,
      "learning_rate": 9.950176599914942e-06,
      "loss": 0.044287413358688354,
      "step": 4708
    },
    {
      "epoch": 0.6366255735156191,
      "grad_norm": 0.2303326278924942,
      "learning_rate": 9.943732219677048e-06,
      "loss": 0.03544317185878754,
      "step": 4709
    },
    {
      "epoch": 0.6367607668843843,
      "grad_norm": 0.5941929817199707,
      "learning_rate": 9.93728889218107e-06,
      "loss": 0.03194409981369972,
      "step": 4710
    },
    {
      "epoch": 0.6368959602531495,
      "grad_norm": 0.40900540351867676,
      "learning_rate": 9.930846618768543e-06,
      "loss": 0.05201096832752228,
      "step": 4711
    },
    {
      "epoch": 0.6370311536219149,
      "grad_norm": 0.6259360909461975,
      "learning_rate": 9.924405400780784e-06,
      "loss": 0.046068064868450165,
      "step": 4712
    },
    {
      "epoch": 0.6371663469906801,
      "grad_norm": 0.5820388793945312,
      "learning_rate": 9.917965239558885e-06,
      "loss": 0.03427599370479584,
      "step": 4713
    },
    {
      "epoch": 0.6373015403594454,
      "grad_norm": 0.33634090423583984,
      "learning_rate": 9.911526136443726e-06,
      "loss": 0.05215923488140106,
      "step": 4714
    },
    {
      "epoch": 0.6374367337282106,
      "grad_norm": 0.8394302129745483,
      "learning_rate": 9.905088092775956e-06,
      "loss": 0.046040382236242294,
      "step": 4715
    },
    {
      "epoch": 0.6375719270969759,
      "grad_norm": 0.8640695214271545,
      "learning_rate": 9.898651109896015e-06,
      "loss": 0.05065205693244934,
      "step": 4716
    },
    {
      "epoch": 0.6377071204657412,
      "grad_norm": 0.9978101253509521,
      "learning_rate": 9.892215189144123e-06,
      "loss": 0.06882886588573456,
      "step": 4717
    },
    {
      "epoch": 0.6378423138345064,
      "grad_norm": 0.4321083724498749,
      "learning_rate": 9.88578033186026e-06,
      "loss": 0.04601617157459259,
      "step": 4718
    },
    {
      "epoch": 0.6379775072032717,
      "grad_norm": 0.35456937551498413,
      "learning_rate": 9.879346539384207e-06,
      "loss": 0.03964271396398544,
      "step": 4719
    },
    {
      "epoch": 0.638112700572037,
      "grad_norm": 0.7505684494972229,
      "learning_rate": 9.87291381305551e-06,
      "loss": 0.053060080856084824,
      "step": 4720
    },
    {
      "epoch": 0.6382478939408022,
      "grad_norm": 0.21678295731544495,
      "learning_rate": 9.866482154213502e-06,
      "loss": 0.04147632420063019,
      "step": 4721
    },
    {
      "epoch": 0.6383830873095675,
      "grad_norm": 0.400951087474823,
      "learning_rate": 9.86005156419728e-06,
      "loss": 0.06529366225004196,
      "step": 4722
    },
    {
      "epoch": 0.6385182806783327,
      "grad_norm": 0.6195996403694153,
      "learning_rate": 9.853622044345732e-06,
      "loss": 0.06088293343782425,
      "step": 4723
    },
    {
      "epoch": 0.6386534740470979,
      "grad_norm": 0.6974532008171082,
      "learning_rate": 9.847193595997522e-06,
      "loss": 0.06185964494943619,
      "step": 4724
    },
    {
      "epoch": 0.6387886674158633,
      "grad_norm": 0.675674319267273,
      "learning_rate": 9.840766220491078e-06,
      "loss": 0.047476764768362045,
      "step": 4725
    },
    {
      "epoch": 0.6389238607846285,
      "grad_norm": 0.5759568214416504,
      "learning_rate": 9.834339919164625e-06,
      "loss": 0.05363106727600098,
      "step": 4726
    },
    {
      "epoch": 0.6390590541533938,
      "grad_norm": 1.741191029548645,
      "learning_rate": 9.827914693356145e-06,
      "loss": 0.05436190217733383,
      "step": 4727
    },
    {
      "epoch": 0.639194247522159,
      "grad_norm": 0.5656270384788513,
      "learning_rate": 9.821490544403403e-06,
      "loss": 0.056480370461940765,
      "step": 4728
    },
    {
      "epoch": 0.6393294408909242,
      "grad_norm": 0.3221462070941925,
      "learning_rate": 9.815067473643951e-06,
      "loss": 0.05631650239229202,
      "step": 4729
    },
    {
      "epoch": 0.6394646342596896,
      "grad_norm": 0.3064517676830292,
      "learning_rate": 9.808645482415097e-06,
      "loss": 0.03556741401553154,
      "step": 4730
    },
    {
      "epoch": 0.6395998276284548,
      "grad_norm": 0.6442540884017944,
      "learning_rate": 9.80222457205394e-06,
      "loss": 0.039395034313201904,
      "step": 4731
    },
    {
      "epoch": 0.6397350209972201,
      "grad_norm": 0.5686182975769043,
      "learning_rate": 9.795804743897341e-06,
      "loss": 0.05261033773422241,
      "step": 4732
    },
    {
      "epoch": 0.6398702143659853,
      "grad_norm": 0.3568746745586395,
      "learning_rate": 9.789385999281948e-06,
      "loss": 0.05085264891386032,
      "step": 4733
    },
    {
      "epoch": 0.6400054077347506,
      "grad_norm": 1.2988499402999878,
      "learning_rate": 9.782968339544179e-06,
      "loss": 0.057544857263565063,
      "step": 4734
    },
    {
      "epoch": 0.6401406011035159,
      "grad_norm": 0.5114401578903198,
      "learning_rate": 9.776551766020219e-06,
      "loss": 0.06983036547899246,
      "step": 4735
    },
    {
      "epoch": 0.6402757944722811,
      "grad_norm": 0.3355632722377777,
      "learning_rate": 9.77013628004604e-06,
      "loss": 0.04666588455438614,
      "step": 4736
    },
    {
      "epoch": 0.6404109878410464,
      "grad_norm": 0.3591151833534241,
      "learning_rate": 9.763721882957371e-06,
      "loss": 0.0373477041721344,
      "step": 4737
    },
    {
      "epoch": 0.6405461812098117,
      "grad_norm": 0.5469942688941956,
      "learning_rate": 9.757308576089732e-06,
      "loss": 0.044421806931495667,
      "step": 4738
    },
    {
      "epoch": 0.6406813745785769,
      "grad_norm": 0.5699624419212341,
      "learning_rate": 9.750896360778404e-06,
      "loss": 0.053258977830410004,
      "step": 4739
    },
    {
      "epoch": 0.6408165679473422,
      "grad_norm": 0.6950234174728394,
      "learning_rate": 9.744485238358448e-06,
      "loss": 0.04999391734600067,
      "step": 4740
    },
    {
      "epoch": 0.6409517613161074,
      "grad_norm": 0.3082069158554077,
      "learning_rate": 9.73807521016469e-06,
      "loss": 0.03505382686853409,
      "step": 4741
    },
    {
      "epoch": 0.6410869546848728,
      "grad_norm": 0.6573958396911621,
      "learning_rate": 9.731666277531732e-06,
      "loss": 0.045720405876636505,
      "step": 4742
    },
    {
      "epoch": 0.641222148053638,
      "grad_norm": 0.31336405873298645,
      "learning_rate": 9.725258441793947e-06,
      "loss": 0.040391504764556885,
      "step": 4743
    },
    {
      "epoch": 0.6413573414224032,
      "grad_norm": 0.3726145923137665,
      "learning_rate": 9.71885170428549e-06,
      "loss": 0.062105149030685425,
      "step": 4744
    },
    {
      "epoch": 0.6414925347911685,
      "grad_norm": 0.24680550396442413,
      "learning_rate": 9.712446066340265e-06,
      "loss": 0.041123002767562866,
      "step": 4745
    },
    {
      "epoch": 0.6416277281599337,
      "grad_norm": 1.074472427368164,
      "learning_rate": 9.70604152929197e-06,
      "loss": 0.06705289334058762,
      "step": 4746
    },
    {
      "epoch": 0.6417629215286991,
      "grad_norm": 1.0206831693649292,
      "learning_rate": 9.699638094474054e-06,
      "loss": 0.061657726764678955,
      "step": 4747
    },
    {
      "epoch": 0.6418981148974643,
      "grad_norm": 0.4847748875617981,
      "learning_rate": 9.693235763219752e-06,
      "loss": 0.036822669208049774,
      "step": 4748
    },
    {
      "epoch": 0.6420333082662295,
      "grad_norm": 0.5083960294723511,
      "learning_rate": 9.68683453686207e-06,
      "loss": 0.06037784367799759,
      "step": 4749
    },
    {
      "epoch": 0.6421685016349948,
      "grad_norm": 0.33866918087005615,
      "learning_rate": 9.680434416733763e-06,
      "loss": 0.05275866761803627,
      "step": 4750
    },
    {
      "epoch": 0.6423036950037601,
      "grad_norm": 0.6149324178695679,
      "learning_rate": 9.674035404167381e-06,
      "loss": 0.04908076301217079,
      "step": 4751
    },
    {
      "epoch": 0.6424388883725253,
      "grad_norm": 0.8515546321868896,
      "learning_rate": 9.66763750049523e-06,
      "loss": 0.06292057782411575,
      "step": 4752
    },
    {
      "epoch": 0.6425740817412906,
      "grad_norm": 1.4440233707427979,
      "learning_rate": 9.66124070704939e-06,
      "loss": 0.07689607888460159,
      "step": 4753
    },
    {
      "epoch": 0.6427092751100558,
      "grad_norm": 0.6262544393539429,
      "learning_rate": 9.654845025161699e-06,
      "loss": 0.03946412354707718,
      "step": 4754
    },
    {
      "epoch": 0.6428444684788212,
      "grad_norm": 0.4633781909942627,
      "learning_rate": 9.648450456163777e-06,
      "loss": 0.04835525155067444,
      "step": 4755
    },
    {
      "epoch": 0.6429796618475864,
      "grad_norm": 0.7444468140602112,
      "learning_rate": 9.64205700138701e-06,
      "loss": 0.04709312319755554,
      "step": 4756
    },
    {
      "epoch": 0.6431148552163516,
      "grad_norm": 0.5160643458366394,
      "learning_rate": 9.635664662162548e-06,
      "loss": 0.05223727226257324,
      "step": 4757
    },
    {
      "epoch": 0.6432500485851169,
      "grad_norm": 0.3359399139881134,
      "learning_rate": 9.629273439821315e-06,
      "loss": 0.054152511060237885,
      "step": 4758
    },
    {
      "epoch": 0.6433852419538821,
      "grad_norm": 0.4551103115081787,
      "learning_rate": 9.622883335693984e-06,
      "loss": 0.05163699388504028,
      "step": 4759
    },
    {
      "epoch": 0.6435204353226475,
      "grad_norm": 0.4248725473880768,
      "learning_rate": 9.616494351111017e-06,
      "loss": 0.04353562742471695,
      "step": 4760
    },
    {
      "epoch": 0.6436556286914127,
      "grad_norm": 0.4266912639141083,
      "learning_rate": 9.610106487402637e-06,
      "loss": 0.056361302733421326,
      "step": 4761
    },
    {
      "epoch": 0.6437908220601779,
      "grad_norm": 0.7169679403305054,
      "learning_rate": 9.603719745898826e-06,
      "loss": 0.059100985527038574,
      "step": 4762
    },
    {
      "epoch": 0.6439260154289432,
      "grad_norm": 0.9839428067207336,
      "learning_rate": 9.597334127929346e-06,
      "loss": 0.05191851034760475,
      "step": 4763
    },
    {
      "epoch": 0.6440612087977085,
      "grad_norm": 0.5494467616081238,
      "learning_rate": 9.590949634823707e-06,
      "loss": 0.05256250873208046,
      "step": 4764
    },
    {
      "epoch": 0.6441964021664738,
      "grad_norm": 0.28033575415611267,
      "learning_rate": 9.584566267911198e-06,
      "loss": 0.0662446916103363,
      "step": 4765
    },
    {
      "epoch": 0.644331595535239,
      "grad_norm": 0.8343197703361511,
      "learning_rate": 9.578184028520874e-06,
      "loss": 0.05676156282424927,
      "step": 4766
    },
    {
      "epoch": 0.6444667889040042,
      "grad_norm": 0.5727573037147522,
      "learning_rate": 9.571802917981548e-06,
      "loss": 0.03550533577799797,
      "step": 4767
    },
    {
      "epoch": 0.6446019822727695,
      "grad_norm": 0.8502346277236938,
      "learning_rate": 9.565422937621798e-06,
      "loss": 0.042402856051921844,
      "step": 4768
    },
    {
      "epoch": 0.6447371756415348,
      "grad_norm": 0.5257234573364258,
      "learning_rate": 9.559044088769971e-06,
      "loss": 0.057161688804626465,
      "step": 4769
    },
    {
      "epoch": 0.6448723690103001,
      "grad_norm": 0.7567318677902222,
      "learning_rate": 9.552666372754182e-06,
      "loss": 0.05372577905654907,
      "step": 4770
    },
    {
      "epoch": 0.6450075623790653,
      "grad_norm": 0.4655241370201111,
      "learning_rate": 9.546289790902307e-06,
      "loss": 0.03939799964427948,
      "step": 4771
    },
    {
      "epoch": 0.6451427557478305,
      "grad_norm": 0.6124745011329651,
      "learning_rate": 9.539914344541976e-06,
      "loss": 0.047296732664108276,
      "step": 4772
    },
    {
      "epoch": 0.6452779491165959,
      "grad_norm": 0.24542281031608582,
      "learning_rate": 9.533540035000598e-06,
      "loss": 0.047902628779411316,
      "step": 4773
    },
    {
      "epoch": 0.6454131424853611,
      "grad_norm": 0.4496416747570038,
      "learning_rate": 9.52716686360533e-06,
      "loss": 0.05026061832904816,
      "step": 4774
    },
    {
      "epoch": 0.6455483358541264,
      "grad_norm": 0.5349769592285156,
      "learning_rate": 9.520794831683108e-06,
      "loss": 0.053541723638772964,
      "step": 4775
    },
    {
      "epoch": 0.6456835292228916,
      "grad_norm": 1.1830024719238281,
      "learning_rate": 9.514423940560627e-06,
      "loss": 0.05755196139216423,
      "step": 4776
    },
    {
      "epoch": 0.6458187225916568,
      "grad_norm": 0.46192553639411926,
      "learning_rate": 9.508054191564326e-06,
      "loss": 0.04821890592575073,
      "step": 4777
    },
    {
      "epoch": 0.6459539159604222,
      "grad_norm": 0.36615103483200073,
      "learning_rate": 9.501685586020434e-06,
      "loss": 0.04434036463499069,
      "step": 4778
    },
    {
      "epoch": 0.6460891093291874,
      "grad_norm": 1.0897423028945923,
      "learning_rate": 9.495318125254919e-06,
      "loss": 0.06512558460235596,
      "step": 4779
    },
    {
      "epoch": 0.6462243026979526,
      "grad_norm": 0.685386598110199,
      "learning_rate": 9.488951810593527e-06,
      "loss": 0.06861716508865356,
      "step": 4780
    },
    {
      "epoch": 0.6463594960667179,
      "grad_norm": 0.6100823879241943,
      "learning_rate": 9.48258664336176e-06,
      "loss": 0.0717424675822258,
      "step": 4781
    },
    {
      "epoch": 0.6464946894354832,
      "grad_norm": 0.46544957160949707,
      "learning_rate": 9.476222624884873e-06,
      "loss": 0.055689986795186996,
      "step": 4782
    },
    {
      "epoch": 0.6466298828042485,
      "grad_norm": 0.4717201590538025,
      "learning_rate": 9.469859756487893e-06,
      "loss": 0.05503963679075241,
      "step": 4783
    },
    {
      "epoch": 0.6467650761730137,
      "grad_norm": 0.6445606350898743,
      "learning_rate": 9.463498039495598e-06,
      "loss": 0.08194366097450256,
      "step": 4784
    },
    {
      "epoch": 0.6469002695417789,
      "grad_norm": 0.6095782518386841,
      "learning_rate": 9.457137475232537e-06,
      "loss": 0.060378074645996094,
      "step": 4785
    },
    {
      "epoch": 0.6470354629105443,
      "grad_norm": 0.4055345058441162,
      "learning_rate": 9.450778065023019e-06,
      "loss": 0.05957464128732681,
      "step": 4786
    },
    {
      "epoch": 0.6471706562793095,
      "grad_norm": 0.547907292842865,
      "learning_rate": 9.444419810191091e-06,
      "loss": 0.05349154770374298,
      "step": 4787
    },
    {
      "epoch": 0.6473058496480748,
      "grad_norm": 0.43033722043037415,
      "learning_rate": 9.43806271206059e-06,
      "loss": 0.04731358587741852,
      "step": 4788
    },
    {
      "epoch": 0.64744104301684,
      "grad_norm": 1.0177619457244873,
      "learning_rate": 9.431706771955089e-06,
      "loss": 0.06784875690937042,
      "step": 4789
    },
    {
      "epoch": 0.6475762363856052,
      "grad_norm": 0.7726863622665405,
      "learning_rate": 9.425351991197937e-06,
      "loss": 0.06031767278909683,
      "step": 4790
    },
    {
      "epoch": 0.6477114297543706,
      "grad_norm": 0.41456326842308044,
      "learning_rate": 9.418998371112221e-06,
      "loss": 0.046337857842445374,
      "step": 4791
    },
    {
      "epoch": 0.6478466231231358,
      "grad_norm": 0.39923757314682007,
      "learning_rate": 9.412645913020807e-06,
      "loss": 0.055342741310596466,
      "step": 4792
    },
    {
      "epoch": 0.6479818164919011,
      "grad_norm": 1.0042386054992676,
      "learning_rate": 9.406294618246313e-06,
      "loss": 0.05199577659368515,
      "step": 4793
    },
    {
      "epoch": 0.6481170098606663,
      "grad_norm": 0.6410531401634216,
      "learning_rate": 9.399944488111103e-06,
      "loss": 0.05791462957859039,
      "step": 4794
    },
    {
      "epoch": 0.6482522032294316,
      "grad_norm": 0.37604424357414246,
      "learning_rate": 9.39359552393732e-06,
      "loss": 0.04712918400764465,
      "step": 4795
    },
    {
      "epoch": 0.6483873965981969,
      "grad_norm": 0.33884933590888977,
      "learning_rate": 9.387247727046845e-06,
      "loss": 0.05989651381969452,
      "step": 4796
    },
    {
      "epoch": 0.6485225899669621,
      "grad_norm": 0.6658311486244202,
      "learning_rate": 9.380901098761319e-06,
      "loss": 0.05861292779445648,
      "step": 4797
    },
    {
      "epoch": 0.6486577833357274,
      "grad_norm": 0.5477061867713928,
      "learning_rate": 9.374555640402153e-06,
      "loss": 0.07044249773025513,
      "step": 4798
    },
    {
      "epoch": 0.6487929767044927,
      "grad_norm": 0.9148610830307007,
      "learning_rate": 9.368211353290503e-06,
      "loss": 0.06327980756759644,
      "step": 4799
    },
    {
      "epoch": 0.6489281700732579,
      "grad_norm": 1.0573673248291016,
      "learning_rate": 9.36186823874728e-06,
      "loss": 0.04950319230556488,
      "step": 4800
    },
    {
      "epoch": 0.6490633634420232,
      "grad_norm": 0.45756450295448303,
      "learning_rate": 9.355526298093152e-06,
      "loss": 0.05846443399786949,
      "step": 4801
    },
    {
      "epoch": 0.6491985568107884,
      "grad_norm": 0.3989078104496002,
      "learning_rate": 9.34918553264855e-06,
      "loss": 0.05459968000650406,
      "step": 4802
    },
    {
      "epoch": 0.6493337501795537,
      "grad_norm": 0.48171260952949524,
      "learning_rate": 9.342845943733658e-06,
      "loss": 0.05709845572710037,
      "step": 4803
    },
    {
      "epoch": 0.649468943548319,
      "grad_norm": 0.8460701704025269,
      "learning_rate": 9.336507532668407e-06,
      "loss": 0.04590034484863281,
      "step": 4804
    },
    {
      "epoch": 0.6496041369170842,
      "grad_norm": 0.6834648847579956,
      "learning_rate": 9.33017030077249e-06,
      "loss": 0.05731251835823059,
      "step": 4805
    },
    {
      "epoch": 0.6497393302858495,
      "grad_norm": 0.4440569579601288,
      "learning_rate": 9.323834249365346e-06,
      "loss": 0.039629578590393066,
      "step": 4806
    },
    {
      "epoch": 0.6498745236546147,
      "grad_norm": 0.35447558760643005,
      "learning_rate": 9.317499379766183e-06,
      "loss": 0.04773441702127457,
      "step": 4807
    },
    {
      "epoch": 0.65000971702338,
      "grad_norm": 0.40892234444618225,
      "learning_rate": 9.311165693293954e-06,
      "loss": 0.05303184688091278,
      "step": 4808
    },
    {
      "epoch": 0.6501449103921453,
      "grad_norm": 0.5478176474571228,
      "learning_rate": 9.304833191267364e-06,
      "loss": 0.051091477274894714,
      "step": 4809
    },
    {
      "epoch": 0.6502801037609105,
      "grad_norm": 0.3483165204524994,
      "learning_rate": 9.298501875004874e-06,
      "loss": 0.04721619188785553,
      "step": 4810
    },
    {
      "epoch": 0.6504152971296758,
      "grad_norm": 0.3029663860797882,
      "learning_rate": 9.292171745824695e-06,
      "loss": 0.05366624891757965,
      "step": 4811
    },
    {
      "epoch": 0.650550490498441,
      "grad_norm": 0.21481314301490784,
      "learning_rate": 9.285842805044797e-06,
      "loss": 0.03443337231874466,
      "step": 4812
    },
    {
      "epoch": 0.6506856838672063,
      "grad_norm": 0.36269816756248474,
      "learning_rate": 9.279515053982905e-06,
      "loss": 0.059001073241233826,
      "step": 4813
    },
    {
      "epoch": 0.6508208772359716,
      "grad_norm": 0.467576801776886,
      "learning_rate": 9.273188493956476e-06,
      "loss": 0.0521400161087513,
      "step": 4814
    },
    {
      "epoch": 0.6509560706047368,
      "grad_norm": 0.4456583559513092,
      "learning_rate": 9.266863126282746e-06,
      "loss": 0.05726826190948486,
      "step": 4815
    },
    {
      "epoch": 0.6510912639735021,
      "grad_norm": 1.0652563571929932,
      "learning_rate": 9.260538952278683e-06,
      "loss": 0.06769102811813354,
      "step": 4816
    },
    {
      "epoch": 0.6512264573422674,
      "grad_norm": 0.3957787752151489,
      "learning_rate": 9.254215973261014e-06,
      "loss": 0.0648040771484375,
      "step": 4817
    },
    {
      "epoch": 0.6513616507110326,
      "grad_norm": 0.7416332364082336,
      "learning_rate": 9.247894190546228e-06,
      "loss": 0.04052651673555374,
      "step": 4818
    },
    {
      "epoch": 0.6514968440797979,
      "grad_norm": 0.7028465270996094,
      "learning_rate": 9.241573605450539e-06,
      "loss": 0.04837683215737343,
      "step": 4819
    },
    {
      "epoch": 0.6516320374485631,
      "grad_norm": 0.6483571529388428,
      "learning_rate": 9.235254219289937e-06,
      "loss": 0.05493997037410736,
      "step": 4820
    },
    {
      "epoch": 0.6517672308173285,
      "grad_norm": 1.38758385181427,
      "learning_rate": 9.228936033380143e-06,
      "loss": 0.05247645080089569,
      "step": 4821
    },
    {
      "epoch": 0.6519024241860937,
      "grad_norm": 0.5732993483543396,
      "learning_rate": 9.222619049036649e-06,
      "loss": 0.04802927002310753,
      "step": 4822
    },
    {
      "epoch": 0.6520376175548589,
      "grad_norm": 0.44083142280578613,
      "learning_rate": 9.216303267574674e-06,
      "loss": 0.05586346238851547,
      "step": 4823
    },
    {
      "epoch": 0.6521728109236242,
      "grad_norm": 0.39318540692329407,
      "learning_rate": 9.209988690309198e-06,
      "loss": 0.05186024308204651,
      "step": 4824
    },
    {
      "epoch": 0.6523080042923894,
      "grad_norm": 0.77596116065979,
      "learning_rate": 9.203675318554956e-06,
      "loss": 0.0651119127869606,
      "step": 4825
    },
    {
      "epoch": 0.6524431976611548,
      "grad_norm": 0.9065393209457397,
      "learning_rate": 9.19736315362642e-06,
      "loss": 0.054299451410770416,
      "step": 4826
    },
    {
      "epoch": 0.65257839102992,
      "grad_norm": 0.5990732312202454,
      "learning_rate": 9.191052196837825e-06,
      "loss": 0.06679188460111618,
      "step": 4827
    },
    {
      "epoch": 0.6527135843986852,
      "grad_norm": 0.8140655159950256,
      "learning_rate": 9.184742449503135e-06,
      "loss": 0.03726750984787941,
      "step": 4828
    },
    {
      "epoch": 0.6528487777674505,
      "grad_norm": 1.070213794708252,
      "learning_rate": 9.178433912936077e-06,
      "loss": 0.06304620951414108,
      "step": 4829
    },
    {
      "epoch": 0.6529839711362158,
      "grad_norm": 1.819206714630127,
      "learning_rate": 9.172126588450125e-06,
      "loss": 0.059059225022792816,
      "step": 4830
    },
    {
      "epoch": 0.653119164504981,
      "grad_norm": 0.282976359128952,
      "learning_rate": 9.165820477358491e-06,
      "loss": 0.03074474073946476,
      "step": 4831
    },
    {
      "epoch": 0.6532543578737463,
      "grad_norm": 0.31543517112731934,
      "learning_rate": 9.159515580974154e-06,
      "loss": 0.04291544109582901,
      "step": 4832
    },
    {
      "epoch": 0.6533895512425115,
      "grad_norm": 0.6946098804473877,
      "learning_rate": 9.15321190060981e-06,
      "loss": 0.051056768745183945,
      "step": 4833
    },
    {
      "epoch": 0.6535247446112769,
      "grad_norm": 0.7878214120864868,
      "learning_rate": 9.14690943757793e-06,
      "loss": 0.048043809831142426,
      "step": 4834
    },
    {
      "epoch": 0.6536599379800421,
      "grad_norm": 0.943025529384613,
      "learning_rate": 9.14060819319072e-06,
      "loss": 0.057242460548877716,
      "step": 4835
    },
    {
      "epoch": 0.6537951313488073,
      "grad_norm": 0.8952677249908447,
      "learning_rate": 9.134308168760127e-06,
      "loss": 0.06574608385562897,
      "step": 4836
    },
    {
      "epoch": 0.6539303247175726,
      "grad_norm": 0.23783697187900543,
      "learning_rate": 9.128009365597854e-06,
      "loss": 0.05044171214103699,
      "step": 4837
    },
    {
      "epoch": 0.6540655180863378,
      "grad_norm": 0.47051429748535156,
      "learning_rate": 9.121711785015342e-06,
      "loss": 0.04344892501831055,
      "step": 4838
    },
    {
      "epoch": 0.6542007114551032,
      "grad_norm": 0.4336937367916107,
      "learning_rate": 9.115415428323787e-06,
      "loss": 0.042462307959795,
      "step": 4839
    },
    {
      "epoch": 0.6543359048238684,
      "grad_norm": 0.4336228668689728,
      "learning_rate": 9.109120296834118e-06,
      "loss": 0.048836223781108856,
      "step": 4840
    },
    {
      "epoch": 0.6544710981926336,
      "grad_norm": 0.520319402217865,
      "learning_rate": 9.10282639185702e-06,
      "loss": 0.05925591289997101,
      "step": 4841
    },
    {
      "epoch": 0.6546062915613989,
      "grad_norm": 0.39654210209846497,
      "learning_rate": 9.096533714702913e-06,
      "loss": 0.055843666195869446,
      "step": 4842
    },
    {
      "epoch": 0.6547414849301642,
      "grad_norm": 0.39380353689193726,
      "learning_rate": 9.090242266681967e-06,
      "loss": 0.061240747570991516,
      "step": 4843
    },
    {
      "epoch": 0.6548766782989295,
      "grad_norm": 0.4190712571144104,
      "learning_rate": 9.083952049104094e-06,
      "loss": 0.050237759947776794,
      "step": 4844
    },
    {
      "epoch": 0.6550118716676947,
      "grad_norm": 0.3751411437988281,
      "learning_rate": 9.07766306327896e-06,
      "loss": 0.04893229529261589,
      "step": 4845
    },
    {
      "epoch": 0.6551470650364599,
      "grad_norm": 0.33267223834991455,
      "learning_rate": 9.071375310515949e-06,
      "loss": 0.06221573054790497,
      "step": 4846
    },
    {
      "epoch": 0.6552822584052252,
      "grad_norm": 0.38737332820892334,
      "learning_rate": 9.065088792124219e-06,
      "loss": 0.04428090900182724,
      "step": 4847
    },
    {
      "epoch": 0.6554174517739905,
      "grad_norm": 0.45950019359588623,
      "learning_rate": 9.058803509412647e-06,
      "loss": 0.05439738184213638,
      "step": 4848
    },
    {
      "epoch": 0.6555526451427558,
      "grad_norm": 0.2903836965560913,
      "learning_rate": 9.05251946368987e-06,
      "loss": 0.04227876663208008,
      "step": 4849
    },
    {
      "epoch": 0.655687838511521,
      "grad_norm": 0.27683648467063904,
      "learning_rate": 9.046236656264258e-06,
      "loss": 0.03420381247997284,
      "step": 4850
    },
    {
      "epoch": 0.6558230318802862,
      "grad_norm": 0.32851892709732056,
      "learning_rate": 9.03995508844392e-06,
      "loss": 0.054434239864349365,
      "step": 4851
    },
    {
      "epoch": 0.6559582252490516,
      "grad_norm": 0.8404786586761475,
      "learning_rate": 9.033674761536718e-06,
      "loss": 0.056378528475761414,
      "step": 4852
    },
    {
      "epoch": 0.6560934186178168,
      "grad_norm": 0.5099725127220154,
      "learning_rate": 9.027395676850244e-06,
      "loss": 0.04750455170869827,
      "step": 4853
    },
    {
      "epoch": 0.6562286119865821,
      "grad_norm": 0.3070297837257385,
      "learning_rate": 9.02111783569184e-06,
      "loss": 0.0587763711810112,
      "step": 4854
    },
    {
      "epoch": 0.6563638053553473,
      "grad_norm": 0.4024368226528168,
      "learning_rate": 9.014841239368591e-06,
      "loss": 0.058742910623550415,
      "step": 4855
    },
    {
      "epoch": 0.6564989987241125,
      "grad_norm": 0.4439047873020172,
      "learning_rate": 9.008565889187308e-06,
      "loss": 0.04431654512882233,
      "step": 4856
    },
    {
      "epoch": 0.6566341920928779,
      "grad_norm": 0.5737723112106323,
      "learning_rate": 9.00229178645456e-06,
      "loss": 0.054135292768478394,
      "step": 4857
    },
    {
      "epoch": 0.6567693854616431,
      "grad_norm": 0.8024532198905945,
      "learning_rate": 8.996018932476641e-06,
      "loss": 0.04513950273394585,
      "step": 4858
    },
    {
      "epoch": 0.6569045788304083,
      "grad_norm": 0.5955286026000977,
      "learning_rate": 8.989747328559606e-06,
      "loss": 0.0479813888669014,
      "step": 4859
    },
    {
      "epoch": 0.6570397721991736,
      "grad_norm": 0.5546364188194275,
      "learning_rate": 8.98347697600922e-06,
      "loss": 0.050392478704452515,
      "step": 4860
    },
    {
      "epoch": 0.6571749655679389,
      "grad_norm": 0.5462738871574402,
      "learning_rate": 8.977207876131013e-06,
      "loss": 0.05263403058052063,
      "step": 4861
    },
    {
      "epoch": 0.6573101589367042,
      "grad_norm": 0.2945302128791809,
      "learning_rate": 8.970940030230245e-06,
      "loss": 0.047712430357933044,
      "step": 4862
    },
    {
      "epoch": 0.6574453523054694,
      "grad_norm": 0.4589152932167053,
      "learning_rate": 8.96467343961191e-06,
      "loss": 0.04600256681442261,
      "step": 4863
    },
    {
      "epoch": 0.6575805456742346,
      "grad_norm": 0.54264235496521,
      "learning_rate": 8.958408105580759e-06,
      "loss": 0.04188507795333862,
      "step": 4864
    },
    {
      "epoch": 0.657715739043,
      "grad_norm": 0.5305918455123901,
      "learning_rate": 8.952144029441248e-06,
      "loss": 0.03736427426338196,
      "step": 4865
    },
    {
      "epoch": 0.6578509324117652,
      "grad_norm": 0.4474351704120636,
      "learning_rate": 8.945881212497603e-06,
      "loss": 0.04515926539897919,
      "step": 4866
    },
    {
      "epoch": 0.6579861257805305,
      "grad_norm": 0.7046924233436584,
      "learning_rate": 8.939619656053777e-06,
      "loss": 0.05059683322906494,
      "step": 4867
    },
    {
      "epoch": 0.6581213191492957,
      "grad_norm": 0.546092689037323,
      "learning_rate": 8.933359361413456e-06,
      "loss": 0.046880945563316345,
      "step": 4868
    },
    {
      "epoch": 0.6582565125180609,
      "grad_norm": 0.5458420515060425,
      "learning_rate": 8.92710032988007e-06,
      "loss": 0.04236246645450592,
      "step": 4869
    },
    {
      "epoch": 0.6583917058868263,
      "grad_norm": 0.6035169363021851,
      "learning_rate": 8.920842562756773e-06,
      "loss": 0.05331994593143463,
      "step": 4870
    },
    {
      "epoch": 0.6585268992555915,
      "grad_norm": 0.38238996267318726,
      "learning_rate": 8.914586061346474e-06,
      "loss": 0.04808778315782547,
      "step": 4871
    },
    {
      "epoch": 0.6586620926243568,
      "grad_norm": 0.5059957504272461,
      "learning_rate": 8.908330826951811e-06,
      "loss": 0.05822131037712097,
      "step": 4872
    },
    {
      "epoch": 0.658797285993122,
      "grad_norm": 0.5722382068634033,
      "learning_rate": 8.902076860875155e-06,
      "loss": 0.052761077880859375,
      "step": 4873
    },
    {
      "epoch": 0.6589324793618873,
      "grad_norm": 0.9064381122589111,
      "learning_rate": 8.895824164418615e-06,
      "loss": 0.053103089332580566,
      "step": 4874
    },
    {
      "epoch": 0.6590676727306526,
      "grad_norm": 0.3827478289604187,
      "learning_rate": 8.889572738884033e-06,
      "loss": 0.04915151745080948,
      "step": 4875
    },
    {
      "epoch": 0.6592028660994178,
      "grad_norm": 1.523366093635559,
      "learning_rate": 8.88332258557299e-06,
      "loss": 0.06489886343479156,
      "step": 4876
    },
    {
      "epoch": 0.6593380594681831,
      "grad_norm": 0.8395458459854126,
      "learning_rate": 8.877073705786806e-06,
      "loss": 0.05081567168235779,
      "step": 4877
    },
    {
      "epoch": 0.6594732528369484,
      "grad_norm": 0.49804267287254333,
      "learning_rate": 8.870826100826527e-06,
      "loss": 0.04341159015893936,
      "step": 4878
    },
    {
      "epoch": 0.6596084462057136,
      "grad_norm": 0.46628883481025696,
      "learning_rate": 8.86457977199294e-06,
      "loss": 0.052684515714645386,
      "step": 4879
    },
    {
      "epoch": 0.6597436395744789,
      "grad_norm": 1.0826929807662964,
      "learning_rate": 8.85833472058656e-06,
      "loss": 0.047316402196884155,
      "step": 4880
    },
    {
      "epoch": 0.6598788329432441,
      "grad_norm": 1.2853542566299438,
      "learning_rate": 8.852090947907643e-06,
      "loss": 0.06989351660013199,
      "step": 4881
    },
    {
      "epoch": 0.6600140263120094,
      "grad_norm": 0.2032260298728943,
      "learning_rate": 8.84584845525618e-06,
      "loss": 0.0314270555973053,
      "step": 4882
    },
    {
      "epoch": 0.6601492196807747,
      "grad_norm": 0.3008894622325897,
      "learning_rate": 8.83960724393188e-06,
      "loss": 0.03849566727876663,
      "step": 4883
    },
    {
      "epoch": 0.6602844130495399,
      "grad_norm": 0.7275771498680115,
      "learning_rate": 8.833367315234206e-06,
      "loss": 0.05365827679634094,
      "step": 4884
    },
    {
      "epoch": 0.6604196064183052,
      "grad_norm": 0.30494412779808044,
      "learning_rate": 8.82712867046234e-06,
      "loss": 0.03763725608587265,
      "step": 4885
    },
    {
      "epoch": 0.6605547997870704,
      "grad_norm": 0.2807183861732483,
      "learning_rate": 8.820891310915203e-06,
      "loss": 0.04110005870461464,
      "step": 4886
    },
    {
      "epoch": 0.6606899931558357,
      "grad_norm": 0.40106067061424255,
      "learning_rate": 8.81465523789145e-06,
      "loss": 0.05846823751926422,
      "step": 4887
    },
    {
      "epoch": 0.660825186524601,
      "grad_norm": 0.8326048254966736,
      "learning_rate": 8.808420452689455e-06,
      "loss": 0.047958508133888245,
      "step": 4888
    },
    {
      "epoch": 0.6609603798933662,
      "grad_norm": 0.7759664058685303,
      "learning_rate": 8.802186956607344e-06,
      "loss": 0.04456786811351776,
      "step": 4889
    },
    {
      "epoch": 0.6610955732621315,
      "grad_norm": 0.5056163668632507,
      "learning_rate": 8.795954750942954e-06,
      "loss": 0.04759174585342407,
      "step": 4890
    },
    {
      "epoch": 0.6612307666308967,
      "grad_norm": 0.40979140996932983,
      "learning_rate": 8.789723836993878e-06,
      "loss": 0.05130434036254883,
      "step": 4891
    },
    {
      "epoch": 0.661365959999662,
      "grad_norm": 1.1193937063217163,
      "learning_rate": 8.783494216057407e-06,
      "loss": 0.06788745522499084,
      "step": 4892
    },
    {
      "epoch": 0.6615011533684273,
      "grad_norm": 0.6738024950027466,
      "learning_rate": 8.777265889430593e-06,
      "loss": 0.03771587088704109,
      "step": 4893
    },
    {
      "epoch": 0.6616363467371925,
      "grad_norm": 0.4568180441856384,
      "learning_rate": 8.771038858410206e-06,
      "loss": 0.06290781497955322,
      "step": 4894
    },
    {
      "epoch": 0.6617715401059578,
      "grad_norm": 0.8085651397705078,
      "learning_rate": 8.764813124292744e-06,
      "loss": 0.0503215417265892,
      "step": 4895
    },
    {
      "epoch": 0.6619067334747231,
      "grad_norm": 0.728347897529602,
      "learning_rate": 8.758588688374445e-06,
      "loss": 0.06116729974746704,
      "step": 4896
    },
    {
      "epoch": 0.6620419268434883,
      "grad_norm": 0.9897721409797668,
      "learning_rate": 8.752365551951262e-06,
      "loss": 0.060762468725442886,
      "step": 4897
    },
    {
      "epoch": 0.6621771202122536,
      "grad_norm": 0.7816612124443054,
      "learning_rate": 8.74614371631888e-06,
      "loss": 0.04762818291783333,
      "step": 4898
    },
    {
      "epoch": 0.6623123135810188,
      "grad_norm": 1.0026624202728271,
      "learning_rate": 8.739923182772732e-06,
      "loss": 0.0729488730430603,
      "step": 4899
    },
    {
      "epoch": 0.6624475069497842,
      "grad_norm": 0.7975327968597412,
      "learning_rate": 8.733703952607956e-06,
      "loss": 0.055999740958213806,
      "step": 4900
    },
    {
      "epoch": 0.6625827003185494,
      "grad_norm": 0.3464163839817047,
      "learning_rate": 8.727486027119443e-06,
      "loss": 0.05318213999271393,
      "step": 4901
    },
    {
      "epoch": 0.6627178936873146,
      "grad_norm": 0.4403265118598938,
      "learning_rate": 8.721269407601783e-06,
      "loss": 0.05413217842578888,
      "step": 4902
    },
    {
      "epoch": 0.6628530870560799,
      "grad_norm": 0.262787401676178,
      "learning_rate": 8.71505409534931e-06,
      "loss": 0.03239549696445465,
      "step": 4903
    },
    {
      "epoch": 0.6629882804248451,
      "grad_norm": 1.250382900238037,
      "learning_rate": 8.708840091656093e-06,
      "loss": 0.07249332964420319,
      "step": 4904
    },
    {
      "epoch": 0.6631234737936105,
      "grad_norm": 0.2127716988325119,
      "learning_rate": 8.70262739781592e-06,
      "loss": 0.03368908166885376,
      "step": 4905
    },
    {
      "epoch": 0.6632586671623757,
      "grad_norm": 0.47799018025398254,
      "learning_rate": 8.696416015122302e-06,
      "loss": 0.04219493269920349,
      "step": 4906
    },
    {
      "epoch": 0.6633938605311409,
      "grad_norm": 0.47827261686325073,
      "learning_rate": 8.690205944868487e-06,
      "loss": 0.05766450986266136,
      "step": 4907
    },
    {
      "epoch": 0.6635290538999062,
      "grad_norm": 0.41965779662132263,
      "learning_rate": 8.683997188347436e-06,
      "loss": 0.06484150886535645,
      "step": 4908
    },
    {
      "epoch": 0.6636642472686715,
      "grad_norm": 0.7600243091583252,
      "learning_rate": 8.677789746851855e-06,
      "loss": 0.05288514122366905,
      "step": 4909
    },
    {
      "epoch": 0.6637994406374368,
      "grad_norm": 0.5731505155563354,
      "learning_rate": 8.671583621674167e-06,
      "loss": 0.056962043046951294,
      "step": 4910
    },
    {
      "epoch": 0.663934634006202,
      "grad_norm": 0.8263890743255615,
      "learning_rate": 8.665378814106512e-06,
      "loss": 0.04230496287345886,
      "step": 4911
    },
    {
      "epoch": 0.6640698273749672,
      "grad_norm": 0.5004463791847229,
      "learning_rate": 8.65917532544077e-06,
      "loss": 0.06371943652629852,
      "step": 4912
    },
    {
      "epoch": 0.6642050207437326,
      "grad_norm": 0.62314373254776,
      "learning_rate": 8.652973156968532e-06,
      "loss": 0.046881675720214844,
      "step": 4913
    },
    {
      "epoch": 0.6643402141124978,
      "grad_norm": 0.8253375887870789,
      "learning_rate": 8.646772309981141e-06,
      "loss": 0.04562593251466751,
      "step": 4914
    },
    {
      "epoch": 0.664475407481263,
      "grad_norm": 0.6556355357170105,
      "learning_rate": 8.640572785769624e-06,
      "loss": 0.0736011266708374,
      "step": 4915
    },
    {
      "epoch": 0.6646106008500283,
      "grad_norm": 0.669047474861145,
      "learning_rate": 8.63437458562477e-06,
      "loss": 0.05263333022594452,
      "step": 4916
    },
    {
      "epoch": 0.6647457942187935,
      "grad_norm": 0.22817447781562805,
      "learning_rate": 8.628177710837068e-06,
      "loss": 0.04105766862630844,
      "step": 4917
    },
    {
      "epoch": 0.6648809875875589,
      "grad_norm": 0.5972623229026794,
      "learning_rate": 8.621982162696752e-06,
      "loss": 0.05957232043147087,
      "step": 4918
    },
    {
      "epoch": 0.6650161809563241,
      "grad_norm": 0.6017746925354004,
      "learning_rate": 8.615787942493766e-06,
      "loss": 0.041902463883161545,
      "step": 4919
    },
    {
      "epoch": 0.6651513743250893,
      "grad_norm": 0.41817542910575867,
      "learning_rate": 8.609595051517765e-06,
      "loss": 0.040782541036605835,
      "step": 4920
    },
    {
      "epoch": 0.6652865676938546,
      "grad_norm": 0.3211895823478699,
      "learning_rate": 8.603403491058157e-06,
      "loss": 0.038159385323524475,
      "step": 4921
    },
    {
      "epoch": 0.6654217610626199,
      "grad_norm": 1.154258131980896,
      "learning_rate": 8.597213262404046e-06,
      "loss": 0.056552689522504807,
      "step": 4922
    },
    {
      "epoch": 0.6655569544313852,
      "grad_norm": 0.20456521213054657,
      "learning_rate": 8.591024366844291e-06,
      "loss": 0.0435362309217453,
      "step": 4923
    },
    {
      "epoch": 0.6656921478001504,
      "grad_norm": 0.8878408670425415,
      "learning_rate": 8.584836805667434e-06,
      "loss": 0.044599395245313644,
      "step": 4924
    },
    {
      "epoch": 0.6658273411689156,
      "grad_norm": 0.32103127241134644,
      "learning_rate": 8.578650580161754e-06,
      "loss": 0.0503804013133049,
      "step": 4925
    },
    {
      "epoch": 0.665962534537681,
      "grad_norm": 0.38880297541618347,
      "learning_rate": 8.572465691615275e-06,
      "loss": 0.05300453305244446,
      "step": 4926
    },
    {
      "epoch": 0.6660977279064462,
      "grad_norm": 0.4174281358718872,
      "learning_rate": 8.56628214131571e-06,
      "loss": 0.06797348707914352,
      "step": 4927
    },
    {
      "epoch": 0.6662329212752115,
      "grad_norm": 0.7094303369522095,
      "learning_rate": 8.560099930550523e-06,
      "loss": 0.05302663892507553,
      "step": 4928
    },
    {
      "epoch": 0.6663681146439767,
      "grad_norm": 0.7040273547172546,
      "learning_rate": 8.553919060606866e-06,
      "loss": 0.07105447351932526,
      "step": 4929
    },
    {
      "epoch": 0.6665033080127419,
      "grad_norm": 0.3127346634864807,
      "learning_rate": 8.54773953277163e-06,
      "loss": 0.045730799436569214,
      "step": 4930
    },
    {
      "epoch": 0.6666385013815073,
      "grad_norm": 0.39840197563171387,
      "learning_rate": 8.541561348331433e-06,
      "loss": 0.04773852229118347,
      "step": 4931
    },
    {
      "epoch": 0.6667736947502725,
      "grad_norm": 0.5970373749732971,
      "learning_rate": 8.535384508572603e-06,
      "loss": 0.06129232794046402,
      "step": 4932
    },
    {
      "epoch": 0.6669088881190378,
      "grad_norm": 0.38890382647514343,
      "learning_rate": 8.529209014781202e-06,
      "loss": 0.062496498227119446,
      "step": 4933
    },
    {
      "epoch": 0.667044081487803,
      "grad_norm": 0.2800302803516388,
      "learning_rate": 8.523034868242984e-06,
      "loss": 0.04507967829704285,
      "step": 4934
    },
    {
      "epoch": 0.6671792748565682,
      "grad_norm": 0.6557746529579163,
      "learning_rate": 8.51686207024344e-06,
      "loss": 0.06638331711292267,
      "step": 4935
    },
    {
      "epoch": 0.6673144682253336,
      "grad_norm": 0.34183457493782043,
      "learning_rate": 8.510690622067792e-06,
      "loss": 0.053346022963523865,
      "step": 4936
    },
    {
      "epoch": 0.6674496615940988,
      "grad_norm": 0.2507877051830292,
      "learning_rate": 8.50452052500096e-06,
      "loss": 0.04421476274728775,
      "step": 4937
    },
    {
      "epoch": 0.667584854962864,
      "grad_norm": 0.7738170027732849,
      "learning_rate": 8.498351780327594e-06,
      "loss": 0.05261988192796707,
      "step": 4938
    },
    {
      "epoch": 0.6677200483316293,
      "grad_norm": 0.4712650179862976,
      "learning_rate": 8.492184389332061e-06,
      "loss": 0.05027683824300766,
      "step": 4939
    },
    {
      "epoch": 0.6678552417003946,
      "grad_norm": 0.3059353530406952,
      "learning_rate": 8.486018353298432e-06,
      "loss": 0.06472623348236084,
      "step": 4940
    },
    {
      "epoch": 0.6679904350691599,
      "grad_norm": 0.711154043674469,
      "learning_rate": 8.479853673510528e-06,
      "loss": 0.049798935651779175,
      "step": 4941
    },
    {
      "epoch": 0.6681256284379251,
      "grad_norm": 0.7332208156585693,
      "learning_rate": 8.473690351251855e-06,
      "loss": 0.05915476754307747,
      "step": 4942
    },
    {
      "epoch": 0.6682608218066903,
      "grad_norm": 0.3910542130470276,
      "learning_rate": 8.467528387805656e-06,
      "loss": 0.04300757497549057,
      "step": 4943
    },
    {
      "epoch": 0.6683960151754557,
      "grad_norm": 0.3801325261592865,
      "learning_rate": 8.461367784454881e-06,
      "loss": 0.039240531623363495,
      "step": 4944
    },
    {
      "epoch": 0.6685312085442209,
      "grad_norm": 0.23581597208976746,
      "learning_rate": 8.455208542482195e-06,
      "loss": 0.04541178047657013,
      "step": 4945
    },
    {
      "epoch": 0.6686664019129862,
      "grad_norm": 0.412605345249176,
      "learning_rate": 8.449050663170004e-06,
      "loss": 0.0451117679476738,
      "step": 4946
    },
    {
      "epoch": 0.6688015952817514,
      "grad_norm": 0.4841384291648865,
      "learning_rate": 8.442894147800387e-06,
      "loss": 0.04456118866801262,
      "step": 4947
    },
    {
      "epoch": 0.6689367886505166,
      "grad_norm": 0.4538283050060272,
      "learning_rate": 8.436738997655184e-06,
      "loss": 0.04737669229507446,
      "step": 4948
    },
    {
      "epoch": 0.669071982019282,
      "grad_norm": 0.4836001694202423,
      "learning_rate": 8.430585214015918e-06,
      "loss": 0.05537992715835571,
      "step": 4949
    },
    {
      "epoch": 0.6692071753880472,
      "grad_norm": 0.385507196187973,
      "learning_rate": 8.424432798163838e-06,
      "loss": 0.04484360292553902,
      "step": 4950
    },
    {
      "epoch": 0.6693423687568125,
      "grad_norm": 1.2376917600631714,
      "learning_rate": 8.418281751379926e-06,
      "loss": 0.06745009124279022,
      "step": 4951
    },
    {
      "epoch": 0.6694775621255777,
      "grad_norm": 0.6492207646369934,
      "learning_rate": 8.41213207494484e-06,
      "loss": 0.06594588607549667,
      "step": 4952
    },
    {
      "epoch": 0.669612755494343,
      "grad_norm": 0.5820128321647644,
      "learning_rate": 8.405983770138992e-06,
      "loss": 0.0351836234331131,
      "step": 4953
    },
    {
      "epoch": 0.6697479488631083,
      "grad_norm": 0.5219746828079224,
      "learning_rate": 8.399836838242479e-06,
      "loss": 0.05730397254228592,
      "step": 4954
    },
    {
      "epoch": 0.6698831422318735,
      "grad_norm": 0.296224445104599,
      "learning_rate": 8.393691280535143e-06,
      "loss": 0.05526592954993248,
      "step": 4955
    },
    {
      "epoch": 0.6700183356006388,
      "grad_norm": 0.6880562901496887,
      "learning_rate": 8.387547098296516e-06,
      "loss": 0.05071261152625084,
      "step": 4956
    },
    {
      "epoch": 0.670153528969404,
      "grad_norm": 0.6879351735115051,
      "learning_rate": 8.38140429280583e-06,
      "loss": 0.03831213712692261,
      "step": 4957
    },
    {
      "epoch": 0.6702887223381693,
      "grad_norm": 0.42694202065467834,
      "learning_rate": 8.375262865342073e-06,
      "loss": 0.049122318625450134,
      "step": 4958
    },
    {
      "epoch": 0.6704239157069346,
      "grad_norm": 0.5570389032363892,
      "learning_rate": 8.36912281718391e-06,
      "loss": 0.053875505924224854,
      "step": 4959
    },
    {
      "epoch": 0.6705591090756998,
      "grad_norm": 0.7798247337341309,
      "learning_rate": 8.362984149609748e-06,
      "loss": 0.049223948270082474,
      "step": 4960
    },
    {
      "epoch": 0.6706943024444652,
      "grad_norm": 0.47196725010871887,
      "learning_rate": 8.356846863897672e-06,
      "loss": 0.06143927201628685,
      "step": 4961
    },
    {
      "epoch": 0.6708294958132304,
      "grad_norm": 0.35774117708206177,
      "learning_rate": 8.350710961325498e-06,
      "loss": 0.06524142622947693,
      "step": 4962
    },
    {
      "epoch": 0.6709646891819956,
      "grad_norm": 0.34857645630836487,
      "learning_rate": 8.344576443170768e-06,
      "loss": 0.05976579710841179,
      "step": 4963
    },
    {
      "epoch": 0.6710998825507609,
      "grad_norm": 0.4516732096672058,
      "learning_rate": 8.338443310710708e-06,
      "loss": 0.04969620704650879,
      "step": 4964
    },
    {
      "epoch": 0.6712350759195261,
      "grad_norm": 1.2974827289581299,
      "learning_rate": 8.332311565222284e-06,
      "loss": 0.05954143404960632,
      "step": 4965
    },
    {
      "epoch": 0.6713702692882914,
      "grad_norm": 0.4259440004825592,
      "learning_rate": 8.326181207982145e-06,
      "loss": 0.04360128194093704,
      "step": 4966
    },
    {
      "epoch": 0.6715054626570567,
      "grad_norm": 0.43535634875297546,
      "learning_rate": 8.32005224026666e-06,
      "loss": 0.06057393550872803,
      "step": 4967
    },
    {
      "epoch": 0.6716406560258219,
      "grad_norm": 0.6019091606140137,
      "learning_rate": 8.313924663351927e-06,
      "loss": 0.042790770530700684,
      "step": 4968
    },
    {
      "epoch": 0.6717758493945872,
      "grad_norm": 0.6441512107849121,
      "learning_rate": 8.307798478513733e-06,
      "loss": 0.06718750298023224,
      "step": 4969
    },
    {
      "epoch": 0.6719110427633525,
      "grad_norm": 0.5350854396820068,
      "learning_rate": 8.301673687027583e-06,
      "loss": 0.05609264224767685,
      "step": 4970
    },
    {
      "epoch": 0.6720462361321177,
      "grad_norm": 1.0968847274780273,
      "learning_rate": 8.295550290168692e-06,
      "loss": 0.054544731974601746,
      "step": 4971
    },
    {
      "epoch": 0.672181429500883,
      "grad_norm": 0.7787512540817261,
      "learning_rate": 8.289428289211977e-06,
      "loss": 0.060831211507320404,
      "step": 4972
    },
    {
      "epoch": 0.6723166228696482,
      "grad_norm": 0.6954808235168457,
      "learning_rate": 8.283307685432083e-06,
      "loss": 0.05581638216972351,
      "step": 4973
    },
    {
      "epoch": 0.6724518162384135,
      "grad_norm": 1.1396392583847046,
      "learning_rate": 8.277188480103348e-06,
      "loss": 0.08411954343318939,
      "step": 4974
    },
    {
      "epoch": 0.6725870096071788,
      "grad_norm": 0.24444527924060822,
      "learning_rate": 8.271070674499821e-06,
      "loss": 0.04333586245775223,
      "step": 4975
    },
    {
      "epoch": 0.672722202975944,
      "grad_norm": 0.30251526832580566,
      "learning_rate": 8.264954269895262e-06,
      "loss": 0.06129070743918419,
      "step": 4976
    },
    {
      "epoch": 0.6728573963447093,
      "grad_norm": 0.4709165692329407,
      "learning_rate": 8.258839267563134e-06,
      "loss": 0.06391668319702148,
      "step": 4977
    },
    {
      "epoch": 0.6729925897134745,
      "grad_norm": 0.7118197679519653,
      "learning_rate": 8.252725668776623e-06,
      "loss": 0.05873258784413338,
      "step": 4978
    },
    {
      "epoch": 0.6731277830822399,
      "grad_norm": 0.9924367666244507,
      "learning_rate": 8.24661347480861e-06,
      "loss": 0.038125358521938324,
      "step": 4979
    },
    {
      "epoch": 0.6732629764510051,
      "grad_norm": 1.1432424783706665,
      "learning_rate": 8.240502686931686e-06,
      "loss": 0.05306572467088699,
      "step": 4980
    },
    {
      "epoch": 0.6733981698197703,
      "grad_norm": 0.39897817373275757,
      "learning_rate": 8.234393306418148e-06,
      "loss": 0.035227708518505096,
      "step": 4981
    },
    {
      "epoch": 0.6735333631885356,
      "grad_norm": 0.5576872229576111,
      "learning_rate": 8.228285334539995e-06,
      "loss": 0.05256688594818115,
      "step": 4982
    },
    {
      "epoch": 0.6736685565573008,
      "grad_norm": 0.5145950317382812,
      "learning_rate": 8.22217877256896e-06,
      "loss": 0.0601494237780571,
      "step": 4983
    },
    {
      "epoch": 0.6738037499260662,
      "grad_norm": 1.218837857246399,
      "learning_rate": 8.216073621776436e-06,
      "loss": 0.07031473517417908,
      "step": 4984
    },
    {
      "epoch": 0.6739389432948314,
      "grad_norm": 0.5454159379005432,
      "learning_rate": 8.209969883433566e-06,
      "loss": 0.054414957761764526,
      "step": 4985
    },
    {
      "epoch": 0.6740741366635966,
      "grad_norm": 0.38681066036224365,
      "learning_rate": 8.203867558811177e-06,
      "loss": 0.043965667486190796,
      "step": 4986
    },
    {
      "epoch": 0.6742093300323619,
      "grad_norm": 0.9242783188819885,
      "learning_rate": 8.197766649179795e-06,
      "loss": 0.06138654798269272,
      "step": 4987
    },
    {
      "epoch": 0.6743445234011272,
      "grad_norm": 0.7326571941375732,
      "learning_rate": 8.191667155809684e-06,
      "loss": 0.05227571725845337,
      "step": 4988
    },
    {
      "epoch": 0.6744797167698925,
      "grad_norm": 0.45089221000671387,
      "learning_rate": 8.185569079970764e-06,
      "loss": 0.07390367984771729,
      "step": 4989
    },
    {
      "epoch": 0.6746149101386577,
      "grad_norm": 0.29904136061668396,
      "learning_rate": 8.179472422932709e-06,
      "loss": 0.038001649081707,
      "step": 4990
    },
    {
      "epoch": 0.6747501035074229,
      "grad_norm": 1.613762378692627,
      "learning_rate": 8.17337718596486e-06,
      "loss": 0.06401367485523224,
      "step": 4991
    },
    {
      "epoch": 0.6748852968761883,
      "grad_norm": 0.8714796900749207,
      "learning_rate": 8.167283370336295e-06,
      "loss": 0.06580977141857147,
      "step": 4992
    },
    {
      "epoch": 0.6750204902449535,
      "grad_norm": 0.274777352809906,
      "learning_rate": 8.161190977315766e-06,
      "loss": 0.03572496771812439,
      "step": 4993
    },
    {
      "epoch": 0.6751556836137187,
      "grad_norm": 0.24242113530635834,
      "learning_rate": 8.155100008171736e-06,
      "loss": 0.03542621433734894,
      "step": 4994
    },
    {
      "epoch": 0.675290876982484,
      "grad_norm": 0.3669450283050537,
      "learning_rate": 8.149010464172392e-06,
      "loss": 0.05259627848863602,
      "step": 4995
    },
    {
      "epoch": 0.6754260703512492,
      "grad_norm": 0.547822892665863,
      "learning_rate": 8.142922346585597e-06,
      "loss": 0.04516976699233055,
      "step": 4996
    },
    {
      "epoch": 0.6755612637200146,
      "grad_norm": 0.8237200379371643,
      "learning_rate": 8.13683565667895e-06,
      "loss": 0.061566174030303955,
      "step": 4997
    },
    {
      "epoch": 0.6756964570887798,
      "grad_norm": 0.19074095785617828,
      "learning_rate": 8.13075039571971e-06,
      "loss": 0.035949528217315674,
      "step": 4998
    },
    {
      "epoch": 0.675831650457545,
      "grad_norm": 0.5671090483665466,
      "learning_rate": 8.124666564974864e-06,
      "loss": 0.07176803052425385,
      "step": 4999
    },
    {
      "epoch": 0.6759668438263103,
      "grad_norm": 0.3654058873653412,
      "learning_rate": 8.11858416571111e-06,
      "loss": 0.05109212547540665,
      "step": 5000
    },
    {
      "epoch": 0.6761020371950756,
      "grad_norm": 0.36814555525779724,
      "learning_rate": 8.112503199194821e-06,
      "loss": 0.05827689170837402,
      "step": 5001
    },
    {
      "epoch": 0.6762372305638409,
      "grad_norm": 0.6010149121284485,
      "learning_rate": 8.106423666692108e-06,
      "loss": 0.04330838471651077,
      "step": 5002
    },
    {
      "epoch": 0.6763724239326061,
      "grad_norm": 0.8253535628318787,
      "learning_rate": 8.100345569468742e-06,
      "loss": 0.04694852977991104,
      "step": 5003
    },
    {
      "epoch": 0.6765076173013713,
      "grad_norm": 0.6037037968635559,
      "learning_rate": 8.094268908790215e-06,
      "loss": 0.06027128919959068,
      "step": 5004
    },
    {
      "epoch": 0.6766428106701367,
      "grad_norm": 0.7514895796775818,
      "learning_rate": 8.088193685921733e-06,
      "loss": 0.04608899727463722,
      "step": 5005
    },
    {
      "epoch": 0.6767780040389019,
      "grad_norm": 0.519709587097168,
      "learning_rate": 8.082119902128185e-06,
      "loss": 0.05882525444030762,
      "step": 5006
    },
    {
      "epoch": 0.6769131974076672,
      "grad_norm": 0.3477354645729065,
      "learning_rate": 8.076047558674164e-06,
      "loss": 0.02975265122950077,
      "step": 5007
    },
    {
      "epoch": 0.6770483907764324,
      "grad_norm": 0.338996022939682,
      "learning_rate": 8.069976656823964e-06,
      "loss": 0.045697540044784546,
      "step": 5008
    },
    {
      "epoch": 0.6771835841451976,
      "grad_norm": 0.4676816463470459,
      "learning_rate": 8.063907197841574e-06,
      "loss": 0.03803064674139023,
      "step": 5009
    },
    {
      "epoch": 0.677318777513963,
      "grad_norm": 0.5746403932571411,
      "learning_rate": 8.057839182990698e-06,
      "loss": 0.057335883378982544,
      "step": 5010
    },
    {
      "epoch": 0.6774539708827282,
      "grad_norm": 0.3573558032512665,
      "learning_rate": 8.051772613534725e-06,
      "loss": 0.061646685004234314,
      "step": 5011
    },
    {
      "epoch": 0.6775891642514935,
      "grad_norm": 1.0433835983276367,
      "learning_rate": 8.045707490736745e-06,
      "loss": 0.07262985408306122,
      "step": 5012
    },
    {
      "epoch": 0.6777243576202587,
      "grad_norm": 0.5547288656234741,
      "learning_rate": 8.039643815859552e-06,
      "loss": 0.05821345001459122,
      "step": 5013
    },
    {
      "epoch": 0.677859550989024,
      "grad_norm": 0.5345027446746826,
      "learning_rate": 8.033581590165627e-06,
      "loss": 0.05726383253931999,
      "step": 5014
    },
    {
      "epoch": 0.6779947443577893,
      "grad_norm": 0.38071736693382263,
      "learning_rate": 8.027520814917175e-06,
      "loss": 0.06746025383472443,
      "step": 5015
    },
    {
      "epoch": 0.6781299377265545,
      "grad_norm": 0.3310386538505554,
      "learning_rate": 8.021461491376064e-06,
      "loss": 0.04701864719390869,
      "step": 5016
    },
    {
      "epoch": 0.6782651310953198,
      "grad_norm": 0.41735756397247314,
      "learning_rate": 8.015403620803885e-06,
      "loss": 0.056895673274993896,
      "step": 5017
    },
    {
      "epoch": 0.678400324464085,
      "grad_norm": 0.2808677852153778,
      "learning_rate": 8.009347204461922e-06,
      "loss": 0.040169231593608856,
      "step": 5018
    },
    {
      "epoch": 0.6785355178328503,
      "grad_norm": 0.31439903378486633,
      "learning_rate": 8.003292243611143e-06,
      "loss": 0.051807381212711334,
      "step": 5019
    },
    {
      "epoch": 0.6786707112016156,
      "grad_norm": 0.4138041138648987,
      "learning_rate": 7.99723873951224e-06,
      "loss": 0.06021660566329956,
      "step": 5020
    },
    {
      "epoch": 0.6788059045703808,
      "grad_norm": 0.6591784358024597,
      "learning_rate": 7.991186693425563e-06,
      "loss": 0.07591414451599121,
      "step": 5021
    },
    {
      "epoch": 0.678941097939146,
      "grad_norm": 0.2700737416744232,
      "learning_rate": 7.9851361066112e-06,
      "loss": 0.054768163710832596,
      "step": 5022
    },
    {
      "epoch": 0.6790762913079114,
      "grad_norm": 0.4354577958583832,
      "learning_rate": 7.979086980328907e-06,
      "loss": 0.04215976595878601,
      "step": 5023
    },
    {
      "epoch": 0.6792114846766766,
      "grad_norm": 0.3354848325252533,
      "learning_rate": 7.973039315838137e-06,
      "loss": 0.053591564297676086,
      "step": 5024
    },
    {
      "epoch": 0.6793466780454419,
      "grad_norm": 0.5004885792732239,
      "learning_rate": 7.966993114398067e-06,
      "loss": 0.05722540616989136,
      "step": 5025
    },
    {
      "epoch": 0.6794818714142071,
      "grad_norm": 0.43668463826179504,
      "learning_rate": 7.960948377267524e-06,
      "loss": 0.06277358531951904,
      "step": 5026
    },
    {
      "epoch": 0.6796170647829723,
      "grad_norm": 0.9970447421073914,
      "learning_rate": 7.954905105705071e-06,
      "loss": 0.05152648687362671,
      "step": 5027
    },
    {
      "epoch": 0.6797522581517377,
      "grad_norm": 0.7500037550926208,
      "learning_rate": 7.948863300968938e-06,
      "loss": 0.05576123297214508,
      "step": 5028
    },
    {
      "epoch": 0.6798874515205029,
      "grad_norm": 0.25165197253227234,
      "learning_rate": 7.942822964317078e-06,
      "loss": 0.04088078439235687,
      "step": 5029
    },
    {
      "epoch": 0.6800226448892682,
      "grad_norm": 0.4969177842140198,
      "learning_rate": 7.936784097007105e-06,
      "loss": 0.051823690533638,
      "step": 5030
    },
    {
      "epoch": 0.6801578382580334,
      "grad_norm": 0.7050125002861023,
      "learning_rate": 7.930746700296344e-06,
      "loss": 0.053785525262355804,
      "step": 5031
    },
    {
      "epoch": 0.6802930316267987,
      "grad_norm": 0.349983811378479,
      "learning_rate": 7.924710775441822e-06,
      "loss": 0.04884212464094162,
      "step": 5032
    },
    {
      "epoch": 0.680428224995564,
      "grad_norm": 0.26775214076042175,
      "learning_rate": 7.918676323700241e-06,
      "loss": 0.04708987474441528,
      "step": 5033
    },
    {
      "epoch": 0.6805634183643292,
      "grad_norm": 0.3257896900177002,
      "learning_rate": 7.912643346328023e-06,
      "loss": 0.043380655348300934,
      "step": 5034
    },
    {
      "epoch": 0.6806986117330945,
      "grad_norm": 0.6439222097396851,
      "learning_rate": 7.906611844581251e-06,
      "loss": 0.05791155621409416,
      "step": 5035
    },
    {
      "epoch": 0.6808338051018598,
      "grad_norm": 1.3756972551345825,
      "learning_rate": 7.900581819715713e-06,
      "loss": 0.07098507136106491,
      "step": 5036
    },
    {
      "epoch": 0.680968998470625,
      "grad_norm": 0.6111595630645752,
      "learning_rate": 7.894553272986901e-06,
      "loss": 0.034134574234485626,
      "step": 5037
    },
    {
      "epoch": 0.6811041918393903,
      "grad_norm": 0.36403319239616394,
      "learning_rate": 7.888526205649993e-06,
      "loss": 0.04534803330898285,
      "step": 5038
    },
    {
      "epoch": 0.6812393852081555,
      "grad_norm": 0.22279338538646698,
      "learning_rate": 7.882500618959849e-06,
      "loss": 0.05339765548706055,
      "step": 5039
    },
    {
      "epoch": 0.6813745785769209,
      "grad_norm": 0.9752552509307861,
      "learning_rate": 7.876476514171033e-06,
      "loss": 0.05286046862602234,
      "step": 5040
    },
    {
      "epoch": 0.6815097719456861,
      "grad_norm": 0.714496910572052,
      "learning_rate": 7.870453892537788e-06,
      "loss": 0.04931214451789856,
      "step": 5041
    },
    {
      "epoch": 0.6816449653144513,
      "grad_norm": 0.6221680045127869,
      "learning_rate": 7.864432755314068e-06,
      "loss": 0.08361977338790894,
      "step": 5042
    },
    {
      "epoch": 0.6817801586832166,
      "grad_norm": 0.4232570230960846,
      "learning_rate": 7.858413103753499e-06,
      "loss": 0.0306400153785944,
      "step": 5043
    },
    {
      "epoch": 0.6819153520519818,
      "grad_norm": 0.7650275230407715,
      "learning_rate": 7.852394939109408e-06,
      "loss": 0.06406787782907486,
      "step": 5044
    },
    {
      "epoch": 0.6820505454207472,
      "grad_norm": 0.3299720287322998,
      "learning_rate": 7.846378262634803e-06,
      "loss": 0.0422019436955452,
      "step": 5045
    },
    {
      "epoch": 0.6821857387895124,
      "grad_norm": 0.2432260811328888,
      "learning_rate": 7.840363075582385e-06,
      "loss": 0.0305987149477005,
      "step": 5046
    },
    {
      "epoch": 0.6823209321582776,
      "grad_norm": 0.507563054561615,
      "learning_rate": 7.834349379204565e-06,
      "loss": 0.06139723211526871,
      "step": 5047
    },
    {
      "epoch": 0.6824561255270429,
      "grad_norm": 0.68167644739151,
      "learning_rate": 7.828337174753411e-06,
      "loss": 0.06552176922559738,
      "step": 5048
    },
    {
      "epoch": 0.6825913188958082,
      "grad_norm": 0.34419867396354675,
      "learning_rate": 7.822326463480703e-06,
      "loss": 0.051542606204748154,
      "step": 5049
    },
    {
      "epoch": 0.6827265122645734,
      "grad_norm": 0.36507555842399597,
      "learning_rate": 7.816317246637901e-06,
      "loss": 0.04676346480846405,
      "step": 5050
    },
    {
      "epoch": 0.6828617056333387,
      "grad_norm": 0.5398001670837402,
      "learning_rate": 7.810309525476152e-06,
      "loss": 0.0544717013835907,
      "step": 5051
    },
    {
      "epoch": 0.6829968990021039,
      "grad_norm": 0.6113669872283936,
      "learning_rate": 7.804303301246311e-06,
      "loss": 0.062158867716789246,
      "step": 5052
    },
    {
      "epoch": 0.6831320923708692,
      "grad_norm": 0.2693955898284912,
      "learning_rate": 7.798298575198884e-06,
      "loss": 0.039754901081323624,
      "step": 5053
    },
    {
      "epoch": 0.6832672857396345,
      "grad_norm": 0.607220470905304,
      "learning_rate": 7.792295348584103e-06,
      "loss": 0.0434325709939003,
      "step": 5054
    },
    {
      "epoch": 0.6834024791083997,
      "grad_norm": 0.3906765282154083,
      "learning_rate": 7.786293622651866e-06,
      "loss": 0.056094035506248474,
      "step": 5055
    },
    {
      "epoch": 0.683537672477165,
      "grad_norm": 0.6982684135437012,
      "learning_rate": 7.78029339865176e-06,
      "loss": 0.06186196580529213,
      "step": 5056
    },
    {
      "epoch": 0.6836728658459302,
      "grad_norm": 0.6677371859550476,
      "learning_rate": 7.774294677833078e-06,
      "loss": 0.060383137315511703,
      "step": 5057
    },
    {
      "epoch": 0.6838080592146956,
      "grad_norm": 0.6132252216339111,
      "learning_rate": 7.768297461444766e-06,
      "loss": 0.054885417222976685,
      "step": 5058
    },
    {
      "epoch": 0.6839432525834608,
      "grad_norm": 0.5411122441291809,
      "learning_rate": 7.762301750735494e-06,
      "loss": 0.06308025121688843,
      "step": 5059
    },
    {
      "epoch": 0.684078445952226,
      "grad_norm": 0.6690797805786133,
      "learning_rate": 7.756307546953592e-06,
      "loss": 0.04824995994567871,
      "step": 5060
    },
    {
      "epoch": 0.6842136393209913,
      "grad_norm": 0.46731314063072205,
      "learning_rate": 7.750314851347087e-06,
      "loss": 0.04488636180758476,
      "step": 5061
    },
    {
      "epoch": 0.6843488326897565,
      "grad_norm": 0.3768135905265808,
      "learning_rate": 7.74432366516369e-06,
      "loss": 0.05284181982278824,
      "step": 5062
    },
    {
      "epoch": 0.6844840260585219,
      "grad_norm": 0.9547297358512878,
      "learning_rate": 7.738333989650794e-06,
      "loss": 0.0503455251455307,
      "step": 5063
    },
    {
      "epoch": 0.6846192194272871,
      "grad_norm": 0.45269399881362915,
      "learning_rate": 7.732345826055487e-06,
      "loss": 0.06963029503822327,
      "step": 5064
    },
    {
      "epoch": 0.6847544127960523,
      "grad_norm": 0.3030494749546051,
      "learning_rate": 7.726359175624537e-06,
      "loss": 0.04146261513233185,
      "step": 5065
    },
    {
      "epoch": 0.6848896061648176,
      "grad_norm": 0.3183102011680603,
      "learning_rate": 7.720374039604395e-06,
      "loss": 0.04418213665485382,
      "step": 5066
    },
    {
      "epoch": 0.6850247995335829,
      "grad_norm": 0.35937291383743286,
      "learning_rate": 7.714390419241198e-06,
      "loss": 0.05535361170768738,
      "step": 5067
    },
    {
      "epoch": 0.6851599929023482,
      "grad_norm": 0.523383617401123,
      "learning_rate": 7.70840831578076e-06,
      "loss": 0.05090562626719475,
      "step": 5068
    },
    {
      "epoch": 0.6852951862711134,
      "grad_norm": 0.7214508056640625,
      "learning_rate": 7.702427730468601e-06,
      "loss": 0.06970679759979248,
      "step": 5069
    },
    {
      "epoch": 0.6854303796398786,
      "grad_norm": 0.1958610862493515,
      "learning_rate": 7.696448664549898e-06,
      "loss": 0.04092078655958176,
      "step": 5070
    },
    {
      "epoch": 0.685565573008644,
      "grad_norm": 0.4784294068813324,
      "learning_rate": 7.690471119269541e-06,
      "loss": 0.06622794270515442,
      "step": 5071
    },
    {
      "epoch": 0.6857007663774092,
      "grad_norm": 0.39987921714782715,
      "learning_rate": 7.684495095872073e-06,
      "loss": 0.05952848494052887,
      "step": 5072
    },
    {
      "epoch": 0.6858359597461744,
      "grad_norm": 0.3583393096923828,
      "learning_rate": 7.678520595601728e-06,
      "loss": 0.0378875732421875,
      "step": 5073
    },
    {
      "epoch": 0.6859711531149397,
      "grad_norm": 0.6153355240821838,
      "learning_rate": 7.672547619702445e-06,
      "loss": 0.06293143332004547,
      "step": 5074
    },
    {
      "epoch": 0.6861063464837049,
      "grad_norm": 0.21090540289878845,
      "learning_rate": 7.666576169417823e-06,
      "loss": 0.026582174003124237,
      "step": 5075
    },
    {
      "epoch": 0.6862415398524703,
      "grad_norm": 0.6142287254333496,
      "learning_rate": 7.660606245991147e-06,
      "loss": 0.05657009780406952,
      "step": 5076
    },
    {
      "epoch": 0.6863767332212355,
      "grad_norm": 1.086599588394165,
      "learning_rate": 7.654637850665393e-06,
      "loss": 0.067495197057724,
      "step": 5077
    },
    {
      "epoch": 0.6865119265900007,
      "grad_norm": 0.5097535252571106,
      "learning_rate": 7.648670984683199e-06,
      "loss": 0.0701148509979248,
      "step": 5078
    },
    {
      "epoch": 0.686647119958766,
      "grad_norm": 0.5392943024635315,
      "learning_rate": 7.642705649286916e-06,
      "loss": 0.05803874135017395,
      "step": 5079
    },
    {
      "epoch": 0.6867823133275313,
      "grad_norm": 0.24042265117168427,
      "learning_rate": 7.63674184571855e-06,
      "loss": 0.050581805408000946,
      "step": 5080
    },
    {
      "epoch": 0.6869175066962966,
      "grad_norm": 0.4314679503440857,
      "learning_rate": 7.630779575219797e-06,
      "loss": 0.07898205518722534,
      "step": 5081
    },
    {
      "epoch": 0.6870527000650618,
      "grad_norm": 0.20744667947292328,
      "learning_rate": 7.6248188390320344e-06,
      "loss": 0.047296687960624695,
      "step": 5082
    },
    {
      "epoch": 0.687187893433827,
      "grad_norm": 0.4768024981021881,
      "learning_rate": 7.6188596383963135e-06,
      "loss": 0.05420637130737305,
      "step": 5083
    },
    {
      "epoch": 0.6873230868025924,
      "grad_norm": 0.28742244839668274,
      "learning_rate": 7.612901974553388e-06,
      "loss": 0.04789707064628601,
      "step": 5084
    },
    {
      "epoch": 0.6874582801713576,
      "grad_norm": 0.3358917236328125,
      "learning_rate": 7.606945848743653e-06,
      "loss": 0.04327031970024109,
      "step": 5085
    },
    {
      "epoch": 0.6875934735401229,
      "grad_norm": 1.0368911027908325,
      "learning_rate": 7.600991262207221e-06,
      "loss": 0.0589791014790535,
      "step": 5086
    },
    {
      "epoch": 0.6877286669088881,
      "grad_norm": 0.335031658411026,
      "learning_rate": 7.595038216183867e-06,
      "loss": 0.05144873261451721,
      "step": 5087
    },
    {
      "epoch": 0.6878638602776533,
      "grad_norm": 0.9198899865150452,
      "learning_rate": 7.589086711913037e-06,
      "loss": 0.060086995363235474,
      "step": 5088
    },
    {
      "epoch": 0.6879990536464187,
      "grad_norm": 0.5198059678077698,
      "learning_rate": 7.583136750633885e-06,
      "loss": 0.049209460616111755,
      "step": 5089
    },
    {
      "epoch": 0.6881342470151839,
      "grad_norm": 0.508225679397583,
      "learning_rate": 7.577188333585202e-06,
      "loss": 0.06779351830482483,
      "step": 5090
    },
    {
      "epoch": 0.6882694403839492,
      "grad_norm": 0.462271511554718,
      "learning_rate": 7.5712414620054975e-06,
      "loss": 0.05050325393676758,
      "step": 5091
    },
    {
      "epoch": 0.6884046337527144,
      "grad_norm": 0.5502487421035767,
      "learning_rate": 7.565296137132935e-06,
      "loss": 0.051377713680267334,
      "step": 5092
    },
    {
      "epoch": 0.6885398271214797,
      "grad_norm": 1.556666374206543,
      "learning_rate": 7.559352360205357e-06,
      "loss": 0.05573230981826782,
      "step": 5093
    },
    {
      "epoch": 0.688675020490245,
      "grad_norm": 0.43901729583740234,
      "learning_rate": 7.553410132460308e-06,
      "loss": 0.04742233455181122,
      "step": 5094
    },
    {
      "epoch": 0.6888102138590102,
      "grad_norm": 0.46726635098457336,
      "learning_rate": 7.547469455134968e-06,
      "loss": 0.05612045153975487,
      "step": 5095
    },
    {
      "epoch": 0.6889454072277755,
      "grad_norm": 0.8445354104042053,
      "learning_rate": 7.541530329466236e-06,
      "loss": 0.051213786005973816,
      "step": 5096
    },
    {
      "epoch": 0.6890806005965407,
      "grad_norm": 0.42055872082710266,
      "learning_rate": 7.535592756690661e-06,
      "loss": 0.04920297861099243,
      "step": 5097
    },
    {
      "epoch": 0.689215793965306,
      "grad_norm": 0.26518017053604126,
      "learning_rate": 7.52965673804448e-06,
      "loss": 0.05163450539112091,
      "step": 5098
    },
    {
      "epoch": 0.6893509873340713,
      "grad_norm": 0.32651418447494507,
      "learning_rate": 7.5237222747636025e-06,
      "loss": 0.05538457632064819,
      "step": 5099
    },
    {
      "epoch": 0.6894861807028365,
      "grad_norm": 0.5737900733947754,
      "learning_rate": 7.517789368083611e-06,
      "loss": 0.05658465251326561,
      "step": 5100
    },
    {
      "epoch": 0.6896213740716017,
      "grad_norm": 0.8376307487487793,
      "learning_rate": 7.511858019239778e-06,
      "loss": 0.049514517188072205,
      "step": 5101
    },
    {
      "epoch": 0.6897565674403671,
      "grad_norm": 0.3505117893218994,
      "learning_rate": 7.505928229467038e-06,
      "loss": 0.0646526962518692,
      "step": 5102
    },
    {
      "epoch": 0.6898917608091323,
      "grad_norm": 0.8085762858390808,
      "learning_rate": 7.500000000000004e-06,
      "loss": 0.06394210457801819,
      "step": 5103
    },
    {
      "epoch": 0.6900269541778976,
      "grad_norm": 0.7817121744155884,
      "learning_rate": 7.494073332072963e-06,
      "loss": 0.06659898161888123,
      "step": 5104
    },
    {
      "epoch": 0.6901621475466628,
      "grad_norm": 0.3992208242416382,
      "learning_rate": 7.488148226919877e-06,
      "loss": 0.030706388875842094,
      "step": 5105
    },
    {
      "epoch": 0.690297340915428,
      "grad_norm": 0.19241000711917877,
      "learning_rate": 7.482224685774393e-06,
      "loss": 0.04445412755012512,
      "step": 5106
    },
    {
      "epoch": 0.6904325342841934,
      "grad_norm": 0.5191207528114319,
      "learning_rate": 7.4763027098698184e-06,
      "loss": 0.051409762352705,
      "step": 5107
    },
    {
      "epoch": 0.6905677276529586,
      "grad_norm": 0.5312533378601074,
      "learning_rate": 7.470382300439143e-06,
      "loss": 0.04130159318447113,
      "step": 5108
    },
    {
      "epoch": 0.6907029210217239,
      "grad_norm": 0.45414721965789795,
      "learning_rate": 7.4644634587150225e-06,
      "loss": 0.05870556831359863,
      "step": 5109
    },
    {
      "epoch": 0.6908381143904891,
      "grad_norm": 0.3957090973854065,
      "learning_rate": 7.4585461859297906e-06,
      "loss": 0.04819725453853607,
      "step": 5110
    },
    {
      "epoch": 0.6909733077592544,
      "grad_norm": 0.5227088928222656,
      "learning_rate": 7.452630483315463e-06,
      "loss": 0.04227384924888611,
      "step": 5111
    },
    {
      "epoch": 0.6911085011280197,
      "grad_norm": 0.29937782883644104,
      "learning_rate": 7.4467163521037186e-06,
      "loss": 0.04523632675409317,
      "step": 5112
    },
    {
      "epoch": 0.6912436944967849,
      "grad_norm": 0.30863818526268005,
      "learning_rate": 7.440803793525907e-06,
      "loss": 0.04430852830410004,
      "step": 5113
    },
    {
      "epoch": 0.6913788878655502,
      "grad_norm": 0.5694001317024231,
      "learning_rate": 7.434892808813056e-06,
      "loss": 0.05520200729370117,
      "step": 5114
    },
    {
      "epoch": 0.6915140812343155,
      "grad_norm": 0.7776246666908264,
      "learning_rate": 7.42898339919586e-06,
      "loss": 0.05291074514389038,
      "step": 5115
    },
    {
      "epoch": 0.6916492746030807,
      "grad_norm": 1.1083894968032837,
      "learning_rate": 7.423075565904698e-06,
      "loss": 0.06358221918344498,
      "step": 5116
    },
    {
      "epoch": 0.691784467971846,
      "grad_norm": 0.6599995493888855,
      "learning_rate": 7.417169310169609e-06,
      "loss": 0.06120622158050537,
      "step": 5117
    },
    {
      "epoch": 0.6919196613406112,
      "grad_norm": 0.9249976873397827,
      "learning_rate": 7.411264633220305e-06,
      "loss": 0.051445357501506805,
      "step": 5118
    },
    {
      "epoch": 0.6920548547093766,
      "grad_norm": 0.4731396734714508,
      "learning_rate": 7.405361536286174e-06,
      "loss": 0.051790982484817505,
      "step": 5119
    },
    {
      "epoch": 0.6921900480781418,
      "grad_norm": 0.4082396328449249,
      "learning_rate": 7.399460020596266e-06,
      "loss": 0.06415332853794098,
      "step": 5120
    },
    {
      "epoch": 0.692325241446907,
      "grad_norm": 0.5769826173782349,
      "learning_rate": 7.393560087379322e-06,
      "loss": 0.05486782267689705,
      "step": 5121
    },
    {
      "epoch": 0.6924604348156723,
      "grad_norm": 0.3882736265659332,
      "learning_rate": 7.3876617378637195e-06,
      "loss": 0.04376395046710968,
      "step": 5122
    },
    {
      "epoch": 0.6925956281844375,
      "grad_norm": 0.3410462737083435,
      "learning_rate": 7.381764973277543e-06,
      "loss": 0.04908747225999832,
      "step": 5123
    },
    {
      "epoch": 0.6927308215532029,
      "grad_norm": 0.5017903447151184,
      "learning_rate": 7.375869794848525e-06,
      "loss": 0.057123810052871704,
      "step": 5124
    },
    {
      "epoch": 0.6928660149219681,
      "grad_norm": 0.5975813269615173,
      "learning_rate": 7.3699762038040654e-06,
      "loss": 0.0574532151222229,
      "step": 5125
    },
    {
      "epoch": 0.6930012082907333,
      "grad_norm": 0.7153971791267395,
      "learning_rate": 7.364084201371261e-06,
      "loss": 0.07156383991241455,
      "step": 5126
    },
    {
      "epoch": 0.6931364016594986,
      "grad_norm": 1.0082093477249146,
      "learning_rate": 7.3581937887768334e-06,
      "loss": 0.05073480308055878,
      "step": 5127
    },
    {
      "epoch": 0.6932715950282639,
      "grad_norm": 1.5162277221679688,
      "learning_rate": 7.352304967247217e-06,
      "loss": 0.07791519165039062,
      "step": 5128
    },
    {
      "epoch": 0.6934067883970291,
      "grad_norm": 0.32575663924217224,
      "learning_rate": 7.346417738008487e-06,
      "loss": 0.0532938614487648,
      "step": 5129
    },
    {
      "epoch": 0.6935419817657944,
      "grad_norm": 0.37175315618515015,
      "learning_rate": 7.340532102286399e-06,
      "loss": 0.0561351478099823,
      "step": 5130
    },
    {
      "epoch": 0.6936771751345596,
      "grad_norm": 0.6612153649330139,
      "learning_rate": 7.3346480613063725e-06,
      "loss": 0.061547353863716125,
      "step": 5131
    },
    {
      "epoch": 0.693812368503325,
      "grad_norm": 0.5450572967529297,
      "learning_rate": 7.328765616293491e-06,
      "loss": 0.04134263098239899,
      "step": 5132
    },
    {
      "epoch": 0.6939475618720902,
      "grad_norm": 0.7792589664459229,
      "learning_rate": 7.322884768472521e-06,
      "loss": 0.05712822079658508,
      "step": 5133
    },
    {
      "epoch": 0.6940827552408554,
      "grad_norm": 1.0409884452819824,
      "learning_rate": 7.317005519067881e-06,
      "loss": 0.05768483877182007,
      "step": 5134
    },
    {
      "epoch": 0.6942179486096207,
      "grad_norm": 0.2901518642902374,
      "learning_rate": 7.311127869303665e-06,
      "loss": 0.046974167227745056,
      "step": 5135
    },
    {
      "epoch": 0.6943531419783859,
      "grad_norm": 0.5461139678955078,
      "learning_rate": 7.305251820403628e-06,
      "loss": 0.055159635841846466,
      "step": 5136
    },
    {
      "epoch": 0.6944883353471513,
      "grad_norm": 0.4399124085903168,
      "learning_rate": 7.299377373591188e-06,
      "loss": 0.07501746714115143,
      "step": 5137
    },
    {
      "epoch": 0.6946235287159165,
      "grad_norm": 1.0696556568145752,
      "learning_rate": 7.29350453008945e-06,
      "loss": 0.060276709496974945,
      "step": 5138
    },
    {
      "epoch": 0.6947587220846817,
      "grad_norm": 0.7495952844619751,
      "learning_rate": 7.287633291121166e-06,
      "loss": 0.053207822144031525,
      "step": 5139
    },
    {
      "epoch": 0.694893915453447,
      "grad_norm": 0.487591028213501,
      "learning_rate": 7.281763657908756e-06,
      "loss": 0.05811338499188423,
      "step": 5140
    },
    {
      "epoch": 0.6950291088222122,
      "grad_norm": 0.38049882650375366,
      "learning_rate": 7.275895631674313e-06,
      "loss": 0.05522039532661438,
      "step": 5141
    },
    {
      "epoch": 0.6951643021909776,
      "grad_norm": 1.051038146018982,
      "learning_rate": 7.2700292136395826e-06,
      "loss": 0.04356321692466736,
      "step": 5142
    },
    {
      "epoch": 0.6952994955597428,
      "grad_norm": 0.38571301102638245,
      "learning_rate": 7.264164405025997e-06,
      "loss": 0.05928133428096771,
      "step": 5143
    },
    {
      "epoch": 0.695434688928508,
      "grad_norm": 0.6754260063171387,
      "learning_rate": 7.2583012070546364e-06,
      "loss": 0.07309995591640472,
      "step": 5144
    },
    {
      "epoch": 0.6955698822972733,
      "grad_norm": 0.4771690368652344,
      "learning_rate": 7.252439620946247e-06,
      "loss": 0.04176047444343567,
      "step": 5145
    },
    {
      "epoch": 0.6957050756660386,
      "grad_norm": 1.246132731437683,
      "learning_rate": 7.246579647921243e-06,
      "loss": 0.06695803999900818,
      "step": 5146
    },
    {
      "epoch": 0.6958402690348039,
      "grad_norm": 0.9741593599319458,
      "learning_rate": 7.240721289199699e-06,
      "loss": 0.057047039270401,
      "step": 5147
    },
    {
      "epoch": 0.6959754624035691,
      "grad_norm": 1.1595772504806519,
      "learning_rate": 7.234864546001364e-06,
      "loss": 0.07157918810844421,
      "step": 5148
    },
    {
      "epoch": 0.6961106557723343,
      "grad_norm": 0.8684871196746826,
      "learning_rate": 7.229009419545638e-06,
      "loss": 0.0670970231294632,
      "step": 5149
    },
    {
      "epoch": 0.6962458491410997,
      "grad_norm": 0.40889328718185425,
      "learning_rate": 7.223155911051593e-06,
      "loss": 0.06505045294761658,
      "step": 5150
    },
    {
      "epoch": 0.6963810425098649,
      "grad_norm": 0.3071252405643463,
      "learning_rate": 7.2173040217379575e-06,
      "loss": 0.05174090340733528,
      "step": 5151
    },
    {
      "epoch": 0.6965162358786302,
      "grad_norm": 0.56471848487854,
      "learning_rate": 7.211453752823122e-06,
      "loss": 0.041987404227256775,
      "step": 5152
    },
    {
      "epoch": 0.6966514292473954,
      "grad_norm": 0.7361147999763489,
      "learning_rate": 7.205605105525161e-06,
      "loss": 0.05496387183666229,
      "step": 5153
    },
    {
      "epoch": 0.6967866226161606,
      "grad_norm": 0.30997464060783386,
      "learning_rate": 7.19975808106177e-06,
      "loss": 0.04422182962298393,
      "step": 5154
    },
    {
      "epoch": 0.696921815984926,
      "grad_norm": 0.3769621253013611,
      "learning_rate": 7.193912680650346e-06,
      "loss": 0.06449800729751587,
      "step": 5155
    },
    {
      "epoch": 0.6970570093536912,
      "grad_norm": 0.44344809651374817,
      "learning_rate": 7.188068905507931e-06,
      "loss": 0.056416332721710205,
      "step": 5156
    },
    {
      "epoch": 0.6971922027224564,
      "grad_norm": 0.7775740027427673,
      "learning_rate": 7.182226756851223e-06,
      "loss": 0.06033511087298393,
      "step": 5157
    },
    {
      "epoch": 0.6973273960912217,
      "grad_norm": 0.2717428207397461,
      "learning_rate": 7.176386235896603e-06,
      "loss": 0.04736998677253723,
      "step": 5158
    },
    {
      "epoch": 0.697462589459987,
      "grad_norm": 0.5783759355545044,
      "learning_rate": 7.170547343860079e-06,
      "loss": 0.0531974732875824,
      "step": 5159
    },
    {
      "epoch": 0.6975977828287523,
      "grad_norm": 1.0207737684249878,
      "learning_rate": 7.164710081957355e-06,
      "loss": 0.054501160979270935,
      "step": 5160
    },
    {
      "epoch": 0.6977329761975175,
      "grad_norm": 0.36375150084495544,
      "learning_rate": 7.158874451403777e-06,
      "loss": 0.05721798911690712,
      "step": 5161
    },
    {
      "epoch": 0.6978681695662827,
      "grad_norm": 1.3351860046386719,
      "learning_rate": 7.15304045341435e-06,
      "loss": 0.06120528280735016,
      "step": 5162
    },
    {
      "epoch": 0.698003362935048,
      "grad_norm": 0.303088515996933,
      "learning_rate": 7.147208089203745e-06,
      "loss": 0.05221192166209221,
      "step": 5163
    },
    {
      "epoch": 0.6981385563038133,
      "grad_norm": 1.5300906896591187,
      "learning_rate": 7.141377359986288e-06,
      "loss": 0.06439690291881561,
      "step": 5164
    },
    {
      "epoch": 0.6982737496725786,
      "grad_norm": 0.3447400629520416,
      "learning_rate": 7.135548266975978e-06,
      "loss": 0.039967283606529236,
      "step": 5165
    },
    {
      "epoch": 0.6984089430413438,
      "grad_norm": 0.665357232093811,
      "learning_rate": 7.129720811386456e-06,
      "loss": 0.057531893253326416,
      "step": 5166
    },
    {
      "epoch": 0.698544136410109,
      "grad_norm": 0.5568000078201294,
      "learning_rate": 7.12389499443103e-06,
      "loss": 0.051739323884248734,
      "step": 5167
    },
    {
      "epoch": 0.6986793297788744,
      "grad_norm": 0.5877439379692078,
      "learning_rate": 7.118070817322668e-06,
      "loss": 0.06013950705528259,
      "step": 5168
    },
    {
      "epoch": 0.6988145231476396,
      "grad_norm": 0.4831162989139557,
      "learning_rate": 7.1122482812739885e-06,
      "loss": 0.06758153438568115,
      "step": 5169
    },
    {
      "epoch": 0.6989497165164049,
      "grad_norm": 0.33590078353881836,
      "learning_rate": 7.106427387497283e-06,
      "loss": 0.04840763285756111,
      "step": 5170
    },
    {
      "epoch": 0.6990849098851701,
      "grad_norm": 0.6922755241394043,
      "learning_rate": 7.10060813720449e-06,
      "loss": 0.07645237445831299,
      "step": 5171
    },
    {
      "epoch": 0.6992201032539354,
      "grad_norm": 0.2970874607563019,
      "learning_rate": 7.094790531607207e-06,
      "loss": 0.045455269515514374,
      "step": 5172
    },
    {
      "epoch": 0.6993552966227007,
      "grad_norm": 0.7809378504753113,
      "learning_rate": 7.088974571916692e-06,
      "loss": 0.05412168800830841,
      "step": 5173
    },
    {
      "epoch": 0.6994904899914659,
      "grad_norm": 0.5885297060012817,
      "learning_rate": 7.0831602593438515e-06,
      "loss": 0.04803713411092758,
      "step": 5174
    },
    {
      "epoch": 0.6996256833602312,
      "grad_norm": 0.4061662554740906,
      "learning_rate": 7.077347595099269e-06,
      "loss": 0.03494921326637268,
      "step": 5175
    },
    {
      "epoch": 0.6997608767289965,
      "grad_norm": 0.6193325519561768,
      "learning_rate": 7.071536580393166e-06,
      "loss": 0.05285486578941345,
      "step": 5176
    },
    {
      "epoch": 0.6998960700977617,
      "grad_norm": 0.35865479707717896,
      "learning_rate": 7.065727216435426e-06,
      "loss": 0.04329541325569153,
      "step": 5177
    },
    {
      "epoch": 0.700031263466527,
      "grad_norm": 0.406389445066452,
      "learning_rate": 7.05991950443559e-06,
      "loss": 0.0642390251159668,
      "step": 5178
    },
    {
      "epoch": 0.7001664568352922,
      "grad_norm": 0.9310538172721863,
      "learning_rate": 7.05411344560285e-06,
      "loss": 0.047193270176649094,
      "step": 5179
    },
    {
      "epoch": 0.7003016502040575,
      "grad_norm": 0.3351069986820221,
      "learning_rate": 7.048309041146069e-06,
      "loss": 0.0356617271900177,
      "step": 5180
    },
    {
      "epoch": 0.7004368435728228,
      "grad_norm": 0.4099974036216736,
      "learning_rate": 7.0425062922737495e-06,
      "loss": 0.04956263303756714,
      "step": 5181
    },
    {
      "epoch": 0.700572036941588,
      "grad_norm": 0.4174969494342804,
      "learning_rate": 7.036705200194053e-06,
      "loss": 0.06671006977558136,
      "step": 5182
    },
    {
      "epoch": 0.7007072303103533,
      "grad_norm": 0.262410044670105,
      "learning_rate": 7.0309057661148e-06,
      "loss": 0.03994864597916603,
      "step": 5183
    },
    {
      "epoch": 0.7008424236791185,
      "grad_norm": 0.31396976113319397,
      "learning_rate": 7.0251079912434565e-06,
      "loss": 0.05015129595994949,
      "step": 5184
    },
    {
      "epoch": 0.7009776170478837,
      "grad_norm": 0.5556950569152832,
      "learning_rate": 7.019311876787169e-06,
      "loss": 0.049565330147743225,
      "step": 5185
    },
    {
      "epoch": 0.7011128104166491,
      "grad_norm": 0.9157549738883972,
      "learning_rate": 7.013517423952696e-06,
      "loss": 0.05310907959938049,
      "step": 5186
    },
    {
      "epoch": 0.7012480037854143,
      "grad_norm": 0.3359881639480591,
      "learning_rate": 7.0077246339464904e-06,
      "loss": 0.04812927544116974,
      "step": 5187
    },
    {
      "epoch": 0.7013831971541796,
      "grad_norm": 0.3641676604747772,
      "learning_rate": 7.001933507974635e-06,
      "loss": 0.04901690036058426,
      "step": 5188
    },
    {
      "epoch": 0.7015183905229448,
      "grad_norm": 0.5193299651145935,
      "learning_rate": 6.996144047242868e-06,
      "loss": 0.057293668389320374,
      "step": 5189
    },
    {
      "epoch": 0.7016535838917101,
      "grad_norm": 0.713130533695221,
      "learning_rate": 6.9903562529566044e-06,
      "loss": 0.059603363275527954,
      "step": 5190
    },
    {
      "epoch": 0.7017887772604754,
      "grad_norm": 0.6724171042442322,
      "learning_rate": 6.984570126320869e-06,
      "loss": 0.05180485546588898,
      "step": 5191
    },
    {
      "epoch": 0.7019239706292406,
      "grad_norm": 0.4026876986026764,
      "learning_rate": 6.978785668540384e-06,
      "loss": 0.06488806009292603,
      "step": 5192
    },
    {
      "epoch": 0.7020591639980059,
      "grad_norm": 0.479975163936615,
      "learning_rate": 6.973002880819496e-06,
      "loss": 0.044665493071079254,
      "step": 5193
    },
    {
      "epoch": 0.7021943573667712,
      "grad_norm": 1.061629295349121,
      "learning_rate": 6.96722176436221e-06,
      "loss": 0.0392318069934845,
      "step": 5194
    },
    {
      "epoch": 0.7023295507355364,
      "grad_norm": 0.34729188680648804,
      "learning_rate": 6.9614423203721975e-06,
      "loss": 0.048036955296993256,
      "step": 5195
    },
    {
      "epoch": 0.7024647441043017,
      "grad_norm": 0.3623282313346863,
      "learning_rate": 6.955664550052749e-06,
      "loss": 0.04847604036331177,
      "step": 5196
    },
    {
      "epoch": 0.7025999374730669,
      "grad_norm": 0.7228285074234009,
      "learning_rate": 6.949888454606847e-06,
      "loss": 0.04531398415565491,
      "step": 5197
    },
    {
      "epoch": 0.7027351308418323,
      "grad_norm": 0.6944763660430908,
      "learning_rate": 6.944114035237095e-06,
      "loss": 0.06294602155685425,
      "step": 5198
    },
    {
      "epoch": 0.7028703242105975,
      "grad_norm": 0.3658258616924286,
      "learning_rate": 6.93834129314576e-06,
      "loss": 0.04628367722034454,
      "step": 5199
    },
    {
      "epoch": 0.7030055175793627,
      "grad_norm": 0.39812472462654114,
      "learning_rate": 6.932570229534759e-06,
      "loss": 0.05470454692840576,
      "step": 5200
    },
    {
      "epoch": 0.703140710948128,
      "grad_norm": 0.3537338078022003,
      "learning_rate": 6.9268008456056505e-06,
      "loss": 0.04562537372112274,
      "step": 5201
    },
    {
      "epoch": 0.7032759043168932,
      "grad_norm": 0.5964049696922302,
      "learning_rate": 6.921033142559664e-06,
      "loss": 0.05565641075372696,
      "step": 5202
    },
    {
      "epoch": 0.7034110976856586,
      "grad_norm": 0.8398289084434509,
      "learning_rate": 6.915267121597659e-06,
      "loss": 0.06396631896495819,
      "step": 5203
    },
    {
      "epoch": 0.7035462910544238,
      "grad_norm": 0.6836197376251221,
      "learning_rate": 6.909502783920153e-06,
      "loss": 0.05060915648937225,
      "step": 5204
    },
    {
      "epoch": 0.703681484423189,
      "grad_norm": 0.7767529487609863,
      "learning_rate": 6.903740130727312e-06,
      "loss": 0.04536814987659454,
      "step": 5205
    },
    {
      "epoch": 0.7038166777919543,
      "grad_norm": 0.8539150953292847,
      "learning_rate": 6.8979791632189425e-06,
      "loss": 0.06451554596424103,
      "step": 5206
    },
    {
      "epoch": 0.7039518711607196,
      "grad_norm": 0.4127066433429718,
      "learning_rate": 6.892219882594523e-06,
      "loss": 0.05285489559173584,
      "step": 5207
    },
    {
      "epoch": 0.7040870645294848,
      "grad_norm": 0.5851449370384216,
      "learning_rate": 6.886462290053159e-06,
      "loss": 0.05206361040472984,
      "step": 5208
    },
    {
      "epoch": 0.7042222578982501,
      "grad_norm": 0.6160390377044678,
      "learning_rate": 6.880706386793614e-06,
      "loss": 0.05865874141454697,
      "step": 5209
    },
    {
      "epoch": 0.7043574512670153,
      "grad_norm": 0.7806863784790039,
      "learning_rate": 6.874952174014298e-06,
      "loss": 0.0446733683347702,
      "step": 5210
    },
    {
      "epoch": 0.7044926446357807,
      "grad_norm": 1.116327166557312,
      "learning_rate": 6.8691996529132585e-06,
      "loss": 0.0803598091006279,
      "step": 5211
    },
    {
      "epoch": 0.7046278380045459,
      "grad_norm": 0.7639861702919006,
      "learning_rate": 6.863448824688217e-06,
      "loss": 0.06711077690124512,
      "step": 5212
    },
    {
      "epoch": 0.7047630313733111,
      "grad_norm": 1.137369990348816,
      "learning_rate": 6.857699690536521e-06,
      "loss": 0.061547838151454926,
      "step": 5213
    },
    {
      "epoch": 0.7048982247420764,
      "grad_norm": 1.2901512384414673,
      "learning_rate": 6.8519522516551685e-06,
      "loss": 0.06110933795571327,
      "step": 5214
    },
    {
      "epoch": 0.7050334181108416,
      "grad_norm": 0.929878830909729,
      "learning_rate": 6.846206509240807e-06,
      "loss": 0.0629815012216568,
      "step": 5215
    },
    {
      "epoch": 0.705168611479607,
      "grad_norm": 0.6455501317977905,
      "learning_rate": 6.840462464489726e-06,
      "loss": 0.05758979916572571,
      "step": 5216
    },
    {
      "epoch": 0.7053038048483722,
      "grad_norm": 0.29729747772216797,
      "learning_rate": 6.834720118597879e-06,
      "loss": 0.041485995054244995,
      "step": 5217
    },
    {
      "epoch": 0.7054389982171374,
      "grad_norm": 1.0058354139328003,
      "learning_rate": 6.828979472760846e-06,
      "loss": 0.07757961750030518,
      "step": 5218
    },
    {
      "epoch": 0.7055741915859027,
      "grad_norm": 0.25626692175865173,
      "learning_rate": 6.823240528173858e-06,
      "loss": 0.05111341178417206,
      "step": 5219
    },
    {
      "epoch": 0.705709384954668,
      "grad_norm": 0.3151022493839264,
      "learning_rate": 6.817503286031797e-06,
      "loss": 0.03189639747142792,
      "step": 5220
    },
    {
      "epoch": 0.7058445783234333,
      "grad_norm": 0.6382620930671692,
      "learning_rate": 6.811767747529181e-06,
      "loss": 0.05536612868309021,
      "step": 5221
    },
    {
      "epoch": 0.7059797716921985,
      "grad_norm": 1.3285009860992432,
      "learning_rate": 6.806033913860195e-06,
      "loss": 0.0520717054605484,
      "step": 5222
    },
    {
      "epoch": 0.7061149650609637,
      "grad_norm": 0.5467832088470459,
      "learning_rate": 6.800301786218634e-06,
      "loss": 0.07821998745203018,
      "step": 5223
    },
    {
      "epoch": 0.706250158429729,
      "grad_norm": 0.32285797595977783,
      "learning_rate": 6.794571365797971e-06,
      "loss": 0.047391682863235474,
      "step": 5224
    },
    {
      "epoch": 0.7063853517984943,
      "grad_norm": 0.620692253112793,
      "learning_rate": 6.788842653791308e-06,
      "loss": 0.04966667294502258,
      "step": 5225
    },
    {
      "epoch": 0.7065205451672596,
      "grad_norm": 1.3107481002807617,
      "learning_rate": 6.7831156513913864e-06,
      "loss": 0.048862241208553314,
      "step": 5226
    },
    {
      "epoch": 0.7066557385360248,
      "grad_norm": 0.8749264478683472,
      "learning_rate": 6.777390359790614e-06,
      "loss": 0.06083528697490692,
      "step": 5227
    },
    {
      "epoch": 0.70679093190479,
      "grad_norm": 0.34699174761772156,
      "learning_rate": 6.771666780181004e-06,
      "loss": 0.05428403615951538,
      "step": 5228
    },
    {
      "epoch": 0.7069261252735554,
      "grad_norm": 0.9546473622322083,
      "learning_rate": 6.765944913754258e-06,
      "loss": 0.04775610566139221,
      "step": 5229
    },
    {
      "epoch": 0.7070613186423206,
      "grad_norm": 0.5657267570495605,
      "learning_rate": 6.7602247617016885e-06,
      "loss": 0.05932777374982834,
      "step": 5230
    },
    {
      "epoch": 0.7071965120110859,
      "grad_norm": 0.40411585569381714,
      "learning_rate": 6.754506325214265e-06,
      "loss": 0.053828880190849304,
      "step": 5231
    },
    {
      "epoch": 0.7073317053798511,
      "grad_norm": 0.5263655781745911,
      "learning_rate": 6.748789605482593e-06,
      "loss": 0.055475205183029175,
      "step": 5232
    },
    {
      "epoch": 0.7074668987486163,
      "grad_norm": 0.4490223526954651,
      "learning_rate": 6.743074603696922e-06,
      "loss": 0.048049196600914,
      "step": 5233
    },
    {
      "epoch": 0.7076020921173817,
      "grad_norm": 0.5384478569030762,
      "learning_rate": 6.737361321047155e-06,
      "loss": 0.05094766616821289,
      "step": 5234
    },
    {
      "epoch": 0.7077372854861469,
      "grad_norm": 0.43504464626312256,
      "learning_rate": 6.731649758722823e-06,
      "loss": 0.04241069406270981,
      "step": 5235
    },
    {
      "epoch": 0.7078724788549121,
      "grad_norm": 0.3702452778816223,
      "learning_rate": 6.725939917913102e-06,
      "loss": 0.04524552822113037,
      "step": 5236
    },
    {
      "epoch": 0.7080076722236774,
      "grad_norm": 0.8604250550270081,
      "learning_rate": 6.720231799806814e-06,
      "loss": 0.057267896831035614,
      "step": 5237
    },
    {
      "epoch": 0.7081428655924427,
      "grad_norm": 0.5495241284370422,
      "learning_rate": 6.7145254055924136e-06,
      "loss": 0.05888666212558746,
      "step": 5238
    },
    {
      "epoch": 0.708278058961208,
      "grad_norm": 0.34099462628364563,
      "learning_rate": 6.70882073645801e-06,
      "loss": 0.03822622448205948,
      "step": 5239
    },
    {
      "epoch": 0.7084132523299732,
      "grad_norm": 0.48479411005973816,
      "learning_rate": 6.703117793591346e-06,
      "loss": 0.04635937884449959,
      "step": 5240
    },
    {
      "epoch": 0.7085484456987384,
      "grad_norm": 0.9159398078918457,
      "learning_rate": 6.6974165781798e-06,
      "loss": 0.041539374738931656,
      "step": 5241
    },
    {
      "epoch": 0.7086836390675038,
      "grad_norm": 0.37235864996910095,
      "learning_rate": 6.691717091410398e-06,
      "loss": 0.05956708639860153,
      "step": 5242
    },
    {
      "epoch": 0.708818832436269,
      "grad_norm": 0.23226448893547058,
      "learning_rate": 6.686019334469797e-06,
      "loss": 0.03829336166381836,
      "step": 5243
    },
    {
      "epoch": 0.7089540258050343,
      "grad_norm": 0.4667544364929199,
      "learning_rate": 6.680323308544312e-06,
      "loss": 0.056112732738256454,
      "step": 5244
    },
    {
      "epoch": 0.7090892191737995,
      "grad_norm": 0.4571991562843323,
      "learning_rate": 6.674629014819879e-06,
      "loss": 0.06417109072208405,
      "step": 5245
    },
    {
      "epoch": 0.7092244125425647,
      "grad_norm": 0.2715124785900116,
      "learning_rate": 6.668936454482082e-06,
      "loss": 0.04070019721984863,
      "step": 5246
    },
    {
      "epoch": 0.7093596059113301,
      "grad_norm": 0.5346907377243042,
      "learning_rate": 6.6632456287161426e-06,
      "loss": 0.06418287754058838,
      "step": 5247
    },
    {
      "epoch": 0.7094947992800953,
      "grad_norm": 1.1269422769546509,
      "learning_rate": 6.657556538706914e-06,
      "loss": 0.05592989921569824,
      "step": 5248
    },
    {
      "epoch": 0.7096299926488606,
      "grad_norm": 0.7274010181427002,
      "learning_rate": 6.651869185638907e-06,
      "loss": 0.042320191860198975,
      "step": 5249
    },
    {
      "epoch": 0.7097651860176258,
      "grad_norm": 0.7881664633750916,
      "learning_rate": 6.646183570696253e-06,
      "loss": 0.06317999213933945,
      "step": 5250
    },
    {
      "epoch": 0.7099003793863911,
      "grad_norm": 0.3684523403644562,
      "learning_rate": 6.6404996950627275e-06,
      "loss": 0.04369693249464035,
      "step": 5251
    },
    {
      "epoch": 0.7100355727551564,
      "grad_norm": 0.39067620038986206,
      "learning_rate": 6.634817559921744e-06,
      "loss": 0.06155034154653549,
      "step": 5252
    },
    {
      "epoch": 0.7101707661239216,
      "grad_norm": 0.21505771577358246,
      "learning_rate": 6.629137166456348e-06,
      "loss": 0.03871676325798035,
      "step": 5253
    },
    {
      "epoch": 0.7103059594926869,
      "grad_norm": 0.9072936773300171,
      "learning_rate": 6.623458515849244e-06,
      "loss": 0.06219252943992615,
      "step": 5254
    },
    {
      "epoch": 0.7104411528614522,
      "grad_norm": 0.764362096786499,
      "learning_rate": 6.6177816092827354e-06,
      "loss": 0.06337004899978638,
      "step": 5255
    },
    {
      "epoch": 0.7105763462302174,
      "grad_norm": 0.58935546875,
      "learning_rate": 6.6121064479388e-06,
      "loss": 0.0512816496193409,
      "step": 5256
    },
    {
      "epoch": 0.7107115395989827,
      "grad_norm": 0.13981947302818298,
      "learning_rate": 6.606433032999031e-06,
      "loss": 0.02731890231370926,
      "step": 5257
    },
    {
      "epoch": 0.7108467329677479,
      "grad_norm": 0.3495858907699585,
      "learning_rate": 6.60076136564466e-06,
      "loss": 0.03778968006372452,
      "step": 5258
    },
    {
      "epoch": 0.7109819263365132,
      "grad_norm": 1.0834962129592896,
      "learning_rate": 6.595091447056574e-06,
      "loss": 0.06191082298755646,
      "step": 5259
    },
    {
      "epoch": 0.7111171197052785,
      "grad_norm": 0.45581018924713135,
      "learning_rate": 6.589423278415259e-06,
      "loss": 0.04199296608567238,
      "step": 5260
    },
    {
      "epoch": 0.7112523130740437,
      "grad_norm": 0.2496291995048523,
      "learning_rate": 6.583756860900872e-06,
      "loss": 0.03917001932859421,
      "step": 5261
    },
    {
      "epoch": 0.711387506442809,
      "grad_norm": 0.4140659272670746,
      "learning_rate": 6.578092195693187e-06,
      "loss": 0.0661146342754364,
      "step": 5262
    },
    {
      "epoch": 0.7115226998115742,
      "grad_norm": 1.0227134227752686,
      "learning_rate": 6.572429283971614e-06,
      "loss": 0.044877439737319946,
      "step": 5263
    },
    {
      "epoch": 0.7116578931803395,
      "grad_norm": 0.615459144115448,
      "learning_rate": 6.566768126915215e-06,
      "loss": 0.05712532624602318,
      "step": 5264
    },
    {
      "epoch": 0.7117930865491048,
      "grad_norm": 0.26087018847465515,
      "learning_rate": 6.561108725702653e-06,
      "loss": 0.04123082756996155,
      "step": 5265
    },
    {
      "epoch": 0.71192827991787,
      "grad_norm": 0.5591263771057129,
      "learning_rate": 6.555451081512262e-06,
      "loss": 0.04603567719459534,
      "step": 5266
    },
    {
      "epoch": 0.7120634732866353,
      "grad_norm": 0.4314917325973511,
      "learning_rate": 6.549795195521988e-06,
      "loss": 0.04503660649061203,
      "step": 5267
    },
    {
      "epoch": 0.7121986666554005,
      "grad_norm": 0.6721104979515076,
      "learning_rate": 6.544141068909416e-06,
      "loss": 0.05728599801659584,
      "step": 5268
    },
    {
      "epoch": 0.7123338600241658,
      "grad_norm": 0.48057857155799866,
      "learning_rate": 6.5384887028517645e-06,
      "loss": 0.0481424480676651,
      "step": 5269
    },
    {
      "epoch": 0.7124690533929311,
      "grad_norm": 0.6880866885185242,
      "learning_rate": 6.532838098525883e-06,
      "loss": 0.04324055835604668,
      "step": 5270
    },
    {
      "epoch": 0.7126042467616963,
      "grad_norm": 0.30941563844680786,
      "learning_rate": 6.5271892571082655e-06,
      "loss": 0.040373966097831726,
      "step": 5271
    },
    {
      "epoch": 0.7127394401304616,
      "grad_norm": 0.28279757499694824,
      "learning_rate": 6.521542179775029e-06,
      "loss": 0.043951280415058136,
      "step": 5272
    },
    {
      "epoch": 0.7128746334992269,
      "grad_norm": 0.553692102432251,
      "learning_rate": 6.515896867701924e-06,
      "loss": 0.05222884938120842,
      "step": 5273
    },
    {
      "epoch": 0.7130098268679921,
      "grad_norm": 0.7409149408340454,
      "learning_rate": 6.510253322064333e-06,
      "loss": 0.07178819179534912,
      "step": 5274
    },
    {
      "epoch": 0.7131450202367574,
      "grad_norm": 0.40606531500816345,
      "learning_rate": 6.504611544037267e-06,
      "loss": 0.06418925523757935,
      "step": 5275
    },
    {
      "epoch": 0.7132802136055226,
      "grad_norm": 0.7129145860671997,
      "learning_rate": 6.498971534795387e-06,
      "loss": 0.05149535834789276,
      "step": 5276
    },
    {
      "epoch": 0.713415406974288,
      "grad_norm": 0.8560460209846497,
      "learning_rate": 6.493333295512965e-06,
      "loss": 0.045469388365745544,
      "step": 5277
    },
    {
      "epoch": 0.7135506003430532,
      "grad_norm": 0.27386897802352905,
      "learning_rate": 6.487696827363916e-06,
      "loss": 0.04209578409790993,
      "step": 5278
    },
    {
      "epoch": 0.7136857937118184,
      "grad_norm": 0.5673085451126099,
      "learning_rate": 6.48206213152178e-06,
      "loss": 0.043784499168395996,
      "step": 5279
    },
    {
      "epoch": 0.7138209870805837,
      "grad_norm": 0.4211233854293823,
      "learning_rate": 6.476429209159725e-06,
      "loss": 0.05123225972056389,
      "step": 5280
    },
    {
      "epoch": 0.7139561804493489,
      "grad_norm": 0.2857317626476288,
      "learning_rate": 6.470798061450568e-06,
      "loss": 0.03570246696472168,
      "step": 5281
    },
    {
      "epoch": 0.7140913738181143,
      "grad_norm": 0.38918983936309814,
      "learning_rate": 6.465168689566738e-06,
      "loss": 0.0379403755068779,
      "step": 5282
    },
    {
      "epoch": 0.7142265671868795,
      "grad_norm": 0.5957227945327759,
      "learning_rate": 6.4595410946803e-06,
      "loss": 0.06253184378147125,
      "step": 5283
    },
    {
      "epoch": 0.7143617605556447,
      "grad_norm": 0.43165212869644165,
      "learning_rate": 6.453915277962948e-06,
      "loss": 0.05619300156831741,
      "step": 5284
    },
    {
      "epoch": 0.71449695392441,
      "grad_norm": 1.7340635061264038,
      "learning_rate": 6.4482912405860055e-06,
      "loss": 0.07015684992074966,
      "step": 5285
    },
    {
      "epoch": 0.7146321472931753,
      "grad_norm": 0.9054259657859802,
      "learning_rate": 6.442668983720434e-06,
      "loss": 0.053550295531749725,
      "step": 5286
    },
    {
      "epoch": 0.7147673406619406,
      "grad_norm": 0.7461398839950562,
      "learning_rate": 6.437048508536813e-06,
      "loss": 0.05409380793571472,
      "step": 5287
    },
    {
      "epoch": 0.7149025340307058,
      "grad_norm": 0.3588312864303589,
      "learning_rate": 6.431429816205357e-06,
      "loss": 0.05562335252761841,
      "step": 5288
    },
    {
      "epoch": 0.715037727399471,
      "grad_norm": 0.19548359513282776,
      "learning_rate": 6.425812907895904e-06,
      "loss": 0.038705095648765564,
      "step": 5289
    },
    {
      "epoch": 0.7151729207682364,
      "grad_norm": 1.8110817670822144,
      "learning_rate": 6.420197784777925e-06,
      "loss": 0.07848978042602539,
      "step": 5290
    },
    {
      "epoch": 0.7153081141370016,
      "grad_norm": 0.19149990379810333,
      "learning_rate": 6.414584448020528e-06,
      "loss": 0.03312406688928604,
      "step": 5291
    },
    {
      "epoch": 0.7154433075057668,
      "grad_norm": 0.27977022528648376,
      "learning_rate": 6.408972898792423e-06,
      "loss": 0.04751484841108322,
      "step": 5292
    },
    {
      "epoch": 0.7155785008745321,
      "grad_norm": 0.27302271127700806,
      "learning_rate": 6.4033631382619766e-06,
      "loss": 0.04971819370985031,
      "step": 5293
    },
    {
      "epoch": 0.7157136942432973,
      "grad_norm": 0.31480973958969116,
      "learning_rate": 6.397755167597171e-06,
      "loss": 0.05296483635902405,
      "step": 5294
    },
    {
      "epoch": 0.7158488876120627,
      "grad_norm": 0.5366674065589905,
      "learning_rate": 6.392148987965603e-06,
      "loss": 0.0478631854057312,
      "step": 5295
    },
    {
      "epoch": 0.7159840809808279,
      "grad_norm": 0.23235802352428436,
      "learning_rate": 6.386544600534532e-06,
      "loss": 0.04729503393173218,
      "step": 5296
    },
    {
      "epoch": 0.7161192743495931,
      "grad_norm": 0.6369978189468384,
      "learning_rate": 6.3809420064707965e-06,
      "loss": 0.05295950174331665,
      "step": 5297
    },
    {
      "epoch": 0.7162544677183584,
      "grad_norm": 0.3334389626979828,
      "learning_rate": 6.375341206940902e-06,
      "loss": 0.05260632932186127,
      "step": 5298
    },
    {
      "epoch": 0.7163896610871237,
      "grad_norm": 0.4831049144268036,
      "learning_rate": 6.369742203110962e-06,
      "loss": 0.059853434562683105,
      "step": 5299
    },
    {
      "epoch": 0.716524854455889,
      "grad_norm": 0.7906429767608643,
      "learning_rate": 6.364144996146716e-06,
      "loss": 0.049375832080841064,
      "step": 5300
    },
    {
      "epoch": 0.7166600478246542,
      "grad_norm": 0.23303119838237762,
      "learning_rate": 6.358549587213534e-06,
      "loss": 0.026674766093492508,
      "step": 5301
    },
    {
      "epoch": 0.7167952411934194,
      "grad_norm": 0.39365312457084656,
      "learning_rate": 6.352955977476405e-06,
      "loss": 0.05765704810619354,
      "step": 5302
    },
    {
      "epoch": 0.7169304345621847,
      "grad_norm": 0.26632797718048096,
      "learning_rate": 6.347364168099959e-06,
      "loss": 0.041059210896492004,
      "step": 5303
    },
    {
      "epoch": 0.71706562793095,
      "grad_norm": 0.5234113931655884,
      "learning_rate": 6.341774160248435e-06,
      "loss": 0.041222766041755676,
      "step": 5304
    },
    {
      "epoch": 0.7172008212997153,
      "grad_norm": 0.3308750092983246,
      "learning_rate": 6.3361859550857e-06,
      "loss": 0.06355522572994232,
      "step": 5305
    },
    {
      "epoch": 0.7173360146684805,
      "grad_norm": 0.8985370993614197,
      "learning_rate": 6.330599553775252e-06,
      "loss": 0.05516377091407776,
      "step": 5306
    },
    {
      "epoch": 0.7174712080372457,
      "grad_norm": 0.4691896140575409,
      "learning_rate": 6.325014957480203e-06,
      "loss": 0.05438847839832306,
      "step": 5307
    },
    {
      "epoch": 0.7176064014060111,
      "grad_norm": 0.31181609630584717,
      "learning_rate": 6.319432167363305e-06,
      "loss": 0.05948318541049957,
      "step": 5308
    },
    {
      "epoch": 0.7177415947747763,
      "grad_norm": 0.599881112575531,
      "learning_rate": 6.313851184586918e-06,
      "loss": 0.04611806944012642,
      "step": 5309
    },
    {
      "epoch": 0.7178767881435416,
      "grad_norm": 0.707398533821106,
      "learning_rate": 6.308272010313037e-06,
      "loss": 0.054347846657037735,
      "step": 5310
    },
    {
      "epoch": 0.7180119815123068,
      "grad_norm": 1.5443148612976074,
      "learning_rate": 6.302694645703273e-06,
      "loss": 0.05726616457104683,
      "step": 5311
    },
    {
      "epoch": 0.718147174881072,
      "grad_norm": 0.3165954649448395,
      "learning_rate": 6.297119091918857e-06,
      "loss": 0.05020739138126373,
      "step": 5312
    },
    {
      "epoch": 0.7182823682498374,
      "grad_norm": 0.3362783193588257,
      "learning_rate": 6.2915453501206634e-06,
      "loss": 0.04550901800394058,
      "step": 5313
    },
    {
      "epoch": 0.7184175616186026,
      "grad_norm": 0.4389294981956482,
      "learning_rate": 6.285973421469166e-06,
      "loss": 0.049336910247802734,
      "step": 5314
    },
    {
      "epoch": 0.7185527549873678,
      "grad_norm": 0.25169846415519714,
      "learning_rate": 6.28040330712447e-06,
      "loss": 0.032390594482421875,
      "step": 5315
    },
    {
      "epoch": 0.7186879483561331,
      "grad_norm": 0.6153716444969177,
      "learning_rate": 6.274835008246304e-06,
      "loss": 0.06711088865995407,
      "step": 5316
    },
    {
      "epoch": 0.7188231417248984,
      "grad_norm": 1.0294628143310547,
      "learning_rate": 6.269268525994013e-06,
      "loss": 0.07531948387622833,
      "step": 5317
    },
    {
      "epoch": 0.7189583350936637,
      "grad_norm": 1.553651213645935,
      "learning_rate": 6.263703861526578e-06,
      "loss": 0.05934993922710419,
      "step": 5318
    },
    {
      "epoch": 0.7190935284624289,
      "grad_norm": 0.3488035500049591,
      "learning_rate": 6.258141016002587e-06,
      "loss": 0.05223589390516281,
      "step": 5319
    },
    {
      "epoch": 0.7192287218311941,
      "grad_norm": 0.4481896162033081,
      "learning_rate": 6.252579990580254e-06,
      "loss": 0.05611022561788559,
      "step": 5320
    },
    {
      "epoch": 0.7193639151999595,
      "grad_norm": 0.5994291305541992,
      "learning_rate": 6.247020786417412e-06,
      "loss": 0.05536480247974396,
      "step": 5321
    },
    {
      "epoch": 0.7194991085687247,
      "grad_norm": 0.7593721151351929,
      "learning_rate": 6.241463404671516e-06,
      "loss": 0.04418239742517471,
      "step": 5322
    },
    {
      "epoch": 0.71963430193749,
      "grad_norm": 0.23984026908874512,
      "learning_rate": 6.235907846499655e-06,
      "loss": 0.03307386487722397,
      "step": 5323
    },
    {
      "epoch": 0.7197694953062552,
      "grad_norm": 0.26421529054641724,
      "learning_rate": 6.230354113058505e-06,
      "loss": 0.03956104815006256,
      "step": 5324
    },
    {
      "epoch": 0.7199046886750204,
      "grad_norm": 0.3595588803291321,
      "learning_rate": 6.2248022055044e-06,
      "loss": 0.03947695344686508,
      "step": 5325
    },
    {
      "epoch": 0.7200398820437858,
      "grad_norm": 0.44018399715423584,
      "learning_rate": 6.219252124993271e-06,
      "loss": 0.049884915351867676,
      "step": 5326
    },
    {
      "epoch": 0.720175075412551,
      "grad_norm": 0.4145936369895935,
      "learning_rate": 6.213703872680668e-06,
      "loss": 0.050020769238471985,
      "step": 5327
    },
    {
      "epoch": 0.7203102687813163,
      "grad_norm": 0.27397283911705017,
      "learning_rate": 6.208157449721785e-06,
      "loss": 0.05111359804868698,
      "step": 5328
    },
    {
      "epoch": 0.7204454621500815,
      "grad_norm": 0.798911988735199,
      "learning_rate": 6.202612857271393e-06,
      "loss": 0.056186266243457794,
      "step": 5329
    },
    {
      "epoch": 0.7205806555188468,
      "grad_norm": 1.0204287767410278,
      "learning_rate": 6.197070096483923e-06,
      "loss": 0.06357286870479584,
      "step": 5330
    },
    {
      "epoch": 0.7207158488876121,
      "grad_norm": 0.359225332736969,
      "learning_rate": 6.191529168513403e-06,
      "loss": 0.0479716956615448,
      "step": 5331
    },
    {
      "epoch": 0.7208510422563773,
      "grad_norm": 0.5838215947151184,
      "learning_rate": 6.1859900745134755e-06,
      "loss": 0.06106770038604736,
      "step": 5332
    },
    {
      "epoch": 0.7209862356251426,
      "grad_norm": 0.6393482089042664,
      "learning_rate": 6.180452815637429e-06,
      "loss": 0.05821022391319275,
      "step": 5333
    },
    {
      "epoch": 0.7211214289939079,
      "grad_norm": 0.39810070395469666,
      "learning_rate": 6.174917393038126e-06,
      "loss": 0.04513488709926605,
      "step": 5334
    },
    {
      "epoch": 0.7212566223626731,
      "grad_norm": 0.8694458603858948,
      "learning_rate": 6.169383807868088e-06,
      "loss": 0.0457916259765625,
      "step": 5335
    },
    {
      "epoch": 0.7213918157314384,
      "grad_norm": 0.3716004490852356,
      "learning_rate": 6.163852061279432e-06,
      "loss": 0.04418657720088959,
      "step": 5336
    },
    {
      "epoch": 0.7215270091002036,
      "grad_norm": 0.44321274757385254,
      "learning_rate": 6.158322154423897e-06,
      "loss": 0.05373102426528931,
      "step": 5337
    },
    {
      "epoch": 0.721662202468969,
      "grad_norm": 0.7104228138923645,
      "learning_rate": 6.15279408845284e-06,
      "loss": 0.062415868043899536,
      "step": 5338
    },
    {
      "epoch": 0.7217973958377342,
      "grad_norm": 1.0298091173171997,
      "learning_rate": 6.147267864517226e-06,
      "loss": 0.058851905167102814,
      "step": 5339
    },
    {
      "epoch": 0.7219325892064994,
      "grad_norm": 0.4179043471813202,
      "learning_rate": 6.141743483767658e-06,
      "loss": 0.05466245114803314,
      "step": 5340
    },
    {
      "epoch": 0.7220677825752647,
      "grad_norm": 0.2920180857181549,
      "learning_rate": 6.136220947354333e-06,
      "loss": 0.03467337787151337,
      "step": 5341
    },
    {
      "epoch": 0.7222029759440299,
      "grad_norm": 1.014169454574585,
      "learning_rate": 6.130700256427075e-06,
      "loss": 0.055091291666030884,
      "step": 5342
    },
    {
      "epoch": 0.7223381693127952,
      "grad_norm": 0.6214808821678162,
      "learning_rate": 6.1251814121353204e-06,
      "loss": 0.0512273907661438,
      "step": 5343
    },
    {
      "epoch": 0.7224733626815605,
      "grad_norm": 0.753602147102356,
      "learning_rate": 6.1196644156281175e-06,
      "loss": 0.045717403292655945,
      "step": 5344
    },
    {
      "epoch": 0.7226085560503257,
      "grad_norm": 0.35925543308258057,
      "learning_rate": 6.114149268054143e-06,
      "loss": 0.04642605781555176,
      "step": 5345
    },
    {
      "epoch": 0.722743749419091,
      "grad_norm": 1.6378378868103027,
      "learning_rate": 6.108635970561679e-06,
      "loss": 0.06745411455631256,
      "step": 5346
    },
    {
      "epoch": 0.7228789427878562,
      "grad_norm": 0.3161318004131317,
      "learning_rate": 6.103124524298617e-06,
      "loss": 0.056581318378448486,
      "step": 5347
    },
    {
      "epoch": 0.7230141361566215,
      "grad_norm": 0.4949035942554474,
      "learning_rate": 6.097614930412475e-06,
      "loss": 0.04653365910053253,
      "step": 5348
    },
    {
      "epoch": 0.7231493295253868,
      "grad_norm": 0.4795159101486206,
      "learning_rate": 6.092107190050371e-06,
      "loss": 0.06439661979675293,
      "step": 5349
    },
    {
      "epoch": 0.723284522894152,
      "grad_norm": 0.2476195991039276,
      "learning_rate": 6.086601304359059e-06,
      "loss": 0.037961263209581375,
      "step": 5350
    },
    {
      "epoch": 0.7234197162629173,
      "grad_norm": 0.9817245602607727,
      "learning_rate": 6.081097274484887e-06,
      "loss": 0.05518001317977905,
      "step": 5351
    },
    {
      "epoch": 0.7235549096316826,
      "grad_norm": 0.415656715631485,
      "learning_rate": 6.075595101573825e-06,
      "loss": 0.05471087992191315,
      "step": 5352
    },
    {
      "epoch": 0.7236901030004478,
      "grad_norm": 0.24990646541118622,
      "learning_rate": 6.070094786771451e-06,
      "loss": 0.03266160190105438,
      "step": 5353
    },
    {
      "epoch": 0.7238252963692131,
      "grad_norm": 0.5751898884773254,
      "learning_rate": 6.06459633122296e-06,
      "loss": 0.05331885814666748,
      "step": 5354
    },
    {
      "epoch": 0.7239604897379783,
      "grad_norm": 0.39063960313796997,
      "learning_rate": 6.059099736073166e-06,
      "loss": 0.04876334220170975,
      "step": 5355
    },
    {
      "epoch": 0.7240956831067437,
      "grad_norm": 0.4047369956970215,
      "learning_rate": 6.0536050024664865e-06,
      "loss": 0.05805745720863342,
      "step": 5356
    },
    {
      "epoch": 0.7242308764755089,
      "grad_norm": 0.6284480690956116,
      "learning_rate": 6.048112131546953e-06,
      "loss": 0.046272601932287216,
      "step": 5357
    },
    {
      "epoch": 0.7243660698442741,
      "grad_norm": 0.24715310335159302,
      "learning_rate": 6.0426211244582105e-06,
      "loss": 0.03717603534460068,
      "step": 5358
    },
    {
      "epoch": 0.7245012632130394,
      "grad_norm": 0.2916569411754608,
      "learning_rate": 6.03713198234351e-06,
      "loss": 0.04737754166126251,
      "step": 5359
    },
    {
      "epoch": 0.7246364565818046,
      "grad_norm": 0.2525986135005951,
      "learning_rate": 6.0316447063457395e-06,
      "loss": 0.04116874933242798,
      "step": 5360
    },
    {
      "epoch": 0.72477164995057,
      "grad_norm": 1.6070128679275513,
      "learning_rate": 6.026159297607356e-06,
      "loss": 0.05991578847169876,
      "step": 5361
    },
    {
      "epoch": 0.7249068433193352,
      "grad_norm": 0.4565029740333557,
      "learning_rate": 6.020675757270466e-06,
      "loss": 0.05490928143262863,
      "step": 5362
    },
    {
      "epoch": 0.7250420366881004,
      "grad_norm": 0.4200201630592346,
      "learning_rate": 6.015194086476766e-06,
      "loss": 0.050050586462020874,
      "step": 5363
    },
    {
      "epoch": 0.7251772300568657,
      "grad_norm": 0.4062773883342743,
      "learning_rate": 6.009714286367565e-06,
      "loss": 0.05584326758980751,
      "step": 5364
    },
    {
      "epoch": 0.725312423425631,
      "grad_norm": 1.1536589860916138,
      "learning_rate": 6.004236358083802e-06,
      "loss": 0.07136119902133942,
      "step": 5365
    },
    {
      "epoch": 0.7254476167943963,
      "grad_norm": 0.45489436388015747,
      "learning_rate": 5.998760302765989e-06,
      "loss": 0.05478912591934204,
      "step": 5366
    },
    {
      "epoch": 0.7255828101631615,
      "grad_norm": 0.5681073665618896,
      "learning_rate": 5.993286121554289e-06,
      "loss": 0.06499084830284119,
      "step": 5367
    },
    {
      "epoch": 0.7257180035319267,
      "grad_norm": 0.3667809069156647,
      "learning_rate": 5.987813815588447e-06,
      "loss": 0.05895799398422241,
      "step": 5368
    },
    {
      "epoch": 0.725853196900692,
      "grad_norm": 0.3511812090873718,
      "learning_rate": 5.982343386007827e-06,
      "loss": 0.048358991742134094,
      "step": 5369
    },
    {
      "epoch": 0.7259883902694573,
      "grad_norm": 0.4450904428958893,
      "learning_rate": 5.976874833951404e-06,
      "loss": 0.06092090532183647,
      "step": 5370
    },
    {
      "epoch": 0.7261235836382225,
      "grad_norm": 1.4003962278366089,
      "learning_rate": 5.971408160557751e-06,
      "loss": 0.07164968550205231,
      "step": 5371
    },
    {
      "epoch": 0.7262587770069878,
      "grad_norm": 0.5836803913116455,
      "learning_rate": 5.965943366965069e-06,
      "loss": 0.04724377393722534,
      "step": 5372
    },
    {
      "epoch": 0.726393970375753,
      "grad_norm": 0.40526553988456726,
      "learning_rate": 5.960480454311155e-06,
      "loss": 0.031082840636372566,
      "step": 5373
    },
    {
      "epoch": 0.7265291637445184,
      "grad_norm": 0.28404921293258667,
      "learning_rate": 5.955019423733416e-06,
      "loss": 0.04713664948940277,
      "step": 5374
    },
    {
      "epoch": 0.7266643571132836,
      "grad_norm": 0.7471832036972046,
      "learning_rate": 5.949560276368866e-06,
      "loss": 0.06546878814697266,
      "step": 5375
    },
    {
      "epoch": 0.7267995504820488,
      "grad_norm": 0.40865886211395264,
      "learning_rate": 5.9441030133541235e-06,
      "loss": 0.05339936912059784,
      "step": 5376
    },
    {
      "epoch": 0.7269347438508141,
      "grad_norm": 0.6689923405647278,
      "learning_rate": 5.938647635825432e-06,
      "loss": 0.05192083865404129,
      "step": 5377
    },
    {
      "epoch": 0.7270699372195794,
      "grad_norm": 0.30280566215515137,
      "learning_rate": 5.933194144918623e-06,
      "loss": 0.041438695043325424,
      "step": 5378
    },
    {
      "epoch": 0.7272051305883447,
      "grad_norm": 0.30500245094299316,
      "learning_rate": 5.927742541769142e-06,
      "loss": 0.04250006377696991,
      "step": 5379
    },
    {
      "epoch": 0.7273403239571099,
      "grad_norm": 0.737541139125824,
      "learning_rate": 5.9222928275120445e-06,
      "loss": 0.057598624378442764,
      "step": 5380
    },
    {
      "epoch": 0.7274755173258751,
      "grad_norm": 1.2554136514663696,
      "learning_rate": 5.916845003281983e-06,
      "loss": 0.0700816661119461,
      "step": 5381
    },
    {
      "epoch": 0.7276107106946405,
      "grad_norm": 0.8446027636528015,
      "learning_rate": 5.911399070213234e-06,
      "loss": 0.04976761341094971,
      "step": 5382
    },
    {
      "epoch": 0.7277459040634057,
      "grad_norm": 0.3828921318054199,
      "learning_rate": 5.905955029439665e-06,
      "loss": 0.04100614786148071,
      "step": 5383
    },
    {
      "epoch": 0.727881097432171,
      "grad_norm": 0.42416954040527344,
      "learning_rate": 5.900512882094754e-06,
      "loss": 0.06199878454208374,
      "step": 5384
    },
    {
      "epoch": 0.7280162908009362,
      "grad_norm": 0.4000873565673828,
      "learning_rate": 5.8950726293115855e-06,
      "loss": 0.05113799124956131,
      "step": 5385
    },
    {
      "epoch": 0.7281514841697014,
      "grad_norm": 0.36619076132774353,
      "learning_rate": 5.889634272222844e-06,
      "loss": 0.05405128374695778,
      "step": 5386
    },
    {
      "epoch": 0.7282866775384668,
      "grad_norm": 0.3242407739162445,
      "learning_rate": 5.8841978119608345e-06,
      "loss": 0.028411703184247017,
      "step": 5387
    },
    {
      "epoch": 0.728421870907232,
      "grad_norm": 0.5705392956733704,
      "learning_rate": 5.878763249657452e-06,
      "loss": 0.053586747497320175,
      "step": 5388
    },
    {
      "epoch": 0.7285570642759973,
      "grad_norm": 0.7255110740661621,
      "learning_rate": 5.873330586444202e-06,
      "loss": 0.04739254713058472,
      "step": 5389
    },
    {
      "epoch": 0.7286922576447625,
      "grad_norm": 0.5211935043334961,
      "learning_rate": 5.867899823452193e-06,
      "loss": 0.05576735734939575,
      "step": 5390
    },
    {
      "epoch": 0.7288274510135277,
      "grad_norm": 0.28238874673843384,
      "learning_rate": 5.862470961812133e-06,
      "loss": 0.03129733353853226,
      "step": 5391
    },
    {
      "epoch": 0.7289626443822931,
      "grad_norm": 0.49288153648376465,
      "learning_rate": 5.857044002654357e-06,
      "loss": 0.04071573540568352,
      "step": 5392
    },
    {
      "epoch": 0.7290978377510583,
      "grad_norm": 0.27442362904548645,
      "learning_rate": 5.851618947108764e-06,
      "loss": 0.04104681685566902,
      "step": 5393
    },
    {
      "epoch": 0.7292330311198236,
      "grad_norm": 1.0412551164627075,
      "learning_rate": 5.8461957963048984e-06,
      "loss": 0.06329309940338135,
      "step": 5394
    },
    {
      "epoch": 0.7293682244885888,
      "grad_norm": 0.862945020198822,
      "learning_rate": 5.840774551371882e-06,
      "loss": 0.038138359785079956,
      "step": 5395
    },
    {
      "epoch": 0.7295034178573541,
      "grad_norm": 1.227352499961853,
      "learning_rate": 5.8353552134384405e-06,
      "loss": 0.04771539568901062,
      "step": 5396
    },
    {
      "epoch": 0.7296386112261194,
      "grad_norm": 0.7169512510299683,
      "learning_rate": 5.829937783632926e-06,
      "loss": 0.07296678423881531,
      "step": 5397
    },
    {
      "epoch": 0.7297738045948846,
      "grad_norm": 0.3882494866847992,
      "learning_rate": 5.824522263083256e-06,
      "loss": 0.042264342308044434,
      "step": 5398
    },
    {
      "epoch": 0.7299089979636498,
      "grad_norm": 0.5537630915641785,
      "learning_rate": 5.8191086529169855e-06,
      "loss": 0.06330940872430801,
      "step": 5399
    },
    {
      "epoch": 0.7300441913324152,
      "grad_norm": 0.5714014768600464,
      "learning_rate": 5.813696954261253e-06,
      "loss": 0.048753999173641205,
      "step": 5400
    },
    {
      "epoch": 0.7301793847011804,
      "grad_norm": 0.6927962303161621,
      "learning_rate": 5.8082871682428e-06,
      "loss": 0.07524000108242035,
      "step": 5401
    },
    {
      "epoch": 0.7303145780699457,
      "grad_norm": 0.41377681493759155,
      "learning_rate": 5.802879295987975e-06,
      "loss": 0.05869010090827942,
      "step": 5402
    },
    {
      "epoch": 0.7304497714387109,
      "grad_norm": 0.6390061378479004,
      "learning_rate": 5.797473338622722e-06,
      "loss": 0.056198105216026306,
      "step": 5403
    },
    {
      "epoch": 0.7305849648074761,
      "grad_norm": 0.33836090564727783,
      "learning_rate": 5.792069297272599e-06,
      "loss": 0.03806114196777344,
      "step": 5404
    },
    {
      "epoch": 0.7307201581762415,
      "grad_norm": 0.44180917739868164,
      "learning_rate": 5.7866671730627485e-06,
      "loss": 0.06397104263305664,
      "step": 5405
    },
    {
      "epoch": 0.7308553515450067,
      "grad_norm": 0.3084310293197632,
      "learning_rate": 5.781266967117925e-06,
      "loss": 0.06115628033876419,
      "step": 5406
    },
    {
      "epoch": 0.730990544913772,
      "grad_norm": 0.6314193606376648,
      "learning_rate": 5.7758686805624815e-06,
      "loss": 0.0561547577381134,
      "step": 5407
    },
    {
      "epoch": 0.7311257382825372,
      "grad_norm": 1.0682895183563232,
      "learning_rate": 5.7704723145203605e-06,
      "loss": 0.045564793050289154,
      "step": 5408
    },
    {
      "epoch": 0.7312609316513025,
      "grad_norm": 0.38729023933410645,
      "learning_rate": 5.765077870115126e-06,
      "loss": 0.04134967178106308,
      "step": 5409
    },
    {
      "epoch": 0.7313961250200678,
      "grad_norm": 0.6590896248817444,
      "learning_rate": 5.759685348469928e-06,
      "loss": 0.05257289111614227,
      "step": 5410
    },
    {
      "epoch": 0.731531318388833,
      "grad_norm": 0.4695395529270172,
      "learning_rate": 5.754294750707514e-06,
      "loss": 0.04570161551237106,
      "step": 5411
    },
    {
      "epoch": 0.7316665117575983,
      "grad_norm": 1.559465765953064,
      "learning_rate": 5.748906077950237e-06,
      "loss": 0.04657503589987755,
      "step": 5412
    },
    {
      "epoch": 0.7318017051263636,
      "grad_norm": 0.7578441500663757,
      "learning_rate": 5.743519331320042e-06,
      "loss": 0.053437016904354095,
      "step": 5413
    },
    {
      "epoch": 0.7319368984951288,
      "grad_norm": 0.7532030940055847,
      "learning_rate": 5.73813451193849e-06,
      "loss": 0.04919366538524628,
      "step": 5414
    },
    {
      "epoch": 0.7320720918638941,
      "grad_norm": 0.6908899545669556,
      "learning_rate": 5.7327516209267225e-06,
      "loss": 0.04903233423829079,
      "step": 5415
    },
    {
      "epoch": 0.7322072852326593,
      "grad_norm": 0.30994343757629395,
      "learning_rate": 5.727370659405486e-06,
      "loss": 0.03947867453098297,
      "step": 5416
    },
    {
      "epoch": 0.7323424786014247,
      "grad_norm": 0.8648495078086853,
      "learning_rate": 5.7219916284951265e-06,
      "loss": 0.057804450392723083,
      "step": 5417
    },
    {
      "epoch": 0.7324776719701899,
      "grad_norm": 0.5805301666259766,
      "learning_rate": 5.716614529315582e-06,
      "loss": 0.0589471310377121,
      "step": 5418
    },
    {
      "epoch": 0.7326128653389551,
      "grad_norm": 0.26550325751304626,
      "learning_rate": 5.711239362986401e-06,
      "loss": 0.035116612911224365,
      "step": 5419
    },
    {
      "epoch": 0.7327480587077204,
      "grad_norm": 0.3858984410762787,
      "learning_rate": 5.705866130626719e-06,
      "loss": 0.059017881751060486,
      "step": 5420
    },
    {
      "epoch": 0.7328832520764856,
      "grad_norm": 0.3619306683540344,
      "learning_rate": 5.700494833355271e-06,
      "loss": 0.03745993226766586,
      "step": 5421
    },
    {
      "epoch": 0.733018445445251,
      "grad_norm": 0.5689400434494019,
      "learning_rate": 5.69512547229039e-06,
      "loss": 0.062376707792282104,
      "step": 5422
    },
    {
      "epoch": 0.7331536388140162,
      "grad_norm": 0.835976243019104,
      "learning_rate": 5.689758048550001e-06,
      "loss": 0.06354323029518127,
      "step": 5423
    },
    {
      "epoch": 0.7332888321827814,
      "grad_norm": 1.202174425125122,
      "learning_rate": 5.684392563251644e-06,
      "loss": 0.051449280232191086,
      "step": 5424
    },
    {
      "epoch": 0.7334240255515467,
      "grad_norm": 0.4447128474712372,
      "learning_rate": 5.679029017512422e-06,
      "loss": 0.04517205059528351,
      "step": 5425
    },
    {
      "epoch": 0.733559218920312,
      "grad_norm": 0.33622944355010986,
      "learning_rate": 5.6736674124490684e-06,
      "loss": 0.032995931804180145,
      "step": 5426
    },
    {
      "epoch": 0.7336944122890772,
      "grad_norm": 0.3162919878959656,
      "learning_rate": 5.6683077491778935e-06,
      "loss": 0.04157820716500282,
      "step": 5427
    },
    {
      "epoch": 0.7338296056578425,
      "grad_norm": 0.3667377829551697,
      "learning_rate": 5.6629500288148044e-06,
      "loss": 0.05399259924888611,
      "step": 5428
    },
    {
      "epoch": 0.7339647990266077,
      "grad_norm": 0.9378178119659424,
      "learning_rate": 5.657594252475319e-06,
      "loss": 0.060357775539159775,
      "step": 5429
    },
    {
      "epoch": 0.734099992395373,
      "grad_norm": 0.38600993156433105,
      "learning_rate": 5.652240421274521e-06,
      "loss": 0.046939365565776825,
      "step": 5430
    },
    {
      "epoch": 0.7342351857641383,
      "grad_norm": 0.6062623858451843,
      "learning_rate": 5.646888536327121e-06,
      "loss": 0.05704595893621445,
      "step": 5431
    },
    {
      "epoch": 0.7343703791329035,
      "grad_norm": 1.3618842363357544,
      "learning_rate": 5.641538598747403e-06,
      "loss": 0.0682445615530014,
      "step": 5432
    },
    {
      "epoch": 0.7345055725016688,
      "grad_norm": 0.37157899141311646,
      "learning_rate": 5.6361906096492495e-06,
      "loss": 0.04282565787434578,
      "step": 5433
    },
    {
      "epoch": 0.734640765870434,
      "grad_norm": 0.5693402290344238,
      "learning_rate": 5.630844570146157e-06,
      "loss": 0.07153460383415222,
      "step": 5434
    },
    {
      "epoch": 0.7347759592391994,
      "grad_norm": 1.0029247999191284,
      "learning_rate": 5.625500481351176e-06,
      "loss": 0.06501775234937668,
      "step": 5435
    },
    {
      "epoch": 0.7349111526079646,
      "grad_norm": 0.43693169951438904,
      "learning_rate": 5.6201583443769895e-06,
      "loss": 0.050835371017456055,
      "step": 5436
    },
    {
      "epoch": 0.7350463459767298,
      "grad_norm": 0.6364457011222839,
      "learning_rate": 5.614818160335857e-06,
      "loss": 0.048228323459625244,
      "step": 5437
    },
    {
      "epoch": 0.7351815393454951,
      "grad_norm": 0.6963417530059814,
      "learning_rate": 5.6094799303396315e-06,
      "loss": 0.045394234359264374,
      "step": 5438
    },
    {
      "epoch": 0.7353167327142603,
      "grad_norm": 1.1840342283248901,
      "learning_rate": 5.6041436554997595e-06,
      "loss": 0.047025345265865326,
      "step": 5439
    },
    {
      "epoch": 0.7354519260830257,
      "grad_norm": 1.1665226221084595,
      "learning_rate": 5.598809336927278e-06,
      "loss": 0.040745411068201065,
      "step": 5440
    },
    {
      "epoch": 0.7355871194517909,
      "grad_norm": 0.7079965472221375,
      "learning_rate": 5.5934769757328325e-06,
      "loss": 0.0491214394569397,
      "step": 5441
    },
    {
      "epoch": 0.7357223128205561,
      "grad_norm": 1.2897837162017822,
      "learning_rate": 5.588146573026642e-06,
      "loss": 0.06136471778154373,
      "step": 5442
    },
    {
      "epoch": 0.7358575061893214,
      "grad_norm": 1.7740062475204468,
      "learning_rate": 5.582818129918525e-06,
      "loss": 0.05140011012554169,
      "step": 5443
    },
    {
      "epoch": 0.7359926995580867,
      "grad_norm": 1.4889432191848755,
      "learning_rate": 5.5774916475178915e-06,
      "loss": 0.08720873296260834,
      "step": 5444
    },
    {
      "epoch": 0.736127892926852,
      "grad_norm": 0.5965356826782227,
      "learning_rate": 5.572167126933738e-06,
      "loss": 0.04267352819442749,
      "step": 5445
    },
    {
      "epoch": 0.7362630862956172,
      "grad_norm": 0.8960931897163391,
      "learning_rate": 5.566844569274669e-06,
      "loss": 0.054328806698322296,
      "step": 5446
    },
    {
      "epoch": 0.7363982796643824,
      "grad_norm": 0.301082044839859,
      "learning_rate": 5.5615239756488665e-06,
      "loss": 0.039215702563524246,
      "step": 5447
    },
    {
      "epoch": 0.7365334730331478,
      "grad_norm": 0.46671172976493835,
      "learning_rate": 5.556205347164104e-06,
      "loss": 0.047793835401535034,
      "step": 5448
    },
    {
      "epoch": 0.736668666401913,
      "grad_norm": 0.6154031157493591,
      "learning_rate": 5.550888684927746e-06,
      "loss": 0.06589873135089874,
      "step": 5449
    },
    {
      "epoch": 0.7368038597706782,
      "grad_norm": 0.28910574316978455,
      "learning_rate": 5.545573990046752e-06,
      "loss": 0.04198184609413147,
      "step": 5450
    },
    {
      "epoch": 0.7369390531394435,
      "grad_norm": 0.4902563989162445,
      "learning_rate": 5.540261263627672e-06,
      "loss": 0.053771331906318665,
      "step": 5451
    },
    {
      "epoch": 0.7370742465082087,
      "grad_norm": 0.48701468110084534,
      "learning_rate": 5.534950506776644e-06,
      "loss": 0.041878048330545425,
      "step": 5452
    },
    {
      "epoch": 0.7372094398769741,
      "grad_norm": 0.4950914978981018,
      "learning_rate": 5.529641720599393e-06,
      "loss": 0.06427861750125885,
      "step": 5453
    },
    {
      "epoch": 0.7373446332457393,
      "grad_norm": 0.11448574811220169,
      "learning_rate": 5.52433490620124e-06,
      "loss": 0.025246743112802505,
      "step": 5454
    },
    {
      "epoch": 0.7374798266145045,
      "grad_norm": 1.035756230354309,
      "learning_rate": 5.519030064687082e-06,
      "loss": 0.04920538514852524,
      "step": 5455
    },
    {
      "epoch": 0.7376150199832698,
      "grad_norm": 1.4957166910171509,
      "learning_rate": 5.51372719716143e-06,
      "loss": 0.05387274548411369,
      "step": 5456
    },
    {
      "epoch": 0.7377502133520351,
      "grad_norm": 0.6066105365753174,
      "learning_rate": 5.508426304728363e-06,
      "loss": 0.062377989292144775,
      "step": 5457
    },
    {
      "epoch": 0.7378854067208004,
      "grad_norm": 0.6234009861946106,
      "learning_rate": 5.503127388491552e-06,
      "loss": 0.05049778148531914,
      "step": 5458
    },
    {
      "epoch": 0.7380206000895656,
      "grad_norm": 0.6380067467689514,
      "learning_rate": 5.497830449554266e-06,
      "loss": 0.060820288956165314,
      "step": 5459
    },
    {
      "epoch": 0.7381557934583308,
      "grad_norm": 0.7095192670822144,
      "learning_rate": 5.492535489019344e-06,
      "loss": 0.05636731535196304,
      "step": 5460
    },
    {
      "epoch": 0.7382909868270962,
      "grad_norm": 0.4989028573036194,
      "learning_rate": 5.4872425079892454e-06,
      "loss": 0.05664017051458359,
      "step": 5461
    },
    {
      "epoch": 0.7384261801958614,
      "grad_norm": 0.41198912262916565,
      "learning_rate": 5.481951507565973e-06,
      "loss": 0.05192377418279648,
      "step": 5462
    },
    {
      "epoch": 0.7385613735646267,
      "grad_norm": 0.4872404932975769,
      "learning_rate": 5.476662488851159e-06,
      "loss": 0.04623319208621979,
      "step": 5463
    },
    {
      "epoch": 0.7386965669333919,
      "grad_norm": 0.6754307150840759,
      "learning_rate": 5.471375452946e-06,
      "loss": 0.06387482583522797,
      "step": 5464
    },
    {
      "epoch": 0.7388317603021571,
      "grad_norm": 0.4037375748157501,
      "learning_rate": 5.466090400951279e-06,
      "loss": 0.03927309811115265,
      "step": 5465
    },
    {
      "epoch": 0.7389669536709225,
      "grad_norm": 0.2377721667289734,
      "learning_rate": 5.460807333967387e-06,
      "loss": 0.04169538989663124,
      "step": 5466
    },
    {
      "epoch": 0.7391021470396877,
      "grad_norm": 0.34345486760139465,
      "learning_rate": 5.455526253094267e-06,
      "loss": 0.04141313210129738,
      "step": 5467
    },
    {
      "epoch": 0.739237340408453,
      "grad_norm": 0.4850858151912689,
      "learning_rate": 5.450247159431486e-06,
      "loss": 0.05807226896286011,
      "step": 5468
    },
    {
      "epoch": 0.7393725337772182,
      "grad_norm": 0.5401327610015869,
      "learning_rate": 5.44497005407817e-06,
      "loss": 0.05638475716114044,
      "step": 5469
    },
    {
      "epoch": 0.7395077271459835,
      "grad_norm": 0.355792373418808,
      "learning_rate": 5.439694938133042e-06,
      "loss": 0.05783563852310181,
      "step": 5470
    },
    {
      "epoch": 0.7396429205147488,
      "grad_norm": 0.5744714736938477,
      "learning_rate": 5.434421812694409e-06,
      "loss": 0.05246324837207794,
      "step": 5471
    },
    {
      "epoch": 0.739778113883514,
      "grad_norm": 0.45978790521621704,
      "learning_rate": 5.4291506788601624e-06,
      "loss": 0.04424493759870529,
      "step": 5472
    },
    {
      "epoch": 0.7399133072522793,
      "grad_norm": 0.31267258524894714,
      "learning_rate": 5.423881537727785e-06,
      "loss": 0.05124234780669212,
      "step": 5473
    },
    {
      "epoch": 0.7400485006210445,
      "grad_norm": 0.4105251431465149,
      "learning_rate": 5.418614390394338e-06,
      "loss": 0.055374015122652054,
      "step": 5474
    },
    {
      "epoch": 0.7401836939898098,
      "grad_norm": 0.23387347161769867,
      "learning_rate": 5.413349237956469e-06,
      "loss": 0.041248027235269547,
      "step": 5475
    },
    {
      "epoch": 0.7403188873585751,
      "grad_norm": 0.6171667575836182,
      "learning_rate": 5.4080860815104125e-06,
      "loss": 0.05616402626037598,
      "step": 5476
    },
    {
      "epoch": 0.7404540807273403,
      "grad_norm": 0.41180717945098877,
      "learning_rate": 5.402824922151977e-06,
      "loss": 0.043565187603235245,
      "step": 5477
    },
    {
      "epoch": 0.7405892740961055,
      "grad_norm": 1.1337555646896362,
      "learning_rate": 5.397565760976577e-06,
      "loss": 0.052514851093292236,
      "step": 5478
    },
    {
      "epoch": 0.7407244674648709,
      "grad_norm": 0.6680711507797241,
      "learning_rate": 5.392308599079193e-06,
      "loss": 0.04831764101982117,
      "step": 5479
    },
    {
      "epoch": 0.7408596608336361,
      "grad_norm": 0.5328572392463684,
      "learning_rate": 5.3870534375543916e-06,
      "loss": 0.047862708568573,
      "step": 5480
    },
    {
      "epoch": 0.7409948542024014,
      "grad_norm": 0.4246765077114105,
      "learning_rate": 5.381800277496328e-06,
      "loss": 0.05253169685602188,
      "step": 5481
    },
    {
      "epoch": 0.7411300475711666,
      "grad_norm": 0.26996853947639465,
      "learning_rate": 5.376549119998731e-06,
      "loss": 0.046397864818573,
      "step": 5482
    },
    {
      "epoch": 0.7412652409399318,
      "grad_norm": 0.5127127170562744,
      "learning_rate": 5.3712999661549314e-06,
      "loss": 0.04420596361160278,
      "step": 5483
    },
    {
      "epoch": 0.7414004343086972,
      "grad_norm": 1.6158188581466675,
      "learning_rate": 5.366052817057826e-06,
      "loss": 0.0653582438826561,
      "step": 5484
    },
    {
      "epoch": 0.7415356276774624,
      "grad_norm": 0.5706088542938232,
      "learning_rate": 5.360807673799899e-06,
      "loss": 0.06273028999567032,
      "step": 5485
    },
    {
      "epoch": 0.7416708210462277,
      "grad_norm": 1.9467107057571411,
      "learning_rate": 5.355564537473214e-06,
      "loss": 0.06342384219169617,
      "step": 5486
    },
    {
      "epoch": 0.7418060144149929,
      "grad_norm": 0.6369274854660034,
      "learning_rate": 5.35032340916942e-06,
      "loss": 0.04993153735995293,
      "step": 5487
    },
    {
      "epoch": 0.7419412077837582,
      "grad_norm": 0.44693541526794434,
      "learning_rate": 5.345084289979755e-06,
      "loss": 0.061812981963157654,
      "step": 5488
    },
    {
      "epoch": 0.7420764011525235,
      "grad_norm": 0.31033197045326233,
      "learning_rate": 5.339847180995026e-06,
      "loss": 0.05872730910778046,
      "step": 5489
    },
    {
      "epoch": 0.7422115945212887,
      "grad_norm": 0.9868068695068359,
      "learning_rate": 5.33461208330563e-06,
      "loss": 0.05589687451720238,
      "step": 5490
    },
    {
      "epoch": 0.742346787890054,
      "grad_norm": 0.5339412093162537,
      "learning_rate": 5.32937899800154e-06,
      "loss": 0.057909876108169556,
      "step": 5491
    },
    {
      "epoch": 0.7424819812588193,
      "grad_norm": 0.29535794258117676,
      "learning_rate": 5.324147926172307e-06,
      "loss": 0.03965213522315025,
      "step": 5492
    },
    {
      "epoch": 0.7426171746275845,
      "grad_norm": 0.3467995822429657,
      "learning_rate": 5.318918868907084e-06,
      "loss": 0.04146319627761841,
      "step": 5493
    },
    {
      "epoch": 0.7427523679963498,
      "grad_norm": 0.32661059498786926,
      "learning_rate": 5.313691827294568e-06,
      "loss": 0.04507768154144287,
      "step": 5494
    },
    {
      "epoch": 0.742887561365115,
      "grad_norm": 0.2553118169307709,
      "learning_rate": 5.308466802423072e-06,
      "loss": 0.037207067012786865,
      "step": 5495
    },
    {
      "epoch": 0.7430227547338804,
      "grad_norm": 0.2132217288017273,
      "learning_rate": 5.303243795380471e-06,
      "loss": 0.035676877945661545,
      "step": 5496
    },
    {
      "epoch": 0.7431579481026456,
      "grad_norm": 0.5859035849571228,
      "learning_rate": 5.298022807254215e-06,
      "loss": 0.06544259190559387,
      "step": 5497
    },
    {
      "epoch": 0.7432931414714108,
      "grad_norm": 0.547611653804779,
      "learning_rate": 5.292803839131358e-06,
      "loss": 0.04794768989086151,
      "step": 5498
    },
    {
      "epoch": 0.7434283348401761,
      "grad_norm": 0.5438745617866516,
      "learning_rate": 5.287586892098496e-06,
      "loss": 0.04789666831493378,
      "step": 5499
    },
    {
      "epoch": 0.7435635282089413,
      "grad_norm": 0.3389785885810852,
      "learning_rate": 5.282371967241842e-06,
      "loss": 0.04364372044801712,
      "step": 5500
    },
    {
      "epoch": 0.7436987215777067,
      "grad_norm": 1.0312086343765259,
      "learning_rate": 5.277159065647164e-06,
      "loss": 0.057772405445575714,
      "step": 5501
    },
    {
      "epoch": 0.7438339149464719,
      "grad_norm": 0.36524340510368347,
      "learning_rate": 5.271948188399814e-06,
      "loss": 0.06014310568571091,
      "step": 5502
    },
    {
      "epoch": 0.7439691083152371,
      "grad_norm": 0.2583935856819153,
      "learning_rate": 5.266739336584735e-06,
      "loss": 0.04450896382331848,
      "step": 5503
    },
    {
      "epoch": 0.7441043016840024,
      "grad_norm": 0.9652280807495117,
      "learning_rate": 5.261532511286422e-06,
      "loss": 0.047141849994659424,
      "step": 5504
    },
    {
      "epoch": 0.7442394950527677,
      "grad_norm": 0.7472803592681885,
      "learning_rate": 5.256327713588977e-06,
      "loss": 0.049192577600479126,
      "step": 5505
    },
    {
      "epoch": 0.7443746884215329,
      "grad_norm": 0.9693590402603149,
      "learning_rate": 5.25112494457606e-06,
      "loss": 0.04846014082431793,
      "step": 5506
    },
    {
      "epoch": 0.7445098817902982,
      "grad_norm": 0.46338823437690735,
      "learning_rate": 5.245924205330919e-06,
      "loss": 0.052900612354278564,
      "step": 5507
    },
    {
      "epoch": 0.7446450751590634,
      "grad_norm": 0.9405046701431274,
      "learning_rate": 5.240725496936373e-06,
      "loss": 0.055649518966674805,
      "step": 5508
    },
    {
      "epoch": 0.7447802685278287,
      "grad_norm": 0.628115713596344,
      "learning_rate": 5.2355288204748145e-06,
      "loss": 0.05605718493461609,
      "step": 5509
    },
    {
      "epoch": 0.744915461896594,
      "grad_norm": 0.4606287181377411,
      "learning_rate": 5.230334177028233e-06,
      "loss": 0.029851146042346954,
      "step": 5510
    },
    {
      "epoch": 0.7450506552653592,
      "grad_norm": 0.7744247317314148,
      "learning_rate": 5.2251415676781726e-06,
      "loss": 0.04809856414794922,
      "step": 5511
    },
    {
      "epoch": 0.7451858486341245,
      "grad_norm": 0.40339159965515137,
      "learning_rate": 5.2199509935057655e-06,
      "loss": 0.06350082159042358,
      "step": 5512
    },
    {
      "epoch": 0.7453210420028897,
      "grad_norm": 1.1671186685562134,
      "learning_rate": 5.214762455591713e-06,
      "loss": 0.06291911005973816,
      "step": 5513
    },
    {
      "epoch": 0.7454562353716551,
      "grad_norm": 0.5708240866661072,
      "learning_rate": 5.209575955016295e-06,
      "loss": 0.059872016310691833,
      "step": 5514
    },
    {
      "epoch": 0.7455914287404203,
      "grad_norm": 0.6044837832450867,
      "learning_rate": 5.204391492859377e-06,
      "loss": 0.037675123661756516,
      "step": 5515
    },
    {
      "epoch": 0.7457266221091855,
      "grad_norm": 0.3899818956851959,
      "learning_rate": 5.199209070200388e-06,
      "loss": 0.04870859906077385,
      "step": 5516
    },
    {
      "epoch": 0.7458618154779508,
      "grad_norm": 0.4968286454677582,
      "learning_rate": 5.194028688118332e-06,
      "loss": 0.04478497803211212,
      "step": 5517
    },
    {
      "epoch": 0.745997008846716,
      "grad_norm": 0.38747715950012207,
      "learning_rate": 5.188850347691797e-06,
      "loss": 0.04321112111210823,
      "step": 5518
    },
    {
      "epoch": 0.7461322022154814,
      "grad_norm": 0.4650703966617584,
      "learning_rate": 5.183674049998934e-06,
      "loss": 0.061393894255161285,
      "step": 5519
    },
    {
      "epoch": 0.7462673955842466,
      "grad_norm": 1.437670111656189,
      "learning_rate": 5.178499796117485e-06,
      "loss": 0.05953851342201233,
      "step": 5520
    },
    {
      "epoch": 0.7464025889530118,
      "grad_norm": 0.7268595695495605,
      "learning_rate": 5.173327587124753e-06,
      "loss": 0.06345154345035553,
      "step": 5521
    },
    {
      "epoch": 0.7465377823217771,
      "grad_norm": 0.6755163669586182,
      "learning_rate": 5.16815742409762e-06,
      "loss": 0.07344269752502441,
      "step": 5522
    },
    {
      "epoch": 0.7466729756905424,
      "grad_norm": 0.2622813880443573,
      "learning_rate": 5.16298930811254e-06,
      "loss": 0.03183408081531525,
      "step": 5523
    },
    {
      "epoch": 0.7468081690593077,
      "grad_norm": 0.4086357355117798,
      "learning_rate": 5.15782324024554e-06,
      "loss": 0.047601908445358276,
      "step": 5524
    },
    {
      "epoch": 0.7469433624280729,
      "grad_norm": 0.5096064209938049,
      "learning_rate": 5.152659221572231e-06,
      "loss": 0.04264703765511513,
      "step": 5525
    },
    {
      "epoch": 0.7470785557968381,
      "grad_norm": 0.3501160144805908,
      "learning_rate": 5.147497253167784e-06,
      "loss": 0.049505382776260376,
      "step": 5526
    },
    {
      "epoch": 0.7472137491656035,
      "grad_norm": 0.29180678725242615,
      "learning_rate": 5.142337336106948e-06,
      "loss": 0.04939737915992737,
      "step": 5527
    },
    {
      "epoch": 0.7473489425343687,
      "grad_norm": 0.3182583451271057,
      "learning_rate": 5.137179471464047e-06,
      "loss": 0.03303126245737076,
      "step": 5528
    },
    {
      "epoch": 0.747484135903134,
      "grad_norm": 0.4041508436203003,
      "learning_rate": 5.13202366031297e-06,
      "loss": 0.06243613362312317,
      "step": 5529
    },
    {
      "epoch": 0.7476193292718992,
      "grad_norm": 0.4116409718990326,
      "learning_rate": 5.1268699037272e-06,
      "loss": 0.03966081142425537,
      "step": 5530
    },
    {
      "epoch": 0.7477545226406644,
      "grad_norm": 0.8240025639533997,
      "learning_rate": 5.121718202779756e-06,
      "loss": 0.06657412648200989,
      "step": 5531
    },
    {
      "epoch": 0.7478897160094298,
      "grad_norm": 0.46376833319664,
      "learning_rate": 5.116568558543264e-06,
      "loss": 0.05384666845202446,
      "step": 5532
    },
    {
      "epoch": 0.748024909378195,
      "grad_norm": 0.8592454791069031,
      "learning_rate": 5.1114209720899025e-06,
      "loss": 0.03947149217128754,
      "step": 5533
    },
    {
      "epoch": 0.7481601027469602,
      "grad_norm": 0.7284401059150696,
      "learning_rate": 5.106275444491423e-06,
      "loss": 0.03800439089536667,
      "step": 5534
    },
    {
      "epoch": 0.7482952961157255,
      "grad_norm": 0.43126651644706726,
      "learning_rate": 5.101131976819165e-06,
      "loss": 0.04651375487446785,
      "step": 5535
    },
    {
      "epoch": 0.7484304894844908,
      "grad_norm": 0.6406553983688354,
      "learning_rate": 5.095990570144008e-06,
      "loss": 0.05920161306858063,
      "step": 5536
    },
    {
      "epoch": 0.7485656828532561,
      "grad_norm": 0.7216412425041199,
      "learning_rate": 5.090851225536432e-06,
      "loss": 0.042914777994155884,
      "step": 5537
    },
    {
      "epoch": 0.7487008762220213,
      "grad_norm": 0.3997083604335785,
      "learning_rate": 5.085713944066474e-06,
      "loss": 0.049191270023584366,
      "step": 5538
    },
    {
      "epoch": 0.7488360695907865,
      "grad_norm": 0.6152087450027466,
      "learning_rate": 5.080578726803741e-06,
      "loss": 0.05972042679786682,
      "step": 5539
    },
    {
      "epoch": 0.7489712629595519,
      "grad_norm": 0.7942944765090942,
      "learning_rate": 5.075445574817415e-06,
      "loss": 0.05272885784506798,
      "step": 5540
    },
    {
      "epoch": 0.7491064563283171,
      "grad_norm": 0.42381125688552856,
      "learning_rate": 5.07031448917624e-06,
      "loss": 0.040979981422424316,
      "step": 5541
    },
    {
      "epoch": 0.7492416496970824,
      "grad_norm": 0.45558756589889526,
      "learning_rate": 5.065185470948544e-06,
      "loss": 0.054111696779727936,
      "step": 5542
    },
    {
      "epoch": 0.7493768430658476,
      "grad_norm": 0.5484164357185364,
      "learning_rate": 5.060058521202211e-06,
      "loss": 0.05012672394514084,
      "step": 5543
    },
    {
      "epoch": 0.7495120364346128,
      "grad_norm": 0.6709420084953308,
      "learning_rate": 5.054933641004703e-06,
      "loss": 0.05820755660533905,
      "step": 5544
    },
    {
      "epoch": 0.7496472298033782,
      "grad_norm": 0.363085001707077,
      "learning_rate": 5.0498108314230425e-06,
      "loss": 0.04157222807407379,
      "step": 5545
    },
    {
      "epoch": 0.7497824231721434,
      "grad_norm": 0.37748420238494873,
      "learning_rate": 5.044690093523823e-06,
      "loss": 0.039329893887043,
      "step": 5546
    },
    {
      "epoch": 0.7499176165409087,
      "grad_norm": 0.7921310663223267,
      "learning_rate": 5.039571428373219e-06,
      "loss": 0.05081334710121155,
      "step": 5547
    },
    {
      "epoch": 0.7500528099096739,
      "grad_norm": 0.5201166868209839,
      "learning_rate": 5.034454837036959e-06,
      "loss": 0.054199472069740295,
      "step": 5548
    },
    {
      "epoch": 0.7501880032784392,
      "grad_norm": 0.47418633103370667,
      "learning_rate": 5.0293403205803455e-06,
      "loss": 0.033734939992427826,
      "step": 5549
    },
    {
      "epoch": 0.7503231966472045,
      "grad_norm": 0.4220098853111267,
      "learning_rate": 5.024227880068247e-06,
      "loss": 0.06173555552959442,
      "step": 5550
    },
    {
      "epoch": 0.7504583900159697,
      "grad_norm": 0.4469703435897827,
      "learning_rate": 5.019117516565096e-06,
      "loss": 0.04640002176165581,
      "step": 5551
    },
    {
      "epoch": 0.750593583384735,
      "grad_norm": 0.7153790593147278,
      "learning_rate": 5.014009231134908e-06,
      "loss": 0.07060258090496063,
      "step": 5552
    },
    {
      "epoch": 0.7507287767535002,
      "grad_norm": 1.1831549406051636,
      "learning_rate": 5.008903024841248e-06,
      "loss": 0.053222209215164185,
      "step": 5553
    },
    {
      "epoch": 0.7508639701222655,
      "grad_norm": 0.7888247966766357,
      "learning_rate": 5.0037988987472595e-06,
      "loss": 0.05328439176082611,
      "step": 5554
    },
    {
      "epoch": 0.7509991634910308,
      "grad_norm": 0.5309898257255554,
      "learning_rate": 4.998696853915646e-06,
      "loss": 0.04646754264831543,
      "step": 5555
    },
    {
      "epoch": 0.751134356859796,
      "grad_norm": 0.5059635043144226,
      "learning_rate": 4.993596891408676e-06,
      "loss": 0.043650705367326736,
      "step": 5556
    },
    {
      "epoch": 0.7512695502285612,
      "grad_norm": 0.4907800853252411,
      "learning_rate": 4.988499012288198e-06,
      "loss": 0.057316869497299194,
      "step": 5557
    },
    {
      "epoch": 0.7514047435973266,
      "grad_norm": 0.9628327488899231,
      "learning_rate": 4.983403217615614e-06,
      "loss": 0.045806288719177246,
      "step": 5558
    },
    {
      "epoch": 0.7515399369660918,
      "grad_norm": 0.989477813243866,
      "learning_rate": 4.978309508451896e-06,
      "loss": 0.050319552421569824,
      "step": 5559
    },
    {
      "epoch": 0.7516751303348571,
      "grad_norm": 0.34060031175613403,
      "learning_rate": 4.973217885857578e-06,
      "loss": 0.03778659552335739,
      "step": 5560
    },
    {
      "epoch": 0.7518103237036223,
      "grad_norm": 0.2964553236961365,
      "learning_rate": 4.968128350892763e-06,
      "loss": 0.06380490958690643,
      "step": 5561
    },
    {
      "epoch": 0.7519455170723875,
      "grad_norm": 0.7513662576675415,
      "learning_rate": 4.963040904617131e-06,
      "loss": 0.05820570886135101,
      "step": 5562
    },
    {
      "epoch": 0.7520807104411529,
      "grad_norm": 0.5941129922866821,
      "learning_rate": 4.9579555480898955e-06,
      "loss": 0.04170685261487961,
      "step": 5563
    },
    {
      "epoch": 0.7522159038099181,
      "grad_norm": 0.333189994096756,
      "learning_rate": 4.952872282369873e-06,
      "loss": 0.05616297572851181,
      "step": 5564
    },
    {
      "epoch": 0.7523510971786834,
      "grad_norm": 0.3728758990764618,
      "learning_rate": 4.947791108515417e-06,
      "loss": 0.054476816207170486,
      "step": 5565
    },
    {
      "epoch": 0.7524862905474486,
      "grad_norm": 0.3514253497123718,
      "learning_rate": 4.942712027584453e-06,
      "loss": 0.04808138310909271,
      "step": 5566
    },
    {
      "epoch": 0.7526214839162139,
      "grad_norm": 1.2773067951202393,
      "learning_rate": 4.937635040634485e-06,
      "loss": 0.061209291219711304,
      "step": 5567
    },
    {
      "epoch": 0.7527566772849792,
      "grad_norm": 1.3021620512008667,
      "learning_rate": 4.9325601487225545e-06,
      "loss": 0.07051263004541397,
      "step": 5568
    },
    {
      "epoch": 0.7528918706537444,
      "grad_norm": 0.6259169578552246,
      "learning_rate": 4.927487352905289e-06,
      "loss": 0.06368683278560638,
      "step": 5569
    },
    {
      "epoch": 0.7530270640225097,
      "grad_norm": 0.5547968745231628,
      "learning_rate": 4.92241665423887e-06,
      "loss": 0.05438164249062538,
      "step": 5570
    },
    {
      "epoch": 0.753162257391275,
      "grad_norm": 0.5523261427879333,
      "learning_rate": 4.917348053779039e-06,
      "loss": 0.05029526352882385,
      "step": 5571
    },
    {
      "epoch": 0.7532974507600402,
      "grad_norm": 0.7147053480148315,
      "learning_rate": 4.912281552581122e-06,
      "loss": 0.03933337330818176,
      "step": 5572
    },
    {
      "epoch": 0.7534326441288055,
      "grad_norm": 0.5926932692527771,
      "learning_rate": 4.907217151699969e-06,
      "loss": 0.03968716412782669,
      "step": 5573
    },
    {
      "epoch": 0.7535678374975707,
      "grad_norm": 0.5737786889076233,
      "learning_rate": 4.9021548521900305e-06,
      "loss": 0.055317461490631104,
      "step": 5574
    },
    {
      "epoch": 0.753703030866336,
      "grad_norm": 0.508925199508667,
      "learning_rate": 4.8970946551053005e-06,
      "loss": 0.050451233983039856,
      "step": 5575
    },
    {
      "epoch": 0.7538382242351013,
      "grad_norm": 0.34880030155181885,
      "learning_rate": 4.892036561499339e-06,
      "loss": 0.04335852712392807,
      "step": 5576
    },
    {
      "epoch": 0.7539734176038665,
      "grad_norm": 0.374421089887619,
      "learning_rate": 4.8869805724252675e-06,
      "loss": 0.033583879470825195,
      "step": 5577
    },
    {
      "epoch": 0.7541086109726318,
      "grad_norm": 0.288886159658432,
      "learning_rate": 4.8819266889357665e-06,
      "loss": 0.0498289130628109,
      "step": 5578
    },
    {
      "epoch": 0.754243804341397,
      "grad_norm": 0.697701632976532,
      "learning_rate": 4.876874912083088e-06,
      "loss": 0.05230463296175003,
      "step": 5579
    },
    {
      "epoch": 0.7543789977101624,
      "grad_norm": 1.3653676509857178,
      "learning_rate": 4.871825242919037e-06,
      "loss": 0.05933082103729248,
      "step": 5580
    },
    {
      "epoch": 0.7545141910789276,
      "grad_norm": 0.3522177040576935,
      "learning_rate": 4.866777682494978e-06,
      "loss": 0.06317342817783356,
      "step": 5581
    },
    {
      "epoch": 0.7546493844476928,
      "grad_norm": 0.3955438733100891,
      "learning_rate": 4.861732231861845e-06,
      "loss": 0.044080331921577454,
      "step": 5582
    },
    {
      "epoch": 0.7547845778164581,
      "grad_norm": 0.46337470412254333,
      "learning_rate": 4.85668889207012e-06,
      "loss": 0.045066311955451965,
      "step": 5583
    },
    {
      "epoch": 0.7549197711852234,
      "grad_norm": 0.60178142786026,
      "learning_rate": 4.851647664169862e-06,
      "loss": 0.04821908473968506,
      "step": 5584
    },
    {
      "epoch": 0.7550549645539886,
      "grad_norm": 0.3464260399341583,
      "learning_rate": 4.846608549210679e-06,
      "loss": 0.05057625100016594,
      "step": 5585
    },
    {
      "epoch": 0.7551901579227539,
      "grad_norm": 0.5496123433113098,
      "learning_rate": 4.841571548241741e-06,
      "loss": 0.052937984466552734,
      "step": 5586
    },
    {
      "epoch": 0.7553253512915191,
      "grad_norm": 0.5719283819198608,
      "learning_rate": 4.836536662311777e-06,
      "loss": 0.045376766473054886,
      "step": 5587
    },
    {
      "epoch": 0.7554605446602844,
      "grad_norm": 0.30261653661727905,
      "learning_rate": 4.8315038924690745e-06,
      "loss": 0.04904872179031372,
      "step": 5588
    },
    {
      "epoch": 0.7555957380290497,
      "grad_norm": 0.5772355198860168,
      "learning_rate": 4.82647323976149e-06,
      "loss": 0.059018149971961975,
      "step": 5589
    },
    {
      "epoch": 0.7557309313978149,
      "grad_norm": 0.37943848967552185,
      "learning_rate": 4.821444705236429e-06,
      "loss": 0.041968148201704025,
      "step": 5590
    },
    {
      "epoch": 0.7558661247665802,
      "grad_norm": 0.44476258754730225,
      "learning_rate": 4.81641828994086e-06,
      "loss": 0.05268823355436325,
      "step": 5591
    },
    {
      "epoch": 0.7560013181353454,
      "grad_norm": 0.3439156413078308,
      "learning_rate": 4.811393994921308e-06,
      "loss": 0.047693632543087006,
      "step": 5592
    },
    {
      "epoch": 0.7561365115041108,
      "grad_norm": 0.28831806778907776,
      "learning_rate": 4.806371821223854e-06,
      "loss": 0.05170748382806778,
      "step": 5593
    },
    {
      "epoch": 0.756271704872876,
      "grad_norm": 0.9020028114318848,
      "learning_rate": 4.801351769894151e-06,
      "loss": 0.043787021189928055,
      "step": 5594
    },
    {
      "epoch": 0.7564068982416412,
      "grad_norm": 0.8622465133666992,
      "learning_rate": 4.796333841977394e-06,
      "loss": 0.06588828563690186,
      "step": 5595
    },
    {
      "epoch": 0.7565420916104065,
      "grad_norm": 0.5422247648239136,
      "learning_rate": 4.791318038518345e-06,
      "loss": 0.04037340357899666,
      "step": 5596
    },
    {
      "epoch": 0.7566772849791717,
      "grad_norm": 1.0936359167099,
      "learning_rate": 4.7863043605613185e-06,
      "loss": 0.058387063443660736,
      "step": 5597
    },
    {
      "epoch": 0.7568124783479371,
      "grad_norm": 0.23124346137046814,
      "learning_rate": 4.7812928091501865e-06,
      "loss": 0.04374666512012482,
      "step": 5598
    },
    {
      "epoch": 0.7569476717167023,
      "grad_norm": 0.36500871181488037,
      "learning_rate": 4.7762833853283935e-06,
      "loss": 0.0665564090013504,
      "step": 5599
    },
    {
      "epoch": 0.7570828650854675,
      "grad_norm": 0.914710283279419,
      "learning_rate": 4.77127609013891e-06,
      "loss": 0.05417275428771973,
      "step": 5600
    },
    {
      "epoch": 0.7572180584542328,
      "grad_norm": 0.4734802544116974,
      "learning_rate": 4.766270924624295e-06,
      "loss": 0.058981843292713165,
      "step": 5601
    },
    {
      "epoch": 0.7573532518229981,
      "grad_norm": 0.8628573417663574,
      "learning_rate": 4.761267889826647e-06,
      "loss": 0.06330180168151855,
      "step": 5602
    },
    {
      "epoch": 0.7574884451917634,
      "grad_norm": 0.7319051027297974,
      "learning_rate": 4.756266986787619e-06,
      "loss": 0.0619068518280983,
      "step": 5603
    },
    {
      "epoch": 0.7576236385605286,
      "grad_norm": 0.6942102909088135,
      "learning_rate": 4.751268216548439e-06,
      "loss": 0.05534280836582184,
      "step": 5604
    },
    {
      "epoch": 0.7577588319292938,
      "grad_norm": 0.1495506763458252,
      "learning_rate": 4.746271580149861e-06,
      "loss": 0.029495157301425934,
      "step": 5605
    },
    {
      "epoch": 0.7578940252980592,
      "grad_norm": 0.6326749920845032,
      "learning_rate": 4.7412770786322244e-06,
      "loss": 0.05507656931877136,
      "step": 5606
    },
    {
      "epoch": 0.7580292186668244,
      "grad_norm": 0.36095789074897766,
      "learning_rate": 4.736284713035406e-06,
      "loss": 0.042208775877952576,
      "step": 5607
    },
    {
      "epoch": 0.7581644120355897,
      "grad_norm": 1.6832358837127686,
      "learning_rate": 4.731294484398843e-06,
      "loss": 0.06477434933185577,
      "step": 5608
    },
    {
      "epoch": 0.7582996054043549,
      "grad_norm": 0.6530440449714661,
      "learning_rate": 4.726306393761526e-06,
      "loss": 0.05780060589313507,
      "step": 5609
    },
    {
      "epoch": 0.7584347987731201,
      "grad_norm": 0.502464234828949,
      "learning_rate": 4.721320442162001e-06,
      "loss": 0.05332181602716446,
      "step": 5610
    },
    {
      "epoch": 0.7585699921418855,
      "grad_norm": 0.44271019101142883,
      "learning_rate": 4.716336630638378e-06,
      "loss": 0.0516282320022583,
      "step": 5611
    },
    {
      "epoch": 0.7587051855106507,
      "grad_norm": 0.6140710711479187,
      "learning_rate": 4.711354960228306e-06,
      "loss": 0.05469930171966553,
      "step": 5612
    },
    {
      "epoch": 0.7588403788794159,
      "grad_norm": 0.4506165087223053,
      "learning_rate": 4.706375431968998e-06,
      "loss": 0.042360689491033554,
      "step": 5613
    },
    {
      "epoch": 0.7589755722481812,
      "grad_norm": 1.0104496479034424,
      "learning_rate": 4.701398046897218e-06,
      "loss": 0.05663909390568733,
      "step": 5614
    },
    {
      "epoch": 0.7591107656169465,
      "grad_norm": 0.4034721851348877,
      "learning_rate": 4.696422806049277e-06,
      "loss": 0.04604434594511986,
      "step": 5615
    },
    {
      "epoch": 0.7592459589857118,
      "grad_norm": 0.5353710651397705,
      "learning_rate": 4.69144971046106e-06,
      "loss": 0.0535174161195755,
      "step": 5616
    },
    {
      "epoch": 0.759381152354477,
      "grad_norm": 0.552902340888977,
      "learning_rate": 4.686478761167984e-06,
      "loss": 0.04603348672389984,
      "step": 5617
    },
    {
      "epoch": 0.7595163457232422,
      "grad_norm": 0.3774576187133789,
      "learning_rate": 4.681509959205028e-06,
      "loss": 0.060617804527282715,
      "step": 5618
    },
    {
      "epoch": 0.7596515390920076,
      "grad_norm": 0.526112973690033,
      "learning_rate": 4.676543305606724e-06,
      "loss": 0.03966321796178818,
      "step": 5619
    },
    {
      "epoch": 0.7597867324607728,
      "grad_norm": 0.32452404499053955,
      "learning_rate": 4.67157880140715e-06,
      "loss": 0.03446151316165924,
      "step": 5620
    },
    {
      "epoch": 0.7599219258295381,
      "grad_norm": 0.8323810696601868,
      "learning_rate": 4.666616447639952e-06,
      "loss": 0.054547131061553955,
      "step": 5621
    },
    {
      "epoch": 0.7600571191983033,
      "grad_norm": 0.9185550212860107,
      "learning_rate": 4.661656245338314e-06,
      "loss": 0.07025061547756195,
      "step": 5622
    },
    {
      "epoch": 0.7601923125670685,
      "grad_norm": 0.5083182454109192,
      "learning_rate": 4.656698195534978e-06,
      "loss": 0.05385303497314453,
      "step": 5623
    },
    {
      "epoch": 0.7603275059358339,
      "grad_norm": 0.2188534438610077,
      "learning_rate": 4.651742299262233e-06,
      "loss": 0.03369933366775513,
      "step": 5624
    },
    {
      "epoch": 0.7604626993045991,
      "grad_norm": 0.4826721251010895,
      "learning_rate": 4.646788557551921e-06,
      "loss": 0.06902796775102615,
      "step": 5625
    },
    {
      "epoch": 0.7605978926733644,
      "grad_norm": 0.5610843300819397,
      "learning_rate": 4.641836971435445e-06,
      "loss": 0.05385258048772812,
      "step": 5626
    },
    {
      "epoch": 0.7607330860421296,
      "grad_norm": 0.3415719270706177,
      "learning_rate": 4.63688754194375e-06,
      "loss": 0.05460992455482483,
      "step": 5627
    },
    {
      "epoch": 0.7608682794108949,
      "grad_norm": 0.5464599132537842,
      "learning_rate": 4.6319402701073295e-06,
      "loss": 0.045591652393341064,
      "step": 5628
    },
    {
      "epoch": 0.7610034727796602,
      "grad_norm": 0.4167063534259796,
      "learning_rate": 4.6269951569562355e-06,
      "loss": 0.0458734929561615,
      "step": 5629
    },
    {
      "epoch": 0.7611386661484254,
      "grad_norm": 0.7480295300483704,
      "learning_rate": 4.622052203520061e-06,
      "loss": 0.04887177050113678,
      "step": 5630
    },
    {
      "epoch": 0.7612738595171907,
      "grad_norm": 0.7025947570800781,
      "learning_rate": 4.617111410827968e-06,
      "loss": 0.06044713407754898,
      "step": 5631
    },
    {
      "epoch": 0.761409052885956,
      "grad_norm": 0.4838401675224304,
      "learning_rate": 4.612172779908639e-06,
      "loss": 0.06201377511024475,
      "step": 5632
    },
    {
      "epoch": 0.7615442462547212,
      "grad_norm": 0.3738311529159546,
      "learning_rate": 4.607236311790335e-06,
      "loss": 0.03970567137002945,
      "step": 5633
    },
    {
      "epoch": 0.7616794396234865,
      "grad_norm": 0.9668474793434143,
      "learning_rate": 4.602302007500854e-06,
      "loss": 0.03886076807975769,
      "step": 5634
    },
    {
      "epoch": 0.7618146329922517,
      "grad_norm": 0.4892688989639282,
      "learning_rate": 4.597369868067537e-06,
      "loss": 0.0554858073592186,
      "step": 5635
    },
    {
      "epoch": 0.761949826361017,
      "grad_norm": 0.4261897802352905,
      "learning_rate": 4.592439894517296e-06,
      "loss": 0.059930577874183655,
      "step": 5636
    },
    {
      "epoch": 0.7620850197297823,
      "grad_norm": 0.8358299732208252,
      "learning_rate": 4.587512087876559e-06,
      "loss": 0.05172547698020935,
      "step": 5637
    },
    {
      "epoch": 0.7622202130985475,
      "grad_norm": 0.2497706413269043,
      "learning_rate": 4.582586449171336e-06,
      "loss": 0.041028089821338654,
      "step": 5638
    },
    {
      "epoch": 0.7623554064673128,
      "grad_norm": 0.7382331490516663,
      "learning_rate": 4.577662979427168e-06,
      "loss": 0.07519373297691345,
      "step": 5639
    },
    {
      "epoch": 0.762490599836078,
      "grad_norm": 0.625405490398407,
      "learning_rate": 4.572741679669147e-06,
      "loss": 0.05997096374630928,
      "step": 5640
    },
    {
      "epoch": 0.7626257932048432,
      "grad_norm": 0.4168659746646881,
      "learning_rate": 4.567822550921912e-06,
      "loss": 0.042681142687797546,
      "step": 5641
    },
    {
      "epoch": 0.7627609865736086,
      "grad_norm": 0.9903636574745178,
      "learning_rate": 4.562905594209647e-06,
      "loss": 0.054901592433452606,
      "step": 5642
    },
    {
      "epoch": 0.7628961799423738,
      "grad_norm": 0.602508008480072,
      "learning_rate": 4.557990810556102e-06,
      "loss": 0.06749507784843445,
      "step": 5643
    },
    {
      "epoch": 0.7630313733111391,
      "grad_norm": 0.7250024080276489,
      "learning_rate": 4.553078200984553e-06,
      "loss": 0.07625281065702438,
      "step": 5644
    },
    {
      "epoch": 0.7631665666799043,
      "grad_norm": 0.3466350734233856,
      "learning_rate": 4.548167766517832e-06,
      "loss": 0.027551140636205673,
      "step": 5645
    },
    {
      "epoch": 0.7633017600486696,
      "grad_norm": 0.46788859367370605,
      "learning_rate": 4.543259508178318e-06,
      "loss": 0.04973139241337776,
      "step": 5646
    },
    {
      "epoch": 0.7634369534174349,
      "grad_norm": 0.3555099666118622,
      "learning_rate": 4.538353426987931e-06,
      "loss": 0.04602910578250885,
      "step": 5647
    },
    {
      "epoch": 0.7635721467862001,
      "grad_norm": 0.5704427361488342,
      "learning_rate": 4.533449523968154e-06,
      "loss": 0.051924578845500946,
      "step": 5648
    },
    {
      "epoch": 0.7637073401549654,
      "grad_norm": 0.814926266670227,
      "learning_rate": 4.528547800140001e-06,
      "loss": 0.07376405596733093,
      "step": 5649
    },
    {
      "epoch": 0.7638425335237307,
      "grad_norm": 0.7312045693397522,
      "learning_rate": 4.523648256524037e-06,
      "loss": 0.047536760568618774,
      "step": 5650
    },
    {
      "epoch": 0.7639777268924959,
      "grad_norm": 0.6229032278060913,
      "learning_rate": 4.518750894140372e-06,
      "loss": 0.05637187883257866,
      "step": 5651
    },
    {
      "epoch": 0.7641129202612612,
      "grad_norm": 1.309388279914856,
      "learning_rate": 4.513855714008659e-06,
      "loss": 0.04895530641078949,
      "step": 5652
    },
    {
      "epoch": 0.7642481136300264,
      "grad_norm": 0.5115934014320374,
      "learning_rate": 4.508962717148111e-06,
      "loss": 0.04286891967058182,
      "step": 5653
    },
    {
      "epoch": 0.7643833069987918,
      "grad_norm": 0.5343486666679382,
      "learning_rate": 4.504071904577469e-06,
      "loss": 0.045203179121017456,
      "step": 5654
    },
    {
      "epoch": 0.764518500367557,
      "grad_norm": 0.6865848898887634,
      "learning_rate": 4.499183277315027e-06,
      "loss": 0.05576498061418533,
      "step": 5655
    },
    {
      "epoch": 0.7646536937363222,
      "grad_norm": 0.5834632515907288,
      "learning_rate": 4.494296836378625e-06,
      "loss": 0.04854917898774147,
      "step": 5656
    },
    {
      "epoch": 0.7647888871050875,
      "grad_norm": 0.44065696001052856,
      "learning_rate": 4.4894125827856415e-06,
      "loss": 0.049645356833934784,
      "step": 5657
    },
    {
      "epoch": 0.7649240804738527,
      "grad_norm": 0.3580378592014313,
      "learning_rate": 4.4845305175530105e-06,
      "loss": 0.03369799628853798,
      "step": 5658
    },
    {
      "epoch": 0.7650592738426181,
      "grad_norm": 0.5414225459098816,
      "learning_rate": 4.479650641697201e-06,
      "loss": 0.0440206453204155,
      "step": 5659
    },
    {
      "epoch": 0.7651944672113833,
      "grad_norm": 0.8068389892578125,
      "learning_rate": 4.4747729562342305e-06,
      "loss": 0.0768054723739624,
      "step": 5660
    },
    {
      "epoch": 0.7653296605801485,
      "grad_norm": 0.508140504360199,
      "learning_rate": 4.469897462179656e-06,
      "loss": 0.06765955686569214,
      "step": 5661
    },
    {
      "epoch": 0.7654648539489138,
      "grad_norm": 0.71355140209198,
      "learning_rate": 4.46502416054858e-06,
      "loss": 0.05604205280542374,
      "step": 5662
    },
    {
      "epoch": 0.765600047317679,
      "grad_norm": 0.8222368955612183,
      "learning_rate": 4.460153052355663e-06,
      "loss": 0.05523822084069252,
      "step": 5663
    },
    {
      "epoch": 0.7657352406864444,
      "grad_norm": 0.8126984238624573,
      "learning_rate": 4.455284138615074e-06,
      "loss": 0.04764276742935181,
      "step": 5664
    },
    {
      "epoch": 0.7658704340552096,
      "grad_norm": 0.6107057332992554,
      "learning_rate": 4.4504174203405656e-06,
      "loss": 0.035220615565776825,
      "step": 5665
    },
    {
      "epoch": 0.7660056274239748,
      "grad_norm": 0.8327341675758362,
      "learning_rate": 4.445552898545407e-06,
      "loss": 0.05195525288581848,
      "step": 5666
    },
    {
      "epoch": 0.7661408207927402,
      "grad_norm": 0.48244842886924744,
      "learning_rate": 4.440690574242413e-06,
      "loss": 0.053544119000434875,
      "step": 5667
    },
    {
      "epoch": 0.7662760141615054,
      "grad_norm": 0.3419904410839081,
      "learning_rate": 4.435830448443961e-06,
      "loss": 0.0494111031293869,
      "step": 5668
    },
    {
      "epoch": 0.7664112075302706,
      "grad_norm": 0.43288305401802063,
      "learning_rate": 4.430972522161934e-06,
      "loss": 0.049941033124923706,
      "step": 5669
    },
    {
      "epoch": 0.7665464008990359,
      "grad_norm": 0.3529101610183716,
      "learning_rate": 4.426116796407794e-06,
      "loss": 0.0279175266623497,
      "step": 5670
    },
    {
      "epoch": 0.7666815942678011,
      "grad_norm": 0.26189494132995605,
      "learning_rate": 4.421263272192523e-06,
      "loss": 0.0417625866830349,
      "step": 5671
    },
    {
      "epoch": 0.7668167876365665,
      "grad_norm": 0.539402425289154,
      "learning_rate": 4.416411950526648e-06,
      "loss": 0.05093270540237427,
      "step": 5672
    },
    {
      "epoch": 0.7669519810053317,
      "grad_norm": 0.8766064643859863,
      "learning_rate": 4.411562832420252e-06,
      "loss": 0.04850289598107338,
      "step": 5673
    },
    {
      "epoch": 0.7670871743740969,
      "grad_norm": 0.3340434730052948,
      "learning_rate": 4.406715918882929e-06,
      "loss": 0.043039269745349884,
      "step": 5674
    },
    {
      "epoch": 0.7672223677428622,
      "grad_norm": 0.49644145369529724,
      "learning_rate": 4.4018712109238475e-06,
      "loss": 0.052819907665252686,
      "step": 5675
    },
    {
      "epoch": 0.7673575611116275,
      "grad_norm": 0.3417988717556,
      "learning_rate": 4.3970287095516965e-06,
      "loss": 0.0382411926984787,
      "step": 5676
    },
    {
      "epoch": 0.7674927544803928,
      "grad_norm": 0.7697397470474243,
      "learning_rate": 4.39218841577471e-06,
      "loss": 0.060521915555000305,
      "step": 5677
    },
    {
      "epoch": 0.767627947849158,
      "grad_norm": 0.7141456604003906,
      "learning_rate": 4.387350330600662e-06,
      "loss": 0.048661477863788605,
      "step": 5678
    },
    {
      "epoch": 0.7677631412179232,
      "grad_norm": 0.9669989347457886,
      "learning_rate": 4.382514455036864e-06,
      "loss": 0.04918695241212845,
      "step": 5679
    },
    {
      "epoch": 0.7678983345866885,
      "grad_norm": 0.5380557775497437,
      "learning_rate": 4.377680790090182e-06,
      "loss": 0.051476217806339264,
      "step": 5680
    },
    {
      "epoch": 0.7680335279554538,
      "grad_norm": 0.5008072853088379,
      "learning_rate": 4.372849336767004e-06,
      "loss": 0.044195614755153656,
      "step": 5681
    },
    {
      "epoch": 0.7681687213242191,
      "grad_norm": 0.5809420943260193,
      "learning_rate": 4.3680200960732645e-06,
      "loss": 0.051470693200826645,
      "step": 5682
    },
    {
      "epoch": 0.7683039146929843,
      "grad_norm": 0.4443592429161072,
      "learning_rate": 4.363193069014439e-06,
      "loss": 0.05219653248786926,
      "step": 5683
    },
    {
      "epoch": 0.7684391080617495,
      "grad_norm": 1.040880799293518,
      "learning_rate": 4.3583682565955325e-06,
      "loss": 0.04851394146680832,
      "step": 5684
    },
    {
      "epoch": 0.7685743014305149,
      "grad_norm": 0.7633553743362427,
      "learning_rate": 4.3535456598211074e-06,
      "loss": 0.059917330741882324,
      "step": 5685
    },
    {
      "epoch": 0.7687094947992801,
      "grad_norm": 0.3302479684352875,
      "learning_rate": 4.348725279695251e-06,
      "loss": 0.034069906920194626,
      "step": 5686
    },
    {
      "epoch": 0.7688446881680454,
      "grad_norm": 0.4986826479434967,
      "learning_rate": 4.343907117221591e-06,
      "loss": 0.04978461563587189,
      "step": 5687
    },
    {
      "epoch": 0.7689798815368106,
      "grad_norm": 0.7870305776596069,
      "learning_rate": 4.339091173403294e-06,
      "loss": 0.07076722383499146,
      "step": 5688
    },
    {
      "epoch": 0.7691150749055758,
      "grad_norm": 0.37520065903663635,
      "learning_rate": 4.334277449243061e-06,
      "loss": 0.0574164092540741,
      "step": 5689
    },
    {
      "epoch": 0.7692502682743412,
      "grad_norm": 1.180557131767273,
      "learning_rate": 4.329465945743144e-06,
      "loss": 0.03569352626800537,
      "step": 5690
    },
    {
      "epoch": 0.7693854616431064,
      "grad_norm": 0.47424882650375366,
      "learning_rate": 4.32465666390532e-06,
      "loss": 0.0582297146320343,
      "step": 5691
    },
    {
      "epoch": 0.7695206550118716,
      "grad_norm": 0.2785654366016388,
      "learning_rate": 4.319849604730905e-06,
      "loss": 0.04558062180876732,
      "step": 5692
    },
    {
      "epoch": 0.7696558483806369,
      "grad_norm": 0.43649712204933167,
      "learning_rate": 4.315044769220758e-06,
      "loss": 0.06203489005565643,
      "step": 5693
    },
    {
      "epoch": 0.7697910417494022,
      "grad_norm": 0.6950480937957764,
      "learning_rate": 4.310242158375264e-06,
      "loss": 0.05505382642149925,
      "step": 5694
    },
    {
      "epoch": 0.7699262351181675,
      "grad_norm": 0.5324304103851318,
      "learning_rate": 4.30544177319436e-06,
      "loss": 0.049787141382694244,
      "step": 5695
    },
    {
      "epoch": 0.7700614284869327,
      "grad_norm": 0.44713354110717773,
      "learning_rate": 4.300643614677511e-06,
      "loss": 0.05204059183597565,
      "step": 5696
    },
    {
      "epoch": 0.7701966218556979,
      "grad_norm": 0.3508920669555664,
      "learning_rate": 4.2958476838237165e-06,
      "loss": 0.0416136234998703,
      "step": 5697
    },
    {
      "epoch": 0.7703318152244633,
      "grad_norm": 0.7075027227401733,
      "learning_rate": 4.2910539816315166e-06,
      "loss": 0.04444601386785507,
      "step": 5698
    },
    {
      "epoch": 0.7704670085932285,
      "grad_norm": 0.790549635887146,
      "learning_rate": 4.286262509098979e-06,
      "loss": 0.05803542211651802,
      "step": 5699
    },
    {
      "epoch": 0.7706022019619938,
      "grad_norm": 1.0832196474075317,
      "learning_rate": 4.28147326722373e-06,
      "loss": 0.03805040568113327,
      "step": 5700
    },
    {
      "epoch": 0.770737395330759,
      "grad_norm": 0.6148602366447449,
      "learning_rate": 4.2766862570028965e-06,
      "loss": 0.07428151369094849,
      "step": 5701
    },
    {
      "epoch": 0.7708725886995242,
      "grad_norm": 0.7821617722511292,
      "learning_rate": 4.2719014794331715e-06,
      "loss": 0.04477187618613243,
      "step": 5702
    },
    {
      "epoch": 0.7710077820682896,
      "grad_norm": 0.6982478499412537,
      "learning_rate": 4.267118935510767e-06,
      "loss": 0.06008635461330414,
      "step": 5703
    },
    {
      "epoch": 0.7711429754370548,
      "grad_norm": 0.43888673186302185,
      "learning_rate": 4.2623386262314306e-06,
      "loss": 0.06447956711053848,
      "step": 5704
    },
    {
      "epoch": 0.7712781688058201,
      "grad_norm": 0.41614997386932373,
      "learning_rate": 4.257560552590461e-06,
      "loss": 0.052055105566978455,
      "step": 5705
    },
    {
      "epoch": 0.7714133621745853,
      "grad_norm": 0.34694284200668335,
      "learning_rate": 4.252784715582661e-06,
      "loss": 0.04736727476119995,
      "step": 5706
    },
    {
      "epoch": 0.7715485555433506,
      "grad_norm": 0.5842350125312805,
      "learning_rate": 4.2480111162024e-06,
      "loss": 0.031814415007829666,
      "step": 5707
    },
    {
      "epoch": 0.7716837489121159,
      "grad_norm": 0.4248739778995514,
      "learning_rate": 4.243239755443561e-06,
      "loss": 0.04953247308731079,
      "step": 5708
    },
    {
      "epoch": 0.7718189422808811,
      "grad_norm": 0.5501111149787903,
      "learning_rate": 4.238470634299567e-06,
      "loss": 0.041759759187698364,
      "step": 5709
    },
    {
      "epoch": 0.7719541356496464,
      "grad_norm": 0.5856011509895325,
      "learning_rate": 4.233703753763375e-06,
      "loss": 0.060414426028728485,
      "step": 5710
    },
    {
      "epoch": 0.7720893290184117,
      "grad_norm": 0.39352521300315857,
      "learning_rate": 4.228939114827469e-06,
      "loss": 0.06005595624446869,
      "step": 5711
    },
    {
      "epoch": 0.7722245223871769,
      "grad_norm": 1.007741093635559,
      "learning_rate": 4.224176718483881e-06,
      "loss": 0.04893037676811218,
      "step": 5712
    },
    {
      "epoch": 0.7723597157559422,
      "grad_norm": 0.6353216171264648,
      "learning_rate": 4.219416565724165e-06,
      "loss": 0.05638323351740837,
      "step": 5713
    },
    {
      "epoch": 0.7724949091247074,
      "grad_norm": 0.47796788811683655,
      "learning_rate": 4.21465865753941e-06,
      "loss": 0.05115942656993866,
      "step": 5714
    },
    {
      "epoch": 0.7726301024934727,
      "grad_norm": 0.7587752938270569,
      "learning_rate": 4.209902994920236e-06,
      "loss": 0.051104605197906494,
      "step": 5715
    },
    {
      "epoch": 0.772765295862238,
      "grad_norm": 0.9474219083786011,
      "learning_rate": 4.205149578856794e-06,
      "loss": 0.05527476966381073,
      "step": 5716
    },
    {
      "epoch": 0.7729004892310032,
      "grad_norm": 1.4634283781051636,
      "learning_rate": 4.200398410338779e-06,
      "loss": 0.05003051832318306,
      "step": 5717
    },
    {
      "epoch": 0.7730356825997685,
      "grad_norm": 0.29603874683380127,
      "learning_rate": 4.1956494903554056e-06,
      "loss": 0.045849766582250595,
      "step": 5718
    },
    {
      "epoch": 0.7731708759685337,
      "grad_norm": 0.5924423336982727,
      "learning_rate": 4.190902819895425e-06,
      "loss": 0.04351881891489029,
      "step": 5719
    },
    {
      "epoch": 0.773306069337299,
      "grad_norm": 0.4455014765262604,
      "learning_rate": 4.186158399947118e-06,
      "loss": 0.04066503047943115,
      "step": 5720
    },
    {
      "epoch": 0.7734412627060643,
      "grad_norm": 0.4561283588409424,
      "learning_rate": 4.181416231498292e-06,
      "loss": 0.043446414172649384,
      "step": 5721
    },
    {
      "epoch": 0.7735764560748295,
      "grad_norm": 0.4668290913105011,
      "learning_rate": 4.176676315536306e-06,
      "loss": 0.04991087689995766,
      "step": 5722
    },
    {
      "epoch": 0.7737116494435948,
      "grad_norm": 0.610687255859375,
      "learning_rate": 4.171938653048027e-06,
      "loss": 0.05808749049901962,
      "step": 5723
    },
    {
      "epoch": 0.77384684281236,
      "grad_norm": 0.5134508013725281,
      "learning_rate": 4.1672032450198616e-06,
      "loss": 0.06386997550725937,
      "step": 5724
    },
    {
      "epoch": 0.7739820361811253,
      "grad_norm": 0.3268800973892212,
      "learning_rate": 4.16247009243775e-06,
      "loss": 0.051299914717674255,
      "step": 5725
    },
    {
      "epoch": 0.7741172295498906,
      "grad_norm": 0.31994378566741943,
      "learning_rate": 4.1577391962871504e-06,
      "loss": 0.04202529788017273,
      "step": 5726
    },
    {
      "epoch": 0.7742524229186558,
      "grad_norm": 1.1951512098312378,
      "learning_rate": 4.153010557553076e-06,
      "loss": 0.056732963770627975,
      "step": 5727
    },
    {
      "epoch": 0.7743876162874211,
      "grad_norm": 0.2234276682138443,
      "learning_rate": 4.148284177220045e-06,
      "loss": 0.05105416476726532,
      "step": 5728
    },
    {
      "epoch": 0.7745228096561864,
      "grad_norm": 0.6800549030303955,
      "learning_rate": 4.143560056272117e-06,
      "loss": 0.039719775319099426,
      "step": 5729
    },
    {
      "epoch": 0.7746580030249516,
      "grad_norm": 0.6751525402069092,
      "learning_rate": 4.1388381956928796e-06,
      "loss": 0.052033185958862305,
      "step": 5730
    },
    {
      "epoch": 0.7747931963937169,
      "grad_norm": 0.39217954874038696,
      "learning_rate": 4.134118596465443e-06,
      "loss": 0.049040380865335464,
      "step": 5731
    },
    {
      "epoch": 0.7749283897624821,
      "grad_norm": 0.5880181193351746,
      "learning_rate": 4.1294012595724675e-06,
      "loss": 0.05009182170033455,
      "step": 5732
    },
    {
      "epoch": 0.7750635831312475,
      "grad_norm": 0.3974986970424652,
      "learning_rate": 4.1246861859961114e-06,
      "loss": 0.05804557725787163,
      "step": 5733
    },
    {
      "epoch": 0.7751987765000127,
      "grad_norm": 0.4695318639278412,
      "learning_rate": 4.119973376718089e-06,
      "loss": 0.053791116923093796,
      "step": 5734
    },
    {
      "epoch": 0.7753339698687779,
      "grad_norm": 0.3485158383846283,
      "learning_rate": 4.115262832719628e-06,
      "loss": 0.05303540080785751,
      "step": 5735
    },
    {
      "epoch": 0.7754691632375432,
      "grad_norm": 0.59285569190979,
      "learning_rate": 4.110554554981486e-06,
      "loss": 0.047546081244945526,
      "step": 5736
    },
    {
      "epoch": 0.7756043566063084,
      "grad_norm": 0.562120258808136,
      "learning_rate": 4.1058485444839655e-06,
      "loss": 0.07138626277446747,
      "step": 5737
    },
    {
      "epoch": 0.7757395499750738,
      "grad_norm": 0.5767315626144409,
      "learning_rate": 4.101144802206862e-06,
      "loss": 0.07587608695030212,
      "step": 5738
    },
    {
      "epoch": 0.775874743343839,
      "grad_norm": 0.5098394155502319,
      "learning_rate": 4.096443329129535e-06,
      "loss": 0.03905707970261574,
      "step": 5739
    },
    {
      "epoch": 0.7760099367126042,
      "grad_norm": 0.6379435658454895,
      "learning_rate": 4.091744126230853e-06,
      "loss": 0.06477160006761551,
      "step": 5740
    },
    {
      "epoch": 0.7761451300813695,
      "grad_norm": 0.42736637592315674,
      "learning_rate": 4.08704719448921e-06,
      "loss": 0.04898831248283386,
      "step": 5741
    },
    {
      "epoch": 0.7762803234501348,
      "grad_norm": 0.4801262319087982,
      "learning_rate": 4.082352534882543e-06,
      "loss": 0.043267395347356796,
      "step": 5742
    },
    {
      "epoch": 0.7764155168189001,
      "grad_norm": 0.3270072340965271,
      "learning_rate": 4.07766014838829e-06,
      "loss": 0.05136318877339363,
      "step": 5743
    },
    {
      "epoch": 0.7765507101876653,
      "grad_norm": 0.46256282925605774,
      "learning_rate": 4.072970035983443e-06,
      "loss": 0.051225755363702774,
      "step": 5744
    },
    {
      "epoch": 0.7766859035564305,
      "grad_norm": 0.7822785377502441,
      "learning_rate": 4.068282198644505e-06,
      "loss": 0.05353916436433792,
      "step": 5745
    },
    {
      "epoch": 0.7768210969251959,
      "grad_norm": 0.40267154574394226,
      "learning_rate": 4.06359663734751e-06,
      "loss": 0.03540894016623497,
      "step": 5746
    },
    {
      "epoch": 0.7769562902939611,
      "grad_norm": 0.2805401086807251,
      "learning_rate": 4.058913353068013e-06,
      "loss": 0.04988737404346466,
      "step": 5747
    },
    {
      "epoch": 0.7770914836627263,
      "grad_norm": 0.654659628868103,
      "learning_rate": 4.0542323467810985e-06,
      "loss": 0.0653553307056427,
      "step": 5748
    },
    {
      "epoch": 0.7772266770314916,
      "grad_norm": 0.57283616065979,
      "learning_rate": 4.049553619461381e-06,
      "loss": 0.06992589682340622,
      "step": 5749
    },
    {
      "epoch": 0.7773618704002568,
      "grad_norm": 0.443215012550354,
      "learning_rate": 4.044877172082997e-06,
      "loss": 0.04163704812526703,
      "step": 5750
    },
    {
      "epoch": 0.7774970637690222,
      "grad_norm": 0.9152970314025879,
      "learning_rate": 4.040203005619604e-06,
      "loss": 0.05481569468975067,
      "step": 5751
    },
    {
      "epoch": 0.7776322571377874,
      "grad_norm": 0.37695738673210144,
      "learning_rate": 4.035531121044392e-06,
      "loss": 0.04760066792368889,
      "step": 5752
    },
    {
      "epoch": 0.7777674505065526,
      "grad_norm": 0.4861292541027069,
      "learning_rate": 4.030861519330065e-06,
      "loss": 0.051728539168834686,
      "step": 5753
    },
    {
      "epoch": 0.7779026438753179,
      "grad_norm": 0.4190061390399933,
      "learning_rate": 4.026194201448868e-06,
      "loss": 0.04368724673986435,
      "step": 5754
    },
    {
      "epoch": 0.7780378372440832,
      "grad_norm": 0.454704612493515,
      "learning_rate": 4.021529168372558e-06,
      "loss": 0.0677175521850586,
      "step": 5755
    },
    {
      "epoch": 0.7781730306128485,
      "grad_norm": 0.8699820041656494,
      "learning_rate": 4.01686642107242e-06,
      "loss": 0.04995008930563927,
      "step": 5756
    },
    {
      "epoch": 0.7783082239816137,
      "grad_norm": 0.5254787802696228,
      "learning_rate": 4.0122059605192624e-06,
      "loss": 0.04606795310974121,
      "step": 5757
    },
    {
      "epoch": 0.7784434173503789,
      "grad_norm": 0.3809521198272705,
      "learning_rate": 4.007547787683412e-06,
      "loss": 0.047455623745918274,
      "step": 5758
    },
    {
      "epoch": 0.7785786107191442,
      "grad_norm": 0.4173734486103058,
      "learning_rate": 4.002891903534736e-06,
      "loss": 0.060535043478012085,
      "step": 5759
    },
    {
      "epoch": 0.7787138040879095,
      "grad_norm": 0.4707096517086029,
      "learning_rate": 3.998238309042611e-06,
      "loss": 0.06532557308673859,
      "step": 5760
    },
    {
      "epoch": 0.7788489974566748,
      "grad_norm": 0.49603208899497986,
      "learning_rate": 3.993587005175937e-06,
      "loss": 0.035698167979717255,
      "step": 5761
    },
    {
      "epoch": 0.77898419082544,
      "grad_norm": 0.6485784649848938,
      "learning_rate": 3.988937992903144e-06,
      "loss": 0.07103504240512848,
      "step": 5762
    },
    {
      "epoch": 0.7791193841942052,
      "grad_norm": 0.8104240894317627,
      "learning_rate": 3.9842912731921716e-06,
      "loss": 0.04359731078147888,
      "step": 5763
    },
    {
      "epoch": 0.7792545775629706,
      "grad_norm": 0.3764703571796417,
      "learning_rate": 3.979646847010506e-06,
      "loss": 0.045068494975566864,
      "step": 5764
    },
    {
      "epoch": 0.7793897709317358,
      "grad_norm": 0.4137573838233948,
      "learning_rate": 3.975004715325134e-06,
      "loss": 0.04682117700576782,
      "step": 5765
    },
    {
      "epoch": 0.7795249643005011,
      "grad_norm": 1.1215702295303345,
      "learning_rate": 3.970364879102572e-06,
      "loss": 0.04776846244931221,
      "step": 5766
    },
    {
      "epoch": 0.7796601576692663,
      "grad_norm": 0.5139415264129639,
      "learning_rate": 3.96572733930886e-06,
      "loss": 0.07105226814746857,
      "step": 5767
    },
    {
      "epoch": 0.7797953510380315,
      "grad_norm": 0.5522721409797668,
      "learning_rate": 3.961092096909552e-06,
      "loss": 0.053970180451869965,
      "step": 5768
    },
    {
      "epoch": 0.7799305444067969,
      "grad_norm": 0.47779467701911926,
      "learning_rate": 3.9564591528697455e-06,
      "loss": 0.03991309180855751,
      "step": 5769
    },
    {
      "epoch": 0.7800657377755621,
      "grad_norm": 0.28264862298965454,
      "learning_rate": 3.9518285081540275e-06,
      "loss": 0.04973236471414566,
      "step": 5770
    },
    {
      "epoch": 0.7802009311443274,
      "grad_norm": 0.4290755093097687,
      "learning_rate": 3.947200163726534e-06,
      "loss": 0.05141142010688782,
      "step": 5771
    },
    {
      "epoch": 0.7803361245130926,
      "grad_norm": 0.614381730556488,
      "learning_rate": 3.9425741205509055e-06,
      "loss": 0.06349176168441772,
      "step": 5772
    },
    {
      "epoch": 0.7804713178818579,
      "grad_norm": 0.5289402604103088,
      "learning_rate": 3.9379503795903065e-06,
      "loss": 0.05813773721456528,
      "step": 5773
    },
    {
      "epoch": 0.7806065112506232,
      "grad_norm": 0.6217710375785828,
      "learning_rate": 3.933328941807439e-06,
      "loss": 0.04701855778694153,
      "step": 5774
    },
    {
      "epoch": 0.7807417046193884,
      "grad_norm": 1.0539789199829102,
      "learning_rate": 3.928709808164491e-06,
      "loss": 0.0448966845870018,
      "step": 5775
    },
    {
      "epoch": 0.7808768979881536,
      "grad_norm": 0.35859671235084534,
      "learning_rate": 3.924092979623203e-06,
      "loss": 0.045018672943115234,
      "step": 5776
    },
    {
      "epoch": 0.781012091356919,
      "grad_norm": 0.5532165169715881,
      "learning_rate": 3.919478457144824e-06,
      "loss": 0.054671987891197205,
      "step": 5777
    },
    {
      "epoch": 0.7811472847256842,
      "grad_norm": 0.28777357935905457,
      "learning_rate": 3.914866241690115e-06,
      "loss": 0.04595384746789932,
      "step": 5778
    },
    {
      "epoch": 0.7812824780944495,
      "grad_norm": 0.3780209720134735,
      "learning_rate": 3.9102563342193695e-06,
      "loss": 0.06887669861316681,
      "step": 5779
    },
    {
      "epoch": 0.7814176714632147,
      "grad_norm": 0.3771103024482727,
      "learning_rate": 3.905648735692389e-06,
      "loss": 0.04795943200588226,
      "step": 5780
    },
    {
      "epoch": 0.7815528648319799,
      "grad_norm": 0.6279870271682739,
      "learning_rate": 3.901043447068508e-06,
      "loss": 0.0433628186583519,
      "step": 5781
    },
    {
      "epoch": 0.7816880582007453,
      "grad_norm": 0.5482854843139648,
      "learning_rate": 3.896440469306567e-06,
      "loss": 0.07337145507335663,
      "step": 5782
    },
    {
      "epoch": 0.7818232515695105,
      "grad_norm": 0.6327605843544006,
      "learning_rate": 3.891839803364934e-06,
      "loss": 0.043239280581474304,
      "step": 5783
    },
    {
      "epoch": 0.7819584449382758,
      "grad_norm": 0.5140485167503357,
      "learning_rate": 3.887241450201487e-06,
      "loss": 0.045149900019168854,
      "step": 5784
    },
    {
      "epoch": 0.782093638307041,
      "grad_norm": 0.703194797039032,
      "learning_rate": 3.882645410773629e-06,
      "loss": 0.06559029966592789,
      "step": 5785
    },
    {
      "epoch": 0.7822288316758063,
      "grad_norm": 0.24512290954589844,
      "learning_rate": 3.878051686038284e-06,
      "loss": 0.05819549784064293,
      "step": 5786
    },
    {
      "epoch": 0.7823640250445716,
      "grad_norm": 0.6596559286117554,
      "learning_rate": 3.873460276951889e-06,
      "loss": 0.05555453896522522,
      "step": 5787
    },
    {
      "epoch": 0.7824992184133368,
      "grad_norm": 0.9214807152748108,
      "learning_rate": 3.868871184470397e-06,
      "loss": 0.03861886262893677,
      "step": 5788
    },
    {
      "epoch": 0.7826344117821021,
      "grad_norm": 0.5653980374336243,
      "learning_rate": 3.864284409549282e-06,
      "loss": 0.05683492124080658,
      "step": 5789
    },
    {
      "epoch": 0.7827696051508674,
      "grad_norm": 0.5974285006523132,
      "learning_rate": 3.859699953143532e-06,
      "loss": 0.054988350719213486,
      "step": 5790
    },
    {
      "epoch": 0.7829047985196326,
      "grad_norm": 0.5913573503494263,
      "learning_rate": 3.855117816207665e-06,
      "loss": 0.048264771699905396,
      "step": 5791
    },
    {
      "epoch": 0.7830399918883979,
      "grad_norm": 0.8017600774765015,
      "learning_rate": 3.850537999695699e-06,
      "loss": 0.046860143542289734,
      "step": 5792
    },
    {
      "epoch": 0.7831751852571631,
      "grad_norm": 0.8446592092514038,
      "learning_rate": 3.845960504561179e-06,
      "loss": 0.05040036886930466,
      "step": 5793
    },
    {
      "epoch": 0.7833103786259284,
      "grad_norm": 0.4981122314929962,
      "learning_rate": 3.841385331757161e-06,
      "loss": 0.04143751040101051,
      "step": 5794
    },
    {
      "epoch": 0.7834455719946937,
      "grad_norm": 0.4162934124469757,
      "learning_rate": 3.8368124822362184e-06,
      "loss": 0.05737932026386261,
      "step": 5795
    },
    {
      "epoch": 0.7835807653634589,
      "grad_norm": 0.661861777305603,
      "learning_rate": 3.832241956950449e-06,
      "loss": 0.048417821526527405,
      "step": 5796
    },
    {
      "epoch": 0.7837159587322242,
      "grad_norm": 0.5370731353759766,
      "learning_rate": 3.82767375685146e-06,
      "loss": 0.056625284254550934,
      "step": 5797
    },
    {
      "epoch": 0.7838511521009894,
      "grad_norm": 0.31087347865104675,
      "learning_rate": 3.823107882890373e-06,
      "loss": 0.04839096963405609,
      "step": 5798
    },
    {
      "epoch": 0.7839863454697548,
      "grad_norm": 0.5822319388389587,
      "learning_rate": 3.8185443360178265e-06,
      "loss": 0.0583285391330719,
      "step": 5799
    },
    {
      "epoch": 0.78412153883852,
      "grad_norm": 0.3733569085597992,
      "learning_rate": 3.813983117183973e-06,
      "loss": 0.0516224168241024,
      "step": 5800
    },
    {
      "epoch": 0.7842567322072852,
      "grad_norm": 0.46516770124435425,
      "learning_rate": 3.8094242273384932e-06,
      "loss": 0.04963213950395584,
      "step": 5801
    },
    {
      "epoch": 0.7843919255760505,
      "grad_norm": 0.36023858189582825,
      "learning_rate": 3.804867667430555e-06,
      "loss": 0.0596126914024353,
      "step": 5802
    },
    {
      "epoch": 0.7845271189448157,
      "grad_norm": 0.5960729122161865,
      "learning_rate": 3.800313438408874e-06,
      "loss": 0.06885658204555511,
      "step": 5803
    },
    {
      "epoch": 0.784662312313581,
      "grad_norm": 0.5354682803153992,
      "learning_rate": 3.7957615412216582e-06,
      "loss": 0.05400829389691353,
      "step": 5804
    },
    {
      "epoch": 0.7847975056823463,
      "grad_norm": 0.8081690073013306,
      "learning_rate": 3.791211976816634e-06,
      "loss": 0.047955870628356934,
      "step": 5805
    },
    {
      "epoch": 0.7849326990511115,
      "grad_norm": 0.7521482706069946,
      "learning_rate": 3.786664746141057e-06,
      "loss": 0.07321782410144806,
      "step": 5806
    },
    {
      "epoch": 0.7850678924198768,
      "grad_norm": 0.656195878982544,
      "learning_rate": 3.782119850141669e-06,
      "loss": 0.04018782824277878,
      "step": 5807
    },
    {
      "epoch": 0.7852030857886421,
      "grad_norm": 0.3345497250556946,
      "learning_rate": 3.777577289764752e-06,
      "loss": 0.06700266897678375,
      "step": 5808
    },
    {
      "epoch": 0.7853382791574073,
      "grad_norm": 0.40408748388290405,
      "learning_rate": 3.7730370659560904e-06,
      "loss": 0.04518834874033928,
      "step": 5809
    },
    {
      "epoch": 0.7854734725261726,
      "grad_norm": 0.5429202318191528,
      "learning_rate": 3.7684991796609746e-06,
      "loss": 0.045586466789245605,
      "step": 5810
    },
    {
      "epoch": 0.7856086658949378,
      "grad_norm": 0.4379957914352417,
      "learning_rate": 3.7639636318242344e-06,
      "loss": 0.06153438985347748,
      "step": 5811
    },
    {
      "epoch": 0.7857438592637032,
      "grad_norm": 0.41803351044654846,
      "learning_rate": 3.7594304233901738e-06,
      "loss": 0.06180073320865631,
      "step": 5812
    },
    {
      "epoch": 0.7858790526324684,
      "grad_norm": 0.37024974822998047,
      "learning_rate": 3.754899555302645e-06,
      "loss": 0.05707992613315582,
      "step": 5813
    },
    {
      "epoch": 0.7860142460012336,
      "grad_norm": 0.9152238965034485,
      "learning_rate": 3.7503710285049964e-06,
      "loss": 0.04836093634366989,
      "step": 5814
    },
    {
      "epoch": 0.7861494393699989,
      "grad_norm": 0.29534226655960083,
      "learning_rate": 3.7458448439400888e-06,
      "loss": 0.05663030594587326,
      "step": 5815
    },
    {
      "epoch": 0.7862846327387641,
      "grad_norm": 0.9970876574516296,
      "learning_rate": 3.7413210025502985e-06,
      "loss": 0.055408746004104614,
      "step": 5816
    },
    {
      "epoch": 0.7864198261075295,
      "grad_norm": 0.9171462655067444,
      "learning_rate": 3.7367995052775123e-06,
      "loss": 0.04693305864930153,
      "step": 5817
    },
    {
      "epoch": 0.7865550194762947,
      "grad_norm": 0.39299800992012024,
      "learning_rate": 3.732280353063133e-06,
      "loss": 0.043476566672325134,
      "step": 5818
    },
    {
      "epoch": 0.7866902128450599,
      "grad_norm": 0.4400605261325836,
      "learning_rate": 3.727763546848074e-06,
      "loss": 0.05251064896583557,
      "step": 5819
    },
    {
      "epoch": 0.7868254062138252,
      "grad_norm": 0.2650921046733856,
      "learning_rate": 3.7232490875727544e-06,
      "loss": 0.04420124739408493,
      "step": 5820
    },
    {
      "epoch": 0.7869605995825905,
      "grad_norm": 0.5161288380622864,
      "learning_rate": 3.718736976177108e-06,
      "loss": 0.05436871945858002,
      "step": 5821
    },
    {
      "epoch": 0.7870957929513558,
      "grad_norm": 0.45217934250831604,
      "learning_rate": 3.71422721360058e-06,
      "loss": 0.051149845123291016,
      "step": 5822
    },
    {
      "epoch": 0.787230986320121,
      "grad_norm": 0.4663630723953247,
      "learning_rate": 3.709719800782133e-06,
      "loss": 0.04806032031774521,
      "step": 5823
    },
    {
      "epoch": 0.7873661796888862,
      "grad_norm": 0.5508677363395691,
      "learning_rate": 3.7052147386602304e-06,
      "loss": 0.04836851358413696,
      "step": 5824
    },
    {
      "epoch": 0.7875013730576516,
      "grad_norm": 0.33200010657310486,
      "learning_rate": 3.700712028172851e-06,
      "loss": 0.05179045349359512,
      "step": 5825
    },
    {
      "epoch": 0.7876365664264168,
      "grad_norm": 0.4456416666507721,
      "learning_rate": 3.696211670257481e-06,
      "loss": 0.06013600155711174,
      "step": 5826
    },
    {
      "epoch": 0.787771759795182,
      "grad_norm": 0.5219780802726746,
      "learning_rate": 3.691713665851117e-06,
      "loss": 0.0494326576590538,
      "step": 5827
    },
    {
      "epoch": 0.7879069531639473,
      "grad_norm": 0.5698608160018921,
      "learning_rate": 3.6872180158902764e-06,
      "loss": 0.05211539566516876,
      "step": 5828
    },
    {
      "epoch": 0.7880421465327125,
      "grad_norm": 0.4790526330471039,
      "learning_rate": 3.6827247213109705e-06,
      "loss": 0.05283583700656891,
      "step": 5829
    },
    {
      "epoch": 0.7881773399014779,
      "grad_norm": 0.3996173143386841,
      "learning_rate": 3.6782337830487294e-06,
      "loss": 0.05189451575279236,
      "step": 5830
    },
    {
      "epoch": 0.7883125332702431,
      "grad_norm": 0.8280686736106873,
      "learning_rate": 3.6737452020385886e-06,
      "loss": 0.04844554513692856,
      "step": 5831
    },
    {
      "epoch": 0.7884477266390083,
      "grad_norm": 0.406904399394989,
      "learning_rate": 3.6692589792150923e-06,
      "loss": 0.0553140714764595,
      "step": 5832
    },
    {
      "epoch": 0.7885829200077736,
      "grad_norm": 0.4483954608440399,
      "learning_rate": 3.6647751155123026e-06,
      "loss": 0.04909306764602661,
      "step": 5833
    },
    {
      "epoch": 0.7887181133765389,
      "grad_norm": 1.0544463396072388,
      "learning_rate": 3.660293611863782e-06,
      "loss": 0.049436070024967194,
      "step": 5834
    },
    {
      "epoch": 0.7888533067453042,
      "grad_norm": 0.558396577835083,
      "learning_rate": 3.655814469202602e-06,
      "loss": 0.04861228168010712,
      "step": 5835
    },
    {
      "epoch": 0.7889885001140694,
      "grad_norm": 0.8497613668441772,
      "learning_rate": 3.6513376884613446e-06,
      "loss": 0.06306800991296768,
      "step": 5836
    },
    {
      "epoch": 0.7891236934828346,
      "grad_norm": 0.47947338223457336,
      "learning_rate": 3.6468632705720934e-06,
      "loss": 0.03444236144423485,
      "step": 5837
    },
    {
      "epoch": 0.7892588868516,
      "grad_norm": 0.7960044741630554,
      "learning_rate": 3.6423912164664606e-06,
      "loss": 0.05294312536716461,
      "step": 5838
    },
    {
      "epoch": 0.7893940802203652,
      "grad_norm": 0.4889637231826782,
      "learning_rate": 3.637921527075534e-06,
      "loss": 0.05175085365772247,
      "step": 5839
    },
    {
      "epoch": 0.7895292735891305,
      "grad_norm": 0.4937731921672821,
      "learning_rate": 3.63345420332994e-06,
      "loss": 0.0552225336432457,
      "step": 5840
    },
    {
      "epoch": 0.7896644669578957,
      "grad_norm": 0.2001790553331375,
      "learning_rate": 3.628989246159795e-06,
      "loss": 0.03912050276994705,
      "step": 5841
    },
    {
      "epoch": 0.7897996603266609,
      "grad_norm": 0.6610744595527649,
      "learning_rate": 3.6245266564947205e-06,
      "loss": 0.06487545371055603,
      "step": 5842
    },
    {
      "epoch": 0.7899348536954263,
      "grad_norm": 0.5012447834014893,
      "learning_rate": 3.620066435263868e-06,
      "loss": 0.05035126209259033,
      "step": 5843
    },
    {
      "epoch": 0.7900700470641915,
      "grad_norm": 0.43423858284950256,
      "learning_rate": 3.6156085833958596e-06,
      "loss": 0.05305560678243637,
      "step": 5844
    },
    {
      "epoch": 0.7902052404329568,
      "grad_norm": 0.3528660833835602,
      "learning_rate": 3.6111531018188584e-06,
      "loss": 0.054082803428173065,
      "step": 5845
    },
    {
      "epoch": 0.790340433801722,
      "grad_norm": 0.4592651128768921,
      "learning_rate": 3.606699991460513e-06,
      "loss": 0.04173130542039871,
      "step": 5846
    },
    {
      "epoch": 0.7904756271704872,
      "grad_norm": 0.4069738984107971,
      "learning_rate": 3.602249253247986e-06,
      "loss": 0.05436393618583679,
      "step": 5847
    },
    {
      "epoch": 0.7906108205392526,
      "grad_norm": 0.375313937664032,
      "learning_rate": 3.5978008881079445e-06,
      "loss": 0.0516459122300148,
      "step": 5848
    },
    {
      "epoch": 0.7907460139080178,
      "grad_norm": 0.3917820155620575,
      "learning_rate": 3.5933548969665587e-06,
      "loss": 0.04290139675140381,
      "step": 5849
    },
    {
      "epoch": 0.7908812072767831,
      "grad_norm": 0.27167415618896484,
      "learning_rate": 3.5889112807495152e-06,
      "loss": 0.046996913850307465,
      "step": 5850
    },
    {
      "epoch": 0.7910164006455483,
      "grad_norm": 0.30987969040870667,
      "learning_rate": 3.5844700403819935e-06,
      "loss": 0.05544118583202362,
      "step": 5851
    },
    {
      "epoch": 0.7911515940143136,
      "grad_norm": 1.262507438659668,
      "learning_rate": 3.5800311767886847e-06,
      "loss": 0.06227273866534233,
      "step": 5852
    },
    {
      "epoch": 0.7912867873830789,
      "grad_norm": 0.5790842771530151,
      "learning_rate": 3.575594690893784e-06,
      "loss": 0.041585374623537064,
      "step": 5853
    },
    {
      "epoch": 0.7914219807518441,
      "grad_norm": 0.7852163910865784,
      "learning_rate": 3.5711605836209853e-06,
      "loss": 0.05485051870346069,
      "step": 5854
    },
    {
      "epoch": 0.7915571741206093,
      "grad_norm": 0.4399736821651459,
      "learning_rate": 3.566728855893505e-06,
      "loss": 0.05603671073913574,
      "step": 5855
    },
    {
      "epoch": 0.7916923674893747,
      "grad_norm": 0.40009763836860657,
      "learning_rate": 3.5622995086340466e-06,
      "loss": 0.04773164168000221,
      "step": 5856
    },
    {
      "epoch": 0.7918275608581399,
      "grad_norm": 0.3449288010597229,
      "learning_rate": 3.5578725427648233e-06,
      "loss": 0.048646703362464905,
      "step": 5857
    },
    {
      "epoch": 0.7919627542269052,
      "grad_norm": 0.28758126497268677,
      "learning_rate": 3.553447959207553e-06,
      "loss": 0.041172951459884644,
      "step": 5858
    },
    {
      "epoch": 0.7920979475956704,
      "grad_norm": 0.44380614161491394,
      "learning_rate": 3.5490257588834552e-06,
      "loss": 0.042261168360710144,
      "step": 5859
    },
    {
      "epoch": 0.7922331409644356,
      "grad_norm": 0.5089505910873413,
      "learning_rate": 3.5446059427132615e-06,
      "loss": 0.05467886105179787,
      "step": 5860
    },
    {
      "epoch": 0.792368334333201,
      "grad_norm": 0.3433547616004944,
      "learning_rate": 3.5401885116171977e-06,
      "loss": 0.04381384328007698,
      "step": 5861
    },
    {
      "epoch": 0.7925035277019662,
      "grad_norm": 0.5133224725723267,
      "learning_rate": 3.5357734665149983e-06,
      "loss": 0.057044968008995056,
      "step": 5862
    },
    {
      "epoch": 0.7926387210707315,
      "grad_norm": 0.5437789559364319,
      "learning_rate": 3.5313608083258975e-06,
      "loss": 0.05475904420018196,
      "step": 5863
    },
    {
      "epoch": 0.7927739144394967,
      "grad_norm": 0.46170106530189514,
      "learning_rate": 3.526950537968629e-06,
      "loss": 0.0631060004234314,
      "step": 5864
    },
    {
      "epoch": 0.792909107808262,
      "grad_norm": 0.4708152413368225,
      "learning_rate": 3.5225426563614466e-06,
      "loss": 0.046782486140728,
      "step": 5865
    },
    {
      "epoch": 0.7930443011770273,
      "grad_norm": 0.5431990027427673,
      "learning_rate": 3.518137164422088e-06,
      "loss": 0.0453324019908905,
      "step": 5866
    },
    {
      "epoch": 0.7931794945457925,
      "grad_norm": 0.8199841976165771,
      "learning_rate": 3.513734063067799e-06,
      "loss": 0.0767545998096466,
      "step": 5867
    },
    {
      "epoch": 0.7933146879145578,
      "grad_norm": 0.3561827540397644,
      "learning_rate": 3.5093333532153316e-06,
      "loss": 0.04865230619907379,
      "step": 5868
    },
    {
      "epoch": 0.793449881283323,
      "grad_norm": 1.4668605327606201,
      "learning_rate": 3.504935035780931e-06,
      "loss": 0.0693330466747284,
      "step": 5869
    },
    {
      "epoch": 0.7935850746520883,
      "grad_norm": 0.6234803199768066,
      "learning_rate": 3.500539111680364e-06,
      "loss": 0.05087678134441376,
      "step": 5870
    },
    {
      "epoch": 0.7937202680208536,
      "grad_norm": 0.4847146272659302,
      "learning_rate": 3.4961455818288683e-06,
      "loss": 0.060263291001319885,
      "step": 5871
    },
    {
      "epoch": 0.7938554613896188,
      "grad_norm": 0.35021916031837463,
      "learning_rate": 3.491754447141212e-06,
      "loss": 0.032333433628082275,
      "step": 5872
    },
    {
      "epoch": 0.7939906547583842,
      "grad_norm": 0.8613430261611938,
      "learning_rate": 3.4873657085316504e-06,
      "loss": 0.05604889616370201,
      "step": 5873
    },
    {
      "epoch": 0.7941258481271494,
      "grad_norm": 0.680248498916626,
      "learning_rate": 3.482979366913935e-06,
      "loss": 0.04539275914430618,
      "step": 5874
    },
    {
      "epoch": 0.7942610414959146,
      "grad_norm": 0.4233610928058624,
      "learning_rate": 3.4785954232013423e-06,
      "loss": 0.045897115021944046,
      "step": 5875
    },
    {
      "epoch": 0.7943962348646799,
      "grad_norm": 1.1722338199615479,
      "learning_rate": 3.4742138783066122e-06,
      "loss": 0.06935595721006393,
      "step": 5876
    },
    {
      "epoch": 0.7945314282334451,
      "grad_norm": 0.495280385017395,
      "learning_rate": 3.4698347331420206e-06,
      "loss": 0.048590466380119324,
      "step": 5877
    },
    {
      "epoch": 0.7946666216022105,
      "grad_norm": 0.6487596035003662,
      "learning_rate": 3.4654579886193223e-06,
      "loss": 0.05463692173361778,
      "step": 5878
    },
    {
      "epoch": 0.7948018149709757,
      "grad_norm": 0.48033493757247925,
      "learning_rate": 3.461083645649782e-06,
      "loss": 0.04509492963552475,
      "step": 5879
    },
    {
      "epoch": 0.7949370083397409,
      "grad_norm": 0.38070330023765564,
      "learning_rate": 3.4567117051441594e-06,
      "loss": 0.05485737323760986,
      "step": 5880
    },
    {
      "epoch": 0.7950722017085062,
      "grad_norm": 0.6430317163467407,
      "learning_rate": 3.4523421680127115e-06,
      "loss": 0.056696128100156784,
      "step": 5881
    },
    {
      "epoch": 0.7952073950772715,
      "grad_norm": 0.5140246748924255,
      "learning_rate": 3.447975035165209e-06,
      "loss": 0.04350459575653076,
      "step": 5882
    },
    {
      "epoch": 0.7953425884460367,
      "grad_norm": 0.6138967871665955,
      "learning_rate": 3.4436103075109076e-06,
      "loss": 0.04122386872768402,
      "step": 5883
    },
    {
      "epoch": 0.795477781814802,
      "grad_norm": 0.4643726944923401,
      "learning_rate": 3.4392479859585642e-06,
      "loss": 0.07189544290304184,
      "step": 5884
    },
    {
      "epoch": 0.7956129751835672,
      "grad_norm": 0.835229754447937,
      "learning_rate": 3.4348880714164416e-06,
      "loss": 0.04200848191976547,
      "step": 5885
    },
    {
      "epoch": 0.7957481685523325,
      "grad_norm": 1.0927886962890625,
      "learning_rate": 3.430530564792289e-06,
      "loss": 0.07564973831176758,
      "step": 5886
    },
    {
      "epoch": 0.7958833619210978,
      "grad_norm": 0.3393499255180359,
      "learning_rate": 3.426175466993374e-06,
      "loss": 0.05649684742093086,
      "step": 5887
    },
    {
      "epoch": 0.796018555289863,
      "grad_norm": 0.723112165927887,
      "learning_rate": 3.4218227789264468e-06,
      "loss": 0.07994869351387024,
      "step": 5888
    },
    {
      "epoch": 0.7961537486586283,
      "grad_norm": 0.47634345293045044,
      "learning_rate": 3.417472501497758e-06,
      "loss": 0.057938359677791595,
      "step": 5889
    },
    {
      "epoch": 0.7962889420273935,
      "grad_norm": 0.27681440114974976,
      "learning_rate": 3.413124635613061e-06,
      "loss": 0.03615274280309677,
      "step": 5890
    },
    {
      "epoch": 0.7964241353961589,
      "grad_norm": 0.2717513144016266,
      "learning_rate": 3.4087791821775986e-06,
      "loss": 0.04536506533622742,
      "step": 5891
    },
    {
      "epoch": 0.7965593287649241,
      "grad_norm": 0.9388107657432556,
      "learning_rate": 3.4044361420961285e-06,
      "loss": 0.05544573441147804,
      "step": 5892
    },
    {
      "epoch": 0.7966945221336893,
      "grad_norm": 0.4876197278499603,
      "learning_rate": 3.4000955162728866e-06,
      "loss": 0.06180986016988754,
      "step": 5893
    },
    {
      "epoch": 0.7968297155024546,
      "grad_norm": 0.5775308012962341,
      "learning_rate": 3.3957573056116164e-06,
      "loss": 0.06050202250480652,
      "step": 5894
    },
    {
      "epoch": 0.7969649088712198,
      "grad_norm": 0.6010247468948364,
      "learning_rate": 3.391421511015558e-06,
      "loss": 0.06660735607147217,
      "step": 5895
    },
    {
      "epoch": 0.7971001022399852,
      "grad_norm": 0.7866582870483398,
      "learning_rate": 3.38708813338744e-06,
      "loss": 0.05013039708137512,
      "step": 5896
    },
    {
      "epoch": 0.7972352956087504,
      "grad_norm": 0.8188081383705139,
      "learning_rate": 3.382757173629506e-06,
      "loss": 0.04270391911268234,
      "step": 5897
    },
    {
      "epoch": 0.7973704889775156,
      "grad_norm": 0.45585179328918457,
      "learning_rate": 3.378428632643478e-06,
      "loss": 0.040556058287620544,
      "step": 5898
    },
    {
      "epoch": 0.7975056823462809,
      "grad_norm": 0.6447826623916626,
      "learning_rate": 3.3741025113305825e-06,
      "loss": 0.040891677141189575,
      "step": 5899
    },
    {
      "epoch": 0.7976408757150462,
      "grad_norm": 0.5508434176445007,
      "learning_rate": 3.369778810591541e-06,
      "loss": 0.05295625329017639,
      "step": 5900
    },
    {
      "epoch": 0.7977760690838115,
      "grad_norm": 0.38005709648132324,
      "learning_rate": 3.3654575313265664e-06,
      "loss": 0.044245921075344086,
      "step": 5901
    },
    {
      "epoch": 0.7979112624525767,
      "grad_norm": 0.2504396140575409,
      "learning_rate": 3.361138674435386e-06,
      "loss": 0.03862909600138664,
      "step": 5902
    },
    {
      "epoch": 0.7980464558213419,
      "grad_norm": 0.38960644602775574,
      "learning_rate": 3.35682224081719e-06,
      "loss": 0.04928061366081238,
      "step": 5903
    },
    {
      "epoch": 0.7981816491901073,
      "grad_norm": 0.7733510136604309,
      "learning_rate": 3.352508231370699e-06,
      "loss": 0.03577747941017151,
      "step": 5904
    },
    {
      "epoch": 0.7983168425588725,
      "grad_norm": 0.2694184184074402,
      "learning_rate": 3.3481966469941044e-06,
      "loss": 0.03993189334869385,
      "step": 5905
    },
    {
      "epoch": 0.7984520359276378,
      "grad_norm": 0.5073070526123047,
      "learning_rate": 3.3438874885850984e-06,
      "loss": 0.06413762271404266,
      "step": 5906
    },
    {
      "epoch": 0.798587229296403,
      "grad_norm": 0.43924349546432495,
      "learning_rate": 3.3395807570408847e-06,
      "loss": 0.04331628605723381,
      "step": 5907
    },
    {
      "epoch": 0.7987224226651682,
      "grad_norm": 0.41936638951301575,
      "learning_rate": 3.33527645325813e-06,
      "loss": 0.049984727054834366,
      "step": 5908
    },
    {
      "epoch": 0.7988576160339336,
      "grad_norm": 0.3911786675453186,
      "learning_rate": 3.3309745781330247e-06,
      "loss": 0.05361741781234741,
      "step": 5909
    },
    {
      "epoch": 0.7989928094026988,
      "grad_norm": 0.35794228315353394,
      "learning_rate": 3.32667513256124e-06,
      "loss": 0.04724178463220596,
      "step": 5910
    },
    {
      "epoch": 0.799128002771464,
      "grad_norm": 0.36457961797714233,
      "learning_rate": 3.3223781174379375e-06,
      "loss": 0.05640942603349686,
      "step": 5911
    },
    {
      "epoch": 0.7992631961402293,
      "grad_norm": 0.7705474495887756,
      "learning_rate": 3.3180835336577917e-06,
      "loss": 0.04085811227560043,
      "step": 5912
    },
    {
      "epoch": 0.7993983895089946,
      "grad_norm": 0.8283811807632446,
      "learning_rate": 3.313791382114943e-06,
      "loss": 0.06525485217571259,
      "step": 5913
    },
    {
      "epoch": 0.7995335828777599,
      "grad_norm": 0.43810907006263733,
      "learning_rate": 3.3095016637030505e-06,
      "loss": 0.05487743392586708,
      "step": 5914
    },
    {
      "epoch": 0.7996687762465251,
      "grad_norm": 0.5616934895515442,
      "learning_rate": 3.3052143793152524e-06,
      "loss": 0.04688303917646408,
      "step": 5915
    },
    {
      "epoch": 0.7998039696152903,
      "grad_norm": 0.6169284582138062,
      "learning_rate": 3.3009295298441855e-06,
      "loss": 0.05722212791442871,
      "step": 5916
    },
    {
      "epoch": 0.7999391629840557,
      "grad_norm": 0.48905086517333984,
      "learning_rate": 3.2966471161819767e-06,
      "loss": 0.05633637309074402,
      "step": 5917
    },
    {
      "epoch": 0.8000743563528209,
      "grad_norm": 0.36363470554351807,
      "learning_rate": 3.292367139220246e-06,
      "loss": 0.05375855416059494,
      "step": 5918
    },
    {
      "epoch": 0.8002095497215862,
      "grad_norm": 0.37660279870033264,
      "learning_rate": 3.288089599850112e-06,
      "loss": 0.04628485441207886,
      "step": 5919
    },
    {
      "epoch": 0.8003447430903514,
      "grad_norm": 0.4696013331413269,
      "learning_rate": 3.2838144989621795e-06,
      "loss": 0.03974946588277817,
      "step": 5920
    },
    {
      "epoch": 0.8004799364591166,
      "grad_norm": 0.6916917562484741,
      "learning_rate": 3.2795418374465458e-06,
      "loss": 0.06132357195019722,
      "step": 5921
    },
    {
      "epoch": 0.800615129827882,
      "grad_norm": 0.3849853575229645,
      "learning_rate": 3.275271616192803e-06,
      "loss": 0.051506709307432175,
      "step": 5922
    },
    {
      "epoch": 0.8007503231966472,
      "grad_norm": 0.5831172466278076,
      "learning_rate": 3.2710038360900303e-06,
      "loss": 0.03692181408405304,
      "step": 5923
    },
    {
      "epoch": 0.8008855165654125,
      "grad_norm": 0.46334871649742126,
      "learning_rate": 3.266738498026808e-06,
      "loss": 0.05871987342834473,
      "step": 5924
    },
    {
      "epoch": 0.8010207099341777,
      "grad_norm": 0.20020557940006256,
      "learning_rate": 3.2624756028912005e-06,
      "loss": 0.03192692995071411,
      "step": 5925
    },
    {
      "epoch": 0.801155903302943,
      "grad_norm": 0.5319243669509888,
      "learning_rate": 3.2582151515707655e-06,
      "loss": 0.05883929133415222,
      "step": 5926
    },
    {
      "epoch": 0.8012910966717083,
      "grad_norm": 0.8536004424095154,
      "learning_rate": 3.253957144952551e-06,
      "loss": 0.04183913394808769,
      "step": 5927
    },
    {
      "epoch": 0.8014262900404735,
      "grad_norm": 0.32520031929016113,
      "learning_rate": 3.249701583923091e-06,
      "loss": 0.05815349891781807,
      "step": 5928
    },
    {
      "epoch": 0.8015614834092388,
      "grad_norm": 0.4152997136116028,
      "learning_rate": 3.2454484693684257e-06,
      "loss": 0.04818055406212807,
      "step": 5929
    },
    {
      "epoch": 0.801696676778004,
      "grad_norm": 0.5073022842407227,
      "learning_rate": 3.2411978021740727e-06,
      "loss": 0.05007811263203621,
      "step": 5930
    },
    {
      "epoch": 0.8018318701467693,
      "grad_norm": 0.34095102548599243,
      "learning_rate": 3.2369495832250434e-06,
      "loss": 0.04873757064342499,
      "step": 5931
    },
    {
      "epoch": 0.8019670635155346,
      "grad_norm": 1.1944684982299805,
      "learning_rate": 3.2327038134058378e-06,
      "loss": 0.06941474229097366,
      "step": 5932
    },
    {
      "epoch": 0.8021022568842998,
      "grad_norm": 0.3666043281555176,
      "learning_rate": 3.228460493600446e-06,
      "loss": 0.04918378219008446,
      "step": 5933
    },
    {
      "epoch": 0.802237450253065,
      "grad_norm": 0.557352602481842,
      "learning_rate": 3.2242196246923554e-06,
      "loss": 0.04282465577125549,
      "step": 5934
    },
    {
      "epoch": 0.8023726436218304,
      "grad_norm": 0.5829573273658752,
      "learning_rate": 3.2199812075645375e-06,
      "loss": 0.062043510377407074,
      "step": 5935
    },
    {
      "epoch": 0.8025078369905956,
      "grad_norm": 0.8321906924247742,
      "learning_rate": 3.2157452430994487e-06,
      "loss": 0.059558261185884476,
      "step": 5936
    },
    {
      "epoch": 0.8026430303593609,
      "grad_norm": 0.55205899477005,
      "learning_rate": 3.2115117321790427e-06,
      "loss": 0.05116770789027214,
      "step": 5937
    },
    {
      "epoch": 0.8027782237281261,
      "grad_norm": 0.4666525721549988,
      "learning_rate": 3.207280675684754e-06,
      "loss": 0.05095399171113968,
      "step": 5938
    },
    {
      "epoch": 0.8029134170968913,
      "grad_norm": 0.34478697180747986,
      "learning_rate": 3.203052074497523e-06,
      "loss": 0.04628527909517288,
      "step": 5939
    },
    {
      "epoch": 0.8030486104656567,
      "grad_norm": 0.7084249258041382,
      "learning_rate": 3.198825929497752e-06,
      "loss": 0.04001776874065399,
      "step": 5940
    },
    {
      "epoch": 0.8031838038344219,
      "grad_norm": 1.0986882448196411,
      "learning_rate": 3.194602241565357e-06,
      "loss": 0.05720539018511772,
      "step": 5941
    },
    {
      "epoch": 0.8033189972031872,
      "grad_norm": 0.4885752499103546,
      "learning_rate": 3.1903810115797282e-06,
      "loss": 0.05838749185204506,
      "step": 5942
    },
    {
      "epoch": 0.8034541905719524,
      "grad_norm": 1.1779735088348389,
      "learning_rate": 3.1861622404197475e-06,
      "loss": 0.06367367506027222,
      "step": 5943
    },
    {
      "epoch": 0.8035893839407177,
      "grad_norm": 0.7139937281608582,
      "learning_rate": 3.181945928963794e-06,
      "loss": 0.049517132341861725,
      "step": 5944
    },
    {
      "epoch": 0.803724577309483,
      "grad_norm": 0.38674503564834595,
      "learning_rate": 3.1777320780897124e-06,
      "loss": 0.05659361183643341,
      "step": 5945
    },
    {
      "epoch": 0.8038597706782482,
      "grad_norm": 0.6774954795837402,
      "learning_rate": 3.1735206886748602e-06,
      "loss": 0.04896688461303711,
      "step": 5946
    },
    {
      "epoch": 0.8039949640470135,
      "grad_norm": 0.5579783916473389,
      "learning_rate": 3.1693117615960665e-06,
      "loss": 0.06495684385299683,
      "step": 5947
    },
    {
      "epoch": 0.8041301574157788,
      "grad_norm": 0.3195686340332031,
      "learning_rate": 3.1651052977296537e-06,
      "loss": 0.03254825621843338,
      "step": 5948
    },
    {
      "epoch": 0.804265350784544,
      "grad_norm": 0.2509482204914093,
      "learning_rate": 3.1609012979514273e-06,
      "loss": 0.038103729486465454,
      "step": 5949
    },
    {
      "epoch": 0.8044005441533093,
      "grad_norm": 0.41426920890808105,
      "learning_rate": 3.156699763136683e-06,
      "loss": 0.03943061828613281,
      "step": 5950
    },
    {
      "epoch": 0.8045357375220745,
      "grad_norm": 0.36887314915657043,
      "learning_rate": 3.152500694160207e-06,
      "loss": 0.05029183626174927,
      "step": 5951
    },
    {
      "epoch": 0.8046709308908399,
      "grad_norm": 1.0817335844039917,
      "learning_rate": 3.148304091896265e-06,
      "loss": 0.05729471892118454,
      "step": 5952
    },
    {
      "epoch": 0.8048061242596051,
      "grad_norm": 0.4055531322956085,
      "learning_rate": 3.144109957218612e-06,
      "loss": 0.050674036145210266,
      "step": 5953
    },
    {
      "epoch": 0.8049413176283703,
      "grad_norm": 0.6952711343765259,
      "learning_rate": 3.1399182910004893e-06,
      "loss": 0.06114784628152847,
      "step": 5954
    },
    {
      "epoch": 0.8050765109971356,
      "grad_norm": 0.3529430329799652,
      "learning_rate": 3.1357290941146215e-06,
      "loss": 0.04761424660682678,
      "step": 5955
    },
    {
      "epoch": 0.8052117043659008,
      "grad_norm": 1.1649551391601562,
      "learning_rate": 3.1315423674332265e-06,
      "loss": 0.06620429456233978,
      "step": 5956
    },
    {
      "epoch": 0.8053468977346662,
      "grad_norm": 0.6548296809196472,
      "learning_rate": 3.127358111828002e-06,
      "loss": 0.055577006191015244,
      "step": 5957
    },
    {
      "epoch": 0.8054820911034314,
      "grad_norm": 0.39968588948249817,
      "learning_rate": 3.123176328170131e-06,
      "loss": 0.05089409649372101,
      "step": 5958
    },
    {
      "epoch": 0.8056172844721966,
      "grad_norm": 0.19584575295448303,
      "learning_rate": 3.1189970173302816e-06,
      "loss": 0.032486554235219955,
      "step": 5959
    },
    {
      "epoch": 0.8057524778409619,
      "grad_norm": 0.4145294427871704,
      "learning_rate": 3.1148201801786085e-06,
      "loss": 0.04014986380934715,
      "step": 5960
    },
    {
      "epoch": 0.8058876712097272,
      "grad_norm": 0.6185129880905151,
      "learning_rate": 3.1106458175847572e-06,
      "loss": 0.05759398639202118,
      "step": 5961
    },
    {
      "epoch": 0.8060228645784924,
      "grad_norm": 0.3994033932685852,
      "learning_rate": 3.106473930417848e-06,
      "loss": 0.039771318435668945,
      "step": 5962
    },
    {
      "epoch": 0.8061580579472577,
      "grad_norm": 0.44405725598335266,
      "learning_rate": 3.1023045195464903e-06,
      "loss": 0.05337781459093094,
      "step": 5963
    },
    {
      "epoch": 0.8062932513160229,
      "grad_norm": 0.36284634470939636,
      "learning_rate": 3.098137585838779e-06,
      "loss": 0.04051637649536133,
      "step": 5964
    },
    {
      "epoch": 0.8064284446847882,
      "grad_norm": 0.2603057026863098,
      "learning_rate": 3.093973130162286e-06,
      "loss": 0.036102838814258575,
      "step": 5965
    },
    {
      "epoch": 0.8065636380535535,
      "grad_norm": 0.7885052561759949,
      "learning_rate": 3.089811153384083e-06,
      "loss": 0.051426973193883896,
      "step": 5966
    },
    {
      "epoch": 0.8066988314223187,
      "grad_norm": 0.8790342211723328,
      "learning_rate": 3.08565165637071e-06,
      "loss": 0.04787473380565643,
      "step": 5967
    },
    {
      "epoch": 0.806834024791084,
      "grad_norm": 0.40639930963516235,
      "learning_rate": 3.081494639988196e-06,
      "loss": 0.055586278438568115,
      "step": 5968
    },
    {
      "epoch": 0.8069692181598492,
      "grad_norm": 0.7877558469772339,
      "learning_rate": 3.077340105102057e-06,
      "loss": 0.03896524757146835,
      "step": 5969
    },
    {
      "epoch": 0.8071044115286146,
      "grad_norm": 0.2484983652830124,
      "learning_rate": 3.0731880525772817e-06,
      "loss": 0.04340030997991562,
      "step": 5970
    },
    {
      "epoch": 0.8072396048973798,
      "grad_norm": 0.4748263359069824,
      "learning_rate": 3.069038483278364e-06,
      "loss": 0.04639517143368721,
      "step": 5971
    },
    {
      "epoch": 0.807374798266145,
      "grad_norm": 0.8538673520088196,
      "learning_rate": 3.0648913980692505e-06,
      "loss": 0.06069638952612877,
      "step": 5972
    },
    {
      "epoch": 0.8075099916349103,
      "grad_norm": 0.4544968008995056,
      "learning_rate": 3.0607467978133985e-06,
      "loss": 0.054083552211523056,
      "step": 5973
    },
    {
      "epoch": 0.8076451850036755,
      "grad_norm": 0.32060369849205017,
      "learning_rate": 3.0566046833737294e-06,
      "loss": 0.04942159727215767,
      "step": 5974
    },
    {
      "epoch": 0.8077803783724409,
      "grad_norm": 0.4709891974925995,
      "learning_rate": 3.0524650556126517e-06,
      "loss": 0.0423416793346405,
      "step": 5975
    },
    {
      "epoch": 0.8079155717412061,
      "grad_norm": 0.48734596371650696,
      "learning_rate": 3.048327915392069e-06,
      "loss": 0.053480833768844604,
      "step": 5976
    },
    {
      "epoch": 0.8080507651099713,
      "grad_norm": 0.3904436528682709,
      "learning_rate": 3.044193263573341e-06,
      "loss": 0.051738277077674866,
      "step": 5977
    },
    {
      "epoch": 0.8081859584787366,
      "grad_norm": 1.3868236541748047,
      "learning_rate": 3.0400611010173355e-06,
      "loss": 0.043892428278923035,
      "step": 5978
    },
    {
      "epoch": 0.8083211518475019,
      "grad_norm": 0.4235009253025055,
      "learning_rate": 3.0359314285843863e-06,
      "loss": 0.04972689598798752,
      "step": 5979
    },
    {
      "epoch": 0.8084563452162672,
      "grad_norm": 0.3021641969680786,
      "learning_rate": 3.0318042471343104e-06,
      "loss": 0.03799622505903244,
      "step": 5980
    },
    {
      "epoch": 0.8085915385850324,
      "grad_norm": 0.6612793207168579,
      "learning_rate": 3.027679557526422e-06,
      "loss": 0.05360596999526024,
      "step": 5981
    },
    {
      "epoch": 0.8087267319537976,
      "grad_norm": 0.1806853860616684,
      "learning_rate": 3.0235573606194844e-06,
      "loss": 0.03642776608467102,
      "step": 5982
    },
    {
      "epoch": 0.808861925322563,
      "grad_norm": 0.35930144786834717,
      "learning_rate": 3.0194376572717743e-06,
      "loss": 0.046458661556243896,
      "step": 5983
    },
    {
      "epoch": 0.8089971186913282,
      "grad_norm": 0.8379183411598206,
      "learning_rate": 3.0153204483410318e-06,
      "loss": 0.035362377762794495,
      "step": 5984
    },
    {
      "epoch": 0.8091323120600935,
      "grad_norm": 0.6114612221717834,
      "learning_rate": 3.0112057346844834e-06,
      "loss": 0.07475194334983826,
      "step": 5985
    },
    {
      "epoch": 0.8092675054288587,
      "grad_norm": 0.3169575333595276,
      "learning_rate": 3.007093517158832e-06,
      "loss": 0.04471041262149811,
      "step": 5986
    },
    {
      "epoch": 0.8094026987976239,
      "grad_norm": 0.25989773869514465,
      "learning_rate": 3.002983796620261e-06,
      "loss": 0.03761632740497589,
      "step": 5987
    },
    {
      "epoch": 0.8095378921663893,
      "grad_norm": 0.8605539798736572,
      "learning_rate": 2.9988765739244427e-06,
      "loss": 0.07086487859487534,
      "step": 5988
    },
    {
      "epoch": 0.8096730855351545,
      "grad_norm": 0.46825146675109863,
      "learning_rate": 2.9947718499265197e-06,
      "loss": 0.0600430965423584,
      "step": 5989
    },
    {
      "epoch": 0.8098082789039197,
      "grad_norm": 0.43604162335395813,
      "learning_rate": 2.9906696254811184e-06,
      "loss": 0.07032966613769531,
      "step": 5990
    },
    {
      "epoch": 0.809943472272685,
      "grad_norm": 0.7523314356803894,
      "learning_rate": 2.9865699014423404e-06,
      "loss": 0.06513407081365585,
      "step": 5991
    },
    {
      "epoch": 0.8100786656414503,
      "grad_norm": 0.39125683903694153,
      "learning_rate": 2.9824726786637698e-06,
      "loss": 0.052056439220905304,
      "step": 5992
    },
    {
      "epoch": 0.8102138590102156,
      "grad_norm": 0.4359718859195709,
      "learning_rate": 2.978377957998477e-06,
      "loss": 0.051102060824632645,
      "step": 5993
    },
    {
      "epoch": 0.8103490523789808,
      "grad_norm": 0.5977135896682739,
      "learning_rate": 2.974285740299001e-06,
      "loss": 0.0511966347694397,
      "step": 5994
    },
    {
      "epoch": 0.810484245747746,
      "grad_norm": 0.7350088357925415,
      "learning_rate": 2.9701960264173612e-06,
      "loss": 0.056402504444122314,
      "step": 5995
    },
    {
      "epoch": 0.8106194391165114,
      "grad_norm": 0.6118856072425842,
      "learning_rate": 2.96610881720506e-06,
      "loss": 0.05361482873558998,
      "step": 5996
    },
    {
      "epoch": 0.8107546324852766,
      "grad_norm": 0.5564828515052795,
      "learning_rate": 2.9620241135130715e-06,
      "loss": 0.06175960600376129,
      "step": 5997
    },
    {
      "epoch": 0.8108898258540419,
      "grad_norm": 0.9695990085601807,
      "learning_rate": 2.9579419161918607e-06,
      "loss": 0.07416443526744843,
      "step": 5998
    },
    {
      "epoch": 0.8110250192228071,
      "grad_norm": 0.34099122881889343,
      "learning_rate": 2.9538622260913595e-06,
      "loss": 0.03728532791137695,
      "step": 5999
    },
    {
      "epoch": 0.8111602125915723,
      "grad_norm": 0.31995654106140137,
      "learning_rate": 2.9497850440609814e-06,
      "loss": 0.05529361963272095,
      "step": 6000
    },
    {
      "epoch": 0.8112954059603377,
      "grad_norm": 0.37622174620628357,
      "learning_rate": 2.945710370949616e-06,
      "loss": 0.032759249210357666,
      "step": 6001
    },
    {
      "epoch": 0.8114305993291029,
      "grad_norm": 0.4569580554962158,
      "learning_rate": 2.941638207605629e-06,
      "loss": 0.05487819015979767,
      "step": 6002
    },
    {
      "epoch": 0.8115657926978682,
      "grad_norm": 0.470242440700531,
      "learning_rate": 2.937568554876873e-06,
      "loss": 0.059556424617767334,
      "step": 6003
    },
    {
      "epoch": 0.8117009860666334,
      "grad_norm": 0.6070911288261414,
      "learning_rate": 2.9335014136106704e-06,
      "loss": 0.07529856264591217,
      "step": 6004
    },
    {
      "epoch": 0.8118361794353987,
      "grad_norm": 0.4584585130214691,
      "learning_rate": 2.929436784653818e-06,
      "loss": 0.04798061400651932,
      "step": 6005
    },
    {
      "epoch": 0.811971372804164,
      "grad_norm": 0.5598738193511963,
      "learning_rate": 2.925374668852597e-06,
      "loss": 0.056647613644599915,
      "step": 6006
    },
    {
      "epoch": 0.8121065661729292,
      "grad_norm": 0.5855300426483154,
      "learning_rate": 2.921315067052754e-06,
      "loss": 0.05834600701928139,
      "step": 6007
    },
    {
      "epoch": 0.8122417595416945,
      "grad_norm": 0.5238317847251892,
      "learning_rate": 2.917257980099535e-06,
      "loss": 0.04600805789232254,
      "step": 6008
    },
    {
      "epoch": 0.8123769529104597,
      "grad_norm": 1.0761687755584717,
      "learning_rate": 2.913203408837629e-06,
      "loss": 0.061650726944208145,
      "step": 6009
    },
    {
      "epoch": 0.812512146279225,
      "grad_norm": 0.408304899930954,
      "learning_rate": 2.909151354111232e-06,
      "loss": 0.0393282026052475,
      "step": 6010
    },
    {
      "epoch": 0.8126473396479903,
      "grad_norm": 0.7101040482521057,
      "learning_rate": 2.905101816763998e-06,
      "loss": 0.05413798987865448,
      "step": 6011
    },
    {
      "epoch": 0.8127825330167555,
      "grad_norm": 0.5555978417396545,
      "learning_rate": 2.9010547976390617e-06,
      "loss": 0.048170894384384155,
      "step": 6012
    },
    {
      "epoch": 0.8129177263855208,
      "grad_norm": 0.4768838882446289,
      "learning_rate": 2.897010297579042e-06,
      "loss": 0.04922903701663017,
      "step": 6013
    },
    {
      "epoch": 0.8130529197542861,
      "grad_norm": 0.7568203210830688,
      "learning_rate": 2.8929683174260133e-06,
      "loss": 0.06784403324127197,
      "step": 6014
    },
    {
      "epoch": 0.8131881131230513,
      "grad_norm": 0.35972142219543457,
      "learning_rate": 2.8889288580215467e-06,
      "loss": 0.05512236803770065,
      "step": 6015
    },
    {
      "epoch": 0.8133233064918166,
      "grad_norm": 0.41098833084106445,
      "learning_rate": 2.8848919202066752e-06,
      "loss": 0.05282828211784363,
      "step": 6016
    },
    {
      "epoch": 0.8134584998605818,
      "grad_norm": 0.2256694734096527,
      "learning_rate": 2.8808575048219123e-06,
      "loss": 0.03642760217189789,
      "step": 6017
    },
    {
      "epoch": 0.813593693229347,
      "grad_norm": 0.7187584638595581,
      "learning_rate": 2.8768256127072436e-06,
      "loss": 0.06820935010910034,
      "step": 6018
    },
    {
      "epoch": 0.8137288865981124,
      "grad_norm": 0.4291542172431946,
      "learning_rate": 2.872796244702128e-06,
      "loss": 0.04829160124063492,
      "step": 6019
    },
    {
      "epoch": 0.8138640799668776,
      "grad_norm": 0.33461013436317444,
      "learning_rate": 2.8687694016455075e-06,
      "loss": 0.056207820773124695,
      "step": 6020
    },
    {
      "epoch": 0.8139992733356429,
      "grad_norm": 0.6442399024963379,
      "learning_rate": 2.86474508437579e-06,
      "loss": 0.07606568932533264,
      "step": 6021
    },
    {
      "epoch": 0.8141344667044081,
      "grad_norm": 0.47667205333709717,
      "learning_rate": 2.8607232937308587e-06,
      "loss": 0.06469475477933884,
      "step": 6022
    },
    {
      "epoch": 0.8142696600731734,
      "grad_norm": 0.6076067686080933,
      "learning_rate": 2.856704030548072e-06,
      "loss": 0.054419368505477905,
      "step": 6023
    },
    {
      "epoch": 0.8144048534419387,
      "grad_norm": 1.0293875932693481,
      "learning_rate": 2.8526872956642568e-06,
      "loss": 0.052383869886398315,
      "step": 6024
    },
    {
      "epoch": 0.8145400468107039,
      "grad_norm": 0.38018274307250977,
      "learning_rate": 2.84867308991573e-06,
      "loss": 0.05584040284156799,
      "step": 6025
    },
    {
      "epoch": 0.8146752401794692,
      "grad_norm": 0.3009570240974426,
      "learning_rate": 2.8446614141382638e-06,
      "loss": 0.04549434781074524,
      "step": 6026
    },
    {
      "epoch": 0.8148104335482345,
      "grad_norm": 0.6425711512565613,
      "learning_rate": 2.8406522691671104e-06,
      "loss": 0.04711001366376877,
      "step": 6027
    },
    {
      "epoch": 0.8149456269169997,
      "grad_norm": 0.4095251262187958,
      "learning_rate": 2.8366456558369975e-06,
      "loss": 0.049842387437820435,
      "step": 6028
    },
    {
      "epoch": 0.815080820285765,
      "grad_norm": 1.0910245180130005,
      "learning_rate": 2.8326415749821186e-06,
      "loss": 0.05321177840232849,
      "step": 6029
    },
    {
      "epoch": 0.8152160136545302,
      "grad_norm": 0.35466185212135315,
      "learning_rate": 2.828640027436151e-06,
      "loss": 0.056677818298339844,
      "step": 6030
    },
    {
      "epoch": 0.8153512070232956,
      "grad_norm": 0.5476468801498413,
      "learning_rate": 2.824641014032235e-06,
      "loss": 0.06131923198699951,
      "step": 6031
    },
    {
      "epoch": 0.8154864003920608,
      "grad_norm": 0.5465874075889587,
      "learning_rate": 2.820644535602987e-06,
      "loss": 0.06629450619220734,
      "step": 6032
    },
    {
      "epoch": 0.815621593760826,
      "grad_norm": 0.37514084577560425,
      "learning_rate": 2.8166505929804953e-06,
      "loss": 0.047612980008125305,
      "step": 6033
    },
    {
      "epoch": 0.8157567871295913,
      "grad_norm": 0.8173540830612183,
      "learning_rate": 2.8126591869963163e-06,
      "loss": 0.04677024483680725,
      "step": 6034
    },
    {
      "epoch": 0.8158919804983565,
      "grad_norm": 0.8117858171463013,
      "learning_rate": 2.8086703184814887e-06,
      "loss": 0.05163833126425743,
      "step": 6035
    },
    {
      "epoch": 0.8160271738671219,
      "grad_norm": 0.6131458878517151,
      "learning_rate": 2.8046839882665134e-06,
      "loss": 0.03967595845460892,
      "step": 6036
    },
    {
      "epoch": 0.8161623672358871,
      "grad_norm": 0.2959958016872406,
      "learning_rate": 2.800700197181364e-06,
      "loss": 0.03668498247861862,
      "step": 6037
    },
    {
      "epoch": 0.8162975606046523,
      "grad_norm": 0.2307182252407074,
      "learning_rate": 2.7967189460554876e-06,
      "loss": 0.03201761469244957,
      "step": 6038
    },
    {
      "epoch": 0.8164327539734176,
      "grad_norm": 0.4182891249656677,
      "learning_rate": 2.792740235717801e-06,
      "loss": 0.04801169037818909,
      "step": 6039
    },
    {
      "epoch": 0.8165679473421829,
      "grad_norm": 0.8194049000740051,
      "learning_rate": 2.7887640669967e-06,
      "loss": 0.0409955233335495,
      "step": 6040
    },
    {
      "epoch": 0.8167031407109482,
      "grad_norm": 0.2697567641735077,
      "learning_rate": 2.7847904407200327e-06,
      "loss": 0.03320390731096268,
      "step": 6041
    },
    {
      "epoch": 0.8168383340797134,
      "grad_norm": 0.3574306070804596,
      "learning_rate": 2.7808193577151363e-06,
      "loss": 0.06256964802742004,
      "step": 6042
    },
    {
      "epoch": 0.8169735274484786,
      "grad_norm": 0.41344887018203735,
      "learning_rate": 2.776850818808812e-06,
      "loss": 0.051427166908979416,
      "step": 6043
    },
    {
      "epoch": 0.817108720817244,
      "grad_norm": 0.5584813952445984,
      "learning_rate": 2.772884824827325e-06,
      "loss": 0.04259277135133743,
      "step": 6044
    },
    {
      "epoch": 0.8172439141860092,
      "grad_norm": 0.49408358335494995,
      "learning_rate": 2.768921376596429e-06,
      "loss": 0.057242006063461304,
      "step": 6045
    },
    {
      "epoch": 0.8173791075547744,
      "grad_norm": 0.2824747860431671,
      "learning_rate": 2.7649604749413176e-06,
      "loss": 0.04084141552448273,
      "step": 6046
    },
    {
      "epoch": 0.8175143009235397,
      "grad_norm": 0.7278298139572144,
      "learning_rate": 2.7610021206866837e-06,
      "loss": 0.058346718549728394,
      "step": 6047
    },
    {
      "epoch": 0.8176494942923049,
      "grad_norm": 0.27571889758110046,
      "learning_rate": 2.757046314656676e-06,
      "loss": 0.04878956824541092,
      "step": 6048
    },
    {
      "epoch": 0.8177846876610703,
      "grad_norm": 0.3876435458660126,
      "learning_rate": 2.753093057674909e-06,
      "loss": 0.046020932495594025,
      "step": 6049
    },
    {
      "epoch": 0.8179198810298355,
      "grad_norm": 1.0057417154312134,
      "learning_rate": 2.749142350564483e-06,
      "loss": 0.07525050640106201,
      "step": 6050
    },
    {
      "epoch": 0.8180550743986007,
      "grad_norm": 1.0023913383483887,
      "learning_rate": 2.7451941941479414e-06,
      "loss": 0.04973903298377991,
      "step": 6051
    },
    {
      "epoch": 0.818190267767366,
      "grad_norm": 1.013225793838501,
      "learning_rate": 2.741248589247323e-06,
      "loss": 0.06751750409603119,
      "step": 6052
    },
    {
      "epoch": 0.8183254611361312,
      "grad_norm": 0.3252927362918854,
      "learning_rate": 2.73730553668412e-06,
      "loss": 0.04444827884435654,
      "step": 6053
    },
    {
      "epoch": 0.8184606545048966,
      "grad_norm": 0.43519914150238037,
      "learning_rate": 2.7333650372792978e-06,
      "loss": 0.05890996754169464,
      "step": 6054
    },
    {
      "epoch": 0.8185958478736618,
      "grad_norm": 0.2914881706237793,
      "learning_rate": 2.7294270918532876e-06,
      "loss": 0.04107041284441948,
      "step": 6055
    },
    {
      "epoch": 0.818731041242427,
      "grad_norm": 0.5293598175048828,
      "learning_rate": 2.7254917012259882e-06,
      "loss": 0.06380148231983185,
      "step": 6056
    },
    {
      "epoch": 0.8188662346111923,
      "grad_norm": 0.8536359667778015,
      "learning_rate": 2.721558866216776e-06,
      "loss": 0.056998737156391144,
      "step": 6057
    },
    {
      "epoch": 0.8190014279799576,
      "grad_norm": 0.7351745367050171,
      "learning_rate": 2.7176285876444846e-06,
      "loss": 0.047584764659404755,
      "step": 6058
    },
    {
      "epoch": 0.8191366213487229,
      "grad_norm": 0.3175927996635437,
      "learning_rate": 2.713700866327417e-06,
      "loss": 0.05166204273700714,
      "step": 6059
    },
    {
      "epoch": 0.8192718147174881,
      "grad_norm": 0.220992311835289,
      "learning_rate": 2.7097757030833497e-06,
      "loss": 0.038770660758018494,
      "step": 6060
    },
    {
      "epoch": 0.8194070080862533,
      "grad_norm": 0.3946404457092285,
      "learning_rate": 2.705853098729517e-06,
      "loss": 0.06518112868070602,
      "step": 6061
    },
    {
      "epoch": 0.8195422014550187,
      "grad_norm": 0.3855142593383789,
      "learning_rate": 2.7019330540826325e-06,
      "loss": 0.045286234468221664,
      "step": 6062
    },
    {
      "epoch": 0.8196773948237839,
      "grad_norm": 0.35774752497673035,
      "learning_rate": 2.6980155699588666e-06,
      "loss": 0.05292356014251709,
      "step": 6063
    },
    {
      "epoch": 0.8198125881925492,
      "grad_norm": 1.1996122598648071,
      "learning_rate": 2.6941006471738633e-06,
      "loss": 0.0711471438407898,
      "step": 6064
    },
    {
      "epoch": 0.8199477815613144,
      "grad_norm": 0.8823511004447937,
      "learning_rate": 2.690188286542726e-06,
      "loss": 0.047625068575143814,
      "step": 6065
    },
    {
      "epoch": 0.8200829749300796,
      "grad_norm": 0.5733317136764526,
      "learning_rate": 2.686278488880029e-06,
      "loss": 0.054057564586400986,
      "step": 6066
    },
    {
      "epoch": 0.820218168298845,
      "grad_norm": 1.0579018592834473,
      "learning_rate": 2.6823712549998187e-06,
      "loss": 0.05836092680692673,
      "step": 6067
    },
    {
      "epoch": 0.8203533616676102,
      "grad_norm": 0.37968772649765015,
      "learning_rate": 2.678466585715599e-06,
      "loss": 0.041866935789585114,
      "step": 6068
    },
    {
      "epoch": 0.8204885550363754,
      "grad_norm": 0.3310588300228119,
      "learning_rate": 2.6745644818403426e-06,
      "loss": 0.05284113064408302,
      "step": 6069
    },
    {
      "epoch": 0.8206237484051407,
      "grad_norm": 0.6838980317115784,
      "learning_rate": 2.6706649441864883e-06,
      "loss": 0.046298783272504807,
      "step": 6070
    },
    {
      "epoch": 0.820758941773906,
      "grad_norm": 0.5420961976051331,
      "learning_rate": 2.666767973565937e-06,
      "loss": 0.04984646290540695,
      "step": 6071
    },
    {
      "epoch": 0.8208941351426713,
      "grad_norm": 1.0090419054031372,
      "learning_rate": 2.6628735707900653e-06,
      "loss": 0.06006111577153206,
      "step": 6072
    },
    {
      "epoch": 0.8210293285114365,
      "grad_norm": 0.8741933703422546,
      "learning_rate": 2.658981736669707e-06,
      "loss": 0.04873526841402054,
      "step": 6073
    },
    {
      "epoch": 0.8211645218802017,
      "grad_norm": 1.103743314743042,
      "learning_rate": 2.655092472015161e-06,
      "loss": 0.06706659495830536,
      "step": 6074
    },
    {
      "epoch": 0.821299715248967,
      "grad_norm": 0.3980848491191864,
      "learning_rate": 2.6512057776361935e-06,
      "loss": 0.0645224079489708,
      "step": 6075
    },
    {
      "epoch": 0.8214349086177323,
      "grad_norm": 0.5778664946556091,
      "learning_rate": 2.64732165434203e-06,
      "loss": 0.04888932406902313,
      "step": 6076
    },
    {
      "epoch": 0.8215701019864976,
      "grad_norm": 1.1038556098937988,
      "learning_rate": 2.6434401029413792e-06,
      "loss": 0.05492448806762695,
      "step": 6077
    },
    {
      "epoch": 0.8217052953552628,
      "grad_norm": 0.45790040493011475,
      "learning_rate": 2.639561124242385e-06,
      "loss": 0.053073152899742126,
      "step": 6078
    },
    {
      "epoch": 0.821840488724028,
      "grad_norm": 0.4845079183578491,
      "learning_rate": 2.635684719052682e-06,
      "loss": 0.0440652072429657,
      "step": 6079
    },
    {
      "epoch": 0.8219756820927934,
      "grad_norm": 0.5969334244728088,
      "learning_rate": 2.631810888179355e-06,
      "loss": 0.06416060030460358,
      "step": 6080
    },
    {
      "epoch": 0.8221108754615586,
      "grad_norm": 0.46688759326934814,
      "learning_rate": 2.627939632428952e-06,
      "loss": 0.038395948708057404,
      "step": 6081
    },
    {
      "epoch": 0.8222460688303239,
      "grad_norm": 0.8864238858222961,
      "learning_rate": 2.624070952607502e-06,
      "loss": 0.04338468983769417,
      "step": 6082
    },
    {
      "epoch": 0.8223812621990891,
      "grad_norm": 0.7390619516372681,
      "learning_rate": 2.620204849520468e-06,
      "loss": 0.06229623034596443,
      "step": 6083
    },
    {
      "epoch": 0.8225164555678544,
      "grad_norm": 0.5434011816978455,
      "learning_rate": 2.616341323972806e-06,
      "loss": 0.04871061444282532,
      "step": 6084
    },
    {
      "epoch": 0.8226516489366197,
      "grad_norm": 0.4083769917488098,
      "learning_rate": 2.612480376768917e-06,
      "loss": 0.033784568309783936,
      "step": 6085
    },
    {
      "epoch": 0.8227868423053849,
      "grad_norm": 0.4263850748538971,
      "learning_rate": 2.608622008712672e-06,
      "loss": 0.05414266884326935,
      "step": 6086
    },
    {
      "epoch": 0.8229220356741502,
      "grad_norm": 0.5287026166915894,
      "learning_rate": 2.6047662206074034e-06,
      "loss": 0.04452309384942055,
      "step": 6087
    },
    {
      "epoch": 0.8230572290429155,
      "grad_norm": 0.36066341400146484,
      "learning_rate": 2.600913013255904e-06,
      "loss": 0.05635611712932587,
      "step": 6088
    },
    {
      "epoch": 0.8231924224116807,
      "grad_norm": 0.6323385238647461,
      "learning_rate": 2.59706238746044e-06,
      "loss": 0.055132389068603516,
      "step": 6089
    },
    {
      "epoch": 0.823327615780446,
      "grad_norm": 0.2682657241821289,
      "learning_rate": 2.593214344022725e-06,
      "loss": 0.028467580676078796,
      "step": 6090
    },
    {
      "epoch": 0.8234628091492112,
      "grad_norm": 0.6920700669288635,
      "learning_rate": 2.5893688837439474e-06,
      "loss": 0.06494811177253723,
      "step": 6091
    },
    {
      "epoch": 0.8235980025179765,
      "grad_norm": 0.49712565541267395,
      "learning_rate": 2.5855260074247473e-06,
      "loss": 0.05038559436798096,
      "step": 6092
    },
    {
      "epoch": 0.8237331958867418,
      "grad_norm": 0.48757442831993103,
      "learning_rate": 2.581685715865232e-06,
      "loss": 0.039091311395168304,
      "step": 6093
    },
    {
      "epoch": 0.823868389255507,
      "grad_norm": 0.8614658713340759,
      "learning_rate": 2.5778480098649766e-06,
      "loss": 0.06300951540470123,
      "step": 6094
    },
    {
      "epoch": 0.8240035826242723,
      "grad_norm": 0.33780935406684875,
      "learning_rate": 2.5740128902230087e-06,
      "loss": 0.03624337911605835,
      "step": 6095
    },
    {
      "epoch": 0.8241387759930375,
      "grad_norm": 0.42048728466033936,
      "learning_rate": 2.5701803577378214e-06,
      "loss": 0.054029226303100586,
      "step": 6096
    },
    {
      "epoch": 0.8242739693618027,
      "grad_norm": 1.1494966745376587,
      "learning_rate": 2.566350413207366e-06,
      "loss": 0.07297086715698242,
      "step": 6097
    },
    {
      "epoch": 0.8244091627305681,
      "grad_norm": 0.5856238603591919,
      "learning_rate": 2.5625230574290554e-06,
      "loss": 0.05555553734302521,
      "step": 6098
    },
    {
      "epoch": 0.8245443560993333,
      "grad_norm": 0.3982914984226227,
      "learning_rate": 2.558698291199773e-06,
      "loss": 0.03345699608325958,
      "step": 6099
    },
    {
      "epoch": 0.8246795494680986,
      "grad_norm": 0.40843665599823,
      "learning_rate": 2.5548761153158524e-06,
      "loss": 0.0372922420501709,
      "step": 6100
    },
    {
      "epoch": 0.8248147428368638,
      "grad_norm": 0.4422677755355835,
      "learning_rate": 2.55105653057309e-06,
      "loss": 0.05058560147881508,
      "step": 6101
    },
    {
      "epoch": 0.8249499362056291,
      "grad_norm": 0.8284547924995422,
      "learning_rate": 2.547239537766743e-06,
      "loss": 0.06245347112417221,
      "step": 6102
    },
    {
      "epoch": 0.8250851295743944,
      "grad_norm": 0.3782658278942108,
      "learning_rate": 2.543425137691526e-06,
      "loss": 0.05809428542852402,
      "step": 6103
    },
    {
      "epoch": 0.8252203229431596,
      "grad_norm": 0.4869481027126312,
      "learning_rate": 2.5396133311416264e-06,
      "loss": 0.03994176536798477,
      "step": 6104
    },
    {
      "epoch": 0.8253555163119249,
      "grad_norm": 0.7174405455589294,
      "learning_rate": 2.5358041189106784e-06,
      "loss": 0.05265609174966812,
      "step": 6105
    },
    {
      "epoch": 0.8254907096806902,
      "grad_norm": 0.548949658870697,
      "learning_rate": 2.531997501791779e-06,
      "loss": 0.053619205951690674,
      "step": 6106
    },
    {
      "epoch": 0.8256259030494554,
      "grad_norm": 0.6912299394607544,
      "learning_rate": 2.528193480577489e-06,
      "loss": 0.037102386355400085,
      "step": 6107
    },
    {
      "epoch": 0.8257610964182207,
      "grad_norm": 0.5926231145858765,
      "learning_rate": 2.5243920560598186e-06,
      "loss": 0.07208763062953949,
      "step": 6108
    },
    {
      "epoch": 0.8258962897869859,
      "grad_norm": 0.6693443655967712,
      "learning_rate": 2.5205932290302598e-06,
      "loss": 0.0516146719455719,
      "step": 6109
    },
    {
      "epoch": 0.8260314831557513,
      "grad_norm": 0.368770569562912,
      "learning_rate": 2.516797000279729e-06,
      "loss": 0.04051916301250458,
      "step": 6110
    },
    {
      "epoch": 0.8261666765245165,
      "grad_norm": 0.4633229076862335,
      "learning_rate": 2.513003370598637e-06,
      "loss": 0.058559875935316086,
      "step": 6111
    },
    {
      "epoch": 0.8263018698932817,
      "grad_norm": 0.4610815644264221,
      "learning_rate": 2.509212340776832e-06,
      "loss": 0.057527586817741394,
      "step": 6112
    },
    {
      "epoch": 0.826437063262047,
      "grad_norm": 0.30278053879737854,
      "learning_rate": 2.505423911603622e-06,
      "loss": 0.03435998409986496,
      "step": 6113
    },
    {
      "epoch": 0.8265722566308122,
      "grad_norm": 0.5012966990470886,
      "learning_rate": 2.501638083867789e-06,
      "loss": 0.042091481387615204,
      "step": 6114
    },
    {
      "epoch": 0.8267074499995776,
      "grad_norm": 0.6133097410202026,
      "learning_rate": 2.497854858357552e-06,
      "loss": 0.051196541637182236,
      "step": 6115
    },
    {
      "epoch": 0.8268426433683428,
      "grad_norm": 0.6618172526359558,
      "learning_rate": 2.494074235860604e-06,
      "loss": 0.05097391456365585,
      "step": 6116
    },
    {
      "epoch": 0.826977836737108,
      "grad_norm": 0.31498780846595764,
      "learning_rate": 2.4902962171640913e-06,
      "loss": 0.04439883306622505,
      "step": 6117
    },
    {
      "epoch": 0.8271130301058733,
      "grad_norm": 0.9481056928634644,
      "learning_rate": 2.4865208030546167e-06,
      "loss": 0.0532616525888443,
      "step": 6118
    },
    {
      "epoch": 0.8272482234746386,
      "grad_norm": 0.4932495355606079,
      "learning_rate": 2.482747994318239e-06,
      "loss": 0.04085560515522957,
      "step": 6119
    },
    {
      "epoch": 0.8273834168434039,
      "grad_norm": 0.49863311648368835,
      "learning_rate": 2.478977791740477e-06,
      "loss": 0.06760723143815994,
      "step": 6120
    },
    {
      "epoch": 0.8275186102121691,
      "grad_norm": 0.5750812888145447,
      "learning_rate": 2.475210196106313e-06,
      "loss": 0.06401936709880829,
      "step": 6121
    },
    {
      "epoch": 0.8276538035809343,
      "grad_norm": 0.8845735788345337,
      "learning_rate": 2.4714452082001753e-06,
      "loss": 0.05270498991012573,
      "step": 6122
    },
    {
      "epoch": 0.8277889969496997,
      "grad_norm": 0.3728500008583069,
      "learning_rate": 2.467682828805956e-06,
      "loss": 0.05247960239648819,
      "step": 6123
    },
    {
      "epoch": 0.8279241903184649,
      "grad_norm": 1.0556254386901855,
      "learning_rate": 2.4639230587070017e-06,
      "loss": 0.06627139449119568,
      "step": 6124
    },
    {
      "epoch": 0.8280593836872301,
      "grad_norm": 1.4212722778320312,
      "learning_rate": 2.460165898686114e-06,
      "loss": 0.07075628638267517,
      "step": 6125
    },
    {
      "epoch": 0.8281945770559954,
      "grad_norm": 0.450106680393219,
      "learning_rate": 2.4564113495255597e-06,
      "loss": 0.052923545241355896,
      "step": 6126
    },
    {
      "epoch": 0.8283297704247606,
      "grad_norm": 1.2226994037628174,
      "learning_rate": 2.4526594120070545e-06,
      "loss": 0.06417983770370483,
      "step": 6127
    },
    {
      "epoch": 0.828464963793526,
      "grad_norm": 0.6285002827644348,
      "learning_rate": 2.4489100869117686e-06,
      "loss": 0.0484108105301857,
      "step": 6128
    },
    {
      "epoch": 0.8286001571622912,
      "grad_norm": 0.5206980109214783,
      "learning_rate": 2.4451633750203344e-06,
      "loss": 0.06516458094120026,
      "step": 6129
    },
    {
      "epoch": 0.8287353505310564,
      "grad_norm": 0.5448839664459229,
      "learning_rate": 2.441419277112831e-06,
      "loss": 0.05673800781369209,
      "step": 6130
    },
    {
      "epoch": 0.8288705438998217,
      "grad_norm": 0.5945592522621155,
      "learning_rate": 2.4376777939688107e-06,
      "loss": 0.05581730604171753,
      "step": 6131
    },
    {
      "epoch": 0.829005737268587,
      "grad_norm": 0.329216867685318,
      "learning_rate": 2.4339389263672625e-06,
      "loss": 0.04142392426729202,
      "step": 6132
    },
    {
      "epoch": 0.8291409306373523,
      "grad_norm": 0.616303026676178,
      "learning_rate": 2.4302026750866406e-06,
      "loss": 0.04276688024401665,
      "step": 6133
    },
    {
      "epoch": 0.8292761240061175,
      "grad_norm": 0.8358241319656372,
      "learning_rate": 2.4264690409048517e-06,
      "loss": 0.05740465223789215,
      "step": 6134
    },
    {
      "epoch": 0.8294113173748827,
      "grad_norm": 1.3148711919784546,
      "learning_rate": 2.4227380245992555e-06,
      "loss": 0.057329535484313965,
      "step": 6135
    },
    {
      "epoch": 0.829546510743648,
      "grad_norm": 0.5319249033927917,
      "learning_rate": 2.4190096269466767e-06,
      "loss": 0.044739410281181335,
      "step": 6136
    },
    {
      "epoch": 0.8296817041124133,
      "grad_norm": 0.5065696239471436,
      "learning_rate": 2.415283848723383e-06,
      "loss": 0.04829120635986328,
      "step": 6137
    },
    {
      "epoch": 0.8298168974811786,
      "grad_norm": 0.7488718032836914,
      "learning_rate": 2.411560690705101e-06,
      "loss": 0.054521623998880386,
      "step": 6138
    },
    {
      "epoch": 0.8299520908499438,
      "grad_norm": 0.23163874447345734,
      "learning_rate": 2.4078401536670146e-06,
      "loss": 0.04687364771962166,
      "step": 6139
    },
    {
      "epoch": 0.830087284218709,
      "grad_norm": 0.9627823233604431,
      "learning_rate": 2.4041222383837538e-06,
      "loss": 0.06781119108200073,
      "step": 6140
    },
    {
      "epoch": 0.8302224775874744,
      "grad_norm": 2.258822441101074,
      "learning_rate": 2.400406945629418e-06,
      "loss": 0.07293403148651123,
      "step": 6141
    },
    {
      "epoch": 0.8303576709562396,
      "grad_norm": 1.251961588859558,
      "learning_rate": 2.3966942761775396e-06,
      "loss": 0.06108919158577919,
      "step": 6142
    },
    {
      "epoch": 0.8304928643250049,
      "grad_norm": 1.0319828987121582,
      "learning_rate": 2.3929842308011263e-06,
      "loss": 0.05969039350748062,
      "step": 6143
    },
    {
      "epoch": 0.8306280576937701,
      "grad_norm": 0.6500529646873474,
      "learning_rate": 2.3892768102726236e-06,
      "loss": 0.03957810252904892,
      "step": 6144
    },
    {
      "epoch": 0.8307632510625353,
      "grad_norm": 0.4057830274105072,
      "learning_rate": 2.3855720153639344e-06,
      "loss": 0.06209491193294525,
      "step": 6145
    },
    {
      "epoch": 0.8308984444313007,
      "grad_norm": 0.8404145836830139,
      "learning_rate": 2.381869846846428e-06,
      "loss": 0.05145217478275299,
      "step": 6146
    },
    {
      "epoch": 0.8310336378000659,
      "grad_norm": 0.3667839467525482,
      "learning_rate": 2.3781703054908993e-06,
      "loss": 0.04141956567764282,
      "step": 6147
    },
    {
      "epoch": 0.8311688311688312,
      "grad_norm": 0.48730531334877014,
      "learning_rate": 2.374473392067624e-06,
      "loss": 0.050588108599185944,
      "step": 6148
    },
    {
      "epoch": 0.8313040245375964,
      "grad_norm": 0.5241788029670715,
      "learning_rate": 2.370779107346317e-06,
      "loss": 0.04770607873797417,
      "step": 6149
    },
    {
      "epoch": 0.8314392179063617,
      "grad_norm": 0.5279482007026672,
      "learning_rate": 2.3670874520961437e-06,
      "loss": 0.055301908403635025,
      "step": 6150
    },
    {
      "epoch": 0.831574411275127,
      "grad_norm": 0.5519096851348877,
      "learning_rate": 2.3633984270857367e-06,
      "loss": 0.04346519708633423,
      "step": 6151
    },
    {
      "epoch": 0.8317096046438922,
      "grad_norm": 0.3969717025756836,
      "learning_rate": 2.359712033083156e-06,
      "loss": 0.0380130298435688,
      "step": 6152
    },
    {
      "epoch": 0.8318447980126574,
      "grad_norm": 0.34964194893836975,
      "learning_rate": 2.35602827085594e-06,
      "loss": 0.04213812202215195,
      "step": 6153
    },
    {
      "epoch": 0.8319799913814228,
      "grad_norm": 0.6710175275802612,
      "learning_rate": 2.3523471411710644e-06,
      "loss": 0.05972909927368164,
      "step": 6154
    },
    {
      "epoch": 0.832115184750188,
      "grad_norm": 0.28571468591690063,
      "learning_rate": 2.3486686447949585e-06,
      "loss": 0.043431565165519714,
      "step": 6155
    },
    {
      "epoch": 0.8322503781189533,
      "grad_norm": 0.6187041401863098,
      "learning_rate": 2.3449927824935075e-06,
      "loss": 0.04155975952744484,
      "step": 6156
    },
    {
      "epoch": 0.8323855714877185,
      "grad_norm": 0.6864646673202515,
      "learning_rate": 2.3413195550320393e-06,
      "loss": 0.047659337520599365,
      "step": 6157
    },
    {
      "epoch": 0.8325207648564837,
      "grad_norm": 0.5197789072990417,
      "learning_rate": 2.3376489631753474e-06,
      "loss": 0.06052897870540619,
      "step": 6158
    },
    {
      "epoch": 0.8326559582252491,
      "grad_norm": 0.4447621703147888,
      "learning_rate": 2.3339810076876665e-06,
      "loss": 0.043431296944618225,
      "step": 6159
    },
    {
      "epoch": 0.8327911515940143,
      "grad_norm": 1.0718086957931519,
      "learning_rate": 2.3303156893326815e-06,
      "loss": 0.0609980933368206,
      "step": 6160
    },
    {
      "epoch": 0.8329263449627796,
      "grad_norm": 0.37900206446647644,
      "learning_rate": 2.326653008873535e-06,
      "loss": 0.04300892353057861,
      "step": 6161
    },
    {
      "epoch": 0.8330615383315448,
      "grad_norm": 0.49139440059661865,
      "learning_rate": 2.3229929670728085e-06,
      "loss": 0.05007103830575943,
      "step": 6162
    },
    {
      "epoch": 0.8331967317003101,
      "grad_norm": 1.430359959602356,
      "learning_rate": 2.319335564692554e-06,
      "loss": 0.05631953850388527,
      "step": 6163
    },
    {
      "epoch": 0.8333319250690754,
      "grad_norm": 0.28598642349243164,
      "learning_rate": 2.315680802494256e-06,
      "loss": 0.056486576795578,
      "step": 6164
    },
    {
      "epoch": 0.8334671184378406,
      "grad_norm": 0.5973681807518005,
      "learning_rate": 2.312028681238856e-06,
      "loss": 0.058661580085754395,
      "step": 6165
    },
    {
      "epoch": 0.8336023118066059,
      "grad_norm": 0.4279343783855438,
      "learning_rate": 2.3083792016867434e-06,
      "loss": 0.04404953122138977,
      "step": 6166
    },
    {
      "epoch": 0.8337375051753712,
      "grad_norm": 0.3314683139324188,
      "learning_rate": 2.304732364597759e-06,
      "loss": 0.0457480326294899,
      "step": 6167
    },
    {
      "epoch": 0.8338726985441364,
      "grad_norm": 0.2974388301372528,
      "learning_rate": 2.3010881707311994e-06,
      "loss": 0.04642082750797272,
      "step": 6168
    },
    {
      "epoch": 0.8340078919129017,
      "grad_norm": 0.3039360046386719,
      "learning_rate": 2.2974466208458017e-06,
      "loss": 0.037277087569236755,
      "step": 6169
    },
    {
      "epoch": 0.8341430852816669,
      "grad_norm": 0.2541879713535309,
      "learning_rate": 2.293807715699755e-06,
      "loss": 0.04402574896812439,
      "step": 6170
    },
    {
      "epoch": 0.8342782786504322,
      "grad_norm": 1.1918162107467651,
      "learning_rate": 2.2901714560507e-06,
      "loss": 0.06334730237722397,
      "step": 6171
    },
    {
      "epoch": 0.8344134720191975,
      "grad_norm": 0.3717571794986725,
      "learning_rate": 2.286537842655722e-06,
      "loss": 0.044879794120788574,
      "step": 6172
    },
    {
      "epoch": 0.8345486653879627,
      "grad_norm": 0.47871437668800354,
      "learning_rate": 2.2829068762713633e-06,
      "loss": 0.05879815295338631,
      "step": 6173
    },
    {
      "epoch": 0.834683858756728,
      "grad_norm": 0.2037031650543213,
      "learning_rate": 2.279278557653611e-06,
      "loss": 0.037516482174396515,
      "step": 6174
    },
    {
      "epoch": 0.8348190521254932,
      "grad_norm": 0.6012883186340332,
      "learning_rate": 2.2756528875578965e-06,
      "loss": 0.042451538145542145,
      "step": 6175
    },
    {
      "epoch": 0.8349542454942585,
      "grad_norm": 0.6637305617332458,
      "learning_rate": 2.2720298667391067e-06,
      "loss": 0.04357580468058586,
      "step": 6176
    },
    {
      "epoch": 0.8350894388630238,
      "grad_norm": 0.4367183744907379,
      "learning_rate": 2.268409495951568e-06,
      "loss": 0.04872041940689087,
      "step": 6177
    },
    {
      "epoch": 0.835224632231789,
      "grad_norm": 0.7653933167457581,
      "learning_rate": 2.2647917759490723e-06,
      "loss": 0.05078501999378204,
      "step": 6178
    },
    {
      "epoch": 0.8353598256005543,
      "grad_norm": 0.5071462392807007,
      "learning_rate": 2.261176707484834e-06,
      "loss": 0.047454267740249634,
      "step": 6179
    },
    {
      "epoch": 0.8354950189693195,
      "grad_norm": 0.5136466026306152,
      "learning_rate": 2.2575642913115408e-06,
      "loss": 0.04026796668767929,
      "step": 6180
    },
    {
      "epoch": 0.8356302123380848,
      "grad_norm": 0.4026106297969818,
      "learning_rate": 2.253954528181313e-06,
      "loss": 0.04408586025238037,
      "step": 6181
    },
    {
      "epoch": 0.8357654057068501,
      "grad_norm": 1.0115156173706055,
      "learning_rate": 2.2503474188457206e-06,
      "loss": 0.04224462807178497,
      "step": 6182
    },
    {
      "epoch": 0.8359005990756153,
      "grad_norm": 0.7170173525810242,
      "learning_rate": 2.2467429640557903e-06,
      "loss": 0.04062668979167938,
      "step": 6183
    },
    {
      "epoch": 0.8360357924443806,
      "grad_norm": 0.32150521874427795,
      "learning_rate": 2.2431411645619776e-06,
      "loss": 0.04180286452174187,
      "step": 6184
    },
    {
      "epoch": 0.8361709858131459,
      "grad_norm": 0.9162803292274475,
      "learning_rate": 2.239542021114205e-06,
      "loss": 0.05110998451709747,
      "step": 6185
    },
    {
      "epoch": 0.8363061791819111,
      "grad_norm": 0.691806435585022,
      "learning_rate": 2.2359455344618306e-06,
      "loss": 0.0493222177028656,
      "step": 6186
    },
    {
      "epoch": 0.8364413725506764,
      "grad_norm": 0.31305885314941406,
      "learning_rate": 2.232351705353663e-06,
      "loss": 0.051948729902505875,
      "step": 6187
    },
    {
      "epoch": 0.8365765659194416,
      "grad_norm": 0.26568931341171265,
      "learning_rate": 2.228760534537955e-06,
      "loss": 0.035367682576179504,
      "step": 6188
    },
    {
      "epoch": 0.836711759288207,
      "grad_norm": 0.3882652521133423,
      "learning_rate": 2.2251720227624044e-06,
      "loss": 0.051023002713918686,
      "step": 6189
    },
    {
      "epoch": 0.8368469526569722,
      "grad_norm": 0.4140540063381195,
      "learning_rate": 2.2215861707741666e-06,
      "loss": 0.03535125404596329,
      "step": 6190
    },
    {
      "epoch": 0.8369821460257374,
      "grad_norm": 0.28733938932418823,
      "learning_rate": 2.2180029793198313e-06,
      "loss": 0.042948298156261444,
      "step": 6191
    },
    {
      "epoch": 0.8371173393945027,
      "grad_norm": 0.3373543620109558,
      "learning_rate": 2.2144224491454363e-06,
      "loss": 0.051349759101867676,
      "step": 6192
    },
    {
      "epoch": 0.8372525327632679,
      "grad_norm": 0.35262036323547363,
      "learning_rate": 2.2108445809964695e-06,
      "loss": 0.04738650470972061,
      "step": 6193
    },
    {
      "epoch": 0.8373877261320333,
      "grad_norm": 0.3689965605735779,
      "learning_rate": 2.2072693756178567e-06,
      "loss": 0.042664796113967896,
      "step": 6194
    },
    {
      "epoch": 0.8375229195007985,
      "grad_norm": 0.9979879856109619,
      "learning_rate": 2.203696833753983e-06,
      "loss": 0.05222224444150925,
      "step": 6195
    },
    {
      "epoch": 0.8376581128695637,
      "grad_norm": 0.35647955536842346,
      "learning_rate": 2.200126956148668e-06,
      "loss": 0.04531312361359596,
      "step": 6196
    },
    {
      "epoch": 0.837793306238329,
      "grad_norm": 0.357084184885025,
      "learning_rate": 2.196559743545177e-06,
      "loss": 0.046202752739191055,
      "step": 6197
    },
    {
      "epoch": 0.8379284996070943,
      "grad_norm": 1.526036262512207,
      "learning_rate": 2.1929951966862233e-06,
      "loss": 0.07071888446807861,
      "step": 6198
    },
    {
      "epoch": 0.8380636929758596,
      "grad_norm": 0.3560903072357178,
      "learning_rate": 2.1894333163139607e-06,
      "loss": 0.031010091304779053,
      "step": 6199
    },
    {
      "epoch": 0.8381988863446248,
      "grad_norm": 0.2677477300167084,
      "learning_rate": 2.1858741031700015e-06,
      "loss": 0.03647442162036896,
      "step": 6200
    },
    {
      "epoch": 0.83833407971339,
      "grad_norm": 1.0474574565887451,
      "learning_rate": 2.1823175579953856e-06,
      "loss": 0.07177795469760895,
      "step": 6201
    },
    {
      "epoch": 0.8384692730821554,
      "grad_norm": 1.179814338684082,
      "learning_rate": 2.1787636815306065e-06,
      "loss": 0.05104741454124451,
      "step": 6202
    },
    {
      "epoch": 0.8386044664509206,
      "grad_norm": 0.4671471118927002,
      "learning_rate": 2.1752124745156005e-06,
      "loss": 0.042914584279060364,
      "step": 6203
    },
    {
      "epoch": 0.8387396598196858,
      "grad_norm": 0.37799084186553955,
      "learning_rate": 2.171663937689744e-06,
      "loss": 0.03742415830492973,
      "step": 6204
    },
    {
      "epoch": 0.8388748531884511,
      "grad_norm": 0.3980090320110321,
      "learning_rate": 2.168118071791868e-06,
      "loss": 0.059833161532878876,
      "step": 6205
    },
    {
      "epoch": 0.8390100465572163,
      "grad_norm": 0.6616215705871582,
      "learning_rate": 2.164574877560237e-06,
      "loss": 0.04415488988161087,
      "step": 6206
    },
    {
      "epoch": 0.8391452399259817,
      "grad_norm": 0.22867345809936523,
      "learning_rate": 2.161034355732564e-06,
      "loss": 0.037374675273895264,
      "step": 6207
    },
    {
      "epoch": 0.8392804332947469,
      "grad_norm": 0.7260293364524841,
      "learning_rate": 2.1574965070460047e-06,
      "loss": 0.032028086483478546,
      "step": 6208
    },
    {
      "epoch": 0.8394156266635121,
      "grad_norm": 0.36259400844573975,
      "learning_rate": 2.1539613322371527e-06,
      "loss": 0.04549260810017586,
      "step": 6209
    },
    {
      "epoch": 0.8395508200322774,
      "grad_norm": 0.4727924168109894,
      "learning_rate": 2.1504288320420613e-06,
      "loss": 0.03847412019968033,
      "step": 6210
    },
    {
      "epoch": 0.8396860134010427,
      "grad_norm": 1.1600937843322754,
      "learning_rate": 2.1468990071962038e-06,
      "loss": 0.0547519326210022,
      "step": 6211
    },
    {
      "epoch": 0.839821206769808,
      "grad_norm": 0.4148150682449341,
      "learning_rate": 2.143371858434515e-06,
      "loss": 0.04619329050183296,
      "step": 6212
    },
    {
      "epoch": 0.8399564001385732,
      "grad_norm": 0.7472928762435913,
      "learning_rate": 2.139847386491367e-06,
      "loss": 0.03591097146272659,
      "step": 6213
    },
    {
      "epoch": 0.8400915935073384,
      "grad_norm": 0.4984283745288849,
      "learning_rate": 2.1363255921005685e-06,
      "loss": 0.057893648743629456,
      "step": 6214
    },
    {
      "epoch": 0.8402267868761037,
      "grad_norm": 0.4796014130115509,
      "learning_rate": 2.1328064759953853e-06,
      "loss": 0.05514446645975113,
      "step": 6215
    },
    {
      "epoch": 0.840361980244869,
      "grad_norm": 0.6901918053627014,
      "learning_rate": 2.129290038908504e-06,
      "loss": 0.03864287957549095,
      "step": 6216
    },
    {
      "epoch": 0.8404971736136343,
      "grad_norm": 0.3759935200214386,
      "learning_rate": 2.1257762815720745e-06,
      "loss": 0.05031343176960945,
      "step": 6217
    },
    {
      "epoch": 0.8406323669823995,
      "grad_norm": 0.801692008972168,
      "learning_rate": 2.122265204717678e-06,
      "loss": 0.05881436914205551,
      "step": 6218
    },
    {
      "epoch": 0.8407675603511647,
      "grad_norm": 0.7840434908866882,
      "learning_rate": 2.1187568090763328e-06,
      "loss": 0.07795971632003784,
      "step": 6219
    },
    {
      "epoch": 0.8409027537199301,
      "grad_norm": 0.6142735481262207,
      "learning_rate": 2.1152510953785196e-06,
      "loss": 0.049477964639663696,
      "step": 6220
    },
    {
      "epoch": 0.8410379470886953,
      "grad_norm": 0.4441145658493042,
      "learning_rate": 2.1117480643541304e-06,
      "loss": 0.05546906590461731,
      "step": 6221
    },
    {
      "epoch": 0.8411731404574606,
      "grad_norm": 0.578981339931488,
      "learning_rate": 2.1082477167325275e-06,
      "loss": 0.050517674535512924,
      "step": 6222
    },
    {
      "epoch": 0.8413083338262258,
      "grad_norm": 0.5397024154663086,
      "learning_rate": 2.1047500532424968e-06,
      "loss": 0.05547235906124115,
      "step": 6223
    },
    {
      "epoch": 0.841443527194991,
      "grad_norm": 1.2395414113998413,
      "learning_rate": 2.1012550746122705e-06,
      "loss": 0.056867048144340515,
      "step": 6224
    },
    {
      "epoch": 0.8415787205637564,
      "grad_norm": 0.7233183979988098,
      "learning_rate": 2.0977627815695217e-06,
      "loss": 0.054838210344314575,
      "step": 6225
    },
    {
      "epoch": 0.8417139139325216,
      "grad_norm": 0.7362427115440369,
      "learning_rate": 2.094273174841362e-06,
      "loss": 0.05307919532060623,
      "step": 6226
    },
    {
      "epoch": 0.8418491073012869,
      "grad_norm": 0.5328913927078247,
      "learning_rate": 2.0907862551543516e-06,
      "loss": 0.043626755475997925,
      "step": 6227
    },
    {
      "epoch": 0.8419843006700521,
      "grad_norm": 0.2717840373516083,
      "learning_rate": 2.087302023234485e-06,
      "loss": 0.037222929298877716,
      "step": 6228
    },
    {
      "epoch": 0.8421194940388174,
      "grad_norm": 0.26456549763679504,
      "learning_rate": 2.083820479807194e-06,
      "loss": 0.03553616255521774,
      "step": 6229
    },
    {
      "epoch": 0.8422546874075827,
      "grad_norm": 0.36989766359329224,
      "learning_rate": 2.0803416255973585e-06,
      "loss": 0.041608355939388275,
      "step": 6230
    },
    {
      "epoch": 0.8423898807763479,
      "grad_norm": 0.3028639853000641,
      "learning_rate": 2.0768654613292887e-06,
      "loss": 0.05160258710384369,
      "step": 6231
    },
    {
      "epoch": 0.8425250741451131,
      "grad_norm": 0.9835982918739319,
      "learning_rate": 2.0733919877267477e-06,
      "loss": 0.045955684036016464,
      "step": 6232
    },
    {
      "epoch": 0.8426602675138785,
      "grad_norm": 0.4995516240596771,
      "learning_rate": 2.0699212055129268e-06,
      "loss": 0.048937439918518066,
      "step": 6233
    },
    {
      "epoch": 0.8427954608826437,
      "grad_norm": 0.32965973019599915,
      "learning_rate": 2.066453115410463e-06,
      "loss": 0.041056472808122635,
      "step": 6234
    },
    {
      "epoch": 0.842930654251409,
      "grad_norm": 0.6493692398071289,
      "learning_rate": 2.062987718141431e-06,
      "loss": 0.046764492988586426,
      "step": 6235
    },
    {
      "epoch": 0.8430658476201742,
      "grad_norm": 0.3514888286590576,
      "learning_rate": 2.0595250144273423e-06,
      "loss": 0.03547742962837219,
      "step": 6236
    },
    {
      "epoch": 0.8432010409889394,
      "grad_norm": 0.2799179255962372,
      "learning_rate": 2.056065004989155e-06,
      "loss": 0.04868411645293236,
      "step": 6237
    },
    {
      "epoch": 0.8433362343577048,
      "grad_norm": 0.4646119475364685,
      "learning_rate": 2.0526076905472585e-06,
      "loss": 0.05488266050815582,
      "step": 6238
    },
    {
      "epoch": 0.84347142772647,
      "grad_norm": 0.7049408555030823,
      "learning_rate": 2.0491530718214855e-06,
      "loss": 0.04533543437719345,
      "step": 6239
    },
    {
      "epoch": 0.8436066210952353,
      "grad_norm": 0.25155574083328247,
      "learning_rate": 2.0457011495311045e-06,
      "loss": 0.05469845235347748,
      "step": 6240
    },
    {
      "epoch": 0.8437418144640005,
      "grad_norm": 0.5906786322593689,
      "learning_rate": 2.0422519243948232e-06,
      "loss": 0.07562047243118286,
      "step": 6241
    },
    {
      "epoch": 0.8438770078327658,
      "grad_norm": 0.6755495071411133,
      "learning_rate": 2.0388053971307927e-06,
      "loss": 0.03171085566282272,
      "step": 6242
    },
    {
      "epoch": 0.8440122012015311,
      "grad_norm": 0.6171776652336121,
      "learning_rate": 2.0353615684565956e-06,
      "loss": 0.06633298099040985,
      "step": 6243
    },
    {
      "epoch": 0.8441473945702963,
      "grad_norm": 0.5486334562301636,
      "learning_rate": 2.0319204390892566e-06,
      "loss": 0.06237971782684326,
      "step": 6244
    },
    {
      "epoch": 0.8442825879390616,
      "grad_norm": 0.7356615662574768,
      "learning_rate": 2.0284820097452374e-06,
      "loss": 0.06647787988185883,
      "step": 6245
    },
    {
      "epoch": 0.8444177813078269,
      "grad_norm": 0.34131887555122375,
      "learning_rate": 2.02504628114043e-06,
      "loss": 0.03561331331729889,
      "step": 6246
    },
    {
      "epoch": 0.8445529746765921,
      "grad_norm": 0.6534212231636047,
      "learning_rate": 2.0216132539901865e-06,
      "loss": 0.054908670485019684,
      "step": 6247
    },
    {
      "epoch": 0.8446881680453574,
      "grad_norm": 0.4225394129753113,
      "learning_rate": 2.0181829290092663e-06,
      "loss": 0.0484328418970108,
      "step": 6248
    },
    {
      "epoch": 0.8448233614141226,
      "grad_norm": 0.4311065375804901,
      "learning_rate": 2.014755306911891e-06,
      "loss": 0.07305488735437393,
      "step": 6249
    },
    {
      "epoch": 0.844958554782888,
      "grad_norm": 1.0475775003433228,
      "learning_rate": 2.0113303884117057e-06,
      "loss": 0.049315690994262695,
      "step": 6250
    },
    {
      "epoch": 0.8450937481516532,
      "grad_norm": 0.3791733980178833,
      "learning_rate": 2.0079081742217957e-06,
      "loss": 0.05666741728782654,
      "step": 6251
    },
    {
      "epoch": 0.8452289415204184,
      "grad_norm": 0.4025326371192932,
      "learning_rate": 2.0044886650546915e-06,
      "loss": 0.04880619794130325,
      "step": 6252
    },
    {
      "epoch": 0.8453641348891837,
      "grad_norm": 0.450590044260025,
      "learning_rate": 2.0010718616223406e-06,
      "loss": 0.06794996559619904,
      "step": 6253
    },
    {
      "epoch": 0.8454993282579489,
      "grad_norm": 0.537303626537323,
      "learning_rate": 1.9976577646361514e-06,
      "loss": 0.05239599943161011,
      "step": 6254
    },
    {
      "epoch": 0.8456345216267143,
      "grad_norm": 0.4125227928161621,
      "learning_rate": 1.994246374806953e-06,
      "loss": 0.04312615841627121,
      "step": 6255
    },
    {
      "epoch": 0.8457697149954795,
      "grad_norm": 0.41425222158432007,
      "learning_rate": 1.9908376928450128e-06,
      "loss": 0.036222536116838455,
      "step": 6256
    },
    {
      "epoch": 0.8459049083642447,
      "grad_norm": 0.43170732259750366,
      "learning_rate": 1.987431719460039e-06,
      "loss": 0.04883016273379326,
      "step": 6257
    },
    {
      "epoch": 0.84604010173301,
      "grad_norm": 0.2581452429294586,
      "learning_rate": 1.9840284553611706e-06,
      "loss": 0.04100232198834419,
      "step": 6258
    },
    {
      "epoch": 0.8461752951017752,
      "grad_norm": 0.7356533408164978,
      "learning_rate": 1.980627901256989e-06,
      "loss": 0.062110066413879395,
      "step": 6259
    },
    {
      "epoch": 0.8463104884705405,
      "grad_norm": 0.6201856136322021,
      "learning_rate": 1.9772300578555062e-06,
      "loss": 0.03643447533249855,
      "step": 6260
    },
    {
      "epoch": 0.8464456818393058,
      "grad_norm": 0.47453224658966064,
      "learning_rate": 1.973834925864172e-06,
      "loss": 0.04004233330488205,
      "step": 6261
    },
    {
      "epoch": 0.846580875208071,
      "grad_norm": 0.9696336388587952,
      "learning_rate": 1.97044250598987e-06,
      "loss": 0.050774574279785156,
      "step": 6262
    },
    {
      "epoch": 0.8467160685768363,
      "grad_norm": 0.8261301517486572,
      "learning_rate": 1.9670527989389177e-06,
      "loss": 0.05598049238324165,
      "step": 6263
    },
    {
      "epoch": 0.8468512619456016,
      "grad_norm": 0.38730388879776,
      "learning_rate": 1.9636658054170747e-06,
      "loss": 0.055252887308597565,
      "step": 6264
    },
    {
      "epoch": 0.8469864553143668,
      "grad_norm": 0.36259526014328003,
      "learning_rate": 1.960281526129531e-06,
      "loss": 0.04671737551689148,
      "step": 6265
    },
    {
      "epoch": 0.8471216486831321,
      "grad_norm": 0.9208915829658508,
      "learning_rate": 1.9568999617809077e-06,
      "loss": 0.06472264230251312,
      "step": 6266
    },
    {
      "epoch": 0.8472568420518973,
      "grad_norm": 0.46292293071746826,
      "learning_rate": 1.9535211130752676e-06,
      "loss": 0.04803779721260071,
      "step": 6267
    },
    {
      "epoch": 0.8473920354206627,
      "grad_norm": 0.6515359878540039,
      "learning_rate": 1.950144980716101e-06,
      "loss": 0.06567519903182983,
      "step": 6268
    },
    {
      "epoch": 0.8475272287894279,
      "grad_norm": 1.0845727920532227,
      "learning_rate": 1.9467715654063444e-06,
      "loss": 0.058678675442934036,
      "step": 6269
    },
    {
      "epoch": 0.8476624221581931,
      "grad_norm": 0.4048415720462799,
      "learning_rate": 1.9434008678483532e-06,
      "loss": 0.049621663987636566,
      "step": 6270
    },
    {
      "epoch": 0.8477976155269584,
      "grad_norm": 0.17586331069469452,
      "learning_rate": 1.9400328887439295e-06,
      "loss": 0.027401769533753395,
      "step": 6271
    },
    {
      "epoch": 0.8479328088957236,
      "grad_norm": 0.6511488556861877,
      "learning_rate": 1.9366676287943038e-06,
      "loss": 0.07533954828977585,
      "step": 6272
    },
    {
      "epoch": 0.848068002264489,
      "grad_norm": 1.032747745513916,
      "learning_rate": 1.9333050887001337e-06,
      "loss": 0.05825965106487274,
      "step": 6273
    },
    {
      "epoch": 0.8482031956332542,
      "grad_norm": 0.867613673210144,
      "learning_rate": 1.9299452691615293e-06,
      "loss": 0.07221711426973343,
      "step": 6274
    },
    {
      "epoch": 0.8483383890020194,
      "grad_norm": 0.1599804013967514,
      "learning_rate": 1.9265881708780182e-06,
      "loss": 0.03129618614912033,
      "step": 6275
    },
    {
      "epoch": 0.8484735823707847,
      "grad_norm": 0.45775869488716125,
      "learning_rate": 1.9232337945485657e-06,
      "loss": 0.04149597883224487,
      "step": 6276
    },
    {
      "epoch": 0.84860877573955,
      "grad_norm": 0.2641725242137909,
      "learning_rate": 1.91988214087157e-06,
      "loss": 0.04288870468735695,
      "step": 6277
    },
    {
      "epoch": 0.8487439691083153,
      "grad_norm": 0.5147126913070679,
      "learning_rate": 1.9165332105448613e-06,
      "loss": 0.04254014417529106,
      "step": 6278
    },
    {
      "epoch": 0.8488791624770805,
      "grad_norm": 0.8537400364875793,
      "learning_rate": 1.913187004265715e-06,
      "loss": 0.0676904171705246,
      "step": 6279
    },
    {
      "epoch": 0.8490143558458457,
      "grad_norm": 0.6239067912101746,
      "learning_rate": 1.909843522730814e-06,
      "loss": 0.03847206383943558,
      "step": 6280
    },
    {
      "epoch": 0.849149549214611,
      "grad_norm": 0.33657991886138916,
      "learning_rate": 1.9065027666363017e-06,
      "loss": 0.031904011964797974,
      "step": 6281
    },
    {
      "epoch": 0.8492847425833763,
      "grad_norm": 1.0744355916976929,
      "learning_rate": 1.903164736677736e-06,
      "loss": 0.061805665493011475,
      "step": 6282
    },
    {
      "epoch": 0.8494199359521416,
      "grad_norm": 0.32329660654067993,
      "learning_rate": 1.8998294335501082e-06,
      "loss": 0.05151509493589401,
      "step": 6283
    },
    {
      "epoch": 0.8495551293209068,
      "grad_norm": 0.6502349376678467,
      "learning_rate": 1.8964968579478592e-06,
      "loss": 0.056619733572006226,
      "step": 6284
    },
    {
      "epoch": 0.849690322689672,
      "grad_norm": 0.845119297504425,
      "learning_rate": 1.893167010564834e-06,
      "loss": 0.04350561648607254,
      "step": 6285
    },
    {
      "epoch": 0.8498255160584374,
      "grad_norm": 0.26181334257125854,
      "learning_rate": 1.8898398920943349e-06,
      "loss": 0.04260338097810745,
      "step": 6286
    },
    {
      "epoch": 0.8499607094272026,
      "grad_norm": 0.5554114580154419,
      "learning_rate": 1.886515503229081e-06,
      "loss": 0.04976817965507507,
      "step": 6287
    },
    {
      "epoch": 0.8500959027959678,
      "grad_norm": 0.38692641258239746,
      "learning_rate": 1.8831938446612269e-06,
      "loss": 0.03918194770812988,
      "step": 6288
    },
    {
      "epoch": 0.8502310961647331,
      "grad_norm": 1.0803706645965576,
      "learning_rate": 1.8798749170823676e-06,
      "loss": 0.04637579619884491,
      "step": 6289
    },
    {
      "epoch": 0.8503662895334984,
      "grad_norm": 0.4428040683269501,
      "learning_rate": 1.8765587211835089e-06,
      "loss": 0.05340223386883736,
      "step": 6290
    },
    {
      "epoch": 0.8505014829022637,
      "grad_norm": 0.47322025895118713,
      "learning_rate": 1.8732452576551102e-06,
      "loss": 0.05617991089820862,
      "step": 6291
    },
    {
      "epoch": 0.8506366762710289,
      "grad_norm": 0.30985206365585327,
      "learning_rate": 1.8699345271870493e-06,
      "loss": 0.03756796196103096,
      "step": 6292
    },
    {
      "epoch": 0.8507718696397941,
      "grad_norm": 0.7566153407096863,
      "learning_rate": 1.8666265304686387e-06,
      "loss": 0.06445176899433136,
      "step": 6293
    },
    {
      "epoch": 0.8509070630085594,
      "grad_norm": 0.530719518661499,
      "learning_rate": 1.8633212681886203e-06,
      "loss": 0.05478502810001373,
      "step": 6294
    },
    {
      "epoch": 0.8510422563773247,
      "grad_norm": 0.34763604402542114,
      "learning_rate": 1.8600187410351621e-06,
      "loss": 0.04491907358169556,
      "step": 6295
    },
    {
      "epoch": 0.85117744974609,
      "grad_norm": 0.35831430554389954,
      "learning_rate": 1.8567189496958776e-06,
      "loss": 0.047058895230293274,
      "step": 6296
    },
    {
      "epoch": 0.8513126431148552,
      "grad_norm": 0.33181583881378174,
      "learning_rate": 1.853421894857797e-06,
      "loss": 0.04101666063070297,
      "step": 6297
    },
    {
      "epoch": 0.8514478364836204,
      "grad_norm": 0.6222859621047974,
      "learning_rate": 1.8501275772073827e-06,
      "loss": 0.05943810194730759,
      "step": 6298
    },
    {
      "epoch": 0.8515830298523858,
      "grad_norm": 0.4641937017440796,
      "learning_rate": 1.8468359974305315e-06,
      "loss": 0.045994631946086884,
      "step": 6299
    },
    {
      "epoch": 0.851718223221151,
      "grad_norm": 0.4015611708164215,
      "learning_rate": 1.8435471562125633e-06,
      "loss": 0.05843416973948479,
      "step": 6300
    },
    {
      "epoch": 0.8518534165899163,
      "grad_norm": 0.35173070430755615,
      "learning_rate": 1.8402610542382386e-06,
      "loss": 0.051927611231803894,
      "step": 6301
    },
    {
      "epoch": 0.8519886099586815,
      "grad_norm": 0.8363332152366638,
      "learning_rate": 1.836977692191742e-06,
      "loss": 0.056589026004076004,
      "step": 6302
    },
    {
      "epoch": 0.8521238033274467,
      "grad_norm": 0.3716745674610138,
      "learning_rate": 1.8336970707566781e-06,
      "loss": 0.03616643697023392,
      "step": 6303
    },
    {
      "epoch": 0.8522589966962121,
      "grad_norm": 0.5818215012550354,
      "learning_rate": 1.8304191906160973e-06,
      "loss": 0.047216322273015976,
      "step": 6304
    },
    {
      "epoch": 0.8523941900649773,
      "grad_norm": 0.7553380131721497,
      "learning_rate": 1.8271440524524668e-06,
      "loss": 0.06350641697645187,
      "step": 6305
    },
    {
      "epoch": 0.8525293834337426,
      "grad_norm": 0.6106654405593872,
      "learning_rate": 1.8238716569476949e-06,
      "loss": 0.039121419191360474,
      "step": 6306
    },
    {
      "epoch": 0.8526645768025078,
      "grad_norm": 0.5820634961128235,
      "learning_rate": 1.8206020047831078e-06,
      "loss": 0.05436358228325844,
      "step": 6307
    },
    {
      "epoch": 0.8527997701712731,
      "grad_norm": 0.4586799442768097,
      "learning_rate": 1.8173350966394648e-06,
      "loss": 0.048718370497226715,
      "step": 6308
    },
    {
      "epoch": 0.8529349635400384,
      "grad_norm": 0.614192545413971,
      "learning_rate": 1.8140709331969513e-06,
      "loss": 0.06158027797937393,
      "step": 6309
    },
    {
      "epoch": 0.8530701569088036,
      "grad_norm": 0.6466459631919861,
      "learning_rate": 1.810809515135184e-06,
      "loss": 0.05154663324356079,
      "step": 6310
    },
    {
      "epoch": 0.8532053502775688,
      "grad_norm": 0.29418498277664185,
      "learning_rate": 1.8075508431332111e-06,
      "loss": 0.036776103079319,
      "step": 6311
    },
    {
      "epoch": 0.8533405436463342,
      "grad_norm": 0.29602229595184326,
      "learning_rate": 1.8042949178695034e-06,
      "loss": 0.03308059275150299,
      "step": 6312
    },
    {
      "epoch": 0.8534757370150994,
      "grad_norm": 0.7528901696205139,
      "learning_rate": 1.8010417400219636e-06,
      "loss": 0.057235248386859894,
      "step": 6313
    },
    {
      "epoch": 0.8536109303838647,
      "grad_norm": 1.0960278511047363,
      "learning_rate": 1.7977913102679167e-06,
      "loss": 0.042923420667648315,
      "step": 6314
    },
    {
      "epoch": 0.8537461237526299,
      "grad_norm": 0.6818878054618835,
      "learning_rate": 1.7945436292841193e-06,
      "loss": 0.06785756349563599,
      "step": 6315
    },
    {
      "epoch": 0.8538813171213951,
      "grad_norm": 0.2882588505744934,
      "learning_rate": 1.791298697746766e-06,
      "loss": 0.05655911564826965,
      "step": 6316
    },
    {
      "epoch": 0.8540165104901605,
      "grad_norm": 0.4256933927536011,
      "learning_rate": 1.7880565163314545e-06,
      "loss": 0.04764436185359955,
      "step": 6317
    },
    {
      "epoch": 0.8541517038589257,
      "grad_norm": 0.6672478318214417,
      "learning_rate": 1.784817085713233e-06,
      "loss": 0.04730550944805145,
      "step": 6318
    },
    {
      "epoch": 0.854286897227691,
      "grad_norm": 0.5017374753952026,
      "learning_rate": 1.7815804065665669e-06,
      "loss": 0.0507778599858284,
      "step": 6319
    },
    {
      "epoch": 0.8544220905964562,
      "grad_norm": 0.46741950511932373,
      "learning_rate": 1.778346479565346e-06,
      "loss": 0.05671544373035431,
      "step": 6320
    },
    {
      "epoch": 0.8545572839652215,
      "grad_norm": 1.0150688886642456,
      "learning_rate": 1.7751153053829011e-06,
      "loss": 0.06213996559381485,
      "step": 6321
    },
    {
      "epoch": 0.8546924773339868,
      "grad_norm": 0.5143606662750244,
      "learning_rate": 1.7718868846919662e-06,
      "loss": 0.06776639074087143,
      "step": 6322
    },
    {
      "epoch": 0.854827670702752,
      "grad_norm": 1.093859314918518,
      "learning_rate": 1.7686612181647266e-06,
      "loss": 0.052836500108242035,
      "step": 6323
    },
    {
      "epoch": 0.8549628640715173,
      "grad_norm": 0.9550508260726929,
      "learning_rate": 1.7654383064727802e-06,
      "loss": 0.04823852330446243,
      "step": 6324
    },
    {
      "epoch": 0.8550980574402826,
      "grad_norm": 0.7983924150466919,
      "learning_rate": 1.762218150287152e-06,
      "loss": 0.05271138250827789,
      "step": 6325
    },
    {
      "epoch": 0.8552332508090478,
      "grad_norm": 0.7791218757629395,
      "learning_rate": 1.759000750278299e-06,
      "loss": 0.042531318962574005,
      "step": 6326
    },
    {
      "epoch": 0.8553684441778131,
      "grad_norm": 0.6210097670555115,
      "learning_rate": 1.7557861071160953e-06,
      "loss": 0.05496615171432495,
      "step": 6327
    },
    {
      "epoch": 0.8555036375465783,
      "grad_norm": 0.20374125242233276,
      "learning_rate": 1.7525742214698538e-06,
      "loss": 0.03397899866104126,
      "step": 6328
    },
    {
      "epoch": 0.8556388309153437,
      "grad_norm": 0.5363911390304565,
      "learning_rate": 1.7493650940083045e-06,
      "loss": 0.04610375314950943,
      "step": 6329
    },
    {
      "epoch": 0.8557740242841089,
      "grad_norm": 1.0519609451293945,
      "learning_rate": 1.746158725399603e-06,
      "loss": 0.04173114895820618,
      "step": 6330
    },
    {
      "epoch": 0.8559092176528741,
      "grad_norm": 0.9720075130462646,
      "learning_rate": 1.7429551163113322e-06,
      "loss": 0.05323171615600586,
      "step": 6331
    },
    {
      "epoch": 0.8560444110216394,
      "grad_norm": 0.5926998853683472,
      "learning_rate": 1.7397542674105e-06,
      "loss": 0.06275883316993713,
      "step": 6332
    },
    {
      "epoch": 0.8561796043904046,
      "grad_norm": 0.4590113162994385,
      "learning_rate": 1.7365561793635431e-06,
      "loss": 0.04344320297241211,
      "step": 6333
    },
    {
      "epoch": 0.85631479775917,
      "grad_norm": 0.28174513578414917,
      "learning_rate": 1.7333608528363227e-06,
      "loss": 0.04860040917992592,
      "step": 6334
    },
    {
      "epoch": 0.8564499911279352,
      "grad_norm": 0.5829482674598694,
      "learning_rate": 1.7301682884941128e-06,
      "loss": 0.05380423367023468,
      "step": 6335
    },
    {
      "epoch": 0.8565851844967004,
      "grad_norm": 0.42195332050323486,
      "learning_rate": 1.726978487001632e-06,
      "loss": 0.04403732717037201,
      "step": 6336
    },
    {
      "epoch": 0.8567203778654657,
      "grad_norm": 0.3213565945625305,
      "learning_rate": 1.7237914490230072e-06,
      "loss": 0.03615276515483856,
      "step": 6337
    },
    {
      "epoch": 0.856855571234231,
      "grad_norm": 1.1198323965072632,
      "learning_rate": 1.7206071752218027e-06,
      "loss": 0.07081387937068939,
      "step": 6338
    },
    {
      "epoch": 0.8569907646029962,
      "grad_norm": 0.3278871774673462,
      "learning_rate": 1.7174256662610032e-06,
      "loss": 0.05254681408405304,
      "step": 6339
    },
    {
      "epoch": 0.8571259579717615,
      "grad_norm": 0.5900607705116272,
      "learning_rate": 1.714246922803004e-06,
      "loss": 0.06021782010793686,
      "step": 6340
    },
    {
      "epoch": 0.8572611513405267,
      "grad_norm": 0.4513183832168579,
      "learning_rate": 1.7110709455096468e-06,
      "loss": 0.049845121800899506,
      "step": 6341
    },
    {
      "epoch": 0.857396344709292,
      "grad_norm": 0.28617408871650696,
      "learning_rate": 1.7078977350421815e-06,
      "loss": 0.044528208673000336,
      "step": 6342
    },
    {
      "epoch": 0.8575315380780573,
      "grad_norm": 0.6556611657142639,
      "learning_rate": 1.7047272920612926e-06,
      "loss": 0.038710109889507294,
      "step": 6343
    },
    {
      "epoch": 0.8576667314468225,
      "grad_norm": 0.3762466311454773,
      "learning_rate": 1.7015596172270841e-06,
      "loss": 0.03848453238606453,
      "step": 6344
    },
    {
      "epoch": 0.8578019248155878,
      "grad_norm": 0.47321709990501404,
      "learning_rate": 1.6983947111990717e-06,
      "loss": 0.04214465618133545,
      "step": 6345
    },
    {
      "epoch": 0.857937118184353,
      "grad_norm": 0.35063162446022034,
      "learning_rate": 1.695232574636218e-06,
      "loss": 0.051957957446575165,
      "step": 6346
    },
    {
      "epoch": 0.8580723115531184,
      "grad_norm": 0.3668923079967499,
      "learning_rate": 1.6920732081968882e-06,
      "loss": 0.05051729083061218,
      "step": 6347
    },
    {
      "epoch": 0.8582075049218836,
      "grad_norm": 0.7809852361679077,
      "learning_rate": 1.6889166125388878e-06,
      "loss": 0.04531104117631912,
      "step": 6348
    },
    {
      "epoch": 0.8583426982906488,
      "grad_norm": 0.735115110874176,
      "learning_rate": 1.6857627883194277e-06,
      "loss": 0.05323927849531174,
      "step": 6349
    },
    {
      "epoch": 0.8584778916594141,
      "grad_norm": 0.2956749200820923,
      "learning_rate": 1.6826117361951577e-06,
      "loss": 0.046146489679813385,
      "step": 6350
    },
    {
      "epoch": 0.8586130850281793,
      "grad_norm": 0.6716263294219971,
      "learning_rate": 1.6794634568221412e-06,
      "loss": 0.07926372438669205,
      "step": 6351
    },
    {
      "epoch": 0.8587482783969447,
      "grad_norm": 0.8734883666038513,
      "learning_rate": 1.676317950855864e-06,
      "loss": 0.055575475096702576,
      "step": 6352
    },
    {
      "epoch": 0.8588834717657099,
      "grad_norm": 0.7916175127029419,
      "learning_rate": 1.6731752189512456e-06,
      "loss": 0.0412934236228466,
      "step": 6353
    },
    {
      "epoch": 0.8590186651344751,
      "grad_norm": 0.5428727269172668,
      "learning_rate": 1.6700352617626092e-06,
      "loss": 0.05648193508386612,
      "step": 6354
    },
    {
      "epoch": 0.8591538585032404,
      "grad_norm": 0.8448138236999512,
      "learning_rate": 1.6668980799437167e-06,
      "loss": 0.04699888080358505,
      "step": 6355
    },
    {
      "epoch": 0.8592890518720057,
      "grad_norm": 0.27350717782974243,
      "learning_rate": 1.6637636741477458e-06,
      "loss": 0.047247543931007385,
      "step": 6356
    },
    {
      "epoch": 0.859424245240771,
      "grad_norm": 0.5769308805465698,
      "learning_rate": 1.6606320450272943e-06,
      "loss": 0.05877112224698067,
      "step": 6357
    },
    {
      "epoch": 0.8595594386095362,
      "grad_norm": 0.3961043655872345,
      "learning_rate": 1.657503193234386e-06,
      "loss": 0.045877233147621155,
      "step": 6358
    },
    {
      "epoch": 0.8596946319783014,
      "grad_norm": 0.647350013256073,
      "learning_rate": 1.654377119420461e-06,
      "loss": 0.047591596841812134,
      "step": 6359
    },
    {
      "epoch": 0.8598298253470668,
      "grad_norm": 0.3304547369480133,
      "learning_rate": 1.6512538242363889e-06,
      "loss": 0.04440970718860626,
      "step": 6360
    },
    {
      "epoch": 0.859965018715832,
      "grad_norm": 0.6236200928688049,
      "learning_rate": 1.6481333083324563e-06,
      "loss": 0.062338974326848984,
      "step": 6361
    },
    {
      "epoch": 0.8601002120845973,
      "grad_norm": 0.4829655885696411,
      "learning_rate": 1.6450155723583698e-06,
      "loss": 0.0645047053694725,
      "step": 6362
    },
    {
      "epoch": 0.8602354054533625,
      "grad_norm": 1.0008617639541626,
      "learning_rate": 1.6419006169632573e-06,
      "loss": 0.058346331119537354,
      "step": 6363
    },
    {
      "epoch": 0.8603705988221277,
      "grad_norm": 0.6507902145385742,
      "learning_rate": 1.638788442795668e-06,
      "loss": 0.055436864495277405,
      "step": 6364
    },
    {
      "epoch": 0.8605057921908931,
      "grad_norm": 1.509135365486145,
      "learning_rate": 1.6356790505035785e-06,
      "loss": 0.06659440696239471,
      "step": 6365
    },
    {
      "epoch": 0.8606409855596583,
      "grad_norm": 0.7501407861709595,
      "learning_rate": 1.6325724407343795e-06,
      "loss": 0.05589187145233154,
      "step": 6366
    },
    {
      "epoch": 0.8607761789284235,
      "grad_norm": 0.7137078642845154,
      "learning_rate": 1.6294686141348801e-06,
      "loss": 0.05699652433395386,
      "step": 6367
    },
    {
      "epoch": 0.8609113722971888,
      "grad_norm": 0.5166784524917603,
      "learning_rate": 1.626367571351317e-06,
      "loss": 0.054060764610767365,
      "step": 6368
    },
    {
      "epoch": 0.8610465656659541,
      "grad_norm": 0.7450024485588074,
      "learning_rate": 1.6232693130293386e-06,
      "loss": 0.04736056178808212,
      "step": 6369
    },
    {
      "epoch": 0.8611817590347194,
      "grad_norm": 0.3280739188194275,
      "learning_rate": 1.6201738398140254e-06,
      "loss": 0.04433522745966911,
      "step": 6370
    },
    {
      "epoch": 0.8613169524034846,
      "grad_norm": 0.5554978847503662,
      "learning_rate": 1.6170811523498718e-06,
      "loss": 0.057081907987594604,
      "step": 6371
    },
    {
      "epoch": 0.8614521457722498,
      "grad_norm": 0.3789294958114624,
      "learning_rate": 1.613991251280783e-06,
      "loss": 0.06383635103702545,
      "step": 6372
    },
    {
      "epoch": 0.8615873391410152,
      "grad_norm": 0.8679389357566833,
      "learning_rate": 1.6109041372501028e-06,
      "loss": 0.049386560916900635,
      "step": 6373
    },
    {
      "epoch": 0.8617225325097804,
      "grad_norm": 0.5927804708480835,
      "learning_rate": 1.6078198109005766e-06,
      "loss": 0.06466534733772278,
      "step": 6374
    },
    {
      "epoch": 0.8618577258785457,
      "grad_norm": 1.431593656539917,
      "learning_rate": 1.6047382728743843e-06,
      "loss": 0.0730823203921318,
      "step": 6375
    },
    {
      "epoch": 0.8619929192473109,
      "grad_norm": 0.36228853464126587,
      "learning_rate": 1.6016595238131176e-06,
      "loss": 0.046502485871315,
      "step": 6376
    },
    {
      "epoch": 0.8621281126160761,
      "grad_norm": 0.4120558798313141,
      "learning_rate": 1.5985835643577824e-06,
      "loss": 0.037470173090696335,
      "step": 6377
    },
    {
      "epoch": 0.8622633059848415,
      "grad_norm": 0.4864238500595093,
      "learning_rate": 1.5955103951488177e-06,
      "loss": 0.04541681706905365,
      "step": 6378
    },
    {
      "epoch": 0.8623984993536067,
      "grad_norm": 0.2943304777145386,
      "learning_rate": 1.5924400168260666e-06,
      "loss": 0.0466192290186882,
      "step": 6379
    },
    {
      "epoch": 0.862533692722372,
      "grad_norm": 1.0102490186691284,
      "learning_rate": 1.5893724300288064e-06,
      "loss": 0.05961532145738602,
      "step": 6380
    },
    {
      "epoch": 0.8626688860911372,
      "grad_norm": 0.5914861559867859,
      "learning_rate": 1.5863076353957196e-06,
      "loss": 0.04195981100201607,
      "step": 6381
    },
    {
      "epoch": 0.8628040794599025,
      "grad_norm": 0.43987593054771423,
      "learning_rate": 1.5832456335649104e-06,
      "loss": 0.0477483868598938,
      "step": 6382
    },
    {
      "epoch": 0.8629392728286678,
      "grad_norm": 0.37542444467544556,
      "learning_rate": 1.580186425173909e-06,
      "loss": 0.04148589074611664,
      "step": 6383
    },
    {
      "epoch": 0.863074466197433,
      "grad_norm": 0.5105854868888855,
      "learning_rate": 1.5771300108596543e-06,
      "loss": 0.04227229952812195,
      "step": 6384
    },
    {
      "epoch": 0.8632096595661983,
      "grad_norm": 0.39576664566993713,
      "learning_rate": 1.5740763912585171e-06,
      "loss": 0.06163744255900383,
      "step": 6385
    },
    {
      "epoch": 0.8633448529349635,
      "grad_norm": 0.43987858295440674,
      "learning_rate": 1.5710255670062657e-06,
      "loss": 0.04137536138296127,
      "step": 6386
    },
    {
      "epoch": 0.8634800463037288,
      "grad_norm": 0.8914985060691833,
      "learning_rate": 1.567977538738105e-06,
      "loss": 0.045304179191589355,
      "step": 6387
    },
    {
      "epoch": 0.8636152396724941,
      "grad_norm": 0.38886401057243347,
      "learning_rate": 1.5649323070886494e-06,
      "loss": 0.04219827055931091,
      "step": 6388
    },
    {
      "epoch": 0.8637504330412593,
      "grad_norm": 0.6032065153121948,
      "learning_rate": 1.5618898726919284e-06,
      "loss": 0.07757360488176346,
      "step": 6389
    },
    {
      "epoch": 0.8638856264100246,
      "grad_norm": 0.6317402124404907,
      "learning_rate": 1.5588502361814032e-06,
      "loss": 0.04318195581436157,
      "step": 6390
    },
    {
      "epoch": 0.8640208197787899,
      "grad_norm": 0.41027575731277466,
      "learning_rate": 1.5558133981899314e-06,
      "loss": 0.04398167133331299,
      "step": 6391
    },
    {
      "epoch": 0.8641560131475551,
      "grad_norm": 0.6707596778869629,
      "learning_rate": 1.5527793593498053e-06,
      "loss": 0.053114041686058044,
      "step": 6392
    },
    {
      "epoch": 0.8642912065163204,
      "grad_norm": 0.6818137168884277,
      "learning_rate": 1.5497481202927244e-06,
      "loss": 0.05459202080965042,
      "step": 6393
    },
    {
      "epoch": 0.8644263998850856,
      "grad_norm": 0.7187246680259705,
      "learning_rate": 1.5467196816498107e-06,
      "loss": 0.05118875205516815,
      "step": 6394
    },
    {
      "epoch": 0.8645615932538508,
      "grad_norm": 0.8068587183952332,
      "learning_rate": 1.5436940440516018e-06,
      "loss": 0.056244686245918274,
      "step": 6395
    },
    {
      "epoch": 0.8646967866226162,
      "grad_norm": 0.5673879981040955,
      "learning_rate": 1.5406712081280484e-06,
      "loss": 0.06307181715965271,
      "step": 6396
    },
    {
      "epoch": 0.8648319799913814,
      "grad_norm": 0.7542120218276978,
      "learning_rate": 1.5376511745085254e-06,
      "loss": 0.06109628081321716,
      "step": 6397
    },
    {
      "epoch": 0.8649671733601467,
      "grad_norm": 0.6209232211112976,
      "learning_rate": 1.5346339438218181e-06,
      "loss": 0.04325780272483826,
      "step": 6398
    },
    {
      "epoch": 0.8651023667289119,
      "grad_norm": 0.5214393734931946,
      "learning_rate": 1.5316195166961295e-06,
      "loss": 0.04177108779549599,
      "step": 6399
    },
    {
      "epoch": 0.8652375600976772,
      "grad_norm": 0.46321913599967957,
      "learning_rate": 1.5286078937590802e-06,
      "loss": 0.04379759728908539,
      "step": 6400
    },
    {
      "epoch": 0.8653727534664425,
      "grad_norm": 0.3907475173473358,
      "learning_rate": 1.5255990756377025e-06,
      "loss": 0.05963176488876343,
      "step": 6401
    },
    {
      "epoch": 0.8655079468352077,
      "grad_norm": 0.32961878180503845,
      "learning_rate": 1.5225930629584534e-06,
      "loss": 0.04474177211523056,
      "step": 6402
    },
    {
      "epoch": 0.865643140203973,
      "grad_norm": 1.0100473165512085,
      "learning_rate": 1.5195898563472038e-06,
      "loss": 0.05553389713168144,
      "step": 6403
    },
    {
      "epoch": 0.8657783335727383,
      "grad_norm": 0.4864867031574249,
      "learning_rate": 1.5165894564292254e-06,
      "loss": 0.044846296310424805,
      "step": 6404
    },
    {
      "epoch": 0.8659135269415035,
      "grad_norm": 0.3903617560863495,
      "learning_rate": 1.5135918638292269e-06,
      "loss": 0.049049973487854004,
      "step": 6405
    },
    {
      "epoch": 0.8660487203102688,
      "grad_norm": 0.393320232629776,
      "learning_rate": 1.5105970791713186e-06,
      "loss": 0.04075436294078827,
      "step": 6406
    },
    {
      "epoch": 0.866183913679034,
      "grad_norm": 0.6641169190406799,
      "learning_rate": 1.5076051030790355e-06,
      "loss": 0.03864654526114464,
      "step": 6407
    },
    {
      "epoch": 0.8663191070477994,
      "grad_norm": 0.3610553741455078,
      "learning_rate": 1.5046159361753226e-06,
      "loss": 0.04762904345989227,
      "step": 6408
    },
    {
      "epoch": 0.8664543004165646,
      "grad_norm": 1.0069036483764648,
      "learning_rate": 1.5016295790825336e-06,
      "loss": 0.07318048179149628,
      "step": 6409
    },
    {
      "epoch": 0.8665894937853298,
      "grad_norm": 0.4771886467933655,
      "learning_rate": 1.4986460324224493e-06,
      "loss": 0.0502270944416523,
      "step": 6410
    },
    {
      "epoch": 0.8667246871540951,
      "grad_norm": 0.47285547852516174,
      "learning_rate": 1.4956652968162582e-06,
      "loss": 0.04290420562028885,
      "step": 6411
    },
    {
      "epoch": 0.8668598805228603,
      "grad_norm": 0.5081998705863953,
      "learning_rate": 1.492687372884567e-06,
      "loss": 0.05049280822277069,
      "step": 6412
    },
    {
      "epoch": 0.8669950738916257,
      "grad_norm": 0.5225005149841309,
      "learning_rate": 1.4897122612473978e-06,
      "loss": 0.03781360387802124,
      "step": 6413
    },
    {
      "epoch": 0.8671302672603909,
      "grad_norm": 0.7593132257461548,
      "learning_rate": 1.4867399625241772e-06,
      "loss": 0.05383706092834473,
      "step": 6414
    },
    {
      "epoch": 0.8672654606291561,
      "grad_norm": 1.5183064937591553,
      "learning_rate": 1.4837704773337602e-06,
      "loss": 0.05239257961511612,
      "step": 6415
    },
    {
      "epoch": 0.8674006539979214,
      "grad_norm": 0.6086822152137756,
      "learning_rate": 1.4808038062944036e-06,
      "loss": 0.05151867866516113,
      "step": 6416
    },
    {
      "epoch": 0.8675358473666867,
      "grad_norm": 1.0132187604904175,
      "learning_rate": 1.4778399500237933e-06,
      "loss": 0.051538825035095215,
      "step": 6417
    },
    {
      "epoch": 0.8676710407354519,
      "grad_norm": 0.4417206943035126,
      "learning_rate": 1.4748789091390124e-06,
      "loss": 0.05564595013856888,
      "step": 6418
    },
    {
      "epoch": 0.8678062341042172,
      "grad_norm": 0.24355541169643402,
      "learning_rate": 1.471920684256563e-06,
      "loss": 0.03250390291213989,
      "step": 6419
    },
    {
      "epoch": 0.8679414274729824,
      "grad_norm": 0.4006502628326416,
      "learning_rate": 1.4689652759923721e-06,
      "loss": 0.051907412707805634,
      "step": 6420
    },
    {
      "epoch": 0.8680766208417477,
      "grad_norm": 0.5877748131752014,
      "learning_rate": 1.4660126849617645e-06,
      "loss": 0.04292129725217819,
      "step": 6421
    },
    {
      "epoch": 0.868211814210513,
      "grad_norm": 0.696816623210907,
      "learning_rate": 1.4630629117794914e-06,
      "loss": 0.07989153265953064,
      "step": 6422
    },
    {
      "epoch": 0.8683470075792782,
      "grad_norm": 0.4741944372653961,
      "learning_rate": 1.4601159570597033e-06,
      "loss": 0.057127028703689575,
      "step": 6423
    },
    {
      "epoch": 0.8684822009480435,
      "grad_norm": 0.4335305690765381,
      "learning_rate": 1.4571718214159795e-06,
      "loss": 0.04525130242109299,
      "step": 6424
    },
    {
      "epoch": 0.8686173943168087,
      "grad_norm": 0.6267746090888977,
      "learning_rate": 1.454230505461303e-06,
      "loss": 0.05341595038771629,
      "step": 6425
    },
    {
      "epoch": 0.8687525876855741,
      "grad_norm": 0.4684426188468933,
      "learning_rate": 1.4512920098080672e-06,
      "loss": 0.05786684900522232,
      "step": 6426
    },
    {
      "epoch": 0.8688877810543393,
      "grad_norm": 0.5344424247741699,
      "learning_rate": 1.4483563350680878e-06,
      "loss": 0.0545182041823864,
      "step": 6427
    },
    {
      "epoch": 0.8690229744231045,
      "grad_norm": 0.33890390396118164,
      "learning_rate": 1.4454234818525824e-06,
      "loss": 0.04585573077201843,
      "step": 6428
    },
    {
      "epoch": 0.8691581677918698,
      "grad_norm": 0.8251439929008484,
      "learning_rate": 1.4424934507721926e-06,
      "loss": 0.05194412171840668,
      "step": 6429
    },
    {
      "epoch": 0.869293361160635,
      "grad_norm": 0.6554752588272095,
      "learning_rate": 1.4395662424369622e-06,
      "loss": 0.04080544039607048,
      "step": 6430
    },
    {
      "epoch": 0.8694285545294004,
      "grad_norm": 0.36110562086105347,
      "learning_rate": 1.436641857456355e-06,
      "loss": 0.055810391902923584,
      "step": 6431
    },
    {
      "epoch": 0.8695637478981656,
      "grad_norm": 0.2818463146686554,
      "learning_rate": 1.4337202964392409e-06,
      "loss": 0.03770819306373596,
      "step": 6432
    },
    {
      "epoch": 0.8696989412669308,
      "grad_norm": 1.177851676940918,
      "learning_rate": 1.4308015599939033e-06,
      "loss": 0.06075853109359741,
      "step": 6433
    },
    {
      "epoch": 0.8698341346356961,
      "grad_norm": 0.5334969758987427,
      "learning_rate": 1.4278856487280428e-06,
      "loss": 0.04215973988175392,
      "step": 6434
    },
    {
      "epoch": 0.8699693280044614,
      "grad_norm": 0.2561131417751312,
      "learning_rate": 1.4249725632487653e-06,
      "loss": 0.03477698192000389,
      "step": 6435
    },
    {
      "epoch": 0.8701045213732267,
      "grad_norm": 0.40459147095680237,
      "learning_rate": 1.4220623041625924e-06,
      "loss": 0.06360536068677902,
      "step": 6436
    },
    {
      "epoch": 0.8702397147419919,
      "grad_norm": 0.41641736030578613,
      "learning_rate": 1.4191548720754527e-06,
      "loss": 0.047386303544044495,
      "step": 6437
    },
    {
      "epoch": 0.8703749081107571,
      "grad_norm": 1.2045555114746094,
      "learning_rate": 1.4162502675926887e-06,
      "loss": 0.04919978976249695,
      "step": 6438
    },
    {
      "epoch": 0.8705101014795225,
      "grad_norm": 0.32616037130355835,
      "learning_rate": 1.4133484913190596e-06,
      "loss": 0.03985454887151718,
      "step": 6439
    },
    {
      "epoch": 0.8706452948482877,
      "grad_norm": 0.6028023362159729,
      "learning_rate": 1.4104495438587295e-06,
      "loss": 0.06465119123458862,
      "step": 6440
    },
    {
      "epoch": 0.870780488217053,
      "grad_norm": 0.649699866771698,
      "learning_rate": 1.4075534258152667e-06,
      "loss": 0.04072011262178421,
      "step": 6441
    },
    {
      "epoch": 0.8709156815858182,
      "grad_norm": 0.3229261636734009,
      "learning_rate": 1.4046601377916673e-06,
      "loss": 0.04177705943584442,
      "step": 6442
    },
    {
      "epoch": 0.8710508749545834,
      "grad_norm": 0.6591719388961792,
      "learning_rate": 1.4017696803903246e-06,
      "loss": 0.046736590564250946,
      "step": 6443
    },
    {
      "epoch": 0.8711860683233488,
      "grad_norm": 0.45171281695365906,
      "learning_rate": 1.3988820542130504e-06,
      "loss": 0.05376102030277252,
      "step": 6444
    },
    {
      "epoch": 0.871321261692114,
      "grad_norm": 0.6127946376800537,
      "learning_rate": 1.395997259861067e-06,
      "loss": 0.05691789835691452,
      "step": 6445
    },
    {
      "epoch": 0.8714564550608792,
      "grad_norm": 0.8426452875137329,
      "learning_rate": 1.3931152979349926e-06,
      "loss": 0.051948148757219315,
      "step": 6446
    },
    {
      "epoch": 0.8715916484296445,
      "grad_norm": 0.42971548438072205,
      "learning_rate": 1.3902361690348769e-06,
      "loss": 0.04130706191062927,
      "step": 6447
    },
    {
      "epoch": 0.8717268417984098,
      "grad_norm": 0.4513098895549774,
      "learning_rate": 1.3873598737601639e-06,
      "loss": 0.056921541690826416,
      "step": 6448
    },
    {
      "epoch": 0.8718620351671751,
      "grad_norm": 0.3034031093120575,
      "learning_rate": 1.3844864127097229e-06,
      "loss": 0.05169004574418068,
      "step": 6449
    },
    {
      "epoch": 0.8719972285359403,
      "grad_norm": 0.3589463531970978,
      "learning_rate": 1.3816157864818151e-06,
      "loss": 0.049613580107688904,
      "step": 6450
    },
    {
      "epoch": 0.8721324219047055,
      "grad_norm": 0.6391415596008301,
      "learning_rate": 1.3787479956741194e-06,
      "loss": 0.04300156235694885,
      "step": 6451
    },
    {
      "epoch": 0.8722676152734709,
      "grad_norm": 0.3037974238395691,
      "learning_rate": 1.3758830408837314e-06,
      "loss": 0.03323987126350403,
      "step": 6452
    },
    {
      "epoch": 0.8724028086422361,
      "grad_norm": 0.7414489984512329,
      "learning_rate": 1.3730209227071439e-06,
      "loss": 0.04776036739349365,
      "step": 6453
    },
    {
      "epoch": 0.8725380020110014,
      "grad_norm": 0.3472529351711273,
      "learning_rate": 1.3701616417402734e-06,
      "loss": 0.04723775386810303,
      "step": 6454
    },
    {
      "epoch": 0.8726731953797666,
      "grad_norm": 0.4649716913700104,
      "learning_rate": 1.367305198578429e-06,
      "loss": 0.040052302181720734,
      "step": 6455
    },
    {
      "epoch": 0.8728083887485318,
      "grad_norm": 0.2935580015182495,
      "learning_rate": 1.36445159381634e-06,
      "loss": 0.04241658002138138,
      "step": 6456
    },
    {
      "epoch": 0.8729435821172972,
      "grad_norm": 0.802302360534668,
      "learning_rate": 1.361600828048144e-06,
      "loss": 0.051356494426727295,
      "step": 6457
    },
    {
      "epoch": 0.8730787754860624,
      "grad_norm": 0.5173913836479187,
      "learning_rate": 1.3587529018673816e-06,
      "loss": 0.04657822474837303,
      "step": 6458
    },
    {
      "epoch": 0.8732139688548277,
      "grad_norm": 0.4374042749404907,
      "learning_rate": 1.3559078158670152e-06,
      "loss": 0.051262274384498596,
      "step": 6459
    },
    {
      "epoch": 0.8733491622235929,
      "grad_norm": 0.8997926115989685,
      "learning_rate": 1.353065570639394e-06,
      "loss": 0.06002334505319595,
      "step": 6460
    },
    {
      "epoch": 0.8734843555923582,
      "grad_norm": 0.5344158411026001,
      "learning_rate": 1.3502261667763e-06,
      "loss": 0.05258147791028023,
      "step": 6461
    },
    {
      "epoch": 0.8736195489611235,
      "grad_norm": 0.42840898036956787,
      "learning_rate": 1.3473896048689067e-06,
      "loss": 0.04621404409408569,
      "step": 6462
    },
    {
      "epoch": 0.8737547423298887,
      "grad_norm": 0.35789334774017334,
      "learning_rate": 1.3445558855078017e-06,
      "loss": 0.05047227442264557,
      "step": 6463
    },
    {
      "epoch": 0.873889935698654,
      "grad_norm": 1.0090200901031494,
      "learning_rate": 1.3417250092829814e-06,
      "loss": 0.05209711194038391,
      "step": 6464
    },
    {
      "epoch": 0.8740251290674192,
      "grad_norm": 0.5702008008956909,
      "learning_rate": 1.338896976783846e-06,
      "loss": 0.04754728078842163,
      "step": 6465
    },
    {
      "epoch": 0.8741603224361845,
      "grad_norm": 0.22192725539207458,
      "learning_rate": 1.336071788599213e-06,
      "loss": 0.03884801268577576,
      "step": 6466
    },
    {
      "epoch": 0.8742955158049498,
      "grad_norm": 0.6642999649047852,
      "learning_rate": 1.3332494453172982e-06,
      "loss": 0.04711092263460159,
      "step": 6467
    },
    {
      "epoch": 0.874430709173715,
      "grad_norm": 0.5399225950241089,
      "learning_rate": 1.3304299475257287e-06,
      "loss": 0.048498328775167465,
      "step": 6468
    },
    {
      "epoch": 0.8745659025424803,
      "grad_norm": 0.5880245566368103,
      "learning_rate": 1.3276132958115394e-06,
      "loss": 0.05429648980498314,
      "step": 6469
    },
    {
      "epoch": 0.8747010959112456,
      "grad_norm": 0.40637123584747314,
      "learning_rate": 1.32479949076117e-06,
      "loss": 0.04147633910179138,
      "step": 6470
    },
    {
      "epoch": 0.8748362892800108,
      "grad_norm": 0.5699285268783569,
      "learning_rate": 1.3219885329604747e-06,
      "loss": 0.04441697895526886,
      "step": 6471
    },
    {
      "epoch": 0.8749714826487761,
      "grad_norm": 0.9453657269477844,
      "learning_rate": 1.319180422994709e-06,
      "loss": 0.04900723695755005,
      "step": 6472
    },
    {
      "epoch": 0.8751066760175413,
      "grad_norm": 0.9205328822135925,
      "learning_rate": 1.3163751614485287e-06,
      "loss": 0.059768013656139374,
      "step": 6473
    },
    {
      "epoch": 0.8752418693863065,
      "grad_norm": 0.6823642253875732,
      "learning_rate": 1.3135727489060113e-06,
      "loss": 0.04388117790222168,
      "step": 6474
    },
    {
      "epoch": 0.8753770627550719,
      "grad_norm": 0.6911574006080627,
      "learning_rate": 1.3107731859506317e-06,
      "loss": 0.03704812377691269,
      "step": 6475
    },
    {
      "epoch": 0.8755122561238371,
      "grad_norm": 0.4610890746116638,
      "learning_rate": 1.3079764731652772e-06,
      "loss": 0.04261764511466026,
      "step": 6476
    },
    {
      "epoch": 0.8756474494926024,
      "grad_norm": 0.36496952176094055,
      "learning_rate": 1.3051826111322368e-06,
      "loss": 0.04335753247141838,
      "step": 6477
    },
    {
      "epoch": 0.8757826428613676,
      "grad_norm": 0.27545037865638733,
      "learning_rate": 1.3023916004332021e-06,
      "loss": 0.0426146499812603,
      "step": 6478
    },
    {
      "epoch": 0.8759178362301329,
      "grad_norm": 0.3519093096256256,
      "learning_rate": 1.2996034416492847e-06,
      "loss": 0.042835038155317307,
      "step": 6479
    },
    {
      "epoch": 0.8760530295988982,
      "grad_norm": 1.1075360774993896,
      "learning_rate": 1.2968181353609854e-06,
      "loss": 0.060977503657341,
      "step": 6480
    },
    {
      "epoch": 0.8761882229676634,
      "grad_norm": 0.6381919384002686,
      "learning_rate": 1.2940356821482285e-06,
      "loss": 0.05944786220788956,
      "step": 6481
    },
    {
      "epoch": 0.8763234163364287,
      "grad_norm": 0.3496914803981781,
      "learning_rate": 1.291256082590334e-06,
      "loss": 0.06015033274888992,
      "step": 6482
    },
    {
      "epoch": 0.876458609705194,
      "grad_norm": 0.657455563545227,
      "learning_rate": 1.2884793372660208e-06,
      "loss": 0.054166048765182495,
      "step": 6483
    },
    {
      "epoch": 0.8765938030739592,
      "grad_norm": 0.5428779125213623,
      "learning_rate": 1.285705446753433e-06,
      "loss": 0.050331417471170425,
      "step": 6484
    },
    {
      "epoch": 0.8767289964427245,
      "grad_norm": 0.5830390453338623,
      "learning_rate": 1.2829344116301e-06,
      "loss": 0.054551899433135986,
      "step": 6485
    },
    {
      "epoch": 0.8768641898114897,
      "grad_norm": 0.6821373701095581,
      "learning_rate": 1.2801662324729774e-06,
      "loss": 0.050336308777332306,
      "step": 6486
    },
    {
      "epoch": 0.876999383180255,
      "grad_norm": 0.3401375114917755,
      "learning_rate": 1.2774009098584055e-06,
      "loss": 0.059595443308353424,
      "step": 6487
    },
    {
      "epoch": 0.8771345765490203,
      "grad_norm": 0.7410391569137573,
      "learning_rate": 1.274638444362139e-06,
      "loss": 0.05255543068051338,
      "step": 6488
    },
    {
      "epoch": 0.8772697699177855,
      "grad_norm": 1.2307811975479126,
      "learning_rate": 1.2718788365593443e-06,
      "loss": 0.04362833499908447,
      "step": 6489
    },
    {
      "epoch": 0.8774049632865508,
      "grad_norm": 0.7255873084068298,
      "learning_rate": 1.26912208702458e-06,
      "loss": 0.06386878341436386,
      "step": 6490
    },
    {
      "epoch": 0.877540156655316,
      "grad_norm": 0.803435206413269,
      "learning_rate": 1.2663681963318242e-06,
      "loss": 0.06775794178247452,
      "step": 6491
    },
    {
      "epoch": 0.8776753500240814,
      "grad_norm": 0.28998297452926636,
      "learning_rate": 1.2636171650544443e-06,
      "loss": 0.03879409283399582,
      "step": 6492
    },
    {
      "epoch": 0.8778105433928466,
      "grad_norm": 1.466269850730896,
      "learning_rate": 1.260868993765219e-06,
      "loss": 0.06270934641361237,
      "step": 6493
    },
    {
      "epoch": 0.8779457367616118,
      "grad_norm": 0.5335553884506226,
      "learning_rate": 1.258123683036339e-06,
      "loss": 0.05277072638273239,
      "step": 6494
    },
    {
      "epoch": 0.8780809301303771,
      "grad_norm": 0.3859676718711853,
      "learning_rate": 1.2553812334393872e-06,
      "loss": 0.04895465075969696,
      "step": 6495
    },
    {
      "epoch": 0.8782161234991424,
      "grad_norm": 0.323969304561615,
      "learning_rate": 1.2526416455453582e-06,
      "loss": 0.055321067571640015,
      "step": 6496
    },
    {
      "epoch": 0.8783513168679077,
      "grad_norm": 0.3847314417362213,
      "learning_rate": 1.249904919924646e-06,
      "loss": 0.051676392555236816,
      "step": 6497
    },
    {
      "epoch": 0.8784865102366729,
      "grad_norm": 0.23739759624004364,
      "learning_rate": 1.2471710571470579e-06,
      "loss": 0.040119320154190063,
      "step": 6498
    },
    {
      "epoch": 0.8786217036054381,
      "grad_norm": 0.3581050932407379,
      "learning_rate": 1.2444400577817922e-06,
      "loss": 0.05266615375876427,
      "step": 6499
    },
    {
      "epoch": 0.8787568969742034,
      "grad_norm": 0.3567295968532562,
      "learning_rate": 1.2417119223974621e-06,
      "loss": 0.03939438238739967,
      "step": 6500
    },
    {
      "epoch": 0.8788920903429687,
      "grad_norm": 0.3675695061683655,
      "learning_rate": 1.2389866515620768e-06,
      "loss": 0.04629841446876526,
      "step": 6501
    },
    {
      "epoch": 0.8790272837117339,
      "grad_norm": 0.3217466175556183,
      "learning_rate": 1.2362642458430505e-06,
      "loss": 0.046079039573669434,
      "step": 6502
    },
    {
      "epoch": 0.8791624770804992,
      "grad_norm": 0.839565098285675,
      "learning_rate": 1.2335447058072103e-06,
      "loss": 0.0968470498919487,
      "step": 6503
    },
    {
      "epoch": 0.8792976704492644,
      "grad_norm": 0.23487775027751923,
      "learning_rate": 1.230828032020771e-06,
      "loss": 0.03534689545631409,
      "step": 6504
    },
    {
      "epoch": 0.8794328638180298,
      "grad_norm": 0.5126654505729675,
      "learning_rate": 1.2281142250493638e-06,
      "loss": 0.05204305052757263,
      "step": 6505
    },
    {
      "epoch": 0.879568057186795,
      "grad_norm": 1.0494384765625,
      "learning_rate": 1.225403285458015e-06,
      "loss": 0.06012066453695297,
      "step": 6506
    },
    {
      "epoch": 0.8797032505555602,
      "grad_norm": 0.41900715231895447,
      "learning_rate": 1.2226952138111546e-06,
      "loss": 0.05904790014028549,
      "step": 6507
    },
    {
      "epoch": 0.8798384439243255,
      "grad_norm": 0.3915879726409912,
      "learning_rate": 1.219990010672622e-06,
      "loss": 0.04632105678319931,
      "step": 6508
    },
    {
      "epoch": 0.8799736372930907,
      "grad_norm": 1.0778716802597046,
      "learning_rate": 1.2172876766056562e-06,
      "loss": 0.038495685905218124,
      "step": 6509
    },
    {
      "epoch": 0.8801088306618561,
      "grad_norm": 0.7172847986221313,
      "learning_rate": 1.2145882121728906e-06,
      "loss": 0.042837344110012054,
      "step": 6510
    },
    {
      "epoch": 0.8802440240306213,
      "grad_norm": 1.080238699913025,
      "learning_rate": 1.2118916179363727e-06,
      "loss": 0.07236932218074799,
      "step": 6511
    },
    {
      "epoch": 0.8803792173993865,
      "grad_norm": 0.26233533024787903,
      "learning_rate": 1.209197894457546e-06,
      "loss": 0.03946400433778763,
      "step": 6512
    },
    {
      "epoch": 0.8805144107681518,
      "grad_norm": 0.4374561309814453,
      "learning_rate": 1.2065070422972606e-06,
      "loss": 0.056719616055488586,
      "step": 6513
    },
    {
      "epoch": 0.8806496041369171,
      "grad_norm": 0.464121550321579,
      "learning_rate": 1.2038190620157685e-06,
      "loss": 0.055736273527145386,
      "step": 6514
    },
    {
      "epoch": 0.8807847975056824,
      "grad_norm": 0.7671204209327698,
      "learning_rate": 1.2011339541727117e-06,
      "loss": 0.049051813781261444,
      "step": 6515
    },
    {
      "epoch": 0.8809199908744476,
      "grad_norm": 0.628445565700531,
      "learning_rate": 1.198451719327155e-06,
      "loss": 0.059822455048561096,
      "step": 6516
    },
    {
      "epoch": 0.8810551842432128,
      "grad_norm": 0.8739883899688721,
      "learning_rate": 1.1957723580375447e-06,
      "loss": 0.06787683814764023,
      "step": 6517
    },
    {
      "epoch": 0.8811903776119782,
      "grad_norm": 0.8604546189308167,
      "learning_rate": 1.193095870861748e-06,
      "loss": 0.05168576538562775,
      "step": 6518
    },
    {
      "epoch": 0.8813255709807434,
      "grad_norm": 0.5475383996963501,
      "learning_rate": 1.1904222583570156e-06,
      "loss": 0.05231267958879471,
      "step": 6519
    },
    {
      "epoch": 0.8814607643495087,
      "grad_norm": 0.43577030301094055,
      "learning_rate": 1.1877515210800077e-06,
      "loss": 0.047853536903858185,
      "step": 6520
    },
    {
      "epoch": 0.8815959577182739,
      "grad_norm": 0.5603119134902954,
      "learning_rate": 1.1850836595867925e-06,
      "loss": 0.051129672676324844,
      "step": 6521
    },
    {
      "epoch": 0.8817311510870391,
      "grad_norm": 0.5466884970664978,
      "learning_rate": 1.1824186744328259e-06,
      "loss": 0.05245116353034973,
      "step": 6522
    },
    {
      "epoch": 0.8818663444558045,
      "grad_norm": 0.2560189664363861,
      "learning_rate": 1.179756566172982e-06,
      "loss": 0.03251160308718681,
      "step": 6523
    },
    {
      "epoch": 0.8820015378245697,
      "grad_norm": 0.32450202107429504,
      "learning_rate": 1.177097335361516e-06,
      "loss": 0.059180498123168945,
      "step": 6524
    },
    {
      "epoch": 0.882136731193335,
      "grad_norm": 0.2926059067249298,
      "learning_rate": 1.1744409825520969e-06,
      "loss": 0.041620850563049316,
      "step": 6525
    },
    {
      "epoch": 0.8822719245621002,
      "grad_norm": 0.39256179332733154,
      "learning_rate": 1.171787508297792e-06,
      "loss": 0.04282285273075104,
      "step": 6526
    },
    {
      "epoch": 0.8824071179308655,
      "grad_norm": 0.3213786780834198,
      "learning_rate": 1.1691369131510676e-06,
      "loss": 0.034612298011779785,
      "step": 6527
    },
    {
      "epoch": 0.8825423112996308,
      "grad_norm": 0.24405010044574738,
      "learning_rate": 1.1664891976637992e-06,
      "loss": 0.04390066862106323,
      "step": 6528
    },
    {
      "epoch": 0.882677504668396,
      "grad_norm": 0.32655808329582214,
      "learning_rate": 1.1638443623872442e-06,
      "loss": 0.03866888955235481,
      "step": 6529
    },
    {
      "epoch": 0.8828126980371612,
      "grad_norm": 0.3981506824493408,
      "learning_rate": 1.1612024078720752e-06,
      "loss": 0.04642610251903534,
      "step": 6530
    },
    {
      "epoch": 0.8829478914059266,
      "grad_norm": 0.9546152353286743,
      "learning_rate": 1.1585633346683655e-06,
      "loss": 0.059473518282175064,
      "step": 6531
    },
    {
      "epoch": 0.8830830847746918,
      "grad_norm": 0.4268510937690735,
      "learning_rate": 1.155927143325579e-06,
      "loss": 0.047990262508392334,
      "step": 6532
    },
    {
      "epoch": 0.8832182781434571,
      "grad_norm": 1.1467900276184082,
      "learning_rate": 1.1532938343925887e-06,
      "loss": 0.05960758030414581,
      "step": 6533
    },
    {
      "epoch": 0.8833534715122223,
      "grad_norm": 0.44528794288635254,
      "learning_rate": 1.1506634084176587e-06,
      "loss": 0.062300823628902435,
      "step": 6534
    },
    {
      "epoch": 0.8834886648809875,
      "grad_norm": 0.3454953134059906,
      "learning_rate": 1.148035865948463e-06,
      "loss": 0.05014472082257271,
      "step": 6535
    },
    {
      "epoch": 0.8836238582497529,
      "grad_norm": 0.6793503165245056,
      "learning_rate": 1.1454112075320688e-06,
      "loss": 0.03962131589651108,
      "step": 6536
    },
    {
      "epoch": 0.8837590516185181,
      "grad_norm": 0.6821101903915405,
      "learning_rate": 1.1427894337149426e-06,
      "loss": 0.059002310037612915,
      "step": 6537
    },
    {
      "epoch": 0.8838942449872834,
      "grad_norm": 0.4089685082435608,
      "learning_rate": 1.1401705450429506e-06,
      "loss": 0.03756380081176758,
      "step": 6538
    },
    {
      "epoch": 0.8840294383560486,
      "grad_norm": 0.8958247900009155,
      "learning_rate": 1.1375545420613586e-06,
      "loss": 0.05500996112823486,
      "step": 6539
    },
    {
      "epoch": 0.8841646317248139,
      "grad_norm": 0.9188216328620911,
      "learning_rate": 1.1349414253148377e-06,
      "loss": 0.04294709861278534,
      "step": 6540
    },
    {
      "epoch": 0.8842998250935792,
      "grad_norm": 1.2239514589309692,
      "learning_rate": 1.1323311953474524e-06,
      "loss": 0.060585539788007736,
      "step": 6541
    },
    {
      "epoch": 0.8844350184623444,
      "grad_norm": 0.5068191885948181,
      "learning_rate": 1.1297238527026582e-06,
      "loss": 0.05804656445980072,
      "step": 6542
    },
    {
      "epoch": 0.8845702118311097,
      "grad_norm": 0.24202004075050354,
      "learning_rate": 1.1271193979233258e-06,
      "loss": 0.033794134855270386,
      "step": 6543
    },
    {
      "epoch": 0.884705405199875,
      "grad_norm": 0.5629062652587891,
      "learning_rate": 1.1245178315517113e-06,
      "loss": 0.045803967863321304,
      "step": 6544
    },
    {
      "epoch": 0.8848405985686402,
      "grad_norm": 0.6934453845024109,
      "learning_rate": 1.1219191541294798e-06,
      "loss": 0.05704040825366974,
      "step": 6545
    },
    {
      "epoch": 0.8849757919374055,
      "grad_norm": 0.4430544078350067,
      "learning_rate": 1.1193233661976887e-06,
      "loss": 0.043864525854587555,
      "step": 6546
    },
    {
      "epoch": 0.8851109853061707,
      "grad_norm": 0.21920977532863617,
      "learning_rate": 1.1167304682967904e-06,
      "loss": 0.037121668457984924,
      "step": 6547
    },
    {
      "epoch": 0.885246178674936,
      "grad_norm": 0.6567691564559937,
      "learning_rate": 1.114140460966645e-06,
      "loss": 0.04725874215364456,
      "step": 6548
    },
    {
      "epoch": 0.8853813720437013,
      "grad_norm": 0.4458910822868347,
      "learning_rate": 1.111553344746501e-06,
      "loss": 0.04362950101494789,
      "step": 6549
    },
    {
      "epoch": 0.8855165654124665,
      "grad_norm": 0.4701639711856842,
      "learning_rate": 1.1089691201750174e-06,
      "loss": 0.05949559062719345,
      "step": 6550
    },
    {
      "epoch": 0.8856517587812318,
      "grad_norm": 0.6279432773590088,
      "learning_rate": 1.106387787790239e-06,
      "loss": 0.05217616260051727,
      "step": 6551
    },
    {
      "epoch": 0.885786952149997,
      "grad_norm": 0.4219288229942322,
      "learning_rate": 1.1038093481296091e-06,
      "loss": 0.04507826268672943,
      "step": 6552
    },
    {
      "epoch": 0.8859221455187622,
      "grad_norm": 0.6901844143867493,
      "learning_rate": 1.10123380172998e-06,
      "loss": 0.06525152176618576,
      "step": 6553
    },
    {
      "epoch": 0.8860573388875276,
      "grad_norm": 0.5920925736427307,
      "learning_rate": 1.098661149127586e-06,
      "loss": 0.03667248785495758,
      "step": 6554
    },
    {
      "epoch": 0.8861925322562928,
      "grad_norm": 0.7695235013961792,
      "learning_rate": 1.0960913908580788e-06,
      "loss": 0.07185602188110352,
      "step": 6555
    },
    {
      "epoch": 0.8863277256250581,
      "grad_norm": 0.5666546821594238,
      "learning_rate": 1.0935245274564852e-06,
      "loss": 0.07072822749614716,
      "step": 6556
    },
    {
      "epoch": 0.8864629189938233,
      "grad_norm": 0.36242297291755676,
      "learning_rate": 1.0909605594572413e-06,
      "loss": 0.05141735076904297,
      "step": 6557
    },
    {
      "epoch": 0.8865981123625886,
      "grad_norm": 0.2691637873649597,
      "learning_rate": 1.0883994873941816e-06,
      "loss": 0.03618592023849487,
      "step": 6558
    },
    {
      "epoch": 0.8867333057313539,
      "grad_norm": 0.40218985080718994,
      "learning_rate": 1.0858413118005345e-06,
      "loss": 0.04321213811635971,
      "step": 6559
    },
    {
      "epoch": 0.8868684991001191,
      "grad_norm": 1.1676558256149292,
      "learning_rate": 1.0832860332089288e-06,
      "loss": 0.06056632101535797,
      "step": 6560
    },
    {
      "epoch": 0.8870036924688844,
      "grad_norm": 0.5845606923103333,
      "learning_rate": 1.0807336521513828e-06,
      "loss": 0.0678192749619484,
      "step": 6561
    },
    {
      "epoch": 0.8871388858376497,
      "grad_norm": 0.42374172806739807,
      "learning_rate": 1.0781841691593142e-06,
      "loss": 0.05576469749212265,
      "step": 6562
    },
    {
      "epoch": 0.8872740792064149,
      "grad_norm": 0.36073818802833557,
      "learning_rate": 1.0756375847635435e-06,
      "loss": 0.03613992780447006,
      "step": 6563
    },
    {
      "epoch": 0.8874092725751802,
      "grad_norm": 0.6461474299430847,
      "learning_rate": 1.0730938994942818e-06,
      "loss": 0.06881657987833023,
      "step": 6564
    },
    {
      "epoch": 0.8875444659439454,
      "grad_norm": 0.395210862159729,
      "learning_rate": 1.070553113881137e-06,
      "loss": 0.04604113847017288,
      "step": 6565
    },
    {
      "epoch": 0.8876796593127108,
      "grad_norm": 0.48403221368789673,
      "learning_rate": 1.0680152284531158e-06,
      "loss": 0.05795672535896301,
      "step": 6566
    },
    {
      "epoch": 0.887814852681476,
      "grad_norm": 0.4987831115722656,
      "learning_rate": 1.0654802437386157e-06,
      "loss": 0.059023723006248474,
      "step": 6567
    },
    {
      "epoch": 0.8879500460502412,
      "grad_norm": 0.4567299485206604,
      "learning_rate": 1.062948160265438e-06,
      "loss": 0.04587674140930176,
      "step": 6568
    },
    {
      "epoch": 0.8880852394190065,
      "grad_norm": 0.5794356465339661,
      "learning_rate": 1.0604189785607772e-06,
      "loss": 0.06764669716358185,
      "step": 6569
    },
    {
      "epoch": 0.8882204327877717,
      "grad_norm": 1.1217278242111206,
      "learning_rate": 1.0578926991512171e-06,
      "loss": 0.046643126755952835,
      "step": 6570
    },
    {
      "epoch": 0.8883556261565371,
      "grad_norm": 0.4777769446372986,
      "learning_rate": 1.0553693225627458e-06,
      "loss": 0.05171160399913788,
      "step": 6571
    },
    {
      "epoch": 0.8884908195253023,
      "grad_norm": 0.19356529414653778,
      "learning_rate": 1.0528488493207444e-06,
      "loss": 0.029933147132396698,
      "step": 6572
    },
    {
      "epoch": 0.8886260128940675,
      "grad_norm": 0.7756878137588501,
      "learning_rate": 1.0503312799499898e-06,
      "loss": 0.06609708070755005,
      "step": 6573
    },
    {
      "epoch": 0.8887612062628328,
      "grad_norm": 0.7496227622032166,
      "learning_rate": 1.0478166149746476e-06,
      "loss": 0.055043645203113556,
      "step": 6574
    },
    {
      "epoch": 0.888896399631598,
      "grad_norm": 0.6214916110038757,
      "learning_rate": 1.0453048549182892e-06,
      "loss": 0.05154578387737274,
      "step": 6575
    },
    {
      "epoch": 0.8890315930003634,
      "grad_norm": 0.4027038514614105,
      "learning_rate": 1.0427960003038744e-06,
      "loss": 0.03429114818572998,
      "step": 6576
    },
    {
      "epoch": 0.8891667863691286,
      "grad_norm": 0.4690026044845581,
      "learning_rate": 1.040290051653764e-06,
      "loss": 0.04975851997733116,
      "step": 6577
    },
    {
      "epoch": 0.8893019797378938,
      "grad_norm": 0.31448978185653687,
      "learning_rate": 1.0377870094897085e-06,
      "loss": 0.05061821639537811,
      "step": 6578
    },
    {
      "epoch": 0.8894371731066592,
      "grad_norm": 0.3992505967617035,
      "learning_rate": 1.0352868743328497e-06,
      "loss": 0.04014480113983154,
      "step": 6579
    },
    {
      "epoch": 0.8895723664754244,
      "grad_norm": 0.39963221549987793,
      "learning_rate": 1.032789646703733e-06,
      "loss": 0.04242967069149017,
      "step": 6580
    },
    {
      "epoch": 0.8897075598441896,
      "grad_norm": 0.5489478707313538,
      "learning_rate": 1.0302953271222938e-06,
      "loss": 0.055710405111312866,
      "step": 6581
    },
    {
      "epoch": 0.8898427532129549,
      "grad_norm": 0.4911177158355713,
      "learning_rate": 1.0278039161078634e-06,
      "loss": 0.04291833937168121,
      "step": 6582
    },
    {
      "epoch": 0.8899779465817201,
      "grad_norm": 0.3042968809604645,
      "learning_rate": 1.0253154141791705e-06,
      "loss": 0.05094408988952637,
      "step": 6583
    },
    {
      "epoch": 0.8901131399504855,
      "grad_norm": 0.7723742127418518,
      "learning_rate": 1.0228298218543253e-06,
      "loss": 0.05948770046234131,
      "step": 6584
    },
    {
      "epoch": 0.8902483333192507,
      "grad_norm": 0.5254697799682617,
      "learning_rate": 1.020347139650849e-06,
      "loss": 0.053774673491716385,
      "step": 6585
    },
    {
      "epoch": 0.8903835266880159,
      "grad_norm": 0.5848549008369446,
      "learning_rate": 1.0178673680856448e-06,
      "loss": 0.04300014674663544,
      "step": 6586
    },
    {
      "epoch": 0.8905187200567812,
      "grad_norm": 0.5533194541931152,
      "learning_rate": 1.0153905076750196e-06,
      "loss": 0.043958112597465515,
      "step": 6587
    },
    {
      "epoch": 0.8906539134255465,
      "grad_norm": 0.6654220819473267,
      "learning_rate": 1.0129165589346644e-06,
      "loss": 0.04274154081940651,
      "step": 6588
    },
    {
      "epoch": 0.8907891067943118,
      "grad_norm": 0.5236038565635681,
      "learning_rate": 1.0104455223796688e-06,
      "loss": 0.0551171712577343,
      "step": 6589
    },
    {
      "epoch": 0.890924300163077,
      "grad_norm": 0.4612497389316559,
      "learning_rate": 1.0079773985245178e-06,
      "loss": 0.0630609393119812,
      "step": 6590
    },
    {
      "epoch": 0.8910594935318422,
      "grad_norm": 0.6932024359703064,
      "learning_rate": 1.0055121878830837e-06,
      "loss": 0.07203590124845505,
      "step": 6591
    },
    {
      "epoch": 0.8911946869006075,
      "grad_norm": 0.6976272463798523,
      "learning_rate": 1.0030498909686458e-06,
      "loss": 0.05027402192354202,
      "step": 6592
    },
    {
      "epoch": 0.8913298802693728,
      "grad_norm": 1.1622633934020996,
      "learning_rate": 1.0005905082938593e-06,
      "loss": 0.05766216665506363,
      "step": 6593
    },
    {
      "epoch": 0.8914650736381381,
      "grad_norm": 0.5363810062408447,
      "learning_rate": 9.981340403707794e-07,
      "loss": 0.05812821537256241,
      "step": 6594
    },
    {
      "epoch": 0.8916002670069033,
      "grad_norm": 0.5079368948936462,
      "learning_rate": 9.956804877108638e-07,
      "loss": 0.05828285962343216,
      "step": 6595
    },
    {
      "epoch": 0.8917354603756685,
      "grad_norm": 0.3242294490337372,
      "learning_rate": 9.932298508249488e-07,
      "loss": 0.03499762713909149,
      "step": 6596
    },
    {
      "epoch": 0.8918706537444339,
      "grad_norm": 0.328448623418808,
      "learning_rate": 9.907821302232729e-07,
      "loss": 0.0437769815325737,
      "step": 6597
    },
    {
      "epoch": 0.8920058471131991,
      "grad_norm": 0.7776864171028137,
      "learning_rate": 9.883373264154633e-07,
      "loss": 0.05071611702442169,
      "step": 6598
    },
    {
      "epoch": 0.8921410404819644,
      "grad_norm": 1.715150237083435,
      "learning_rate": 9.858954399105397e-07,
      "loss": 0.07196488976478577,
      "step": 6599
    },
    {
      "epoch": 0.8922762338507296,
      "grad_norm": 0.566561222076416,
      "learning_rate": 9.834564712169202e-07,
      "loss": 0.047789350152015686,
      "step": 6600
    },
    {
      "epoch": 0.8924114272194948,
      "grad_norm": 0.8820871710777283,
      "learning_rate": 9.81020420842409e-07,
      "loss": 0.054968468844890594,
      "step": 6601
    },
    {
      "epoch": 0.8925466205882602,
      "grad_norm": 0.2990488111972809,
      "learning_rate": 9.785872892942033e-07,
      "loss": 0.049929387867450714,
      "step": 6602
    },
    {
      "epoch": 0.8926818139570254,
      "grad_norm": 0.34779053926467896,
      "learning_rate": 9.761570770788964e-07,
      "loss": 0.04305468499660492,
      "step": 6603
    },
    {
      "epoch": 0.8928170073257907,
      "grad_norm": 0.2624479830265045,
      "learning_rate": 9.737297847024685e-07,
      "loss": 0.03659946471452713,
      "step": 6604
    },
    {
      "epoch": 0.8929522006945559,
      "grad_norm": 0.6984428763389587,
      "learning_rate": 9.713054126702968e-07,
      "loss": 0.056578680872917175,
      "step": 6605
    },
    {
      "epoch": 0.8930873940633212,
      "grad_norm": 0.2594917416572571,
      "learning_rate": 9.688839614871497e-07,
      "loss": 0.041214700788259506,
      "step": 6606
    },
    {
      "epoch": 0.8932225874320865,
      "grad_norm": 0.6019291281700134,
      "learning_rate": 9.664654316571852e-07,
      "loss": 0.038591306656599045,
      "step": 6607
    },
    {
      "epoch": 0.8933577808008517,
      "grad_norm": 0.37569570541381836,
      "learning_rate": 9.640498236839507e-07,
      "loss": 0.050752900540828705,
      "step": 6608
    },
    {
      "epoch": 0.8934929741696169,
      "grad_norm": 0.5007606744766235,
      "learning_rate": 9.616371380703953e-07,
      "loss": 0.044734060764312744,
      "step": 6609
    },
    {
      "epoch": 0.8936281675383823,
      "grad_norm": 0.3374848961830139,
      "learning_rate": 9.592273753188507e-07,
      "loss": 0.06034480035305023,
      "step": 6610
    },
    {
      "epoch": 0.8937633609071475,
      "grad_norm": 0.957298994064331,
      "learning_rate": 9.568205359310372e-07,
      "loss": 0.0423714816570282,
      "step": 6611
    },
    {
      "epoch": 0.8938985542759128,
      "grad_norm": 0.7147104144096375,
      "learning_rate": 9.544166204080772e-07,
      "loss": 0.04316665232181549,
      "step": 6612
    },
    {
      "epoch": 0.894033747644678,
      "grad_norm": 0.3773733973503113,
      "learning_rate": 9.520156292504739e-07,
      "loss": 0.03957603871822357,
      "step": 6613
    },
    {
      "epoch": 0.8941689410134432,
      "grad_norm": 0.5294026136398315,
      "learning_rate": 9.496175629581322e-07,
      "loss": 0.047300323843955994,
      "step": 6614
    },
    {
      "epoch": 0.8943041343822086,
      "grad_norm": 0.40021997690200806,
      "learning_rate": 9.472224220303427e-07,
      "loss": 0.04881554841995239,
      "step": 6615
    },
    {
      "epoch": 0.8944393277509738,
      "grad_norm": 0.9231643676757812,
      "learning_rate": 9.448302069657799e-07,
      "loss": 0.04621110111474991,
      "step": 6616
    },
    {
      "epoch": 0.8945745211197391,
      "grad_norm": 0.40651726722717285,
      "learning_rate": 9.424409182625205e-07,
      "loss": 0.05630484223365784,
      "step": 6617
    },
    {
      "epoch": 0.8947097144885043,
      "grad_norm": 0.686684787273407,
      "learning_rate": 9.40054556418023e-07,
      "loss": 0.049347031861543655,
      "step": 6618
    },
    {
      "epoch": 0.8948449078572696,
      "grad_norm": 0.4266158640384674,
      "learning_rate": 9.376711219291483e-07,
      "loss": 0.05713128298521042,
      "step": 6619
    },
    {
      "epoch": 0.8949801012260349,
      "grad_norm": 0.2672691345214844,
      "learning_rate": 9.352906152921348e-07,
      "loss": 0.05233769863843918,
      "step": 6620
    },
    {
      "epoch": 0.8951152945948001,
      "grad_norm": 0.7209280729293823,
      "learning_rate": 9.32913037002614e-07,
      "loss": 0.04346020147204399,
      "step": 6621
    },
    {
      "epoch": 0.8952504879635654,
      "grad_norm": 0.788275420665741,
      "learning_rate": 9.30538387555615e-07,
      "loss": 0.04618893563747406,
      "step": 6622
    },
    {
      "epoch": 0.8953856813323307,
      "grad_norm": 0.6408319473266602,
      "learning_rate": 9.281666674455508e-07,
      "loss": 0.04850313067436218,
      "step": 6623
    },
    {
      "epoch": 0.8955208747010959,
      "grad_norm": 0.6206988096237183,
      "learning_rate": 9.257978771662295e-07,
      "loss": 0.054072365164756775,
      "step": 6624
    },
    {
      "epoch": 0.8956560680698612,
      "grad_norm": 0.3479498028755188,
      "learning_rate": 9.234320172108418e-07,
      "loss": 0.06309705972671509,
      "step": 6625
    },
    {
      "epoch": 0.8957912614386264,
      "grad_norm": 1.3672723770141602,
      "learning_rate": 9.210690880719719e-07,
      "loss": 0.05391441285610199,
      "step": 6626
    },
    {
      "epoch": 0.8959264548073917,
      "grad_norm": 0.3593200445175171,
      "learning_rate": 9.187090902415962e-07,
      "loss": 0.04264727234840393,
      "step": 6627
    },
    {
      "epoch": 0.896061648176157,
      "grad_norm": 0.5666792988777161,
      "learning_rate": 9.163520242110784e-07,
      "loss": 0.04747620224952698,
      "step": 6628
    },
    {
      "epoch": 0.8961968415449222,
      "grad_norm": 0.4805433452129364,
      "learning_rate": 9.13997890471176e-07,
      "loss": 0.04367192089557648,
      "step": 6629
    },
    {
      "epoch": 0.8963320349136875,
      "grad_norm": 0.607454240322113,
      "learning_rate": 9.116466895120251e-07,
      "loss": 0.0377173125743866,
      "step": 6630
    },
    {
      "epoch": 0.8964672282824527,
      "grad_norm": 0.8191885352134705,
      "learning_rate": 9.092984218231609e-07,
      "loss": 0.052013371139764786,
      "step": 6631
    },
    {
      "epoch": 0.8966024216512181,
      "grad_norm": 0.7560107707977295,
      "learning_rate": 9.069530878935072e-07,
      "loss": 0.050231315195560455,
      "step": 6632
    },
    {
      "epoch": 0.8967376150199833,
      "grad_norm": 0.40043050050735474,
      "learning_rate": 9.046106882113753e-07,
      "loss": 0.04714782536029816,
      "step": 6633
    },
    {
      "epoch": 0.8968728083887485,
      "grad_norm": 0.46625396609306335,
      "learning_rate": 9.022712232644631e-07,
      "loss": 0.04685495048761368,
      "step": 6634
    },
    {
      "epoch": 0.8970080017575138,
      "grad_norm": 0.6381518840789795,
      "learning_rate": 8.999346935398611e-07,
      "loss": 0.06336129456758499,
      "step": 6635
    },
    {
      "epoch": 0.897143195126279,
      "grad_norm": 0.6549854278564453,
      "learning_rate": 8.976010995240436e-07,
      "loss": 0.06682717800140381,
      "step": 6636
    },
    {
      "epoch": 0.8972783884950443,
      "grad_norm": 0.9094131588935852,
      "learning_rate": 8.952704417028818e-07,
      "loss": 0.054925426840782166,
      "step": 6637
    },
    {
      "epoch": 0.8974135818638096,
      "grad_norm": 0.8842827081680298,
      "learning_rate": 8.929427205616308e-07,
      "loss": 0.061176784336566925,
      "step": 6638
    },
    {
      "epoch": 0.8975487752325748,
      "grad_norm": 0.6502466797828674,
      "learning_rate": 8.906179365849332e-07,
      "loss": 0.04725296422839165,
      "step": 6639
    },
    {
      "epoch": 0.8976839686013401,
      "grad_norm": 0.5638745427131653,
      "learning_rate": 8.882960902568216e-07,
      "loss": 0.04483944922685623,
      "step": 6640
    },
    {
      "epoch": 0.8978191619701054,
      "grad_norm": 0.5545969605445862,
      "learning_rate": 8.85977182060716e-07,
      "loss": 0.04482412338256836,
      "step": 6641
    },
    {
      "epoch": 0.8979543553388706,
      "grad_norm": 0.5826829671859741,
      "learning_rate": 8.836612124794285e-07,
      "loss": 0.061395928263664246,
      "step": 6642
    },
    {
      "epoch": 0.8980895487076359,
      "grad_norm": 0.9112701416015625,
      "learning_rate": 8.813481819951502e-07,
      "loss": 0.05944044142961502,
      "step": 6643
    },
    {
      "epoch": 0.8982247420764011,
      "grad_norm": 0.33140549063682556,
      "learning_rate": 8.790380910894724e-07,
      "loss": 0.04346524924039841,
      "step": 6644
    },
    {
      "epoch": 0.8983599354451665,
      "grad_norm": 0.8436567187309265,
      "learning_rate": 8.767309402433671e-07,
      "loss": 0.052305594086647034,
      "step": 6645
    },
    {
      "epoch": 0.8984951288139317,
      "grad_norm": 0.32180801033973694,
      "learning_rate": 8.744267299371917e-07,
      "loss": 0.04961639642715454,
      "step": 6646
    },
    {
      "epoch": 0.8986303221826969,
      "grad_norm": 0.40363961458206177,
      "learning_rate": 8.721254606507023e-07,
      "loss": 0.052960917353630066,
      "step": 6647
    },
    {
      "epoch": 0.8987655155514622,
      "grad_norm": 0.8495690822601318,
      "learning_rate": 8.698271328630275e-07,
      "loss": 0.07176923006772995,
      "step": 6648
    },
    {
      "epoch": 0.8989007089202274,
      "grad_norm": 0.30365681648254395,
      "learning_rate": 8.675317470526961e-07,
      "loss": 0.04673558101058006,
      "step": 6649
    },
    {
      "epoch": 0.8990359022889928,
      "grad_norm": 0.7263216972351074,
      "learning_rate": 8.652393036976159e-07,
      "loss": 0.04724418371915817,
      "step": 6650
    },
    {
      "epoch": 0.899171095657758,
      "grad_norm": 0.47684288024902344,
      "learning_rate": 8.629498032750916e-07,
      "loss": 0.03681980073451996,
      "step": 6651
    },
    {
      "epoch": 0.8993062890265232,
      "grad_norm": 1.1259374618530273,
      "learning_rate": 8.606632462618069e-07,
      "loss": 0.07840201258659363,
      "step": 6652
    },
    {
      "epoch": 0.8994414823952885,
      "grad_norm": 0.8104427456855774,
      "learning_rate": 8.583796331338311e-07,
      "loss": 0.058555424213409424,
      "step": 6653
    },
    {
      "epoch": 0.8995766757640538,
      "grad_norm": 0.6018322706222534,
      "learning_rate": 8.560989643666306e-07,
      "loss": 0.05421757698059082,
      "step": 6654
    },
    {
      "epoch": 0.8997118691328191,
      "grad_norm": 0.478898823261261,
      "learning_rate": 8.538212404350471e-07,
      "loss": 0.048683859407901764,
      "step": 6655
    },
    {
      "epoch": 0.8998470625015843,
      "grad_norm": 0.6437026262283325,
      "learning_rate": 8.515464618133228e-07,
      "loss": 0.048388078808784485,
      "step": 6656
    },
    {
      "epoch": 0.8999822558703495,
      "grad_norm": 0.513083815574646,
      "learning_rate": 8.492746289750725e-07,
      "loss": 0.028460489585995674,
      "step": 6657
    },
    {
      "epoch": 0.9001174492391149,
      "grad_norm": 0.2022758424282074,
      "learning_rate": 8.470057423933026e-07,
      "loss": 0.039095520973205566,
      "step": 6658
    },
    {
      "epoch": 0.9002526426078801,
      "grad_norm": 0.21069441735744476,
      "learning_rate": 8.447398025404118e-07,
      "loss": 0.048804871737957,
      "step": 6659
    },
    {
      "epoch": 0.9003878359766454,
      "grad_norm": 0.31124478578567505,
      "learning_rate": 8.42476809888178e-07,
      "loss": 0.04619381204247475,
      "step": 6660
    },
    {
      "epoch": 0.9005230293454106,
      "grad_norm": 0.6880770921707153,
      "learning_rate": 8.402167649077725e-07,
      "loss": 0.05387970805168152,
      "step": 6661
    },
    {
      "epoch": 0.9006582227141758,
      "grad_norm": 0.9850658178329468,
      "learning_rate": 8.379596680697454e-07,
      "loss": 0.058292463421821594,
      "step": 6662
    },
    {
      "epoch": 0.9007934160829412,
      "grad_norm": 0.7874597311019897,
      "learning_rate": 8.357055198440328e-07,
      "loss": 0.0476127564907074,
      "step": 6663
    },
    {
      "epoch": 0.9009286094517064,
      "grad_norm": 0.9424549341201782,
      "learning_rate": 8.334543206999673e-07,
      "loss": 0.06544631719589233,
      "step": 6664
    },
    {
      "epoch": 0.9010638028204716,
      "grad_norm": 0.8563365340232849,
      "learning_rate": 8.312060711062558e-07,
      "loss": 0.07613970339298248,
      "step": 6665
    },
    {
      "epoch": 0.9011989961892369,
      "grad_norm": 0.9180619120597839,
      "learning_rate": 8.289607715309988e-07,
      "loss": 0.058623701333999634,
      "step": 6666
    },
    {
      "epoch": 0.9013341895580022,
      "grad_norm": 0.31008583307266235,
      "learning_rate": 8.267184224416791e-07,
      "loss": 0.06269200891256332,
      "step": 6667
    },
    {
      "epoch": 0.9014693829267675,
      "grad_norm": 0.631394624710083,
      "learning_rate": 8.244790243051614e-07,
      "loss": 0.05464514344930649,
      "step": 6668
    },
    {
      "epoch": 0.9016045762955327,
      "grad_norm": 0.3147026300430298,
      "learning_rate": 8.222425775877079e-07,
      "loss": 0.03973004221916199,
      "step": 6669
    },
    {
      "epoch": 0.9017397696642979,
      "grad_norm": 1.0921937227249146,
      "learning_rate": 8.200090827549527e-07,
      "loss": 0.06183578073978424,
      "step": 6670
    },
    {
      "epoch": 0.9018749630330632,
      "grad_norm": 0.43494585156440735,
      "learning_rate": 8.17778540271924e-07,
      "loss": 0.03694123029708862,
      "step": 6671
    },
    {
      "epoch": 0.9020101564018285,
      "grad_norm": 0.34973108768463135,
      "learning_rate": 8.155509506030334e-07,
      "loss": 0.04598543792963028,
      "step": 6672
    },
    {
      "epoch": 0.9021453497705938,
      "grad_norm": 0.7542912364006042,
      "learning_rate": 8.133263142120717e-07,
      "loss": 0.05929616093635559,
      "step": 6673
    },
    {
      "epoch": 0.902280543139359,
      "grad_norm": 0.28905144333839417,
      "learning_rate": 8.111046315622284e-07,
      "loss": 0.05378742143511772,
      "step": 6674
    },
    {
      "epoch": 0.9024157365081242,
      "grad_norm": 0.5028477907180786,
      "learning_rate": 8.088859031160633e-07,
      "loss": 0.06173387169837952,
      "step": 6675
    },
    {
      "epoch": 0.9025509298768896,
      "grad_norm": 0.5621204376220703,
      "learning_rate": 8.066701293355288e-07,
      "loss": 0.04248432815074921,
      "step": 6676
    },
    {
      "epoch": 0.9026861232456548,
      "grad_norm": 0.40853118896484375,
      "learning_rate": 8.044573106819625e-07,
      "loss": 0.06189492344856262,
      "step": 6677
    },
    {
      "epoch": 0.9028213166144201,
      "grad_norm": 1.013044834136963,
      "learning_rate": 8.022474476160824e-07,
      "loss": 0.05431247875094414,
      "step": 6678
    },
    {
      "epoch": 0.9029565099831853,
      "grad_norm": 0.3616791069507599,
      "learning_rate": 8.000405405979988e-07,
      "loss": 0.040154121816158295,
      "step": 6679
    },
    {
      "epoch": 0.9030917033519505,
      "grad_norm": 0.3997938334941864,
      "learning_rate": 7.978365900871943e-07,
      "loss": 0.042385224252939224,
      "step": 6680
    },
    {
      "epoch": 0.9032268967207159,
      "grad_norm": 0.6154190301895142,
      "learning_rate": 7.956355965425482e-07,
      "loss": 0.04839048534631729,
      "step": 6681
    },
    {
      "epoch": 0.9033620900894811,
      "grad_norm": 0.43059805035591125,
      "learning_rate": 7.934375604223193e-07,
      "loss": 0.05832874774932861,
      "step": 6682
    },
    {
      "epoch": 0.9034972834582464,
      "grad_norm": 0.7557125091552734,
      "learning_rate": 7.912424821841463e-07,
      "loss": 0.050268713384866714,
      "step": 6683
    },
    {
      "epoch": 0.9036324768270116,
      "grad_norm": 0.39958083629608154,
      "learning_rate": 7.89050362285062e-07,
      "loss": 0.046776190400123596,
      "step": 6684
    },
    {
      "epoch": 0.9037676701957769,
      "grad_norm": 0.45176929235458374,
      "learning_rate": 7.868612011814713e-07,
      "loss": 0.06302172690629959,
      "step": 6685
    },
    {
      "epoch": 0.9039028635645422,
      "grad_norm": 0.8538007140159607,
      "learning_rate": 7.846749993291746e-07,
      "loss": 0.03994462639093399,
      "step": 6686
    },
    {
      "epoch": 0.9040380569333074,
      "grad_norm": 0.4270612299442291,
      "learning_rate": 7.824917571833445e-07,
      "loss": 0.0590907484292984,
      "step": 6687
    },
    {
      "epoch": 0.9041732503020726,
      "grad_norm": 0.5902412533760071,
      "learning_rate": 7.80311475198554e-07,
      "loss": 0.048745639622211456,
      "step": 6688
    },
    {
      "epoch": 0.904308443670838,
      "grad_norm": 0.642102837562561,
      "learning_rate": 7.781341538287384e-07,
      "loss": 0.04547517001628876,
      "step": 6689
    },
    {
      "epoch": 0.9044436370396032,
      "grad_norm": 0.8880202174186707,
      "learning_rate": 7.759597935272316e-07,
      "loss": 0.052772752940654755,
      "step": 6690
    },
    {
      "epoch": 0.9045788304083685,
      "grad_norm": 0.4674089848995209,
      "learning_rate": 7.7378839474675e-07,
      "loss": 0.05677032470703125,
      "step": 6691
    },
    {
      "epoch": 0.9047140237771337,
      "grad_norm": 0.9701254963874817,
      "learning_rate": 7.716199579393851e-07,
      "loss": 0.06401762366294861,
      "step": 6692
    },
    {
      "epoch": 0.9048492171458989,
      "grad_norm": 0.5477458834648132,
      "learning_rate": 7.694544835566259e-07,
      "loss": 0.05173102766275406,
      "step": 6693
    },
    {
      "epoch": 0.9049844105146643,
      "grad_norm": 0.9523248076438904,
      "learning_rate": 7.672919720493249e-07,
      "loss": 0.06883459538221359,
      "step": 6694
    },
    {
      "epoch": 0.9051196038834295,
      "grad_norm": 0.44439083337783813,
      "learning_rate": 7.651324238677338e-07,
      "loss": 0.0566340833902359,
      "step": 6695
    },
    {
      "epoch": 0.9052547972521948,
      "grad_norm": 0.6270071268081665,
      "learning_rate": 7.629758394614828e-07,
      "loss": 0.0413985475897789,
      "step": 6696
    },
    {
      "epoch": 0.90538999062096,
      "grad_norm": 0.5257025361061096,
      "learning_rate": 7.608222192795794e-07,
      "loss": 0.051652226597070694,
      "step": 6697
    },
    {
      "epoch": 0.9055251839897253,
      "grad_norm": 0.5392767786979675,
      "learning_rate": 7.586715637704284e-07,
      "loss": 0.051113277673721313,
      "step": 6698
    },
    {
      "epoch": 0.9056603773584906,
      "grad_norm": 0.5225648283958435,
      "learning_rate": 7.565238733817998e-07,
      "loss": 0.05162392556667328,
      "step": 6699
    },
    {
      "epoch": 0.9057955707272558,
      "grad_norm": 0.845072865486145,
      "learning_rate": 7.543791485608542e-07,
      "loss": 0.05560246855020523,
      "step": 6700
    },
    {
      "epoch": 0.9059307640960211,
      "grad_norm": 0.5046133995056152,
      "learning_rate": 7.52237389754138e-07,
      "loss": 0.0519113764166832,
      "step": 6701
    },
    {
      "epoch": 0.9060659574647864,
      "grad_norm": 0.6372543573379517,
      "learning_rate": 7.500985974075758e-07,
      "loss": 0.06785418838262558,
      "step": 6702
    },
    {
      "epoch": 0.9062011508335516,
      "grad_norm": 0.48460158705711365,
      "learning_rate": 7.479627719664767e-07,
      "loss": 0.05035112053155899,
      "step": 6703
    },
    {
      "epoch": 0.9063363442023169,
      "grad_norm": 0.39503371715545654,
      "learning_rate": 7.458299138755281e-07,
      "loss": 0.05464310199022293,
      "step": 6704
    },
    {
      "epoch": 0.9064715375710821,
      "grad_norm": 0.5971314907073975,
      "learning_rate": 7.437000235788033e-07,
      "loss": 0.0529952310025692,
      "step": 6705
    },
    {
      "epoch": 0.9066067309398474,
      "grad_norm": 0.768519401550293,
      "learning_rate": 7.415731015197575e-07,
      "loss": 0.045520246028900146,
      "step": 6706
    },
    {
      "epoch": 0.9067419243086127,
      "grad_norm": 0.21944670379161835,
      "learning_rate": 7.39449148141228e-07,
      "loss": 0.03883674368262291,
      "step": 6707
    },
    {
      "epoch": 0.9068771176773779,
      "grad_norm": 0.5327578186988831,
      "learning_rate": 7.373281638854329e-07,
      "loss": 0.04125240445137024,
      "step": 6708
    },
    {
      "epoch": 0.9070123110461432,
      "grad_norm": 0.4574945867061615,
      "learning_rate": 7.352101491939722e-07,
      "loss": 0.05063946545124054,
      "step": 6709
    },
    {
      "epoch": 0.9071475044149084,
      "grad_norm": 0.954848051071167,
      "learning_rate": 7.330951045078249e-07,
      "loss": 0.07672044634819031,
      "step": 6710
    },
    {
      "epoch": 0.9072826977836738,
      "grad_norm": 0.28551214933395386,
      "learning_rate": 7.309830302673621e-07,
      "loss": 0.04225584492087364,
      "step": 6711
    },
    {
      "epoch": 0.907417891152439,
      "grad_norm": 0.7695335149765015,
      "learning_rate": 7.288739269123184e-07,
      "loss": 0.06044097617268562,
      "step": 6712
    },
    {
      "epoch": 0.9075530845212042,
      "grad_norm": 0.5745910406112671,
      "learning_rate": 7.267677948818296e-07,
      "loss": 0.04879125952720642,
      "step": 6713
    },
    {
      "epoch": 0.9076882778899695,
      "grad_norm": 0.2723865807056427,
      "learning_rate": 7.246646346143997e-07,
      "loss": 0.03120660036802292,
      "step": 6714
    },
    {
      "epoch": 0.9078234712587347,
      "grad_norm": 0.568640947341919,
      "learning_rate": 7.225644465479153e-07,
      "loss": 0.04885981231927872,
      "step": 6715
    },
    {
      "epoch": 0.9079586646275,
      "grad_norm": 0.988162636756897,
      "learning_rate": 7.204672311196547e-07,
      "loss": 0.05720759928226471,
      "step": 6716
    },
    {
      "epoch": 0.9080938579962653,
      "grad_norm": 0.6753554940223694,
      "learning_rate": 7.183729887662604e-07,
      "loss": 0.044138528406620026,
      "step": 6717
    },
    {
      "epoch": 0.9082290513650305,
      "grad_norm": 0.2969352900981903,
      "learning_rate": 7.162817199237703e-07,
      "loss": 0.041479967534542084,
      "step": 6718
    },
    {
      "epoch": 0.9083642447337958,
      "grad_norm": 0.5062270164489746,
      "learning_rate": 7.141934250275978e-07,
      "loss": 0.07195169478654861,
      "step": 6719
    },
    {
      "epoch": 0.9084994381025611,
      "grad_norm": 1.2973712682724,
      "learning_rate": 7.121081045125316e-07,
      "loss": 0.05476060509681702,
      "step": 6720
    },
    {
      "epoch": 0.9086346314713263,
      "grad_norm": 0.7226802706718445,
      "learning_rate": 7.100257588127545e-07,
      "loss": 0.0396604910492897,
      "step": 6721
    },
    {
      "epoch": 0.9087698248400916,
      "grad_norm": 0.5772003531455994,
      "learning_rate": 7.079463883618148e-07,
      "loss": 0.032029759138822556,
      "step": 6722
    },
    {
      "epoch": 0.9089050182088568,
      "grad_norm": 0.5488506555557251,
      "learning_rate": 7.058699935926527e-07,
      "loss": 0.05793087184429169,
      "step": 6723
    },
    {
      "epoch": 0.9090402115776222,
      "grad_norm": 0.39535918831825256,
      "learning_rate": 7.037965749375808e-07,
      "loss": 0.04278813302516937,
      "step": 6724
    },
    {
      "epoch": 0.9091754049463874,
      "grad_norm": 0.7367969155311584,
      "learning_rate": 7.017261328283037e-07,
      "loss": 0.06994271278381348,
      "step": 6725
    },
    {
      "epoch": 0.9093105983151526,
      "grad_norm": 0.43338915705680847,
      "learning_rate": 6.996586676958916e-07,
      "loss": 0.039698973298072815,
      "step": 6726
    },
    {
      "epoch": 0.9094457916839179,
      "grad_norm": 0.3456937074661255,
      "learning_rate": 6.975941799708019e-07,
      "loss": 0.03696325793862343,
      "step": 6727
    },
    {
      "epoch": 0.9095809850526831,
      "grad_norm": 1.5028387308120728,
      "learning_rate": 6.955326700828757e-07,
      "loss": 0.07935422658920288,
      "step": 6728
    },
    {
      "epoch": 0.9097161784214485,
      "grad_norm": 0.29515862464904785,
      "learning_rate": 6.934741384613246e-07,
      "loss": 0.04717147350311279,
      "step": 6729
    },
    {
      "epoch": 0.9098513717902137,
      "grad_norm": 0.20951049029827118,
      "learning_rate": 6.91418585534756e-07,
      "loss": 0.041094474494457245,
      "step": 6730
    },
    {
      "epoch": 0.9099865651589789,
      "grad_norm": 0.29584041237831116,
      "learning_rate": 6.893660117311373e-07,
      "loss": 0.04478374123573303,
      "step": 6731
    },
    {
      "epoch": 0.9101217585277442,
      "grad_norm": 0.4299290180206299,
      "learning_rate": 6.873164174778252e-07,
      "loss": 0.05629752576351166,
      "step": 6732
    },
    {
      "epoch": 0.9102569518965095,
      "grad_norm": 1.201492190361023,
      "learning_rate": 6.852698032015631e-07,
      "loss": 0.0653962567448616,
      "step": 6733
    },
    {
      "epoch": 0.9103921452652748,
      "grad_norm": 0.7219903469085693,
      "learning_rate": 6.832261693284636e-07,
      "loss": 0.06231025606393814,
      "step": 6734
    },
    {
      "epoch": 0.91052733863404,
      "grad_norm": 1.453986406326294,
      "learning_rate": 6.811855162840214e-07,
      "loss": 0.06157355010509491,
      "step": 6735
    },
    {
      "epoch": 0.9106625320028052,
      "grad_norm": 0.48053452372550964,
      "learning_rate": 6.791478444931132e-07,
      "loss": 0.04321175813674927,
      "step": 6736
    },
    {
      "epoch": 0.9107977253715706,
      "grad_norm": 1.0145496129989624,
      "learning_rate": 6.77113154379988e-07,
      "loss": 0.049371764063835144,
      "step": 6737
    },
    {
      "epoch": 0.9109329187403358,
      "grad_norm": 0.3684634864330292,
      "learning_rate": 6.75081446368287e-07,
      "loss": 0.05167236179113388,
      "step": 6738
    },
    {
      "epoch": 0.9110681121091011,
      "grad_norm": 1.0186936855316162,
      "learning_rate": 6.730527208810166e-07,
      "loss": 0.05943696200847626,
      "step": 6739
    },
    {
      "epoch": 0.9112033054778663,
      "grad_norm": 1.1795024871826172,
      "learning_rate": 6.710269783405709e-07,
      "loss": 0.07245517522096634,
      "step": 6740
    },
    {
      "epoch": 0.9113384988466315,
      "grad_norm": 1.3816347122192383,
      "learning_rate": 6.690042191687206e-07,
      "loss": 0.07006622105836868,
      "step": 6741
    },
    {
      "epoch": 0.9114736922153969,
      "grad_norm": 0.36937466263771057,
      "learning_rate": 6.669844437866124e-07,
      "loss": 0.04919500648975372,
      "step": 6742
    },
    {
      "epoch": 0.9116088855841621,
      "grad_norm": 0.8501532077789307,
      "learning_rate": 6.649676526147764e-07,
      "loss": 0.06431706994771957,
      "step": 6743
    },
    {
      "epoch": 0.9117440789529273,
      "grad_norm": 0.3945702314376831,
      "learning_rate": 6.629538460731199e-07,
      "loss": 0.05623989552259445,
      "step": 6744
    },
    {
      "epoch": 0.9118792723216926,
      "grad_norm": 0.4328218698501587,
      "learning_rate": 6.609430245809261e-07,
      "loss": 0.04234835505485535,
      "step": 6745
    },
    {
      "epoch": 0.9120144656904579,
      "grad_norm": 0.42859089374542236,
      "learning_rate": 6.589351885568617e-07,
      "loss": 0.07000689953565598,
      "step": 6746
    },
    {
      "epoch": 0.9121496590592232,
      "grad_norm": 0.29595860838890076,
      "learning_rate": 6.569303384189624e-07,
      "loss": 0.030823007225990295,
      "step": 6747
    },
    {
      "epoch": 0.9122848524279884,
      "grad_norm": 0.6928752064704895,
      "learning_rate": 6.54928474584659e-07,
      "loss": 0.04525993764400482,
      "step": 6748
    },
    {
      "epoch": 0.9124200457967536,
      "grad_norm": 0.5369593501091003,
      "learning_rate": 6.5292959747074e-07,
      "loss": 0.045181408524513245,
      "step": 6749
    },
    {
      "epoch": 0.912555239165519,
      "grad_norm": 0.24417948722839355,
      "learning_rate": 6.509337074933891e-07,
      "loss": 0.04770684242248535,
      "step": 6750
    },
    {
      "epoch": 0.9126904325342842,
      "grad_norm": 0.6596533060073853,
      "learning_rate": 6.489408050681589e-07,
      "loss": 0.05600264295935631,
      "step": 6751
    },
    {
      "epoch": 0.9128256259030495,
      "grad_norm": 0.49990686774253845,
      "learning_rate": 6.469508906099792e-07,
      "loss": 0.05889815092086792,
      "step": 6752
    },
    {
      "epoch": 0.9129608192718147,
      "grad_norm": 0.7795887589454651,
      "learning_rate": 6.449639645331684e-07,
      "loss": 0.05389513075351715,
      "step": 6753
    },
    {
      "epoch": 0.9130960126405799,
      "grad_norm": 0.3981202244758606,
      "learning_rate": 6.429800272514058e-07,
      "loss": 0.04817391186952591,
      "step": 6754
    },
    {
      "epoch": 0.9132312060093453,
      "grad_norm": 0.2686627209186554,
      "learning_rate": 6.409990791777659e-07,
      "loss": 0.03644941747188568,
      "step": 6755
    },
    {
      "epoch": 0.9133663993781105,
      "grad_norm": 0.5846071243286133,
      "learning_rate": 6.390211207246888e-07,
      "loss": 0.06171203777194023,
      "step": 6756
    },
    {
      "epoch": 0.9135015927468758,
      "grad_norm": 0.24295461177825928,
      "learning_rate": 6.370461523039967e-07,
      "loss": 0.038843024522066116,
      "step": 6757
    },
    {
      "epoch": 0.913636786115641,
      "grad_norm": 0.4861011207103729,
      "learning_rate": 6.350741743268873e-07,
      "loss": 0.0579574853181839,
      "step": 6758
    },
    {
      "epoch": 0.9137719794844062,
      "grad_norm": 0.49936577677726746,
      "learning_rate": 6.331051872039373e-07,
      "loss": 0.07107189297676086,
      "step": 6759
    },
    {
      "epoch": 0.9139071728531716,
      "grad_norm": 0.5456699132919312,
      "learning_rate": 6.31139191345102e-07,
      "loss": 0.06423240900039673,
      "step": 6760
    },
    {
      "epoch": 0.9140423662219368,
      "grad_norm": 0.3858813941478729,
      "learning_rate": 6.291761871597091e-07,
      "loss": 0.05905080586671829,
      "step": 6761
    },
    {
      "epoch": 0.9141775595907021,
      "grad_norm": 0.8678537607192993,
      "learning_rate": 6.272161750564731e-07,
      "loss": 0.04089200496673584,
      "step": 6762
    },
    {
      "epoch": 0.9143127529594673,
      "grad_norm": 0.25096315145492554,
      "learning_rate": 6.252591554434728e-07,
      "loss": 0.03662240505218506,
      "step": 6763
    },
    {
      "epoch": 0.9144479463282326,
      "grad_norm": 0.9206839799880981,
      "learning_rate": 6.233051287281688e-07,
      "loss": 0.06119590997695923,
      "step": 6764
    },
    {
      "epoch": 0.9145831396969979,
      "grad_norm": 0.43097183108329773,
      "learning_rate": 6.213540953174057e-07,
      "loss": 0.06014223396778107,
      "step": 6765
    },
    {
      "epoch": 0.9147183330657631,
      "grad_norm": 0.441585510969162,
      "learning_rate": 6.194060556173953e-07,
      "loss": 0.044751644134521484,
      "step": 6766
    },
    {
      "epoch": 0.9148535264345284,
      "grad_norm": 0.6441513895988464,
      "learning_rate": 6.17461010033733e-07,
      "loss": 0.06827010214328766,
      "step": 6767
    },
    {
      "epoch": 0.9149887198032937,
      "grad_norm": 0.22242850065231323,
      "learning_rate": 6.155189589713833e-07,
      "loss": 0.04135541617870331,
      "step": 6768
    },
    {
      "epoch": 0.9151239131720589,
      "grad_norm": 0.4358827471733093,
      "learning_rate": 6.135799028346928e-07,
      "loss": 0.048567309975624084,
      "step": 6769
    },
    {
      "epoch": 0.9152591065408242,
      "grad_norm": 0.3070225417613983,
      "learning_rate": 6.116438420273868e-07,
      "loss": 0.052805133163928986,
      "step": 6770
    },
    {
      "epoch": 0.9153942999095894,
      "grad_norm": 0.6133350729942322,
      "learning_rate": 6.097107769525595e-07,
      "loss": 0.05564849451184273,
      "step": 6771
    },
    {
      "epoch": 0.9155294932783546,
      "grad_norm": 0.32048967480659485,
      "learning_rate": 6.077807080126873e-07,
      "loss": 0.05249929428100586,
      "step": 6772
    },
    {
      "epoch": 0.91566468664712,
      "grad_norm": 0.4419393241405487,
      "learning_rate": 6.058536356096206e-07,
      "loss": 0.05552881956100464,
      "step": 6773
    },
    {
      "epoch": 0.9157998800158852,
      "grad_norm": 0.4304952323436737,
      "learning_rate": 6.039295601445833e-07,
      "loss": 0.059729404747486115,
      "step": 6774
    },
    {
      "epoch": 0.9159350733846505,
      "grad_norm": 0.6956761479377747,
      "learning_rate": 6.020084820181831e-07,
      "loss": 0.051030222326517105,
      "step": 6775
    },
    {
      "epoch": 0.9160702667534157,
      "grad_norm": 0.27552148699760437,
      "learning_rate": 6.000904016303971e-07,
      "loss": 0.03568500652909279,
      "step": 6776
    },
    {
      "epoch": 0.916205460122181,
      "grad_norm": 0.6241278648376465,
      "learning_rate": 5.981753193805789e-07,
      "loss": 0.04082517698407173,
      "step": 6777
    },
    {
      "epoch": 0.9163406534909463,
      "grad_norm": 0.3620690703392029,
      "learning_rate": 5.962632356674597e-07,
      "loss": 0.04518449306488037,
      "step": 6778
    },
    {
      "epoch": 0.9164758468597115,
      "grad_norm": 0.4352056384086609,
      "learning_rate": 5.94354150889141e-07,
      "loss": 0.05573827028274536,
      "step": 6779
    },
    {
      "epoch": 0.9166110402284768,
      "grad_norm": 0.5349669456481934,
      "learning_rate": 5.924480654431147e-07,
      "loss": 0.0477016344666481,
      "step": 6780
    },
    {
      "epoch": 0.916746233597242,
      "grad_norm": 0.7191454172134399,
      "learning_rate": 5.905449797262252e-07,
      "loss": 0.05366852134466171,
      "step": 6781
    },
    {
      "epoch": 0.9168814269660073,
      "grad_norm": 0.9488183856010437,
      "learning_rate": 5.886448941347156e-07,
      "loss": 0.06195848435163498,
      "step": 6782
    },
    {
      "epoch": 0.9170166203347726,
      "grad_norm": 0.3767930865287781,
      "learning_rate": 5.867478090641892e-07,
      "loss": 0.04960257560014725,
      "step": 6783
    },
    {
      "epoch": 0.9171518137035378,
      "grad_norm": 0.26893851161003113,
      "learning_rate": 5.848537249096269e-07,
      "loss": 0.03197500482201576,
      "step": 6784
    },
    {
      "epoch": 0.9172870070723032,
      "grad_norm": 0.5486786365509033,
      "learning_rate": 5.829626420653949e-07,
      "loss": 0.04590778425335884,
      "step": 6785
    },
    {
      "epoch": 0.9174222004410684,
      "grad_norm": 0.43994349241256714,
      "learning_rate": 5.810745609252166e-07,
      "loss": 0.04933256655931473,
      "step": 6786
    },
    {
      "epoch": 0.9175573938098336,
      "grad_norm": 0.6018949747085571,
      "learning_rate": 5.791894818822091e-07,
      "loss": 0.06430856883525848,
      "step": 6787
    },
    {
      "epoch": 0.9176925871785989,
      "grad_norm": 0.3604245185852051,
      "learning_rate": 5.773074053288519e-07,
      "loss": 0.04646119475364685,
      "step": 6788
    },
    {
      "epoch": 0.9178277805473641,
      "grad_norm": 0.7268880009651184,
      "learning_rate": 5.75428331657003e-07,
      "loss": 0.04490852355957031,
      "step": 6789
    },
    {
      "epoch": 0.9179629739161295,
      "grad_norm": 0.5224449038505554,
      "learning_rate": 5.735522612578998e-07,
      "loss": 0.040009189397096634,
      "step": 6790
    },
    {
      "epoch": 0.9180981672848947,
      "grad_norm": 0.696461021900177,
      "learning_rate": 5.716791945221444e-07,
      "loss": 0.042265623807907104,
      "step": 6791
    },
    {
      "epoch": 0.9182333606536599,
      "grad_norm": 0.45371294021606445,
      "learning_rate": 5.698091318397219e-07,
      "loss": 0.035211462527513504,
      "step": 6792
    },
    {
      "epoch": 0.9183685540224252,
      "grad_norm": 0.6046249866485596,
      "learning_rate": 5.679420735999908e-07,
      "loss": 0.04693195968866348,
      "step": 6793
    },
    {
      "epoch": 0.9185037473911905,
      "grad_norm": 0.25992992520332336,
      "learning_rate": 5.660780201916799e-07,
      "loss": 0.041408687829971313,
      "step": 6794
    },
    {
      "epoch": 0.9186389407599557,
      "grad_norm": 1.2105790376663208,
      "learning_rate": 5.642169720028973e-07,
      "loss": 0.062398843467235565,
      "step": 6795
    },
    {
      "epoch": 0.918774134128721,
      "grad_norm": 0.7734612226486206,
      "learning_rate": 5.623589294211196e-07,
      "loss": 0.05671662092208862,
      "step": 6796
    },
    {
      "epoch": 0.9189093274974862,
      "grad_norm": 0.7736897468566895,
      "learning_rate": 5.605038928332057e-07,
      "loss": 0.06395828723907471,
      "step": 6797
    },
    {
      "epoch": 0.9190445208662515,
      "grad_norm": 1.4590506553649902,
      "learning_rate": 5.586518626253817e-07,
      "loss": 0.06466203927993774,
      "step": 6798
    },
    {
      "epoch": 0.9191797142350168,
      "grad_norm": 0.279774010181427,
      "learning_rate": 5.568028391832524e-07,
      "loss": 0.03793752193450928,
      "step": 6799
    },
    {
      "epoch": 0.919314907603782,
      "grad_norm": 0.6834036111831665,
      "learning_rate": 5.549568228917917e-07,
      "loss": 0.05344633385539055,
      "step": 6800
    },
    {
      "epoch": 0.9194501009725473,
      "grad_norm": 0.437775582075119,
      "learning_rate": 5.531138141353486e-07,
      "loss": 0.04165481775999069,
      "step": 6801
    },
    {
      "epoch": 0.9195852943413125,
      "grad_norm": 0.5745537877082825,
      "learning_rate": 5.512738132976514e-07,
      "loss": 0.037696562707424164,
      "step": 6802
    },
    {
      "epoch": 0.9197204877100779,
      "grad_norm": 0.3792494535446167,
      "learning_rate": 5.49436820761795e-07,
      "loss": 0.04074833169579506,
      "step": 6803
    },
    {
      "epoch": 0.9198556810788431,
      "grad_norm": 0.5810791850090027,
      "learning_rate": 5.476028369102537e-07,
      "loss": 0.06212163716554642,
      "step": 6804
    },
    {
      "epoch": 0.9199908744476083,
      "grad_norm": 1.3440566062927246,
      "learning_rate": 5.45771862124872e-07,
      "loss": 0.06217864900827408,
      "step": 6805
    },
    {
      "epoch": 0.9201260678163736,
      "grad_norm": 0.669025182723999,
      "learning_rate": 5.439438967868649e-07,
      "loss": 0.05776963010430336,
      "step": 6806
    },
    {
      "epoch": 0.9202612611851388,
      "grad_norm": 0.3603413999080658,
      "learning_rate": 5.421189412768296e-07,
      "loss": 0.06129508838057518,
      "step": 6807
    },
    {
      "epoch": 0.9203964545539042,
      "grad_norm": 0.5741078853607178,
      "learning_rate": 5.402969959747306e-07,
      "loss": 0.046145543456077576,
      "step": 6808
    },
    {
      "epoch": 0.9205316479226694,
      "grad_norm": 0.3716714680194855,
      "learning_rate": 5.384780612599044e-07,
      "loss": 0.041075631976127625,
      "step": 6809
    },
    {
      "epoch": 0.9206668412914346,
      "grad_norm": 0.6136707067489624,
      "learning_rate": 5.366621375110647e-07,
      "loss": 0.06171710789203644,
      "step": 6810
    },
    {
      "epoch": 0.9208020346601999,
      "grad_norm": 0.4962771236896515,
      "learning_rate": 5.348492251062942e-07,
      "loss": 0.04525885730981827,
      "step": 6811
    },
    {
      "epoch": 0.9209372280289652,
      "grad_norm": 0.3042999804019928,
      "learning_rate": 5.330393244230558e-07,
      "loss": 0.04088454693555832,
      "step": 6812
    },
    {
      "epoch": 0.9210724213977305,
      "grad_norm": 0.23977521061897278,
      "learning_rate": 5.312324358381731e-07,
      "loss": 0.032902006059885025,
      "step": 6813
    },
    {
      "epoch": 0.9212076147664957,
      "grad_norm": 0.41756364703178406,
      "learning_rate": 5.29428559727857e-07,
      "loss": 0.05185117572546005,
      "step": 6814
    },
    {
      "epoch": 0.9213428081352609,
      "grad_norm": 0.6165286302566528,
      "learning_rate": 5.276276964676802e-07,
      "loss": 0.06036829203367233,
      "step": 6815
    },
    {
      "epoch": 0.9214780015040263,
      "grad_norm": 0.5815603733062744,
      "learning_rate": 5.258298464325928e-07,
      "loss": 0.05971301719546318,
      "step": 6816
    },
    {
      "epoch": 0.9216131948727915,
      "grad_norm": 0.6722524166107178,
      "learning_rate": 5.240350099969204e-07,
      "loss": 0.031716879457235336,
      "step": 6817
    },
    {
      "epoch": 0.9217483882415568,
      "grad_norm": 0.3034622073173523,
      "learning_rate": 5.222431875343492e-07,
      "loss": 0.03890479356050491,
      "step": 6818
    },
    {
      "epoch": 0.921883581610322,
      "grad_norm": 0.39960381388664246,
      "learning_rate": 5.204543794179539e-07,
      "loss": 0.042162250727415085,
      "step": 6819
    },
    {
      "epoch": 0.9220187749790872,
      "grad_norm": 0.3323359191417694,
      "learning_rate": 5.186685860201717e-07,
      "loss": 0.04835645854473114,
      "step": 6820
    },
    {
      "epoch": 0.9221539683478526,
      "grad_norm": 0.5833526849746704,
      "learning_rate": 5.16885807712812e-07,
      "loss": 0.05948370322585106,
      "step": 6821
    },
    {
      "epoch": 0.9222891617166178,
      "grad_norm": 0.2958061397075653,
      "learning_rate": 5.151060448670625e-07,
      "loss": 0.03802281990647316,
      "step": 6822
    },
    {
      "epoch": 0.922424355085383,
      "grad_norm": 0.5264478325843811,
      "learning_rate": 5.133292978534754e-07,
      "loss": 0.06669776141643524,
      "step": 6823
    },
    {
      "epoch": 0.9225595484541483,
      "grad_norm": 0.37217971682548523,
      "learning_rate": 5.115555670419814e-07,
      "loss": 0.04286617413163185,
      "step": 6824
    },
    {
      "epoch": 0.9226947418229136,
      "grad_norm": 0.4074537456035614,
      "learning_rate": 5.097848528018817e-07,
      "loss": 0.03774375468492508,
      "step": 6825
    },
    {
      "epoch": 0.9228299351916789,
      "grad_norm": 0.890584409236908,
      "learning_rate": 5.080171555018448e-07,
      "loss": 0.045653946697711945,
      "step": 6826
    },
    {
      "epoch": 0.9229651285604441,
      "grad_norm": 0.4125596880912781,
      "learning_rate": 5.06252475509918e-07,
      "loss": 0.07558906078338623,
      "step": 6827
    },
    {
      "epoch": 0.9231003219292093,
      "grad_norm": 0.22560857236385345,
      "learning_rate": 5.044908131935139e-07,
      "loss": 0.0393093004822731,
      "step": 6828
    },
    {
      "epoch": 0.9232355152979747,
      "grad_norm": 0.37901556491851807,
      "learning_rate": 5.027321689194242e-07,
      "loss": 0.04411853104829788,
      "step": 6829
    },
    {
      "epoch": 0.9233707086667399,
      "grad_norm": 0.8121209144592285,
      "learning_rate": 5.009765430538061e-07,
      "loss": 0.06160545349121094,
      "step": 6830
    },
    {
      "epoch": 0.9235059020355052,
      "grad_norm": 0.38927438855171204,
      "learning_rate": 4.992239359621886e-07,
      "loss": 0.04794158786535263,
      "step": 6831
    },
    {
      "epoch": 0.9236410954042704,
      "grad_norm": 0.718383252620697,
      "learning_rate": 4.974743480094767e-07,
      "loss": 0.05739040672779083,
      "step": 6832
    },
    {
      "epoch": 0.9237762887730356,
      "grad_norm": 0.5825828313827515,
      "learning_rate": 4.957277795599407e-07,
      "loss": 0.054212428629398346,
      "step": 6833
    },
    {
      "epoch": 0.923911482141801,
      "grad_norm": 0.9849809408187866,
      "learning_rate": 4.93984230977228e-07,
      "loss": 0.052510667592287064,
      "step": 6834
    },
    {
      "epoch": 0.9240466755105662,
      "grad_norm": 0.5457069277763367,
      "learning_rate": 4.922437026243531e-07,
      "loss": 0.05190357193350792,
      "step": 6835
    },
    {
      "epoch": 0.9241818688793315,
      "grad_norm": 0.33216261863708496,
      "learning_rate": 4.905061948637063e-07,
      "loss": 0.048580102622509,
      "step": 6836
    },
    {
      "epoch": 0.9243170622480967,
      "grad_norm": 0.28173625469207764,
      "learning_rate": 4.887717080570431e-07,
      "loss": 0.04640397056937218,
      "step": 6837
    },
    {
      "epoch": 0.924452255616862,
      "grad_norm": 0.5957367420196533,
      "learning_rate": 4.870402425654913e-07,
      "loss": 0.044471945613622665,
      "step": 6838
    },
    {
      "epoch": 0.9245874489856273,
      "grad_norm": 0.29059484601020813,
      "learning_rate": 4.853117987495542e-07,
      "loss": 0.04338253289461136,
      "step": 6839
    },
    {
      "epoch": 0.9247226423543925,
      "grad_norm": 0.333125501871109,
      "learning_rate": 4.83586376969104e-07,
      "loss": 0.04301752895116806,
      "step": 6840
    },
    {
      "epoch": 0.9248578357231578,
      "grad_norm": 0.4836156368255615,
      "learning_rate": 4.818639775833816e-07,
      "loss": 0.06459888815879822,
      "step": 6841
    },
    {
      "epoch": 0.924993029091923,
      "grad_norm": 0.762301504611969,
      "learning_rate": 4.801446009509969e-07,
      "loss": 0.0763884037733078,
      "step": 6842
    },
    {
      "epoch": 0.9251282224606883,
      "grad_norm": 1.4248297214508057,
      "learning_rate": 4.784282474299367e-07,
      "loss": 0.07100367546081543,
      "step": 6843
    },
    {
      "epoch": 0.9252634158294536,
      "grad_norm": 0.7939831018447876,
      "learning_rate": 4.767149173775537e-07,
      "loss": 0.04749560356140137,
      "step": 6844
    },
    {
      "epoch": 0.9253986091982188,
      "grad_norm": 0.36788618564605713,
      "learning_rate": 4.750046111505724e-07,
      "loss": 0.04334796592593193,
      "step": 6845
    },
    {
      "epoch": 0.9255338025669841,
      "grad_norm": 0.8742966055870056,
      "learning_rate": 4.732973291050896e-07,
      "loss": 0.05864344537258148,
      "step": 6846
    },
    {
      "epoch": 0.9256689959357494,
      "grad_norm": 0.6088741421699524,
      "learning_rate": 4.7159307159656607e-07,
      "loss": 0.06100534647703171,
      "step": 6847
    },
    {
      "epoch": 0.9258041893045146,
      "grad_norm": 0.34826967120170593,
      "learning_rate": 4.6989183897983954e-07,
      "loss": 0.045671992003917694,
      "step": 6848
    },
    {
      "epoch": 0.9259393826732799,
      "grad_norm": 1.2595014572143555,
      "learning_rate": 4.681936316091201e-07,
      "loss": 0.07075631618499756,
      "step": 6849
    },
    {
      "epoch": 0.9260745760420451,
      "grad_norm": 0.4086087942123413,
      "learning_rate": 4.664984498379765e-07,
      "loss": 0.05260862782597542,
      "step": 6850
    },
    {
      "epoch": 0.9262097694108103,
      "grad_norm": 0.4885626435279846,
      "learning_rate": 4.6480629401935814e-07,
      "loss": 0.036779481917619705,
      "step": 6851
    },
    {
      "epoch": 0.9263449627795757,
      "grad_norm": 0.5012714862823486,
      "learning_rate": 4.631171645055815e-07,
      "loss": 0.06439995765686035,
      "step": 6852
    },
    {
      "epoch": 0.9264801561483409,
      "grad_norm": 0.4285504221916199,
      "learning_rate": 4.614310616483286e-07,
      "loss": 0.04551964998245239,
      "step": 6853
    },
    {
      "epoch": 0.9266153495171062,
      "grad_norm": 1.070753812789917,
      "learning_rate": 4.5974798579866193e-07,
      "loss": 0.05979108065366745,
      "step": 6854
    },
    {
      "epoch": 0.9267505428858714,
      "grad_norm": 0.7485949993133545,
      "learning_rate": 4.580679373069996e-07,
      "loss": 0.055185697972774506,
      "step": 6855
    },
    {
      "epoch": 0.9268857362546367,
      "grad_norm": 0.34306201338768005,
      "learning_rate": 4.5639091652314e-07,
      "loss": 0.051580943167209625,
      "step": 6856
    },
    {
      "epoch": 0.927020929623402,
      "grad_norm": 0.29891863465309143,
      "learning_rate": 4.54716923796249e-07,
      "loss": 0.03212370723485947,
      "step": 6857
    },
    {
      "epoch": 0.9271561229921672,
      "grad_norm": 0.823025107383728,
      "learning_rate": 4.5304595947485927e-07,
      "loss": 0.05347749590873718,
      "step": 6858
    },
    {
      "epoch": 0.9272913163609325,
      "grad_norm": 0.6613633632659912,
      "learning_rate": 4.5137802390687433e-07,
      "loss": 0.05161743983626366,
      "step": 6859
    },
    {
      "epoch": 0.9274265097296978,
      "grad_norm": 0.6856628656387329,
      "learning_rate": 4.497131174395663e-07,
      "loss": 0.06462137401103973,
      "step": 6860
    },
    {
      "epoch": 0.927561703098463,
      "grad_norm": 0.35582128167152405,
      "learning_rate": 4.4805124041957967e-07,
      "loss": 0.051616791635751724,
      "step": 6861
    },
    {
      "epoch": 0.9276968964672283,
      "grad_norm": 1.0053256750106812,
      "learning_rate": 4.463923931929259e-07,
      "loss": 0.04513093829154968,
      "step": 6862
    },
    {
      "epoch": 0.9278320898359935,
      "grad_norm": 0.2146720439195633,
      "learning_rate": 4.4473657610498377e-07,
      "loss": 0.03712264075875282,
      "step": 6863
    },
    {
      "epoch": 0.9279672832047589,
      "grad_norm": 0.38348299264907837,
      "learning_rate": 4.430837895005058e-07,
      "loss": 0.03526635468006134,
      "step": 6864
    },
    {
      "epoch": 0.9281024765735241,
      "grad_norm": 0.5687447190284729,
      "learning_rate": 4.4143403372360836e-07,
      "loss": 0.06411556899547577,
      "step": 6865
    },
    {
      "epoch": 0.9282376699422893,
      "grad_norm": 0.8411898612976074,
      "learning_rate": 4.3978730911778176e-07,
      "loss": 0.041140396147966385,
      "step": 6866
    },
    {
      "epoch": 0.9283728633110546,
      "grad_norm": 0.3413739502429962,
      "learning_rate": 4.381436160258834e-07,
      "loss": 0.053307853639125824,
      "step": 6867
    },
    {
      "epoch": 0.9285080566798198,
      "grad_norm": 0.593433678150177,
      "learning_rate": 4.3650295479013615e-07,
      "loss": 0.04913664609193802,
      "step": 6868
    },
    {
      "epoch": 0.9286432500485852,
      "grad_norm": 0.9128335118293762,
      "learning_rate": 4.348653257521351e-07,
      "loss": 0.086134172976017,
      "step": 6869
    },
    {
      "epoch": 0.9287784434173504,
      "grad_norm": 1.4322556257247925,
      "learning_rate": 4.332307292528442e-07,
      "loss": 0.0616513267159462,
      "step": 6870
    },
    {
      "epoch": 0.9289136367861156,
      "grad_norm": 0.6071394681930542,
      "learning_rate": 4.315991656325946e-07,
      "loss": 0.04312963783740997,
      "step": 6871
    },
    {
      "epoch": 0.9290488301548809,
      "grad_norm": 0.6859742999076843,
      "learning_rate": 4.299706352310895e-07,
      "loss": 0.060514338314533234,
      "step": 6872
    },
    {
      "epoch": 0.9291840235236462,
      "grad_norm": 0.36952733993530273,
      "learning_rate": 4.283451383873926e-07,
      "loss": 0.05173899978399277,
      "step": 6873
    },
    {
      "epoch": 0.9293192168924115,
      "grad_norm": 1.0553873777389526,
      "learning_rate": 4.26722675439945e-07,
      "loss": 0.06947000324726105,
      "step": 6874
    },
    {
      "epoch": 0.9294544102611767,
      "grad_norm": 0.5726354122161865,
      "learning_rate": 4.251032467265481e-07,
      "loss": 0.0537940114736557,
      "step": 6875
    },
    {
      "epoch": 0.9295896036299419,
      "grad_norm": 0.43401291966438293,
      "learning_rate": 4.234868525843805e-07,
      "loss": 0.05457330867648125,
      "step": 6876
    },
    {
      "epoch": 0.9297247969987072,
      "grad_norm": 0.5494489669799805,
      "learning_rate": 4.218734933499796e-07,
      "loss": 0.04692947864532471,
      "step": 6877
    },
    {
      "epoch": 0.9298599903674725,
      "grad_norm": 0.41966408491134644,
      "learning_rate": 4.202631693592601e-07,
      "loss": 0.036715637892484665,
      "step": 6878
    },
    {
      "epoch": 0.9299951837362377,
      "grad_norm": 0.32652556896209717,
      "learning_rate": 4.186558809474955e-07,
      "loss": 0.037756599485874176,
      "step": 6879
    },
    {
      "epoch": 0.930130377105003,
      "grad_norm": 1.1343594789505005,
      "learning_rate": 4.170516284493331e-07,
      "loss": 0.06886908411979675,
      "step": 6880
    },
    {
      "epoch": 0.9302655704737682,
      "grad_norm": 0.4337814152240753,
      "learning_rate": 4.1545041219879063e-07,
      "loss": 0.05352926254272461,
      "step": 6881
    },
    {
      "epoch": 0.9304007638425336,
      "grad_norm": 0.30627673864364624,
      "learning_rate": 4.138522325292432e-07,
      "loss": 0.04135363921523094,
      "step": 6882
    },
    {
      "epoch": 0.9305359572112988,
      "grad_norm": 0.41036850214004517,
      "learning_rate": 4.1225708977344457e-07,
      "loss": 0.04702620208263397,
      "step": 6883
    },
    {
      "epoch": 0.930671150580064,
      "grad_norm": 0.6200788617134094,
      "learning_rate": 4.106649842635124e-07,
      "loss": 0.0588408038020134,
      "step": 6884
    },
    {
      "epoch": 0.9308063439488293,
      "grad_norm": 0.3638732135295868,
      "learning_rate": 4.090759163309282e-07,
      "loss": 0.04563349857926369,
      "step": 6885
    },
    {
      "epoch": 0.9309415373175945,
      "grad_norm": 0.40585437417030334,
      "learning_rate": 4.07489886306549e-07,
      "loss": 0.06302759051322937,
      "step": 6886
    },
    {
      "epoch": 0.9310767306863599,
      "grad_norm": 1.2502042055130005,
      "learning_rate": 4.059068945205907e-07,
      "loss": 0.054131630808115005,
      "step": 6887
    },
    {
      "epoch": 0.9312119240551251,
      "grad_norm": 0.4585094451904297,
      "learning_rate": 4.043269413026429e-07,
      "loss": 0.04849252104759216,
      "step": 6888
    },
    {
      "epoch": 0.9313471174238903,
      "grad_norm": 0.3046894669532776,
      "learning_rate": 4.027500269816592e-07,
      "loss": 0.04442837834358215,
      "step": 6889
    },
    {
      "epoch": 0.9314823107926556,
      "grad_norm": 0.410313218832016,
      "learning_rate": 4.011761518859619e-07,
      "loss": 0.06582614034414291,
      "step": 6890
    },
    {
      "epoch": 0.9316175041614209,
      "grad_norm": 0.38509640097618103,
      "learning_rate": 3.996053163432406e-07,
      "loss": 0.052085891366004944,
      "step": 6891
    },
    {
      "epoch": 0.9317526975301862,
      "grad_norm": 0.6463513374328613,
      "learning_rate": 3.980375206805503e-07,
      "loss": 0.05951409786939621,
      "step": 6892
    },
    {
      "epoch": 0.9318878908989514,
      "grad_norm": 0.6272176504135132,
      "learning_rate": 3.9647276522431664e-07,
      "loss": 0.053722187876701355,
      "step": 6893
    },
    {
      "epoch": 0.9320230842677166,
      "grad_norm": 0.36715465784072876,
      "learning_rate": 3.949110503003289e-07,
      "loss": 0.04019009321928024,
      "step": 6894
    },
    {
      "epoch": 0.932158277636482,
      "grad_norm": 0.7486395239830017,
      "learning_rate": 3.9335237623374377e-07,
      "loss": 0.04625747725367546,
      "step": 6895
    },
    {
      "epoch": 0.9322934710052472,
      "grad_norm": 0.3732355535030365,
      "learning_rate": 3.917967433490849e-07,
      "loss": 0.04526018351316452,
      "step": 6896
    },
    {
      "epoch": 0.9324286643740125,
      "grad_norm": 0.4516226351261139,
      "learning_rate": 3.902441519702449e-07,
      "loss": 0.06396698951721191,
      "step": 6897
    },
    {
      "epoch": 0.9325638577427777,
      "grad_norm": 0.6238554120063782,
      "learning_rate": 3.886946024204818e-07,
      "loss": 0.049856893718242645,
      "step": 6898
    },
    {
      "epoch": 0.9326990511115429,
      "grad_norm": 0.262699156999588,
      "learning_rate": 3.871480950224193e-07,
      "loss": 0.035853028297424316,
      "step": 6899
    },
    {
      "epoch": 0.9328342444803083,
      "grad_norm": 0.3124576210975647,
      "learning_rate": 3.856046300980498e-07,
      "loss": 0.05443498492240906,
      "step": 6900
    },
    {
      "epoch": 0.9329694378490735,
      "grad_norm": 0.6546632647514343,
      "learning_rate": 3.8406420796872953e-07,
      "loss": 0.06108447164297104,
      "step": 6901
    },
    {
      "epoch": 0.9331046312178388,
      "grad_norm": 0.38207605481147766,
      "learning_rate": 3.825268289551803e-07,
      "loss": 0.036447420716285706,
      "step": 6902
    },
    {
      "epoch": 0.933239824586604,
      "grad_norm": 0.5642752647399902,
      "learning_rate": 3.8099249337749777e-07,
      "loss": 0.0575939416885376,
      "step": 6903
    },
    {
      "epoch": 0.9333750179553693,
      "grad_norm": 0.28798946738243103,
      "learning_rate": 3.7946120155513465e-07,
      "loss": 0.03988340497016907,
      "step": 6904
    },
    {
      "epoch": 0.9335102113241346,
      "grad_norm": 0.5857437252998352,
      "learning_rate": 3.7793295380691595e-07,
      "loss": 0.04093194007873535,
      "step": 6905
    },
    {
      "epoch": 0.9336454046928998,
      "grad_norm": 0.7235367894172668,
      "learning_rate": 3.7640775045103214e-07,
      "loss": 0.05671316385269165,
      "step": 6906
    },
    {
      "epoch": 0.933780598061665,
      "grad_norm": 0.7842723727226257,
      "learning_rate": 3.7488559180503423e-07,
      "loss": 0.07119694352149963,
      "step": 6907
    },
    {
      "epoch": 0.9339157914304304,
      "grad_norm": 0.3284439146518707,
      "learning_rate": 3.7336647818584866e-07,
      "loss": 0.052975937724113464,
      "step": 6908
    },
    {
      "epoch": 0.9340509847991956,
      "grad_norm": 0.7204075455665588,
      "learning_rate": 3.718504099097625e-07,
      "loss": 0.04168041795492172,
      "step": 6909
    },
    {
      "epoch": 0.9341861781679609,
      "grad_norm": 0.2687974274158478,
      "learning_rate": 3.703373872924265e-07,
      "loss": 0.03667091578245163,
      "step": 6910
    },
    {
      "epoch": 0.9343213715367261,
      "grad_norm": 0.4961684048175812,
      "learning_rate": 3.688274106488604e-07,
      "loss": 0.05535639822483063,
      "step": 6911
    },
    {
      "epoch": 0.9344565649054913,
      "grad_norm": 0.7571970820426941,
      "learning_rate": 3.67320480293451e-07,
      "loss": 0.04825403541326523,
      "step": 6912
    },
    {
      "epoch": 0.9345917582742567,
      "grad_norm": 0.4126468300819397,
      "learning_rate": 3.6581659653994736e-07,
      "loss": 0.04247164726257324,
      "step": 6913
    },
    {
      "epoch": 0.9347269516430219,
      "grad_norm": 1.5049036741256714,
      "learning_rate": 3.64315759701469e-07,
      "loss": 0.05699065327644348,
      "step": 6914
    },
    {
      "epoch": 0.9348621450117872,
      "grad_norm": 0.5831983089447021,
      "learning_rate": 3.6281797009049765e-07,
      "loss": 0.04462004452943802,
      "step": 6915
    },
    {
      "epoch": 0.9349973383805524,
      "grad_norm": 0.36182349920272827,
      "learning_rate": 3.613232280188772e-07,
      "loss": 0.03545515984296799,
      "step": 6916
    },
    {
      "epoch": 0.9351325317493177,
      "grad_norm": 0.5539961457252502,
      "learning_rate": 3.5983153379782363e-07,
      "loss": 0.06775589287281036,
      "step": 6917
    },
    {
      "epoch": 0.935267725118083,
      "grad_norm": 0.36537250876426697,
      "learning_rate": 3.5834288773791854e-07,
      "loss": 0.03607872873544693,
      "step": 6918
    },
    {
      "epoch": 0.9354029184868482,
      "grad_norm": 0.3558017611503601,
      "learning_rate": 3.568572901491007e-07,
      "loss": 0.045041970908641815,
      "step": 6919
    },
    {
      "epoch": 0.9355381118556135,
      "grad_norm": 0.9121891856193542,
      "learning_rate": 3.553747413406827e-07,
      "loss": 0.04132611304521561,
      "step": 6920
    },
    {
      "epoch": 0.9356733052243787,
      "grad_norm": 0.3153373599052429,
      "learning_rate": 3.538952416213376e-07,
      "loss": 0.03956308960914612,
      "step": 6921
    },
    {
      "epoch": 0.935808498593144,
      "grad_norm": 0.35840272903442383,
      "learning_rate": 3.524187912991056e-07,
      "loss": 0.04421333223581314,
      "step": 6922
    },
    {
      "epoch": 0.9359436919619093,
      "grad_norm": 0.3468698263168335,
      "learning_rate": 3.5094539068139254e-07,
      "loss": 0.049999285489320755,
      "step": 6923
    },
    {
      "epoch": 0.9360788853306745,
      "grad_norm": 0.6103076934814453,
      "learning_rate": 3.494750400749663e-07,
      "loss": 0.041032955050468445,
      "step": 6924
    },
    {
      "epoch": 0.9362140786994398,
      "grad_norm": 0.4706680178642273,
      "learning_rate": 3.480077397859638e-07,
      "loss": 0.056187525391578674,
      "step": 6925
    },
    {
      "epoch": 0.9363492720682051,
      "grad_norm": 0.3376704156398773,
      "learning_rate": 3.4654349011988384e-07,
      "loss": 0.045402638614177704,
      "step": 6926
    },
    {
      "epoch": 0.9364844654369703,
      "grad_norm": 0.40775826573371887,
      "learning_rate": 3.4508229138159095e-07,
      "loss": 0.04560445249080658,
      "step": 6927
    },
    {
      "epoch": 0.9366196588057356,
      "grad_norm": 0.23570431768894196,
      "learning_rate": 3.4362414387531516e-07,
      "loss": 0.04377678781747818,
      "step": 6928
    },
    {
      "epoch": 0.9367548521745008,
      "grad_norm": 0.5702465772628784,
      "learning_rate": 3.4216904790464854e-07,
      "loss": 0.03975013643503189,
      "step": 6929
    },
    {
      "epoch": 0.936890045543266,
      "grad_norm": 0.6928509473800659,
      "learning_rate": 3.407170037725521e-07,
      "loss": 0.062287162989377975,
      "step": 6930
    },
    {
      "epoch": 0.9370252389120314,
      "grad_norm": 0.22119970619678497,
      "learning_rate": 3.3926801178134737e-07,
      "loss": 0.03406835347414017,
      "step": 6931
    },
    {
      "epoch": 0.9371604322807966,
      "grad_norm": 0.3057532012462616,
      "learning_rate": 3.3782207223272467e-07,
      "loss": 0.03304586559534073,
      "step": 6932
    },
    {
      "epoch": 0.9372956256495619,
      "grad_norm": 0.7759184241294861,
      "learning_rate": 3.363791854277348e-07,
      "loss": 0.04740680754184723,
      "step": 6933
    },
    {
      "epoch": 0.9374308190183271,
      "grad_norm": 0.4912753403186798,
      "learning_rate": 3.349393516667926e-07,
      "loss": 0.06619034707546234,
      "step": 6934
    },
    {
      "epoch": 0.9375660123870924,
      "grad_norm": 0.7480354309082031,
      "learning_rate": 3.335025712496814e-07,
      "loss": 0.04873591661453247,
      "step": 6935
    },
    {
      "epoch": 0.9377012057558577,
      "grad_norm": 1.266300082206726,
      "learning_rate": 3.320688444755471e-07,
      "loss": 0.05261167511343956,
      "step": 6936
    },
    {
      "epoch": 0.9378363991246229,
      "grad_norm": 0.5463661551475525,
      "learning_rate": 3.306381716428991e-07,
      "loss": 0.06616979837417603,
      "step": 6937
    },
    {
      "epoch": 0.9379715924933882,
      "grad_norm": 0.4364592432975769,
      "learning_rate": 3.2921055304960925e-07,
      "loss": 0.05242602527141571,
      "step": 6938
    },
    {
      "epoch": 0.9381067858621535,
      "grad_norm": 0.4664572477340698,
      "learning_rate": 3.277859889929147e-07,
      "loss": 0.05553281307220459,
      "step": 6939
    },
    {
      "epoch": 0.9382419792309187,
      "grad_norm": 0.3258020579814911,
      "learning_rate": 3.263644797694215e-07,
      "loss": 0.05469992756843567,
      "step": 6940
    },
    {
      "epoch": 0.938377172599684,
      "grad_norm": 0.3591028153896332,
      "learning_rate": 3.2494602567509303e-07,
      "loss": 0.04743524640798569,
      "step": 6941
    },
    {
      "epoch": 0.9385123659684492,
      "grad_norm": 0.6162520051002502,
      "learning_rate": 3.2353062700525794e-07,
      "loss": 0.05535570904612541,
      "step": 6942
    },
    {
      "epoch": 0.9386475593372146,
      "grad_norm": 0.9031170010566711,
      "learning_rate": 3.221182840546122e-07,
      "loss": 0.046951621770858765,
      "step": 6943
    },
    {
      "epoch": 0.9387827527059798,
      "grad_norm": 0.8480353355407715,
      "learning_rate": 3.207089971172089e-07,
      "loss": 0.06944916397333145,
      "step": 6944
    },
    {
      "epoch": 0.938917946074745,
      "grad_norm": 0.5644265413284302,
      "learning_rate": 3.1930276648647504e-07,
      "loss": 0.05255895107984543,
      "step": 6945
    },
    {
      "epoch": 0.9390531394435103,
      "grad_norm": 0.5231180191040039,
      "learning_rate": 3.178995924551914e-07,
      "loss": 0.05253271386027336,
      "step": 6946
    },
    {
      "epoch": 0.9391883328122755,
      "grad_norm": 0.3365705609321594,
      "learning_rate": 3.164994753155059e-07,
      "loss": 0.03835735097527504,
      "step": 6947
    },
    {
      "epoch": 0.9393235261810409,
      "grad_norm": 0.6025895476341248,
      "learning_rate": 3.1510241535893215e-07,
      "loss": 0.052831098437309265,
      "step": 6948
    },
    {
      "epoch": 0.9394587195498061,
      "grad_norm": 0.46395939588546753,
      "learning_rate": 3.1370841287634567e-07,
      "loss": 0.04790782928466797,
      "step": 6949
    },
    {
      "epoch": 0.9395939129185713,
      "grad_norm": 0.26668328046798706,
      "learning_rate": 3.1231746815798436e-07,
      "loss": 0.036391712725162506,
      "step": 6950
    },
    {
      "epoch": 0.9397291062873366,
      "grad_norm": 0.882217288017273,
      "learning_rate": 3.1092958149344985e-07,
      "loss": 0.06984452903270721,
      "step": 6951
    },
    {
      "epoch": 0.9398642996561019,
      "grad_norm": 0.6008186340332031,
      "learning_rate": 3.095447531717077e-07,
      "loss": 0.05031910538673401,
      "step": 6952
    },
    {
      "epoch": 0.9399994930248672,
      "grad_norm": 0.4434886574745178,
      "learning_rate": 3.08162983481089e-07,
      "loss": 0.04873112589120865,
      "step": 6953
    },
    {
      "epoch": 0.9401346863936324,
      "grad_norm": 0.7196475267410278,
      "learning_rate": 3.067842727092801e-07,
      "loss": 0.04900206997990608,
      "step": 6954
    },
    {
      "epoch": 0.9402698797623976,
      "grad_norm": 0.26183605194091797,
      "learning_rate": 3.0540862114334323e-07,
      "loss": 0.038414888083934784,
      "step": 6955
    },
    {
      "epoch": 0.940405073131163,
      "grad_norm": 0.35962367057800293,
      "learning_rate": 3.0403602906969086e-07,
      "loss": 0.03842412680387497,
      "step": 6956
    },
    {
      "epoch": 0.9405402664999282,
      "grad_norm": 0.6452633738517761,
      "learning_rate": 3.0266649677410605e-07,
      "loss": 0.0571509413421154,
      "step": 6957
    },
    {
      "epoch": 0.9406754598686934,
      "grad_norm": 0.6688090562820435,
      "learning_rate": 3.0130002454173243e-07,
      "loss": 0.04868298023939133,
      "step": 6958
    },
    {
      "epoch": 0.9408106532374587,
      "grad_norm": 0.6871205568313599,
      "learning_rate": 2.9993661265707407e-07,
      "loss": 0.05186744034290314,
      "step": 6959
    },
    {
      "epoch": 0.9409458466062239,
      "grad_norm": 0.6805575489997864,
      "learning_rate": 2.985762614040072e-07,
      "loss": 0.048685796558856964,
      "step": 6960
    },
    {
      "epoch": 0.9410810399749893,
      "grad_norm": 0.38146278262138367,
      "learning_rate": 2.972189710657586e-07,
      "loss": 0.03637781739234924,
      "step": 6961
    },
    {
      "epoch": 0.9412162333437545,
      "grad_norm": 0.6755660772323608,
      "learning_rate": 2.958647419249255e-07,
      "loss": 0.06114444509148598,
      "step": 6962
    },
    {
      "epoch": 0.9413514267125197,
      "grad_norm": 0.29800012707710266,
      "learning_rate": 2.9451357426346415e-07,
      "loss": 0.04383504390716553,
      "step": 6963
    },
    {
      "epoch": 0.941486620081285,
      "grad_norm": 0.46357131004333496,
      "learning_rate": 2.9316546836269776e-07,
      "loss": 0.045333947986364365,
      "step": 6964
    },
    {
      "epoch": 0.9416218134500502,
      "grad_norm": 1.0406869649887085,
      "learning_rate": 2.9182042450330516e-07,
      "loss": 0.04631621390581131,
      "step": 6965
    },
    {
      "epoch": 0.9417570068188156,
      "grad_norm": 0.6998670697212219,
      "learning_rate": 2.9047844296533397e-07,
      "loss": 0.06134534627199173,
      "step": 6966
    },
    {
      "epoch": 0.9418922001875808,
      "grad_norm": 0.43813422322273254,
      "learning_rate": 2.8913952402819246e-07,
      "loss": 0.037293940782547,
      "step": 6967
    },
    {
      "epoch": 0.942027393556346,
      "grad_norm": 0.44376295804977417,
      "learning_rate": 2.878036679706492e-07,
      "loss": 0.06385214626789093,
      "step": 6968
    },
    {
      "epoch": 0.9421625869251113,
      "grad_norm": 0.4596906900405884,
      "learning_rate": 2.8647087507083837e-07,
      "loss": 0.03660021722316742,
      "step": 6969
    },
    {
      "epoch": 0.9422977802938766,
      "grad_norm": 0.29545512795448303,
      "learning_rate": 2.8514114560625303e-07,
      "loss": 0.03944259136915207,
      "step": 6970
    },
    {
      "epoch": 0.9424329736626419,
      "grad_norm": 0.7356157898902893,
      "learning_rate": 2.8381447985375007e-07,
      "loss": 0.04027798771858215,
      "step": 6971
    },
    {
      "epoch": 0.9425681670314071,
      "grad_norm": 0.6049381494522095,
      "learning_rate": 2.8249087808954853e-07,
      "loss": 0.04712092876434326,
      "step": 6972
    },
    {
      "epoch": 0.9427033604001723,
      "grad_norm": 0.27560052275657654,
      "learning_rate": 2.811703405892296e-07,
      "loss": 0.036072008311748505,
      "step": 6973
    },
    {
      "epoch": 0.9428385537689377,
      "grad_norm": 0.5239424705505371,
      "learning_rate": 2.798528676277368e-07,
      "loss": 0.05494248867034912,
      "step": 6974
    },
    {
      "epoch": 0.9429737471377029,
      "grad_norm": 0.20207205414772034,
      "learning_rate": 2.785384594793738e-07,
      "loss": 0.03431403264403343,
      "step": 6975
    },
    {
      "epoch": 0.9431089405064682,
      "grad_norm": 0.8590747117996216,
      "learning_rate": 2.772271164178086e-07,
      "loss": 0.04887323081493378,
      "step": 6976
    },
    {
      "epoch": 0.9432441338752334,
      "grad_norm": 0.45627400279045105,
      "learning_rate": 2.759188387160677e-07,
      "loss": 0.04969898611307144,
      "step": 6977
    },
    {
      "epoch": 0.9433793272439986,
      "grad_norm": 0.35985031723976135,
      "learning_rate": 2.746136266465449e-07,
      "loss": 0.04580122232437134,
      "step": 6978
    },
    {
      "epoch": 0.943514520612764,
      "grad_norm": 0.5015252232551575,
      "learning_rate": 2.7331148048098943e-07,
      "loss": 0.05370482802391052,
      "step": 6979
    },
    {
      "epoch": 0.9436497139815292,
      "grad_norm": 0.5873124599456787,
      "learning_rate": 2.7201240049051613e-07,
      "loss": 0.05057905986905098,
      "step": 6980
    },
    {
      "epoch": 0.9437849073502945,
      "grad_norm": 0.5648605227470398,
      "learning_rate": 2.707163869455986e-07,
      "loss": 0.04481428116559982,
      "step": 6981
    },
    {
      "epoch": 0.9439201007190597,
      "grad_norm": 0.5584554672241211,
      "learning_rate": 2.694234401160778e-07,
      "loss": 0.04871741682291031,
      "step": 6982
    },
    {
      "epoch": 0.944055294087825,
      "grad_norm": 0.4979788661003113,
      "learning_rate": 2.6813356027114986e-07,
      "loss": 0.058538466691970825,
      "step": 6983
    },
    {
      "epoch": 0.9441904874565903,
      "grad_norm": 0.6593074202537537,
      "learning_rate": 2.6684674767937346e-07,
      "loss": 0.06859508156776428,
      "step": 6984
    },
    {
      "epoch": 0.9443256808253555,
      "grad_norm": 0.5325539708137512,
      "learning_rate": 2.655630026086708e-07,
      "loss": 0.053475797176361084,
      "step": 6985
    },
    {
      "epoch": 0.9444608741941207,
      "grad_norm": 1.0206910371780396,
      "learning_rate": 2.642823253263249e-07,
      "loss": 0.06128133833408356,
      "step": 6986
    },
    {
      "epoch": 0.944596067562886,
      "grad_norm": 0.5368931293487549,
      "learning_rate": 2.630047160989807e-07,
      "loss": 0.05463109910488129,
      "step": 6987
    },
    {
      "epoch": 0.9447312609316513,
      "grad_norm": 0.23954050242900848,
      "learning_rate": 2.6173017519263875e-07,
      "loss": 0.04308106005191803,
      "step": 6988
    },
    {
      "epoch": 0.9448664543004166,
      "grad_norm": 0.8664786219596863,
      "learning_rate": 2.6045870287267014e-07,
      "loss": 0.06628285348415375,
      "step": 6989
    },
    {
      "epoch": 0.9450016476691818,
      "grad_norm": 0.5322293043136597,
      "learning_rate": 2.5919029940380147e-07,
      "loss": 0.07109948992729187,
      "step": 6990
    },
    {
      "epoch": 0.945136841037947,
      "grad_norm": 1.1322778463363647,
      "learning_rate": 2.5792496505011807e-07,
      "loss": 0.035068489611148834,
      "step": 6991
    },
    {
      "epoch": 0.9452720344067124,
      "grad_norm": 0.6810746192932129,
      "learning_rate": 2.5666270007507266e-07,
      "loss": 0.07524842023849487,
      "step": 6992
    },
    {
      "epoch": 0.9454072277754776,
      "grad_norm": 0.7723119854927063,
      "learning_rate": 2.5540350474147324e-07,
      "loss": 0.056723274290561676,
      "step": 6993
    },
    {
      "epoch": 0.9455424211442429,
      "grad_norm": 0.5633419752120972,
      "learning_rate": 2.5414737931149346e-07,
      "loss": 0.06861861050128937,
      "step": 6994
    },
    {
      "epoch": 0.9456776145130081,
      "grad_norm": 0.7289638519287109,
      "learning_rate": 2.5289432404666246e-07,
      "loss": 0.06425695866346359,
      "step": 6995
    },
    {
      "epoch": 0.9458128078817734,
      "grad_norm": 0.7342117428779602,
      "learning_rate": 2.5164433920787487e-07,
      "loss": 0.045791588723659515,
      "step": 6996
    },
    {
      "epoch": 0.9459480012505387,
      "grad_norm": 0.6309106349945068,
      "learning_rate": 2.503974250553842e-07,
      "loss": 0.044482797384262085,
      "step": 6997
    },
    {
      "epoch": 0.9460831946193039,
      "grad_norm": 0.6806005835533142,
      "learning_rate": 2.491535818488011e-07,
      "loss": 0.04941936582326889,
      "step": 6998
    },
    {
      "epoch": 0.9462183879880692,
      "grad_norm": 0.4786623418331146,
      "learning_rate": 2.479128098471067e-07,
      "loss": 0.04707799479365349,
      "step": 6999
    },
    {
      "epoch": 0.9463535813568345,
      "grad_norm": 0.5309879183769226,
      "learning_rate": 2.466751093086328e-07,
      "loss": 0.04728540778160095,
      "step": 7000
    },
    {
      "epoch": 0.9464887747255997,
      "grad_norm": 0.511669397354126,
      "learning_rate": 2.454404804910748e-07,
      "loss": 0.03843013942241669,
      "step": 7001
    },
    {
      "epoch": 0.946623968094365,
      "grad_norm": 0.22840344905853271,
      "learning_rate": 2.442089236514888e-07,
      "loss": 0.048430219292640686,
      "step": 7002
    },
    {
      "epoch": 0.9467591614631302,
      "grad_norm": 0.37283170223236084,
      "learning_rate": 2.429804390462931e-07,
      "loss": 0.052125778049230576,
      "step": 7003
    },
    {
      "epoch": 0.9468943548318955,
      "grad_norm": 0.6303684115409851,
      "learning_rate": 2.4175502693126293e-07,
      "loss": 0.03049170784652233,
      "step": 7004
    },
    {
      "epoch": 0.9470295482006608,
      "grad_norm": 1.169367790222168,
      "learning_rate": 2.4053268756153933e-07,
      "loss": 0.05231369659304619,
      "step": 7005
    },
    {
      "epoch": 0.947164741569426,
      "grad_norm": 0.2846252918243408,
      "learning_rate": 2.393134211916154e-07,
      "loss": 0.04040795564651489,
      "step": 7006
    },
    {
      "epoch": 0.9472999349381913,
      "grad_norm": 0.5095453858375549,
      "learning_rate": 2.3809722807535128e-07,
      "loss": 0.05332716554403305,
      "step": 7007
    },
    {
      "epoch": 0.9474351283069565,
      "grad_norm": 0.5822315216064453,
      "learning_rate": 2.3688410846596287e-07,
      "loss": 0.04308965429663658,
      "step": 7008
    },
    {
      "epoch": 0.9475703216757219,
      "grad_norm": 0.46818676590919495,
      "learning_rate": 2.3567406261603143e-07,
      "loss": 0.06459400057792664,
      "step": 7009
    },
    {
      "epoch": 0.9477055150444871,
      "grad_norm": 0.9598819613456726,
      "learning_rate": 2.3446709077749206e-07,
      "loss": 0.07035110890865326,
      "step": 7010
    },
    {
      "epoch": 0.9478407084132523,
      "grad_norm": 0.8856511116027832,
      "learning_rate": 2.3326319320164546e-07,
      "loss": 0.050109125673770905,
      "step": 7011
    },
    {
      "epoch": 0.9479759017820176,
      "grad_norm": 0.8328272700309753,
      "learning_rate": 2.320623701391461e-07,
      "loss": 0.060020022094249725,
      "step": 7012
    },
    {
      "epoch": 0.9481110951507828,
      "grad_norm": 0.5795353651046753,
      "learning_rate": 2.30864621840014e-07,
      "loss": 0.04656301438808441,
      "step": 7013
    },
    {
      "epoch": 0.9482462885195481,
      "grad_norm": 0.3655354380607605,
      "learning_rate": 2.2966994855362633e-07,
      "loss": 0.030804309993982315,
      "step": 7014
    },
    {
      "epoch": 0.9483814818883134,
      "grad_norm": 0.26527708768844604,
      "learning_rate": 2.2847835052872079e-07,
      "loss": 0.03650180250406265,
      "step": 7015
    },
    {
      "epoch": 0.9485166752570786,
      "grad_norm": 0.4255734086036682,
      "learning_rate": 2.2728982801339392e-07,
      "loss": 0.04098595678806305,
      "step": 7016
    },
    {
      "epoch": 0.9486518686258439,
      "grad_norm": 1.1527483463287354,
      "learning_rate": 2.261043812551028e-07,
      "loss": 0.0618458166718483,
      "step": 7017
    },
    {
      "epoch": 0.9487870619946092,
      "grad_norm": 0.22363893687725067,
      "learning_rate": 2.249220105006633e-07,
      "loss": 0.03161557763814926,
      "step": 7018
    },
    {
      "epoch": 0.9489222553633744,
      "grad_norm": 0.7646569609642029,
      "learning_rate": 2.2374271599625185e-07,
      "loss": 0.06502489745616913,
      "step": 7019
    },
    {
      "epoch": 0.9490574487321397,
      "grad_norm": 0.6240152716636658,
      "learning_rate": 2.2256649798740204e-07,
      "loss": 0.048737481236457825,
      "step": 7020
    },
    {
      "epoch": 0.9491926421009049,
      "grad_norm": 0.3745671510696411,
      "learning_rate": 2.2139335671901294e-07,
      "loss": 0.06698073446750641,
      "step": 7021
    },
    {
      "epoch": 0.9493278354696703,
      "grad_norm": 0.3783004879951477,
      "learning_rate": 2.2022329243533422e-07,
      "loss": 0.05842578411102295,
      "step": 7022
    },
    {
      "epoch": 0.9494630288384355,
      "grad_norm": 0.7516889572143555,
      "learning_rate": 2.19056305379981e-07,
      "loss": 0.05865636467933655,
      "step": 7023
    },
    {
      "epoch": 0.9495982222072007,
      "grad_norm": 0.27407386898994446,
      "learning_rate": 2.178923957959289e-07,
      "loss": 0.042855437844991684,
      "step": 7024
    },
    {
      "epoch": 0.949733415575966,
      "grad_norm": 0.24665077030658722,
      "learning_rate": 2.1673156392550408e-07,
      "loss": 0.03831611946225166,
      "step": 7025
    },
    {
      "epoch": 0.9498686089447312,
      "grad_norm": 0.6068966388702393,
      "learning_rate": 2.155738100104049e-07,
      "loss": 0.046118929982185364,
      "step": 7026
    },
    {
      "epoch": 0.9500038023134966,
      "grad_norm": 0.4965992867946625,
      "learning_rate": 2.1441913429167682e-07,
      "loss": 0.057642895728349686,
      "step": 7027
    },
    {
      "epoch": 0.9501389956822618,
      "grad_norm": 0.24034655094146729,
      "learning_rate": 2.1326753700973256e-07,
      "loss": 0.03208189085125923,
      "step": 7028
    },
    {
      "epoch": 0.950274189051027,
      "grad_norm": 0.8269594311714172,
      "learning_rate": 2.1211901840434034e-07,
      "loss": 0.0472969189286232,
      "step": 7029
    },
    {
      "epoch": 0.9504093824197923,
      "grad_norm": 1.1255297660827637,
      "learning_rate": 2.1097357871462386e-07,
      "loss": 0.043352872133255005,
      "step": 7030
    },
    {
      "epoch": 0.9505445757885576,
      "grad_norm": 0.23174862563610077,
      "learning_rate": 2.098312181790757e-07,
      "loss": 0.04218031466007233,
      "step": 7031
    },
    {
      "epoch": 0.9506797691573229,
      "grad_norm": 0.5022928714752197,
      "learning_rate": 2.086919370355389e-07,
      "loss": 0.04457200691103935,
      "step": 7032
    },
    {
      "epoch": 0.9508149625260881,
      "grad_norm": 1.0669138431549072,
      "learning_rate": 2.075557355212171e-07,
      "loss": 0.06347168236970901,
      "step": 7033
    },
    {
      "epoch": 0.9509501558948533,
      "grad_norm": 0.7271129488945007,
      "learning_rate": 2.0642261387267268e-07,
      "loss": 0.060640715062618256,
      "step": 7034
    },
    {
      "epoch": 0.9510853492636187,
      "grad_norm": 0.8298430442810059,
      "learning_rate": 2.0529257232583033e-07,
      "loss": 0.03710779547691345,
      "step": 7035
    },
    {
      "epoch": 0.9512205426323839,
      "grad_norm": 0.6621572375297546,
      "learning_rate": 2.0416561111596844e-07,
      "loss": 0.04770224541425705,
      "step": 7036
    },
    {
      "epoch": 0.9513557360011491,
      "grad_norm": 0.9314915537834167,
      "learning_rate": 2.0304173047772933e-07,
      "loss": 0.04345601797103882,
      "step": 7037
    },
    {
      "epoch": 0.9514909293699144,
      "grad_norm": 0.4792274534702301,
      "learning_rate": 2.0192093064510753e-07,
      "loss": 0.03735507279634476,
      "step": 7038
    },
    {
      "epoch": 0.9516261227386796,
      "grad_norm": 0.6797763705253601,
      "learning_rate": 2.0080321185146134e-07,
      "loss": 0.054027535021305084,
      "step": 7039
    },
    {
      "epoch": 0.951761316107445,
      "grad_norm": 0.21267877519130707,
      "learning_rate": 1.996885743295046e-07,
      "loss": 0.03521095588803291,
      "step": 7040
    },
    {
      "epoch": 0.9518965094762102,
      "grad_norm": 0.707499086856842,
      "learning_rate": 1.985770183113117e-07,
      "loss": 0.06317204236984253,
      "step": 7041
    },
    {
      "epoch": 0.9520317028449754,
      "grad_norm": 0.3947928547859192,
      "learning_rate": 1.9746854402831583e-07,
      "loss": 0.053580671548843384,
      "step": 7042
    },
    {
      "epoch": 0.9521668962137407,
      "grad_norm": 0.631397008895874,
      "learning_rate": 1.963631517113057e-07,
      "loss": 0.049400266259908676,
      "step": 7043
    },
    {
      "epoch": 0.952302089582506,
      "grad_norm": 0.7041942477226257,
      "learning_rate": 1.952608415904289e-07,
      "loss": 0.04562880098819733,
      "step": 7044
    },
    {
      "epoch": 0.9524372829512713,
      "grad_norm": 0.778236985206604,
      "learning_rate": 1.9416161389519348e-07,
      "loss": 0.05960611253976822,
      "step": 7045
    },
    {
      "epoch": 0.9525724763200365,
      "grad_norm": 0.47910770773887634,
      "learning_rate": 1.9306546885446475e-07,
      "loss": 0.04913066327571869,
      "step": 7046
    },
    {
      "epoch": 0.9527076696888017,
      "grad_norm": 0.3509472906589508,
      "learning_rate": 1.919724066964651e-07,
      "loss": 0.04112546890974045,
      "step": 7047
    },
    {
      "epoch": 0.952842863057567,
      "grad_norm": 0.3091643750667572,
      "learning_rate": 1.908824276487775e-07,
      "loss": 0.03803032636642456,
      "step": 7048
    },
    {
      "epoch": 0.9529780564263323,
      "grad_norm": 0.5412918329238892,
      "learning_rate": 1.8979553193833876e-07,
      "loss": 0.053448278456926346,
      "step": 7049
    },
    {
      "epoch": 0.9531132497950976,
      "grad_norm": 0.3901522755622864,
      "learning_rate": 1.8871171979144786e-07,
      "loss": 0.04777824878692627,
      "step": 7050
    },
    {
      "epoch": 0.9532484431638628,
      "grad_norm": 0.7053197622299194,
      "learning_rate": 1.8763099143376262e-07,
      "loss": 0.0679606944322586,
      "step": 7051
    },
    {
      "epoch": 0.953383636532628,
      "grad_norm": 0.3503236770629883,
      "learning_rate": 1.8655334709029303e-07,
      "loss": 0.052497513592243195,
      "step": 7052
    },
    {
      "epoch": 0.9535188299013934,
      "grad_norm": 0.36755189299583435,
      "learning_rate": 1.8547878698541132e-07,
      "loss": 0.038585394620895386,
      "step": 7053
    },
    {
      "epoch": 0.9536540232701586,
      "grad_norm": 0.4827481210231781,
      "learning_rate": 1.8440731134284684e-07,
      "loss": 0.05043557658791542,
      "step": 7054
    },
    {
      "epoch": 0.9537892166389239,
      "grad_norm": 0.3423571288585663,
      "learning_rate": 1.833389203856861e-07,
      "loss": 0.06038116663694382,
      "step": 7055
    },
    {
      "epoch": 0.9539244100076891,
      "grad_norm": 0.19178050756454468,
      "learning_rate": 1.8227361433637625e-07,
      "loss": 0.018407421186566353,
      "step": 7056
    },
    {
      "epoch": 0.9540596033764543,
      "grad_norm": 0.5763635039329529,
      "learning_rate": 1.812113934167148e-07,
      "loss": 0.05605198070406914,
      "step": 7057
    },
    {
      "epoch": 0.9541947967452197,
      "grad_norm": 0.5263624787330627,
      "learning_rate": 1.8015225784786483e-07,
      "loss": 0.0454728789627552,
      "step": 7058
    },
    {
      "epoch": 0.9543299901139849,
      "grad_norm": 0.32723468542099,
      "learning_rate": 1.7909620785034663e-07,
      "loss": 0.04476713389158249,
      "step": 7059
    },
    {
      "epoch": 0.9544651834827502,
      "grad_norm": 0.3506423234939575,
      "learning_rate": 1.7804324364402936e-07,
      "loss": 0.05381958559155464,
      "step": 7060
    },
    {
      "epoch": 0.9546003768515154,
      "grad_norm": 0.5271151661872864,
      "learning_rate": 1.769933654481526e-07,
      "loss": 0.04244597256183624,
      "step": 7061
    },
    {
      "epoch": 0.9547355702202807,
      "grad_norm": 0.4287923574447632,
      "learning_rate": 1.7594657348129984e-07,
      "loss": 0.04769602417945862,
      "step": 7062
    },
    {
      "epoch": 0.954870763589046,
      "grad_norm": 0.3924559950828552,
      "learning_rate": 1.749028679614234e-07,
      "loss": 0.04283842816948891,
      "step": 7063
    },
    {
      "epoch": 0.9550059569578112,
      "grad_norm": 0.5396997928619385,
      "learning_rate": 1.7386224910582615e-07,
      "loss": 0.045995138585567474,
      "step": 7064
    },
    {
      "epoch": 0.9551411503265764,
      "grad_norm": 0.5138243436813354,
      "learning_rate": 1.728247171311731e-07,
      "loss": 0.05613095313310623,
      "step": 7065
    },
    {
      "epoch": 0.9552763436953418,
      "grad_norm": 0.36675021052360535,
      "learning_rate": 1.7179027225348142e-07,
      "loss": 0.049286842346191406,
      "step": 7066
    },
    {
      "epoch": 0.955411537064107,
      "grad_norm": 0.4823135733604431,
      "learning_rate": 1.7075891468812722e-07,
      "loss": 0.0472954660654068,
      "step": 7067
    },
    {
      "epoch": 0.9555467304328723,
      "grad_norm": 0.27447083592414856,
      "learning_rate": 1.69730644649847e-07,
      "loss": 0.04970790445804596,
      "step": 7068
    },
    {
      "epoch": 0.9556819238016375,
      "grad_norm": 0.2845061719417572,
      "learning_rate": 1.687054623527312e-07,
      "loss": 0.043370600789785385,
      "step": 7069
    },
    {
      "epoch": 0.9558171171704027,
      "grad_norm": 0.7848108410835266,
      "learning_rate": 1.676833680102291e-07,
      "loss": 0.04972783476114273,
      "step": 7070
    },
    {
      "epoch": 0.9559523105391681,
      "grad_norm": 0.3588605225086212,
      "learning_rate": 1.6666436183514378e-07,
      "loss": 0.04506923258304596,
      "step": 7071
    },
    {
      "epoch": 0.9560875039079333,
      "grad_norm": 0.5329291224479675,
      "learning_rate": 1.6564844403964053e-07,
      "loss": 0.05122965946793556,
      "step": 7072
    },
    {
      "epoch": 0.9562226972766986,
      "grad_norm": 0.38017287850379944,
      "learning_rate": 1.6463561483523682e-07,
      "loss": 0.04808839410543442,
      "step": 7073
    },
    {
      "epoch": 0.9563578906454638,
      "grad_norm": 0.31512629985809326,
      "learning_rate": 1.6362587443281063e-07,
      "loss": 0.036521174013614655,
      "step": 7074
    },
    {
      "epoch": 0.9564930840142291,
      "grad_norm": 1.4295697212219238,
      "learning_rate": 1.626192230425938e-07,
      "loss": 0.06929187476634979,
      "step": 7075
    },
    {
      "epoch": 0.9566282773829944,
      "grad_norm": 1.0130048990249634,
      "learning_rate": 1.6161566087417868e-07,
      "loss": 0.05602887272834778,
      "step": 7076
    },
    {
      "epoch": 0.9567634707517596,
      "grad_norm": 0.5397262573242188,
      "learning_rate": 1.6061518813650977e-07,
      "loss": 0.054977670311927795,
      "step": 7077
    },
    {
      "epoch": 0.9568986641205249,
      "grad_norm": 0.6468890905380249,
      "learning_rate": 1.5961780503789215e-07,
      "loss": 0.057840749621391296,
      "step": 7078
    },
    {
      "epoch": 0.9570338574892902,
      "grad_norm": 0.49301132559776306,
      "learning_rate": 1.5862351178598633e-07,
      "loss": 0.04878881573677063,
      "step": 7079
    },
    {
      "epoch": 0.9571690508580554,
      "grad_norm": 0.9630247950553894,
      "learning_rate": 1.5763230858781008e-07,
      "loss": 0.0603552907705307,
      "step": 7080
    },
    {
      "epoch": 0.9573042442268207,
      "grad_norm": 1.3121906518936157,
      "learning_rate": 1.5664419564973497e-07,
      "loss": 0.04599609971046448,
      "step": 7081
    },
    {
      "epoch": 0.9574394375955859,
      "grad_norm": 0.875666618347168,
      "learning_rate": 1.5565917317749146e-07,
      "loss": 0.054941367357969284,
      "step": 7082
    },
    {
      "epoch": 0.9575746309643512,
      "grad_norm": 0.5639085173606873,
      "learning_rate": 1.5467724137617046e-07,
      "loss": 0.03386823832988739,
      "step": 7083
    },
    {
      "epoch": 0.9577098243331165,
      "grad_norm": 0.37282270193099976,
      "learning_rate": 1.5369840045021178e-07,
      "loss": 0.0650518536567688,
      "step": 7084
    },
    {
      "epoch": 0.9578450177018817,
      "grad_norm": 0.6280562281608582,
      "learning_rate": 1.5272265060341572e-07,
      "loss": 0.06313121318817139,
      "step": 7085
    },
    {
      "epoch": 0.957980211070647,
      "grad_norm": 0.6540583968162537,
      "learning_rate": 1.517499920389398e-07,
      "loss": 0.0653175413608551,
      "step": 7086
    },
    {
      "epoch": 0.9581154044394122,
      "grad_norm": 0.8435095548629761,
      "learning_rate": 1.5078042495929534e-07,
      "loss": 0.06455972790718079,
      "step": 7087
    },
    {
      "epoch": 0.9582505978081776,
      "grad_norm": 0.4780527353286743,
      "learning_rate": 1.498139495663542e-07,
      "loss": 0.053380683064460754,
      "step": 7088
    },
    {
      "epoch": 0.9583857911769428,
      "grad_norm": 0.709287166595459,
      "learning_rate": 1.4885056606133707e-07,
      "loss": 0.04134759306907654,
      "step": 7089
    },
    {
      "epoch": 0.958520984545708,
      "grad_norm": 0.820174515247345,
      "learning_rate": 1.478902746448302e-07,
      "loss": 0.049540046602487564,
      "step": 7090
    },
    {
      "epoch": 0.9586561779144733,
      "grad_norm": 0.6051679253578186,
      "learning_rate": 1.469330755167686e-07,
      "loss": 0.042211562395095825,
      "step": 7091
    },
    {
      "epoch": 0.9587913712832385,
      "grad_norm": 0.39660871028900146,
      "learning_rate": 1.4597896887644458e-07,
      "loss": 0.0401935800909996,
      "step": 7092
    },
    {
      "epoch": 0.9589265646520038,
      "grad_norm": 0.529358983039856,
      "learning_rate": 1.4502795492251418e-07,
      "loss": 0.06666940450668335,
      "step": 7093
    },
    {
      "epoch": 0.9590617580207691,
      "grad_norm": 0.45158839225769043,
      "learning_rate": 1.4408003385297742e-07,
      "loss": 0.043891970068216324,
      "step": 7094
    },
    {
      "epoch": 0.9591969513895343,
      "grad_norm": 0.29564619064331055,
      "learning_rate": 1.4313520586519968e-07,
      "loss": 0.04253445565700531,
      "step": 7095
    },
    {
      "epoch": 0.9593321447582996,
      "grad_norm": 0.8895026445388794,
      "learning_rate": 1.4219347115589863e-07,
      "loss": 0.04468303546309471,
      "step": 7096
    },
    {
      "epoch": 0.9594673381270649,
      "grad_norm": 0.3830501437187195,
      "learning_rate": 1.4125482992114914e-07,
      "loss": 0.03943002223968506,
      "step": 7097
    },
    {
      "epoch": 0.9596025314958301,
      "grad_norm": 1.0268203020095825,
      "learning_rate": 1.403192823563798e-07,
      "loss": 0.04540792852640152,
      "step": 7098
    },
    {
      "epoch": 0.9597377248645954,
      "grad_norm": 0.26480627059936523,
      "learning_rate": 1.3938682865637654e-07,
      "loss": 0.05046455189585686,
      "step": 7099
    },
    {
      "epoch": 0.9598729182333606,
      "grad_norm": 0.42334091663360596,
      "learning_rate": 1.38457469015284e-07,
      "loss": 0.04434821382164955,
      "step": 7100
    },
    {
      "epoch": 0.960008111602126,
      "grad_norm": 0.47198688983917236,
      "learning_rate": 1.3753120362659576e-07,
      "loss": 0.04567451775074005,
      "step": 7101
    },
    {
      "epoch": 0.9601433049708912,
      "grad_norm": 0.4214702844619751,
      "learning_rate": 1.3660803268316925e-07,
      "loss": 0.052460767328739166,
      "step": 7102
    },
    {
      "epoch": 0.9602784983396564,
      "grad_norm": 0.3947446048259735,
      "learning_rate": 1.3568795637721065e-07,
      "loss": 0.04251347482204437,
      "step": 7103
    },
    {
      "epoch": 0.9604136917084217,
      "grad_norm": 0.5492876172065735,
      "learning_rate": 1.347709749002851e-07,
      "loss": 0.048519060015678406,
      "step": 7104
    },
    {
      "epoch": 0.9605488850771869,
      "grad_norm": 0.7770055532455444,
      "learning_rate": 1.338570884433149e-07,
      "loss": 0.040394216775894165,
      "step": 7105
    },
    {
      "epoch": 0.9606840784459523,
      "grad_norm": 0.38754794001579285,
      "learning_rate": 1.3294629719657448e-07,
      "loss": 0.04493887722492218,
      "step": 7106
    },
    {
      "epoch": 0.9608192718147175,
      "grad_norm": 0.64454185962677,
      "learning_rate": 1.3203860134969548e-07,
      "loss": 0.07401294261217117,
      "step": 7107
    },
    {
      "epoch": 0.9609544651834827,
      "grad_norm": 0.6045639514923096,
      "learning_rate": 1.3113400109166508e-07,
      "loss": 0.03798668086528778,
      "step": 7108
    },
    {
      "epoch": 0.961089658552248,
      "grad_norm": 0.6573352217674255,
      "learning_rate": 1.3023249661082592e-07,
      "loss": 0.052593767642974854,
      "step": 7109
    },
    {
      "epoch": 0.9612248519210133,
      "grad_norm": 0.7348781824111938,
      "learning_rate": 1.2933408809487623e-07,
      "loss": 0.054810114204883575,
      "step": 7110
    },
    {
      "epoch": 0.9613600452897786,
      "grad_norm": 0.7497594356536865,
      "learning_rate": 1.2843877573086972e-07,
      "loss": 0.07394813001155853,
      "step": 7111
    },
    {
      "epoch": 0.9614952386585438,
      "grad_norm": 0.7536496520042419,
      "learning_rate": 1.2754655970521556e-07,
      "loss": 0.0443432480096817,
      "step": 7112
    },
    {
      "epoch": 0.961630432027309,
      "grad_norm": 0.37718063592910767,
      "learning_rate": 1.2665744020367686e-07,
      "loss": 0.04746967554092407,
      "step": 7113
    },
    {
      "epoch": 0.9617656253960744,
      "grad_norm": 0.4490594267845154,
      "learning_rate": 1.2577141741137388e-07,
      "loss": 0.05376119539141655,
      "step": 7114
    },
    {
      "epoch": 0.9619008187648396,
      "grad_norm": 0.3001408278942108,
      "learning_rate": 1.248884915127807e-07,
      "loss": 0.03593166172504425,
      "step": 7115
    },
    {
      "epoch": 0.9620360121336049,
      "grad_norm": 0.586151123046875,
      "learning_rate": 1.2400866269172694e-07,
      "loss": 0.05728622153401375,
      "step": 7116
    },
    {
      "epoch": 0.9621712055023701,
      "grad_norm": 0.241166353225708,
      "learning_rate": 1.2313193113139777e-07,
      "loss": 0.037278253585100174,
      "step": 7117
    },
    {
      "epoch": 0.9623063988711353,
      "grad_norm": 0.3017977476119995,
      "learning_rate": 1.2225829701433545e-07,
      "loss": 0.0425596684217453,
      "step": 7118
    },
    {
      "epoch": 0.9624415922399007,
      "grad_norm": 0.27473878860473633,
      "learning_rate": 1.2138776052243116e-07,
      "loss": 0.046478137373924255,
      "step": 7119
    },
    {
      "epoch": 0.9625767856086659,
      "grad_norm": 0.616244375705719,
      "learning_rate": 1.2052032183693996e-07,
      "loss": 0.05840327590703964,
      "step": 7120
    },
    {
      "epoch": 0.9627119789774311,
      "grad_norm": 0.43871229887008667,
      "learning_rate": 1.196559811384623e-07,
      "loss": 0.05451767519116402,
      "step": 7121
    },
    {
      "epoch": 0.9628471723461964,
      "grad_norm": 0.3072042465209961,
      "learning_rate": 1.1879473860696266e-07,
      "loss": 0.04456089437007904,
      "step": 7122
    },
    {
      "epoch": 0.9629823657149617,
      "grad_norm": 0.33909162878990173,
      "learning_rate": 1.179365944217542e-07,
      "loss": 0.05005289614200592,
      "step": 7123
    },
    {
      "epoch": 0.963117559083727,
      "grad_norm": 0.5105652213096619,
      "learning_rate": 1.1708154876150735e-07,
      "loss": 0.06939264386892319,
      "step": 7124
    },
    {
      "epoch": 0.9632527524524922,
      "grad_norm": 0.45631834864616394,
      "learning_rate": 1.1622960180424801e-07,
      "loss": 0.062086015939712524,
      "step": 7125
    },
    {
      "epoch": 0.9633879458212574,
      "grad_norm": 0.2777315676212311,
      "learning_rate": 1.1538075372735435e-07,
      "loss": 0.04628264904022217,
      "step": 7126
    },
    {
      "epoch": 0.9635231391900227,
      "grad_norm": 0.4102330207824707,
      "learning_rate": 1.1453500470756328e-07,
      "loss": 0.047382205724716187,
      "step": 7127
    },
    {
      "epoch": 0.963658332558788,
      "grad_norm": 0.4439198672771454,
      "learning_rate": 1.1369235492096397e-07,
      "loss": 0.03975337743759155,
      "step": 7128
    },
    {
      "epoch": 0.9637935259275533,
      "grad_norm": 0.33770933747291565,
      "learning_rate": 1.1285280454299774e-07,
      "loss": 0.05452870577573776,
      "step": 7129
    },
    {
      "epoch": 0.9639287192963185,
      "grad_norm": 0.2832639515399933,
      "learning_rate": 1.1201635374846808e-07,
      "loss": 0.05067891627550125,
      "step": 7130
    },
    {
      "epoch": 0.9640639126650837,
      "grad_norm": 0.5231163501739502,
      "learning_rate": 1.1118300271152404e-07,
      "loss": 0.05393754318356514,
      "step": 7131
    },
    {
      "epoch": 0.9641991060338491,
      "grad_norm": 0.6559244394302368,
      "learning_rate": 1.1035275160567682e-07,
      "loss": 0.04904860258102417,
      "step": 7132
    },
    {
      "epoch": 0.9643342994026143,
      "grad_norm": 0.38227379322052,
      "learning_rate": 1.0952560060378813e-07,
      "loss": 0.04571916162967682,
      "step": 7133
    },
    {
      "epoch": 0.9644694927713796,
      "grad_norm": 0.7259240746498108,
      "learning_rate": 1.0870154987807523e-07,
      "loss": 0.04718397557735443,
      "step": 7134
    },
    {
      "epoch": 0.9646046861401448,
      "grad_norm": 1.2006630897521973,
      "learning_rate": 1.0788059960010921e-07,
      "loss": 0.05734813213348389,
      "step": 7135
    },
    {
      "epoch": 0.96473987950891,
      "grad_norm": 0.5730720162391663,
      "learning_rate": 1.0706274994081499e-07,
      "loss": 0.03975367546081543,
      "step": 7136
    },
    {
      "epoch": 0.9648750728776754,
      "grad_norm": 0.6688952445983887,
      "learning_rate": 1.0624800107047805e-07,
      "loss": 0.05123265087604523,
      "step": 7137
    },
    {
      "epoch": 0.9650102662464406,
      "grad_norm": 0.5579339265823364,
      "learning_rate": 1.0543635315872934e-07,
      "loss": 0.056104280054569244,
      "step": 7138
    },
    {
      "epoch": 0.9651454596152059,
      "grad_norm": 0.38516145944595337,
      "learning_rate": 1.0462780637455871e-07,
      "loss": 0.05657939240336418,
      "step": 7139
    },
    {
      "epoch": 0.9652806529839711,
      "grad_norm": 0.45473095774650574,
      "learning_rate": 1.0382236088631148e-07,
      "loss": 0.0546586811542511,
      "step": 7140
    },
    {
      "epoch": 0.9654158463527364,
      "grad_norm": 0.47887957096099854,
      "learning_rate": 1.0302001686168349e-07,
      "loss": 0.0643608570098877,
      "step": 7141
    },
    {
      "epoch": 0.9655510397215017,
      "grad_norm": 0.4869871139526367,
      "learning_rate": 1.0222077446772949e-07,
      "loss": 0.05020198971033096,
      "step": 7142
    },
    {
      "epoch": 0.9656862330902669,
      "grad_norm": 1.108733892440796,
      "learning_rate": 1.0142463387085465e-07,
      "loss": 0.049881063401699066,
      "step": 7143
    },
    {
      "epoch": 0.9658214264590322,
      "grad_norm": 0.43896281719207764,
      "learning_rate": 1.0063159523682142e-07,
      "loss": 0.05548959970474243,
      "step": 7144
    },
    {
      "epoch": 0.9659566198277975,
      "grad_norm": 0.7983781695365906,
      "learning_rate": 9.984165873074102e-08,
      "loss": 0.051128603518009186,
      "step": 7145
    },
    {
      "epoch": 0.9660918131965627,
      "grad_norm": 1.0796408653259277,
      "learning_rate": 9.905482451708526e-08,
      "loss": 0.05442555248737335,
      "step": 7146
    },
    {
      "epoch": 0.966227006565328,
      "grad_norm": 0.6328137516975403,
      "learning_rate": 9.827109275967638e-08,
      "loss": 0.06515852361917496,
      "step": 7147
    },
    {
      "epoch": 0.9663621999340932,
      "grad_norm": 0.4078194499015808,
      "learning_rate": 9.749046362169223e-08,
      "loss": 0.043637827038764954,
      "step": 7148
    },
    {
      "epoch": 0.9664973933028584,
      "grad_norm": 1.3172117471694946,
      "learning_rate": 9.671293726566443e-08,
      "loss": 0.06495068967342377,
      "step": 7149
    },
    {
      "epoch": 0.9666325866716238,
      "grad_norm": 0.3541133403778076,
      "learning_rate": 9.593851385347518e-08,
      "loss": 0.04651781916618347,
      "step": 7150
    },
    {
      "epoch": 0.966767780040389,
      "grad_norm": 0.3603734076023102,
      "learning_rate": 9.516719354636716e-08,
      "loss": 0.05565696954727173,
      "step": 7151
    },
    {
      "epoch": 0.9669029734091543,
      "grad_norm": 0.39890196919441223,
      "learning_rate": 9.439897650493024e-08,
      "loss": 0.05666684731841087,
      "step": 7152
    },
    {
      "epoch": 0.9670381667779195,
      "grad_norm": 0.28428933024406433,
      "learning_rate": 9.363386288911313e-08,
      "loss": 0.04845336079597473,
      "step": 7153
    },
    {
      "epoch": 0.9671733601466848,
      "grad_norm": 0.36257874965667725,
      "learning_rate": 9.287185285821675e-08,
      "loss": 0.04322141036391258,
      "step": 7154
    },
    {
      "epoch": 0.9673085535154501,
      "grad_norm": 0.7991976141929626,
      "learning_rate": 9.211294657089587e-08,
      "loss": 0.06890275329351425,
      "step": 7155
    },
    {
      "epoch": 0.9674437468842153,
      "grad_norm": 0.44612377882003784,
      "learning_rate": 9.135714418515573e-08,
      "loss": 0.050186991691589355,
      "step": 7156
    },
    {
      "epoch": 0.9675789402529806,
      "grad_norm": 0.645582377910614,
      "learning_rate": 9.060444585836381e-08,
      "loss": 0.04344116523861885,
      "step": 7157
    },
    {
      "epoch": 0.9677141336217459,
      "grad_norm": 0.705514132976532,
      "learning_rate": 8.985485174722974e-08,
      "loss": 0.047943323850631714,
      "step": 7158
    },
    {
      "epoch": 0.9678493269905111,
      "grad_norm": 0.3527999520301819,
      "learning_rate": 8.910836200782868e-08,
      "loss": 0.044599514454603195,
      "step": 7159
    },
    {
      "epoch": 0.9679845203592764,
      "grad_norm": 0.5106353163719177,
      "learning_rate": 8.836497679557964e-08,
      "loss": 0.0542168915271759,
      "step": 7160
    },
    {
      "epoch": 0.9681197137280416,
      "grad_norm": 0.4286550283432007,
      "learning_rate": 8.762469626526048e-08,
      "loss": 0.0699646919965744,
      "step": 7161
    },
    {
      "epoch": 0.968254907096807,
      "grad_norm": 0.41920235753059387,
      "learning_rate": 8.688752057100457e-08,
      "loss": 0.03273099660873413,
      "step": 7162
    },
    {
      "epoch": 0.9683901004655722,
      "grad_norm": 0.38292351365089417,
      "learning_rate": 8.615344986629082e-08,
      "loss": 0.03990975767374039,
      "step": 7163
    },
    {
      "epoch": 0.9685252938343374,
      "grad_norm": 0.4461629092693329,
      "learning_rate": 8.542248430396027e-08,
      "loss": 0.04121055454015732,
      "step": 7164
    },
    {
      "epoch": 0.9686604872031027,
      "grad_norm": 0.3781512975692749,
      "learning_rate": 8.469462403620287e-08,
      "loss": 0.04370203614234924,
      "step": 7165
    },
    {
      "epoch": 0.9687956805718679,
      "grad_norm": 0.470559298992157,
      "learning_rate": 8.39698692145624e-08,
      "loss": 0.05675918236374855,
      "step": 7166
    },
    {
      "epoch": 0.9689308739406333,
      "grad_norm": 0.39212754368782043,
      "learning_rate": 8.324821998993648e-08,
      "loss": 0.060417890548706055,
      "step": 7167
    },
    {
      "epoch": 0.9690660673093985,
      "grad_norm": 0.5165414810180664,
      "learning_rate": 8.252967651257826e-08,
      "loss": 0.054235540330410004,
      "step": 7168
    },
    {
      "epoch": 0.9692012606781637,
      "grad_norm": 0.6558708548545837,
      "learning_rate": 8.181423893208973e-08,
      "loss": 0.056982845067977905,
      "step": 7169
    },
    {
      "epoch": 0.969336454046929,
      "grad_norm": 0.44899582862854004,
      "learning_rate": 8.110190739743172e-08,
      "loss": 0.038351260125637054,
      "step": 7170
    },
    {
      "epoch": 0.9694716474156942,
      "grad_norm": 0.3913290500640869,
      "learning_rate": 8.03926820569123e-08,
      "loss": 0.05124525725841522,
      "step": 7171
    },
    {
      "epoch": 0.9696068407844595,
      "grad_norm": 0.391361266374588,
      "learning_rate": 7.968656305819833e-08,
      "loss": 0.04919546842575073,
      "step": 7172
    },
    {
      "epoch": 0.9697420341532248,
      "grad_norm": 0.48596450686454773,
      "learning_rate": 7.898355054830719e-08,
      "loss": 0.05240263417363167,
      "step": 7173
    },
    {
      "epoch": 0.96987722752199,
      "grad_norm": 0.9218450784683228,
      "learning_rate": 7.828364467360849e-08,
      "loss": 0.05866679549217224,
      "step": 7174
    },
    {
      "epoch": 0.9700124208907553,
      "grad_norm": 0.5449346303939819,
      "learning_rate": 7.758684557982731e-08,
      "loss": 0.05991728603839874,
      "step": 7175
    },
    {
      "epoch": 0.9701476142595206,
      "grad_norm": 0.5859743356704712,
      "learning_rate": 7.689315341204262e-08,
      "loss": 0.04650276526808739,
      "step": 7176
    },
    {
      "epoch": 0.9702828076282858,
      "grad_norm": 1.184984803199768,
      "learning_rate": 7.62025683146822e-08,
      "loss": 0.05737435445189476,
      "step": 7177
    },
    {
      "epoch": 0.9704180009970511,
      "grad_norm": 0.4459211528301239,
      "learning_rate": 7.551509043152937e-08,
      "loss": 0.04736095666885376,
      "step": 7178
    },
    {
      "epoch": 0.9705531943658163,
      "grad_norm": 0.3969394564628601,
      "learning_rate": 7.483071990572132e-08,
      "loss": 0.04486561939120293,
      "step": 7179
    },
    {
      "epoch": 0.9706883877345817,
      "grad_norm": 0.3110262155532837,
      "learning_rate": 7.414945687975072e-08,
      "loss": 0.03385123610496521,
      "step": 7180
    },
    {
      "epoch": 0.9708235811033469,
      "grad_norm": 0.46259990334510803,
      "learning_rate": 7.347130149545578e-08,
      "loss": 0.0484342947602272,
      "step": 7181
    },
    {
      "epoch": 0.9709587744721121,
      "grad_norm": 0.48633572459220886,
      "learning_rate": 7.279625389403355e-08,
      "loss": 0.04074862599372864,
      "step": 7182
    },
    {
      "epoch": 0.9710939678408774,
      "grad_norm": 0.40412136912345886,
      "learning_rate": 7.212431421603327e-08,
      "loss": 0.053075142204761505,
      "step": 7183
    },
    {
      "epoch": 0.9712291612096426,
      "grad_norm": 0.7840222716331482,
      "learning_rate": 7.145548260135638e-08,
      "loss": 0.04890257492661476,
      "step": 7184
    },
    {
      "epoch": 0.971364354578408,
      "grad_norm": 0.6063414216041565,
      "learning_rate": 7.078975918925645e-08,
      "loss": 0.051963746547698975,
      "step": 7185
    },
    {
      "epoch": 0.9714995479471732,
      "grad_norm": 0.6529886722564697,
      "learning_rate": 7.012714411834098e-08,
      "loss": 0.056936487555503845,
      "step": 7186
    },
    {
      "epoch": 0.9716347413159384,
      "grad_norm": 0.3436601161956787,
      "learning_rate": 6.946763752656959e-08,
      "loss": 0.03833438456058502,
      "step": 7187
    },
    {
      "epoch": 0.9717699346847037,
      "grad_norm": 1.0106734037399292,
      "learning_rate": 6.881123955125579e-08,
      "loss": 0.0608120858669281,
      "step": 7188
    },
    {
      "epoch": 0.971905128053469,
      "grad_norm": 0.377293199300766,
      "learning_rate": 6.815795032906524e-08,
      "loss": 0.05356192961335182,
      "step": 7189
    },
    {
      "epoch": 0.9720403214222343,
      "grad_norm": 0.45921361446380615,
      "learning_rate": 6.750776999601415e-08,
      "loss": 0.04495931789278984,
      "step": 7190
    },
    {
      "epoch": 0.9721755147909995,
      "grad_norm": 0.5022160410881042,
      "learning_rate": 6.68606986874759e-08,
      "loss": 0.07296773791313171,
      "step": 7191
    },
    {
      "epoch": 0.9723107081597647,
      "grad_norm": 0.5256549715995789,
      "learning_rate": 6.62167365381744e-08,
      "loss": 0.058248236775398254,
      "step": 7192
    },
    {
      "epoch": 0.97244590152853,
      "grad_norm": 0.44644707441329956,
      "learning_rate": 6.557588368218237e-08,
      "loss": 0.05250312760472298,
      "step": 7193
    },
    {
      "epoch": 0.9725810948972953,
      "grad_norm": 0.546574592590332,
      "learning_rate": 6.493814025293476e-08,
      "loss": 0.042590174823999405,
      "step": 7194
    },
    {
      "epoch": 0.9727162882660606,
      "grad_norm": 0.36539652943611145,
      "learning_rate": 6.430350638320704e-08,
      "loss": 0.04494466632604599,
      "step": 7195
    },
    {
      "epoch": 0.9728514816348258,
      "grad_norm": 0.41228771209716797,
      "learning_rate": 6.367198220513848e-08,
      "loss": 0.05681285262107849,
      "step": 7196
    },
    {
      "epoch": 0.972986675003591,
      "grad_norm": 0.3799850344657898,
      "learning_rate": 6.304356785021226e-08,
      "loss": 0.04737436771392822,
      "step": 7197
    },
    {
      "epoch": 0.9731218683723564,
      "grad_norm": 0.5011929869651794,
      "learning_rate": 6.241826344926704e-08,
      "loss": 0.058452725410461426,
      "step": 7198
    },
    {
      "epoch": 0.9732570617411216,
      "grad_norm": 0.4156959354877472,
      "learning_rate": 6.17960691324987e-08,
      "loss": 0.04878932610154152,
      "step": 7199
    },
    {
      "epoch": 0.9733922551098868,
      "grad_norm": 0.5496264696121216,
      "learning_rate": 6.117698502944857e-08,
      "loss": 0.05052957683801651,
      "step": 7200
    },
    {
      "epoch": 0.9735274484786521,
      "grad_norm": 0.523495078086853,
      "learning_rate": 6.056101126901358e-08,
      "loss": 0.05461758375167847,
      "step": 7201
    },
    {
      "epoch": 0.9736626418474174,
      "grad_norm": 0.6230342984199524,
      "learning_rate": 5.994814797944281e-08,
      "loss": 0.0575384721159935,
      "step": 7202
    },
    {
      "epoch": 0.9737978352161827,
      "grad_norm": 0.3428310453891754,
      "learning_rate": 5.933839528833751e-08,
      "loss": 0.04423064738512039,
      "step": 7203
    },
    {
      "epoch": 0.9739330285849479,
      "grad_norm": 0.2559383809566498,
      "learning_rate": 5.873175332265279e-08,
      "loss": 0.039129048585891724,
      "step": 7204
    },
    {
      "epoch": 0.9740682219537131,
      "grad_norm": 0.8830430507659912,
      "learning_rate": 5.812822220869096e-08,
      "loss": 0.04372261092066765,
      "step": 7205
    },
    {
      "epoch": 0.9742034153224784,
      "grad_norm": 0.4651319682598114,
      "learning_rate": 5.752780207211483e-08,
      "loss": 0.05418848991394043,
      "step": 7206
    },
    {
      "epoch": 0.9743386086912437,
      "grad_norm": 0.6852202415466309,
      "learning_rate": 5.693049303793274e-08,
      "loss": 0.054455917328596115,
      "step": 7207
    },
    {
      "epoch": 0.974473802060009,
      "grad_norm": 0.3617790639400482,
      "learning_rate": 5.6336295230508536e-08,
      "loss": 0.05614317208528519,
      "step": 7208
    },
    {
      "epoch": 0.9746089954287742,
      "grad_norm": 0.4937783181667328,
      "learning_rate": 5.5745208773558266e-08,
      "loss": 0.04383295774459839,
      "step": 7209
    },
    {
      "epoch": 0.9747441887975394,
      "grad_norm": 1.3043458461761475,
      "learning_rate": 5.515723379014681e-08,
      "loss": 0.06821008026599884,
      "step": 7210
    },
    {
      "epoch": 0.9748793821663048,
      "grad_norm": 0.593988835811615,
      "learning_rate": 5.4572370402694583e-08,
      "loss": 0.06013453006744385,
      "step": 7211
    },
    {
      "epoch": 0.97501457553507,
      "grad_norm": 1.361351490020752,
      "learning_rate": 5.399061873297417e-08,
      "loss": 0.04968570917844772,
      "step": 7212
    },
    {
      "epoch": 0.9751497689038353,
      "grad_norm": 0.6637198328971863,
      "learning_rate": 5.341197890210869e-08,
      "loss": 0.051711756736040115,
      "step": 7213
    },
    {
      "epoch": 0.9752849622726005,
      "grad_norm": 0.5340448617935181,
      "learning_rate": 5.283645103057344e-08,
      "loss": 0.0720614492893219,
      "step": 7214
    },
    {
      "epoch": 0.9754201556413657,
      "grad_norm": 0.3676905035972595,
      "learning_rate": 5.226403523819756e-08,
      "loss": 0.04148132726550102,
      "step": 7215
    },
    {
      "epoch": 0.9755553490101311,
      "grad_norm": 0.7414591312408447,
      "learning_rate": 5.169473164416072e-08,
      "loss": 0.06510700285434723,
      "step": 7216
    },
    {
      "epoch": 0.9756905423788963,
      "grad_norm": 0.48349106311798096,
      "learning_rate": 5.112854036699477e-08,
      "loss": 0.0573880709707737,
      "step": 7217
    },
    {
      "epoch": 0.9758257357476616,
      "grad_norm": 0.37112605571746826,
      "learning_rate": 5.0565461524583745e-08,
      "loss": 0.053566765040159225,
      "step": 7218
    },
    {
      "epoch": 0.9759609291164268,
      "grad_norm": 0.5526524186134338,
      "learning_rate": 5.0005495234163865e-08,
      "loss": 0.048564400523900986,
      "step": 7219
    },
    {
      "epoch": 0.9760961224851921,
      "grad_norm": 1.4637776613235474,
      "learning_rate": 4.9448641612321874e-08,
      "loss": 0.05839678645133972,
      "step": 7220
    },
    {
      "epoch": 0.9762313158539574,
      "grad_norm": 0.49976494908332825,
      "learning_rate": 4.889490077500003e-08,
      "loss": 0.04154215753078461,
      "step": 7221
    },
    {
      "epoch": 0.9763665092227226,
      "grad_norm": 0.3476382791996002,
      "learning_rate": 4.8344272837489434e-08,
      "loss": 0.05859503149986267,
      "step": 7222
    },
    {
      "epoch": 0.9765017025914879,
      "grad_norm": 0.4435412585735321,
      "learning_rate": 4.779675791443172e-08,
      "loss": 0.044656869024038315,
      "step": 7223
    },
    {
      "epoch": 0.9766368959602532,
      "grad_norm": 0.6044033169746399,
      "learning_rate": 4.72523561198257e-08,
      "loss": 0.04710233211517334,
      "step": 7224
    },
    {
      "epoch": 0.9767720893290184,
      "grad_norm": 0.3266955316066742,
      "learning_rate": 4.6711067567014044e-08,
      "loss": 0.035173751413822174,
      "step": 7225
    },
    {
      "epoch": 0.9769072826977837,
      "grad_norm": 0.25391367077827454,
      "learning_rate": 4.6172892368701595e-08,
      "loss": 0.03522966429591179,
      "step": 7226
    },
    {
      "epoch": 0.9770424760665489,
      "grad_norm": 0.5141711831092834,
      "learning_rate": 4.5637830636935385e-08,
      "loss": 0.037283703684806824,
      "step": 7227
    },
    {
      "epoch": 0.9771776694353141,
      "grad_norm": 0.4566790461540222,
      "learning_rate": 4.5105882483119643e-08,
      "loss": 0.07286709547042847,
      "step": 7228
    },
    {
      "epoch": 0.9773128628040795,
      "grad_norm": 0.4071083664894104,
      "learning_rate": 4.4577048018007436e-08,
      "loss": 0.04396776109933853,
      "step": 7229
    },
    {
      "epoch": 0.9774480561728447,
      "grad_norm": 0.5844578742980957,
      "learning_rate": 4.405132735170569e-08,
      "loss": 0.07535350322723389,
      "step": 7230
    },
    {
      "epoch": 0.97758324954161,
      "grad_norm": 1.004151463508606,
      "learning_rate": 4.3528720593675184e-08,
      "loss": 0.06708019971847534,
      "step": 7231
    },
    {
      "epoch": 0.9777184429103752,
      "grad_norm": 0.24426449835300446,
      "learning_rate": 4.300922785271888e-08,
      "loss": 0.05083953216671944,
      "step": 7232
    },
    {
      "epoch": 0.9778536362791405,
      "grad_norm": 0.5798628926277161,
      "learning_rate": 4.249284923700358e-08,
      "loss": 0.051040276885032654,
      "step": 7233
    },
    {
      "epoch": 0.9779888296479058,
      "grad_norm": 0.7276411652565002,
      "learning_rate": 4.197958485404163e-08,
      "loss": 0.057106249034404755,
      "step": 7234
    },
    {
      "epoch": 0.978124023016671,
      "grad_norm": 0.36441540718078613,
      "learning_rate": 4.1469434810694206e-08,
      "loss": 0.044418517500162125,
      "step": 7235
    },
    {
      "epoch": 0.9782592163854363,
      "grad_norm": 1.381124496459961,
      "learning_rate": 4.096239921317968e-08,
      "loss": 0.05549690127372742,
      "step": 7236
    },
    {
      "epoch": 0.9783944097542016,
      "grad_norm": 0.371625155210495,
      "learning_rate": 4.045847816706361e-08,
      "loss": 0.0550915002822876,
      "step": 7237
    },
    {
      "epoch": 0.9785296031229668,
      "grad_norm": 0.3654465973377228,
      "learning_rate": 3.9957671777268724e-08,
      "loss": 0.03545597568154335,
      "step": 7238
    },
    {
      "epoch": 0.9786647964917321,
      "grad_norm": 1.3275188207626343,
      "learning_rate": 3.945998014806163e-08,
      "loss": 0.056324053555727005,
      "step": 7239
    },
    {
      "epoch": 0.9787999898604973,
      "grad_norm": 0.6294799447059631,
      "learning_rate": 3.896540338306609e-08,
      "loss": 0.05544418841600418,
      "step": 7240
    },
    {
      "epoch": 0.9789351832292627,
      "grad_norm": 0.429246187210083,
      "learning_rate": 3.847394158525641e-08,
      "loss": 0.05259708687663078,
      "step": 7241
    },
    {
      "epoch": 0.9790703765980279,
      "grad_norm": 0.6167306900024414,
      "learning_rate": 3.798559485695574e-08,
      "loss": 0.0666983425617218,
      "step": 7242
    },
    {
      "epoch": 0.9792055699667931,
      "grad_norm": 0.4300607442855835,
      "learning_rate": 3.7500363299842746e-08,
      "loss": 0.02788853645324707,
      "step": 7243
    },
    {
      "epoch": 0.9793407633355584,
      "grad_norm": 0.3931945264339447,
      "learning_rate": 3.701824701494327e-08,
      "loss": 0.022445425391197205,
      "step": 7244
    },
    {
      "epoch": 0.9794759567043236,
      "grad_norm": 0.3176533281803131,
      "learning_rate": 3.653924610263703e-08,
      "loss": 0.03755953907966614,
      "step": 7245
    },
    {
      "epoch": 0.979611150073089,
      "grad_norm": 0.4524339735507965,
      "learning_rate": 3.6063360662654255e-08,
      "loss": 0.04998893290758133,
      "step": 7246
    },
    {
      "epoch": 0.9797463434418542,
      "grad_norm": 0.85714191198349,
      "learning_rate": 3.559059079407734e-08,
      "loss": 0.05953269824385643,
      "step": 7247
    },
    {
      "epoch": 0.9798815368106194,
      "grad_norm": 0.5563637614250183,
      "learning_rate": 3.512093659533922e-08,
      "loss": 0.04398603364825249,
      "step": 7248
    },
    {
      "epoch": 0.9800167301793847,
      "grad_norm": 0.5393894910812378,
      "learning_rate": 3.4654398164225e-08,
      "loss": 0.056876182556152344,
      "step": 7249
    },
    {
      "epoch": 0.98015192354815,
      "grad_norm": 0.6972790956497192,
      "learning_rate": 3.4190975597870325e-08,
      "loss": 0.06028540059924126,
      "step": 7250
    },
    {
      "epoch": 0.9802871169169153,
      "grad_norm": 0.3229658603668213,
      "learning_rate": 3.373066899276134e-08,
      "loss": 0.03550612926483154,
      "step": 7251
    },
    {
      "epoch": 0.9804223102856805,
      "grad_norm": 0.2454824298620224,
      "learning_rate": 3.3273478444736386e-08,
      "loss": 0.03997718542814255,
      "step": 7252
    },
    {
      "epoch": 0.9805575036544457,
      "grad_norm": 0.39571788907051086,
      "learning_rate": 3.281940404898764e-08,
      "loss": 0.05327256768941879,
      "step": 7253
    },
    {
      "epoch": 0.980692697023211,
      "grad_norm": 0.4503594636917114,
      "learning_rate": 3.236844590005117e-08,
      "loss": 0.05812521651387215,
      "step": 7254
    },
    {
      "epoch": 0.9808278903919763,
      "grad_norm": 0.4966477155685425,
      "learning_rate": 3.192060409182351e-08,
      "loss": 0.05891026556491852,
      "step": 7255
    },
    {
      "epoch": 0.9809630837607415,
      "grad_norm": 0.3571639955043793,
      "learning_rate": 3.147587871754509e-08,
      "loss": 0.05026090145111084,
      "step": 7256
    },
    {
      "epoch": 0.9810982771295068,
      "grad_norm": 0.27719926834106445,
      "learning_rate": 3.1034269869810174e-08,
      "loss": 0.03677530586719513,
      "step": 7257
    },
    {
      "epoch": 0.981233470498272,
      "grad_norm": 0.29793864488601685,
      "learning_rate": 3.05957776405652e-08,
      "loss": 0.04622465372085571,
      "step": 7258
    },
    {
      "epoch": 0.9813686638670374,
      "grad_norm": 0.7389929294586182,
      "learning_rate": 3.016040212110549e-08,
      "loss": 0.061945840716362,
      "step": 7259
    },
    {
      "epoch": 0.9815038572358026,
      "grad_norm": 1.515032172203064,
      "learning_rate": 2.9728143402078522e-08,
      "loss": 0.0556643009185791,
      "step": 7260
    },
    {
      "epoch": 0.9816390506045678,
      "grad_norm": 0.5657045245170593,
      "learning_rate": 2.9299001573483975e-08,
      "loss": 0.05606083199381828,
      "step": 7261
    },
    {
      "epoch": 0.9817742439733331,
      "grad_norm": 0.537739634513855,
      "learning_rate": 2.8872976724670375e-08,
      "loss": 0.06261782348155975,
      "step": 7262
    },
    {
      "epoch": 0.9819094373420983,
      "grad_norm": 0.5054749846458435,
      "learning_rate": 2.8450068944338436e-08,
      "loss": 0.04781985282897949,
      "step": 7263
    },
    {
      "epoch": 0.9820446307108637,
      "grad_norm": 0.5242066383361816,
      "learning_rate": 2.803027832054106e-08,
      "loss": 0.06356542557477951,
      "step": 7264
    },
    {
      "epoch": 0.9821798240796289,
      "grad_norm": 0.4420137107372284,
      "learning_rate": 2.7613604940679995e-08,
      "loss": 0.033886268734931946,
      "step": 7265
    },
    {
      "epoch": 0.9823150174483941,
      "grad_norm": 0.37018951773643494,
      "learning_rate": 2.7200048891509176e-08,
      "loss": 0.057096101343631744,
      "step": 7266
    },
    {
      "epoch": 0.9824502108171594,
      "grad_norm": 0.3043867349624634,
      "learning_rate": 2.67896102591314e-08,
      "loss": 0.04797044396400452,
      "step": 7267
    },
    {
      "epoch": 0.9825854041859247,
      "grad_norm": 0.3050529956817627,
      "learning_rate": 2.6382289129004978e-08,
      "loss": 0.057027190923690796,
      "step": 7268
    },
    {
      "epoch": 0.98272059755469,
      "grad_norm": 0.3412294387817383,
      "learning_rate": 2.5978085585935395e-08,
      "loss": 0.040994659066200256,
      "step": 7269
    },
    {
      "epoch": 0.9828557909234552,
      "grad_norm": 0.43741175532341003,
      "learning_rate": 2.5576999714078676e-08,
      "loss": 0.051810264587402344,
      "step": 7270
    },
    {
      "epoch": 0.9829909842922204,
      "grad_norm": 0.20743736624717712,
      "learning_rate": 2.517903159694468e-08,
      "loss": 0.04013640806078911,
      "step": 7271
    },
    {
      "epoch": 0.9831261776609858,
      "grad_norm": 0.7194942235946655,
      "learning_rate": 2.4784181317390465e-08,
      "loss": 0.0506615936756134,
      "step": 7272
    },
    {
      "epoch": 0.983261371029751,
      "grad_norm": 0.7069981694221497,
      "learning_rate": 2.4392448957628598e-08,
      "loss": 0.06042274460196495,
      "step": 7273
    },
    {
      "epoch": 0.9833965643985163,
      "grad_norm": 0.47161486744880676,
      "learning_rate": 2.4003834599217177e-08,
      "loss": 0.05850675702095032,
      "step": 7274
    },
    {
      "epoch": 0.9835317577672815,
      "grad_norm": 0.4753275513648987,
      "learning_rate": 2.3618338323071474e-08,
      "loss": 0.05263178050518036,
      "step": 7275
    },
    {
      "epoch": 0.9836669511360467,
      "grad_norm": 0.6359841823577881,
      "learning_rate": 2.3235960209448958e-08,
      "loss": 0.0439833328127861,
      "step": 7276
    },
    {
      "epoch": 0.9838021445048121,
      "grad_norm": 0.2887997329235077,
      "learning_rate": 2.2856700337967606e-08,
      "loss": 0.03405828773975372,
      "step": 7277
    },
    {
      "epoch": 0.9839373378735773,
      "grad_norm": 0.3563024699687958,
      "learning_rate": 2.2480558787587592e-08,
      "loss": 0.0542268306016922,
      "step": 7278
    },
    {
      "epoch": 0.9840725312423426,
      "grad_norm": 0.5194105505943298,
      "learning_rate": 2.2107535636626263e-08,
      "loss": 0.04978272318840027,
      "step": 7279
    },
    {
      "epoch": 0.9842077246111078,
      "grad_norm": 0.3321051299571991,
      "learning_rate": 2.1737630962746502e-08,
      "loss": 0.03914583474397659,
      "step": 7280
    },
    {
      "epoch": 0.984342917979873,
      "grad_norm": 0.5193450450897217,
      "learning_rate": 2.1370844842966696e-08,
      "loss": 0.05838894844055176,
      "step": 7281
    },
    {
      "epoch": 0.9844781113486384,
      "grad_norm": 0.4899522066116333,
      "learning_rate": 2.100717735365243e-08,
      "loss": 0.05410711467266083,
      "step": 7282
    },
    {
      "epoch": 0.9846133047174036,
      "grad_norm": 0.998993456363678,
      "learning_rate": 2.0646628570521464e-08,
      "loss": 0.04839441180229187,
      "step": 7283
    },
    {
      "epoch": 0.9847484980861688,
      "grad_norm": 0.34424448013305664,
      "learning_rate": 2.028919856864375e-08,
      "loss": 0.03327178955078125,
      "step": 7284
    },
    {
      "epoch": 0.9848836914549342,
      "grad_norm": 0.7587503790855408,
      "learning_rate": 1.9934887422434766e-08,
      "loss": 0.04919227957725525,
      "step": 7285
    },
    {
      "epoch": 0.9850188848236994,
      "grad_norm": 0.3807278275489807,
      "learning_rate": 1.9583695205665496e-08,
      "loss": 0.04222244769334793,
      "step": 7286
    },
    {
      "epoch": 0.9851540781924647,
      "grad_norm": 0.918523907661438,
      "learning_rate": 1.9235621991457454e-08,
      "loss": 0.05059755593538284,
      "step": 7287
    },
    {
      "epoch": 0.9852892715612299,
      "grad_norm": 0.6230126023292542,
      "learning_rate": 1.889066785227933e-08,
      "loss": 0.0633402019739151,
      "step": 7288
    },
    {
      "epoch": 0.9854244649299951,
      "grad_norm": 0.3672540485858917,
      "learning_rate": 1.854883285995368e-08,
      "loss": 0.0501851849257946,
      "step": 7289
    },
    {
      "epoch": 0.9855596582987605,
      "grad_norm": 0.6481570601463318,
      "learning_rate": 1.8210117085651902e-08,
      "loss": 0.04359808564186096,
      "step": 7290
    },
    {
      "epoch": 0.9856948516675257,
      "grad_norm": 0.344828337430954,
      "learning_rate": 1.7874520599894252e-08,
      "loss": 0.03428077697753906,
      "step": 7291
    },
    {
      "epoch": 0.985830045036291,
      "grad_norm": 0.23351861536502838,
      "learning_rate": 1.7542043472558166e-08,
      "loss": 0.038654983043670654,
      "step": 7292
    },
    {
      "epoch": 0.9859652384050562,
      "grad_norm": 0.39137551188468933,
      "learning_rate": 1.7212685772864945e-08,
      "loss": 0.046825893223285675,
      "step": 7293
    },
    {
      "epoch": 0.9861004317738215,
      "grad_norm": 0.562049388885498,
      "learning_rate": 1.68864475693864e-08,
      "loss": 0.0369291752576828,
      "step": 7294
    },
    {
      "epoch": 0.9862356251425868,
      "grad_norm": 0.30741167068481445,
      "learning_rate": 1.6563328930051526e-08,
      "loss": 0.04968493431806564,
      "step": 7295
    },
    {
      "epoch": 0.986370818511352,
      "grad_norm": 0.5242623090744019,
      "learning_rate": 1.624332992213151e-08,
      "loss": 0.03748743236064911,
      "step": 7296
    },
    {
      "epoch": 0.9865060118801173,
      "grad_norm": 0.499889612197876,
      "learning_rate": 1.5926450612254728e-08,
      "loss": 0.044719040393829346,
      "step": 7297
    },
    {
      "epoch": 0.9866412052488825,
      "grad_norm": 1.1401457786560059,
      "learning_rate": 1.5612691066395068e-08,
      "loss": 0.05591359734535217,
      "step": 7298
    },
    {
      "epoch": 0.9867763986176478,
      "grad_norm": 0.27081045508384705,
      "learning_rate": 1.530205134987861e-08,
      "loss": 0.03869227319955826,
      "step": 7299
    },
    {
      "epoch": 0.9869115919864131,
      "grad_norm": 0.46528851985931396,
      "learning_rate": 1.499453152738528e-08,
      "loss": 0.04791422188282013,
      "step": 7300
    },
    {
      "epoch": 0.9870467853551783,
      "grad_norm": 0.404367595911026,
      "learning_rate": 1.4690131662938866e-08,
      "loss": 0.037147656083106995,
      "step": 7301
    },
    {
      "epoch": 0.9871819787239436,
      "grad_norm": 0.41163381934165955,
      "learning_rate": 1.438885181991867e-08,
      "loss": 0.03467421233654022,
      "step": 7302
    },
    {
      "epoch": 0.9873171720927089,
      "grad_norm": 0.5218137502670288,
      "learning_rate": 1.4090692061052846e-08,
      "loss": 0.04829150438308716,
      "step": 7303
    },
    {
      "epoch": 0.9874523654614741,
      "grad_norm": 0.31034398078918457,
      "learning_rate": 1.3795652448420071e-08,
      "loss": 0.03866346925497055,
      "step": 7304
    },
    {
      "epoch": 0.9875875588302394,
      "grad_norm": 0.34017080068588257,
      "learning_rate": 1.3503733043447874e-08,
      "loss": 0.040759533643722534,
      "step": 7305
    },
    {
      "epoch": 0.9877227521990046,
      "grad_norm": 0.2991945147514343,
      "learning_rate": 1.3214933906915971e-08,
      "loss": 0.04314536601305008,
      "step": 7306
    },
    {
      "epoch": 0.9878579455677698,
      "grad_norm": 0.48050668835639954,
      "learning_rate": 1.2929255098954596e-08,
      "loss": 0.05446822568774223,
      "step": 7307
    },
    {
      "epoch": 0.9879931389365352,
      "grad_norm": 0.6828495264053345,
      "learning_rate": 1.2646696679042835e-08,
      "loss": 0.04760288447141647,
      "step": 7308
    },
    {
      "epoch": 0.9881283323053004,
      "grad_norm": 0.5227488279342651,
      "learning_rate": 1.2367258706010298e-08,
      "loss": 0.04110647737979889,
      "step": 7309
    },
    {
      "epoch": 0.9882635256740657,
      "grad_norm": 0.495807945728302,
      "learning_rate": 1.2090941238040443e-08,
      "loss": 0.03955543041229248,
      "step": 7310
    },
    {
      "epoch": 0.9883987190428309,
      "grad_norm": 0.7100663781166077,
      "learning_rate": 1.1817744332660584e-08,
      "loss": 0.032148346304893494,
      "step": 7311
    },
    {
      "epoch": 0.9885339124115962,
      "grad_norm": 0.7787774205207825,
      "learning_rate": 1.1547668046751891e-08,
      "loss": 0.04325483739376068,
      "step": 7312
    },
    {
      "epoch": 0.9886691057803615,
      "grad_norm": 0.5496942400932312,
      "learning_rate": 1.1280712436549379e-08,
      "loss": 0.04591679573059082,
      "step": 7313
    },
    {
      "epoch": 0.9888042991491267,
      "grad_norm": 0.47885429859161377,
      "learning_rate": 1.1016877557630257e-08,
      "loss": 0.05158200114965439,
      "step": 7314
    },
    {
      "epoch": 0.988939492517892,
      "grad_norm": 0.42680996656417847,
      "learning_rate": 1.0756163464928915e-08,
      "loss": 0.05484379827976227,
      "step": 7315
    },
    {
      "epoch": 0.9890746858866573,
      "grad_norm": 0.7590004205703735,
      "learning_rate": 1.0498570212726932e-08,
      "loss": 0.05555178225040436,
      "step": 7316
    },
    {
      "epoch": 0.9892098792554225,
      "grad_norm": 0.32244136929512024,
      "learning_rate": 1.024409785465641e-08,
      "loss": 0.039124730974435806,
      "step": 7317
    },
    {
      "epoch": 0.9893450726241878,
      "grad_norm": 0.7999218106269836,
      "learning_rate": 9.992746443699962e-09,
      "loss": 0.06894140690565109,
      "step": 7318
    },
    {
      "epoch": 0.989480265992953,
      "grad_norm": 0.7047271728515625,
      "learning_rate": 9.744516032190731e-09,
      "loss": 0.06663292646408081,
      "step": 7319
    },
    {
      "epoch": 0.9896154593617184,
      "grad_norm": 0.92720627784729,
      "learning_rate": 9.499406671809041e-09,
      "loss": 0.058796972036361694,
      "step": 7320
    },
    {
      "epoch": 0.9897506527304836,
      "grad_norm": 0.615882933139801,
      "learning_rate": 9.2574184135924e-09,
      "loss": 0.05456940829753876,
      "step": 7321
    },
    {
      "epoch": 0.9898858460992488,
      "grad_norm": 0.49301859736442566,
      "learning_rate": 9.018551307920508e-09,
      "loss": 0.05672220140695572,
      "step": 7322
    },
    {
      "epoch": 0.9900210394680141,
      "grad_norm": 1.0820434093475342,
      "learning_rate": 8.782805404526917e-09,
      "loss": 0.04235593229532242,
      "step": 7323
    },
    {
      "epoch": 0.9901562328367793,
      "grad_norm": 0.39856624603271484,
      "learning_rate": 8.55018075249736e-09,
      "loss": 0.061092808842659,
      "step": 7324
    },
    {
      "epoch": 0.9902914262055447,
      "grad_norm": 0.46612387895584106,
      "learning_rate": 8.320677400264764e-09,
      "loss": 0.049432266503572464,
      "step": 7325
    },
    {
      "epoch": 0.9904266195743099,
      "grad_norm": 1.114904522895813,
      "learning_rate": 8.094295395610906e-09,
      "loss": 0.048036329448223114,
      "step": 7326
    },
    {
      "epoch": 0.9905618129430751,
      "grad_norm": 0.7147827744483948,
      "learning_rate": 7.87103478567308e-09,
      "loss": 0.05065537989139557,
      "step": 7327
    },
    {
      "epoch": 0.9906970063118404,
      "grad_norm": 0.4050542414188385,
      "learning_rate": 7.65089561693244e-09,
      "loss": 0.050760142505168915,
      "step": 7328
    },
    {
      "epoch": 0.9908321996806057,
      "grad_norm": 0.7346510887145996,
      "learning_rate": 7.433877935225652e-09,
      "loss": 0.04969797655940056,
      "step": 7329
    },
    {
      "epoch": 0.990967393049371,
      "grad_norm": 0.38470184803009033,
      "learning_rate": 7.219981785733243e-09,
      "loss": 0.05018511041998863,
      "step": 7330
    },
    {
      "epoch": 0.9911025864181362,
      "grad_norm": 0.2662709057331085,
      "learning_rate": 7.009207212992919e-09,
      "loss": 0.040013380348682404,
      "step": 7331
    },
    {
      "epoch": 0.9912377797869014,
      "grad_norm": 0.8572955131530762,
      "learning_rate": 6.801554260889575e-09,
      "loss": 0.05664151906967163,
      "step": 7332
    },
    {
      "epoch": 0.9913729731556667,
      "grad_norm": 0.4344453811645508,
      "learning_rate": 6.5970229726552976e-09,
      "loss": 0.044277727603912354,
      "step": 7333
    },
    {
      "epoch": 0.991508166524432,
      "grad_norm": 0.7193469405174255,
      "learning_rate": 6.3956133908743556e-09,
      "loss": 0.04339740052819252,
      "step": 7334
    },
    {
      "epoch": 0.9916433598931972,
      "grad_norm": 0.33341583609580994,
      "learning_rate": 6.197325557483202e-09,
      "loss": 0.04601786285638809,
      "step": 7335
    },
    {
      "epoch": 0.9917785532619625,
      "grad_norm": 0.4186420738697052,
      "learning_rate": 6.002159513765482e-09,
      "loss": 0.05796504020690918,
      "step": 7336
    },
    {
      "epoch": 0.9919137466307277,
      "grad_norm": 0.5352140665054321,
      "learning_rate": 5.810115300355357e-09,
      "loss": 0.0758514553308487,
      "step": 7337
    },
    {
      "epoch": 0.9920489399994931,
      "grad_norm": 0.312301903963089,
      "learning_rate": 5.621192957239174e-09,
      "loss": 0.05176236480474472,
      "step": 7338
    },
    {
      "epoch": 0.9921841333682583,
      "grad_norm": 0.5666396021842957,
      "learning_rate": 5.435392523748806e-09,
      "loss": 0.049953341484069824,
      "step": 7339
    },
    {
      "epoch": 0.9923193267370235,
      "grad_norm": 0.6477637887001038,
      "learning_rate": 5.252714038571638e-09,
      "loss": 0.046043433248996735,
      "step": 7340
    },
    {
      "epoch": 0.9924545201057888,
      "grad_norm": 0.44991573691368103,
      "learning_rate": 5.073157539742246e-09,
      "loss": 0.04542979598045349,
      "step": 7341
    },
    {
      "epoch": 0.992589713474554,
      "grad_norm": 0.42734628915786743,
      "learning_rate": 4.896723064642394e-09,
      "loss": 0.05863310396671295,
      "step": 7342
    },
    {
      "epoch": 0.9927249068433194,
      "grad_norm": 0.7168428301811218,
      "learning_rate": 4.723410650009363e-09,
      "loss": 0.03537128120660782,
      "step": 7343
    },
    {
      "epoch": 0.9928601002120846,
      "grad_norm": 0.33194559812545776,
      "learning_rate": 4.553220331925956e-09,
      "loss": 0.044486235827207565,
      "step": 7344
    },
    {
      "epoch": 0.9929952935808498,
      "grad_norm": 0.7351543307304382,
      "learning_rate": 4.38615214582716e-09,
      "loss": 0.045731037855148315,
      "step": 7345
    },
    {
      "epoch": 0.9931304869496151,
      "grad_norm": 0.40694817900657654,
      "learning_rate": 4.2222061265001496e-09,
      "loss": 0.047724440693855286,
      "step": 7346
    },
    {
      "epoch": 0.9932656803183804,
      "grad_norm": 0.5287313461303711,
      "learning_rate": 4.0613823080742905e-09,
      "loss": 0.04364553838968277,
      "step": 7347
    },
    {
      "epoch": 0.9934008736871457,
      "grad_norm": 0.5800458192825317,
      "learning_rate": 3.903680724037795e-09,
      "loss": 0.06558582186698914,
      "step": 7348
    },
    {
      "epoch": 0.9935360670559109,
      "grad_norm": 0.46178337931632996,
      "learning_rate": 3.749101407224398e-09,
      "loss": 0.04394470155239105,
      "step": 7349
    },
    {
      "epoch": 0.9936712604246761,
      "grad_norm": 0.4155826270580292,
      "learning_rate": 3.597644389818355e-09,
      "loss": 0.03946772962808609,
      "step": 7350
    },
    {
      "epoch": 0.9938064537934415,
      "grad_norm": 0.3430546224117279,
      "learning_rate": 3.4493097033527767e-09,
      "loss": 0.04388341307640076,
      "step": 7351
    },
    {
      "epoch": 0.9939416471622067,
      "grad_norm": 0.7640652060508728,
      "learning_rate": 3.3040973787112904e-09,
      "loss": 0.045907557010650635,
      "step": 7352
    },
    {
      "epoch": 0.994076840530972,
      "grad_norm": 0.5630175471305847,
      "learning_rate": 3.162007446129711e-09,
      "loss": 0.06320783495903015,
      "step": 7353
    },
    {
      "epoch": 0.9942120338997372,
      "grad_norm": 0.5912644863128662,
      "learning_rate": 3.023039935191041e-09,
      "loss": 0.047093722969293594,
      "step": 7354
    },
    {
      "epoch": 0.9943472272685024,
      "grad_norm": 0.46691006422042847,
      "learning_rate": 2.887194874830468e-09,
      "loss": 0.04925764724612236,
      "step": 7355
    },
    {
      "epoch": 0.9944824206372678,
      "grad_norm": 0.7257679104804993,
      "learning_rate": 2.7544722933287026e-09,
      "loss": 0.054604671895504,
      "step": 7356
    },
    {
      "epoch": 0.994617614006033,
      "grad_norm": 0.5620920062065125,
      "learning_rate": 2.6248722183203066e-09,
      "loss": 0.045784588903188705,
      "step": 7357
    },
    {
      "epoch": 0.9947528073747983,
      "grad_norm": 0.2936725616455078,
      "learning_rate": 2.498394676790361e-09,
      "loss": 0.039904750883579254,
      "step": 7358
    },
    {
      "epoch": 0.9948880007435635,
      "grad_norm": 0.34694939851760864,
      "learning_rate": 2.375039695071135e-09,
      "loss": 0.052568770945072174,
      "step": 7359
    },
    {
      "epoch": 0.9950231941123288,
      "grad_norm": 0.4866679310798645,
      "learning_rate": 2.2548072988454184e-09,
      "loss": 0.05424186587333679,
      "step": 7360
    },
    {
      "epoch": 0.9951583874810941,
      "grad_norm": 0.2569344639778137,
      "learning_rate": 2.1376975131465194e-09,
      "loss": 0.04274074733257294,
      "step": 7361
    },
    {
      "epoch": 0.9952935808498593,
      "grad_norm": 0.5606085062026978,
      "learning_rate": 2.023710362356601e-09,
      "loss": 0.047488436102867126,
      "step": 7362
    },
    {
      "epoch": 0.9954287742186245,
      "grad_norm": 1.2286579608917236,
      "learning_rate": 1.9128458702100117e-09,
      "loss": 0.05946382135152817,
      "step": 7363
    },
    {
      "epoch": 0.9955639675873899,
      "grad_norm": 0.8543102741241455,
      "learning_rate": 1.8051040597882873e-09,
      "loss": 0.06839191913604736,
      "step": 7364
    },
    {
      "epoch": 0.9956991609561551,
      "grad_norm": 0.40325114130973816,
      "learning_rate": 1.70048495352515e-09,
      "loss": 0.06515640020370483,
      "step": 7365
    },
    {
      "epoch": 0.9958343543249204,
      "grad_norm": 0.7575832009315491,
      "learning_rate": 1.5989885731998443e-09,
      "loss": 0.05808024853467941,
      "step": 7366
    },
    {
      "epoch": 0.9959695476936856,
      "grad_norm": 0.5866435766220093,
      "learning_rate": 1.5006149399487966e-09,
      "loss": 0.054545819759368896,
      "step": 7367
    },
    {
      "epoch": 0.9961047410624508,
      "grad_norm": 0.41578561067581177,
      "learning_rate": 1.4053640742489604e-09,
      "loss": 0.06123358756303787,
      "step": 7368
    },
    {
      "epoch": 0.9962399344312162,
      "grad_norm": 0.5700839757919312,
      "learning_rate": 1.3132359959361351e-09,
      "loss": 0.043250590562820435,
      "step": 7369
    },
    {
      "epoch": 0.9963751277999814,
      "grad_norm": 0.3385492265224457,
      "learning_rate": 1.2242307241899787e-09,
      "loss": 0.05139123648405075,
      "step": 7370
    },
    {
      "epoch": 0.9965103211687467,
      "grad_norm": 0.3921484053134918,
      "learning_rate": 1.1383482775406685e-09,
      "loss": 0.05334556847810745,
      "step": 7371
    },
    {
      "epoch": 0.9966455145375119,
      "grad_norm": 0.6032009124755859,
      "learning_rate": 1.0555886738738973e-09,
      "loss": 0.06666934490203857,
      "step": 7372
    },
    {
      "epoch": 0.9967807079062772,
      "grad_norm": 0.3122635781764984,
      "learning_rate": 9.75951930415886e-10,
      "loss": 0.0473453551530838,
      "step": 7373
    },
    {
      "epoch": 0.9969159012750425,
      "grad_norm": 0.7638356685638428,
      "learning_rate": 8.994380637483701e-10,
      "loss": 0.049236640334129333,
      "step": 7374
    },
    {
      "epoch": 0.9970510946438077,
      "grad_norm": 0.5076047778129578,
      "learning_rate": 8.260470898036054e-10,
      "loss": 0.06502559781074524,
      "step": 7375
    },
    {
      "epoch": 0.997186288012573,
      "grad_norm": 0.37870505452156067,
      "learning_rate": 7.557790238627016e-10,
      "loss": 0.05796322226524353,
      "step": 7376
    },
    {
      "epoch": 0.9973214813813382,
      "grad_norm": 1.0089482069015503,
      "learning_rate": 6.886338805522918e-10,
      "loss": 0.05884271115064621,
      "step": 7377
    },
    {
      "epoch": 0.9974566747501035,
      "grad_norm": 0.9313562512397766,
      "learning_rate": 6.246116738561903e-10,
      "loss": 0.05640513077378273,
      "step": 7378
    },
    {
      "epoch": 0.9975918681188688,
      "grad_norm": 0.3580704629421234,
      "learning_rate": 5.637124171004038e-10,
      "loss": 0.05452676862478256,
      "step": 7379
    },
    {
      "epoch": 0.997727061487634,
      "grad_norm": 0.5312066078186035,
      "learning_rate": 5.059361229681203e-10,
      "loss": 0.05440554767847061,
      "step": 7380
    },
    {
      "epoch": 0.9978622548563993,
      "grad_norm": 0.5021882653236389,
      "learning_rate": 4.5128280348638583e-10,
      "loss": 0.03760679066181183,
      "step": 7381
    },
    {
      "epoch": 0.9979974482251646,
      "grad_norm": 0.4162086844444275,
      "learning_rate": 3.9975247003443127e-10,
      "loss": 0.04282546043395996,
      "step": 7382
    },
    {
      "epoch": 0.9981326415939298,
      "grad_norm": 0.33936846256256104,
      "learning_rate": 3.51345133342007e-10,
      "loss": 0.041175272315740585,
      "step": 7383
    },
    {
      "epoch": 0.9982678349626951,
      "grad_norm": 0.3153817355632782,
      "learning_rate": 3.060608034877177e-10,
      "loss": 0.04674961790442467,
      "step": 7384
    },
    {
      "epoch": 0.9984030283314603,
      "grad_norm": 0.3183005452156067,
      "learning_rate": 2.638994898990221e-10,
      "loss": 0.053490906953811646,
      "step": 7385
    },
    {
      "epoch": 0.9985382217002257,
      "grad_norm": 0.5567049980163574,
      "learning_rate": 2.2486120135556398e-10,
      "loss": 0.060450583696365356,
      "step": 7386
    },
    {
      "epoch": 0.9986734150689909,
      "grad_norm": 0.5934447050094604,
      "learning_rate": 1.889459459841758e-10,
      "loss": 0.05848591774702072,
      "step": 7387
    },
    {
      "epoch": 0.9988086084377561,
      "grad_norm": 0.319583535194397,
      "learning_rate": 1.56153731263875e-10,
      "loss": 0.05173078551888466,
      "step": 7388
    },
    {
      "epoch": 0.9989438018065214,
      "grad_norm": 0.7217959761619568,
      "learning_rate": 1.2648456402086784e-10,
      "loss": 0.05946068465709686,
      "step": 7389
    },
    {
      "epoch": 0.9990789951752866,
      "grad_norm": 0.866469144821167,
      "learning_rate": 9.99384504318801e-11,
      "loss": 0.05586791783571243,
      "step": 7390
    },
    {
      "epoch": 0.9992141885440519,
      "grad_norm": 1.0748348236083984,
      "learning_rate": 7.651539602582247e-11,
      "loss": 0.05418063700199127,
      "step": 7391
    },
    {
      "epoch": 0.9993493819128172,
      "grad_norm": 0.4735819697380066,
      "learning_rate": 5.6215405678794464e-11,
      "loss": 0.056563325226306915,
      "step": 7392
    },
    {
      "epoch": 0.9994845752815824,
      "grad_norm": 0.43438920378685,
      "learning_rate": 3.9038483615749795e-11,
      "loss": 0.0597258135676384,
      "step": 7393
    },
    {
      "epoch": 0.9996197686503477,
      "grad_norm": 0.21488040685653687,
      "learning_rate": 2.4984633415492398e-11,
      "loss": 0.03498855233192444,
      "step": 7394
    },
    {
      "epoch": 0.999754962019113,
      "grad_norm": 0.5102145671844482,
      "learning_rate": 1.4053858004015041e-11,
      "loss": 0.05587794631719589,
      "step": 7395
    },
    {
      "epoch": 0.9998901553878782,
      "grad_norm": 0.25195667147636414,
      "learning_rate": 6.246159654499373e-12,
      "loss": 0.03818102926015854,
      "step": 7396
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4720384180545807,
      "learning_rate": 1.561539995642569e-12,
      "loss": 0.048416413366794586,
      "step": 7397
    },
    {
      "epoch": 1.0,
      "step": 7397,
      "total_flos": 9.247070602262269e+19,
      "train_loss": 0.06301417348261007,
      "train_runtime": 121181.5701,
      "train_samples_per_second": 15.626,
      "train_steps_per_second": 0.061
    }
  ],
  "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": 9.247070602262269e+19,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}