{
  "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": 0.0001351922264469793,
      "grad_norm": 113.21495056152344,
      "learning_rate": 0.0,
      "loss": 0.748291015625,
      "step": 1
    },
    {
      "epoch": 0.0002703844528939586,
      "grad_norm": 108.36476135253906,
      "learning_rate": 5.859375e-08,
      "loss": 0.74407958984375,
      "step": 2
    },
    {
      "epoch": 0.0004055766793409379,
      "grad_norm": 106.5140609741211,
      "learning_rate": 1.171875e-07,
      "loss": 0.74310302734375,
      "step": 3
    },
    {
      "epoch": 0.0005407689057879172,
      "grad_norm": 107.39302825927734,
      "learning_rate": 1.7578125e-07,
      "loss": 0.74542236328125,
      "step": 4
    },
    {
      "epoch": 0.0006759611322348965,
      "grad_norm": 103.53854370117188,
      "learning_rate": 2.34375e-07,
      "loss": 0.74053955078125,
      "step": 5
    },
    {
      "epoch": 0.0008111533586818758,
      "grad_norm": 113.02201080322266,
      "learning_rate": 2.9296875000000003e-07,
      "loss": 0.75213623046875,
      "step": 6
    },
    {
      "epoch": 0.0009463455851288551,
      "grad_norm": 107.06986999511719,
      "learning_rate": 3.515625e-07,
      "loss": 0.7452392578125,
      "step": 7
    },
    {
      "epoch": 0.0010815378115758344,
      "grad_norm": 107.04864501953125,
      "learning_rate": 4.1015625e-07,
      "loss": 0.741455078125,
      "step": 8
    },
    {
      "epoch": 0.0012167300380228137,
      "grad_norm": 107.50235748291016,
      "learning_rate": 4.6875e-07,
      "loss": 0.74114990234375,
      "step": 9
    },
    {
      "epoch": 0.001351922264469793,
      "grad_norm": 110.0961685180664,
      "learning_rate": 5.2734375e-07,
      "loss": 0.742431640625,
      "step": 10
    },
    {
      "epoch": 0.0014871144909167722,
      "grad_norm": 107.9870376586914,
      "learning_rate": 5.859375000000001e-07,
      "loss": 0.74407958984375,
      "step": 11
    },
    {
      "epoch": 0.0016223067173637515,
      "grad_norm": 106.07730102539062,
      "learning_rate": 6.4453125e-07,
      "loss": 0.73370361328125,
      "step": 12
    },
    {
      "epoch": 0.0017574989438107309,
      "grad_norm": 111.1282730102539,
      "learning_rate": 7.03125e-07,
      "loss": 0.735595703125,
      "step": 13
    },
    {
      "epoch": 0.0018926911702577102,
      "grad_norm": 108.10362243652344,
      "learning_rate": 7.6171875e-07,
      "loss": 0.7298583984375,
      "step": 14
    },
    {
      "epoch": 0.0020278833967046894,
      "grad_norm": 112.204833984375,
      "learning_rate": 8.203125e-07,
      "loss": 0.73345947265625,
      "step": 15
    },
    {
      "epoch": 0.0021630756231516687,
      "grad_norm": 110.4712905883789,
      "learning_rate": 8.7890625e-07,
      "loss": 0.714111328125,
      "step": 16
    },
    {
      "epoch": 0.002298267849598648,
      "grad_norm": 105.55673217773438,
      "learning_rate": 9.375e-07,
      "loss": 0.7093505859375,
      "step": 17
    },
    {
      "epoch": 0.0024334600760456274,
      "grad_norm": 92.90567779541016,
      "learning_rate": 9.9609375e-07,
      "loss": 0.68585205078125,
      "step": 18
    },
    {
      "epoch": 0.0025686523024926067,
      "grad_norm": 105.00279235839844,
      "learning_rate": 1.0546875e-06,
      "loss": 0.6875,
      "step": 19
    },
    {
      "epoch": 0.002703844528939586,
      "grad_norm": 96.80097198486328,
      "learning_rate": 1.11328125e-06,
      "loss": 0.68011474609375,
      "step": 20
    },
    {
      "epoch": 0.0028390367553865654,
      "grad_norm": 92.1880111694336,
      "learning_rate": 1.1718750000000001e-06,
      "loss": 0.6771240234375,
      "step": 21
    },
    {
      "epoch": 0.0029742289818335444,
      "grad_norm": 98.32208251953125,
      "learning_rate": 1.23046875e-06,
      "loss": 0.6190185546875,
      "step": 22
    },
    {
      "epoch": 0.0031094212082805237,
      "grad_norm": 94.18339538574219,
      "learning_rate": 1.2890625e-06,
      "loss": 0.611083984375,
      "step": 23
    },
    {
      "epoch": 0.003244613434727503,
      "grad_norm": 92.8238525390625,
      "learning_rate": 1.34765625e-06,
      "loss": 0.6107177734375,
      "step": 24
    },
    {
      "epoch": 0.0033798056611744824,
      "grad_norm": 84.07279205322266,
      "learning_rate": 1.40625e-06,
      "loss": 0.56585693359375,
      "step": 25
    },
    {
      "epoch": 0.0035149978876214617,
      "grad_norm": 82.3076400756836,
      "learning_rate": 1.46484375e-06,
      "loss": 0.55584716796875,
      "step": 26
    },
    {
      "epoch": 0.003650190114068441,
      "grad_norm": 81.03410339355469,
      "learning_rate": 1.5234375e-06,
      "loss": 0.55029296875,
      "step": 27
    },
    {
      "epoch": 0.0037853823405154204,
      "grad_norm": 84.79804229736328,
      "learning_rate": 1.5820312500000001e-06,
      "loss": 0.54022216796875,
      "step": 28
    },
    {
      "epoch": 0.003920574566962399,
      "grad_norm": 74.59902954101562,
      "learning_rate": 1.640625e-06,
      "loss": 0.54278564453125,
      "step": 29
    },
    {
      "epoch": 0.004055766793409379,
      "grad_norm": 58.30272674560547,
      "learning_rate": 1.69921875e-06,
      "loss": 0.419342041015625,
      "step": 30
    },
    {
      "epoch": 0.004190959019856358,
      "grad_norm": 62.59189224243164,
      "learning_rate": 1.7578125e-06,
      "loss": 0.3858642578125,
      "step": 31
    },
    {
      "epoch": 0.004326151246303337,
      "grad_norm": 58.085750579833984,
      "learning_rate": 1.81640625e-06,
      "loss": 0.390350341796875,
      "step": 32
    },
    {
      "epoch": 0.004461343472750317,
      "grad_norm": 57.22941970825195,
      "learning_rate": 1.875e-06,
      "loss": 0.3809814453125,
      "step": 33
    },
    {
      "epoch": 0.004596535699197296,
      "grad_norm": 50.55696487426758,
      "learning_rate": 1.93359375e-06,
      "loss": 0.371734619140625,
      "step": 34
    },
    {
      "epoch": 0.0047317279256442754,
      "grad_norm": 51.165592193603516,
      "learning_rate": 1.9921875e-06,
      "loss": 0.346527099609375,
      "step": 35
    },
    {
      "epoch": 0.004866920152091255,
      "grad_norm": 48.22037124633789,
      "learning_rate": 2.05078125e-06,
      "loss": 0.338653564453125,
      "step": 36
    },
    {
      "epoch": 0.005002112378538234,
      "grad_norm": 40.0262565612793,
      "learning_rate": 2.109375e-06,
      "loss": 0.355743408203125,
      "step": 37
    },
    {
      "epoch": 0.0051373046049852135,
      "grad_norm": 38.58692932128906,
      "learning_rate": 2.16796875e-06,
      "loss": 0.340789794921875,
      "step": 38
    },
    {
      "epoch": 0.005272496831432193,
      "grad_norm": 32.50050735473633,
      "learning_rate": 2.2265625e-06,
      "loss": 0.33416748046875,
      "step": 39
    },
    {
      "epoch": 0.005407689057879172,
      "grad_norm": 22.643756866455078,
      "learning_rate": 2.28515625e-06,
      "loss": 0.3324127197265625,
      "step": 40
    },
    {
      "epoch": 0.0055428812843261515,
      "grad_norm": 20.40488624572754,
      "learning_rate": 2.3437500000000002e-06,
      "loss": 0.2858123779296875,
      "step": 41
    },
    {
      "epoch": 0.005678073510773131,
      "grad_norm": 17.645462036132812,
      "learning_rate": 2.40234375e-06,
      "loss": 0.262664794921875,
      "step": 42
    },
    {
      "epoch": 0.00581326573722011,
      "grad_norm": 13.637259483337402,
      "learning_rate": 2.4609375e-06,
      "loss": 0.2672119140625,
      "step": 43
    },
    {
      "epoch": 0.005948457963667089,
      "grad_norm": 15.169034004211426,
      "learning_rate": 2.5195312500000003e-06,
      "loss": 0.23325347900390625,
      "step": 44
    },
    {
      "epoch": 0.006083650190114068,
      "grad_norm": 14.01146411895752,
      "learning_rate": 2.578125e-06,
      "loss": 0.2244873046875,
      "step": 45
    },
    {
      "epoch": 0.006218842416561047,
      "grad_norm": 9.481780052185059,
      "learning_rate": 2.63671875e-06,
      "loss": 0.24816131591796875,
      "step": 46
    },
    {
      "epoch": 0.006354034643008027,
      "grad_norm": 10.433836936950684,
      "learning_rate": 2.6953125e-06,
      "loss": 0.21952056884765625,
      "step": 47
    },
    {
      "epoch": 0.006489226869455006,
      "grad_norm": 8.479501724243164,
      "learning_rate": 2.75390625e-06,
      "loss": 0.21712493896484375,
      "step": 48
    },
    {
      "epoch": 0.0066244190959019855,
      "grad_norm": 2.557535171508789,
      "learning_rate": 2.8125e-06,
      "loss": 0.259521484375,
      "step": 49
    },
    {
      "epoch": 0.006759611322348965,
      "grad_norm": 2.974193572998047,
      "learning_rate": 2.87109375e-06,
      "loss": 0.2401885986328125,
      "step": 50
    },
    {
      "epoch": 0.006894803548795944,
      "grad_norm": 2.2221181392669678,
      "learning_rate": 2.9296875e-06,
      "loss": 0.26718902587890625,
      "step": 51
    },
    {
      "epoch": 0.0070299957752429235,
      "grad_norm": 1.9755792617797852,
      "learning_rate": 2.9882812500000002e-06,
      "loss": 0.23004150390625,
      "step": 52
    },
    {
      "epoch": 0.007165188001689903,
      "grad_norm": 2.6573526859283447,
      "learning_rate": 3.046875e-06,
      "loss": 0.21501922607421875,
      "step": 53
    },
    {
      "epoch": 0.007300380228136882,
      "grad_norm": 3.352895975112915,
      "learning_rate": 3.10546875e-06,
      "loss": 0.20934295654296875,
      "step": 54
    },
    {
      "epoch": 0.0074355724545838615,
      "grad_norm": 2.83560848236084,
      "learning_rate": 3.1640625000000003e-06,
      "loss": 0.22216796875,
      "step": 55
    },
    {
      "epoch": 0.007570764681030841,
      "grad_norm": 5.9283127784729,
      "learning_rate": 3.22265625e-06,
      "loss": 0.15530014038085938,
      "step": 56
    },
    {
      "epoch": 0.00770595690747782,
      "grad_norm": 3.6991169452667236,
      "learning_rate": 3.28125e-06,
      "loss": 0.1821136474609375,
      "step": 57
    },
    {
      "epoch": 0.007841149133924799,
      "grad_norm": 4.471269607543945,
      "learning_rate": 3.3398437500000003e-06,
      "loss": 0.232086181640625,
      "step": 58
    },
    {
      "epoch": 0.007976341360371779,
      "grad_norm": 5.5343217849731445,
      "learning_rate": 3.3984375e-06,
      "loss": 0.23752593994140625,
      "step": 59
    },
    {
      "epoch": 0.008111533586818757,
      "grad_norm": 3.6134774684906006,
      "learning_rate": 3.45703125e-06,
      "loss": 0.2616615295410156,
      "step": 60
    },
    {
      "epoch": 0.008246725813265738,
      "grad_norm": 5.2135910987854,
      "learning_rate": 3.515625e-06,
      "loss": 0.145904541015625,
      "step": 61
    },
    {
      "epoch": 0.008381918039712716,
      "grad_norm": 4.801625728607178,
      "learning_rate": 3.57421875e-06,
      "loss": 0.15020751953125,
      "step": 62
    },
    {
      "epoch": 0.008517110266159696,
      "grad_norm": 4.166983127593994,
      "learning_rate": 3.6328125e-06,
      "loss": 0.19959259033203125,
      "step": 63
    },
    {
      "epoch": 0.008652302492606675,
      "grad_norm": 4.064243793487549,
      "learning_rate": 3.69140625e-06,
      "loss": 0.13083839416503906,
      "step": 64
    },
    {
      "epoch": 0.008787494719053655,
      "grad_norm": 7.325190544128418,
      "learning_rate": 3.75e-06,
      "loss": 0.16757583618164062,
      "step": 65
    },
    {
      "epoch": 0.008922686945500634,
      "grad_norm": 9.485154151916504,
      "learning_rate": 3.8085937500000002e-06,
      "loss": 0.22097015380859375,
      "step": 66
    },
    {
      "epoch": 0.009057879171947614,
      "grad_norm": 9.843076705932617,
      "learning_rate": 3.8671875e-06,
      "loss": 0.13802337646484375,
      "step": 67
    },
    {
      "epoch": 0.009193071398394592,
      "grad_norm": 8.675374031066895,
      "learning_rate": 3.92578125e-06,
      "loss": 0.1808319091796875,
      "step": 68
    },
    {
      "epoch": 0.009328263624841572,
      "grad_norm": 3.483614921569824,
      "learning_rate": 3.984375e-06,
      "loss": 0.10939407348632812,
      "step": 69
    },
    {
      "epoch": 0.009463455851288551,
      "grad_norm": 3.5544822216033936,
      "learning_rate": 4.0429687500000004e-06,
      "loss": 0.13950729370117188,
      "step": 70
    },
    {
      "epoch": 0.00959864807773553,
      "grad_norm": 3.999521017074585,
      "learning_rate": 4.1015625e-06,
      "loss": 0.17206954956054688,
      "step": 71
    },
    {
      "epoch": 0.00973384030418251,
      "grad_norm": 9.9075927734375,
      "learning_rate": 4.16015625e-06,
      "loss": 0.22394180297851562,
      "step": 72
    },
    {
      "epoch": 0.009869032530629488,
      "grad_norm": 3.8207085132598877,
      "learning_rate": 4.21875e-06,
      "loss": 0.12192153930664062,
      "step": 73
    },
    {
      "epoch": 0.010004224757076468,
      "grad_norm": 8.090127944946289,
      "learning_rate": 4.27734375e-06,
      "loss": 0.21240997314453125,
      "step": 74
    },
    {
      "epoch": 0.010139416983523447,
      "grad_norm": 6.701347827911377,
      "learning_rate": 4.3359375e-06,
      "loss": 0.13440704345703125,
      "step": 75
    },
    {
      "epoch": 0.010274609209970427,
      "grad_norm": 2.4383575916290283,
      "learning_rate": 4.3945312500000005e-06,
      "loss": 0.13167572021484375,
      "step": 76
    },
    {
      "epoch": 0.010409801436417405,
      "grad_norm": 3.738039493560791,
      "learning_rate": 4.453125e-06,
      "loss": 0.19110107421875,
      "step": 77
    },
    {
      "epoch": 0.010544993662864386,
      "grad_norm": 7.909547328948975,
      "learning_rate": 4.51171875e-06,
      "loss": 0.13581275939941406,
      "step": 78
    },
    {
      "epoch": 0.010680185889311364,
      "grad_norm": 2.761085271835327,
      "learning_rate": 4.5703125e-06,
      "loss": 0.1324901580810547,
      "step": 79
    },
    {
      "epoch": 0.010815378115758344,
      "grad_norm": 11.46563720703125,
      "learning_rate": 4.62890625e-06,
      "loss": 0.1734781265258789,
      "step": 80
    },
    {
      "epoch": 0.010950570342205323,
      "grad_norm": 11.595213890075684,
      "learning_rate": 4.6875000000000004e-06,
      "loss": 0.22270774841308594,
      "step": 81
    },
    {
      "epoch": 0.011085762568652303,
      "grad_norm": 2.31451153755188,
      "learning_rate": 4.74609375e-06,
      "loss": 0.13615036010742188,
      "step": 82
    },
    {
      "epoch": 0.011220954795099282,
      "grad_norm": 6.179954528808594,
      "learning_rate": 4.8046875e-06,
      "loss": 0.23222732543945312,
      "step": 83
    },
    {
      "epoch": 0.011356147021546262,
      "grad_norm": 6.969549179077148,
      "learning_rate": 4.86328125e-06,
      "loss": 0.12405014038085938,
      "step": 84
    },
    {
      "epoch": 0.01149133924799324,
      "grad_norm": 4.883032321929932,
      "learning_rate": 4.921875e-06,
      "loss": 0.18102645874023438,
      "step": 85
    },
    {
      "epoch": 0.01162653147444022,
      "grad_norm": 3.645406484603882,
      "learning_rate": 4.98046875e-06,
      "loss": 0.12834930419921875,
      "step": 86
    },
    {
      "epoch": 0.011761723700887199,
      "grad_norm": 2.0353078842163086,
      "learning_rate": 5.0390625000000005e-06,
      "loss": 0.062042236328125,
      "step": 87
    },
    {
      "epoch": 0.011896915927334177,
      "grad_norm": 8.733626365661621,
      "learning_rate": 5.09765625e-06,
      "loss": 0.16191673278808594,
      "step": 88
    },
    {
      "epoch": 0.012032108153781158,
      "grad_norm": 7.236554145812988,
      "learning_rate": 5.15625e-06,
      "loss": 0.1753101348876953,
      "step": 89
    },
    {
      "epoch": 0.012167300380228136,
      "grad_norm": 5.710174083709717,
      "learning_rate": 5.21484375e-06,
      "loss": 0.18563079833984375,
      "step": 90
    },
    {
      "epoch": 0.012302492606675116,
      "grad_norm": 2.9105305671691895,
      "learning_rate": 5.2734375e-06,
      "loss": 0.1434478759765625,
      "step": 91
    },
    {
      "epoch": 0.012437684833122095,
      "grad_norm": 8.332087516784668,
      "learning_rate": 5.3320312500000004e-06,
      "loss": 0.10818862915039062,
      "step": 92
    },
    {
      "epoch": 0.012572877059569075,
      "grad_norm": 2.86965012550354,
      "learning_rate": 5.390625e-06,
      "loss": 0.16751861572265625,
      "step": 93
    },
    {
      "epoch": 0.012708069286016054,
      "grad_norm": 6.664602756500244,
      "learning_rate": 5.44921875e-06,
      "loss": 0.13586997985839844,
      "step": 94
    },
    {
      "epoch": 0.012843261512463034,
      "grad_norm": 3.1009819507598877,
      "learning_rate": 5.5078125e-06,
      "loss": 0.12674665451049805,
      "step": 95
    },
    {
      "epoch": 0.012978453738910012,
      "grad_norm": 8.342328071594238,
      "learning_rate": 5.56640625e-06,
      "loss": 0.14958953857421875,
      "step": 96
    },
    {
      "epoch": 0.013113645965356992,
      "grad_norm": 10.464272499084473,
      "learning_rate": 5.625e-06,
      "loss": 0.14819717407226562,
      "step": 97
    },
    {
      "epoch": 0.013248838191803971,
      "grad_norm": 6.0037736892700195,
      "learning_rate": 5.6835937500000005e-06,
      "loss": 0.10096168518066406,
      "step": 98
    },
    {
      "epoch": 0.013384030418250951,
      "grad_norm": 2.2893693447113037,
      "learning_rate": 5.7421875e-06,
      "loss": 0.13230514526367188,
      "step": 99
    },
    {
      "epoch": 0.01351922264469793,
      "grad_norm": 7.123652935028076,
      "learning_rate": 5.80078125e-06,
      "loss": 0.11539888381958008,
      "step": 100
    },
    {
      "epoch": 0.01365441487114491,
      "grad_norm": 10.961258888244629,
      "learning_rate": 5.859375e-06,
      "loss": 0.1784381866455078,
      "step": 101
    },
    {
      "epoch": 0.013789607097591888,
      "grad_norm": 4.8647990226745605,
      "learning_rate": 5.91796875e-06,
      "loss": 0.11674118041992188,
      "step": 102
    },
    {
      "epoch": 0.013924799324038869,
      "grad_norm": 7.338796615600586,
      "learning_rate": 5.9765625000000004e-06,
      "loss": 0.13860511779785156,
      "step": 103
    },
    {
      "epoch": 0.014059991550485847,
      "grad_norm": 8.011110305786133,
      "learning_rate": 6.03515625e-06,
      "loss": 0.16212081909179688,
      "step": 104
    },
    {
      "epoch": 0.014195183776932827,
      "grad_norm": 5.4134440422058105,
      "learning_rate": 6.09375e-06,
      "loss": 0.10045814514160156,
      "step": 105
    },
    {
      "epoch": 0.014330376003379806,
      "grad_norm": 2.763641834259033,
      "learning_rate": 6.15234375e-06,
      "loss": 0.1777801513671875,
      "step": 106
    },
    {
      "epoch": 0.014465568229826784,
      "grad_norm": 1.190101146697998,
      "learning_rate": 6.2109375e-06,
      "loss": 0.07335090637207031,
      "step": 107
    },
    {
      "epoch": 0.014600760456273764,
      "grad_norm": 8.644311904907227,
      "learning_rate": 6.26953125e-06,
      "loss": 0.1813502311706543,
      "step": 108
    },
    {
      "epoch": 0.014735952682720743,
      "grad_norm": 4.509277820587158,
      "learning_rate": 6.3281250000000005e-06,
      "loss": 0.1643209457397461,
      "step": 109
    },
    {
      "epoch": 0.014871144909167723,
      "grad_norm": 1.0009852647781372,
      "learning_rate": 6.38671875e-06,
      "loss": 0.08414649963378906,
      "step": 110
    },
    {
      "epoch": 0.015006337135614702,
      "grad_norm": 2.962554693222046,
      "learning_rate": 6.4453125e-06,
      "loss": 0.1264801025390625,
      "step": 111
    },
    {
      "epoch": 0.015141529362061682,
      "grad_norm": 2.0783417224884033,
      "learning_rate": 6.50390625e-06,
      "loss": 0.11137199401855469,
      "step": 112
    },
    {
      "epoch": 0.01527672158850866,
      "grad_norm": 4.764652729034424,
      "learning_rate": 6.5625e-06,
      "loss": 0.08538532257080078,
      "step": 113
    },
    {
      "epoch": 0.01541191381495564,
      "grad_norm": 2.847435474395752,
      "learning_rate": 6.6210937500000004e-06,
      "loss": 0.18856430053710938,
      "step": 114
    },
    {
      "epoch": 0.015547106041402619,
      "grad_norm": 2.094693660736084,
      "learning_rate": 6.679687500000001e-06,
      "loss": 0.1180562973022461,
      "step": 115
    },
    {
      "epoch": 0.015682298267849597,
      "grad_norm": 4.287889003753662,
      "learning_rate": 6.73828125e-06,
      "loss": 0.121612548828125,
      "step": 116
    },
    {
      "epoch": 0.01581749049429658,
      "grad_norm": 2.730125665664673,
      "learning_rate": 6.796875e-06,
      "loss": 0.10434532165527344,
      "step": 117
    },
    {
      "epoch": 0.015952682720743558,
      "grad_norm": 2.53700852394104,
      "learning_rate": 6.85546875e-06,
      "loss": 0.1632843017578125,
      "step": 118
    },
    {
      "epoch": 0.016087874947190536,
      "grad_norm": 8.600790977478027,
      "learning_rate": 6.9140625e-06,
      "loss": 0.12584304809570312,
      "step": 119
    },
    {
      "epoch": 0.016223067173637515,
      "grad_norm": 8.225876808166504,
      "learning_rate": 6.9726562500000005e-06,
      "loss": 0.16274023056030273,
      "step": 120
    },
    {
      "epoch": 0.016358259400084497,
      "grad_norm": 7.000145435333252,
      "learning_rate": 7.03125e-06,
      "loss": 0.1355915069580078,
      "step": 121
    },
    {
      "epoch": 0.016493451626531475,
      "grad_norm": 1.2663960456848145,
      "learning_rate": 7.08984375e-06,
      "loss": 0.10964584350585938,
      "step": 122
    },
    {
      "epoch": 0.016628643852978454,
      "grad_norm": 4.407687187194824,
      "learning_rate": 7.1484375e-06,
      "loss": 0.1642303466796875,
      "step": 123
    },
    {
      "epoch": 0.016763836079425432,
      "grad_norm": 4.05107307434082,
      "learning_rate": 7.20703125e-06,
      "loss": 0.1561412811279297,
      "step": 124
    },
    {
      "epoch": 0.01689902830587241,
      "grad_norm": 4.977163791656494,
      "learning_rate": 7.265625e-06,
      "loss": 0.12038421630859375,
      "step": 125
    },
    {
      "epoch": 0.017034220532319393,
      "grad_norm": 2.574216604232788,
      "learning_rate": 7.3242187500000006e-06,
      "loss": 0.18793296813964844,
      "step": 126
    },
    {
      "epoch": 0.01716941275876637,
      "grad_norm": 1.0781961679458618,
      "learning_rate": 7.3828125e-06,
      "loss": 0.12640953063964844,
      "step": 127
    },
    {
      "epoch": 0.01730460498521335,
      "grad_norm": 1.6773957014083862,
      "learning_rate": 7.44140625e-06,
      "loss": 0.0811920166015625,
      "step": 128
    },
    {
      "epoch": 0.017439797211660328,
      "grad_norm": 0.7164351344108582,
      "learning_rate": 7.5e-06,
      "loss": 0.06787109375,
      "step": 129
    },
    {
      "epoch": 0.01757498943810731,
      "grad_norm": 3.4869437217712402,
      "learning_rate": 7.55859375e-06,
      "loss": 0.11167430877685547,
      "step": 130
    },
    {
      "epoch": 0.01771018166455429,
      "grad_norm": 1.7625919580459595,
      "learning_rate": 7.6171875000000005e-06,
      "loss": 0.16495513916015625,
      "step": 131
    },
    {
      "epoch": 0.017845373891001267,
      "grad_norm": 0.7602922320365906,
      "learning_rate": 7.67578125e-06,
      "loss": 0.13075828552246094,
      "step": 132
    },
    {
      "epoch": 0.017980566117448245,
      "grad_norm": 2.4049386978149414,
      "learning_rate": 7.734375e-06,
      "loss": 0.15265655517578125,
      "step": 133
    },
    {
      "epoch": 0.018115758343895227,
      "grad_norm": 1.0546514987945557,
      "learning_rate": 7.792968750000001e-06,
      "loss": 0.14289093017578125,
      "step": 134
    },
    {
      "epoch": 0.018250950570342206,
      "grad_norm": 1.0986353158950806,
      "learning_rate": 7.8515625e-06,
      "loss": 0.14336681365966797,
      "step": 135
    },
    {
      "epoch": 0.018386142796789184,
      "grad_norm": 3.7656426429748535,
      "learning_rate": 7.91015625e-06,
      "loss": 0.1464376449584961,
      "step": 136
    },
    {
      "epoch": 0.018521335023236163,
      "grad_norm": 1.7089734077453613,
      "learning_rate": 7.96875e-06,
      "loss": 0.12820720672607422,
      "step": 137
    },
    {
      "epoch": 0.018656527249683145,
      "grad_norm": 1.987181305885315,
      "learning_rate": 8.02734375e-06,
      "loss": 0.11230659484863281,
      "step": 138
    },
    {
      "epoch": 0.018791719476130123,
      "grad_norm": 0.8318939208984375,
      "learning_rate": 8.085937500000001e-06,
      "loss": 0.09685516357421875,
      "step": 139
    },
    {
      "epoch": 0.018926911702577102,
      "grad_norm": 1.2632231712341309,
      "learning_rate": 8.14453125e-06,
      "loss": 0.09587240219116211,
      "step": 140
    },
    {
      "epoch": 0.01906210392902408,
      "grad_norm": 1.4492000341415405,
      "learning_rate": 8.203125e-06,
      "loss": 0.12637901306152344,
      "step": 141
    },
    {
      "epoch": 0.01919729615547106,
      "grad_norm": 3.6630232334136963,
      "learning_rate": 8.26171875e-06,
      "loss": 0.1568431854248047,
      "step": 142
    },
    {
      "epoch": 0.01933248838191804,
      "grad_norm": 2.887187957763672,
      "learning_rate": 8.3203125e-06,
      "loss": 0.12235832214355469,
      "step": 143
    },
    {
      "epoch": 0.01946768060836502,
      "grad_norm": 2.018618583679199,
      "learning_rate": 8.37890625e-06,
      "loss": 0.08384561538696289,
      "step": 144
    },
    {
      "epoch": 0.019602872834811998,
      "grad_norm": 2.8993093967437744,
      "learning_rate": 8.4375e-06,
      "loss": 0.13108062744140625,
      "step": 145
    },
    {
      "epoch": 0.019738065061258976,
      "grad_norm": 3.5131022930145264,
      "learning_rate": 8.49609375e-06,
      "loss": 0.17909622192382812,
      "step": 146
    },
    {
      "epoch": 0.019873257287705958,
      "grad_norm": 2.9853014945983887,
      "learning_rate": 8.5546875e-06,
      "loss": 0.08828163146972656,
      "step": 147
    },
    {
      "epoch": 0.020008449514152937,
      "grad_norm": 4.798372745513916,
      "learning_rate": 8.61328125e-06,
      "loss": 0.14518165588378906,
      "step": 148
    },
    {
      "epoch": 0.020143641740599915,
      "grad_norm": 4.102869033813477,
      "learning_rate": 8.671875e-06,
      "loss": 0.12208366394042969,
      "step": 149
    },
    {
      "epoch": 0.020278833967046894,
      "grad_norm": 1.452130675315857,
      "learning_rate": 8.73046875e-06,
      "loss": 0.09453487396240234,
      "step": 150
    },
    {
      "epoch": 0.020414026193493875,
      "grad_norm": 3.580275535583496,
      "learning_rate": 8.789062500000001e-06,
      "loss": 0.11167144775390625,
      "step": 151
    },
    {
      "epoch": 0.020549218419940854,
      "grad_norm": 2.5855484008789062,
      "learning_rate": 8.84765625e-06,
      "loss": 0.08226871490478516,
      "step": 152
    },
    {
      "epoch": 0.020684410646387832,
      "grad_norm": 3.079184055328369,
      "learning_rate": 8.90625e-06,
      "loss": 0.08919048309326172,
      "step": 153
    },
    {
      "epoch": 0.02081960287283481,
      "grad_norm": 1.7718374729156494,
      "learning_rate": 8.96484375e-06,
      "loss": 0.1257932186126709,
      "step": 154
    },
    {
      "epoch": 0.020954795099281793,
      "grad_norm": 1.7795010805130005,
      "learning_rate": 9.0234375e-06,
      "loss": 0.10280036926269531,
      "step": 155
    },
    {
      "epoch": 0.02108998732572877,
      "grad_norm": 6.808331489562988,
      "learning_rate": 9.082031250000001e-06,
      "loss": 0.15737152099609375,
      "step": 156
    },
    {
      "epoch": 0.02122517955217575,
      "grad_norm": 1.3567724227905273,
      "learning_rate": 9.140625e-06,
      "loss": 0.15021324157714844,
      "step": 157
    },
    {
      "epoch": 0.02136037177862273,
      "grad_norm": 1.3221532106399536,
      "learning_rate": 9.19921875e-06,
      "loss": 0.07552814483642578,
      "step": 158
    },
    {
      "epoch": 0.021495564005069707,
      "grad_norm": 1.1455117464065552,
      "learning_rate": 9.2578125e-06,
      "loss": 0.10880470275878906,
      "step": 159
    },
    {
      "epoch": 0.02163075623151669,
      "grad_norm": 1.018518090248108,
      "learning_rate": 9.31640625e-06,
      "loss": 0.06482505798339844,
      "step": 160
    },
    {
      "epoch": 0.021765948457963667,
      "grad_norm": 1.9412482976913452,
      "learning_rate": 9.375000000000001e-06,
      "loss": 0.06850290298461914,
      "step": 161
    },
    {
      "epoch": 0.021901140684410646,
      "grad_norm": 3.4968411922454834,
      "learning_rate": 9.43359375e-06,
      "loss": 0.12122619152069092,
      "step": 162
    },
    {
      "epoch": 0.022036332910857624,
      "grad_norm": 4.4286417961120605,
      "learning_rate": 9.4921875e-06,
      "loss": 0.1243133544921875,
      "step": 163
    },
    {
      "epoch": 0.022171525137304606,
      "grad_norm": 5.524990558624268,
      "learning_rate": 9.55078125e-06,
      "loss": 0.14089202880859375,
      "step": 164
    },
    {
      "epoch": 0.022306717363751585,
      "grad_norm": 3.6802821159362793,
      "learning_rate": 9.609375e-06,
      "loss": 0.08651351928710938,
      "step": 165
    },
    {
      "epoch": 0.022441909590198563,
      "grad_norm": 3.1190967559814453,
      "learning_rate": 9.66796875e-06,
      "loss": 0.08302116394042969,
      "step": 166
    },
    {
      "epoch": 0.02257710181664554,
      "grad_norm": 4.092504024505615,
      "learning_rate": 9.7265625e-06,
      "loss": 0.13624954223632812,
      "step": 167
    },
    {
      "epoch": 0.022712294043092524,
      "grad_norm": 2.8123655319213867,
      "learning_rate": 9.78515625e-06,
      "loss": 0.09840965270996094,
      "step": 168
    },
    {
      "epoch": 0.022847486269539502,
      "grad_norm": 2.038686513900757,
      "learning_rate": 9.84375e-06,
      "loss": 0.13901329040527344,
      "step": 169
    },
    {
      "epoch": 0.02298267849598648,
      "grad_norm": 1.851515293121338,
      "learning_rate": 9.90234375e-06,
      "loss": 0.1355876922607422,
      "step": 170
    },
    {
      "epoch": 0.02311787072243346,
      "grad_norm": 4.631291389465332,
      "learning_rate": 9.9609375e-06,
      "loss": 0.181976318359375,
      "step": 171
    },
    {
      "epoch": 0.02325306294888044,
      "grad_norm": 7.575974941253662,
      "learning_rate": 1.001953125e-05,
      "loss": 0.2035970687866211,
      "step": 172
    },
    {
      "epoch": 0.02338825517532742,
      "grad_norm": 9.929818153381348,
      "learning_rate": 1.0078125000000001e-05,
      "loss": 0.13532638549804688,
      "step": 173
    },
    {
      "epoch": 0.023523447401774398,
      "grad_norm": 1.531161904335022,
      "learning_rate": 1.013671875e-05,
      "loss": 0.11621952056884766,
      "step": 174
    },
    {
      "epoch": 0.023658639628221376,
      "grad_norm": 2.7926032543182373,
      "learning_rate": 1.01953125e-05,
      "loss": 0.08643817901611328,
      "step": 175
    },
    {
      "epoch": 0.023793831854668355,
      "grad_norm": 6.364810466766357,
      "learning_rate": 1.025390625e-05,
      "loss": 0.16801691055297852,
      "step": 176
    },
    {
      "epoch": 0.023929024081115337,
      "grad_norm": 4.558557033538818,
      "learning_rate": 1.03125e-05,
      "loss": 0.12172222137451172,
      "step": 177
    },
    {
      "epoch": 0.024064216307562315,
      "grad_norm": 13.089761734008789,
      "learning_rate": 1.0371093750000001e-05,
      "loss": 0.1399097442626953,
      "step": 178
    },
    {
      "epoch": 0.024199408534009294,
      "grad_norm": 12.028462409973145,
      "learning_rate": 1.04296875e-05,
      "loss": 0.11944293975830078,
      "step": 179
    },
    {
      "epoch": 0.024334600760456272,
      "grad_norm": 7.456510543823242,
      "learning_rate": 1.048828125e-05,
      "loss": 0.09870719909667969,
      "step": 180
    },
    {
      "epoch": 0.024469792986903254,
      "grad_norm": 1.4857101440429688,
      "learning_rate": 1.0546875e-05,
      "loss": 0.09162235260009766,
      "step": 181
    },
    {
      "epoch": 0.024604985213350233,
      "grad_norm": 4.810478687286377,
      "learning_rate": 1.060546875e-05,
      "loss": 0.1052408218383789,
      "step": 182
    },
    {
      "epoch": 0.02474017743979721,
      "grad_norm": 4.571746826171875,
      "learning_rate": 1.0664062500000001e-05,
      "loss": 0.12206459045410156,
      "step": 183
    },
    {
      "epoch": 0.02487536966624419,
      "grad_norm": 4.826739311218262,
      "learning_rate": 1.072265625e-05,
      "loss": 0.14027976989746094,
      "step": 184
    },
    {
      "epoch": 0.02501056189269117,
      "grad_norm": 1.68911874294281,
      "learning_rate": 1.078125e-05,
      "loss": 0.13982009887695312,
      "step": 185
    },
    {
      "epoch": 0.02514575411913815,
      "grad_norm": 5.710495948791504,
      "learning_rate": 1.083984375e-05,
      "loss": 0.11030769348144531,
      "step": 186
    },
    {
      "epoch": 0.02528094634558513,
      "grad_norm": 2.2882609367370605,
      "learning_rate": 1.08984375e-05,
      "loss": 0.09025359153747559,
      "step": 187
    },
    {
      "epoch": 0.025416138572032107,
      "grad_norm": 3.9504992961883545,
      "learning_rate": 1.095703125e-05,
      "loss": 0.13927841186523438,
      "step": 188
    },
    {
      "epoch": 0.02555133079847909,
      "grad_norm": 2.2836692333221436,
      "learning_rate": 1.1015625e-05,
      "loss": 0.1269536018371582,
      "step": 189
    },
    {
      "epoch": 0.025686523024926067,
      "grad_norm": 2.6040592193603516,
      "learning_rate": 1.1074218750000001e-05,
      "loss": 0.11598014831542969,
      "step": 190
    },
    {
      "epoch": 0.025821715251373046,
      "grad_norm": 5.9299092292785645,
      "learning_rate": 1.11328125e-05,
      "loss": 0.14548873901367188,
      "step": 191
    },
    {
      "epoch": 0.025956907477820024,
      "grad_norm": 2.255737066268921,
      "learning_rate": 1.119140625e-05,
      "loss": 0.11804676055908203,
      "step": 192
    },
    {
      "epoch": 0.026092099704267006,
      "grad_norm": 1.8716999292373657,
      "learning_rate": 1.125e-05,
      "loss": 0.06808090209960938,
      "step": 193
    },
    {
      "epoch": 0.026227291930713985,
      "grad_norm": 2.497042655944824,
      "learning_rate": 1.130859375e-05,
      "loss": 0.10706710815429688,
      "step": 194
    },
    {
      "epoch": 0.026362484157160963,
      "grad_norm": 3.482166051864624,
      "learning_rate": 1.1367187500000001e-05,
      "loss": 0.15346050262451172,
      "step": 195
    },
    {
      "epoch": 0.026497676383607942,
      "grad_norm": 4.400440216064453,
      "learning_rate": 1.142578125e-05,
      "loss": 0.10824441909790039,
      "step": 196
    },
    {
      "epoch": 0.02663286861005492,
      "grad_norm": 6.114097595214844,
      "learning_rate": 1.1484375e-05,
      "loss": 0.11834430694580078,
      "step": 197
    },
    {
      "epoch": 0.026768060836501902,
      "grad_norm": 2.279799699783325,
      "learning_rate": 1.154296875e-05,
      "loss": 0.12312889099121094,
      "step": 198
    },
    {
      "epoch": 0.02690325306294888,
      "grad_norm": 0.9420679211616516,
      "learning_rate": 1.16015625e-05,
      "loss": 0.06853199005126953,
      "step": 199
    },
    {
      "epoch": 0.02703844528939586,
      "grad_norm": 1.9888229370117188,
      "learning_rate": 1.1660156250000001e-05,
      "loss": 0.0968637466430664,
      "step": 200
    },
    {
      "epoch": 0.027173637515842838,
      "grad_norm": 1.231702208518982,
      "learning_rate": 1.171875e-05,
      "loss": 0.10464668273925781,
      "step": 201
    },
    {
      "epoch": 0.02730882974228982,
      "grad_norm": 1.6797163486480713,
      "learning_rate": 1.177734375e-05,
      "loss": 0.11948394775390625,
      "step": 202
    },
    {
      "epoch": 0.027444021968736798,
      "grad_norm": 2.0518569946289062,
      "learning_rate": 1.18359375e-05,
      "loss": 0.09188961982727051,
      "step": 203
    },
    {
      "epoch": 0.027579214195183777,
      "grad_norm": 2.415738821029663,
      "learning_rate": 1.189453125e-05,
      "loss": 0.09293174743652344,
      "step": 204
    },
    {
      "epoch": 0.027714406421630755,
      "grad_norm": 0.9479292035102844,
      "learning_rate": 1.1953125000000001e-05,
      "loss": 0.10342216491699219,
      "step": 205
    },
    {
      "epoch": 0.027849598648077737,
      "grad_norm": 0.7605688571929932,
      "learning_rate": 1.201171875e-05,
      "loss": 0.06887245178222656,
      "step": 206
    },
    {
      "epoch": 0.027984790874524716,
      "grad_norm": 1.1083003282546997,
      "learning_rate": 1.20703125e-05,
      "loss": 0.08589363098144531,
      "step": 207
    },
    {
      "epoch": 0.028119983100971694,
      "grad_norm": 3.79036283493042,
      "learning_rate": 1.212890625e-05,
      "loss": 0.19014549255371094,
      "step": 208
    },
    {
      "epoch": 0.028255175327418672,
      "grad_norm": 3.4152753353118896,
      "learning_rate": 1.21875e-05,
      "loss": 0.12618637084960938,
      "step": 209
    },
    {
      "epoch": 0.028390367553865654,
      "grad_norm": 2.4498088359832764,
      "learning_rate": 1.224609375e-05,
      "loss": 0.13806629180908203,
      "step": 210
    },
    {
      "epoch": 0.028525559780312633,
      "grad_norm": 5.081950664520264,
      "learning_rate": 1.23046875e-05,
      "loss": 0.14824390411376953,
      "step": 211
    },
    {
      "epoch": 0.02866075200675961,
      "grad_norm": 3.4185314178466797,
      "learning_rate": 1.2363281250000001e-05,
      "loss": 0.1659564971923828,
      "step": 212
    },
    {
      "epoch": 0.02879594423320659,
      "grad_norm": 5.166901588439941,
      "learning_rate": 1.2421875e-05,
      "loss": 0.08841896057128906,
      "step": 213
    },
    {
      "epoch": 0.02893113645965357,
      "grad_norm": 4.72791862487793,
      "learning_rate": 1.248046875e-05,
      "loss": 0.11452770233154297,
      "step": 214
    },
    {
      "epoch": 0.02906632868610055,
      "grad_norm": 5.573087692260742,
      "learning_rate": 1.25390625e-05,
      "loss": 0.08337831497192383,
      "step": 215
    },
    {
      "epoch": 0.02920152091254753,
      "grad_norm": 4.106476783752441,
      "learning_rate": 1.259765625e-05,
      "loss": 0.07535362243652344,
      "step": 216
    },
    {
      "epoch": 0.029336713138994507,
      "grad_norm": 2.1564137935638428,
      "learning_rate": 1.2656250000000001e-05,
      "loss": 0.08741140365600586,
      "step": 217
    },
    {
      "epoch": 0.029471905365441486,
      "grad_norm": 1.2396667003631592,
      "learning_rate": 1.271484375e-05,
      "loss": 0.1266641616821289,
      "step": 218
    },
    {
      "epoch": 0.029607097591888468,
      "grad_norm": 2.104222536087036,
      "learning_rate": 1.27734375e-05,
      "loss": 0.14081764221191406,
      "step": 219
    },
    {
      "epoch": 0.029742289818335446,
      "grad_norm": 2.055009365081787,
      "learning_rate": 1.283203125e-05,
      "loss": 0.1470928192138672,
      "step": 220
    },
    {
      "epoch": 0.029877482044782425,
      "grad_norm": 1.493439793586731,
      "learning_rate": 1.2890625e-05,
      "loss": 0.1262054443359375,
      "step": 221
    },
    {
      "epoch": 0.030012674271229403,
      "grad_norm": 0.9046214818954468,
      "learning_rate": 1.2949218750000001e-05,
      "loss": 0.0701284408569336,
      "step": 222
    },
    {
      "epoch": 0.030147866497676385,
      "grad_norm": 2.089538097381592,
      "learning_rate": 1.30078125e-05,
      "loss": 0.10072708129882812,
      "step": 223
    },
    {
      "epoch": 0.030283058724123364,
      "grad_norm": 3.5991060733795166,
      "learning_rate": 1.306640625e-05,
      "loss": 0.11789846420288086,
      "step": 224
    },
    {
      "epoch": 0.030418250950570342,
      "grad_norm": 3.6529898643493652,
      "learning_rate": 1.3125e-05,
      "loss": 0.10924339294433594,
      "step": 225
    },
    {
      "epoch": 0.03055344317701732,
      "grad_norm": 8.038374900817871,
      "learning_rate": 1.318359375e-05,
      "loss": 0.1462249755859375,
      "step": 226
    },
    {
      "epoch": 0.030688635403464302,
      "grad_norm": 5.628714084625244,
      "learning_rate": 1.3242187500000001e-05,
      "loss": 0.10403728485107422,
      "step": 227
    },
    {
      "epoch": 0.03082382762991128,
      "grad_norm": 2.750640630722046,
      "learning_rate": 1.330078125e-05,
      "loss": 0.13383102416992188,
      "step": 228
    },
    {
      "epoch": 0.03095901985635826,
      "grad_norm": 2.91054630279541,
      "learning_rate": 1.3359375000000001e-05,
      "loss": 0.11275577545166016,
      "step": 229
    },
    {
      "epoch": 0.031094212082805238,
      "grad_norm": 2.669840097427368,
      "learning_rate": 1.341796875e-05,
      "loss": 0.0904231071472168,
      "step": 230
    },
    {
      "epoch": 0.031229404309252216,
      "grad_norm": 2.7783708572387695,
      "learning_rate": 1.34765625e-05,
      "loss": 0.10542774200439453,
      "step": 231
    },
    {
      "epoch": 0.031364596535699195,
      "grad_norm": 1.4299713373184204,
      "learning_rate": 1.353515625e-05,
      "loss": 0.0928964614868164,
      "step": 232
    },
    {
      "epoch": 0.03149978876214617,
      "grad_norm": 2.016174554824829,
      "learning_rate": 1.359375e-05,
      "loss": 0.114776611328125,
      "step": 233
    },
    {
      "epoch": 0.03163498098859316,
      "grad_norm": 4.8079423904418945,
      "learning_rate": 1.3652343750000001e-05,
      "loss": 0.11942386627197266,
      "step": 234
    },
    {
      "epoch": 0.03177017321504014,
      "grad_norm": 1.7228970527648926,
      "learning_rate": 1.37109375e-05,
      "loss": 0.0829782485961914,
      "step": 235
    },
    {
      "epoch": 0.031905365441487116,
      "grad_norm": 0.42919689416885376,
      "learning_rate": 1.376953125e-05,
      "loss": 0.04460620880126953,
      "step": 236
    },
    {
      "epoch": 0.032040557667934094,
      "grad_norm": 3.3028643131256104,
      "learning_rate": 1.3828125e-05,
      "loss": 0.09788990020751953,
      "step": 237
    },
    {
      "epoch": 0.03217574989438107,
      "grad_norm": 2.778446912765503,
      "learning_rate": 1.388671875e-05,
      "loss": 0.10048770904541016,
      "step": 238
    },
    {
      "epoch": 0.03231094212082805,
      "grad_norm": 1.2364140748977661,
      "learning_rate": 1.3945312500000001e-05,
      "loss": 0.08328843116760254,
      "step": 239
    },
    {
      "epoch": 0.03244613434727503,
      "grad_norm": 1.6592155694961548,
      "learning_rate": 1.400390625e-05,
      "loss": 0.10396099090576172,
      "step": 240
    },
    {
      "epoch": 0.03258132657372201,
      "grad_norm": 0.6079725623130798,
      "learning_rate": 1.40625e-05,
      "loss": 0.060886383056640625,
      "step": 241
    },
    {
      "epoch": 0.032716518800168994,
      "grad_norm": 5.930914878845215,
      "learning_rate": 1.412109375e-05,
      "loss": 0.11059761047363281,
      "step": 242
    },
    {
      "epoch": 0.03285171102661597,
      "grad_norm": 4.0349626541137695,
      "learning_rate": 1.41796875e-05,
      "loss": 0.11272239685058594,
      "step": 243
    },
    {
      "epoch": 0.03298690325306295,
      "grad_norm": 1.2537180185317993,
      "learning_rate": 1.4238281250000001e-05,
      "loss": 0.10306453704833984,
      "step": 244
    },
    {
      "epoch": 0.03312209547950993,
      "grad_norm": 1.9081717729568481,
      "learning_rate": 1.4296875e-05,
      "loss": 0.13697052001953125,
      "step": 245
    },
    {
      "epoch": 0.03325728770595691,
      "grad_norm": 1.496750831604004,
      "learning_rate": 1.435546875e-05,
      "loss": 0.08673858642578125,
      "step": 246
    },
    {
      "epoch": 0.033392479932403886,
      "grad_norm": 2.706200361251831,
      "learning_rate": 1.44140625e-05,
      "loss": 0.1148829460144043,
      "step": 247
    },
    {
      "epoch": 0.033527672158850864,
      "grad_norm": 1.9627279043197632,
      "learning_rate": 1.447265625e-05,
      "loss": 0.08620643615722656,
      "step": 248
    },
    {
      "epoch": 0.03366286438529784,
      "grad_norm": 1.650659203529358,
      "learning_rate": 1.453125e-05,
      "loss": 0.12858200073242188,
      "step": 249
    },
    {
      "epoch": 0.03379805661174482,
      "grad_norm": 1.805548906326294,
      "learning_rate": 1.458984375e-05,
      "loss": 0.09543418884277344,
      "step": 250
    },
    {
      "epoch": 0.03393324883819181,
      "grad_norm": 2.7181215286254883,
      "learning_rate": 1.4648437500000001e-05,
      "loss": 0.07841825485229492,
      "step": 251
    },
    {
      "epoch": 0.034068441064638785,
      "grad_norm": 2.5783114433288574,
      "learning_rate": 1.470703125e-05,
      "loss": 0.12348270416259766,
      "step": 252
    },
    {
      "epoch": 0.034203633291085764,
      "grad_norm": 0.9260610938072205,
      "learning_rate": 1.4765625e-05,
      "loss": 0.06962871551513672,
      "step": 253
    },
    {
      "epoch": 0.03433882551753274,
      "grad_norm": 1.6076881885528564,
      "learning_rate": 1.482421875e-05,
      "loss": 0.11204147338867188,
      "step": 254
    },
    {
      "epoch": 0.03447401774397972,
      "grad_norm": 2.5134031772613525,
      "learning_rate": 1.48828125e-05,
      "loss": 0.11007165908813477,
      "step": 255
    },
    {
      "epoch": 0.0346092099704267,
      "grad_norm": 1.9173743724822998,
      "learning_rate": 1.4941406250000001e-05,
      "loss": 0.11146736145019531,
      "step": 256
    },
    {
      "epoch": 0.03474440219687368,
      "grad_norm": 1.3722951412200928,
      "learning_rate": 1.5e-05,
      "loss": 0.09109020233154297,
      "step": 257
    },
    {
      "epoch": 0.034879594423320656,
      "grad_norm": 5.6338629722595215,
      "learning_rate": 1.505859375e-05,
      "loss": 0.14918041229248047,
      "step": 258
    },
    {
      "epoch": 0.03501478664976764,
      "grad_norm": 3.58321213722229,
      "learning_rate": 1.51171875e-05,
      "loss": 0.1257801055908203,
      "step": 259
    },
    {
      "epoch": 0.03514997887621462,
      "grad_norm": 2.2501776218414307,
      "learning_rate": 1.517578125e-05,
      "loss": 0.1175689697265625,
      "step": 260
    },
    {
      "epoch": 0.0352851711026616,
      "grad_norm": 3.7236227989196777,
      "learning_rate": 1.5234375000000001e-05,
      "loss": 0.12977886199951172,
      "step": 261
    },
    {
      "epoch": 0.03542036332910858,
      "grad_norm": 2.549156427383423,
      "learning_rate": 1.529296875e-05,
      "loss": 0.10477447509765625,
      "step": 262
    },
    {
      "epoch": 0.035555555555555556,
      "grad_norm": 1.5814710855484009,
      "learning_rate": 1.53515625e-05,
      "loss": 0.07183074951171875,
      "step": 263
    },
    {
      "epoch": 0.035690747782002534,
      "grad_norm": 0.9356978535652161,
      "learning_rate": 1.541015625e-05,
      "loss": 0.08452892303466797,
      "step": 264
    },
    {
      "epoch": 0.03582594000844951,
      "grad_norm": 1.1948210000991821,
      "learning_rate": 1.546875e-05,
      "loss": 0.11107063293457031,
      "step": 265
    },
    {
      "epoch": 0.03596113223489649,
      "grad_norm": 0.8266691565513611,
      "learning_rate": 1.552734375e-05,
      "loss": 0.09385013580322266,
      "step": 266
    },
    {
      "epoch": 0.03609632446134347,
      "grad_norm": 5.6417646408081055,
      "learning_rate": 1.5585937500000002e-05,
      "loss": 0.11809539794921875,
      "step": 267
    },
    {
      "epoch": 0.036231516687790455,
      "grad_norm": 1.6528130769729614,
      "learning_rate": 1.564453125e-05,
      "loss": 0.05729341506958008,
      "step": 268
    },
    {
      "epoch": 0.03636670891423743,
      "grad_norm": 3.9556565284729004,
      "learning_rate": 1.5703125e-05,
      "loss": 0.09425687789916992,
      "step": 269
    },
    {
      "epoch": 0.03650190114068441,
      "grad_norm": 4.515077590942383,
      "learning_rate": 1.576171875e-05,
      "loss": 0.1512136459350586,
      "step": 270
    },
    {
      "epoch": 0.03663709336713139,
      "grad_norm": 1.4441699981689453,
      "learning_rate": 1.58203125e-05,
      "loss": 0.12055397033691406,
      "step": 271
    },
    {
      "epoch": 0.03677228559357837,
      "grad_norm": 1.241927981376648,
      "learning_rate": 1.587890625e-05,
      "loss": 0.1313619613647461,
      "step": 272
    },
    {
      "epoch": 0.03690747782002535,
      "grad_norm": 4.403429985046387,
      "learning_rate": 1.59375e-05,
      "loss": 0.16307449340820312,
      "step": 273
    },
    {
      "epoch": 0.037042670046472326,
      "grad_norm": 2.729262351989746,
      "learning_rate": 1.599609375e-05,
      "loss": 0.08340573310852051,
      "step": 274
    },
    {
      "epoch": 0.037177862272919304,
      "grad_norm": 1.3819665908813477,
      "learning_rate": 1.60546875e-05,
      "loss": 0.16694259643554688,
      "step": 275
    },
    {
      "epoch": 0.03731305449936629,
      "grad_norm": 1.756886601448059,
      "learning_rate": 1.611328125e-05,
      "loss": 0.11464977264404297,
      "step": 276
    },
    {
      "epoch": 0.03744824672581327,
      "grad_norm": 0.8777771592140198,
      "learning_rate": 1.6171875000000002e-05,
      "loss": 0.08952999114990234,
      "step": 277
    },
    {
      "epoch": 0.03758343895226025,
      "grad_norm": 2.6637778282165527,
      "learning_rate": 1.623046875e-05,
      "loss": 0.1071014404296875,
      "step": 278
    },
    {
      "epoch": 0.037718631178707225,
      "grad_norm": 0.8831402063369751,
      "learning_rate": 1.62890625e-05,
      "loss": 0.09461021423339844,
      "step": 279
    },
    {
      "epoch": 0.037853823405154204,
      "grad_norm": 2.507668972015381,
      "learning_rate": 1.634765625e-05,
      "loss": 0.12186670303344727,
      "step": 280
    },
    {
      "epoch": 0.03798901563160118,
      "grad_norm": 2.0636186599731445,
      "learning_rate": 1.640625e-05,
      "loss": 0.10665130615234375,
      "step": 281
    },
    {
      "epoch": 0.03812420785804816,
      "grad_norm": 3.099966287612915,
      "learning_rate": 1.646484375e-05,
      "loss": 0.11927509307861328,
      "step": 282
    },
    {
      "epoch": 0.03825940008449514,
      "grad_norm": 1.2251938581466675,
      "learning_rate": 1.65234375e-05,
      "loss": 0.10716438293457031,
      "step": 283
    },
    {
      "epoch": 0.03839459231094212,
      "grad_norm": 1.5043450593948364,
      "learning_rate": 1.6582031250000002e-05,
      "loss": 0.0943303108215332,
      "step": 284
    },
    {
      "epoch": 0.0385297845373891,
      "grad_norm": 1.2747001647949219,
      "learning_rate": 1.6640625e-05,
      "loss": 0.08355045318603516,
      "step": 285
    },
    {
      "epoch": 0.03866497676383608,
      "grad_norm": 2.7686452865600586,
      "learning_rate": 1.669921875e-05,
      "loss": 0.09137535095214844,
      "step": 286
    },
    {
      "epoch": 0.03880016899028306,
      "grad_norm": 1.7852996587753296,
      "learning_rate": 1.67578125e-05,
      "loss": 0.09190750122070312,
      "step": 287
    },
    {
      "epoch": 0.03893536121673004,
      "grad_norm": 1.1456475257873535,
      "learning_rate": 1.681640625e-05,
      "loss": 0.1054677963256836,
      "step": 288
    },
    {
      "epoch": 0.03907055344317702,
      "grad_norm": 0.7634332180023193,
      "learning_rate": 1.6875e-05,
      "loss": 0.06901717185974121,
      "step": 289
    },
    {
      "epoch": 0.039205745669623995,
      "grad_norm": 2.010342836380005,
      "learning_rate": 1.693359375e-05,
      "loss": 0.06561470031738281,
      "step": 290
    },
    {
      "epoch": 0.039340937896070974,
      "grad_norm": 3.22692608833313,
      "learning_rate": 1.69921875e-05,
      "loss": 0.07588434219360352,
      "step": 291
    },
    {
      "epoch": 0.03947613012251795,
      "grad_norm": 2.1595005989074707,
      "learning_rate": 1.705078125e-05,
      "loss": 0.14582252502441406,
      "step": 292
    },
    {
      "epoch": 0.03961132234896494,
      "grad_norm": 1.0591042041778564,
      "learning_rate": 1.7109375e-05,
      "loss": 0.09330081939697266,
      "step": 293
    },
    {
      "epoch": 0.039746514575411916,
      "grad_norm": 1.6329455375671387,
      "learning_rate": 1.7167968750000002e-05,
      "loss": 0.09622001647949219,
      "step": 294
    },
    {
      "epoch": 0.039881706801858895,
      "grad_norm": 1.439279556274414,
      "learning_rate": 1.72265625e-05,
      "loss": 0.12204265594482422,
      "step": 295
    },
    {
      "epoch": 0.04001689902830587,
      "grad_norm": 1.1472564935684204,
      "learning_rate": 1.728515625e-05,
      "loss": 0.048249244689941406,
      "step": 296
    },
    {
      "epoch": 0.04015209125475285,
      "grad_norm": 1.2087730169296265,
      "learning_rate": 1.734375e-05,
      "loss": 0.11009597778320312,
      "step": 297
    },
    {
      "epoch": 0.04028728348119983,
      "grad_norm": 1.881333827972412,
      "learning_rate": 1.740234375e-05,
      "loss": 0.1336355209350586,
      "step": 298
    },
    {
      "epoch": 0.04042247570764681,
      "grad_norm": 1.5699728727340698,
      "learning_rate": 1.74609375e-05,
      "loss": 0.11317634582519531,
      "step": 299
    },
    {
      "epoch": 0.04055766793409379,
      "grad_norm": 3.4574596881866455,
      "learning_rate": 1.751953125e-05,
      "loss": 0.09913825988769531,
      "step": 300
    },
    {
      "epoch": 0.040692860160540766,
      "grad_norm": 2.1715574264526367,
      "learning_rate": 1.7578125000000002e-05,
      "loss": 0.09168434143066406,
      "step": 301
    },
    {
      "epoch": 0.04082805238698775,
      "grad_norm": 0.5499640107154846,
      "learning_rate": 1.763671875e-05,
      "loss": 0.0647573471069336,
      "step": 302
    },
    {
      "epoch": 0.04096324461343473,
      "grad_norm": 1.5822196006774902,
      "learning_rate": 1.76953125e-05,
      "loss": 0.06601619720458984,
      "step": 303
    },
    {
      "epoch": 0.04109843683988171,
      "grad_norm": 0.46376851201057434,
      "learning_rate": 1.775390625e-05,
      "loss": 0.08853340148925781,
      "step": 304
    },
    {
      "epoch": 0.041233629066328686,
      "grad_norm": 1.0824228525161743,
      "learning_rate": 1.78125e-05,
      "loss": 0.1126103401184082,
      "step": 305
    },
    {
      "epoch": 0.041368821292775665,
      "grad_norm": 1.072993278503418,
      "learning_rate": 1.787109375e-05,
      "loss": 0.12059402465820312,
      "step": 306
    },
    {
      "epoch": 0.04150401351922264,
      "grad_norm": 1.2036006450653076,
      "learning_rate": 1.79296875e-05,
      "loss": 0.061367034912109375,
      "step": 307
    },
    {
      "epoch": 0.04163920574566962,
      "grad_norm": 0.7556990385055542,
      "learning_rate": 1.798828125e-05,
      "loss": 0.10070037841796875,
      "step": 308
    },
    {
      "epoch": 0.0417743979721166,
      "grad_norm": 1.9353309869766235,
      "learning_rate": 1.8046875e-05,
      "loss": 0.07542753219604492,
      "step": 309
    },
    {
      "epoch": 0.041909590198563586,
      "grad_norm": 1.5867691040039062,
      "learning_rate": 1.810546875e-05,
      "loss": 0.08946514129638672,
      "step": 310
    },
    {
      "epoch": 0.042044782425010564,
      "grad_norm": 1.0343576669692993,
      "learning_rate": 1.8164062500000002e-05,
      "loss": 0.15060901641845703,
      "step": 311
    },
    {
      "epoch": 0.04217997465145754,
      "grad_norm": 1.1521289348602295,
      "learning_rate": 1.822265625e-05,
      "loss": 0.10712528228759766,
      "step": 312
    },
    {
      "epoch": 0.04231516687790452,
      "grad_norm": 1.2042758464813232,
      "learning_rate": 1.828125e-05,
      "loss": 0.11998486518859863,
      "step": 313
    },
    {
      "epoch": 0.0424503591043515,
      "grad_norm": 1.1514405012130737,
      "learning_rate": 1.833984375e-05,
      "loss": 0.0639185905456543,
      "step": 314
    },
    {
      "epoch": 0.04258555133079848,
      "grad_norm": 0.6923871636390686,
      "learning_rate": 1.83984375e-05,
      "loss": 0.09173774719238281,
      "step": 315
    },
    {
      "epoch": 0.04272074355724546,
      "grad_norm": 0.998264491558075,
      "learning_rate": 1.845703125e-05,
      "loss": 0.12207412719726562,
      "step": 316
    },
    {
      "epoch": 0.042855935783692435,
      "grad_norm": 2.2990736961364746,
      "learning_rate": 1.8515625e-05,
      "loss": 0.07476043701171875,
      "step": 317
    },
    {
      "epoch": 0.042991128010139414,
      "grad_norm": 1.482649564743042,
      "learning_rate": 1.8574218750000002e-05,
      "loss": 0.108489990234375,
      "step": 318
    },
    {
      "epoch": 0.0431263202365864,
      "grad_norm": 2.9330074787139893,
      "learning_rate": 1.86328125e-05,
      "loss": 0.06729316711425781,
      "step": 319
    },
    {
      "epoch": 0.04326151246303338,
      "grad_norm": 3.1757407188415527,
      "learning_rate": 1.869140625e-05,
      "loss": 0.07111549377441406,
      "step": 320
    },
    {
      "epoch": 0.043396704689480356,
      "grad_norm": 1.5996159315109253,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 0.10639572143554688,
      "step": 321
    },
    {
      "epoch": 0.043531896915927334,
      "grad_norm": 1.336880087852478,
      "learning_rate": 1.880859375e-05,
      "loss": 0.07693672180175781,
      "step": 322
    },
    {
      "epoch": 0.04366708914237431,
      "grad_norm": 0.7539777755737305,
      "learning_rate": 1.88671875e-05,
      "loss": 0.07776546478271484,
      "step": 323
    },
    {
      "epoch": 0.04380228136882129,
      "grad_norm": 1.1080620288848877,
      "learning_rate": 1.892578125e-05,
      "loss": 0.09150028228759766,
      "step": 324
    },
    {
      "epoch": 0.04393747359526827,
      "grad_norm": 1.0383034944534302,
      "learning_rate": 1.8984375e-05,
      "loss": 0.08949756622314453,
      "step": 325
    },
    {
      "epoch": 0.04407266582171525,
      "grad_norm": 1.2968395948410034,
      "learning_rate": 1.904296875e-05,
      "loss": 0.0734872817993164,
      "step": 326
    },
    {
      "epoch": 0.044207858048162234,
      "grad_norm": 1.0070537328720093,
      "learning_rate": 1.91015625e-05,
      "loss": 0.06596803665161133,
      "step": 327
    },
    {
      "epoch": 0.04434305027460921,
      "grad_norm": 0.8868251442909241,
      "learning_rate": 1.9160156250000002e-05,
      "loss": 0.10872650146484375,
      "step": 328
    },
    {
      "epoch": 0.04447824250105619,
      "grad_norm": 1.2364088296890259,
      "learning_rate": 1.921875e-05,
      "loss": 0.08606529235839844,
      "step": 329
    },
    {
      "epoch": 0.04461343472750317,
      "grad_norm": 0.915959894657135,
      "learning_rate": 1.927734375e-05,
      "loss": 0.091278076171875,
      "step": 330
    },
    {
      "epoch": 0.04474862695395015,
      "grad_norm": 1.4230494499206543,
      "learning_rate": 1.93359375e-05,
      "loss": 0.10262727737426758,
      "step": 331
    },
    {
      "epoch": 0.044883819180397126,
      "grad_norm": 2.434359550476074,
      "learning_rate": 1.939453125e-05,
      "loss": 0.10689163208007812,
      "step": 332
    },
    {
      "epoch": 0.045019011406844105,
      "grad_norm": 0.5046936273574829,
      "learning_rate": 1.9453125e-05,
      "loss": 0.05563926696777344,
      "step": 333
    },
    {
      "epoch": 0.04515420363329108,
      "grad_norm": 1.49765944480896,
      "learning_rate": 1.951171875e-05,
      "loss": 0.11042022705078125,
      "step": 334
    },
    {
      "epoch": 0.04528939585973806,
      "grad_norm": 1.485113501548767,
      "learning_rate": 1.95703125e-05,
      "loss": 0.06636238098144531,
      "step": 335
    },
    {
      "epoch": 0.04542458808618505,
      "grad_norm": 3.4412622451782227,
      "learning_rate": 1.962890625e-05,
      "loss": 0.11137104034423828,
      "step": 336
    },
    {
      "epoch": 0.045559780312632026,
      "grad_norm": 1.7870159149169922,
      "learning_rate": 1.96875e-05,
      "loss": 0.06279373168945312,
      "step": 337
    },
    {
      "epoch": 0.045694972539079004,
      "grad_norm": 1.9660131931304932,
      "learning_rate": 1.9746093750000002e-05,
      "loss": 0.13500404357910156,
      "step": 338
    },
    {
      "epoch": 0.04583016476552598,
      "grad_norm": 1.4462882280349731,
      "learning_rate": 1.98046875e-05,
      "loss": 0.08622026443481445,
      "step": 339
    },
    {
      "epoch": 0.04596535699197296,
      "grad_norm": 0.8645960092544556,
      "learning_rate": 1.986328125e-05,
      "loss": 0.034657955169677734,
      "step": 340
    },
    {
      "epoch": 0.04610054921841994,
      "grad_norm": 9.027908325195312,
      "learning_rate": 1.9921875e-05,
      "loss": 0.1717844009399414,
      "step": 341
    },
    {
      "epoch": 0.04623574144486692,
      "grad_norm": 2.525975227355957,
      "learning_rate": 1.998046875e-05,
      "loss": 0.0699930191040039,
      "step": 342
    },
    {
      "epoch": 0.046370933671313896,
      "grad_norm": 1.6707751750946045,
      "learning_rate": 2.00390625e-05,
      "loss": 0.08284282684326172,
      "step": 343
    },
    {
      "epoch": 0.04650612589776088,
      "grad_norm": 0.4815549850463867,
      "learning_rate": 2.009765625e-05,
      "loss": 0.06775760650634766,
      "step": 344
    },
    {
      "epoch": 0.04664131812420786,
      "grad_norm": 1.3863520622253418,
      "learning_rate": 2.0156250000000002e-05,
      "loss": 0.07233905792236328,
      "step": 345
    },
    {
      "epoch": 0.04677651035065484,
      "grad_norm": 1.3405689001083374,
      "learning_rate": 2.021484375e-05,
      "loss": 0.06769132614135742,
      "step": 346
    },
    {
      "epoch": 0.04691170257710182,
      "grad_norm": 0.4060302972793579,
      "learning_rate": 2.02734375e-05,
      "loss": 0.057675838470458984,
      "step": 347
    },
    {
      "epoch": 0.047046894803548796,
      "grad_norm": 0.8686118721961975,
      "learning_rate": 2.033203125e-05,
      "loss": 0.12314987182617188,
      "step": 348
    },
    {
      "epoch": 0.047182087029995774,
      "grad_norm": 1.619500994682312,
      "learning_rate": 2.0390625e-05,
      "loss": 0.11477994918823242,
      "step": 349
    },
    {
      "epoch": 0.04731727925644275,
      "grad_norm": 0.8877691626548767,
      "learning_rate": 2.044921875e-05,
      "loss": 0.11084270477294922,
      "step": 350
    },
    {
      "epoch": 0.04745247148288973,
      "grad_norm": 1.370432734489441,
      "learning_rate": 2.05078125e-05,
      "loss": 0.08461380004882812,
      "step": 351
    },
    {
      "epoch": 0.04758766370933671,
      "grad_norm": 0.5697827935218811,
      "learning_rate": 2.056640625e-05,
      "loss": 0.07113170623779297,
      "step": 352
    },
    {
      "epoch": 0.047722855935783695,
      "grad_norm": 0.863343358039856,
      "learning_rate": 2.0625e-05,
      "loss": 0.06659460067749023,
      "step": 353
    },
    {
      "epoch": 0.047858048162230674,
      "grad_norm": 0.5409537553787231,
      "learning_rate": 2.068359375e-05,
      "loss": 0.06585693359375,
      "step": 354
    },
    {
      "epoch": 0.04799324038867765,
      "grad_norm": 1.2570327520370483,
      "learning_rate": 2.0742187500000002e-05,
      "loss": 0.06066417694091797,
      "step": 355
    },
    {
      "epoch": 0.04812843261512463,
      "grad_norm": 0.9887579083442688,
      "learning_rate": 2.080078125e-05,
      "loss": 0.07993316650390625,
      "step": 356
    },
    {
      "epoch": 0.04826362484157161,
      "grad_norm": 1.390712857246399,
      "learning_rate": 2.0859375e-05,
      "loss": 0.1002206802368164,
      "step": 357
    },
    {
      "epoch": 0.04839881706801859,
      "grad_norm": 2.349412441253662,
      "learning_rate": 2.091796875e-05,
      "loss": 0.08269977569580078,
      "step": 358
    },
    {
      "epoch": 0.048534009294465566,
      "grad_norm": 2.8065319061279297,
      "learning_rate": 2.09765625e-05,
      "loss": 0.1237192153930664,
      "step": 359
    },
    {
      "epoch": 0.048669201520912544,
      "grad_norm": 0.7555651068687439,
      "learning_rate": 2.103515625e-05,
      "loss": 0.06234538555145264,
      "step": 360
    },
    {
      "epoch": 0.04880439374735953,
      "grad_norm": 2.7890522480010986,
      "learning_rate": 2.109375e-05,
      "loss": 0.11326980590820312,
      "step": 361
    },
    {
      "epoch": 0.04893958597380651,
      "grad_norm": 2.958678722381592,
      "learning_rate": 2.1152343750000002e-05,
      "loss": 0.10531044006347656,
      "step": 362
    },
    {
      "epoch": 0.04907477820025349,
      "grad_norm": 0.7495773434638977,
      "learning_rate": 2.12109375e-05,
      "loss": 0.09909248352050781,
      "step": 363
    },
    {
      "epoch": 0.049209970426700465,
      "grad_norm": 1.170852541923523,
      "learning_rate": 2.126953125e-05,
      "loss": 0.118499755859375,
      "step": 364
    },
    {
      "epoch": 0.049345162653147444,
      "grad_norm": 1.6964222192764282,
      "learning_rate": 2.1328125000000002e-05,
      "loss": 0.09187841415405273,
      "step": 365
    },
    {
      "epoch": 0.04948035487959442,
      "grad_norm": 2.094663381576538,
      "learning_rate": 2.138671875e-05,
      "loss": 0.1099853515625,
      "step": 366
    },
    {
      "epoch": 0.0496155471060414,
      "grad_norm": 0.7523356080055237,
      "learning_rate": 2.14453125e-05,
      "loss": 0.11729049682617188,
      "step": 367
    },
    {
      "epoch": 0.04975073933248838,
      "grad_norm": 2.4578635692596436,
      "learning_rate": 2.150390625e-05,
      "loss": 0.10619497299194336,
      "step": 368
    },
    {
      "epoch": 0.049885931558935365,
      "grad_norm": 3.7947957515716553,
      "learning_rate": 2.15625e-05,
      "loss": 0.12723445892333984,
      "step": 369
    },
    {
      "epoch": 0.05002112378538234,
      "grad_norm": 1.2074638605117798,
      "learning_rate": 2.162109375e-05,
      "loss": 0.04551982879638672,
      "step": 370
    },
    {
      "epoch": 0.05015631601182932,
      "grad_norm": 0.4961264431476593,
      "learning_rate": 2.16796875e-05,
      "loss": 0.06826496124267578,
      "step": 371
    },
    {
      "epoch": 0.0502915082382763,
      "grad_norm": 0.8231317400932312,
      "learning_rate": 2.1738281250000002e-05,
      "loss": 0.07170963287353516,
      "step": 372
    },
    {
      "epoch": 0.05042670046472328,
      "grad_norm": 2.801870822906494,
      "learning_rate": 2.1796875e-05,
      "loss": 0.12439346313476562,
      "step": 373
    },
    {
      "epoch": 0.05056189269117026,
      "grad_norm": 1.2951346635818481,
      "learning_rate": 2.185546875e-05,
      "loss": 0.093963623046875,
      "step": 374
    },
    {
      "epoch": 0.050697084917617236,
      "grad_norm": 0.5000646710395813,
      "learning_rate": 2.19140625e-05,
      "loss": 0.08770179748535156,
      "step": 375
    },
    {
      "epoch": 0.050832277144064214,
      "grad_norm": 0.9898878931999207,
      "learning_rate": 2.197265625e-05,
      "loss": 0.11335372924804688,
      "step": 376
    },
    {
      "epoch": 0.05096746937051119,
      "grad_norm": 2.055985927581787,
      "learning_rate": 2.203125e-05,
      "loss": 0.1132669448852539,
      "step": 377
    },
    {
      "epoch": 0.05110266159695818,
      "grad_norm": 1.0124309062957764,
      "learning_rate": 2.208984375e-05,
      "loss": 0.11417198181152344,
      "step": 378
    },
    {
      "epoch": 0.051237853823405156,
      "grad_norm": 1.9621731042861938,
      "learning_rate": 2.2148437500000002e-05,
      "loss": 0.07587289810180664,
      "step": 379
    },
    {
      "epoch": 0.051373046049852135,
      "grad_norm": 3.2625293731689453,
      "learning_rate": 2.220703125e-05,
      "loss": 0.1237955093383789,
      "step": 380
    },
    {
      "epoch": 0.05150823827629911,
      "grad_norm": 1.7194373607635498,
      "learning_rate": 2.2265625e-05,
      "loss": 0.12086820602416992,
      "step": 381
    },
    {
      "epoch": 0.05164343050274609,
      "grad_norm": 1.6917966604232788,
      "learning_rate": 2.2324218750000002e-05,
      "loss": 0.08769416809082031,
      "step": 382
    },
    {
      "epoch": 0.05177862272919307,
      "grad_norm": 1.7035692930221558,
      "learning_rate": 2.23828125e-05,
      "loss": 0.13475990295410156,
      "step": 383
    },
    {
      "epoch": 0.05191381495564005,
      "grad_norm": 1.4051333665847778,
      "learning_rate": 2.244140625e-05,
      "loss": 0.06197834014892578,
      "step": 384
    },
    {
      "epoch": 0.05204900718208703,
      "grad_norm": 0.9884541034698486,
      "learning_rate": 2.25e-05,
      "loss": 0.09939861297607422,
      "step": 385
    },
    {
      "epoch": 0.05218419940853401,
      "grad_norm": 0.5881649851799011,
      "learning_rate": 2.255859375e-05,
      "loss": 0.06579351425170898,
      "step": 386
    },
    {
      "epoch": 0.05231939163498099,
      "grad_norm": 0.9547070264816284,
      "learning_rate": 2.26171875e-05,
      "loss": 0.07893943786621094,
      "step": 387
    },
    {
      "epoch": 0.05245458386142797,
      "grad_norm": 0.5487326383590698,
      "learning_rate": 2.267578125e-05,
      "loss": 0.08156633377075195,
      "step": 388
    },
    {
      "epoch": 0.05258977608787495,
      "grad_norm": 0.8872190713882446,
      "learning_rate": 2.2734375000000002e-05,
      "loss": 0.07341861724853516,
      "step": 389
    },
    {
      "epoch": 0.05272496831432193,
      "grad_norm": 1.4447367191314697,
      "learning_rate": 2.279296875e-05,
      "loss": 0.11373615264892578,
      "step": 390
    },
    {
      "epoch": 0.052860160540768905,
      "grad_norm": 1.5204302072525024,
      "learning_rate": 2.28515625e-05,
      "loss": 0.05935955047607422,
      "step": 391
    },
    {
      "epoch": 0.052995352767215884,
      "grad_norm": 1.0304832458496094,
      "learning_rate": 2.291015625e-05,
      "loss": 0.07056713104248047,
      "step": 392
    },
    {
      "epoch": 0.05313054499366286,
      "grad_norm": 1.544775128364563,
      "learning_rate": 2.296875e-05,
      "loss": 0.05887651443481445,
      "step": 393
    },
    {
      "epoch": 0.05326573722010984,
      "grad_norm": 0.6929005980491638,
      "learning_rate": 2.302734375e-05,
      "loss": 0.06090068817138672,
      "step": 394
    },
    {
      "epoch": 0.053400929446556826,
      "grad_norm": 0.5169737935066223,
      "learning_rate": 2.30859375e-05,
      "loss": 0.08879148960113525,
      "step": 395
    },
    {
      "epoch": 0.053536121673003804,
      "grad_norm": 0.952409565448761,
      "learning_rate": 2.3144531250000002e-05,
      "loss": 0.07410049438476562,
      "step": 396
    },
    {
      "epoch": 0.05367131389945078,
      "grad_norm": 0.9403892159461975,
      "learning_rate": 2.3203125e-05,
      "loss": 0.1253814697265625,
      "step": 397
    },
    {
      "epoch": 0.05380650612589776,
      "grad_norm": 0.6398392915725708,
      "learning_rate": 2.326171875e-05,
      "loss": 0.12230050563812256,
      "step": 398
    },
    {
      "epoch": 0.05394169835234474,
      "grad_norm": 2.4542503356933594,
      "learning_rate": 2.3320312500000002e-05,
      "loss": 0.08579373359680176,
      "step": 399
    },
    {
      "epoch": 0.05407689057879172,
      "grad_norm": 1.6023656129837036,
      "learning_rate": 2.337890625e-05,
      "loss": 0.14700698852539062,
      "step": 400
    },
    {
      "epoch": 0.0542120828052387,
      "grad_norm": 1.851426601409912,
      "learning_rate": 2.34375e-05,
      "loss": 0.09523773193359375,
      "step": 401
    },
    {
      "epoch": 0.054347275031685675,
      "grad_norm": 0.9876133799552917,
      "learning_rate": 2.349609375e-05,
      "loss": 0.09573745727539062,
      "step": 402
    },
    {
      "epoch": 0.05448246725813266,
      "grad_norm": 0.7136473655700684,
      "learning_rate": 2.35546875e-05,
      "loss": 0.07299518585205078,
      "step": 403
    },
    {
      "epoch": 0.05461765948457964,
      "grad_norm": 1.0516611337661743,
      "learning_rate": 2.361328125e-05,
      "loss": 0.09270524978637695,
      "step": 404
    },
    {
      "epoch": 0.05475285171102662,
      "grad_norm": 0.8308907747268677,
      "learning_rate": 2.3671875e-05,
      "loss": 0.05745649337768555,
      "step": 405
    },
    {
      "epoch": 0.054888043937473596,
      "grad_norm": 0.47025609016418457,
      "learning_rate": 2.3730468750000002e-05,
      "loss": 0.08206653594970703,
      "step": 406
    },
    {
      "epoch": 0.055023236163920575,
      "grad_norm": 0.8204769492149353,
      "learning_rate": 2.37890625e-05,
      "loss": 0.06442022323608398,
      "step": 407
    },
    {
      "epoch": 0.05515842839036755,
      "grad_norm": 0.8910301923751831,
      "learning_rate": 2.384765625e-05,
      "loss": 0.08228254318237305,
      "step": 408
    },
    {
      "epoch": 0.05529362061681453,
      "grad_norm": 1.1042304039001465,
      "learning_rate": 2.3906250000000002e-05,
      "loss": 0.10942554473876953,
      "step": 409
    },
    {
      "epoch": 0.05542881284326151,
      "grad_norm": 1.2192918062210083,
      "learning_rate": 2.396484375e-05,
      "loss": 0.08187246322631836,
      "step": 410
    },
    {
      "epoch": 0.05556400506970849,
      "grad_norm": 1.1344430446624756,
      "learning_rate": 2.40234375e-05,
      "loss": 0.1009821891784668,
      "step": 411
    },
    {
      "epoch": 0.055699197296155474,
      "grad_norm": 0.9855709671974182,
      "learning_rate": 2.408203125e-05,
      "loss": 0.09816837310791016,
      "step": 412
    },
    {
      "epoch": 0.05583438952260245,
      "grad_norm": 1.6689993143081665,
      "learning_rate": 2.4140625e-05,
      "loss": 0.09561729431152344,
      "step": 413
    },
    {
      "epoch": 0.05596958174904943,
      "grad_norm": 3.0833699703216553,
      "learning_rate": 2.419921875e-05,
      "loss": 0.10691452026367188,
      "step": 414
    },
    {
      "epoch": 0.05610477397549641,
      "grad_norm": 0.7255855202674866,
      "learning_rate": 2.42578125e-05,
      "loss": 0.04642772674560547,
      "step": 415
    },
    {
      "epoch": 0.05623996620194339,
      "grad_norm": 1.850696325302124,
      "learning_rate": 2.4316406250000002e-05,
      "loss": 0.1008908748626709,
      "step": 416
    },
    {
      "epoch": 0.056375158428390366,
      "grad_norm": 2.2940292358398438,
      "learning_rate": 2.4375e-05,
      "loss": 0.11898565292358398,
      "step": 417
    },
    {
      "epoch": 0.056510350654837345,
      "grad_norm": 0.6170001029968262,
      "learning_rate": 2.443359375e-05,
      "loss": 0.059403419494628906,
      "step": 418
    },
    {
      "epoch": 0.05664554288128432,
      "grad_norm": 4.38313102722168,
      "learning_rate": 2.44921875e-05,
      "loss": 0.1190958023071289,
      "step": 419
    },
    {
      "epoch": 0.05678073510773131,
      "grad_norm": 1.159766435623169,
      "learning_rate": 2.455078125e-05,
      "loss": 0.04065656661987305,
      "step": 420
    },
    {
      "epoch": 0.05691592733417829,
      "grad_norm": 2.5286099910736084,
      "learning_rate": 2.4609375e-05,
      "loss": 0.07746124267578125,
      "step": 421
    },
    {
      "epoch": 0.057051119560625266,
      "grad_norm": 1.0204311609268188,
      "learning_rate": 2.466796875e-05,
      "loss": 0.11473560333251953,
      "step": 422
    },
    {
      "epoch": 0.057186311787072244,
      "grad_norm": 2.9297640323638916,
      "learning_rate": 2.4726562500000002e-05,
      "loss": 0.10918998718261719,
      "step": 423
    },
    {
      "epoch": 0.05732150401351922,
      "grad_norm": 2.293121576309204,
      "learning_rate": 2.478515625e-05,
      "loss": 0.0850973129272461,
      "step": 424
    },
    {
      "epoch": 0.0574566962399662,
      "grad_norm": 2.8647897243499756,
      "learning_rate": 2.484375e-05,
      "loss": 0.07971739768981934,
      "step": 425
    },
    {
      "epoch": 0.05759188846641318,
      "grad_norm": 1.8048826456069946,
      "learning_rate": 2.4902343750000002e-05,
      "loss": 0.10360240936279297,
      "step": 426
    },
    {
      "epoch": 0.05772708069286016,
      "grad_norm": 1.325268268585205,
      "learning_rate": 2.49609375e-05,
      "loss": 0.0766916275024414,
      "step": 427
    },
    {
      "epoch": 0.05786227291930714,
      "grad_norm": 0.8526578545570374,
      "learning_rate": 2.501953125e-05,
      "loss": 0.10811042785644531,
      "step": 428
    },
    {
      "epoch": 0.05799746514575412,
      "grad_norm": 0.8991419076919556,
      "learning_rate": 2.5078125e-05,
      "loss": 0.06367301940917969,
      "step": 429
    },
    {
      "epoch": 0.0581326573722011,
      "grad_norm": 1.5593284368515015,
      "learning_rate": 2.513671875e-05,
      "loss": 0.09201622009277344,
      "step": 430
    },
    {
      "epoch": 0.05826784959864808,
      "grad_norm": 0.803339421749115,
      "learning_rate": 2.51953125e-05,
      "loss": 0.06955718994140625,
      "step": 431
    },
    {
      "epoch": 0.05840304182509506,
      "grad_norm": 1.2454171180725098,
      "learning_rate": 2.525390625e-05,
      "loss": 0.058101654052734375,
      "step": 432
    },
    {
      "epoch": 0.058538234051542036,
      "grad_norm": 1.0445276498794556,
      "learning_rate": 2.5312500000000002e-05,
      "loss": 0.09754657745361328,
      "step": 433
    },
    {
      "epoch": 0.058673426277989014,
      "grad_norm": 1.1726139783859253,
      "learning_rate": 2.537109375e-05,
      "loss": 0.0903177261352539,
      "step": 434
    },
    {
      "epoch": 0.05880861850443599,
      "grad_norm": 2.145634174346924,
      "learning_rate": 2.54296875e-05,
      "loss": 0.0947575569152832,
      "step": 435
    },
    {
      "epoch": 0.05894381073088297,
      "grad_norm": 2.7240688800811768,
      "learning_rate": 2.548828125e-05,
      "loss": 0.09523487091064453,
      "step": 436
    },
    {
      "epoch": 0.05907900295732996,
      "grad_norm": 1.097942590713501,
      "learning_rate": 2.5546875e-05,
      "loss": 0.06391668319702148,
      "step": 437
    },
    {
      "epoch": 0.059214195183776935,
      "grad_norm": 2.3833532333374023,
      "learning_rate": 2.560546875e-05,
      "loss": 0.10647773742675781,
      "step": 438
    },
    {
      "epoch": 0.059349387410223914,
      "grad_norm": 3.94865083694458,
      "learning_rate": 2.56640625e-05,
      "loss": 0.11064720153808594,
      "step": 439
    },
    {
      "epoch": 0.05948457963667089,
      "grad_norm": 2.3660669326782227,
      "learning_rate": 2.5722656250000002e-05,
      "loss": 0.10022509098052979,
      "step": 440
    },
    {
      "epoch": 0.05961977186311787,
      "grad_norm": 1.014296531677246,
      "learning_rate": 2.578125e-05,
      "loss": 0.06762075424194336,
      "step": 441
    },
    {
      "epoch": 0.05975496408956485,
      "grad_norm": 0.9329538941383362,
      "learning_rate": 2.583984375e-05,
      "loss": 0.07561779022216797,
      "step": 442
    },
    {
      "epoch": 0.05989015631601183,
      "grad_norm": 2.7418301105499268,
      "learning_rate": 2.5898437500000002e-05,
      "loss": 0.10692977905273438,
      "step": 443
    },
    {
      "epoch": 0.060025348542458806,
      "grad_norm": 1.9461236000061035,
      "learning_rate": 2.595703125e-05,
      "loss": 0.08919239044189453,
      "step": 444
    },
    {
      "epoch": 0.060160540768905785,
      "grad_norm": 0.5753958821296692,
      "learning_rate": 2.6015625e-05,
      "loss": 0.1025991439819336,
      "step": 445
    },
    {
      "epoch": 0.06029573299535277,
      "grad_norm": 0.7839550375938416,
      "learning_rate": 2.607421875e-05,
      "loss": 0.06124305725097656,
      "step": 446
    },
    {
      "epoch": 0.06043092522179975,
      "grad_norm": 1.8102830648422241,
      "learning_rate": 2.61328125e-05,
      "loss": 0.12053298950195312,
      "step": 447
    },
    {
      "epoch": 0.06056611744824673,
      "grad_norm": 1.0977753400802612,
      "learning_rate": 2.619140625e-05,
      "loss": 0.07920074462890625,
      "step": 448
    },
    {
      "epoch": 0.060701309674693706,
      "grad_norm": 0.3934282064437866,
      "learning_rate": 2.625e-05,
      "loss": 0.09472990036010742,
      "step": 449
    },
    {
      "epoch": 0.060836501901140684,
      "grad_norm": 3.4103548526763916,
      "learning_rate": 2.6308593750000002e-05,
      "loss": 0.08573007583618164,
      "step": 450
    },
    {
      "epoch": 0.06097169412758766,
      "grad_norm": 4.015721321105957,
      "learning_rate": 2.63671875e-05,
      "loss": 0.09718036651611328,
      "step": 451
    },
    {
      "epoch": 0.06110688635403464,
      "grad_norm": 0.9889941215515137,
      "learning_rate": 2.642578125e-05,
      "loss": 0.06290102005004883,
      "step": 452
    },
    {
      "epoch": 0.06124207858048162,
      "grad_norm": 2.233210563659668,
      "learning_rate": 2.6484375000000002e-05,
      "loss": 0.11217117309570312,
      "step": 453
    },
    {
      "epoch": 0.061377270806928605,
      "grad_norm": 0.9601654410362244,
      "learning_rate": 2.654296875e-05,
      "loss": 0.07436752319335938,
      "step": 454
    },
    {
      "epoch": 0.06151246303337558,
      "grad_norm": 1.0864660739898682,
      "learning_rate": 2.66015625e-05,
      "loss": 0.10283088684082031,
      "step": 455
    },
    {
      "epoch": 0.06164765525982256,
      "grad_norm": 0.8739877939224243,
      "learning_rate": 2.666015625e-05,
      "loss": 0.09707832336425781,
      "step": 456
    },
    {
      "epoch": 0.06178284748626954,
      "grad_norm": 1.828110694885254,
      "learning_rate": 2.6718750000000002e-05,
      "loss": 0.09204626083374023,
      "step": 457
    },
    {
      "epoch": 0.06191803971271652,
      "grad_norm": 1.8099445104599,
      "learning_rate": 2.677734375e-05,
      "loss": 0.084808349609375,
      "step": 458
    },
    {
      "epoch": 0.0620532319391635,
      "grad_norm": 1.5391427278518677,
      "learning_rate": 2.68359375e-05,
      "loss": 0.06592464447021484,
      "step": 459
    },
    {
      "epoch": 0.062188424165610476,
      "grad_norm": 1.0284428596496582,
      "learning_rate": 2.6894531250000002e-05,
      "loss": 0.0708761215209961,
      "step": 460
    },
    {
      "epoch": 0.062323616392057454,
      "grad_norm": 1.2352813482284546,
      "learning_rate": 2.6953125e-05,
      "loss": 0.10225677490234375,
      "step": 461
    },
    {
      "epoch": 0.06245880861850443,
      "grad_norm": 0.6165651679039001,
      "learning_rate": 2.701171875e-05,
      "loss": 0.06885337829589844,
      "step": 462
    },
    {
      "epoch": 0.06259400084495141,
      "grad_norm": 1.559402346611023,
      "learning_rate": 2.70703125e-05,
      "loss": 0.11022090911865234,
      "step": 463
    },
    {
      "epoch": 0.06272919307139839,
      "grad_norm": 0.6769663691520691,
      "learning_rate": 2.712890625e-05,
      "loss": 0.0871734619140625,
      "step": 464
    },
    {
      "epoch": 0.06286438529784537,
      "grad_norm": 1.1884297132492065,
      "learning_rate": 2.71875e-05,
      "loss": 0.12218093872070312,
      "step": 465
    },
    {
      "epoch": 0.06299957752429235,
      "grad_norm": 1.8552953004837036,
      "learning_rate": 2.724609375e-05,
      "loss": 0.09126853942871094,
      "step": 466
    },
    {
      "epoch": 0.06313476975073934,
      "grad_norm": 3.1213207244873047,
      "learning_rate": 2.7304687500000002e-05,
      "loss": 0.12496566772460938,
      "step": 467
    },
    {
      "epoch": 0.06326996197718632,
      "grad_norm": 0.7156252861022949,
      "learning_rate": 2.736328125e-05,
      "loss": 0.07934856414794922,
      "step": 468
    },
    {
      "epoch": 0.0634051542036333,
      "grad_norm": 0.7846072912216187,
      "learning_rate": 2.7421875e-05,
      "loss": 0.08756685256958008,
      "step": 469
    },
    {
      "epoch": 0.06354034643008027,
      "grad_norm": 2.3148536682128906,
      "learning_rate": 2.7480468750000002e-05,
      "loss": 0.07439613342285156,
      "step": 470
    },
    {
      "epoch": 0.06367553865652725,
      "grad_norm": 5.831628799438477,
      "learning_rate": 2.75390625e-05,
      "loss": 0.11340999603271484,
      "step": 471
    },
    {
      "epoch": 0.06381073088297423,
      "grad_norm": 1.986263632774353,
      "learning_rate": 2.759765625e-05,
      "loss": 0.06171226501464844,
      "step": 472
    },
    {
      "epoch": 0.06394592310942121,
      "grad_norm": 3.016879081726074,
      "learning_rate": 2.765625e-05,
      "loss": 0.08681917190551758,
      "step": 473
    },
    {
      "epoch": 0.06408111533586819,
      "grad_norm": 0.9517513513565063,
      "learning_rate": 2.7714843750000002e-05,
      "loss": 0.07973384857177734,
      "step": 474
    },
    {
      "epoch": 0.06421630756231517,
      "grad_norm": 1.4492366313934326,
      "learning_rate": 2.77734375e-05,
      "loss": 0.057143211364746094,
      "step": 475
    },
    {
      "epoch": 0.06435149978876215,
      "grad_norm": 2.5386431217193604,
      "learning_rate": 2.783203125e-05,
      "loss": 0.0702970027923584,
      "step": 476
    },
    {
      "epoch": 0.06448669201520912,
      "grad_norm": 2.8083841800689697,
      "learning_rate": 2.7890625000000002e-05,
      "loss": 0.09427070617675781,
      "step": 477
    },
    {
      "epoch": 0.0646218842416561,
      "grad_norm": 2.863194227218628,
      "learning_rate": 2.794921875e-05,
      "loss": 0.1324458122253418,
      "step": 478
    },
    {
      "epoch": 0.06475707646810308,
      "grad_norm": 1.1017186641693115,
      "learning_rate": 2.80078125e-05,
      "loss": 0.07815027236938477,
      "step": 479
    },
    {
      "epoch": 0.06489226869455006,
      "grad_norm": 1.798361897468567,
      "learning_rate": 2.806640625e-05,
      "loss": 0.09364700317382812,
      "step": 480
    },
    {
      "epoch": 0.06502746092099704,
      "grad_norm": 0.5833501815795898,
      "learning_rate": 2.8125e-05,
      "loss": 0.058249473571777344,
      "step": 481
    },
    {
      "epoch": 0.06516265314744402,
      "grad_norm": 2.1234350204467773,
      "learning_rate": 2.818359375e-05,
      "loss": 0.08389091491699219,
      "step": 482
    },
    {
      "epoch": 0.065297845373891,
      "grad_norm": 2.4928934574127197,
      "learning_rate": 2.82421875e-05,
      "loss": 0.13441944122314453,
      "step": 483
    },
    {
      "epoch": 0.06543303760033799,
      "grad_norm": 2.3466925621032715,
      "learning_rate": 2.8300781250000002e-05,
      "loss": 0.15199065208435059,
      "step": 484
    },
    {
      "epoch": 0.06556822982678497,
      "grad_norm": 1.0683932304382324,
      "learning_rate": 2.8359375e-05,
      "loss": 0.12136125564575195,
      "step": 485
    },
    {
      "epoch": 0.06570342205323194,
      "grad_norm": 1.038776159286499,
      "learning_rate": 2.841796875e-05,
      "loss": 0.07250213623046875,
      "step": 486
    },
    {
      "epoch": 0.06583861427967892,
      "grad_norm": 1.1282142400741577,
      "learning_rate": 2.8476562500000002e-05,
      "loss": 0.09813117980957031,
      "step": 487
    },
    {
      "epoch": 0.0659738065061259,
      "grad_norm": 2.1990718841552734,
      "learning_rate": 2.853515625e-05,
      "loss": 0.1062326431274414,
      "step": 488
    },
    {
      "epoch": 0.06610899873257288,
      "grad_norm": 1.2486003637313843,
      "learning_rate": 2.859375e-05,
      "loss": 0.08194351196289062,
      "step": 489
    },
    {
      "epoch": 0.06624419095901986,
      "grad_norm": 1.2207564115524292,
      "learning_rate": 2.865234375e-05,
      "loss": 0.08218860626220703,
      "step": 490
    },
    {
      "epoch": 0.06637938318546684,
      "grad_norm": 0.8832204937934875,
      "learning_rate": 2.87109375e-05,
      "loss": 0.1184835433959961,
      "step": 491
    },
    {
      "epoch": 0.06651457541191381,
      "grad_norm": 0.8516805171966553,
      "learning_rate": 2.876953125e-05,
      "loss": 0.06719732284545898,
      "step": 492
    },
    {
      "epoch": 0.0666497676383608,
      "grad_norm": 3.5853073596954346,
      "learning_rate": 2.8828125e-05,
      "loss": 0.09199237823486328,
      "step": 493
    },
    {
      "epoch": 0.06678495986480777,
      "grad_norm": 2.824306011199951,
      "learning_rate": 2.8886718750000002e-05,
      "loss": 0.07332611083984375,
      "step": 494
    },
    {
      "epoch": 0.06692015209125475,
      "grad_norm": 2.6137425899505615,
      "learning_rate": 2.89453125e-05,
      "loss": 0.08916282653808594,
      "step": 495
    },
    {
      "epoch": 0.06705534431770173,
      "grad_norm": 0.6165325045585632,
      "learning_rate": 2.900390625e-05,
      "loss": 0.0583348274230957,
      "step": 496
    },
    {
      "epoch": 0.06719053654414871,
      "grad_norm": 1.457873821258545,
      "learning_rate": 2.90625e-05,
      "loss": 0.1300978660583496,
      "step": 497
    },
    {
      "epoch": 0.06732572877059569,
      "grad_norm": 0.6833505630493164,
      "learning_rate": 2.912109375e-05,
      "loss": 0.05837535858154297,
      "step": 498
    },
    {
      "epoch": 0.06746092099704266,
      "grad_norm": 0.8221961855888367,
      "learning_rate": 2.91796875e-05,
      "loss": 0.07027196884155273,
      "step": 499
    },
    {
      "epoch": 0.06759611322348964,
      "grad_norm": 0.7736613154411316,
      "learning_rate": 2.923828125e-05,
      "loss": 0.08208274841308594,
      "step": 500
    },
    {
      "epoch": 0.06773130544993664,
      "grad_norm": 0.4670673608779907,
      "learning_rate": 2.9296875000000002e-05,
      "loss": 0.0636606216430664,
      "step": 501
    },
    {
      "epoch": 0.06786649767638361,
      "grad_norm": 1.6846320629119873,
      "learning_rate": 2.935546875e-05,
      "loss": 0.07015705108642578,
      "step": 502
    },
    {
      "epoch": 0.06800168990283059,
      "grad_norm": 1.551141381263733,
      "learning_rate": 2.94140625e-05,
      "loss": 0.06168341636657715,
      "step": 503
    },
    {
      "epoch": 0.06813688212927757,
      "grad_norm": 0.6758229732513428,
      "learning_rate": 2.9472656250000002e-05,
      "loss": 0.09338092803955078,
      "step": 504
    },
    {
      "epoch": 0.06827207435572455,
      "grad_norm": 0.5300410985946655,
      "learning_rate": 2.953125e-05,
      "loss": 0.07401514053344727,
      "step": 505
    },
    {
      "epoch": 0.06840726658217153,
      "grad_norm": 1.1444934606552124,
      "learning_rate": 2.958984375e-05,
      "loss": 0.0963125228881836,
      "step": 506
    },
    {
      "epoch": 0.0685424588086185,
      "grad_norm": 0.7125914096832275,
      "learning_rate": 2.96484375e-05,
      "loss": 0.051430702209472656,
      "step": 507
    },
    {
      "epoch": 0.06867765103506548,
      "grad_norm": 0.7984012961387634,
      "learning_rate": 2.970703125e-05,
      "loss": 0.06642436981201172,
      "step": 508
    },
    {
      "epoch": 0.06881284326151246,
      "grad_norm": 1.0330177545547485,
      "learning_rate": 2.9765625e-05,
      "loss": 0.08015632629394531,
      "step": 509
    },
    {
      "epoch": 0.06894803548795944,
      "grad_norm": 0.9897464513778687,
      "learning_rate": 2.982421875e-05,
      "loss": 0.05432319641113281,
      "step": 510
    },
    {
      "epoch": 0.06908322771440642,
      "grad_norm": 1.3147339820861816,
      "learning_rate": 2.9882812500000002e-05,
      "loss": 0.09183740615844727,
      "step": 511
    },
    {
      "epoch": 0.0692184199408534,
      "grad_norm": 2.923064708709717,
      "learning_rate": 2.994140625e-05,
      "loss": 0.0987863540649414,
      "step": 512
    },
    {
      "epoch": 0.06935361216730038,
      "grad_norm": 2.062232732772827,
      "learning_rate": 3e-05,
      "loss": 0.06314706802368164,
      "step": 513
    },
    {
      "epoch": 0.06948880439374736,
      "grad_norm": 3.0970215797424316,
      "learning_rate": 2.9999998438460004e-05,
      "loss": 0.10694122314453125,
      "step": 514
    },
    {
      "epoch": 0.06962399662019433,
      "grad_norm": 0.9646059274673462,
      "learning_rate": 2.9999993753840344e-05,
      "loss": 0.04927396774291992,
      "step": 515
    },
    {
      "epoch": 0.06975918884664131,
      "grad_norm": 1.4721053838729858,
      "learning_rate": 2.9999985946141995e-05,
      "loss": 0.09446072578430176,
      "step": 516
    },
    {
      "epoch": 0.06989438107308829,
      "grad_norm": 1.6983895301818848,
      "learning_rate": 2.9999975015366586e-05,
      "loss": 0.07980012893676758,
      "step": 517
    },
    {
      "epoch": 0.07002957329953528,
      "grad_norm": 0.6937089562416077,
      "learning_rate": 2.9999960961516384e-05,
      "loss": 0.09464740753173828,
      "step": 518
    },
    {
      "epoch": 0.07016476552598226,
      "grad_norm": 1.892868995666504,
      "learning_rate": 2.9999943784594325e-05,
      "loss": 0.07539939880371094,
      "step": 519
    },
    {
      "epoch": 0.07029995775242924,
      "grad_norm": 3.4821112155914307,
      "learning_rate": 2.9999923484603975e-05,
      "loss": 0.08524513244628906,
      "step": 520
    },
    {
      "epoch": 0.07043514997887622,
      "grad_norm": 1.4239543676376343,
      "learning_rate": 2.999990006154957e-05,
      "loss": 0.09918570518493652,
      "step": 521
    },
    {
      "epoch": 0.0705703422053232,
      "grad_norm": 1.6605324745178223,
      "learning_rate": 2.9999873515435977e-05,
      "loss": 0.08107471466064453,
      "step": 522
    },
    {
      "epoch": 0.07070553443177018,
      "grad_norm": 3.454179286956787,
      "learning_rate": 2.9999843846268735e-05,
      "loss": 0.11911773681640625,
      "step": 523
    },
    {
      "epoch": 0.07084072665821715,
      "grad_norm": 3.45165753364563,
      "learning_rate": 2.9999811054054018e-05,
      "loss": 0.11735010147094727,
      "step": 524
    },
    {
      "epoch": 0.07097591888466413,
      "grad_norm": 4.04960298538208,
      "learning_rate": 2.9999775138798646e-05,
      "loss": 0.13319826126098633,
      "step": 525
    },
    {
      "epoch": 0.07111111111111111,
      "grad_norm": 1.6464530229568481,
      "learning_rate": 2.99997361005101e-05,
      "loss": 0.0993800163269043,
      "step": 526
    },
    {
      "epoch": 0.07124630333755809,
      "grad_norm": 1.0972402095794678,
      "learning_rate": 2.9999693939196513e-05,
      "loss": 0.09639739990234375,
      "step": 527
    },
    {
      "epoch": 0.07138149556400507,
      "grad_norm": 0.29994040727615356,
      "learning_rate": 2.999964865486666e-05,
      "loss": 0.04704856872558594,
      "step": 528
    },
    {
      "epoch": 0.07151668779045205,
      "grad_norm": 1.53493332862854,
      "learning_rate": 2.999960024752997e-05,
      "loss": 0.12227725982666016,
      "step": 529
    },
    {
      "epoch": 0.07165188001689902,
      "grad_norm": 1.67615807056427,
      "learning_rate": 2.9999548717196516e-05,
      "loss": 0.10790061950683594,
      "step": 530
    },
    {
      "epoch": 0.071787072243346,
      "grad_norm": 0.4506801962852478,
      "learning_rate": 2.999949406387703e-05,
      "loss": 0.06563091278076172,
      "step": 531
    },
    {
      "epoch": 0.07192226446979298,
      "grad_norm": 0.9371463060379028,
      "learning_rate": 2.9999436287582903e-05,
      "loss": 0.10554313659667969,
      "step": 532
    },
    {
      "epoch": 0.07205745669623996,
      "grad_norm": 2.043168783187866,
      "learning_rate": 2.9999375388326145e-05,
      "loss": 0.09876775741577148,
      "step": 533
    },
    {
      "epoch": 0.07219264892268694,
      "grad_norm": 1.5399434566497803,
      "learning_rate": 2.9999311366119447e-05,
      "loss": 0.057539939880371094,
      "step": 534
    },
    {
      "epoch": 0.07232784114913393,
      "grad_norm": 2.689091920852661,
      "learning_rate": 2.9999244220976137e-05,
      "loss": 0.11822700500488281,
      "step": 535
    },
    {
      "epoch": 0.07246303337558091,
      "grad_norm": 2.3212358951568604,
      "learning_rate": 2.9999173952910197e-05,
      "loss": 0.12038135528564453,
      "step": 536
    },
    {
      "epoch": 0.07259822560202789,
      "grad_norm": 1.0485259294509888,
      "learning_rate": 2.9999100561936252e-05,
      "loss": 0.10058116912841797,
      "step": 537
    },
    {
      "epoch": 0.07273341782847487,
      "grad_norm": 1.9497294425964355,
      "learning_rate": 2.9999024048069585e-05,
      "loss": 0.05536365509033203,
      "step": 538
    },
    {
      "epoch": 0.07286861005492185,
      "grad_norm": 2.2156689167022705,
      "learning_rate": 2.9998944411326127e-05,
      "loss": 0.07976913452148438,
      "step": 539
    },
    {
      "epoch": 0.07300380228136882,
      "grad_norm": 1.8322675228118896,
      "learning_rate": 2.999886165172246e-05,
      "loss": 0.07038617134094238,
      "step": 540
    },
    {
      "epoch": 0.0731389945078158,
      "grad_norm": 2.59919810295105,
      "learning_rate": 2.9998775769275814e-05,
      "loss": 0.1098175048828125,
      "step": 541
    },
    {
      "epoch": 0.07327418673426278,
      "grad_norm": 0.9037202000617981,
      "learning_rate": 2.9998686764004067e-05,
      "loss": 0.10444879531860352,
      "step": 542
    },
    {
      "epoch": 0.07340937896070976,
      "grad_norm": 1.077059030532837,
      "learning_rate": 2.9998594635925755e-05,
      "loss": 0.07860016822814941,
      "step": 543
    },
    {
      "epoch": 0.07354457118715674,
      "grad_norm": 1.3018779754638672,
      "learning_rate": 2.999849938506005e-05,
      "loss": 0.0791769027709961,
      "step": 544
    },
    {
      "epoch": 0.07367976341360372,
      "grad_norm": 0.6867576241493225,
      "learning_rate": 2.99984010114268e-05,
      "loss": 0.07667970657348633,
      "step": 545
    },
    {
      "epoch": 0.0738149556400507,
      "grad_norm": 0.7375957369804382,
      "learning_rate": 2.9998299515046475e-05,
      "loss": 0.07518553733825684,
      "step": 546
    },
    {
      "epoch": 0.07395014786649767,
      "grad_norm": 0.5696337223052979,
      "learning_rate": 2.9998194895940213e-05,
      "loss": 0.09038829803466797,
      "step": 547
    },
    {
      "epoch": 0.07408534009294465,
      "grad_norm": 0.7057362794876099,
      "learning_rate": 2.9998087154129792e-05,
      "loss": 0.06936836242675781,
      "step": 548
    },
    {
      "epoch": 0.07422053231939163,
      "grad_norm": 2.4675979614257812,
      "learning_rate": 2.9997976289637645e-05,
      "loss": 0.08766460418701172,
      "step": 549
    },
    {
      "epoch": 0.07435572454583861,
      "grad_norm": 0.5624940395355225,
      "learning_rate": 2.9997862302486855e-05,
      "loss": 0.06978332996368408,
      "step": 550
    },
    {
      "epoch": 0.07449091677228559,
      "grad_norm": 0.6041243672370911,
      "learning_rate": 2.9997745192701153e-05,
      "loss": 0.07013988494873047,
      "step": 551
    },
    {
      "epoch": 0.07462610899873258,
      "grad_norm": 0.5061952471733093,
      "learning_rate": 2.9997624960304926e-05,
      "loss": 0.061765193939208984,
      "step": 552
    },
    {
      "epoch": 0.07476130122517956,
      "grad_norm": 0.9293647408485413,
      "learning_rate": 2.9997501605323214e-05,
      "loss": 0.052285194396972656,
      "step": 553
    },
    {
      "epoch": 0.07489649345162654,
      "grad_norm": 2.0838630199432373,
      "learning_rate": 2.999737512778168e-05,
      "loss": 0.08160591125488281,
      "step": 554
    },
    {
      "epoch": 0.07503168567807351,
      "grad_norm": 0.5081800818443298,
      "learning_rate": 2.9997245527706674e-05,
      "loss": 0.07192039489746094,
      "step": 555
    },
    {
      "epoch": 0.0751668779045205,
      "grad_norm": 0.5362699031829834,
      "learning_rate": 2.999711280512517e-05,
      "loss": 0.03781700134277344,
      "step": 556
    },
    {
      "epoch": 0.07530207013096747,
      "grad_norm": 0.8696369528770447,
      "learning_rate": 2.9996976960064807e-05,
      "loss": 0.11155176162719727,
      "step": 557
    },
    {
      "epoch": 0.07543726235741445,
      "grad_norm": 0.7809555530548096,
      "learning_rate": 2.999683799255387e-05,
      "loss": 0.09868621826171875,
      "step": 558
    },
    {
      "epoch": 0.07557245458386143,
      "grad_norm": 0.7617474794387817,
      "learning_rate": 2.999669590262129e-05,
      "loss": 0.06615638732910156,
      "step": 559
    },
    {
      "epoch": 0.07570764681030841,
      "grad_norm": 1.0787569284439087,
      "learning_rate": 2.999655069029665e-05,
      "loss": 0.0860823392868042,
      "step": 560
    },
    {
      "epoch": 0.07584283903675539,
      "grad_norm": 0.6820898056030273,
      "learning_rate": 2.9996402355610183e-05,
      "loss": 0.068115234375,
      "step": 561
    },
    {
      "epoch": 0.07597803126320236,
      "grad_norm": 0.7910268306732178,
      "learning_rate": 2.9996250898592777e-05,
      "loss": 0.15956687927246094,
      "step": 562
    },
    {
      "epoch": 0.07611322348964934,
      "grad_norm": 2.779169797897339,
      "learning_rate": 2.9996096319275962e-05,
      "loss": 0.10142803192138672,
      "step": 563
    },
    {
      "epoch": 0.07624841571609632,
      "grad_norm": 2.24088454246521,
      "learning_rate": 2.9995938617691925e-05,
      "loss": 0.10528564453125,
      "step": 564
    },
    {
      "epoch": 0.0763836079425433,
      "grad_norm": 1.1034523248672485,
      "learning_rate": 2.9995777793873504e-05,
      "loss": 0.062438011169433594,
      "step": 565
    },
    {
      "epoch": 0.07651880016899028,
      "grad_norm": 0.7434017658233643,
      "learning_rate": 2.9995613847854176e-05,
      "loss": 0.08405876159667969,
      "step": 566
    },
    {
      "epoch": 0.07665399239543726,
      "grad_norm": 1.4872692823410034,
      "learning_rate": 2.9995446779668078e-05,
      "loss": 0.07397747039794922,
      "step": 567
    },
    {
      "epoch": 0.07678918462188423,
      "grad_norm": 0.9276036620140076,
      "learning_rate": 2.9995276589349992e-05,
      "loss": 0.08028507232666016,
      "step": 568
    },
    {
      "epoch": 0.07692437684833123,
      "grad_norm": 2.156606912612915,
      "learning_rate": 2.9995103276935357e-05,
      "loss": 0.11402225494384766,
      "step": 569
    },
    {
      "epoch": 0.0770595690747782,
      "grad_norm": 0.6237265467643738,
      "learning_rate": 2.9994926842460258e-05,
      "loss": 0.062169551849365234,
      "step": 570
    },
    {
      "epoch": 0.07719476130122518,
      "grad_norm": 1.6970484256744385,
      "learning_rate": 2.9994747285961428e-05,
      "loss": 0.15056133270263672,
      "step": 571
    },
    {
      "epoch": 0.07732995352767216,
      "grad_norm": 0.6036470532417297,
      "learning_rate": 2.9994564607476255e-05,
      "loss": 0.06957483291625977,
      "step": 572
    },
    {
      "epoch": 0.07746514575411914,
      "grad_norm": 1.0580317974090576,
      "learning_rate": 2.9994378807042762e-05,
      "loss": 0.07123374938964844,
      "step": 573
    },
    {
      "epoch": 0.07760033798056612,
      "grad_norm": 1.689844012260437,
      "learning_rate": 2.9994189884699647e-05,
      "loss": 0.0945281982421875,
      "step": 574
    },
    {
      "epoch": 0.0777355302070131,
      "grad_norm": 1.0948611497879028,
      "learning_rate": 2.9993997840486233e-05,
      "loss": 0.07328128814697266,
      "step": 575
    },
    {
      "epoch": 0.07787072243346008,
      "grad_norm": 2.237011432647705,
      "learning_rate": 2.9993802674442516e-05,
      "loss": 0.0851888656616211,
      "step": 576
    },
    {
      "epoch": 0.07800591465990706,
      "grad_norm": 0.6065527200698853,
      "learning_rate": 2.999360438660913e-05,
      "loss": 0.10452127456665039,
      "step": 577
    },
    {
      "epoch": 0.07814110688635403,
      "grad_norm": 1.1571671962738037,
      "learning_rate": 2.9993402977027346e-05,
      "loss": 0.09570884704589844,
      "step": 578
    },
    {
      "epoch": 0.07827629911280101,
      "grad_norm": 1.6818926334381104,
      "learning_rate": 2.999319844573911e-05,
      "loss": 0.10058212280273438,
      "step": 579
    },
    {
      "epoch": 0.07841149133924799,
      "grad_norm": 1.6151068210601807,
      "learning_rate": 2.9992990792787007e-05,
      "loss": 0.10038471221923828,
      "step": 580
    },
    {
      "epoch": 0.07854668356569497,
      "grad_norm": 0.8078140616416931,
      "learning_rate": 2.999278001821427e-05,
      "loss": 0.08383822441101074,
      "step": 581
    },
    {
      "epoch": 0.07868187579214195,
      "grad_norm": 1.0977928638458252,
      "learning_rate": 2.9992566122064775e-05,
      "loss": 0.10461854934692383,
      "step": 582
    },
    {
      "epoch": 0.07881706801858893,
      "grad_norm": 0.49909019470214844,
      "learning_rate": 2.999234910438307e-05,
      "loss": 0.0581972599029541,
      "step": 583
    },
    {
      "epoch": 0.0789522602450359,
      "grad_norm": 1.8760097026824951,
      "learning_rate": 2.999212896521433e-05,
      "loss": 0.08929443359375,
      "step": 584
    },
    {
      "epoch": 0.07908745247148288,
      "grad_norm": 2.0165154933929443,
      "learning_rate": 2.999190570460439e-05,
      "loss": 0.10365009307861328,
      "step": 585
    },
    {
      "epoch": 0.07922264469792988,
      "grad_norm": 0.3590303957462311,
      "learning_rate": 2.9991679322599734e-05,
      "loss": 0.0414276123046875,
      "step": 586
    },
    {
      "epoch": 0.07935783692437685,
      "grad_norm": 1.8026360273361206,
      "learning_rate": 2.9991449819247505e-05,
      "loss": 0.08381783962249756,
      "step": 587
    },
    {
      "epoch": 0.07949302915082383,
      "grad_norm": 1.1399129629135132,
      "learning_rate": 2.9991217194595474e-05,
      "loss": 0.07933759689331055,
      "step": 588
    },
    {
      "epoch": 0.07962822137727081,
      "grad_norm": 1.8184621334075928,
      "learning_rate": 2.9990981448692078e-05,
      "loss": 0.08505058288574219,
      "step": 589
    },
    {
      "epoch": 0.07976341360371779,
      "grad_norm": 0.7170165777206421,
      "learning_rate": 2.999074258158641e-05,
      "loss": 0.08453178405761719,
      "step": 590
    },
    {
      "epoch": 0.07989860583016477,
      "grad_norm": 1.4948357343673706,
      "learning_rate": 2.9990500593328192e-05,
      "loss": 0.08392477035522461,
      "step": 591
    },
    {
      "epoch": 0.08003379805661175,
      "grad_norm": 2.0151801109313965,
      "learning_rate": 2.999025548396781e-05,
      "loss": 0.06732606887817383,
      "step": 592
    },
    {
      "epoch": 0.08016899028305872,
      "grad_norm": 1.5872349739074707,
      "learning_rate": 2.9990007253556302e-05,
      "loss": 0.08662223815917969,
      "step": 593
    },
    {
      "epoch": 0.0803041825095057,
      "grad_norm": 0.9635038375854492,
      "learning_rate": 2.9989755902145345e-05,
      "loss": 0.08410263061523438,
      "step": 594
    },
    {
      "epoch": 0.08043937473595268,
      "grad_norm": 0.6852844953536987,
      "learning_rate": 2.9989501429787273e-05,
      "loss": 0.07874822616577148,
      "step": 595
    },
    {
      "epoch": 0.08057456696239966,
      "grad_norm": 0.37786686420440674,
      "learning_rate": 2.9989243836535073e-05,
      "loss": 0.0660257339477539,
      "step": 596
    },
    {
      "epoch": 0.08070975918884664,
      "grad_norm": 0.45019659399986267,
      "learning_rate": 2.998898312244237e-05,
      "loss": 0.07022261619567871,
      "step": 597
    },
    {
      "epoch": 0.08084495141529362,
      "grad_norm": 0.33326104283332825,
      "learning_rate": 2.9988719287563452e-05,
      "loss": 0.062285423278808594,
      "step": 598
    },
    {
      "epoch": 0.0809801436417406,
      "grad_norm": 0.47920557856559753,
      "learning_rate": 2.998845233195325e-05,
      "loss": 0.06540918350219727,
      "step": 599
    },
    {
      "epoch": 0.08111533586818757,
      "grad_norm": 0.6676343083381653,
      "learning_rate": 2.998818225566734e-05,
      "loss": 0.10930967330932617,
      "step": 600
    },
    {
      "epoch": 0.08125052809463455,
      "grad_norm": 1.1839637756347656,
      "learning_rate": 2.998790905876196e-05,
      "loss": 0.12560462951660156,
      "step": 601
    },
    {
      "epoch": 0.08138572032108153,
      "grad_norm": 0.7003291845321655,
      "learning_rate": 2.9987632741293987e-05,
      "loss": 0.07976245880126953,
      "step": 602
    },
    {
      "epoch": 0.08152091254752852,
      "grad_norm": 0.6434202194213867,
      "learning_rate": 2.998735330332096e-05,
      "loss": 0.09475851058959961,
      "step": 603
    },
    {
      "epoch": 0.0816561047739755,
      "grad_norm": 1.0487202405929565,
      "learning_rate": 2.9987070744901046e-05,
      "loss": 0.08417701721191406,
      "step": 604
    },
    {
      "epoch": 0.08179129700042248,
      "grad_norm": 1.0852630138397217,
      "learning_rate": 2.9986785066093084e-05,
      "loss": 0.10083460807800293,
      "step": 605
    },
    {
      "epoch": 0.08192648922686946,
      "grad_norm": 0.45663532614707947,
      "learning_rate": 2.9986496266956556e-05,
      "loss": 0.06810188293457031,
      "step": 606
    },
    {
      "epoch": 0.08206168145331644,
      "grad_norm": 1.3907688856124878,
      "learning_rate": 2.9986204347551583e-05,
      "loss": 0.11536788940429688,
      "step": 607
    },
    {
      "epoch": 0.08219687367976342,
      "grad_norm": 0.9038923382759094,
      "learning_rate": 2.9985909307938948e-05,
      "loss": 0.06625890731811523,
      "step": 608
    },
    {
      "epoch": 0.0823320659062104,
      "grad_norm": 0.8186825513839722,
      "learning_rate": 2.9985611148180082e-05,
      "loss": 0.08992385864257812,
      "step": 609
    },
    {
      "epoch": 0.08246725813265737,
      "grad_norm": 0.49454689025878906,
      "learning_rate": 2.9985309868337063e-05,
      "loss": 0.08014965057373047,
      "step": 610
    },
    {
      "epoch": 0.08260245035910435,
      "grad_norm": 0.7636528015136719,
      "learning_rate": 2.9985005468472617e-05,
      "loss": 0.06519794464111328,
      "step": 611
    },
    {
      "epoch": 0.08273764258555133,
      "grad_norm": 0.5517367720603943,
      "learning_rate": 2.9984697948650124e-05,
      "loss": 0.0753469467163086,
      "step": 612
    },
    {
      "epoch": 0.08287283481199831,
      "grad_norm": 1.6862266063690186,
      "learning_rate": 2.998438730893361e-05,
      "loss": 0.14296340942382812,
      "step": 613
    },
    {
      "epoch": 0.08300802703844529,
      "grad_norm": 1.0018746852874756,
      "learning_rate": 2.9984073549387747e-05,
      "loss": 0.09372568130493164,
      "step": 614
    },
    {
      "epoch": 0.08314321926489227,
      "grad_norm": 0.5324258804321289,
      "learning_rate": 2.998375667007787e-05,
      "loss": 0.10657310485839844,
      "step": 615
    },
    {
      "epoch": 0.08327841149133924,
      "grad_norm": 0.5407150983810425,
      "learning_rate": 2.998343667106995e-05,
      "loss": 0.06703996658325195,
      "step": 616
    },
    {
      "epoch": 0.08341360371778622,
      "grad_norm": 0.7302781343460083,
      "learning_rate": 2.9983113552430616e-05,
      "loss": 0.08555793762207031,
      "step": 617
    },
    {
      "epoch": 0.0835487959442332,
      "grad_norm": 1.0875376462936401,
      "learning_rate": 2.9982787314227134e-05,
      "loss": 0.0678713321685791,
      "step": 618
    },
    {
      "epoch": 0.08368398817068018,
      "grad_norm": 0.9090947508811951,
      "learning_rate": 2.998245795652744e-05,
      "loss": 0.06933879852294922,
      "step": 619
    },
    {
      "epoch": 0.08381918039712717,
      "grad_norm": 1.000343680381775,
      "learning_rate": 2.9982125479400106e-05,
      "loss": 0.07668018341064453,
      "step": 620
    },
    {
      "epoch": 0.08395437262357415,
      "grad_norm": 1.2728577852249146,
      "learning_rate": 2.9981789882914352e-05,
      "loss": 0.05016660690307617,
      "step": 621
    },
    {
      "epoch": 0.08408956485002113,
      "grad_norm": 0.5821818709373474,
      "learning_rate": 2.9981451167140048e-05,
      "loss": 0.07403802871704102,
      "step": 622
    },
    {
      "epoch": 0.08422475707646811,
      "grad_norm": 0.7069630026817322,
      "learning_rate": 2.9981109332147722e-05,
      "loss": 0.09298419952392578,
      "step": 623
    },
    {
      "epoch": 0.08435994930291509,
      "grad_norm": 2.1008784770965576,
      "learning_rate": 2.9980764378008545e-05,
      "loss": 0.09131908416748047,
      "step": 624
    },
    {
      "epoch": 0.08449514152936206,
      "grad_norm": 0.5936189293861389,
      "learning_rate": 2.9980416304794332e-05,
      "loss": 0.07242202758789062,
      "step": 625
    },
    {
      "epoch": 0.08463033375580904,
      "grad_norm": 0.5418726801872253,
      "learning_rate": 2.9980065112577565e-05,
      "loss": 0.07906627655029297,
      "step": 626
    },
    {
      "epoch": 0.08476552598225602,
      "grad_norm": 1.0324081182479858,
      "learning_rate": 2.9979710801431357e-05,
      "loss": 0.07719612121582031,
      "step": 627
    },
    {
      "epoch": 0.084900718208703,
      "grad_norm": 0.7520241737365723,
      "learning_rate": 2.997935337142948e-05,
      "loss": 0.06949472427368164,
      "step": 628
    },
    {
      "epoch": 0.08503591043514998,
      "grad_norm": 0.7859668135643005,
      "learning_rate": 2.9978992822646347e-05,
      "loss": 0.08112668991088867,
      "step": 629
    },
    {
      "epoch": 0.08517110266159696,
      "grad_norm": 0.7527902126312256,
      "learning_rate": 2.9978629155157036e-05,
      "loss": 0.08897209167480469,
      "step": 630
    },
    {
      "epoch": 0.08530629488804393,
      "grad_norm": 1.6249278783798218,
      "learning_rate": 2.9978262369037252e-05,
      "loss": 0.10718822479248047,
      "step": 631
    },
    {
      "epoch": 0.08544148711449091,
      "grad_norm": 1.6737232208251953,
      "learning_rate": 2.9977892464363375e-05,
      "loss": 0.09795284271240234,
      "step": 632
    },
    {
      "epoch": 0.08557667934093789,
      "grad_norm": 0.856467604637146,
      "learning_rate": 2.9977519441212412e-05,
      "loss": 0.12327003479003906,
      "step": 633
    },
    {
      "epoch": 0.08571187156738487,
      "grad_norm": 1.5771981477737427,
      "learning_rate": 2.9977143299662034e-05,
      "loss": 0.08191108703613281,
      "step": 634
    },
    {
      "epoch": 0.08584706379383185,
      "grad_norm": 1.770072102546692,
      "learning_rate": 2.997676403979055e-05,
      "loss": 0.11481571197509766,
      "step": 635
    },
    {
      "epoch": 0.08598225602027883,
      "grad_norm": 1.2036272287368774,
      "learning_rate": 2.997638166167693e-05,
      "loss": 0.10878849029541016,
      "step": 636
    },
    {
      "epoch": 0.08611744824672582,
      "grad_norm": 0.8349308967590332,
      "learning_rate": 2.9975996165400786e-05,
      "loss": 0.061860084533691406,
      "step": 637
    },
    {
      "epoch": 0.0862526404731728,
      "grad_norm": 1.7418475151062012,
      "learning_rate": 2.9975607551042373e-05,
      "loss": 0.06311798095703125,
      "step": 638
    },
    {
      "epoch": 0.08638783269961978,
      "grad_norm": 1.278281569480896,
      "learning_rate": 2.9975215818682607e-05,
      "loss": 0.09191131591796875,
      "step": 639
    },
    {
      "epoch": 0.08652302492606675,
      "grad_norm": 1.303808569908142,
      "learning_rate": 2.9974820968403056e-05,
      "loss": 0.1275162696838379,
      "step": 640
    },
    {
      "epoch": 0.08665821715251373,
      "grad_norm": 3.023681640625,
      "learning_rate": 2.9974423000285923e-05,
      "loss": 0.12473678588867188,
      "step": 641
    },
    {
      "epoch": 0.08679340937896071,
      "grad_norm": 2.977800130844116,
      "learning_rate": 2.9974021914414068e-05,
      "loss": 0.07272624969482422,
      "step": 642
    },
    {
      "epoch": 0.08692860160540769,
      "grad_norm": 3.0921504497528076,
      "learning_rate": 2.9973617710871e-05,
      "loss": 0.14291000366210938,
      "step": 643
    },
    {
      "epoch": 0.08706379383185467,
      "grad_norm": 0.6507948637008667,
      "learning_rate": 2.997321038974087e-05,
      "loss": 0.093414306640625,
      "step": 644
    },
    {
      "epoch": 0.08719898605830165,
      "grad_norm": 1.0436996221542358,
      "learning_rate": 2.997279995110849e-05,
      "loss": 0.10811042785644531,
      "step": 645
    },
    {
      "epoch": 0.08733417828474863,
      "grad_norm": 0.9280104041099548,
      "learning_rate": 2.997238639505932e-05,
      "loss": 0.08973407745361328,
      "step": 646
    },
    {
      "epoch": 0.0874693705111956,
      "grad_norm": 0.880110502243042,
      "learning_rate": 2.997196972167946e-05,
      "loss": 0.06839847564697266,
      "step": 647
    },
    {
      "epoch": 0.08760456273764258,
      "grad_norm": 0.6661954522132874,
      "learning_rate": 2.9971549931055665e-05,
      "loss": 0.09742164611816406,
      "step": 648
    },
    {
      "epoch": 0.08773975496408956,
      "grad_norm": 0.5180718302726746,
      "learning_rate": 2.997112702327533e-05,
      "loss": 0.0813608169555664,
      "step": 649
    },
    {
      "epoch": 0.08787494719053654,
      "grad_norm": 0.32702475786209106,
      "learning_rate": 2.9970700998426518e-05,
      "loss": 0.05808305740356445,
      "step": 650
    },
    {
      "epoch": 0.08801013941698352,
      "grad_norm": 2.2169876098632812,
      "learning_rate": 2.9970271856597925e-05,
      "loss": 0.12061214447021484,
      "step": 651
    },
    {
      "epoch": 0.0881453316434305,
      "grad_norm": 0.3961057662963867,
      "learning_rate": 2.9969839597878896e-05,
      "loss": 0.04576587677001953,
      "step": 652
    },
    {
      "epoch": 0.08828052386987748,
      "grad_norm": 0.618942141532898,
      "learning_rate": 2.9969404222359436e-05,
      "loss": 0.08178424835205078,
      "step": 653
    },
    {
      "epoch": 0.08841571609632447,
      "grad_norm": 0.7832603454589844,
      "learning_rate": 2.9968965730130188e-05,
      "loss": 0.06732845306396484,
      "step": 654
    },
    {
      "epoch": 0.08855090832277145,
      "grad_norm": 1.4064054489135742,
      "learning_rate": 2.9968524121282455e-05,
      "loss": 0.09905624389648438,
      "step": 655
    },
    {
      "epoch": 0.08868610054921842,
      "grad_norm": 0.720722496509552,
      "learning_rate": 2.9968079395908178e-05,
      "loss": 0.04322624206542969,
      "step": 656
    },
    {
      "epoch": 0.0888212927756654,
      "grad_norm": 0.9421080350875854,
      "learning_rate": 2.9967631554099947e-05,
      "loss": 0.059384822845458984,
      "step": 657
    },
    {
      "epoch": 0.08895648500211238,
      "grad_norm": 1.0023547410964966,
      "learning_rate": 2.996718059595101e-05,
      "loss": 0.11681890487670898,
      "step": 658
    },
    {
      "epoch": 0.08909167722855936,
      "grad_norm": 0.48367545008659363,
      "learning_rate": 2.9966726521555265e-05,
      "loss": 0.053379058837890625,
      "step": 659
    },
    {
      "epoch": 0.08922686945500634,
      "grad_norm": 0.7467041611671448,
      "learning_rate": 2.996626933100724e-05,
      "loss": 0.09331703186035156,
      "step": 660
    },
    {
      "epoch": 0.08936206168145332,
      "grad_norm": 0.46111801266670227,
      "learning_rate": 2.996580902440213e-05,
      "loss": 0.05901908874511719,
      "step": 661
    },
    {
      "epoch": 0.0894972539079003,
      "grad_norm": 0.40190592408180237,
      "learning_rate": 2.9965345601835773e-05,
      "loss": 0.05935859680175781,
      "step": 662
    },
    {
      "epoch": 0.08963244613434727,
      "grad_norm": 1.0780490636825562,
      "learning_rate": 2.996487906340466e-05,
      "loss": 0.08236527442932129,
      "step": 663
    },
    {
      "epoch": 0.08976763836079425,
      "grad_norm": 0.5780622959136963,
      "learning_rate": 2.996440940920592e-05,
      "loss": 0.04762125015258789,
      "step": 664
    },
    {
      "epoch": 0.08990283058724123,
      "grad_norm": 0.5727117657661438,
      "learning_rate": 2.996393663933735e-05,
      "loss": 0.08030319213867188,
      "step": 665
    },
    {
      "epoch": 0.09003802281368821,
      "grad_norm": 1.2621744871139526,
      "learning_rate": 2.9963460753897364e-05,
      "loss": 0.08369064331054688,
      "step": 666
    },
    {
      "epoch": 0.09017321504013519,
      "grad_norm": 1.0983774662017822,
      "learning_rate": 2.996298175298506e-05,
      "loss": 0.0863642692565918,
      "step": 667
    },
    {
      "epoch": 0.09030840726658217,
      "grad_norm": 0.49387016892433167,
      "learning_rate": 2.996249963670016e-05,
      "loss": 0.09246349334716797,
      "step": 668
    },
    {
      "epoch": 0.09044359949302914,
      "grad_norm": 0.6467683911323547,
      "learning_rate": 2.9962014405143042e-05,
      "loss": 0.09410476684570312,
      "step": 669
    },
    {
      "epoch": 0.09057879171947612,
      "grad_norm": 1.23124361038208,
      "learning_rate": 2.9961526058414745e-05,
      "loss": 0.09699106216430664,
      "step": 670
    },
    {
      "epoch": 0.09071398394592312,
      "grad_norm": 0.7838457822799683,
      "learning_rate": 2.9961034596616936e-05,
      "loss": 0.0928945541381836,
      "step": 671
    },
    {
      "epoch": 0.0908491761723701,
      "grad_norm": 0.22324122488498688,
      "learning_rate": 2.996054001985194e-05,
      "loss": 0.03599286079406738,
      "step": 672
    },
    {
      "epoch": 0.09098436839881707,
      "grad_norm": 0.8104805946350098,
      "learning_rate": 2.9960042328222732e-05,
      "loss": 0.0686802864074707,
      "step": 673
    },
    {
      "epoch": 0.09111956062526405,
      "grad_norm": 0.5143696665763855,
      "learning_rate": 2.995954152183294e-05,
      "loss": 0.04830813407897949,
      "step": 674
    },
    {
      "epoch": 0.09125475285171103,
      "grad_norm": 1.3214099407196045,
      "learning_rate": 2.9959037600786822e-05,
      "loss": 0.10090303421020508,
      "step": 675
    },
    {
      "epoch": 0.09138994507815801,
      "grad_norm": 0.6124051809310913,
      "learning_rate": 2.9958530565189307e-05,
      "loss": 0.07726764678955078,
      "step": 676
    },
    {
      "epoch": 0.09152513730460499,
      "grad_norm": 1.5081795454025269,
      "learning_rate": 2.995802041514596e-05,
      "loss": 0.08885860443115234,
      "step": 677
    },
    {
      "epoch": 0.09166032953105197,
      "grad_norm": 0.8456224799156189,
      "learning_rate": 2.9957507150762996e-05,
      "loss": 0.06396722793579102,
      "step": 678
    },
    {
      "epoch": 0.09179552175749894,
      "grad_norm": 2.4486069679260254,
      "learning_rate": 2.9956990772147283e-05,
      "loss": 0.09366321563720703,
      "step": 679
    },
    {
      "epoch": 0.09193071398394592,
      "grad_norm": 0.7520796060562134,
      "learning_rate": 2.9956471279406324e-05,
      "loss": 0.08934783935546875,
      "step": 680
    },
    {
      "epoch": 0.0920659062103929,
      "grad_norm": 1.2322204113006592,
      "learning_rate": 2.9955948672648298e-05,
      "loss": 0.06775808334350586,
      "step": 681
    },
    {
      "epoch": 0.09220109843683988,
      "grad_norm": 2.079535961151123,
      "learning_rate": 2.9955422951981994e-05,
      "loss": 0.09503650665283203,
      "step": 682
    },
    {
      "epoch": 0.09233629066328686,
      "grad_norm": 2.127537250518799,
      "learning_rate": 2.995489411751688e-05,
      "loss": 0.08817386627197266,
      "step": 683
    },
    {
      "epoch": 0.09247148288973384,
      "grad_norm": 1.3528122901916504,
      "learning_rate": 2.9954362169363064e-05,
      "loss": 0.17700481414794922,
      "step": 684
    },
    {
      "epoch": 0.09260667511618081,
      "grad_norm": 2.128573179244995,
      "learning_rate": 2.99538271076313e-05,
      "loss": 0.10328292846679688,
      "step": 685
    },
    {
      "epoch": 0.09274186734262779,
      "grad_norm": 0.7017504572868347,
      "learning_rate": 2.9953288932432985e-05,
      "loss": 0.08335733413696289,
      "step": 686
    },
    {
      "epoch": 0.09287705956907477,
      "grad_norm": 1.8741538524627686,
      "learning_rate": 2.995274764388018e-05,
      "loss": 0.09028053283691406,
      "step": 687
    },
    {
      "epoch": 0.09301225179552176,
      "grad_norm": 0.9078563451766968,
      "learning_rate": 2.9952203242085566e-05,
      "loss": 0.06004953384399414,
      "step": 688
    },
    {
      "epoch": 0.09314744402196874,
      "grad_norm": 1.0448877811431885,
      "learning_rate": 2.995165572716251e-05,
      "loss": 0.09534072875976562,
      "step": 689
    },
    {
      "epoch": 0.09328263624841572,
      "grad_norm": 0.7695907950401306,
      "learning_rate": 2.9951105099225003e-05,
      "loss": 0.08654117584228516,
      "step": 690
    },
    {
      "epoch": 0.0934178284748627,
      "grad_norm": 0.4433549642562866,
      "learning_rate": 2.995055135838768e-05,
      "loss": 0.05588960647583008,
      "step": 691
    },
    {
      "epoch": 0.09355302070130968,
      "grad_norm": 0.5443503856658936,
      "learning_rate": 2.994999450476584e-05,
      "loss": 0.10817337036132812,
      "step": 692
    },
    {
      "epoch": 0.09368821292775666,
      "grad_norm": 0.6996601223945618,
      "learning_rate": 2.9949434538475414e-05,
      "loss": 0.08685922622680664,
      "step": 693
    },
    {
      "epoch": 0.09382340515420363,
      "grad_norm": 0.4138985574245453,
      "learning_rate": 2.9948871459633008e-05,
      "loss": 0.07134532928466797,
      "step": 694
    },
    {
      "epoch": 0.09395859738065061,
      "grad_norm": 0.5843410491943359,
      "learning_rate": 2.994830526835584e-05,
      "loss": 0.09716653823852539,
      "step": 695
    },
    {
      "epoch": 0.09409378960709759,
      "grad_norm": 0.43319714069366455,
      "learning_rate": 2.9947735964761803e-05,
      "loss": 0.08893775939941406,
      "step": 696
    },
    {
      "epoch": 0.09422898183354457,
      "grad_norm": 0.7768867015838623,
      "learning_rate": 2.9947163548969428e-05,
      "loss": 0.08929061889648438,
      "step": 697
    },
    {
      "epoch": 0.09436417405999155,
      "grad_norm": 0.18085744976997375,
      "learning_rate": 2.9946588021097893e-05,
      "loss": 0.03959512710571289,
      "step": 698
    },
    {
      "epoch": 0.09449936628643853,
      "grad_norm": 0.6254941821098328,
      "learning_rate": 2.9946009381267028e-05,
      "loss": 0.03760814666748047,
      "step": 699
    },
    {
      "epoch": 0.0946345585128855,
      "grad_norm": 0.48299524188041687,
      "learning_rate": 2.9945427629597306e-05,
      "loss": 0.0595393180847168,
      "step": 700
    },
    {
      "epoch": 0.09476975073933248,
      "grad_norm": 1.3369132280349731,
      "learning_rate": 2.9944842766209853e-05,
      "loss": 0.08656501770019531,
      "step": 701
    },
    {
      "epoch": 0.09490494296577946,
      "grad_norm": 0.4847789704799652,
      "learning_rate": 2.9944254791226444e-05,
      "loss": 0.09263086318969727,
      "step": 702
    },
    {
      "epoch": 0.09504013519222644,
      "grad_norm": 0.9162623882293701,
      "learning_rate": 2.994366370476949e-05,
      "loss": 0.11437129974365234,
      "step": 703
    },
    {
      "epoch": 0.09517532741867342,
      "grad_norm": 1.4687793254852295,
      "learning_rate": 2.9943069506962067e-05,
      "loss": 0.09958362579345703,
      "step": 704
    },
    {
      "epoch": 0.09531051964512041,
      "grad_norm": 0.5143060684204102,
      "learning_rate": 2.9942472197927886e-05,
      "loss": 0.05807340145111084,
      "step": 705
    },
    {
      "epoch": 0.09544571187156739,
      "grad_norm": 1.066678524017334,
      "learning_rate": 2.994187177779131e-05,
      "loss": 0.11996746063232422,
      "step": 706
    },
    {
      "epoch": 0.09558090409801437,
      "grad_norm": 0.4829036593437195,
      "learning_rate": 2.9941268246677353e-05,
      "loss": 0.05228614807128906,
      "step": 707
    },
    {
      "epoch": 0.09571609632446135,
      "grad_norm": 0.8587409853935242,
      "learning_rate": 2.9940661604711664e-05,
      "loss": 0.0664682388305664,
      "step": 708
    },
    {
      "epoch": 0.09585128855090833,
      "grad_norm": 1.2800363302230835,
      "learning_rate": 2.994005185202056e-05,
      "loss": 0.07017230987548828,
      "step": 709
    },
    {
      "epoch": 0.0959864807773553,
      "grad_norm": 1.7786747217178345,
      "learning_rate": 2.9939438988730986e-05,
      "loss": 0.08867454528808594,
      "step": 710
    },
    {
      "epoch": 0.09612167300380228,
      "grad_norm": 0.5550428628921509,
      "learning_rate": 2.9938823014970553e-05,
      "loss": 0.08253812789916992,
      "step": 711
    },
    {
      "epoch": 0.09625686523024926,
      "grad_norm": 3.2433102130889893,
      "learning_rate": 2.99382039308675e-05,
      "loss": 0.10310506820678711,
      "step": 712
    },
    {
      "epoch": 0.09639205745669624,
      "grad_norm": 2.4590277671813965,
      "learning_rate": 2.993758173655073e-05,
      "loss": 0.12313079833984375,
      "step": 713
    },
    {
      "epoch": 0.09652724968314322,
      "grad_norm": 1.180315375328064,
      "learning_rate": 2.993695643214979e-05,
      "loss": 0.05230712890625,
      "step": 714
    },
    {
      "epoch": 0.0966624419095902,
      "grad_norm": 1.1300855875015259,
      "learning_rate": 2.9936328017794864e-05,
      "loss": 0.09466934204101562,
      "step": 715
    },
    {
      "epoch": 0.09679763413603718,
      "grad_norm": 1.2884372472763062,
      "learning_rate": 2.9935696493616796e-05,
      "loss": 0.08497238159179688,
      "step": 716
    },
    {
      "epoch": 0.09693282636248415,
      "grad_norm": 1.4714183807373047,
      "learning_rate": 2.9935061859747065e-05,
      "loss": 0.08114802837371826,
      "step": 717
    },
    {
      "epoch": 0.09706801858893113,
      "grad_norm": 1.5180643796920776,
      "learning_rate": 2.993442411631782e-05,
      "loss": 0.06940174102783203,
      "step": 718
    },
    {
      "epoch": 0.09720321081537811,
      "grad_norm": 1.6803812980651855,
      "learning_rate": 2.9933783263461827e-05,
      "loss": 0.0589139461517334,
      "step": 719
    },
    {
      "epoch": 0.09733840304182509,
      "grad_norm": 1.2122961282730103,
      "learning_rate": 2.9933139301312526e-05,
      "loss": 0.0832834243774414,
      "step": 720
    },
    {
      "epoch": 0.09747359526827207,
      "grad_norm": 0.6643701791763306,
      "learning_rate": 2.9932492230003984e-05,
      "loss": 0.051471710205078125,
      "step": 721
    },
    {
      "epoch": 0.09760878749471906,
      "grad_norm": 0.9927834868431091,
      "learning_rate": 2.993184204967094e-05,
      "loss": 0.0924386978149414,
      "step": 722
    },
    {
      "epoch": 0.09774397972116604,
      "grad_norm": 0.8113033771514893,
      "learning_rate": 2.9931188760448748e-05,
      "loss": 0.0583806037902832,
      "step": 723
    },
    {
      "epoch": 0.09787917194761302,
      "grad_norm": 1.3883392810821533,
      "learning_rate": 2.9930532362473433e-05,
      "loss": 0.09034204483032227,
      "step": 724
    },
    {
      "epoch": 0.09801436417406,
      "grad_norm": 0.8473360538482666,
      "learning_rate": 2.9929872855881663e-05,
      "loss": 0.06917667388916016,
      "step": 725
    },
    {
      "epoch": 0.09814955640050697,
      "grad_norm": 0.36346328258514404,
      "learning_rate": 2.9929210240810744e-05,
      "loss": 0.08127927780151367,
      "step": 726
    },
    {
      "epoch": 0.09828474862695395,
      "grad_norm": 0.8365683555603027,
      "learning_rate": 2.9928544517398644e-05,
      "loss": 0.0607752799987793,
      "step": 727
    },
    {
      "epoch": 0.09841994085340093,
      "grad_norm": 0.430542916059494,
      "learning_rate": 2.9927875685783966e-05,
      "loss": 0.06436395645141602,
      "step": 728
    },
    {
      "epoch": 0.09855513307984791,
      "grad_norm": 0.6196367740631104,
      "learning_rate": 2.9927203746105968e-05,
      "loss": 0.07642745971679688,
      "step": 729
    },
    {
      "epoch": 0.09869032530629489,
      "grad_norm": 2.0761096477508545,
      "learning_rate": 2.9926528698504546e-05,
      "loss": 0.10896492004394531,
      "step": 730
    },
    {
      "epoch": 0.09882551753274187,
      "grad_norm": 0.8039914965629578,
      "learning_rate": 2.992585054312025e-05,
      "loss": 0.09295988082885742,
      "step": 731
    },
    {
      "epoch": 0.09896070975918884,
      "grad_norm": 0.9510341286659241,
      "learning_rate": 2.9925169280094278e-05,
      "loss": 0.1031789779663086,
      "step": 732
    },
    {
      "epoch": 0.09909590198563582,
      "grad_norm": 0.2380317896604538,
      "learning_rate": 2.9924484909568472e-05,
      "loss": 0.034404754638671875,
      "step": 733
    },
    {
      "epoch": 0.0992310942120828,
      "grad_norm": 0.2911379933357239,
      "learning_rate": 2.9923797431685322e-05,
      "loss": 0.08263731002807617,
      "step": 734
    },
    {
      "epoch": 0.09936628643852978,
      "grad_norm": 0.30056285858154297,
      "learning_rate": 2.992310684658796e-05,
      "loss": 0.057534217834472656,
      "step": 735
    },
    {
      "epoch": 0.09950147866497676,
      "grad_norm": 0.9018310308456421,
      "learning_rate": 2.9922413154420173e-05,
      "loss": 0.0929574966430664,
      "step": 736
    },
    {
      "epoch": 0.09963667089142374,
      "grad_norm": 0.37030431628227234,
      "learning_rate": 2.9921716355326393e-05,
      "loss": 0.06572103500366211,
      "step": 737
    },
    {
      "epoch": 0.09977186311787073,
      "grad_norm": 0.59752357006073,
      "learning_rate": 2.9921016449451695e-05,
      "loss": 0.08301687240600586,
      "step": 738
    },
    {
      "epoch": 0.09990705534431771,
      "grad_norm": 1.3171252012252808,
      "learning_rate": 2.9920313436941805e-05,
      "loss": 0.0990591049194336,
      "step": 739
    },
    {
      "epoch": 0.10004224757076469,
      "grad_norm": 0.27812743186950684,
      "learning_rate": 2.991960731794309e-05,
      "loss": 0.055756568908691406,
      "step": 740
    },
    {
      "epoch": 0.10017743979721166,
      "grad_norm": 0.3967611789703369,
      "learning_rate": 2.991889809260257e-05,
      "loss": 0.07887506484985352,
      "step": 741
    },
    {
      "epoch": 0.10031263202365864,
      "grad_norm": 0.5462791323661804,
      "learning_rate": 2.9918185761067912e-05,
      "loss": 0.08826637268066406,
      "step": 742
    },
    {
      "epoch": 0.10044782425010562,
      "grad_norm": 0.430637389421463,
      "learning_rate": 2.9917470323487423e-05,
      "loss": 0.08333444595336914,
      "step": 743
    },
    {
      "epoch": 0.1005830164765526,
      "grad_norm": 0.7959412932395935,
      "learning_rate": 2.9916751780010063e-05,
      "loss": 0.05894875526428223,
      "step": 744
    },
    {
      "epoch": 0.10071820870299958,
      "grad_norm": 0.6657803058624268,
      "learning_rate": 2.9916030130785436e-05,
      "loss": 0.0435028076171875,
      "step": 745
    },
    {
      "epoch": 0.10085340092944656,
      "grad_norm": 0.5945706963539124,
      "learning_rate": 2.99153053759638e-05,
      "loss": 0.10613536834716797,
      "step": 746
    },
    {
      "epoch": 0.10098859315589354,
      "grad_norm": 0.40436336398124695,
      "learning_rate": 2.991457751569604e-05,
      "loss": 0.06596684455871582,
      "step": 747
    },
    {
      "epoch": 0.10112378538234051,
      "grad_norm": 1.403603196144104,
      "learning_rate": 2.991384655013371e-05,
      "loss": 0.1388988494873047,
      "step": 748
    },
    {
      "epoch": 0.10125897760878749,
      "grad_norm": 0.8612577319145203,
      "learning_rate": 2.9913112479429e-05,
      "loss": 0.10822081565856934,
      "step": 749
    },
    {
      "epoch": 0.10139416983523447,
      "grad_norm": 0.4296482801437378,
      "learning_rate": 2.991237530373474e-05,
      "loss": 0.08353328704833984,
      "step": 750
    },
    {
      "epoch": 0.10152936206168145,
      "grad_norm": 1.9338592290878296,
      "learning_rate": 2.9911635023204423e-05,
      "loss": 0.12401151657104492,
      "step": 751
    },
    {
      "epoch": 0.10166455428812843,
      "grad_norm": 0.8141088485717773,
      "learning_rate": 2.9910891637992172e-05,
      "loss": 0.08863067626953125,
      "step": 752
    },
    {
      "epoch": 0.1017997465145754,
      "grad_norm": 0.4315641224384308,
      "learning_rate": 2.991014514825277e-05,
      "loss": 0.08157825469970703,
      "step": 753
    },
    {
      "epoch": 0.10193493874102239,
      "grad_norm": 0.6997880339622498,
      "learning_rate": 2.9909395554141638e-05,
      "loss": 0.06783151626586914,
      "step": 754
    },
    {
      "epoch": 0.10207013096746938,
      "grad_norm": 1.4606138467788696,
      "learning_rate": 2.9908642855814844e-05,
      "loss": 0.08960342407226562,
      "step": 755
    },
    {
      "epoch": 0.10220532319391636,
      "grad_norm": 0.4545760452747345,
      "learning_rate": 2.9907887053429107e-05,
      "loss": 0.0668487548828125,
      "step": 756
    },
    {
      "epoch": 0.10234051542036333,
      "grad_norm": 0.4696957767009735,
      "learning_rate": 2.9907128147141783e-05,
      "loss": 0.07184481620788574,
      "step": 757
    },
    {
      "epoch": 0.10247570764681031,
      "grad_norm": 0.5667471885681152,
      "learning_rate": 2.990636613711089e-05,
      "loss": 0.08581137657165527,
      "step": 758
    },
    {
      "epoch": 0.10261089987325729,
      "grad_norm": 0.37433671951293945,
      "learning_rate": 2.990560102349507e-05,
      "loss": 0.03728783130645752,
      "step": 759
    },
    {
      "epoch": 0.10274609209970427,
      "grad_norm": 1.2046866416931152,
      "learning_rate": 2.9904832806453635e-05,
      "loss": 0.07507181167602539,
      "step": 760
    },
    {
      "epoch": 0.10288128432615125,
      "grad_norm": 0.5006322860717773,
      "learning_rate": 2.9904061486146524e-05,
      "loss": 0.0571751594543457,
      "step": 761
    },
    {
      "epoch": 0.10301647655259823,
      "grad_norm": 1.884344220161438,
      "learning_rate": 2.9903287062734333e-05,
      "loss": 0.10209131240844727,
      "step": 762
    },
    {
      "epoch": 0.1031516687790452,
      "grad_norm": 0.3833383321762085,
      "learning_rate": 2.990250953637831e-05,
      "loss": 0.07235002517700195,
      "step": 763
    },
    {
      "epoch": 0.10328686100549218,
      "grad_norm": 1.8122638463974,
      "learning_rate": 2.9901728907240326e-05,
      "loss": 0.07389068603515625,
      "step": 764
    },
    {
      "epoch": 0.10342205323193916,
      "grad_norm": 1.4250800609588623,
      "learning_rate": 2.9900945175482916e-05,
      "loss": 0.07757329940795898,
      "step": 765
    },
    {
      "epoch": 0.10355724545838614,
      "grad_norm": 1.2942662239074707,
      "learning_rate": 2.990015834126926e-05,
      "loss": 0.1118459701538086,
      "step": 766
    },
    {
      "epoch": 0.10369243768483312,
      "grad_norm": 0.535193145275116,
      "learning_rate": 2.989936840476318e-05,
      "loss": 0.07560348510742188,
      "step": 767
    },
    {
      "epoch": 0.1038276299112801,
      "grad_norm": 1.116136908531189,
      "learning_rate": 2.9898575366129145e-05,
      "loss": 0.07685685157775879,
      "step": 768
    },
    {
      "epoch": 0.10396282213772708,
      "grad_norm": 1.636549949645996,
      "learning_rate": 2.9897779225532273e-05,
      "loss": 0.06873607635498047,
      "step": 769
    },
    {
      "epoch": 0.10409801436417405,
      "grad_norm": 0.851189374923706,
      "learning_rate": 2.989697998313832e-05,
      "loss": 0.07567405700683594,
      "step": 770
    },
    {
      "epoch": 0.10423320659062103,
      "grad_norm": 0.8681046962738037,
      "learning_rate": 2.989617763911369e-05,
      "loss": 0.08973884582519531,
      "step": 771
    },
    {
      "epoch": 0.10436839881706803,
      "grad_norm": 0.6807007193565369,
      "learning_rate": 2.9895372193625442e-05,
      "loss": 0.09481048583984375,
      "step": 772
    },
    {
      "epoch": 0.104503591043515,
      "grad_norm": 1.2062404155731201,
      "learning_rate": 2.9894563646841273e-05,
      "loss": 0.07929611206054688,
      "step": 773
    },
    {
      "epoch": 0.10463878326996198,
      "grad_norm": 1.149382472038269,
      "learning_rate": 2.9893751998929523e-05,
      "loss": 0.08654499053955078,
      "step": 774
    },
    {
      "epoch": 0.10477397549640896,
      "grad_norm": 1.5561285018920898,
      "learning_rate": 2.9892937250059187e-05,
      "loss": 0.09778118133544922,
      "step": 775
    },
    {
      "epoch": 0.10490916772285594,
      "grad_norm": 0.5485138297080994,
      "learning_rate": 2.9892119400399894e-05,
      "loss": 0.07680368423461914,
      "step": 776
    },
    {
      "epoch": 0.10504435994930292,
      "grad_norm": 0.43881773948669434,
      "learning_rate": 2.989129845012193e-05,
      "loss": 0.07778644561767578,
      "step": 777
    },
    {
      "epoch": 0.1051795521757499,
      "grad_norm": 1.080625057220459,
      "learning_rate": 2.989047439939621e-05,
      "loss": 0.08548808097839355,
      "step": 778
    },
    {
      "epoch": 0.10531474440219687,
      "grad_norm": 0.38331812620162964,
      "learning_rate": 2.9889647248394324e-05,
      "loss": 0.08388328552246094,
      "step": 779
    },
    {
      "epoch": 0.10544993662864385,
      "grad_norm": 1.2832900285720825,
      "learning_rate": 2.9888816997288475e-05,
      "loss": 0.09711027145385742,
      "step": 780
    },
    {
      "epoch": 0.10558512885509083,
      "grad_norm": 0.751434862613678,
      "learning_rate": 2.988798364625153e-05,
      "loss": 0.08174657821655273,
      "step": 781
    },
    {
      "epoch": 0.10572032108153781,
      "grad_norm": 1.0784751176834106,
      "learning_rate": 2.9887147195457002e-05,
      "loss": 0.11003780364990234,
      "step": 782
    },
    {
      "epoch": 0.10585551330798479,
      "grad_norm": 0.450376957654953,
      "learning_rate": 2.9886307645079037e-05,
      "loss": 0.09079742431640625,
      "step": 783
    },
    {
      "epoch": 0.10599070553443177,
      "grad_norm": 1.3440659046173096,
      "learning_rate": 2.9885464995292436e-05,
      "loss": 0.1103525161743164,
      "step": 784
    },
    {
      "epoch": 0.10612589776087875,
      "grad_norm": 0.4448350667953491,
      "learning_rate": 2.9884619246272648e-05,
      "loss": 0.07683658599853516,
      "step": 785
    },
    {
      "epoch": 0.10626108998732572,
      "grad_norm": 1.2078944444656372,
      "learning_rate": 2.988377039819575e-05,
      "loss": 0.06465387344360352,
      "step": 786
    },
    {
      "epoch": 0.1063962822137727,
      "grad_norm": 1.4633591175079346,
      "learning_rate": 2.9882918451238494e-05,
      "loss": 0.049219608306884766,
      "step": 787
    },
    {
      "epoch": 0.10653147444021968,
      "grad_norm": 1.1847048997879028,
      "learning_rate": 2.988206340557825e-05,
      "loss": 0.04823780059814453,
      "step": 788
    },
    {
      "epoch": 0.10666666666666667,
      "grad_norm": 0.5670963525772095,
      "learning_rate": 2.9881205261393037e-05,
      "loss": 0.10341835021972656,
      "step": 789
    },
    {
      "epoch": 0.10680185889311365,
      "grad_norm": 0.8065640926361084,
      "learning_rate": 2.988034401886154e-05,
      "loss": 0.08946609497070312,
      "step": 790
    },
    {
      "epoch": 0.10693705111956063,
      "grad_norm": 1.5516135692596436,
      "learning_rate": 2.9879479678163065e-05,
      "loss": 0.15784835815429688,
      "step": 791
    },
    {
      "epoch": 0.10707224334600761,
      "grad_norm": 0.7580604553222656,
      "learning_rate": 2.9878612239477568e-05,
      "loss": 0.08668923377990723,
      "step": 792
    },
    {
      "epoch": 0.10720743557245459,
      "grad_norm": 1.5237256288528442,
      "learning_rate": 2.9877741702985666e-05,
      "loss": 0.11019420623779297,
      "step": 793
    },
    {
      "epoch": 0.10734262779890157,
      "grad_norm": 0.6032150387763977,
      "learning_rate": 2.98768680688686e-05,
      "loss": 0.07812643051147461,
      "step": 794
    },
    {
      "epoch": 0.10747782002534854,
      "grad_norm": 0.8539298176765442,
      "learning_rate": 2.9875991337308274e-05,
      "loss": 0.07448577880859375,
      "step": 795
    },
    {
      "epoch": 0.10761301225179552,
      "grad_norm": 0.5719842314720154,
      "learning_rate": 2.987511150848722e-05,
      "loss": 0.05310940742492676,
      "step": 796
    },
    {
      "epoch": 0.1077482044782425,
      "grad_norm": 0.7173001170158386,
      "learning_rate": 2.9874228582588627e-05,
      "loss": 0.07994794845581055,
      "step": 797
    },
    {
      "epoch": 0.10788339670468948,
      "grad_norm": 0.686671257019043,
      "learning_rate": 2.9873342559796325e-05,
      "loss": 0.09230518341064453,
      "step": 798
    },
    {
      "epoch": 0.10801858893113646,
      "grad_norm": 0.7101694345474243,
      "learning_rate": 2.9872453440294787e-05,
      "loss": 0.07753467559814453,
      "step": 799
    },
    {
      "epoch": 0.10815378115758344,
      "grad_norm": 1.979806900024414,
      "learning_rate": 2.9871561224269134e-05,
      "loss": 0.13216018676757812,
      "step": 800
    },
    {
      "epoch": 0.10828897338403042,
      "grad_norm": 0.41649532318115234,
      "learning_rate": 2.9870665911905127e-05,
      "loss": 0.06540203094482422,
      "step": 801
    },
    {
      "epoch": 0.1084241656104774,
      "grad_norm": 0.6844872236251831,
      "learning_rate": 2.9869767503389176e-05,
      "loss": 0.07187485694885254,
      "step": 802
    },
    {
      "epoch": 0.10855935783692437,
      "grad_norm": 0.8601712584495544,
      "learning_rate": 2.986886599890834e-05,
      "loss": 0.08647775650024414,
      "step": 803
    },
    {
      "epoch": 0.10869455006337135,
      "grad_norm": 1.345866084098816,
      "learning_rate": 2.9867961398650306e-05,
      "loss": 0.06264054775238037,
      "step": 804
    },
    {
      "epoch": 0.10882974228981833,
      "grad_norm": 1.1698988676071167,
      "learning_rate": 2.9867053702803425e-05,
      "loss": 0.07759284973144531,
      "step": 805
    },
    {
      "epoch": 0.10896493451626532,
      "grad_norm": 0.3186224400997162,
      "learning_rate": 2.9866142911556685e-05,
      "loss": 0.07211780548095703,
      "step": 806
    },
    {
      "epoch": 0.1091001267427123,
      "grad_norm": 1.0271692276000977,
      "learning_rate": 2.9865229025099713e-05,
      "loss": 0.08307266235351562,
      "step": 807
    },
    {
      "epoch": 0.10923531896915928,
      "grad_norm": 0.8040181398391724,
      "learning_rate": 2.986431204362279e-05,
      "loss": 0.062419891357421875,
      "step": 808
    },
    {
      "epoch": 0.10937051119560626,
      "grad_norm": 1.5733109712600708,
      "learning_rate": 2.9863391967316835e-05,
      "loss": 0.08540821075439453,
      "step": 809
    },
    {
      "epoch": 0.10950570342205324,
      "grad_norm": 1.0525751113891602,
      "learning_rate": 2.9862468796373404e-05,
      "loss": 0.09246659278869629,
      "step": 810
    },
    {
      "epoch": 0.10964089564850021,
      "grad_norm": 1.5251206159591675,
      "learning_rate": 2.9861542530984718e-05,
      "loss": 0.0996556282043457,
      "step": 811
    },
    {
      "epoch": 0.10977608787494719,
      "grad_norm": 0.38562726974487305,
      "learning_rate": 2.9860613171343624e-05,
      "loss": 0.07490324974060059,
      "step": 812
    },
    {
      "epoch": 0.10991128010139417,
      "grad_norm": 0.6965151429176331,
      "learning_rate": 2.9859680717643623e-05,
      "loss": 0.08292293548583984,
      "step": 813
    },
    {
      "epoch": 0.11004647232784115,
      "grad_norm": 0.5826792120933533,
      "learning_rate": 2.985874517007885e-05,
      "loss": 0.06980037689208984,
      "step": 814
    },
    {
      "epoch": 0.11018166455428813,
      "grad_norm": 0.6408787369728088,
      "learning_rate": 2.98578065288441e-05,
      "loss": 0.0714254379272461,
      "step": 815
    },
    {
      "epoch": 0.1103168567807351,
      "grad_norm": 0.33082664012908936,
      "learning_rate": 2.9856864794134798e-05,
      "loss": 0.07348203659057617,
      "step": 816
    },
    {
      "epoch": 0.11045204900718208,
      "grad_norm": 0.5304411053657532,
      "learning_rate": 2.9855919966147025e-05,
      "loss": 0.0860593318939209,
      "step": 817
    },
    {
      "epoch": 0.11058724123362906,
      "grad_norm": 0.653801441192627,
      "learning_rate": 2.9854972045077485e-05,
      "loss": 0.0687417984008789,
      "step": 818
    },
    {
      "epoch": 0.11072243346007604,
      "grad_norm": 0.3759799003601074,
      "learning_rate": 2.9854021031123555e-05,
      "loss": 0.06639289855957031,
      "step": 819
    },
    {
      "epoch": 0.11085762568652302,
      "grad_norm": 1.0197070837020874,
      "learning_rate": 2.9853066924483232e-05,
      "loss": 0.06349563598632812,
      "step": 820
    },
    {
      "epoch": 0.11099281791297,
      "grad_norm": 0.19180253148078918,
      "learning_rate": 2.9852109725355173e-05,
      "loss": 0.039182186126708984,
      "step": 821
    },
    {
      "epoch": 0.11112801013941698,
      "grad_norm": 0.3551379442214966,
      "learning_rate": 2.9851149433938662e-05,
      "loss": 0.050806283950805664,
      "step": 822
    },
    {
      "epoch": 0.11126320236586397,
      "grad_norm": 0.4571462869644165,
      "learning_rate": 2.9850186050433645e-05,
      "loss": 0.0558934211730957,
      "step": 823
    },
    {
      "epoch": 0.11139839459231095,
      "grad_norm": 0.4020529091358185,
      "learning_rate": 2.9849219575040708e-05,
      "loss": 0.06947755813598633,
      "step": 824
    },
    {
      "epoch": 0.11153358681875793,
      "grad_norm": 0.38608673214912415,
      "learning_rate": 2.984825000796106e-05,
      "loss": 0.06527233123779297,
      "step": 825
    },
    {
      "epoch": 0.1116687790452049,
      "grad_norm": 0.6543492078781128,
      "learning_rate": 2.9847277349396586e-05,
      "loss": 0.12175655364990234,
      "step": 826
    },
    {
      "epoch": 0.11180397127165188,
      "grad_norm": 0.39084070920944214,
      "learning_rate": 2.984630159954979e-05,
      "loss": 0.07764720916748047,
      "step": 827
    },
    {
      "epoch": 0.11193916349809886,
      "grad_norm": 0.6771588325500488,
      "learning_rate": 2.9845322758623833e-05,
      "loss": 0.11553192138671875,
      "step": 828
    },
    {
      "epoch": 0.11207435572454584,
      "grad_norm": 1.0467925071716309,
      "learning_rate": 2.984434082682251e-05,
      "loss": 0.11504268646240234,
      "step": 829
    },
    {
      "epoch": 0.11220954795099282,
      "grad_norm": 0.32621562480926514,
      "learning_rate": 2.984335580435027e-05,
      "loss": 0.06376457214355469,
      "step": 830
    },
    {
      "epoch": 0.1123447401774398,
      "grad_norm": 0.9813115000724792,
      "learning_rate": 2.9842367691412192e-05,
      "loss": 0.0780191421508789,
      "step": 831
    },
    {
      "epoch": 0.11247993240388678,
      "grad_norm": 0.4412747025489807,
      "learning_rate": 2.9841376488214015e-05,
      "loss": 0.05402421951293945,
      "step": 832
    },
    {
      "epoch": 0.11261512463033375,
      "grad_norm": 0.511574387550354,
      "learning_rate": 2.984038219496211e-05,
      "loss": 0.1097269058227539,
      "step": 833
    },
    {
      "epoch": 0.11275031685678073,
      "grad_norm": 0.6203243136405945,
      "learning_rate": 2.9839384811863493e-05,
      "loss": 0.0794076919555664,
      "step": 834
    },
    {
      "epoch": 0.11288550908322771,
      "grad_norm": 0.48098886013031006,
      "learning_rate": 2.9838384339125824e-05,
      "loss": 0.06277823448181152,
      "step": 835
    },
    {
      "epoch": 0.11302070130967469,
      "grad_norm": 0.30106276273727417,
      "learning_rate": 2.9837380776957405e-05,
      "loss": 0.07796001434326172,
      "step": 836
    },
    {
      "epoch": 0.11315589353612167,
      "grad_norm": 0.6713687777519226,
      "learning_rate": 2.9836374125567193e-05,
      "loss": 0.11228370666503906,
      "step": 837
    },
    {
      "epoch": 0.11329108576256865,
      "grad_norm": 0.8277329802513123,
      "learning_rate": 2.9835364385164764e-05,
      "loss": 0.057244300842285156,
      "step": 838
    },
    {
      "epoch": 0.11342627798901563,
      "grad_norm": 0.5808928608894348,
      "learning_rate": 2.983435155596036e-05,
      "loss": 0.09830093383789062,
      "step": 839
    },
    {
      "epoch": 0.11356147021546262,
      "grad_norm": 0.46376940608024597,
      "learning_rate": 2.9833335638164858e-05,
      "loss": 0.08458471298217773,
      "step": 840
    },
    {
      "epoch": 0.1136966624419096,
      "grad_norm": 0.3028194010257721,
      "learning_rate": 2.9832316631989774e-05,
      "loss": 0.06242990493774414,
      "step": 841
    },
    {
      "epoch": 0.11383185466835657,
      "grad_norm": 0.3183048665523529,
      "learning_rate": 2.9831294537647272e-05,
      "loss": 0.07565617561340332,
      "step": 842
    },
    {
      "epoch": 0.11396704689480355,
      "grad_norm": 1.0704361200332642,
      "learning_rate": 2.9830269355350155e-05,
      "loss": 0.07597064971923828,
      "step": 843
    },
    {
      "epoch": 0.11410223912125053,
      "grad_norm": 0.7910330891609192,
      "learning_rate": 2.9829241085311872e-05,
      "loss": 0.06637001037597656,
      "step": 844
    },
    {
      "epoch": 0.11423743134769751,
      "grad_norm": 0.4403345286846161,
      "learning_rate": 2.9828209727746522e-05,
      "loss": 0.08481502532958984,
      "step": 845
    },
    {
      "epoch": 0.11437262357414449,
      "grad_norm": 0.9092098474502563,
      "learning_rate": 2.982717528286883e-05,
      "loss": 0.09525442123413086,
      "step": 846
    },
    {
      "epoch": 0.11450781580059147,
      "grad_norm": 1.2730127573013306,
      "learning_rate": 2.9826137750894176e-05,
      "loss": 0.10476016998291016,
      "step": 847
    },
    {
      "epoch": 0.11464300802703845,
      "grad_norm": 0.3732256293296814,
      "learning_rate": 2.9825097132038578e-05,
      "loss": 0.05859994888305664,
      "step": 848
    },
    {
      "epoch": 0.11477820025348542,
      "grad_norm": 0.41927069425582886,
      "learning_rate": 2.9824053426518703e-05,
      "loss": 0.09224319458007812,
      "step": 849
    },
    {
      "epoch": 0.1149133924799324,
      "grad_norm": 1.1012687683105469,
      "learning_rate": 2.9823006634551848e-05,
      "loss": 0.0856013298034668,
      "step": 850
    },
    {
      "epoch": 0.11504858470637938,
      "grad_norm": 0.5237995386123657,
      "learning_rate": 2.9821956756355973e-05,
      "loss": 0.07183837890625,
      "step": 851
    },
    {
      "epoch": 0.11518377693282636,
      "grad_norm": 0.3141275942325592,
      "learning_rate": 2.9820903792149653e-05,
      "loss": 0.04247856140136719,
      "step": 852
    },
    {
      "epoch": 0.11531896915927334,
      "grad_norm": 0.8342028260231018,
      "learning_rate": 2.981984774215214e-05,
      "loss": 0.0800313949584961,
      "step": 853
    },
    {
      "epoch": 0.11545416138572032,
      "grad_norm": 0.6760586500167847,
      "learning_rate": 2.9818788606583286e-05,
      "loss": 0.09170246124267578,
      "step": 854
    },
    {
      "epoch": 0.1155893536121673,
      "grad_norm": 0.7063131332397461,
      "learning_rate": 2.9817726385663627e-05,
      "loss": 0.09348559379577637,
      "step": 855
    },
    {
      "epoch": 0.11572454583861427,
      "grad_norm": 0.3852894604206085,
      "learning_rate": 2.9816661079614316e-05,
      "loss": 0.0653526782989502,
      "step": 856
    },
    {
      "epoch": 0.11585973806506127,
      "grad_norm": 0.7683351635932922,
      "learning_rate": 2.9815592688657154e-05,
      "loss": 0.0905923843383789,
      "step": 857
    },
    {
      "epoch": 0.11599493029150824,
      "grad_norm": 0.5725550651550293,
      "learning_rate": 2.9814521213014588e-05,
      "loss": 0.09342670440673828,
      "step": 858
    },
    {
      "epoch": 0.11613012251795522,
      "grad_norm": 0.7682245969772339,
      "learning_rate": 2.9813446652909707e-05,
      "loss": 0.08133316040039062,
      "step": 859
    },
    {
      "epoch": 0.1162653147444022,
      "grad_norm": 0.7258708477020264,
      "learning_rate": 2.981236900856624e-05,
      "loss": 0.10254907608032227,
      "step": 860
    },
    {
      "epoch": 0.11640050697084918,
      "grad_norm": 0.7408928871154785,
      "learning_rate": 2.9811288280208552e-05,
      "loss": 0.0868387222290039,
      "step": 861
    },
    {
      "epoch": 0.11653569919729616,
      "grad_norm": 0.8647645711898804,
      "learning_rate": 2.9810204468061664e-05,
      "loss": 0.09180545806884766,
      "step": 862
    },
    {
      "epoch": 0.11667089142374314,
      "grad_norm": 0.5420779585838318,
      "learning_rate": 2.9809117572351223e-05,
      "loss": 0.06743431091308594,
      "step": 863
    },
    {
      "epoch": 0.11680608365019012,
      "grad_norm": 0.6346800327301025,
      "learning_rate": 2.9808027593303537e-05,
      "loss": 0.06043863296508789,
      "step": 864
    },
    {
      "epoch": 0.1169412758766371,
      "grad_norm": 1.154919147491455,
      "learning_rate": 2.980693453114554e-05,
      "loss": 0.08947134017944336,
      "step": 865
    },
    {
      "epoch": 0.11707646810308407,
      "grad_norm": 0.7358927726745605,
      "learning_rate": 2.980583838610481e-05,
      "loss": 0.055572509765625,
      "step": 866
    },
    {
      "epoch": 0.11721166032953105,
      "grad_norm": 0.6965184807777405,
      "learning_rate": 2.980473915840957e-05,
      "loss": 0.07071423530578613,
      "step": 867
    },
    {
      "epoch": 0.11734685255597803,
      "grad_norm": 2.4694793224334717,
      "learning_rate": 2.9803636848288696e-05,
      "loss": 0.11431694030761719,
      "step": 868
    },
    {
      "epoch": 0.11748204478242501,
      "grad_norm": 1.4147696495056152,
      "learning_rate": 2.9802531455971686e-05,
      "loss": 0.07610034942626953,
      "step": 869
    },
    {
      "epoch": 0.11761723700887199,
      "grad_norm": 1.7168290615081787,
      "learning_rate": 2.980142298168869e-05,
      "loss": 0.09935760498046875,
      "step": 870
    },
    {
      "epoch": 0.11775242923531896,
      "grad_norm": 1.0734306573867798,
      "learning_rate": 2.9800311425670495e-05,
      "loss": 0.09269189834594727,
      "step": 871
    },
    {
      "epoch": 0.11788762146176594,
      "grad_norm": 0.946284294128418,
      "learning_rate": 2.9799196788148538e-05,
      "loss": 0.07281208038330078,
      "step": 872
    },
    {
      "epoch": 0.11802281368821292,
      "grad_norm": 1.427246332168579,
      "learning_rate": 2.9798079069354893e-05,
      "loss": 0.11725389957427979,
      "step": 873
    },
    {
      "epoch": 0.11815800591465991,
      "grad_norm": 1.4104865789413452,
      "learning_rate": 2.9796958269522273e-05,
      "loss": 0.06522738933563232,
      "step": 874
    },
    {
      "epoch": 0.11829319814110689,
      "grad_norm": 0.8461236953735352,
      "learning_rate": 2.9795834388884034e-05,
      "loss": 0.04470396041870117,
      "step": 875
    },
    {
      "epoch": 0.11842839036755387,
      "grad_norm": 0.7273390293121338,
      "learning_rate": 2.979470742767417e-05,
      "loss": 0.09412288665771484,
      "step": 876
    },
    {
      "epoch": 0.11856358259400085,
      "grad_norm": 0.4011588394641876,
      "learning_rate": 2.9793577386127327e-05,
      "loss": 0.07372426986694336,
      "step": 877
    },
    {
      "epoch": 0.11869877482044783,
      "grad_norm": 0.4491732716560364,
      "learning_rate": 2.9792444264478784e-05,
      "loss": 0.08272123336791992,
      "step": 878
    },
    {
      "epoch": 0.1188339670468948,
      "grad_norm": 0.5146609544754028,
      "learning_rate": 2.979130806296446e-05,
      "loss": 0.08355045318603516,
      "step": 879
    },
    {
      "epoch": 0.11896915927334178,
      "grad_norm": 0.8118272423744202,
      "learning_rate": 2.9790168781820925e-05,
      "loss": 0.1037912368774414,
      "step": 880
    },
    {
      "epoch": 0.11910435149978876,
      "grad_norm": 1.625578761100769,
      "learning_rate": 2.9789026421285375e-05,
      "loss": 0.073211669921875,
      "step": 881
    },
    {
      "epoch": 0.11923954372623574,
      "grad_norm": 0.3579215705394745,
      "learning_rate": 2.9787880981595663e-05,
      "loss": 0.04483389854431152,
      "step": 882
    },
    {
      "epoch": 0.11937473595268272,
      "grad_norm": 0.6862708926200867,
      "learning_rate": 2.9786732462990267e-05,
      "loss": 0.07068490982055664,
      "step": 883
    },
    {
      "epoch": 0.1195099281791297,
      "grad_norm": 0.37091052532196045,
      "learning_rate": 2.9785580865708323e-05,
      "loss": 0.09660053253173828,
      "step": 884
    },
    {
      "epoch": 0.11964512040557668,
      "grad_norm": 0.5847275257110596,
      "learning_rate": 2.97844261899896e-05,
      "loss": 0.11170244216918945,
      "step": 885
    },
    {
      "epoch": 0.11978031263202366,
      "grad_norm": 0.4958404302597046,
      "learning_rate": 2.9783268436074495e-05,
      "loss": 0.06901073455810547,
      "step": 886
    },
    {
      "epoch": 0.11991550485847063,
      "grad_norm": 1.4115339517593384,
      "learning_rate": 2.978210760420407e-05,
      "loss": 0.10162115097045898,
      "step": 887
    },
    {
      "epoch": 0.12005069708491761,
      "grad_norm": 0.8564049601554871,
      "learning_rate": 2.978094369462002e-05,
      "loss": 0.05538368225097656,
      "step": 888
    },
    {
      "epoch": 0.12018588931136459,
      "grad_norm": 0.5000080466270447,
      "learning_rate": 2.977977670756467e-05,
      "loss": 0.0643768310546875,
      "step": 889
    },
    {
      "epoch": 0.12032108153781157,
      "grad_norm": 0.5024836659431458,
      "learning_rate": 2.9778606643280987e-05,
      "loss": 0.07504701614379883,
      "step": 890
    },
    {
      "epoch": 0.12045627376425856,
      "grad_norm": 0.6458346843719482,
      "learning_rate": 2.97774335020126e-05,
      "loss": 0.09175586700439453,
      "step": 891
    },
    {
      "epoch": 0.12059146599070554,
      "grad_norm": 0.6736217737197876,
      "learning_rate": 2.9776257284003748e-05,
      "loss": 0.07339859008789062,
      "step": 892
    },
    {
      "epoch": 0.12072665821715252,
      "grad_norm": 0.8182859420776367,
      "learning_rate": 2.9775077989499338e-05,
      "loss": 0.1093292236328125,
      "step": 893
    },
    {
      "epoch": 0.1208618504435995,
      "grad_norm": 0.5426833033561707,
      "learning_rate": 2.97738956187449e-05,
      "loss": 0.09246015548706055,
      "step": 894
    },
    {
      "epoch": 0.12099704267004648,
      "grad_norm": 0.6667600870132446,
      "learning_rate": 2.9772710171986605e-05,
      "loss": 0.054932594299316406,
      "step": 895
    },
    {
      "epoch": 0.12113223489649345,
      "grad_norm": 0.4842645525932312,
      "learning_rate": 2.977152164947128e-05,
      "loss": 0.0806894302368164,
      "step": 896
    },
    {
      "epoch": 0.12126742712294043,
      "grad_norm": 1.1293233633041382,
      "learning_rate": 2.9770330051446373e-05,
      "loss": 0.09871912002563477,
      "step": 897
    },
    {
      "epoch": 0.12140261934938741,
      "grad_norm": 0.521162211894989,
      "learning_rate": 2.976913537815999e-05,
      "loss": 0.07866871356964111,
      "step": 898
    },
    {
      "epoch": 0.12153781157583439,
      "grad_norm": 0.48777225613594055,
      "learning_rate": 2.9767937629860853e-05,
      "loss": 0.06045722961425781,
      "step": 899
    },
    {
      "epoch": 0.12167300380228137,
      "grad_norm": 0.654708206653595,
      "learning_rate": 2.9766736806798353e-05,
      "loss": 0.08138012886047363,
      "step": 900
    },
    {
      "epoch": 0.12180819602872835,
      "grad_norm": 0.6570261716842651,
      "learning_rate": 2.9765532909222512e-05,
      "loss": 0.07046079635620117,
      "step": 901
    },
    {
      "epoch": 0.12194338825517533,
      "grad_norm": 0.8348051309585571,
      "learning_rate": 2.976432593738397e-05,
      "loss": 0.06657648086547852,
      "step": 902
    },
    {
      "epoch": 0.1220785804816223,
      "grad_norm": 0.42024993896484375,
      "learning_rate": 2.9763115891534036e-05,
      "loss": 0.08733034133911133,
      "step": 903
    },
    {
      "epoch": 0.12221377270806928,
      "grad_norm": 0.9708475470542908,
      "learning_rate": 2.9761902771924648e-05,
      "loss": 0.07796192169189453,
      "step": 904
    },
    {
      "epoch": 0.12234896493451626,
      "grad_norm": 0.6434724926948547,
      "learning_rate": 2.9760686578808387e-05,
      "loss": 0.0627903938293457,
      "step": 905
    },
    {
      "epoch": 0.12248415716096324,
      "grad_norm": 0.8117854595184326,
      "learning_rate": 2.9759467312438462e-05,
      "loss": 0.0890951156616211,
      "step": 906
    },
    {
      "epoch": 0.12261934938741022,
      "grad_norm": 0.44805821776390076,
      "learning_rate": 2.975824497306874e-05,
      "loss": 0.0842752456665039,
      "step": 907
    },
    {
      "epoch": 0.12275454161385721,
      "grad_norm": 0.8622244000434875,
      "learning_rate": 2.9757019560953707e-05,
      "loss": 0.05887746810913086,
      "step": 908
    },
    {
      "epoch": 0.12288973384030419,
      "grad_norm": 0.5789123177528381,
      "learning_rate": 2.9755791076348517e-05,
      "loss": 0.07620429992675781,
      "step": 909
    },
    {
      "epoch": 0.12302492606675117,
      "grad_norm": 1.0534026622772217,
      "learning_rate": 2.9754559519508924e-05,
      "loss": 0.07729911804199219,
      "step": 910
    },
    {
      "epoch": 0.12316011829319815,
      "grad_norm": 1.0735238790512085,
      "learning_rate": 2.975332489069137e-05,
      "loss": 0.07388782501220703,
      "step": 911
    },
    {
      "epoch": 0.12329531051964512,
      "grad_norm": 1.1803566217422485,
      "learning_rate": 2.9752087190152893e-05,
      "loss": 0.10381555557250977,
      "step": 912
    },
    {
      "epoch": 0.1234305027460921,
      "grad_norm": 1.0864958763122559,
      "learning_rate": 2.97508464181512e-05,
      "loss": 0.07347965240478516,
      "step": 913
    },
    {
      "epoch": 0.12356569497253908,
      "grad_norm": 0.71308833360672,
      "learning_rate": 2.9749602574944615e-05,
      "loss": 0.05932760238647461,
      "step": 914
    },
    {
      "epoch": 0.12370088719898606,
      "grad_norm": 0.8423773646354675,
      "learning_rate": 2.9748355660792125e-05,
      "loss": 0.07395792007446289,
      "step": 915
    },
    {
      "epoch": 0.12383607942543304,
      "grad_norm": 0.5618091225624084,
      "learning_rate": 2.9747105675953338e-05,
      "loss": 0.09624385833740234,
      "step": 916
    },
    {
      "epoch": 0.12397127165188002,
      "grad_norm": 0.3507370054721832,
      "learning_rate": 2.9745852620688506e-05,
      "loss": 0.0835719108581543,
      "step": 917
    },
    {
      "epoch": 0.124106463878327,
      "grad_norm": 0.8579305410385132,
      "learning_rate": 2.974459649525853e-05,
      "loss": 0.10347938537597656,
      "step": 918
    },
    {
      "epoch": 0.12424165610477397,
      "grad_norm": 1.5583200454711914,
      "learning_rate": 2.9743337299924925e-05,
      "loss": 0.08157157897949219,
      "step": 919
    },
    {
      "epoch": 0.12437684833122095,
      "grad_norm": 1.6886459589004517,
      "learning_rate": 2.9742075034949883e-05,
      "loss": 0.13440799713134766,
      "step": 920
    },
    {
      "epoch": 0.12451204055766793,
      "grad_norm": 0.4353358745574951,
      "learning_rate": 2.97408097005962e-05,
      "loss": 0.06841325759887695,
      "step": 921
    },
    {
      "epoch": 0.12464723278411491,
      "grad_norm": 0.9028810858726501,
      "learning_rate": 2.973954129712733e-05,
      "loss": 0.10115814208984375,
      "step": 922
    },
    {
      "epoch": 0.12478242501056189,
      "grad_norm": 0.4330396056175232,
      "learning_rate": 2.973826982480736e-05,
      "loss": 0.07902836799621582,
      "step": 923
    },
    {
      "epoch": 0.12491761723700887,
      "grad_norm": 1.4390037059783936,
      "learning_rate": 2.9736995283901022e-05,
      "loss": 0.07700300216674805,
      "step": 924
    },
    {
      "epoch": 0.12505280946345584,
      "grad_norm": 0.6106219291687012,
      "learning_rate": 2.9735717674673676e-05,
      "loss": 0.09856510162353516,
      "step": 925
    },
    {
      "epoch": 0.12518800168990282,
      "grad_norm": 0.40409108996391296,
      "learning_rate": 2.973443699739133e-05,
      "loss": 0.08372306823730469,
      "step": 926
    },
    {
      "epoch": 0.1253231939163498,
      "grad_norm": 0.5649853944778442,
      "learning_rate": 2.973315325232063e-05,
      "loss": 0.05322265625,
      "step": 927
    },
    {
      "epoch": 0.12545838614279678,
      "grad_norm": 0.33068177103996277,
      "learning_rate": 2.9731866439728853e-05,
      "loss": 0.053058624267578125,
      "step": 928
    },
    {
      "epoch": 0.12559357836924376,
      "grad_norm": 0.5357032418251038,
      "learning_rate": 2.9730576559883924e-05,
      "loss": 0.07047724723815918,
      "step": 929
    },
    {
      "epoch": 0.12572877059569074,
      "grad_norm": 1.169585108757019,
      "learning_rate": 2.97292836130544e-05,
      "loss": 0.09200382232666016,
      "step": 930
    },
    {
      "epoch": 0.12586396282213771,
      "grad_norm": 0.6226010918617249,
      "learning_rate": 2.9727987599509485e-05,
      "loss": 0.07495355606079102,
      "step": 931
    },
    {
      "epoch": 0.1259991550485847,
      "grad_norm": 0.34719395637512207,
      "learning_rate": 2.972668851951901e-05,
      "loss": 0.061835527420043945,
      "step": 932
    },
    {
      "epoch": 0.12613434727503167,
      "grad_norm": 0.43001851439476013,
      "learning_rate": 2.9725386373353455e-05,
      "loss": 0.09643840789794922,
      "step": 933
    },
    {
      "epoch": 0.12626953950147868,
      "grad_norm": 0.632045328617096,
      "learning_rate": 2.972408116128393e-05,
      "loss": 0.07325077056884766,
      "step": 934
    },
    {
      "epoch": 0.12640473172792566,
      "grad_norm": 0.6281790733337402,
      "learning_rate": 2.972277288358219e-05,
      "loss": 0.05544567108154297,
      "step": 935
    },
    {
      "epoch": 0.12653992395437264,
      "grad_norm": 0.5434910655021667,
      "learning_rate": 2.9721461540520628e-05,
      "loss": 0.0962514877319336,
      "step": 936
    },
    {
      "epoch": 0.1266751161808196,
      "grad_norm": 0.338144451379776,
      "learning_rate": 2.9720147132372265e-05,
      "loss": 0.07651424407958984,
      "step": 937
    },
    {
      "epoch": 0.1268103084072666,
      "grad_norm": 0.3626541793346405,
      "learning_rate": 2.9718829659410772e-05,
      "loss": 0.059327125549316406,
      "step": 938
    },
    {
      "epoch": 0.12694550063371357,
      "grad_norm": 0.7962621450424194,
      "learning_rate": 2.9717509121910453e-05,
      "loss": 0.07543087005615234,
      "step": 939
    },
    {
      "epoch": 0.12708069286016055,
      "grad_norm": 0.3682408630847931,
      "learning_rate": 2.971618552014625e-05,
      "loss": 0.0697183609008789,
      "step": 940
    },
    {
      "epoch": 0.12721588508660753,
      "grad_norm": 0.334225058555603,
      "learning_rate": 2.971485885439375e-05,
      "loss": 0.04205179214477539,
      "step": 941
    },
    {
      "epoch": 0.1273510773130545,
      "grad_norm": 0.7661436200141907,
      "learning_rate": 2.9713529124929163e-05,
      "loss": 0.10788249969482422,
      "step": 942
    },
    {
      "epoch": 0.12748626953950148,
      "grad_norm": 0.4361717700958252,
      "learning_rate": 2.9712196332029352e-05,
      "loss": 0.060984134674072266,
      "step": 943
    },
    {
      "epoch": 0.12762146176594846,
      "grad_norm": 0.5988253951072693,
      "learning_rate": 2.971086047597181e-05,
      "loss": 0.1024322509765625,
      "step": 944
    },
    {
      "epoch": 0.12775665399239544,
      "grad_norm": 0.4751938283443451,
      "learning_rate": 2.9709521557034668e-05,
      "loss": 0.0825343132019043,
      "step": 945
    },
    {
      "epoch": 0.12789184621884242,
      "grad_norm": 0.26907584071159363,
      "learning_rate": 2.9708179575496696e-05,
      "loss": 0.05469369888305664,
      "step": 946
    },
    {
      "epoch": 0.1280270384452894,
      "grad_norm": 0.2958700954914093,
      "learning_rate": 2.9706834531637303e-05,
      "loss": 0.06627678871154785,
      "step": 947
    },
    {
      "epoch": 0.12816223067173638,
      "grad_norm": 0.3210330009460449,
      "learning_rate": 2.9705486425736537e-05,
      "loss": 0.06341791152954102,
      "step": 948
    },
    {
      "epoch": 0.12829742289818336,
      "grad_norm": 0.46173107624053955,
      "learning_rate": 2.9704135258075077e-05,
      "loss": 0.07384729385375977,
      "step": 949
    },
    {
      "epoch": 0.12843261512463033,
      "grad_norm": 0.48776519298553467,
      "learning_rate": 2.970278102893424e-05,
      "loss": 0.05780982971191406,
      "step": 950
    },
    {
      "epoch": 0.1285678073510773,
      "grad_norm": 0.23254428803920746,
      "learning_rate": 2.9701423738595992e-05,
      "loss": 0.037735939025878906,
      "step": 951
    },
    {
      "epoch": 0.1287029995775243,
      "grad_norm": 0.31969594955444336,
      "learning_rate": 2.9700063387342925e-05,
      "loss": 0.07977175712585449,
      "step": 952
    },
    {
      "epoch": 0.12883819180397127,
      "grad_norm": 0.27571967244148254,
      "learning_rate": 2.969869997545827e-05,
      "loss": 0.059992313385009766,
      "step": 953
    },
    {
      "epoch": 0.12897338403041825,
      "grad_norm": 0.734204888343811,
      "learning_rate": 2.9697333503225897e-05,
      "loss": 0.08205127716064453,
      "step": 954
    },
    {
      "epoch": 0.12910857625686523,
      "grad_norm": 0.296566367149353,
      "learning_rate": 2.969596397093031e-05,
      "loss": 0.06202888488769531,
      "step": 955
    },
    {
      "epoch": 0.1292437684833122,
      "grad_norm": 0.5508058667182922,
      "learning_rate": 2.969459137885666e-05,
      "loss": 0.1157083511352539,
      "step": 956
    },
    {
      "epoch": 0.12937896070975918,
      "grad_norm": 0.7532540559768677,
      "learning_rate": 2.969321572729072e-05,
      "loss": 0.07169175148010254,
      "step": 957
    },
    {
      "epoch": 0.12951415293620616,
      "grad_norm": 0.40385115146636963,
      "learning_rate": 2.9691837016518915e-05,
      "loss": 0.05229759216308594,
      "step": 958
    },
    {
      "epoch": 0.12964934516265314,
      "grad_norm": 0.44017350673675537,
      "learning_rate": 2.9690455246828294e-05,
      "loss": 0.06218385696411133,
      "step": 959
    },
    {
      "epoch": 0.12978453738910012,
      "grad_norm": 1.5186437368392944,
      "learning_rate": 2.968907041850655e-05,
      "loss": 0.07077431678771973,
      "step": 960
    },
    {
      "epoch": 0.1299197296155471,
      "grad_norm": 0.28694963455200195,
      "learning_rate": 2.968768253184202e-05,
      "loss": 0.05970430374145508,
      "step": 961
    },
    {
      "epoch": 0.13005492184199408,
      "grad_norm": 1.0450797080993652,
      "learning_rate": 2.9686291587123655e-05,
      "loss": 0.09966850280761719,
      "step": 962
    },
    {
      "epoch": 0.13019011406844105,
      "grad_norm": 0.2038547694683075,
      "learning_rate": 2.968489758464107e-05,
      "loss": 0.0590057373046875,
      "step": 963
    },
    {
      "epoch": 0.13032530629488803,
      "grad_norm": 0.43276289105415344,
      "learning_rate": 2.9683500524684494e-05,
      "loss": 0.08506965637207031,
      "step": 964
    },
    {
      "epoch": 0.130460498521335,
      "grad_norm": 0.5809456706047058,
      "learning_rate": 2.9682100407544812e-05,
      "loss": 0.05703449249267578,
      "step": 965
    },
    {
      "epoch": 0.130595690747782,
      "grad_norm": 0.41581490635871887,
      "learning_rate": 2.9680697233513526e-05,
      "loss": 0.07561635971069336,
      "step": 966
    },
    {
      "epoch": 0.13073088297422897,
      "grad_norm": 0.473766028881073,
      "learning_rate": 2.9679291002882793e-05,
      "loss": 0.06410503387451172,
      "step": 967
    },
    {
      "epoch": 0.13086607520067597,
      "grad_norm": 0.35944005846977234,
      "learning_rate": 2.967788171594539e-05,
      "loss": 0.05506134033203125,
      "step": 968
    },
    {
      "epoch": 0.13100126742712295,
      "grad_norm": 0.3487021028995514,
      "learning_rate": 2.967646937299474e-05,
      "loss": 0.059519290924072266,
      "step": 969
    },
    {
      "epoch": 0.13113645965356993,
      "grad_norm": 0.7498253583908081,
      "learning_rate": 2.9675053974324907e-05,
      "loss": 0.1223001480102539,
      "step": 970
    },
    {
      "epoch": 0.1312716518800169,
      "grad_norm": 0.3644426763057709,
      "learning_rate": 2.9673635520230576e-05,
      "loss": 0.06449270248413086,
      "step": 971
    },
    {
      "epoch": 0.1314068441064639,
      "grad_norm": 0.3195389211177826,
      "learning_rate": 2.9672214011007087e-05,
      "loss": 0.06031656265258789,
      "step": 972
    },
    {
      "epoch": 0.13154203633291087,
      "grad_norm": 0.9045985341072083,
      "learning_rate": 2.9670789446950396e-05,
      "loss": 0.1047515869140625,
      "step": 973
    },
    {
      "epoch": 0.13167722855935785,
      "grad_norm": 0.521141767501831,
      "learning_rate": 2.9669361828357105e-05,
      "loss": 0.07935333251953125,
      "step": 974
    },
    {
      "epoch": 0.13181242078580482,
      "grad_norm": 0.6143845319747925,
      "learning_rate": 2.9667931155524454e-05,
      "loss": 0.06447744369506836,
      "step": 975
    },
    {
      "epoch": 0.1319476130122518,
      "grad_norm": 0.5948580503463745,
      "learning_rate": 2.966649742875032e-05,
      "loss": 0.07799911499023438,
      "step": 976
    },
    {
      "epoch": 0.13208280523869878,
      "grad_norm": 0.6051185131072998,
      "learning_rate": 2.9665060648333206e-05,
      "loss": 0.05740940570831299,
      "step": 977
    },
    {
      "epoch": 0.13221799746514576,
      "grad_norm": 0.34990349411964417,
      "learning_rate": 2.9663620814572266e-05,
      "loss": 0.05348944664001465,
      "step": 978
    },
    {
      "epoch": 0.13235318969159274,
      "grad_norm": 0.9583558440208435,
      "learning_rate": 2.966217792776728e-05,
      "loss": 0.0702972412109375,
      "step": 979
    },
    {
      "epoch": 0.13248838191803972,
      "grad_norm": 0.45510542392730713,
      "learning_rate": 2.9660731988218652e-05,
      "loss": 0.05802106857299805,
      "step": 980
    },
    {
      "epoch": 0.1326235741444867,
      "grad_norm": 0.821110188961029,
      "learning_rate": 2.965928299622745e-05,
      "loss": 0.06091594696044922,
      "step": 981
    },
    {
      "epoch": 0.13275876637093367,
      "grad_norm": 0.4617513120174408,
      "learning_rate": 2.965783095209535e-05,
      "loss": 0.08968257904052734,
      "step": 982
    },
    {
      "epoch": 0.13289395859738065,
      "grad_norm": 1.1345301866531372,
      "learning_rate": 2.965637585612469e-05,
      "loss": 0.0957021713256836,
      "step": 983
    },
    {
      "epoch": 0.13302915082382763,
      "grad_norm": 1.1611295938491821,
      "learning_rate": 2.965491770861841e-05,
      "loss": 0.08506393432617188,
      "step": 984
    },
    {
      "epoch": 0.1331643430502746,
      "grad_norm": 0.5205171704292297,
      "learning_rate": 2.965345650988012e-05,
      "loss": 0.08159351348876953,
      "step": 985
    },
    {
      "epoch": 0.1332995352767216,
      "grad_norm": 0.4934007227420807,
      "learning_rate": 2.9651992260214035e-05,
      "loss": 0.07059884071350098,
      "step": 986
    },
    {
      "epoch": 0.13343472750316857,
      "grad_norm": 0.580374538898468,
      "learning_rate": 2.9650524959925037e-05,
      "loss": 0.1042940616607666,
      "step": 987
    },
    {
      "epoch": 0.13356991972961554,
      "grad_norm": 0.6456085443496704,
      "learning_rate": 2.9649054609318607e-05,
      "loss": 0.08425045013427734,
      "step": 988
    },
    {
      "epoch": 0.13370511195606252,
      "grad_norm": 0.5819563269615173,
      "learning_rate": 2.9647581208700894e-05,
      "loss": 0.12133598327636719,
      "step": 989
    },
    {
      "epoch": 0.1338403041825095,
      "grad_norm": 0.6495242714881897,
      "learning_rate": 2.9646104758378666e-05,
      "loss": 0.08918619155883789,
      "step": 990
    },
    {
      "epoch": 0.13397549640895648,
      "grad_norm": 0.5270083546638489,
      "learning_rate": 2.964462525865932e-05,
      "loss": 0.07123589515686035,
      "step": 991
    },
    {
      "epoch": 0.13411068863540346,
      "grad_norm": 0.361344575881958,
      "learning_rate": 2.96431427098509e-05,
      "loss": 0.06732702255249023,
      "step": 992
    },
    {
      "epoch": 0.13424588086185044,
      "grad_norm": 1.39292311668396,
      "learning_rate": 2.9641657112262084e-05,
      "loss": 0.12557554244995117,
      "step": 993
    },
    {
      "epoch": 0.13438107308829741,
      "grad_norm": 0.5411080121994019,
      "learning_rate": 2.9640168466202174e-05,
      "loss": 0.05614137649536133,
      "step": 994
    },
    {
      "epoch": 0.1345162653147444,
      "grad_norm": 0.24641117453575134,
      "learning_rate": 2.9638676771981124e-05,
      "loss": 0.03980302810668945,
      "step": 995
    },
    {
      "epoch": 0.13465145754119137,
      "grad_norm": 0.4695040285587311,
      "learning_rate": 2.9637182029909508e-05,
      "loss": 0.06638145446777344,
      "step": 996
    },
    {
      "epoch": 0.13478664976763835,
      "grad_norm": 0.8481863737106323,
      "learning_rate": 2.9635684240298532e-05,
      "loss": 0.09592628479003906,
      "step": 997
    },
    {
      "epoch": 0.13492184199408533,
      "grad_norm": 0.4733298122882843,
      "learning_rate": 2.9634183403460053e-05,
      "loss": 0.06079983711242676,
      "step": 998
    },
    {
      "epoch": 0.1350570342205323,
      "grad_norm": 0.7947366237640381,
      "learning_rate": 2.9632679519706553e-05,
      "loss": 0.06237435340881348,
      "step": 999
    },
    {
      "epoch": 0.13519222644697929,
      "grad_norm": 0.41676557064056396,
      "learning_rate": 2.9631172589351137e-05,
      "loss": 0.10515213012695312,
      "step": 1000
    },
    {
      "epoch": 0.1353274186734263,
      "grad_norm": 1.2089643478393555,
      "learning_rate": 2.962966261270758e-05,
      "loss": 0.11839675903320312,
      "step": 1001
    },
    {
      "epoch": 0.13546261089987327,
      "grad_norm": 0.3339465856552124,
      "learning_rate": 2.962814959009024e-05,
      "loss": 0.04536879062652588,
      "step": 1002
    },
    {
      "epoch": 0.13559780312632025,
      "grad_norm": 0.6421874761581421,
      "learning_rate": 2.962663352181415e-05,
      "loss": 0.08237600326538086,
      "step": 1003
    },
    {
      "epoch": 0.13573299535276723,
      "grad_norm": 0.8473460078239441,
      "learning_rate": 2.9625114408194966e-05,
      "loss": 0.08166265487670898,
      "step": 1004
    },
    {
      "epoch": 0.1358681875792142,
      "grad_norm": 0.45747676491737366,
      "learning_rate": 2.962359224954897e-05,
      "loss": 0.054253578186035156,
      "step": 1005
    },
    {
      "epoch": 0.13600337980566118,
      "grad_norm": 0.5543858408927917,
      "learning_rate": 2.9622067046193086e-05,
      "loss": 0.06447505950927734,
      "step": 1006
    },
    {
      "epoch": 0.13613857203210816,
      "grad_norm": 0.41753020882606506,
      "learning_rate": 2.9620538798444867e-05,
      "loss": 0.060455322265625,
      "step": 1007
    },
    {
      "epoch": 0.13627376425855514,
      "grad_norm": 0.5391937494277954,
      "learning_rate": 2.9619007506622506e-05,
      "loss": 0.12227058410644531,
      "step": 1008
    },
    {
      "epoch": 0.13640895648500212,
      "grad_norm": 0.2970508337020874,
      "learning_rate": 2.961747317104482e-05,
      "loss": 0.05994606018066406,
      "step": 1009
    },
    {
      "epoch": 0.1365441487114491,
      "grad_norm": 0.5311251878738403,
      "learning_rate": 2.9615935792031274e-05,
      "loss": 0.07413721084594727,
      "step": 1010
    },
    {
      "epoch": 0.13667934093789608,
      "grad_norm": 0.6452791690826416,
      "learning_rate": 2.9614395369901953e-05,
      "loss": 0.10640430450439453,
      "step": 1011
    },
    {
      "epoch": 0.13681453316434306,
      "grad_norm": 0.8512495160102844,
      "learning_rate": 2.9612851904977582e-05,
      "loss": 0.09710979461669922,
      "step": 1012
    },
    {
      "epoch": 0.13694972539079003,
      "grad_norm": 0.396820068359375,
      "learning_rate": 2.9611305397579518e-05,
      "loss": 0.10082101821899414,
      "step": 1013
    },
    {
      "epoch": 0.137084917617237,
      "grad_norm": 0.4907481372356415,
      "learning_rate": 2.9609755848029755e-05,
      "loss": 0.0582963228225708,
      "step": 1014
    },
    {
      "epoch": 0.137220109843684,
      "grad_norm": 0.48300814628601074,
      "learning_rate": 2.9608203256650916e-05,
      "loss": 0.07283365726470947,
      "step": 1015
    },
    {
      "epoch": 0.13735530207013097,
      "grad_norm": 0.401167094707489,
      "learning_rate": 2.9606647623766257e-05,
      "loss": 0.06013607978820801,
      "step": 1016
    },
    {
      "epoch": 0.13749049429657795,
      "grad_norm": 1.058401107788086,
      "learning_rate": 2.9605088949699672e-05,
      "loss": 0.08579373359680176,
      "step": 1017
    },
    {
      "epoch": 0.13762568652302493,
      "grad_norm": 0.44358688592910767,
      "learning_rate": 2.9603527234775682e-05,
      "loss": 0.08293008804321289,
      "step": 1018
    },
    {
      "epoch": 0.1377608787494719,
      "grad_norm": 0.3435921370983124,
      "learning_rate": 2.960196247931945e-05,
      "loss": 0.06587696075439453,
      "step": 1019
    },
    {
      "epoch": 0.13789607097591888,
      "grad_norm": 0.5250937342643738,
      "learning_rate": 2.960039468365676e-05,
      "loss": 0.09907770156860352,
      "step": 1020
    },
    {
      "epoch": 0.13803126320236586,
      "grad_norm": 0.4432773292064667,
      "learning_rate": 2.959882384811404e-05,
      "loss": 0.07782220840454102,
      "step": 1021
    },
    {
      "epoch": 0.13816645542881284,
      "grad_norm": 0.3793444335460663,
      "learning_rate": 2.9597249973018343e-05,
      "loss": 0.07200241088867188,
      "step": 1022
    },
    {
      "epoch": 0.13830164765525982,
      "grad_norm": 0.7196708917617798,
      "learning_rate": 2.959567305869736e-05,
      "loss": 0.07615351676940918,
      "step": 1023
    },
    {
      "epoch": 0.1384368398817068,
      "grad_norm": 0.5667684674263,
      "learning_rate": 2.9594093105479413e-05,
      "loss": 0.0909280776977539,
      "step": 1024
    },
    {
      "epoch": 0.13857203210815378,
      "grad_norm": 0.42288902401924133,
      "learning_rate": 2.959251011369345e-05,
      "loss": 0.05932807922363281,
      "step": 1025
    },
    {
      "epoch": 0.13870722433460075,
      "grad_norm": 0.4394952058792114,
      "learning_rate": 2.959092408366907e-05,
      "loss": 0.05600166320800781,
      "step": 1026
    },
    {
      "epoch": 0.13884241656104773,
      "grad_norm": 1.555326223373413,
      "learning_rate": 2.958933501573649e-05,
      "loss": 0.10719490051269531,
      "step": 1027
    },
    {
      "epoch": 0.1389776087874947,
      "grad_norm": 0.6938232183456421,
      "learning_rate": 2.9587742910226555e-05,
      "loss": 0.0561060905456543,
      "step": 1028
    },
    {
      "epoch": 0.1391128010139417,
      "grad_norm": 0.2859027087688446,
      "learning_rate": 2.958614776747076e-05,
      "loss": 0.05535411834716797,
      "step": 1029
    },
    {
      "epoch": 0.13924799324038867,
      "grad_norm": 0.5560369491577148,
      "learning_rate": 2.9584549587801213e-05,
      "loss": 0.105712890625,
      "step": 1030
    },
    {
      "epoch": 0.13938318546683565,
      "grad_norm": 0.6921489238739014,
      "learning_rate": 2.958294837155067e-05,
      "loss": 0.05954265594482422,
      "step": 1031
    },
    {
      "epoch": 0.13951837769328262,
      "grad_norm": 0.28092825412750244,
      "learning_rate": 2.9581344119052508e-05,
      "loss": 0.023319005966186523,
      "step": 1032
    },
    {
      "epoch": 0.1396535699197296,
      "grad_norm": 1.4917162656784058,
      "learning_rate": 2.957973683064074e-05,
      "loss": 0.11414146423339844,
      "step": 1033
    },
    {
      "epoch": 0.13978876214617658,
      "grad_norm": 0.5399127006530762,
      "learning_rate": 2.957812650665002e-05,
      "loss": 0.09021472930908203,
      "step": 1034
    },
    {
      "epoch": 0.1399239543726236,
      "grad_norm": 0.3196600675582886,
      "learning_rate": 2.957651314741562e-05,
      "loss": 0.06321144104003906,
      "step": 1035
    },
    {
      "epoch": 0.14005914659907057,
      "grad_norm": 0.5471014380455017,
      "learning_rate": 2.9574896753273454e-05,
      "loss": 0.07691097259521484,
      "step": 1036
    },
    {
      "epoch": 0.14019433882551754,
      "grad_norm": 0.7765745520591736,
      "learning_rate": 2.9573277324560058e-05,
      "loss": 0.10283279418945312,
      "step": 1037
    },
    {
      "epoch": 0.14032953105196452,
      "grad_norm": 0.5622662901878357,
      "learning_rate": 2.9571654861612608e-05,
      "loss": 0.04420614242553711,
      "step": 1038
    },
    {
      "epoch": 0.1404647232784115,
      "grad_norm": 0.3749459683895111,
      "learning_rate": 2.957002936476891e-05,
      "loss": 0.08757567405700684,
      "step": 1039
    },
    {
      "epoch": 0.14059991550485848,
      "grad_norm": 0.4691568911075592,
      "learning_rate": 2.9568400834367406e-05,
      "loss": 0.07784152030944824,
      "step": 1040
    },
    {
      "epoch": 0.14073510773130546,
      "grad_norm": 0.6638574600219727,
      "learning_rate": 2.9566769270747158e-05,
      "loss": 0.07189655303955078,
      "step": 1041
    },
    {
      "epoch": 0.14087029995775244,
      "grad_norm": 0.31720519065856934,
      "learning_rate": 2.9565134674247864e-05,
      "loss": 0.06950515508651733,
      "step": 1042
    },
    {
      "epoch": 0.14100549218419942,
      "grad_norm": 0.25245657563209534,
      "learning_rate": 2.9563497045209866e-05,
      "loss": 0.05792236328125,
      "step": 1043
    },
    {
      "epoch": 0.1411406844106464,
      "grad_norm": 0.7169446349143982,
      "learning_rate": 2.9561856383974118e-05,
      "loss": 0.10790157318115234,
      "step": 1044
    },
    {
      "epoch": 0.14127587663709337,
      "grad_norm": 0.22160176932811737,
      "learning_rate": 2.9560212690882218e-05,
      "loss": 0.05505990982055664,
      "step": 1045
    },
    {
      "epoch": 0.14141106886354035,
      "grad_norm": 0.5142950415611267,
      "learning_rate": 2.9558565966276395e-05,
      "loss": 0.07321357727050781,
      "step": 1046
    },
    {
      "epoch": 0.14154626108998733,
      "grad_norm": 0.5482933521270752,
      "learning_rate": 2.9556916210499497e-05,
      "loss": 0.08965778350830078,
      "step": 1047
    },
    {
      "epoch": 0.1416814533164343,
      "grad_norm": 1.486504077911377,
      "learning_rate": 2.9555263423895016e-05,
      "loss": 0.08441925048828125,
      "step": 1048
    },
    {
      "epoch": 0.1418166455428813,
      "grad_norm": 0.40179309248924255,
      "learning_rate": 2.955360760680708e-05,
      "loss": 0.07248306274414062,
      "step": 1049
    },
    {
      "epoch": 0.14195183776932827,
      "grad_norm": 0.19929270446300507,
      "learning_rate": 2.9551948759580423e-05,
      "loss": 0.04222297668457031,
      "step": 1050
    },
    {
      "epoch": 0.14208702999577524,
      "grad_norm": 1.3443044424057007,
      "learning_rate": 2.9550286882560435e-05,
      "loss": 0.08512163162231445,
      "step": 1051
    },
    {
      "epoch": 0.14222222222222222,
      "grad_norm": 0.6884334683418274,
      "learning_rate": 2.9548621976093126e-05,
      "loss": 0.08371925354003906,
      "step": 1052
    },
    {
      "epoch": 0.1423574144486692,
      "grad_norm": 0.5100288391113281,
      "learning_rate": 2.9546954040525144e-05,
      "loss": 0.07264518737792969,
      "step": 1053
    },
    {
      "epoch": 0.14249260667511618,
      "grad_norm": 0.3990485370159149,
      "learning_rate": 2.9545283076203753e-05,
      "loss": 0.060115814208984375,
      "step": 1054
    },
    {
      "epoch": 0.14262779890156316,
      "grad_norm": 1.017004132270813,
      "learning_rate": 2.954360908347686e-05,
      "loss": 0.09311389923095703,
      "step": 1055
    },
    {
      "epoch": 0.14276299112801014,
      "grad_norm": 0.6060172319412231,
      "learning_rate": 2.9541932062693e-05,
      "loss": 0.05374288558959961,
      "step": 1056
    },
    {
      "epoch": 0.14289818335445711,
      "grad_norm": 1.6509504318237305,
      "learning_rate": 2.954025201420134e-05,
      "loss": 0.0938262939453125,
      "step": 1057
    },
    {
      "epoch": 0.1430333755809041,
      "grad_norm": 1.1378108263015747,
      "learning_rate": 2.9538568938351672e-05,
      "loss": 0.07957077026367188,
      "step": 1058
    },
    {
      "epoch": 0.14316856780735107,
      "grad_norm": 0.31916308403015137,
      "learning_rate": 2.953688283549442e-05,
      "loss": 0.05724906921386719,
      "step": 1059
    },
    {
      "epoch": 0.14330376003379805,
      "grad_norm": 0.3913947343826294,
      "learning_rate": 2.9535193705980642e-05,
      "loss": 0.05498838424682617,
      "step": 1060
    },
    {
      "epoch": 0.14343895226024503,
      "grad_norm": 0.7348978519439697,
      "learning_rate": 2.9533501550162028e-05,
      "loss": 0.0771627426147461,
      "step": 1061
    },
    {
      "epoch": 0.143574144486692,
      "grad_norm": 0.5150848627090454,
      "learning_rate": 2.9531806368390882e-05,
      "loss": 0.07370626926422119,
      "step": 1062
    },
    {
      "epoch": 0.14370933671313899,
      "grad_norm": 1.1939650774002075,
      "learning_rate": 2.953010816102016e-05,
      "loss": 0.06557035446166992,
      "step": 1063
    },
    {
      "epoch": 0.14384452893958596,
      "grad_norm": 0.4618939757347107,
      "learning_rate": 2.952840692840343e-05,
      "loss": 0.09060502052307129,
      "step": 1064
    },
    {
      "epoch": 0.14397972116603294,
      "grad_norm": 0.4631308913230896,
      "learning_rate": 2.9526702670894914e-05,
      "loss": 0.09955406188964844,
      "step": 1065
    },
    {
      "epoch": 0.14411491339247992,
      "grad_norm": 0.4638875722885132,
      "learning_rate": 2.952499538884943e-05,
      "loss": 0.07295036315917969,
      "step": 1066
    },
    {
      "epoch": 0.1442501056189269,
      "grad_norm": 0.2289997935295105,
      "learning_rate": 2.9523285082622448e-05,
      "loss": 0.05170297622680664,
      "step": 1067
    },
    {
      "epoch": 0.14438529784537388,
      "grad_norm": 0.28436052799224854,
      "learning_rate": 2.9521571752570064e-05,
      "loss": 0.05417752265930176,
      "step": 1068
    },
    {
      "epoch": 0.14452049007182088,
      "grad_norm": 0.5871800184249878,
      "learning_rate": 2.9519855399049004e-05,
      "loss": 0.060787200927734375,
      "step": 1069
    },
    {
      "epoch": 0.14465568229826786,
      "grad_norm": 0.8053838014602661,
      "learning_rate": 2.951813602241662e-05,
      "loss": 0.09525346755981445,
      "step": 1070
    },
    {
      "epoch": 0.14479087452471484,
      "grad_norm": 0.5935882329940796,
      "learning_rate": 2.9516413623030896e-05,
      "loss": 0.09336566925048828,
      "step": 1071
    },
    {
      "epoch": 0.14492606675116182,
      "grad_norm": 0.7615500092506409,
      "learning_rate": 2.951468820125045e-05,
      "loss": 0.07540607452392578,
      "step": 1072
    },
    {
      "epoch": 0.1450612589776088,
      "grad_norm": 0.8269580006599426,
      "learning_rate": 2.9512959757434508e-05,
      "loss": 0.056622982025146484,
      "step": 1073
    },
    {
      "epoch": 0.14519645120405578,
      "grad_norm": 0.5052419304847717,
      "learning_rate": 2.951122829194296e-05,
      "loss": 0.08120441436767578,
      "step": 1074
    },
    {
      "epoch": 0.14533164343050275,
      "grad_norm": 0.4648810625076294,
      "learning_rate": 2.9509493805136296e-05,
      "loss": 0.07573890686035156,
      "step": 1075
    },
    {
      "epoch": 0.14546683565694973,
      "grad_norm": 0.43185341358184814,
      "learning_rate": 2.9507756297375648e-05,
      "loss": 0.10009193420410156,
      "step": 1076
    },
    {
      "epoch": 0.1456020278833967,
      "grad_norm": 0.5906679034233093,
      "learning_rate": 2.9506015769022778e-05,
      "loss": 0.1013031005859375,
      "step": 1077
    },
    {
      "epoch": 0.1457372201098437,
      "grad_norm": 0.5084944367408752,
      "learning_rate": 2.950427222044006e-05,
      "loss": 0.08063554763793945,
      "step": 1078
    },
    {
      "epoch": 0.14587241233629067,
      "grad_norm": 0.5056231021881104,
      "learning_rate": 2.9502525651990525e-05,
      "loss": 0.1089935302734375,
      "step": 1079
    },
    {
      "epoch": 0.14600760456273765,
      "grad_norm": 0.4640476703643799,
      "learning_rate": 2.9500776064037813e-05,
      "loss": 0.05266237258911133,
      "step": 1080
    },
    {
      "epoch": 0.14614279678918463,
      "grad_norm": 0.37210893630981445,
      "learning_rate": 2.9499023456946194e-05,
      "loss": 0.06466007232666016,
      "step": 1081
    },
    {
      "epoch": 0.1462779890156316,
      "grad_norm": 0.5744715929031372,
      "learning_rate": 2.9497267831080575e-05,
      "loss": 0.065887451171875,
      "step": 1082
    },
    {
      "epoch": 0.14641318124207858,
      "grad_norm": 0.5396444201469421,
      "learning_rate": 2.949550918680649e-05,
      "loss": 0.07608509063720703,
      "step": 1083
    },
    {
      "epoch": 0.14654837346852556,
      "grad_norm": 0.3733367919921875,
      "learning_rate": 2.9493747524490086e-05,
      "loss": 0.07193183898925781,
      "step": 1084
    },
    {
      "epoch": 0.14668356569497254,
      "grad_norm": 0.3441978693008423,
      "learning_rate": 2.9491982844498156e-05,
      "loss": 0.06608152389526367,
      "step": 1085
    },
    {
      "epoch": 0.14681875792141952,
      "grad_norm": 0.49852749705314636,
      "learning_rate": 2.949021514719812e-05,
      "loss": 0.08067667484283447,
      "step": 1086
    },
    {
      "epoch": 0.1469539501478665,
      "grad_norm": 0.4173051118850708,
      "learning_rate": 2.948844443295802e-05,
      "loss": 0.0680074691772461,
      "step": 1087
    },
    {
      "epoch": 0.14708914237431348,
      "grad_norm": 0.7929953336715698,
      "learning_rate": 2.9486670702146526e-05,
      "loss": 0.11616706848144531,
      "step": 1088
    },
    {
      "epoch": 0.14722433460076045,
      "grad_norm": 0.8731603026390076,
      "learning_rate": 2.948489395513294e-05,
      "loss": 0.0901632308959961,
      "step": 1089
    },
    {
      "epoch": 0.14735952682720743,
      "grad_norm": 0.7319706678390503,
      "learning_rate": 2.948311419228719e-05,
      "loss": 0.11249923706054688,
      "step": 1090
    },
    {
      "epoch": 0.1474947190536544,
      "grad_norm": 0.907683789730072,
      "learning_rate": 2.948133141397983e-05,
      "loss": 0.0464324951171875,
      "step": 1091
    },
    {
      "epoch": 0.1476299112801014,
      "grad_norm": 0.5078043937683105,
      "learning_rate": 2.9479545620582047e-05,
      "loss": 0.0528569221496582,
      "step": 1092
    },
    {
      "epoch": 0.14776510350654837,
      "grad_norm": 0.421255886554718,
      "learning_rate": 2.9477756812465652e-05,
      "loss": 0.08293342590332031,
      "step": 1093
    },
    {
      "epoch": 0.14790029573299535,
      "grad_norm": 0.40994036197662354,
      "learning_rate": 2.9475964990003085e-05,
      "loss": 0.03383135795593262,
      "step": 1094
    },
    {
      "epoch": 0.14803548795944232,
      "grad_norm": 0.4158625900745392,
      "learning_rate": 2.9474170153567406e-05,
      "loss": 0.07805156707763672,
      "step": 1095
    },
    {
      "epoch": 0.1481706801858893,
      "grad_norm": 0.7317113876342773,
      "learning_rate": 2.947237230353232e-05,
      "loss": 0.1040964126586914,
      "step": 1096
    },
    {
      "epoch": 0.14830587241233628,
      "grad_norm": 0.23181234300136566,
      "learning_rate": 2.9470571440272147e-05,
      "loss": 0.05652332305908203,
      "step": 1097
    },
    {
      "epoch": 0.14844106463878326,
      "grad_norm": 1.2110581398010254,
      "learning_rate": 2.946876756416183e-05,
      "loss": 0.06962728500366211,
      "step": 1098
    },
    {
      "epoch": 0.14857625686523024,
      "grad_norm": 0.9232327342033386,
      "learning_rate": 2.946696067557695e-05,
      "loss": 0.07483863830566406,
      "step": 1099
    },
    {
      "epoch": 0.14871144909167722,
      "grad_norm": 0.2956840395927429,
      "learning_rate": 2.9465150774893706e-05,
      "loss": 0.06692075729370117,
      "step": 1100
    },
    {
      "epoch": 0.1488466413181242,
      "grad_norm": 0.21126961708068848,
      "learning_rate": 2.9463337862488938e-05,
      "loss": 0.04665088653564453,
      "step": 1101
    },
    {
      "epoch": 0.14898183354457117,
      "grad_norm": 1.7077643871307373,
      "learning_rate": 2.9461521938740096e-05,
      "loss": 0.09197998046875,
      "step": 1102
    },
    {
      "epoch": 0.14911702577101818,
      "grad_norm": 0.6724783182144165,
      "learning_rate": 2.9459703004025273e-05,
      "loss": 0.08422660827636719,
      "step": 1103
    },
    {
      "epoch": 0.14925221799746516,
      "grad_norm": 1.316791296005249,
      "learning_rate": 2.9457881058723174e-05,
      "loss": 0.10529279708862305,
      "step": 1104
    },
    {
      "epoch": 0.14938741022391214,
      "grad_norm": 0.42805215716362,
      "learning_rate": 2.9456056103213137e-05,
      "loss": 0.09313654899597168,
      "step": 1105
    },
    {
      "epoch": 0.14952260245035912,
      "grad_norm": 0.657561182975769,
      "learning_rate": 2.945422813787513e-05,
      "loss": 0.052576541900634766,
      "step": 1106
    },
    {
      "epoch": 0.1496577946768061,
      "grad_norm": 0.41958561539649963,
      "learning_rate": 2.9452397163089748e-05,
      "loss": 0.04102897644042969,
      "step": 1107
    },
    {
      "epoch": 0.14979298690325307,
      "grad_norm": 0.6558903455734253,
      "learning_rate": 2.9450563179238207e-05,
      "loss": 0.0634920597076416,
      "step": 1108
    },
    {
      "epoch": 0.14992817912970005,
      "grad_norm": 0.9969330430030823,
      "learning_rate": 2.9448726186702354e-05,
      "loss": 0.07057714462280273,
      "step": 1109
    },
    {
      "epoch": 0.15006337135614703,
      "grad_norm": 0.4252939224243164,
      "learning_rate": 2.9446886185864652e-05,
      "loss": 0.0529322624206543,
      "step": 1110
    },
    {
      "epoch": 0.150198563582594,
      "grad_norm": 0.46789953112602234,
      "learning_rate": 2.944504317710821e-05,
      "loss": 0.07893943786621094,
      "step": 1111
    },
    {
      "epoch": 0.150333755809041,
      "grad_norm": 1.2787641286849976,
      "learning_rate": 2.944319716081675e-05,
      "loss": 0.08886337280273438,
      "step": 1112
    },
    {
      "epoch": 0.15046894803548796,
      "grad_norm": 0.5518364906311035,
      "learning_rate": 2.944134813737462e-05,
      "loss": 0.08701801300048828,
      "step": 1113
    },
    {
      "epoch": 0.15060414026193494,
      "grad_norm": 0.2912185490131378,
      "learning_rate": 2.9439496107166796e-05,
      "loss": 0.0686025619506836,
      "step": 1114
    },
    {
      "epoch": 0.15073933248838192,
      "grad_norm": 0.4945691227912903,
      "learning_rate": 2.943764107057888e-05,
      "loss": 0.10154247283935547,
      "step": 1115
    },
    {
      "epoch": 0.1508745247148289,
      "grad_norm": 0.4553638696670532,
      "learning_rate": 2.9435783027997106e-05,
      "loss": 0.09839916229248047,
      "step": 1116
    },
    {
      "epoch": 0.15100971694127588,
      "grad_norm": 0.43696364760398865,
      "learning_rate": 2.9433921979808323e-05,
      "loss": 0.0523834228515625,
      "step": 1117
    },
    {
      "epoch": 0.15114490916772286,
      "grad_norm": 0.49646279215812683,
      "learning_rate": 2.9432057926400014e-05,
      "loss": 0.06965899467468262,
      "step": 1118
    },
    {
      "epoch": 0.15128010139416984,
      "grad_norm": 0.37695229053497314,
      "learning_rate": 2.943019086816028e-05,
      "loss": 0.06361579895019531,
      "step": 1119
    },
    {
      "epoch": 0.15141529362061681,
      "grad_norm": 0.34943926334381104,
      "learning_rate": 2.9428320805477855e-05,
      "loss": 0.059767723083496094,
      "step": 1120
    },
    {
      "epoch": 0.1515504858470638,
      "grad_norm": 0.72733074426651,
      "learning_rate": 2.9426447738742104e-05,
      "loss": 0.09514474868774414,
      "step": 1121
    },
    {
      "epoch": 0.15168567807351077,
      "grad_norm": 0.23378986120224,
      "learning_rate": 2.9424571668343e-05,
      "loss": 0.05317068099975586,
      "step": 1122
    },
    {
      "epoch": 0.15182087029995775,
      "grad_norm": 0.3431641757488251,
      "learning_rate": 2.942269259467115e-05,
      "loss": 0.07488346099853516,
      "step": 1123
    },
    {
      "epoch": 0.15195606252640473,
      "grad_norm": 0.5362439751625061,
      "learning_rate": 2.9420810518117794e-05,
      "loss": 0.04477691650390625,
      "step": 1124
    },
    {
      "epoch": 0.1520912547528517,
      "grad_norm": 0.5341426134109497,
      "learning_rate": 2.9418925439074784e-05,
      "loss": 0.09751033782958984,
      "step": 1125
    },
    {
      "epoch": 0.15222644697929869,
      "grad_norm": 0.393439918756485,
      "learning_rate": 2.9417037357934606e-05,
      "loss": 0.0592045783996582,
      "step": 1126
    },
    {
      "epoch": 0.15236163920574566,
      "grad_norm": 0.8095723390579224,
      "learning_rate": 2.9415146275090373e-05,
      "loss": 0.0764150619506836,
      "step": 1127
    },
    {
      "epoch": 0.15249683143219264,
      "grad_norm": 0.9617822170257568,
      "learning_rate": 2.9413252190935813e-05,
      "loss": 0.07028436660766602,
      "step": 1128
    },
    {
      "epoch": 0.15263202365863962,
      "grad_norm": 0.29926449060440063,
      "learning_rate": 2.9411355105865286e-05,
      "loss": 0.07842016220092773,
      "step": 1129
    },
    {
      "epoch": 0.1527672158850866,
      "grad_norm": 0.4743107259273529,
      "learning_rate": 2.9409455020273775e-05,
      "loss": 0.10325860977172852,
      "step": 1130
    },
    {
      "epoch": 0.15290240811153358,
      "grad_norm": 0.37752461433410645,
      "learning_rate": 2.940755193455689e-05,
      "loss": 0.06924009323120117,
      "step": 1131
    },
    {
      "epoch": 0.15303760033798056,
      "grad_norm": 0.901729166507721,
      "learning_rate": 2.940564584911086e-05,
      "loss": 0.07530999183654785,
      "step": 1132
    },
    {
      "epoch": 0.15317279256442753,
      "grad_norm": 0.4585990011692047,
      "learning_rate": 2.9403736764332543e-05,
      "loss": 0.06424522399902344,
      "step": 1133
    },
    {
      "epoch": 0.1533079847908745,
      "grad_norm": 0.5100941061973572,
      "learning_rate": 2.9401824680619423e-05,
      "loss": 0.09717559814453125,
      "step": 1134
    },
    {
      "epoch": 0.1534431770173215,
      "grad_norm": 0.38628143072128296,
      "learning_rate": 2.9399909598369604e-05,
      "loss": 0.07665491104125977,
      "step": 1135
    },
    {
      "epoch": 0.15357836924376847,
      "grad_norm": 0.6921217441558838,
      "learning_rate": 2.939799151798182e-05,
      "loss": 0.07894515991210938,
      "step": 1136
    },
    {
      "epoch": 0.15371356147021548,
      "grad_norm": 0.510297417640686,
      "learning_rate": 2.9396070439855417e-05,
      "loss": 0.09485530853271484,
      "step": 1137
    },
    {
      "epoch": 0.15384875369666245,
      "grad_norm": 0.42599210143089294,
      "learning_rate": 2.9394146364390382e-05,
      "loss": 0.08612918853759766,
      "step": 1138
    },
    {
      "epoch": 0.15398394592310943,
      "grad_norm": 0.4523150622844696,
      "learning_rate": 2.9392219291987315e-05,
      "loss": 0.08692288398742676,
      "step": 1139
    },
    {
      "epoch": 0.1541191381495564,
      "grad_norm": 0.7739888429641724,
      "learning_rate": 2.939028922304744e-05,
      "loss": 0.06455564498901367,
      "step": 1140
    },
    {
      "epoch": 0.1542543303760034,
      "grad_norm": 0.7813996076583862,
      "learning_rate": 2.9388356157972615e-05,
      "loss": 0.07198441028594971,
      "step": 1141
    },
    {
      "epoch": 0.15438952260245037,
      "grad_norm": 0.3456525206565857,
      "learning_rate": 2.938642009716531e-05,
      "loss": 0.0829610824584961,
      "step": 1142
    },
    {
      "epoch": 0.15452471482889735,
      "grad_norm": 0.3205811381340027,
      "learning_rate": 2.938448104102862e-05,
      "loss": 0.06907987594604492,
      "step": 1143
    },
    {
      "epoch": 0.15465990705534433,
      "grad_norm": 0.3885345160961151,
      "learning_rate": 2.9382538989966267e-05,
      "loss": 0.08256053924560547,
      "step": 1144
    },
    {
      "epoch": 0.1547950992817913,
      "grad_norm": 0.29711881279945374,
      "learning_rate": 2.9380593944382605e-05,
      "loss": 0.03909134864807129,
      "step": 1145
    },
    {
      "epoch": 0.15493029150823828,
      "grad_norm": 0.47773250937461853,
      "learning_rate": 2.9378645904682596e-05,
      "loss": 0.05340099334716797,
      "step": 1146
    },
    {
      "epoch": 0.15506548373468526,
      "grad_norm": 0.7147215604782104,
      "learning_rate": 2.937669487127183e-05,
      "loss": 0.07366132736206055,
      "step": 1147
    },
    {
      "epoch": 0.15520067596113224,
      "grad_norm": 0.7375534176826477,
      "learning_rate": 2.9374740844556532e-05,
      "loss": 0.08477926254272461,
      "step": 1148
    },
    {
      "epoch": 0.15533586818757922,
      "grad_norm": 0.4000887870788574,
      "learning_rate": 2.937278382494353e-05,
      "loss": 0.08346080780029297,
      "step": 1149
    },
    {
      "epoch": 0.1554710604140262,
      "grad_norm": 0.3509191870689392,
      "learning_rate": 2.9370823812840287e-05,
      "loss": 0.08596360683441162,
      "step": 1150
    },
    {
      "epoch": 0.15560625264047317,
      "grad_norm": 0.8168505430221558,
      "learning_rate": 2.93688608086549e-05,
      "loss": 0.07916259765625,
      "step": 1151
    },
    {
      "epoch": 0.15574144486692015,
      "grad_norm": 0.22475913166999817,
      "learning_rate": 2.9366894812796064e-05,
      "loss": 0.06238365173339844,
      "step": 1152
    },
    {
      "epoch": 0.15587663709336713,
      "grad_norm": 0.5967375040054321,
      "learning_rate": 2.9364925825673117e-05,
      "loss": 0.06840991973876953,
      "step": 1153
    },
    {
      "epoch": 0.1560118293198141,
      "grad_norm": 0.6725037693977356,
      "learning_rate": 2.9362953847696006e-05,
      "loss": 0.0830678939819336,
      "step": 1154
    },
    {
      "epoch": 0.1561470215462611,
      "grad_norm": 0.9402245283126831,
      "learning_rate": 2.9360978879275313e-05,
      "loss": 0.08952546119689941,
      "step": 1155
    },
    {
      "epoch": 0.15628221377270807,
      "grad_norm": 0.571772038936615,
      "learning_rate": 2.9359000920822237e-05,
      "loss": 0.06865835189819336,
      "step": 1156
    },
    {
      "epoch": 0.15641740599915505,
      "grad_norm": 0.5571675300598145,
      "learning_rate": 2.9357019972748594e-05,
      "loss": 0.06963920593261719,
      "step": 1157
    },
    {
      "epoch": 0.15655259822560202,
      "grad_norm": 0.4392727017402649,
      "learning_rate": 2.9355036035466836e-05,
      "loss": 0.0725712776184082,
      "step": 1158
    },
    {
      "epoch": 0.156687790452049,
      "grad_norm": 0.9154953956604004,
      "learning_rate": 2.935304910939002e-05,
      "loss": 0.08074283599853516,
      "step": 1159
    },
    {
      "epoch": 0.15682298267849598,
      "grad_norm": 1.5962560176849365,
      "learning_rate": 2.935105919493184e-05,
      "loss": 0.09226322174072266,
      "step": 1160
    },
    {
      "epoch": 0.15695817490494296,
      "grad_norm": 0.28233760595321655,
      "learning_rate": 2.9349066292506613e-05,
      "loss": 0.07190513610839844,
      "step": 1161
    },
    {
      "epoch": 0.15709336713138994,
      "grad_norm": 0.43321600556373596,
      "learning_rate": 2.934707040252926e-05,
      "loss": 0.08964920043945312,
      "step": 1162
    },
    {
      "epoch": 0.15722855935783692,
      "grad_norm": 0.37806591391563416,
      "learning_rate": 2.9345071525415342e-05,
      "loss": 0.10108518600463867,
      "step": 1163
    },
    {
      "epoch": 0.1573637515842839,
      "grad_norm": 0.6121034026145935,
      "learning_rate": 2.9343069661581035e-05,
      "loss": 0.0862569808959961,
      "step": 1164
    },
    {
      "epoch": 0.15749894381073087,
      "grad_norm": 1.3534986972808838,
      "learning_rate": 2.9341064811443138e-05,
      "loss": 0.10041403770446777,
      "step": 1165
    },
    {
      "epoch": 0.15763413603717785,
      "grad_norm": 0.6392109394073486,
      "learning_rate": 2.9339056975419078e-05,
      "loss": 0.0789952278137207,
      "step": 1166
    },
    {
      "epoch": 0.15776932826362483,
      "grad_norm": 1.1749188899993896,
      "learning_rate": 2.9337046153926882e-05,
      "loss": 0.10968422889709473,
      "step": 1167
    },
    {
      "epoch": 0.1579045204900718,
      "grad_norm": 0.2618032693862915,
      "learning_rate": 2.9335032347385224e-05,
      "loss": 0.04763603210449219,
      "step": 1168
    },
    {
      "epoch": 0.1580397127165188,
      "grad_norm": 0.8895334005355835,
      "learning_rate": 2.933301555621339e-05,
      "loss": 0.10931873321533203,
      "step": 1169
    },
    {
      "epoch": 0.15817490494296577,
      "grad_norm": 1.1451064348220825,
      "learning_rate": 2.933099578083128e-05,
      "loss": 0.08714914321899414,
      "step": 1170
    },
    {
      "epoch": 0.15831009716941277,
      "grad_norm": 1.0426045656204224,
      "learning_rate": 2.932897302165943e-05,
      "loss": 0.12275981903076172,
      "step": 1171
    },
    {
      "epoch": 0.15844528939585975,
      "grad_norm": 1.392737627029419,
      "learning_rate": 2.9326947279118983e-05,
      "loss": 0.08675193786621094,
      "step": 1172
    },
    {
      "epoch": 0.15858048162230673,
      "grad_norm": 0.6134676337242126,
      "learning_rate": 2.9324918553631716e-05,
      "loss": 0.06651067733764648,
      "step": 1173
    },
    {
      "epoch": 0.1587156738487537,
      "grad_norm": 0.4304104447364807,
      "learning_rate": 2.9322886845620013e-05,
      "loss": 0.07193470001220703,
      "step": 1174
    },
    {
      "epoch": 0.1588508660752007,
      "grad_norm": 0.47915732860565186,
      "learning_rate": 2.932085215550689e-05,
      "loss": 0.06539511680603027,
      "step": 1175
    },
    {
      "epoch": 0.15898605830164766,
      "grad_norm": 0.35044535994529724,
      "learning_rate": 2.9318814483715982e-05,
      "loss": 0.09577655792236328,
      "step": 1176
    },
    {
      "epoch": 0.15912125052809464,
      "grad_norm": 0.5289446115493774,
      "learning_rate": 2.9316773830671537e-05,
      "loss": 0.05567789077758789,
      "step": 1177
    },
    {
      "epoch": 0.15925644275454162,
      "grad_norm": 1.16676926612854,
      "learning_rate": 2.9314730196798437e-05,
      "loss": 0.07228612899780273,
      "step": 1178
    },
    {
      "epoch": 0.1593916349809886,
      "grad_norm": 0.33163681626319885,
      "learning_rate": 2.9312683582522178e-05,
      "loss": 0.06267356872558594,
      "step": 1179
    },
    {
      "epoch": 0.15952682720743558,
      "grad_norm": 0.47903668880462646,
      "learning_rate": 2.9310633988268868e-05,
      "loss": 0.07154417037963867,
      "step": 1180
    },
    {
      "epoch": 0.15966201943388256,
      "grad_norm": 0.445639431476593,
      "learning_rate": 2.9308581414465246e-05,
      "loss": 0.05153632164001465,
      "step": 1181
    },
    {
      "epoch": 0.15979721166032954,
      "grad_norm": 0.2278568148612976,
      "learning_rate": 2.9306525861538674e-05,
      "loss": 0.06707310676574707,
      "step": 1182
    },
    {
      "epoch": 0.15993240388677651,
      "grad_norm": 0.36722877621650696,
      "learning_rate": 2.9304467329917127e-05,
      "loss": 0.06810903549194336,
      "step": 1183
    },
    {
      "epoch": 0.1600675961132235,
      "grad_norm": 0.5402987599372864,
      "learning_rate": 2.9302405820029198e-05,
      "loss": 0.07645750045776367,
      "step": 1184
    },
    {
      "epoch": 0.16020278833967047,
      "grad_norm": 0.7956507205963135,
      "learning_rate": 2.9300341332304114e-05,
      "loss": 0.102630615234375,
      "step": 1185
    },
    {
      "epoch": 0.16033798056611745,
      "grad_norm": 0.47227296233177185,
      "learning_rate": 2.9298273867171697e-05,
      "loss": 0.0795888900756836,
      "step": 1186
    },
    {
      "epoch": 0.16047317279256443,
      "grad_norm": 0.48212334513664246,
      "learning_rate": 2.929620342506242e-05,
      "loss": 0.07121658325195312,
      "step": 1187
    },
    {
      "epoch": 0.1606083650190114,
      "grad_norm": 0.5086870789527893,
      "learning_rate": 2.929413000640735e-05,
      "loss": 0.07348346710205078,
      "step": 1188
    },
    {
      "epoch": 0.16074355724545838,
      "grad_norm": 0.40089017152786255,
      "learning_rate": 2.9292053611638187e-05,
      "loss": 0.06230783462524414,
      "step": 1189
    },
    {
      "epoch": 0.16087874947190536,
      "grad_norm": 0.7533932328224182,
      "learning_rate": 2.928997424118725e-05,
      "loss": 0.08992528915405273,
      "step": 1190
    },
    {
      "epoch": 0.16101394169835234,
      "grad_norm": 0.37295597791671753,
      "learning_rate": 2.928789189548747e-05,
      "loss": 0.09594249725341797,
      "step": 1191
    },
    {
      "epoch": 0.16114913392479932,
      "grad_norm": 0.5290022492408752,
      "learning_rate": 2.9285806574972405e-05,
      "loss": 0.08588194847106934,
      "step": 1192
    },
    {
      "epoch": 0.1612843261512463,
      "grad_norm": 0.3387167155742645,
      "learning_rate": 2.928371828007623e-05,
      "loss": 0.04946088790893555,
      "step": 1193
    },
    {
      "epoch": 0.16141951837769328,
      "grad_norm": 0.3090049624443054,
      "learning_rate": 2.928162701123374e-05,
      "loss": 0.06363749504089355,
      "step": 1194
    },
    {
      "epoch": 0.16155471060414026,
      "grad_norm": 0.7510243654251099,
      "learning_rate": 2.9279532768880345e-05,
      "loss": 0.08713436126708984,
      "step": 1195
    },
    {
      "epoch": 0.16168990283058723,
      "grad_norm": 0.2118089497089386,
      "learning_rate": 2.9277435553452084e-05,
      "loss": 0.04293251037597656,
      "step": 1196
    },
    {
      "epoch": 0.1618250950570342,
      "grad_norm": 0.2888762056827545,
      "learning_rate": 2.9275335365385602e-05,
      "loss": 0.07376575469970703,
      "step": 1197
    },
    {
      "epoch": 0.1619602872834812,
      "grad_norm": 0.5952282547950745,
      "learning_rate": 2.927323220511817e-05,
      "loss": 0.08271503448486328,
      "step": 1198
    },
    {
      "epoch": 0.16209547950992817,
      "grad_norm": 0.39375627040863037,
      "learning_rate": 2.9271126073087684e-05,
      "loss": 0.0668494701385498,
      "step": 1199
    },
    {
      "epoch": 0.16223067173637515,
      "grad_norm": 0.591256856918335,
      "learning_rate": 2.926901696973264e-05,
      "loss": 0.0479205846786499,
      "step": 1200
    },
    {
      "epoch": 0.16236586396282213,
      "grad_norm": 0.5941855311393738,
      "learning_rate": 2.9266904895492177e-05,
      "loss": 0.07880735397338867,
      "step": 1201
    },
    {
      "epoch": 0.1625010561892691,
      "grad_norm": 0.5759038329124451,
      "learning_rate": 2.926478985080603e-05,
      "loss": 0.07346010208129883,
      "step": 1202
    },
    {
      "epoch": 0.16263624841571608,
      "grad_norm": 0.39038926362991333,
      "learning_rate": 2.9262671836114568e-05,
      "loss": 0.05943942070007324,
      "step": 1203
    },
    {
      "epoch": 0.16277144064216306,
      "grad_norm": 0.5903465747833252,
      "learning_rate": 2.9260550851858774e-05,
      "loss": 0.07378315925598145,
      "step": 1204
    },
    {
      "epoch": 0.16290663286861007,
      "grad_norm": 0.5197759866714478,
      "learning_rate": 2.9258426898480243e-05,
      "loss": 0.09259724617004395,
      "step": 1205
    },
    {
      "epoch": 0.16304182509505705,
      "grad_norm": 0.46826356649398804,
      "learning_rate": 2.9256299976421198e-05,
      "loss": 0.07935118675231934,
      "step": 1206
    },
    {
      "epoch": 0.16317701732150403,
      "grad_norm": 0.39055630564689636,
      "learning_rate": 2.9254170086124474e-05,
      "loss": 0.07779121398925781,
      "step": 1207
    },
    {
      "epoch": 0.163312209547951,
      "grad_norm": 0.5419468283653259,
      "learning_rate": 2.9252037228033526e-05,
      "loss": 0.1074373722076416,
      "step": 1208
    },
    {
      "epoch": 0.16344740177439798,
      "grad_norm": 0.4405940771102905,
      "learning_rate": 2.9249901402592424e-05,
      "loss": 0.0735478401184082,
      "step": 1209
    },
    {
      "epoch": 0.16358259400084496,
      "grad_norm": 0.7484593987464905,
      "learning_rate": 2.9247762610245863e-05,
      "loss": 0.08356046676635742,
      "step": 1210
    },
    {
      "epoch": 0.16371778622729194,
      "grad_norm": 0.418267160654068,
      "learning_rate": 2.9245620851439146e-05,
      "loss": 0.04102802276611328,
      "step": 1211
    },
    {
      "epoch": 0.16385297845373892,
      "grad_norm": 0.3406740427017212,
      "learning_rate": 2.92434761266182e-05,
      "loss": 0.07135295867919922,
      "step": 1212
    },
    {
      "epoch": 0.1639881706801859,
      "grad_norm": 0.4251808524131775,
      "learning_rate": 2.924132843622957e-05,
      "loss": 0.07227134704589844,
      "step": 1213
    },
    {
      "epoch": 0.16412336290663287,
      "grad_norm": 0.8093040585517883,
      "learning_rate": 2.9239177780720418e-05,
      "loss": 0.079986572265625,
      "step": 1214
    },
    {
      "epoch": 0.16425855513307985,
      "grad_norm": 0.4330202639102936,
      "learning_rate": 2.923702416053852e-05,
      "loss": 0.08249950408935547,
      "step": 1215
    },
    {
      "epoch": 0.16439374735952683,
      "grad_norm": 0.2535877525806427,
      "learning_rate": 2.9234867576132268e-05,
      "loss": 0.0570220947265625,
      "step": 1216
    },
    {
      "epoch": 0.1645289395859738,
      "grad_norm": 0.5611778497695923,
      "learning_rate": 2.923270802795068e-05,
      "loss": 0.10200023651123047,
      "step": 1217
    },
    {
      "epoch": 0.1646641318124208,
      "grad_norm": 0.2409171760082245,
      "learning_rate": 2.9230545516443378e-05,
      "loss": 0.059731483459472656,
      "step": 1218
    },
    {
      "epoch": 0.16479932403886777,
      "grad_norm": 0.22265183925628662,
      "learning_rate": 2.9228380042060615e-05,
      "loss": 0.06401443481445312,
      "step": 1219
    },
    {
      "epoch": 0.16493451626531475,
      "grad_norm": 0.18647074699401855,
      "learning_rate": 2.9226211605253252e-05,
      "loss": 0.036158084869384766,
      "step": 1220
    },
    {
      "epoch": 0.16506970849176172,
      "grad_norm": 0.6751832365989685,
      "learning_rate": 2.922404020647277e-05,
      "loss": 0.055684566497802734,
      "step": 1221
    },
    {
      "epoch": 0.1652049007182087,
      "grad_norm": 0.5978150367736816,
      "learning_rate": 2.9221865846171264e-05,
      "loss": 0.07187366485595703,
      "step": 1222
    },
    {
      "epoch": 0.16534009294465568,
      "grad_norm": 0.23081234097480774,
      "learning_rate": 2.9219688524801446e-05,
      "loss": 0.04249095916748047,
      "step": 1223
    },
    {
      "epoch": 0.16547528517110266,
      "grad_norm": 0.2462223619222641,
      "learning_rate": 2.9217508242816653e-05,
      "loss": 0.05764305591583252,
      "step": 1224
    },
    {
      "epoch": 0.16561047739754964,
      "grad_norm": 0.2645401954650879,
      "learning_rate": 2.921532500067083e-05,
      "loss": 0.04351234436035156,
      "step": 1225
    },
    {
      "epoch": 0.16574566962399662,
      "grad_norm": 0.30472224950790405,
      "learning_rate": 2.9213138798818528e-05,
      "loss": 0.05962371826171875,
      "step": 1226
    },
    {
      "epoch": 0.1658808618504436,
      "grad_norm": 0.7413991093635559,
      "learning_rate": 2.921094963771494e-05,
      "loss": 0.08212137222290039,
      "step": 1227
    },
    {
      "epoch": 0.16601605407689057,
      "grad_norm": 0.5718117356300354,
      "learning_rate": 2.9208757517815855e-05,
      "loss": 0.07232332229614258,
      "step": 1228
    },
    {
      "epoch": 0.16615124630333755,
      "grad_norm": 0.5255526900291443,
      "learning_rate": 2.9206562439577684e-05,
      "loss": 0.06885433197021484,
      "step": 1229
    },
    {
      "epoch": 0.16628643852978453,
      "grad_norm": 0.2876903712749481,
      "learning_rate": 2.9204364403457452e-05,
      "loss": 0.07418346405029297,
      "step": 1230
    },
    {
      "epoch": 0.1664216307562315,
      "grad_norm": 0.577975332736969,
      "learning_rate": 2.9202163409912808e-05,
      "loss": 0.06223487854003906,
      "step": 1231
    },
    {
      "epoch": 0.1665568229826785,
      "grad_norm": 1.10220205783844,
      "learning_rate": 2.9199959459402003e-05,
      "loss": 0.08066082000732422,
      "step": 1232
    },
    {
      "epoch": 0.16669201520912547,
      "grad_norm": 0.407647967338562,
      "learning_rate": 2.919775255238392e-05,
      "loss": 0.07311725616455078,
      "step": 1233
    },
    {
      "epoch": 0.16682720743557244,
      "grad_norm": 0.6401172280311584,
      "learning_rate": 2.919554268931804e-05,
      "loss": 0.0740213394165039,
      "step": 1234
    },
    {
      "epoch": 0.16696239966201942,
      "grad_norm": 0.5067819356918335,
      "learning_rate": 2.9193329870664475e-05,
      "loss": 0.07566213607788086,
      "step": 1235
    },
    {
      "epoch": 0.1670975918884664,
      "grad_norm": 0.45964089035987854,
      "learning_rate": 2.9191114096883938e-05,
      "loss": 0.07619571685791016,
      "step": 1236
    },
    {
      "epoch": 0.16723278411491338,
      "grad_norm": 0.9318496584892273,
      "learning_rate": 2.9188895368437774e-05,
      "loss": 0.0842432975769043,
      "step": 1237
    },
    {
      "epoch": 0.16736797634136036,
      "grad_norm": 0.8267168998718262,
      "learning_rate": 2.9186673685787926e-05,
      "loss": 0.09326457977294922,
      "step": 1238
    },
    {
      "epoch": 0.16750316856780736,
      "grad_norm": 0.5304529666900635,
      "learning_rate": 2.918444904939697e-05,
      "loss": 0.08791971206665039,
      "step": 1239
    },
    {
      "epoch": 0.16763836079425434,
      "grad_norm": 0.4358448386192322,
      "learning_rate": 2.9182221459728078e-05,
      "loss": 0.07210993766784668,
      "step": 1240
    },
    {
      "epoch": 0.16777355302070132,
      "grad_norm": 0.5428539514541626,
      "learning_rate": 2.917999091724505e-05,
      "loss": 0.08456897735595703,
      "step": 1241
    },
    {
      "epoch": 0.1679087452471483,
      "grad_norm": 0.6845792531967163,
      "learning_rate": 2.9177757422412294e-05,
      "loss": 0.07977914810180664,
      "step": 1242
    },
    {
      "epoch": 0.16804393747359528,
      "grad_norm": 0.8218401670455933,
      "learning_rate": 2.917552097569484e-05,
      "loss": 0.09357595443725586,
      "step": 1243
    },
    {
      "epoch": 0.16817912970004226,
      "grad_norm": 0.5126569867134094,
      "learning_rate": 2.917328157755832e-05,
      "loss": 0.07925701141357422,
      "step": 1244
    },
    {
      "epoch": 0.16831432192648924,
      "grad_norm": 0.41536298394203186,
      "learning_rate": 2.9171039228469003e-05,
      "loss": 0.06977081298828125,
      "step": 1245
    },
    {
      "epoch": 0.16844951415293621,
      "grad_norm": 0.4484061598777771,
      "learning_rate": 2.9168793928893747e-05,
      "loss": 0.07013273239135742,
      "step": 1246
    },
    {
      "epoch": 0.1685847063793832,
      "grad_norm": 0.34745410084724426,
      "learning_rate": 2.9166545679300036e-05,
      "loss": 0.05951499938964844,
      "step": 1247
    },
    {
      "epoch": 0.16871989860583017,
      "grad_norm": 0.9694186449050903,
      "learning_rate": 2.9164294480155966e-05,
      "loss": 0.09905719757080078,
      "step": 1248
    },
    {
      "epoch": 0.16885509083227715,
      "grad_norm": 0.38170325756073,
      "learning_rate": 2.9162040331930256e-05,
      "loss": 0.0772395133972168,
      "step": 1249
    },
    {
      "epoch": 0.16899028305872413,
      "grad_norm": 0.47763752937316895,
      "learning_rate": 2.915978323509223e-05,
      "loss": 0.08076238632202148,
      "step": 1250
    },
    {
      "epoch": 0.1691254752851711,
      "grad_norm": 0.3137807250022888,
      "learning_rate": 2.915752319011182e-05,
      "loss": 0.07265853881835938,
      "step": 1251
    },
    {
      "epoch": 0.16926066751161808,
      "grad_norm": 0.22357763350009918,
      "learning_rate": 2.9155260197459588e-05,
      "loss": 0.049147605895996094,
      "step": 1252
    },
    {
      "epoch": 0.16939585973806506,
      "grad_norm": 0.29378217458724976,
      "learning_rate": 2.91529942576067e-05,
      "loss": 0.05040168762207031,
      "step": 1253
    },
    {
      "epoch": 0.16953105196451204,
      "grad_norm": 0.7124695777893066,
      "learning_rate": 2.915072537102493e-05,
      "loss": 0.09065961837768555,
      "step": 1254
    },
    {
      "epoch": 0.16966624419095902,
      "grad_norm": 0.44418132305145264,
      "learning_rate": 2.914845353818668e-05,
      "loss": 0.08211135864257812,
      "step": 1255
    },
    {
      "epoch": 0.169801436417406,
      "grad_norm": 0.3166307806968689,
      "learning_rate": 2.9146178759564953e-05,
      "loss": 0.07910871505737305,
      "step": 1256
    },
    {
      "epoch": 0.16993662864385298,
      "grad_norm": 0.42782777547836304,
      "learning_rate": 2.914390103563337e-05,
      "loss": 0.0905752182006836,
      "step": 1257
    },
    {
      "epoch": 0.17007182087029996,
      "grad_norm": 0.6515711545944214,
      "learning_rate": 2.914162036686617e-05,
      "loss": 0.11740922927856445,
      "step": 1258
    },
    {
      "epoch": 0.17020701309674693,
      "grad_norm": 0.44100797176361084,
      "learning_rate": 2.9139336753738196e-05,
      "loss": 0.06583881378173828,
      "step": 1259
    },
    {
      "epoch": 0.1703422053231939,
      "grad_norm": 0.6195662021636963,
      "learning_rate": 2.913705019672491e-05,
      "loss": 0.09023380279541016,
      "step": 1260
    },
    {
      "epoch": 0.1704773975496409,
      "grad_norm": 0.3901897370815277,
      "learning_rate": 2.9134760696302386e-05,
      "loss": 0.048100948333740234,
      "step": 1261
    },
    {
      "epoch": 0.17061258977608787,
      "grad_norm": 0.6318403482437134,
      "learning_rate": 2.9132468252947306e-05,
      "loss": 0.08879256248474121,
      "step": 1262
    },
    {
      "epoch": 0.17074778200253485,
      "grad_norm": 0.27793169021606445,
      "learning_rate": 2.9130172867136974e-05,
      "loss": 0.04077959060668945,
      "step": 1263
    },
    {
      "epoch": 0.17088297422898183,
      "grad_norm": 1.3737980127334595,
      "learning_rate": 2.91278745393493e-05,
      "loss": 0.11382007598876953,
      "step": 1264
    },
    {
      "epoch": 0.1710181664554288,
      "grad_norm": 0.8151664733886719,
      "learning_rate": 2.9125573270062812e-05,
      "loss": 0.0523991584777832,
      "step": 1265
    },
    {
      "epoch": 0.17115335868187578,
      "grad_norm": 0.40451499819755554,
      "learning_rate": 2.9123269059756634e-05,
      "loss": 0.04644775390625,
      "step": 1266
    },
    {
      "epoch": 0.17128855090832276,
      "grad_norm": 0.4913591742515564,
      "learning_rate": 2.9120961908910528e-05,
      "loss": 0.10492610931396484,
      "step": 1267
    },
    {
      "epoch": 0.17142374313476974,
      "grad_norm": 0.45436978340148926,
      "learning_rate": 2.911865181800485e-05,
      "loss": 0.05088472366333008,
      "step": 1268
    },
    {
      "epoch": 0.17155893536121672,
      "grad_norm": 0.5139414072036743,
      "learning_rate": 2.9116338787520577e-05,
      "loss": 0.059630393981933594,
      "step": 1269
    },
    {
      "epoch": 0.1716941275876637,
      "grad_norm": 1.0088963508605957,
      "learning_rate": 2.9114022817939283e-05,
      "loss": 0.0697479248046875,
      "step": 1270
    },
    {
      "epoch": 0.17182931981411068,
      "grad_norm": 0.6361703276634216,
      "learning_rate": 2.911170390974318e-05,
      "loss": 0.05156564712524414,
      "step": 1271
    },
    {
      "epoch": 0.17196451204055765,
      "grad_norm": 0.5383647084236145,
      "learning_rate": 2.9109382063415067e-05,
      "loss": 0.07959222793579102,
      "step": 1272
    },
    {
      "epoch": 0.17209970426700466,
      "grad_norm": 0.2548418641090393,
      "learning_rate": 2.9107057279438372e-05,
      "loss": 0.05634641647338867,
      "step": 1273
    },
    {
      "epoch": 0.17223489649345164,
      "grad_norm": 0.7601280808448792,
      "learning_rate": 2.910472955829712e-05,
      "loss": 0.07485723495483398,
      "step": 1274
    },
    {
      "epoch": 0.17237008871989862,
      "grad_norm": 0.819226086139679,
      "learning_rate": 2.9102398900475958e-05,
      "loss": 0.0662851333618164,
      "step": 1275
    },
    {
      "epoch": 0.1725052809463456,
      "grad_norm": 0.4801798462867737,
      "learning_rate": 2.910006530646014e-05,
      "loss": 0.06519842147827148,
      "step": 1276
    },
    {
      "epoch": 0.17264047317279257,
      "grad_norm": 0.7828125953674316,
      "learning_rate": 2.909772877673554e-05,
      "loss": 0.05161428451538086,
      "step": 1277
    },
    {
      "epoch": 0.17277566539923955,
      "grad_norm": 0.5340220332145691,
      "learning_rate": 2.9095389311788626e-05,
      "loss": 0.08268213272094727,
      "step": 1278
    },
    {
      "epoch": 0.17291085762568653,
      "grad_norm": 0.644839346408844,
      "learning_rate": 2.9093046912106494e-05,
      "loss": 0.1143331527709961,
      "step": 1279
    },
    {
      "epoch": 0.1730460498521335,
      "grad_norm": 0.6397328972816467,
      "learning_rate": 2.909070157817684e-05,
      "loss": 0.05923748016357422,
      "step": 1280
    },
    {
      "epoch": 0.1731812420785805,
      "grad_norm": 0.6796171069145203,
      "learning_rate": 2.9088353310487976e-05,
      "loss": 0.053072452545166016,
      "step": 1281
    },
    {
      "epoch": 0.17331643430502747,
      "grad_norm": 1.5975652933120728,
      "learning_rate": 2.9086002109528825e-05,
      "loss": 0.09544229507446289,
      "step": 1282
    },
    {
      "epoch": 0.17345162653147445,
      "grad_norm": 0.5910240411758423,
      "learning_rate": 2.908364797578892e-05,
      "loss": 0.04752826690673828,
      "step": 1283
    },
    {
      "epoch": 0.17358681875792142,
      "grad_norm": 0.37046805024147034,
      "learning_rate": 2.9081290909758405e-05,
      "loss": 0.06883811950683594,
      "step": 1284
    },
    {
      "epoch": 0.1737220109843684,
      "grad_norm": 0.9193475246429443,
      "learning_rate": 2.9078930911928033e-05,
      "loss": 0.09998798370361328,
      "step": 1285
    },
    {
      "epoch": 0.17385720321081538,
      "grad_norm": 0.3985954523086548,
      "learning_rate": 2.907656798278916e-05,
      "loss": 0.09383535385131836,
      "step": 1286
    },
    {
      "epoch": 0.17399239543726236,
      "grad_norm": 0.4473966956138611,
      "learning_rate": 2.9074202122833773e-05,
      "loss": 0.07605195045471191,
      "step": 1287
    },
    {
      "epoch": 0.17412758766370934,
      "grad_norm": 0.4351816177368164,
      "learning_rate": 2.907183333255445e-05,
      "loss": 0.09464550018310547,
      "step": 1288
    },
    {
      "epoch": 0.17426277989015632,
      "grad_norm": 1.3260295391082764,
      "learning_rate": 2.9069461612444384e-05,
      "loss": 0.1009683609008789,
      "step": 1289
    },
    {
      "epoch": 0.1743979721166033,
      "grad_norm": 1.2182968854904175,
      "learning_rate": 2.9067086962997385e-05,
      "loss": 0.07427072525024414,
      "step": 1290
    },
    {
      "epoch": 0.17453316434305027,
      "grad_norm": 0.37077826261520386,
      "learning_rate": 2.9064709384707868e-05,
      "loss": 0.056021690368652344,
      "step": 1291
    },
    {
      "epoch": 0.17466835656949725,
      "grad_norm": 1.2460367679595947,
      "learning_rate": 2.9062328878070855e-05,
      "loss": 0.15209388732910156,
      "step": 1292
    },
    {
      "epoch": 0.17480354879594423,
      "grad_norm": 0.9635239839553833,
      "learning_rate": 2.905994544358198e-05,
      "loss": 0.07224225997924805,
      "step": 1293
    },
    {
      "epoch": 0.1749387410223912,
      "grad_norm": 0.28591299057006836,
      "learning_rate": 2.9057559081737482e-05,
      "loss": 0.06308269500732422,
      "step": 1294
    },
    {
      "epoch": 0.1750739332488382,
      "grad_norm": 0.6258271932601929,
      "learning_rate": 2.9055169793034225e-05,
      "loss": 0.09959697723388672,
      "step": 1295
    },
    {
      "epoch": 0.17520912547528517,
      "grad_norm": 0.4155055284500122,
      "learning_rate": 2.9052777577969656e-05,
      "loss": 0.08299922943115234,
      "step": 1296
    },
    {
      "epoch": 0.17534431770173214,
      "grad_norm": 0.22354209423065186,
      "learning_rate": 2.9050382437041868e-05,
      "loss": 0.04102611541748047,
      "step": 1297
    },
    {
      "epoch": 0.17547950992817912,
      "grad_norm": 0.3546565771102905,
      "learning_rate": 2.9047984370749526e-05,
      "loss": 0.09758901596069336,
      "step": 1298
    },
    {
      "epoch": 0.1756147021546261,
      "grad_norm": 0.34732240438461304,
      "learning_rate": 2.9045583379591925e-05,
      "loss": 0.058268070220947266,
      "step": 1299
    },
    {
      "epoch": 0.17574989438107308,
      "grad_norm": 0.4467600882053375,
      "learning_rate": 2.9043179464068965e-05,
      "loss": 0.09261894226074219,
      "step": 1300
    },
    {
      "epoch": 0.17588508660752006,
      "grad_norm": 0.2954292297363281,
      "learning_rate": 2.9040772624681152e-05,
      "loss": 0.06338787078857422,
      "step": 1301
    },
    {
      "epoch": 0.17602027883396704,
      "grad_norm": 0.49611571431159973,
      "learning_rate": 2.9038362861929603e-05,
      "loss": 0.0854954719543457,
      "step": 1302
    },
    {
      "epoch": 0.17615547106041402,
      "grad_norm": 0.506494402885437,
      "learning_rate": 2.903595017631605e-05,
      "loss": 0.06277728080749512,
      "step": 1303
    },
    {
      "epoch": 0.176290663286861,
      "grad_norm": 0.20389805734157562,
      "learning_rate": 2.903353456834282e-05,
      "loss": 0.0464324951171875,
      "step": 1304
    },
    {
      "epoch": 0.17642585551330797,
      "grad_norm": 0.25398895144462585,
      "learning_rate": 2.903111603851285e-05,
      "loss": 0.06668472290039062,
      "step": 1305
    },
    {
      "epoch": 0.17656104773975495,
      "grad_norm": 0.2984888255596161,
      "learning_rate": 2.9028694587329704e-05,
      "loss": 0.07360184192657471,
      "step": 1306
    },
    {
      "epoch": 0.17669623996620196,
      "grad_norm": 0.14982306957244873,
      "learning_rate": 2.902627021529753e-05,
      "loss": 0.04499363899230957,
      "step": 1307
    },
    {
      "epoch": 0.17683143219264894,
      "grad_norm": 0.24880695343017578,
      "learning_rate": 2.9023842922921105e-05,
      "loss": 0.06262731552124023,
      "step": 1308
    },
    {
      "epoch": 0.1769666244190959,
      "grad_norm": 0.24707214534282684,
      "learning_rate": 2.90214127107058e-05,
      "loss": 0.07372760772705078,
      "step": 1309
    },
    {
      "epoch": 0.1771018166455429,
      "grad_norm": 0.5720853209495544,
      "learning_rate": 2.9018979579157592e-05,
      "loss": 0.09578704833984375,
      "step": 1310
    },
    {
      "epoch": 0.17723700887198987,
      "grad_norm": 0.3586277961730957,
      "learning_rate": 2.901654352878308e-05,
      "loss": 0.06875896453857422,
      "step": 1311
    },
    {
      "epoch": 0.17737220109843685,
      "grad_norm": 0.23985342681407928,
      "learning_rate": 2.9014104560089462e-05,
      "loss": 0.036717891693115234,
      "step": 1312
    },
    {
      "epoch": 0.17750739332488383,
      "grad_norm": 0.6026805639266968,
      "learning_rate": 2.9011662673584538e-05,
      "loss": 0.08232450485229492,
      "step": 1313
    },
    {
      "epoch": 0.1776425855513308,
      "grad_norm": 0.25450122356414795,
      "learning_rate": 2.900921786977673e-05,
      "loss": 0.07907867431640625,
      "step": 1314
    },
    {
      "epoch": 0.17777777777777778,
      "grad_norm": 0.6614646911621094,
      "learning_rate": 2.900677014917505e-05,
      "loss": 0.07966995239257812,
      "step": 1315
    },
    {
      "epoch": 0.17791297000422476,
      "grad_norm": 0.2466992437839508,
      "learning_rate": 2.9004319512289136e-05,
      "loss": 0.06562530994415283,
      "step": 1316
    },
    {
      "epoch": 0.17804816223067174,
      "grad_norm": 0.4602111279964447,
      "learning_rate": 2.9001865959629222e-05,
      "loss": 0.0880885124206543,
      "step": 1317
    },
    {
      "epoch": 0.17818335445711872,
      "grad_norm": 0.8172971606254578,
      "learning_rate": 2.8999409491706143e-05,
      "loss": 0.0840153694152832,
      "step": 1318
    },
    {
      "epoch": 0.1783185466835657,
      "grad_norm": 0.4828501343727112,
      "learning_rate": 2.8996950109031355e-05,
      "loss": 0.07391500473022461,
      "step": 1319
    },
    {
      "epoch": 0.17845373891001268,
      "grad_norm": 0.44382625818252563,
      "learning_rate": 2.8994487812116917e-05,
      "loss": 0.05877399444580078,
      "step": 1320
    },
    {
      "epoch": 0.17858893113645966,
      "grad_norm": 0.43092894554138184,
      "learning_rate": 2.8992022601475483e-05,
      "loss": 0.08002662658691406,
      "step": 1321
    },
    {
      "epoch": 0.17872412336290663,
      "grad_norm": 0.5200045704841614,
      "learning_rate": 2.8989554477620332e-05,
      "loss": 0.05827474594116211,
      "step": 1322
    },
    {
      "epoch": 0.1788593155893536,
      "grad_norm": 0.2672748863697052,
      "learning_rate": 2.8987083441065335e-05,
      "loss": 0.07176971435546875,
      "step": 1323
    },
    {
      "epoch": 0.1789945078158006,
      "grad_norm": 0.42366379499435425,
      "learning_rate": 2.8984609492324983e-05,
      "loss": 0.06212615966796875,
      "step": 1324
    },
    {
      "epoch": 0.17912970004224757,
      "grad_norm": 0.6650789976119995,
      "learning_rate": 2.8982132631914357e-05,
      "loss": 0.1201324462890625,
      "step": 1325
    },
    {
      "epoch": 0.17926489226869455,
      "grad_norm": 0.6314013004302979,
      "learning_rate": 2.8979652860349154e-05,
      "loss": 0.1195363998413086,
      "step": 1326
    },
    {
      "epoch": 0.17940008449514153,
      "grad_norm": 0.37290480732917786,
      "learning_rate": 2.8977170178145675e-05,
      "loss": 0.0666818618774414,
      "step": 1327
    },
    {
      "epoch": 0.1795352767215885,
      "grad_norm": 0.14347536861896515,
      "learning_rate": 2.8974684585820833e-05,
      "loss": 0.03762531280517578,
      "step": 1328
    },
    {
      "epoch": 0.17967046894803548,
      "grad_norm": 0.33331525325775146,
      "learning_rate": 2.8972196083892138e-05,
      "loss": 0.06098461151123047,
      "step": 1329
    },
    {
      "epoch": 0.17980566117448246,
      "grad_norm": 0.36663979291915894,
      "learning_rate": 2.8969704672877707e-05,
      "loss": 0.059617042541503906,
      "step": 1330
    },
    {
      "epoch": 0.17994085340092944,
      "grad_norm": 0.4675559997558594,
      "learning_rate": 2.896721035329627e-05,
      "loss": 0.07706928253173828,
      "step": 1331
    },
    {
      "epoch": 0.18007604562737642,
      "grad_norm": 0.23798486590385437,
      "learning_rate": 2.8964713125667153e-05,
      "loss": 0.03920292854309082,
      "step": 1332
    },
    {
      "epoch": 0.1802112378538234,
      "grad_norm": 0.643490731716156,
      "learning_rate": 2.8962212990510294e-05,
      "loss": 0.08657097816467285,
      "step": 1333
    },
    {
      "epoch": 0.18034643008027038,
      "grad_norm": 0.4887504577636719,
      "learning_rate": 2.8959709948346237e-05,
      "loss": 0.08656644821166992,
      "step": 1334
    },
    {
      "epoch": 0.18048162230671735,
      "grad_norm": 0.27658864855766296,
      "learning_rate": 2.8957203999696124e-05,
      "loss": 0.05988359451293945,
      "step": 1335
    },
    {
      "epoch": 0.18061681453316433,
      "grad_norm": 0.5200914740562439,
      "learning_rate": 2.8954695145081713e-05,
      "loss": 0.055371761322021484,
      "step": 1336
    },
    {
      "epoch": 0.1807520067596113,
      "grad_norm": 0.3747258484363556,
      "learning_rate": 2.8952183385025356e-05,
      "loss": 0.07856559753417969,
      "step": 1337
    },
    {
      "epoch": 0.1808871989860583,
      "grad_norm": 0.8470179438591003,
      "learning_rate": 2.8949668720050014e-05,
      "loss": 0.10687398910522461,
      "step": 1338
    },
    {
      "epoch": 0.18102239121250527,
      "grad_norm": 0.38438788056373596,
      "learning_rate": 2.8947151150679256e-05,
      "loss": 0.0556793212890625,
      "step": 1339
    },
    {
      "epoch": 0.18115758343895225,
      "grad_norm": 0.40534600615501404,
      "learning_rate": 2.8944630677437255e-05,
      "loss": 0.06749582290649414,
      "step": 1340
    },
    {
      "epoch": 0.18129277566539925,
      "grad_norm": 0.3961862325668335,
      "learning_rate": 2.8942107300848784e-05,
      "loss": 0.04695749282836914,
      "step": 1341
    },
    {
      "epoch": 0.18142796789184623,
      "grad_norm": 1.1106945276260376,
      "learning_rate": 2.8939581021439225e-05,
      "loss": 0.09221553802490234,
      "step": 1342
    },
    {
      "epoch": 0.1815631601182932,
      "grad_norm": 0.3141094446182251,
      "learning_rate": 2.8937051839734563e-05,
      "loss": 0.07662510871887207,
      "step": 1343
    },
    {
      "epoch": 0.1816983523447402,
      "grad_norm": 0.7614315152168274,
      "learning_rate": 2.8934519756261384e-05,
      "loss": 0.09467506408691406,
      "step": 1344
    },
    {
      "epoch": 0.18183354457118717,
      "grad_norm": 0.27528730034828186,
      "learning_rate": 2.8931984771546885e-05,
      "loss": 0.03645491600036621,
      "step": 1345
    },
    {
      "epoch": 0.18196873679763415,
      "grad_norm": 0.30013570189476013,
      "learning_rate": 2.8929446886118866e-05,
      "loss": 0.06853103637695312,
      "step": 1346
    },
    {
      "epoch": 0.18210392902408112,
      "grad_norm": 0.6001962423324585,
      "learning_rate": 2.892690610050572e-05,
      "loss": 0.08995652198791504,
      "step": 1347
    },
    {
      "epoch": 0.1822391212505281,
      "grad_norm": 0.2631484568119049,
      "learning_rate": 2.892436241523646e-05,
      "loss": 0.05412912368774414,
      "step": 1348
    },
    {
      "epoch": 0.18237431347697508,
      "grad_norm": 0.25234776735305786,
      "learning_rate": 2.8921815830840685e-05,
      "loss": 0.0598912239074707,
      "step": 1349
    },
    {
      "epoch": 0.18250950570342206,
      "grad_norm": 0.2799901068210602,
      "learning_rate": 2.891926634784862e-05,
      "loss": 0.07377505302429199,
      "step": 1350
    },
    {
      "epoch": 0.18264469792986904,
      "grad_norm": 0.33253398537635803,
      "learning_rate": 2.8916713966791076e-05,
      "loss": 0.0797414779663086,
      "step": 1351
    },
    {
      "epoch": 0.18277989015631602,
      "grad_norm": 0.3313521146774292,
      "learning_rate": 2.8914158688199464e-05,
      "loss": 0.0764913558959961,
      "step": 1352
    },
    {
      "epoch": 0.182915082382763,
      "grad_norm": 1.5569920539855957,
      "learning_rate": 2.891160051260582e-05,
      "loss": 0.08131265640258789,
      "step": 1353
    },
    {
      "epoch": 0.18305027460920997,
      "grad_norm": 0.662760317325592,
      "learning_rate": 2.8909039440542758e-05,
      "loss": 0.09003162384033203,
      "step": 1354
    },
    {
      "epoch": 0.18318546683565695,
      "grad_norm": 0.42132049798965454,
      "learning_rate": 2.890647547254352e-05,
      "loss": 0.0610203742980957,
      "step": 1355
    },
    {
      "epoch": 0.18332065906210393,
      "grad_norm": 0.369578093290329,
      "learning_rate": 2.8903908609141923e-05,
      "loss": 0.07361698150634766,
      "step": 1356
    },
    {
      "epoch": 0.1834558512885509,
      "grad_norm": 0.6707038283348083,
      "learning_rate": 2.8901338850872413e-05,
      "loss": 0.09276676177978516,
      "step": 1357
    },
    {
      "epoch": 0.1835910435149979,
      "grad_norm": 0.6248319149017334,
      "learning_rate": 2.8898766198270022e-05,
      "loss": 0.09302616119384766,
      "step": 1358
    },
    {
      "epoch": 0.18372623574144487,
      "grad_norm": 0.5640145540237427,
      "learning_rate": 2.8896190651870392e-05,
      "loss": 0.06710171699523926,
      "step": 1359
    },
    {
      "epoch": 0.18386142796789184,
      "grad_norm": 0.5199601650238037,
      "learning_rate": 2.8893612212209763e-05,
      "loss": 0.08800101280212402,
      "step": 1360
    },
    {
      "epoch": 0.18399662019433882,
      "grad_norm": 0.6198127269744873,
      "learning_rate": 2.8891030879824985e-05,
      "loss": 0.05260324478149414,
      "step": 1361
    },
    {
      "epoch": 0.1841318124207858,
      "grad_norm": 0.33992907404899597,
      "learning_rate": 2.88884466552535e-05,
      "loss": 0.05054807662963867,
      "step": 1362
    },
    {
      "epoch": 0.18426700464723278,
      "grad_norm": 0.43843114376068115,
      "learning_rate": 2.888585953903336e-05,
      "loss": 0.10338830947875977,
      "step": 1363
    },
    {
      "epoch": 0.18440219687367976,
      "grad_norm": 0.40042394399642944,
      "learning_rate": 2.888326953170321e-05,
      "loss": 0.06149578094482422,
      "step": 1364
    },
    {
      "epoch": 0.18453738910012674,
      "grad_norm": 0.8685281276702881,
      "learning_rate": 2.8880676633802314e-05,
      "loss": 0.07832813262939453,
      "step": 1365
    },
    {
      "epoch": 0.18467258132657371,
      "grad_norm": 0.1962835192680359,
      "learning_rate": 2.8878080845870522e-05,
      "loss": 0.04717904329299927,
      "step": 1366
    },
    {
      "epoch": 0.1848077735530207,
      "grad_norm": 0.3657790422439575,
      "learning_rate": 2.887548216844829e-05,
      "loss": 0.0577082633972168,
      "step": 1367
    },
    {
      "epoch": 0.18494296577946767,
      "grad_norm": 0.27183419466018677,
      "learning_rate": 2.8872880602076675e-05,
      "loss": 0.054390907287597656,
      "step": 1368
    },
    {
      "epoch": 0.18507815800591465,
      "grad_norm": 0.6341437697410583,
      "learning_rate": 2.8870276147297344e-05,
      "loss": 0.10355615615844727,
      "step": 1369
    },
    {
      "epoch": 0.18521335023236163,
      "grad_norm": 0.5695599317550659,
      "learning_rate": 2.8867668804652552e-05,
      "loss": 0.07115697860717773,
      "step": 1370
    },
    {
      "epoch": 0.1853485424588086,
      "grad_norm": 0.35412582755088806,
      "learning_rate": 2.886505857468516e-05,
      "loss": 0.06946957111358643,
      "step": 1371
    },
    {
      "epoch": 0.18548373468525559,
      "grad_norm": 0.6484407782554626,
      "learning_rate": 2.8862445457938642e-05,
      "loss": 0.08343744277954102,
      "step": 1372
    },
    {
      "epoch": 0.18561892691170256,
      "grad_norm": 0.2549974322319031,
      "learning_rate": 2.8859829454957053e-05,
      "loss": 0.08511996269226074,
      "step": 1373
    },
    {
      "epoch": 0.18575411913814954,
      "grad_norm": 0.24268680810928345,
      "learning_rate": 2.8857210566285062e-05,
      "loss": 0.0642235279083252,
      "step": 1374
    },
    {
      "epoch": 0.18588931136459655,
      "grad_norm": 0.3865174949169159,
      "learning_rate": 2.8854588792467932e-05,
      "loss": 0.08867883682250977,
      "step": 1375
    },
    {
      "epoch": 0.18602450359104353,
      "grad_norm": 1.0227934122085571,
      "learning_rate": 2.8851964134051535e-05,
      "loss": 0.08764886856079102,
      "step": 1376
    },
    {
      "epoch": 0.1861596958174905,
      "grad_norm": 0.9363678097724915,
      "learning_rate": 2.884933659158234e-05,
      "loss": 0.11895275115966797,
      "step": 1377
    },
    {
      "epoch": 0.18629488804393748,
      "grad_norm": 0.6783412098884583,
      "learning_rate": 2.8846706165607415e-05,
      "loss": 0.06989693641662598,
      "step": 1378
    },
    {
      "epoch": 0.18643008027038446,
      "grad_norm": 0.49864137172698975,
      "learning_rate": 2.8844072856674422e-05,
      "loss": 0.07854223251342773,
      "step": 1379
    },
    {
      "epoch": 0.18656527249683144,
      "grad_norm": 0.8806300759315491,
      "learning_rate": 2.8841436665331634e-05,
      "loss": 0.07428121566772461,
      "step": 1380
    },
    {
      "epoch": 0.18670046472327842,
      "grad_norm": 0.36095812916755676,
      "learning_rate": 2.8838797592127927e-05,
      "loss": 0.057088375091552734,
      "step": 1381
    },
    {
      "epoch": 0.1868356569497254,
      "grad_norm": 0.21287067234516144,
      "learning_rate": 2.883615563761276e-05,
      "loss": 0.056174278259277344,
      "step": 1382
    },
    {
      "epoch": 0.18697084917617238,
      "grad_norm": 0.7511873245239258,
      "learning_rate": 2.8833510802336206e-05,
      "loss": 0.1345386505126953,
      "step": 1383
    },
    {
      "epoch": 0.18710604140261936,
      "grad_norm": 0.5694673657417297,
      "learning_rate": 2.883086308684893e-05,
      "loss": 0.0859684944152832,
      "step": 1384
    },
    {
      "epoch": 0.18724123362906633,
      "grad_norm": 0.32220032811164856,
      "learning_rate": 2.882821249170221e-05,
      "loss": 0.09302234649658203,
      "step": 1385
    },
    {
      "epoch": 0.1873764258555133,
      "grad_norm": 0.6958578824996948,
      "learning_rate": 2.8825559017447905e-05,
      "loss": 0.08498334884643555,
      "step": 1386
    },
    {
      "epoch": 0.1875116180819603,
      "grad_norm": 0.43010157346725464,
      "learning_rate": 2.8822902664638487e-05,
      "loss": 0.08757972717285156,
      "step": 1387
    },
    {
      "epoch": 0.18764681030840727,
      "grad_norm": 0.5990293622016907,
      "learning_rate": 2.882024343382702e-05,
      "loss": 0.06763720512390137,
      "step": 1388
    },
    {
      "epoch": 0.18778200253485425,
      "grad_norm": 0.3788200616836548,
      "learning_rate": 2.8817581325567174e-05,
      "loss": 0.09396004676818848,
      "step": 1389
    },
    {
      "epoch": 0.18791719476130123,
      "grad_norm": 0.25358206033706665,
      "learning_rate": 2.8814916340413205e-05,
      "loss": 0.05852937698364258,
      "step": 1390
    },
    {
      "epoch": 0.1880523869877482,
      "grad_norm": 0.3131154179573059,
      "learning_rate": 2.881224847891999e-05,
      "loss": 0.0705723762512207,
      "step": 1391
    },
    {
      "epoch": 0.18818757921419518,
      "grad_norm": 0.3784100413322449,
      "learning_rate": 2.8809577741642987e-05,
      "loss": 0.09511137008666992,
      "step": 1392
    },
    {
      "epoch": 0.18832277144064216,
      "grad_norm": 0.22725166380405426,
      "learning_rate": 2.8806904129138255e-05,
      "loss": 0.0523834228515625,
      "step": 1393
    },
    {
      "epoch": 0.18845796366708914,
      "grad_norm": 0.49162107706069946,
      "learning_rate": 2.8804227641962457e-05,
      "loss": 0.08653831481933594,
      "step": 1394
    },
    {
      "epoch": 0.18859315589353612,
      "grad_norm": 0.30303943157196045,
      "learning_rate": 2.8801548280672847e-05,
      "loss": 0.04696464538574219,
      "step": 1395
    },
    {
      "epoch": 0.1887283481199831,
      "grad_norm": 0.2844439744949341,
      "learning_rate": 2.8798866045827288e-05,
      "loss": 0.061730384826660156,
      "step": 1396
    },
    {
      "epoch": 0.18886354034643008,
      "grad_norm": 0.21750833094120026,
      "learning_rate": 2.8796180937984234e-05,
      "loss": 0.04741954803466797,
      "step": 1397
    },
    {
      "epoch": 0.18899873257287705,
      "grad_norm": 0.359749436378479,
      "learning_rate": 2.8793492957702738e-05,
      "loss": 0.05258369445800781,
      "step": 1398
    },
    {
      "epoch": 0.18913392479932403,
      "grad_norm": 0.4648842215538025,
      "learning_rate": 2.8790802105542454e-05,
      "loss": 0.05172157287597656,
      "step": 1399
    },
    {
      "epoch": 0.189269117025771,
      "grad_norm": 0.6724797487258911,
      "learning_rate": 2.8788108382063628e-05,
      "loss": 0.08913755416870117,
      "step": 1400
    },
    {
      "epoch": 0.189404309252218,
      "grad_norm": 0.2939850687980652,
      "learning_rate": 2.878541178782711e-05,
      "loss": 0.06687545776367188,
      "step": 1401
    },
    {
      "epoch": 0.18953950147866497,
      "grad_norm": 1.0107295513153076,
      "learning_rate": 2.8782712323394344e-05,
      "loss": 0.10088801383972168,
      "step": 1402
    },
    {
      "epoch": 0.18967469370511195,
      "grad_norm": 0.5084041357040405,
      "learning_rate": 2.878000998932738e-05,
      "loss": 0.06899881362915039,
      "step": 1403
    },
    {
      "epoch": 0.18980988593155892,
      "grad_norm": 0.23239442706108093,
      "learning_rate": 2.8777304786188847e-05,
      "loss": 0.04545164108276367,
      "step": 1404
    },
    {
      "epoch": 0.1899450781580059,
      "grad_norm": 0.5677239894866943,
      "learning_rate": 2.8774596714541988e-05,
      "loss": 0.08278656005859375,
      "step": 1405
    },
    {
      "epoch": 0.19008027038445288,
      "grad_norm": 0.598884105682373,
      "learning_rate": 2.8771885774950637e-05,
      "loss": 0.05105018615722656,
      "step": 1406
    },
    {
      "epoch": 0.19021546261089986,
      "grad_norm": 0.4362331032752991,
      "learning_rate": 2.876917196797923e-05,
      "loss": 0.08715939521789551,
      "step": 1407
    },
    {
      "epoch": 0.19035065483734684,
      "grad_norm": 0.4212772846221924,
      "learning_rate": 2.876645529419279e-05,
      "loss": 0.10063362121582031,
      "step": 1408
    },
    {
      "epoch": 0.19048584706379385,
      "grad_norm": 0.28369805216789246,
      "learning_rate": 2.876373575415695e-05,
      "loss": 0.03939962387084961,
      "step": 1409
    },
    {
      "epoch": 0.19062103929024082,
      "grad_norm": 0.47209009528160095,
      "learning_rate": 2.8761013348437926e-05,
      "loss": 0.07767105102539062,
      "step": 1410
    },
    {
      "epoch": 0.1907562315166878,
      "grad_norm": 0.3788244426250458,
      "learning_rate": 2.875828807760254e-05,
      "loss": 0.09695816040039062,
      "step": 1411
    },
    {
      "epoch": 0.19089142374313478,
      "grad_norm": 0.39608728885650635,
      "learning_rate": 2.875555994221821e-05,
      "loss": 0.08797359466552734,
      "step": 1412
    },
    {
      "epoch": 0.19102661596958176,
      "grad_norm": 0.27540138363838196,
      "learning_rate": 2.8752828942852943e-05,
      "loss": 0.05115818977355957,
      "step": 1413
    },
    {
      "epoch": 0.19116180819602874,
      "grad_norm": 0.23669497668743134,
      "learning_rate": 2.875009508007535e-05,
      "loss": 0.05493497848510742,
      "step": 1414
    },
    {
      "epoch": 0.19129700042247572,
      "grad_norm": 0.40007221698760986,
      "learning_rate": 2.8747358354454642e-05,
      "loss": 0.07413768768310547,
      "step": 1415
    },
    {
      "epoch": 0.1914321926489227,
      "grad_norm": 0.4296001195907593,
      "learning_rate": 2.8744618766560614e-05,
      "loss": 0.10266685485839844,
      "step": 1416
    },
    {
      "epoch": 0.19156738487536967,
      "grad_norm": 0.500477135181427,
      "learning_rate": 2.8741876316963664e-05,
      "loss": 0.11459064483642578,
      "step": 1417
    },
    {
      "epoch": 0.19170257710181665,
      "grad_norm": 0.3222072124481201,
      "learning_rate": 2.873913100623478e-05,
      "loss": 0.06590056419372559,
      "step": 1418
    },
    {
      "epoch": 0.19183776932826363,
      "grad_norm": 0.5025795698165894,
      "learning_rate": 2.873638283494556e-05,
      "loss": 0.09156131744384766,
      "step": 1419
    },
    {
      "epoch": 0.1919729615547106,
      "grad_norm": 0.26444897055625916,
      "learning_rate": 2.8733631803668178e-05,
      "loss": 0.050261080265045166,
      "step": 1420
    },
    {
      "epoch": 0.1921081537811576,
      "grad_norm": 0.24569398164749146,
      "learning_rate": 2.8730877912975418e-05,
      "loss": 0.055072784423828125,
      "step": 1421
    },
    {
      "epoch": 0.19224334600760457,
      "grad_norm": 0.3167657256126404,
      "learning_rate": 2.8728121163440656e-05,
      "loss": 0.0781550407409668,
      "step": 1422
    },
    {
      "epoch": 0.19237853823405154,
      "grad_norm": 0.661824643611908,
      "learning_rate": 2.8725361555637863e-05,
      "loss": 0.06928849220275879,
      "step": 1423
    },
    {
      "epoch": 0.19251373046049852,
      "grad_norm": 0.373063325881958,
      "learning_rate": 2.8722599090141598e-05,
      "loss": 0.07135629653930664,
      "step": 1424
    },
    {
      "epoch": 0.1926489226869455,
      "grad_norm": 0.4678906500339508,
      "learning_rate": 2.8719833767527026e-05,
      "loss": 0.07494425773620605,
      "step": 1425
    },
    {
      "epoch": 0.19278411491339248,
      "grad_norm": 0.47001031041145325,
      "learning_rate": 2.8717065588369896e-05,
      "loss": 0.09882116317749023,
      "step": 1426
    },
    {
      "epoch": 0.19291930713983946,
      "grad_norm": 0.3049338757991791,
      "learning_rate": 2.871429455324657e-05,
      "loss": 0.04781341552734375,
      "step": 1427
    },
    {
      "epoch": 0.19305449936628644,
      "grad_norm": 0.5737128257751465,
      "learning_rate": 2.871152066273398e-05,
      "loss": 0.11860418319702148,
      "step": 1428
    },
    {
      "epoch": 0.19318969159273341,
      "grad_norm": 0.272352397441864,
      "learning_rate": 2.870874391740967e-05,
      "loss": 0.06917500495910645,
      "step": 1429
    },
    {
      "epoch": 0.1933248838191804,
      "grad_norm": 0.3731553852558136,
      "learning_rate": 2.8705964317851774e-05,
      "loss": 0.08531379699707031,
      "step": 1430
    },
    {
      "epoch": 0.19346007604562737,
      "grad_norm": 0.5229251980781555,
      "learning_rate": 2.8703181864639013e-05,
      "loss": 0.08930301666259766,
      "step": 1431
    },
    {
      "epoch": 0.19359526827207435,
      "grad_norm": 0.24590151011943817,
      "learning_rate": 2.870039655835072e-05,
      "loss": 0.06226539611816406,
      "step": 1432
    },
    {
      "epoch": 0.19373046049852133,
      "grad_norm": 0.4491427540779114,
      "learning_rate": 2.8697608399566796e-05,
      "loss": 0.060721397399902344,
      "step": 1433
    },
    {
      "epoch": 0.1938656527249683,
      "grad_norm": 0.4435080587863922,
      "learning_rate": 2.869481738886777e-05,
      "loss": 0.0764169692993164,
      "step": 1434
    },
    {
      "epoch": 0.19400084495141529,
      "grad_norm": 0.2470765858888626,
      "learning_rate": 2.8692023526834725e-05,
      "loss": 0.05698108673095703,
      "step": 1435
    },
    {
      "epoch": 0.19413603717786226,
      "grad_norm": 0.27826058864593506,
      "learning_rate": 2.8689226814049367e-05,
      "loss": 0.06575632095336914,
      "step": 1436
    },
    {
      "epoch": 0.19427122940430924,
      "grad_norm": 0.7404970526695251,
      "learning_rate": 2.868642725109399e-05,
      "loss": 0.1010732650756836,
      "step": 1437
    },
    {
      "epoch": 0.19440642163075622,
      "grad_norm": 0.648781955242157,
      "learning_rate": 2.868362483855147e-05,
      "loss": 0.09947490692138672,
      "step": 1438
    },
    {
      "epoch": 0.1945416138572032,
      "grad_norm": 0.4128603935241699,
      "learning_rate": 2.8680819577005295e-05,
      "loss": 0.07432889938354492,
      "step": 1439
    },
    {
      "epoch": 0.19467680608365018,
      "grad_norm": 0.5470665097236633,
      "learning_rate": 2.8678011467039526e-05,
      "loss": 0.0687570571899414,
      "step": 1440
    },
    {
      "epoch": 0.19481199831009716,
      "grad_norm": 0.32682204246520996,
      "learning_rate": 2.867520050923883e-05,
      "loss": 0.06057882308959961,
      "step": 1441
    },
    {
      "epoch": 0.19494719053654413,
      "grad_norm": 0.3395327627658844,
      "learning_rate": 2.8672386704188466e-05,
      "loss": 0.040506839752197266,
      "step": 1442
    },
    {
      "epoch": 0.19508238276299114,
      "grad_norm": 0.42658281326293945,
      "learning_rate": 2.8669570052474273e-05,
      "loss": 0.07293081283569336,
      "step": 1443
    },
    {
      "epoch": 0.19521757498943812,
      "grad_norm": 0.823521614074707,
      "learning_rate": 2.86667505546827e-05,
      "loss": 0.07235860824584961,
      "step": 1444
    },
    {
      "epoch": 0.1953527672158851,
      "grad_norm": 0.7458866238594055,
      "learning_rate": 2.866392821140079e-05,
      "loss": 0.08974456787109375,
      "step": 1445
    },
    {
      "epoch": 0.19548795944233208,
      "grad_norm": 0.37601667642593384,
      "learning_rate": 2.8661103023216154e-05,
      "loss": 0.061406612396240234,
      "step": 1446
    },
    {
      "epoch": 0.19562315166877906,
      "grad_norm": 0.333290696144104,
      "learning_rate": 2.8658274990717018e-05,
      "loss": 0.040827274322509766,
      "step": 1447
    },
    {
      "epoch": 0.19575834389522603,
      "grad_norm": 0.3625883162021637,
      "learning_rate": 2.86554441144922e-05,
      "loss": 0.08270835876464844,
      "step": 1448
    },
    {
      "epoch": 0.195893536121673,
      "grad_norm": 0.9148176312446594,
      "learning_rate": 2.8652610395131097e-05,
      "loss": 0.08868408203125,
      "step": 1449
    },
    {
      "epoch": 0.19602872834812,
      "grad_norm": 0.32081353664398193,
      "learning_rate": 2.8649773833223702e-05,
      "loss": 0.08452033996582031,
      "step": 1450
    },
    {
      "epoch": 0.19616392057456697,
      "grad_norm": 0.18311938643455505,
      "learning_rate": 2.8646934429360606e-05,
      "loss": 0.04708433151245117,
      "step": 1451
    },
    {
      "epoch": 0.19629911280101395,
      "grad_norm": 0.7025923728942871,
      "learning_rate": 2.8644092184132986e-05,
      "loss": 0.07661104202270508,
      "step": 1452
    },
    {
      "epoch": 0.19643430502746093,
      "grad_norm": 0.6821320652961731,
      "learning_rate": 2.864124709813262e-05,
      "loss": 0.1329793930053711,
      "step": 1453
    },
    {
      "epoch": 0.1965694972539079,
      "grad_norm": 0.19413632154464722,
      "learning_rate": 2.8638399171951856e-05,
      "loss": 0.04824209213256836,
      "step": 1454
    },
    {
      "epoch": 0.19670468948035488,
      "grad_norm": 0.36810097098350525,
      "learning_rate": 2.8635548406183664e-05,
      "loss": 0.06572818756103516,
      "step": 1455
    },
    {
      "epoch": 0.19683988170680186,
      "grad_norm": 0.35278984904289246,
      "learning_rate": 2.8632694801421576e-05,
      "loss": 0.055222511291503906,
      "step": 1456
    },
    {
      "epoch": 0.19697507393324884,
      "grad_norm": 0.7974708676338196,
      "learning_rate": 2.862983835825973e-05,
      "loss": 0.08074188232421875,
      "step": 1457
    },
    {
      "epoch": 0.19711026615969582,
      "grad_norm": 0.6072102785110474,
      "learning_rate": 2.8626979077292856e-05,
      "loss": 0.07314491271972656,
      "step": 1458
    },
    {
      "epoch": 0.1972454583861428,
      "grad_norm": 0.43032896518707275,
      "learning_rate": 2.862411695911627e-05,
      "loss": 0.07599592208862305,
      "step": 1459
    },
    {
      "epoch": 0.19738065061258978,
      "grad_norm": 0.17891408503055573,
      "learning_rate": 2.862125200432588e-05,
      "loss": 0.04436969757080078,
      "step": 1460
    },
    {
      "epoch": 0.19751584283903675,
      "grad_norm": 0.5310320258140564,
      "learning_rate": 2.8618384213518188e-05,
      "loss": 0.06681764125823975,
      "step": 1461
    },
    {
      "epoch": 0.19765103506548373,
      "grad_norm": 0.2765582799911499,
      "learning_rate": 2.861551358729028e-05,
      "loss": 0.06185102462768555,
      "step": 1462
    },
    {
      "epoch": 0.1977862272919307,
      "grad_norm": 0.4849787950515747,
      "learning_rate": 2.8612640126239836e-05,
      "loss": 0.05442094802856445,
      "step": 1463
    },
    {
      "epoch": 0.1979214195183777,
      "grad_norm": 0.937890350818634,
      "learning_rate": 2.8609763830965126e-05,
      "loss": 0.09646177291870117,
      "step": 1464
    },
    {
      "epoch": 0.19805661174482467,
      "grad_norm": 0.5472166538238525,
      "learning_rate": 2.860688470206501e-05,
      "loss": 0.08281135559082031,
      "step": 1465
    },
    {
      "epoch": 0.19819180397127165,
      "grad_norm": 0.4535694420337677,
      "learning_rate": 2.8604002740138936e-05,
      "loss": 0.10628128051757812,
      "step": 1466
    },
    {
      "epoch": 0.19832699619771862,
      "grad_norm": 0.3947764039039612,
      "learning_rate": 2.860111794578695e-05,
      "loss": 0.05362558364868164,
      "step": 1467
    },
    {
      "epoch": 0.1984621884241656,
      "grad_norm": 0.35460296273231506,
      "learning_rate": 2.8598230319609677e-05,
      "loss": 0.0586855411529541,
      "step": 1468
    },
    {
      "epoch": 0.19859738065061258,
      "grad_norm": 0.43101951479911804,
      "learning_rate": 2.8595339862208336e-05,
      "loss": 0.06691217422485352,
      "step": 1469
    },
    {
      "epoch": 0.19873257287705956,
      "grad_norm": 0.529268205165863,
      "learning_rate": 2.8592446574184733e-05,
      "loss": 0.0815577507019043,
      "step": 1470
    },
    {
      "epoch": 0.19886776510350654,
      "grad_norm": 0.6762465238571167,
      "learning_rate": 2.8589550456141274e-05,
      "loss": 0.11527824401855469,
      "step": 1471
    },
    {
      "epoch": 0.19900295732995352,
      "grad_norm": 0.661712110042572,
      "learning_rate": 2.8586651508680942e-05,
      "loss": 0.08480548858642578,
      "step": 1472
    },
    {
      "epoch": 0.1991381495564005,
      "grad_norm": 0.4358825385570526,
      "learning_rate": 2.8583749732407312e-05,
      "loss": 0.07483959197998047,
      "step": 1473
    },
    {
      "epoch": 0.19927334178284747,
      "grad_norm": 0.359188973903656,
      "learning_rate": 2.8580845127924546e-05,
      "loss": 0.08134269714355469,
      "step": 1474
    },
    {
      "epoch": 0.19940853400929445,
      "grad_norm": 0.35140374302864075,
      "learning_rate": 2.8577937695837408e-05,
      "loss": 0.05597877502441406,
      "step": 1475
    },
    {
      "epoch": 0.19954372623574146,
      "grad_norm": 0.7786426544189453,
      "learning_rate": 2.8575027436751235e-05,
      "loss": 0.06444978713989258,
      "step": 1476
    },
    {
      "epoch": 0.19967891846218844,
      "grad_norm": 0.2603908181190491,
      "learning_rate": 2.8572114351271955e-05,
      "loss": 0.06761407852172852,
      "step": 1477
    },
    {
      "epoch": 0.19981411068863542,
      "grad_norm": 0.7338619232177734,
      "learning_rate": 2.85691984400061e-05,
      "loss": 0.07612419128417969,
      "step": 1478
    },
    {
      "epoch": 0.1999493029150824,
      "grad_norm": 0.2574705183506012,
      "learning_rate": 2.8566279703560762e-05,
      "loss": 0.0436396598815918,
      "step": 1479
    },
    {
      "epoch": 0.20008449514152937,
      "grad_norm": 0.2783486545085907,
      "learning_rate": 2.8563358142543648e-05,
      "loss": 0.06040048599243164,
      "step": 1480
    },
    {
      "epoch": 0.20021968736797635,
      "grad_norm": 0.5714578628540039,
      "learning_rate": 2.856043375756304e-05,
      "loss": 0.08622550964355469,
      "step": 1481
    },
    {
      "epoch": 0.20035487959442333,
      "grad_norm": 0.6847553253173828,
      "learning_rate": 2.855750654922781e-05,
      "loss": 0.10153627395629883,
      "step": 1482
    },
    {
      "epoch": 0.2004900718208703,
      "grad_norm": 0.328132688999176,
      "learning_rate": 2.855457651814742e-05,
      "loss": 0.0824136734008789,
      "step": 1483
    },
    {
      "epoch": 0.2006252640473173,
      "grad_norm": 0.346091628074646,
      "learning_rate": 2.8551643664931916e-05,
      "loss": 0.08436298370361328,
      "step": 1484
    },
    {
      "epoch": 0.20076045627376427,
      "grad_norm": 0.47722601890563965,
      "learning_rate": 2.8548707990191933e-05,
      "loss": 0.10326790809631348,
      "step": 1485
    },
    {
      "epoch": 0.20089564850021124,
      "grad_norm": 0.41707944869995117,
      "learning_rate": 2.8545769494538698e-05,
      "loss": 0.06589221954345703,
      "step": 1486
    },
    {
      "epoch": 0.20103084072665822,
      "grad_norm": 0.5870721340179443,
      "learning_rate": 2.854282817858402e-05,
      "loss": 0.06289863586425781,
      "step": 1487
    },
    {
      "epoch": 0.2011660329531052,
      "grad_norm": 0.5807158350944519,
      "learning_rate": 2.85398840429403e-05,
      "loss": 0.11984395980834961,
      "step": 1488
    },
    {
      "epoch": 0.20130122517955218,
      "grad_norm": 0.4396728575229645,
      "learning_rate": 2.853693708822051e-05,
      "loss": 0.0967855453491211,
      "step": 1489
    },
    {
      "epoch": 0.20143641740599916,
      "grad_norm": 0.5051938891410828,
      "learning_rate": 2.8533987315038234e-05,
      "loss": 0.07933425903320312,
      "step": 1490
    },
    {
      "epoch": 0.20157160963244614,
      "grad_norm": 0.31764644384384155,
      "learning_rate": 2.8531034724007627e-05,
      "loss": 0.0799555778503418,
      "step": 1491
    },
    {
      "epoch": 0.20170680185889311,
      "grad_norm": 0.17717859148979187,
      "learning_rate": 2.8528079315743435e-05,
      "loss": 0.03476667404174805,
      "step": 1492
    },
    {
      "epoch": 0.2018419940853401,
      "grad_norm": 0.3987855613231659,
      "learning_rate": 2.852512109086099e-05,
      "loss": 0.06544375419616699,
      "step": 1493
    },
    {
      "epoch": 0.20197718631178707,
      "grad_norm": 0.6518162488937378,
      "learning_rate": 2.8522160049976208e-05,
      "loss": 0.0998842716217041,
      "step": 1494
    },
    {
      "epoch": 0.20211237853823405,
      "grad_norm": 0.5000519156455994,
      "learning_rate": 2.8519196193705595e-05,
      "loss": 0.05768442153930664,
      "step": 1495
    },
    {
      "epoch": 0.20224757076468103,
      "grad_norm": 0.6162581443786621,
      "learning_rate": 2.8516229522666243e-05,
      "loss": 0.0803002417087555,
      "step": 1496
    },
    {
      "epoch": 0.202382762991128,
      "grad_norm": 0.4996272623538971,
      "learning_rate": 2.8513260037475825e-05,
      "loss": 0.05600053071975708,
      "step": 1497
    },
    {
      "epoch": 0.20251795521757499,
      "grad_norm": 0.44334933161735535,
      "learning_rate": 2.8510287738752604e-05,
      "loss": 0.09145450592041016,
      "step": 1498
    },
    {
      "epoch": 0.20265314744402196,
      "grad_norm": 0.4339718520641327,
      "learning_rate": 2.8507312627115435e-05,
      "loss": 0.06299620866775513,
      "step": 1499
    },
    {
      "epoch": 0.20278833967046894,
      "grad_norm": 0.41716277599334717,
      "learning_rate": 2.850433470318374e-05,
      "loss": 0.052942752838134766,
      "step": 1500
    },
    {
      "epoch": 0.20292353189691592,
      "grad_norm": 0.4254150986671448,
      "learning_rate": 2.8501353967577556e-05,
      "loss": 0.07092094421386719,
      "step": 1501
    },
    {
      "epoch": 0.2030587241233629,
      "grad_norm": 0.314262330532074,
      "learning_rate": 2.8498370420917468e-05,
      "loss": 0.06371545791625977,
      "step": 1502
    },
    {
      "epoch": 0.20319391634980988,
      "grad_norm": 0.41127675771713257,
      "learning_rate": 2.8495384063824683e-05,
      "loss": 0.07239151000976562,
      "step": 1503
    },
    {
      "epoch": 0.20332910857625686,
      "grad_norm": 0.23433437943458557,
      "learning_rate": 2.8492394896920964e-05,
      "loss": 0.05965924263000488,
      "step": 1504
    },
    {
      "epoch": 0.20346430080270383,
      "grad_norm": 0.3009444773197174,
      "learning_rate": 2.848940292082868e-05,
      "loss": 0.06430244445800781,
      "step": 1505
    },
    {
      "epoch": 0.2035994930291508,
      "grad_norm": 0.48559582233428955,
      "learning_rate": 2.8486408136170772e-05,
      "loss": 0.0673685073852539,
      "step": 1506
    },
    {
      "epoch": 0.2037346852555978,
      "grad_norm": 0.443024218082428,
      "learning_rate": 2.8483410543570776e-05,
      "loss": 0.07785987854003906,
      "step": 1507
    },
    {
      "epoch": 0.20386987748204477,
      "grad_norm": 0.5127843022346497,
      "learning_rate": 2.8480410143652803e-05,
      "loss": 0.08401298522949219,
      "step": 1508
    },
    {
      "epoch": 0.20400506970849175,
      "grad_norm": 1.0754501819610596,
      "learning_rate": 2.8477406937041547e-05,
      "loss": 0.08812069892883301,
      "step": 1509
    },
    {
      "epoch": 0.20414026193493875,
      "grad_norm": 0.4283447563648224,
      "learning_rate": 2.8474400924362298e-05,
      "loss": 0.1021566390991211,
      "step": 1510
    },
    {
      "epoch": 0.20427545416138573,
      "grad_norm": 0.4467059373855591,
      "learning_rate": 2.847139210624092e-05,
      "loss": 0.05289793014526367,
      "step": 1511
    },
    {
      "epoch": 0.2044106463878327,
      "grad_norm": 0.46175262331962585,
      "learning_rate": 2.8468380483303873e-05,
      "loss": 0.08680331707000732,
      "step": 1512
    },
    {
      "epoch": 0.2045458386142797,
      "grad_norm": 0.7817302942276001,
      "learning_rate": 2.8465366056178183e-05,
      "loss": 0.056456804275512695,
      "step": 1513
    },
    {
      "epoch": 0.20468103084072667,
      "grad_norm": 0.6416090130805969,
      "learning_rate": 2.8462348825491475e-05,
      "loss": 0.06924200057983398,
      "step": 1514
    },
    {
      "epoch": 0.20481622306717365,
      "grad_norm": 0.4183678925037384,
      "learning_rate": 2.8459328791871953e-05,
      "loss": 0.10460662841796875,
      "step": 1515
    },
    {
      "epoch": 0.20495141529362063,
      "grad_norm": 0.4831472635269165,
      "learning_rate": 2.8456305955948402e-05,
      "loss": 0.06530570983886719,
      "step": 1516
    },
    {
      "epoch": 0.2050866075200676,
      "grad_norm": 0.5936373472213745,
      "learning_rate": 2.845328031835019e-05,
      "loss": 0.08108949661254883,
      "step": 1517
    },
    {
      "epoch": 0.20522179974651458,
      "grad_norm": 0.21675732731819153,
      "learning_rate": 2.8450251879707277e-05,
      "loss": 0.06800222396850586,
      "step": 1518
    },
    {
      "epoch": 0.20535699197296156,
      "grad_norm": 0.6820784211158752,
      "learning_rate": 2.8447220640650194e-05,
      "loss": 0.06974172592163086,
      "step": 1519
    },
    {
      "epoch": 0.20549218419940854,
      "grad_norm": 0.42916199564933777,
      "learning_rate": 2.8444186601810068e-05,
      "loss": 0.077606201171875,
      "step": 1520
    },
    {
      "epoch": 0.20562737642585552,
      "grad_norm": 0.5398802757263184,
      "learning_rate": 2.84411497638186e-05,
      "loss": 0.06264090538024902,
      "step": 1521
    },
    {
      "epoch": 0.2057625686523025,
      "grad_norm": 0.574612557888031,
      "learning_rate": 2.843811012730807e-05,
      "loss": 0.0909883975982666,
      "step": 1522
    },
    {
      "epoch": 0.20589776087874948,
      "grad_norm": 0.6658186912536621,
      "learning_rate": 2.8435067692911353e-05,
      "loss": 0.07992124557495117,
      "step": 1523
    },
    {
      "epoch": 0.20603295310519645,
      "grad_norm": 0.8445116877555847,
      "learning_rate": 2.8432022461261897e-05,
      "loss": 0.07997274398803711,
      "step": 1524
    },
    {
      "epoch": 0.20616814533164343,
      "grad_norm": 0.35964757204055786,
      "learning_rate": 2.8428974432993736e-05,
      "loss": 0.07648134231567383,
      "step": 1525
    },
    {
      "epoch": 0.2063033375580904,
      "grad_norm": 0.4669344127178192,
      "learning_rate": 2.8425923608741486e-05,
      "loss": 0.08966922760009766,
      "step": 1526
    },
    {
      "epoch": 0.2064385297845374,
      "grad_norm": 0.5233417749404907,
      "learning_rate": 2.8422869989140343e-05,
      "loss": 0.06952023506164551,
      "step": 1527
    },
    {
      "epoch": 0.20657372201098437,
      "grad_norm": 0.3246438205242157,
      "learning_rate": 2.8419813574826093e-05,
      "loss": 0.04368925094604492,
      "step": 1528
    },
    {
      "epoch": 0.20670891423743135,
      "grad_norm": 0.6084328889846802,
      "learning_rate": 2.8416754366435092e-05,
      "loss": 0.0888967514038086,
      "step": 1529
    },
    {
      "epoch": 0.20684410646387832,
      "grad_norm": 0.4416901767253876,
      "learning_rate": 2.8413692364604285e-05,
      "loss": 0.07677769660949707,
      "step": 1530
    },
    {
      "epoch": 0.2069792986903253,
      "grad_norm": 1.4339102506637573,
      "learning_rate": 2.8410627569971197e-05,
      "loss": 0.12319564819335938,
      "step": 1531
    },
    {
      "epoch": 0.20711449091677228,
      "grad_norm": 0.6371877193450928,
      "learning_rate": 2.8407559983173934e-05,
      "loss": 0.1146383285522461,
      "step": 1532
    },
    {
      "epoch": 0.20724968314321926,
      "grad_norm": 0.3475244641304016,
      "learning_rate": 2.8404489604851186e-05,
      "loss": 0.07754826545715332,
      "step": 1533
    },
    {
      "epoch": 0.20738487536966624,
      "grad_norm": 0.3506496250629425,
      "learning_rate": 2.840141643564222e-05,
      "loss": 0.055328369140625,
      "step": 1534
    },
    {
      "epoch": 0.20752006759611322,
      "grad_norm": 0.6898812055587769,
      "learning_rate": 2.8398340476186885e-05,
      "loss": 0.08753299713134766,
      "step": 1535
    },
    {
      "epoch": 0.2076552598225602,
      "grad_norm": 0.6379316449165344,
      "learning_rate": 2.8395261727125617e-05,
      "loss": 0.08071374893188477,
      "step": 1536
    },
    {
      "epoch": 0.20779045204900717,
      "grad_norm": 1.1470060348510742,
      "learning_rate": 2.8392180189099425e-05,
      "loss": 0.09624767303466797,
      "step": 1537
    },
    {
      "epoch": 0.20792564427545415,
      "grad_norm": 0.7213916182518005,
      "learning_rate": 2.83890958627499e-05,
      "loss": 0.07874691486358643,
      "step": 1538
    },
    {
      "epoch": 0.20806083650190113,
      "grad_norm": 0.25094074010849,
      "learning_rate": 2.8386008748719216e-05,
      "loss": 0.04964780807495117,
      "step": 1539
    },
    {
      "epoch": 0.2081960287283481,
      "grad_norm": 0.29679808020591736,
      "learning_rate": 2.838291884765013e-05,
      "loss": 0.07548999786376953,
      "step": 1540
    },
    {
      "epoch": 0.2083312209547951,
      "grad_norm": 0.6929604411125183,
      "learning_rate": 2.8379826160185975e-05,
      "loss": 0.09263134002685547,
      "step": 1541
    },
    {
      "epoch": 0.20846641318124207,
      "grad_norm": 0.5786704421043396,
      "learning_rate": 2.8376730686970664e-05,
      "loss": 0.0851593017578125,
      "step": 1542
    },
    {
      "epoch": 0.20860160540768904,
      "grad_norm": 0.1921105831861496,
      "learning_rate": 2.8373632428648683e-05,
      "loss": 0.054320335388183594,
      "step": 1543
    },
    {
      "epoch": 0.20873679763413605,
      "grad_norm": 0.3362789750099182,
      "learning_rate": 2.8370531385865124e-05,
      "loss": 0.05617380142211914,
      "step": 1544
    },
    {
      "epoch": 0.20887198986058303,
      "grad_norm": 0.5695424675941467,
      "learning_rate": 2.8367427559265622e-05,
      "loss": 0.05769157409667969,
      "step": 1545
    },
    {
      "epoch": 0.20900718208703,
      "grad_norm": 0.3748445212841034,
      "learning_rate": 2.836432094949642e-05,
      "loss": 0.08294010162353516,
      "step": 1546
    },
    {
      "epoch": 0.209142374313477,
      "grad_norm": 0.3538089096546173,
      "learning_rate": 2.836121155720433e-05,
      "loss": 0.06462764739990234,
      "step": 1547
    },
    {
      "epoch": 0.20927756653992396,
      "grad_norm": 0.42894431948661804,
      "learning_rate": 2.8358099383036745e-05,
      "loss": 0.0638885498046875,
      "step": 1548
    },
    {
      "epoch": 0.20941275876637094,
      "grad_norm": 0.47183895111083984,
      "learning_rate": 2.8354984427641634e-05,
      "loss": 0.06286382675170898,
      "step": 1549
    },
    {
      "epoch": 0.20954795099281792,
      "grad_norm": 0.2746015191078186,
      "learning_rate": 2.8351866691667544e-05,
      "loss": 0.0701746940612793,
      "step": 1550
    },
    {
      "epoch": 0.2096831432192649,
      "grad_norm": 0.28003767132759094,
      "learning_rate": 2.8348746175763613e-05,
      "loss": 0.0535731315612793,
      "step": 1551
    },
    {
      "epoch": 0.20981833544571188,
      "grad_norm": 0.4062836468219757,
      "learning_rate": 2.8345622880579537e-05,
      "loss": 0.07574129104614258,
      "step": 1552
    },
    {
      "epoch": 0.20995352767215886,
      "grad_norm": 0.6867547035217285,
      "learning_rate": 2.8342496806765615e-05,
      "loss": 0.08504390716552734,
      "step": 1553
    },
    {
      "epoch": 0.21008871989860584,
      "grad_norm": 0.14522002637386322,
      "learning_rate": 2.833936795497271e-05,
      "loss": 0.047537803649902344,
      "step": 1554
    },
    {
      "epoch": 0.21022391212505281,
      "grad_norm": 0.47608858346939087,
      "learning_rate": 2.8336236325852257e-05,
      "loss": 0.07120323181152344,
      "step": 1555
    },
    {
      "epoch": 0.2103591043514998,
      "grad_norm": 0.4264245927333832,
      "learning_rate": 2.8333101920056285e-05,
      "loss": 0.055731892585754395,
      "step": 1556
    },
    {
      "epoch": 0.21049429657794677,
      "grad_norm": 0.6420731544494629,
      "learning_rate": 2.8329964738237392e-05,
      "loss": 0.08244848251342773,
      "step": 1557
    },
    {
      "epoch": 0.21062948880439375,
      "grad_norm": 0.534919023513794,
      "learning_rate": 2.8326824781048756e-05,
      "loss": 0.09705448150634766,
      "step": 1558
    },
    {
      "epoch": 0.21076468103084073,
      "grad_norm": 0.9707580804824829,
      "learning_rate": 2.8323682049144135e-05,
      "loss": 0.09867000579833984,
      "step": 1559
    },
    {
      "epoch": 0.2108998732572877,
      "grad_norm": 0.37556371092796326,
      "learning_rate": 2.832053654317786e-05,
      "loss": 0.06475949287414551,
      "step": 1560
    },
    {
      "epoch": 0.21103506548373469,
      "grad_norm": 0.7855553030967712,
      "learning_rate": 2.8317388263804842e-05,
      "loss": 0.11359071731567383,
      "step": 1561
    },
    {
      "epoch": 0.21117025771018166,
      "grad_norm": 0.3888871669769287,
      "learning_rate": 2.8314237211680573e-05,
      "loss": 0.0841970443725586,
      "step": 1562
    },
    {
      "epoch": 0.21130544993662864,
      "grad_norm": 0.495535671710968,
      "learning_rate": 2.8311083387461118e-05,
      "loss": 0.07243537902832031,
      "step": 1563
    },
    {
      "epoch": 0.21144064216307562,
      "grad_norm": 0.18253490328788757,
      "learning_rate": 2.8307926791803114e-05,
      "loss": 0.05575847625732422,
      "step": 1564
    },
    {
      "epoch": 0.2115758343895226,
      "grad_norm": 0.5124557614326477,
      "learning_rate": 2.8304767425363785e-05,
      "loss": 0.07570600509643555,
      "step": 1565
    },
    {
      "epoch": 0.21171102661596958,
      "grad_norm": 0.28501805663108826,
      "learning_rate": 2.830160528880093e-05,
      "loss": 0.0680689811706543,
      "step": 1566
    },
    {
      "epoch": 0.21184621884241656,
      "grad_norm": 0.3124721646308899,
      "learning_rate": 2.829844038277292e-05,
      "loss": 0.07054710388183594,
      "step": 1567
    },
    {
      "epoch": 0.21198141106886353,
      "grad_norm": 0.35767754912376404,
      "learning_rate": 2.8295272707938706e-05,
      "loss": 0.07143974304199219,
      "step": 1568
    },
    {
      "epoch": 0.2121166032953105,
      "grad_norm": 0.6448807716369629,
      "learning_rate": 2.8292102264957817e-05,
      "loss": 0.0975027084350586,
      "step": 1569
    },
    {
      "epoch": 0.2122517955217575,
      "grad_norm": 0.5385596752166748,
      "learning_rate": 2.8288929054490357e-05,
      "loss": 0.08779716491699219,
      "step": 1570
    },
    {
      "epoch": 0.21238698774820447,
      "grad_norm": 0.35227349400520325,
      "learning_rate": 2.8285753077196998e-05,
      "loss": 0.0719599723815918,
      "step": 1571
    },
    {
      "epoch": 0.21252217997465145,
      "grad_norm": 0.4539269208908081,
      "learning_rate": 2.8282574333739006e-05,
      "loss": 0.06339693069458008,
      "step": 1572
    },
    {
      "epoch": 0.21265737220109843,
      "grad_norm": 0.8397480249404907,
      "learning_rate": 2.8279392824778197e-05,
      "loss": 0.0697946548461914,
      "step": 1573
    },
    {
      "epoch": 0.2127925644275454,
      "grad_norm": 0.6460810899734497,
      "learning_rate": 2.8276208550976993e-05,
      "loss": 0.06089729070663452,
      "step": 1574
    },
    {
      "epoch": 0.21292775665399238,
      "grad_norm": 0.4288603961467743,
      "learning_rate": 2.8273021512998372e-05,
      "loss": 0.087799072265625,
      "step": 1575
    },
    {
      "epoch": 0.21306294888043936,
      "grad_norm": 0.8934928774833679,
      "learning_rate": 2.826983171150589e-05,
      "loss": 0.07850122451782227,
      "step": 1576
    },
    {
      "epoch": 0.21319814110688634,
      "grad_norm": 0.47060516476631165,
      "learning_rate": 2.826663914716368e-05,
      "loss": 0.0952754020690918,
      "step": 1577
    },
    {
      "epoch": 0.21333333333333335,
      "grad_norm": 0.39937278628349304,
      "learning_rate": 2.826344382063646e-05,
      "loss": 0.07137489318847656,
      "step": 1578
    },
    {
      "epoch": 0.21346852555978033,
      "grad_norm": 0.8838233351707458,
      "learning_rate": 2.8260245732589503e-05,
      "loss": 0.08221101760864258,
      "step": 1579
    },
    {
      "epoch": 0.2136037177862273,
      "grad_norm": 0.6140885353088379,
      "learning_rate": 2.8257044883688672e-05,
      "loss": 0.06524944305419922,
      "step": 1580
    },
    {
      "epoch": 0.21373891001267428,
      "grad_norm": 0.33716484904289246,
      "learning_rate": 2.82538412746004e-05,
      "loss": 0.06624317169189453,
      "step": 1581
    },
    {
      "epoch": 0.21387410223912126,
      "grad_norm": 0.35295796394348145,
      "learning_rate": 2.8250634905991695e-05,
      "loss": 0.0657663345336914,
      "step": 1582
    },
    {
      "epoch": 0.21400929446556824,
      "grad_norm": 0.2585761845111847,
      "learning_rate": 2.824742577853015e-05,
      "loss": 0.046559810638427734,
      "step": 1583
    },
    {
      "epoch": 0.21414448669201522,
      "grad_norm": 0.34351542592048645,
      "learning_rate": 2.8244213892883907e-05,
      "loss": 0.07097244262695312,
      "step": 1584
    },
    {
      "epoch": 0.2142796789184622,
      "grad_norm": 0.20833387970924377,
      "learning_rate": 2.82409992497217e-05,
      "loss": 0.045954227447509766,
      "step": 1585
    },
    {
      "epoch": 0.21441487114490917,
      "grad_norm": 0.46430692076683044,
      "learning_rate": 2.8237781849712852e-05,
      "loss": 0.07614660263061523,
      "step": 1586
    },
    {
      "epoch": 0.21455006337135615,
      "grad_norm": 0.27918508648872375,
      "learning_rate": 2.8234561693527222e-05,
      "loss": 0.054819583892822266,
      "step": 1587
    },
    {
      "epoch": 0.21468525559780313,
      "grad_norm": 0.412960410118103,
      "learning_rate": 2.8231338781835275e-05,
      "loss": 0.07755041122436523,
      "step": 1588
    },
    {
      "epoch": 0.2148204478242501,
      "grad_norm": 0.4582512378692627,
      "learning_rate": 2.8228113115308032e-05,
      "loss": 0.10426759719848633,
      "step": 1589
    },
    {
      "epoch": 0.2149556400506971,
      "grad_norm": 0.3591464161872864,
      "learning_rate": 2.82248846946171e-05,
      "loss": 0.06746578216552734,
      "step": 1590
    },
    {
      "epoch": 0.21509083227714407,
      "grad_norm": 0.5530175566673279,
      "learning_rate": 2.822165352043465e-05,
      "loss": 0.06784582138061523,
      "step": 1591
    },
    {
      "epoch": 0.21522602450359105,
      "grad_norm": 0.8045269846916199,
      "learning_rate": 2.8218419593433437e-05,
      "loss": 0.08627581596374512,
      "step": 1592
    },
    {
      "epoch": 0.21536121673003802,
      "grad_norm": 0.20450256764888763,
      "learning_rate": 2.8215182914286768e-05,
      "loss": 0.0470733642578125,
      "step": 1593
    },
    {
      "epoch": 0.215496408956485,
      "grad_norm": 0.38736405968666077,
      "learning_rate": 2.8211943483668546e-05,
      "loss": 0.08690547943115234,
      "step": 1594
    },
    {
      "epoch": 0.21563160118293198,
      "grad_norm": 0.15070442855358124,
      "learning_rate": 2.8208701302253237e-05,
      "loss": 0.04215669631958008,
      "step": 1595
    },
    {
      "epoch": 0.21576679340937896,
      "grad_norm": 0.4521065354347229,
      "learning_rate": 2.820545637071588e-05,
      "loss": 0.08721351623535156,
      "step": 1596
    },
    {
      "epoch": 0.21590198563582594,
      "grad_norm": 0.3432156443595886,
      "learning_rate": 2.8202208689732083e-05,
      "loss": 0.056870460510253906,
      "step": 1597
    },
    {
      "epoch": 0.21603717786227292,
      "grad_norm": 0.39810627698898315,
      "learning_rate": 2.819895825997804e-05,
      "loss": 0.03982114791870117,
      "step": 1598
    },
    {
      "epoch": 0.2161723700887199,
      "grad_norm": 0.4973376393318176,
      "learning_rate": 2.81957050821305e-05,
      "loss": 0.07609415054321289,
      "step": 1599
    },
    {
      "epoch": 0.21630756231516687,
      "grad_norm": 0.6641979217529297,
      "learning_rate": 2.8192449156866787e-05,
      "loss": 0.09238433837890625,
      "step": 1600
    },
    {
      "epoch": 0.21644275454161385,
      "grad_norm": 0.7539771795272827,
      "learning_rate": 2.8189190484864814e-05,
      "loss": 0.09517097473144531,
      "step": 1601
    },
    {
      "epoch": 0.21657794676806083,
      "grad_norm": 0.3724718987941742,
      "learning_rate": 2.8185929066803052e-05,
      "loss": 0.05844688415527344,
      "step": 1602
    },
    {
      "epoch": 0.2167131389945078,
      "grad_norm": 0.37707340717315674,
      "learning_rate": 2.818266490336054e-05,
      "loss": 0.06653022766113281,
      "step": 1603
    },
    {
      "epoch": 0.2168483312209548,
      "grad_norm": 0.3917069435119629,
      "learning_rate": 2.817939799521689e-05,
      "loss": 0.07715415954589844,
      "step": 1604
    },
    {
      "epoch": 0.21698352344740177,
      "grad_norm": 0.4703545570373535,
      "learning_rate": 2.8176128343052304e-05,
      "loss": 0.07887458801269531,
      "step": 1605
    },
    {
      "epoch": 0.21711871567384874,
      "grad_norm": 0.5951199531555176,
      "learning_rate": 2.817285594754753e-05,
      "loss": 0.10241222381591797,
      "step": 1606
    },
    {
      "epoch": 0.21725390790029572,
      "grad_norm": 0.28272005915641785,
      "learning_rate": 2.8169580809383902e-05,
      "loss": 0.06273269653320312,
      "step": 1607
    },
    {
      "epoch": 0.2173891001267427,
      "grad_norm": 0.3859010636806488,
      "learning_rate": 2.8166302929243326e-05,
      "loss": 0.07463669776916504,
      "step": 1608
    },
    {
      "epoch": 0.21752429235318968,
      "grad_norm": 0.4278121888637543,
      "learning_rate": 2.8163022307808264e-05,
      "loss": 0.05964255332946777,
      "step": 1609
    },
    {
      "epoch": 0.21765948457963666,
      "grad_norm": 0.8931379318237305,
      "learning_rate": 2.8159738945761764e-05,
      "loss": 0.09549093246459961,
      "step": 1610
    },
    {
      "epoch": 0.21779467680608364,
      "grad_norm": 0.41075560450553894,
      "learning_rate": 2.8156452843787438e-05,
      "loss": 0.06782150268554688,
      "step": 1611
    },
    {
      "epoch": 0.21792986903253064,
      "grad_norm": 0.3972751498222351,
      "learning_rate": 2.815316400256947e-05,
      "loss": 0.10162878036499023,
      "step": 1612
    },
    {
      "epoch": 0.21806506125897762,
      "grad_norm": 0.5238962173461914,
      "learning_rate": 2.814987242279262e-05,
      "loss": 0.08918380737304688,
      "step": 1613
    },
    {
      "epoch": 0.2182002534854246,
      "grad_norm": 0.5222808122634888,
      "learning_rate": 2.8146578105142202e-05,
      "loss": 0.09671163558959961,
      "step": 1614
    },
    {
      "epoch": 0.21833544571187158,
      "grad_norm": 0.3641374111175537,
      "learning_rate": 2.814328105030412e-05,
      "loss": 0.06192207336425781,
      "step": 1615
    },
    {
      "epoch": 0.21847063793831856,
      "grad_norm": 0.2877354621887207,
      "learning_rate": 2.8139981258964836e-05,
      "loss": 0.0652461051940918,
      "step": 1616
    },
    {
      "epoch": 0.21860583016476554,
      "grad_norm": 0.4902604818344116,
      "learning_rate": 2.8136678731811385e-05,
      "loss": 0.053981781005859375,
      "step": 1617
    },
    {
      "epoch": 0.21874102239121251,
      "grad_norm": 0.6506996750831604,
      "learning_rate": 2.8133373469531362e-05,
      "loss": 0.0872354507446289,
      "step": 1618
    },
    {
      "epoch": 0.2188762146176595,
      "grad_norm": 0.22775883972644806,
      "learning_rate": 2.8130065472812952e-05,
      "loss": 0.0394824743270874,
      "step": 1619
    },
    {
      "epoch": 0.21901140684410647,
      "grad_norm": 0.20165197551250458,
      "learning_rate": 2.812675474234489e-05,
      "loss": 0.04616832733154297,
      "step": 1620
    },
    {
      "epoch": 0.21914659907055345,
      "grad_norm": 0.6121368408203125,
      "learning_rate": 2.812344127881649e-05,
      "loss": 0.1114950180053711,
      "step": 1621
    },
    {
      "epoch": 0.21928179129700043,
      "grad_norm": 0.844108521938324,
      "learning_rate": 2.8120125082917638e-05,
      "loss": 0.07841968536376953,
      "step": 1622
    },
    {
      "epoch": 0.2194169835234474,
      "grad_norm": 0.48570796847343445,
      "learning_rate": 2.8116806155338773e-05,
      "loss": 0.07401180267333984,
      "step": 1623
    },
    {
      "epoch": 0.21955217574989438,
      "grad_norm": 0.6665063500404358,
      "learning_rate": 2.8113484496770923e-05,
      "loss": 0.05527907609939575,
      "step": 1624
    },
    {
      "epoch": 0.21968736797634136,
      "grad_norm": 1.430572748184204,
      "learning_rate": 2.811016010790567e-05,
      "loss": 0.095458984375,
      "step": 1625
    },
    {
      "epoch": 0.21982256020278834,
      "grad_norm": 0.8118380308151245,
      "learning_rate": 2.8106832989435165e-05,
      "loss": 0.0848240852355957,
      "step": 1626
    },
    {
      "epoch": 0.21995775242923532,
      "grad_norm": 0.3135979175567627,
      "learning_rate": 2.8103503142052146e-05,
      "loss": 0.0822296142578125,
      "step": 1627
    },
    {
      "epoch": 0.2200929446556823,
      "grad_norm": 0.32224681973457336,
      "learning_rate": 2.8100170566449892e-05,
      "loss": 0.06378364562988281,
      "step": 1628
    },
    {
      "epoch": 0.22022813688212928,
      "grad_norm": 0.4593490660190582,
      "learning_rate": 2.8096835263322266e-05,
      "loss": 0.0670623779296875,
      "step": 1629
    },
    {
      "epoch": 0.22036332910857626,
      "grad_norm": 0.3958438038825989,
      "learning_rate": 2.8093497233363702e-05,
      "loss": 0.0702054500579834,
      "step": 1630
    },
    {
      "epoch": 0.22049852133502323,
      "grad_norm": 0.44892239570617676,
      "learning_rate": 2.8090156477269185e-05,
      "loss": 0.054297685623168945,
      "step": 1631
    },
    {
      "epoch": 0.2206337135614702,
      "grad_norm": 0.6927587985992432,
      "learning_rate": 2.808681299573429e-05,
      "loss": 0.0662841796875,
      "step": 1632
    },
    {
      "epoch": 0.2207689057879172,
      "grad_norm": 0.444137305021286,
      "learning_rate": 2.8083466789455137e-05,
      "loss": 0.07687711715698242,
      "step": 1633
    },
    {
      "epoch": 0.22090409801436417,
      "grad_norm": 0.39730507135391235,
      "learning_rate": 2.808011785912843e-05,
      "loss": 0.0796651840209961,
      "step": 1634
    },
    {
      "epoch": 0.22103929024081115,
      "grad_norm": 0.2972898483276367,
      "learning_rate": 2.8076766205451435e-05,
      "loss": 0.055123090744018555,
      "step": 1635
    },
    {
      "epoch": 0.22117448246725813,
      "grad_norm": 0.18792884051799774,
      "learning_rate": 2.8073411829121983e-05,
      "loss": 0.0420079231262207,
      "step": 1636
    },
    {
      "epoch": 0.2213096746937051,
      "grad_norm": 0.36996304988861084,
      "learning_rate": 2.8070054730838467e-05,
      "loss": 0.0798807144165039,
      "step": 1637
    },
    {
      "epoch": 0.22144486692015208,
      "grad_norm": 0.4978063702583313,
      "learning_rate": 2.8066694911299865e-05,
      "loss": 0.1259927749633789,
      "step": 1638
    },
    {
      "epoch": 0.22158005914659906,
      "grad_norm": 0.5838686227798462,
      "learning_rate": 2.8063332371205698e-05,
      "loss": 0.08549308776855469,
      "step": 1639
    },
    {
      "epoch": 0.22171525137304604,
      "grad_norm": 0.36738187074661255,
      "learning_rate": 2.8059967111256072e-05,
      "loss": 0.05120038986206055,
      "step": 1640
    },
    {
      "epoch": 0.22185044359949302,
      "grad_norm": 0.2991965413093567,
      "learning_rate": 2.8056599132151647e-05,
      "loss": 0.07931041717529297,
      "step": 1641
    },
    {
      "epoch": 0.22198563582594,
      "grad_norm": 0.4698663055896759,
      "learning_rate": 2.8053228434593656e-05,
      "loss": 0.06342899799346924,
      "step": 1642
    },
    {
      "epoch": 0.22212082805238698,
      "grad_norm": 0.4911721348762512,
      "learning_rate": 2.8049855019283895e-05,
      "loss": 0.09208106994628906,
      "step": 1643
    },
    {
      "epoch": 0.22225602027883395,
      "grad_norm": 0.4793623089790344,
      "learning_rate": 2.8046478886924736e-05,
      "loss": 0.06871199607849121,
      "step": 1644
    },
    {
      "epoch": 0.22239121250528093,
      "grad_norm": 0.28812339901924133,
      "learning_rate": 2.804310003821909e-05,
      "loss": 0.0816950798034668,
      "step": 1645
    },
    {
      "epoch": 0.22252640473172794,
      "grad_norm": 0.515812337398529,
      "learning_rate": 2.8039718473870473e-05,
      "loss": 0.09521055221557617,
      "step": 1646
    },
    {
      "epoch": 0.22266159695817492,
      "grad_norm": 0.3579423725605011,
      "learning_rate": 2.8036334194582924e-05,
      "loss": 0.06319046020507812,
      "step": 1647
    },
    {
      "epoch": 0.2227967891846219,
      "grad_norm": 0.2613706588745117,
      "learning_rate": 2.8032947201061084e-05,
      "loss": 0.052468299865722656,
      "step": 1648
    },
    {
      "epoch": 0.22293198141106887,
      "grad_norm": 0.54111647605896,
      "learning_rate": 2.8029557494010132e-05,
      "loss": 0.0862126350402832,
      "step": 1649
    },
    {
      "epoch": 0.22306717363751585,
      "grad_norm": 0.5427489876747131,
      "learning_rate": 2.802616507413583e-05,
      "loss": 0.07244467735290527,
      "step": 1650
    },
    {
      "epoch": 0.22320236586396283,
      "grad_norm": 0.3109368681907654,
      "learning_rate": 2.8022769942144492e-05,
      "loss": 0.08296823501586914,
      "step": 1651
    },
    {
      "epoch": 0.2233375580904098,
      "grad_norm": 0.3801068961620331,
      "learning_rate": 2.801937209874301e-05,
      "loss": 0.08475017547607422,
      "step": 1652
    },
    {
      "epoch": 0.2234727503168568,
      "grad_norm": 0.30102646350860596,
      "learning_rate": 2.8015971544638832e-05,
      "loss": 0.05987656116485596,
      "step": 1653
    },
    {
      "epoch": 0.22360794254330377,
      "grad_norm": 0.42254638671875,
      "learning_rate": 2.8012568280539964e-05,
      "loss": 0.0904998779296875,
      "step": 1654
    },
    {
      "epoch": 0.22374313476975075,
      "grad_norm": 0.22949378192424774,
      "learning_rate": 2.800916230715499e-05,
      "loss": 0.0606616735458374,
      "step": 1655
    },
    {
      "epoch": 0.22387832699619772,
      "grad_norm": 0.9997631907463074,
      "learning_rate": 2.800575362519305e-05,
      "loss": 0.08878564834594727,
      "step": 1656
    },
    {
      "epoch": 0.2240135192226447,
      "grad_norm": 0.28981316089630127,
      "learning_rate": 2.800234223536385e-05,
      "loss": 0.04744911193847656,
      "step": 1657
    },
    {
      "epoch": 0.22414871144909168,
      "grad_norm": 0.8237676620483398,
      "learning_rate": 2.799892813837766e-05,
      "loss": 0.09437572956085205,
      "step": 1658
    },
    {
      "epoch": 0.22428390367553866,
      "grad_norm": 0.45472365617752075,
      "learning_rate": 2.7995511334945315e-05,
      "loss": 0.08826065063476562,
      "step": 1659
    },
    {
      "epoch": 0.22441909590198564,
      "grad_norm": 0.4593025743961334,
      "learning_rate": 2.7992091825778202e-05,
      "loss": 0.06944799423217773,
      "step": 1660
    },
    {
      "epoch": 0.22455428812843262,
      "grad_norm": 0.7557271122932434,
      "learning_rate": 2.7988669611588295e-05,
      "loss": 0.0856008529663086,
      "step": 1661
    },
    {
      "epoch": 0.2246894803548796,
      "grad_norm": 0.29140231013298035,
      "learning_rate": 2.7985244693088112e-05,
      "loss": 0.028855204582214355,
      "step": 1662
    },
    {
      "epoch": 0.22482467258132657,
      "grad_norm": 0.5522841811180115,
      "learning_rate": 2.7981817070990736e-05,
      "loss": 0.07449054718017578,
      "step": 1663
    },
    {
      "epoch": 0.22495986480777355,
      "grad_norm": 0.3269520401954651,
      "learning_rate": 2.7978386746009813e-05,
      "loss": 0.060117244720458984,
      "step": 1664
    },
    {
      "epoch": 0.22509505703422053,
      "grad_norm": 0.7490774989128113,
      "learning_rate": 2.797495371885957e-05,
      "loss": 0.08272123336791992,
      "step": 1665
    },
    {
      "epoch": 0.2252302492606675,
      "grad_norm": 0.5157455205917358,
      "learning_rate": 2.7971517990254768e-05,
      "loss": 0.06467342376708984,
      "step": 1666
    },
    {
      "epoch": 0.2253654414871145,
      "grad_norm": 0.6056908965110779,
      "learning_rate": 2.7968079560910744e-05,
      "loss": 0.0678703784942627,
      "step": 1667
    },
    {
      "epoch": 0.22550063371356147,
      "grad_norm": 0.8360289931297302,
      "learning_rate": 2.7964638431543402e-05,
      "loss": 0.062014102935791016,
      "step": 1668
    },
    {
      "epoch": 0.22563582594000844,
      "grad_norm": 0.650890588760376,
      "learning_rate": 2.7961194602869208e-05,
      "loss": 0.07659244537353516,
      "step": 1669
    },
    {
      "epoch": 0.22577101816645542,
      "grad_norm": 0.4710063636302948,
      "learning_rate": 2.7957748075605178e-05,
      "loss": 0.107208251953125,
      "step": 1670
    },
    {
      "epoch": 0.2259062103929024,
      "grad_norm": 0.3793172836303711,
      "learning_rate": 2.7954298850468898e-05,
      "loss": 0.07379865646362305,
      "step": 1671
    },
    {
      "epoch": 0.22604140261934938,
      "grad_norm": 0.5283216238021851,
      "learning_rate": 2.7950846928178517e-05,
      "loss": 0.09037399291992188,
      "step": 1672
    },
    {
      "epoch": 0.22617659484579636,
      "grad_norm": 0.5233229994773865,
      "learning_rate": 2.7947392309452744e-05,
      "loss": 0.06802845001220703,
      "step": 1673
    },
    {
      "epoch": 0.22631178707224334,
      "grad_norm": 0.5740306973457336,
      "learning_rate": 2.7943934995010845e-05,
      "loss": 0.07895135879516602,
      "step": 1674
    },
    {
      "epoch": 0.22644697929869032,
      "grad_norm": 0.22923940420150757,
      "learning_rate": 2.7940474985572657e-05,
      "loss": 0.04120969772338867,
      "step": 1675
    },
    {
      "epoch": 0.2265821715251373,
      "grad_norm": 0.8615206480026245,
      "learning_rate": 2.793701228185857e-05,
      "loss": 0.05845475196838379,
      "step": 1676
    },
    {
      "epoch": 0.22671736375158427,
      "grad_norm": 0.25676557421684265,
      "learning_rate": 2.7933546884589536e-05,
      "loss": 0.053670406341552734,
      "step": 1677
    },
    {
      "epoch": 0.22685255597803125,
      "grad_norm": 0.4690135419368744,
      "learning_rate": 2.7930078794487077e-05,
      "loss": 0.0667567253112793,
      "step": 1678
    },
    {
      "epoch": 0.22698774820447823,
      "grad_norm": 0.3204026222229004,
      "learning_rate": 2.7926608012273253e-05,
      "loss": 0.06004142761230469,
      "step": 1679
    },
    {
      "epoch": 0.22712294043092524,
      "grad_norm": 0.6366397738456726,
      "learning_rate": 2.7923134538670715e-05,
      "loss": 0.08652496337890625,
      "step": 1680
    },
    {
      "epoch": 0.2272581326573722,
      "grad_norm": 0.31071799993515015,
      "learning_rate": 2.7919658374402645e-05,
      "loss": 0.04079151153564453,
      "step": 1681
    },
    {
      "epoch": 0.2273933248838192,
      "grad_norm": 0.3043774664402008,
      "learning_rate": 2.7916179520192807e-05,
      "loss": 0.06686258316040039,
      "step": 1682
    },
    {
      "epoch": 0.22752851711026617,
      "grad_norm": 0.26399528980255127,
      "learning_rate": 2.7912697976765516e-05,
      "loss": 0.07759618759155273,
      "step": 1683
    },
    {
      "epoch": 0.22766370933671315,
      "grad_norm": 0.34320852160453796,
      "learning_rate": 2.790921374484565e-05,
      "loss": 0.06002044677734375,
      "step": 1684
    },
    {
      "epoch": 0.22779890156316013,
      "grad_norm": 0.2710177004337311,
      "learning_rate": 2.7905726825158637e-05,
      "loss": 0.08067035675048828,
      "step": 1685
    },
    {
      "epoch": 0.2279340937896071,
      "grad_norm": 0.32556334137916565,
      "learning_rate": 2.7902237218430485e-05,
      "loss": 0.05906534194946289,
      "step": 1686
    },
    {
      "epoch": 0.22806928601605408,
      "grad_norm": 0.27595052123069763,
      "learning_rate": 2.7898744925387735e-05,
      "loss": 0.061810970306396484,
      "step": 1687
    },
    {
      "epoch": 0.22820447824250106,
      "grad_norm": 0.2853275537490845,
      "learning_rate": 2.7895249946757505e-05,
      "loss": 0.0636434555053711,
      "step": 1688
    },
    {
      "epoch": 0.22833967046894804,
      "grad_norm": 0.5397588014602661,
      "learning_rate": 2.7891752283267474e-05,
      "loss": 0.09965991973876953,
      "step": 1689
    },
    {
      "epoch": 0.22847486269539502,
      "grad_norm": 0.1745423674583435,
      "learning_rate": 2.788825193564587e-05,
      "loss": 0.03982114791870117,
      "step": 1690
    },
    {
      "epoch": 0.228610054921842,
      "grad_norm": 0.13676410913467407,
      "learning_rate": 2.7884748904621483e-05,
      "loss": 0.027382850646972656,
      "step": 1691
    },
    {
      "epoch": 0.22874524714828898,
      "grad_norm": 0.2747132480144501,
      "learning_rate": 2.7881243190923667e-05,
      "loss": 0.06270027160644531,
      "step": 1692
    },
    {
      "epoch": 0.22888043937473596,
      "grad_norm": 0.7449238300323486,
      "learning_rate": 2.7877734795282326e-05,
      "loss": 0.07269144058227539,
      "step": 1693
    },
    {
      "epoch": 0.22901563160118293,
      "grad_norm": 0.67307448387146,
      "learning_rate": 2.7874223718427926e-05,
      "loss": 0.09056568145751953,
      "step": 1694
    },
    {
      "epoch": 0.2291508238276299,
      "grad_norm": 0.3201822340488434,
      "learning_rate": 2.78707099610915e-05,
      "loss": 0.04983329772949219,
      "step": 1695
    },
    {
      "epoch": 0.2292860160540769,
      "grad_norm": 0.3325496017932892,
      "learning_rate": 2.7867193524004618e-05,
      "loss": 0.06076478958129883,
      "step": 1696
    },
    {
      "epoch": 0.22942120828052387,
      "grad_norm": 0.4683951437473297,
      "learning_rate": 2.786367440789943e-05,
      "loss": 0.05465984344482422,
      "step": 1697
    },
    {
      "epoch": 0.22955640050697085,
      "grad_norm": 0.3375021815299988,
      "learning_rate": 2.7860152613508634e-05,
      "loss": 0.06065797805786133,
      "step": 1698
    },
    {
      "epoch": 0.22969159273341783,
      "grad_norm": 1.1951357126235962,
      "learning_rate": 2.7856628141565484e-05,
      "loss": 0.10079479217529297,
      "step": 1699
    },
    {
      "epoch": 0.2298267849598648,
      "grad_norm": 0.1562008112668991,
      "learning_rate": 2.7853100992803797e-05,
      "loss": 0.05537271499633789,
      "step": 1700
    },
    {
      "epoch": 0.22996197718631178,
      "grad_norm": 0.3340827524662018,
      "learning_rate": 2.7849571167957942e-05,
      "loss": 0.08322334289550781,
      "step": 1701
    },
    {
      "epoch": 0.23009716941275876,
      "grad_norm": 0.4469181001186371,
      "learning_rate": 2.784603866776285e-05,
      "loss": 0.10549259185791016,
      "step": 1702
    },
    {
      "epoch": 0.23023236163920574,
      "grad_norm": 0.42036375403404236,
      "learning_rate": 2.7842503492953996e-05,
      "loss": 0.07880759239196777,
      "step": 1703
    },
    {
      "epoch": 0.23036755386565272,
      "grad_norm": 0.4152294099330902,
      "learning_rate": 2.7838965644267435e-05,
      "loss": 0.07308101654052734,
      "step": 1704
    },
    {
      "epoch": 0.2305027460920997,
      "grad_norm": 0.30345314741134644,
      "learning_rate": 2.7835425122439764e-05,
      "loss": 0.050995826721191406,
      "step": 1705
    },
    {
      "epoch": 0.23063793831854668,
      "grad_norm": 0.5056906342506409,
      "learning_rate": 2.7831881928208128e-05,
      "loss": 0.06304597854614258,
      "step": 1706
    },
    {
      "epoch": 0.23077313054499365,
      "grad_norm": 0.32819733023643494,
      "learning_rate": 2.7828336062310252e-05,
      "loss": 0.07190895080566406,
      "step": 1707
    },
    {
      "epoch": 0.23090832277144063,
      "grad_norm": 0.5538195371627808,
      "learning_rate": 2.7824787525484403e-05,
      "loss": 0.07466936111450195,
      "step": 1708
    },
    {
      "epoch": 0.2310435149978876,
      "grad_norm": 0.5046731233596802,
      "learning_rate": 2.7821236318469395e-05,
      "loss": 0.08994722366333008,
      "step": 1709
    },
    {
      "epoch": 0.2311787072243346,
      "grad_norm": 0.400930792093277,
      "learning_rate": 2.7817682442004615e-05,
      "loss": 0.08697342872619629,
      "step": 1710
    },
    {
      "epoch": 0.23131389945078157,
      "grad_norm": 0.3092827796936035,
      "learning_rate": 2.781412589683e-05,
      "loss": 0.07714653015136719,
      "step": 1711
    },
    {
      "epoch": 0.23144909167722855,
      "grad_norm": 0.4910641312599182,
      "learning_rate": 2.781056668368604e-05,
      "loss": 0.06731009483337402,
      "step": 1712
    },
    {
      "epoch": 0.23158428390367553,
      "grad_norm": 0.3656303286552429,
      "learning_rate": 2.780700480331378e-05,
      "loss": 0.06447601318359375,
      "step": 1713
    },
    {
      "epoch": 0.23171947613012253,
      "grad_norm": 0.3311465382575989,
      "learning_rate": 2.7803440256454825e-05,
      "loss": 0.07804012298583984,
      "step": 1714
    },
    {
      "epoch": 0.2318546683565695,
      "grad_norm": 0.6129467487335205,
      "learning_rate": 2.7799873043851337e-05,
      "loss": 0.059590816497802734,
      "step": 1715
    },
    {
      "epoch": 0.2319898605830165,
      "grad_norm": 0.7785227298736572,
      "learning_rate": 2.7796303166246016e-05,
      "loss": 0.08037972450256348,
      "step": 1716
    },
    {
      "epoch": 0.23212505280946347,
      "grad_norm": 0.7429837584495544,
      "learning_rate": 2.7792730624382142e-05,
      "loss": 0.09062862396240234,
      "step": 1717
    },
    {
      "epoch": 0.23226024503591045,
      "grad_norm": 0.42086148262023926,
      "learning_rate": 2.778915541900353e-05,
      "loss": 0.10192680358886719,
      "step": 1718
    },
    {
      "epoch": 0.23239543726235742,
      "grad_norm": 0.3895472586154938,
      "learning_rate": 2.7785577550854566e-05,
      "loss": 0.07087564468383789,
      "step": 1719
    },
    {
      "epoch": 0.2325306294888044,
      "grad_norm": 0.9704498648643494,
      "learning_rate": 2.778199702068017e-05,
      "loss": 0.07492542266845703,
      "step": 1720
    },
    {
      "epoch": 0.23266582171525138,
      "grad_norm": 0.6616615653038025,
      "learning_rate": 2.777841382922583e-05,
      "loss": 0.07393407821655273,
      "step": 1721
    },
    {
      "epoch": 0.23280101394169836,
      "grad_norm": 0.6751999258995056,
      "learning_rate": 2.7774827977237596e-05,
      "loss": 0.067596435546875,
      "step": 1722
    },
    {
      "epoch": 0.23293620616814534,
      "grad_norm": 0.22478896379470825,
      "learning_rate": 2.777123946546205e-05,
      "loss": 0.05893135070800781,
      "step": 1723
    },
    {
      "epoch": 0.23307139839459232,
      "grad_norm": 0.4378553628921509,
      "learning_rate": 2.776764829464634e-05,
      "loss": 0.06891918182373047,
      "step": 1724
    },
    {
      "epoch": 0.2332065906210393,
      "grad_norm": 0.5406183004379272,
      "learning_rate": 2.7764054465538173e-05,
      "loss": 0.0537571907043457,
      "step": 1725
    },
    {
      "epoch": 0.23334178284748627,
      "grad_norm": 0.5965280532836914,
      "learning_rate": 2.7760457978885794e-05,
      "loss": 0.06615829467773438,
      "step": 1726
    },
    {
      "epoch": 0.23347697507393325,
      "grad_norm": 0.4570900499820709,
      "learning_rate": 2.7756858835438022e-05,
      "loss": 0.07481217384338379,
      "step": 1727
    },
    {
      "epoch": 0.23361216730038023,
      "grad_norm": 0.4667648673057556,
      "learning_rate": 2.7753257035944216e-05,
      "loss": 0.07844877243041992,
      "step": 1728
    },
    {
      "epoch": 0.2337473595268272,
      "grad_norm": 0.2384541779756546,
      "learning_rate": 2.7749652581154277e-05,
      "loss": 0.047064781188964844,
      "step": 1729
    },
    {
      "epoch": 0.2338825517532742,
      "grad_norm": 0.2297324538230896,
      "learning_rate": 2.7746045471818685e-05,
      "loss": 0.049700260162353516,
      "step": 1730
    },
    {
      "epoch": 0.23401774397972117,
      "grad_norm": 0.4015132784843445,
      "learning_rate": 2.7742435708688458e-05,
      "loss": 0.06686007976531982,
      "step": 1731
    },
    {
      "epoch": 0.23415293620616814,
      "grad_norm": 0.6935897469520569,
      "learning_rate": 2.7738823292515167e-05,
      "loss": 0.09031033515930176,
      "step": 1732
    },
    {
      "epoch": 0.23428812843261512,
      "grad_norm": 1.1761062145233154,
      "learning_rate": 2.773520822405093e-05,
      "loss": 0.09777450561523438,
      "step": 1733
    },
    {
      "epoch": 0.2344233206590621,
      "grad_norm": 0.39272940158843994,
      "learning_rate": 2.7731590504048433e-05,
      "loss": 0.06181478500366211,
      "step": 1734
    },
    {
      "epoch": 0.23455851288550908,
      "grad_norm": 0.33699238300323486,
      "learning_rate": 2.7727970133260896e-05,
      "loss": 0.07762861251831055,
      "step": 1735
    },
    {
      "epoch": 0.23469370511195606,
      "grad_norm": 0.2577589452266693,
      "learning_rate": 2.7724347112442106e-05,
      "loss": 0.04999685287475586,
      "step": 1736
    },
    {
      "epoch": 0.23482889733840304,
      "grad_norm": 0.4214746654033661,
      "learning_rate": 2.772072144234639e-05,
      "loss": 0.05493927001953125,
      "step": 1737
    },
    {
      "epoch": 0.23496408956485001,
      "grad_norm": 0.23358161747455597,
      "learning_rate": 2.7717093123728634e-05,
      "loss": 0.06423664093017578,
      "step": 1738
    },
    {
      "epoch": 0.235099281791297,
      "grad_norm": 0.4160080552101135,
      "learning_rate": 2.771346215734428e-05,
      "loss": 0.07363462448120117,
      "step": 1739
    },
    {
      "epoch": 0.23523447401774397,
      "grad_norm": 0.5243673324584961,
      "learning_rate": 2.7709828543949302e-05,
      "loss": 0.07423591613769531,
      "step": 1740
    },
    {
      "epoch": 0.23536966624419095,
      "grad_norm": 0.5465552806854248,
      "learning_rate": 2.770619228430025e-05,
      "loss": 0.0896914005279541,
      "step": 1741
    },
    {
      "epoch": 0.23550485847063793,
      "grad_norm": 0.4041135013103485,
      "learning_rate": 2.77025533791542e-05,
      "loss": 0.06969207525253296,
      "step": 1742
    },
    {
      "epoch": 0.2356400506970849,
      "grad_norm": 0.33528199791908264,
      "learning_rate": 2.76989118292688e-05,
      "loss": 0.055374860763549805,
      "step": 1743
    },
    {
      "epoch": 0.23577524292353189,
      "grad_norm": 0.37491998076438904,
      "learning_rate": 2.7695267635402242e-05,
      "loss": 0.08498287200927734,
      "step": 1744
    },
    {
      "epoch": 0.23591043514997886,
      "grad_norm": 0.5534172058105469,
      "learning_rate": 2.7691620798313258e-05,
      "loss": 0.05818986892700195,
      "step": 1745
    },
    {
      "epoch": 0.23604562737642584,
      "grad_norm": 0.5367288589477539,
      "learning_rate": 2.7687971318761145e-05,
      "loss": 0.054492950439453125,
      "step": 1746
    },
    {
      "epoch": 0.23618081960287282,
      "grad_norm": 0.2770763039588928,
      "learning_rate": 2.7684319197505746e-05,
      "loss": 0.060950636863708496,
      "step": 1747
    },
    {
      "epoch": 0.23631601182931983,
      "grad_norm": 0.20724846422672272,
      "learning_rate": 2.7680664435307446e-05,
      "loss": 0.04057741165161133,
      "step": 1748
    },
    {
      "epoch": 0.2364512040557668,
      "grad_norm": 0.587720513343811,
      "learning_rate": 2.767700703292719e-05,
      "loss": 0.059874534606933594,
      "step": 1749
    },
    {
      "epoch": 0.23658639628221378,
      "grad_norm": 0.4023610055446625,
      "learning_rate": 2.767334699112647e-05,
      "loss": 0.09053611755371094,
      "step": 1750
    },
    {
      "epoch": 0.23672158850866076,
      "grad_norm": 0.4142555892467499,
      "learning_rate": 2.7669684310667318e-05,
      "loss": 0.038106441497802734,
      "step": 1751
    },
    {
      "epoch": 0.23685678073510774,
      "grad_norm": 0.2784874439239502,
      "learning_rate": 2.7666018992312333e-05,
      "loss": 0.06513595581054688,
      "step": 1752
    },
    {
      "epoch": 0.23699197296155472,
      "grad_norm": 0.3611937463283539,
      "learning_rate": 2.7662351036824653e-05,
      "loss": 0.07321715354919434,
      "step": 1753
    },
    {
      "epoch": 0.2371271651880017,
      "grad_norm": 0.3308567702770233,
      "learning_rate": 2.7658680444967964e-05,
      "loss": 0.06647872924804688,
      "step": 1754
    },
    {
      "epoch": 0.23726235741444868,
      "grad_norm": 0.4900350272655487,
      "learning_rate": 2.76550072175065e-05,
      "loss": 0.08567094802856445,
      "step": 1755
    },
    {
      "epoch": 0.23739754964089566,
      "grad_norm": 0.30941900610923767,
      "learning_rate": 2.7651331355205044e-05,
      "loss": 0.07367181777954102,
      "step": 1756
    },
    {
      "epoch": 0.23753274186734263,
      "grad_norm": 0.39354702830314636,
      "learning_rate": 2.7647652858828936e-05,
      "loss": 0.06630945205688477,
      "step": 1757
    },
    {
      "epoch": 0.2376679340937896,
      "grad_norm": 0.41852644085884094,
      "learning_rate": 2.764397172914406e-05,
      "loss": 0.09297752380371094,
      "step": 1758
    },
    {
      "epoch": 0.2378031263202366,
      "grad_norm": 0.2813112139701843,
      "learning_rate": 2.7640287966916845e-05,
      "loss": 0.05391216278076172,
      "step": 1759
    },
    {
      "epoch": 0.23793831854668357,
      "grad_norm": 0.3069011867046356,
      "learning_rate": 2.7636601572914266e-05,
      "loss": 0.06717586517333984,
      "step": 1760
    },
    {
      "epoch": 0.23807351077313055,
      "grad_norm": 0.3833242952823639,
      "learning_rate": 2.7632912547903855e-05,
      "loss": 0.05843085050582886,
      "step": 1761
    },
    {
      "epoch": 0.23820870299957753,
      "grad_norm": 0.3839249610900879,
      "learning_rate": 2.7629220892653685e-05,
      "loss": 0.06429481506347656,
      "step": 1762
    },
    {
      "epoch": 0.2383438952260245,
      "grad_norm": 0.22615458071231842,
      "learning_rate": 2.7625526607932378e-05,
      "loss": 0.05566549301147461,
      "step": 1763
    },
    {
      "epoch": 0.23847908745247148,
      "grad_norm": 0.5173673033714294,
      "learning_rate": 2.76218296945091e-05,
      "loss": 0.06610107421875,
      "step": 1764
    },
    {
      "epoch": 0.23861427967891846,
      "grad_norm": 0.25685936212539673,
      "learning_rate": 2.7618130153153577e-05,
      "loss": 0.056848883628845215,
      "step": 1765
    },
    {
      "epoch": 0.23874947190536544,
      "grad_norm": 0.1981397122144699,
      "learning_rate": 2.7614427984636063e-05,
      "loss": 0.040857791900634766,
      "step": 1766
    },
    {
      "epoch": 0.23888466413181242,
      "grad_norm": 0.25202375650405884,
      "learning_rate": 2.7610723189727377e-05,
      "loss": 0.05035686492919922,
      "step": 1767
    },
    {
      "epoch": 0.2390198563582594,
      "grad_norm": 0.47765684127807617,
      "learning_rate": 2.760701576919888e-05,
      "loss": 0.06260061264038086,
      "step": 1768
    },
    {
      "epoch": 0.23915504858470638,
      "grad_norm": 0.29305171966552734,
      "learning_rate": 2.760330572382246e-05,
      "loss": 0.07577896118164062,
      "step": 1769
    },
    {
      "epoch": 0.23929024081115335,
      "grad_norm": 0.21195778250694275,
      "learning_rate": 2.7599593054370584e-05,
      "loss": 0.050594329833984375,
      "step": 1770
    },
    {
      "epoch": 0.23942543303760033,
      "grad_norm": 0.31517109274864197,
      "learning_rate": 2.7595877761616246e-05,
      "loss": 0.07395100593566895,
      "step": 1771
    },
    {
      "epoch": 0.2395606252640473,
      "grad_norm": 0.42024487257003784,
      "learning_rate": 2.759215984633299e-05,
      "loss": 0.06496191024780273,
      "step": 1772
    },
    {
      "epoch": 0.2396958174904943,
      "grad_norm": 0.4478231966495514,
      "learning_rate": 2.7588439309294902e-05,
      "loss": 0.051871299743652344,
      "step": 1773
    },
    {
      "epoch": 0.23983100971694127,
      "grad_norm": 0.4402880072593689,
      "learning_rate": 2.7584716151276623e-05,
      "loss": 0.08018779754638672,
      "step": 1774
    },
    {
      "epoch": 0.23996620194338825,
      "grad_norm": 0.26607251167297363,
      "learning_rate": 2.7580990373053325e-05,
      "loss": 0.056462764739990234,
      "step": 1775
    },
    {
      "epoch": 0.24010139416983522,
      "grad_norm": 0.3490714132785797,
      "learning_rate": 2.7577261975400747e-05,
      "loss": 0.0739583969116211,
      "step": 1776
    },
    {
      "epoch": 0.2402365863962822,
      "grad_norm": 0.5762818455696106,
      "learning_rate": 2.7573530959095154e-05,
      "loss": 0.06903731822967529,
      "step": 1777
    },
    {
      "epoch": 0.24037177862272918,
      "grad_norm": 0.3279045820236206,
      "learning_rate": 2.756979732491336e-05,
      "loss": 0.06824302673339844,
      "step": 1778
    },
    {
      "epoch": 0.24050697084917616,
      "grad_norm": 0.3041938245296478,
      "learning_rate": 2.756606107363274e-05,
      "loss": 0.042186737060546875,
      "step": 1779
    },
    {
      "epoch": 0.24064216307562314,
      "grad_norm": 0.22868023812770844,
      "learning_rate": 2.7562322206031192e-05,
      "loss": 0.05701124668121338,
      "step": 1780
    },
    {
      "epoch": 0.24077735530207012,
      "grad_norm": 0.5037597417831421,
      "learning_rate": 2.7558580722887166e-05,
      "loss": 0.07582688331604004,
      "step": 1781
    },
    {
      "epoch": 0.24091254752851712,
      "grad_norm": 0.24145066738128662,
      "learning_rate": 2.7554836624979666e-05,
      "loss": 0.054518699645996094,
      "step": 1782
    },
    {
      "epoch": 0.2410477397549641,
      "grad_norm": 0.6626120209693909,
      "learning_rate": 2.7551089913088233e-05,
      "loss": 0.07623720169067383,
      "step": 1783
    },
    {
      "epoch": 0.24118293198141108,
      "grad_norm": 0.5188962817192078,
      "learning_rate": 2.7547340587992948e-05,
      "loss": 0.08408737182617188,
      "step": 1784
    },
    {
      "epoch": 0.24131812420785806,
      "grad_norm": 0.4455700218677521,
      "learning_rate": 2.754358865047444e-05,
      "loss": 0.06111133098602295,
      "step": 1785
    },
    {
      "epoch": 0.24145331643430504,
      "grad_norm": 0.5123838186264038,
      "learning_rate": 2.7539834101313885e-05,
      "loss": 0.086700439453125,
      "step": 1786
    },
    {
      "epoch": 0.24158850866075202,
      "grad_norm": 0.9112415909767151,
      "learning_rate": 2.7536076941293003e-05,
      "loss": 0.0907973051071167,
      "step": 1787
    },
    {
      "epoch": 0.241723700887199,
      "grad_norm": 0.20850281417369843,
      "learning_rate": 2.753231717119405e-05,
      "loss": 0.04787421226501465,
      "step": 1788
    },
    {
      "epoch": 0.24185889311364597,
      "grad_norm": 0.24106447398662567,
      "learning_rate": 2.7528554791799826e-05,
      "loss": 0.04145240783691406,
      "step": 1789
    },
    {
      "epoch": 0.24199408534009295,
      "grad_norm": 0.19539408385753632,
      "learning_rate": 2.7524789803893686e-05,
      "loss": 0.04960012435913086,
      "step": 1790
    },
    {
      "epoch": 0.24212927756653993,
      "grad_norm": 0.44446125626564026,
      "learning_rate": 2.7521022208259526e-05,
      "loss": 0.0738372802734375,
      "step": 1791
    },
    {
      "epoch": 0.2422644697929869,
      "grad_norm": 0.5379305481910706,
      "learning_rate": 2.7517252005681762e-05,
      "loss": 0.08936834335327148,
      "step": 1792
    },
    {
      "epoch": 0.2423996620194339,
      "grad_norm": 0.4252738952636719,
      "learning_rate": 2.7513479196945385e-05,
      "loss": 0.0784444808959961,
      "step": 1793
    },
    {
      "epoch": 0.24253485424588087,
      "grad_norm": 0.32151463627815247,
      "learning_rate": 2.750970378283591e-05,
      "loss": 0.07025814056396484,
      "step": 1794
    },
    {
      "epoch": 0.24267004647232784,
      "grad_norm": 0.3168907165527344,
      "learning_rate": 2.7505925764139398e-05,
      "loss": 0.05860328674316406,
      "step": 1795
    },
    {
      "epoch": 0.24280523869877482,
      "grad_norm": 0.18452592194080353,
      "learning_rate": 2.7502145141642447e-05,
      "loss": 0.03277397155761719,
      "step": 1796
    },
    {
      "epoch": 0.2429404309252218,
      "grad_norm": 0.5366207361221313,
      "learning_rate": 2.7498361916132212e-05,
      "loss": 0.06469535827636719,
      "step": 1797
    },
    {
      "epoch": 0.24307562315166878,
      "grad_norm": 0.3622032105922699,
      "learning_rate": 2.7494576088396376e-05,
      "loss": 0.06041765213012695,
      "step": 1798
    },
    {
      "epoch": 0.24321081537811576,
      "grad_norm": 0.5622307062149048,
      "learning_rate": 2.749078765922317e-05,
      "loss": 0.08621740341186523,
      "step": 1799
    },
    {
      "epoch": 0.24334600760456274,
      "grad_norm": 0.47309064865112305,
      "learning_rate": 2.7486996629401366e-05,
      "loss": 0.07869148254394531,
      "step": 1800
    },
    {
      "epoch": 0.24348119983100971,
      "grad_norm": 0.4613445997238159,
      "learning_rate": 2.7483202999720272e-05,
      "loss": 0.0908050537109375,
      "step": 1801
    },
    {
      "epoch": 0.2436163920574567,
      "grad_norm": 0.4424324035644531,
      "learning_rate": 2.7479406770969747e-05,
      "loss": 0.06970596313476562,
      "step": 1802
    },
    {
      "epoch": 0.24375158428390367,
      "grad_norm": 0.2568359375,
      "learning_rate": 2.7475607943940182e-05,
      "loss": 0.05686187744140625,
      "step": 1803
    },
    {
      "epoch": 0.24388677651035065,
      "grad_norm": 0.3203972280025482,
      "learning_rate": 2.7471806519422514e-05,
      "loss": 0.08163166046142578,
      "step": 1804
    },
    {
      "epoch": 0.24402196873679763,
      "grad_norm": 0.2624284625053406,
      "learning_rate": 2.746800249820822e-05,
      "loss": 0.038188934326171875,
      "step": 1805
    },
    {
      "epoch": 0.2441571609632446,
      "grad_norm": 0.7388049960136414,
      "learning_rate": 2.7464195881089323e-05,
      "loss": 0.10936880111694336,
      "step": 1806
    },
    {
      "epoch": 0.24429235318969159,
      "grad_norm": 0.41418957710266113,
      "learning_rate": 2.746038666885837e-05,
      "loss": 0.06546401977539062,
      "step": 1807
    },
    {
      "epoch": 0.24442754541613856,
      "grad_norm": 0.3561664819717407,
      "learning_rate": 2.7456574862308474e-05,
      "loss": 0.12682628631591797,
      "step": 1808
    },
    {
      "epoch": 0.24456273764258554,
      "grad_norm": 0.2145920991897583,
      "learning_rate": 2.745276046223326e-05,
      "loss": 0.041588783264160156,
      "step": 1809
    },
    {
      "epoch": 0.24469792986903252,
      "grad_norm": 0.8844074010848999,
      "learning_rate": 2.744894346942691e-05,
      "loss": 0.07974708080291748,
      "step": 1810
    },
    {
      "epoch": 0.2448331220954795,
      "grad_norm": 0.38096532225608826,
      "learning_rate": 2.744512388468415e-05,
      "loss": 0.07446169853210449,
      "step": 1811
    },
    {
      "epoch": 0.24496831432192648,
      "grad_norm": 0.3055894374847412,
      "learning_rate": 2.7441301708800227e-05,
      "loss": 0.058235883712768555,
      "step": 1812
    },
    {
      "epoch": 0.24510350654837346,
      "grad_norm": 0.23488426208496094,
      "learning_rate": 2.7437476942570942e-05,
      "loss": 0.048816680908203125,
      "step": 1813
    },
    {
      "epoch": 0.24523869877482044,
      "grad_norm": 0.316966712474823,
      "learning_rate": 2.7433649586792637e-05,
      "loss": 0.03723907470703125,
      "step": 1814
    },
    {
      "epoch": 0.2453738910012674,
      "grad_norm": 0.5726296305656433,
      "learning_rate": 2.7429819642262178e-05,
      "loss": 0.05982494354248047,
      "step": 1815
    },
    {
      "epoch": 0.24550908322771442,
      "grad_norm": 0.21773973107337952,
      "learning_rate": 2.7425987109776994e-05,
      "loss": 0.056328773498535156,
      "step": 1816
    },
    {
      "epoch": 0.2456442754541614,
      "grad_norm": 1.0662094354629517,
      "learning_rate": 2.7422151990135022e-05,
      "loss": 0.08037757873535156,
      "step": 1817
    },
    {
      "epoch": 0.24577946768060838,
      "grad_norm": 0.6202515959739685,
      "learning_rate": 2.741831428413477e-05,
      "loss": 0.09058094024658203,
      "step": 1818
    },
    {
      "epoch": 0.24591465990705536,
      "grad_norm": 0.29102861881256104,
      "learning_rate": 2.7414473992575257e-05,
      "loss": 0.06721258163452148,
      "step": 1819
    },
    {
      "epoch": 0.24604985213350233,
      "grad_norm": 0.5228894352912903,
      "learning_rate": 2.7410631116256054e-05,
      "loss": 0.07520627975463867,
      "step": 1820
    },
    {
      "epoch": 0.2461850443599493,
      "grad_norm": 0.7071523666381836,
      "learning_rate": 2.7406785655977275e-05,
      "loss": 0.06624984741210938,
      "step": 1821
    },
    {
      "epoch": 0.2463202365863963,
      "grad_norm": 0.6517912745475769,
      "learning_rate": 2.7402937612539563e-05,
      "loss": 0.08530855178833008,
      "step": 1822
    },
    {
      "epoch": 0.24645542881284327,
      "grad_norm": 0.2140028029680252,
      "learning_rate": 2.7399086986744095e-05,
      "loss": 0.058829307556152344,
      "step": 1823
    },
    {
      "epoch": 0.24659062103929025,
      "grad_norm": 0.24932214617729187,
      "learning_rate": 2.7395233779392598e-05,
      "loss": 0.056301116943359375,
      "step": 1824
    },
    {
      "epoch": 0.24672581326573723,
      "grad_norm": 1.0421650409698486,
      "learning_rate": 2.739137799128733e-05,
      "loss": 0.07622814178466797,
      "step": 1825
    },
    {
      "epoch": 0.2468610054921842,
      "grad_norm": 0.2780892550945282,
      "learning_rate": 2.7387519623231085e-05,
      "loss": 0.05814766883850098,
      "step": 1826
    },
    {
      "epoch": 0.24699619771863118,
      "grad_norm": 0.29341742396354675,
      "learning_rate": 2.7383658676027195e-05,
      "loss": 0.06454992294311523,
      "step": 1827
    },
    {
      "epoch": 0.24713138994507816,
      "grad_norm": 0.3836059868335724,
      "learning_rate": 2.7379795150479535e-05,
      "loss": 0.06870555877685547,
      "step": 1828
    },
    {
      "epoch": 0.24726658217152514,
      "grad_norm": 0.20684897899627686,
      "learning_rate": 2.73759290473925e-05,
      "loss": 0.053259849548339844,
      "step": 1829
    },
    {
      "epoch": 0.24740177439797212,
      "grad_norm": 0.6141616702079773,
      "learning_rate": 2.7372060367571044e-05,
      "loss": 0.09223604202270508,
      "step": 1830
    },
    {
      "epoch": 0.2475369666244191,
      "grad_norm": 0.5665993094444275,
      "learning_rate": 2.7368189111820648e-05,
      "loss": 0.0918726921081543,
      "step": 1831
    },
    {
      "epoch": 0.24767215885086608,
      "grad_norm": 0.5302583575248718,
      "learning_rate": 2.736431528094732e-05,
      "loss": 0.05847883224487305,
      "step": 1832
    },
    {
      "epoch": 0.24780735107731305,
      "grad_norm": 0.508675754070282,
      "learning_rate": 2.7360438875757614e-05,
      "loss": 0.07937335968017578,
      "step": 1833
    },
    {
      "epoch": 0.24794254330376003,
      "grad_norm": 0.39748477935791016,
      "learning_rate": 2.7356559897058624e-05,
      "loss": 0.07612800598144531,
      "step": 1834
    },
    {
      "epoch": 0.248077735530207,
      "grad_norm": 0.3022492229938507,
      "learning_rate": 2.735267834565797e-05,
      "loss": 0.052533626556396484,
      "step": 1835
    },
    {
      "epoch": 0.248212927756654,
      "grad_norm": 0.8229854702949524,
      "learning_rate": 2.734879422236381e-05,
      "loss": 0.07980060577392578,
      "step": 1836
    },
    {
      "epoch": 0.24834811998310097,
      "grad_norm": 0.8079227209091187,
      "learning_rate": 2.734490752798484e-05,
      "loss": 0.11629676818847656,
      "step": 1837
    },
    {
      "epoch": 0.24848331220954795,
      "grad_norm": 0.24918903410434723,
      "learning_rate": 2.7341018263330296e-05,
      "loss": 0.04086637496948242,
      "step": 1838
    },
    {
      "epoch": 0.24861850443599492,
      "grad_norm": 0.30910560488700867,
      "learning_rate": 2.7337126429209935e-05,
      "loss": 0.06302356719970703,
      "step": 1839
    },
    {
      "epoch": 0.2487536966624419,
      "grad_norm": 0.5225250124931335,
      "learning_rate": 2.7333232026434064e-05,
      "loss": 0.07401227951049805,
      "step": 1840
    },
    {
      "epoch": 0.24888888888888888,
      "grad_norm": 0.2577703297138214,
      "learning_rate": 2.7329335055813517e-05,
      "loss": 0.06687498092651367,
      "step": 1841
    },
    {
      "epoch": 0.24902408111533586,
      "grad_norm": 0.18020059168338776,
      "learning_rate": 2.732543551815966e-05,
      "loss": 0.049753665924072266,
      "step": 1842
    },
    {
      "epoch": 0.24915927334178284,
      "grad_norm": 0.8789290189743042,
      "learning_rate": 2.7321533414284404e-05,
      "loss": 0.06753873825073242,
      "step": 1843
    },
    {
      "epoch": 0.24929446556822982,
      "grad_norm": 0.3964804708957672,
      "learning_rate": 2.731762874500018e-05,
      "loss": 0.09026479721069336,
      "step": 1844
    },
    {
      "epoch": 0.2494296577946768,
      "grad_norm": 0.7734335660934448,
      "learning_rate": 2.7313721511119972e-05,
      "loss": 0.06596851348876953,
      "step": 1845
    },
    {
      "epoch": 0.24956485002112377,
      "grad_norm": 0.5335857272148132,
      "learning_rate": 2.7309811713457275e-05,
      "loss": 0.06571030616760254,
      "step": 1846
    },
    {
      "epoch": 0.24970004224757075,
      "grad_norm": 0.3945905566215515,
      "learning_rate": 2.730589935282614e-05,
      "loss": 0.06314945220947266,
      "step": 1847
    },
    {
      "epoch": 0.24983523447401773,
      "grad_norm": 0.7017887234687805,
      "learning_rate": 2.7301984430041135e-05,
      "loss": 0.06937360763549805,
      "step": 1848
    },
    {
      "epoch": 0.2499704267004647,
      "grad_norm": 0.31867197155952454,
      "learning_rate": 2.7298066945917368e-05,
      "loss": 0.0736551284790039,
      "step": 1849
    },
    {
      "epoch": 0.2501056189269117,
      "grad_norm": 0.31929993629455566,
      "learning_rate": 2.7294146901270482e-05,
      "loss": 0.06351780891418457,
      "step": 1850
    },
    {
      "epoch": 0.25024081115335867,
      "grad_norm": 0.41168221831321716,
      "learning_rate": 2.7290224296916653e-05,
      "loss": 0.07644474506378174,
      "step": 1851
    },
    {
      "epoch": 0.25037600337980565,
      "grad_norm": 0.38052400946617126,
      "learning_rate": 2.7286299133672584e-05,
      "loss": 0.060194969177246094,
      "step": 1852
    },
    {
      "epoch": 0.2505111956062526,
      "grad_norm": 0.7387686967849731,
      "learning_rate": 2.728237141235552e-05,
      "loss": 0.05945873260498047,
      "step": 1853
    },
    {
      "epoch": 0.2506463878326996,
      "grad_norm": 0.5313815474510193,
      "learning_rate": 2.727844113378322e-05,
      "loss": 0.06591439247131348,
      "step": 1854
    },
    {
      "epoch": 0.2507815800591466,
      "grad_norm": 0.5547090172767639,
      "learning_rate": 2.7274508298774013e-05,
      "loss": 0.06976318359375,
      "step": 1855
    },
    {
      "epoch": 0.25091677228559356,
      "grad_norm": 0.3434223234653473,
      "learning_rate": 2.727057290814672e-05,
      "loss": 0.08040809631347656,
      "step": 1856
    },
    {
      "epoch": 0.25105196451204054,
      "grad_norm": 0.4341026246547699,
      "learning_rate": 2.7266634962720704e-05,
      "loss": 0.04680442810058594,
      "step": 1857
    },
    {
      "epoch": 0.2511871567384875,
      "grad_norm": 0.4787321984767914,
      "learning_rate": 2.726269446331588e-05,
      "loss": 0.07925605773925781,
      "step": 1858
    },
    {
      "epoch": 0.2513223489649345,
      "grad_norm": 0.3505917489528656,
      "learning_rate": 2.7258751410752676e-05,
      "loss": 0.07482719421386719,
      "step": 1859
    },
    {
      "epoch": 0.2514575411913815,
      "grad_norm": 0.27366140484809875,
      "learning_rate": 2.725480580585206e-05,
      "loss": 0.05466175079345703,
      "step": 1860
    },
    {
      "epoch": 0.25159273341782845,
      "grad_norm": 0.4962812662124634,
      "learning_rate": 2.7250857649435522e-05,
      "loss": 0.07026433944702148,
      "step": 1861
    },
    {
      "epoch": 0.25172792564427543,
      "grad_norm": 0.41084927320480347,
      "learning_rate": 2.724690694232509e-05,
      "loss": 0.05806446075439453,
      "step": 1862
    },
    {
      "epoch": 0.2518631178707224,
      "grad_norm": 0.5985238552093506,
      "learning_rate": 2.7242953685343327e-05,
      "loss": 0.08927154541015625,
      "step": 1863
    },
    {
      "epoch": 0.2519983100971694,
      "grad_norm": 0.6761572360992432,
      "learning_rate": 2.723899787931332e-05,
      "loss": 0.07208061218261719,
      "step": 1864
    },
    {
      "epoch": 0.25213350232361637,
      "grad_norm": 0.29676154255867004,
      "learning_rate": 2.7235039525058684e-05,
      "loss": 0.06667649745941162,
      "step": 1865
    },
    {
      "epoch": 0.25226869455006334,
      "grad_norm": 0.33624520897865295,
      "learning_rate": 2.7231078623403575e-05,
      "loss": 0.05866742134094238,
      "step": 1866
    },
    {
      "epoch": 0.2524038867765104,
      "grad_norm": 0.3481168746948242,
      "learning_rate": 2.722711517517267e-05,
      "loss": 0.07056522369384766,
      "step": 1867
    },
    {
      "epoch": 0.25253907900295736,
      "grad_norm": 0.9190338850021362,
      "learning_rate": 2.7223149181191187e-05,
      "loss": 0.10958647727966309,
      "step": 1868
    },
    {
      "epoch": 0.25267427122940433,
      "grad_norm": 0.6682137250900269,
      "learning_rate": 2.7219180642284864e-05,
      "loss": 0.08075964450836182,
      "step": 1869
    },
    {
      "epoch": 0.2528094634558513,
      "grad_norm": 0.5249190926551819,
      "learning_rate": 2.721520955927997e-05,
      "loss": 0.07635307312011719,
      "step": 1870
    },
    {
      "epoch": 0.2529446556822983,
      "grad_norm": 0.583173394203186,
      "learning_rate": 2.7211235933003302e-05,
      "loss": 0.09204816818237305,
      "step": 1871
    },
    {
      "epoch": 0.25307984790874527,
      "grad_norm": 0.29399970173835754,
      "learning_rate": 2.72072597642822e-05,
      "loss": 0.044519901275634766,
      "step": 1872
    },
    {
      "epoch": 0.25321504013519225,
      "grad_norm": 0.4802243709564209,
      "learning_rate": 2.7203281053944512e-05,
      "loss": 0.09960746765136719,
      "step": 1873
    },
    {
      "epoch": 0.2533502323616392,
      "grad_norm": 0.3301074206829071,
      "learning_rate": 2.719929980281864e-05,
      "loss": 0.06789875030517578,
      "step": 1874
    },
    {
      "epoch": 0.2534854245880862,
      "grad_norm": 0.5058059692382812,
      "learning_rate": 2.719531601173349e-05,
      "loss": 0.08098983764648438,
      "step": 1875
    },
    {
      "epoch": 0.2536206168145332,
      "grad_norm": 0.3935385048389435,
      "learning_rate": 2.7191329681518512e-05,
      "loss": 0.0812234878540039,
      "step": 1876
    },
    {
      "epoch": 0.25375580904098016,
      "grad_norm": 0.3673516511917114,
      "learning_rate": 2.7187340813003682e-05,
      "loss": 0.04439401626586914,
      "step": 1877
    },
    {
      "epoch": 0.25389100126742714,
      "grad_norm": 0.31120064854621887,
      "learning_rate": 2.718334940701951e-05,
      "loss": 0.05800420045852661,
      "step": 1878
    },
    {
      "epoch": 0.2540261934938741,
      "grad_norm": 0.2556016743183136,
      "learning_rate": 2.7179355464397014e-05,
      "loss": 0.033266305923461914,
      "step": 1879
    },
    {
      "epoch": 0.2541613857203211,
      "grad_norm": 0.24700996279716492,
      "learning_rate": 2.7175358985967763e-05,
      "loss": 0.08549118041992188,
      "step": 1880
    },
    {
      "epoch": 0.2542965779467681,
      "grad_norm": 0.2749456763267517,
      "learning_rate": 2.717135997256385e-05,
      "loss": 0.060967206954956055,
      "step": 1881
    },
    {
      "epoch": 0.25443177017321505,
      "grad_norm": 0.6183525323867798,
      "learning_rate": 2.7167358425017882e-05,
      "loss": 0.07579755783081055,
      "step": 1882
    },
    {
      "epoch": 0.25456696239966203,
      "grad_norm": 0.21796010434627533,
      "learning_rate": 2.7163354344163004e-05,
      "loss": 0.06002235412597656,
      "step": 1883
    },
    {
      "epoch": 0.254702154626109,
      "grad_norm": 0.24536041915416718,
      "learning_rate": 2.715934773083289e-05,
      "loss": 0.04146528244018555,
      "step": 1884
    },
    {
      "epoch": 0.254837346852556,
      "grad_norm": 0.3926338851451874,
      "learning_rate": 2.715533858586174e-05,
      "loss": 0.08445155620574951,
      "step": 1885
    },
    {
      "epoch": 0.25497253907900297,
      "grad_norm": 0.1608278751373291,
      "learning_rate": 2.715132691008427e-05,
      "loss": 0.04503297805786133,
      "step": 1886
    },
    {
      "epoch": 0.25510773130544995,
      "grad_norm": 0.31084445118904114,
      "learning_rate": 2.714731270433574e-05,
      "loss": 0.07880449295043945,
      "step": 1887
    },
    {
      "epoch": 0.2552429235318969,
      "grad_norm": 0.5509818196296692,
      "learning_rate": 2.7143295969451933e-05,
      "loss": 0.05958402156829834,
      "step": 1888
    },
    {
      "epoch": 0.2553781157583439,
      "grad_norm": 0.3491239845752716,
      "learning_rate": 2.7139276706269147e-05,
      "loss": 0.04583573341369629,
      "step": 1889
    },
    {
      "epoch": 0.2555133079847909,
      "grad_norm": 0.3660595417022705,
      "learning_rate": 2.7135254915624213e-05,
      "loss": 0.05745553970336914,
      "step": 1890
    },
    {
      "epoch": 0.25564850021123786,
      "grad_norm": 0.3180834650993347,
      "learning_rate": 2.7131230598354497e-05,
      "loss": 0.07939624786376953,
      "step": 1891
    },
    {
      "epoch": 0.25578369243768484,
      "grad_norm": 0.3096720576286316,
      "learning_rate": 2.712720375529787e-05,
      "loss": 0.06689834594726562,
      "step": 1892
    },
    {
      "epoch": 0.2559188846641318,
      "grad_norm": 0.8833802938461304,
      "learning_rate": 2.7123174387292758e-05,
      "loss": 0.10140687227249146,
      "step": 1893
    },
    {
      "epoch": 0.2560540768905788,
      "grad_norm": 0.5502578616142273,
      "learning_rate": 2.7119142495178088e-05,
      "loss": 0.09019899368286133,
      "step": 1894
    },
    {
      "epoch": 0.2561892691170258,
      "grad_norm": 0.2593752443790436,
      "learning_rate": 2.711510807979333e-05,
      "loss": 0.052825212478637695,
      "step": 1895
    },
    {
      "epoch": 0.25632446134347275,
      "grad_norm": 0.5093520283699036,
      "learning_rate": 2.7111071141978452e-05,
      "loss": 0.06151437759399414,
      "step": 1896
    },
    {
      "epoch": 0.25645965356991973,
      "grad_norm": 0.3877309262752533,
      "learning_rate": 2.7107031682573987e-05,
      "loss": 0.043610453605651855,
      "step": 1897
    },
    {
      "epoch": 0.2565948457963667,
      "grad_norm": 0.3071913719177246,
      "learning_rate": 2.710298970242096e-05,
      "loss": 0.07933568954467773,
      "step": 1898
    },
    {
      "epoch": 0.2567300380228137,
      "grad_norm": 0.2128777951002121,
      "learning_rate": 2.7098945202360937e-05,
      "loss": 0.059468746185302734,
      "step": 1899
    },
    {
      "epoch": 0.25686523024926067,
      "grad_norm": 0.40854331851005554,
      "learning_rate": 2.7094898183236e-05,
      "loss": 0.093597412109375,
      "step": 1900
    },
    {
      "epoch": 0.25700042247570765,
      "grad_norm": 0.48837989568710327,
      "learning_rate": 2.709084864588877e-05,
      "loss": 0.08818244934082031,
      "step": 1901
    },
    {
      "epoch": 0.2571356147021546,
      "grad_norm": 0.22727102041244507,
      "learning_rate": 2.708679659116237e-05,
      "loss": 0.034447431564331055,
      "step": 1902
    },
    {
      "epoch": 0.2572708069286016,
      "grad_norm": 0.2908400595188141,
      "learning_rate": 2.708274201990047e-05,
      "loss": 0.06956076622009277,
      "step": 1903
    },
    {
      "epoch": 0.2574059991550486,
      "grad_norm": 0.29992619156837463,
      "learning_rate": 2.7078684932947247e-05,
      "loss": 0.07156085968017578,
      "step": 1904
    },
    {
      "epoch": 0.25754119138149556,
      "grad_norm": 0.490669846534729,
      "learning_rate": 2.7074625331147407e-05,
      "loss": 0.09055781364440918,
      "step": 1905
    },
    {
      "epoch": 0.25767638360794254,
      "grad_norm": 0.474657267332077,
      "learning_rate": 2.7070563215346184e-05,
      "loss": 0.0897374153137207,
      "step": 1906
    },
    {
      "epoch": 0.2578115758343895,
      "grad_norm": 0.4891012907028198,
      "learning_rate": 2.7066498586389332e-05,
      "loss": 0.10694503784179688,
      "step": 1907
    },
    {
      "epoch": 0.2579467680608365,
      "grad_norm": 0.6720229983329773,
      "learning_rate": 2.7062431445123127e-05,
      "loss": 0.10859489440917969,
      "step": 1908
    },
    {
      "epoch": 0.2580819602872835,
      "grad_norm": 0.163092240691185,
      "learning_rate": 2.705836179239437e-05,
      "loss": 0.03721427917480469,
      "step": 1909
    },
    {
      "epoch": 0.25821715251373045,
      "grad_norm": 0.2246158868074417,
      "learning_rate": 2.705428962905039e-05,
      "loss": 0.05305147171020508,
      "step": 1910
    },
    {
      "epoch": 0.25835234474017743,
      "grad_norm": 0.4187816381454468,
      "learning_rate": 2.705021495593902e-05,
      "loss": 0.06217479705810547,
      "step": 1911
    },
    {
      "epoch": 0.2584875369666244,
      "grad_norm": 0.23612040281295776,
      "learning_rate": 2.704613777390864e-05,
      "loss": 0.05178356170654297,
      "step": 1912
    },
    {
      "epoch": 0.2586227291930714,
      "grad_norm": 0.2466345876455307,
      "learning_rate": 2.7042058083808135e-05,
      "loss": 0.06723594665527344,
      "step": 1913
    },
    {
      "epoch": 0.25875792141951837,
      "grad_norm": 0.3143467605113983,
      "learning_rate": 2.7037975886486928e-05,
      "loss": 0.07501935958862305,
      "step": 1914
    },
    {
      "epoch": 0.25889311364596534,
      "grad_norm": 0.7473633885383606,
      "learning_rate": 2.7033891182794942e-05,
      "loss": 0.07051324844360352,
      "step": 1915
    },
    {
      "epoch": 0.2590283058724123,
      "grad_norm": 0.3036815822124481,
      "learning_rate": 2.7029803973582642e-05,
      "loss": 0.09438610076904297,
      "step": 1916
    },
    {
      "epoch": 0.2591634980988593,
      "grad_norm": 0.7471302151679993,
      "learning_rate": 2.7025714259701e-05,
      "loss": 0.11596965789794922,
      "step": 1917
    },
    {
      "epoch": 0.2592986903253063,
      "grad_norm": 0.2975316345691681,
      "learning_rate": 2.7021622042001524e-05,
      "loss": 0.060982704162597656,
      "step": 1918
    },
    {
      "epoch": 0.25943388255175326,
      "grad_norm": 0.2315310388803482,
      "learning_rate": 2.701752732133623e-05,
      "loss": 0.06364107131958008,
      "step": 1919
    },
    {
      "epoch": 0.25956907477820024,
      "grad_norm": 0.36252063512802124,
      "learning_rate": 2.7013430098557664e-05,
      "loss": 0.07845568656921387,
      "step": 1920
    },
    {
      "epoch": 0.2597042670046472,
      "grad_norm": 0.29441624879837036,
      "learning_rate": 2.7009330374518885e-05,
      "loss": 0.07388079166412354,
      "step": 1921
    },
    {
      "epoch": 0.2598394592310942,
      "grad_norm": 0.3996450901031494,
      "learning_rate": 2.7005228150073483e-05,
      "loss": 0.06785202026367188,
      "step": 1922
    },
    {
      "epoch": 0.2599746514575412,
      "grad_norm": 0.42293208837509155,
      "learning_rate": 2.7001123426075558e-05,
      "loss": 0.07725763320922852,
      "step": 1923
    },
    {
      "epoch": 0.26010984368398815,
      "grad_norm": 0.36804306507110596,
      "learning_rate": 2.699701620337974e-05,
      "loss": 0.08327007293701172,
      "step": 1924
    },
    {
      "epoch": 0.26024503591043513,
      "grad_norm": 0.20941795408725739,
      "learning_rate": 2.699290648284117e-05,
      "loss": 0.04230928421020508,
      "step": 1925
    },
    {
      "epoch": 0.2603802281368821,
      "grad_norm": 0.41908788681030273,
      "learning_rate": 2.6988794265315522e-05,
      "loss": 0.07233428955078125,
      "step": 1926
    },
    {
      "epoch": 0.2605154203633291,
      "grad_norm": 0.648202121257782,
      "learning_rate": 2.698467955165897e-05,
      "loss": 0.07257938385009766,
      "step": 1927
    },
    {
      "epoch": 0.26065061258977607,
      "grad_norm": 0.6043091416358948,
      "learning_rate": 2.6980562342728226e-05,
      "loss": 0.07048463821411133,
      "step": 1928
    },
    {
      "epoch": 0.26078580481622304,
      "grad_norm": 0.27515342831611633,
      "learning_rate": 2.6976442639380516e-05,
      "loss": 0.05468153953552246,
      "step": 1929
    },
    {
      "epoch": 0.26092099704267,
      "grad_norm": 0.39143988490104675,
      "learning_rate": 2.6972320442473583e-05,
      "loss": 0.11225557327270508,
      "step": 1930
    },
    {
      "epoch": 0.261056189269117,
      "grad_norm": 0.34398409724235535,
      "learning_rate": 2.6968195752865686e-05,
      "loss": 0.06856608390808105,
      "step": 1931
    },
    {
      "epoch": 0.261191381495564,
      "grad_norm": 0.5425191521644592,
      "learning_rate": 2.6964068571415613e-05,
      "loss": 0.05328011512756348,
      "step": 1932
    },
    {
      "epoch": 0.26132657372201096,
      "grad_norm": 0.5468816161155701,
      "learning_rate": 2.6959938898982667e-05,
      "loss": 0.05895841121673584,
      "step": 1933
    },
    {
      "epoch": 0.26146176594845794,
      "grad_norm": 0.7372881770133972,
      "learning_rate": 2.6955806736426657e-05,
      "loss": 0.055188655853271484,
      "step": 1934
    },
    {
      "epoch": 0.26159695817490497,
      "grad_norm": 0.6510387659072876,
      "learning_rate": 2.6951672084607937e-05,
      "loss": 0.07172107696533203,
      "step": 1935
    },
    {
      "epoch": 0.26173215040135195,
      "grad_norm": 0.4357663094997406,
      "learning_rate": 2.694753494438735e-05,
      "loss": 0.08475422859191895,
      "step": 1936
    },
    {
      "epoch": 0.2618673426277989,
      "grad_norm": 0.36257466673851013,
      "learning_rate": 2.6943395316626272e-05,
      "loss": 0.07384920120239258,
      "step": 1937
    },
    {
      "epoch": 0.2620025348542459,
      "grad_norm": 0.6099448800086975,
      "learning_rate": 2.69392532021866e-05,
      "loss": 0.06450843811035156,
      "step": 1938
    },
    {
      "epoch": 0.2621377270806929,
      "grad_norm": 0.5269374847412109,
      "learning_rate": 2.693510860193075e-05,
      "loss": 0.04517483711242676,
      "step": 1939
    },
    {
      "epoch": 0.26227291930713986,
      "grad_norm": 0.3653620779514313,
      "learning_rate": 2.6930961516721638e-05,
      "loss": 0.07281732559204102,
      "step": 1940
    },
    {
      "epoch": 0.26240811153358684,
      "grad_norm": 0.34565868973731995,
      "learning_rate": 2.6926811947422717e-05,
      "loss": 0.07845878601074219,
      "step": 1941
    },
    {
      "epoch": 0.2625433037600338,
      "grad_norm": 0.4030977487564087,
      "learning_rate": 2.6922659894897946e-05,
      "loss": 0.09230232238769531,
      "step": 1942
    },
    {
      "epoch": 0.2626784959864808,
      "grad_norm": 0.1322583258152008,
      "learning_rate": 2.6918505360011805e-05,
      "loss": 0.04181957244873047,
      "step": 1943
    },
    {
      "epoch": 0.2628136882129278,
      "grad_norm": 0.3881249725818634,
      "learning_rate": 2.6914348343629292e-05,
      "loss": 0.06250715255737305,
      "step": 1944
    },
    {
      "epoch": 0.26294888043937475,
      "grad_norm": 0.3524782955646515,
      "learning_rate": 2.6910188846615918e-05,
      "loss": 0.06326961517333984,
      "step": 1945
    },
    {
      "epoch": 0.26308407266582173,
      "grad_norm": 0.18269595503807068,
      "learning_rate": 2.6906026869837714e-05,
      "loss": 0.05874431133270264,
      "step": 1946
    },
    {
      "epoch": 0.2632192648922687,
      "grad_norm": 0.2870374917984009,
      "learning_rate": 2.6901862414161222e-05,
      "loss": 0.07396650314331055,
      "step": 1947
    },
    {
      "epoch": 0.2633544571187157,
      "grad_norm": 0.2652686834335327,
      "learning_rate": 2.689769548045351e-05,
      "loss": 0.05239880084991455,
      "step": 1948
    },
    {
      "epoch": 0.26348964934516267,
      "grad_norm": 0.4712505340576172,
      "learning_rate": 2.6893526069582154e-05,
      "loss": 0.067535400390625,
      "step": 1949
    },
    {
      "epoch": 0.26362484157160965,
      "grad_norm": 0.2714921236038208,
      "learning_rate": 2.6889354182415245e-05,
      "loss": 0.04259657859802246,
      "step": 1950
    },
    {
      "epoch": 0.2637600337980566,
      "grad_norm": 0.21372658014297485,
      "learning_rate": 2.688517981982139e-05,
      "loss": 0.06403589248657227,
      "step": 1951
    },
    {
      "epoch": 0.2638952260245036,
      "grad_norm": 0.45280420780181885,
      "learning_rate": 2.6881002982669723e-05,
      "loss": 0.06479430198669434,
      "step": 1952
    },
    {
      "epoch": 0.2640304182509506,
      "grad_norm": 0.3762657642364502,
      "learning_rate": 2.6876823671829874e-05,
      "loss": 0.06491422653198242,
      "step": 1953
    },
    {
      "epoch": 0.26416561047739756,
      "grad_norm": 0.33864468336105347,
      "learning_rate": 2.6872641888172e-05,
      "loss": 0.0740060806274414,
      "step": 1954
    },
    {
      "epoch": 0.26430080270384454,
      "grad_norm": 0.3459004759788513,
      "learning_rate": 2.6868457632566774e-05,
      "loss": 0.08530616760253906,
      "step": 1955
    },
    {
      "epoch": 0.2644359949302915,
      "grad_norm": 0.3440328538417816,
      "learning_rate": 2.6864270905885377e-05,
      "loss": 0.07538318634033203,
      "step": 1956
    },
    {
      "epoch": 0.2645711871567385,
      "grad_norm": 0.5170190334320068,
      "learning_rate": 2.6860081708999515e-05,
      "loss": 0.05870771408081055,
      "step": 1957
    },
    {
      "epoch": 0.2647063793831855,
      "grad_norm": 0.4916151762008667,
      "learning_rate": 2.685589004278139e-05,
      "loss": 0.1157674789428711,
      "step": 1958
    },
    {
      "epoch": 0.26484157160963245,
      "grad_norm": 0.31486639380455017,
      "learning_rate": 2.6851695908103737e-05,
      "loss": 0.07006168365478516,
      "step": 1959
    },
    {
      "epoch": 0.26497676383607943,
      "grad_norm": 0.7468628287315369,
      "learning_rate": 2.6847499305839796e-05,
      "loss": 0.07650637626647949,
      "step": 1960
    },
    {
      "epoch": 0.2651119560625264,
      "grad_norm": 0.2336750477552414,
      "learning_rate": 2.684330023686332e-05,
      "loss": 0.05476212501525879,
      "step": 1961
    },
    {
      "epoch": 0.2652471482889734,
      "grad_norm": 0.2948932945728302,
      "learning_rate": 2.6839098702048577e-05,
      "loss": 0.05624723434448242,
      "step": 1962
    },
    {
      "epoch": 0.26538234051542037,
      "grad_norm": 0.46952325105667114,
      "learning_rate": 2.683489470227035e-05,
      "loss": 0.09190154075622559,
      "step": 1963
    },
    {
      "epoch": 0.26551753274186735,
      "grad_norm": 0.3077831566333771,
      "learning_rate": 2.6830688238403936e-05,
      "loss": 0.05613231658935547,
      "step": 1964
    },
    {
      "epoch": 0.2656527249683143,
      "grad_norm": 0.20866359770298004,
      "learning_rate": 2.682647931132514e-05,
      "loss": 0.04815483093261719,
      "step": 1965
    },
    {
      "epoch": 0.2657879171947613,
      "grad_norm": 0.33155784010887146,
      "learning_rate": 2.682226792191029e-05,
      "loss": 0.05470585823059082,
      "step": 1966
    },
    {
      "epoch": 0.2659231094212083,
      "grad_norm": 0.35803017020225525,
      "learning_rate": 2.681805407103621e-05,
      "loss": 0.057668209075927734,
      "step": 1967
    },
    {
      "epoch": 0.26605830164765526,
      "grad_norm": 0.2577706575393677,
      "learning_rate": 2.6813837759580253e-05,
      "loss": 0.04786479473114014,
      "step": 1968
    },
    {
      "epoch": 0.26619349387410224,
      "grad_norm": 0.7822843194007874,
      "learning_rate": 2.6809618988420274e-05,
      "loss": 0.08090686798095703,
      "step": 1969
    },
    {
      "epoch": 0.2663286861005492,
      "grad_norm": 0.39716994762420654,
      "learning_rate": 2.6805397758434647e-05,
      "loss": 0.04407310485839844,
      "step": 1970
    },
    {
      "epoch": 0.2664638783269962,
      "grad_norm": 0.3969775438308716,
      "learning_rate": 2.6801174070502248e-05,
      "loss": 0.06796550750732422,
      "step": 1971
    },
    {
      "epoch": 0.2665990705534432,
      "grad_norm": 0.5066125392913818,
      "learning_rate": 2.679694792550248e-05,
      "loss": 0.1039915680885315,
      "step": 1972
    },
    {
      "epoch": 0.26673426277989015,
      "grad_norm": 0.3619740605354309,
      "learning_rate": 2.6792719324315248e-05,
      "loss": 0.07608604431152344,
      "step": 1973
    },
    {
      "epoch": 0.26686945500633713,
      "grad_norm": 0.46979114413261414,
      "learning_rate": 2.678848826782096e-05,
      "loss": 0.07393312454223633,
      "step": 1974
    },
    {
      "epoch": 0.2670046472327841,
      "grad_norm": 0.2395251840353012,
      "learning_rate": 2.678425475690055e-05,
      "loss": 0.06704425811767578,
      "step": 1975
    },
    {
      "epoch": 0.2671398394592311,
      "grad_norm": 0.3646530508995056,
      "learning_rate": 2.6780018792435464e-05,
      "loss": 0.05572080612182617,
      "step": 1976
    },
    {
      "epoch": 0.26727503168567807,
      "grad_norm": 0.5919400453567505,
      "learning_rate": 2.6775780375307645e-05,
      "loss": 0.08342146873474121,
      "step": 1977
    },
    {
      "epoch": 0.26741022391212504,
      "grad_norm": 0.5902851819992065,
      "learning_rate": 2.6771539506399555e-05,
      "loss": 0.09327661991119385,
      "step": 1978
    },
    {
      "epoch": 0.267545416138572,
      "grad_norm": 0.19895702600479126,
      "learning_rate": 2.6767296186594165e-05,
      "loss": 0.03602182865142822,
      "step": 1979
    },
    {
      "epoch": 0.267680608365019,
      "grad_norm": 0.27372804284095764,
      "learning_rate": 2.676305041677496e-05,
      "loss": 0.04723072052001953,
      "step": 1980
    },
    {
      "epoch": 0.267815800591466,
      "grad_norm": 0.3940640985965729,
      "learning_rate": 2.675880219782593e-05,
      "loss": 0.07594537734985352,
      "step": 1981
    },
    {
      "epoch": 0.26795099281791296,
      "grad_norm": 0.512906551361084,
      "learning_rate": 2.6754551530631575e-05,
      "loss": 0.046251535415649414,
      "step": 1982
    },
    {
      "epoch": 0.26808618504435994,
      "grad_norm": 0.2687968313694,
      "learning_rate": 2.6750298416076907e-05,
      "loss": 0.05890655517578125,
      "step": 1983
    },
    {
      "epoch": 0.2682213772708069,
      "grad_norm": 0.3124563992023468,
      "learning_rate": 2.674604285504745e-05,
      "loss": 0.07229232788085938,
      "step": 1984
    },
    {
      "epoch": 0.2683565694972539,
      "grad_norm": 0.29689836502075195,
      "learning_rate": 2.6741784848429235e-05,
      "loss": 0.07830238342285156,
      "step": 1985
    },
    {
      "epoch": 0.2684917617237009,
      "grad_norm": 0.41927099227905273,
      "learning_rate": 2.67375243971088e-05,
      "loss": 0.08188128471374512,
      "step": 1986
    },
    {
      "epoch": 0.26862695395014785,
      "grad_norm": 0.36072978377342224,
      "learning_rate": 2.6733261501973192e-05,
      "loss": 0.06334662437438965,
      "step": 1987
    },
    {
      "epoch": 0.26876214617659483,
      "grad_norm": 0.26369619369506836,
      "learning_rate": 2.672899616390997e-05,
      "loss": 0.045501708984375,
      "step": 1988
    },
    {
      "epoch": 0.2688973384030418,
      "grad_norm": 0.3635765314102173,
      "learning_rate": 2.67247283838072e-05,
      "loss": 0.07266116142272949,
      "step": 1989
    },
    {
      "epoch": 0.2690325306294888,
      "grad_norm": 0.43948882818222046,
      "learning_rate": 2.6720458162553457e-05,
      "loss": 0.07158470153808594,
      "step": 1990
    },
    {
      "epoch": 0.26916772285593576,
      "grad_norm": 0.31330564618110657,
      "learning_rate": 2.6716185501037822e-05,
      "loss": 0.05783557891845703,
      "step": 1991
    },
    {
      "epoch": 0.26930291508238274,
      "grad_norm": 0.35045677423477173,
      "learning_rate": 2.671191040014989e-05,
      "loss": 0.05690479278564453,
      "step": 1992
    },
    {
      "epoch": 0.2694381073088297,
      "grad_norm": 0.6701129674911499,
      "learning_rate": 2.6707632860779756e-05,
      "loss": 0.06578874588012695,
      "step": 1993
    },
    {
      "epoch": 0.2695732995352767,
      "grad_norm": 0.2643255889415741,
      "learning_rate": 2.6703352883818024e-05,
      "loss": 0.07078218460083008,
      "step": 1994
    },
    {
      "epoch": 0.2697084917617237,
      "grad_norm": 0.2178814560174942,
      "learning_rate": 2.6699070470155816e-05,
      "loss": 0.059894561767578125,
      "step": 1995
    },
    {
      "epoch": 0.26984368398817066,
      "grad_norm": 0.33343926072120667,
      "learning_rate": 2.669478562068475e-05,
      "loss": 0.0846400260925293,
      "step": 1996
    },
    {
      "epoch": 0.26997887621461764,
      "grad_norm": 0.2248917669057846,
      "learning_rate": 2.6690498336296955e-05,
      "loss": 0.05446124076843262,
      "step": 1997
    },
    {
      "epoch": 0.2701140684410646,
      "grad_norm": 0.5515278577804565,
      "learning_rate": 2.6686208617885057e-05,
      "loss": 0.09082603454589844,
      "step": 1998
    },
    {
      "epoch": 0.2702492606675116,
      "grad_norm": 0.20593951642513275,
      "learning_rate": 2.668191646634221e-05,
      "loss": 0.02307724952697754,
      "step": 1999
    },
    {
      "epoch": 0.27038445289395857,
      "grad_norm": 0.270003080368042,
      "learning_rate": 2.667762188256206e-05,
      "loss": 0.065643310546875,
      "step": 2000
    },
    {
      "epoch": 0.27051964512040555,
      "grad_norm": 0.21622775495052338,
      "learning_rate": 2.6673324867438764e-05,
      "loss": 0.05423140525817871,
      "step": 2001
    },
    {
      "epoch": 0.2706548373468526,
      "grad_norm": 0.351639986038208,
      "learning_rate": 2.666902542186698e-05,
      "loss": 0.054004669189453125,
      "step": 2002
    },
    {
      "epoch": 0.27079002957329956,
      "grad_norm": 0.47563913464546204,
      "learning_rate": 2.666472354674187e-05,
      "loss": 0.12000083923339844,
      "step": 2003
    },
    {
      "epoch": 0.27092522179974654,
      "grad_norm": 0.4176982045173645,
      "learning_rate": 2.666041924295912e-05,
      "loss": 0.0824594497680664,
      "step": 2004
    },
    {
      "epoch": 0.2710604140261935,
      "grad_norm": 0.3982656002044678,
      "learning_rate": 2.6656112511414902e-05,
      "loss": 0.08656144142150879,
      "step": 2005
    },
    {
      "epoch": 0.2711956062526405,
      "grad_norm": 0.5034632086753845,
      "learning_rate": 2.6651803353005896e-05,
      "loss": 0.05389070510864258,
      "step": 2006
    },
    {
      "epoch": 0.2713307984790875,
      "grad_norm": 0.5947905778884888,
      "learning_rate": 2.66474917686293e-05,
      "loss": 0.09485912322998047,
      "step": 2007
    },
    {
      "epoch": 0.27146599070553445,
      "grad_norm": 0.6628316044807434,
      "learning_rate": 2.664317775918281e-05,
      "loss": 0.06086158752441406,
      "step": 2008
    },
    {
      "epoch": 0.27160118293198143,
      "grad_norm": 0.31264886260032654,
      "learning_rate": 2.6638861325564615e-05,
      "loss": 0.058119773864746094,
      "step": 2009
    },
    {
      "epoch": 0.2717363751584284,
      "grad_norm": 0.3556406795978546,
      "learning_rate": 2.6634542468673432e-05,
      "loss": 0.0729360580444336,
      "step": 2010
    },
    {
      "epoch": 0.2718715673848754,
      "grad_norm": 0.4376427233219147,
      "learning_rate": 2.663022118940846e-05,
      "loss": 0.07712364196777344,
      "step": 2011
    },
    {
      "epoch": 0.27200675961132237,
      "grad_norm": 0.49218839406967163,
      "learning_rate": 2.662589748866942e-05,
      "loss": 0.06293153762817383,
      "step": 2012
    },
    {
      "epoch": 0.27214195183776935,
      "grad_norm": 0.5261367559432983,
      "learning_rate": 2.6621571367356522e-05,
      "loss": 0.057858943939208984,
      "step": 2013
    },
    {
      "epoch": 0.2722771440642163,
      "grad_norm": 0.455219030380249,
      "learning_rate": 2.6617242826370495e-05,
      "loss": 0.08683300018310547,
      "step": 2014
    },
    {
      "epoch": 0.2724123362906633,
      "grad_norm": 0.24125950038433075,
      "learning_rate": 2.661291186661256e-05,
      "loss": 0.05467796325683594,
      "step": 2015
    },
    {
      "epoch": 0.2725475285171103,
      "grad_norm": 1.090306043624878,
      "learning_rate": 2.6608578488984444e-05,
      "loss": 0.09870195388793945,
      "step": 2016
    },
    {
      "epoch": 0.27268272074355726,
      "grad_norm": 0.6628405451774597,
      "learning_rate": 2.6604242694388388e-05,
      "loss": 0.1032710075378418,
      "step": 2017
    },
    {
      "epoch": 0.27281791297000424,
      "grad_norm": 0.5847178101539612,
      "learning_rate": 2.6599904483727116e-05,
      "loss": 0.05669689178466797,
      "step": 2018
    },
    {
      "epoch": 0.2729531051964512,
      "grad_norm": 0.3156111240386963,
      "learning_rate": 2.6595563857903872e-05,
      "loss": 0.0486445426940918,
      "step": 2019
    },
    {
      "epoch": 0.2730882974228982,
      "grad_norm": 0.45537737011909485,
      "learning_rate": 2.6591220817822405e-05,
      "loss": 0.08964300155639648,
      "step": 2020
    },
    {
      "epoch": 0.2732234896493452,
      "grad_norm": 0.6203433871269226,
      "learning_rate": 2.658687536438694e-05,
      "loss": 0.07988929748535156,
      "step": 2021
    },
    {
      "epoch": 0.27335868187579215,
      "grad_norm": 0.25784873962402344,
      "learning_rate": 2.6582527498502243e-05,
      "loss": 0.04907023906707764,
      "step": 2022
    },
    {
      "epoch": 0.27349387410223913,
      "grad_norm": 0.6224722266197205,
      "learning_rate": 2.6578177221073556e-05,
      "loss": 0.07737493515014648,
      "step": 2023
    },
    {
      "epoch": 0.2736290663286861,
      "grad_norm": 0.2951950132846832,
      "learning_rate": 2.6573824533006628e-05,
      "loss": 0.08441162109375,
      "step": 2024
    },
    {
      "epoch": 0.2737642585551331,
      "grad_norm": 0.2897282838821411,
      "learning_rate": 2.6569469435207712e-05,
      "loss": 0.06387567520141602,
      "step": 2025
    },
    {
      "epoch": 0.27389945078158007,
      "grad_norm": 0.47775185108184814,
      "learning_rate": 2.656511192858356e-05,
      "loss": 0.04537701606750488,
      "step": 2026
    },
    {
      "epoch": 0.27403464300802705,
      "grad_norm": 0.587962806224823,
      "learning_rate": 2.6560752014041438e-05,
      "loss": 0.062053680419921875,
      "step": 2027
    },
    {
      "epoch": 0.274169835234474,
      "grad_norm": 0.3327188491821289,
      "learning_rate": 2.6556389692489098e-05,
      "loss": 0.07045269012451172,
      "step": 2028
    },
    {
      "epoch": 0.274305027460921,
      "grad_norm": 0.25019848346710205,
      "learning_rate": 2.6552024964834795e-05,
      "loss": 0.08115196228027344,
      "step": 2029
    },
    {
      "epoch": 0.274440219687368,
      "grad_norm": 0.5601955056190491,
      "learning_rate": 2.6547657831987286e-05,
      "loss": 0.09872913360595703,
      "step": 2030
    },
    {
      "epoch": 0.27457541191381496,
      "grad_norm": 0.5355115532875061,
      "learning_rate": 2.6543288294855843e-05,
      "loss": 0.07593822479248047,
      "step": 2031
    },
    {
      "epoch": 0.27471060414026194,
      "grad_norm": 0.37519311904907227,
      "learning_rate": 2.653891635435022e-05,
      "loss": 0.07274961471557617,
      "step": 2032
    },
    {
      "epoch": 0.2748457963667089,
      "grad_norm": 0.2885375916957855,
      "learning_rate": 2.653454201138068e-05,
      "loss": 0.06522369384765625,
      "step": 2033
    },
    {
      "epoch": 0.2749809885931559,
      "grad_norm": 0.504676342010498,
      "learning_rate": 2.653016526685798e-05,
      "loss": 0.08700704574584961,
      "step": 2034
    },
    {
      "epoch": 0.2751161808196029,
      "grad_norm": 0.4324076175689697,
      "learning_rate": 2.6525786121693387e-05,
      "loss": 0.0872201919555664,
      "step": 2035
    },
    {
      "epoch": 0.27525137304604985,
      "grad_norm": 0.6423711776733398,
      "learning_rate": 2.652140457679866e-05,
      "loss": 0.08132648468017578,
      "step": 2036
    },
    {
      "epoch": 0.27538656527249683,
      "grad_norm": 0.24215656518936157,
      "learning_rate": 2.6517020633086064e-05,
      "loss": 0.059365272521972656,
      "step": 2037
    },
    {
      "epoch": 0.2755217574989438,
      "grad_norm": 0.30316051840782166,
      "learning_rate": 2.6512634291468354e-05,
      "loss": 0.07519912719726562,
      "step": 2038
    },
    {
      "epoch": 0.2756569497253908,
      "grad_norm": 0.16837528347969055,
      "learning_rate": 2.6508245552858792e-05,
      "loss": 0.04826211929321289,
      "step": 2039
    },
    {
      "epoch": 0.27579214195183777,
      "grad_norm": 0.25674694776535034,
      "learning_rate": 2.6503854418171133e-05,
      "loss": 0.05576038360595703,
      "step": 2040
    },
    {
      "epoch": 0.27592733417828474,
      "grad_norm": 0.30849340558052063,
      "learning_rate": 2.6499460888319644e-05,
      "loss": 0.03331947326660156,
      "step": 2041
    },
    {
      "epoch": 0.2760625264047317,
      "grad_norm": 0.7548207640647888,
      "learning_rate": 2.6495064964219073e-05,
      "loss": 0.08566570281982422,
      "step": 2042
    },
    {
      "epoch": 0.2761977186311787,
      "grad_norm": 0.4818129539489746,
      "learning_rate": 2.649066664678467e-05,
      "loss": 0.06218671798706055,
      "step": 2043
    },
    {
      "epoch": 0.2763329108576257,
      "grad_norm": 0.2774481177330017,
      "learning_rate": 2.6486265936932205e-05,
      "loss": 0.06703591346740723,
      "step": 2044
    },
    {
      "epoch": 0.27646810308407266,
      "grad_norm": 0.21266894042491913,
      "learning_rate": 2.6481862835577915e-05,
      "loss": 0.05988788604736328,
      "step": 2045
    },
    {
      "epoch": 0.27660329531051964,
      "grad_norm": 0.47143039107322693,
      "learning_rate": 2.6477457343638557e-05,
      "loss": 0.0818624496459961,
      "step": 2046
    },
    {
      "epoch": 0.2767384875369666,
      "grad_norm": 1.2861460447311401,
      "learning_rate": 2.647304946203137e-05,
      "loss": 0.10152912139892578,
      "step": 2047
    },
    {
      "epoch": 0.2768736797634136,
      "grad_norm": 0.23386774957180023,
      "learning_rate": 2.6468639191674106e-05,
      "loss": 0.04874753952026367,
      "step": 2048
    },
    {
      "epoch": 0.2770088719898606,
      "grad_norm": 0.3961261808872223,
      "learning_rate": 2.6464226533485007e-05,
      "loss": 0.08870506286621094,
      "step": 2049
    },
    {
      "epoch": 0.27714406421630755,
      "grad_norm": 0.2394411414861679,
      "learning_rate": 2.6459811488382806e-05,
      "loss": 0.059814453125,
      "step": 2050
    },
    {
      "epoch": 0.27727925644275453,
      "grad_norm": 0.35150033235549927,
      "learning_rate": 2.645539405728674e-05,
      "loss": 0.07271122932434082,
      "step": 2051
    },
    {
      "epoch": 0.2774144486692015,
      "grad_norm": 0.21934662759304047,
      "learning_rate": 2.6450974241116545e-05,
      "loss": 0.04949212074279785,
      "step": 2052
    },
    {
      "epoch": 0.2775496408956485,
      "grad_norm": 0.33216625452041626,
      "learning_rate": 2.644655204079245e-05,
      "loss": 0.056897759437561035,
      "step": 2053
    },
    {
      "epoch": 0.27768483312209546,
      "grad_norm": 0.2515887916088104,
      "learning_rate": 2.6442127457235177e-05,
      "loss": 0.051683902740478516,
      "step": 2054
    },
    {
      "epoch": 0.27782002534854244,
      "grad_norm": 0.4746667742729187,
      "learning_rate": 2.6437700491365957e-05,
      "loss": 0.06813335418701172,
      "step": 2055
    },
    {
      "epoch": 0.2779552175749894,
      "grad_norm": 0.4989183843135834,
      "learning_rate": 2.6433271144106495e-05,
      "loss": 0.053537845611572266,
      "step": 2056
    },
    {
      "epoch": 0.2780904098014364,
      "grad_norm": 1.0948442220687866,
      "learning_rate": 2.6428839416379015e-05,
      "loss": 0.07752418518066406,
      "step": 2057
    },
    {
      "epoch": 0.2782256020278834,
      "grad_norm": 0.5345621705055237,
      "learning_rate": 2.642440530910622e-05,
      "loss": 0.06610107421875,
      "step": 2058
    },
    {
      "epoch": 0.27836079425433036,
      "grad_norm": 0.6258991956710815,
      "learning_rate": 2.6419968823211318e-05,
      "loss": 0.09168887138366699,
      "step": 2059
    },
    {
      "epoch": 0.27849598648077734,
      "grad_norm": 0.42434990406036377,
      "learning_rate": 2.641552995961801e-05,
      "loss": 0.057613372802734375,
      "step": 2060
    },
    {
      "epoch": 0.2786311787072243,
      "grad_norm": 0.3588142693042755,
      "learning_rate": 2.6411088719250484e-05,
      "loss": 0.06316852569580078,
      "step": 2061
    },
    {
      "epoch": 0.2787663709336713,
      "grad_norm": 0.2978154420852661,
      "learning_rate": 2.6406645103033442e-05,
      "loss": 0.0728154182434082,
      "step": 2062
    },
    {
      "epoch": 0.27890156316011827,
      "grad_norm": 0.35572904348373413,
      "learning_rate": 2.640219911189206e-05,
      "loss": 0.046875953674316406,
      "step": 2063
    },
    {
      "epoch": 0.27903675538656525,
      "grad_norm": 0.22884875535964966,
      "learning_rate": 2.6397750746752015e-05,
      "loss": 0.0498356819152832,
      "step": 2064
    },
    {
      "epoch": 0.27917194761301223,
      "grad_norm": 0.7761141061782837,
      "learning_rate": 2.6393300008539488e-05,
      "loss": 0.08547496795654297,
      "step": 2065
    },
    {
      "epoch": 0.2793071398394592,
      "grad_norm": 0.38458916544914246,
      "learning_rate": 2.6388846898181143e-05,
      "loss": 0.10342693328857422,
      "step": 2066
    },
    {
      "epoch": 0.2794423320659062,
      "grad_norm": 0.43140697479248047,
      "learning_rate": 2.6384391416604142e-05,
      "loss": 0.09862852096557617,
      "step": 2067
    },
    {
      "epoch": 0.27957752429235316,
      "grad_norm": 0.3888213634490967,
      "learning_rate": 2.6379933564736136e-05,
      "loss": 0.059099674224853516,
      "step": 2068
    },
    {
      "epoch": 0.27971271651880014,
      "grad_norm": 0.24982337653636932,
      "learning_rate": 2.637547334350528e-05,
      "loss": 0.06424570083618164,
      "step": 2069
    },
    {
      "epoch": 0.2798479087452472,
      "grad_norm": 0.27125102281570435,
      "learning_rate": 2.637101075384021e-05,
      "loss": 0.06197500228881836,
      "step": 2070
    },
    {
      "epoch": 0.27998310097169415,
      "grad_norm": 0.31710943579673767,
      "learning_rate": 2.636654579667006e-05,
      "loss": 0.06935620307922363,
      "step": 2071
    },
    {
      "epoch": 0.28011829319814113,
      "grad_norm": 0.4451105296611786,
      "learning_rate": 2.6362078472924467e-05,
      "loss": 0.1120138168334961,
      "step": 2072
    },
    {
      "epoch": 0.2802534854245881,
      "grad_norm": 0.6225810647010803,
      "learning_rate": 2.6357608783533545e-05,
      "loss": 0.10464096069335938,
      "step": 2073
    },
    {
      "epoch": 0.2803886776510351,
      "grad_norm": 0.4876011312007904,
      "learning_rate": 2.6353136729427907e-05,
      "loss": 0.11539840698242188,
      "step": 2074
    },
    {
      "epoch": 0.28052386987748207,
      "grad_norm": 0.8319481015205383,
      "learning_rate": 2.6348662311538657e-05,
      "loss": 0.10995841026306152,
      "step": 2075
    },
    {
      "epoch": 0.28065906210392905,
      "grad_norm": 0.3177359104156494,
      "learning_rate": 2.6344185530797398e-05,
      "loss": 0.08544445037841797,
      "step": 2076
    },
    {
      "epoch": 0.280794254330376,
      "grad_norm": 0.4108535051345825,
      "learning_rate": 2.633970638813622e-05,
      "loss": 0.06264829635620117,
      "step": 2077
    },
    {
      "epoch": 0.280929446556823,
      "grad_norm": 0.8796353340148926,
      "learning_rate": 2.6335224884487698e-05,
      "loss": 0.10674858093261719,
      "step": 2078
    },
    {
      "epoch": 0.28106463878327,
      "grad_norm": 0.6604017615318298,
      "learning_rate": 2.6330741020784905e-05,
      "loss": 0.08148258924484253,
      "step": 2079
    },
    {
      "epoch": 0.28119983100971696,
      "grad_norm": 0.3304895758628845,
      "learning_rate": 2.6326254797961415e-05,
      "loss": 0.07968354225158691,
      "step": 2080
    },
    {
      "epoch": 0.28133502323616394,
      "grad_norm": 0.28452926874160767,
      "learning_rate": 2.6321766216951273e-05,
      "loss": 0.04527163505554199,
      "step": 2081
    },
    {
      "epoch": 0.2814702154626109,
      "grad_norm": 0.18396013975143433,
      "learning_rate": 2.631727527868903e-05,
      "loss": 0.040405988693237305,
      "step": 2082
    },
    {
      "epoch": 0.2816054076890579,
      "grad_norm": 0.4389074146747589,
      "learning_rate": 2.6312781984109727e-05,
      "loss": 0.0720205307006836,
      "step": 2083
    },
    {
      "epoch": 0.2817405999155049,
      "grad_norm": 0.4166303277015686,
      "learning_rate": 2.6308286334148882e-05,
      "loss": 0.06545305252075195,
      "step": 2084
    },
    {
      "epoch": 0.28187579214195185,
      "grad_norm": 0.5256975889205933,
      "learning_rate": 2.630378832974252e-05,
      "loss": 0.07265925407409668,
      "step": 2085
    },
    {
      "epoch": 0.28201098436839883,
      "grad_norm": 0.4502474367618561,
      "learning_rate": 2.6299287971827154e-05,
      "loss": 0.0744485855102539,
      "step": 2086
    },
    {
      "epoch": 0.2821461765948458,
      "grad_norm": 0.4261792004108429,
      "learning_rate": 2.629478526133977e-05,
      "loss": 0.061431884765625,
      "step": 2087
    },
    {
      "epoch": 0.2822813688212928,
      "grad_norm": 0.6339582204818726,
      "learning_rate": 2.6290280199217867e-05,
      "loss": 0.09616327285766602,
      "step": 2088
    },
    {
      "epoch": 0.28241656104773977,
      "grad_norm": 0.6434690952301025,
      "learning_rate": 2.6285772786399424e-05,
      "loss": 0.06148338317871094,
      "step": 2089
    },
    {
      "epoch": 0.28255175327418675,
      "grad_norm": 0.30254992842674255,
      "learning_rate": 2.6281263023822894e-05,
      "loss": 0.06952857971191406,
      "step": 2090
    },
    {
      "epoch": 0.2826869455006337,
      "grad_norm": 0.37189915776252747,
      "learning_rate": 2.627675091242725e-05,
      "loss": 0.06276750564575195,
      "step": 2091
    },
    {
      "epoch": 0.2828221377270807,
      "grad_norm": 0.29637306928634644,
      "learning_rate": 2.627223645315193e-05,
      "loss": 0.06892728805541992,
      "step": 2092
    },
    {
      "epoch": 0.2829573299535277,
      "grad_norm": 0.3086879253387451,
      "learning_rate": 2.6267719646936868e-05,
      "loss": 0.05177867412567139,
      "step": 2093
    },
    {
      "epoch": 0.28309252217997466,
      "grad_norm": 0.23939448595046997,
      "learning_rate": 2.626320049472249e-05,
      "loss": 0.06682491302490234,
      "step": 2094
    },
    {
      "epoch": 0.28322771440642164,
      "grad_norm": 0.5124935507774353,
      "learning_rate": 2.6258678997449705e-05,
      "loss": 0.08545827865600586,
      "step": 2095
    },
    {
      "epoch": 0.2833629066328686,
      "grad_norm": 0.5773860216140747,
      "learning_rate": 2.6254155156059912e-05,
      "loss": 0.09173870086669922,
      "step": 2096
    },
    {
      "epoch": 0.2834980988593156,
      "grad_norm": 0.46175697445869446,
      "learning_rate": 2.6249628971495006e-05,
      "loss": 0.0669870376586914,
      "step": 2097
    },
    {
      "epoch": 0.2836332910857626,
      "grad_norm": 0.2535857856273651,
      "learning_rate": 2.6245100444697353e-05,
      "loss": 0.07016611099243164,
      "step": 2098
    },
    {
      "epoch": 0.28376848331220955,
      "grad_norm": 0.20684558153152466,
      "learning_rate": 2.6240569576609824e-05,
      "loss": 0.04796886444091797,
      "step": 2099
    },
    {
      "epoch": 0.28390367553865653,
      "grad_norm": 0.20866382122039795,
      "learning_rate": 2.623603636817577e-05,
      "loss": 0.0501713752746582,
      "step": 2100
    },
    {
      "epoch": 0.2840388677651035,
      "grad_norm": 0.2257915437221527,
      "learning_rate": 2.6231500820339024e-05,
      "loss": 0.05210685729980469,
      "step": 2101
    },
    {
      "epoch": 0.2841740599915505,
      "grad_norm": 0.30343684554100037,
      "learning_rate": 2.6226962934043913e-05,
      "loss": 0.06421279907226562,
      "step": 2102
    },
    {
      "epoch": 0.28430925221799747,
      "grad_norm": 0.3582315742969513,
      "learning_rate": 2.622242271023525e-05,
      "loss": 0.054036617279052734,
      "step": 2103
    },
    {
      "epoch": 0.28444444444444444,
      "grad_norm": 0.523270845413208,
      "learning_rate": 2.6217880149858333e-05,
      "loss": 0.07139182090759277,
      "step": 2104
    },
    {
      "epoch": 0.2845796366708914,
      "grad_norm": 0.23183846473693848,
      "learning_rate": 2.621333525385895e-05,
      "loss": 0.06073260307312012,
      "step": 2105
    },
    {
      "epoch": 0.2847148288973384,
      "grad_norm": 0.30376967787742615,
      "learning_rate": 2.6208788023183366e-05,
      "loss": 0.08838939666748047,
      "step": 2106
    },
    {
      "epoch": 0.2848500211237854,
      "grad_norm": 0.5577009916305542,
      "learning_rate": 2.6204238458778346e-05,
      "loss": 0.09938955307006836,
      "step": 2107
    },
    {
      "epoch": 0.28498521335023236,
      "grad_norm": 0.3254562020301819,
      "learning_rate": 2.619968656159113e-05,
      "loss": 0.04650545120239258,
      "step": 2108
    },
    {
      "epoch": 0.28512040557667934,
      "grad_norm": 0.3568468391895294,
      "learning_rate": 2.6195132332569445e-05,
      "loss": 0.09248828887939453,
      "step": 2109
    },
    {
      "epoch": 0.2852555978031263,
      "grad_norm": 0.2698529064655304,
      "learning_rate": 2.619057577266151e-05,
      "loss": 0.04743695259094238,
      "step": 2110
    },
    {
      "epoch": 0.2853907900295733,
      "grad_norm": 0.3744509220123291,
      "learning_rate": 2.6186016882816027e-05,
      "loss": 0.08064508438110352,
      "step": 2111
    },
    {
      "epoch": 0.28552598225602027,
      "grad_norm": 0.31511595845222473,
      "learning_rate": 2.6181455663982175e-05,
      "loss": 0.06336402893066406,
      "step": 2112
    },
    {
      "epoch": 0.28566117448246725,
      "grad_norm": 0.7368386387825012,
      "learning_rate": 2.6176892117109628e-05,
      "loss": 0.08360576629638672,
      "step": 2113
    },
    {
      "epoch": 0.28579636670891423,
      "grad_norm": 0.2987616956233978,
      "learning_rate": 2.617232624314854e-05,
      "loss": 0.07652616500854492,
      "step": 2114
    },
    {
      "epoch": 0.2859315589353612,
      "grad_norm": 0.2696145176887512,
      "learning_rate": 2.616775804304955e-05,
      "loss": 0.05968737602233887,
      "step": 2115
    },
    {
      "epoch": 0.2860667511618082,
      "grad_norm": 0.23673371970653534,
      "learning_rate": 2.616318751776378e-05,
      "loss": 0.03470277786254883,
      "step": 2116
    },
    {
      "epoch": 0.28620194338825516,
      "grad_norm": 0.3378492295742035,
      "learning_rate": 2.615861466824284e-05,
      "loss": 0.04506063461303711,
      "step": 2117
    },
    {
      "epoch": 0.28633713561470214,
      "grad_norm": 0.5445701479911804,
      "learning_rate": 2.6154039495438825e-05,
      "loss": 0.058304548263549805,
      "step": 2118
    },
    {
      "epoch": 0.2864723278411491,
      "grad_norm": 0.31142300367355347,
      "learning_rate": 2.6149462000304302e-05,
      "loss": 0.06822836399078369,
      "step": 2119
    },
    {
      "epoch": 0.2866075200675961,
      "grad_norm": 0.14689652621746063,
      "learning_rate": 2.6144882183792335e-05,
      "loss": 0.04671072959899902,
      "step": 2120
    },
    {
      "epoch": 0.2867427122940431,
      "grad_norm": 0.29540175199508667,
      "learning_rate": 2.6140300046856468e-05,
      "loss": 0.06613868474960327,
      "step": 2121
    },
    {
      "epoch": 0.28687790452049006,
      "grad_norm": 0.43784645199775696,
      "learning_rate": 2.6135715590450722e-05,
      "loss": 0.07046675682067871,
      "step": 2122
    },
    {
      "epoch": 0.28701309674693704,
      "grad_norm": 0.34985587000846863,
      "learning_rate": 2.6131128815529608e-05,
      "loss": 0.07567620277404785,
      "step": 2123
    },
    {
      "epoch": 0.287148288973384,
      "grad_norm": 0.32104727625846863,
      "learning_rate": 2.6126539723048115e-05,
      "loss": 0.0802912712097168,
      "step": 2124
    },
    {
      "epoch": 0.287283481199831,
      "grad_norm": 0.33987459540367126,
      "learning_rate": 2.612194831396172e-05,
      "loss": 0.05086708068847656,
      "step": 2125
    },
    {
      "epoch": 0.28741867342627797,
      "grad_norm": 0.2595984935760498,
      "learning_rate": 2.611735458922637e-05,
      "loss": 0.05791115760803223,
      "step": 2126
    },
    {
      "epoch": 0.28755386565272495,
      "grad_norm": 0.7451924085617065,
      "learning_rate": 2.6112758549798515e-05,
      "loss": 0.0922541618347168,
      "step": 2127
    },
    {
      "epoch": 0.2876890578791719,
      "grad_norm": 0.3935089409351349,
      "learning_rate": 2.610816019663507e-05,
      "loss": 0.09931564331054688,
      "step": 2128
    },
    {
      "epoch": 0.2878242501056189,
      "grad_norm": 0.4284669756889343,
      "learning_rate": 2.6103559530693436e-05,
      "loss": 0.0782618522644043,
      "step": 2129
    },
    {
      "epoch": 0.2879594423320659,
      "grad_norm": 0.2335512638092041,
      "learning_rate": 2.6098956552931495e-05,
      "loss": 0.0554194450378418,
      "step": 2130
    },
    {
      "epoch": 0.28809463455851286,
      "grad_norm": 0.4409209191799164,
      "learning_rate": 2.6094351264307613e-05,
      "loss": 0.09137964248657227,
      "step": 2131
    },
    {
      "epoch": 0.28822982678495984,
      "grad_norm": 0.33390122652053833,
      "learning_rate": 2.6089743665780635e-05,
      "loss": 0.07697200775146484,
      "step": 2132
    },
    {
      "epoch": 0.2883650190114068,
      "grad_norm": 0.31019923090934753,
      "learning_rate": 2.6085133758309887e-05,
      "loss": 0.08348321914672852,
      "step": 2133
    },
    {
      "epoch": 0.2885002112378538,
      "grad_norm": 0.24304573237895966,
      "learning_rate": 2.6080521542855182e-05,
      "loss": 0.06161618232727051,
      "step": 2134
    },
    {
      "epoch": 0.2886354034643008,
      "grad_norm": 0.505469560623169,
      "learning_rate": 2.60759070203768e-05,
      "loss": 0.06316828727722168,
      "step": 2135
    },
    {
      "epoch": 0.28877059569074776,
      "grad_norm": 0.14189660549163818,
      "learning_rate": 2.607129019183551e-05,
      "loss": 0.03790163993835449,
      "step": 2136
    },
    {
      "epoch": 0.28890578791719473,
      "grad_norm": 0.4363943636417389,
      "learning_rate": 2.6066671058192566e-05,
      "loss": 0.0781412124633789,
      "step": 2137
    },
    {
      "epoch": 0.28904098014364177,
      "grad_norm": 0.4986313283443451,
      "learning_rate": 2.606204962040969e-05,
      "loss": 0.11483097076416016,
      "step": 2138
    },
    {
      "epoch": 0.28917617237008875,
      "grad_norm": 0.37120503187179565,
      "learning_rate": 2.6057425879449095e-05,
      "loss": 0.06394124031066895,
      "step": 2139
    },
    {
      "epoch": 0.2893113645965357,
      "grad_norm": 0.5296469330787659,
      "learning_rate": 2.605279983627347e-05,
      "loss": 0.05107307434082031,
      "step": 2140
    },
    {
      "epoch": 0.2894465568229827,
      "grad_norm": 0.85414057970047,
      "learning_rate": 2.6048171491845974e-05,
      "loss": 0.09617280960083008,
      "step": 2141
    },
    {
      "epoch": 0.2895817490494297,
      "grad_norm": 0.6802937388420105,
      "learning_rate": 2.604354084713026e-05,
      "loss": 0.06146836280822754,
      "step": 2142
    },
    {
      "epoch": 0.28971694127587666,
      "grad_norm": 0.1778324544429779,
      "learning_rate": 2.6038907903090446e-05,
      "loss": 0.04071521759033203,
      "step": 2143
    },
    {
      "epoch": 0.28985213350232364,
      "grad_norm": 0.44937044382095337,
      "learning_rate": 2.6034272660691143e-05,
      "loss": 0.08974599838256836,
      "step": 2144
    },
    {
      "epoch": 0.2899873257287706,
      "grad_norm": 0.3087543249130249,
      "learning_rate": 2.6029635120897434e-05,
      "loss": 0.05942583084106445,
      "step": 2145
    },
    {
      "epoch": 0.2901225179552176,
      "grad_norm": 0.587215781211853,
      "learning_rate": 2.6024995284674867e-05,
      "loss": 0.0678091049194336,
      "step": 2146
    },
    {
      "epoch": 0.2902577101816646,
      "grad_norm": 0.6224071979522705,
      "learning_rate": 2.6020353152989496e-05,
      "loss": 0.06561803817749023,
      "step": 2147
    },
    {
      "epoch": 0.29039290240811155,
      "grad_norm": 0.29002130031585693,
      "learning_rate": 2.601570872680783e-05,
      "loss": 0.05462241172790527,
      "step": 2148
    },
    {
      "epoch": 0.29052809463455853,
      "grad_norm": 0.25732851028442383,
      "learning_rate": 2.6011062007096857e-05,
      "loss": 0.050579190254211426,
      "step": 2149
    },
    {
      "epoch": 0.2906632868610055,
      "grad_norm": 0.25392258167266846,
      "learning_rate": 2.6006412994824067e-05,
      "loss": 0.08182525634765625,
      "step": 2150
    },
    {
      "epoch": 0.2907984790874525,
      "grad_norm": 0.3276512622833252,
      "learning_rate": 2.6001761690957388e-05,
      "loss": 0.058895111083984375,
      "step": 2151
    },
    {
      "epoch": 0.29093367131389947,
      "grad_norm": 0.27798473834991455,
      "learning_rate": 2.5997108096465263e-05,
      "loss": 0.05332183837890625,
      "step": 2152
    },
    {
      "epoch": 0.29106886354034645,
      "grad_norm": 0.6182574033737183,
      "learning_rate": 2.599245221231659e-05,
      "loss": 0.10705089569091797,
      "step": 2153
    },
    {
      "epoch": 0.2912040557667934,
      "grad_norm": 0.4207114279270172,
      "learning_rate": 2.5987794039480743e-05,
      "loss": 0.07284879684448242,
      "step": 2154
    },
    {
      "epoch": 0.2913392479932404,
      "grad_norm": 0.28705453872680664,
      "learning_rate": 2.5983133578927584e-05,
      "loss": 0.08143019676208496,
      "step": 2155
    },
    {
      "epoch": 0.2914744402196874,
      "grad_norm": 0.31432029604911804,
      "learning_rate": 2.5978470831627444e-05,
      "loss": 0.062462806701660156,
      "step": 2156
    },
    {
      "epoch": 0.29160963244613436,
      "grad_norm": 0.327834814786911,
      "learning_rate": 2.597380579855113e-05,
      "loss": 0.05445361137390137,
      "step": 2157
    },
    {
      "epoch": 0.29174482467258134,
      "grad_norm": 0.19253234565258026,
      "learning_rate": 2.5969138480669936e-05,
      "loss": 0.04073488712310791,
      "step": 2158
    },
    {
      "epoch": 0.2918800168990283,
      "grad_norm": 0.35155487060546875,
      "learning_rate": 2.5964468878955614e-05,
      "loss": 0.09739065170288086,
      "step": 2159
    },
    {
      "epoch": 0.2920152091254753,
      "grad_norm": 0.8452580571174622,
      "learning_rate": 2.5959796994380397e-05,
      "loss": 0.09544754028320312,
      "step": 2160
    },
    {
      "epoch": 0.2921504013519223,
      "grad_norm": 0.5045964121818542,
      "learning_rate": 2.5955122827917004e-05,
      "loss": 0.06953799724578857,
      "step": 2161
    },
    {
      "epoch": 0.29228559357836925,
      "grad_norm": 0.28637591004371643,
      "learning_rate": 2.595044638053862e-05,
      "loss": 0.06774139404296875,
      "step": 2162
    },
    {
      "epoch": 0.29242078580481623,
      "grad_norm": 0.358574241399765,
      "learning_rate": 2.59457676532189e-05,
      "loss": 0.07538914680480957,
      "step": 2163
    },
    {
      "epoch": 0.2925559780312632,
      "grad_norm": 0.18713060021400452,
      "learning_rate": 2.594108664693199e-05,
      "loss": 0.04306435585021973,
      "step": 2164
    },
    {
      "epoch": 0.2926911702577102,
      "grad_norm": 0.2859300374984741,
      "learning_rate": 2.5936403362652494e-05,
      "loss": 0.05262899398803711,
      "step": 2165
    },
    {
      "epoch": 0.29282636248415717,
      "grad_norm": 0.22714892029762268,
      "learning_rate": 2.5931717801355497e-05,
      "loss": 0.057476043701171875,
      "step": 2166
    },
    {
      "epoch": 0.29296155471060414,
      "grad_norm": 0.4381961524486542,
      "learning_rate": 2.5927029964016556e-05,
      "loss": 0.08054375648498535,
      "step": 2167
    },
    {
      "epoch": 0.2930967469370511,
      "grad_norm": 0.5504139065742493,
      "learning_rate": 2.592233985161171e-05,
      "loss": 0.11686086654663086,
      "step": 2168
    },
    {
      "epoch": 0.2932319391634981,
      "grad_norm": 0.44838497042655945,
      "learning_rate": 2.5917647465117463e-05,
      "loss": 0.05600118637084961,
      "step": 2169
    },
    {
      "epoch": 0.2933671313899451,
      "grad_norm": 0.3687882721424103,
      "learning_rate": 2.591295280551079e-05,
      "loss": 0.09234809875488281,
      "step": 2170
    },
    {
      "epoch": 0.29350232361639206,
      "grad_norm": 0.3056941330432892,
      "learning_rate": 2.590825587376915e-05,
      "loss": 0.07508683204650879,
      "step": 2171
    },
    {
      "epoch": 0.29363751584283904,
      "grad_norm": 0.49831512570381165,
      "learning_rate": 2.5903556670870464e-05,
      "loss": 0.08393239974975586,
      "step": 2172
    },
    {
      "epoch": 0.293772708069286,
      "grad_norm": 0.40404629707336426,
      "learning_rate": 2.589885519779314e-05,
      "loss": 0.07346916198730469,
      "step": 2173
    },
    {
      "epoch": 0.293907900295733,
      "grad_norm": 0.40110450983047485,
      "learning_rate": 2.5894151455516043e-05,
      "loss": 0.048647165298461914,
      "step": 2174
    },
    {
      "epoch": 0.29404309252217997,
      "grad_norm": 0.8366751074790955,
      "learning_rate": 2.5889445445018513e-05,
      "loss": 0.08992195129394531,
      "step": 2175
    },
    {
      "epoch": 0.29417828474862695,
      "grad_norm": 0.5584591031074524,
      "learning_rate": 2.5884737167280375e-05,
      "loss": 0.08277320861816406,
      "step": 2176
    },
    {
      "epoch": 0.29431347697507393,
      "grad_norm": 0.3746296465396881,
      "learning_rate": 2.5880026623281914e-05,
      "loss": 0.051294565200805664,
      "step": 2177
    },
    {
      "epoch": 0.2944486692015209,
      "grad_norm": 0.6029298901557922,
      "learning_rate": 2.5875313814003892e-05,
      "loss": 0.09947824478149414,
      "step": 2178
    },
    {
      "epoch": 0.2945838614279679,
      "grad_norm": 0.3427838683128357,
      "learning_rate": 2.587059874042754e-05,
      "loss": 0.04492020606994629,
      "step": 2179
    },
    {
      "epoch": 0.29471905365441486,
      "grad_norm": 0.325602650642395,
      "learning_rate": 2.5865881403534557e-05,
      "loss": 0.05062389373779297,
      "step": 2180
    },
    {
      "epoch": 0.29485424588086184,
      "grad_norm": 0.5873908400535583,
      "learning_rate": 2.5861161804307124e-05,
      "loss": 0.07400941848754883,
      "step": 2181
    },
    {
      "epoch": 0.2949894381073088,
      "grad_norm": 0.5059689879417419,
      "learning_rate": 2.5856439943727886e-05,
      "loss": 0.07167625427246094,
      "step": 2182
    },
    {
      "epoch": 0.2951246303337558,
      "grad_norm": 0.5450960993766785,
      "learning_rate": 2.5851715822779954e-05,
      "loss": 0.06877422332763672,
      "step": 2183
    },
    {
      "epoch": 0.2952598225602028,
      "grad_norm": 0.6907166242599487,
      "learning_rate": 2.5846989442446926e-05,
      "loss": 0.08760929107666016,
      "step": 2184
    },
    {
      "epoch": 0.29539501478664976,
      "grad_norm": 0.43607136607170105,
      "learning_rate": 2.584226080371285e-05,
      "loss": 0.06807327270507812,
      "step": 2185
    },
    {
      "epoch": 0.29553020701309674,
      "grad_norm": 0.43293964862823486,
      "learning_rate": 2.5837529907562258e-05,
      "loss": 0.07244682312011719,
      "step": 2186
    },
    {
      "epoch": 0.2956653992395437,
      "grad_norm": 0.4194972515106201,
      "learning_rate": 2.5832796754980138e-05,
      "loss": 0.06484603881835938,
      "step": 2187
    },
    {
      "epoch": 0.2958005914659907,
      "grad_norm": 0.8278330564498901,
      "learning_rate": 2.5828061346951974e-05,
      "loss": 0.08581256866455078,
      "step": 2188
    },
    {
      "epoch": 0.29593578369243767,
      "grad_norm": 0.37330037355422974,
      "learning_rate": 2.5823323684463693e-05,
      "loss": 0.07082855701446533,
      "step": 2189
    },
    {
      "epoch": 0.29607097591888465,
      "grad_norm": 0.2774165868759155,
      "learning_rate": 2.5818583768501708e-05,
      "loss": 0.06768918037414551,
      "step": 2190
    },
    {
      "epoch": 0.2962061681453316,
      "grad_norm": 0.23821696639060974,
      "learning_rate": 2.5813841600052887e-05,
      "loss": 0.060115814208984375,
      "step": 2191
    },
    {
      "epoch": 0.2963413603717786,
      "grad_norm": 0.4634384512901306,
      "learning_rate": 2.580909718010458e-05,
      "loss": 0.08556747436523438,
      "step": 2192
    },
    {
      "epoch": 0.2964765525982256,
      "grad_norm": 0.3078383803367615,
      "learning_rate": 2.58043505096446e-05,
      "loss": 0.07528328895568848,
      "step": 2193
    },
    {
      "epoch": 0.29661174482467256,
      "grad_norm": 0.4035947620868683,
      "learning_rate": 2.5799601589661223e-05,
      "loss": 0.09781312942504883,
      "step": 2194
    },
    {
      "epoch": 0.29674693705111954,
      "grad_norm": 0.21481718122959137,
      "learning_rate": 2.579485042114321e-05,
      "loss": 0.032877445220947266,
      "step": 2195
    },
    {
      "epoch": 0.2968821292775665,
      "grad_norm": 0.19529657065868378,
      "learning_rate": 2.5790097005079766e-05,
      "loss": 0.04868650436401367,
      "step": 2196
    },
    {
      "epoch": 0.2970173215040135,
      "grad_norm": 0.37735462188720703,
      "learning_rate": 2.5785341342460595e-05,
      "loss": 0.06742596626281738,
      "step": 2197
    },
    {
      "epoch": 0.2971525137304605,
      "grad_norm": 0.634418249130249,
      "learning_rate": 2.5780583434275837e-05,
      "loss": 0.06396782398223877,
      "step": 2198
    },
    {
      "epoch": 0.29728770595690746,
      "grad_norm": 0.2229696363210678,
      "learning_rate": 2.577582328151612e-05,
      "loss": 0.05273914337158203,
      "step": 2199
    },
    {
      "epoch": 0.29742289818335443,
      "grad_norm": 0.3070399761199951,
      "learning_rate": 2.5771060885172532e-05,
      "loss": 0.048244476318359375,
      "step": 2200
    },
    {
      "epoch": 0.2975580904098014,
      "grad_norm": 0.4507209062576294,
      "learning_rate": 2.5766296246236628e-05,
      "loss": 0.07697296142578125,
      "step": 2201
    },
    {
      "epoch": 0.2976932826362484,
      "grad_norm": 0.3317039906978607,
      "learning_rate": 2.5761529365700437e-05,
      "loss": 0.06859683990478516,
      "step": 2202
    },
    {
      "epoch": 0.29782847486269537,
      "grad_norm": 0.24445976316928864,
      "learning_rate": 2.5756760244556445e-05,
      "loss": 0.06505942344665527,
      "step": 2203
    },
    {
      "epoch": 0.29796366708914235,
      "grad_norm": 0.3230881094932556,
      "learning_rate": 2.5751988883797603e-05,
      "loss": 0.045443058013916016,
      "step": 2204
    },
    {
      "epoch": 0.2980988593155893,
      "grad_norm": 0.5358401536941528,
      "learning_rate": 2.574721528441734e-05,
      "loss": 0.08615493774414062,
      "step": 2205
    },
    {
      "epoch": 0.29823405154203636,
      "grad_norm": 0.5425900816917419,
      "learning_rate": 2.5742439447409545e-05,
      "loss": 0.0880136489868164,
      "step": 2206
    },
    {
      "epoch": 0.29836924376848334,
      "grad_norm": 0.29647156596183777,
      "learning_rate": 2.5737661373768568e-05,
      "loss": 0.07267105579376221,
      "step": 2207
    },
    {
      "epoch": 0.2985044359949303,
      "grad_norm": 0.20283657312393188,
      "learning_rate": 2.5732881064489237e-05,
      "loss": 0.05887603759765625,
      "step": 2208
    },
    {
      "epoch": 0.2986396282213773,
      "grad_norm": 0.31360796093940735,
      "learning_rate": 2.572809852056683e-05,
      "loss": 0.06426763534545898,
      "step": 2209
    },
    {
      "epoch": 0.2987748204478243,
      "grad_norm": 0.32841983437538147,
      "learning_rate": 2.572331374299711e-05,
      "loss": 0.08529233932495117,
      "step": 2210
    },
    {
      "epoch": 0.29891001267427125,
      "grad_norm": 0.5919017791748047,
      "learning_rate": 2.5718526732776276e-05,
      "loss": 0.10047054290771484,
      "step": 2211
    },
    {
      "epoch": 0.29904520490071823,
      "grad_norm": 0.4126860797405243,
      "learning_rate": 2.5713737490901023e-05,
      "loss": 0.10191154479980469,
      "step": 2212
    },
    {
      "epoch": 0.2991803971271652,
      "grad_norm": 0.3165905475616455,
      "learning_rate": 2.570894601836849e-05,
      "loss": 0.07066822052001953,
      "step": 2213
    },
    {
      "epoch": 0.2993155893536122,
      "grad_norm": 0.6350536942481995,
      "learning_rate": 2.5704152316176287e-05,
      "loss": 0.08958101272583008,
      "step": 2214
    },
    {
      "epoch": 0.29945078158005917,
      "grad_norm": 0.2643260657787323,
      "learning_rate": 2.5699356385322487e-05,
      "loss": 0.09143209457397461,
      "step": 2215
    },
    {
      "epoch": 0.29958597380650615,
      "grad_norm": 0.22539664804935455,
      "learning_rate": 2.5694558226805643e-05,
      "loss": 0.04373311996459961,
      "step": 2216
    },
    {
      "epoch": 0.2997211660329531,
      "grad_norm": 0.3142715394496918,
      "learning_rate": 2.568975784162474e-05,
      "loss": 0.06304371356964111,
      "step": 2217
    },
    {
      "epoch": 0.2998563582594001,
      "grad_norm": 0.13800500333309174,
      "learning_rate": 2.5684955230779245e-05,
      "loss": 0.037569522857666016,
      "step": 2218
    },
    {
      "epoch": 0.2999915504858471,
      "grad_norm": 0.343159019947052,
      "learning_rate": 2.5680150395269096e-05,
      "loss": 0.08579206466674805,
      "step": 2219
    },
    {
      "epoch": 0.30012674271229406,
      "grad_norm": 0.32813340425491333,
      "learning_rate": 2.5675343336094683e-05,
      "loss": 0.0595088005065918,
      "step": 2220
    },
    {
      "epoch": 0.30026193493874104,
      "grad_norm": 0.2609620988368988,
      "learning_rate": 2.5670534054256855e-05,
      "loss": 0.057581186294555664,
      "step": 2221
    },
    {
      "epoch": 0.300397127165188,
      "grad_norm": 0.16500750184059143,
      "learning_rate": 2.5665722550756937e-05,
      "loss": 0.0329890251159668,
      "step": 2222
    },
    {
      "epoch": 0.300532319391635,
      "grad_norm": 0.2694440484046936,
      "learning_rate": 2.5660908826596707e-05,
      "loss": 0.06969094276428223,
      "step": 2223
    },
    {
      "epoch": 0.300667511618082,
      "grad_norm": 0.3647930324077606,
      "learning_rate": 2.5656092882778413e-05,
      "loss": 0.07144689559936523,
      "step": 2224
    },
    {
      "epoch": 0.30080270384452895,
      "grad_norm": 0.3648439645767212,
      "learning_rate": 2.565127472030475e-05,
      "loss": 0.09749078750610352,
      "step": 2225
    },
    {
      "epoch": 0.30093789607097593,
      "grad_norm": 0.189004048705101,
      "learning_rate": 2.5646454340178894e-05,
      "loss": 0.034863948822021484,
      "step": 2226
    },
    {
      "epoch": 0.3010730882974229,
      "grad_norm": 0.2944033741950989,
      "learning_rate": 2.564163174340447e-05,
      "loss": 0.046622276306152344,
      "step": 2227
    },
    {
      "epoch": 0.3012082805238699,
      "grad_norm": 0.4640463590621948,
      "learning_rate": 2.5636806930985565e-05,
      "loss": 0.10419654846191406,
      "step": 2228
    },
    {
      "epoch": 0.30134347275031687,
      "grad_norm": 0.44318172335624695,
      "learning_rate": 2.5631979903926738e-05,
      "loss": 0.05919647216796875,
      "step": 2229
    },
    {
      "epoch": 0.30147866497676384,
      "grad_norm": 0.6915941834449768,
      "learning_rate": 2.5627150663233e-05,
      "loss": 0.0716242790222168,
      "step": 2230
    },
    {
      "epoch": 0.3016138572032108,
      "grad_norm": 0.23209555447101593,
      "learning_rate": 2.5622319209909817e-05,
      "loss": 0.06306028366088867,
      "step": 2231
    },
    {
      "epoch": 0.3017490494296578,
      "grad_norm": 0.3597882091999054,
      "learning_rate": 2.5617485544963135e-05,
      "loss": 0.06644511222839355,
      "step": 2232
    },
    {
      "epoch": 0.3018842416561048,
      "grad_norm": 0.32142817974090576,
      "learning_rate": 2.561264966939934e-05,
      "loss": 0.0655069351196289,
      "step": 2233
    },
    {
      "epoch": 0.30201943388255176,
      "grad_norm": 0.5331870913505554,
      "learning_rate": 2.5607811584225294e-05,
      "loss": 0.07327508926391602,
      "step": 2234
    },
    {
      "epoch": 0.30215462610899874,
      "grad_norm": 0.38085871934890747,
      "learning_rate": 2.5602971290448305e-05,
      "loss": 0.05700969696044922,
      "step": 2235
    },
    {
      "epoch": 0.3022898183354457,
      "grad_norm": 0.5730758309364319,
      "learning_rate": 2.5598128789076152e-05,
      "loss": 0.09220600128173828,
      "step": 2236
    },
    {
      "epoch": 0.3024250105618927,
      "grad_norm": 0.24458611011505127,
      "learning_rate": 2.559328408111707e-05,
      "loss": 0.06153547763824463,
      "step": 2237
    },
    {
      "epoch": 0.30256020278833967,
      "grad_norm": 0.24315804243087769,
      "learning_rate": 2.5588437167579755e-05,
      "loss": 0.06522560119628906,
      "step": 2238
    },
    {
      "epoch": 0.30269539501478665,
      "grad_norm": 0.49711737036705017,
      "learning_rate": 2.558358804947335e-05,
      "loss": 0.0958859920501709,
      "step": 2239
    },
    {
      "epoch": 0.30283058724123363,
      "grad_norm": 0.1731814742088318,
      "learning_rate": 2.557873672780748e-05,
      "loss": 0.044841468334198,
      "step": 2240
    },
    {
      "epoch": 0.3029657794676806,
      "grad_norm": 0.40775710344314575,
      "learning_rate": 2.557388320359221e-05,
      "loss": 0.0728074312210083,
      "step": 2241
    },
    {
      "epoch": 0.3031009716941276,
      "grad_norm": 0.35549649596214294,
      "learning_rate": 2.5569027477838068e-05,
      "loss": 0.08945083618164062,
      "step": 2242
    },
    {
      "epoch": 0.30323616392057456,
      "grad_norm": 0.6588951349258423,
      "learning_rate": 2.5564169551556044e-05,
      "loss": 0.06209409236907959,
      "step": 2243
    },
    {
      "epoch": 0.30337135614702154,
      "grad_norm": 0.5763364434242249,
      "learning_rate": 2.5559309425757586e-05,
      "loss": 0.08461761474609375,
      "step": 2244
    },
    {
      "epoch": 0.3035065483734685,
      "grad_norm": 0.5224930047988892,
      "learning_rate": 2.5554447101454597e-05,
      "loss": 0.07148313522338867,
      "step": 2245
    },
    {
      "epoch": 0.3036417405999155,
      "grad_norm": 0.22069016098976135,
      "learning_rate": 2.554958257965944e-05,
      "loss": 0.05002164840698242,
      "step": 2246
    },
    {
      "epoch": 0.3037769328263625,
      "grad_norm": 0.2523531913757324,
      "learning_rate": 2.554471586138493e-05,
      "loss": 0.04245567321777344,
      "step": 2247
    },
    {
      "epoch": 0.30391212505280946,
      "grad_norm": 0.5251622796058655,
      "learning_rate": 2.5539846947644342e-05,
      "loss": 0.05673050880432129,
      "step": 2248
    },
    {
      "epoch": 0.30404731727925643,
      "grad_norm": 0.7554552555084229,
      "learning_rate": 2.5534975839451416e-05,
      "loss": 0.07118797302246094,
      "step": 2249
    },
    {
      "epoch": 0.3041825095057034,
      "grad_norm": 0.39777493476867676,
      "learning_rate": 2.5530102537820348e-05,
      "loss": 0.07718515396118164,
      "step": 2250
    },
    {
      "epoch": 0.3043177017321504,
      "grad_norm": 0.32851290702819824,
      "learning_rate": 2.5525227043765774e-05,
      "loss": 0.07964611053466797,
      "step": 2251
    },
    {
      "epoch": 0.30445289395859737,
      "grad_norm": 0.47780588269233704,
      "learning_rate": 2.55203493583028e-05,
      "loss": 0.05550628900527954,
      "step": 2252
    },
    {
      "epoch": 0.30458808618504435,
      "grad_norm": 0.217255637049675,
      "learning_rate": 2.551546948244699e-05,
      "loss": 0.05336284637451172,
      "step": 2253
    },
    {
      "epoch": 0.3047232784114913,
      "grad_norm": 0.5100780129432678,
      "learning_rate": 2.551058741721436e-05,
      "loss": 0.08109474182128906,
      "step": 2254
    },
    {
      "epoch": 0.3048584706379383,
      "grad_norm": 0.4553520083427429,
      "learning_rate": 2.550570316362138e-05,
      "loss": 0.07550406455993652,
      "step": 2255
    },
    {
      "epoch": 0.3049936628643853,
      "grad_norm": 0.3899535536766052,
      "learning_rate": 2.5500816722684975e-05,
      "loss": 0.05846214294433594,
      "step": 2256
    },
    {
      "epoch": 0.30512885509083226,
      "grad_norm": 0.4410453736782074,
      "learning_rate": 2.549592809542253e-05,
      "loss": 0.06292164325714111,
      "step": 2257
    },
    {
      "epoch": 0.30526404731727924,
      "grad_norm": 0.3141182065010071,
      "learning_rate": 2.549103728285189e-05,
      "loss": 0.07396173477172852,
      "step": 2258
    },
    {
      "epoch": 0.3053992395437262,
      "grad_norm": 0.4648004472255707,
      "learning_rate": 2.548614428599134e-05,
      "loss": 0.08410477638244629,
      "step": 2259
    },
    {
      "epoch": 0.3055344317701732,
      "grad_norm": 0.3898724615573883,
      "learning_rate": 2.5481249105859633e-05,
      "loss": 0.05596303939819336,
      "step": 2260
    },
    {
      "epoch": 0.3056696239966202,
      "grad_norm": 0.3171229660511017,
      "learning_rate": 2.5476351743475964e-05,
      "loss": 0.06770706176757812,
      "step": 2261
    },
    {
      "epoch": 0.30580481622306716,
      "grad_norm": 0.3378964364528656,
      "learning_rate": 2.547145219986e-05,
      "loss": 0.06552982330322266,
      "step": 2262
    },
    {
      "epoch": 0.30594000844951413,
      "grad_norm": 0.288042277097702,
      "learning_rate": 2.5466550476031846e-05,
      "loss": 0.057799339294433594,
      "step": 2263
    },
    {
      "epoch": 0.3060752006759611,
      "grad_norm": 0.8326659202575684,
      "learning_rate": 2.5461646573012072e-05,
      "loss": 0.09737968444824219,
      "step": 2264
    },
    {
      "epoch": 0.3062103929024081,
      "grad_norm": 0.615618109703064,
      "learning_rate": 2.5456740491821687e-05,
      "loss": 0.0564122200012207,
      "step": 2265
    },
    {
      "epoch": 0.30634558512885507,
      "grad_norm": 0.278624027967453,
      "learning_rate": 2.5451832233482172e-05,
      "loss": 0.07700729370117188,
      "step": 2266
    },
    {
      "epoch": 0.30648077735530205,
      "grad_norm": 0.7243728637695312,
      "learning_rate": 2.544692179901545e-05,
      "loss": 0.08532524108886719,
      "step": 2267
    },
    {
      "epoch": 0.306615969581749,
      "grad_norm": 0.31695494055747986,
      "learning_rate": 2.5442009189443902e-05,
      "loss": 0.04893636703491211,
      "step": 2268
    },
    {
      "epoch": 0.306751161808196,
      "grad_norm": 0.41937071084976196,
      "learning_rate": 2.5437094405790355e-05,
      "loss": 0.050414085388183594,
      "step": 2269
    },
    {
      "epoch": 0.306886354034643,
      "grad_norm": 0.7431758642196655,
      "learning_rate": 2.5432177449078096e-05,
      "loss": 0.08933520317077637,
      "step": 2270
    },
    {
      "epoch": 0.30702154626108996,
      "grad_norm": 0.32833048701286316,
      "learning_rate": 2.5427258320330857e-05,
      "loss": 0.03657245635986328,
      "step": 2271
    },
    {
      "epoch": 0.30715673848753694,
      "grad_norm": 0.326952189207077,
      "learning_rate": 2.5422337020572835e-05,
      "loss": 0.057470083236694336,
      "step": 2272
    },
    {
      "epoch": 0.3072919307139839,
      "grad_norm": 0.21216987073421478,
      "learning_rate": 2.5417413550828667e-05,
      "loss": 0.052736759185791016,
      "step": 2273
    },
    {
      "epoch": 0.30742712294043095,
      "grad_norm": 0.2672852575778961,
      "learning_rate": 2.5412487912123444e-05,
      "loss": 0.06456470489501953,
      "step": 2274
    },
    {
      "epoch": 0.30756231516687793,
      "grad_norm": 0.2392459213733673,
      "learning_rate": 2.5407560105482708e-05,
      "loss": 0.06494379043579102,
      "step": 2275
    },
    {
      "epoch": 0.3076975073933249,
      "grad_norm": 0.20050475001335144,
      "learning_rate": 2.540263013193246e-05,
      "loss": 0.0528712272644043,
      "step": 2276
    },
    {
      "epoch": 0.3078326996197719,
      "grad_norm": 0.5630680322647095,
      "learning_rate": 2.539769799249915e-05,
      "loss": 0.0738229751586914,
      "step": 2277
    },
    {
      "epoch": 0.30796789184621887,
      "grad_norm": 0.27651989459991455,
      "learning_rate": 2.5392763688209666e-05,
      "loss": 0.06695342063903809,
      "step": 2278
    },
    {
      "epoch": 0.30810308407266584,
      "grad_norm": 0.5245217084884644,
      "learning_rate": 2.5387827220091362e-05,
      "loss": 0.07639026641845703,
      "step": 2279
    },
    {
      "epoch": 0.3082382762991128,
      "grad_norm": 0.5639858245849609,
      "learning_rate": 2.538288858917204e-05,
      "loss": 0.0779733657836914,
      "step": 2280
    },
    {
      "epoch": 0.3083734685255598,
      "grad_norm": 50.24871826171875,
      "learning_rate": 2.5377947796479936e-05,
      "loss": 0.07575416564941406,
      "step": 2281
    },
    {
      "epoch": 0.3085086607520068,
      "grad_norm": 0.19175417721271515,
      "learning_rate": 2.537300484304377e-05,
      "loss": 0.049895405769348145,
      "step": 2282
    },
    {
      "epoch": 0.30864385297845376,
      "grad_norm": 0.5713838934898376,
      "learning_rate": 2.536805972989267e-05,
      "loss": 0.05373960733413696,
      "step": 2283
    },
    {
      "epoch": 0.30877904520490074,
      "grad_norm": 0.43302640318870544,
      "learning_rate": 2.5363112458056252e-05,
      "loss": 0.061408042907714844,
      "step": 2284
    },
    {
      "epoch": 0.3089142374313477,
      "grad_norm": 0.41834351420402527,
      "learning_rate": 2.5358163028564552e-05,
      "loss": 0.09498310089111328,
      "step": 2285
    },
    {
      "epoch": 0.3090494296577947,
      "grad_norm": 0.1747867912054062,
      "learning_rate": 2.535321144244808e-05,
      "loss": 0.053493499755859375,
      "step": 2286
    },
    {
      "epoch": 0.3091846218842417,
      "grad_norm": 0.26045501232147217,
      "learning_rate": 2.534825770073777e-05,
      "loss": 0.06392765045166016,
      "step": 2287
    },
    {
      "epoch": 0.30931981411068865,
      "grad_norm": 0.43064865469932556,
      "learning_rate": 2.5343301804465026e-05,
      "loss": 0.08998680114746094,
      "step": 2288
    },
    {
      "epoch": 0.30945500633713563,
      "grad_norm": 0.46299099922180176,
      "learning_rate": 2.533834375466169e-05,
      "loss": 0.08238840103149414,
      "step": 2289
    },
    {
      "epoch": 0.3095901985635826,
      "grad_norm": 0.37767377495765686,
      "learning_rate": 2.533338355236005e-05,
      "loss": 0.06478118896484375,
      "step": 2290
    },
    {
      "epoch": 0.3097253907900296,
      "grad_norm": 0.31063926219940186,
      "learning_rate": 2.532842119859285e-05,
      "loss": 0.06270742416381836,
      "step": 2291
    },
    {
      "epoch": 0.30986058301647657,
      "grad_norm": 0.7849839925765991,
      "learning_rate": 2.532345669439328e-05,
      "loss": 0.09780025482177734,
      "step": 2292
    },
    {
      "epoch": 0.30999577524292354,
      "grad_norm": 0.24539461731910706,
      "learning_rate": 2.5318490040794975e-05,
      "loss": 0.04785728454589844,
      "step": 2293
    },
    {
      "epoch": 0.3101309674693705,
      "grad_norm": 0.3790227472782135,
      "learning_rate": 2.531352123883202e-05,
      "loss": 0.07446098327636719,
      "step": 2294
    },
    {
      "epoch": 0.3102661596958175,
      "grad_norm": 0.33788323402404785,
      "learning_rate": 2.530855028953894e-05,
      "loss": 0.07399749755859375,
      "step": 2295
    },
    {
      "epoch": 0.3104013519222645,
      "grad_norm": 0.3270670473575592,
      "learning_rate": 2.5303577193950724e-05,
      "loss": 0.058358192443847656,
      "step": 2296
    },
    {
      "epoch": 0.31053654414871146,
      "grad_norm": 0.4766222834587097,
      "learning_rate": 2.5298601953102785e-05,
      "loss": 0.0890495777130127,
      "step": 2297
    },
    {
      "epoch": 0.31067173637515844,
      "grad_norm": 0.5018221139907837,
      "learning_rate": 2.5293624568031008e-05,
      "loss": 0.06609797477722168,
      "step": 2298
    },
    {
      "epoch": 0.3108069286016054,
      "grad_norm": 0.4079321622848511,
      "learning_rate": 2.5288645039771697e-05,
      "loss": 0.05998802185058594,
      "step": 2299
    },
    {
      "epoch": 0.3109421208280524,
      "grad_norm": 0.39015671610832214,
      "learning_rate": 2.5283663369361624e-05,
      "loss": 0.04808831214904785,
      "step": 2300
    },
    {
      "epoch": 0.31107731305449937,
      "grad_norm": 0.28397712111473083,
      "learning_rate": 2.5278679557837998e-05,
      "loss": 0.04781508445739746,
      "step": 2301
    },
    {
      "epoch": 0.31121250528094635,
      "grad_norm": 0.8658705949783325,
      "learning_rate": 2.5273693606238474e-05,
      "loss": 0.09870147705078125,
      "step": 2302
    },
    {
      "epoch": 0.31134769750739333,
      "grad_norm": 0.43585294485092163,
      "learning_rate": 2.5268705515601164e-05,
      "loss": 0.07146501541137695,
      "step": 2303
    },
    {
      "epoch": 0.3114828897338403,
      "grad_norm": 0.20357875525951385,
      "learning_rate": 2.5263715286964596e-05,
      "loss": 0.04984736442565918,
      "step": 2304
    },
    {
      "epoch": 0.3116180819602873,
      "grad_norm": 0.3360837996006012,
      "learning_rate": 2.525872292136778e-05,
      "loss": 0.07769489288330078,
      "step": 2305
    },
    {
      "epoch": 0.31175327418673426,
      "grad_norm": 0.30891481041908264,
      "learning_rate": 2.525372841985014e-05,
      "loss": 0.0885767936706543,
      "step": 2306
    },
    {
      "epoch": 0.31188846641318124,
      "grad_norm": 0.19803576171398163,
      "learning_rate": 2.5248731783451567e-05,
      "loss": 0.04516786336898804,
      "step": 2307
    },
    {
      "epoch": 0.3120236586396282,
      "grad_norm": 0.33781781792640686,
      "learning_rate": 2.524373301321238e-05,
      "loss": 0.07804155349731445,
      "step": 2308
    },
    {
      "epoch": 0.3121588508660752,
      "grad_norm": 0.21870151162147522,
      "learning_rate": 2.5238732110173356e-05,
      "loss": 0.06848716735839844,
      "step": 2309
    },
    {
      "epoch": 0.3122940430925222,
      "grad_norm": 0.30236539244651794,
      "learning_rate": 2.5233729075375708e-05,
      "loss": 0.06928348541259766,
      "step": 2310
    },
    {
      "epoch": 0.31242923531896916,
      "grad_norm": 0.350146621465683,
      "learning_rate": 2.522872390986109e-05,
      "loss": 0.0879526138305664,
      "step": 2311
    },
    {
      "epoch": 0.31256442754541613,
      "grad_norm": 0.34718042612075806,
      "learning_rate": 2.522371661467161e-05,
      "loss": 0.03924298286437988,
      "step": 2312
    },
    {
      "epoch": 0.3126996197718631,
      "grad_norm": 0.39739125967025757,
      "learning_rate": 2.521870719084981e-05,
      "loss": 0.08159017562866211,
      "step": 2313
    },
    {
      "epoch": 0.3128348119983101,
      "grad_norm": 0.5475897789001465,
      "learning_rate": 2.5213695639438686e-05,
      "loss": 0.07092475891113281,
      "step": 2314
    },
    {
      "epoch": 0.31297000422475707,
      "grad_norm": 0.6640769839286804,
      "learning_rate": 2.5208681961481657e-05,
      "loss": 0.10393238067626953,
      "step": 2315
    },
    {
      "epoch": 0.31310519645120405,
      "grad_norm": 0.28634682297706604,
      "learning_rate": 2.5203666158022607e-05,
      "loss": 0.04479193687438965,
      "step": 2316
    },
    {
      "epoch": 0.313240388677651,
      "grad_norm": 0.238839790225029,
      "learning_rate": 2.519864823010585e-05,
      "loss": 0.05104637145996094,
      "step": 2317
    },
    {
      "epoch": 0.313375580904098,
      "grad_norm": 0.20142219960689545,
      "learning_rate": 2.5193628178776148e-05,
      "loss": 0.04235410690307617,
      "step": 2318
    },
    {
      "epoch": 0.313510773130545,
      "grad_norm": 0.2510052025318146,
      "learning_rate": 2.5188606005078695e-05,
      "loss": 0.07171773910522461,
      "step": 2319
    },
    {
      "epoch": 0.31364596535699196,
      "grad_norm": 0.23331953585147858,
      "learning_rate": 2.518358171005914e-05,
      "loss": 0.0742340087890625,
      "step": 2320
    },
    {
      "epoch": 0.31378115758343894,
      "grad_norm": 0.5179891586303711,
      "learning_rate": 2.517855529476357e-05,
      "loss": 0.09337329864501953,
      "step": 2321
    },
    {
      "epoch": 0.3139163498098859,
      "grad_norm": 0.3570202589035034,
      "learning_rate": 2.517352676023851e-05,
      "loss": 0.06017327308654785,
      "step": 2322
    },
    {
      "epoch": 0.3140515420363329,
      "grad_norm": 0.5075674653053284,
      "learning_rate": 2.5168496107530925e-05,
      "loss": 0.06083083152770996,
      "step": 2323
    },
    {
      "epoch": 0.3141867342627799,
      "grad_norm": 0.3785947859287262,
      "learning_rate": 2.5163463337688224e-05,
      "loss": 0.06925487518310547,
      "step": 2324
    },
    {
      "epoch": 0.31432192648922685,
      "grad_norm": 0.35839200019836426,
      "learning_rate": 2.515842845175826e-05,
      "loss": 0.05210590362548828,
      "step": 2325
    },
    {
      "epoch": 0.31445711871567383,
      "grad_norm": 0.24263793230056763,
      "learning_rate": 2.5153391450789326e-05,
      "loss": 0.04543876647949219,
      "step": 2326
    },
    {
      "epoch": 0.3145923109421208,
      "grad_norm": 0.2730138301849365,
      "learning_rate": 2.514835233583014e-05,
      "loss": 0.0621342658996582,
      "step": 2327
    },
    {
      "epoch": 0.3147275031685678,
      "grad_norm": 0.2266179472208023,
      "learning_rate": 2.514331110792988e-05,
      "loss": 0.054215431213378906,
      "step": 2328
    },
    {
      "epoch": 0.31486269539501477,
      "grad_norm": 0.18004214763641357,
      "learning_rate": 2.513826776813816e-05,
      "loss": 0.060683250427246094,
      "step": 2329
    },
    {
      "epoch": 0.31499788762146175,
      "grad_norm": 0.3396877944469452,
      "learning_rate": 2.5133222317505024e-05,
      "loss": 0.07338428497314453,
      "step": 2330
    },
    {
      "epoch": 0.3151330798479087,
      "grad_norm": 0.28569263219833374,
      "learning_rate": 2.5128174757080965e-05,
      "loss": 0.07358694076538086,
      "step": 2331
    },
    {
      "epoch": 0.3152682720743557,
      "grad_norm": 0.5643778443336487,
      "learning_rate": 2.5123125087916916e-05,
      "loss": 0.06362366676330566,
      "step": 2332
    },
    {
      "epoch": 0.3154034643008027,
      "grad_norm": 0.54227614402771,
      "learning_rate": 2.5118073311064236e-05,
      "loss": 0.06790781021118164,
      "step": 2333
    },
    {
      "epoch": 0.31553865652724966,
      "grad_norm": 0.39101892709732056,
      "learning_rate": 2.5113019427574734e-05,
      "loss": 0.06030726432800293,
      "step": 2334
    },
    {
      "epoch": 0.31567384875369664,
      "grad_norm": 0.44047442078590393,
      "learning_rate": 2.5107963438500666e-05,
      "loss": 0.0663461685180664,
      "step": 2335
    },
    {
      "epoch": 0.3158090409801436,
      "grad_norm": 0.21675944328308105,
      "learning_rate": 2.51029053448947e-05,
      "loss": 0.05660057067871094,
      "step": 2336
    },
    {
      "epoch": 0.3159442332065906,
      "grad_norm": 0.3622889816761017,
      "learning_rate": 2.509784514780997e-05,
      "loss": 0.057483673095703125,
      "step": 2337
    },
    {
      "epoch": 0.3160794254330376,
      "grad_norm": 0.5427830815315247,
      "learning_rate": 2.5092782848300033e-05,
      "loss": 0.0858011245727539,
      "step": 2338
    },
    {
      "epoch": 0.31621461765948455,
      "grad_norm": 0.7293141484260559,
      "learning_rate": 2.5087718447418886e-05,
      "loss": 0.07434439659118652,
      "step": 2339
    },
    {
      "epoch": 0.31634980988593153,
      "grad_norm": 0.22585107386112213,
      "learning_rate": 2.5082651946220958e-05,
      "loss": 0.07416868209838867,
      "step": 2340
    },
    {
      "epoch": 0.3164850021123785,
      "grad_norm": 0.2874718904495239,
      "learning_rate": 2.507758334576113e-05,
      "loss": 0.06458187103271484,
      "step": 2341
    },
    {
      "epoch": 0.31662019433882554,
      "grad_norm": 0.40999042987823486,
      "learning_rate": 2.5072512647094713e-05,
      "loss": 0.07197344303131104,
      "step": 2342
    },
    {
      "epoch": 0.3167553865652725,
      "grad_norm": 0.2622562348842621,
      "learning_rate": 2.506743985127745e-05,
      "loss": 0.08288097381591797,
      "step": 2343
    },
    {
      "epoch": 0.3168905787917195,
      "grad_norm": 0.3014708459377289,
      "learning_rate": 2.506236495936552e-05,
      "loss": 0.05206465721130371,
      "step": 2344
    },
    {
      "epoch": 0.3170257710181665,
      "grad_norm": 0.22873209416866302,
      "learning_rate": 2.5057287972415547e-05,
      "loss": 0.06554126739501953,
      "step": 2345
    },
    {
      "epoch": 0.31716096324461346,
      "grad_norm": 0.2425178438425064,
      "learning_rate": 2.5052208891484588e-05,
      "loss": 0.06679439544677734,
      "step": 2346
    },
    {
      "epoch": 0.31729615547106044,
      "grad_norm": 0.22180858254432678,
      "learning_rate": 2.504712771763013e-05,
      "loss": 0.059993743896484375,
      "step": 2347
    },
    {
      "epoch": 0.3174313476975074,
      "grad_norm": 0.31056347489356995,
      "learning_rate": 2.5042044451910108e-05,
      "loss": 0.08017396926879883,
      "step": 2348
    },
    {
      "epoch": 0.3175665399239544,
      "grad_norm": 0.5533975958824158,
      "learning_rate": 2.5036959095382875e-05,
      "loss": 0.0951075553894043,
      "step": 2349
    },
    {
      "epoch": 0.3177017321504014,
      "grad_norm": 0.26910534501075745,
      "learning_rate": 2.5031871649107233e-05,
      "loss": 0.050804853439331055,
      "step": 2350
    },
    {
      "epoch": 0.31783692437684835,
      "grad_norm": 0.4528810679912567,
      "learning_rate": 2.5026782114142426e-05,
      "loss": 0.08442306518554688,
      "step": 2351
    },
    {
      "epoch": 0.31797211660329533,
      "grad_norm": 0.6761053204536438,
      "learning_rate": 2.5021690491548107e-05,
      "loss": 0.10964083671569824,
      "step": 2352
    },
    {
      "epoch": 0.3181073088297423,
      "grad_norm": 0.2603733241558075,
      "learning_rate": 2.5016596782384387e-05,
      "loss": 0.04936408996582031,
      "step": 2353
    },
    {
      "epoch": 0.3182425010561893,
      "grad_norm": 0.4187442362308502,
      "learning_rate": 2.5011500987711804e-05,
      "loss": 0.09982156753540039,
      "step": 2354
    },
    {
      "epoch": 0.31837769328263626,
      "grad_norm": 0.3205567002296448,
      "learning_rate": 2.5006403108591325e-05,
      "loss": 0.07063543796539307,
      "step": 2355
    },
    {
      "epoch": 0.31851288550908324,
      "grad_norm": 0.35416197776794434,
      "learning_rate": 2.500130314608436e-05,
      "loss": 0.07037878036499023,
      "step": 2356
    },
    {
      "epoch": 0.3186480777355302,
      "grad_norm": 0.3131636381149292,
      "learning_rate": 2.4996201101252742e-05,
      "loss": 0.06651270389556885,
      "step": 2357
    },
    {
      "epoch": 0.3187832699619772,
      "grad_norm": 1.0636065006256104,
      "learning_rate": 2.4991096975158757e-05,
      "loss": 0.07877635955810547,
      "step": 2358
    },
    {
      "epoch": 0.3189184621884242,
      "grad_norm": 0.706203818321228,
      "learning_rate": 2.4985990768865095e-05,
      "loss": 0.07495427131652832,
      "step": 2359
    },
    {
      "epoch": 0.31905365441487116,
      "grad_norm": 0.3577408194541931,
      "learning_rate": 2.4980882483434904e-05,
      "loss": 0.06362533569335938,
      "step": 2360
    },
    {
      "epoch": 0.31918884664131814,
      "grad_norm": 0.2690145671367645,
      "learning_rate": 2.497577211993176e-05,
      "loss": 0.07851815223693848,
      "step": 2361
    },
    {
      "epoch": 0.3193240388677651,
      "grad_norm": 0.2870801091194153,
      "learning_rate": 2.4970659679419658e-05,
      "loss": 0.06454581022262573,
      "step": 2362
    },
    {
      "epoch": 0.3194592310942121,
      "grad_norm": 0.5231418013572693,
      "learning_rate": 2.496554516296304e-05,
      "loss": 0.08087480068206787,
      "step": 2363
    },
    {
      "epoch": 0.31959442332065907,
      "grad_norm": 0.5498780608177185,
      "learning_rate": 2.4960428571626784e-05,
      "loss": 0.08754825592041016,
      "step": 2364
    },
    {
      "epoch": 0.31972961554710605,
      "grad_norm": 0.290738046169281,
      "learning_rate": 2.4955309906476177e-05,
      "loss": 0.07848024368286133,
      "step": 2365
    },
    {
      "epoch": 0.31986480777355303,
      "grad_norm": 0.21460847556591034,
      "learning_rate": 2.495018916857696e-05,
      "loss": 0.06312370300292969,
      "step": 2366
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6291492581367493,
      "learning_rate": 2.4945066358995304e-05,
      "loss": 0.07329463958740234,
      "step": 2367
    },
    {
      "epoch": 0.320135192226447,
      "grad_norm": 0.21784338355064392,
      "learning_rate": 2.493994147879779e-05,
      "loss": 0.056948184967041016,
      "step": 2368
    },
    {
      "epoch": 0.32027038445289396,
      "grad_norm": 0.6110391616821289,
      "learning_rate": 2.4934814529051458e-05,
      "loss": 0.0796196460723877,
      "step": 2369
    },
    {
      "epoch": 0.32040557667934094,
      "grad_norm": 0.31653475761413574,
      "learning_rate": 2.4929685510823763e-05,
      "loss": 0.0922393798828125,
      "step": 2370
    },
    {
      "epoch": 0.3205407689057879,
      "grad_norm": 0.44890958070755005,
      "learning_rate": 2.492455442518259e-05,
      "loss": 0.08061408996582031,
      "step": 2371
    },
    {
      "epoch": 0.3206759611322349,
      "grad_norm": 0.3351346552371979,
      "learning_rate": 2.4919421273196262e-05,
      "loss": 0.058281898498535156,
      "step": 2372
    },
    {
      "epoch": 0.3208111533586819,
      "grad_norm": 0.2324884682893753,
      "learning_rate": 2.4914286055933527e-05,
      "loss": 0.064056396484375,
      "step": 2373
    },
    {
      "epoch": 0.32094634558512886,
      "grad_norm": 0.4020441472530365,
      "learning_rate": 2.4909148774463572e-05,
      "loss": 0.09259605407714844,
      "step": 2374
    },
    {
      "epoch": 0.32108153781157583,
      "grad_norm": 0.30229443311691284,
      "learning_rate": 2.4904009429855992e-05,
      "loss": 0.046872496604919434,
      "step": 2375
    },
    {
      "epoch": 0.3212167300380228,
      "grad_norm": 0.43443962931632996,
      "learning_rate": 2.4898868023180844e-05,
      "loss": 0.06595730781555176,
      "step": 2376
    },
    {
      "epoch": 0.3213519222644698,
      "grad_norm": 0.39085280895233154,
      "learning_rate": 2.4893724555508575e-05,
      "loss": 0.10126161575317383,
      "step": 2377
    },
    {
      "epoch": 0.32148711449091677,
      "grad_norm": 0.25762200355529785,
      "learning_rate": 2.4888579027910105e-05,
      "loss": 0.07214164733886719,
      "step": 2378
    },
    {
      "epoch": 0.32162230671736375,
      "grad_norm": 0.2287529706954956,
      "learning_rate": 2.4883431441456738e-05,
      "loss": 0.07157754898071289,
      "step": 2379
    },
    {
      "epoch": 0.3217574989438107,
      "grad_norm": 0.3518243432044983,
      "learning_rate": 2.4878281797220244e-05,
      "loss": 0.06694567203521729,
      "step": 2380
    },
    {
      "epoch": 0.3218926911702577,
      "grad_norm": 0.4456416070461273,
      "learning_rate": 2.4873130096272805e-05,
      "loss": 0.07938957214355469,
      "step": 2381
    },
    {
      "epoch": 0.3220278833967047,
      "grad_norm": 0.26300668716430664,
      "learning_rate": 2.4867976339687026e-05,
      "loss": 0.06231880187988281,
      "step": 2382
    },
    {
      "epoch": 0.32216307562315166,
      "grad_norm": 0.8203449249267578,
      "learning_rate": 2.4862820528535955e-05,
      "loss": 0.0864250659942627,
      "step": 2383
    },
    {
      "epoch": 0.32229826784959864,
      "grad_norm": 0.43552136421203613,
      "learning_rate": 2.4857662663893054e-05,
      "loss": 0.0946817398071289,
      "step": 2384
    },
    {
      "epoch": 0.3224334600760456,
      "grad_norm": 0.18367503583431244,
      "learning_rate": 2.485250274683222e-05,
      "loss": 0.02898305654525757,
      "step": 2385
    },
    {
      "epoch": 0.3225686523024926,
      "grad_norm": 0.18898211419582367,
      "learning_rate": 2.4847340778427772e-05,
      "loss": 0.05438041687011719,
      "step": 2386
    },
    {
      "epoch": 0.3227038445289396,
      "grad_norm": 0.33147570490837097,
      "learning_rate": 2.484217675975446e-05,
      "loss": 0.05866193771362305,
      "step": 2387
    },
    {
      "epoch": 0.32283903675538655,
      "grad_norm": 0.9009811282157898,
      "learning_rate": 2.4837010691887466e-05,
      "loss": 0.09663963317871094,
      "step": 2388
    },
    {
      "epoch": 0.32297422898183353,
      "grad_norm": 0.7079839706420898,
      "learning_rate": 2.4831842575902383e-05,
      "loss": 0.08978748321533203,
      "step": 2389
    },
    {
      "epoch": 0.3231094212082805,
      "grad_norm": 0.27973049879074097,
      "learning_rate": 2.482667241287525e-05,
      "loss": 0.0802769660949707,
      "step": 2390
    },
    {
      "epoch": 0.3232446134347275,
      "grad_norm": 0.21611611545085907,
      "learning_rate": 2.4821500203882517e-05,
      "loss": 0.051239967346191406,
      "step": 2391
    },
    {
      "epoch": 0.32337980566117447,
      "grad_norm": 0.4308067858219147,
      "learning_rate": 2.4816325950001067e-05,
      "loss": 0.08449077606201172,
      "step": 2392
    },
    {
      "epoch": 0.32351499788762145,
      "grad_norm": 0.18825387954711914,
      "learning_rate": 2.4811149652308205e-05,
      "loss": 0.06111717224121094,
      "step": 2393
    },
    {
      "epoch": 0.3236501901140684,
      "grad_norm": 0.42325976490974426,
      "learning_rate": 2.480597131188167e-05,
      "loss": 0.09814834594726562,
      "step": 2394
    },
    {
      "epoch": 0.3237853823405154,
      "grad_norm": 0.27803319692611694,
      "learning_rate": 2.4800790929799614e-05,
      "loss": 0.07898426055908203,
      "step": 2395
    },
    {
      "epoch": 0.3239205745669624,
      "grad_norm": 0.3575000464916229,
      "learning_rate": 2.4795608507140623e-05,
      "loss": 0.07029938697814941,
      "step": 2396
    },
    {
      "epoch": 0.32405576679340936,
      "grad_norm": 0.5440062880516052,
      "learning_rate": 2.4790424044983705e-05,
      "loss": 0.09047508239746094,
      "step": 2397
    },
    {
      "epoch": 0.32419095901985634,
      "grad_norm": 0.30727916955947876,
      "learning_rate": 2.4785237544408288e-05,
      "loss": 0.09341716766357422,
      "step": 2398
    },
    {
      "epoch": 0.3243261512463033,
      "grad_norm": 0.3869057297706604,
      "learning_rate": 2.478004900649424e-05,
      "loss": 0.07088875770568848,
      "step": 2399
    },
    {
      "epoch": 0.3244613434727503,
      "grad_norm": 0.3073582351207733,
      "learning_rate": 2.477485843232183e-05,
      "loss": 0.05738997459411621,
      "step": 2400
    },
    {
      "epoch": 0.3245965356991973,
      "grad_norm": 0.26189202070236206,
      "learning_rate": 2.476966582297177e-05,
      "loss": 0.08462309837341309,
      "step": 2401
    },
    {
      "epoch": 0.32473172792564425,
      "grad_norm": 0.4069730341434479,
      "learning_rate": 2.4764471179525188e-05,
      "loss": 0.08410358428955078,
      "step": 2402
    },
    {
      "epoch": 0.32486692015209123,
      "grad_norm": 0.3748754858970642,
      "learning_rate": 2.4759274503063632e-05,
      "loss": 0.09998178482055664,
      "step": 2403
    },
    {
      "epoch": 0.3250021123785382,
      "grad_norm": 0.760499894618988,
      "learning_rate": 2.4754075794669088e-05,
      "loss": 0.0803825855255127,
      "step": 2404
    },
    {
      "epoch": 0.3251373046049852,
      "grad_norm": 0.3767063319683075,
      "learning_rate": 2.4748875055423942e-05,
      "loss": 0.08749151229858398,
      "step": 2405
    },
    {
      "epoch": 0.32527249683143217,
      "grad_norm": 0.5090206265449524,
      "learning_rate": 2.4743672286411027e-05,
      "loss": 0.07190060615539551,
      "step": 2406
    },
    {
      "epoch": 0.32540768905787915,
      "grad_norm": 0.4415806531906128,
      "learning_rate": 2.4738467488713582e-05,
      "loss": 0.08155441284179688,
      "step": 2407
    },
    {
      "epoch": 0.3255428812843261,
      "grad_norm": 0.5361815690994263,
      "learning_rate": 2.473326066341527e-05,
      "loss": 0.08111095428466797,
      "step": 2408
    },
    {
      "epoch": 0.3256780735107731,
      "grad_norm": 0.246195986866951,
      "learning_rate": 2.4728051811600184e-05,
      "loss": 0.06524133682250977,
      "step": 2409
    },
    {
      "epoch": 0.32581326573722014,
      "grad_norm": 0.4579744338989258,
      "learning_rate": 2.4722840934352838e-05,
      "loss": 0.09024429321289062,
      "step": 2410
    },
    {
      "epoch": 0.3259484579636671,
      "grad_norm": 0.4055515229701996,
      "learning_rate": 2.471762803275816e-05,
      "loss": 0.059591054916381836,
      "step": 2411
    },
    {
      "epoch": 0.3260836501901141,
      "grad_norm": 0.4381348490715027,
      "learning_rate": 2.4712413107901504e-05,
      "loss": 0.07091140747070312,
      "step": 2412
    },
    {
      "epoch": 0.3262188424165611,
      "grad_norm": 0.30810537934303284,
      "learning_rate": 2.470719616086865e-05,
      "loss": 0.0596085786819458,
      "step": 2413
    },
    {
      "epoch": 0.32635403464300805,
      "grad_norm": 0.34375977516174316,
      "learning_rate": 2.4701977192745785e-05,
      "loss": 0.0819559097290039,
      "step": 2414
    },
    {
      "epoch": 0.32648922686945503,
      "grad_norm": 0.3275398313999176,
      "learning_rate": 2.4696756204619535e-05,
      "loss": 0.07708930969238281,
      "step": 2415
    },
    {
      "epoch": 0.326624419095902,
      "grad_norm": 0.5567411184310913,
      "learning_rate": 2.469153319757693e-05,
      "loss": 0.10406804084777832,
      "step": 2416
    },
    {
      "epoch": 0.326759611322349,
      "grad_norm": 0.2049816995859146,
      "learning_rate": 2.4686308172705433e-05,
      "loss": 0.03955411911010742,
      "step": 2417
    },
    {
      "epoch": 0.32689480354879596,
      "grad_norm": 0.18643677234649658,
      "learning_rate": 2.4681081131092926e-05,
      "loss": 0.054799556732177734,
      "step": 2418
    },
    {
      "epoch": 0.32702999577524294,
      "grad_norm": 0.29502829909324646,
      "learning_rate": 2.467585207382769e-05,
      "loss": 0.07020759582519531,
      "step": 2419
    },
    {
      "epoch": 0.3271651880016899,
      "grad_norm": 0.2715286612510681,
      "learning_rate": 2.4670621001998467e-05,
      "loss": 0.05517768859863281,
      "step": 2420
    },
    {
      "epoch": 0.3273003802281369,
      "grad_norm": 0.30594679713249207,
      "learning_rate": 2.466538791669437e-05,
      "loss": 0.06382042169570923,
      "step": 2421
    },
    {
      "epoch": 0.3274355724545839,
      "grad_norm": 0.4444860517978668,
      "learning_rate": 2.4660152819004973e-05,
      "loss": 0.08150219917297363,
      "step": 2422
    },
    {
      "epoch": 0.32757076468103086,
      "grad_norm": 0.4667988419532776,
      "learning_rate": 2.4654915710020246e-05,
      "loss": 0.05817985534667969,
      "step": 2423
    },
    {
      "epoch": 0.32770595690747784,
      "grad_norm": 0.42448729276657104,
      "learning_rate": 2.464967659083058e-05,
      "loss": 0.07563543319702148,
      "step": 2424
    },
    {
      "epoch": 0.3278411491339248,
      "grad_norm": 0.32104459404945374,
      "learning_rate": 2.464443546252679e-05,
      "loss": 0.06878280639648438,
      "step": 2425
    },
    {
      "epoch": 0.3279763413603718,
      "grad_norm": 0.3184334635734558,
      "learning_rate": 2.4639192326200104e-05,
      "loss": 0.07930433750152588,
      "step": 2426
    },
    {
      "epoch": 0.32811153358681877,
      "grad_norm": 0.1668136864900589,
      "learning_rate": 2.463394718294218e-05,
      "loss": 0.030389070510864258,
      "step": 2427
    },
    {
      "epoch": 0.32824672581326575,
      "grad_norm": 0.43874040246009827,
      "learning_rate": 2.4628700033845072e-05,
      "loss": 0.07626056671142578,
      "step": 2428
    },
    {
      "epoch": 0.32838191803971273,
      "grad_norm": 0.6310276985168457,
      "learning_rate": 2.4623450880001268e-05,
      "loss": 0.0951528549194336,
      "step": 2429
    },
    {
      "epoch": 0.3285171102661597,
      "grad_norm": 0.21510225534439087,
      "learning_rate": 2.4618199722503676e-05,
      "loss": 0.04945993423461914,
      "step": 2430
    },
    {
      "epoch": 0.3286523024926067,
      "grad_norm": 0.6062254905700684,
      "learning_rate": 2.4612946562445613e-05,
      "loss": 0.0645909309387207,
      "step": 2431
    },
    {
      "epoch": 0.32878749471905366,
      "grad_norm": 0.3155181407928467,
      "learning_rate": 2.460769140092081e-05,
      "loss": 0.07437801361083984,
      "step": 2432
    },
    {
      "epoch": 0.32892268694550064,
      "grad_norm": 0.2749468684196472,
      "learning_rate": 2.460243423902342e-05,
      "loss": 0.06017649173736572,
      "step": 2433
    },
    {
      "epoch": 0.3290578791719476,
      "grad_norm": 0.3019416034221649,
      "learning_rate": 2.459717507784802e-05,
      "loss": 0.07065987586975098,
      "step": 2434
    },
    {
      "epoch": 0.3291930713983946,
      "grad_norm": 0.4691430926322937,
      "learning_rate": 2.459191391848959e-05,
      "loss": 0.06434822082519531,
      "step": 2435
    },
    {
      "epoch": 0.3293282636248416,
      "grad_norm": 0.3791032135486603,
      "learning_rate": 2.4586650762043538e-05,
      "loss": 0.07087516784667969,
      "step": 2436
    },
    {
      "epoch": 0.32946345585128856,
      "grad_norm": 0.6622600555419922,
      "learning_rate": 2.4581385609605665e-05,
      "loss": 0.0833359956741333,
      "step": 2437
    },
    {
      "epoch": 0.32959864807773553,
      "grad_norm": 0.46688854694366455,
      "learning_rate": 2.4576118462272218e-05,
      "loss": 0.08251190185546875,
      "step": 2438
    },
    {
      "epoch": 0.3297338403041825,
      "grad_norm": 0.3251646161079407,
      "learning_rate": 2.4570849321139836e-05,
      "loss": 0.09423828125,
      "step": 2439
    },
    {
      "epoch": 0.3298690325306295,
      "grad_norm": 0.27186331152915955,
      "learning_rate": 2.4565578187305596e-05,
      "loss": 0.06013679504394531,
      "step": 2440
    },
    {
      "epoch": 0.33000422475707647,
      "grad_norm": 0.23624005913734436,
      "learning_rate": 2.456030506186696e-05,
      "loss": 0.05769085884094238,
      "step": 2441
    },
    {
      "epoch": 0.33013941698352345,
      "grad_norm": 0.20653310418128967,
      "learning_rate": 2.4555029945921832e-05,
      "loss": 0.049443721771240234,
      "step": 2442
    },
    {
      "epoch": 0.3302746092099704,
      "grad_norm": 0.3757876455783844,
      "learning_rate": 2.4549752840568516e-05,
      "loss": 0.0664370059967041,
      "step": 2443
    },
    {
      "epoch": 0.3304098014364174,
      "grad_norm": 0.3367663621902466,
      "learning_rate": 2.4544473746905733e-05,
      "loss": 0.07474899291992188,
      "step": 2444
    },
    {
      "epoch": 0.3305449936628644,
      "grad_norm": 0.23202677071094513,
      "learning_rate": 2.4539192666032617e-05,
      "loss": 0.06258010864257812,
      "step": 2445
    },
    {
      "epoch": 0.33068018588931136,
      "grad_norm": 0.1665327548980713,
      "learning_rate": 2.4533909599048718e-05,
      "loss": 0.03667253255844116,
      "step": 2446
    },
    {
      "epoch": 0.33081537811575834,
      "grad_norm": 0.42714065313339233,
      "learning_rate": 2.4528624547054003e-05,
      "loss": 0.058437228202819824,
      "step": 2447
    },
    {
      "epoch": 0.3309505703422053,
      "grad_norm": 0.4131702482700348,
      "learning_rate": 2.4523337511148843e-05,
      "loss": 0.058091163635253906,
      "step": 2448
    },
    {
      "epoch": 0.3310857625686523,
      "grad_norm": 0.5783988833427429,
      "learning_rate": 2.4518048492434028e-05,
      "loss": 0.0949258804321289,
      "step": 2449
    },
    {
      "epoch": 0.3312209547950993,
      "grad_norm": 0.24023276567459106,
      "learning_rate": 2.4512757492010762e-05,
      "loss": 0.0540919303894043,
      "step": 2450
    },
    {
      "epoch": 0.33135614702154625,
      "grad_norm": 0.24015502631664276,
      "learning_rate": 2.4507464510980652e-05,
      "loss": 0.07108378410339355,
      "step": 2451
    },
    {
      "epoch": 0.33149133924799323,
      "grad_norm": 0.2364378720521927,
      "learning_rate": 2.450216955044574e-05,
      "loss": 0.06290721893310547,
      "step": 2452
    },
    {
      "epoch": 0.3316265314744402,
      "grad_norm": 0.5855490565299988,
      "learning_rate": 2.449687261150845e-05,
      "loss": 0.10780525207519531,
      "step": 2453
    },
    {
      "epoch": 0.3317617237008872,
      "grad_norm": 0.6197882294654846,
      "learning_rate": 2.449157369527164e-05,
      "loss": 0.09061050415039062,
      "step": 2454
    },
    {
      "epoch": 0.33189691592733417,
      "grad_norm": 0.33625009655952454,
      "learning_rate": 2.448627280283857e-05,
      "loss": 0.06360673904418945,
      "step": 2455
    },
    {
      "epoch": 0.33203210815378115,
      "grad_norm": 0.3676844537258148,
      "learning_rate": 2.4480969935312917e-05,
      "loss": 0.07337427139282227,
      "step": 2456
    },
    {
      "epoch": 0.3321673003802281,
      "grad_norm": 0.20737534761428833,
      "learning_rate": 2.4475665093798766e-05,
      "loss": 0.059857845306396484,
      "step": 2457
    },
    {
      "epoch": 0.3323024926066751,
      "grad_norm": 0.6505117416381836,
      "learning_rate": 2.447035827940061e-05,
      "loss": 0.07700157165527344,
      "step": 2458
    },
    {
      "epoch": 0.3324376848331221,
      "grad_norm": 0.4976421594619751,
      "learning_rate": 2.4465049493223356e-05,
      "loss": 0.0570683479309082,
      "step": 2459
    },
    {
      "epoch": 0.33257287705956906,
      "grad_norm": 0.3306278884410858,
      "learning_rate": 2.4459738736372327e-05,
      "loss": 0.0829916000366211,
      "step": 2460
    },
    {
      "epoch": 0.33270806928601604,
      "grad_norm": 0.3618042469024658,
      "learning_rate": 2.4454426009953252e-05,
      "loss": 0.04579496383666992,
      "step": 2461
    },
    {
      "epoch": 0.332843261512463,
      "grad_norm": 0.31327348947525024,
      "learning_rate": 2.4449111315072254e-05,
      "loss": 0.07779788970947266,
      "step": 2462
    },
    {
      "epoch": 0.33297845373891,
      "grad_norm": 0.5885433554649353,
      "learning_rate": 2.44437946528359e-05,
      "loss": 0.09245491027832031,
      "step": 2463
    },
    {
      "epoch": 0.333113645965357,
      "grad_norm": 0.38821515440940857,
      "learning_rate": 2.4438476024351138e-05,
      "loss": 0.06781506538391113,
      "step": 2464
    },
    {
      "epoch": 0.33324883819180395,
      "grad_norm": 0.3372174799442291,
      "learning_rate": 2.4433155430725333e-05,
      "loss": 0.07224130630493164,
      "step": 2465
    },
    {
      "epoch": 0.33338403041825093,
      "grad_norm": 0.34986966848373413,
      "learning_rate": 2.4427832873066262e-05,
      "loss": 0.07378792762756348,
      "step": 2466
    },
    {
      "epoch": 0.3335192226446979,
      "grad_norm": 0.291404664516449,
      "learning_rate": 2.4422508352482113e-05,
      "loss": 0.05615878105163574,
      "step": 2467
    },
    {
      "epoch": 0.3336544148711449,
      "grad_norm": 0.4609013795852661,
      "learning_rate": 2.441718187008148e-05,
      "loss": 0.06363534927368164,
      "step": 2468
    },
    {
      "epoch": 0.33378960709759187,
      "grad_norm": 0.18457138538360596,
      "learning_rate": 2.441185342697336e-05,
      "loss": 0.04251909255981445,
      "step": 2469
    },
    {
      "epoch": 0.33392479932403885,
      "grad_norm": 0.2803070545196533,
      "learning_rate": 2.440652302426717e-05,
      "loss": 0.055945396423339844,
      "step": 2470
    },
    {
      "epoch": 0.3340599915504858,
      "grad_norm": 0.45884576439857483,
      "learning_rate": 2.440119066307272e-05,
      "loss": 0.08475303649902344,
      "step": 2471
    },
    {
      "epoch": 0.3341951837769328,
      "grad_norm": 0.28993961215019226,
      "learning_rate": 2.4395856344500244e-05,
      "loss": 0.07215166091918945,
      "step": 2472
    },
    {
      "epoch": 0.3343303760033798,
      "grad_norm": 0.4510860741138458,
      "learning_rate": 2.4390520069660377e-05,
      "loss": 0.07471084594726562,
      "step": 2473
    },
    {
      "epoch": 0.33446556822982676,
      "grad_norm": 0.33531898260116577,
      "learning_rate": 2.4385181839664146e-05,
      "loss": 0.06544113159179688,
      "step": 2474
    },
    {
      "epoch": 0.33460076045627374,
      "grad_norm": 0.4310208559036255,
      "learning_rate": 2.437984165562301e-05,
      "loss": 0.09584617614746094,
      "step": 2475
    },
    {
      "epoch": 0.3347359526827207,
      "grad_norm": 0.33662885427474976,
      "learning_rate": 2.4374499518648827e-05,
      "loss": 0.052356719970703125,
      "step": 2476
    },
    {
      "epoch": 0.33487114490916775,
      "grad_norm": 0.2527439296245575,
      "learning_rate": 2.436915542985385e-05,
      "loss": 0.05551505088806152,
      "step": 2477
    },
    {
      "epoch": 0.33500633713561473,
      "grad_norm": 0.3310180604457855,
      "learning_rate": 2.436380939035075e-05,
      "loss": 0.08473348617553711,
      "step": 2478
    },
    {
      "epoch": 0.3351415293620617,
      "grad_norm": 0.2368757724761963,
      "learning_rate": 2.43584614012526e-05,
      "loss": 0.042256832122802734,
      "step": 2479
    },
    {
      "epoch": 0.3352767215885087,
      "grad_norm": 0.17077018320560455,
      "learning_rate": 2.4353111463672882e-05,
      "loss": 0.03796041011810303,
      "step": 2480
    },
    {
      "epoch": 0.33541191381495566,
      "grad_norm": 0.5071596503257751,
      "learning_rate": 2.4347759578725482e-05,
      "loss": 0.08613014221191406,
      "step": 2481
    },
    {
      "epoch": 0.33554710604140264,
      "grad_norm": 0.4547046720981598,
      "learning_rate": 2.4342405747524685e-05,
      "loss": 0.08476781845092773,
      "step": 2482
    },
    {
      "epoch": 0.3356822982678496,
      "grad_norm": 0.46177658438682556,
      "learning_rate": 2.4337049971185194e-05,
      "loss": 0.1083369255065918,
      "step": 2483
    },
    {
      "epoch": 0.3358174904942966,
      "grad_norm": 0.36410853266716003,
      "learning_rate": 2.433169225082211e-05,
      "loss": 0.07090163230895996,
      "step": 2484
    },
    {
      "epoch": 0.3359526827207436,
      "grad_norm": 0.20836199820041656,
      "learning_rate": 2.432633258755093e-05,
      "loss": 0.05080842971801758,
      "step": 2485
    },
    {
      "epoch": 0.33608787494719056,
      "grad_norm": 0.3155006170272827,
      "learning_rate": 2.432097098248758e-05,
      "loss": 0.06565451622009277,
      "step": 2486
    },
    {
      "epoch": 0.33622306717363754,
      "grad_norm": 0.40902048349380493,
      "learning_rate": 2.4315607436748362e-05,
      "loss": 0.0543973445892334,
      "step": 2487
    },
    {
      "epoch": 0.3363582594000845,
      "grad_norm": 0.6134740710258484,
      "learning_rate": 2.4310241951449997e-05,
      "loss": 0.060318946838378906,
      "step": 2488
    },
    {
      "epoch": 0.3364934516265315,
      "grad_norm": 0.3207819163799286,
      "learning_rate": 2.4304874527709614e-05,
      "loss": 0.06946825981140137,
      "step": 2489
    },
    {
      "epoch": 0.33662864385297847,
      "grad_norm": 0.31709375977516174,
      "learning_rate": 2.429950516664473e-05,
      "loss": 0.06220436096191406,
      "step": 2490
    },
    {
      "epoch": 0.33676383607942545,
      "grad_norm": 0.24454593658447266,
      "learning_rate": 2.4294133869373284e-05,
      "loss": 0.05053210258483887,
      "step": 2491
    },
    {
      "epoch": 0.33689902830587243,
      "grad_norm": 0.4360541105270386,
      "learning_rate": 2.42887606370136e-05,
      "loss": 0.06767749786376953,
      "step": 2492
    },
    {
      "epoch": 0.3370342205323194,
      "grad_norm": 0.25928908586502075,
      "learning_rate": 2.428338547068442e-05,
      "loss": 0.05216515064239502,
      "step": 2493
    },
    {
      "epoch": 0.3371694127587664,
      "grad_norm": 0.9997990727424622,
      "learning_rate": 2.427800837150488e-05,
      "loss": 0.07999038696289062,
      "step": 2494
    },
    {
      "epoch": 0.33730460498521336,
      "grad_norm": 0.4855858385562897,
      "learning_rate": 2.4272629340594518e-05,
      "loss": 0.0854107141494751,
      "step": 2495
    },
    {
      "epoch": 0.33743979721166034,
      "grad_norm": 0.3172101676464081,
      "learning_rate": 2.426724837907328e-05,
      "loss": 0.06275415420532227,
      "step": 2496
    },
    {
      "epoch": 0.3375749894381073,
      "grad_norm": 0.45540010929107666,
      "learning_rate": 2.4261865488061512e-05,
      "loss": 0.06249189376831055,
      "step": 2497
    },
    {
      "epoch": 0.3377101816645543,
      "grad_norm": 0.5258225202560425,
      "learning_rate": 2.4256480668679958e-05,
      "loss": 0.0672459602355957,
      "step": 2498
    },
    {
      "epoch": 0.3378453738910013,
      "grad_norm": 0.22932447493076324,
      "learning_rate": 2.4251093922049766e-05,
      "loss": 0.07059991359710693,
      "step": 2499
    },
    {
      "epoch": 0.33798056611744826,
      "grad_norm": 0.2417266070842743,
      "learning_rate": 2.4245705249292494e-05,
      "loss": 0.05587363243103027,
      "step": 2500
    },
    {
      "epoch": 0.33811575834389523,
      "grad_norm": 0.5009039044380188,
      "learning_rate": 2.4240314651530073e-05,
      "loss": 0.0823812484741211,
      "step": 2501
    },
    {
      "epoch": 0.3382509505703422,
      "grad_norm": 0.4212474524974823,
      "learning_rate": 2.4234922129884873e-05,
      "loss": 0.06021285057067871,
      "step": 2502
    },
    {
      "epoch": 0.3383861427967892,
      "grad_norm": 0.5608205199241638,
      "learning_rate": 2.4229527685479644e-05,
      "loss": 0.08831501007080078,
      "step": 2503
    },
    {
      "epoch": 0.33852133502323617,
      "grad_norm": 0.28817611932754517,
      "learning_rate": 2.4224131319437523e-05,
      "loss": 0.05916786193847656,
      "step": 2504
    },
    {
      "epoch": 0.33865652724968315,
      "grad_norm": 0.37260687351226807,
      "learning_rate": 2.421873303288208e-05,
      "loss": 0.07591724395751953,
      "step": 2505
    },
    {
      "epoch": 0.3387917194761301,
      "grad_norm": 0.7916498780250549,
      "learning_rate": 2.4213332826937255e-05,
      "loss": 0.10970115661621094,
      "step": 2506
    },
    {
      "epoch": 0.3389269117025771,
      "grad_norm": 0.17081116139888763,
      "learning_rate": 2.4207930702727404e-05,
      "loss": 0.05327153205871582,
      "step": 2507
    },
    {
      "epoch": 0.3390621039290241,
      "grad_norm": 0.3214297592639923,
      "learning_rate": 2.420252666137728e-05,
      "loss": 0.05479860305786133,
      "step": 2508
    },
    {
      "epoch": 0.33919729615547106,
      "grad_norm": 0.23765519261360168,
      "learning_rate": 2.419712070401203e-05,
      "loss": 0.06476879119873047,
      "step": 2509
    },
    {
      "epoch": 0.33933248838191804,
      "grad_norm": 0.493564635515213,
      "learning_rate": 2.4191712831757203e-05,
      "loss": 0.08979177474975586,
      "step": 2510
    },
    {
      "epoch": 0.339467680608365,
      "grad_norm": 0.35108911991119385,
      "learning_rate": 2.418630304573875e-05,
      "loss": 0.06929492950439453,
      "step": 2511
    },
    {
      "epoch": 0.339602872834812,
      "grad_norm": 0.2620317339897156,
      "learning_rate": 2.418089134708302e-05,
      "loss": 0.06982278823852539,
      "step": 2512
    },
    {
      "epoch": 0.339738065061259,
      "grad_norm": 0.3687685430049896,
      "learning_rate": 2.4175477736916743e-05,
      "loss": 0.10542488098144531,
      "step": 2513
    },
    {
      "epoch": 0.33987325728770595,
      "grad_norm": 0.7251862287521362,
      "learning_rate": 2.4170062216367082e-05,
      "loss": 0.08716773986816406,
      "step": 2514
    },
    {
      "epoch": 0.34000844951415293,
      "grad_norm": 0.7256414890289307,
      "learning_rate": 2.416464478656156e-05,
      "loss": 0.08104228973388672,
      "step": 2515
    },
    {
      "epoch": 0.3401436417405999,
      "grad_norm": 0.3080953061580658,
      "learning_rate": 2.4159225448628123e-05,
      "loss": 0.058271169662475586,
      "step": 2516
    },
    {
      "epoch": 0.3402788339670469,
      "grad_norm": 0.5783313512802124,
      "learning_rate": 2.4153804203695103e-05,
      "loss": 0.08752673864364624,
      "step": 2517
    },
    {
      "epoch": 0.34041402619349387,
      "grad_norm": 0.6522753834724426,
      "learning_rate": 2.4148381052891236e-05,
      "loss": 0.06243085861206055,
      "step": 2518
    },
    {
      "epoch": 0.34054921841994085,
      "grad_norm": 0.30998751521110535,
      "learning_rate": 2.4142955997345648e-05,
      "loss": 0.0700232982635498,
      "step": 2519
    },
    {
      "epoch": 0.3406844106463878,
      "grad_norm": 0.21559259295463562,
      "learning_rate": 2.4137529038187864e-05,
      "loss": 0.06367969512939453,
      "step": 2520
    },
    {
      "epoch": 0.3408196028728348,
      "grad_norm": 0.35696348547935486,
      "learning_rate": 2.413210017654781e-05,
      "loss": 0.05223369598388672,
      "step": 2521
    },
    {
      "epoch": 0.3409547950992818,
      "grad_norm": 0.45169007778167725,
      "learning_rate": 2.4126669413555802e-05,
      "loss": 0.08095884323120117,
      "step": 2522
    },
    {
      "epoch": 0.34108998732572876,
      "grad_norm": 0.6773762106895447,
      "learning_rate": 2.4121236750342548e-05,
      "loss": 0.0945596694946289,
      "step": 2523
    },
    {
      "epoch": 0.34122517955217574,
      "grad_norm": 0.2786770462989807,
      "learning_rate": 2.4115802188039165e-05,
      "loss": 0.07101106643676758,
      "step": 2524
    },
    {
      "epoch": 0.3413603717786227,
      "grad_norm": 0.3666849434375763,
      "learning_rate": 2.4110365727777156e-05,
      "loss": 0.06577682495117188,
      "step": 2525
    },
    {
      "epoch": 0.3414955640050697,
      "grad_norm": 0.3651229739189148,
      "learning_rate": 2.410492737068842e-05,
      "loss": 0.06389713287353516,
      "step": 2526
    },
    {
      "epoch": 0.3416307562315167,
      "grad_norm": 0.3447984755039215,
      "learning_rate": 2.409948711790525e-05,
      "loss": 0.09207653999328613,
      "step": 2527
    },
    {
      "epoch": 0.34176594845796365,
      "grad_norm": 0.5862815380096436,
      "learning_rate": 2.4094044970560336e-05,
      "loss": 0.09639883041381836,
      "step": 2528
    },
    {
      "epoch": 0.34190114068441063,
      "grad_norm": 0.3143575191497803,
      "learning_rate": 2.4088600929786767e-05,
      "loss": 0.06098484992980957,
      "step": 2529
    },
    {
      "epoch": 0.3420363329108576,
      "grad_norm": 0.5567917823791504,
      "learning_rate": 2.408315499671802e-05,
      "loss": 0.07166743278503418,
      "step": 2530
    },
    {
      "epoch": 0.3421715251373046,
      "grad_norm": 0.24091428518295288,
      "learning_rate": 2.407770717248796e-05,
      "loss": 0.05601072311401367,
      "step": 2531
    },
    {
      "epoch": 0.34230671736375157,
      "grad_norm": 0.4519946277141571,
      "learning_rate": 2.407225745823086e-05,
      "loss": 0.07663154602050781,
      "step": 2532
    },
    {
      "epoch": 0.34244190959019855,
      "grad_norm": 0.43207940459251404,
      "learning_rate": 2.4066805855081378e-05,
      "loss": 0.06264829635620117,
      "step": 2533
    },
    {
      "epoch": 0.3425771018166455,
      "grad_norm": 0.4547562897205353,
      "learning_rate": 2.406135236417457e-05,
      "loss": 0.10232162475585938,
      "step": 2534
    },
    {
      "epoch": 0.3427122940430925,
      "grad_norm": 0.5358670949935913,
      "learning_rate": 2.4055896986645875e-05,
      "loss": 0.08292484283447266,
      "step": 2535
    },
    {
      "epoch": 0.3428474862695395,
      "grad_norm": 0.1853201687335968,
      "learning_rate": 2.4050439723631136e-05,
      "loss": 0.050823330879211426,
      "step": 2536
    },
    {
      "epoch": 0.34298267849598646,
      "grad_norm": 0.19585396349430084,
      "learning_rate": 2.404498057626659e-05,
      "loss": 0.05597734451293945,
      "step": 2537
    },
    {
      "epoch": 0.34311787072243344,
      "grad_norm": 0.33710625767707825,
      "learning_rate": 2.4039519545688848e-05,
      "loss": 0.08285927772521973,
      "step": 2538
    },
    {
      "epoch": 0.3432530629488804,
      "grad_norm": 0.7918868660926819,
      "learning_rate": 2.4034056633034932e-05,
      "loss": 0.08617329597473145,
      "step": 2539
    },
    {
      "epoch": 0.3433882551753274,
      "grad_norm": 0.3767209053039551,
      "learning_rate": 2.402859183944225e-05,
      "loss": 0.08967018127441406,
      "step": 2540
    },
    {
      "epoch": 0.3435234474017744,
      "grad_norm": 0.26449450850486755,
      "learning_rate": 2.4023125166048597e-05,
      "loss": 0.056416988372802734,
      "step": 2541
    },
    {
      "epoch": 0.34365863962822135,
      "grad_norm": 0.35602036118507385,
      "learning_rate": 2.401765661399218e-05,
      "loss": 0.06450843811035156,
      "step": 2542
    },
    {
      "epoch": 0.34379383185466833,
      "grad_norm": 0.4066449999809265,
      "learning_rate": 2.4012186184411556e-05,
      "loss": 0.08829712867736816,
      "step": 2543
    },
    {
      "epoch": 0.3439290240811153,
      "grad_norm": 0.3011881113052368,
      "learning_rate": 2.400671387844571e-05,
      "loss": 0.06794929504394531,
      "step": 2544
    },
    {
      "epoch": 0.34406421630756234,
      "grad_norm": 0.43036797642707825,
      "learning_rate": 2.4001239697234008e-05,
      "loss": 0.0652916431427002,
      "step": 2545
    },
    {
      "epoch": 0.3441994085340093,
      "grad_norm": 0.1733134686946869,
      "learning_rate": 2.3995763641916205e-05,
      "loss": 0.040465593338012695,
      "step": 2546
    },
    {
      "epoch": 0.3443346007604563,
      "grad_norm": 0.28372931480407715,
      "learning_rate": 2.3990285713632436e-05,
      "loss": 0.06408858299255371,
      "step": 2547
    },
    {
      "epoch": 0.3444697929869033,
      "grad_norm": 0.31761035323143005,
      "learning_rate": 2.398480591352324e-05,
      "loss": 0.060512423515319824,
      "step": 2548
    },
    {
      "epoch": 0.34460498521335026,
      "grad_norm": 0.1383335292339325,
      "learning_rate": 2.3979324242729537e-05,
      "loss": 0.043040931224823,
      "step": 2549
    },
    {
      "epoch": 0.34474017743979724,
      "grad_norm": 0.26285022497177124,
      "learning_rate": 2.3973840702392646e-05,
      "loss": 0.07085132598876953,
      "step": 2550
    },
    {
      "epoch": 0.3448753696662442,
      "grad_norm": 0.12210826575756073,
      "learning_rate": 2.3968355293654267e-05,
      "loss": 0.0397639274597168,
      "step": 2551
    },
    {
      "epoch": 0.3450105618926912,
      "grad_norm": 0.18168649077415466,
      "learning_rate": 2.396286801765649e-05,
      "loss": 0.047493934631347656,
      "step": 2552
    },
    {
      "epoch": 0.34514575411913817,
      "grad_norm": 0.31178680062294006,
      "learning_rate": 2.3957378875541795e-05,
      "loss": 0.06428849697113037,
      "step": 2553
    },
    {
      "epoch": 0.34528094634558515,
      "grad_norm": 0.32699131965637207,
      "learning_rate": 2.395188786845305e-05,
      "loss": 0.0743408203125,
      "step": 2554
    },
    {
      "epoch": 0.3454161385720321,
      "grad_norm": 0.23083823919296265,
      "learning_rate": 2.3946394997533516e-05,
      "loss": 0.06639689207077026,
      "step": 2555
    },
    {
      "epoch": 0.3455513307984791,
      "grad_norm": 0.38130173087120056,
      "learning_rate": 2.3940900263926833e-05,
      "loss": 0.06683635711669922,
      "step": 2556
    },
    {
      "epoch": 0.3456865230249261,
      "grad_norm": 0.43939515948295593,
      "learning_rate": 2.393540366877704e-05,
      "loss": 0.09607124328613281,
      "step": 2557
    },
    {
      "epoch": 0.34582171525137306,
      "grad_norm": 0.6057755947113037,
      "learning_rate": 2.392990521322855e-05,
      "loss": 0.06909465789794922,
      "step": 2558
    },
    {
      "epoch": 0.34595690747782004,
      "grad_norm": 0.6586816906929016,
      "learning_rate": 2.392440489842618e-05,
      "loss": 0.09536075592041016,
      "step": 2559
    },
    {
      "epoch": 0.346092099704267,
      "grad_norm": 0.3285783529281616,
      "learning_rate": 2.3918902725515118e-05,
      "loss": 0.0948324203491211,
      "step": 2560
    },
    {
      "epoch": 0.346227291930714,
      "grad_norm": 0.2755489945411682,
      "learning_rate": 2.391339869564094e-05,
      "loss": 0.07570266723632812,
      "step": 2561
    },
    {
      "epoch": 0.346362484157161,
      "grad_norm": 0.25667473673820496,
      "learning_rate": 2.3907892809949628e-05,
      "loss": 0.038190245628356934,
      "step": 2562
    },
    {
      "epoch": 0.34649767638360796,
      "grad_norm": 0.2906736731529236,
      "learning_rate": 2.390238506958753e-05,
      "loss": 0.06618642807006836,
      "step": 2563
    },
    {
      "epoch": 0.34663286861005493,
      "grad_norm": 0.5905830264091492,
      "learning_rate": 2.3896875475701387e-05,
      "loss": 0.09708178043365479,
      "step": 2564
    },
    {
      "epoch": 0.3467680608365019,
      "grad_norm": 0.2868516743183136,
      "learning_rate": 2.3891364029438323e-05,
      "loss": 0.0561293363571167,
      "step": 2565
    },
    {
      "epoch": 0.3469032530629489,
      "grad_norm": 0.29824212193489075,
      "learning_rate": 2.3885850731945857e-05,
      "loss": 0.06819438934326172,
      "step": 2566
    },
    {
      "epoch": 0.34703844528939587,
      "grad_norm": 0.49924513697624207,
      "learning_rate": 2.3880335584371884e-05,
      "loss": 0.0761265754699707,
      "step": 2567
    },
    {
      "epoch": 0.34717363751584285,
      "grad_norm": 0.5289023518562317,
      "learning_rate": 2.387481858786468e-05,
      "loss": 0.08414745330810547,
      "step": 2568
    },
    {
      "epoch": 0.3473088297422898,
      "grad_norm": 0.3754660487174988,
      "learning_rate": 2.386929974357293e-05,
      "loss": 0.05096358060836792,
      "step": 2569
    },
    {
      "epoch": 0.3474440219687368,
      "grad_norm": 0.29868337512016296,
      "learning_rate": 2.386377905264567e-05,
      "loss": 0.0711207389831543,
      "step": 2570
    },
    {
      "epoch": 0.3475792141951838,
      "grad_norm": 0.4866524040699005,
      "learning_rate": 2.3858256516232346e-05,
      "loss": 0.10391855239868164,
      "step": 2571
    },
    {
      "epoch": 0.34771440642163076,
      "grad_norm": 0.2132396697998047,
      "learning_rate": 2.3852732135482775e-05,
      "loss": 0.060329437255859375,
      "step": 2572
    },
    {
      "epoch": 0.34784959864807774,
      "grad_norm": 0.3209385573863983,
      "learning_rate": 2.3847205911547166e-05,
      "loss": 0.05642986297607422,
      "step": 2573
    },
    {
      "epoch": 0.3479847908745247,
      "grad_norm": 0.29649993777275085,
      "learning_rate": 2.3841677845576108e-05,
      "loss": 0.037259817123413086,
      "step": 2574
    },
    {
      "epoch": 0.3481199831009717,
      "grad_norm": 0.4815114140510559,
      "learning_rate": 2.383614793872057e-05,
      "loss": 0.10496234893798828,
      "step": 2575
    },
    {
      "epoch": 0.3482551753274187,
      "grad_norm": 0.2752836346626282,
      "learning_rate": 2.3830616192131913e-05,
      "loss": 0.07256031036376953,
      "step": 2576
    },
    {
      "epoch": 0.34839036755386565,
      "grad_norm": 0.21755489706993103,
      "learning_rate": 2.3825082606961876e-05,
      "loss": 0.06976318359375,
      "step": 2577
    },
    {
      "epoch": 0.34852555978031263,
      "grad_norm": 0.21390248835086823,
      "learning_rate": 2.3819547184362575e-05,
      "loss": 0.040138840675354004,
      "step": 2578
    },
    {
      "epoch": 0.3486607520067596,
      "grad_norm": 0.40627220273017883,
      "learning_rate": 2.3814009925486522e-05,
      "loss": 0.0750131607055664,
      "step": 2579
    },
    {
      "epoch": 0.3487959442332066,
      "grad_norm": 0.22322294116020203,
      "learning_rate": 2.38084708314866e-05,
      "loss": 0.067047119140625,
      "step": 2580
    },
    {
      "epoch": 0.34893113645965357,
      "grad_norm": 0.31564804911613464,
      "learning_rate": 2.380292990351608e-05,
      "loss": 0.08023738861083984,
      "step": 2581
    },
    {
      "epoch": 0.34906632868610055,
      "grad_norm": 0.29696330428123474,
      "learning_rate": 2.3797387142728607e-05,
      "loss": 0.06979990005493164,
      "step": 2582
    },
    {
      "epoch": 0.3492015209125475,
      "grad_norm": 0.2819051146507263,
      "learning_rate": 2.379184255027822e-05,
      "loss": 0.0637967586517334,
      "step": 2583
    },
    {
      "epoch": 0.3493367131389945,
      "grad_norm": 0.40146055817604065,
      "learning_rate": 2.378629612731933e-05,
      "loss": 0.08315134048461914,
      "step": 2584
    },
    {
      "epoch": 0.3494719053654415,
      "grad_norm": 0.31889259815216064,
      "learning_rate": 2.3780747875006735e-05,
      "loss": 0.05870974063873291,
      "step": 2585
    },
    {
      "epoch": 0.34960709759188846,
      "grad_norm": 0.3140102028846741,
      "learning_rate": 2.37751977944956e-05,
      "loss": 0.0748147964477539,
      "step": 2586
    },
    {
      "epoch": 0.34974228981833544,
      "grad_norm": 0.37359362840652466,
      "learning_rate": 2.3769645886941497e-05,
      "loss": 0.07153606414794922,
      "step": 2587
    },
    {
      "epoch": 0.3498774820447824,
      "grad_norm": 0.3564010262489319,
      "learning_rate": 2.376409215350035e-05,
      "loss": 0.0818319320678711,
      "step": 2588
    },
    {
      "epoch": 0.3500126742712294,
      "grad_norm": 0.3945556581020355,
      "learning_rate": 2.3758536595328486e-05,
      "loss": 0.07982826232910156,
      "step": 2589
    },
    {
      "epoch": 0.3501478664976764,
      "grad_norm": 0.4441947638988495,
      "learning_rate": 2.375297921358259e-05,
      "loss": 0.06662988662719727,
      "step": 2590
    },
    {
      "epoch": 0.35028305872412335,
      "grad_norm": 0.2367473840713501,
      "learning_rate": 2.3747420009419745e-05,
      "loss": 0.043430328369140625,
      "step": 2591
    },
    {
      "epoch": 0.35041825095057033,
      "grad_norm": 0.4292266368865967,
      "learning_rate": 2.3741858983997415e-05,
      "loss": 0.07572412490844727,
      "step": 2592
    },
    {
      "epoch": 0.3505534431770173,
      "grad_norm": 0.31039565801620483,
      "learning_rate": 2.373629613847342e-05,
      "loss": 0.053858041763305664,
      "step": 2593
    },
    {
      "epoch": 0.3506886354034643,
      "grad_norm": 0.5356007218360901,
      "learning_rate": 2.3730731474005988e-05,
      "loss": 0.0842585563659668,
      "step": 2594
    },
    {
      "epoch": 0.35082382762991127,
      "grad_norm": 0.10539377480745316,
      "learning_rate": 2.37251649917537e-05,
      "loss": 0.027021288871765137,
      "step": 2595
    },
    {
      "epoch": 0.35095901985635825,
      "grad_norm": 0.18666324019432068,
      "learning_rate": 2.3719596692875534e-05,
      "loss": 0.05283641815185547,
      "step": 2596
    },
    {
      "epoch": 0.3510942120828052,
      "grad_norm": 0.6938058137893677,
      "learning_rate": 2.3714026578530836e-05,
      "loss": 0.08938217163085938,
      "step": 2597
    },
    {
      "epoch": 0.3512294043092522,
      "grad_norm": 0.29282745718955994,
      "learning_rate": 2.370845464987934e-05,
      "loss": 0.054965972900390625,
      "step": 2598
    },
    {
      "epoch": 0.3513645965356992,
      "grad_norm": 0.17689616978168488,
      "learning_rate": 2.370288090808114e-05,
      "loss": 0.03647947311401367,
      "step": 2599
    },
    {
      "epoch": 0.35149978876214616,
      "grad_norm": 0.35268816351890564,
      "learning_rate": 2.369730535429673e-05,
      "loss": 0.053194522857666016,
      "step": 2600
    },
    {
      "epoch": 0.35163498098859314,
      "grad_norm": 0.5388017892837524,
      "learning_rate": 2.369172798968697e-05,
      "loss": 0.06907200813293457,
      "step": 2601
    },
    {
      "epoch": 0.3517701732150401,
      "grad_norm": 0.25129905343055725,
      "learning_rate": 2.3686148815413083e-05,
      "loss": 0.07374000549316406,
      "step": 2602
    },
    {
      "epoch": 0.3519053654414871,
      "grad_norm": 0.5436261892318726,
      "learning_rate": 2.3680567832636695e-05,
      "loss": 0.10884809494018555,
      "step": 2603
    },
    {
      "epoch": 0.3520405576679341,
      "grad_norm": 0.5323015451431274,
      "learning_rate": 2.3674985042519795e-05,
      "loss": 0.10357809066772461,
      "step": 2604
    },
    {
      "epoch": 0.35217574989438105,
      "grad_norm": 0.3448658883571625,
      "learning_rate": 2.366940044622475e-05,
      "loss": 0.07070350646972656,
      "step": 2605
    },
    {
      "epoch": 0.35231094212082803,
      "grad_norm": 2.245497941970825,
      "learning_rate": 2.3663814044914302e-05,
      "loss": 0.0849158763885498,
      "step": 2606
    },
    {
      "epoch": 0.352446134347275,
      "grad_norm": 0.4565022587776184,
      "learning_rate": 2.3658225839751566e-05,
      "loss": 0.05966758728027344,
      "step": 2607
    },
    {
      "epoch": 0.352581326573722,
      "grad_norm": 0.25623318552970886,
      "learning_rate": 2.3652635831900043e-05,
      "loss": 0.05603742599487305,
      "step": 2608
    },
    {
      "epoch": 0.35271651880016897,
      "grad_norm": 0.7350065112113953,
      "learning_rate": 2.3647044022523595e-05,
      "loss": 0.09103775024414062,
      "step": 2609
    },
    {
      "epoch": 0.35285171102661594,
      "grad_norm": 0.3453652262687683,
      "learning_rate": 2.364145041278647e-05,
      "loss": 0.09230566024780273,
      "step": 2610
    },
    {
      "epoch": 0.3529869032530629,
      "grad_norm": 0.28571489453315735,
      "learning_rate": 2.3635855003853287e-05,
      "loss": 0.06862258911132812,
      "step": 2611
    },
    {
      "epoch": 0.3531220954795099,
      "grad_norm": 0.38044893741607666,
      "learning_rate": 2.363025779688904e-05,
      "loss": 0.0712437629699707,
      "step": 2612
    },
    {
      "epoch": 0.35325728770595693,
      "grad_norm": 0.3779444396495819,
      "learning_rate": 2.3624658793059103e-05,
      "loss": 0.058063507080078125,
      "step": 2613
    },
    {
      "epoch": 0.3533924799324039,
      "grad_norm": 0.42958560585975647,
      "learning_rate": 2.3619057993529204e-05,
      "loss": 0.05938148498535156,
      "step": 2614
    },
    {
      "epoch": 0.3535276721588509,
      "grad_norm": 0.3385521471500397,
      "learning_rate": 2.3613455399465475e-05,
      "loss": 0.07525253295898438,
      "step": 2615
    },
    {
      "epoch": 0.35366286438529787,
      "grad_norm": 0.35580816864967346,
      "learning_rate": 2.3607851012034394e-05,
      "loss": 0.09200477600097656,
      "step": 2616
    },
    {
      "epoch": 0.35379805661174485,
      "grad_norm": 0.4908611476421356,
      "learning_rate": 2.3602244832402838e-05,
      "loss": 0.08040714263916016,
      "step": 2617
    },
    {
      "epoch": 0.3539332488381918,
      "grad_norm": 0.4449275732040405,
      "learning_rate": 2.3596636861738024e-05,
      "loss": 0.08589982986450195,
      "step": 2618
    },
    {
      "epoch": 0.3540684410646388,
      "grad_norm": 0.22156795859336853,
      "learning_rate": 2.3591027101207578e-05,
      "loss": 0.062267184257507324,
      "step": 2619
    },
    {
      "epoch": 0.3542036332910858,
      "grad_norm": 0.4815278947353363,
      "learning_rate": 2.3585415551979476e-05,
      "loss": 0.06960344314575195,
      "step": 2620
    },
    {
      "epoch": 0.35433882551753276,
      "grad_norm": 0.24333488941192627,
      "learning_rate": 2.3579802215222076e-05,
      "loss": 0.047625064849853516,
      "step": 2621
    },
    {
      "epoch": 0.35447401774397974,
      "grad_norm": 0.23683685064315796,
      "learning_rate": 2.35741870921041e-05,
      "loss": 0.05717003345489502,
      "step": 2622
    },
    {
      "epoch": 0.3546092099704267,
      "grad_norm": 0.9289461970329285,
      "learning_rate": 2.3568570183794645e-05,
      "loss": 0.06751084327697754,
      "step": 2623
    },
    {
      "epoch": 0.3547444021968737,
      "grad_norm": 0.24210478365421295,
      "learning_rate": 2.356295149146319e-05,
      "loss": 0.06833076477050781,
      "step": 2624
    },
    {
      "epoch": 0.3548795944233207,
      "grad_norm": 0.31110408902168274,
      "learning_rate": 2.3557331016279567e-05,
      "loss": 0.048918724060058594,
      "step": 2625
    },
    {
      "epoch": 0.35501478664976766,
      "grad_norm": 0.4204086661338806,
      "learning_rate": 2.3551708759413998e-05,
      "loss": 0.10658454895019531,
      "step": 2626
    },
    {
      "epoch": 0.35514997887621463,
      "grad_norm": 0.627342700958252,
      "learning_rate": 2.354608472203706e-05,
      "loss": 0.057012081146240234,
      "step": 2627
    },
    {
      "epoch": 0.3552851711026616,
      "grad_norm": 0.6301606893539429,
      "learning_rate": 2.3540458905319705e-05,
      "loss": 0.07567739486694336,
      "step": 2628
    },
    {
      "epoch": 0.3554203633291086,
      "grad_norm": 0.8095700740814209,
      "learning_rate": 2.3534831310433264e-05,
      "loss": 0.06254839897155762,
      "step": 2629
    },
    {
      "epoch": 0.35555555555555557,
      "grad_norm": 0.41683945059776306,
      "learning_rate": 2.3529201938549434e-05,
      "loss": 0.07245302200317383,
      "step": 2630
    },
    {
      "epoch": 0.35569074778200255,
      "grad_norm": 0.2649560570716858,
      "learning_rate": 2.3523570790840274e-05,
      "loss": 0.07520246505737305,
      "step": 2631
    },
    {
      "epoch": 0.3558259400084495,
      "grad_norm": 0.3291664719581604,
      "learning_rate": 2.3517937868478228e-05,
      "loss": 0.06602144241333008,
      "step": 2632
    },
    {
      "epoch": 0.3559611322348965,
      "grad_norm": 0.45417171716690063,
      "learning_rate": 2.3512303172636092e-05,
      "loss": 0.10899162292480469,
      "step": 2633
    },
    {
      "epoch": 0.3560963244613435,
      "grad_norm": 0.47209101915359497,
      "learning_rate": 2.3506666704487033e-05,
      "loss": 0.08172464370727539,
      "step": 2634
    },
    {
      "epoch": 0.35623151668779046,
      "grad_norm": 0.2024308145046234,
      "learning_rate": 2.3501028465204614e-05,
      "loss": 0.05172157287597656,
      "step": 2635
    },
    {
      "epoch": 0.35636670891423744,
      "grad_norm": 0.2909577786922455,
      "learning_rate": 2.3495388455962734e-05,
      "loss": 0.06408214569091797,
      "step": 2636
    },
    {
      "epoch": 0.3565019011406844,
      "grad_norm": 0.30045104026794434,
      "learning_rate": 2.3489746677935673e-05,
      "loss": 0.06735086441040039,
      "step": 2637
    },
    {
      "epoch": 0.3566370933671314,
      "grad_norm": 0.29983365535736084,
      "learning_rate": 2.3484103132298082e-05,
      "loss": 0.09738349914550781,
      "step": 2638
    },
    {
      "epoch": 0.3567722855935784,
      "grad_norm": 0.5098674297332764,
      "learning_rate": 2.347845782022497e-05,
      "loss": 0.07211780548095703,
      "step": 2639
    },
    {
      "epoch": 0.35690747782002535,
      "grad_norm": 0.45709922909736633,
      "learning_rate": 2.3472810742891734e-05,
      "loss": 0.08301448822021484,
      "step": 2640
    },
    {
      "epoch": 0.35704267004647233,
      "grad_norm": 0.3854307234287262,
      "learning_rate": 2.3467161901474118e-05,
      "loss": 0.06582474708557129,
      "step": 2641
    },
    {
      "epoch": 0.3571778622729193,
      "grad_norm": 1.0477949380874634,
      "learning_rate": 2.346151129714824e-05,
      "loss": 0.11779594421386719,
      "step": 2642
    },
    {
      "epoch": 0.3573130544993663,
      "grad_norm": 0.5089383125305176,
      "learning_rate": 2.3455858931090588e-05,
      "loss": 0.07087850570678711,
      "step": 2643
    },
    {
      "epoch": 0.35744824672581327,
      "grad_norm": 0.28413328528404236,
      "learning_rate": 2.3450204804478014e-05,
      "loss": 0.06539106369018555,
      "step": 2644
    },
    {
      "epoch": 0.35758343895226025,
      "grad_norm": 0.3827768862247467,
      "learning_rate": 2.344454891848774e-05,
      "loss": 0.06665468215942383,
      "step": 2645
    },
    {
      "epoch": 0.3577186311787072,
      "grad_norm": 0.22406728565692902,
      "learning_rate": 2.3438891274297348e-05,
      "loss": 0.05580329895019531,
      "step": 2646
    },
    {
      "epoch": 0.3578538234051542,
      "grad_norm": 0.2666054964065552,
      "learning_rate": 2.343323187308479e-05,
      "loss": 0.036884307861328125,
      "step": 2647
    },
    {
      "epoch": 0.3579890156316012,
      "grad_norm": 0.32149770855903625,
      "learning_rate": 2.342757071602839e-05,
      "loss": 0.05465364456176758,
      "step": 2648
    },
    {
      "epoch": 0.35812420785804816,
      "grad_norm": 0.4001545011997223,
      "learning_rate": 2.3421907804306816e-05,
      "loss": 0.07889938354492188,
      "step": 2649
    },
    {
      "epoch": 0.35825940008449514,
      "grad_norm": 0.4604174494743347,
      "learning_rate": 2.341624313909913e-05,
      "loss": 0.07190680503845215,
      "step": 2650
    },
    {
      "epoch": 0.3583945923109421,
      "grad_norm": 0.4112929105758667,
      "learning_rate": 2.3410576721584742e-05,
      "loss": 0.060394287109375,
      "step": 2651
    },
    {
      "epoch": 0.3585297845373891,
      "grad_norm": 0.6505587100982666,
      "learning_rate": 2.3404908552943435e-05,
      "loss": 0.0673673152923584,
      "step": 2652
    },
    {
      "epoch": 0.3586649767638361,
      "grad_norm": 0.20990322530269623,
      "learning_rate": 2.339923863435534e-05,
      "loss": 0.04921913146972656,
      "step": 2653
    },
    {
      "epoch": 0.35880016899028305,
      "grad_norm": 0.29582133889198303,
      "learning_rate": 2.3393566967000974e-05,
      "loss": 0.08043354749679565,
      "step": 2654
    },
    {
      "epoch": 0.35893536121673003,
      "grad_norm": 0.540556788444519,
      "learning_rate": 2.3387893552061202e-05,
      "loss": 0.0742948055267334,
      "step": 2655
    },
    {
      "epoch": 0.359070553443177,
      "grad_norm": 0.5607966780662537,
      "learning_rate": 2.3382218390717268e-05,
      "loss": 0.05429863929748535,
      "step": 2656
    },
    {
      "epoch": 0.359205745669624,
      "grad_norm": 0.3039640486240387,
      "learning_rate": 2.3376541484150762e-05,
      "loss": 0.054396629333496094,
      "step": 2657
    },
    {
      "epoch": 0.35934093789607097,
      "grad_norm": 0.28045663237571716,
      "learning_rate": 2.3370862833543652e-05,
      "loss": 0.05309939384460449,
      "step": 2658
    },
    {
      "epoch": 0.35947613012251795,
      "grad_norm": 0.20683476328849792,
      "learning_rate": 2.336518244007826e-05,
      "loss": 0.05039215087890625,
      "step": 2659
    },
    {
      "epoch": 0.3596113223489649,
      "grad_norm": 0.5803465843200684,
      "learning_rate": 2.3359500304937274e-05,
      "loss": 0.1053009033203125,
      "step": 2660
    },
    {
      "epoch": 0.3597465145754119,
      "grad_norm": 0.3155378997325897,
      "learning_rate": 2.335381642930375e-05,
      "loss": 0.05952921509742737,
      "step": 2661
    },
    {
      "epoch": 0.3598817068018589,
      "grad_norm": 0.3066279888153076,
      "learning_rate": 2.3348130814361094e-05,
      "loss": 0.08416175842285156,
      "step": 2662
    },
    {
      "epoch": 0.36001689902830586,
      "grad_norm": 0.3454969525337219,
      "learning_rate": 2.334244346129309e-05,
      "loss": 0.08137035369873047,
      "step": 2663
    },
    {
      "epoch": 0.36015209125475284,
      "grad_norm": 0.3854672312736511,
      "learning_rate": 2.3336754371283862e-05,
      "loss": 0.06069779396057129,
      "step": 2664
    },
    {
      "epoch": 0.3602872834811998,
      "grad_norm": 0.336407333612442,
      "learning_rate": 2.333106354551792e-05,
      "loss": 0.07065081596374512,
      "step": 2665
    },
    {
      "epoch": 0.3604224757076468,
      "grad_norm": 0.22606343030929565,
      "learning_rate": 2.332537098518012e-05,
      "loss": 0.056944847106933594,
      "step": 2666
    },
    {
      "epoch": 0.3605576679340938,
      "grad_norm": 0.23767049610614777,
      "learning_rate": 2.3319676691455686e-05,
      "loss": 0.05093240737915039,
      "step": 2667
    },
    {
      "epoch": 0.36069286016054075,
      "grad_norm": 0.34505635499954224,
      "learning_rate": 2.3313980665530205e-05,
      "loss": 0.07906246185302734,
      "step": 2668
    },
    {
      "epoch": 0.36082805238698773,
      "grad_norm": 0.2485184669494629,
      "learning_rate": 2.3308282908589606e-05,
      "loss": 0.0647132396697998,
      "step": 2669
    },
    {
      "epoch": 0.3609632446134347,
      "grad_norm": 0.8621676564216614,
      "learning_rate": 2.330258342182021e-05,
      "loss": 0.08985209465026855,
      "step": 2670
    },
    {
      "epoch": 0.3610984368398817,
      "grad_norm": 0.21045689284801483,
      "learning_rate": 2.329688220640866e-05,
      "loss": 0.05572819709777832,
      "step": 2671
    },
    {
      "epoch": 0.36123362906632867,
      "grad_norm": 0.2865007519721985,
      "learning_rate": 2.329117926354199e-05,
      "loss": 0.07909107208251953,
      "step": 2672
    },
    {
      "epoch": 0.36136882129277564,
      "grad_norm": 0.35110726952552795,
      "learning_rate": 2.3285474594407588e-05,
      "loss": 0.08241653442382812,
      "step": 2673
    },
    {
      "epoch": 0.3615040135192226,
      "grad_norm": 0.3314117193222046,
      "learning_rate": 2.327976820019319e-05,
      "loss": 0.07957983016967773,
      "step": 2674
    },
    {
      "epoch": 0.3616392057456696,
      "grad_norm": 0.4399498403072357,
      "learning_rate": 2.32740600820869e-05,
      "loss": 0.1187286376953125,
      "step": 2675
    },
    {
      "epoch": 0.3617743979721166,
      "grad_norm": 0.5442261695861816,
      "learning_rate": 2.326835024127718e-05,
      "loss": 0.07625532150268555,
      "step": 2676
    },
    {
      "epoch": 0.36190959019856356,
      "grad_norm": 0.29945287108421326,
      "learning_rate": 2.326263867895285e-05,
      "loss": 0.052216529846191406,
      "step": 2677
    },
    {
      "epoch": 0.36204478242501054,
      "grad_norm": 0.31563931703567505,
      "learning_rate": 2.3256925396303076e-05,
      "loss": 0.05013012886047363,
      "step": 2678
    },
    {
      "epoch": 0.3621799746514575,
      "grad_norm": 0.3799682557582855,
      "learning_rate": 2.3251210394517412e-05,
      "loss": 0.09147167205810547,
      "step": 2679
    },
    {
      "epoch": 0.3623151668779045,
      "grad_norm": 0.13053618371486664,
      "learning_rate": 2.3245493674785742e-05,
      "loss": 0.03233528137207031,
      "step": 2680
    },
    {
      "epoch": 0.3624503591043515,
      "grad_norm": 0.31542643904685974,
      "learning_rate": 2.3239775238298316e-05,
      "loss": 0.06459903717041016,
      "step": 2681
    },
    {
      "epoch": 0.3625855513307985,
      "grad_norm": 0.34938162565231323,
      "learning_rate": 2.3234055086245744e-05,
      "loss": 0.06383705139160156,
      "step": 2682
    },
    {
      "epoch": 0.3627207435572455,
      "grad_norm": 0.49902424216270447,
      "learning_rate": 2.3228333219818998e-05,
      "loss": 0.055005550384521484,
      "step": 2683
    },
    {
      "epoch": 0.36285593578369246,
      "grad_norm": 0.6347744464874268,
      "learning_rate": 2.3222609640209397e-05,
      "loss": 0.0708012580871582,
      "step": 2684
    },
    {
      "epoch": 0.36299112801013944,
      "grad_norm": 0.2834555208683014,
      "learning_rate": 2.3216884348608614e-05,
      "loss": 0.0907144546508789,
      "step": 2685
    },
    {
      "epoch": 0.3631263202365864,
      "grad_norm": 0.4300340414047241,
      "learning_rate": 2.32111573462087e-05,
      "loss": 0.06136035919189453,
      "step": 2686
    },
    {
      "epoch": 0.3632615124630334,
      "grad_norm": 0.46756449341773987,
      "learning_rate": 2.3205428634202028e-05,
      "loss": 0.06485748291015625,
      "step": 2687
    },
    {
      "epoch": 0.3633967046894804,
      "grad_norm": 0.42676329612731934,
      "learning_rate": 2.3199698213781367e-05,
      "loss": 0.08089709281921387,
      "step": 2688
    },
    {
      "epoch": 0.36353189691592735,
      "grad_norm": 0.617524266242981,
      "learning_rate": 2.319396608613981e-05,
      "loss": 0.09272432327270508,
      "step": 2689
    },
    {
      "epoch": 0.36366708914237433,
      "grad_norm": 0.39837923645973206,
      "learning_rate": 2.318823225247082e-05,
      "loss": 0.0822758674621582,
      "step": 2690
    },
    {
      "epoch": 0.3638022813688213,
      "grad_norm": 0.43281444907188416,
      "learning_rate": 2.3182496713968208e-05,
      "loss": 0.08322620391845703,
      "step": 2691
    },
    {
      "epoch": 0.3639374735952683,
      "grad_norm": 0.16380058228969574,
      "learning_rate": 2.3176759471826143e-05,
      "loss": 0.04537546634674072,
      "step": 2692
    },
    {
      "epoch": 0.36407266582171527,
      "grad_norm": 0.39458754658699036,
      "learning_rate": 2.3171020527239155e-05,
      "loss": 0.07996082305908203,
      "step": 2693
    },
    {
      "epoch": 0.36420785804816225,
      "grad_norm": 0.4488235414028168,
      "learning_rate": 2.316527988140212e-05,
      "loss": 0.08432817459106445,
      "step": 2694
    },
    {
      "epoch": 0.3643430502746092,
      "grad_norm": 0.1847764551639557,
      "learning_rate": 2.315953753551027e-05,
      "loss": 0.043015480041503906,
      "step": 2695
    },
    {
      "epoch": 0.3644782425010562,
      "grad_norm": 0.23556862771511078,
      "learning_rate": 2.3153793490759197e-05,
      "loss": 0.0568547248840332,
      "step": 2696
    },
    {
      "epoch": 0.3646134347275032,
      "grad_norm": 0.2863653302192688,
      "learning_rate": 2.3148047748344835e-05,
      "loss": 0.05669593811035156,
      "step": 2697
    },
    {
      "epoch": 0.36474862695395016,
      "grad_norm": 0.23151041567325592,
      "learning_rate": 2.314230030946348e-05,
      "loss": 0.04804563522338867,
      "step": 2698
    },
    {
      "epoch": 0.36488381918039714,
      "grad_norm": 0.2547624409198761,
      "learning_rate": 2.3136551175311782e-05,
      "loss": 0.06856918334960938,
      "step": 2699
    },
    {
      "epoch": 0.3650190114068441,
      "grad_norm": 0.19527021050453186,
      "learning_rate": 2.313080034708674e-05,
      "loss": 0.04534578323364258,
      "step": 2700
    },
    {
      "epoch": 0.3651542036332911,
      "grad_norm": 0.3311425745487213,
      "learning_rate": 2.312504782598571e-05,
      "loss": 0.06702995300292969,
      "step": 2701
    },
    {
      "epoch": 0.3652893958597381,
      "grad_norm": 0.31967419385910034,
      "learning_rate": 2.311929361320639e-05,
      "loss": 0.06995820999145508,
      "step": 2702
    },
    {
      "epoch": 0.36542458808618505,
      "grad_norm": 0.30657532811164856,
      "learning_rate": 2.311353770994684e-05,
      "loss": 0.0912940502166748,
      "step": 2703
    },
    {
      "epoch": 0.36555978031263203,
      "grad_norm": 1.057766079902649,
      "learning_rate": 2.310778011740548e-05,
      "loss": 0.11150264739990234,
      "step": 2704
    },
    {
      "epoch": 0.365694972539079,
      "grad_norm": 0.4057547152042389,
      "learning_rate": 2.310202083678106e-05,
      "loss": 0.08566856384277344,
      "step": 2705
    },
    {
      "epoch": 0.365830164765526,
      "grad_norm": 0.32117873430252075,
      "learning_rate": 2.3096259869272694e-05,
      "loss": 0.0414891242980957,
      "step": 2706
    },
    {
      "epoch": 0.36596535699197297,
      "grad_norm": 0.304269403219223,
      "learning_rate": 2.309049721607985e-05,
      "loss": 0.05008554458618164,
      "step": 2707
    },
    {
      "epoch": 0.36610054921841995,
      "grad_norm": 0.30867356061935425,
      "learning_rate": 2.3084732878402342e-05,
      "loss": 0.06404685974121094,
      "step": 2708
    },
    {
      "epoch": 0.3662357414448669,
      "grad_norm": 0.4098702371120453,
      "learning_rate": 2.307896685744034e-05,
      "loss": 0.0984644889831543,
      "step": 2709
    },
    {
      "epoch": 0.3663709336713139,
      "grad_norm": 0.3380717933177948,
      "learning_rate": 2.3073199154394352e-05,
      "loss": 0.059273719787597656,
      "step": 2710
    },
    {
      "epoch": 0.3665061258977609,
      "grad_norm": 0.28665271401405334,
      "learning_rate": 2.3067429770465246e-05,
      "loss": 0.06628751754760742,
      "step": 2711
    },
    {
      "epoch": 0.36664131812420786,
      "grad_norm": 0.27406877279281616,
      "learning_rate": 2.3061658706854244e-05,
      "loss": 0.06778573989868164,
      "step": 2712
    },
    {
      "epoch": 0.36677651035065484,
      "grad_norm": 0.21890553832054138,
      "learning_rate": 2.3055885964762907e-05,
      "loss": 0.0660862922668457,
      "step": 2713
    },
    {
      "epoch": 0.3669117025771018,
      "grad_norm": 0.22885750234127045,
      "learning_rate": 2.3050111545393156e-05,
      "loss": 0.06143760681152344,
      "step": 2714
    },
    {
      "epoch": 0.3670468948035488,
      "grad_norm": 0.43521565198898315,
      "learning_rate": 2.304433544994725e-05,
      "loss": 0.09455585479736328,
      "step": 2715
    },
    {
      "epoch": 0.3671820870299958,
      "grad_norm": 0.4448837637901306,
      "learning_rate": 2.303855767962781e-05,
      "loss": 0.08457422256469727,
      "step": 2716
    },
    {
      "epoch": 0.36731727925644275,
      "grad_norm": 0.30835726857185364,
      "learning_rate": 2.303277823563779e-05,
      "loss": 0.0513911247253418,
      "step": 2717
    },
    {
      "epoch": 0.36745247148288973,
      "grad_norm": 0.42902329564094543,
      "learning_rate": 2.3026997119180507e-05,
      "loss": 0.06301069259643555,
      "step": 2718
    },
    {
      "epoch": 0.3675876637093367,
      "grad_norm": 0.15876011550426483,
      "learning_rate": 2.3021214331459616e-05,
      "loss": 0.039977073669433594,
      "step": 2719
    },
    {
      "epoch": 0.3677228559357837,
      "grad_norm": 0.3574123978614807,
      "learning_rate": 2.301542987367913e-05,
      "loss": 0.07372152805328369,
      "step": 2720
    },
    {
      "epoch": 0.36785804816223067,
      "grad_norm": 0.27513548731803894,
      "learning_rate": 2.3009643747043403e-05,
      "loss": 0.05409669876098633,
      "step": 2721
    },
    {
      "epoch": 0.36799324038867764,
      "grad_norm": 0.4079545736312866,
      "learning_rate": 2.3003855952757132e-05,
      "loss": 0.06157541275024414,
      "step": 2722
    },
    {
      "epoch": 0.3681284326151246,
      "grad_norm": 0.4085012972354889,
      "learning_rate": 2.2998066492025372e-05,
      "loss": 0.08626556396484375,
      "step": 2723
    },
    {
      "epoch": 0.3682636248415716,
      "grad_norm": 0.28132185339927673,
      "learning_rate": 2.2992275366053513e-05,
      "loss": 0.0478367805480957,
      "step": 2724
    },
    {
      "epoch": 0.3683988170680186,
      "grad_norm": 0.40833789110183716,
      "learning_rate": 2.2986482576047305e-05,
      "loss": 0.08284187316894531,
      "step": 2725
    },
    {
      "epoch": 0.36853400929446556,
      "grad_norm": 0.34776440262794495,
      "learning_rate": 2.298068812321284e-05,
      "loss": 0.06432294845581055,
      "step": 2726
    },
    {
      "epoch": 0.36866920152091254,
      "grad_norm": 0.2508021295070648,
      "learning_rate": 2.297489200875654e-05,
      "loss": 0.05548906326293945,
      "step": 2727
    },
    {
      "epoch": 0.3688043937473595,
      "grad_norm": 0.24563243985176086,
      "learning_rate": 2.2969094233885204e-05,
      "loss": 0.04309797286987305,
      "step": 2728
    },
    {
      "epoch": 0.3689395859738065,
      "grad_norm": 0.23610103130340576,
      "learning_rate": 2.296329479980595e-05,
      "loss": 0.05093646049499512,
      "step": 2729
    },
    {
      "epoch": 0.3690747782002535,
      "grad_norm": 0.22633366286754608,
      "learning_rate": 2.2957493707726252e-05,
      "loss": 0.04903268814086914,
      "step": 2730
    },
    {
      "epoch": 0.36920997042670045,
      "grad_norm": 0.4889355003833771,
      "learning_rate": 2.2951690958853932e-05,
      "loss": 0.08364629745483398,
      "step": 2731
    },
    {
      "epoch": 0.36934516265314743,
      "grad_norm": 0.4934959411621094,
      "learning_rate": 2.2945886554397154e-05,
      "loss": 0.07945013046264648,
      "step": 2732
    },
    {
      "epoch": 0.3694803548795944,
      "grad_norm": 0.5263916850090027,
      "learning_rate": 2.294008049556441e-05,
      "loss": 0.08485126495361328,
      "step": 2733
    },
    {
      "epoch": 0.3696155471060414,
      "grad_norm": 0.3724130094051361,
      "learning_rate": 2.2934272783564577e-05,
      "loss": 0.09107446670532227,
      "step": 2734
    },
    {
      "epoch": 0.36975073933248837,
      "grad_norm": 0.2591290771961212,
      "learning_rate": 2.2928463419606835e-05,
      "loss": 0.05467033386230469,
      "step": 2735
    },
    {
      "epoch": 0.36988593155893534,
      "grad_norm": 0.36286965012550354,
      "learning_rate": 2.292265240490073e-05,
      "loss": 0.06613016128540039,
      "step": 2736
    },
    {
      "epoch": 0.3700211237853823,
      "grad_norm": 0.3119962811470032,
      "learning_rate": 2.2916839740656154e-05,
      "loss": 0.06099367141723633,
      "step": 2737
    },
    {
      "epoch": 0.3701563160118293,
      "grad_norm": 0.23577842116355896,
      "learning_rate": 2.2911025428083316e-05,
      "loss": 0.04093170166015625,
      "step": 2738
    },
    {
      "epoch": 0.3702915082382763,
      "grad_norm": 0.2807522118091583,
      "learning_rate": 2.2905209468392798e-05,
      "loss": 0.0690145492553711,
      "step": 2739
    },
    {
      "epoch": 0.37042670046472326,
      "grad_norm": 0.22506266832351685,
      "learning_rate": 2.2899391862795514e-05,
      "loss": 0.058818817138671875,
      "step": 2740
    },
    {
      "epoch": 0.37056189269117024,
      "grad_norm": 0.15967266261577606,
      "learning_rate": 2.2893572612502718e-05,
      "loss": 0.03066563606262207,
      "step": 2741
    },
    {
      "epoch": 0.3706970849176172,
      "grad_norm": 0.24224527180194855,
      "learning_rate": 2.2887751718726013e-05,
      "loss": 0.060857564210891724,
      "step": 2742
    },
    {
      "epoch": 0.3708322771440642,
      "grad_norm": 0.3642336428165436,
      "learning_rate": 2.288192918267734e-05,
      "loss": 0.04384136199951172,
      "step": 2743
    },
    {
      "epoch": 0.37096746937051117,
      "grad_norm": 0.33621320128440857,
      "learning_rate": 2.2876105005568974e-05,
      "loss": 0.07636129856109619,
      "step": 2744
    },
    {
      "epoch": 0.37110266159695815,
      "grad_norm": 0.37995463609695435,
      "learning_rate": 2.287027918861355e-05,
      "loss": 0.09354114532470703,
      "step": 2745
    },
    {
      "epoch": 0.37123785382340513,
      "grad_norm": 0.2660060226917267,
      "learning_rate": 2.2864451733024024e-05,
      "loss": 0.049910545349121094,
      "step": 2746
    },
    {
      "epoch": 0.3713730460498521,
      "grad_norm": 0.23012079298496246,
      "learning_rate": 2.2858622640013716e-05,
      "loss": 0.06747961044311523,
      "step": 2747
    },
    {
      "epoch": 0.3715082382762991,
      "grad_norm": 0.35055288672447205,
      "learning_rate": 2.285279191079626e-05,
      "loss": 0.06960487365722656,
      "step": 2748
    },
    {
      "epoch": 0.3716434305027461,
      "grad_norm": 0.2419525682926178,
      "learning_rate": 2.2846959546585656e-05,
      "loss": 0.06652379035949707,
      "step": 2749
    },
    {
      "epoch": 0.3717786227291931,
      "grad_norm": 0.2012145072221756,
      "learning_rate": 2.2841125548596225e-05,
      "loss": 0.0448918342590332,
      "step": 2750
    },
    {
      "epoch": 0.3719138149556401,
      "grad_norm": 0.35305511951446533,
      "learning_rate": 2.2835289918042648e-05,
      "loss": 0.08181023597717285,
      "step": 2751
    },
    {
      "epoch": 0.37204900718208705,
      "grad_norm": 0.24050933122634888,
      "learning_rate": 2.282945265613992e-05,
      "loss": 0.06129741668701172,
      "step": 2752
    },
    {
      "epoch": 0.37218419940853403,
      "grad_norm": 0.34321218729019165,
      "learning_rate": 2.2823613764103406e-05,
      "loss": 0.09194564819335938,
      "step": 2753
    },
    {
      "epoch": 0.372319391634981,
      "grad_norm": 0.4667815864086151,
      "learning_rate": 2.2817773243148776e-05,
      "loss": 0.08031225204467773,
      "step": 2754
    },
    {
      "epoch": 0.372454583861428,
      "grad_norm": 0.2626888155937195,
      "learning_rate": 2.2811931094492074e-05,
      "loss": 0.048987627029418945,
      "step": 2755
    },
    {
      "epoch": 0.37258977608787497,
      "grad_norm": 0.5321769118309021,
      "learning_rate": 2.280608731934966e-05,
      "loss": 0.09866523742675781,
      "step": 2756
    },
    {
      "epoch": 0.37272496831432195,
      "grad_norm": 0.4468326270580292,
      "learning_rate": 2.280024191893823e-05,
      "loss": 0.08239603042602539,
      "step": 2757
    },
    {
      "epoch": 0.3728601605407689,
      "grad_norm": 0.36021310091018677,
      "learning_rate": 2.279439489447485e-05,
      "loss": 0.07276153564453125,
      "step": 2758
    },
    {
      "epoch": 0.3729953527672159,
      "grad_norm": 0.17463645339012146,
      "learning_rate": 2.278854624717688e-05,
      "loss": 0.043268680572509766,
      "step": 2759
    },
    {
      "epoch": 0.3731305449936629,
      "grad_norm": 0.28230324387550354,
      "learning_rate": 2.2782695978262045e-05,
      "loss": 0.07775115966796875,
      "step": 2760
    },
    {
      "epoch": 0.37326573722010986,
      "grad_norm": 0.28289878368377686,
      "learning_rate": 2.2776844088948406e-05,
      "loss": 0.07141399383544922,
      "step": 2761
    },
    {
      "epoch": 0.37340092944655684,
      "grad_norm": 0.2537488341331482,
      "learning_rate": 2.2770990580454364e-05,
      "loss": 0.06646251678466797,
      "step": 2762
    },
    {
      "epoch": 0.3735361216730038,
      "grad_norm": 0.3188614845275879,
      "learning_rate": 2.276513545399864e-05,
      "loss": 0.06700706481933594,
      "step": 2763
    },
    {
      "epoch": 0.3736713138994508,
      "grad_norm": 0.3226417601108551,
      "learning_rate": 2.2759278710800306e-05,
      "loss": 0.0826559066772461,
      "step": 2764
    },
    {
      "epoch": 0.3738065061258978,
      "grad_norm": 0.7408802509307861,
      "learning_rate": 2.275342035207876e-05,
      "loss": 0.09273117780685425,
      "step": 2765
    },
    {
      "epoch": 0.37394169835234475,
      "grad_norm": 0.35370224714279175,
      "learning_rate": 2.2747560379053752e-05,
      "loss": 0.07136058807373047,
      "step": 2766
    },
    {
      "epoch": 0.37407689057879173,
      "grad_norm": 0.7217914462089539,
      "learning_rate": 2.2741698792945364e-05,
      "loss": 0.07427978515625,
      "step": 2767
    },
    {
      "epoch": 0.3742120828052387,
      "grad_norm": 0.25097402930259705,
      "learning_rate": 2.2735835594974003e-05,
      "loss": 0.05281949043273926,
      "step": 2768
    },
    {
      "epoch": 0.3743472750316857,
      "grad_norm": 0.27837270498275757,
      "learning_rate": 2.272997078636042e-05,
      "loss": 0.06299638748168945,
      "step": 2769
    },
    {
      "epoch": 0.37448246725813267,
      "grad_norm": 0.43518730998039246,
      "learning_rate": 2.272410436832569e-05,
      "loss": 0.0702052116394043,
      "step": 2770
    },
    {
      "epoch": 0.37461765948457965,
      "grad_norm": 0.2612428069114685,
      "learning_rate": 2.2718236342091248e-05,
      "loss": 0.04570198059082031,
      "step": 2771
    },
    {
      "epoch": 0.3747528517110266,
      "grad_norm": 0.4023292064666748,
      "learning_rate": 2.2712366708878838e-05,
      "loss": 0.07886171340942383,
      "step": 2772
    },
    {
      "epoch": 0.3748880439374736,
      "grad_norm": 0.3393609821796417,
      "learning_rate": 2.2706495469910552e-05,
      "loss": 0.048750877380371094,
      "step": 2773
    },
    {
      "epoch": 0.3750232361639206,
      "grad_norm": 0.3709828555583954,
      "learning_rate": 2.2700622626408814e-05,
      "loss": 0.05859220027923584,
      "step": 2774
    },
    {
      "epoch": 0.37515842839036756,
      "grad_norm": 0.5168056488037109,
      "learning_rate": 2.2694748179596375e-05,
      "loss": 0.07851588726043701,
      "step": 2775
    },
    {
      "epoch": 0.37529362061681454,
      "grad_norm": 0.43829160928726196,
      "learning_rate": 2.2688872130696342e-05,
      "loss": 0.0841212272644043,
      "step": 2776
    },
    {
      "epoch": 0.3754288128432615,
      "grad_norm": 0.640561044216156,
      "learning_rate": 2.268299448093212e-05,
      "loss": 0.07652139663696289,
      "step": 2777
    },
    {
      "epoch": 0.3755640050697085,
      "grad_norm": 0.20402050018310547,
      "learning_rate": 2.2677115231527482e-05,
      "loss": 0.05471146106719971,
      "step": 2778
    },
    {
      "epoch": 0.3756991972961555,
      "grad_norm": 0.22976543009281158,
      "learning_rate": 2.267123438370651e-05,
      "loss": 0.05427813529968262,
      "step": 2779
    },
    {
      "epoch": 0.37583438952260245,
      "grad_norm": 0.6341850757598877,
      "learning_rate": 2.266535193869363e-05,
      "loss": 0.08101654052734375,
      "step": 2780
    },
    {
      "epoch": 0.37596958174904943,
      "grad_norm": 0.5343198776245117,
      "learning_rate": 2.2659467897713604e-05,
      "loss": 0.058493614196777344,
      "step": 2781
    },
    {
      "epoch": 0.3761047739754964,
      "grad_norm": 0.23260414600372314,
      "learning_rate": 2.2653582261991516e-05,
      "loss": 0.05374574661254883,
      "step": 2782
    },
    {
      "epoch": 0.3762399662019434,
      "grad_norm": 0.13126878440380096,
      "learning_rate": 2.2647695032752785e-05,
      "loss": 0.03887307643890381,
      "step": 2783
    },
    {
      "epoch": 0.37637515842839037,
      "grad_norm": 0.37174203991889954,
      "learning_rate": 2.264180621122317e-05,
      "loss": 0.08713150024414062,
      "step": 2784
    },
    {
      "epoch": 0.37651035065483734,
      "grad_norm": 0.3188643753528595,
      "learning_rate": 2.2635915798628747e-05,
      "loss": 0.07477092742919922,
      "step": 2785
    },
    {
      "epoch": 0.3766455428812843,
      "grad_norm": 0.22900445759296417,
      "learning_rate": 2.2630023796195932e-05,
      "loss": 0.060585975646972656,
      "step": 2786
    },
    {
      "epoch": 0.3767807351077313,
      "grad_norm": 0.2584836483001709,
      "learning_rate": 2.262413020515148e-05,
      "loss": 0.049193382263183594,
      "step": 2787
    },
    {
      "epoch": 0.3769159273341783,
      "grad_norm": 0.3672921061515808,
      "learning_rate": 2.261823502672246e-05,
      "loss": 0.0650472640991211,
      "step": 2788
    },
    {
      "epoch": 0.37705111956062526,
      "grad_norm": 0.2643785774707794,
      "learning_rate": 2.261233826213628e-05,
      "loss": 0.05163264274597168,
      "step": 2789
    },
    {
      "epoch": 0.37718631178707224,
      "grad_norm": 0.9889642596244812,
      "learning_rate": 2.2606439912620688e-05,
      "loss": 0.11510229110717773,
      "step": 2790
    },
    {
      "epoch": 0.3773215040135192,
      "grad_norm": 0.3425293266773224,
      "learning_rate": 2.2600539979403734e-05,
      "loss": 0.09545040130615234,
      "step": 2791
    },
    {
      "epoch": 0.3774566962399662,
      "grad_norm": 0.21810343861579895,
      "learning_rate": 2.259463846371383e-05,
      "loss": 0.06734371185302734,
      "step": 2792
    },
    {
      "epoch": 0.3775918884664132,
      "grad_norm": 0.2664501368999481,
      "learning_rate": 2.2588735366779698e-05,
      "loss": 0.057622671127319336,
      "step": 2793
    },
    {
      "epoch": 0.37772708069286015,
      "grad_norm": 0.27079471945762634,
      "learning_rate": 2.2582830689830394e-05,
      "loss": 0.04458427429199219,
      "step": 2794
    },
    {
      "epoch": 0.37786227291930713,
      "grad_norm": 0.23766346275806427,
      "learning_rate": 2.2576924434095305e-05,
      "loss": 0.05955839157104492,
      "step": 2795
    },
    {
      "epoch": 0.3779974651457541,
      "grad_norm": 0.38281500339508057,
      "learning_rate": 2.257101660080414e-05,
      "loss": 0.06927323341369629,
      "step": 2796
    },
    {
      "epoch": 0.3781326573722011,
      "grad_norm": 0.29529163241386414,
      "learning_rate": 2.256510719118695e-05,
      "loss": 0.04836249351501465,
      "step": 2797
    },
    {
      "epoch": 0.37826784959864806,
      "grad_norm": 0.20660586655139923,
      "learning_rate": 2.2559196206474094e-05,
      "loss": 0.056540489196777344,
      "step": 2798
    },
    {
      "epoch": 0.37840304182509504,
      "grad_norm": 0.4235142767429352,
      "learning_rate": 2.2553283647896287e-05,
      "loss": 0.043947458267211914,
      "step": 2799
    },
    {
      "epoch": 0.378538234051542,
      "grad_norm": 0.6575372815132141,
      "learning_rate": 2.254736951668454e-05,
      "loss": 0.07664811611175537,
      "step": 2800
    },
    {
      "epoch": 0.378673426277989,
      "grad_norm": 0.39030948281288147,
      "learning_rate": 2.2541453814070212e-05,
      "loss": 0.06126832962036133,
      "step": 2801
    },
    {
      "epoch": 0.378808618504436,
      "grad_norm": 0.32001739740371704,
      "learning_rate": 2.2535536541284983e-05,
      "loss": 0.07765579223632812,
      "step": 2802
    },
    {
      "epoch": 0.37894381073088296,
      "grad_norm": 0.3228089511394501,
      "learning_rate": 2.2529617699560857e-05,
      "loss": 0.04865312576293945,
      "step": 2803
    },
    {
      "epoch": 0.37907900295732994,
      "grad_norm": 0.2946130931377411,
      "learning_rate": 2.2523697290130185e-05,
      "loss": 0.06833744049072266,
      "step": 2804
    },
    {
      "epoch": 0.3792141951837769,
      "grad_norm": 0.3752012848854065,
      "learning_rate": 2.251777531422561e-05,
      "loss": 0.07851028442382812,
      "step": 2805
    },
    {
      "epoch": 0.3793493874102239,
      "grad_norm": 0.586348831653595,
      "learning_rate": 2.2511851773080127e-05,
      "loss": 0.10017108917236328,
      "step": 2806
    },
    {
      "epoch": 0.37948457963667087,
      "grad_norm": 0.5851037502288818,
      "learning_rate": 2.2505926667927043e-05,
      "loss": 0.07167577743530273,
      "step": 2807
    },
    {
      "epoch": 0.37961977186311785,
      "grad_norm": 0.31651997566223145,
      "learning_rate": 2.25e-05,
      "loss": 0.08237814903259277,
      "step": 2808
    },
    {
      "epoch": 0.37975496408956483,
      "grad_norm": 0.2083374410867691,
      "learning_rate": 2.2494071770532966e-05,
      "loss": 0.05229926109313965,
      "step": 2809
    },
    {
      "epoch": 0.3798901563160118,
      "grad_norm": 0.20976102352142334,
      "learning_rate": 2.2488141980760223e-05,
      "loss": 0.050457000732421875,
      "step": 2810
    },
    {
      "epoch": 0.3800253485424588,
      "grad_norm": 0.28908151388168335,
      "learning_rate": 2.248221063191639e-05,
      "loss": 0.0801401138305664,
      "step": 2811
    },
    {
      "epoch": 0.38016054076890576,
      "grad_norm": 0.29635336995124817,
      "learning_rate": 2.24762777252364e-05,
      "loss": 0.05572652816772461,
      "step": 2812
    },
    {
      "epoch": 0.38029573299535274,
      "grad_norm": 0.43832477927207947,
      "learning_rate": 2.2470343261955525e-05,
      "loss": 0.04839181900024414,
      "step": 2813
    },
    {
      "epoch": 0.3804309252217997,
      "grad_norm": 0.2038612812757492,
      "learning_rate": 2.246440724330934e-05,
      "loss": 0.052689552307128906,
      "step": 2814
    },
    {
      "epoch": 0.3805661174482467,
      "grad_norm": 0.2589648962020874,
      "learning_rate": 2.2458469670533765e-05,
      "loss": 0.06357526779174805,
      "step": 2815
    },
    {
      "epoch": 0.3807013096746937,
      "grad_norm": 0.33596086502075195,
      "learning_rate": 2.2452530544865034e-05,
      "loss": 0.08633661270141602,
      "step": 2816
    },
    {
      "epoch": 0.3808365019011407,
      "grad_norm": 0.2913959324359894,
      "learning_rate": 2.24465898675397e-05,
      "loss": 0.08229923248291016,
      "step": 2817
    },
    {
      "epoch": 0.3809716941275877,
      "grad_norm": 0.53569096326828,
      "learning_rate": 2.244064763979464e-05,
      "loss": 0.06948637962341309,
      "step": 2818
    },
    {
      "epoch": 0.38110688635403467,
      "grad_norm": 0.23542706668376923,
      "learning_rate": 2.2434703862867068e-05,
      "loss": 0.06728863716125488,
      "step": 2819
    },
    {
      "epoch": 0.38124207858048165,
      "grad_norm": 0.35748404264450073,
      "learning_rate": 2.2428758537994504e-05,
      "loss": 0.059915125370025635,
      "step": 2820
    },
    {
      "epoch": 0.3813772708069286,
      "grad_norm": 0.5053253173828125,
      "learning_rate": 2.24228116664148e-05,
      "loss": 0.0852808952331543,
      "step": 2821
    },
    {
      "epoch": 0.3815124630333756,
      "grad_norm": 0.45190510153770447,
      "learning_rate": 2.2416863249366125e-05,
      "loss": 0.08397531509399414,
      "step": 2822
    },
    {
      "epoch": 0.3816476552598226,
      "grad_norm": 0.40200597047805786,
      "learning_rate": 2.241091328808696e-05,
      "loss": 0.05139565467834473,
      "step": 2823
    },
    {
      "epoch": 0.38178284748626956,
      "grad_norm": 0.3294566571712494,
      "learning_rate": 2.240496178381614e-05,
      "loss": 0.051227569580078125,
      "step": 2824
    },
    {
      "epoch": 0.38191803971271654,
      "grad_norm": 0.5295678377151489,
      "learning_rate": 2.239900873779278e-05,
      "loss": 0.05830860137939453,
      "step": 2825
    },
    {
      "epoch": 0.3820532319391635,
      "grad_norm": 0.33402496576309204,
      "learning_rate": 2.2393054151256352e-05,
      "loss": 0.05664205551147461,
      "step": 2826
    },
    {
      "epoch": 0.3821884241656105,
      "grad_norm": 0.37120258808135986,
      "learning_rate": 2.238709802544662e-05,
      "loss": 0.09269380569458008,
      "step": 2827
    },
    {
      "epoch": 0.3823236163920575,
      "grad_norm": 0.5080986618995667,
      "learning_rate": 2.2381140361603686e-05,
      "loss": 0.07447850704193115,
      "step": 2828
    },
    {
      "epoch": 0.38245880861850445,
      "grad_norm": 0.3413444459438324,
      "learning_rate": 2.237518116096797e-05,
      "loss": 0.063995361328125,
      "step": 2829
    },
    {
      "epoch": 0.38259400084495143,
      "grad_norm": 0.25153154134750366,
      "learning_rate": 2.2369220424780203e-05,
      "loss": 0.06133317947387695,
      "step": 2830
    },
    {
      "epoch": 0.3827291930713984,
      "grad_norm": 0.316782683134079,
      "learning_rate": 2.2363258154281452e-05,
      "loss": 0.09371709823608398,
      "step": 2831
    },
    {
      "epoch": 0.3828643852978454,
      "grad_norm": 0.35821273922920227,
      "learning_rate": 2.2357294350713088e-05,
      "loss": 0.056162118911743164,
      "step": 2832
    },
    {
      "epoch": 0.38299957752429237,
      "grad_norm": 0.33244818449020386,
      "learning_rate": 2.2351329015316802e-05,
      "loss": 0.06569063663482666,
      "step": 2833
    },
    {
      "epoch": 0.38313476975073935,
      "grad_norm": 0.7121259570121765,
      "learning_rate": 2.2345362149334613e-05,
      "loss": 0.07290458679199219,
      "step": 2834
    },
    {
      "epoch": 0.3832699619771863,
      "grad_norm": 0.3778751790523529,
      "learning_rate": 2.2339393754008854e-05,
      "loss": 0.07805752754211426,
      "step": 2835
    },
    {
      "epoch": 0.3834051542036333,
      "grad_norm": 0.46667975187301636,
      "learning_rate": 2.233342383058218e-05,
      "loss": 0.09800231456756592,
      "step": 2836
    },
    {
      "epoch": 0.3835403464300803,
      "grad_norm": 0.1805945634841919,
      "learning_rate": 2.2327452380297554e-05,
      "loss": 0.042798519134521484,
      "step": 2837
    },
    {
      "epoch": 0.38367553865652726,
      "grad_norm": 0.26929038763046265,
      "learning_rate": 2.232147940439827e-05,
      "loss": 0.06944727897644043,
      "step": 2838
    },
    {
      "epoch": 0.38381073088297424,
      "grad_norm": 0.3745846450328827,
      "learning_rate": 2.2315504904127936e-05,
      "loss": 0.08572673797607422,
      "step": 2839
    },
    {
      "epoch": 0.3839459231094212,
      "grad_norm": 0.25058165192604065,
      "learning_rate": 2.2309528880730463e-05,
      "loss": 0.06348562240600586,
      "step": 2840
    },
    {
      "epoch": 0.3840811153358682,
      "grad_norm": 0.20035450160503387,
      "learning_rate": 2.2303551335450096e-05,
      "loss": 0.03886604309082031,
      "step": 2841
    },
    {
      "epoch": 0.3842163075623152,
      "grad_norm": 0.43715858459472656,
      "learning_rate": 2.2297572269531398e-05,
      "loss": 0.05697441101074219,
      "step": 2842
    },
    {
      "epoch": 0.38435149978876215,
      "grad_norm": 0.23630057275295258,
      "learning_rate": 2.2291591684219243e-05,
      "loss": 0.06506586074829102,
      "step": 2843
    },
    {
      "epoch": 0.38448669201520913,
      "grad_norm": 0.3400982618331909,
      "learning_rate": 2.2285609580758806e-05,
      "loss": 0.04954862594604492,
      "step": 2844
    },
    {
      "epoch": 0.3846218842416561,
      "grad_norm": 0.2969959080219269,
      "learning_rate": 2.227962596039561e-05,
      "loss": 0.06925249099731445,
      "step": 2845
    },
    {
      "epoch": 0.3847570764681031,
      "grad_norm": 0.44228595495224,
      "learning_rate": 2.2273640824375462e-05,
      "loss": 0.06440496444702148,
      "step": 2846
    },
    {
      "epoch": 0.38489226869455007,
      "grad_norm": 0.43075305223464966,
      "learning_rate": 2.2267654173944515e-05,
      "loss": 0.0472872257232666,
      "step": 2847
    },
    {
      "epoch": 0.38502746092099704,
      "grad_norm": 0.24376831948757172,
      "learning_rate": 2.2261666010349212e-05,
      "loss": 0.04905128479003906,
      "step": 2848
    },
    {
      "epoch": 0.385162653147444,
      "grad_norm": 0.31551626324653625,
      "learning_rate": 2.2255676334836317e-05,
      "loss": 0.07819223403930664,
      "step": 2849
    },
    {
      "epoch": 0.385297845373891,
      "grad_norm": 0.21241162717342377,
      "learning_rate": 2.2249685148652917e-05,
      "loss": 0.05205106735229492,
      "step": 2850
    },
    {
      "epoch": 0.385433037600338,
      "grad_norm": 0.5643239617347717,
      "learning_rate": 2.224369245304641e-05,
      "loss": 0.0682516098022461,
      "step": 2851
    },
    {
      "epoch": 0.38556822982678496,
      "grad_norm": 0.7587423920631409,
      "learning_rate": 2.2237698249264507e-05,
      "loss": 0.07445049285888672,
      "step": 2852
    },
    {
      "epoch": 0.38570342205323194,
      "grad_norm": 0.22339507937431335,
      "learning_rate": 2.2231702538555235e-05,
      "loss": 0.06548738479614258,
      "step": 2853
    },
    {
      "epoch": 0.3858386142796789,
      "grad_norm": 0.4586602747440338,
      "learning_rate": 2.2225705322166928e-05,
      "loss": 0.0638885498046875,
      "step": 2854
    },
    {
      "epoch": 0.3859738065061259,
      "grad_norm": 0.7055124640464783,
      "learning_rate": 2.2219706601348242e-05,
      "loss": 0.06122446060180664,
      "step": 2855
    },
    {
      "epoch": 0.3861089987325729,
      "grad_norm": 0.5644597411155701,
      "learning_rate": 2.221370637734814e-05,
      "loss": 0.071929931640625,
      "step": 2856
    },
    {
      "epoch": 0.38624419095901985,
      "grad_norm": 0.4933735132217407,
      "learning_rate": 2.22077046514159e-05,
      "loss": 0.08907699584960938,
      "step": 2857
    },
    {
      "epoch": 0.38637938318546683,
      "grad_norm": 0.31833040714263916,
      "learning_rate": 2.220170142480112e-05,
      "loss": 0.06456661224365234,
      "step": 2858
    },
    {
      "epoch": 0.3865145754119138,
      "grad_norm": 0.21775972843170166,
      "learning_rate": 2.2195696698753695e-05,
      "loss": 0.0671839714050293,
      "step": 2859
    },
    {
      "epoch": 0.3866497676383608,
      "grad_norm": 0.21280226111412048,
      "learning_rate": 2.2189690474523844e-05,
      "loss": 0.03300642967224121,
      "step": 2860
    },
    {
      "epoch": 0.38678495986480776,
      "grad_norm": 0.39132893085479736,
      "learning_rate": 2.21836827533621e-05,
      "loss": 0.07460212707519531,
      "step": 2861
    },
    {
      "epoch": 0.38692015209125474,
      "grad_norm": 0.24174456298351288,
      "learning_rate": 2.2177673536519297e-05,
      "loss": 0.04238176345825195,
      "step": 2862
    },
    {
      "epoch": 0.3870553443177017,
      "grad_norm": 0.7087098360061646,
      "learning_rate": 2.217166282524659e-05,
      "loss": 0.08649265766143799,
      "step": 2863
    },
    {
      "epoch": 0.3871905365441487,
      "grad_norm": 0.44046109914779663,
      "learning_rate": 2.216565062079544e-05,
      "loss": 0.07009744644165039,
      "step": 2864
    },
    {
      "epoch": 0.3873257287705957,
      "grad_norm": 0.6048197150230408,
      "learning_rate": 2.2159636924417612e-05,
      "loss": 0.05726814270019531,
      "step": 2865
    },
    {
      "epoch": 0.38746092099704266,
      "grad_norm": 0.48509982228279114,
      "learning_rate": 2.2153621737365205e-05,
      "loss": 0.08022212982177734,
      "step": 2866
    },
    {
      "epoch": 0.38759611322348964,
      "grad_norm": 0.3172224462032318,
      "learning_rate": 2.2147605060890598e-05,
      "loss": 0.05950450897216797,
      "step": 2867
    },
    {
      "epoch": 0.3877313054499366,
      "grad_norm": 0.3830418586730957,
      "learning_rate": 2.2141586896246503e-05,
      "loss": 0.04602622985839844,
      "step": 2868
    },
    {
      "epoch": 0.3878664976763836,
      "grad_norm": 0.3074946999549866,
      "learning_rate": 2.2135567244685933e-05,
      "loss": 0.057450294494628906,
      "step": 2869
    },
    {
      "epoch": 0.38800168990283057,
      "grad_norm": 0.4040433168411255,
      "learning_rate": 2.2129546107462214e-05,
      "loss": 0.09191322326660156,
      "step": 2870
    },
    {
      "epoch": 0.38813688212927755,
      "grad_norm": 0.26097050309181213,
      "learning_rate": 2.212352348582897e-05,
      "loss": 0.07602787017822266,
      "step": 2871
    },
    {
      "epoch": 0.38827207435572453,
      "grad_norm": 0.21725687384605408,
      "learning_rate": 2.2117499381040157e-05,
      "loss": 0.06807947158813477,
      "step": 2872
    },
    {
      "epoch": 0.3884072665821715,
      "grad_norm": 0.3893558382987976,
      "learning_rate": 2.211147379435001e-05,
      "loss": 0.07411360740661621,
      "step": 2873
    },
    {
      "epoch": 0.3885424588086185,
      "grad_norm": 0.41507139801979065,
      "learning_rate": 2.2105446727013098e-05,
      "loss": 0.08133411407470703,
      "step": 2874
    },
    {
      "epoch": 0.38867765103506546,
      "grad_norm": 0.5074864625930786,
      "learning_rate": 2.209941818028429e-05,
      "loss": 0.08805179595947266,
      "step": 2875
    },
    {
      "epoch": 0.38881284326151244,
      "grad_norm": 0.3369874358177185,
      "learning_rate": 2.2093388155418757e-05,
      "loss": 0.05635690689086914,
      "step": 2876
    },
    {
      "epoch": 0.3889480354879594,
      "grad_norm": 0.42361295223236084,
      "learning_rate": 2.2087356653671982e-05,
      "loss": 0.07360386848449707,
      "step": 2877
    },
    {
      "epoch": 0.3890832277144064,
      "grad_norm": 0.13845252990722656,
      "learning_rate": 2.2081323676299756e-05,
      "loss": 0.04274892807006836,
      "step": 2878
    },
    {
      "epoch": 0.3892184199408534,
      "grad_norm": 0.317159503698349,
      "learning_rate": 2.207528922455818e-05,
      "loss": 0.053198814392089844,
      "step": 2879
    },
    {
      "epoch": 0.38935361216730036,
      "grad_norm": 0.19099321961402893,
      "learning_rate": 2.206925329970366e-05,
      "loss": 0.04053235054016113,
      "step": 2880
    },
    {
      "epoch": 0.38948880439374733,
      "grad_norm": 0.28317663073539734,
      "learning_rate": 2.20632159029929e-05,
      "loss": 0.08286499977111816,
      "step": 2881
    },
    {
      "epoch": 0.3896239966201943,
      "grad_norm": 0.401815265417099,
      "learning_rate": 2.2057177035682926e-05,
      "loss": 0.07370281219482422,
      "step": 2882
    },
    {
      "epoch": 0.3897591888466413,
      "grad_norm": 0.42706337571144104,
      "learning_rate": 2.2051136699031058e-05,
      "loss": 0.07823848724365234,
      "step": 2883
    },
    {
      "epoch": 0.38989438107308827,
      "grad_norm": 0.4053654074668884,
      "learning_rate": 2.2045094894294933e-05,
      "loss": 0.07383918762207031,
      "step": 2884
    },
    {
      "epoch": 0.3900295732995353,
      "grad_norm": 0.3096252977848053,
      "learning_rate": 2.203905162273248e-05,
      "loss": 0.06415963172912598,
      "step": 2885
    },
    {
      "epoch": 0.3901647655259823,
      "grad_norm": 0.5765342116355896,
      "learning_rate": 2.203300688560194e-05,
      "loss": 0.08324623107910156,
      "step": 2886
    },
    {
      "epoch": 0.39029995775242926,
      "grad_norm": 0.3492313623428345,
      "learning_rate": 2.2026960684161862e-05,
      "loss": 0.052733659744262695,
      "step": 2887
    },
    {
      "epoch": 0.39043514997887624,
      "grad_norm": 0.2543022930622101,
      "learning_rate": 2.2020913019671097e-05,
      "loss": 0.056778907775878906,
      "step": 2888
    },
    {
      "epoch": 0.3905703422053232,
      "grad_norm": 0.22441573441028595,
      "learning_rate": 2.20148638933888e-05,
      "loss": 0.05511680245399475,
      "step": 2889
    },
    {
      "epoch": 0.3907055344317702,
      "grad_norm": 0.2703585922718048,
      "learning_rate": 2.2008813306574438e-05,
      "loss": 0.05136585235595703,
      "step": 2890
    },
    {
      "epoch": 0.3908407266582172,
      "grad_norm": 0.5141708254814148,
      "learning_rate": 2.200276126048777e-05,
      "loss": 0.08012866973876953,
      "step": 2891
    },
    {
      "epoch": 0.39097591888466415,
      "grad_norm": 0.33381491899490356,
      "learning_rate": 2.199670775638886e-05,
      "loss": 0.0890340805053711,
      "step": 2892
    },
    {
      "epoch": 0.39111111111111113,
      "grad_norm": 0.34990614652633667,
      "learning_rate": 2.1990652795538085e-05,
      "loss": 0.05481719970703125,
      "step": 2893
    },
    {
      "epoch": 0.3912463033375581,
      "grad_norm": 0.28374919295310974,
      "learning_rate": 2.1984596379196117e-05,
      "loss": 0.054431915283203125,
      "step": 2894
    },
    {
      "epoch": 0.3913814955640051,
      "grad_norm": 0.2645837068557739,
      "learning_rate": 2.1978538508623942e-05,
      "loss": 0.08042633533477783,
      "step": 2895
    },
    {
      "epoch": 0.39151668779045207,
      "grad_norm": 0.42274656891822815,
      "learning_rate": 2.197247918508283e-05,
      "loss": 0.0625143051147461,
      "step": 2896
    },
    {
      "epoch": 0.39165188001689905,
      "grad_norm": 0.49335598945617676,
      "learning_rate": 2.1966418409834374e-05,
      "loss": 0.06671142578125,
      "step": 2897
    },
    {
      "epoch": 0.391787072243346,
      "grad_norm": 0.3548906445503235,
      "learning_rate": 2.1960356184140453e-05,
      "loss": 0.0783998966217041,
      "step": 2898
    },
    {
      "epoch": 0.391922264469793,
      "grad_norm": 0.2631221413612366,
      "learning_rate": 2.1954292509263258e-05,
      "loss": 0.055816650390625,
      "step": 2899
    },
    {
      "epoch": 0.39205745669624,
      "grad_norm": 0.18816456198692322,
      "learning_rate": 2.194822738646528e-05,
      "loss": 0.046355485916137695,
      "step": 2900
    },
    {
      "epoch": 0.39219264892268696,
      "grad_norm": 0.41053277254104614,
      "learning_rate": 2.1942160817009304e-05,
      "loss": 0.07032132148742676,
      "step": 2901
    },
    {
      "epoch": 0.39232784114913394,
      "grad_norm": 0.7297241687774658,
      "learning_rate": 2.193609280215843e-05,
      "loss": 0.08548355102539062,
      "step": 2902
    },
    {
      "epoch": 0.3924630333755809,
      "grad_norm": 0.5028813481330872,
      "learning_rate": 2.1930023343176044e-05,
      "loss": 0.07327556610107422,
      "step": 2903
    },
    {
      "epoch": 0.3925982256020279,
      "grad_norm": 0.2076469361782074,
      "learning_rate": 2.1923952441325837e-05,
      "loss": 0.07343053817749023,
      "step": 2904
    },
    {
      "epoch": 0.3927334178284749,
      "grad_norm": 0.26775214076042175,
      "learning_rate": 2.191788009787182e-05,
      "loss": 0.05754566192626953,
      "step": 2905
    },
    {
      "epoch": 0.39286861005492185,
      "grad_norm": 0.3015051782131195,
      "learning_rate": 2.1911806314078267e-05,
      "loss": 0.0705561637878418,
      "step": 2906
    },
    {
      "epoch": 0.39300380228136883,
      "grad_norm": 0.34772127866744995,
      "learning_rate": 2.1905731091209786e-05,
      "loss": 0.081695556640625,
      "step": 2907
    },
    {
      "epoch": 0.3931389945078158,
      "grad_norm": 0.3780077397823334,
      "learning_rate": 2.1899654430531262e-05,
      "loss": 0.10386800765991211,
      "step": 2908
    },
    {
      "epoch": 0.3932741867342628,
      "grad_norm": 0.37237271666526794,
      "learning_rate": 2.18935763333079e-05,
      "loss": 0.06761503219604492,
      "step": 2909
    },
    {
      "epoch": 0.39340937896070977,
      "grad_norm": 0.22529982030391693,
      "learning_rate": 2.1887496800805175e-05,
      "loss": 0.06626415252685547,
      "step": 2910
    },
    {
      "epoch": 0.39354457118715674,
      "grad_norm": 0.3191796541213989,
      "learning_rate": 2.188141583428889e-05,
      "loss": 0.0892951488494873,
      "step": 2911
    },
    {
      "epoch": 0.3936797634136037,
      "grad_norm": 0.14784513413906097,
      "learning_rate": 2.1875333435025138e-05,
      "loss": 0.026790857315063477,
      "step": 2912
    },
    {
      "epoch": 0.3938149556400507,
      "grad_norm": 0.12898798286914825,
      "learning_rate": 2.1869249604280296e-05,
      "loss": 0.034829139709472656,
      "step": 2913
    },
    {
      "epoch": 0.3939501478664977,
      "grad_norm": 0.39526745676994324,
      "learning_rate": 2.1863164343321057e-05,
      "loss": 0.06908988952636719,
      "step": 2914
    },
    {
      "epoch": 0.39408534009294466,
      "grad_norm": 0.38009893894195557,
      "learning_rate": 2.1857077653414397e-05,
      "loss": 0.08252692222595215,
      "step": 2915
    },
    {
      "epoch": 0.39422053231939164,
      "grad_norm": 0.2364497035741806,
      "learning_rate": 2.185098953582761e-05,
      "loss": 0.0792229175567627,
      "step": 2916
    },
    {
      "epoch": 0.3943557245458386,
      "grad_norm": 0.35587725043296814,
      "learning_rate": 2.1844899991828265e-05,
      "loss": 0.0800466537475586,
      "step": 2917
    },
    {
      "epoch": 0.3944909167722856,
      "grad_norm": 0.5487125515937805,
      "learning_rate": 2.1838809022684247e-05,
      "loss": 0.0685577392578125,
      "step": 2918
    },
    {
      "epoch": 0.39462610899873257,
      "grad_norm": 0.3930009603500366,
      "learning_rate": 2.1832716629663712e-05,
      "loss": 0.05109715461730957,
      "step": 2919
    },
    {
      "epoch": 0.39476130122517955,
      "grad_norm": 0.3796512484550476,
      "learning_rate": 2.1826622814035138e-05,
      "loss": 0.0814366340637207,
      "step": 2920
    },
    {
      "epoch": 0.39489649345162653,
      "grad_norm": 0.29777342081069946,
      "learning_rate": 2.1820527577067293e-05,
      "loss": 0.07606244087219238,
      "step": 2921
    },
    {
      "epoch": 0.3950316856780735,
      "grad_norm": 0.24324575066566467,
      "learning_rate": 2.1814430920029238e-05,
      "loss": 0.0659322738647461,
      "step": 2922
    },
    {
      "epoch": 0.3951668779045205,
      "grad_norm": 0.28217342495918274,
      "learning_rate": 2.1808332844190325e-05,
      "loss": 0.07102012634277344,
      "step": 2923
    },
    {
      "epoch": 0.39530207013096746,
      "grad_norm": 0.3461306095123291,
      "learning_rate": 2.1802233350820203e-05,
      "loss": 0.06911420822143555,
      "step": 2924
    },
    {
      "epoch": 0.39543726235741444,
      "grad_norm": 0.40410807728767395,
      "learning_rate": 2.179613244118883e-05,
      "loss": 0.0623631477355957,
      "step": 2925
    },
    {
      "epoch": 0.3955724545838614,
      "grad_norm": 0.3982985019683838,
      "learning_rate": 2.1790030116566436e-05,
      "loss": 0.06379342079162598,
      "step": 2926
    },
    {
      "epoch": 0.3957076468103084,
      "grad_norm": 0.31446436047554016,
      "learning_rate": 2.1783926378223563e-05,
      "loss": 0.07854652404785156,
      "step": 2927
    },
    {
      "epoch": 0.3958428390367554,
      "grad_norm": 0.26369568705558777,
      "learning_rate": 2.1777821227431048e-05,
      "loss": 0.06229257583618164,
      "step": 2928
    },
    {
      "epoch": 0.39597803126320236,
      "grad_norm": 0.372574120759964,
      "learning_rate": 2.1771714665460005e-05,
      "loss": 0.07358336448669434,
      "step": 2929
    },
    {
      "epoch": 0.39611322348964934,
      "grad_norm": 0.27585744857788086,
      "learning_rate": 2.1765606693581857e-05,
      "loss": 0.0627450942993164,
      "step": 2930
    },
    {
      "epoch": 0.3962484157160963,
      "grad_norm": 0.3051319420337677,
      "learning_rate": 2.1759497313068316e-05,
      "loss": 0.09266376495361328,
      "step": 2931
    },
    {
      "epoch": 0.3963836079425433,
      "grad_norm": 0.421512633562088,
      "learning_rate": 2.175338652519139e-05,
      "loss": 0.07506465911865234,
      "step": 2932
    },
    {
      "epoch": 0.39651880016899027,
      "grad_norm": 0.4446544051170349,
      "learning_rate": 2.1747274331223377e-05,
      "loss": 0.07260560989379883,
      "step": 2933
    },
    {
      "epoch": 0.39665399239543725,
      "grad_norm": 0.2291414588689804,
      "learning_rate": 2.1741160732436865e-05,
      "loss": 0.06614303588867188,
      "step": 2934
    },
    {
      "epoch": 0.39678918462188423,
      "grad_norm": 0.34969449043273926,
      "learning_rate": 2.1735045730104746e-05,
      "loss": 0.10336685180664062,
      "step": 2935
    },
    {
      "epoch": 0.3969243768483312,
      "grad_norm": 0.22155889868736267,
      "learning_rate": 2.1728929325500183e-05,
      "loss": 0.046473026275634766,
      "step": 2936
    },
    {
      "epoch": 0.3970595690747782,
      "grad_norm": 0.3308123052120209,
      "learning_rate": 2.1722811519896654e-05,
      "loss": 0.061977386474609375,
      "step": 2937
    },
    {
      "epoch": 0.39719476130122516,
      "grad_norm": 0.3256804943084717,
      "learning_rate": 2.171669231456792e-05,
      "loss": 0.06390666961669922,
      "step": 2938
    },
    {
      "epoch": 0.39732995352767214,
      "grad_norm": 0.35173648595809937,
      "learning_rate": 2.1710571710788025e-05,
      "loss": 0.07680559158325195,
      "step": 2939
    },
    {
      "epoch": 0.3974651457541191,
      "grad_norm": 0.36809730529785156,
      "learning_rate": 2.1704449709831312e-05,
      "loss": 0.040982723236083984,
      "step": 2940
    },
    {
      "epoch": 0.3976003379805661,
      "grad_norm": 0.29116567969322205,
      "learning_rate": 2.1698326312972423e-05,
      "loss": 0.05270671844482422,
      "step": 2941
    },
    {
      "epoch": 0.3977355302070131,
      "grad_norm": 0.16820156574249268,
      "learning_rate": 2.1692201521486268e-05,
      "loss": 0.0484539270401001,
      "step": 2942
    },
    {
      "epoch": 0.39787072243346006,
      "grad_norm": 0.29799380898475647,
      "learning_rate": 2.1686075336648075e-05,
      "loss": 0.07322835922241211,
      "step": 2943
    },
    {
      "epoch": 0.39800591465990703,
      "grad_norm": 0.43065980076789856,
      "learning_rate": 2.167994775973334e-05,
      "loss": 0.06460905075073242,
      "step": 2944
    },
    {
      "epoch": 0.398141106886354,
      "grad_norm": 0.37359461188316345,
      "learning_rate": 2.167381879201786e-05,
      "loss": 0.06734895706176758,
      "step": 2945
    },
    {
      "epoch": 0.398276299112801,
      "grad_norm": 0.45527586340904236,
      "learning_rate": 2.166768843477772e-05,
      "loss": 0.06209278106689453,
      "step": 2946
    },
    {
      "epoch": 0.39841149133924797,
      "grad_norm": 0.3737458288669586,
      "learning_rate": 2.166155668928929e-05,
      "loss": 0.06509113311767578,
      "step": 2947
    },
    {
      "epoch": 0.39854668356569495,
      "grad_norm": 0.2930571734905243,
      "learning_rate": 2.1655423556829233e-05,
      "loss": 0.04776430130004883,
      "step": 2948
    },
    {
      "epoch": 0.3986818757921419,
      "grad_norm": 0.2965002954006195,
      "learning_rate": 2.1649289038674504e-05,
      "loss": 0.07193899154663086,
      "step": 2949
    },
    {
      "epoch": 0.3988170680185889,
      "grad_norm": 0.20768171548843384,
      "learning_rate": 2.1643153136102333e-05,
      "loss": 0.05431318283081055,
      "step": 2950
    },
    {
      "epoch": 0.3989522602450359,
      "grad_norm": 0.2334291785955429,
      "learning_rate": 2.1637015850390255e-05,
      "loss": 0.05737948417663574,
      "step": 2951
    },
    {
      "epoch": 0.3990874524714829,
      "grad_norm": 0.27689775824546814,
      "learning_rate": 2.1630877182816087e-05,
      "loss": 0.06384515762329102,
      "step": 2952
    },
    {
      "epoch": 0.3992226446979299,
      "grad_norm": 0.3566146790981293,
      "learning_rate": 2.162473713465793e-05,
      "loss": 0.07159090042114258,
      "step": 2953
    },
    {
      "epoch": 0.3993578369243769,
      "grad_norm": 0.34134575724601746,
      "learning_rate": 2.161859570719417e-05,
      "loss": 0.07120990753173828,
      "step": 2954
    },
    {
      "epoch": 0.39949302915082385,
      "grad_norm": 0.49748843908309937,
      "learning_rate": 2.161245290170349e-05,
      "loss": 0.10651493072509766,
      "step": 2955
    },
    {
      "epoch": 0.39962822137727083,
      "grad_norm": 0.3313961327075958,
      "learning_rate": 2.1606308719464858e-05,
      "loss": 0.062203407287597656,
      "step": 2956
    },
    {
      "epoch": 0.3997634136037178,
      "grad_norm": 0.304697722196579,
      "learning_rate": 2.160016316175752e-05,
      "loss": 0.06562328338623047,
      "step": 2957
    },
    {
      "epoch": 0.3998986058301648,
      "grad_norm": 0.20526127517223358,
      "learning_rate": 2.159401622986101e-05,
      "loss": 0.05479693412780762,
      "step": 2958
    },
    {
      "epoch": 0.40003379805661177,
      "grad_norm": 0.1999506652355194,
      "learning_rate": 2.1587867925055165e-05,
      "loss": 0.05121326446533203,
      "step": 2959
    },
    {
      "epoch": 0.40016899028305875,
      "grad_norm": 0.3261958360671997,
      "learning_rate": 2.158171824862008e-05,
      "loss": 0.06938362121582031,
      "step": 2960
    },
    {
      "epoch": 0.4003041825095057,
      "grad_norm": 0.538030743598938,
      "learning_rate": 2.157556720183616e-05,
      "loss": 0.0921485424041748,
      "step": 2961
    },
    {
      "epoch": 0.4004393747359527,
      "grad_norm": 0.28004932403564453,
      "learning_rate": 2.156941478598409e-05,
      "loss": 0.05296158790588379,
      "step": 2962
    },
    {
      "epoch": 0.4005745669623997,
      "grad_norm": 0.2439052164554596,
      "learning_rate": 2.156326100234482e-05,
      "loss": 0.05390024185180664,
      "step": 2963
    },
    {
      "epoch": 0.40070975918884666,
      "grad_norm": 0.30761170387268066,
      "learning_rate": 2.1557105852199612e-05,
      "loss": 0.0789484977722168,
      "step": 2964
    },
    {
      "epoch": 0.40084495141529364,
      "grad_norm": 0.45547664165496826,
      "learning_rate": 2.155094933683e-05,
      "loss": 0.07648730278015137,
      "step": 2965
    },
    {
      "epoch": 0.4009801436417406,
      "grad_norm": 0.22745536267757416,
      "learning_rate": 2.1544791457517802e-05,
      "loss": 0.05563545227050781,
      "step": 2966
    },
    {
      "epoch": 0.4011153358681876,
      "grad_norm": 0.2178754061460495,
      "learning_rate": 2.1538632215545126e-05,
      "loss": 0.05500507354736328,
      "step": 2967
    },
    {
      "epoch": 0.4012505280946346,
      "grad_norm": 0.34506499767303467,
      "learning_rate": 2.153247161219435e-05,
      "loss": 0.061418771743774414,
      "step": 2968
    },
    {
      "epoch": 0.40138572032108155,
      "grad_norm": 0.6026394963264465,
      "learning_rate": 2.1526309648748147e-05,
      "loss": 0.08974194526672363,
      "step": 2969
    },
    {
      "epoch": 0.40152091254752853,
      "grad_norm": 0.2664487361907959,
      "learning_rate": 2.1520146326489476e-05,
      "loss": 0.07912063598632812,
      "step": 2970
    },
    {
      "epoch": 0.4016561047739755,
      "grad_norm": 0.39691248536109924,
      "learning_rate": 2.151398164670157e-05,
      "loss": 0.08194828033447266,
      "step": 2971
    },
    {
      "epoch": 0.4017912970004225,
      "grad_norm": 0.39125579595565796,
      "learning_rate": 2.1507815610667948e-05,
      "loss": 0.056514739990234375,
      "step": 2972
    },
    {
      "epoch": 0.40192648922686947,
      "grad_norm": 0.34392523765563965,
      "learning_rate": 2.1501648219672407e-05,
      "loss": 0.06239795684814453,
      "step": 2973
    },
    {
      "epoch": 0.40206168145331644,
      "grad_norm": 0.6169334650039673,
      "learning_rate": 2.149547947499904e-05,
      "loss": 0.09360408782958984,
      "step": 2974
    },
    {
      "epoch": 0.4021968736797634,
      "grad_norm": 0.5909185409545898,
      "learning_rate": 2.1489309377932212e-05,
      "loss": 0.07928919792175293,
      "step": 2975
    },
    {
      "epoch": 0.4023320659062104,
      "grad_norm": 0.5892732739448547,
      "learning_rate": 2.1483137929756562e-05,
      "loss": 0.08875560760498047,
      "step": 2976
    },
    {
      "epoch": 0.4024672581326574,
      "grad_norm": 0.3181554973125458,
      "learning_rate": 2.147696513175702e-05,
      "loss": 0.06846928596496582,
      "step": 2977
    },
    {
      "epoch": 0.40260245035910436,
      "grad_norm": 0.25439995527267456,
      "learning_rate": 2.1470790985218804e-05,
      "loss": 0.05667448043823242,
      "step": 2978
    },
    {
      "epoch": 0.40273764258555134,
      "grad_norm": 0.40789681673049927,
      "learning_rate": 2.1464615491427393e-05,
      "loss": 0.05379605293273926,
      "step": 2979
    },
    {
      "epoch": 0.4028728348119983,
      "grad_norm": 0.16324302554130554,
      "learning_rate": 2.1458438651668567e-05,
      "loss": 0.04113578796386719,
      "step": 2980
    },
    {
      "epoch": 0.4030080270384453,
      "grad_norm": 0.44761011004447937,
      "learning_rate": 2.1452260467228376e-05,
      "loss": 0.04816389083862305,
      "step": 2981
    },
    {
      "epoch": 0.40314321926489227,
      "grad_norm": 1.3353536128997803,
      "learning_rate": 2.144608093939314e-05,
      "loss": 0.14235496520996094,
      "step": 2982
    },
    {
      "epoch": 0.40327841149133925,
      "grad_norm": 0.37307024002075195,
      "learning_rate": 2.1439900069449483e-05,
      "loss": 0.05567169189453125,
      "step": 2983
    },
    {
      "epoch": 0.40341360371778623,
      "grad_norm": 0.25737079977989197,
      "learning_rate": 2.1433717858684286e-05,
      "loss": 0.06741142272949219,
      "step": 2984
    },
    {
      "epoch": 0.4035487959442332,
      "grad_norm": 0.4321408271789551,
      "learning_rate": 2.1427534308384724e-05,
      "loss": 0.07166290283203125,
      "step": 2985
    },
    {
      "epoch": 0.4036839881706802,
      "grad_norm": 0.4518645703792572,
      "learning_rate": 2.1421349419838245e-05,
      "loss": 0.09037590026855469,
      "step": 2986
    },
    {
      "epoch": 0.40381918039712716,
      "grad_norm": 0.391391396522522,
      "learning_rate": 2.1415163194332574e-05,
      "loss": 0.08176469802856445,
      "step": 2987
    },
    {
      "epoch": 0.40395437262357414,
      "grad_norm": 0.27435001730918884,
      "learning_rate": 2.1408975633155715e-05,
      "loss": 0.07512962818145752,
      "step": 2988
    },
    {
      "epoch": 0.4040895648500211,
      "grad_norm": 0.4362889230251312,
      "learning_rate": 2.140278673759595e-05,
      "loss": 0.0748453140258789,
      "step": 2989
    },
    {
      "epoch": 0.4042247570764681,
      "grad_norm": 0.656727135181427,
      "learning_rate": 2.1396596508941847e-05,
      "loss": 0.0941476821899414,
      "step": 2990
    },
    {
      "epoch": 0.4043599493029151,
      "grad_norm": 0.34294772148132324,
      "learning_rate": 2.1390404948482238e-05,
      "loss": 0.0550762414932251,
      "step": 2991
    },
    {
      "epoch": 0.40449514152936206,
      "grad_norm": 0.2868243157863617,
      "learning_rate": 2.1384212057506243e-05,
      "loss": 0.06558704376220703,
      "step": 2992
    },
    {
      "epoch": 0.40463033375580904,
      "grad_norm": 0.18572582304477692,
      "learning_rate": 2.137801783730325e-05,
      "loss": 0.059659481048583984,
      "step": 2993
    },
    {
      "epoch": 0.404765525982256,
      "grad_norm": 0.3745408058166504,
      "learning_rate": 2.137182228916293e-05,
      "loss": 0.0683131217956543,
      "step": 2994
    },
    {
      "epoch": 0.404900718208703,
      "grad_norm": 0.2456149309873581,
      "learning_rate": 2.136562541437523e-05,
      "loss": 0.05441570281982422,
      "step": 2995
    },
    {
      "epoch": 0.40503591043514997,
      "grad_norm": 0.18396519124507904,
      "learning_rate": 2.135942721423038e-05,
      "loss": 0.0449833869934082,
      "step": 2996
    },
    {
      "epoch": 0.40517110266159695,
      "grad_norm": 0.3149019777774811,
      "learning_rate": 2.1353227690018865e-05,
      "loss": 0.059523582458496094,
      "step": 2997
    },
    {
      "epoch": 0.4053062948880439,
      "grad_norm": 0.3205578625202179,
      "learning_rate": 2.1347026843031467e-05,
      "loss": 0.07893157005310059,
      "step": 2998
    },
    {
      "epoch": 0.4054414871144909,
      "grad_norm": 0.6945152282714844,
      "learning_rate": 2.1340824674559238e-05,
      "loss": 0.08907890319824219,
      "step": 2999
    },
    {
      "epoch": 0.4055766793409379,
      "grad_norm": 0.2924824357032776,
      "learning_rate": 2.133462118589349e-05,
      "loss": 0.06456422805786133,
      "step": 3000
    },
    {
      "epoch": 0.40571187156738486,
      "grad_norm": 0.13423191010951996,
      "learning_rate": 2.1328416378325837e-05,
      "loss": 0.03351247310638428,
      "step": 3001
    },
    {
      "epoch": 0.40584706379383184,
      "grad_norm": 0.2060016691684723,
      "learning_rate": 2.1322210253148144e-05,
      "loss": 0.04142570495605469,
      "step": 3002
    },
    {
      "epoch": 0.4059822560202788,
      "grad_norm": 0.20809145271778107,
      "learning_rate": 2.131600281165257e-05,
      "loss": 0.04394388198852539,
      "step": 3003
    },
    {
      "epoch": 0.4061174482467258,
      "grad_norm": 0.3027481734752655,
      "learning_rate": 2.130979405513152e-05,
      "loss": 0.051453351974487305,
      "step": 3004
    },
    {
      "epoch": 0.4062526404731728,
      "grad_norm": 0.40413138270378113,
      "learning_rate": 2.1303583984877697e-05,
      "loss": 0.07988333702087402,
      "step": 3005
    },
    {
      "epoch": 0.40638783269961976,
      "grad_norm": 0.24105243384838104,
      "learning_rate": 2.1297372602184085e-05,
      "loss": 0.05415964126586914,
      "step": 3006
    },
    {
      "epoch": 0.40652302492606673,
      "grad_norm": 0.597404956817627,
      "learning_rate": 2.1291159908343907e-05,
      "loss": 0.06812691688537598,
      "step": 3007
    },
    {
      "epoch": 0.4066582171525137,
      "grad_norm": 0.4122142493724823,
      "learning_rate": 2.1284945904650693e-05,
      "loss": 0.0854644775390625,
      "step": 3008
    },
    {
      "epoch": 0.4067934093789607,
      "grad_norm": 0.3772042691707611,
      "learning_rate": 2.127873059239822e-05,
      "loss": 0.08576107025146484,
      "step": 3009
    },
    {
      "epoch": 0.40692860160540767,
      "grad_norm": 0.17306067049503326,
      "learning_rate": 2.127251397288056e-05,
      "loss": 0.046257853507995605,
      "step": 3010
    },
    {
      "epoch": 0.40706379383185465,
      "grad_norm": 0.3024364709854126,
      "learning_rate": 2.126629604739204e-05,
      "loss": 0.06950187683105469,
      "step": 3011
    },
    {
      "epoch": 0.4071989860583016,
      "grad_norm": 0.2842211425304413,
      "learning_rate": 2.1260076817227268e-05,
      "loss": 0.07561922073364258,
      "step": 3012
    },
    {
      "epoch": 0.4073341782847486,
      "grad_norm": 0.29641711711883545,
      "learning_rate": 2.1253856283681122e-05,
      "loss": 0.05509376525878906,
      "step": 3013
    },
    {
      "epoch": 0.4074693705111956,
      "grad_norm": 0.3248545527458191,
      "learning_rate": 2.1247634448048743e-05,
      "loss": 0.0809168815612793,
      "step": 3014
    },
    {
      "epoch": 0.40760456273764256,
      "grad_norm": 0.18395265936851501,
      "learning_rate": 2.1241411311625562e-05,
      "loss": 0.03696107864379883,
      "step": 3015
    },
    {
      "epoch": 0.40773975496408954,
      "grad_norm": 0.4544692933559418,
      "learning_rate": 2.1235186875707257e-05,
      "loss": 0.06669068336486816,
      "step": 3016
    },
    {
      "epoch": 0.4078749471905365,
      "grad_norm": 0.28700846433639526,
      "learning_rate": 2.1228961141589797e-05,
      "loss": 0.06910467147827148,
      "step": 3017
    },
    {
      "epoch": 0.4080101394169835,
      "grad_norm": 0.305080771446228,
      "learning_rate": 2.122273411056941e-05,
      "loss": 0.0659780502319336,
      "step": 3018
    },
    {
      "epoch": 0.4081453316434305,
      "grad_norm": 0.49413424730300903,
      "learning_rate": 2.1216505783942592e-05,
      "loss": 0.08311891555786133,
      "step": 3019
    },
    {
      "epoch": 0.4082805238698775,
      "grad_norm": 0.23321709036827087,
      "learning_rate": 2.121027616300613e-05,
      "loss": 0.06351089477539062,
      "step": 3020
    },
    {
      "epoch": 0.4084157160963245,
      "grad_norm": 0.45711737871170044,
      "learning_rate": 2.1204045249057043e-05,
      "loss": 0.09911465644836426,
      "step": 3021
    },
    {
      "epoch": 0.40855090832277147,
      "grad_norm": 0.24758939445018768,
      "learning_rate": 2.119781304339266e-05,
      "loss": 0.0674586296081543,
      "step": 3022
    },
    {
      "epoch": 0.40868610054921845,
      "grad_norm": 0.223337322473526,
      "learning_rate": 2.1191579547310547e-05,
      "loss": 0.06248021125793457,
      "step": 3023
    },
    {
      "epoch": 0.4088212927756654,
      "grad_norm": 0.41172075271606445,
      "learning_rate": 2.1185344762108556e-05,
      "loss": 0.06467294692993164,
      "step": 3024
    },
    {
      "epoch": 0.4089564850021124,
      "grad_norm": 0.22348926961421967,
      "learning_rate": 2.11791086890848e-05,
      "loss": 0.0626610517501831,
      "step": 3025
    },
    {
      "epoch": 0.4090916772285594,
      "grad_norm": 0.21511821448802948,
      "learning_rate": 2.1172871329537662e-05,
      "loss": 0.05102062225341797,
      "step": 3026
    },
    {
      "epoch": 0.40922686945500636,
      "grad_norm": 0.2693392336368561,
      "learning_rate": 2.1166632684765794e-05,
      "loss": 0.05055880546569824,
      "step": 3027
    },
    {
      "epoch": 0.40936206168145334,
      "grad_norm": 0.37640175223350525,
      "learning_rate": 2.1160392756068124e-05,
      "loss": 0.07921767234802246,
      "step": 3028
    },
    {
      "epoch": 0.4094972539079003,
      "grad_norm": 0.2543674409389496,
      "learning_rate": 2.1154151544743826e-05,
      "loss": 0.06111800670623779,
      "step": 3029
    },
    {
      "epoch": 0.4096324461343473,
      "grad_norm": 0.14939340949058533,
      "learning_rate": 2.114790905209236e-05,
      "loss": 0.04032135009765625,
      "step": 3030
    },
    {
      "epoch": 0.4097676383607943,
      "grad_norm": 0.500823438167572,
      "learning_rate": 2.1141665279413444e-05,
      "loss": 0.06637167930603027,
      "step": 3031
    },
    {
      "epoch": 0.40990283058724125,
      "grad_norm": 0.36694619059562683,
      "learning_rate": 2.1135420228007062e-05,
      "loss": 0.06368553638458252,
      "step": 3032
    },
    {
      "epoch": 0.41003802281368823,
      "grad_norm": 0.31600624322891235,
      "learning_rate": 2.1129173899173474e-05,
      "loss": 0.039308786392211914,
      "step": 3033
    },
    {
      "epoch": 0.4101732150401352,
      "grad_norm": 0.4579591155052185,
      "learning_rate": 2.11229262942132e-05,
      "loss": 0.09085893630981445,
      "step": 3034
    },
    {
      "epoch": 0.4103084072665822,
      "grad_norm": 0.47570300102233887,
      "learning_rate": 2.1116677414427008e-05,
      "loss": 0.1028285026550293,
      "step": 3035
    },
    {
      "epoch": 0.41044359949302917,
      "grad_norm": 0.3296443223953247,
      "learning_rate": 2.1110427261115972e-05,
      "loss": 0.09491825103759766,
      "step": 3036
    },
    {
      "epoch": 0.41057879171947614,
      "grad_norm": 0.2338595688343048,
      "learning_rate": 2.1104175835581386e-05,
      "loss": 0.07488346099853516,
      "step": 3037
    },
    {
      "epoch": 0.4107139839459231,
      "grad_norm": 0.2658068537712097,
      "learning_rate": 2.1097923139124846e-05,
      "loss": 0.06888961791992188,
      "step": 3038
    },
    {
      "epoch": 0.4108491761723701,
      "grad_norm": 0.24706095457077026,
      "learning_rate": 2.109166917304819e-05,
      "loss": 0.07400083541870117,
      "step": 3039
    },
    {
      "epoch": 0.4109843683988171,
      "grad_norm": 0.2957863509654999,
      "learning_rate": 2.1085413938653532e-05,
      "loss": 0.05295455455780029,
      "step": 3040
    },
    {
      "epoch": 0.41111956062526406,
      "grad_norm": 0.3177264332771301,
      "learning_rate": 2.107915743724323e-05,
      "loss": 0.06435346603393555,
      "step": 3041
    },
    {
      "epoch": 0.41125475285171104,
      "grad_norm": 0.2976820766925812,
      "learning_rate": 2.1072899670119935e-05,
      "loss": 0.0619511604309082,
      "step": 3042
    },
    {
      "epoch": 0.411389945078158,
      "grad_norm": 0.4411100149154663,
      "learning_rate": 2.1066640638586543e-05,
      "loss": 0.0731658935546875,
      "step": 3043
    },
    {
      "epoch": 0.411525137304605,
      "grad_norm": 0.1755773276090622,
      "learning_rate": 2.1060380343946223e-05,
      "loss": 0.031377315521240234,
      "step": 3044
    },
    {
      "epoch": 0.41166032953105197,
      "grad_norm": 0.3265551030635834,
      "learning_rate": 2.10541187875024e-05,
      "loss": 0.06280660629272461,
      "step": 3045
    },
    {
      "epoch": 0.41179552175749895,
      "grad_norm": 0.16660244762897491,
      "learning_rate": 2.1047855970558753e-05,
      "loss": 0.03179144859313965,
      "step": 3046
    },
    {
      "epoch": 0.41193071398394593,
      "grad_norm": 0.25354695320129395,
      "learning_rate": 2.1041591894419244e-05,
      "loss": 0.06399154663085938,
      "step": 3047
    },
    {
      "epoch": 0.4120659062103929,
      "grad_norm": 0.34272727370262146,
      "learning_rate": 2.1035326560388087e-05,
      "loss": 0.09914398193359375,
      "step": 3048
    },
    {
      "epoch": 0.4122010984368399,
      "grad_norm": 0.4404015839099884,
      "learning_rate": 2.1029059969769756e-05,
      "loss": 0.08698320388793945,
      "step": 3049
    },
    {
      "epoch": 0.41233629066328686,
      "grad_norm": 0.6310895085334778,
      "learning_rate": 2.1022792123868986e-05,
      "loss": 0.11017704010009766,
      "step": 3050
    },
    {
      "epoch": 0.41247148288973384,
      "grad_norm": 0.4720802307128906,
      "learning_rate": 2.1016523023990783e-05,
      "loss": 0.0863032341003418,
      "step": 3051
    },
    {
      "epoch": 0.4126066751161808,
      "grad_norm": 0.15391072630882263,
      "learning_rate": 2.1010252671440398e-05,
      "loss": 0.03853416442871094,
      "step": 3052
    },
    {
      "epoch": 0.4127418673426278,
      "grad_norm": 0.49569839239120483,
      "learning_rate": 2.1003981067523358e-05,
      "loss": 0.07751250267028809,
      "step": 3053
    },
    {
      "epoch": 0.4128770595690748,
      "grad_norm": 0.2507389187812805,
      "learning_rate": 2.099770821354544e-05,
      "loss": 0.05987739562988281,
      "step": 3054
    },
    {
      "epoch": 0.41301225179552176,
      "grad_norm": 0.286739319562912,
      "learning_rate": 2.0991434110812692e-05,
      "loss": 0.050983428955078125,
      "step": 3055
    },
    {
      "epoch": 0.41314744402196873,
      "grad_norm": 0.6455079913139343,
      "learning_rate": 2.0985158760631415e-05,
      "loss": 0.09912776947021484,
      "step": 3056
    },
    {
      "epoch": 0.4132826362484157,
      "grad_norm": 0.6160709857940674,
      "learning_rate": 2.0978882164308157e-05,
      "loss": 0.0709223747253418,
      "step": 3057
    },
    {
      "epoch": 0.4134178284748627,
      "grad_norm": 0.4100304841995239,
      "learning_rate": 2.0972604323149755e-05,
      "loss": 0.08180618286132812,
      "step": 3058
    },
    {
      "epoch": 0.41355302070130967,
      "grad_norm": 0.34018391370773315,
      "learning_rate": 2.0966325238463283e-05,
      "loss": 0.08721542358398438,
      "step": 3059
    },
    {
      "epoch": 0.41368821292775665,
      "grad_norm": 0.19317318499088287,
      "learning_rate": 2.096004491155608e-05,
      "loss": 0.06745672225952148,
      "step": 3060
    },
    {
      "epoch": 0.4138234051542036,
      "grad_norm": 0.3779497444629669,
      "learning_rate": 2.0953763343735746e-05,
      "loss": 0.08286094665527344,
      "step": 3061
    },
    {
      "epoch": 0.4139585973806506,
      "grad_norm": 0.31215447187423706,
      "learning_rate": 2.0947480536310133e-05,
      "loss": 0.05769228935241699,
      "step": 3062
    },
    {
      "epoch": 0.4140937896070976,
      "grad_norm": 0.47123974561691284,
      "learning_rate": 2.0941196490587352e-05,
      "loss": 0.08304882049560547,
      "step": 3063
    },
    {
      "epoch": 0.41422898183354456,
      "grad_norm": 0.28059065341949463,
      "learning_rate": 2.0934911207875782e-05,
      "loss": 0.0818033218383789,
      "step": 3064
    },
    {
      "epoch": 0.41436417405999154,
      "grad_norm": 0.48526760935783386,
      "learning_rate": 2.092862468948405e-05,
      "loss": 0.09539604187011719,
      "step": 3065
    },
    {
      "epoch": 0.4144993662864385,
      "grad_norm": 0.21677564084529877,
      "learning_rate": 2.0922336936721044e-05,
      "loss": 0.0581052303314209,
      "step": 3066
    },
    {
      "epoch": 0.4146345585128855,
      "grad_norm": 0.21911999583244324,
      "learning_rate": 2.0916047950895907e-05,
      "loss": 0.05284690856933594,
      "step": 3067
    },
    {
      "epoch": 0.4147697507393325,
      "grad_norm": 0.523078441619873,
      "learning_rate": 2.0909757733318035e-05,
      "loss": 0.07018661499023438,
      "step": 3068
    },
    {
      "epoch": 0.41490494296577946,
      "grad_norm": 0.2608419358730316,
      "learning_rate": 2.090346628529709e-05,
      "loss": 0.06709098815917969,
      "step": 3069
    },
    {
      "epoch": 0.41504013519222643,
      "grad_norm": 0.4290355443954468,
      "learning_rate": 2.089717360814298e-05,
      "loss": 0.09755420684814453,
      "step": 3070
    },
    {
      "epoch": 0.4151753274186734,
      "grad_norm": 0.209347203373909,
      "learning_rate": 2.0890879703165885e-05,
      "loss": 0.055274009704589844,
      "step": 3071
    },
    {
      "epoch": 0.4153105196451204,
      "grad_norm": 0.4696345925331116,
      "learning_rate": 2.0884584571676217e-05,
      "loss": 0.0701138973236084,
      "step": 3072
    },
    {
      "epoch": 0.41544571187156737,
      "grad_norm": 0.29664644598960876,
      "learning_rate": 2.0878288214984657e-05,
      "loss": 0.06930828094482422,
      "step": 3073
    },
    {
      "epoch": 0.41558090409801435,
      "grad_norm": 0.22433871030807495,
      "learning_rate": 2.0871990634402147e-05,
      "loss": 0.0613095760345459,
      "step": 3074
    },
    {
      "epoch": 0.4157160963244613,
      "grad_norm": 0.31830698251724243,
      "learning_rate": 2.0865691831239877e-05,
      "loss": 0.066925048828125,
      "step": 3075
    },
    {
      "epoch": 0.4158512885509083,
      "grad_norm": 0.4611347019672394,
      "learning_rate": 2.0859391806809285e-05,
      "loss": 0.08638906478881836,
      "step": 3076
    },
    {
      "epoch": 0.4159864807773553,
      "grad_norm": 0.3555915951728821,
      "learning_rate": 2.0853090562422072e-05,
      "loss": 0.06160902976989746,
      "step": 3077
    },
    {
      "epoch": 0.41612167300380226,
      "grad_norm": 0.20421913266181946,
      "learning_rate": 2.084678809939019e-05,
      "loss": 0.05112624168395996,
      "step": 3078
    },
    {
      "epoch": 0.41625686523024924,
      "grad_norm": 0.5702481269836426,
      "learning_rate": 2.084048441902585e-05,
      "loss": 0.08750391006469727,
      "step": 3079
    },
    {
      "epoch": 0.4163920574566962,
      "grad_norm": 0.6666382551193237,
      "learning_rate": 2.0834179522641508e-05,
      "loss": 0.08485496044158936,
      "step": 3080
    },
    {
      "epoch": 0.4165272496831432,
      "grad_norm": 0.6690225005149841,
      "learning_rate": 2.0827873411549877e-05,
      "loss": 0.10233592987060547,
      "step": 3081
    },
    {
      "epoch": 0.4166624419095902,
      "grad_norm": 0.49933552742004395,
      "learning_rate": 2.0821566087063926e-05,
      "loss": 0.09810972213745117,
      "step": 3082
    },
    {
      "epoch": 0.41679763413603715,
      "grad_norm": 0.6558201909065247,
      "learning_rate": 2.081525755049687e-05,
      "loss": 0.08875274658203125,
      "step": 3083
    },
    {
      "epoch": 0.41693282636248413,
      "grad_norm": 0.506667971611023,
      "learning_rate": 2.0808947803162182e-05,
      "loss": 0.05261152982711792,
      "step": 3084
    },
    {
      "epoch": 0.4170680185889311,
      "grad_norm": 0.7671221494674683,
      "learning_rate": 2.0802636846373578e-05,
      "loss": 0.06038618087768555,
      "step": 3085
    },
    {
      "epoch": 0.4172032108153781,
      "grad_norm": 0.3433699309825897,
      "learning_rate": 2.0796324681445045e-05,
      "loss": 0.10524749755859375,
      "step": 3086
    },
    {
      "epoch": 0.41733840304182507,
      "grad_norm": 0.2865079343318939,
      "learning_rate": 2.0790011309690806e-05,
      "loss": 0.053057193756103516,
      "step": 3087
    },
    {
      "epoch": 0.4174735952682721,
      "grad_norm": 0.28365615010261536,
      "learning_rate": 2.0783696732425332e-05,
      "loss": 0.09075212478637695,
      "step": 3088
    },
    {
      "epoch": 0.4176087874947191,
      "grad_norm": 0.1373298317193985,
      "learning_rate": 2.0777380950963355e-05,
      "loss": 0.03651416301727295,
      "step": 3089
    },
    {
      "epoch": 0.41774397972116606,
      "grad_norm": 0.5920847058296204,
      "learning_rate": 2.0771063966619854e-05,
      "loss": 0.08390545845031738,
      "step": 3090
    },
    {
      "epoch": 0.41787917194761304,
      "grad_norm": 1.0154945850372314,
      "learning_rate": 2.0764745780710065e-05,
      "loss": 0.11055469512939453,
      "step": 3091
    },
    {
      "epoch": 0.41801436417406,
      "grad_norm": 0.23824970424175262,
      "learning_rate": 2.075842639454946e-05,
      "loss": 0.04407072067260742,
      "step": 3092
    },
    {
      "epoch": 0.418149556400507,
      "grad_norm": 0.49458593130111694,
      "learning_rate": 2.075210580945378e-05,
      "loss": 0.06061124801635742,
      "step": 3093
    },
    {
      "epoch": 0.418284748626954,
      "grad_norm": 0.32972678542137146,
      "learning_rate": 2.0745784026738984e-05,
      "loss": 0.06857776641845703,
      "step": 3094
    },
    {
      "epoch": 0.41841994085340095,
      "grad_norm": 0.3052026629447937,
      "learning_rate": 2.073946104772132e-05,
      "loss": 0.0900566577911377,
      "step": 3095
    },
    {
      "epoch": 0.41855513307984793,
      "grad_norm": 0.36622339487075806,
      "learning_rate": 2.0733136873717258e-05,
      "loss": 0.05018901824951172,
      "step": 3096
    },
    {
      "epoch": 0.4186903253062949,
      "grad_norm": 0.2580163776874542,
      "learning_rate": 2.0726811506043527e-05,
      "loss": 0.07453489303588867,
      "step": 3097
    },
    {
      "epoch": 0.4188255175327419,
      "grad_norm": 0.3222690224647522,
      "learning_rate": 2.0720484946017104e-05,
      "loss": 0.06111764907836914,
      "step": 3098
    },
    {
      "epoch": 0.41896070975918887,
      "grad_norm": 0.3600582480430603,
      "learning_rate": 2.0714157194955202e-05,
      "loss": 0.06430578231811523,
      "step": 3099
    },
    {
      "epoch": 0.41909590198563584,
      "grad_norm": 0.34537774324417114,
      "learning_rate": 2.070782825417531e-05,
      "loss": 0.05692088603973389,
      "step": 3100
    },
    {
      "epoch": 0.4192310942120828,
      "grad_norm": 0.4973081946372986,
      "learning_rate": 2.0701498124995127e-05,
      "loss": 0.0888824462890625,
      "step": 3101
    },
    {
      "epoch": 0.4193662864385298,
      "grad_norm": 0.31936827301979065,
      "learning_rate": 2.069516680873264e-05,
      "loss": 0.07187843322753906,
      "step": 3102
    },
    {
      "epoch": 0.4195014786649768,
      "grad_norm": 0.2811417579650879,
      "learning_rate": 2.0688834306706047e-05,
      "loss": 0.06684207916259766,
      "step": 3103
    },
    {
      "epoch": 0.41963667089142376,
      "grad_norm": 0.3405231833457947,
      "learning_rate": 2.0682500620233815e-05,
      "loss": 0.0851449966430664,
      "step": 3104
    },
    {
      "epoch": 0.41977186311787074,
      "grad_norm": 0.26490673422813416,
      "learning_rate": 2.0676165750634656e-05,
      "loss": 0.03852987289428711,
      "step": 3105
    },
    {
      "epoch": 0.4199070553443177,
      "grad_norm": 0.18394184112548828,
      "learning_rate": 2.0669829699227513e-05,
      "loss": 0.04652690887451172,
      "step": 3106
    },
    {
      "epoch": 0.4200422475707647,
      "grad_norm": 0.40319758653640747,
      "learning_rate": 2.06634924673316e-05,
      "loss": 0.06999492645263672,
      "step": 3107
    },
    {
      "epoch": 0.42017743979721167,
      "grad_norm": 0.4493744671344757,
      "learning_rate": 2.0657154056266346e-05,
      "loss": 0.09916114807128906,
      "step": 3108
    },
    {
      "epoch": 0.42031263202365865,
      "grad_norm": 0.7170290946960449,
      "learning_rate": 2.0650814467351452e-05,
      "loss": 0.063210129737854,
      "step": 3109
    },
    {
      "epoch": 0.42044782425010563,
      "grad_norm": 0.2866007685661316,
      "learning_rate": 2.064447370190685e-05,
      "loss": 0.04627275466918945,
      "step": 3110
    },
    {
      "epoch": 0.4205830164765526,
      "grad_norm": 0.39962875843048096,
      "learning_rate": 2.0638131761252724e-05,
      "loss": 0.07088196277618408,
      "step": 3111
    },
    {
      "epoch": 0.4207182087029996,
      "grad_norm": 0.5062935948371887,
      "learning_rate": 2.06317886467095e-05,
      "loss": 0.09743881225585938,
      "step": 3112
    },
    {
      "epoch": 0.42085340092944656,
      "grad_norm": 0.18093198537826538,
      "learning_rate": 2.0625444359597847e-05,
      "loss": 0.04445075988769531,
      "step": 3113
    },
    {
      "epoch": 0.42098859315589354,
      "grad_norm": 0.7550981640815735,
      "learning_rate": 2.0619098901238684e-05,
      "loss": 0.07025456428527832,
      "step": 3114
    },
    {
      "epoch": 0.4211237853823405,
      "grad_norm": 0.3525427281856537,
      "learning_rate": 2.0612752272953158e-05,
      "loss": 0.05990409851074219,
      "step": 3115
    },
    {
      "epoch": 0.4212589776087875,
      "grad_norm": 0.48233169317245483,
      "learning_rate": 2.060640447606268e-05,
      "loss": 0.0691070556640625,
      "step": 3116
    },
    {
      "epoch": 0.4213941698352345,
      "grad_norm": 0.37629425525665283,
      "learning_rate": 2.0600055511888895e-05,
      "loss": 0.06235241889953613,
      "step": 3117
    },
    {
      "epoch": 0.42152936206168146,
      "grad_norm": 0.38955265283584595,
      "learning_rate": 2.059370538175369e-05,
      "loss": 0.07288360595703125,
      "step": 3118
    },
    {
      "epoch": 0.42166455428812843,
      "grad_norm": 0.3728174567222595,
      "learning_rate": 2.0587354086979194e-05,
      "loss": 0.07150459289550781,
      "step": 3119
    },
    {
      "epoch": 0.4217997465145754,
      "grad_norm": 0.22158977389335632,
      "learning_rate": 2.0581001628887785e-05,
      "loss": 0.05391716957092285,
      "step": 3120
    },
    {
      "epoch": 0.4219349387410224,
      "grad_norm": 0.5192732214927673,
      "learning_rate": 2.057464800880207e-05,
      "loss": 0.08232378959655762,
      "step": 3121
    },
    {
      "epoch": 0.42207013096746937,
      "grad_norm": 0.3635179102420807,
      "learning_rate": 2.0568293228044914e-05,
      "loss": 0.07523679733276367,
      "step": 3122
    },
    {
      "epoch": 0.42220532319391635,
      "grad_norm": 0.16220387816429138,
      "learning_rate": 2.0561937287939413e-05,
      "loss": 0.04125869274139404,
      "step": 3123
    },
    {
      "epoch": 0.4223405154203633,
      "grad_norm": 0.23348811268806458,
      "learning_rate": 2.055558018980891e-05,
      "loss": 0.0638723373413086,
      "step": 3124
    },
    {
      "epoch": 0.4224757076468103,
      "grad_norm": 0.334791362285614,
      "learning_rate": 2.0549221934976987e-05,
      "loss": 0.08350467681884766,
      "step": 3125
    },
    {
      "epoch": 0.4226108998732573,
      "grad_norm": 0.3081171214580536,
      "learning_rate": 2.054286252476746e-05,
      "loss": 0.07777190208435059,
      "step": 3126
    },
    {
      "epoch": 0.42274609209970426,
      "grad_norm": 0.3021962344646454,
      "learning_rate": 2.05365019605044e-05,
      "loss": 0.07677555084228516,
      "step": 3127
    },
    {
      "epoch": 0.42288128432615124,
      "grad_norm": 0.2252826690673828,
      "learning_rate": 2.053014024351211e-05,
      "loss": 0.06123685836791992,
      "step": 3128
    },
    {
      "epoch": 0.4230164765525982,
      "grad_norm": 0.17544758319854736,
      "learning_rate": 2.0523777375115133e-05,
      "loss": 0.047986507415771484,
      "step": 3129
    },
    {
      "epoch": 0.4231516687790452,
      "grad_norm": 0.25181517004966736,
      "learning_rate": 2.0517413356638245e-05,
      "loss": 0.04684591293334961,
      "step": 3130
    },
    {
      "epoch": 0.4232868610054922,
      "grad_norm": 0.6005727648735046,
      "learning_rate": 2.0511048189406472e-05,
      "loss": 0.10750675201416016,
      "step": 3131
    },
    {
      "epoch": 0.42342205323193916,
      "grad_norm": 0.4774467349052429,
      "learning_rate": 2.0504681874745082e-05,
      "loss": 0.0865640640258789,
      "step": 3132
    },
    {
      "epoch": 0.42355724545838613,
      "grad_norm": 0.33371490240097046,
      "learning_rate": 2.049831441397957e-05,
      "loss": 0.07904577255249023,
      "step": 3133
    },
    {
      "epoch": 0.4236924376848331,
      "grad_norm": 0.6800884008407593,
      "learning_rate": 2.0491945808435674e-05,
      "loss": 0.08980846405029297,
      "step": 3134
    },
    {
      "epoch": 0.4238276299112801,
      "grad_norm": 0.2587440609931946,
      "learning_rate": 2.048557605943938e-05,
      "loss": 0.03163552284240723,
      "step": 3135
    },
    {
      "epoch": 0.42396282213772707,
      "grad_norm": 0.28416770696640015,
      "learning_rate": 2.047920516831689e-05,
      "loss": 0.060605525970458984,
      "step": 3136
    },
    {
      "epoch": 0.42409801436417405,
      "grad_norm": 0.37363213300704956,
      "learning_rate": 2.047283313639467e-05,
      "loss": 0.10064458847045898,
      "step": 3137
    },
    {
      "epoch": 0.424233206590621,
      "grad_norm": 0.32091838121414185,
      "learning_rate": 2.0466459964999408e-05,
      "loss": 0.04574716091156006,
      "step": 3138
    },
    {
      "epoch": 0.424368398817068,
      "grad_norm": 0.3039965033531189,
      "learning_rate": 2.0460085655458025e-05,
      "loss": 0.05404996871948242,
      "step": 3139
    },
    {
      "epoch": 0.424503591043515,
      "grad_norm": 0.3287688195705414,
      "learning_rate": 2.0453710209097697e-05,
      "loss": 0.08441638946533203,
      "step": 3140
    },
    {
      "epoch": 0.42463878326996196,
      "grad_norm": 0.2386762499809265,
      "learning_rate": 2.044733362724582e-05,
      "loss": 0.06625723838806152,
      "step": 3141
    },
    {
      "epoch": 0.42477397549640894,
      "grad_norm": 0.21166974306106567,
      "learning_rate": 2.0440955911230028e-05,
      "loss": 0.06540441513061523,
      "step": 3142
    },
    {
      "epoch": 0.4249091677228559,
      "grad_norm": 0.2828345000743866,
      "learning_rate": 2.0434577062378203e-05,
      "loss": 0.09438323974609375,
      "step": 3143
    },
    {
      "epoch": 0.4250443599493029,
      "grad_norm": 0.2404760867357254,
      "learning_rate": 2.0428197082018458e-05,
      "loss": 0.05620098114013672,
      "step": 3144
    },
    {
      "epoch": 0.4251795521757499,
      "grad_norm": 0.33218252658843994,
      "learning_rate": 2.042181597147913e-05,
      "loss": 0.07691001892089844,
      "step": 3145
    },
    {
      "epoch": 0.42531474440219685,
      "grad_norm": 0.31662729382514954,
      "learning_rate": 2.0415433732088806e-05,
      "loss": 0.0770573616027832,
      "step": 3146
    },
    {
      "epoch": 0.42544993662864383,
      "grad_norm": 0.2540580630302429,
      "learning_rate": 2.0409050365176294e-05,
      "loss": 0.06334209442138672,
      "step": 3147
    },
    {
      "epoch": 0.4255851288550908,
      "grad_norm": 0.6130107045173645,
      "learning_rate": 2.0402665872070656e-05,
      "loss": 0.08261871337890625,
      "step": 3148
    },
    {
      "epoch": 0.4257203210815378,
      "grad_norm": 0.2526322603225708,
      "learning_rate": 2.0396280254101172e-05,
      "loss": 0.03117823600769043,
      "step": 3149
    },
    {
      "epoch": 0.42585551330798477,
      "grad_norm": 0.2598215341567993,
      "learning_rate": 2.0389893512597364e-05,
      "loss": 0.06298279762268066,
      "step": 3150
    },
    {
      "epoch": 0.42599070553443175,
      "grad_norm": 0.2517451345920563,
      "learning_rate": 2.0383505648888986e-05,
      "loss": 0.05115222930908203,
      "step": 3151
    },
    {
      "epoch": 0.4261258977608787,
      "grad_norm": 0.4152345359325409,
      "learning_rate": 2.037711666430602e-05,
      "loss": 0.06592178344726562,
      "step": 3152
    },
    {
      "epoch": 0.4262610899873257,
      "grad_norm": 0.21208588778972626,
      "learning_rate": 2.0370726560178693e-05,
      "loss": 0.06539249420166016,
      "step": 3153
    },
    {
      "epoch": 0.4263962822137727,
      "grad_norm": 0.3017374575138092,
      "learning_rate": 2.036433533783745e-05,
      "loss": 0.0654914379119873,
      "step": 3154
    },
    {
      "epoch": 0.42653147444021966,
      "grad_norm": 0.4026988744735718,
      "learning_rate": 2.0357942998612988e-05,
      "loss": 0.06147575378417969,
      "step": 3155
    },
    {
      "epoch": 0.4266666666666667,
      "grad_norm": 0.2728249728679657,
      "learning_rate": 2.0351549543836224e-05,
      "loss": 0.07100391387939453,
      "step": 3156
    },
    {
      "epoch": 0.4268018588931137,
      "grad_norm": 0.2211737185716629,
      "learning_rate": 2.0345154974838307e-05,
      "loss": 0.05893146991729736,
      "step": 3157
    },
    {
      "epoch": 0.42693705111956065,
      "grad_norm": 0.3204041123390198,
      "learning_rate": 2.0338759292950618e-05,
      "loss": 0.06424188613891602,
      "step": 3158
    },
    {
      "epoch": 0.42707224334600763,
      "grad_norm": 0.23772023618221283,
      "learning_rate": 2.033236249950477e-05,
      "loss": 0.04523515701293945,
      "step": 3159
    },
    {
      "epoch": 0.4272074355724546,
      "grad_norm": 0.24550877511501312,
      "learning_rate": 2.0325964595832618e-05,
      "loss": 0.055190324783325195,
      "step": 3160
    },
    {
      "epoch": 0.4273426277989016,
      "grad_norm": 0.3470247685909271,
      "learning_rate": 2.031956558326624e-05,
      "loss": 0.07321572303771973,
      "step": 3161
    },
    {
      "epoch": 0.42747782002534856,
      "grad_norm": 0.3611384928226471,
      "learning_rate": 2.0313165463137935e-05,
      "loss": 0.07228946685791016,
      "step": 3162
    },
    {
      "epoch": 0.42761301225179554,
      "grad_norm": 0.34616079926490784,
      "learning_rate": 2.030676423678025e-05,
      "loss": 0.07276630401611328,
      "step": 3163
    },
    {
      "epoch": 0.4277482044782425,
      "grad_norm": 0.18685780465602875,
      "learning_rate": 2.030036190552595e-05,
      "loss": 0.07238388061523438,
      "step": 3164
    },
    {
      "epoch": 0.4278833967046895,
      "grad_norm": 0.4286274313926697,
      "learning_rate": 2.029395847070803e-05,
      "loss": 0.0866384506225586,
      "step": 3165
    },
    {
      "epoch": 0.4280185889311365,
      "grad_norm": 0.34199005365371704,
      "learning_rate": 2.0287553933659735e-05,
      "loss": 0.06592819094657898,
      "step": 3166
    },
    {
      "epoch": 0.42815378115758346,
      "grad_norm": 0.4649122357368469,
      "learning_rate": 2.0281148295714512e-05,
      "loss": 0.06458044052124023,
      "step": 3167
    },
    {
      "epoch": 0.42828897338403044,
      "grad_norm": 0.2594263553619385,
      "learning_rate": 2.027474155820605e-05,
      "loss": 0.060778021812438965,
      "step": 3168
    },
    {
      "epoch": 0.4284241656104774,
      "grad_norm": 0.8002752661705017,
      "learning_rate": 2.026833372246827e-05,
      "loss": 0.10184288024902344,
      "step": 3169
    },
    {
      "epoch": 0.4285593578369244,
      "grad_norm": 0.37460413575172424,
      "learning_rate": 2.026192478983531e-05,
      "loss": 0.07912003993988037,
      "step": 3170
    },
    {
      "epoch": 0.42869455006337137,
      "grad_norm": 0.49152660369873047,
      "learning_rate": 2.0255514761641555e-05,
      "loss": 0.08054447174072266,
      "step": 3171
    },
    {
      "epoch": 0.42882974228981835,
      "grad_norm": 0.4222905933856964,
      "learning_rate": 2.0249103639221597e-05,
      "loss": 0.0560762882232666,
      "step": 3172
    },
    {
      "epoch": 0.42896493451626533,
      "grad_norm": 0.4432280361652374,
      "learning_rate": 2.024269142391027e-05,
      "loss": 0.057440757751464844,
      "step": 3173
    },
    {
      "epoch": 0.4291001267427123,
      "grad_norm": 0.3203895688056946,
      "learning_rate": 2.023627811704263e-05,
      "loss": 0.07187581062316895,
      "step": 3174
    },
    {
      "epoch": 0.4292353189691593,
      "grad_norm": 0.8808659315109253,
      "learning_rate": 2.0229863719953963e-05,
      "loss": 0.09990119934082031,
      "step": 3175
    },
    {
      "epoch": 0.42937051119560626,
      "grad_norm": 0.4019676148891449,
      "learning_rate": 2.0223448233979785e-05,
      "loss": 0.0510178804397583,
      "step": 3176
    },
    {
      "epoch": 0.42950570342205324,
      "grad_norm": 0.5386053919792175,
      "learning_rate": 2.0217031660455825e-05,
      "loss": 0.06357765197753906,
      "step": 3177
    },
    {
      "epoch": 0.4296408956485002,
      "grad_norm": 0.24994978308677673,
      "learning_rate": 2.0210614000718054e-05,
      "loss": 0.06611442565917969,
      "step": 3178
    },
    {
      "epoch": 0.4297760878749472,
      "grad_norm": 0.41393131017684937,
      "learning_rate": 2.020419525610266e-05,
      "loss": 0.06168675422668457,
      "step": 3179
    },
    {
      "epoch": 0.4299112801013942,
      "grad_norm": 0.2493085414171219,
      "learning_rate": 2.0197775427946066e-05,
      "loss": 0.06556558609008789,
      "step": 3180
    },
    {
      "epoch": 0.43004647232784116,
      "grad_norm": 0.3049503564834595,
      "learning_rate": 2.0191354517584902e-05,
      "loss": 0.09117889404296875,
      "step": 3181
    },
    {
      "epoch": 0.43018166455428813,
      "grad_norm": 0.5270341038703918,
      "learning_rate": 2.018493252635605e-05,
      "loss": 0.052628517150878906,
      "step": 3182
    },
    {
      "epoch": 0.4303168567807351,
      "grad_norm": 0.33322596549987793,
      "learning_rate": 2.0178509455596598e-05,
      "loss": 0.06375730037689209,
      "step": 3183
    },
    {
      "epoch": 0.4304520490071821,
      "grad_norm": 0.21174496412277222,
      "learning_rate": 2.017208530664386e-05,
      "loss": 0.05694985389709473,
      "step": 3184
    },
    {
      "epoch": 0.43058724123362907,
      "grad_norm": 0.23592528700828552,
      "learning_rate": 2.016566008083538e-05,
      "loss": 0.06638085842132568,
      "step": 3185
    },
    {
      "epoch": 0.43072243346007605,
      "grad_norm": 0.15597030520439148,
      "learning_rate": 2.0159233779508923e-05,
      "loss": 0.03405046463012695,
      "step": 3186
    },
    {
      "epoch": 0.430857625686523,
      "grad_norm": 0.31485578417778015,
      "learning_rate": 2.0152806404002482e-05,
      "loss": 0.06981182098388672,
      "step": 3187
    },
    {
      "epoch": 0.43099281791297,
      "grad_norm": 0.1980515718460083,
      "learning_rate": 2.014637795565427e-05,
      "loss": 0.056336402893066406,
      "step": 3188
    },
    {
      "epoch": 0.431128010139417,
      "grad_norm": 0.29888081550598145,
      "learning_rate": 2.0139948435802722e-05,
      "loss": 0.0712289810180664,
      "step": 3189
    },
    {
      "epoch": 0.43126320236586396,
      "grad_norm": 0.4977204203605652,
      "learning_rate": 2.0133517845786504e-05,
      "loss": 0.08624553680419922,
      "step": 3190
    },
    {
      "epoch": 0.43139839459231094,
      "grad_norm": 0.34856224060058594,
      "learning_rate": 2.012708618694449e-05,
      "loss": 0.10265088081359863,
      "step": 3191
    },
    {
      "epoch": 0.4315335868187579,
      "grad_norm": 0.2021336406469345,
      "learning_rate": 2.0120653460615795e-05,
      "loss": 0.048661231994628906,
      "step": 3192
    },
    {
      "epoch": 0.4316687790452049,
      "grad_norm": 0.26064735651016235,
      "learning_rate": 2.011421966813974e-05,
      "loss": 0.06618070602416992,
      "step": 3193
    },
    {
      "epoch": 0.4318039712716519,
      "grad_norm": 0.19883018732070923,
      "learning_rate": 2.0107784810855882e-05,
      "loss": 0.06342935562133789,
      "step": 3194
    },
    {
      "epoch": 0.43193916349809885,
      "grad_norm": 0.4939442574977875,
      "learning_rate": 2.0101348890103985e-05,
      "loss": 0.09639501571655273,
      "step": 3195
    },
    {
      "epoch": 0.43207435572454583,
      "grad_norm": 0.32421886920928955,
      "learning_rate": 2.0094911907224043e-05,
      "loss": 0.05952805280685425,
      "step": 3196
    },
    {
      "epoch": 0.4322095479509928,
      "grad_norm": 0.3773537278175354,
      "learning_rate": 2.008847386355628e-05,
      "loss": 0.05793642997741699,
      "step": 3197
    },
    {
      "epoch": 0.4323447401774398,
      "grad_norm": 0.45993101596832275,
      "learning_rate": 2.008203476044112e-05,
      "loss": 0.08371472358703613,
      "step": 3198
    },
    {
      "epoch": 0.43247993240388677,
      "grad_norm": 0.6151641607284546,
      "learning_rate": 2.007559459921922e-05,
      "loss": 0.1029520034790039,
      "step": 3199
    },
    {
      "epoch": 0.43261512463033375,
      "grad_norm": 0.30804434418678284,
      "learning_rate": 2.0069153381231456e-05,
      "loss": 0.07819843292236328,
      "step": 3200
    },
    {
      "epoch": 0.4327503168567807,
      "grad_norm": 0.30804434418678284,
      "learning_rate": 2.0062711107818933e-05,
      "loss": 0.06537199020385742,
      "step": 3201
    },
    {
      "epoch": 0.4328855090832277,
      "grad_norm": 0.37973085045814514,
      "learning_rate": 2.0056267780322953e-05,
      "loss": 0.07285261154174805,
      "step": 3202
    },
    {
      "epoch": 0.4330207013096747,
      "grad_norm": 0.3804246187210083,
      "learning_rate": 2.004982340008506e-05,
      "loss": 0.07038390636444092,
      "step": 3203
    },
    {
      "epoch": 0.43315589353612166,
      "grad_norm": 0.1897815465927124,
      "learning_rate": 2.004337796844701e-05,
      "loss": 0.05299806594848633,
      "step": 3204
    },
    {
      "epoch": 0.43329108576256864,
      "grad_norm": 0.5176883339881897,
      "learning_rate": 2.003693148675077e-05,
      "loss": 0.06395530700683594,
      "step": 3205
    },
    {
      "epoch": 0.4334262779890156,
      "grad_norm": 0.33159148693084717,
      "learning_rate": 2.003048395633853e-05,
      "loss": 0.06426000595092773,
      "step": 3206
    },
    {
      "epoch": 0.4335614702154626,
      "grad_norm": 0.4107198119163513,
      "learning_rate": 2.0024035378552708e-05,
      "loss": 0.08121180534362793,
      "step": 3207
    },
    {
      "epoch": 0.4336966624419096,
      "grad_norm": 0.215802401304245,
      "learning_rate": 2.001758575473593e-05,
      "loss": 0.053858280181884766,
      "step": 3208
    },
    {
      "epoch": 0.43383185466835655,
      "grad_norm": 0.1386597603559494,
      "learning_rate": 2.0011135086231042e-05,
      "loss": 0.029804706573486328,
      "step": 3209
    },
    {
      "epoch": 0.43396704689480353,
      "grad_norm": 0.28065362572669983,
      "learning_rate": 2.0004683374381104e-05,
      "loss": 0.07149219512939453,
      "step": 3210
    },
    {
      "epoch": 0.4341022391212505,
      "grad_norm": 0.1404481679201126,
      "learning_rate": 1.9998230620529395e-05,
      "loss": 0.03793168067932129,
      "step": 3211
    },
    {
      "epoch": 0.4342374313476975,
      "grad_norm": 0.2699950933456421,
      "learning_rate": 1.999177682601942e-05,
      "loss": 0.072113037109375,
      "step": 3212
    },
    {
      "epoch": 0.43437262357414447,
      "grad_norm": 0.3491198718547821,
      "learning_rate": 1.9985321992194896e-05,
      "loss": 0.06703424453735352,
      "step": 3213
    },
    {
      "epoch": 0.43450781580059145,
      "grad_norm": 0.22347047924995422,
      "learning_rate": 1.9978866120399746e-05,
      "loss": 0.044330596923828125,
      "step": 3214
    },
    {
      "epoch": 0.4346430080270384,
      "grad_norm": 0.44929203391075134,
      "learning_rate": 1.9972409211978116e-05,
      "loss": 0.08514881134033203,
      "step": 3215
    },
    {
      "epoch": 0.4347782002534854,
      "grad_norm": 0.3418920338153839,
      "learning_rate": 1.9965951268274373e-05,
      "loss": 0.08155107498168945,
      "step": 3216
    },
    {
      "epoch": 0.4349133924799324,
      "grad_norm": 0.34604474902153015,
      "learning_rate": 1.9959492290633093e-05,
      "loss": 0.06706786155700684,
      "step": 3217
    },
    {
      "epoch": 0.43504858470637936,
      "grad_norm": 0.4195420742034912,
      "learning_rate": 1.995303228039907e-05,
      "loss": 0.05675840377807617,
      "step": 3218
    },
    {
      "epoch": 0.43518377693282634,
      "grad_norm": 0.4107212424278259,
      "learning_rate": 1.994657123891732e-05,
      "loss": 0.061762332916259766,
      "step": 3219
    },
    {
      "epoch": 0.4353189691592733,
      "grad_norm": 0.2144065499305725,
      "learning_rate": 1.9940109167533055e-05,
      "loss": 0.05501842498779297,
      "step": 3220
    },
    {
      "epoch": 0.4354541613857203,
      "grad_norm": 0.23718760907649994,
      "learning_rate": 1.9933646067591716e-05,
      "loss": 0.0576167106628418,
      "step": 3221
    },
    {
      "epoch": 0.4355893536121673,
      "grad_norm": 0.543972373008728,
      "learning_rate": 1.992718194043896e-05,
      "loss": 0.0962066650390625,
      "step": 3222
    },
    {
      "epoch": 0.43572454583861425,
      "grad_norm": 0.2456401288509369,
      "learning_rate": 1.9920716787420643e-05,
      "loss": 0.06356048583984375,
      "step": 3223
    },
    {
      "epoch": 0.4358597380650613,
      "grad_norm": 0.4617266356945038,
      "learning_rate": 1.9914250609882858e-05,
      "loss": 0.09366726875305176,
      "step": 3224
    },
    {
      "epoch": 0.43599493029150826,
      "grad_norm": 0.5732520818710327,
      "learning_rate": 1.9907783409171885e-05,
      "loss": 0.12010478973388672,
      "step": 3225
    },
    {
      "epoch": 0.43613012251795524,
      "grad_norm": 0.3567442297935486,
      "learning_rate": 1.990131518663424e-05,
      "loss": 0.09979486465454102,
      "step": 3226
    },
    {
      "epoch": 0.4362653147444022,
      "grad_norm": 0.23667088150978088,
      "learning_rate": 1.9894845943616632e-05,
      "loss": 0.05428934097290039,
      "step": 3227
    },
    {
      "epoch": 0.4364005069708492,
      "grad_norm": 0.34958022832870483,
      "learning_rate": 1.988837568146599e-05,
      "loss": 0.08516740798950195,
      "step": 3228
    },
    {
      "epoch": 0.4365356991972962,
      "grad_norm": 0.34907564520835876,
      "learning_rate": 1.988190440152947e-05,
      "loss": 0.08010435104370117,
      "step": 3229
    },
    {
      "epoch": 0.43667089142374316,
      "grad_norm": 0.44578802585601807,
      "learning_rate": 1.9875432105154424e-05,
      "loss": 0.08909749984741211,
      "step": 3230
    },
    {
      "epoch": 0.43680608365019014,
      "grad_norm": 0.34709492325782776,
      "learning_rate": 1.9868958793688412e-05,
      "loss": 0.08998554944992065,
      "step": 3231
    },
    {
      "epoch": 0.4369412758766371,
      "grad_norm": 0.24281227588653564,
      "learning_rate": 1.9862484468479213e-05,
      "loss": 0.05726742744445801,
      "step": 3232
    },
    {
      "epoch": 0.4370764681030841,
      "grad_norm": 0.24110205471515656,
      "learning_rate": 1.985600913087482e-05,
      "loss": 0.038764238357543945,
      "step": 3233
    },
    {
      "epoch": 0.43721166032953107,
      "grad_norm": 0.210011288523674,
      "learning_rate": 1.9849532782223425e-05,
      "loss": 0.05803108215332031,
      "step": 3234
    },
    {
      "epoch": 0.43734685255597805,
      "grad_norm": 0.2844734191894531,
      "learning_rate": 1.9843055423873447e-05,
      "loss": 0.054401397705078125,
      "step": 3235
    },
    {
      "epoch": 0.43748204478242503,
      "grad_norm": 0.2790979743003845,
      "learning_rate": 1.9836577057173507e-05,
      "loss": 0.050627708435058594,
      "step": 3236
    },
    {
      "epoch": 0.437617237008872,
      "grad_norm": 0.20133322477340698,
      "learning_rate": 1.9830097683472427e-05,
      "loss": 0.06956779956817627,
      "step": 3237
    },
    {
      "epoch": 0.437752429235319,
      "grad_norm": 0.26941344141960144,
      "learning_rate": 1.9823617304119252e-05,
      "loss": 0.06499528884887695,
      "step": 3238
    },
    {
      "epoch": 0.43788762146176596,
      "grad_norm": 0.383767694234848,
      "learning_rate": 1.9817135920463232e-05,
      "loss": 0.07185077667236328,
      "step": 3239
    },
    {
      "epoch": 0.43802281368821294,
      "grad_norm": 0.2848849892616272,
      "learning_rate": 1.9810653533853826e-05,
      "loss": 0.07364368438720703,
      "step": 3240
    },
    {
      "epoch": 0.4381580059146599,
      "grad_norm": 0.2620222270488739,
      "learning_rate": 1.9804170145640706e-05,
      "loss": 0.06521177291870117,
      "step": 3241
    },
    {
      "epoch": 0.4382931981411069,
      "grad_norm": 0.7277887463569641,
      "learning_rate": 1.9797685757173737e-05,
      "loss": 0.0758504867553711,
      "step": 3242
    },
    {
      "epoch": 0.4384283903675539,
      "grad_norm": 0.21800507605075836,
      "learning_rate": 1.979120036980301e-05,
      "loss": 0.056427717208862305,
      "step": 3243
    },
    {
      "epoch": 0.43856358259400086,
      "grad_norm": 0.2958162724971771,
      "learning_rate": 1.9784713984878814e-05,
      "loss": 0.07983338832855225,
      "step": 3244
    },
    {
      "epoch": 0.43869877482044783,
      "grad_norm": 0.37943050265312195,
      "learning_rate": 1.9778226603751652e-05,
      "loss": 0.08130407333374023,
      "step": 3245
    },
    {
      "epoch": 0.4388339670468948,
      "grad_norm": 0.2472180873155594,
      "learning_rate": 1.9771738227772235e-05,
      "loss": 0.04934966564178467,
      "step": 3246
    },
    {
      "epoch": 0.4389691592733418,
      "grad_norm": 0.29032063484191895,
      "learning_rate": 1.976524885829147e-05,
      "loss": 0.05225777626037598,
      "step": 3247
    },
    {
      "epoch": 0.43910435149978877,
      "grad_norm": 0.21247722208499908,
      "learning_rate": 1.975875849666048e-05,
      "loss": 0.042376160621643066,
      "step": 3248
    },
    {
      "epoch": 0.43923954372623575,
      "grad_norm": 0.3725825250148773,
      "learning_rate": 1.9752267144230595e-05,
      "loss": 0.08426475524902344,
      "step": 3249
    },
    {
      "epoch": 0.4393747359526827,
      "grad_norm": 0.2907467782497406,
      "learning_rate": 1.9745774802353347e-05,
      "loss": 0.06331539154052734,
      "step": 3250
    },
    {
      "epoch": 0.4395099281791297,
      "grad_norm": 0.3804432451725006,
      "learning_rate": 1.973928147238048e-05,
      "loss": 0.07320070266723633,
      "step": 3251
    },
    {
      "epoch": 0.4396451204055767,
      "grad_norm": 0.21498610079288483,
      "learning_rate": 1.973278715566394e-05,
      "loss": 0.06844902038574219,
      "step": 3252
    },
    {
      "epoch": 0.43978031263202366,
      "grad_norm": 0.31550881266593933,
      "learning_rate": 1.972629185355587e-05,
      "loss": 0.055769920349121094,
      "step": 3253
    },
    {
      "epoch": 0.43991550485847064,
      "grad_norm": 1.6527247428894043,
      "learning_rate": 1.971979556740864e-05,
      "loss": 0.07448101043701172,
      "step": 3254
    },
    {
      "epoch": 0.4400506970849176,
      "grad_norm": 0.18900066614151,
      "learning_rate": 1.971329829857479e-05,
      "loss": 0.057601213455200195,
      "step": 3255
    },
    {
      "epoch": 0.4401858893113646,
      "grad_norm": 0.4061093032360077,
      "learning_rate": 1.9706800048407112e-05,
      "loss": 0.07447242736816406,
      "step": 3256
    },
    {
      "epoch": 0.4403210815378116,
      "grad_norm": 0.3093827962875366,
      "learning_rate": 1.9700300818258566e-05,
      "loss": 0.06330156326293945,
      "step": 3257
    },
    {
      "epoch": 0.44045627376425855,
      "grad_norm": 0.4635726511478424,
      "learning_rate": 1.9693800609482318e-05,
      "loss": 0.06341552734375,
      "step": 3258
    },
    {
      "epoch": 0.44059146599070553,
      "grad_norm": 0.36550334095954895,
      "learning_rate": 1.9687299423431754e-05,
      "loss": 0.07641315460205078,
      "step": 3259
    },
    {
      "epoch": 0.4407266582171525,
      "grad_norm": 0.400806725025177,
      "learning_rate": 1.968079726146045e-05,
      "loss": 0.05738019943237305,
      "step": 3260
    },
    {
      "epoch": 0.4408618504435995,
      "grad_norm": 0.30217820405960083,
      "learning_rate": 1.9674294124922204e-05,
      "loss": 0.06429815292358398,
      "step": 3261
    },
    {
      "epoch": 0.44099704267004647,
      "grad_norm": 0.2563503384590149,
      "learning_rate": 1.966779001517099e-05,
      "loss": 0.04334282875061035,
      "step": 3262
    },
    {
      "epoch": 0.44113223489649345,
      "grad_norm": 0.34525415301322937,
      "learning_rate": 1.9661284933561006e-05,
      "loss": 0.09828376770019531,
      "step": 3263
    },
    {
      "epoch": 0.4412674271229404,
      "grad_norm": 0.3323579728603363,
      "learning_rate": 1.9654778881446636e-05,
      "loss": 0.07301616668701172,
      "step": 3264
    },
    {
      "epoch": 0.4414026193493874,
      "grad_norm": 0.2360515296459198,
      "learning_rate": 1.9648271860182487e-05,
      "loss": 0.06273889541625977,
      "step": 3265
    },
    {
      "epoch": 0.4415378115758344,
      "grad_norm": 0.314397931098938,
      "learning_rate": 1.9641763871123345e-05,
      "loss": 0.053080081939697266,
      "step": 3266
    },
    {
      "epoch": 0.44167300380228136,
      "grad_norm": 0.2568463087081909,
      "learning_rate": 1.963525491562421e-05,
      "loss": 0.07741570472717285,
      "step": 3267
    },
    {
      "epoch": 0.44180819602872834,
      "grad_norm": 0.5656931400299072,
      "learning_rate": 1.9628744995040287e-05,
      "loss": 0.1472454071044922,
      "step": 3268
    },
    {
      "epoch": 0.4419433882551753,
      "grad_norm": 0.24615713953971863,
      "learning_rate": 1.9622234110726976e-05,
      "loss": 0.049279212951660156,
      "step": 3269
    },
    {
      "epoch": 0.4420785804816223,
      "grad_norm": 0.3402886390686035,
      "learning_rate": 1.9615722264039868e-05,
      "loss": 0.07629966735839844,
      "step": 3270
    },
    {
      "epoch": 0.4422137727080693,
      "grad_norm": 0.25823819637298584,
      "learning_rate": 1.9609209456334772e-05,
      "loss": 0.05963563919067383,
      "step": 3271
    },
    {
      "epoch": 0.44234896493451625,
      "grad_norm": 0.420064777135849,
      "learning_rate": 1.960269568896769e-05,
      "loss": 0.06271743774414062,
      "step": 3272
    },
    {
      "epoch": 0.44248415716096323,
      "grad_norm": 0.33517250418663025,
      "learning_rate": 1.9596180963294822e-05,
      "loss": 0.07872200012207031,
      "step": 3273
    },
    {
      "epoch": 0.4426193493874102,
      "grad_norm": 0.2894724905490875,
      "learning_rate": 1.9589665280672564e-05,
      "loss": 0.0593571662902832,
      "step": 3274
    },
    {
      "epoch": 0.4427545416138572,
      "grad_norm": 0.317278653383255,
      "learning_rate": 1.958314864245752e-05,
      "loss": 0.07319831848144531,
      "step": 3275
    },
    {
      "epoch": 0.44288973384030417,
      "grad_norm": 0.30626583099365234,
      "learning_rate": 1.957663105000649e-05,
      "loss": 0.06558942794799805,
      "step": 3276
    },
    {
      "epoch": 0.44302492606675115,
      "grad_norm": 0.37172073125839233,
      "learning_rate": 1.957011250467647e-05,
      "loss": 0.04667782783508301,
      "step": 3277
    },
    {
      "epoch": 0.4431601182931981,
      "grad_norm": 0.44745346903800964,
      "learning_rate": 1.9563593007824658e-05,
      "loss": 0.07155179977416992,
      "step": 3278
    },
    {
      "epoch": 0.4432953105196451,
      "grad_norm": 0.30978548526763916,
      "learning_rate": 1.9557072560808442e-05,
      "loss": 0.08234691619873047,
      "step": 3279
    },
    {
      "epoch": 0.4434305027460921,
      "grad_norm": 0.26966768503189087,
      "learning_rate": 1.9550551164985418e-05,
      "loss": 0.06200361251831055,
      "step": 3280
    },
    {
      "epoch": 0.44356569497253906,
      "grad_norm": 0.4869910776615143,
      "learning_rate": 1.9544028821713372e-05,
      "loss": 0.08661699295043945,
      "step": 3281
    },
    {
      "epoch": 0.44370088719898604,
      "grad_norm": 0.25459226965904236,
      "learning_rate": 1.9537505532350298e-05,
      "loss": 0.07809257507324219,
      "step": 3282
    },
    {
      "epoch": 0.443836079425433,
      "grad_norm": 0.3783869445323944,
      "learning_rate": 1.9530981298254376e-05,
      "loss": 0.06965303421020508,
      "step": 3283
    },
    {
      "epoch": 0.44397127165188,
      "grad_norm": 0.17915265262126923,
      "learning_rate": 1.9524456120783983e-05,
      "loss": 0.06292033195495605,
      "step": 3284
    },
    {
      "epoch": 0.444106463878327,
      "grad_norm": 0.2549220621585846,
      "learning_rate": 1.95179300012977e-05,
      "loss": 0.06046426296234131,
      "step": 3285
    },
    {
      "epoch": 0.44424165610477395,
      "grad_norm": 0.3083202540874481,
      "learning_rate": 1.9511402941154296e-05,
      "loss": 0.07697391510009766,
      "step": 3286
    },
    {
      "epoch": 0.44437684833122093,
      "grad_norm": 0.6192878484725952,
      "learning_rate": 1.950487494171274e-05,
      "loss": 0.09878063201904297,
      "step": 3287
    },
    {
      "epoch": 0.4445120405576679,
      "grad_norm": 0.4199250042438507,
      "learning_rate": 1.9498346004332203e-05,
      "loss": 0.07847166061401367,
      "step": 3288
    },
    {
      "epoch": 0.4446472327841149,
      "grad_norm": 0.2877029776573181,
      "learning_rate": 1.949181613037204e-05,
      "loss": 0.08483314514160156,
      "step": 3289
    },
    {
      "epoch": 0.44478242501056187,
      "grad_norm": 0.17623817920684814,
      "learning_rate": 1.9485285321191804e-05,
      "loss": 0.04591083526611328,
      "step": 3290
    },
    {
      "epoch": 0.44491761723700884,
      "grad_norm": 0.42056602239608765,
      "learning_rate": 1.9478753578151244e-05,
      "loss": 0.06670105457305908,
      "step": 3291
    },
    {
      "epoch": 0.4450528094634559,
      "grad_norm": 0.4309781491756439,
      "learning_rate": 1.9472220902610304e-05,
      "loss": 0.070465087890625,
      "step": 3292
    },
    {
      "epoch": 0.44518800168990286,
      "grad_norm": 0.14935748279094696,
      "learning_rate": 1.9465687295929127e-05,
      "loss": 0.03504657745361328,
      "step": 3293
    },
    {
      "epoch": 0.44532319391634984,
      "grad_norm": 0.21144822239875793,
      "learning_rate": 1.945915275946804e-05,
      "loss": 0.042441487312316895,
      "step": 3294
    },
    {
      "epoch": 0.4454583861427968,
      "grad_norm": 0.30384671688079834,
      "learning_rate": 1.9452617294587573e-05,
      "loss": 0.07517242431640625,
      "step": 3295
    },
    {
      "epoch": 0.4455935783692438,
      "grad_norm": 0.18729279935359955,
      "learning_rate": 1.9446080902648435e-05,
      "loss": 0.04670882225036621,
      "step": 3296
    },
    {
      "epoch": 0.44572877059569077,
      "grad_norm": 0.5563015937805176,
      "learning_rate": 1.943954358501154e-05,
      "loss": 0.0895843505859375,
      "step": 3297
    },
    {
      "epoch": 0.44586396282213775,
      "grad_norm": 0.342330664396286,
      "learning_rate": 1.943300534303801e-05,
      "loss": 0.05006873607635498,
      "step": 3298
    },
    {
      "epoch": 0.44599915504858473,
      "grad_norm": 0.30710843205451965,
      "learning_rate": 1.9426466178089116e-05,
      "loss": 0.06839275360107422,
      "step": 3299
    },
    {
      "epoch": 0.4461343472750317,
      "grad_norm": 0.3760884404182434,
      "learning_rate": 1.9419926091526367e-05,
      "loss": 0.09142160415649414,
      "step": 3300
    },
    {
      "epoch": 0.4462695395014787,
      "grad_norm": 0.5989261269569397,
      "learning_rate": 1.9413385084711425e-05,
      "loss": 0.09687232971191406,
      "step": 3301
    },
    {
      "epoch": 0.44640473172792566,
      "grad_norm": 0.2804635763168335,
      "learning_rate": 1.9406843159006183e-05,
      "loss": 0.064483642578125,
      "step": 3302
    },
    {
      "epoch": 0.44653992395437264,
      "grad_norm": 0.34838199615478516,
      "learning_rate": 1.940030031577269e-05,
      "loss": 0.057769775390625,
      "step": 3303
    },
    {
      "epoch": 0.4466751161808196,
      "grad_norm": 0.6450948715209961,
      "learning_rate": 1.9393756556373212e-05,
      "loss": 0.09352350234985352,
      "step": 3304
    },
    {
      "epoch": 0.4468103084072666,
      "grad_norm": 0.4629974365234375,
      "learning_rate": 1.9387211882170184e-05,
      "loss": 0.06857538223266602,
      "step": 3305
    },
    {
      "epoch": 0.4469455006337136,
      "grad_norm": 0.38311535120010376,
      "learning_rate": 1.9380666294526243e-05,
      "loss": 0.06781387329101562,
      "step": 3306
    },
    {
      "epoch": 0.44708069286016056,
      "grad_norm": 0.3830869495868683,
      "learning_rate": 1.9374119794804228e-05,
      "loss": 0.10775327682495117,
      "step": 3307
    },
    {
      "epoch": 0.44721588508660753,
      "grad_norm": 0.2984815835952759,
      "learning_rate": 1.936757238436714e-05,
      "loss": 0.0723876953125,
      "step": 3308
    },
    {
      "epoch": 0.4473510773130545,
      "grad_norm": 0.407243937253952,
      "learning_rate": 1.936102406457819e-05,
      "loss": 0.0800161361694336,
      "step": 3309
    },
    {
      "epoch": 0.4474862695395015,
      "grad_norm": 0.5078486204147339,
      "learning_rate": 1.935447483680078e-05,
      "loss": 0.09047174453735352,
      "step": 3310
    },
    {
      "epoch": 0.44762146176594847,
      "grad_norm": 0.2887296676635742,
      "learning_rate": 1.9347924702398484e-05,
      "loss": 0.04402494430541992,
      "step": 3311
    },
    {
      "epoch": 0.44775665399239545,
      "grad_norm": 0.3636722266674042,
      "learning_rate": 1.9341373662735075e-05,
      "loss": 0.07357645034790039,
      "step": 3312
    },
    {
      "epoch": 0.4478918462188424,
      "grad_norm": 0.22289150953292847,
      "learning_rate": 1.9334821719174524e-05,
      "loss": 0.054033517837524414,
      "step": 3313
    },
    {
      "epoch": 0.4480270384452894,
      "grad_norm": 0.35751694440841675,
      "learning_rate": 1.9328268873080974e-05,
      "loss": 0.10008716583251953,
      "step": 3314
    },
    {
      "epoch": 0.4481622306717364,
      "grad_norm": 0.2627313733100891,
      "learning_rate": 1.9321715125818765e-05,
      "loss": 0.07291555404663086,
      "step": 3315
    },
    {
      "epoch": 0.44829742289818336,
      "grad_norm": 0.24357503652572632,
      "learning_rate": 1.931516047875242e-05,
      "loss": 0.0700075626373291,
      "step": 3316
    },
    {
      "epoch": 0.44843261512463034,
      "grad_norm": 0.39955195784568787,
      "learning_rate": 1.930860493324665e-05,
      "loss": 0.07883906364440918,
      "step": 3317
    },
    {
      "epoch": 0.4485678073510773,
      "grad_norm": 0.22202704846858978,
      "learning_rate": 1.9302048490666356e-05,
      "loss": 0.059880733489990234,
      "step": 3318
    },
    {
      "epoch": 0.4487029995775243,
      "grad_norm": 0.42034509778022766,
      "learning_rate": 1.9295491152376623e-05,
      "loss": 0.06959772109985352,
      "step": 3319
    },
    {
      "epoch": 0.4488381918039713,
      "grad_norm": 0.4925479590892792,
      "learning_rate": 1.928893291974273e-05,
      "loss": 0.06459498405456543,
      "step": 3320
    },
    {
      "epoch": 0.44897338403041825,
      "grad_norm": 0.38164782524108887,
      "learning_rate": 1.9282373794130132e-05,
      "loss": 0.06588971614837646,
      "step": 3321
    },
    {
      "epoch": 0.44910857625686523,
      "grad_norm": 0.23238146305084229,
      "learning_rate": 1.9275813776904472e-05,
      "loss": 0.04213905334472656,
      "step": 3322
    },
    {
      "epoch": 0.4492437684833122,
      "grad_norm": 0.4110434949398041,
      "learning_rate": 1.9269252869431582e-05,
      "loss": 0.08344841003417969,
      "step": 3323
    },
    {
      "epoch": 0.4493789607097592,
      "grad_norm": 0.3393557667732239,
      "learning_rate": 1.9262691073077476e-05,
      "loss": 0.045320987701416016,
      "step": 3324
    },
    {
      "epoch": 0.44951415293620617,
      "grad_norm": 0.19677022099494934,
      "learning_rate": 1.9256128389208363e-05,
      "loss": 0.04906785488128662,
      "step": 3325
    },
    {
      "epoch": 0.44964934516265315,
      "grad_norm": 0.21272487938404083,
      "learning_rate": 1.924956481919062e-05,
      "loss": 0.056756019592285156,
      "step": 3326
    },
    {
      "epoch": 0.4497845373891001,
      "grad_norm": 0.28534969687461853,
      "learning_rate": 1.9243000364390825e-05,
      "loss": 0.07583838701248169,
      "step": 3327
    },
    {
      "epoch": 0.4499197296155471,
      "grad_norm": 0.22729046642780304,
      "learning_rate": 1.9236435026175717e-05,
      "loss": 0.044859886169433594,
      "step": 3328
    },
    {
      "epoch": 0.4500549218419941,
      "grad_norm": 0.32950401306152344,
      "learning_rate": 1.9229868805912248e-05,
      "loss": 0.07035517692565918,
      "step": 3329
    },
    {
      "epoch": 0.45019011406844106,
      "grad_norm": 0.285521924495697,
      "learning_rate": 1.9223301704967543e-05,
      "loss": 0.06944990158081055,
      "step": 3330
    },
    {
      "epoch": 0.45032530629488804,
      "grad_norm": 0.5533238053321838,
      "learning_rate": 1.92167337247089e-05,
      "loss": 0.09995520114898682,
      "step": 3331
    },
    {
      "epoch": 0.450460498521335,
      "grad_norm": 0.15855996310710907,
      "learning_rate": 1.9210164866503808e-05,
      "loss": 0.047154903411865234,
      "step": 3332
    },
    {
      "epoch": 0.450595690747782,
      "grad_norm": 0.19494955241680145,
      "learning_rate": 1.9203595131719935e-05,
      "loss": 0.049901723861694336,
      "step": 3333
    },
    {
      "epoch": 0.450730882974229,
      "grad_norm": 0.31304165720939636,
      "learning_rate": 1.9197024521725148e-05,
      "loss": 0.08922672271728516,
      "step": 3334
    },
    {
      "epoch": 0.45086607520067595,
      "grad_norm": 0.4357836842536926,
      "learning_rate": 1.9190453037887465e-05,
      "loss": 0.1002645492553711,
      "step": 3335
    },
    {
      "epoch": 0.45100126742712293,
      "grad_norm": 0.33210405707359314,
      "learning_rate": 1.918388068157512e-05,
      "loss": 0.05168336629867554,
      "step": 3336
    },
    {
      "epoch": 0.4511364596535699,
      "grad_norm": 0.2215290665626526,
      "learning_rate": 1.9177307454156507e-05,
      "loss": 0.06288552284240723,
      "step": 3337
    },
    {
      "epoch": 0.4512716518800169,
      "grad_norm": 0.4807816743850708,
      "learning_rate": 1.9170733357000202e-05,
      "loss": 0.0641336441040039,
      "step": 3338
    },
    {
      "epoch": 0.45140684410646387,
      "grad_norm": 0.2763362228870392,
      "learning_rate": 1.916415839147497e-05,
      "loss": 0.06133127212524414,
      "step": 3339
    },
    {
      "epoch": 0.45154203633291085,
      "grad_norm": 0.3626943826675415,
      "learning_rate": 1.9157582558949756e-05,
      "loss": 0.04251909255981445,
      "step": 3340
    },
    {
      "epoch": 0.4516772285593578,
      "grad_norm": 0.409523069858551,
      "learning_rate": 1.9151005860793682e-05,
      "loss": 0.07260751724243164,
      "step": 3341
    },
    {
      "epoch": 0.4518124207858048,
      "grad_norm": 0.3742903769016266,
      "learning_rate": 1.9144428298376056e-05,
      "loss": 0.09168076515197754,
      "step": 3342
    },
    {
      "epoch": 0.4519476130122518,
      "grad_norm": 0.3158626854419708,
      "learning_rate": 1.9137849873066355e-05,
      "loss": 0.07207798957824707,
      "step": 3343
    },
    {
      "epoch": 0.45208280523869876,
      "grad_norm": 0.5290866494178772,
      "learning_rate": 1.9131270586234243e-05,
      "loss": 0.08540534973144531,
      "step": 3344
    },
    {
      "epoch": 0.45221799746514574,
      "grad_norm": 0.33754172921180725,
      "learning_rate": 1.9124690439249564e-05,
      "loss": 0.06018531322479248,
      "step": 3345
    },
    {
      "epoch": 0.4523531896915927,
      "grad_norm": 0.21841798722743988,
      "learning_rate": 1.9118109433482342e-05,
      "loss": 0.04831266403198242,
      "step": 3346
    },
    {
      "epoch": 0.4524883819180397,
      "grad_norm": 0.3793192505836487,
      "learning_rate": 1.911152757030278e-05,
      "loss": 0.06748485565185547,
      "step": 3347
    },
    {
      "epoch": 0.4526235741444867,
      "grad_norm": 0.17827394604682922,
      "learning_rate": 1.9104944851081247e-05,
      "loss": 0.044198036193847656,
      "step": 3348
    },
    {
      "epoch": 0.45275876637093365,
      "grad_norm": 0.27998778223991394,
      "learning_rate": 1.9098361277188303e-05,
      "loss": 0.06831824779510498,
      "step": 3349
    },
    {
      "epoch": 0.45289395859738063,
      "grad_norm": 0.34288692474365234,
      "learning_rate": 1.909177684999469e-05,
      "loss": 0.05758047103881836,
      "step": 3350
    },
    {
      "epoch": 0.4530291508238276,
      "grad_norm": 0.2111118584871292,
      "learning_rate": 1.9085191570871316e-05,
      "loss": 0.056126147508621216,
      "step": 3351
    },
    {
      "epoch": 0.4531643430502746,
      "grad_norm": 0.3934229910373688,
      "learning_rate": 1.9078605441189275e-05,
      "loss": 0.07675504684448242,
      "step": 3352
    },
    {
      "epoch": 0.45329953527672157,
      "grad_norm": 0.23766179382801056,
      "learning_rate": 1.9072018462319828e-05,
      "loss": 0.054659128189086914,
      "step": 3353
    },
    {
      "epoch": 0.45343472750316854,
      "grad_norm": 0.530798077583313,
      "learning_rate": 1.9065430635634422e-05,
      "loss": 0.08081603050231934,
      "step": 3354
    },
    {
      "epoch": 0.4535699197296155,
      "grad_norm": 0.244573712348938,
      "learning_rate": 1.9058841962504677e-05,
      "loss": 0.057622432708740234,
      "step": 3355
    },
    {
      "epoch": 0.4537051119560625,
      "grad_norm": 0.16016773879528046,
      "learning_rate": 1.9052252444302394e-05,
      "loss": 0.04063546657562256,
      "step": 3356
    },
    {
      "epoch": 0.4538403041825095,
      "grad_norm": 0.15886633098125458,
      "learning_rate": 1.904566208239954e-05,
      "loss": 0.05351591110229492,
      "step": 3357
    },
    {
      "epoch": 0.45397549640895646,
      "grad_norm": 0.556342363357544,
      "learning_rate": 1.903907087816827e-05,
      "loss": 0.07323169708251953,
      "step": 3358
    },
    {
      "epoch": 0.45411068863540344,
      "grad_norm": 0.33654287457466125,
      "learning_rate": 1.9032478832980902e-05,
      "loss": 0.08651256561279297,
      "step": 3359
    },
    {
      "epoch": 0.45424588086185047,
      "grad_norm": 0.23699747025966644,
      "learning_rate": 1.9025885948209938e-05,
      "loss": 0.06017589569091797,
      "step": 3360
    },
    {
      "epoch": 0.45438107308829745,
      "grad_norm": 0.5729971528053284,
      "learning_rate": 1.901929222522805e-05,
      "loss": 0.1022958755493164,
      "step": 3361
    },
    {
      "epoch": 0.4545162653147444,
      "grad_norm": 0.28968188166618347,
      "learning_rate": 1.901269766540809e-05,
      "loss": 0.0852656364440918,
      "step": 3362
    },
    {
      "epoch": 0.4546514575411914,
      "grad_norm": 0.15169383585453033,
      "learning_rate": 1.9006102270123076e-05,
      "loss": 0.03434610366821289,
      "step": 3363
    },
    {
      "epoch": 0.4547866497676384,
      "grad_norm": 0.33873650431632996,
      "learning_rate": 1.8999506040746208e-05,
      "loss": 0.054129600524902344,
      "step": 3364
    },
    {
      "epoch": 0.45492184199408536,
      "grad_norm": 0.30081894993782043,
      "learning_rate": 1.899290897865085e-05,
      "loss": 0.064056396484375,
      "step": 3365
    },
    {
      "epoch": 0.45505703422053234,
      "grad_norm": 0.21700730919837952,
      "learning_rate": 1.898631108521055e-05,
      "loss": 0.06077003479003906,
      "step": 3366
    },
    {
      "epoch": 0.4551922264469793,
      "grad_norm": 0.16849905252456665,
      "learning_rate": 1.8979712361799027e-05,
      "loss": 0.050501346588134766,
      "step": 3367
    },
    {
      "epoch": 0.4553274186734263,
      "grad_norm": 0.46449747681617737,
      "learning_rate": 1.8973112809790168e-05,
      "loss": 0.0909113883972168,
      "step": 3368
    },
    {
      "epoch": 0.4554626108998733,
      "grad_norm": 0.3328554630279541,
      "learning_rate": 1.8966512430558036e-05,
      "loss": 0.08860397338867188,
      "step": 3369
    },
    {
      "epoch": 0.45559780312632026,
      "grad_norm": 0.3225538432598114,
      "learning_rate": 1.8959911225476858e-05,
      "loss": 0.05960369110107422,
      "step": 3370
    },
    {
      "epoch": 0.45573299535276723,
      "grad_norm": 0.29006409645080566,
      "learning_rate": 1.895330919592105e-05,
      "loss": 0.05233478546142578,
      "step": 3371
    },
    {
      "epoch": 0.4558681875792142,
      "grad_norm": 0.30673539638519287,
      "learning_rate": 1.8946706343265187e-05,
      "loss": 0.07146930694580078,
      "step": 3372
    },
    {
      "epoch": 0.4560033798056612,
      "grad_norm": 0.3561408519744873,
      "learning_rate": 1.8940102668884016e-05,
      "loss": 0.09210491180419922,
      "step": 3373
    },
    {
      "epoch": 0.45613857203210817,
      "grad_norm": 0.30946722626686096,
      "learning_rate": 1.893349817415246e-05,
      "loss": 0.06884050369262695,
      "step": 3374
    },
    {
      "epoch": 0.45627376425855515,
      "grad_norm": 0.20836809277534485,
      "learning_rate": 1.8926892860445607e-05,
      "loss": 0.05988717079162598,
      "step": 3375
    },
    {
      "epoch": 0.4564089564850021,
      "grad_norm": 0.42764633893966675,
      "learning_rate": 1.8920286729138718e-05,
      "loss": 0.06943464279174805,
      "step": 3376
    },
    {
      "epoch": 0.4565441487114491,
      "grad_norm": 0.2909482717514038,
      "learning_rate": 1.8913679781607225e-05,
      "loss": 0.0875391960144043,
      "step": 3377
    },
    {
      "epoch": 0.4566793409378961,
      "grad_norm": 0.4012003540992737,
      "learning_rate": 1.8907072019226734e-05,
      "loss": 0.07361888885498047,
      "step": 3378
    },
    {
      "epoch": 0.45681453316434306,
      "grad_norm": 0.33011263608932495,
      "learning_rate": 1.8900463443373015e-05,
      "loss": 0.06682038307189941,
      "step": 3379
    },
    {
      "epoch": 0.45694972539079004,
      "grad_norm": 0.273349791765213,
      "learning_rate": 1.889385405542201e-05,
      "loss": 0.06336069107055664,
      "step": 3380
    },
    {
      "epoch": 0.457084917617237,
      "grad_norm": 0.4372940957546234,
      "learning_rate": 1.8887243856749816e-05,
      "loss": 0.09548282623291016,
      "step": 3381
    },
    {
      "epoch": 0.457220109843684,
      "grad_norm": 0.269704669713974,
      "learning_rate": 1.8880632848732723e-05,
      "loss": 0.06457805633544922,
      "step": 3382
    },
    {
      "epoch": 0.457355302070131,
      "grad_norm": 0.447592556476593,
      "learning_rate": 1.8874021032747185e-05,
      "loss": 0.0651707649230957,
      "step": 3383
    },
    {
      "epoch": 0.45749049429657795,
      "grad_norm": 0.36614465713500977,
      "learning_rate": 1.8867408410169803e-05,
      "loss": 0.04779016971588135,
      "step": 3384
    },
    {
      "epoch": 0.45762568652302493,
      "grad_norm": 0.5801144242286682,
      "learning_rate": 1.886079498237737e-05,
      "loss": 0.10569000244140625,
      "step": 3385
    },
    {
      "epoch": 0.4577608787494719,
      "grad_norm": 0.3168382942676544,
      "learning_rate": 1.885418075074683e-05,
      "loss": 0.07435190677642822,
      "step": 3386
    },
    {
      "epoch": 0.4578960709759189,
      "grad_norm": 0.5302079916000366,
      "learning_rate": 1.884756571665531e-05,
      "loss": 0.08707332611083984,
      "step": 3387
    },
    {
      "epoch": 0.45803126320236587,
      "grad_norm": 0.3065987527370453,
      "learning_rate": 1.8840949881480085e-05,
      "loss": 0.08809423446655273,
      "step": 3388
    },
    {
      "epoch": 0.45816645542881285,
      "grad_norm": 0.25234851241111755,
      "learning_rate": 1.8834333246598613e-05,
      "loss": 0.05010044574737549,
      "step": 3389
    },
    {
      "epoch": 0.4583016476552598,
      "grad_norm": 0.2220805585384369,
      "learning_rate": 1.8827715813388514e-05,
      "loss": 0.0434725284576416,
      "step": 3390
    },
    {
      "epoch": 0.4584368398817068,
      "grad_norm": 0.5903912782669067,
      "learning_rate": 1.8821097583227572e-05,
      "loss": 0.07428836822509766,
      "step": 3391
    },
    {
      "epoch": 0.4585720321081538,
      "grad_norm": 0.38121965527534485,
      "learning_rate": 1.8814478557493732e-05,
      "loss": 0.07053303718566895,
      "step": 3392
    },
    {
      "epoch": 0.45870722433460076,
      "grad_norm": 0.45263683795928955,
      "learning_rate": 1.8807858737565118e-05,
      "loss": 0.07894706726074219,
      "step": 3393
    },
    {
      "epoch": 0.45884241656104774,
      "grad_norm": 0.4440828561782837,
      "learning_rate": 1.880123812482001e-05,
      "loss": 0.09063720703125,
      "step": 3394
    },
    {
      "epoch": 0.4589776087874947,
      "grad_norm": 0.3302997350692749,
      "learning_rate": 1.8794616720636853e-05,
      "loss": 0.05445122718811035,
      "step": 3395
    },
    {
      "epoch": 0.4591128010139417,
      "grad_norm": 0.2598004639148712,
      "learning_rate": 1.8787994526394257e-05,
      "loss": 0.06152057647705078,
      "step": 3396
    },
    {
      "epoch": 0.4592479932403887,
      "grad_norm": 0.3922182619571686,
      "learning_rate": 1.8781371543471e-05,
      "loss": 0.057865142822265625,
      "step": 3397
    },
    {
      "epoch": 0.45938318546683565,
      "grad_norm": 0.21379107236862183,
      "learning_rate": 1.8774747773246024e-05,
      "loss": 0.03534579277038574,
      "step": 3398
    },
    {
      "epoch": 0.45951837769328263,
      "grad_norm": 0.3878244459629059,
      "learning_rate": 1.8768123217098438e-05,
      "loss": 0.057530879974365234,
      "step": 3399
    },
    {
      "epoch": 0.4596535699197296,
      "grad_norm": 0.5795989036560059,
      "learning_rate": 1.8761497876407496e-05,
      "loss": 0.07978522777557373,
      "step": 3400
    },
    {
      "epoch": 0.4597887621461766,
      "grad_norm": 0.23303039371967316,
      "learning_rate": 1.8754871752552646e-05,
      "loss": 0.06266593933105469,
      "step": 3401
    },
    {
      "epoch": 0.45992395437262357,
      "grad_norm": 0.17354288697242737,
      "learning_rate": 1.8748244846913463e-05,
      "loss": 0.036647796630859375,
      "step": 3402
    },
    {
      "epoch": 0.46005914659907055,
      "grad_norm": 0.2165619432926178,
      "learning_rate": 1.874161716086972e-05,
      "loss": 0.0589756965637207,
      "step": 3403
    },
    {
      "epoch": 0.4601943388255175,
      "grad_norm": 0.4398226737976074,
      "learning_rate": 1.8734988695801333e-05,
      "loss": 0.08697986602783203,
      "step": 3404
    },
    {
      "epoch": 0.4603295310519645,
      "grad_norm": 0.3662380874156952,
      "learning_rate": 1.8728359453088382e-05,
      "loss": 0.05473041534423828,
      "step": 3405
    },
    {
      "epoch": 0.4604647232784115,
      "grad_norm": 0.16858461499214172,
      "learning_rate": 1.8721729434111108e-05,
      "loss": 0.037944793701171875,
      "step": 3406
    },
    {
      "epoch": 0.46059991550485846,
      "grad_norm": 0.31228041648864746,
      "learning_rate": 1.871509864024992e-05,
      "loss": 0.08854866027832031,
      "step": 3407
    },
    {
      "epoch": 0.46073510773130544,
      "grad_norm": 0.25661909580230713,
      "learning_rate": 1.8708467072885385e-05,
      "loss": 0.07062864303588867,
      "step": 3408
    },
    {
      "epoch": 0.4608702999577524,
      "grad_norm": 0.2618213891983032,
      "learning_rate": 1.8701834733398227e-05,
      "loss": 0.07455968856811523,
      "step": 3409
    },
    {
      "epoch": 0.4610054921841994,
      "grad_norm": 0.3570909798145294,
      "learning_rate": 1.8695201623169335e-05,
      "loss": 0.049992918968200684,
      "step": 3410
    },
    {
      "epoch": 0.4611406844106464,
      "grad_norm": 0.34414079785346985,
      "learning_rate": 1.868856774357977e-05,
      "loss": 0.07537531852722168,
      "step": 3411
    },
    {
      "epoch": 0.46127587663709335,
      "grad_norm": 0.5453190803527832,
      "learning_rate": 1.868193309601072e-05,
      "loss": 0.06511664390563965,
      "step": 3412
    },
    {
      "epoch": 0.46141106886354033,
      "grad_norm": 0.16751492023468018,
      "learning_rate": 1.867529768184357e-05,
      "loss": 0.05666553974151611,
      "step": 3413
    },
    {
      "epoch": 0.4615462610899873,
      "grad_norm": 0.17946866154670715,
      "learning_rate": 1.8668661502459842e-05,
      "loss": 0.04185080528259277,
      "step": 3414
    },
    {
      "epoch": 0.4616814533164343,
      "grad_norm": 0.17078571021556854,
      "learning_rate": 1.866202455924123e-05,
      "loss": 0.04868030548095703,
      "step": 3415
    },
    {
      "epoch": 0.46181664554288127,
      "grad_norm": 0.22714941203594208,
      "learning_rate": 1.865538685356957e-05,
      "loss": 0.06744003295898438,
      "step": 3416
    },
    {
      "epoch": 0.46195183776932824,
      "grad_norm": 0.3168025314807892,
      "learning_rate": 1.864874838682688e-05,
      "loss": 0.08167481422424316,
      "step": 3417
    },
    {
      "epoch": 0.4620870299957752,
      "grad_norm": 0.3245197832584381,
      "learning_rate": 1.864210916039531e-05,
      "loss": 0.08807134628295898,
      "step": 3418
    },
    {
      "epoch": 0.4622222222222222,
      "grad_norm": 0.6082145571708679,
      "learning_rate": 1.86354691756572e-05,
      "loss": 0.07672929763793945,
      "step": 3419
    },
    {
      "epoch": 0.4623574144486692,
      "grad_norm": 0.4219649136066437,
      "learning_rate": 1.8628828433995013e-05,
      "loss": 0.061176180839538574,
      "step": 3420
    },
    {
      "epoch": 0.46249260667511616,
      "grad_norm": 0.5252535939216614,
      "learning_rate": 1.86221869367914e-05,
      "loss": 0.10416841506958008,
      "step": 3421
    },
    {
      "epoch": 0.46262779890156314,
      "grad_norm": 0.16591042280197144,
      "learning_rate": 1.8615544685429153e-05,
      "loss": 0.0382232666015625,
      "step": 3422
    },
    {
      "epoch": 0.4627629911280101,
      "grad_norm": 0.3893396854400635,
      "learning_rate": 1.860890168129122e-05,
      "loss": 0.0856313705444336,
      "step": 3423
    },
    {
      "epoch": 0.4628981833544571,
      "grad_norm": 0.2738642990589142,
      "learning_rate": 1.8602257925760708e-05,
      "loss": 0.05086398124694824,
      "step": 3424
    },
    {
      "epoch": 0.46303337558090407,
      "grad_norm": 0.22418640553951263,
      "learning_rate": 1.859561342022089e-05,
      "loss": 0.05542564392089844,
      "step": 3425
    },
    {
      "epoch": 0.46316856780735105,
      "grad_norm": 0.2757858633995056,
      "learning_rate": 1.8588968166055185e-05,
      "loss": 0.04795694351196289,
      "step": 3426
    },
    {
      "epoch": 0.4633037600337981,
      "grad_norm": 0.37701842188835144,
      "learning_rate": 1.8582322164647166e-05,
      "loss": 0.06206989288330078,
      "step": 3427
    },
    {
      "epoch": 0.46343895226024506,
      "grad_norm": 0.29527583718299866,
      "learning_rate": 1.8575675417380568e-05,
      "loss": 0.05865907669067383,
      "step": 3428
    },
    {
      "epoch": 0.46357414448669204,
      "grad_norm": 0.554803192615509,
      "learning_rate": 1.856902792563928e-05,
      "loss": 0.08922004699707031,
      "step": 3429
    },
    {
      "epoch": 0.463709336713139,
      "grad_norm": 0.4446096420288086,
      "learning_rate": 1.856237969080735e-05,
      "loss": 0.09359264373779297,
      "step": 3430
    },
    {
      "epoch": 0.463844528939586,
      "grad_norm": 0.2654801607131958,
      "learning_rate": 1.8555730714268967e-05,
      "loss": 0.04187750816345215,
      "step": 3431
    },
    {
      "epoch": 0.463979721166033,
      "grad_norm": 0.29031580686569214,
      "learning_rate": 1.8549080997408492e-05,
      "loss": 0.0662527084350586,
      "step": 3432
    },
    {
      "epoch": 0.46411491339247996,
      "grad_norm": 0.3369176387786865,
      "learning_rate": 1.8542430541610426e-05,
      "loss": 0.06745195388793945,
      "step": 3433
    },
    {
      "epoch": 0.46425010561892693,
      "grad_norm": 0.32542991638183594,
      "learning_rate": 1.8535779348259428e-05,
      "loss": 0.06024777889251709,
      "step": 3434
    },
    {
      "epoch": 0.4643852978453739,
      "grad_norm": 0.19965492188930511,
      "learning_rate": 1.852912741874032e-05,
      "loss": 0.040612220764160156,
      "step": 3435
    },
    {
      "epoch": 0.4645204900718209,
      "grad_norm": 0.23951001465320587,
      "learning_rate": 1.8522474754438056e-05,
      "loss": 0.06544429063796997,
      "step": 3436
    },
    {
      "epoch": 0.46465568229826787,
      "grad_norm": 0.351089209318161,
      "learning_rate": 1.851582135673777e-05,
      "loss": 0.05725383758544922,
      "step": 3437
    },
    {
      "epoch": 0.46479087452471485,
      "grad_norm": 0.2836418151855469,
      "learning_rate": 1.850916722702473e-05,
      "loss": 0.06226921081542969,
      "step": 3438
    },
    {
      "epoch": 0.4649260667511618,
      "grad_norm": 0.3764338791370392,
      "learning_rate": 1.8502512366684355e-05,
      "loss": 0.05451464653015137,
      "step": 3439
    },
    {
      "epoch": 0.4650612589776088,
      "grad_norm": 0.25200191140174866,
      "learning_rate": 1.8495856777102232e-05,
      "loss": 0.0630868673324585,
      "step": 3440
    },
    {
      "epoch": 0.4651964512040558,
      "grad_norm": 0.35389888286590576,
      "learning_rate": 1.848920045966408e-05,
      "loss": 0.05103874206542969,
      "step": 3441
    },
    {
      "epoch": 0.46533164343050276,
      "grad_norm": 0.23646841943264008,
      "learning_rate": 1.8482543415755797e-05,
      "loss": 0.05875825881958008,
      "step": 3442
    },
    {
      "epoch": 0.46546683565694974,
      "grad_norm": 0.30320239067077637,
      "learning_rate": 1.8475885646763394e-05,
      "loss": 0.05823349952697754,
      "step": 3443
    },
    {
      "epoch": 0.4656020278833967,
      "grad_norm": 0.18037791550159454,
      "learning_rate": 1.8469227154073064e-05,
      "loss": 0.04271292686462402,
      "step": 3444
    },
    {
      "epoch": 0.4657372201098437,
      "grad_norm": 0.41756686568260193,
      "learning_rate": 1.8462567939071142e-05,
      "loss": 0.05832624435424805,
      "step": 3445
    },
    {
      "epoch": 0.4658724123362907,
      "grad_norm": 0.5267950296401978,
      "learning_rate": 1.845590800314411e-05,
      "loss": 0.07138395309448242,
      "step": 3446
    },
    {
      "epoch": 0.46600760456273765,
      "grad_norm": 0.30034756660461426,
      "learning_rate": 1.8449247347678607e-05,
      "loss": 0.055419921875,
      "step": 3447
    },
    {
      "epoch": 0.46614279678918463,
      "grad_norm": 0.27279147505760193,
      "learning_rate": 1.8442585974061405e-05,
      "loss": 0.0683819055557251,
      "step": 3448
    },
    {
      "epoch": 0.4662779890156316,
      "grad_norm": 0.28086161613464355,
      "learning_rate": 1.8435923883679452e-05,
      "loss": 0.051483154296875,
      "step": 3449
    },
    {
      "epoch": 0.4664131812420786,
      "grad_norm": 0.2908973693847656,
      "learning_rate": 1.8429261077919818e-05,
      "loss": 0.06496095657348633,
      "step": 3450
    },
    {
      "epoch": 0.46654837346852557,
      "grad_norm": 0.5070411562919617,
      "learning_rate": 1.8422597558169742e-05,
      "loss": 0.07939910888671875,
      "step": 3451
    },
    {
      "epoch": 0.46668356569497255,
      "grad_norm": 0.4878532588481903,
      "learning_rate": 1.84159333258166e-05,
      "loss": 0.08448600769042969,
      "step": 3452
    },
    {
      "epoch": 0.4668187579214195,
      "grad_norm": 0.49835166335105896,
      "learning_rate": 1.8409268382247933e-05,
      "loss": 0.06260395050048828,
      "step": 3453
    },
    {
      "epoch": 0.4669539501478665,
      "grad_norm": 0.24069352447986603,
      "learning_rate": 1.8402602728851405e-05,
      "loss": 0.042691707611083984,
      "step": 3454
    },
    {
      "epoch": 0.4670891423743135,
      "grad_norm": 0.13667982816696167,
      "learning_rate": 1.839593636701484e-05,
      "loss": 0.02815401554107666,
      "step": 3455
    },
    {
      "epoch": 0.46722433460076046,
      "grad_norm": 0.25796830654144287,
      "learning_rate": 1.8389269298126214e-05,
      "loss": 0.047643184661865234,
      "step": 3456
    },
    {
      "epoch": 0.46735952682720744,
      "grad_norm": 0.4226883053779602,
      "learning_rate": 1.838260152357365e-05,
      "loss": 0.0767965316772461,
      "step": 3457
    },
    {
      "epoch": 0.4674947190536544,
      "grad_norm": 0.4205719828605652,
      "learning_rate": 1.837593304474541e-05,
      "loss": 0.0598905086517334,
      "step": 3458
    },
    {
      "epoch": 0.4676299112801014,
      "grad_norm": 0.27018600702285767,
      "learning_rate": 1.836926386302991e-05,
      "loss": 0.0561213493347168,
      "step": 3459
    },
    {
      "epoch": 0.4677651035065484,
      "grad_norm": 0.520823061466217,
      "learning_rate": 1.8362593979815696e-05,
      "loss": 0.06141042709350586,
      "step": 3460
    },
    {
      "epoch": 0.46790029573299535,
      "grad_norm": 0.17237596213817596,
      "learning_rate": 1.8355923396491496e-05,
      "loss": 0.04550826549530029,
      "step": 3461
    },
    {
      "epoch": 0.46803548795944233,
      "grad_norm": 0.21555861830711365,
      "learning_rate": 1.8349252114446138e-05,
      "loss": 0.03649139404296875,
      "step": 3462
    },
    {
      "epoch": 0.4681706801858893,
      "grad_norm": 0.3433397710323334,
      "learning_rate": 1.834258013506864e-05,
      "loss": 0.07322883605957031,
      "step": 3463
    },
    {
      "epoch": 0.4683058724123363,
      "grad_norm": 0.27201583981513977,
      "learning_rate": 1.833590745974813e-05,
      "loss": 0.042952537536621094,
      "step": 3464
    },
    {
      "epoch": 0.46844106463878327,
      "grad_norm": 0.23703953623771667,
      "learning_rate": 1.8329234089873892e-05,
      "loss": 0.0563054084777832,
      "step": 3465
    },
    {
      "epoch": 0.46857625686523025,
      "grad_norm": 0.3415151834487915,
      "learning_rate": 1.8322560026835366e-05,
      "loss": 0.08687686920166016,
      "step": 3466
    },
    {
      "epoch": 0.4687114490916772,
      "grad_norm": 0.22360317409038544,
      "learning_rate": 1.8315885272022125e-05,
      "loss": 0.04466962814331055,
      "step": 3467
    },
    {
      "epoch": 0.4688466413181242,
      "grad_norm": 0.24439865350723267,
      "learning_rate": 1.830920982682389e-05,
      "loss": 0.058303654193878174,
      "step": 3468
    },
    {
      "epoch": 0.4689818335445712,
      "grad_norm": 0.1943369358778,
      "learning_rate": 1.830253369263052e-05,
      "loss": 0.04276081919670105,
      "step": 3469
    },
    {
      "epoch": 0.46911702577101816,
      "grad_norm": 0.39875328540802,
      "learning_rate": 1.8295856870832024e-05,
      "loss": 0.07946610450744629,
      "step": 3470
    },
    {
      "epoch": 0.46925221799746514,
      "grad_norm": 0.16692864894866943,
      "learning_rate": 1.828917936281855e-05,
      "loss": 0.04334312677383423,
      "step": 3471
    },
    {
      "epoch": 0.4693874102239121,
      "grad_norm": 0.2114153653383255,
      "learning_rate": 1.8282501169980396e-05,
      "loss": 0.03536951541900635,
      "step": 3472
    },
    {
      "epoch": 0.4695226024503591,
      "grad_norm": 0.36615613102912903,
      "learning_rate": 1.8275822293707992e-05,
      "loss": 0.08063077926635742,
      "step": 3473
    },
    {
      "epoch": 0.4696577946768061,
      "grad_norm": 0.24879881739616394,
      "learning_rate": 1.8269142735391917e-05,
      "loss": 0.05558371543884277,
      "step": 3474
    },
    {
      "epoch": 0.46979298690325305,
      "grad_norm": 0.316879004240036,
      "learning_rate": 1.8262462496422893e-05,
      "loss": 0.050569355487823486,
      "step": 3475
    },
    {
      "epoch": 0.46992817912970003,
      "grad_norm": 0.5125378370285034,
      "learning_rate": 1.8255781578191778e-05,
      "loss": 0.07561302185058594,
      "step": 3476
    },
    {
      "epoch": 0.470063371356147,
      "grad_norm": 0.4138695299625397,
      "learning_rate": 1.824909998208958e-05,
      "loss": 0.08216714859008789,
      "step": 3477
    },
    {
      "epoch": 0.470198563582594,
      "grad_norm": 0.25304940342903137,
      "learning_rate": 1.8242417709507434e-05,
      "loss": 0.046431541442871094,
      "step": 3478
    },
    {
      "epoch": 0.47033375580904097,
      "grad_norm": 0.3585839569568634,
      "learning_rate": 1.8235734761836637e-05,
      "loss": 0.06269121170043945,
      "step": 3479
    },
    {
      "epoch": 0.47046894803548794,
      "grad_norm": 0.19518537819385529,
      "learning_rate": 1.82290511404686e-05,
      "loss": 0.04918479919433594,
      "step": 3480
    },
    {
      "epoch": 0.4706041402619349,
      "grad_norm": 0.4402546286582947,
      "learning_rate": 1.8222366846794904e-05,
      "loss": 0.08723926544189453,
      "step": 3481
    },
    {
      "epoch": 0.4707393324883819,
      "grad_norm": 0.27418094873428345,
      "learning_rate": 1.8215681882207238e-05,
      "loss": 0.06601524353027344,
      "step": 3482
    },
    {
      "epoch": 0.4708745247148289,
      "grad_norm": 0.24436897039413452,
      "learning_rate": 1.8208996248097462e-05,
      "loss": 0.050644636154174805,
      "step": 3483
    },
    {
      "epoch": 0.47100971694127586,
      "grad_norm": 0.3325497806072235,
      "learning_rate": 1.8202309945857557e-05,
      "loss": 0.04599761962890625,
      "step": 3484
    },
    {
      "epoch": 0.47114490916772284,
      "grad_norm": 0.33765271306037903,
      "learning_rate": 1.8195622976879638e-05,
      "loss": 0.04256880283355713,
      "step": 3485
    },
    {
      "epoch": 0.4712801013941698,
      "grad_norm": 0.20836441218852997,
      "learning_rate": 1.8188935342555977e-05,
      "loss": 0.04271090030670166,
      "step": 3486
    },
    {
      "epoch": 0.4714152936206168,
      "grad_norm": 0.1930689960718155,
      "learning_rate": 1.818224704427897e-05,
      "loss": 0.05206632614135742,
      "step": 3487
    },
    {
      "epoch": 0.47155048584706377,
      "grad_norm": 0.5138967037200928,
      "learning_rate": 1.8175558083441162e-05,
      "loss": 0.08325481414794922,
      "step": 3488
    },
    {
      "epoch": 0.47168567807351075,
      "grad_norm": 0.23282721638679504,
      "learning_rate": 1.8168868461435225e-05,
      "loss": 0.06511831283569336,
      "step": 3489
    },
    {
      "epoch": 0.47182087029995773,
      "grad_norm": 0.3169877231121063,
      "learning_rate": 1.8162178179653977e-05,
      "loss": 0.06747961044311523,
      "step": 3490
    },
    {
      "epoch": 0.4719560625264047,
      "grad_norm": 0.1433219462633133,
      "learning_rate": 1.815548723949037e-05,
      "loss": 0.03924417495727539,
      "step": 3491
    },
    {
      "epoch": 0.4720912547528517,
      "grad_norm": 0.5342644453048706,
      "learning_rate": 1.814879564233749e-05,
      "loss": 0.06855583190917969,
      "step": 3492
    },
    {
      "epoch": 0.47222644697929866,
      "grad_norm": 0.2186155617237091,
      "learning_rate": 1.8142103389588567e-05,
      "loss": 0.06804943084716797,
      "step": 3493
    },
    {
      "epoch": 0.47236163920574564,
      "grad_norm": 0.3618812561035156,
      "learning_rate": 1.813541048263696e-05,
      "loss": 0.05611443519592285,
      "step": 3494
    },
    {
      "epoch": 0.4724968314321927,
      "grad_norm": 0.4073510468006134,
      "learning_rate": 1.8128716922876178e-05,
      "loss": 0.071868896484375,
      "step": 3495
    },
    {
      "epoch": 0.47263202365863966,
      "grad_norm": 0.24067722260951996,
      "learning_rate": 1.812202271169984e-05,
      "loss": 0.052373528480529785,
      "step": 3496
    },
    {
      "epoch": 0.47276721588508663,
      "grad_norm": 0.42137137055397034,
      "learning_rate": 1.8115327850501726e-05,
      "loss": 0.07637220621109009,
      "step": 3497
    },
    {
      "epoch": 0.4729024081115336,
      "grad_norm": 0.2202090471982956,
      "learning_rate": 1.8108632340675746e-05,
      "loss": 0.06374931335449219,
      "step": 3498
    },
    {
      "epoch": 0.4730376003379806,
      "grad_norm": 0.49998244643211365,
      "learning_rate": 1.810193618361593e-05,
      "loss": 0.06397366523742676,
      "step": 3499
    },
    {
      "epoch": 0.47317279256442757,
      "grad_norm": 0.44209524989128113,
      "learning_rate": 1.8095239380716464e-05,
      "loss": 0.07425355911254883,
      "step": 3500
    },
    {
      "epoch": 0.47330798479087455,
      "grad_norm": 0.26591023802757263,
      "learning_rate": 1.808854193337165e-05,
      "loss": 0.07446479797363281,
      "step": 3501
    },
    {
      "epoch": 0.4734431770173215,
      "grad_norm": 0.33701515197753906,
      "learning_rate": 1.8081843842975935e-05,
      "loss": 0.07298707962036133,
      "step": 3502
    },
    {
      "epoch": 0.4735783692437685,
      "grad_norm": 0.31315985321998596,
      "learning_rate": 1.80751451109239e-05,
      "loss": 0.05843305587768555,
      "step": 3503
    },
    {
      "epoch": 0.4737135614702155,
      "grad_norm": 0.33350905776023865,
      "learning_rate": 1.806844573861025e-05,
      "loss": 0.06082010269165039,
      "step": 3504
    },
    {
      "epoch": 0.47384875369666246,
      "grad_norm": 0.3647076189517975,
      "learning_rate": 1.806174572742984e-05,
      "loss": 0.05537247657775879,
      "step": 3505
    },
    {
      "epoch": 0.47398394592310944,
      "grad_norm": 0.25711965560913086,
      "learning_rate": 1.8055045078777634e-05,
      "loss": 0.057493627071380615,
      "step": 3506
    },
    {
      "epoch": 0.4741191381495564,
      "grad_norm": 0.26133498549461365,
      "learning_rate": 1.8048343794048762e-05,
      "loss": 0.07339620590209961,
      "step": 3507
    },
    {
      "epoch": 0.4742543303760034,
      "grad_norm": 0.2561602294445038,
      "learning_rate": 1.8041641874638445e-05,
      "loss": 0.0631871223449707,
      "step": 3508
    },
    {
      "epoch": 0.4743895226024504,
      "grad_norm": 0.31776508688926697,
      "learning_rate": 1.8034939321942077e-05,
      "loss": 0.060988783836364746,
      "step": 3509
    },
    {
      "epoch": 0.47452471482889735,
      "grad_norm": 0.258253812789917,
      "learning_rate": 1.8028236137355154e-05,
      "loss": 0.0681147575378418,
      "step": 3510
    },
    {
      "epoch": 0.47465990705534433,
      "grad_norm": 0.2523079216480255,
      "learning_rate": 1.8021532322273327e-05,
      "loss": 0.04806375503540039,
      "step": 3511
    },
    {
      "epoch": 0.4747950992817913,
      "grad_norm": 0.2886520028114319,
      "learning_rate": 1.8014827878092347e-05,
      "loss": 0.056687116622924805,
      "step": 3512
    },
    {
      "epoch": 0.4749302915082383,
      "grad_norm": 0.3062235116958618,
      "learning_rate": 1.800812280620813e-05,
      "loss": 0.059404850006103516,
      "step": 3513
    },
    {
      "epoch": 0.47506548373468527,
      "grad_norm": 0.4281168282032013,
      "learning_rate": 1.80014171080167e-05,
      "loss": 0.08860492706298828,
      "step": 3514
    },
    {
      "epoch": 0.47520067596113225,
      "grad_norm": 0.3539980351924896,
      "learning_rate": 1.7994710784914227e-05,
      "loss": 0.07063722610473633,
      "step": 3515
    },
    {
      "epoch": 0.4753358681875792,
      "grad_norm": 0.3549995422363281,
      "learning_rate": 1.7988003838297e-05,
      "loss": 0.053465962409973145,
      "step": 3516
    },
    {
      "epoch": 0.4754710604140262,
      "grad_norm": 0.3343566060066223,
      "learning_rate": 1.7981296269561438e-05,
      "loss": 0.06540298461914062,
      "step": 3517
    },
    {
      "epoch": 0.4756062526404732,
      "grad_norm": 0.3658030331134796,
      "learning_rate": 1.7974588080104095e-05,
      "loss": 0.06323480606079102,
      "step": 3518
    },
    {
      "epoch": 0.47574144486692016,
      "grad_norm": 0.23306547105312347,
      "learning_rate": 1.7967879271321648e-05,
      "loss": 0.05868053436279297,
      "step": 3519
    },
    {
      "epoch": 0.47587663709336714,
      "grad_norm": 0.22199088335037231,
      "learning_rate": 1.7961169844610913e-05,
      "loss": 0.06950092315673828,
      "step": 3520
    },
    {
      "epoch": 0.4760118293198141,
      "grad_norm": 0.25085461139678955,
      "learning_rate": 1.795445980136883e-05,
      "loss": 0.06714701652526855,
      "step": 3521
    },
    {
      "epoch": 0.4761470215462611,
      "grad_norm": 0.2103465348482132,
      "learning_rate": 1.794774914299245e-05,
      "loss": 0.04891157150268555,
      "step": 3522
    },
    {
      "epoch": 0.4762822137727081,
      "grad_norm": 0.42396554350852966,
      "learning_rate": 1.794103787087899e-05,
      "loss": 0.06476306915283203,
      "step": 3523
    },
    {
      "epoch": 0.47641740599915505,
      "grad_norm": 0.2529670000076294,
      "learning_rate": 1.7934325986425755e-05,
      "loss": 0.05025029182434082,
      "step": 3524
    },
    {
      "epoch": 0.47655259822560203,
      "grad_norm": 0.25131848454475403,
      "learning_rate": 1.7927613491030204e-05,
      "loss": 0.05663418769836426,
      "step": 3525
    },
    {
      "epoch": 0.476687790452049,
      "grad_norm": 0.3443480432033539,
      "learning_rate": 1.7920900386089913e-05,
      "loss": 0.06392145156860352,
      "step": 3526
    },
    {
      "epoch": 0.476822982678496,
      "grad_norm": 0.37560203671455383,
      "learning_rate": 1.7914186673002588e-05,
      "loss": 0.09542560577392578,
      "step": 3527
    },
    {
      "epoch": 0.47695817490494297,
      "grad_norm": 0.2930920720100403,
      "learning_rate": 1.790747235316605e-05,
      "loss": 0.06230950355529785,
      "step": 3528
    },
    {
      "epoch": 0.47709336713138994,
      "grad_norm": 0.4717599153518677,
      "learning_rate": 1.790075742797827e-05,
      "loss": 0.08771228790283203,
      "step": 3529
    },
    {
      "epoch": 0.4772285593578369,
      "grad_norm": 0.2221323400735855,
      "learning_rate": 1.789404189883732e-05,
      "loss": 0.07851028442382812,
      "step": 3530
    },
    {
      "epoch": 0.4773637515842839,
      "grad_norm": 0.6136850118637085,
      "learning_rate": 1.7887325767141415e-05,
      "loss": 0.10535526275634766,
      "step": 3531
    },
    {
      "epoch": 0.4774989438107309,
      "grad_norm": 0.35282716155052185,
      "learning_rate": 1.7880609034288894e-05,
      "loss": 0.0712127685546875,
      "step": 3532
    },
    {
      "epoch": 0.47763413603717786,
      "grad_norm": 0.3388783037662506,
      "learning_rate": 1.7873891701678208e-05,
      "loss": 0.05991506576538086,
      "step": 3533
    },
    {
      "epoch": 0.47776932826362484,
      "grad_norm": 0.3645012080669403,
      "learning_rate": 1.786717377070794e-05,
      "loss": 0.0606842041015625,
      "step": 3534
    },
    {
      "epoch": 0.4779045204900718,
      "grad_norm": 0.3051288425922394,
      "learning_rate": 1.7860455242776803e-05,
      "loss": 0.06638526916503906,
      "step": 3535
    },
    {
      "epoch": 0.4780397127165188,
      "grad_norm": 0.2696172893047333,
      "learning_rate": 1.7853736119283635e-05,
      "loss": 0.06256294250488281,
      "step": 3536
    },
    {
      "epoch": 0.4781749049429658,
      "grad_norm": 0.21427100896835327,
      "learning_rate": 1.7847016401627388e-05,
      "loss": 0.06046009063720703,
      "step": 3537
    },
    {
      "epoch": 0.47831009716941275,
      "grad_norm": 0.3176337480545044,
      "learning_rate": 1.7840296091207144e-05,
      "loss": 0.061319828033447266,
      "step": 3538
    },
    {
      "epoch": 0.47844528939585973,
      "grad_norm": 0.47059541940689087,
      "learning_rate": 1.7833575189422107e-05,
      "loss": 0.06873583793640137,
      "step": 3539
    },
    {
      "epoch": 0.4785804816223067,
      "grad_norm": 0.2860632538795471,
      "learning_rate": 1.7826853697671604e-05,
      "loss": 0.04511648416519165,
      "step": 3540
    },
    {
      "epoch": 0.4787156738487537,
      "grad_norm": 0.4167746603488922,
      "learning_rate": 1.782013161735509e-05,
      "loss": 0.07744717597961426,
      "step": 3541
    },
    {
      "epoch": 0.47885086607520067,
      "grad_norm": 0.33337217569351196,
      "learning_rate": 1.781340894987213e-05,
      "loss": 0.06517982482910156,
      "step": 3542
    },
    {
      "epoch": 0.47898605830164764,
      "grad_norm": 0.24096325039863586,
      "learning_rate": 1.7806685696622426e-05,
      "loss": 0.06062459945678711,
      "step": 3543
    },
    {
      "epoch": 0.4791212505280946,
      "grad_norm": 0.36147910356521606,
      "learning_rate": 1.779996185900579e-05,
      "loss": 0.0786590576171875,
      "step": 3544
    },
    {
      "epoch": 0.4792564427545416,
      "grad_norm": 0.3622245490550995,
      "learning_rate": 1.7793237438422165e-05,
      "loss": 0.06263208389282227,
      "step": 3545
    },
    {
      "epoch": 0.4793916349809886,
      "grad_norm": 0.7660185098648071,
      "learning_rate": 1.7786512436271617e-05,
      "loss": 0.09419870376586914,
      "step": 3546
    },
    {
      "epoch": 0.47952682720743556,
      "grad_norm": 0.4656512141227722,
      "learning_rate": 1.777978685395431e-05,
      "loss": 0.0979766845703125,
      "step": 3547
    },
    {
      "epoch": 0.47966201943388254,
      "grad_norm": 0.39631086587905884,
      "learning_rate": 1.7773060692870564e-05,
      "loss": 0.06427288055419922,
      "step": 3548
    },
    {
      "epoch": 0.4797972116603295,
      "grad_norm": 0.3351927399635315,
      "learning_rate": 1.7766333954420794e-05,
      "loss": 0.0591578483581543,
      "step": 3549
    },
    {
      "epoch": 0.4799324038867765,
      "grad_norm": 0.3812809884548187,
      "learning_rate": 1.775960664000554e-05,
      "loss": 0.07298469543457031,
      "step": 3550
    },
    {
      "epoch": 0.48006759611322347,
      "grad_norm": 0.2618882358074188,
      "learning_rate": 1.7752878751025463e-05,
      "loss": 0.06702661514282227,
      "step": 3551
    },
    {
      "epoch": 0.48020278833967045,
      "grad_norm": 0.41424980759620667,
      "learning_rate": 1.7746150288881352e-05,
      "loss": 0.07321524620056152,
      "step": 3552
    },
    {
      "epoch": 0.48033798056611743,
      "grad_norm": 0.32623255252838135,
      "learning_rate": 1.7739421254974114e-05,
      "loss": 0.07372140884399414,
      "step": 3553
    },
    {
      "epoch": 0.4804731727925644,
      "grad_norm": 0.5553512573242188,
      "learning_rate": 1.7732691650704756e-05,
      "loss": 0.06461763381958008,
      "step": 3554
    },
    {
      "epoch": 0.4806083650190114,
      "grad_norm": 0.36241671442985535,
      "learning_rate": 1.7725961477474423e-05,
      "loss": 0.05710554122924805,
      "step": 3555
    },
    {
      "epoch": 0.48074355724545836,
      "grad_norm": 0.36378782987594604,
      "learning_rate": 1.7719230736684375e-05,
      "loss": 0.061231374740600586,
      "step": 3556
    },
    {
      "epoch": 0.48087874947190534,
      "grad_norm": 0.16820940375328064,
      "learning_rate": 1.771249942973599e-05,
      "loss": 0.04599440097808838,
      "step": 3557
    },
    {
      "epoch": 0.4810139416983523,
      "grad_norm": 0.27629798650741577,
      "learning_rate": 1.7705767558030756e-05,
      "loss": 0.05648040771484375,
      "step": 3558
    },
    {
      "epoch": 0.4811491339247993,
      "grad_norm": 0.33989623188972473,
      "learning_rate": 1.769903512297029e-05,
      "loss": 0.06895923614501953,
      "step": 3559
    },
    {
      "epoch": 0.4812843261512463,
      "grad_norm": 0.48969021439552307,
      "learning_rate": 1.7692302125956315e-05,
      "loss": 0.0887598991394043,
      "step": 3560
    },
    {
      "epoch": 0.48141951837769326,
      "grad_norm": 0.4426460862159729,
      "learning_rate": 1.768556856839068e-05,
      "loss": 0.07474339008331299,
      "step": 3561
    },
    {
      "epoch": 0.48155471060414023,
      "grad_norm": 0.6918957233428955,
      "learning_rate": 1.767883445167535e-05,
      "loss": 0.0795588493347168,
      "step": 3562
    },
    {
      "epoch": 0.48168990283058727,
      "grad_norm": 0.47357305884361267,
      "learning_rate": 1.7672099777212398e-05,
      "loss": 0.0653543472290039,
      "step": 3563
    },
    {
      "epoch": 0.48182509505703425,
      "grad_norm": 0.3981247544288635,
      "learning_rate": 1.7665364546404034e-05,
      "loss": 0.08488988876342773,
      "step": 3564
    },
    {
      "epoch": 0.4819602872834812,
      "grad_norm": 0.23753295838832855,
      "learning_rate": 1.7658628760652548e-05,
      "loss": 0.04964733123779297,
      "step": 3565
    },
    {
      "epoch": 0.4820954795099282,
      "grad_norm": 0.3822995126247406,
      "learning_rate": 1.765189242136038e-05,
      "loss": 0.08352327346801758,
      "step": 3566
    },
    {
      "epoch": 0.4822306717363752,
      "grad_norm": 0.20389318466186523,
      "learning_rate": 1.7645155529930065e-05,
      "loss": 0.05323982238769531,
      "step": 3567
    },
    {
      "epoch": 0.48236586396282216,
      "grad_norm": 0.6372861266136169,
      "learning_rate": 1.763841808776426e-05,
      "loss": 0.08947277069091797,
      "step": 3568
    },
    {
      "epoch": 0.48250105618926914,
      "grad_norm": 0.3404562175273895,
      "learning_rate": 1.763168009626575e-05,
      "loss": 0.07779765129089355,
      "step": 3569
    },
    {
      "epoch": 0.4826362484157161,
      "grad_norm": 0.38316822052001953,
      "learning_rate": 1.7624941556837406e-05,
      "loss": 0.07512998580932617,
      "step": 3570
    },
    {
      "epoch": 0.4827714406421631,
      "grad_norm": 0.2137921005487442,
      "learning_rate": 1.7618202470882233e-05,
      "loss": 0.04875946044921875,
      "step": 3571
    },
    {
      "epoch": 0.4829066328686101,
      "grad_norm": 0.21961045265197754,
      "learning_rate": 1.7611462839803336e-05,
      "loss": 0.05422782897949219,
      "step": 3572
    },
    {
      "epoch": 0.48304182509505705,
      "grad_norm": 0.5259507894515991,
      "learning_rate": 1.760472266500396e-05,
      "loss": 0.0807027816772461,
      "step": 3573
    },
    {
      "epoch": 0.48317701732150403,
      "grad_norm": 0.24510490894317627,
      "learning_rate": 1.759798194788743e-05,
      "loss": 0.047423362731933594,
      "step": 3574
    },
    {
      "epoch": 0.483312209547951,
      "grad_norm": 0.4417347013950348,
      "learning_rate": 1.75912406898572e-05,
      "loss": 0.06473278999328613,
      "step": 3575
    },
    {
      "epoch": 0.483447401774398,
      "grad_norm": 0.317496657371521,
      "learning_rate": 1.758449889231685e-05,
      "loss": 0.08271288871765137,
      "step": 3576
    },
    {
      "epoch": 0.48358259400084497,
      "grad_norm": 0.2920406758785248,
      "learning_rate": 1.757775655667004e-05,
      "loss": 0.08787870407104492,
      "step": 3577
    },
    {
      "epoch": 0.48371778622729195,
      "grad_norm": 0.25630223751068115,
      "learning_rate": 1.757101368432057e-05,
      "loss": 0.0801386833190918,
      "step": 3578
    },
    {
      "epoch": 0.4838529784537389,
      "grad_norm": 0.33489561080932617,
      "learning_rate": 1.7564270276672343e-05,
      "loss": 0.08162117004394531,
      "step": 3579
    },
    {
      "epoch": 0.4839881706801859,
      "grad_norm": 0.31049832701683044,
      "learning_rate": 1.7557526335129372e-05,
      "loss": 0.07106292247772217,
      "step": 3580
    },
    {
      "epoch": 0.4841233629066329,
      "grad_norm": 0.3761488199234009,
      "learning_rate": 1.7550781861095774e-05,
      "loss": 0.08405113220214844,
      "step": 3581
    },
    {
      "epoch": 0.48425855513307986,
      "grad_norm": 0.28159165382385254,
      "learning_rate": 1.754403685597579e-05,
      "loss": 0.051313042640686035,
      "step": 3582
    },
    {
      "epoch": 0.48439374735952684,
      "grad_norm": 0.2896674871444702,
      "learning_rate": 1.7537291321173773e-05,
      "loss": 0.0895681381225586,
      "step": 3583
    },
    {
      "epoch": 0.4845289395859738,
      "grad_norm": 0.27373892068862915,
      "learning_rate": 1.7530545258094165e-05,
      "loss": 0.06352818012237549,
      "step": 3584
    },
    {
      "epoch": 0.4846641318124208,
      "grad_norm": 0.2834480404853821,
      "learning_rate": 1.7523798668141548e-05,
      "loss": 0.07228279113769531,
      "step": 3585
    },
    {
      "epoch": 0.4847993240388678,
      "grad_norm": 0.38150838017463684,
      "learning_rate": 1.7517051552720584e-05,
      "loss": 0.08531951904296875,
      "step": 3586
    },
    {
      "epoch": 0.48493451626531475,
      "grad_norm": 0.2771611213684082,
      "learning_rate": 1.7510303913236066e-05,
      "loss": 0.0728762149810791,
      "step": 3587
    },
    {
      "epoch": 0.48506970849176173,
      "grad_norm": 0.22732122242450714,
      "learning_rate": 1.7503555751092883e-05,
      "loss": 0.04701519012451172,
      "step": 3588
    },
    {
      "epoch": 0.4852049007182087,
      "grad_norm": 0.4139147102832794,
      "learning_rate": 1.7496807067696046e-05,
      "loss": 0.08082056045532227,
      "step": 3589
    },
    {
      "epoch": 0.4853400929446557,
      "grad_norm": 0.3019137978553772,
      "learning_rate": 1.7490057864450665e-05,
      "loss": 0.06302165985107422,
      "step": 3590
    },
    {
      "epoch": 0.48547528517110267,
      "grad_norm": 0.34268927574157715,
      "learning_rate": 1.748330814276195e-05,
      "loss": 0.07494640350341797,
      "step": 3591
    },
    {
      "epoch": 0.48561047739754964,
      "grad_norm": 0.24578368663787842,
      "learning_rate": 1.7476557904035243e-05,
      "loss": 0.055861711502075195,
      "step": 3592
    },
    {
      "epoch": 0.4857456696239966,
      "grad_norm": 0.3556198179721832,
      "learning_rate": 1.7469807149675973e-05,
      "loss": 0.07169103622436523,
      "step": 3593
    },
    {
      "epoch": 0.4858808618504436,
      "grad_norm": 0.33735206723213196,
      "learning_rate": 1.7463055881089685e-05,
      "loss": 0.08289265632629395,
      "step": 3594
    },
    {
      "epoch": 0.4860160540768906,
      "grad_norm": 0.23104017972946167,
      "learning_rate": 1.7456304099682024e-05,
      "loss": 0.06075930595397949,
      "step": 3595
    },
    {
      "epoch": 0.48615124630333756,
      "grad_norm": 0.23543255031108856,
      "learning_rate": 1.7449551806858756e-05,
      "loss": 0.040868282318115234,
      "step": 3596
    },
    {
      "epoch": 0.48628643852978454,
      "grad_norm": 0.23597250878810883,
      "learning_rate": 1.7442799004025733e-05,
      "loss": 0.055483102798461914,
      "step": 3597
    },
    {
      "epoch": 0.4864216307562315,
      "grad_norm": 0.13425995409488678,
      "learning_rate": 1.7436045692588934e-05,
      "loss": 0.02841949462890625,
      "step": 3598
    },
    {
      "epoch": 0.4865568229826785,
      "grad_norm": 0.3073100745677948,
      "learning_rate": 1.742929187395443e-05,
      "loss": 0.07666206359863281,
      "step": 3599
    },
    {
      "epoch": 0.4866920152091255,
      "grad_norm": 0.21240779757499695,
      "learning_rate": 1.7422537549528402e-05,
      "loss": 0.07803821563720703,
      "step": 3600
    },
    {
      "epoch": 0.48682720743557245,
      "grad_norm": 0.2068926990032196,
      "learning_rate": 1.7415782720717147e-05,
      "loss": 0.0429685115814209,
      "step": 3601
    },
    {
      "epoch": 0.48696239966201943,
      "grad_norm": 0.2717747390270233,
      "learning_rate": 1.740902738892704e-05,
      "loss": 0.07506906986236572,
      "step": 3602
    },
    {
      "epoch": 0.4870975918884664,
      "grad_norm": 0.21194171905517578,
      "learning_rate": 1.7402271555564585e-05,
      "loss": 0.054779052734375,
      "step": 3603
    },
    {
      "epoch": 0.4872327841149134,
      "grad_norm": 0.23456545174121857,
      "learning_rate": 1.739551522203638e-05,
      "loss": 0.06736063957214355,
      "step": 3604
    },
    {
      "epoch": 0.48736797634136036,
      "grad_norm": 0.16652098298072815,
      "learning_rate": 1.738875838974913e-05,
      "loss": 0.045694947242736816,
      "step": 3605
    },
    {
      "epoch": 0.48750316856780734,
      "grad_norm": 0.3408462405204773,
      "learning_rate": 1.7382001060109652e-05,
      "loss": 0.060717105865478516,
      "step": 3606
    },
    {
      "epoch": 0.4876383607942543,
      "grad_norm": 0.33881875872612,
      "learning_rate": 1.7375243234524843e-05,
      "loss": 0.07779502868652344,
      "step": 3607
    },
    {
      "epoch": 0.4877735530207013,
      "grad_norm": 0.4297168254852295,
      "learning_rate": 1.736848491440173e-05,
      "loss": 0.06330251693725586,
      "step": 3608
    },
    {
      "epoch": 0.4879087452471483,
      "grad_norm": 0.3366488814353943,
      "learning_rate": 1.7361726101147424e-05,
      "loss": 0.049425363540649414,
      "step": 3609
    },
    {
      "epoch": 0.48804393747359526,
      "grad_norm": 0.36104920506477356,
      "learning_rate": 1.7354966796169157e-05,
      "loss": 0.07415008544921875,
      "step": 3610
    },
    {
      "epoch": 0.48817912970004224,
      "grad_norm": 0.2824004292488098,
      "learning_rate": 1.734820700087424e-05,
      "loss": 0.0686182975769043,
      "step": 3611
    },
    {
      "epoch": 0.4883143219264892,
      "grad_norm": 0.2889813184738159,
      "learning_rate": 1.7341446716670103e-05,
      "loss": 0.06698274612426758,
      "step": 3612
    },
    {
      "epoch": 0.4884495141529362,
      "grad_norm": 0.3684225082397461,
      "learning_rate": 1.7334685944964272e-05,
      "loss": 0.0602569580078125,
      "step": 3613
    },
    {
      "epoch": 0.48858470637938317,
      "grad_norm": 0.3210425078868866,
      "learning_rate": 1.732792468716438e-05,
      "loss": 0.05359315872192383,
      "step": 3614
    },
    {
      "epoch": 0.48871989860583015,
      "grad_norm": 0.40614524483680725,
      "learning_rate": 1.7321162944678155e-05,
      "loss": 0.06607246398925781,
      "step": 3615
    },
    {
      "epoch": 0.48885509083227713,
      "grad_norm": 0.4153936207294464,
      "learning_rate": 1.7314400718913425e-05,
      "loss": 0.052260637283325195,
      "step": 3616
    },
    {
      "epoch": 0.4889902830587241,
      "grad_norm": 0.25126591324806213,
      "learning_rate": 1.7307638011278126e-05,
      "loss": 0.06325578689575195,
      "step": 3617
    },
    {
      "epoch": 0.4891254752851711,
      "grad_norm": 0.16747547686100006,
      "learning_rate": 1.7300874823180284e-05,
      "loss": 0.03872513771057129,
      "step": 3618
    },
    {
      "epoch": 0.48926066751161806,
      "grad_norm": 0.7757872343063354,
      "learning_rate": 1.7294111156028034e-05,
      "loss": 0.09938335418701172,
      "step": 3619
    },
    {
      "epoch": 0.48939585973806504,
      "grad_norm": 0.2757006585597992,
      "learning_rate": 1.7287347011229605e-05,
      "loss": 0.07147598266601562,
      "step": 3620
    },
    {
      "epoch": 0.489531051964512,
      "grad_norm": 0.20310567319393158,
      "learning_rate": 1.7280582390193333e-05,
      "loss": 0.041744232177734375,
      "step": 3621
    },
    {
      "epoch": 0.489666244190959,
      "grad_norm": 0.2866188585758209,
      "learning_rate": 1.7273817294327653e-05,
      "loss": 0.052403926849365234,
      "step": 3622
    },
    {
      "epoch": 0.489801436417406,
      "grad_norm": 0.25933730602264404,
      "learning_rate": 1.726705172504108e-05,
      "loss": 0.08164167404174805,
      "step": 3623
    },
    {
      "epoch": 0.48993662864385296,
      "grad_norm": 0.2824884355068207,
      "learning_rate": 1.7260285683742248e-05,
      "loss": 0.046981215476989746,
      "step": 3624
    },
    {
      "epoch": 0.49007182087029993,
      "grad_norm": 0.12598471343517303,
      "learning_rate": 1.7253519171839883e-05,
      "loss": 0.03424215316772461,
      "step": 3625
    },
    {
      "epoch": 0.4902070130967469,
      "grad_norm": 0.38121935725212097,
      "learning_rate": 1.724675219074281e-05,
      "loss": 0.06019115447998047,
      "step": 3626
    },
    {
      "epoch": 0.4903422053231939,
      "grad_norm": 0.14721521735191345,
      "learning_rate": 1.7239984741859957e-05,
      "loss": 0.027959585189819336,
      "step": 3627
    },
    {
      "epoch": 0.49047739754964087,
      "grad_norm": 0.3771783113479614,
      "learning_rate": 1.7233216826600324e-05,
      "loss": 0.07523393630981445,
      "step": 3628
    },
    {
      "epoch": 0.49061258977608785,
      "grad_norm": 0.1825200766324997,
      "learning_rate": 1.7226448446373047e-05,
      "loss": 0.047032833099365234,
      "step": 3629
    },
    {
      "epoch": 0.4907477820025348,
      "grad_norm": 0.637860119342804,
      "learning_rate": 1.7219679602587326e-05,
      "loss": 0.1150738000869751,
      "step": 3630
    },
    {
      "epoch": 0.49088297422898186,
      "grad_norm": 0.311662882566452,
      "learning_rate": 1.7212910296652476e-05,
      "loss": 0.06574392318725586,
      "step": 3631
    },
    {
      "epoch": 0.49101816645542884,
      "grad_norm": 0.16438733041286469,
      "learning_rate": 1.7206140529977904e-05,
      "loss": 0.04515564441680908,
      "step": 3632
    },
    {
      "epoch": 0.4911533586818758,
      "grad_norm": 0.22129812836647034,
      "learning_rate": 1.719937030397311e-05,
      "loss": 0.05523681640625,
      "step": 3633
    },
    {
      "epoch": 0.4912885509083228,
      "grad_norm": 0.15538308024406433,
      "learning_rate": 1.7192599620047683e-05,
      "loss": 0.0331113338470459,
      "step": 3634
    },
    {
      "epoch": 0.4914237431347698,
      "grad_norm": 0.3123125731945038,
      "learning_rate": 1.718582847961133e-05,
      "loss": 0.06272220611572266,
      "step": 3635
    },
    {
      "epoch": 0.49155893536121675,
      "grad_norm": 0.19490057229995728,
      "learning_rate": 1.7179056884073826e-05,
      "loss": 0.06317687034606934,
      "step": 3636
    },
    {
      "epoch": 0.49169412758766373,
      "grad_norm": 0.45507940649986267,
      "learning_rate": 1.717228483484506e-05,
      "loss": 0.055414438247680664,
      "step": 3637
    },
    {
      "epoch": 0.4918293198141107,
      "grad_norm": 0.5364943742752075,
      "learning_rate": 1.7165512333335013e-05,
      "loss": 0.06654071807861328,
      "step": 3638
    },
    {
      "epoch": 0.4919645120405577,
      "grad_norm": 0.370889276266098,
      "learning_rate": 1.715873938095374e-05,
      "loss": 0.07144880294799805,
      "step": 3639
    },
    {
      "epoch": 0.49209970426700467,
      "grad_norm": 0.21666491031646729,
      "learning_rate": 1.7151965979111427e-05,
      "loss": 0.06777763366699219,
      "step": 3640
    },
    {
      "epoch": 0.49223489649345165,
      "grad_norm": 0.24387285113334656,
      "learning_rate": 1.7145192129218313e-05,
      "loss": 0.06093788146972656,
      "step": 3641
    },
    {
      "epoch": 0.4923700887198986,
      "grad_norm": 0.344928503036499,
      "learning_rate": 1.7138417832684763e-05,
      "loss": 0.07947254180908203,
      "step": 3642
    },
    {
      "epoch": 0.4925052809463456,
      "grad_norm": 0.22023750841617584,
      "learning_rate": 1.7131643090921216e-05,
      "loss": 0.053901731967926025,
      "step": 3643
    },
    {
      "epoch": 0.4926404731727926,
      "grad_norm": 0.30977097153663635,
      "learning_rate": 1.712486790533821e-05,
      "loss": 0.08480358123779297,
      "step": 3644
    },
    {
      "epoch": 0.49277566539923956,
      "grad_norm": 0.2903079688549042,
      "learning_rate": 1.7118092277346372e-05,
      "loss": 0.06642723083496094,
      "step": 3645
    },
    {
      "epoch": 0.49291085762568654,
      "grad_norm": 0.3772442638874054,
      "learning_rate": 1.7111316208356428e-05,
      "loss": 0.08548742532730103,
      "step": 3646
    },
    {
      "epoch": 0.4930460498521335,
      "grad_norm": 0.12240912765264511,
      "learning_rate": 1.7104539699779192e-05,
      "loss": 0.0373530387878418,
      "step": 3647
    },
    {
      "epoch": 0.4931812420785805,
      "grad_norm": 0.2652791142463684,
      "learning_rate": 1.709776275302557e-05,
      "loss": 0.0529026985168457,
      "step": 3648
    },
    {
      "epoch": 0.4933164343050275,
      "grad_norm": 0.33920103311538696,
      "learning_rate": 1.7090985369506555e-05,
      "loss": 0.09222221374511719,
      "step": 3649
    },
    {
      "epoch": 0.49345162653147445,
      "grad_norm": 0.2678898870944977,
      "learning_rate": 1.708420755063323e-05,
      "loss": 0.05522608757019043,
      "step": 3650
    },
    {
      "epoch": 0.49358681875792143,
      "grad_norm": 0.18864497542381287,
      "learning_rate": 1.707742929781678e-05,
      "loss": 0.04535555839538574,
      "step": 3651
    },
    {
      "epoch": 0.4937220109843684,
      "grad_norm": 0.16273167729377747,
      "learning_rate": 1.707065061246848e-05,
      "loss": 0.047704219818115234,
      "step": 3652
    },
    {
      "epoch": 0.4938572032108154,
      "grad_norm": 0.3690492510795593,
      "learning_rate": 1.7063871495999677e-05,
      "loss": 0.07541942596435547,
      "step": 3653
    },
    {
      "epoch": 0.49399239543726237,
      "grad_norm": 0.26042553782463074,
      "learning_rate": 1.705709194982182e-05,
      "loss": 0.06457853317260742,
      "step": 3654
    },
    {
      "epoch": 0.49412758766370934,
      "grad_norm": 0.32687294483184814,
      "learning_rate": 1.7050311975346447e-05,
      "loss": 0.08898448944091797,
      "step": 3655
    },
    {
      "epoch": 0.4942627798901563,
      "grad_norm": 0.3517727851867676,
      "learning_rate": 1.704353157398519e-05,
      "loss": 0.06552386283874512,
      "step": 3656
    },
    {
      "epoch": 0.4943979721166033,
      "grad_norm": 0.26627597212791443,
      "learning_rate": 1.7036750747149764e-05,
      "loss": 0.05442667007446289,
      "step": 3657
    },
    {
      "epoch": 0.4945331643430503,
      "grad_norm": 0.2553832232952118,
      "learning_rate": 1.702996949625197e-05,
      "loss": 0.05390191078186035,
      "step": 3658
    },
    {
      "epoch": 0.49466835656949726,
      "grad_norm": 0.26399871706962585,
      "learning_rate": 1.7023187822703702e-05,
      "loss": 0.06447291374206543,
      "step": 3659
    },
    {
      "epoch": 0.49480354879594424,
      "grad_norm": 0.3089481294155121,
      "learning_rate": 1.7016405727916936e-05,
      "loss": 0.06996846199035645,
      "step": 3660
    },
    {
      "epoch": 0.4949387410223912,
      "grad_norm": 0.4111063778400421,
      "learning_rate": 1.700962321330375e-05,
      "loss": 0.08368968963623047,
      "step": 3661
    },
    {
      "epoch": 0.4950739332488382,
      "grad_norm": 0.33295831084251404,
      "learning_rate": 1.700284028027629e-05,
      "loss": 0.07135665416717529,
      "step": 3662
    },
    {
      "epoch": 0.4952091254752852,
      "grad_norm": 0.24843953549861908,
      "learning_rate": 1.6996056930246807e-05,
      "loss": 0.056128501892089844,
      "step": 3663
    },
    {
      "epoch": 0.49534431770173215,
      "grad_norm": 0.2934357523918152,
      "learning_rate": 1.6989273164627626e-05,
      "loss": 0.0641700029373169,
      "step": 3664
    },
    {
      "epoch": 0.49547950992817913,
      "grad_norm": 0.5144562125205994,
      "learning_rate": 1.6982488984831163e-05,
      "loss": 0.06073331832885742,
      "step": 3665
    },
    {
      "epoch": 0.4956147021546261,
      "grad_norm": 0.42985787987709045,
      "learning_rate": 1.697570439226992e-05,
      "loss": 0.04372763633728027,
      "step": 3666
    },
    {
      "epoch": 0.4957498943810731,
      "grad_norm": 0.7460087537765503,
      "learning_rate": 1.6968919388356486e-05,
      "loss": 0.10422563552856445,
      "step": 3667
    },
    {
      "epoch": 0.49588508660752006,
      "grad_norm": 0.2547072768211365,
      "learning_rate": 1.696213397450354e-05,
      "loss": 0.05795717239379883,
      "step": 3668
    },
    {
      "epoch": 0.49602027883396704,
      "grad_norm": 0.5143445730209351,
      "learning_rate": 1.695534815212384e-05,
      "loss": 0.07362556457519531,
      "step": 3669
    },
    {
      "epoch": 0.496155471060414,
      "grad_norm": 0.5136200785636902,
      "learning_rate": 1.6948561922630223e-05,
      "loss": 0.08347606658935547,
      "step": 3670
    },
    {
      "epoch": 0.496290663286861,
      "grad_norm": 0.2744808793067932,
      "learning_rate": 1.694177528743562e-05,
      "loss": 0.06880831718444824,
      "step": 3671
    },
    {
      "epoch": 0.496425855513308,
      "grad_norm": 0.43338385224342346,
      "learning_rate": 1.6934988247953053e-05,
      "loss": 0.09651374816894531,
      "step": 3672
    },
    {
      "epoch": 0.49656104773975496,
      "grad_norm": 0.47571730613708496,
      "learning_rate": 1.6928200805595606e-05,
      "loss": 0.07631063461303711,
      "step": 3673
    },
    {
      "epoch": 0.49669623996620194,
      "grad_norm": 0.2577090859413147,
      "learning_rate": 1.6921412961776475e-05,
      "loss": 0.06802845001220703,
      "step": 3674
    },
    {
      "epoch": 0.4968314321926489,
      "grad_norm": 0.3719443380832672,
      "learning_rate": 1.6914624717908922e-05,
      "loss": 0.07479619979858398,
      "step": 3675
    },
    {
      "epoch": 0.4969666244190959,
      "grad_norm": 0.4411269724369049,
      "learning_rate": 1.6907836075406288e-05,
      "loss": 0.06498908996582031,
      "step": 3676
    },
    {
      "epoch": 0.49710181664554287,
      "grad_norm": 0.16571342945098877,
      "learning_rate": 1.690104703568201e-05,
      "loss": 0.04548358917236328,
      "step": 3677
    },
    {
      "epoch": 0.49723700887198985,
      "grad_norm": 0.3210400640964508,
      "learning_rate": 1.68942576001496e-05,
      "loss": 0.060634732246398926,
      "step": 3678
    },
    {
      "epoch": 0.49737220109843683,
      "grad_norm": 0.2716442346572876,
      "learning_rate": 1.6887467770222658e-05,
      "loss": 0.05343437194824219,
      "step": 3679
    },
    {
      "epoch": 0.4975073933248838,
      "grad_norm": 0.3123975396156311,
      "learning_rate": 1.6880677547314865e-05,
      "loss": 0.06958913803100586,
      "step": 3680
    },
    {
      "epoch": 0.4976425855513308,
      "grad_norm": 0.5601609945297241,
      "learning_rate": 1.6873886932839973e-05,
      "loss": 0.06643915176391602,
      "step": 3681
    },
    {
      "epoch": 0.49777777777777776,
      "grad_norm": 0.32763174176216125,
      "learning_rate": 1.686709592821183e-05,
      "loss": 0.045107483863830566,
      "step": 3682
    },
    {
      "epoch": 0.49791297000422474,
      "grad_norm": 0.2631048560142517,
      "learning_rate": 1.6860304534844355e-05,
      "loss": 0.06543707847595215,
      "step": 3683
    },
    {
      "epoch": 0.4980481622306717,
      "grad_norm": 0.45689377188682556,
      "learning_rate": 1.6853512754151556e-05,
      "loss": 0.09626007080078125,
      "step": 3684
    },
    {
      "epoch": 0.4981833544571187,
      "grad_norm": 0.18219901621341705,
      "learning_rate": 1.684672058754752e-05,
      "loss": 0.06010627746582031,
      "step": 3685
    },
    {
      "epoch": 0.4983185466835657,
      "grad_norm": 0.28514334559440613,
      "learning_rate": 1.6839928036446416e-05,
      "loss": 0.08095073699951172,
      "step": 3686
    },
    {
      "epoch": 0.49845373891001266,
      "grad_norm": 0.40925583243370056,
      "learning_rate": 1.6833135102262473e-05,
      "loss": 0.08321571350097656,
      "step": 3687
    },
    {
      "epoch": 0.49858893113645963,
      "grad_norm": 0.5032579302787781,
      "learning_rate": 1.682634178641003e-05,
      "loss": 0.06683099269866943,
      "step": 3688
    },
    {
      "epoch": 0.4987241233629066,
      "grad_norm": 0.2649838626384735,
      "learning_rate": 1.6819548090303485e-05,
      "loss": 0.08434677124023438,
      "step": 3689
    },
    {
      "epoch": 0.4988593155893536,
      "grad_norm": 0.292099267244339,
      "learning_rate": 1.6812754015357328e-05,
      "loss": 0.08489036560058594,
      "step": 3690
    },
    {
      "epoch": 0.49899450781580057,
      "grad_norm": 0.2250075489282608,
      "learning_rate": 1.680595956298612e-05,
      "loss": 0.054587364196777344,
      "step": 3691
    },
    {
      "epoch": 0.49912970004224755,
      "grad_norm": 0.23876000940799713,
      "learning_rate": 1.6799164734604497e-05,
      "loss": 0.05211949348449707,
      "step": 3692
    },
    {
      "epoch": 0.4992648922686945,
      "grad_norm": 0.5526087880134583,
      "learning_rate": 1.6792369531627186e-05,
      "loss": 0.07991886138916016,
      "step": 3693
    },
    {
      "epoch": 0.4994000844951415,
      "grad_norm": 0.21991519629955292,
      "learning_rate": 1.6785573955468974e-05,
      "loss": 0.06328868865966797,
      "step": 3694
    },
    {
      "epoch": 0.4995352767215885,
      "grad_norm": 0.25740939378738403,
      "learning_rate": 1.6778778007544745e-05,
      "loss": 0.05763810873031616,
      "step": 3695
    },
    {
      "epoch": 0.49967046894803546,
      "grad_norm": 0.22257405519485474,
      "learning_rate": 1.6771981689269452e-05,
      "loss": 0.05683422088623047,
      "step": 3696
    },
    {
      "epoch": 0.49980566117448244,
      "grad_norm": 0.23424720764160156,
      "learning_rate": 1.6765185002058123e-05,
      "loss": 0.0545077919960022,
      "step": 3697
    },
    {
      "epoch": 0.4999408534009294,
      "grad_norm": 0.24864928424358368,
      "learning_rate": 1.6758387947325856e-05,
      "loss": 0.07478094100952148,
      "step": 3698
    },
    {
      "epoch": 0.5000760456273764,
      "grad_norm": 0.26618900895118713,
      "learning_rate": 1.6751590526487843e-05,
      "loss": 0.044876694679260254,
      "step": 3699
    },
    {
      "epoch": 0.5002112378538234,
      "grad_norm": 0.3026556670665741,
      "learning_rate": 1.6744792740959347e-05,
      "loss": 0.06730842590332031,
      "step": 3700
    },
    {
      "epoch": 0.5003464300802704,
      "grad_norm": 0.573049783706665,
      "learning_rate": 1.6737994592155697e-05,
      "loss": 0.0917515754699707,
      "step": 3701
    },
    {
      "epoch": 0.5004816223067173,
      "grad_norm": 0.15614494681358337,
      "learning_rate": 1.6731196081492307e-05,
      "loss": 0.04000115394592285,
      "step": 3702
    },
    {
      "epoch": 0.5006168145331643,
      "grad_norm": 0.24104918539524078,
      "learning_rate": 1.6724397210384655e-05,
      "loss": 0.04947304725646973,
      "step": 3703
    },
    {
      "epoch": 0.5007520067596113,
      "grad_norm": 0.2229125201702118,
      "learning_rate": 1.6717597980248308e-05,
      "loss": 0.0579066276550293,
      "step": 3704
    },
    {
      "epoch": 0.5008871989860583,
      "grad_norm": 0.3010246455669403,
      "learning_rate": 1.6710798392498904e-05,
      "loss": 0.06881427764892578,
      "step": 3705
    },
    {
      "epoch": 0.5010223912125052,
      "grad_norm": 0.6752832531929016,
      "learning_rate": 1.6703998448552154e-05,
      "loss": 0.10321807861328125,
      "step": 3706
    },
    {
      "epoch": 0.5011575834389522,
      "grad_norm": 0.3871508240699768,
      "learning_rate": 1.669719814982384e-05,
      "loss": 0.06765806674957275,
      "step": 3707
    },
    {
      "epoch": 0.5012927756653992,
      "grad_norm": 0.47764483094215393,
      "learning_rate": 1.6690397497729818e-05,
      "loss": 0.08770179748535156,
      "step": 3708
    },
    {
      "epoch": 0.5014279678918462,
      "grad_norm": 0.2824943959712982,
      "learning_rate": 1.6683596493686028e-05,
      "loss": 0.060486793518066406,
      "step": 3709
    },
    {
      "epoch": 0.5015631601182932,
      "grad_norm": 0.26470473408699036,
      "learning_rate": 1.667679513910846e-05,
      "loss": 0.06539344787597656,
      "step": 3710
    },
    {
      "epoch": 0.5016983523447401,
      "grad_norm": 0.34304139018058777,
      "learning_rate": 1.666999343541321e-05,
      "loss": 0.04815506935119629,
      "step": 3711
    },
    {
      "epoch": 0.5018335445711871,
      "grad_norm": 0.28561848402023315,
      "learning_rate": 1.6663191384016422e-05,
      "loss": 0.06753778457641602,
      "step": 3712
    },
    {
      "epoch": 0.5019687367976341,
      "grad_norm": 0.285433828830719,
      "learning_rate": 1.6656388986334315e-05,
      "loss": 0.0729752779006958,
      "step": 3713
    },
    {
      "epoch": 0.5021039290240811,
      "grad_norm": 0.32891544699668884,
      "learning_rate": 1.6649586243783186e-05,
      "loss": 0.0743703842163086,
      "step": 3714
    },
    {
      "epoch": 0.502239121250528,
      "grad_norm": 0.3816950023174286,
      "learning_rate": 1.6642783157779405e-05,
      "loss": 0.07209134101867676,
      "step": 3715
    },
    {
      "epoch": 0.502374313476975,
      "grad_norm": 0.263702392578125,
      "learning_rate": 1.6635979729739417e-05,
      "loss": 0.05934882164001465,
      "step": 3716
    },
    {
      "epoch": 0.502509505703422,
      "grad_norm": 0.6623087525367737,
      "learning_rate": 1.662917596107972e-05,
      "loss": 0.10495758056640625,
      "step": 3717
    },
    {
      "epoch": 0.502644697929869,
      "grad_norm": 0.40381714701652527,
      "learning_rate": 1.6622371853216904e-05,
      "loss": 0.06665229797363281,
      "step": 3718
    },
    {
      "epoch": 0.502779890156316,
      "grad_norm": 0.28729915618896484,
      "learning_rate": 1.661556740756761e-05,
      "loss": 0.04879331588745117,
      "step": 3719
    },
    {
      "epoch": 0.502915082382763,
      "grad_norm": 0.18054580688476562,
      "learning_rate": 1.6608762625548572e-05,
      "loss": 0.049221038818359375,
      "step": 3720
    },
    {
      "epoch": 0.5030502746092099,
      "grad_norm": 0.24564173817634583,
      "learning_rate": 1.6601957508576573e-05,
      "loss": 0.06606054306030273,
      "step": 3721
    },
    {
      "epoch": 0.5031854668356569,
      "grad_norm": 0.33070167899131775,
      "learning_rate": 1.659515205806848e-05,
      "loss": 0.07685279846191406,
      "step": 3722
    },
    {
      "epoch": 0.5033206590621039,
      "grad_norm": 0.2144303321838379,
      "learning_rate": 1.6588346275441224e-05,
      "loss": 0.0387420654296875,
      "step": 3723
    },
    {
      "epoch": 0.5034558512885509,
      "grad_norm": 0.3891025483608246,
      "learning_rate": 1.65815401621118e-05,
      "loss": 0.07223653793334961,
      "step": 3724
    },
    {
      "epoch": 0.5035910435149978,
      "grad_norm": 0.2672344148159027,
      "learning_rate": 1.657473371949728e-05,
      "loss": 0.06636923551559448,
      "step": 3725
    },
    {
      "epoch": 0.5037262357414448,
      "grad_norm": 0.23694884777069092,
      "learning_rate": 1.6567926949014805e-05,
      "loss": 0.06660652160644531,
      "step": 3726
    },
    {
      "epoch": 0.5038614279678918,
      "grad_norm": 0.21389798820018768,
      "learning_rate": 1.6561119852081574e-05,
      "loss": 0.06486654281616211,
      "step": 3727
    },
    {
      "epoch": 0.5039966201943388,
      "grad_norm": 0.19706472754478455,
      "learning_rate": 1.6554312430114868e-05,
      "loss": 0.05663442611694336,
      "step": 3728
    },
    {
      "epoch": 0.5041318124207858,
      "grad_norm": 0.12402030825614929,
      "learning_rate": 1.6547504684532026e-05,
      "loss": 0.027398109436035156,
      "step": 3729
    },
    {
      "epoch": 0.5042670046472327,
      "grad_norm": 0.17284323275089264,
      "learning_rate": 1.6540696616750454e-05,
      "loss": 0.0577009916305542,
      "step": 3730
    },
    {
      "epoch": 0.5044021968736797,
      "grad_norm": 0.22955511510372162,
      "learning_rate": 1.6533888228187628e-05,
      "loss": 0.05922102928161621,
      "step": 3731
    },
    {
      "epoch": 0.5045373891001267,
      "grad_norm": 0.23436099290847778,
      "learning_rate": 1.6527079520261103e-05,
      "loss": 0.06249058246612549,
      "step": 3732
    },
    {
      "epoch": 0.5046725813265738,
      "grad_norm": 0.4458872377872467,
      "learning_rate": 1.6520270494388472e-05,
      "loss": 0.09632301330566406,
      "step": 3733
    },
    {
      "epoch": 0.5048077735530208,
      "grad_norm": 0.2628108859062195,
      "learning_rate": 1.6513461151987418e-05,
      "loss": 0.06567001342773438,
      "step": 3734
    },
    {
      "epoch": 0.5049429657794677,
      "grad_norm": 0.2242846041917801,
      "learning_rate": 1.6506651494475678e-05,
      "loss": 0.05688738822937012,
      "step": 3735
    },
    {
      "epoch": 0.5050781580059147,
      "grad_norm": 0.18838833272457123,
      "learning_rate": 1.6499841523271062e-05,
      "loss": 0.05216526985168457,
      "step": 3736
    },
    {
      "epoch": 0.5052133502323617,
      "grad_norm": 0.5324968099594116,
      "learning_rate": 1.649303123979145e-05,
      "loss": 0.09590935707092285,
      "step": 3737
    },
    {
      "epoch": 0.5053485424588087,
      "grad_norm": 0.2790025472640991,
      "learning_rate": 1.648622064545477e-05,
      "loss": 0.06006813049316406,
      "step": 3738
    },
    {
      "epoch": 0.5054837346852556,
      "grad_norm": 0.23045334219932556,
      "learning_rate": 1.6479409741679025e-05,
      "loss": 0.044718146324157715,
      "step": 3739
    },
    {
      "epoch": 0.5056189269117026,
      "grad_norm": 0.44952282309532166,
      "learning_rate": 1.6472598529882277e-05,
      "loss": 0.06872034072875977,
      "step": 3740
    },
    {
      "epoch": 0.5057541191381496,
      "grad_norm": 0.15813127160072327,
      "learning_rate": 1.646578701148267e-05,
      "loss": 0.04623889923095703,
      "step": 3741
    },
    {
      "epoch": 0.5058893113645966,
      "grad_norm": 0.4009650647640228,
      "learning_rate": 1.6458975187898384e-05,
      "loss": 0.07753944396972656,
      "step": 3742
    },
    {
      "epoch": 0.5060245035910436,
      "grad_norm": 0.3411290943622589,
      "learning_rate": 1.6452163060547687e-05,
      "loss": 0.08057403564453125,
      "step": 3743
    },
    {
      "epoch": 0.5061596958174905,
      "grad_norm": 0.20344440639019012,
      "learning_rate": 1.64453506308489e-05,
      "loss": 0.052147626876831055,
      "step": 3744
    },
    {
      "epoch": 0.5062948880439375,
      "grad_norm": 0.3297085464000702,
      "learning_rate": 1.64385379002204e-05,
      "loss": 0.0858154296875,
      "step": 3745
    },
    {
      "epoch": 0.5064300802703845,
      "grad_norm": 0.26425638794898987,
      "learning_rate": 1.643172487008064e-05,
      "loss": 0.058515310287475586,
      "step": 3746
    },
    {
      "epoch": 0.5065652724968315,
      "grad_norm": 0.25333982706069946,
      "learning_rate": 1.6424911541848124e-05,
      "loss": 0.04899787902832031,
      "step": 3747
    },
    {
      "epoch": 0.5067004647232785,
      "grad_norm": 0.403068870306015,
      "learning_rate": 1.641809791694143e-05,
      "loss": 0.05936574935913086,
      "step": 3748
    },
    {
      "epoch": 0.5068356569497254,
      "grad_norm": 0.3366691768169403,
      "learning_rate": 1.6411283996779184e-05,
      "loss": 0.07269346714019775,
      "step": 3749
    },
    {
      "epoch": 0.5069708491761724,
      "grad_norm": 0.4272845685482025,
      "learning_rate": 1.6404469782780088e-05,
      "loss": 0.0903940200805664,
      "step": 3750
    },
    {
      "epoch": 0.5071060414026194,
      "grad_norm": 0.26125475764274597,
      "learning_rate": 1.639765527636289e-05,
      "loss": 0.06279802322387695,
      "step": 3751
    },
    {
      "epoch": 0.5072412336290664,
      "grad_norm": 0.2773575484752655,
      "learning_rate": 1.639084047894641e-05,
      "loss": 0.06950044631958008,
      "step": 3752
    },
    {
      "epoch": 0.5073764258555133,
      "grad_norm": 0.335889607667923,
      "learning_rate": 1.638402539194953e-05,
      "loss": 0.09420204162597656,
      "step": 3753
    },
    {
      "epoch": 0.5075116180819603,
      "grad_norm": 0.5095440745353699,
      "learning_rate": 1.6377210016791182e-05,
      "loss": 0.07118093967437744,
      "step": 3754
    },
    {
      "epoch": 0.5076468103084073,
      "grad_norm": 0.31784096360206604,
      "learning_rate": 1.6370394354890364e-05,
      "loss": 0.0667428970336914,
      "step": 3755
    },
    {
      "epoch": 0.5077820025348543,
      "grad_norm": 0.4596634805202484,
      "learning_rate": 1.636357840766613e-05,
      "loss": 0.0998077392578125,
      "step": 3756
    },
    {
      "epoch": 0.5079171947613013,
      "grad_norm": 0.16048742830753326,
      "learning_rate": 1.6356762176537606e-05,
      "loss": 0.03633636236190796,
      "step": 3757
    },
    {
      "epoch": 0.5080523869877482,
      "grad_norm": 0.2662222683429718,
      "learning_rate": 1.6349945662923953e-05,
      "loss": 0.0802011489868164,
      "step": 3758
    },
    {
      "epoch": 0.5081875792141952,
      "grad_norm": 0.23478074371814728,
      "learning_rate": 1.634312886824442e-05,
      "loss": 0.0711979866027832,
      "step": 3759
    },
    {
      "epoch": 0.5083227714406422,
      "grad_norm": 0.21493028104305267,
      "learning_rate": 1.6336311793918298e-05,
      "loss": 0.05287933349609375,
      "step": 3760
    },
    {
      "epoch": 0.5084579636670892,
      "grad_norm": 0.42671141028404236,
      "learning_rate": 1.6329494441364925e-05,
      "loss": 0.10504579544067383,
      "step": 3761
    },
    {
      "epoch": 0.5085931558935362,
      "grad_norm": 0.18574018776416779,
      "learning_rate": 1.6322676812003727e-05,
      "loss": 0.043202877044677734,
      "step": 3762
    },
    {
      "epoch": 0.5087283481199831,
      "grad_norm": 0.2294776737689972,
      "learning_rate": 1.631585890725416e-05,
      "loss": 0.048957228660583496,
      "step": 3763
    },
    {
      "epoch": 0.5088635403464301,
      "grad_norm": 0.18813537061214447,
      "learning_rate": 1.630904072853575e-05,
      "loss": 0.04468345642089844,
      "step": 3764
    },
    {
      "epoch": 0.5089987325728771,
      "grad_norm": 0.4334390163421631,
      "learning_rate": 1.6302222277268085e-05,
      "loss": 0.10156059265136719,
      "step": 3765
    },
    {
      "epoch": 0.5091339247993241,
      "grad_norm": 0.2519180476665497,
      "learning_rate": 1.6295403554870794e-05,
      "loss": 0.05563712120056152,
      "step": 3766
    },
    {
      "epoch": 0.509269117025771,
      "grad_norm": 0.33654114603996277,
      "learning_rate": 1.6288584562763572e-05,
      "loss": 0.0795598030090332,
      "step": 3767
    },
    {
      "epoch": 0.509404309252218,
      "grad_norm": 0.30152785778045654,
      "learning_rate": 1.6281765302366176e-05,
      "loss": 0.08346986770629883,
      "step": 3768
    },
    {
      "epoch": 0.509539501478665,
      "grad_norm": 0.348947674036026,
      "learning_rate": 1.6274945775098412e-05,
      "loss": 0.07816386222839355,
      "step": 3769
    },
    {
      "epoch": 0.509674693705112,
      "grad_norm": 0.5325326919555664,
      "learning_rate": 1.6268125982380135e-05,
      "loss": 0.06723415851593018,
      "step": 3770
    },
    {
      "epoch": 0.509809885931559,
      "grad_norm": 0.2995305061340332,
      "learning_rate": 1.626130592563127e-05,
      "loss": 0.07673883438110352,
      "step": 3771
    },
    {
      "epoch": 0.5099450781580059,
      "grad_norm": 0.2491988092660904,
      "learning_rate": 1.6254485606271778e-05,
      "loss": 0.04816484451293945,
      "step": 3772
    },
    {
      "epoch": 0.5100802703844529,
      "grad_norm": 0.27081945538520813,
      "learning_rate": 1.6247665025721698e-05,
      "loss": 0.08287811279296875,
      "step": 3773
    },
    {
      "epoch": 0.5102154626108999,
      "grad_norm": 0.3316383361816406,
      "learning_rate": 1.62408441854011e-05,
      "loss": 0.0939030647277832,
      "step": 3774
    },
    {
      "epoch": 0.5103506548373469,
      "grad_norm": 0.2686176598072052,
      "learning_rate": 1.6234023086730136e-05,
      "loss": 0.05086064338684082,
      "step": 3775
    },
    {
      "epoch": 0.5104858470637939,
      "grad_norm": 0.1850663274526596,
      "learning_rate": 1.622720173112898e-05,
      "loss": 0.04773569107055664,
      "step": 3776
    },
    {
      "epoch": 0.5106210392902408,
      "grad_norm": 0.46319863200187683,
      "learning_rate": 1.6220380120017874e-05,
      "loss": 0.06171417236328125,
      "step": 3777
    },
    {
      "epoch": 0.5107562315166878,
      "grad_norm": 0.2455659657716751,
      "learning_rate": 1.6213558254817128e-05,
      "loss": 0.06778955459594727,
      "step": 3778
    },
    {
      "epoch": 0.5108914237431348,
      "grad_norm": 0.19397512078285217,
      "learning_rate": 1.6206736136947074e-05,
      "loss": 0.04960966110229492,
      "step": 3779
    },
    {
      "epoch": 0.5110266159695818,
      "grad_norm": 0.2613491714000702,
      "learning_rate": 1.6199913767828126e-05,
      "loss": 0.07055473327636719,
      "step": 3780
    },
    {
      "epoch": 0.5111618081960287,
      "grad_norm": 0.2777951955795288,
      "learning_rate": 1.6193091148880733e-05,
      "loss": 0.08042335510253906,
      "step": 3781
    },
    {
      "epoch": 0.5112970004224757,
      "grad_norm": 0.3295799791812897,
      "learning_rate": 1.61862682815254e-05,
      "loss": 0.07029294967651367,
      "step": 3782
    },
    {
      "epoch": 0.5114321926489227,
      "grad_norm": 0.2006811946630478,
      "learning_rate": 1.617944516718268e-05,
      "loss": 0.055863380432128906,
      "step": 3783
    },
    {
      "epoch": 0.5115673848753697,
      "grad_norm": 0.2747129499912262,
      "learning_rate": 1.617262180727319e-05,
      "loss": 0.07995939254760742,
      "step": 3784
    },
    {
      "epoch": 0.5117025771018167,
      "grad_norm": 0.3211976885795593,
      "learning_rate": 1.6165798203217588e-05,
      "loss": 0.10124373435974121,
      "step": 3785
    },
    {
      "epoch": 0.5118377693282636,
      "grad_norm": 0.501947283744812,
      "learning_rate": 1.6158974356436585e-05,
      "loss": 0.08261764049530029,
      "step": 3786
    },
    {
      "epoch": 0.5119729615547106,
      "grad_norm": 0.1718253791332245,
      "learning_rate": 1.6152150268350938e-05,
      "loss": 0.04965782165527344,
      "step": 3787
    },
    {
      "epoch": 0.5121081537811576,
      "grad_norm": 0.2696118652820587,
      "learning_rate": 1.6145325940381458e-05,
      "loss": 0.04480338096618652,
      "step": 3788
    },
    {
      "epoch": 0.5122433460076046,
      "grad_norm": 0.23200277984142303,
      "learning_rate": 1.6138501373949018e-05,
      "loss": 0.06024932861328125,
      "step": 3789
    },
    {
      "epoch": 0.5123785382340516,
      "grad_norm": 0.28233274817466736,
      "learning_rate": 1.613167657047451e-05,
      "loss": 0.05148577690124512,
      "step": 3790
    },
    {
      "epoch": 0.5125137304604985,
      "grad_norm": 0.16529662907123566,
      "learning_rate": 1.612485153137891e-05,
      "loss": 0.04063296318054199,
      "step": 3791
    },
    {
      "epoch": 0.5126489226869455,
      "grad_norm": 0.3903208076953888,
      "learning_rate": 1.611802625808323e-05,
      "loss": 0.09176206588745117,
      "step": 3792
    },
    {
      "epoch": 0.5127841149133925,
      "grad_norm": 0.36968398094177246,
      "learning_rate": 1.611120075200851e-05,
      "loss": 0.08179950714111328,
      "step": 3793
    },
    {
      "epoch": 0.5129193071398395,
      "grad_norm": 0.3314633369445801,
      "learning_rate": 1.610437501457587e-05,
      "loss": 0.07626128196716309,
      "step": 3794
    },
    {
      "epoch": 0.5130544993662864,
      "grad_norm": 0.2416071742773056,
      "learning_rate": 1.6097549047206464e-05,
      "loss": 0.057384490966796875,
      "step": 3795
    },
    {
      "epoch": 0.5131896915927334,
      "grad_norm": 0.3267790973186493,
      "learning_rate": 1.6090722851321497e-05,
      "loss": 0.05952334403991699,
      "step": 3796
    },
    {
      "epoch": 0.5133248838191804,
      "grad_norm": 0.3248332142829895,
      "learning_rate": 1.6083896428342213e-05,
      "loss": 0.07203388214111328,
      "step": 3797
    },
    {
      "epoch": 0.5134600760456274,
      "grad_norm": 0.7720645666122437,
      "learning_rate": 1.6077069779689915e-05,
      "loss": 0.0859231948852539,
      "step": 3798
    },
    {
      "epoch": 0.5135952682720744,
      "grad_norm": 0.21609126031398773,
      "learning_rate": 1.607024290678594e-05,
      "loss": 0.04471468925476074,
      "step": 3799
    },
    {
      "epoch": 0.5137304604985213,
      "grad_norm": 0.20270587503910065,
      "learning_rate": 1.6063415811051686e-05,
      "loss": 0.06509971618652344,
      "step": 3800
    },
    {
      "epoch": 0.5138656527249683,
      "grad_norm": 0.28665196895599365,
      "learning_rate": 1.6056588493908596e-05,
      "loss": 0.0832204818725586,
      "step": 3801
    },
    {
      "epoch": 0.5140008449514153,
      "grad_norm": 0.3154524266719818,
      "learning_rate": 1.604976095677814e-05,
      "loss": 0.07967185974121094,
      "step": 3802
    },
    {
      "epoch": 0.5141360371778623,
      "grad_norm": 0.21033187210559845,
      "learning_rate": 1.604293320108186e-05,
      "loss": 0.044440269470214844,
      "step": 3803
    },
    {
      "epoch": 0.5142712294043092,
      "grad_norm": 0.4083731174468994,
      "learning_rate": 1.603610522824132e-05,
      "loss": 0.06813383102416992,
      "step": 3804
    },
    {
      "epoch": 0.5144064216307562,
      "grad_norm": 0.3776145577430725,
      "learning_rate": 1.6029277039678153e-05,
      "loss": 0.08827733993530273,
      "step": 3805
    },
    {
      "epoch": 0.5145416138572032,
      "grad_norm": 0.2450273483991623,
      "learning_rate": 1.602244863681401e-05,
      "loss": 0.05036449432373047,
      "step": 3806
    },
    {
      "epoch": 0.5146768060836502,
      "grad_norm": 0.2197941392660141,
      "learning_rate": 1.6015620021070613e-05,
      "loss": 0.054108619689941406,
      "step": 3807
    },
    {
      "epoch": 0.5148119983100972,
      "grad_norm": 0.19557338953018188,
      "learning_rate": 1.6008791193869714e-05,
      "loss": 0.04666924476623535,
      "step": 3808
    },
    {
      "epoch": 0.5149471905365441,
      "grad_norm": 0.31723499298095703,
      "learning_rate": 1.6001962156633102e-05,
      "loss": 0.06638240814208984,
      "step": 3809
    },
    {
      "epoch": 0.5150823827629911,
      "grad_norm": 0.33990269899368286,
      "learning_rate": 1.5995132910782632e-05,
      "loss": 0.08015680313110352,
      "step": 3810
    },
    {
      "epoch": 0.5152175749894381,
      "grad_norm": 0.2806377410888672,
      "learning_rate": 1.5988303457740178e-05,
      "loss": 0.06101036071777344,
      "step": 3811
    },
    {
      "epoch": 0.5153527672158851,
      "grad_norm": 0.2810553014278412,
      "learning_rate": 1.598147379892768e-05,
      "loss": 0.05346107482910156,
      "step": 3812
    },
    {
      "epoch": 0.515487959442332,
      "grad_norm": 0.2763311266899109,
      "learning_rate": 1.5974643935767098e-05,
      "loss": 0.05693769454956055,
      "step": 3813
    },
    {
      "epoch": 0.515623151668779,
      "grad_norm": 0.31364405155181885,
      "learning_rate": 1.5967813869680452e-05,
      "loss": 0.061385154724121094,
      "step": 3814
    },
    {
      "epoch": 0.515758343895226,
      "grad_norm": 0.3968309760093689,
      "learning_rate": 1.59609836020898e-05,
      "loss": 0.06057929992675781,
      "step": 3815
    },
    {
      "epoch": 0.515893536121673,
      "grad_norm": 0.2200632244348526,
      "learning_rate": 1.5954153134417236e-05,
      "loss": 0.060900211334228516,
      "step": 3816
    },
    {
      "epoch": 0.51602872834812,
      "grad_norm": 0.33449482917785645,
      "learning_rate": 1.59473224680849e-05,
      "loss": 0.05581164360046387,
      "step": 3817
    },
    {
      "epoch": 0.516163920574567,
      "grad_norm": 0.2559302747249603,
      "learning_rate": 1.5940491604514976e-05,
      "loss": 0.06303215026855469,
      "step": 3818
    },
    {
      "epoch": 0.5162991128010139,
      "grad_norm": 0.2785590887069702,
      "learning_rate": 1.5933660545129683e-05,
      "loss": 0.0716848373413086,
      "step": 3819
    },
    {
      "epoch": 0.5164343050274609,
      "grad_norm": 0.6427760124206543,
      "learning_rate": 1.5926829291351288e-05,
      "loss": 0.0704050064086914,
      "step": 3820
    },
    {
      "epoch": 0.5165694972539079,
      "grad_norm": 0.3818990886211395,
      "learning_rate": 1.591999784460209e-05,
      "loss": 0.05056571960449219,
      "step": 3821
    },
    {
      "epoch": 0.5167046894803549,
      "grad_norm": 0.3075590431690216,
      "learning_rate": 1.5913166206304435e-05,
      "loss": 0.06289792060852051,
      "step": 3822
    },
    {
      "epoch": 0.5168398817068018,
      "grad_norm": 0.24053780734539032,
      "learning_rate": 1.5906334377880707e-05,
      "loss": 0.05351996421813965,
      "step": 3823
    },
    {
      "epoch": 0.5169750739332488,
      "grad_norm": 0.39546748995780945,
      "learning_rate": 1.589950236075333e-05,
      "loss": 0.07835102081298828,
      "step": 3824
    },
    {
      "epoch": 0.5171102661596958,
      "grad_norm": 0.2950100898742676,
      "learning_rate": 1.5892670156344764e-05,
      "loss": 0.06669855117797852,
      "step": 3825
    },
    {
      "epoch": 0.5172454583861428,
      "grad_norm": 0.30787771940231323,
      "learning_rate": 1.588583776607751e-05,
      "loss": 0.051708221435546875,
      "step": 3826
    },
    {
      "epoch": 0.5173806506125898,
      "grad_norm": 0.2573969066143036,
      "learning_rate": 1.5879005191374106e-05,
      "loss": 0.054888248443603516,
      "step": 3827
    },
    {
      "epoch": 0.5175158428390367,
      "grad_norm": 0.31818974018096924,
      "learning_rate": 1.587217243365714e-05,
      "loss": 0.08167243003845215,
      "step": 3828
    },
    {
      "epoch": 0.5176510350654837,
      "grad_norm": 0.20823635160923004,
      "learning_rate": 1.586533949434922e-05,
      "loss": 0.04093742370605469,
      "step": 3829
    },
    {
      "epoch": 0.5177862272919307,
      "grad_norm": 0.29374828934669495,
      "learning_rate": 1.5858506374872998e-05,
      "loss": 0.06679534912109375,
      "step": 3830
    },
    {
      "epoch": 0.5179214195183777,
      "grad_norm": 0.3115184009075165,
      "learning_rate": 1.5851673076651178e-05,
      "loss": 0.04941272735595703,
      "step": 3831
    },
    {
      "epoch": 0.5180566117448246,
      "grad_norm": 0.15480205416679382,
      "learning_rate": 1.5844839601106477e-05,
      "loss": 0.041136980056762695,
      "step": 3832
    },
    {
      "epoch": 0.5181918039712716,
      "grad_norm": 0.3272755742073059,
      "learning_rate": 1.583800594966167e-05,
      "loss": 0.06614065170288086,
      "step": 3833
    },
    {
      "epoch": 0.5183269961977186,
      "grad_norm": 0.2249755561351776,
      "learning_rate": 1.583117212373955e-05,
      "loss": 0.05387401580810547,
      "step": 3834
    },
    {
      "epoch": 0.5184621884241656,
      "grad_norm": 0.2950299084186554,
      "learning_rate": 1.5824338124762967e-05,
      "loss": 0.05718231201171875,
      "step": 3835
    },
    {
      "epoch": 0.5185973806506126,
      "grad_norm": 0.28259211778640747,
      "learning_rate": 1.581750395415479e-05,
      "loss": 0.04875826835632324,
      "step": 3836
    },
    {
      "epoch": 0.5187325728770595,
      "grad_norm": 0.6511593461036682,
      "learning_rate": 1.5810669613337922e-05,
      "loss": 0.10650253295898438,
      "step": 3837
    },
    {
      "epoch": 0.5188677651035065,
      "grad_norm": 0.18748058378696442,
      "learning_rate": 1.5803835103735327e-05,
      "loss": 0.05097150802612305,
      "step": 3838
    },
    {
      "epoch": 0.5190029573299535,
      "grad_norm": 0.2657163441181183,
      "learning_rate": 1.5797000426769973e-05,
      "loss": 0.06188774108886719,
      "step": 3839
    },
    {
      "epoch": 0.5191381495564005,
      "grad_norm": 0.3725571632385254,
      "learning_rate": 1.579016558386488e-05,
      "loss": 0.08385348320007324,
      "step": 3840
    },
    {
      "epoch": 0.5192733417828475,
      "grad_norm": 0.5250045657157898,
      "learning_rate": 1.5783330576443096e-05,
      "loss": 0.10803556442260742,
      "step": 3841
    },
    {
      "epoch": 0.5194085340092944,
      "grad_norm": 0.4862484335899353,
      "learning_rate": 1.5776495405927716e-05,
      "loss": 0.06979751586914062,
      "step": 3842
    },
    {
      "epoch": 0.5195437262357414,
      "grad_norm": 0.52033931016922,
      "learning_rate": 1.5769660073741844e-05,
      "loss": 0.10051965713500977,
      "step": 3843
    },
    {
      "epoch": 0.5196789184621884,
      "grad_norm": 0.265226274728775,
      "learning_rate": 1.5762824581308645e-05,
      "loss": 0.0498957633972168,
      "step": 3844
    },
    {
      "epoch": 0.5198141106886354,
      "grad_norm": 0.21396085619926453,
      "learning_rate": 1.5755988930051304e-05,
      "loss": 0.05698704719543457,
      "step": 3845
    },
    {
      "epoch": 0.5199493029150823,
      "grad_norm": 0.10826153308153152,
      "learning_rate": 1.5749153121393025e-05,
      "loss": 0.03523445129394531,
      "step": 3846
    },
    {
      "epoch": 0.5200844951415293,
      "grad_norm": 0.3097763955593109,
      "learning_rate": 1.574231715675708e-05,
      "loss": 0.06415605545043945,
      "step": 3847
    },
    {
      "epoch": 0.5202196873679763,
      "grad_norm": 0.23440732061862946,
      "learning_rate": 1.573548103756674e-05,
      "loss": 0.05231189727783203,
      "step": 3848
    },
    {
      "epoch": 0.5203548795944233,
      "grad_norm": 0.35284194350242615,
      "learning_rate": 1.572864476524533e-05,
      "loss": 0.0575556755065918,
      "step": 3849
    },
    {
      "epoch": 0.5204900718208703,
      "grad_norm": 0.5734379291534424,
      "learning_rate": 1.5721808341216195e-05,
      "loss": 0.043912649154663086,
      "step": 3850
    },
    {
      "epoch": 0.5206252640473172,
      "grad_norm": 0.2837241590023041,
      "learning_rate": 1.571497176690271e-05,
      "loss": 0.060446739196777344,
      "step": 3851
    },
    {
      "epoch": 0.5207604562737642,
      "grad_norm": 0.2905561029911041,
      "learning_rate": 1.570813504372829e-05,
      "loss": 0.06624746322631836,
      "step": 3852
    },
    {
      "epoch": 0.5208956485002112,
      "grad_norm": 0.2592106759548187,
      "learning_rate": 1.570129817311638e-05,
      "loss": 0.05748152732849121,
      "step": 3853
    },
    {
      "epoch": 0.5210308407266582,
      "grad_norm": 0.4550934135913849,
      "learning_rate": 1.5694461156490452e-05,
      "loss": 0.07835555076599121,
      "step": 3854
    },
    {
      "epoch": 0.5211660329531052,
      "grad_norm": 0.28268054127693176,
      "learning_rate": 1.5687623995274008e-05,
      "loss": 0.04772043228149414,
      "step": 3855
    },
    {
      "epoch": 0.5213012251795521,
      "grad_norm": 0.3123287558555603,
      "learning_rate": 1.568078669089058e-05,
      "loss": 0.07772183418273926,
      "step": 3856
    },
    {
      "epoch": 0.5214364174059991,
      "grad_norm": 0.32369691133499146,
      "learning_rate": 1.567394924476373e-05,
      "loss": 0.08242559432983398,
      "step": 3857
    },
    {
      "epoch": 0.5215716096324461,
      "grad_norm": 0.5650047659873962,
      "learning_rate": 1.5667111658317057e-05,
      "loss": 0.08599567413330078,
      "step": 3858
    },
    {
      "epoch": 0.5217068018588931,
      "grad_norm": 0.232393279671669,
      "learning_rate": 1.5660273932974177e-05,
      "loss": 0.04138898849487305,
      "step": 3859
    },
    {
      "epoch": 0.52184199408534,
      "grad_norm": 0.5171011090278625,
      "learning_rate": 1.5653436070158743e-05,
      "loss": 0.07064485549926758,
      "step": 3860
    },
    {
      "epoch": 0.521977186311787,
      "grad_norm": 0.2689336836338043,
      "learning_rate": 1.564659807129444e-05,
      "loss": 0.08013391494750977,
      "step": 3861
    },
    {
      "epoch": 0.522112378538234,
      "grad_norm": 0.4905410706996918,
      "learning_rate": 1.5639759937804962e-05,
      "loss": 0.07244324684143066,
      "step": 3862
    },
    {
      "epoch": 0.522247570764681,
      "grad_norm": 0.40969574451446533,
      "learning_rate": 1.5632921671114055e-05,
      "loss": 0.09016799926757812,
      "step": 3863
    },
    {
      "epoch": 0.522382762991128,
      "grad_norm": 0.40808987617492676,
      "learning_rate": 1.5626083272645485e-05,
      "loss": 0.06834697723388672,
      "step": 3864
    },
    {
      "epoch": 0.5225179552175749,
      "grad_norm": 0.55250483751297,
      "learning_rate": 1.5619244743823038e-05,
      "loss": 0.06690692901611328,
      "step": 3865
    },
    {
      "epoch": 0.5226531474440219,
      "grad_norm": 0.3186188042163849,
      "learning_rate": 1.5612406086070534e-05,
      "loss": 0.08028411865234375,
      "step": 3866
    },
    {
      "epoch": 0.5227883396704689,
      "grad_norm": 0.21495214104652405,
      "learning_rate": 1.560556730081181e-05,
      "loss": 0.05619668960571289,
      "step": 3867
    },
    {
      "epoch": 0.5229235318969159,
      "grad_norm": 0.4362373650074005,
      "learning_rate": 1.5598728389470754e-05,
      "loss": 0.10913562774658203,
      "step": 3868
    },
    {
      "epoch": 0.523058724123363,
      "grad_norm": 0.2426283210515976,
      "learning_rate": 1.5591889353471245e-05,
      "loss": 0.06658697128295898,
      "step": 3869
    },
    {
      "epoch": 0.5231939163498099,
      "grad_norm": 0.23084065318107605,
      "learning_rate": 1.5585050194237226e-05,
      "loss": 0.054060935974121094,
      "step": 3870
    },
    {
      "epoch": 0.5233291085762569,
      "grad_norm": 0.4734134376049042,
      "learning_rate": 1.557821091319263e-05,
      "loss": 0.10500526428222656,
      "step": 3871
    },
    {
      "epoch": 0.5234643008027039,
      "grad_norm": 0.33723270893096924,
      "learning_rate": 1.5571371511761446e-05,
      "loss": 0.0969386100769043,
      "step": 3872
    },
    {
      "epoch": 0.5235994930291509,
      "grad_norm": 0.24272193014621735,
      "learning_rate": 1.5564531991367658e-05,
      "loss": 0.055225372314453125,
      "step": 3873
    },
    {
      "epoch": 0.5237346852555979,
      "grad_norm": 0.27921679615974426,
      "learning_rate": 1.5557692353435302e-05,
      "loss": 0.060835838317871094,
      "step": 3874
    },
    {
      "epoch": 0.5238698774820448,
      "grad_norm": 0.2498018741607666,
      "learning_rate": 1.5550852599388424e-05,
      "loss": 0.04842883348464966,
      "step": 3875
    },
    {
      "epoch": 0.5240050697084918,
      "grad_norm": 0.36984783411026,
      "learning_rate": 1.5544012730651096e-05,
      "loss": 0.08437395095825195,
      "step": 3876
    },
    {
      "epoch": 0.5241402619349388,
      "grad_norm": 0.23008708655834198,
      "learning_rate": 1.5537172748647422e-05,
      "loss": 0.050740718841552734,
      "step": 3877
    },
    {
      "epoch": 0.5242754541613858,
      "grad_norm": 0.3785080909729004,
      "learning_rate": 1.553033265480151e-05,
      "loss": 0.0830850601196289,
      "step": 3878
    },
    {
      "epoch": 0.5244106463878327,
      "grad_norm": 0.2892637550830841,
      "learning_rate": 1.552349245053752e-05,
      "loss": 0.0842580795288086,
      "step": 3879
    },
    {
      "epoch": 0.5245458386142797,
      "grad_norm": 0.21594905853271484,
      "learning_rate": 1.5516652137279597e-05,
      "loss": 0.05982255935668945,
      "step": 3880
    },
    {
      "epoch": 0.5246810308407267,
      "grad_norm": 0.31781211495399475,
      "learning_rate": 1.5509811716451955e-05,
      "loss": 0.053493499755859375,
      "step": 3881
    },
    {
      "epoch": 0.5248162230671737,
      "grad_norm": 0.23139196634292603,
      "learning_rate": 1.550297118947879e-05,
      "loss": 0.059580087661743164,
      "step": 3882
    },
    {
      "epoch": 0.5249514152936207,
      "grad_norm": 0.2394331991672516,
      "learning_rate": 1.5496130557784343e-05,
      "loss": 0.04545712471008301,
      "step": 3883
    },
    {
      "epoch": 0.5250866075200676,
      "grad_norm": 0.2936137616634369,
      "learning_rate": 1.5489289822792868e-05,
      "loss": 0.06028890609741211,
      "step": 3884
    },
    {
      "epoch": 0.5252217997465146,
      "grad_norm": 0.6374660134315491,
      "learning_rate": 1.5482448985928645e-05,
      "loss": 0.09602594375610352,
      "step": 3885
    },
    {
      "epoch": 0.5253569919729616,
      "grad_norm": 0.20281396806240082,
      "learning_rate": 1.5475608048615964e-05,
      "loss": 0.05556201934814453,
      "step": 3886
    },
    {
      "epoch": 0.5254921841994086,
      "grad_norm": 0.16149945557117462,
      "learning_rate": 1.546876701227916e-05,
      "loss": 0.04952716827392578,
      "step": 3887
    },
    {
      "epoch": 0.5256273764258556,
      "grad_norm": 0.3811986446380615,
      "learning_rate": 1.5461925878342558e-05,
      "loss": 0.07009601593017578,
      "step": 3888
    },
    {
      "epoch": 0.5257625686523025,
      "grad_norm": 0.26125630736351013,
      "learning_rate": 1.5455084648230527e-05,
      "loss": 0.06839418411254883,
      "step": 3889
    },
    {
      "epoch": 0.5258977608787495,
      "grad_norm": 0.24328996241092682,
      "learning_rate": 1.5448243323367438e-05,
      "loss": 0.06770992279052734,
      "step": 3890
    },
    {
      "epoch": 0.5260329531051965,
      "grad_norm": 0.37833523750305176,
      "learning_rate": 1.544140190517771e-05,
      "loss": 0.07322216033935547,
      "step": 3891
    },
    {
      "epoch": 0.5261681453316435,
      "grad_norm": 0.2731761634349823,
      "learning_rate": 1.5434560395085745e-05,
      "loss": 0.06831550598144531,
      "step": 3892
    },
    {
      "epoch": 0.5263033375580904,
      "grad_norm": 0.29786181449890137,
      "learning_rate": 1.542771879451599e-05,
      "loss": 0.05793619155883789,
      "step": 3893
    },
    {
      "epoch": 0.5264385297845374,
      "grad_norm": 0.3301241993904114,
      "learning_rate": 1.54208771048929e-05,
      "loss": 0.055074214935302734,
      "step": 3894
    },
    {
      "epoch": 0.5265737220109844,
      "grad_norm": 0.31042030453681946,
      "learning_rate": 1.5414035327640958e-05,
      "loss": 0.07587242126464844,
      "step": 3895
    },
    {
      "epoch": 0.5267089142374314,
      "grad_norm": 0.24964457750320435,
      "learning_rate": 1.5407193464184644e-05,
      "loss": 0.047933101654052734,
      "step": 3896
    },
    {
      "epoch": 0.5268441064638784,
      "grad_norm": 0.5645866990089417,
      "learning_rate": 1.5400351515948485e-05,
      "loss": 0.07502710819244385,
      "step": 3897
    },
    {
      "epoch": 0.5269792986903253,
      "grad_norm": 0.5430082082748413,
      "learning_rate": 1.5393509484357006e-05,
      "loss": 0.09406805038452148,
      "step": 3898
    },
    {
      "epoch": 0.5271144909167723,
      "grad_norm": 0.3937346935272217,
      "learning_rate": 1.538666737083475e-05,
      "loss": 0.07242584228515625,
      "step": 3899
    },
    {
      "epoch": 0.5272496831432193,
      "grad_norm": 0.1799725741147995,
      "learning_rate": 1.537982517680629e-05,
      "loss": 0.048305511474609375,
      "step": 3900
    },
    {
      "epoch": 0.5273848753696663,
      "grad_norm": 0.45096421241760254,
      "learning_rate": 1.5372982903696196e-05,
      "loss": 0.08168745040893555,
      "step": 3901
    },
    {
      "epoch": 0.5275200675961133,
      "grad_norm": 0.2164062261581421,
      "learning_rate": 1.536614055292908e-05,
      "loss": 0.05948066711425781,
      "step": 3902
    },
    {
      "epoch": 0.5276552598225602,
      "grad_norm": 0.3785545229911804,
      "learning_rate": 1.535929812592955e-05,
      "loss": 0.0536121129989624,
      "step": 3903
    },
    {
      "epoch": 0.5277904520490072,
      "grad_norm": 0.45809656381607056,
      "learning_rate": 1.5352455624122227e-05,
      "loss": 0.06551361083984375,
      "step": 3904
    },
    {
      "epoch": 0.5279256442754542,
      "grad_norm": 0.35153132677078247,
      "learning_rate": 1.5345613048931765e-05,
      "loss": 0.09002065658569336,
      "step": 3905
    },
    {
      "epoch": 0.5280608365019012,
      "grad_norm": 0.6052442193031311,
      "learning_rate": 1.5338770401782822e-05,
      "loss": 0.09721851348876953,
      "step": 3906
    },
    {
      "epoch": 0.5281960287283481,
      "grad_norm": 0.24167391657829285,
      "learning_rate": 1.5331927684100077e-05,
      "loss": 0.05780363082885742,
      "step": 3907
    },
    {
      "epoch": 0.5283312209547951,
      "grad_norm": 0.15925996005535126,
      "learning_rate": 1.5325084897308218e-05,
      "loss": 0.04418158531188965,
      "step": 3908
    },
    {
      "epoch": 0.5284664131812421,
      "grad_norm": 0.32850953936576843,
      "learning_rate": 1.5318242042831952e-05,
      "loss": 0.06540441513061523,
      "step": 3909
    },
    {
      "epoch": 0.5286016054076891,
      "grad_norm": 0.28783658146858215,
      "learning_rate": 1.5311399122095992e-05,
      "loss": 0.060230255126953125,
      "step": 3910
    },
    {
      "epoch": 0.5287367976341361,
      "grad_norm": 0.3889015018939972,
      "learning_rate": 1.5304556136525074e-05,
      "loss": 0.10230636596679688,
      "step": 3911
    },
    {
      "epoch": 0.528871989860583,
      "grad_norm": 0.3231453597545624,
      "learning_rate": 1.5297713087543948e-05,
      "loss": 0.06882619857788086,
      "step": 3912
    },
    {
      "epoch": 0.52900718208703,
      "grad_norm": 0.2304292768239975,
      "learning_rate": 1.5290869976577365e-05,
      "loss": 0.052089691162109375,
      "step": 3913
    },
    {
      "epoch": 0.529142374313477,
      "grad_norm": 0.24516160786151886,
      "learning_rate": 1.5284026805050107e-05,
      "loss": 0.06309914588928223,
      "step": 3914
    },
    {
      "epoch": 0.529277566539924,
      "grad_norm": 0.37919315695762634,
      "learning_rate": 1.5277183574386947e-05,
      "loss": 0.07090353965759277,
      "step": 3915
    },
    {
      "epoch": 0.529412758766371,
      "grad_norm": 0.5468193292617798,
      "learning_rate": 1.5270340286012694e-05,
      "loss": 0.0863804817199707,
      "step": 3916
    },
    {
      "epoch": 0.5295479509928179,
      "grad_norm": 0.3777635097503662,
      "learning_rate": 1.526349694135215e-05,
      "loss": 0.09398460388183594,
      "step": 3917
    },
    {
      "epoch": 0.5296831432192649,
      "grad_norm": 0.29108086228370667,
      "learning_rate": 1.525665354183014e-05,
      "loss": 0.06123161315917969,
      "step": 3918
    },
    {
      "epoch": 0.5298183354457119,
      "grad_norm": 0.3264821469783783,
      "learning_rate": 1.5249810088871493e-05,
      "loss": 0.0648965835571289,
      "step": 3919
    },
    {
      "epoch": 0.5299535276721589,
      "grad_norm": 0.2696027159690857,
      "learning_rate": 1.5242966583901052e-05,
      "loss": 0.07930850982666016,
      "step": 3920
    },
    {
      "epoch": 0.5300887198986058,
      "grad_norm": 0.27279606461524963,
      "learning_rate": 1.523612302834367e-05,
      "loss": 0.07864618301391602,
      "step": 3921
    },
    {
      "epoch": 0.5302239121250528,
      "grad_norm": 0.2681496739387512,
      "learning_rate": 1.5229279423624217e-05,
      "loss": 0.04422426223754883,
      "step": 3922
    },
    {
      "epoch": 0.5303591043514998,
      "grad_norm": 0.19210343062877655,
      "learning_rate": 1.5222435771167566e-05,
      "loss": 0.0578460693359375,
      "step": 3923
    },
    {
      "epoch": 0.5304942965779468,
      "grad_norm": 0.23293140530586243,
      "learning_rate": 1.5215592072398602e-05,
      "loss": 0.06531012058258057,
      "step": 3924
    },
    {
      "epoch": 0.5306294888043938,
      "grad_norm": 0.3658544719219208,
      "learning_rate": 1.520874832874222e-05,
      "loss": 0.07075285911560059,
      "step": 3925
    },
    {
      "epoch": 0.5307646810308407,
      "grad_norm": 0.48965251445770264,
      "learning_rate": 1.5201904541623318e-05,
      "loss": 0.06623554229736328,
      "step": 3926
    },
    {
      "epoch": 0.5308998732572877,
      "grad_norm": 0.19527631998062134,
      "learning_rate": 1.5195060712466817e-05,
      "loss": 0.041087985038757324,
      "step": 3927
    },
    {
      "epoch": 0.5310350654837347,
      "grad_norm": 0.36395376920700073,
      "learning_rate": 1.5188216842697635e-05,
      "loss": 0.07523655891418457,
      "step": 3928
    },
    {
      "epoch": 0.5311702577101817,
      "grad_norm": 0.372333824634552,
      "learning_rate": 1.5181372933740703e-05,
      "loss": 0.06962180137634277,
      "step": 3929
    },
    {
      "epoch": 0.5313054499366286,
      "grad_norm": 0.17896510660648346,
      "learning_rate": 1.5174528987020958e-05,
      "loss": 0.04320979118347168,
      "step": 3930
    },
    {
      "epoch": 0.5314406421630756,
      "grad_norm": 0.2079179286956787,
      "learning_rate": 1.5167685003963345e-05,
      "loss": 0.05081343650817871,
      "step": 3931
    },
    {
      "epoch": 0.5315758343895226,
      "grad_norm": 0.2520752251148224,
      "learning_rate": 1.5160840985992824e-05,
      "loss": 0.0585322380065918,
      "step": 3932
    },
    {
      "epoch": 0.5317110266159696,
      "grad_norm": 0.2267618179321289,
      "learning_rate": 1.515399693453435e-05,
      "loss": 0.055820465087890625,
      "step": 3933
    },
    {
      "epoch": 0.5318462188424166,
      "grad_norm": 0.44096824526786804,
      "learning_rate": 1.5147152851012894e-05,
      "loss": 0.07697486877441406,
      "step": 3934
    },
    {
      "epoch": 0.5319814110688635,
      "grad_norm": 0.3692184388637543,
      "learning_rate": 1.514030873685343e-05,
      "loss": 0.08869409561157227,
      "step": 3935
    },
    {
      "epoch": 0.5321166032953105,
      "grad_norm": 0.37288644909858704,
      "learning_rate": 1.513346459348094e-05,
      "loss": 0.06119251251220703,
      "step": 3936
    },
    {
      "epoch": 0.5322517955217575,
      "grad_norm": 0.2925654351711273,
      "learning_rate": 1.5126620422320405e-05,
      "loss": 0.0448918342590332,
      "step": 3937
    },
    {
      "epoch": 0.5323869877482045,
      "grad_norm": 0.22279894351959229,
      "learning_rate": 1.5119776224796823e-05,
      "loss": 0.04670119285583496,
      "step": 3938
    },
    {
      "epoch": 0.5325221799746515,
      "grad_norm": 0.49064162373542786,
      "learning_rate": 1.5112932002335195e-05,
      "loss": 0.09360694885253906,
      "step": 3939
    },
    {
      "epoch": 0.5326573722010984,
      "grad_norm": 0.4237077832221985,
      "learning_rate": 1.5106087756360524e-05,
      "loss": 0.07799720764160156,
      "step": 3940
    },
    {
      "epoch": 0.5327925644275454,
      "grad_norm": 0.2859136760234833,
      "learning_rate": 1.5099243488297816e-05,
      "loss": 0.0719451904296875,
      "step": 3941
    },
    {
      "epoch": 0.5329277566539924,
      "grad_norm": 0.3216422200202942,
      "learning_rate": 1.5092399199572083e-05,
      "loss": 0.0841364860534668,
      "step": 3942
    },
    {
      "epoch": 0.5330629488804394,
      "grad_norm": 0.19512711465358734,
      "learning_rate": 1.5085554891608343e-05,
      "loss": 0.04526257514953613,
      "step": 3943
    },
    {
      "epoch": 0.5331981411068863,
      "grad_norm": 0.25813478231430054,
      "learning_rate": 1.5078710565831616e-05,
      "loss": 0.05426192283630371,
      "step": 3944
    },
    {
      "epoch": 0.5333333333333333,
      "grad_norm": 0.3508923351764679,
      "learning_rate": 1.5071866223666935e-05,
      "loss": 0.058115243911743164,
      "step": 3945
    },
    {
      "epoch": 0.5334685255597803,
      "grad_norm": 0.37193968892097473,
      "learning_rate": 1.5065021866539323e-05,
      "loss": 0.07733488082885742,
      "step": 3946
    },
    {
      "epoch": 0.5336037177862273,
      "grad_norm": 0.3799670934677124,
      "learning_rate": 1.5058177495873805e-05,
      "loss": 0.0815572738647461,
      "step": 3947
    },
    {
      "epoch": 0.5337389100126743,
      "grad_norm": 0.22421643137931824,
      "learning_rate": 1.5051333113095429e-05,
      "loss": 0.06677722930908203,
      "step": 3948
    },
    {
      "epoch": 0.5338741022391212,
      "grad_norm": 0.18863984942436218,
      "learning_rate": 1.5044488719629218e-05,
      "loss": 0.04372763633728027,
      "step": 3949
    },
    {
      "epoch": 0.5340092944655682,
      "grad_norm": 0.1745264083147049,
      "learning_rate": 1.5037644316900227e-05,
      "loss": 0.04700195789337158,
      "step": 3950
    },
    {
      "epoch": 0.5341444866920152,
      "grad_norm": 0.42912304401397705,
      "learning_rate": 1.5030799906333484e-05,
      "loss": 0.09153890609741211,
      "step": 3951
    },
    {
      "epoch": 0.5342796789184622,
      "grad_norm": 0.2364966869354248,
      "learning_rate": 1.5023955489354031e-05,
      "loss": 0.0482325553894043,
      "step": 3952
    },
    {
      "epoch": 0.5344148711449092,
      "grad_norm": 0.29457613825798035,
      "learning_rate": 1.5017111067386927e-05,
      "loss": 0.08064079284667969,
      "step": 3953
    },
    {
      "epoch": 0.5345500633713561,
      "grad_norm": 0.2295532077550888,
      "learning_rate": 1.50102666418572e-05,
      "loss": 0.0716547966003418,
      "step": 3954
    },
    {
      "epoch": 0.5346852555978031,
      "grad_norm": 0.2165478616952896,
      "learning_rate": 1.500342221418991e-05,
      "loss": 0.06681489944458008,
      "step": 3955
    },
    {
      "epoch": 0.5348204478242501,
      "grad_norm": 0.307179719209671,
      "learning_rate": 1.4996577785810094e-05,
      "loss": 0.07864665985107422,
      "step": 3956
    },
    {
      "epoch": 0.5349556400506971,
      "grad_norm": 0.25438976287841797,
      "learning_rate": 1.4989733358142798e-05,
      "loss": 0.06771159172058105,
      "step": 3957
    },
    {
      "epoch": 0.535090832277144,
      "grad_norm": 0.18071763217449188,
      "learning_rate": 1.498288893261308e-05,
      "loss": 0.05147981643676758,
      "step": 3958
    },
    {
      "epoch": 0.535226024503591,
      "grad_norm": 0.21865326166152954,
      "learning_rate": 1.497604451064597e-05,
      "loss": 0.06647729873657227,
      "step": 3959
    },
    {
      "epoch": 0.535361216730038,
      "grad_norm": 0.2598375976085663,
      "learning_rate": 1.496920009366652e-05,
      "loss": 0.05479311943054199,
      "step": 3960
    },
    {
      "epoch": 0.535496408956485,
      "grad_norm": 0.31111493706703186,
      "learning_rate": 1.4962355683099777e-05,
      "loss": 0.08464837074279785,
      "step": 3961
    },
    {
      "epoch": 0.535631601182932,
      "grad_norm": 0.19168411195278168,
      "learning_rate": 1.4955511280370782e-05,
      "loss": 0.050550222396850586,
      "step": 3962
    },
    {
      "epoch": 0.5357667934093789,
      "grad_norm": 0.3476887345314026,
      "learning_rate": 1.4948666886904579e-05,
      "loss": 0.07009601593017578,
      "step": 3963
    },
    {
      "epoch": 0.5359019856358259,
      "grad_norm": 0.32077741622924805,
      "learning_rate": 1.4941822504126199e-05,
      "loss": 0.08756351470947266,
      "step": 3964
    },
    {
      "epoch": 0.5360371778622729,
      "grad_norm": 0.22124578058719635,
      "learning_rate": 1.4934978133460681e-05,
      "loss": 0.05996525287628174,
      "step": 3965
    },
    {
      "epoch": 0.5361723700887199,
      "grad_norm": 0.2110230028629303,
      "learning_rate": 1.4928133776333068e-05,
      "loss": 0.06747722625732422,
      "step": 3966
    },
    {
      "epoch": 0.5363075623151669,
      "grad_norm": 0.29011982679367065,
      "learning_rate": 1.4921289434168386e-05,
      "loss": 0.05568695068359375,
      "step": 3967
    },
    {
      "epoch": 0.5364427545416138,
      "grad_norm": 0.36574438214302063,
      "learning_rate": 1.4914445108391663e-05,
      "loss": 0.06701183319091797,
      "step": 3968
    },
    {
      "epoch": 0.5365779467680608,
      "grad_norm": 0.2853558659553528,
      "learning_rate": 1.4907600800427922e-05,
      "loss": 0.06931304931640625,
      "step": 3969
    },
    {
      "epoch": 0.5367131389945078,
      "grad_norm": 0.746479332447052,
      "learning_rate": 1.4900756511702188e-05,
      "loss": 0.12049102783203125,
      "step": 3970
    },
    {
      "epoch": 0.5368483312209548,
      "grad_norm": 0.20287103950977325,
      "learning_rate": 1.4893912243639479e-05,
      "loss": 0.06114983558654785,
      "step": 3971
    },
    {
      "epoch": 0.5369835234474017,
      "grad_norm": 0.38798338174819946,
      "learning_rate": 1.4887067997664807e-05,
      "loss": 0.0707712173461914,
      "step": 3972
    },
    {
      "epoch": 0.5371187156738487,
      "grad_norm": 0.340192973613739,
      "learning_rate": 1.488022377520318e-05,
      "loss": 0.09295082092285156,
      "step": 3973
    },
    {
      "epoch": 0.5372539079002957,
      "grad_norm": 0.3362419903278351,
      "learning_rate": 1.4873379577679599e-05,
      "loss": 0.06299209594726562,
      "step": 3974
    },
    {
      "epoch": 0.5373891001267427,
      "grad_norm": 0.44043293595314026,
      "learning_rate": 1.4866535406519063e-05,
      "loss": 0.06498146057128906,
      "step": 3975
    },
    {
      "epoch": 0.5375242923531897,
      "grad_norm": 0.12804383039474487,
      "learning_rate": 1.4859691263146574e-05,
      "loss": 0.03209352493286133,
      "step": 3976
    },
    {
      "epoch": 0.5376594845796366,
      "grad_norm": 0.3819577395915985,
      "learning_rate": 1.485284714898711e-05,
      "loss": 0.06276702880859375,
      "step": 3977
    },
    {
      "epoch": 0.5377946768060836,
      "grad_norm": 0.24549607932567596,
      "learning_rate": 1.4846003065465653e-05,
      "loss": 0.055055320262908936,
      "step": 3978
    },
    {
      "epoch": 0.5379298690325306,
      "grad_norm": 0.6252531409263611,
      "learning_rate": 1.4839159014007179e-05,
      "loss": 0.07657742500305176,
      "step": 3979
    },
    {
      "epoch": 0.5380650612589776,
      "grad_norm": 0.2426566630601883,
      "learning_rate": 1.4832314996036653e-05,
      "loss": 0.05623602867126465,
      "step": 3980
    },
    {
      "epoch": 0.5382002534854246,
      "grad_norm": 0.3851320743560791,
      "learning_rate": 1.4825471012979048e-05,
      "loss": 0.07387828826904297,
      "step": 3981
    },
    {
      "epoch": 0.5383354457118715,
      "grad_norm": 0.34674301743507385,
      "learning_rate": 1.4818627066259301e-05,
      "loss": 0.07103347778320312,
      "step": 3982
    },
    {
      "epoch": 0.5384706379383185,
      "grad_norm": 0.22041839361190796,
      "learning_rate": 1.481178315730237e-05,
      "loss": 0.06679677963256836,
      "step": 3983
    },
    {
      "epoch": 0.5386058301647655,
      "grad_norm": 0.19270789623260498,
      "learning_rate": 1.4804939287533184e-05,
      "loss": 0.07610785961151123,
      "step": 3984
    },
    {
      "epoch": 0.5387410223912125,
      "grad_norm": 0.3698720932006836,
      "learning_rate": 1.4798095458376682e-05,
      "loss": 0.08079910278320312,
      "step": 3985
    },
    {
      "epoch": 0.5388762146176594,
      "grad_norm": 0.40295422077178955,
      "learning_rate": 1.4791251671257788e-05,
      "loss": 0.07804924249649048,
      "step": 3986
    },
    {
      "epoch": 0.5390114068441064,
      "grad_norm": 0.273698091506958,
      "learning_rate": 1.4784407927601401e-05,
      "loss": 0.07843509316444397,
      "step": 3987
    },
    {
      "epoch": 0.5391465990705534,
      "grad_norm": 0.5676789879798889,
      "learning_rate": 1.4777564228832436e-05,
      "loss": 0.0808095932006836,
      "step": 3988
    },
    {
      "epoch": 0.5392817912970004,
      "grad_norm": 0.2204323559999466,
      "learning_rate": 1.4770720576375782e-05,
      "loss": 0.045033931732177734,
      "step": 3989
    },
    {
      "epoch": 0.5394169835234474,
      "grad_norm": 0.2715097665786743,
      "learning_rate": 1.4763876971656334e-05,
      "loss": 0.06560802459716797,
      "step": 3990
    },
    {
      "epoch": 0.5395521757498943,
      "grad_norm": 0.6461181640625,
      "learning_rate": 1.4757033416098953e-05,
      "loss": 0.09101676940917969,
      "step": 3991
    },
    {
      "epoch": 0.5396873679763413,
      "grad_norm": 0.48632514476776123,
      "learning_rate": 1.4750189911128511e-05,
      "loss": 0.08342933654785156,
      "step": 3992
    },
    {
      "epoch": 0.5398225602027883,
      "grad_norm": 0.19051408767700195,
      "learning_rate": 1.4743346458169863e-05,
      "loss": 0.033000946044921875,
      "step": 3993
    },
    {
      "epoch": 0.5399577524292353,
      "grad_norm": 0.5683019757270813,
      "learning_rate": 1.473650305864785e-05,
      "loss": 0.09323787689208984,
      "step": 3994
    },
    {
      "epoch": 0.5400929446556822,
      "grad_norm": 0.4190486967563629,
      "learning_rate": 1.472965971398731e-05,
      "loss": 0.06127762794494629,
      "step": 3995
    },
    {
      "epoch": 0.5402281368821292,
      "grad_norm": 0.555079996585846,
      "learning_rate": 1.4722816425613054e-05,
      "loss": 0.07727217674255371,
      "step": 3996
    },
    {
      "epoch": 0.5403633291085762,
      "grad_norm": 0.9168782234191895,
      "learning_rate": 1.4715973194949895e-05,
      "loss": 0.07502365112304688,
      "step": 3997
    },
    {
      "epoch": 0.5404985213350232,
      "grad_norm": 0.27501973509788513,
      "learning_rate": 1.4709130023422636e-05,
      "loss": 0.07636690139770508,
      "step": 3998
    },
    {
      "epoch": 0.5406337135614702,
      "grad_norm": 0.3995406925678253,
      "learning_rate": 1.4702286912456052e-05,
      "loss": 0.07165002822875977,
      "step": 3999
    },
    {
      "epoch": 0.5407689057879171,
      "grad_norm": 1.017317771911621,
      "learning_rate": 1.4695443863474928e-05,
      "loss": 0.12094640731811523,
      "step": 4000
    },
    {
      "epoch": 0.5409040980143641,
      "grad_norm": 0.23867133259773254,
      "learning_rate": 1.4688600877904012e-05,
      "loss": 0.057695865631103516,
      "step": 4001
    },
    {
      "epoch": 0.5410392902408111,
      "grad_norm": 1.027727723121643,
      "learning_rate": 1.468175795716805e-05,
      "loss": 0.07674694061279297,
      "step": 4002
    },
    {
      "epoch": 0.5411744824672581,
      "grad_norm": 0.28664782643318176,
      "learning_rate": 1.4674915102691783e-05,
      "loss": 0.06580352783203125,
      "step": 4003
    },
    {
      "epoch": 0.5413096746937052,
      "grad_norm": 0.20941662788391113,
      "learning_rate": 1.4668072315899926e-05,
      "loss": 0.04004549980163574,
      "step": 4004
    },
    {
      "epoch": 0.5414448669201521,
      "grad_norm": 0.266179621219635,
      "learning_rate": 1.466122959821718e-05,
      "loss": 0.05678606033325195,
      "step": 4005
    },
    {
      "epoch": 0.5415800591465991,
      "grad_norm": 0.3486866056919098,
      "learning_rate": 1.4654386951068239e-05,
      "loss": 0.08021783828735352,
      "step": 4006
    },
    {
      "epoch": 0.5417152513730461,
      "grad_norm": 0.18772508203983307,
      "learning_rate": 1.4647544375877776e-05,
      "loss": 0.04320788383483887,
      "step": 4007
    },
    {
      "epoch": 0.5418504435994931,
      "grad_norm": 0.2924988865852356,
      "learning_rate": 1.4640701874070457e-05,
      "loss": 0.07803106307983398,
      "step": 4008
    },
    {
      "epoch": 0.5419856358259401,
      "grad_norm": 0.2681550979614258,
      "learning_rate": 1.4633859447070922e-05,
      "loss": 0.08400964736938477,
      "step": 4009
    },
    {
      "epoch": 0.542120828052387,
      "grad_norm": 0.4184032678604126,
      "learning_rate": 1.4627017096303805e-05,
      "loss": 0.07605552673339844,
      "step": 4010
    },
    {
      "epoch": 0.542256020278834,
      "grad_norm": 0.6821161508560181,
      "learning_rate": 1.4620174823193711e-05,
      "loss": 0.0885004997253418,
      "step": 4011
    },
    {
      "epoch": 0.542391212505281,
      "grad_norm": 0.8827657699584961,
      "learning_rate": 1.4613332629165249e-05,
      "loss": 0.06777095794677734,
      "step": 4012
    },
    {
      "epoch": 0.542526404731728,
      "grad_norm": 0.32371920347213745,
      "learning_rate": 1.4606490515642998e-05,
      "loss": 0.05782461166381836,
      "step": 4013
    },
    {
      "epoch": 0.542661596958175,
      "grad_norm": 0.5198997855186462,
      "learning_rate": 1.4599648484051516e-05,
      "loss": 0.10658121109008789,
      "step": 4014
    },
    {
      "epoch": 0.5427967891846219,
      "grad_norm": 0.4386010468006134,
      "learning_rate": 1.4592806535815358e-05,
      "loss": 0.07213687896728516,
      "step": 4015
    },
    {
      "epoch": 0.5429319814110689,
      "grad_norm": 0.3345072865486145,
      "learning_rate": 1.4585964672359045e-05,
      "loss": 0.0564727783203125,
      "step": 4016
    },
    {
      "epoch": 0.5430671736375159,
      "grad_norm": 0.36038216948509216,
      "learning_rate": 1.4579122895107098e-05,
      "loss": 0.07569026947021484,
      "step": 4017
    },
    {
      "epoch": 0.5432023658639629,
      "grad_norm": 0.3713933527469635,
      "learning_rate": 1.4572281205484012e-05,
      "loss": 0.07886934280395508,
      "step": 4018
    },
    {
      "epoch": 0.5433375580904098,
      "grad_norm": 0.7414966225624084,
      "learning_rate": 1.4565439604914256e-05,
      "loss": 0.08749055862426758,
      "step": 4019
    },
    {
      "epoch": 0.5434727503168568,
      "grad_norm": 0.6894304752349854,
      "learning_rate": 1.4558598094822294e-05,
      "loss": 0.07340073585510254,
      "step": 4020
    },
    {
      "epoch": 0.5436079425433038,
      "grad_norm": 0.3799217939376831,
      "learning_rate": 1.455175667663256e-05,
      "loss": 0.04095315933227539,
      "step": 4021
    },
    {
      "epoch": 0.5437431347697508,
      "grad_norm": 0.2783501446247101,
      "learning_rate": 1.4544915351769476e-05,
      "loss": 0.04219430685043335,
      "step": 4022
    },
    {
      "epoch": 0.5438783269961978,
      "grad_norm": 0.3897891342639923,
      "learning_rate": 1.4538074121657448e-05,
      "loss": 0.04985332489013672,
      "step": 4023
    },
    {
      "epoch": 0.5440135192226447,
      "grad_norm": 0.3952426612377167,
      "learning_rate": 1.4531232987720846e-05,
      "loss": 0.0752558708190918,
      "step": 4024
    },
    {
      "epoch": 0.5441487114490917,
      "grad_norm": 0.4929008185863495,
      "learning_rate": 1.4524391951384037e-05,
      "loss": 0.07437896728515625,
      "step": 4025
    },
    {
      "epoch": 0.5442839036755387,
      "grad_norm": 2.2381513118743896,
      "learning_rate": 1.4517551014071358e-05,
      "loss": 0.10858821868896484,
      "step": 4026
    },
    {
      "epoch": 0.5444190959019857,
      "grad_norm": 0.3457643985748291,
      "learning_rate": 1.4510710177207137e-05,
      "loss": 0.06877613067626953,
      "step": 4027
    },
    {
      "epoch": 0.5445542881284327,
      "grad_norm": 0.3232475817203522,
      "learning_rate": 1.450386944221566e-05,
      "loss": 0.05397367477416992,
      "step": 4028
    },
    {
      "epoch": 0.5446894803548796,
      "grad_norm": 0.4774819612503052,
      "learning_rate": 1.449702881052121e-05,
      "loss": 0.07984447479248047,
      "step": 4029
    },
    {
      "epoch": 0.5448246725813266,
      "grad_norm": 0.31248894333839417,
      "learning_rate": 1.4490188283548048e-05,
      "loss": 0.06699800491333008,
      "step": 4030
    },
    {
      "epoch": 0.5449598648077736,
      "grad_norm": 0.6607853770256042,
      "learning_rate": 1.44833478627204e-05,
      "loss": 0.07941150665283203,
      "step": 4031
    },
    {
      "epoch": 0.5450950570342206,
      "grad_norm": 0.5920162200927734,
      "learning_rate": 1.447650754946249e-05,
      "loss": 0.07666754722595215,
      "step": 4032
    },
    {
      "epoch": 0.5452302492606675,
      "grad_norm": 0.3527291417121887,
      "learning_rate": 1.4469667345198492e-05,
      "loss": 0.08711528778076172,
      "step": 4033
    },
    {
      "epoch": 0.5453654414871145,
      "grad_norm": 0.47891005873680115,
      "learning_rate": 1.446282725135258e-05,
      "loss": 0.08446550369262695,
      "step": 4034
    },
    {
      "epoch": 0.5455006337135615,
      "grad_norm": 0.6416643261909485,
      "learning_rate": 1.4455987269348904e-05,
      "loss": 0.06750345230102539,
      "step": 4035
    },
    {
      "epoch": 0.5456358259400085,
      "grad_norm": 0.37330344319343567,
      "learning_rate": 1.4449147400611578e-05,
      "loss": 0.07050609588623047,
      "step": 4036
    },
    {
      "epoch": 0.5457710181664555,
      "grad_norm": 0.4069422781467438,
      "learning_rate": 1.4442307646564702e-05,
      "loss": 0.07738596200942993,
      "step": 4037
    },
    {
      "epoch": 0.5459062103929024,
      "grad_norm": 0.29950210452079773,
      "learning_rate": 1.4435468008632345e-05,
      "loss": 0.06700122356414795,
      "step": 4038
    },
    {
      "epoch": 0.5460414026193494,
      "grad_norm": 0.35566675662994385,
      "learning_rate": 1.4428628488238557e-05,
      "loss": 0.08124065399169922,
      "step": 4039
    },
    {
      "epoch": 0.5461765948457964,
      "grad_norm": 0.3127792775630951,
      "learning_rate": 1.442178908680737e-05,
      "loss": 0.08421540260314941,
      "step": 4040
    },
    {
      "epoch": 0.5463117870722434,
      "grad_norm": 0.38258448243141174,
      "learning_rate": 1.4414949805762779e-05,
      "loss": 0.06313610076904297,
      "step": 4041
    },
    {
      "epoch": 0.5464469792986903,
      "grad_norm": 0.181279718875885,
      "learning_rate": 1.4408110646528757e-05,
      "loss": 0.04441499710083008,
      "step": 4042
    },
    {
      "epoch": 0.5465821715251373,
      "grad_norm": 0.3570832312107086,
      "learning_rate": 1.440127161052925e-05,
      "loss": 0.06917190551757812,
      "step": 4043
    },
    {
      "epoch": 0.5467173637515843,
      "grad_norm": 0.3481096625328064,
      "learning_rate": 1.4394432699188188e-05,
      "loss": 0.07164525985717773,
      "step": 4044
    },
    {
      "epoch": 0.5468525559780313,
      "grad_norm": 0.13638432323932648,
      "learning_rate": 1.4387593913929472e-05,
      "loss": 0.033329010009765625,
      "step": 4045
    },
    {
      "epoch": 0.5469877482044783,
      "grad_norm": 0.4036140441894531,
      "learning_rate": 1.4380755256176968e-05,
      "loss": 0.06957507133483887,
      "step": 4046
    },
    {
      "epoch": 0.5471229404309252,
      "grad_norm": 0.18465276062488556,
      "learning_rate": 1.437391672735452e-05,
      "loss": 0.04264044761657715,
      "step": 4047
    },
    {
      "epoch": 0.5472581326573722,
      "grad_norm": 0.32538607716560364,
      "learning_rate": 1.4367078328885946e-05,
      "loss": 0.06711673736572266,
      "step": 4048
    },
    {
      "epoch": 0.5473933248838192,
      "grad_norm": 0.4988720118999481,
      "learning_rate": 1.4360240062195039e-05,
      "loss": 0.05175590515136719,
      "step": 4049
    },
    {
      "epoch": 0.5475285171102662,
      "grad_norm": 0.1915602684020996,
      "learning_rate": 1.435340192870557e-05,
      "loss": 0.061510562896728516,
      "step": 4050
    },
    {
      "epoch": 0.5476637093367132,
      "grad_norm": 0.22108495235443115,
      "learning_rate": 1.434656392984126e-05,
      "loss": 0.053974151611328125,
      "step": 4051
    },
    {
      "epoch": 0.5477989015631601,
      "grad_norm": 0.4570392370223999,
      "learning_rate": 1.4339726067025828e-05,
      "loss": 0.0910940170288086,
      "step": 4052
    },
    {
      "epoch": 0.5479340937896071,
      "grad_norm": 0.43323811888694763,
      "learning_rate": 1.4332888341682947e-05,
      "loss": 0.06746768951416016,
      "step": 4053
    },
    {
      "epoch": 0.5480692860160541,
      "grad_norm": 0.5113841891288757,
      "learning_rate": 1.432605075523627e-05,
      "loss": 0.09701395034790039,
      "step": 4054
    },
    {
      "epoch": 0.5482044782425011,
      "grad_norm": 0.3196141719818115,
      "learning_rate": 1.4319213309109426e-05,
      "loss": 0.057730674743652344,
      "step": 4055
    },
    {
      "epoch": 0.548339670468948,
      "grad_norm": 0.27028515934944153,
      "learning_rate": 1.4312376004725996e-05,
      "loss": 0.07694220542907715,
      "step": 4056
    },
    {
      "epoch": 0.548474862695395,
      "grad_norm": 0.33444029092788696,
      "learning_rate": 1.430553884350955e-05,
      "loss": 0.058648109436035156,
      "step": 4057
    },
    {
      "epoch": 0.548610054921842,
      "grad_norm": 0.410898894071579,
      "learning_rate": 1.429870182688362e-05,
      "loss": 0.0593876838684082,
      "step": 4058
    },
    {
      "epoch": 0.548745247148289,
      "grad_norm": 0.19892758131027222,
      "learning_rate": 1.4291864956271713e-05,
      "loss": 0.04279899597167969,
      "step": 4059
    },
    {
      "epoch": 0.548880439374736,
      "grad_norm": 0.2936065196990967,
      "learning_rate": 1.4285028233097293e-05,
      "loss": 0.07022547721862793,
      "step": 4060
    },
    {
      "epoch": 0.5490156316011829,
      "grad_norm": 1.709394097328186,
      "learning_rate": 1.4278191658783809e-05,
      "loss": 0.07623481750488281,
      "step": 4061
    },
    {
      "epoch": 0.5491508238276299,
      "grad_norm": 0.38031327724456787,
      "learning_rate": 1.427135523475467e-05,
      "loss": 0.08426761627197266,
      "step": 4062
    },
    {
      "epoch": 0.5492860160540769,
      "grad_norm": 0.3300657272338867,
      "learning_rate": 1.4264518962433258e-05,
      "loss": 0.08411979675292969,
      "step": 4063
    },
    {
      "epoch": 0.5494212082805239,
      "grad_norm": 0.27288737893104553,
      "learning_rate": 1.4257682843242925e-05,
      "loss": 0.047863006591796875,
      "step": 4064
    },
    {
      "epoch": 0.5495564005069709,
      "grad_norm": 0.3293284773826599,
      "learning_rate": 1.4250846878606974e-05,
      "loss": 0.06148529052734375,
      "step": 4065
    },
    {
      "epoch": 0.5496915927334178,
      "grad_norm": 0.28542405366897583,
      "learning_rate": 1.4244011069948702e-05,
      "loss": 0.056427955627441406,
      "step": 4066
    },
    {
      "epoch": 0.5498267849598648,
      "grad_norm": 0.34654128551483154,
      "learning_rate": 1.4237175418691357e-05,
      "loss": 0.05337119102478027,
      "step": 4067
    },
    {
      "epoch": 0.5499619771863118,
      "grad_norm": 0.309227854013443,
      "learning_rate": 1.4230339926258153e-05,
      "loss": 0.05243372917175293,
      "step": 4068
    },
    {
      "epoch": 0.5500971694127588,
      "grad_norm": 0.36314624547958374,
      "learning_rate": 1.422350459407229e-05,
      "loss": 0.07169866561889648,
      "step": 4069
    },
    {
      "epoch": 0.5502323616392057,
      "grad_norm": 0.44304680824279785,
      "learning_rate": 1.4216669423556903e-05,
      "loss": 0.08015775680541992,
      "step": 4070
    },
    {
      "epoch": 0.5503675538656527,
      "grad_norm": 0.20376355946063995,
      "learning_rate": 1.420983441613512e-05,
      "loss": 0.033985912799835205,
      "step": 4071
    },
    {
      "epoch": 0.5505027460920997,
      "grad_norm": 0.2869951128959656,
      "learning_rate": 1.420299957323003e-05,
      "loss": 0.043447256088256836,
      "step": 4072
    },
    {
      "epoch": 0.5506379383185467,
      "grad_norm": 0.7137879729270935,
      "learning_rate": 1.4196164896264679e-05,
      "loss": 0.10390281677246094,
      "step": 4073
    },
    {
      "epoch": 0.5507731305449937,
      "grad_norm": 0.25919097661972046,
      "learning_rate": 1.418933038666208e-05,
      "loss": 0.057084083557128906,
      "step": 4074
    },
    {
      "epoch": 0.5509083227714406,
      "grad_norm": 0.24649231135845184,
      "learning_rate": 1.4182496045845217e-05,
      "loss": 0.06522965431213379,
      "step": 4075
    },
    {
      "epoch": 0.5510435149978876,
      "grad_norm": 0.28304094076156616,
      "learning_rate": 1.4175661875237036e-05,
      "loss": 0.05105435848236084,
      "step": 4076
    },
    {
      "epoch": 0.5511787072243346,
      "grad_norm": 0.18977323174476624,
      "learning_rate": 1.416882787626045e-05,
      "loss": 0.036368727684020996,
      "step": 4077
    },
    {
      "epoch": 0.5513138994507816,
      "grad_norm": 0.20677489042282104,
      "learning_rate": 1.4161994050338334e-05,
      "loss": 0.04225122928619385,
      "step": 4078
    },
    {
      "epoch": 0.5514490916772286,
      "grad_norm": 0.24968455731868744,
      "learning_rate": 1.4155160398893528e-05,
      "loss": 0.05938720703125,
      "step": 4079
    },
    {
      "epoch": 0.5515842839036755,
      "grad_norm": 0.620607852935791,
      "learning_rate": 1.4148326923348824e-05,
      "loss": 0.10203838348388672,
      "step": 4080
    },
    {
      "epoch": 0.5517194761301225,
      "grad_norm": 0.1507611721754074,
      "learning_rate": 1.4141493625127e-05,
      "loss": 0.03885054588317871,
      "step": 4081
    },
    {
      "epoch": 0.5518546683565695,
      "grad_norm": 0.46643996238708496,
      "learning_rate": 1.4134660505650786e-05,
      "loss": 0.05947399139404297,
      "step": 4082
    },
    {
      "epoch": 0.5519898605830165,
      "grad_norm": 0.3163384795188904,
      "learning_rate": 1.4127827566342864e-05,
      "loss": 0.04616665840148926,
      "step": 4083
    },
    {
      "epoch": 0.5521250528094634,
      "grad_norm": 0.2818586826324463,
      "learning_rate": 1.4120994808625896e-05,
      "loss": 0.05243968963623047,
      "step": 4084
    },
    {
      "epoch": 0.5522602450359104,
      "grad_norm": 0.3116128742694855,
      "learning_rate": 1.4114162233922494e-05,
      "loss": 0.062319278717041016,
      "step": 4085
    },
    {
      "epoch": 0.5523954372623574,
      "grad_norm": 0.2428111433982849,
      "learning_rate": 1.4107329843655238e-05,
      "loss": 0.054234862327575684,
      "step": 4086
    },
    {
      "epoch": 0.5525306294888044,
      "grad_norm": 0.48988455533981323,
      "learning_rate": 1.4100497639246675e-05,
      "loss": 0.08711624145507812,
      "step": 4087
    },
    {
      "epoch": 0.5526658217152514,
      "grad_norm": 0.30226606130599976,
      "learning_rate": 1.4093665622119294e-05,
      "loss": 0.07283449172973633,
      "step": 4088
    },
    {
      "epoch": 0.5528010139416983,
      "grad_norm": 0.5557790994644165,
      "learning_rate": 1.4086833793695566e-05,
      "loss": 0.06707191467285156,
      "step": 4089
    },
    {
      "epoch": 0.5529362061681453,
      "grad_norm": 0.2503361701965332,
      "learning_rate": 1.408000215539791e-05,
      "loss": 0.04856252670288086,
      "step": 4090
    },
    {
      "epoch": 0.5530713983945923,
      "grad_norm": 0.45031601190567017,
      "learning_rate": 1.4073170708648711e-05,
      "loss": 0.07099366188049316,
      "step": 4091
    },
    {
      "epoch": 0.5532065906210393,
      "grad_norm": 0.6956272721290588,
      "learning_rate": 1.406633945487032e-05,
      "loss": 0.0933542251586914,
      "step": 4092
    },
    {
      "epoch": 0.5533417828474863,
      "grad_norm": 0.2610994875431061,
      "learning_rate": 1.4059508395485026e-05,
      "loss": 0.04047584533691406,
      "step": 4093
    },
    {
      "epoch": 0.5534769750739332,
      "grad_norm": 0.19905126094818115,
      "learning_rate": 1.4052677531915102e-05,
      "loss": 0.04932081699371338,
      "step": 4094
    },
    {
      "epoch": 0.5536121673003802,
      "grad_norm": 0.4185316264629364,
      "learning_rate": 1.4045846865582765e-05,
      "loss": 0.09696292877197266,
      "step": 4095
    },
    {
      "epoch": 0.5537473595268272,
      "grad_norm": 0.2882910668849945,
      "learning_rate": 1.4039016397910206e-05,
      "loss": 0.06048727035522461,
      "step": 4096
    },
    {
      "epoch": 0.5538825517532742,
      "grad_norm": 0.28646066784858704,
      "learning_rate": 1.403218613031955e-05,
      "loss": 0.051027774810791016,
      "step": 4097
    },
    {
      "epoch": 0.5540177439797211,
      "grad_norm": 0.30673086643218994,
      "learning_rate": 1.4025356064232903e-05,
      "loss": 0.0428166389465332,
      "step": 4098
    },
    {
      "epoch": 0.5541529362061681,
      "grad_norm": 0.2924477159976959,
      "learning_rate": 1.4018526201072324e-05,
      "loss": 0.06857061386108398,
      "step": 4099
    },
    {
      "epoch": 0.5542881284326151,
      "grad_norm": 0.3255417048931122,
      "learning_rate": 1.4011696542259821e-05,
      "loss": 0.0710136890411377,
      "step": 4100
    },
    {
      "epoch": 0.5544233206590621,
      "grad_norm": 0.5660695433616638,
      "learning_rate": 1.4004867089217376e-05,
      "loss": 0.06834745407104492,
      "step": 4101
    },
    {
      "epoch": 0.5545585128855091,
      "grad_norm": 0.3874152600765228,
      "learning_rate": 1.39980378433669e-05,
      "loss": 0.06563949584960938,
      "step": 4102
    },
    {
      "epoch": 0.554693705111956,
      "grad_norm": 0.397693008184433,
      "learning_rate": 1.399120880613029e-05,
      "loss": 0.0656123161315918,
      "step": 4103
    },
    {
      "epoch": 0.554828897338403,
      "grad_norm": 0.2028006762266159,
      "learning_rate": 1.3984379978929388e-05,
      "loss": 0.036795616149902344,
      "step": 4104
    },
    {
      "epoch": 0.55496408956485,
      "grad_norm": 0.45990800857543945,
      "learning_rate": 1.3977551363185995e-05,
      "loss": 0.0576939582824707,
      "step": 4105
    },
    {
      "epoch": 0.555099281791297,
      "grad_norm": 0.38087111711502075,
      "learning_rate": 1.3970722960321854e-05,
      "loss": 0.07435894012451172,
      "step": 4106
    },
    {
      "epoch": 0.555234474017744,
      "grad_norm": 0.17435480654239655,
      "learning_rate": 1.3963894771758682e-05,
      "loss": 0.031667470932006836,
      "step": 4107
    },
    {
      "epoch": 0.5553696662441909,
      "grad_norm": 0.1557110846042633,
      "learning_rate": 1.3957066798918143e-05,
      "loss": 0.040529847145080566,
      "step": 4108
    },
    {
      "epoch": 0.5555048584706379,
      "grad_norm": 0.4714154005050659,
      "learning_rate": 1.3950239043221861e-05,
      "loss": 0.05920982360839844,
      "step": 4109
    },
    {
      "epoch": 0.5556400506970849,
      "grad_norm": 0.49100977182388306,
      "learning_rate": 1.3943411506091408e-05,
      "loss": 0.06539535522460938,
      "step": 4110
    },
    {
      "epoch": 0.5557752429235319,
      "grad_norm": 0.23461778461933136,
      "learning_rate": 1.3936584188948313e-05,
      "loss": 0.039727210998535156,
      "step": 4111
    },
    {
      "epoch": 0.5559104351499788,
      "grad_norm": 0.5778148174285889,
      "learning_rate": 1.3929757093214059e-05,
      "loss": 0.08493161201477051,
      "step": 4112
    },
    {
      "epoch": 0.5560456273764258,
      "grad_norm": 0.5185837745666504,
      "learning_rate": 1.3922930220310085e-05,
      "loss": 0.07986998558044434,
      "step": 4113
    },
    {
      "epoch": 0.5561808196028728,
      "grad_norm": 0.3703401982784271,
      "learning_rate": 1.3916103571657786e-05,
      "loss": 0.0818014144897461,
      "step": 4114
    },
    {
      "epoch": 0.5563160118293198,
      "grad_norm": 0.41620615124702454,
      "learning_rate": 1.3909277148678504e-05,
      "loss": 0.08162307739257812,
      "step": 4115
    },
    {
      "epoch": 0.5564512040557668,
      "grad_norm": 0.31471601128578186,
      "learning_rate": 1.3902450952793536e-05,
      "loss": 0.06625914573669434,
      "step": 4116
    },
    {
      "epoch": 0.5565863962822137,
      "grad_norm": 0.5340596437454224,
      "learning_rate": 1.389562498542413e-05,
      "loss": 0.045081496238708496,
      "step": 4117
    },
    {
      "epoch": 0.5567215885086607,
      "grad_norm": 0.45191457867622375,
      "learning_rate": 1.388879924799149e-05,
      "loss": 0.09464693069458008,
      "step": 4118
    },
    {
      "epoch": 0.5568567807351077,
      "grad_norm": 0.2838112711906433,
      "learning_rate": 1.388197374191678e-05,
      "loss": 0.061566829681396484,
      "step": 4119
    },
    {
      "epoch": 0.5569919729615547,
      "grad_norm": 0.4281176030635834,
      "learning_rate": 1.387514846862109e-05,
      "loss": 0.07387542724609375,
      "step": 4120
    },
    {
      "epoch": 0.5571271651880016,
      "grad_norm": 0.1971270740032196,
      "learning_rate": 1.3868323429525492e-05,
      "loss": 0.051799774169921875,
      "step": 4121
    },
    {
      "epoch": 0.5572623574144486,
      "grad_norm": 0.3478848934173584,
      "learning_rate": 1.3861498626050986e-05,
      "loss": 0.05951738357543945,
      "step": 4122
    },
    {
      "epoch": 0.5573975496408956,
      "grad_norm": 0.1675407886505127,
      "learning_rate": 1.385467405961854e-05,
      "loss": 0.042083740234375,
      "step": 4123
    },
    {
      "epoch": 0.5575327418673426,
      "grad_norm": 0.3506906032562256,
      "learning_rate": 1.3847849731649066e-05,
      "loss": 0.05516231060028076,
      "step": 4124
    },
    {
      "epoch": 0.5576679340937896,
      "grad_norm": 0.31182748079299927,
      "learning_rate": 1.3841025643563418e-05,
      "loss": 0.06922054290771484,
      "step": 4125
    },
    {
      "epoch": 0.5578031263202365,
      "grad_norm": 0.30659523606300354,
      "learning_rate": 1.3834201796782413e-05,
      "loss": 0.07000207901000977,
      "step": 4126
    },
    {
      "epoch": 0.5579383185466835,
      "grad_norm": 0.39148402214050293,
      "learning_rate": 1.3827378192726808e-05,
      "loss": 0.09355020523071289,
      "step": 4127
    },
    {
      "epoch": 0.5580735107731305,
      "grad_norm": 0.22750145196914673,
      "learning_rate": 1.3820554832817324e-05,
      "loss": 0.06097602844238281,
      "step": 4128
    },
    {
      "epoch": 0.5582087029995775,
      "grad_norm": 0.26270392537117004,
      "learning_rate": 1.3813731718474606e-05,
      "loss": 0.045249104499816895,
      "step": 4129
    },
    {
      "epoch": 0.5583438952260245,
      "grad_norm": 0.280582994222641,
      "learning_rate": 1.380690885111927e-05,
      "loss": 0.07164716720581055,
      "step": 4130
    },
    {
      "epoch": 0.5584790874524714,
      "grad_norm": 0.2737008035182953,
      "learning_rate": 1.3800086232171877e-05,
      "loss": 0.06156015396118164,
      "step": 4131
    },
    {
      "epoch": 0.5586142796789184,
      "grad_norm": 0.31067678332328796,
      "learning_rate": 1.3793263863052926e-05,
      "loss": 0.053952932357788086,
      "step": 4132
    },
    {
      "epoch": 0.5587494719053654,
      "grad_norm": 0.20502115786075592,
      "learning_rate": 1.3786441745182881e-05,
      "loss": 0.05380678176879883,
      "step": 4133
    },
    {
      "epoch": 0.5588846641318124,
      "grad_norm": 0.16852745413780212,
      "learning_rate": 1.3779619879982127e-05,
      "loss": 0.046740055084228516,
      "step": 4134
    },
    {
      "epoch": 0.5590198563582593,
      "grad_norm": 0.2714245021343231,
      "learning_rate": 1.3772798268871025e-05,
      "loss": 0.05841350555419922,
      "step": 4135
    },
    {
      "epoch": 0.5591550485847063,
      "grad_norm": 0.178030863404274,
      "learning_rate": 1.376597691326987e-05,
      "loss": 0.04270339012145996,
      "step": 4136
    },
    {
      "epoch": 0.5592902408111533,
      "grad_norm": 0.49629151821136475,
      "learning_rate": 1.3759155814598898e-05,
      "loss": 0.08703851699829102,
      "step": 4137
    },
    {
      "epoch": 0.5594254330376003,
      "grad_norm": 0.399566113948822,
      "learning_rate": 1.3752334974278308e-05,
      "loss": 0.08135366439819336,
      "step": 4138
    },
    {
      "epoch": 0.5595606252640473,
      "grad_norm": 0.22032016515731812,
      "learning_rate": 1.3745514393728225e-05,
      "loss": 0.04446768760681152,
      "step": 4139
    },
    {
      "epoch": 0.5596958174904944,
      "grad_norm": 0.1380263715982437,
      "learning_rate": 1.3738694074368735e-05,
      "loss": 0.025764435529708862,
      "step": 4140
    },
    {
      "epoch": 0.5598310097169413,
      "grad_norm": 1.0428732633590698,
      "learning_rate": 1.3731874017619868e-05,
      "loss": 0.08602285385131836,
      "step": 4141
    },
    {
      "epoch": 0.5599662019433883,
      "grad_norm": 0.23632653057575226,
      "learning_rate": 1.3725054224901597e-05,
      "loss": 0.050508737564086914,
      "step": 4142
    },
    {
      "epoch": 0.5601013941698353,
      "grad_norm": 0.3460511863231659,
      "learning_rate": 1.3718234697633826e-05,
      "loss": 0.0711984634399414,
      "step": 4143
    },
    {
      "epoch": 0.5602365863962823,
      "grad_norm": 0.3127269744873047,
      "learning_rate": 1.3711415437236427e-05,
      "loss": 0.06948280334472656,
      "step": 4144
    },
    {
      "epoch": 0.5603717786227292,
      "grad_norm": 0.3011138141155243,
      "learning_rate": 1.3704596445129207e-05,
      "loss": 0.06205272674560547,
      "step": 4145
    },
    {
      "epoch": 0.5605069708491762,
      "grad_norm": 0.25340166687965393,
      "learning_rate": 1.369777772273192e-05,
      "loss": 0.04357600212097168,
      "step": 4146
    },
    {
      "epoch": 0.5606421630756232,
      "grad_norm": 0.3297371566295624,
      "learning_rate": 1.369095927146425e-05,
      "loss": 0.054175376892089844,
      "step": 4147
    },
    {
      "epoch": 0.5607773553020702,
      "grad_norm": 0.21068519353866577,
      "learning_rate": 1.3684141092745846e-05,
      "loss": 0.051644086837768555,
      "step": 4148
    },
    {
      "epoch": 0.5609125475285172,
      "grad_norm": 0.38605958223342896,
      "learning_rate": 1.3677323187996276e-05,
      "loss": 0.05482792854309082,
      "step": 4149
    },
    {
      "epoch": 0.5610477397549641,
      "grad_norm": 0.5324007868766785,
      "learning_rate": 1.3670505558635074e-05,
      "loss": 0.07591867446899414,
      "step": 4150
    },
    {
      "epoch": 0.5611829319814111,
      "grad_norm": 0.39704835414886475,
      "learning_rate": 1.366368820608171e-05,
      "loss": 0.07014083862304688,
      "step": 4151
    },
    {
      "epoch": 0.5613181242078581,
      "grad_norm": 0.2284613996744156,
      "learning_rate": 1.365687113175558e-05,
      "loss": 0.04727625846862793,
      "step": 4152
    },
    {
      "epoch": 0.5614533164343051,
      "grad_norm": 0.17523261904716492,
      "learning_rate": 1.3650054337076049e-05,
      "loss": 0.049777984619140625,
      "step": 4153
    },
    {
      "epoch": 0.561588508660752,
      "grad_norm": 0.2719191610813141,
      "learning_rate": 1.3643237823462398e-05,
      "loss": 0.06025516986846924,
      "step": 4154
    },
    {
      "epoch": 0.561723700887199,
      "grad_norm": 0.21693992614746094,
      "learning_rate": 1.363642159233387e-05,
      "loss": 0.05410635471343994,
      "step": 4155
    },
    {
      "epoch": 0.561858893113646,
      "grad_norm": 0.2796480357646942,
      "learning_rate": 1.3629605645109642e-05,
      "loss": 0.045880913734436035,
      "step": 4156
    },
    {
      "epoch": 0.561994085340093,
      "grad_norm": 0.2789880931377411,
      "learning_rate": 1.362278998320882e-05,
      "loss": 0.08246898651123047,
      "step": 4157
    },
    {
      "epoch": 0.56212927756654,
      "grad_norm": 0.4399503767490387,
      "learning_rate": 1.3615974608050472e-05,
      "loss": 0.06561732292175293,
      "step": 4158
    },
    {
      "epoch": 0.5622644697929869,
      "grad_norm": 0.6177327036857605,
      "learning_rate": 1.3609159521053588e-05,
      "loss": 0.10786080360412598,
      "step": 4159
    },
    {
      "epoch": 0.5623996620194339,
      "grad_norm": 0.49539682269096375,
      "learning_rate": 1.3602344723637107e-05,
      "loss": 0.08597803115844727,
      "step": 4160
    },
    {
      "epoch": 0.5625348542458809,
      "grad_norm": 0.17610320448875427,
      "learning_rate": 1.3595530217219916e-05,
      "loss": 0.04601788520812988,
      "step": 4161
    },
    {
      "epoch": 0.5626700464723279,
      "grad_norm": 0.19023549556732178,
      "learning_rate": 1.3588716003220815e-05,
      "loss": 0.06010150909423828,
      "step": 4162
    },
    {
      "epoch": 0.5628052386987749,
      "grad_norm": 0.45934152603149414,
      "learning_rate": 1.3581902083058574e-05,
      "loss": 0.06469964981079102,
      "step": 4163
    },
    {
      "epoch": 0.5629404309252218,
      "grad_norm": 0.15902107954025269,
      "learning_rate": 1.3575088458151877e-05,
      "loss": 0.041898250579833984,
      "step": 4164
    },
    {
      "epoch": 0.5630756231516688,
      "grad_norm": 1.0006210803985596,
      "learning_rate": 1.3568275129919367e-05,
      "loss": 0.07063019275665283,
      "step": 4165
    },
    {
      "epoch": 0.5632108153781158,
      "grad_norm": 0.4762057363986969,
      "learning_rate": 1.3561462099779604e-05,
      "loss": 0.07624053955078125,
      "step": 4166
    },
    {
      "epoch": 0.5633460076045628,
      "grad_norm": 0.35259366035461426,
      "learning_rate": 1.3554649369151104e-05,
      "loss": 0.06088662147521973,
      "step": 4167
    },
    {
      "epoch": 0.5634811998310097,
      "grad_norm": 0.33427998423576355,
      "learning_rate": 1.3547836939452315e-05,
      "loss": 0.06496810913085938,
      "step": 4168
    },
    {
      "epoch": 0.5636163920574567,
      "grad_norm": 0.22689493000507355,
      "learning_rate": 1.3541024812101615e-05,
      "loss": 0.0627431869506836,
      "step": 4169
    },
    {
      "epoch": 0.5637515842839037,
      "grad_norm": 0.33148717880249023,
      "learning_rate": 1.3534212988517339e-05,
      "loss": 0.057376861572265625,
      "step": 4170
    },
    {
      "epoch": 0.5638867765103507,
      "grad_norm": 0.44841817021369934,
      "learning_rate": 1.3527401470117726e-05,
      "loss": 0.06867027282714844,
      "step": 4171
    },
    {
      "epoch": 0.5640219687367977,
      "grad_norm": 0.14598503708839417,
      "learning_rate": 1.3520590258320981e-05,
      "loss": 0.03270459175109863,
      "step": 4172
    },
    {
      "epoch": 0.5641571609632446,
      "grad_norm": 0.1817558854818344,
      "learning_rate": 1.3513779354545235e-05,
      "loss": 0.051550865173339844,
      "step": 4173
    },
    {
      "epoch": 0.5642923531896916,
      "grad_norm": 0.2841106951236725,
      "learning_rate": 1.3506968760208557e-05,
      "loss": 0.055995941162109375,
      "step": 4174
    },
    {
      "epoch": 0.5644275454161386,
      "grad_norm": 0.40993207693099976,
      "learning_rate": 1.3500158476728938e-05,
      "loss": 0.06054532527923584,
      "step": 4175
    },
    {
      "epoch": 0.5645627376425856,
      "grad_norm": 0.17506608366966248,
      "learning_rate": 1.3493348505524325e-05,
      "loss": 0.05012226104736328,
      "step": 4176
    },
    {
      "epoch": 0.5646979298690326,
      "grad_norm": 0.437804639339447,
      "learning_rate": 1.3486538848012586e-05,
      "loss": 0.06559109687805176,
      "step": 4177
    },
    {
      "epoch": 0.5648331220954795,
      "grad_norm": 0.3530019223690033,
      "learning_rate": 1.3479729505611532e-05,
      "loss": 0.08833551406860352,
      "step": 4178
    },
    {
      "epoch": 0.5649683143219265,
      "grad_norm": 0.4161841571331024,
      "learning_rate": 1.3472920479738906e-05,
      "loss": 0.06860017776489258,
      "step": 4179
    },
    {
      "epoch": 0.5651035065483735,
      "grad_norm": 0.2763068377971649,
      "learning_rate": 1.346611177181237e-05,
      "loss": 0.056932687759399414,
      "step": 4180
    },
    {
      "epoch": 0.5652386987748205,
      "grad_norm": 0.307731956243515,
      "learning_rate": 1.3459303383249547e-05,
      "loss": 0.053565025329589844,
      "step": 4181
    },
    {
      "epoch": 0.5653738910012674,
      "grad_norm": 0.5274808406829834,
      "learning_rate": 1.3452495315467975e-05,
      "loss": 0.07222676277160645,
      "step": 4182
    },
    {
      "epoch": 0.5655090832277144,
      "grad_norm": 0.25629836320877075,
      "learning_rate": 1.3445687569885132e-05,
      "loss": 0.056728363037109375,
      "step": 4183
    },
    {
      "epoch": 0.5656442754541614,
      "grad_norm": 0.19897255301475525,
      "learning_rate": 1.3438880147918429e-05,
      "loss": 0.03602886199951172,
      "step": 4184
    },
    {
      "epoch": 0.5657794676806084,
      "grad_norm": 0.2085980474948883,
      "learning_rate": 1.3432073050985201e-05,
      "loss": 0.03757810592651367,
      "step": 4185
    },
    {
      "epoch": 0.5659146599070554,
      "grad_norm": 0.6800008416175842,
      "learning_rate": 1.3425266280502721e-05,
      "loss": 0.058939218521118164,
      "step": 4186
    },
    {
      "epoch": 0.5660498521335023,
      "grad_norm": 0.3324357867240906,
      "learning_rate": 1.3418459837888202e-05,
      "loss": 0.07649612426757812,
      "step": 4187
    },
    {
      "epoch": 0.5661850443599493,
      "grad_norm": 0.4090869724750519,
      "learning_rate": 1.3411653724558784e-05,
      "loss": 0.07547235488891602,
      "step": 4188
    },
    {
      "epoch": 0.5663202365863963,
      "grad_norm": 0.18743127584457397,
      "learning_rate": 1.3404847941931523e-05,
      "loss": 0.05271005630493164,
      "step": 4189
    },
    {
      "epoch": 0.5664554288128433,
      "grad_norm": 0.29850631952285767,
      "learning_rate": 1.339804249142343e-05,
      "loss": 0.08056211471557617,
      "step": 4190
    },
    {
      "epoch": 0.5665906210392903,
      "grad_norm": 0.348247230052948,
      "learning_rate": 1.3391237374451429e-05,
      "loss": 0.042586326599121094,
      "step": 4191
    },
    {
      "epoch": 0.5667258132657372,
      "grad_norm": 0.4494287073612213,
      "learning_rate": 1.3384432592432388e-05,
      "loss": 0.07525444030761719,
      "step": 4192
    },
    {
      "epoch": 0.5668610054921842,
      "grad_norm": 0.7001757621765137,
      "learning_rate": 1.3377628146783102e-05,
      "loss": 0.11281776428222656,
      "step": 4193
    },
    {
      "epoch": 0.5669961977186312,
      "grad_norm": 0.3711039423942566,
      "learning_rate": 1.3370824038920281e-05,
      "loss": 0.04892301559448242,
      "step": 4194
    },
    {
      "epoch": 0.5671313899450782,
      "grad_norm": 0.25460025668144226,
      "learning_rate": 1.3364020270260586e-05,
      "loss": 0.07265377044677734,
      "step": 4195
    },
    {
      "epoch": 0.5672665821715251,
      "grad_norm": 0.3742760121822357,
      "learning_rate": 1.335721684222059e-05,
      "loss": 0.0734403133392334,
      "step": 4196
    },
    {
      "epoch": 0.5674017743979721,
      "grad_norm": 0.26575955748558044,
      "learning_rate": 1.3350413756216816e-05,
      "loss": 0.049715638160705566,
      "step": 4197
    },
    {
      "epoch": 0.5675369666244191,
      "grad_norm": 0.2327270656824112,
      "learning_rate": 1.334361101366569e-05,
      "loss": 0.06168937683105469,
      "step": 4198
    },
    {
      "epoch": 0.5676721588508661,
      "grad_norm": 0.341900110244751,
      "learning_rate": 1.3336808615983582e-05,
      "loss": 0.04771614074707031,
      "step": 4199
    },
    {
      "epoch": 0.5678073510773131,
      "grad_norm": 0.31476742029190063,
      "learning_rate": 1.3330006564586791e-05,
      "loss": 0.06220579147338867,
      "step": 4200
    },
    {
      "epoch": 0.56794254330376,
      "grad_norm": 0.48586228489875793,
      "learning_rate": 1.3323204860891539e-05,
      "loss": 0.09546470642089844,
      "step": 4201
    },
    {
      "epoch": 0.568077735530207,
      "grad_norm": 0.24473726749420166,
      "learning_rate": 1.3316403506313981e-05,
      "loss": 0.04458355903625488,
      "step": 4202
    },
    {
      "epoch": 0.568212927756654,
      "grad_norm": 0.39096829295158386,
      "learning_rate": 1.3309602502270184e-05,
      "loss": 0.04271745681762695,
      "step": 4203
    },
    {
      "epoch": 0.568348119983101,
      "grad_norm": 0.1461983174085617,
      "learning_rate": 1.3302801850176161e-05,
      "loss": 0.03778576850891113,
      "step": 4204
    },
    {
      "epoch": 0.568483312209548,
      "grad_norm": 0.3260999321937561,
      "learning_rate": 1.3296001551447848e-05,
      "loss": 0.050777435302734375,
      "step": 4205
    },
    {
      "epoch": 0.5686185044359949,
      "grad_norm": 0.24449004232883453,
      "learning_rate": 1.32892016075011e-05,
      "loss": 0.07197999954223633,
      "step": 4206
    },
    {
      "epoch": 0.5687536966624419,
      "grad_norm": 0.21209828555583954,
      "learning_rate": 1.3282402019751694e-05,
      "loss": 0.05071306228637695,
      "step": 4207
    },
    {
      "epoch": 0.5688888888888889,
      "grad_norm": 0.30274760723114014,
      "learning_rate": 1.327560278961535e-05,
      "loss": 0.04763948917388916,
      "step": 4208
    },
    {
      "epoch": 0.5690240811153359,
      "grad_norm": 0.36460569500923157,
      "learning_rate": 1.3268803918507699e-05,
      "loss": 0.07638049125671387,
      "step": 4209
    },
    {
      "epoch": 0.5691592733417828,
      "grad_norm": 0.40261271595954895,
      "learning_rate": 1.3262005407844306e-05,
      "loss": 0.06149876117706299,
      "step": 4210
    },
    {
      "epoch": 0.5692944655682298,
      "grad_norm": 0.24275512993335724,
      "learning_rate": 1.325520725904066e-05,
      "loss": 0.05030655860900879,
      "step": 4211
    },
    {
      "epoch": 0.5694296577946768,
      "grad_norm": 0.2505589425563812,
      "learning_rate": 1.3248409473512158e-05,
      "loss": 0.05789375305175781,
      "step": 4212
    },
    {
      "epoch": 0.5695648500211238,
      "grad_norm": 0.49377721548080444,
      "learning_rate": 1.3241612052674146e-05,
      "loss": 0.10123944282531738,
      "step": 4213
    },
    {
      "epoch": 0.5697000422475708,
      "grad_norm": 0.34104618430137634,
      "learning_rate": 1.3234814997941883e-05,
      "loss": 0.08298385143280029,
      "step": 4214
    },
    {
      "epoch": 0.5698352344740177,
      "grad_norm": 0.3067338764667511,
      "learning_rate": 1.322801831073055e-05,
      "loss": 0.06383585929870605,
      "step": 4215
    },
    {
      "epoch": 0.5699704267004647,
      "grad_norm": 0.32947397232055664,
      "learning_rate": 1.322122199245526e-05,
      "loss": 0.06831598281860352,
      "step": 4216
    },
    {
      "epoch": 0.5701056189269117,
      "grad_norm": 0.2474450170993805,
      "learning_rate": 1.321442604453103e-05,
      "loss": 0.06218385696411133,
      "step": 4217
    },
    {
      "epoch": 0.5702408111533587,
      "grad_norm": 0.3361920118331909,
      "learning_rate": 1.320763046837282e-05,
      "loss": 0.06545305252075195,
      "step": 4218
    },
    {
      "epoch": 0.5703760033798057,
      "grad_norm": 0.44064798951148987,
      "learning_rate": 1.3200835265395504e-05,
      "loss": 0.0945591926574707,
      "step": 4219
    },
    {
      "epoch": 0.5705111956062526,
      "grad_norm": 0.4850946068763733,
      "learning_rate": 1.3194040437013885e-05,
      "loss": 0.08764171600341797,
      "step": 4220
    },
    {
      "epoch": 0.5706463878326996,
      "grad_norm": 0.3688046634197235,
      "learning_rate": 1.3187245984642673e-05,
      "loss": 0.05492281913757324,
      "step": 4221
    },
    {
      "epoch": 0.5707815800591466,
      "grad_norm": 0.8212231397628784,
      "learning_rate": 1.3180451909696517e-05,
      "loss": 0.04170036315917969,
      "step": 4222
    },
    {
      "epoch": 0.5709167722855936,
      "grad_norm": 0.2588346600532532,
      "learning_rate": 1.3173658213589972e-05,
      "loss": 0.057157278060913086,
      "step": 4223
    },
    {
      "epoch": 0.5710519645120405,
      "grad_norm": 0.7691304683685303,
      "learning_rate": 1.3166864897737526e-05,
      "loss": 0.0939931869506836,
      "step": 4224
    },
    {
      "epoch": 0.5711871567384875,
      "grad_norm": 0.22323811054229736,
      "learning_rate": 1.3160071963553593e-05,
      "loss": 0.039671897888183594,
      "step": 4225
    },
    {
      "epoch": 0.5713223489649345,
      "grad_norm": 0.4611817002296448,
      "learning_rate": 1.315327941245248e-05,
      "loss": 0.09499216079711914,
      "step": 4226
    },
    {
      "epoch": 0.5714575411913815,
      "grad_norm": 0.3247144818305969,
      "learning_rate": 1.3146487245848445e-05,
      "loss": 0.07895421981811523,
      "step": 4227
    },
    {
      "epoch": 0.5715927334178285,
      "grad_norm": 0.2039884477853775,
      "learning_rate": 1.3139695465155645e-05,
      "loss": 0.0419316291809082,
      "step": 4228
    },
    {
      "epoch": 0.5717279256442754,
      "grad_norm": 0.3020569980144501,
      "learning_rate": 1.3132904071788177e-05,
      "loss": 0.07380104064941406,
      "step": 4229
    },
    {
      "epoch": 0.5718631178707224,
      "grad_norm": 0.421318382024765,
      "learning_rate": 1.3126113067160031e-05,
      "loss": 0.07337641716003418,
      "step": 4230
    },
    {
      "epoch": 0.5719983100971694,
      "grad_norm": 0.26076003909111023,
      "learning_rate": 1.3119322452685139e-05,
      "loss": 0.07502365112304688,
      "step": 4231
    },
    {
      "epoch": 0.5721335023236164,
      "grad_norm": 0.456095814704895,
      "learning_rate": 1.3112532229777344e-05,
      "loss": 0.09958291053771973,
      "step": 4232
    },
    {
      "epoch": 0.5722686945500634,
      "grad_norm": 0.3706338405609131,
      "learning_rate": 1.3105742399850399e-05,
      "loss": 0.0721426010131836,
      "step": 4233
    },
    {
      "epoch": 0.5724038867765103,
      "grad_norm": 0.1377008855342865,
      "learning_rate": 1.3098952964317996e-05,
      "loss": 0.04186058044433594,
      "step": 4234
    },
    {
      "epoch": 0.5725390790029573,
      "grad_norm": 0.22118398547172546,
      "learning_rate": 1.3092163924593717e-05,
      "loss": 0.04795026779174805,
      "step": 4235
    },
    {
      "epoch": 0.5726742712294043,
      "grad_norm": 0.2733802795410156,
      "learning_rate": 1.308537528209108e-05,
      "loss": 0.07524681091308594,
      "step": 4236
    },
    {
      "epoch": 0.5728094634558513,
      "grad_norm": 0.3269210755825043,
      "learning_rate": 1.3078587038223525e-05,
      "loss": 0.07893228530883789,
      "step": 4237
    },
    {
      "epoch": 0.5729446556822982,
      "grad_norm": 0.37420061230659485,
      "learning_rate": 1.3071799194404392e-05,
      "loss": 0.1178579330444336,
      "step": 4238
    },
    {
      "epoch": 0.5730798479087452,
      "grad_norm": 0.4048102796077728,
      "learning_rate": 1.3065011752046955e-05,
      "loss": 0.10272502899169922,
      "step": 4239
    },
    {
      "epoch": 0.5732150401351922,
      "grad_norm": 0.20261088013648987,
      "learning_rate": 1.3058224712564382e-05,
      "loss": 0.05172419548034668,
      "step": 4240
    },
    {
      "epoch": 0.5733502323616392,
      "grad_norm": 0.2715659439563751,
      "learning_rate": 1.305143807736978e-05,
      "loss": 0.07482242584228516,
      "step": 4241
    },
    {
      "epoch": 0.5734854245880862,
      "grad_norm": 0.3544654846191406,
      "learning_rate": 1.3044651847876163e-05,
      "loss": 0.08915042877197266,
      "step": 4242
    },
    {
      "epoch": 0.5736206168145331,
      "grad_norm": 0.6068330407142639,
      "learning_rate": 1.3037866025496466e-05,
      "loss": 0.09934806823730469,
      "step": 4243
    },
    {
      "epoch": 0.5737558090409801,
      "grad_norm": 0.3568841516971588,
      "learning_rate": 1.3031080611643514e-05,
      "loss": 0.06970405578613281,
      "step": 4244
    },
    {
      "epoch": 0.5738910012674271,
      "grad_norm": 0.27211013436317444,
      "learning_rate": 1.3024295607730083e-05,
      "loss": 0.07442402839660645,
      "step": 4245
    },
    {
      "epoch": 0.5740261934938741,
      "grad_norm": 0.3491658568382263,
      "learning_rate": 1.301751101516884e-05,
      "loss": 0.0696868896484375,
      "step": 4246
    },
    {
      "epoch": 0.574161385720321,
      "grad_norm": 0.4164893925189972,
      "learning_rate": 1.3010726835372377e-05,
      "loss": 0.06960058212280273,
      "step": 4247
    },
    {
      "epoch": 0.574296577946768,
      "grad_norm": 0.667211651802063,
      "learning_rate": 1.30039430697532e-05,
      "loss": 0.12839508056640625,
      "step": 4248
    },
    {
      "epoch": 0.574431770173215,
      "grad_norm": 0.35223665833473206,
      "learning_rate": 1.2997159719723713e-05,
      "loss": 0.05519413948059082,
      "step": 4249
    },
    {
      "epoch": 0.574566962399662,
      "grad_norm": 0.196187362074852,
      "learning_rate": 1.2990376786696254e-05,
      "loss": 0.049340248107910156,
      "step": 4250
    },
    {
      "epoch": 0.574702154626109,
      "grad_norm": 0.2209187150001526,
      "learning_rate": 1.2983594272083063e-05,
      "loss": 0.06000876426696777,
      "step": 4251
    },
    {
      "epoch": 0.5748373468525559,
      "grad_norm": 0.6193345785140991,
      "learning_rate": 1.2976812177296307e-05,
      "loss": 0.08017492294311523,
      "step": 4252
    },
    {
      "epoch": 0.5749725390790029,
      "grad_norm": 0.21420758962631226,
      "learning_rate": 1.2970030503748039e-05,
      "loss": 0.05553770065307617,
      "step": 4253
    },
    {
      "epoch": 0.5751077313054499,
      "grad_norm": 0.24513158202171326,
      "learning_rate": 1.2963249252850242e-05,
      "loss": 0.04099082946777344,
      "step": 4254
    },
    {
      "epoch": 0.5752429235318969,
      "grad_norm": 0.29867297410964966,
      "learning_rate": 1.295646842601481e-05,
      "loss": 0.05138349533081055,
      "step": 4255
    },
    {
      "epoch": 0.5753781157583439,
      "grad_norm": 0.5102136731147766,
      "learning_rate": 1.294968802465355e-05,
      "loss": 0.10262441635131836,
      "step": 4256
    },
    {
      "epoch": 0.5755133079847908,
      "grad_norm": 0.24540355801582336,
      "learning_rate": 1.2942908050178187e-05,
      "loss": 0.05449390411376953,
      "step": 4257
    },
    {
      "epoch": 0.5756485002112378,
      "grad_norm": 0.24239884316921234,
      "learning_rate": 1.293612850400033e-05,
      "loss": 0.05982780456542969,
      "step": 4258
    },
    {
      "epoch": 0.5757836924376848,
      "grad_norm": 0.3142111599445343,
      "learning_rate": 1.2929349387531525e-05,
      "loss": 0.08676910400390625,
      "step": 4259
    },
    {
      "epoch": 0.5759188846641318,
      "grad_norm": 0.28510600328445435,
      "learning_rate": 1.2922570702183217e-05,
      "loss": 0.054271697998046875,
      "step": 4260
    },
    {
      "epoch": 0.5760540768905787,
      "grad_norm": 0.2281937450170517,
      "learning_rate": 1.2915792449366768e-05,
      "loss": 0.05469083786010742,
      "step": 4261
    },
    {
      "epoch": 0.5761892691170257,
      "grad_norm": 0.3259390592575073,
      "learning_rate": 1.2909014630493451e-05,
      "loss": 0.06585693359375,
      "step": 4262
    },
    {
      "epoch": 0.5763244613434727,
      "grad_norm": 0.2396048605442047,
      "learning_rate": 1.2902237246974432e-05,
      "loss": 0.06091499328613281,
      "step": 4263
    },
    {
      "epoch": 0.5764596535699197,
      "grad_norm": 0.34529733657836914,
      "learning_rate": 1.289546030022081e-05,
      "loss": 0.06329727172851562,
      "step": 4264
    },
    {
      "epoch": 0.5765948457963667,
      "grad_norm": 0.2951788306236267,
      "learning_rate": 1.2888683791643572e-05,
      "loss": 0.05420076847076416,
      "step": 4265
    },
    {
      "epoch": 0.5767300380228136,
      "grad_norm": 0.27161645889282227,
      "learning_rate": 1.2881907722653633e-05,
      "loss": 0.044722557067871094,
      "step": 4266
    },
    {
      "epoch": 0.5768652302492606,
      "grad_norm": 0.19044330716133118,
      "learning_rate": 1.2875132094661796e-05,
      "loss": 0.04693293571472168,
      "step": 4267
    },
    {
      "epoch": 0.5770004224757076,
      "grad_norm": 0.42991766333580017,
      "learning_rate": 1.2868356909078787e-05,
      "loss": 0.08460688591003418,
      "step": 4268
    },
    {
      "epoch": 0.5771356147021546,
      "grad_norm": 0.23017805814743042,
      "learning_rate": 1.286158216731524e-05,
      "loss": 0.0533832311630249,
      "step": 4269
    },
    {
      "epoch": 0.5772708069286016,
      "grad_norm": 0.21773308515548706,
      "learning_rate": 1.2854807870781686e-05,
      "loss": 0.06833410263061523,
      "step": 4270
    },
    {
      "epoch": 0.5774059991550485,
      "grad_norm": 0.2180422693490982,
      "learning_rate": 1.284803402088858e-05,
      "loss": 0.058348894119262695,
      "step": 4271
    },
    {
      "epoch": 0.5775411913814955,
      "grad_norm": 0.39264976978302,
      "learning_rate": 1.284126061904626e-05,
      "loss": 0.06962013244628906,
      "step": 4272
    },
    {
      "epoch": 0.5776763836079425,
      "grad_norm": 0.40345919132232666,
      "learning_rate": 1.283448766666499e-05,
      "loss": 0.10006141662597656,
      "step": 4273
    },
    {
      "epoch": 0.5778115758343895,
      "grad_norm": 0.25123125314712524,
      "learning_rate": 1.282771516515494e-05,
      "loss": 0.0528331995010376,
      "step": 4274
    },
    {
      "epoch": 0.5779467680608364,
      "grad_norm": 0.3381307125091553,
      "learning_rate": 1.282094311592618e-05,
      "loss": 0.05719423294067383,
      "step": 4275
    },
    {
      "epoch": 0.5780819602872835,
      "grad_norm": 0.26049450039863586,
      "learning_rate": 1.2814171520388676e-05,
      "loss": 0.05408525466918945,
      "step": 4276
    },
    {
      "epoch": 0.5782171525137305,
      "grad_norm": 0.39702731370925903,
      "learning_rate": 1.2807400379952318e-05,
      "loss": 0.07463216781616211,
      "step": 4277
    },
    {
      "epoch": 0.5783523447401775,
      "grad_norm": 0.5777104496955872,
      "learning_rate": 1.2800629696026895e-05,
      "loss": 0.09904742240905762,
      "step": 4278
    },
    {
      "epoch": 0.5784875369666245,
      "grad_norm": 0.4103236198425293,
      "learning_rate": 1.2793859470022098e-05,
      "loss": 0.07288527488708496,
      "step": 4279
    },
    {
      "epoch": 0.5786227291930715,
      "grad_norm": 0.32955893874168396,
      "learning_rate": 1.278708970334753e-05,
      "loss": 0.07037973403930664,
      "step": 4280
    },
    {
      "epoch": 0.5787579214195184,
      "grad_norm": 0.3472101092338562,
      "learning_rate": 1.2780320397412678e-05,
      "loss": 0.07021617889404297,
      "step": 4281
    },
    {
      "epoch": 0.5788931136459654,
      "grad_norm": 0.6170212030410767,
      "learning_rate": 1.2773551553626957e-05,
      "loss": 0.06719493865966797,
      "step": 4282
    },
    {
      "epoch": 0.5790283058724124,
      "grad_norm": 0.4556148648262024,
      "learning_rate": 1.2766783173399675e-05,
      "loss": 0.07308840751647949,
      "step": 4283
    },
    {
      "epoch": 0.5791634980988594,
      "grad_norm": 0.5287730097770691,
      "learning_rate": 1.276001525814005e-05,
      "loss": 0.07259821891784668,
      "step": 4284
    },
    {
      "epoch": 0.5792986903253063,
      "grad_norm": 0.3877529799938202,
      "learning_rate": 1.2753247809257192e-05,
      "loss": 0.06958603858947754,
      "step": 4285
    },
    {
      "epoch": 0.5794338825517533,
      "grad_norm": 0.31652677059173584,
      "learning_rate": 1.2746480828160119e-05,
      "loss": 0.05983781814575195,
      "step": 4286
    },
    {
      "epoch": 0.5795690747782003,
      "grad_norm": 0.2596035897731781,
      "learning_rate": 1.2739714316257753e-05,
      "loss": 0.06442499160766602,
      "step": 4287
    },
    {
      "epoch": 0.5797042670046473,
      "grad_norm": 0.4206702411174774,
      "learning_rate": 1.273294827495892e-05,
      "loss": 0.08462643623352051,
      "step": 4288
    },
    {
      "epoch": 0.5798394592310943,
      "grad_norm": 0.3511323630809784,
      "learning_rate": 1.2726182705672352e-05,
      "loss": 0.06094837188720703,
      "step": 4289
    },
    {
      "epoch": 0.5799746514575412,
      "grad_norm": 0.2860526144504547,
      "learning_rate": 1.271941760980667e-05,
      "loss": 0.0730443000793457,
      "step": 4290
    },
    {
      "epoch": 0.5801098436839882,
      "grad_norm": 0.578201174736023,
      "learning_rate": 1.2712652988770396e-05,
      "loss": 0.07014274597167969,
      "step": 4291
    },
    {
      "epoch": 0.5802450359104352,
      "grad_norm": 0.2719123065471649,
      "learning_rate": 1.2705888843971967e-05,
      "loss": 0.05372893810272217,
      "step": 4292
    },
    {
      "epoch": 0.5803802281368822,
      "grad_norm": 0.37068089842796326,
      "learning_rate": 1.2699125176819717e-05,
      "loss": 0.07972216606140137,
      "step": 4293
    },
    {
      "epoch": 0.5805154203633291,
      "grad_norm": 0.14126096665859222,
      "learning_rate": 1.269236198872188e-05,
      "loss": 0.04064035415649414,
      "step": 4294
    },
    {
      "epoch": 0.5806506125897761,
      "grad_norm": 0.2503538727760315,
      "learning_rate": 1.2685599281086577e-05,
      "loss": 0.06243181228637695,
      "step": 4295
    },
    {
      "epoch": 0.5807858048162231,
      "grad_norm": 0.15925723314285278,
      "learning_rate": 1.2678837055321849e-05,
      "loss": 0.04054218530654907,
      "step": 4296
    },
    {
      "epoch": 0.5809209970426701,
      "grad_norm": 0.3280768394470215,
      "learning_rate": 1.267207531283562e-05,
      "loss": 0.07500123977661133,
      "step": 4297
    },
    {
      "epoch": 0.5810561892691171,
      "grad_norm": 0.17902933061122894,
      "learning_rate": 1.266531405503573e-05,
      "loss": 0.035021305084228516,
      "step": 4298
    },
    {
      "epoch": 0.581191381495564,
      "grad_norm": 0.30209484696388245,
      "learning_rate": 1.26585532833299e-05,
      "loss": 0.078033447265625,
      "step": 4299
    },
    {
      "epoch": 0.581326573722011,
      "grad_norm": 0.4677204489707947,
      "learning_rate": 1.2651792999125763e-05,
      "loss": 0.08407258987426758,
      "step": 4300
    },
    {
      "epoch": 0.581461765948458,
      "grad_norm": 0.18018262088298798,
      "learning_rate": 1.2645033203830846e-05,
      "loss": 0.04263925552368164,
      "step": 4301
    },
    {
      "epoch": 0.581596958174905,
      "grad_norm": 0.23889626562595367,
      "learning_rate": 1.2638273898852573e-05,
      "loss": 0.06528663635253906,
      "step": 4302
    },
    {
      "epoch": 0.581732150401352,
      "grad_norm": 0.22320957481861115,
      "learning_rate": 1.2631515085598275e-05,
      "loss": 0.054903507232666016,
      "step": 4303
    },
    {
      "epoch": 0.5818673426277989,
      "grad_norm": 0.30559372901916504,
      "learning_rate": 1.262475676547516e-05,
      "loss": 0.08128738403320312,
      "step": 4304
    },
    {
      "epoch": 0.5820025348542459,
      "grad_norm": 0.32230517268180847,
      "learning_rate": 1.2617998939890352e-05,
      "loss": 0.05758810043334961,
      "step": 4305
    },
    {
      "epoch": 0.5821377270806929,
      "grad_norm": 0.3014519214630127,
      "learning_rate": 1.261124161025087e-05,
      "loss": 0.07735538482666016,
      "step": 4306
    },
    {
      "epoch": 0.5822729193071399,
      "grad_norm": 0.2579478919506073,
      "learning_rate": 1.260448477796362e-05,
      "loss": 0.06965446472167969,
      "step": 4307
    },
    {
      "epoch": 0.5824081115335868,
      "grad_norm": 0.16503727436065674,
      "learning_rate": 1.259772844443542e-05,
      "loss": 0.03880476951599121,
      "step": 4308
    },
    {
      "epoch": 0.5825433037600338,
      "grad_norm": 0.47773122787475586,
      "learning_rate": 1.2590972611072964e-05,
      "loss": 0.0793771743774414,
      "step": 4309
    },
    {
      "epoch": 0.5826784959864808,
      "grad_norm": 0.5266780257225037,
      "learning_rate": 1.2584217279282855e-05,
      "loss": 0.08047771453857422,
      "step": 4310
    },
    {
      "epoch": 0.5828136882129278,
      "grad_norm": 0.19990146160125732,
      "learning_rate": 1.2577462450471593e-05,
      "loss": 0.04231917858123779,
      "step": 4311
    },
    {
      "epoch": 0.5829488804393748,
      "grad_norm": 0.40081676840782166,
      "learning_rate": 1.2570708126045574e-05,
      "loss": 0.09717178344726562,
      "step": 4312
    },
    {
      "epoch": 0.5830840726658217,
      "grad_norm": 0.23123854398727417,
      "learning_rate": 1.256395430741107e-05,
      "loss": 0.0459747314453125,
      "step": 4313
    },
    {
      "epoch": 0.5832192648922687,
      "grad_norm": 0.2861766815185547,
      "learning_rate": 1.2557200995974268e-05,
      "loss": 0.06968975067138672,
      "step": 4314
    },
    {
      "epoch": 0.5833544571187157,
      "grad_norm": 0.3242214620113373,
      "learning_rate": 1.2550448193141248e-05,
      "loss": 0.05453753471374512,
      "step": 4315
    },
    {
      "epoch": 0.5834896493451627,
      "grad_norm": 0.33520573377609253,
      "learning_rate": 1.2543695900317977e-05,
      "loss": 0.07026195526123047,
      "step": 4316
    },
    {
      "epoch": 0.5836248415716097,
      "grad_norm": 0.353973388671875,
      "learning_rate": 1.2536944118910323e-05,
      "loss": 0.05172041058540344,
      "step": 4317
    },
    {
      "epoch": 0.5837600337980566,
      "grad_norm": 0.20054422318935394,
      "learning_rate": 1.2530192850324032e-05,
      "loss": 0.048375606536865234,
      "step": 4318
    },
    {
      "epoch": 0.5838952260245036,
      "grad_norm": 0.18036530911922455,
      "learning_rate": 1.252344209596476e-05,
      "loss": 0.041887491941452026,
      "step": 4319
    },
    {
      "epoch": 0.5840304182509506,
      "grad_norm": 0.41878896951675415,
      "learning_rate": 1.251669185723805e-05,
      "loss": 0.08332812786102295,
      "step": 4320
    },
    {
      "epoch": 0.5841656104773976,
      "grad_norm": 0.3873213827610016,
      "learning_rate": 1.2509942135549344e-05,
      "loss": 0.06663715839385986,
      "step": 4321
    },
    {
      "epoch": 0.5843008027038445,
      "grad_norm": 0.6192523837089539,
      "learning_rate": 1.250319293230396e-05,
      "loss": 0.07886505126953125,
      "step": 4322
    },
    {
      "epoch": 0.5844359949302915,
      "grad_norm": 0.4124995172023773,
      "learning_rate": 1.2496444248907121e-05,
      "loss": 0.07800483703613281,
      "step": 4323
    },
    {
      "epoch": 0.5845711871567385,
      "grad_norm": 0.14207519590854645,
      "learning_rate": 1.2489696086763939e-05,
      "loss": 0.03564596176147461,
      "step": 4324
    },
    {
      "epoch": 0.5847063793831855,
      "grad_norm": 0.31678009033203125,
      "learning_rate": 1.2482948447279417e-05,
      "loss": 0.059558868408203125,
      "step": 4325
    },
    {
      "epoch": 0.5848415716096325,
      "grad_norm": 0.3399200439453125,
      "learning_rate": 1.2476201331858458e-05,
      "loss": 0.07506275177001953,
      "step": 4326
    },
    {
      "epoch": 0.5849767638360794,
      "grad_norm": 0.3999166488647461,
      "learning_rate": 1.2469454741905839e-05,
      "loss": 0.06264925003051758,
      "step": 4327
    },
    {
      "epoch": 0.5851119560625264,
      "grad_norm": 0.4690093398094177,
      "learning_rate": 1.2462708678826233e-05,
      "loss": 0.08655166625976562,
      "step": 4328
    },
    {
      "epoch": 0.5852471482889734,
      "grad_norm": 0.31304731965065,
      "learning_rate": 1.245596314402421e-05,
      "loss": 0.08196377754211426,
      "step": 4329
    },
    {
      "epoch": 0.5853823405154204,
      "grad_norm": 0.23181632161140442,
      "learning_rate": 1.2449218138904225e-05,
      "loss": 0.05222761631011963,
      "step": 4330
    },
    {
      "epoch": 0.5855175327418674,
      "grad_norm": 0.22853532433509827,
      "learning_rate": 1.2442473664870636e-05,
      "loss": 0.06299638748168945,
      "step": 4331
    },
    {
      "epoch": 0.5856527249683143,
      "grad_norm": 0.3060552775859833,
      "learning_rate": 1.2435729723327661e-05,
      "loss": 0.06338882446289062,
      "step": 4332
    },
    {
      "epoch": 0.5857879171947613,
      "grad_norm": 0.38231733441352844,
      "learning_rate": 1.2428986315679433e-05,
      "loss": 0.07598447799682617,
      "step": 4333
    },
    {
      "epoch": 0.5859231094212083,
      "grad_norm": 0.4449533522129059,
      "learning_rate": 1.2422243443329962e-05,
      "loss": 0.07172644138336182,
      "step": 4334
    },
    {
      "epoch": 0.5860583016476553,
      "grad_norm": 0.3609060049057007,
      "learning_rate": 1.241550110768316e-05,
      "loss": 0.07144355773925781,
      "step": 4335
    },
    {
      "epoch": 0.5861934938741022,
      "grad_norm": 0.4742407202720642,
      "learning_rate": 1.2408759310142803e-05,
      "loss": 0.08863067626953125,
      "step": 4336
    },
    {
      "epoch": 0.5863286861005492,
      "grad_norm": 0.40765735507011414,
      "learning_rate": 1.2402018052112576e-05,
      "loss": 0.0685725212097168,
      "step": 4337
    },
    {
      "epoch": 0.5864638783269962,
      "grad_norm": 0.2166009247303009,
      "learning_rate": 1.2395277334996045e-05,
      "loss": 0.05163097381591797,
      "step": 4338
    },
    {
      "epoch": 0.5865990705534432,
      "grad_norm": 0.2810015380382538,
      "learning_rate": 1.2388537160196663e-05,
      "loss": 0.07212471961975098,
      "step": 4339
    },
    {
      "epoch": 0.5867342627798902,
      "grad_norm": 0.8222311735153198,
      "learning_rate": 1.2381797529117776e-05,
      "loss": 0.09038233757019043,
      "step": 4340
    },
    {
      "epoch": 0.5868694550063371,
      "grad_norm": 0.4227246344089508,
      "learning_rate": 1.23750584431626e-05,
      "loss": 0.08516407012939453,
      "step": 4341
    },
    {
      "epoch": 0.5870046472327841,
      "grad_norm": 0.278315007686615,
      "learning_rate": 1.236831990373425e-05,
      "loss": 0.07106304168701172,
      "step": 4342
    },
    {
      "epoch": 0.5871398394592311,
      "grad_norm": 0.46818864345550537,
      "learning_rate": 1.2361581912235736e-05,
      "loss": 0.09428846836090088,
      "step": 4343
    },
    {
      "epoch": 0.5872750316856781,
      "grad_norm": 0.42657315731048584,
      "learning_rate": 1.235484447006994e-05,
      "loss": 0.06872272491455078,
      "step": 4344
    },
    {
      "epoch": 0.587410223912125,
      "grad_norm": 0.341647744178772,
      "learning_rate": 1.2348107578639627e-05,
      "loss": 0.05318880081176758,
      "step": 4345
    },
    {
      "epoch": 0.587545416138572,
      "grad_norm": 0.28094810247421265,
      "learning_rate": 1.2341371239347454e-05,
      "loss": 0.07716250419616699,
      "step": 4346
    },
    {
      "epoch": 0.587680608365019,
      "grad_norm": 0.29405146837234497,
      "learning_rate": 1.233463545359597e-05,
      "loss": 0.07787227630615234,
      "step": 4347
    },
    {
      "epoch": 0.587815800591466,
      "grad_norm": 0.26101914048194885,
      "learning_rate": 1.23279002227876e-05,
      "loss": 0.06612205505371094,
      "step": 4348
    },
    {
      "epoch": 0.587950992817913,
      "grad_norm": 0.40927189588546753,
      "learning_rate": 1.2321165548324655e-05,
      "loss": 0.08002543449401855,
      "step": 4349
    },
    {
      "epoch": 0.5880861850443599,
      "grad_norm": 0.3589462339878082,
      "learning_rate": 1.2314431431609323e-05,
      "loss": 0.05216789245605469,
      "step": 4350
    },
    {
      "epoch": 0.5882213772708069,
      "grad_norm": 0.268644779920578,
      "learning_rate": 1.2307697874043687e-05,
      "loss": 0.0704643726348877,
      "step": 4351
    },
    {
      "epoch": 0.5883565694972539,
      "grad_norm": 0.2965880334377289,
      "learning_rate": 1.2300964877029712e-05,
      "loss": 0.05938005447387695,
      "step": 4352
    },
    {
      "epoch": 0.5884917617237009,
      "grad_norm": 0.3145478069782257,
      "learning_rate": 1.2294232441969246e-05,
      "loss": 0.06465482711791992,
      "step": 4353
    },
    {
      "epoch": 0.5886269539501479,
      "grad_norm": 0.28410008549690247,
      "learning_rate": 1.2287500570264017e-05,
      "loss": 0.05022144317626953,
      "step": 4354
    },
    {
      "epoch": 0.5887621461765948,
      "grad_norm": 0.23661217093467712,
      "learning_rate": 1.2280769263315628e-05,
      "loss": 0.042916297912597656,
      "step": 4355
    },
    {
      "epoch": 0.5888973384030418,
      "grad_norm": 0.39567339420318604,
      "learning_rate": 1.2274038522525577e-05,
      "loss": 0.08017563819885254,
      "step": 4356
    },
    {
      "epoch": 0.5890325306294888,
      "grad_norm": 0.31014779210090637,
      "learning_rate": 1.2267308349295246e-05,
      "loss": 0.07112038135528564,
      "step": 4357
    },
    {
      "epoch": 0.5891677228559358,
      "grad_norm": 0.1551205962896347,
      "learning_rate": 1.2260578745025892e-05,
      "loss": 0.02809929847717285,
      "step": 4358
    },
    {
      "epoch": 0.5893029150823828,
      "grad_norm": 0.40543797612190247,
      "learning_rate": 1.225384971111865e-05,
      "loss": 0.09670495986938477,
      "step": 4359
    },
    {
      "epoch": 0.5894381073088297,
      "grad_norm": 0.3169974088668823,
      "learning_rate": 1.224712124897454e-05,
      "loss": 0.06278276443481445,
      "step": 4360
    },
    {
      "epoch": 0.5895732995352767,
      "grad_norm": 0.3309388756752014,
      "learning_rate": 1.2240393359994466e-05,
      "loss": 0.07917428016662598,
      "step": 4361
    },
    {
      "epoch": 0.5897084917617237,
      "grad_norm": 0.48446932435035706,
      "learning_rate": 1.2233666045579209e-05,
      "loss": 0.0726470947265625,
      "step": 4362
    },
    {
      "epoch": 0.5898436839881707,
      "grad_norm": 0.32095468044281006,
      "learning_rate": 1.222693930712944e-05,
      "loss": 0.08367252349853516,
      "step": 4363
    },
    {
      "epoch": 0.5899788762146176,
      "grad_norm": 1.0433701276779175,
      "learning_rate": 1.2220213146045691e-05,
      "loss": 0.0549013614654541,
      "step": 4364
    },
    {
      "epoch": 0.5901140684410646,
      "grad_norm": 0.48906344175338745,
      "learning_rate": 1.2213487563728389e-05,
      "loss": 0.04992187023162842,
      "step": 4365
    },
    {
      "epoch": 0.5902492606675116,
      "grad_norm": 0.6377118825912476,
      "learning_rate": 1.220676256157783e-05,
      "loss": 0.07468461990356445,
      "step": 4366
    },
    {
      "epoch": 0.5903844528939586,
      "grad_norm": 0.2501105070114136,
      "learning_rate": 1.2200038140994212e-05,
      "loss": 0.04983377456665039,
      "step": 4367
    },
    {
      "epoch": 0.5905196451204056,
      "grad_norm": 0.15972237288951874,
      "learning_rate": 1.2193314303377578e-05,
      "loss": 0.03852057456970215,
      "step": 4368
    },
    {
      "epoch": 0.5906548373468525,
      "grad_norm": 0.2540685832500458,
      "learning_rate": 1.2186591050127874e-05,
      "loss": 0.052690982818603516,
      "step": 4369
    },
    {
      "epoch": 0.5907900295732995,
      "grad_norm": 0.30108579993247986,
      "learning_rate": 1.2179868382644916e-05,
      "loss": 0.07224893569946289,
      "step": 4370
    },
    {
      "epoch": 0.5909252217997465,
      "grad_norm": 0.38752955198287964,
      "learning_rate": 1.2173146302328396e-05,
      "loss": 0.06566667556762695,
      "step": 4371
    },
    {
      "epoch": 0.5910604140261935,
      "grad_norm": 0.37940678000450134,
      "learning_rate": 1.21664248105779e-05,
      "loss": 0.06433296203613281,
      "step": 4372
    },
    {
      "epoch": 0.5911956062526404,
      "grad_norm": 0.9298123717308044,
      "learning_rate": 1.2159703908792858e-05,
      "loss": 0.11943340301513672,
      "step": 4373
    },
    {
      "epoch": 0.5913307984790874,
      "grad_norm": 0.4051517844200134,
      "learning_rate": 1.2152983598372613e-05,
      "loss": 0.08199405670166016,
      "step": 4374
    },
    {
      "epoch": 0.5914659907055344,
      "grad_norm": 0.4235312342643738,
      "learning_rate": 1.2146263880716366e-05,
      "loss": 0.07595133781433105,
      "step": 4375
    },
    {
      "epoch": 0.5916011829319814,
      "grad_norm": 0.43915656208992004,
      "learning_rate": 1.2139544757223194e-05,
      "loss": 0.11987781524658203,
      "step": 4376
    },
    {
      "epoch": 0.5917363751584284,
      "grad_norm": 0.40207841992378235,
      "learning_rate": 1.2132826229292066e-05,
      "loss": 0.08121633529663086,
      "step": 4377
    },
    {
      "epoch": 0.5918715673848753,
      "grad_norm": 0.47785353660583496,
      "learning_rate": 1.2126108298321798e-05,
      "loss": 0.0492706298828125,
      "step": 4378
    },
    {
      "epoch": 0.5920067596113223,
      "grad_norm": 0.3230375647544861,
      "learning_rate": 1.2119390965711107e-05,
      "loss": 0.07749605178833008,
      "step": 4379
    },
    {
      "epoch": 0.5921419518377693,
      "grad_norm": 0.29489728808403015,
      "learning_rate": 1.2112674232858582e-05,
      "loss": 0.05100727081298828,
      "step": 4380
    },
    {
      "epoch": 0.5922771440642163,
      "grad_norm": 0.6030992865562439,
      "learning_rate": 1.2105958101162684e-05,
      "loss": 0.09671342372894287,
      "step": 4381
    },
    {
      "epoch": 0.5924123362906633,
      "grad_norm": 0.2506835162639618,
      "learning_rate": 1.2099242572021735e-05,
      "loss": 0.055449843406677246,
      "step": 4382
    },
    {
      "epoch": 0.5925475285171102,
      "grad_norm": 0.17606131732463837,
      "learning_rate": 1.209252764683395e-05,
      "loss": 0.04475855827331543,
      "step": 4383
    },
    {
      "epoch": 0.5926827207435572,
      "grad_norm": 0.3181266188621521,
      "learning_rate": 1.2085813326997414e-05,
      "loss": 0.06715583801269531,
      "step": 4384
    },
    {
      "epoch": 0.5928179129700042,
      "grad_norm": 0.31322917342185974,
      "learning_rate": 1.2079099613910088e-05,
      "loss": 0.07672882080078125,
      "step": 4385
    },
    {
      "epoch": 0.5929531051964512,
      "grad_norm": 0.41575130820274353,
      "learning_rate": 1.20723865089698e-05,
      "loss": 0.062352657318115234,
      "step": 4386
    },
    {
      "epoch": 0.5930882974228981,
      "grad_norm": 0.22439508140087128,
      "learning_rate": 1.2065674013574248e-05,
      "loss": 0.05141806602478027,
      "step": 4387
    },
    {
      "epoch": 0.5932234896493451,
      "grad_norm": 0.3074764609336853,
      "learning_rate": 1.2058962129121013e-05,
      "loss": 0.06819868087768555,
      "step": 4388
    },
    {
      "epoch": 0.5933586818757921,
      "grad_norm": 0.4605916142463684,
      "learning_rate": 1.2052250857007548e-05,
      "loss": 0.0662388801574707,
      "step": 4389
    },
    {
      "epoch": 0.5934938741022391,
      "grad_norm": 0.22463290393352509,
      "learning_rate": 1.2045540198631177e-05,
      "loss": 0.048723697662353516,
      "step": 4390
    },
    {
      "epoch": 0.5936290663286861,
      "grad_norm": 0.2734999656677246,
      "learning_rate": 1.2038830155389091e-05,
      "loss": 0.06160116195678711,
      "step": 4391
    },
    {
      "epoch": 0.593764258555133,
      "grad_norm": 0.5060994029045105,
      "learning_rate": 1.2032120728678354e-05,
      "loss": 0.06933116912841797,
      "step": 4392
    },
    {
      "epoch": 0.59389945078158,
      "grad_norm": 0.3359962999820709,
      "learning_rate": 1.2025411919895907e-05,
      "loss": 0.06267333030700684,
      "step": 4393
    },
    {
      "epoch": 0.594034643008027,
      "grad_norm": 0.3087286949157715,
      "learning_rate": 1.2018703730438561e-05,
      "loss": 0.07016444206237793,
      "step": 4394
    },
    {
      "epoch": 0.594169835234474,
      "grad_norm": 0.37510600686073303,
      "learning_rate": 1.2011996161703003e-05,
      "loss": 0.07643699645996094,
      "step": 4395
    },
    {
      "epoch": 0.594305027460921,
      "grad_norm": 0.5119466185569763,
      "learning_rate": 1.2005289215085775e-05,
      "loss": 0.10711193084716797,
      "step": 4396
    },
    {
      "epoch": 0.5944402196873679,
      "grad_norm": 0.33476558327674866,
      "learning_rate": 1.19985828919833e-05,
      "loss": 0.05661749839782715,
      "step": 4397
    },
    {
      "epoch": 0.5945754119138149,
      "grad_norm": 0.19884833693504333,
      "learning_rate": 1.1991877193791872e-05,
      "loss": 0.048090994358062744,
      "step": 4398
    },
    {
      "epoch": 0.5947106041402619,
      "grad_norm": 0.26273098587989807,
      "learning_rate": 1.1985172121907653e-05,
      "loss": 0.06650567054748535,
      "step": 4399
    },
    {
      "epoch": 0.5948457963667089,
      "grad_norm": 0.6869086623191833,
      "learning_rate": 1.1978467677726682e-05,
      "loss": 0.0905604362487793,
      "step": 4400
    },
    {
      "epoch": 0.5949809885931558,
      "grad_norm": 0.5525241494178772,
      "learning_rate": 1.197176386264485e-05,
      "loss": 0.08647680282592773,
      "step": 4401
    },
    {
      "epoch": 0.5951161808196028,
      "grad_norm": 0.17063087224960327,
      "learning_rate": 1.1965060678057927e-05,
      "loss": 0.04663515090942383,
      "step": 4402
    },
    {
      "epoch": 0.5952513730460498,
      "grad_norm": 0.5738041996955872,
      "learning_rate": 1.1958358125361554e-05,
      "loss": 0.10048294067382812,
      "step": 4403
    },
    {
      "epoch": 0.5953865652724968,
      "grad_norm": 0.18927814066410065,
      "learning_rate": 1.1951656205951247e-05,
      "loss": 0.04247361421585083,
      "step": 4404
    },
    {
      "epoch": 0.5955217574989438,
      "grad_norm": 0.4425094425678253,
      "learning_rate": 1.1944954921222367e-05,
      "loss": 0.06999385356903076,
      "step": 4405
    },
    {
      "epoch": 0.5956569497253907,
      "grad_norm": 0.2552849352359772,
      "learning_rate": 1.1938254272570167e-05,
      "loss": 0.03948771953582764,
      "step": 4406
    },
    {
      "epoch": 0.5957921419518377,
      "grad_norm": 0.337760329246521,
      "learning_rate": 1.1931554261389751e-05,
      "loss": 0.06551945209503174,
      "step": 4407
    },
    {
      "epoch": 0.5959273341782847,
      "grad_norm": 0.35250124335289,
      "learning_rate": 1.1924854889076103e-05,
      "loss": 0.07527542114257812,
      "step": 4408
    },
    {
      "epoch": 0.5960625264047317,
      "grad_norm": 0.29173189401626587,
      "learning_rate": 1.191815615702407e-05,
      "loss": 0.057657837867736816,
      "step": 4409
    },
    {
      "epoch": 0.5961977186311787,
      "grad_norm": 0.33585357666015625,
      "learning_rate": 1.1911458066628353e-05,
      "loss": 0.06293225288391113,
      "step": 4410
    },
    {
      "epoch": 0.5963329108576257,
      "grad_norm": 0.41833505034446716,
      "learning_rate": 1.1904760619283537e-05,
      "loss": 0.0571436882019043,
      "step": 4411
    },
    {
      "epoch": 0.5964681030840727,
      "grad_norm": 0.48114222288131714,
      "learning_rate": 1.1898063816384069e-05,
      "loss": 0.09582042694091797,
      "step": 4412
    },
    {
      "epoch": 0.5966032953105197,
      "grad_norm": 0.2538919150829315,
      "learning_rate": 1.189136765932426e-05,
      "loss": 0.0637664794921875,
      "step": 4413
    },
    {
      "epoch": 0.5967384875369667,
      "grad_norm": 0.1911185383796692,
      "learning_rate": 1.1884672149498276e-05,
      "loss": 0.05926513671875,
      "step": 4414
    },
    {
      "epoch": 0.5968736797634137,
      "grad_norm": 0.37829163670539856,
      "learning_rate": 1.187797728830016e-05,
      "loss": 0.04589354991912842,
      "step": 4415
    },
    {
      "epoch": 0.5970088719898606,
      "grad_norm": 0.560752272605896,
      "learning_rate": 1.1871283077123823e-05,
      "loss": 0.09465360641479492,
      "step": 4416
    },
    {
      "epoch": 0.5971440642163076,
      "grad_norm": 0.23581284284591675,
      "learning_rate": 1.1864589517363038e-05,
      "loss": 0.04889726638793945,
      "step": 4417
    },
    {
      "epoch": 0.5972792564427546,
      "grad_norm": 0.2395762801170349,
      "learning_rate": 1.185789661041144e-05,
      "loss": 0.0637204647064209,
      "step": 4418
    },
    {
      "epoch": 0.5974144486692016,
      "grad_norm": 0.24712955951690674,
      "learning_rate": 1.1851204357662513e-05,
      "loss": 0.06745243072509766,
      "step": 4419
    },
    {
      "epoch": 0.5975496408956485,
      "grad_norm": 0.2869206666946411,
      "learning_rate": 1.1844512760509634e-05,
      "loss": 0.050952911376953125,
      "step": 4420
    },
    {
      "epoch": 0.5976848331220955,
      "grad_norm": 0.36108118295669556,
      "learning_rate": 1.1837821820346022e-05,
      "loss": 0.07438981533050537,
      "step": 4421
    },
    {
      "epoch": 0.5978200253485425,
      "grad_norm": 0.4791282117366791,
      "learning_rate": 1.1831131538564775e-05,
      "loss": 0.09142017364501953,
      "step": 4422
    },
    {
      "epoch": 0.5979552175749895,
      "grad_norm": 0.1886630803346634,
      "learning_rate": 1.1824441916558843e-05,
      "loss": 0.04613995552062988,
      "step": 4423
    },
    {
      "epoch": 0.5980904098014365,
      "grad_norm": 0.27580684423446655,
      "learning_rate": 1.1817752955721031e-05,
      "loss": 0.0632939338684082,
      "step": 4424
    },
    {
      "epoch": 0.5982256020278834,
      "grad_norm": 0.35286441445350647,
      "learning_rate": 1.1811064657444023e-05,
      "loss": 0.07007050514221191,
      "step": 4425
    },
    {
      "epoch": 0.5983607942543304,
      "grad_norm": 0.5837258696556091,
      "learning_rate": 1.1804377023120361e-05,
      "loss": 0.09163427352905273,
      "step": 4426
    },
    {
      "epoch": 0.5984959864807774,
      "grad_norm": 0.3789232075214386,
      "learning_rate": 1.1797690054142451e-05,
      "loss": 0.05483245849609375,
      "step": 4427
    },
    {
      "epoch": 0.5986311787072244,
      "grad_norm": 0.17387795448303223,
      "learning_rate": 1.1791003751902542e-05,
      "loss": 0.04521942138671875,
      "step": 4428
    },
    {
      "epoch": 0.5987663709336714,
      "grad_norm": 0.36982420086860657,
      "learning_rate": 1.1784318117792763e-05,
      "loss": 0.05423092842102051,
      "step": 4429
    },
    {
      "epoch": 0.5989015631601183,
      "grad_norm": 0.4229053556919098,
      "learning_rate": 1.17776331532051e-05,
      "loss": 0.06500959396362305,
      "step": 4430
    },
    {
      "epoch": 0.5990367553865653,
      "grad_norm": 0.416337788105011,
      "learning_rate": 1.1770948859531397e-05,
      "loss": 0.07033443450927734,
      "step": 4431
    },
    {
      "epoch": 0.5991719476130123,
      "grad_norm": 0.40159526467323303,
      "learning_rate": 1.1764265238163369e-05,
      "loss": 0.07051944732666016,
      "step": 4432
    },
    {
      "epoch": 0.5993071398394593,
      "grad_norm": 0.23113057017326355,
      "learning_rate": 1.1757582290492568e-05,
      "loss": 0.0779874324798584,
      "step": 4433
    },
    {
      "epoch": 0.5994423320659062,
      "grad_norm": 0.41072410345077515,
      "learning_rate": 1.1750900017910425e-05,
      "loss": 0.0892038345336914,
      "step": 4434
    },
    {
      "epoch": 0.5995775242923532,
      "grad_norm": 0.2522313892841339,
      "learning_rate": 1.1744218421808221e-05,
      "loss": 0.05801105499267578,
      "step": 4435
    },
    {
      "epoch": 0.5997127165188002,
      "grad_norm": 0.15272016823291779,
      "learning_rate": 1.1737537503577112e-05,
      "loss": 0.04536318778991699,
      "step": 4436
    },
    {
      "epoch": 0.5998479087452472,
      "grad_norm": 0.3802728056907654,
      "learning_rate": 1.1730857264608086e-05,
      "loss": 0.06792402267456055,
      "step": 4437
    },
    {
      "epoch": 0.5999831009716942,
      "grad_norm": 0.33153486251831055,
      "learning_rate": 1.1724177706292013e-05,
      "loss": 0.067108154296875,
      "step": 4438
    },
    {
      "epoch": 0.6001182931981411,
      "grad_norm": 0.2935335338115692,
      "learning_rate": 1.1717498830019607e-05,
      "loss": 0.050164103507995605,
      "step": 4439
    },
    {
      "epoch": 0.6002534854245881,
      "grad_norm": 0.2432565540075302,
      "learning_rate": 1.1710820637181449e-05,
      "loss": 0.06165719032287598,
      "step": 4440
    },
    {
      "epoch": 0.6003886776510351,
      "grad_norm": 0.22496061027050018,
      "learning_rate": 1.170414312916798e-05,
      "loss": 0.07255363464355469,
      "step": 4441
    },
    {
      "epoch": 0.6005238698774821,
      "grad_norm": 0.295534610748291,
      "learning_rate": 1.1697466307369484e-05,
      "loss": 0.0618743896484375,
      "step": 4442
    },
    {
      "epoch": 0.600659062103929,
      "grad_norm": 0.20630301535129547,
      "learning_rate": 1.1690790173176116e-05,
      "loss": 0.05825996398925781,
      "step": 4443
    },
    {
      "epoch": 0.600794254330376,
      "grad_norm": 0.13719916343688965,
      "learning_rate": 1.1684114727977876e-05,
      "loss": 0.035945743322372437,
      "step": 4444
    },
    {
      "epoch": 0.600929446556823,
      "grad_norm": 0.34321242570877075,
      "learning_rate": 1.167743997316464e-05,
      "loss": 0.07386445999145508,
      "step": 4445
    },
    {
      "epoch": 0.60106463878327,
      "grad_norm": 0.1980847269296646,
      "learning_rate": 1.1670765910126112e-05,
      "loss": 0.05725669860839844,
      "step": 4446
    },
    {
      "epoch": 0.601199831009717,
      "grad_norm": 0.19179165363311768,
      "learning_rate": 1.1664092540251877e-05,
      "loss": 0.03864544630050659,
      "step": 4447
    },
    {
      "epoch": 0.601335023236164,
      "grad_norm": 0.36496564745903015,
      "learning_rate": 1.1657419864931361e-05,
      "loss": 0.06734848022460938,
      "step": 4448
    },
    {
      "epoch": 0.6014702154626109,
      "grad_norm": 0.42936888337135315,
      "learning_rate": 1.165074788555386e-05,
      "loss": 0.0985708236694336,
      "step": 4449
    },
    {
      "epoch": 0.6016054076890579,
      "grad_norm": 0.36141806840896606,
      "learning_rate": 1.1644076603508514e-05,
      "loss": 0.07548117637634277,
      "step": 4450
    },
    {
      "epoch": 0.6017405999155049,
      "grad_norm": 0.16163617372512817,
      "learning_rate": 1.1637406020184305e-05,
      "loss": 0.02950739860534668,
      "step": 4451
    },
    {
      "epoch": 0.6018757921419519,
      "grad_norm": 0.32977062463760376,
      "learning_rate": 1.1630736136970097e-05,
      "loss": 0.06973505020141602,
      "step": 4452
    },
    {
      "epoch": 0.6020109843683988,
      "grad_norm": 0.23191730678081512,
      "learning_rate": 1.162406695525459e-05,
      "loss": 0.03450584411621094,
      "step": 4453
    },
    {
      "epoch": 0.6021461765948458,
      "grad_norm": 0.26966822147369385,
      "learning_rate": 1.161739847642635e-05,
      "loss": 0.044283151626586914,
      "step": 4454
    },
    {
      "epoch": 0.6022813688212928,
      "grad_norm": 0.2920010983943939,
      "learning_rate": 1.1610730701873788e-05,
      "loss": 0.04284381866455078,
      "step": 4455
    },
    {
      "epoch": 0.6024165610477398,
      "grad_norm": 0.22126412391662598,
      "learning_rate": 1.1604063632985163e-05,
      "loss": 0.06336522102355957,
      "step": 4456
    },
    {
      "epoch": 0.6025517532741868,
      "grad_norm": 0.2864057421684265,
      "learning_rate": 1.1597397271148598e-05,
      "loss": 0.06883811950683594,
      "step": 4457
    },
    {
      "epoch": 0.6026869455006337,
      "grad_norm": 0.5990896224975586,
      "learning_rate": 1.1590731617752067e-05,
      "loss": 0.1023101806640625,
      "step": 4458
    },
    {
      "epoch": 0.6028221377270807,
      "grad_norm": 0.2607079744338989,
      "learning_rate": 1.1584066674183398e-05,
      "loss": 0.07424402236938477,
      "step": 4459
    },
    {
      "epoch": 0.6029573299535277,
      "grad_norm": 0.13554170727729797,
      "learning_rate": 1.1577402441830262e-05,
      "loss": 0.03943014144897461,
      "step": 4460
    },
    {
      "epoch": 0.6030925221799747,
      "grad_norm": 0.4893263280391693,
      "learning_rate": 1.1570738922080185e-05,
      "loss": 0.08084595203399658,
      "step": 4461
    },
    {
      "epoch": 0.6032277144064216,
      "grad_norm": 0.3303582966327667,
      "learning_rate": 1.1564076116320552e-05,
      "loss": 0.05921870470046997,
      "step": 4462
    },
    {
      "epoch": 0.6033629066328686,
      "grad_norm": 0.26532837748527527,
      "learning_rate": 1.1557414025938592e-05,
      "loss": 0.05894160270690918,
      "step": 4463
    },
    {
      "epoch": 0.6034980988593156,
      "grad_norm": 0.2555946409702301,
      "learning_rate": 1.15507526523214e-05,
      "loss": 0.06483268737792969,
      "step": 4464
    },
    {
      "epoch": 0.6036332910857626,
      "grad_norm": 0.1355355978012085,
      "learning_rate": 1.1544091996855895e-05,
      "loss": 0.042125701904296875,
      "step": 4465
    },
    {
      "epoch": 0.6037684833122096,
      "grad_norm": 1.1640866994857788,
      "learning_rate": 1.153743206092886e-05,
      "loss": 0.07763147354125977,
      "step": 4466
    },
    {
      "epoch": 0.6039036755386565,
      "grad_norm": 0.3471051752567291,
      "learning_rate": 1.1530772845926936e-05,
      "loss": 0.06594324111938477,
      "step": 4467
    },
    {
      "epoch": 0.6040388677651035,
      "grad_norm": 0.14810936152935028,
      "learning_rate": 1.1524114353236614e-05,
      "loss": 0.038870811462402344,
      "step": 4468
    },
    {
      "epoch": 0.6041740599915505,
      "grad_norm": 0.315115749835968,
      "learning_rate": 1.151745658424421e-05,
      "loss": 0.06246590614318848,
      "step": 4469
    },
    {
      "epoch": 0.6043092522179975,
      "grad_norm": 0.2308308631181717,
      "learning_rate": 1.151079954033592e-05,
      "loss": 0.06431865692138672,
      "step": 4470
    },
    {
      "epoch": 0.6044444444444445,
      "grad_norm": 0.3533167541027069,
      "learning_rate": 1.150414322289777e-05,
      "loss": 0.06346344947814941,
      "step": 4471
    },
    {
      "epoch": 0.6045796366708914,
      "grad_norm": 0.26321864128112793,
      "learning_rate": 1.1497487633315643e-05,
      "loss": 0.06212043762207031,
      "step": 4472
    },
    {
      "epoch": 0.6047148288973384,
      "grad_norm": 0.23988419771194458,
      "learning_rate": 1.1490832772975275e-05,
      "loss": 0.05938410758972168,
      "step": 4473
    },
    {
      "epoch": 0.6048500211237854,
      "grad_norm": 1.596050500869751,
      "learning_rate": 1.148417864326223e-05,
      "loss": 0.06836318969726562,
      "step": 4474
    },
    {
      "epoch": 0.6049852133502324,
      "grad_norm": 0.18255159258842468,
      "learning_rate": 1.1477525245561944e-05,
      "loss": 0.041760921478271484,
      "step": 4475
    },
    {
      "epoch": 0.6051204055766793,
      "grad_norm": 0.35769742727279663,
      "learning_rate": 1.1470872581259684e-05,
      "loss": 0.06638026237487793,
      "step": 4476
    },
    {
      "epoch": 0.6052555978031263,
      "grad_norm": 0.2494780272245407,
      "learning_rate": 1.146422065174057e-05,
      "loss": 0.05401611328125,
      "step": 4477
    },
    {
      "epoch": 0.6053907900295733,
      "grad_norm": 0.25412124395370483,
      "learning_rate": 1.1457569458389578e-05,
      "loss": 0.055255889892578125,
      "step": 4478
    },
    {
      "epoch": 0.6055259822560203,
      "grad_norm": 0.3694517910480499,
      "learning_rate": 1.145091900259151e-05,
      "loss": 0.07106637954711914,
      "step": 4479
    },
    {
      "epoch": 0.6056611744824673,
      "grad_norm": 0.3325493633747101,
      "learning_rate": 1.1444269285731032e-05,
      "loss": 0.0693817138671875,
      "step": 4480
    },
    {
      "epoch": 0.6057963667089142,
      "grad_norm": 0.34629350900650024,
      "learning_rate": 1.1437620309192652e-05,
      "loss": 0.07529926300048828,
      "step": 4481
    },
    {
      "epoch": 0.6059315589353612,
      "grad_norm": 0.35613033175468445,
      "learning_rate": 1.1430972074360722e-05,
      "loss": 0.09632658958435059,
      "step": 4482
    },
    {
      "epoch": 0.6060667511618082,
      "grad_norm": 0.49247199296951294,
      "learning_rate": 1.1424324582619435e-05,
      "loss": 0.08773231506347656,
      "step": 4483
    },
    {
      "epoch": 0.6062019433882552,
      "grad_norm": 0.17535701394081116,
      "learning_rate": 1.1417677835352837e-05,
      "loss": 0.047327399253845215,
      "step": 4484
    },
    {
      "epoch": 0.6063371356147021,
      "grad_norm": 0.3107914328575134,
      "learning_rate": 1.1411031833944816e-05,
      "loss": 0.08040523529052734,
      "step": 4485
    },
    {
      "epoch": 0.6064723278411491,
      "grad_norm": 0.3590696156024933,
      "learning_rate": 1.1404386579779111e-05,
      "loss": 0.07081890106201172,
      "step": 4486
    },
    {
      "epoch": 0.6066075200675961,
      "grad_norm": 0.37001606822013855,
      "learning_rate": 1.1397742074239296e-05,
      "loss": 0.11190128326416016,
      "step": 4487
    },
    {
      "epoch": 0.6067427122940431,
      "grad_norm": 0.37231770157814026,
      "learning_rate": 1.1391098318708785e-05,
      "loss": 0.06675958633422852,
      "step": 4488
    },
    {
      "epoch": 0.6068779045204901,
      "grad_norm": 0.7272331714630127,
      "learning_rate": 1.1384455314570848e-05,
      "loss": 0.10275650024414062,
      "step": 4489
    },
    {
      "epoch": 0.607013096746937,
      "grad_norm": 0.3578093349933624,
      "learning_rate": 1.1377813063208596e-05,
      "loss": 0.091766357421875,
      "step": 4490
    },
    {
      "epoch": 0.607148288973384,
      "grad_norm": 0.40044039487838745,
      "learning_rate": 1.1371171566004986e-05,
      "loss": 0.08358001708984375,
      "step": 4491
    },
    {
      "epoch": 0.607283481199831,
      "grad_norm": 0.2737836241722107,
      "learning_rate": 1.1364530824342806e-05,
      "loss": 0.06424522399902344,
      "step": 4492
    },
    {
      "epoch": 0.607418673426278,
      "grad_norm": 0.17177489399909973,
      "learning_rate": 1.1357890839604688e-05,
      "loss": 0.03950929641723633,
      "step": 4493
    },
    {
      "epoch": 0.607553865652725,
      "grad_norm": 0.2997469902038574,
      "learning_rate": 1.1351251613173122e-05,
      "loss": 0.06270122528076172,
      "step": 4494
    },
    {
      "epoch": 0.6076890578791719,
      "grad_norm": 0.2717417776584625,
      "learning_rate": 1.1344613146430428e-05,
      "loss": 0.0774831771850586,
      "step": 4495
    },
    {
      "epoch": 0.6078242501056189,
      "grad_norm": 0.3255259692668915,
      "learning_rate": 1.1337975440758775e-05,
      "loss": 0.07410800457000732,
      "step": 4496
    },
    {
      "epoch": 0.6079594423320659,
      "grad_norm": 0.20851555466651917,
      "learning_rate": 1.133133849754016e-05,
      "loss": 0.047528743743896484,
      "step": 4497
    },
    {
      "epoch": 0.6080946345585129,
      "grad_norm": 0.4493762254714966,
      "learning_rate": 1.1324702318156431e-05,
      "loss": 0.05563664436340332,
      "step": 4498
    },
    {
      "epoch": 0.6082298267849598,
      "grad_norm": 0.4611785113811493,
      "learning_rate": 1.1318066903989279e-05,
      "loss": 0.05272743105888367,
      "step": 4499
    },
    {
      "epoch": 0.6083650190114068,
      "grad_norm": 0.30822551250457764,
      "learning_rate": 1.1311432256420232e-05,
      "loss": 0.05843162536621094,
      "step": 4500
    },
    {
      "epoch": 0.6085002112378538,
      "grad_norm": 0.6020203828811646,
      "learning_rate": 1.1304798376830664e-05,
      "loss": 0.13279247283935547,
      "step": 4501
    },
    {
      "epoch": 0.6086354034643008,
      "grad_norm": 0.41823306679725647,
      "learning_rate": 1.1298165266601778e-05,
      "loss": 0.0845341682434082,
      "step": 4502
    },
    {
      "epoch": 0.6087705956907478,
      "grad_norm": 0.33516818284988403,
      "learning_rate": 1.129153292711462e-05,
      "loss": 0.06222248077392578,
      "step": 4503
    },
    {
      "epoch": 0.6089057879171947,
      "grad_norm": 0.32954806089401245,
      "learning_rate": 1.1284901359750082e-05,
      "loss": 0.07540035247802734,
      "step": 4504
    },
    {
      "epoch": 0.6090409801436417,
      "grad_norm": 0.2039014995098114,
      "learning_rate": 1.1278270565888897e-05,
      "loss": 0.05766725540161133,
      "step": 4505
    },
    {
      "epoch": 0.6091761723700887,
      "grad_norm": 0.24326585233211517,
      "learning_rate": 1.1271640546911624e-05,
      "loss": 0.06625604629516602,
      "step": 4506
    },
    {
      "epoch": 0.6093113645965357,
      "grad_norm": 0.27929040789604187,
      "learning_rate": 1.1265011304198672e-05,
      "loss": 0.046045541763305664,
      "step": 4507
    },
    {
      "epoch": 0.6094465568229827,
      "grad_norm": 0.3388141691684723,
      "learning_rate": 1.1258382839130282e-05,
      "loss": 0.0822591781616211,
      "step": 4508
    },
    {
      "epoch": 0.6095817490494296,
      "grad_norm": 0.2273472398519516,
      "learning_rate": 1.1251755153086536e-05,
      "loss": 0.05302000045776367,
      "step": 4509
    },
    {
      "epoch": 0.6097169412758766,
      "grad_norm": 0.5747863054275513,
      "learning_rate": 1.1245128247447362e-05,
      "loss": 0.05986785888671875,
      "step": 4510
    },
    {
      "epoch": 0.6098521335023236,
      "grad_norm": 0.17942984402179718,
      "learning_rate": 1.1238502123592507e-05,
      "loss": 0.056525230407714844,
      "step": 4511
    },
    {
      "epoch": 0.6099873257287706,
      "grad_norm": 0.5319517254829407,
      "learning_rate": 1.1231876782901568e-05,
      "loss": 0.06388139724731445,
      "step": 4512
    },
    {
      "epoch": 0.6101225179552175,
      "grad_norm": 0.17188724875450134,
      "learning_rate": 1.1225252226753975e-05,
      "loss": 0.045076847076416016,
      "step": 4513
    },
    {
      "epoch": 0.6102577101816645,
      "grad_norm": 0.8894361853599548,
      "learning_rate": 1.1218628456529005e-05,
      "loss": 0.10037994384765625,
      "step": 4514
    },
    {
      "epoch": 0.6103929024081115,
      "grad_norm": 0.26242193579673767,
      "learning_rate": 1.1212005473605746e-05,
      "loss": 0.0639791488647461,
      "step": 4515
    },
    {
      "epoch": 0.6105280946345585,
      "grad_norm": 0.24523401260375977,
      "learning_rate": 1.120538327936315e-05,
      "loss": 0.08405637741088867,
      "step": 4516
    },
    {
      "epoch": 0.6106632868610055,
      "grad_norm": 0.3713717460632324,
      "learning_rate": 1.1198761875179993e-05,
      "loss": 0.0932474136352539,
      "step": 4517
    },
    {
      "epoch": 0.6107984790874524,
      "grad_norm": 0.42905890941619873,
      "learning_rate": 1.1192141262434883e-05,
      "loss": 0.06979227066040039,
      "step": 4518
    },
    {
      "epoch": 0.6109336713138994,
      "grad_norm": 0.29989415407180786,
      "learning_rate": 1.1185521442506272e-05,
      "loss": 0.07082176208496094,
      "step": 4519
    },
    {
      "epoch": 0.6110688635403464,
      "grad_norm": 0.5955882668495178,
      "learning_rate": 1.1178902416772432e-05,
      "loss": 0.09388101100921631,
      "step": 4520
    },
    {
      "epoch": 0.6112040557667934,
      "grad_norm": 0.3061008155345917,
      "learning_rate": 1.1172284186611485e-05,
      "loss": 0.07376432418823242,
      "step": 4521
    },
    {
      "epoch": 0.6113392479932404,
      "grad_norm": 0.2999645173549652,
      "learning_rate": 1.1165666753401384e-05,
      "loss": 0.06786715984344482,
      "step": 4522
    },
    {
      "epoch": 0.6114744402196873,
      "grad_norm": 0.36514514684677124,
      "learning_rate": 1.1159050118519914e-05,
      "loss": 0.06745219230651855,
      "step": 4523
    },
    {
      "epoch": 0.6116096324461343,
      "grad_norm": 0.19906677305698395,
      "learning_rate": 1.1152434283344696e-05,
      "loss": 0.0386357307434082,
      "step": 4524
    },
    {
      "epoch": 0.6117448246725813,
      "grad_norm": 0.4493810832500458,
      "learning_rate": 1.114581924925317e-05,
      "loss": 0.06978821754455566,
      "step": 4525
    },
    {
      "epoch": 0.6118800168990283,
      "grad_norm": 0.2112336903810501,
      "learning_rate": 1.113920501762263e-05,
      "loss": 0.051206111907958984,
      "step": 4526
    },
    {
      "epoch": 0.6120152091254752,
      "grad_norm": 0.48546716570854187,
      "learning_rate": 1.1132591589830193e-05,
      "loss": 0.09076404571533203,
      "step": 4527
    },
    {
      "epoch": 0.6121504013519222,
      "grad_norm": 0.42367085814476013,
      "learning_rate": 1.1125978967252818e-05,
      "loss": 0.07500863075256348,
      "step": 4528
    },
    {
      "epoch": 0.6122855935783692,
      "grad_norm": 0.3923182189464569,
      "learning_rate": 1.1119367151267278e-05,
      "loss": 0.059465885162353516,
      "step": 4529
    },
    {
      "epoch": 0.6124207858048162,
      "grad_norm": 0.25272053480148315,
      "learning_rate": 1.1112756143250186e-05,
      "loss": 0.05335044860839844,
      "step": 4530
    },
    {
      "epoch": 0.6125559780312632,
      "grad_norm": 0.2617141306400299,
      "learning_rate": 1.1106145944577995e-05,
      "loss": 0.06698143482208252,
      "step": 4531
    },
    {
      "epoch": 0.6126911702577101,
      "grad_norm": 0.24244795739650726,
      "learning_rate": 1.1099536556626984e-05,
      "loss": 0.062398672103881836,
      "step": 4532
    },
    {
      "epoch": 0.6128263624841571,
      "grad_norm": 0.3359515964984894,
      "learning_rate": 1.1092927980773269e-05,
      "loss": 0.06544971466064453,
      "step": 4533
    },
    {
      "epoch": 0.6129615547106041,
      "grad_norm": 0.34295231103897095,
      "learning_rate": 1.1086320218392777e-05,
      "loss": 0.0650484561920166,
      "step": 4534
    },
    {
      "epoch": 0.6130967469370511,
      "grad_norm": 0.5967460870742798,
      "learning_rate": 1.1079713270861286e-05,
      "loss": 0.07013285160064697,
      "step": 4535
    },
    {
      "epoch": 0.613231939163498,
      "grad_norm": 0.27164527773857117,
      "learning_rate": 1.1073107139554395e-05,
      "loss": 0.056929588317871094,
      "step": 4536
    },
    {
      "epoch": 0.613367131389945,
      "grad_norm": 0.2666477859020233,
      "learning_rate": 1.1066501825847545e-05,
      "loss": 0.07135677337646484,
      "step": 4537
    },
    {
      "epoch": 0.613502323616392,
      "grad_norm": 0.8466290235519409,
      "learning_rate": 1.1059897331115985e-05,
      "loss": 0.07445192337036133,
      "step": 4538
    },
    {
      "epoch": 0.613637515842839,
      "grad_norm": 0.5046213269233704,
      "learning_rate": 1.1053293656734816e-05,
      "loss": 0.06831479072570801,
      "step": 4539
    },
    {
      "epoch": 0.613772708069286,
      "grad_norm": 0.2886885702610016,
      "learning_rate": 1.1046690804078949e-05,
      "loss": 0.06088685989379883,
      "step": 4540
    },
    {
      "epoch": 0.613907900295733,
      "grad_norm": 0.2978242337703705,
      "learning_rate": 1.1040088774523139e-05,
      "loss": 0.05490458011627197,
      "step": 4541
    },
    {
      "epoch": 0.6140430925221799,
      "grad_norm": 0.14407029747962952,
      "learning_rate": 1.1033487569441971e-05,
      "loss": 0.02333986759185791,
      "step": 4542
    },
    {
      "epoch": 0.6141782847486269,
      "grad_norm": 0.3460027575492859,
      "learning_rate": 1.1026887190209834e-05,
      "loss": 0.09065914154052734,
      "step": 4543
    },
    {
      "epoch": 0.6143134769750739,
      "grad_norm": 0.17839789390563965,
      "learning_rate": 1.1020287638200977e-05,
      "loss": 0.03936338424682617,
      "step": 4544
    },
    {
      "epoch": 0.6144486692015209,
      "grad_norm": 0.3785291612148285,
      "learning_rate": 1.1013688914789452e-05,
      "loss": 0.07198762893676758,
      "step": 4545
    },
    {
      "epoch": 0.6145838614279678,
      "grad_norm": 0.4057964086532593,
      "learning_rate": 1.100709102134915e-05,
      "loss": 0.08894228935241699,
      "step": 4546
    },
    {
      "epoch": 0.6147190536544149,
      "grad_norm": 0.3648238778114319,
      "learning_rate": 1.10004939592538e-05,
      "loss": 0.07256269454956055,
      "step": 4547
    },
    {
      "epoch": 0.6148542458808619,
      "grad_norm": 0.19871188700199127,
      "learning_rate": 1.0993897729876927e-05,
      "loss": 0.0429837703704834,
      "step": 4548
    },
    {
      "epoch": 0.6149894381073089,
      "grad_norm": 0.36041560769081116,
      "learning_rate": 1.0987302334591915e-05,
      "loss": 0.08607816696166992,
      "step": 4549
    },
    {
      "epoch": 0.6151246303337559,
      "grad_norm": 0.4003594219684601,
      "learning_rate": 1.098070777477195e-05,
      "loss": 0.05415606498718262,
      "step": 4550
    },
    {
      "epoch": 0.6152598225602028,
      "grad_norm": 0.2990688979625702,
      "learning_rate": 1.0974114051790067e-05,
      "loss": 0.0651865005493164,
      "step": 4551
    },
    {
      "epoch": 0.6153950147866498,
      "grad_norm": 0.17521131038665771,
      "learning_rate": 1.09675211670191e-05,
      "loss": 0.03891599178314209,
      "step": 4552
    },
    {
      "epoch": 0.6155302070130968,
      "grad_norm": 0.3220004439353943,
      "learning_rate": 1.0960929121831732e-05,
      "loss": 0.05576133728027344,
      "step": 4553
    },
    {
      "epoch": 0.6156653992395438,
      "grad_norm": 0.28507575392723083,
      "learning_rate": 1.095433791760046e-05,
      "loss": 0.05747842788696289,
      "step": 4554
    },
    {
      "epoch": 0.6158005914659908,
      "grad_norm": 0.3530092239379883,
      "learning_rate": 1.0947747555697609e-05,
      "loss": 0.0827333927154541,
      "step": 4555
    },
    {
      "epoch": 0.6159357836924377,
      "grad_norm": 0.17005878686904907,
      "learning_rate": 1.0941158037495328e-05,
      "loss": 0.04477977752685547,
      "step": 4556
    },
    {
      "epoch": 0.6160709759188847,
      "grad_norm": 0.1905655860900879,
      "learning_rate": 1.0934569364365583e-05,
      "loss": 0.056057095527648926,
      "step": 4557
    },
    {
      "epoch": 0.6162061681453317,
      "grad_norm": 0.44855692982673645,
      "learning_rate": 1.0927981537680176e-05,
      "loss": 0.08228158950805664,
      "step": 4558
    },
    {
      "epoch": 0.6163413603717787,
      "grad_norm": 0.43794751167297363,
      "learning_rate": 1.0921394558810726e-05,
      "loss": 0.07080614566802979,
      "step": 4559
    },
    {
      "epoch": 0.6164765525982256,
      "grad_norm": 0.484383761882782,
      "learning_rate": 1.0914808429128688e-05,
      "loss": 0.06489837169647217,
      "step": 4560
    },
    {
      "epoch": 0.6166117448246726,
      "grad_norm": 0.3000618517398834,
      "learning_rate": 1.0908223150005315e-05,
      "loss": 0.07498455047607422,
      "step": 4561
    },
    {
      "epoch": 0.6167469370511196,
      "grad_norm": 0.24319665133953094,
      "learning_rate": 1.09016387228117e-05,
      "loss": 0.04724609851837158,
      "step": 4562
    },
    {
      "epoch": 0.6168821292775666,
      "grad_norm": 0.20916372537612915,
      "learning_rate": 1.0895055148918758e-05,
      "loss": 0.03911542892456055,
      "step": 4563
    },
    {
      "epoch": 0.6170173215040136,
      "grad_norm": 0.36333584785461426,
      "learning_rate": 1.0888472429697223e-05,
      "loss": 0.051422834396362305,
      "step": 4564
    },
    {
      "epoch": 0.6171525137304605,
      "grad_norm": 0.14120206236839294,
      "learning_rate": 1.088189056651766e-05,
      "loss": 0.05162477493286133,
      "step": 4565
    },
    {
      "epoch": 0.6172877059569075,
      "grad_norm": 0.2763865292072296,
      "learning_rate": 1.0875309560750438e-05,
      "loss": 0.06842708587646484,
      "step": 4566
    },
    {
      "epoch": 0.6174228981833545,
      "grad_norm": 0.34497779607772827,
      "learning_rate": 1.086872941376576e-05,
      "loss": 0.07580947875976562,
      "step": 4567
    },
    {
      "epoch": 0.6175580904098015,
      "grad_norm": 0.3612481951713562,
      "learning_rate": 1.0862150126933648e-05,
      "loss": 0.06434965133666992,
      "step": 4568
    },
    {
      "epoch": 0.6176932826362485,
      "grad_norm": 0.4365478456020355,
      "learning_rate": 1.0855571701623942e-05,
      "loss": 0.06622171401977539,
      "step": 4569
    },
    {
      "epoch": 0.6178284748626954,
      "grad_norm": 0.28336986899375916,
      "learning_rate": 1.0848994139206317e-05,
      "loss": 0.054013967514038086,
      "step": 4570
    },
    {
      "epoch": 0.6179636670891424,
      "grad_norm": 0.2518436908721924,
      "learning_rate": 1.0842417441050247e-05,
      "loss": 0.06433439254760742,
      "step": 4571
    },
    {
      "epoch": 0.6180988593155894,
      "grad_norm": 0.33681032061576843,
      "learning_rate": 1.0835841608525031e-05,
      "loss": 0.06585288047790527,
      "step": 4572
    },
    {
      "epoch": 0.6182340515420364,
      "grad_norm": 0.31533190608024597,
      "learning_rate": 1.08292666429998e-05,
      "loss": 0.06459236145019531,
      "step": 4573
    },
    {
      "epoch": 0.6183692437684833,
      "grad_norm": 0.2309371531009674,
      "learning_rate": 1.08226925458435e-05,
      "loss": 0.05120372772216797,
      "step": 4574
    },
    {
      "epoch": 0.6185044359949303,
      "grad_norm": 0.21744850277900696,
      "learning_rate": 1.0816119318424882e-05,
      "loss": 0.06373906135559082,
      "step": 4575
    },
    {
      "epoch": 0.6186396282213773,
      "grad_norm": 0.2714594006538391,
      "learning_rate": 1.0809546962112535e-05,
      "loss": 0.07249265909194946,
      "step": 4576
    },
    {
      "epoch": 0.6187748204478243,
      "grad_norm": 0.2508978545665741,
      "learning_rate": 1.0802975478274856e-05,
      "loss": 0.058229684829711914,
      "step": 4577
    },
    {
      "epoch": 0.6189100126742713,
      "grad_norm": 0.352032333612442,
      "learning_rate": 1.0796404868280062e-05,
      "loss": 0.05313920974731445,
      "step": 4578
    },
    {
      "epoch": 0.6190452049007182,
      "grad_norm": 0.24912215769290924,
      "learning_rate": 1.07898351334962e-05,
      "loss": 0.05867719650268555,
      "step": 4579
    },
    {
      "epoch": 0.6191803971271652,
      "grad_norm": 0.2457178831100464,
      "learning_rate": 1.0783266275291103e-05,
      "loss": 0.06145477294921875,
      "step": 4580
    },
    {
      "epoch": 0.6193155893536122,
      "grad_norm": 0.22022317349910736,
      "learning_rate": 1.077669829503246e-05,
      "loss": 0.054735422134399414,
      "step": 4581
    },
    {
      "epoch": 0.6194507815800592,
      "grad_norm": 0.30525389313697815,
      "learning_rate": 1.077013119408775e-05,
      "loss": 0.07169342041015625,
      "step": 4582
    },
    {
      "epoch": 0.6195859738065062,
      "grad_norm": 0.212591752409935,
      "learning_rate": 1.0763564973824289e-05,
      "loss": 0.05258166790008545,
      "step": 4583
    },
    {
      "epoch": 0.6197211660329531,
      "grad_norm": 0.3430812358856201,
      "learning_rate": 1.0756999635609185e-05,
      "loss": 0.05687212944030762,
      "step": 4584
    },
    {
      "epoch": 0.6198563582594001,
      "grad_norm": 0.27148669958114624,
      "learning_rate": 1.0750435180809381e-05,
      "loss": 0.0537867546081543,
      "step": 4585
    },
    {
      "epoch": 0.6199915504858471,
      "grad_norm": 0.2941371202468872,
      "learning_rate": 1.074387161079164e-05,
      "loss": 0.05854785442352295,
      "step": 4586
    },
    {
      "epoch": 0.6201267427122941,
      "grad_norm": 0.16351230442523956,
      "learning_rate": 1.0737308926922521e-05,
      "loss": 0.03908205032348633,
      "step": 4587
    },
    {
      "epoch": 0.620261934938741,
      "grad_norm": 0.13048654794692993,
      "learning_rate": 1.0730747130568424e-05,
      "loss": 0.034320831298828125,
      "step": 4588
    },
    {
      "epoch": 0.620397127165188,
      "grad_norm": 0.29218488931655884,
      "learning_rate": 1.0724186223095532e-05,
      "loss": 0.06438088417053223,
      "step": 4589
    },
    {
      "epoch": 0.620532319391635,
      "grad_norm": 0.17921486496925354,
      "learning_rate": 1.071762620586987e-05,
      "loss": 0.04278111457824707,
      "step": 4590
    },
    {
      "epoch": 0.620667511618082,
      "grad_norm": 0.3059333860874176,
      "learning_rate": 1.0711067080257273e-05,
      "loss": 0.05582815408706665,
      "step": 4591
    },
    {
      "epoch": 0.620802703844529,
      "grad_norm": 0.3186451494693756,
      "learning_rate": 1.0704508847623374e-05,
      "loss": 0.07403016090393066,
      "step": 4592
    },
    {
      "epoch": 0.6209378960709759,
      "grad_norm": 0.30739936232566833,
      "learning_rate": 1.069795150933365e-05,
      "loss": 0.05772209167480469,
      "step": 4593
    },
    {
      "epoch": 0.6210730882974229,
      "grad_norm": 0.21873977780342102,
      "learning_rate": 1.0691395066753357e-05,
      "loss": 0.05909430980682373,
      "step": 4594
    },
    {
      "epoch": 0.6212082805238699,
      "grad_norm": 0.3679015636444092,
      "learning_rate": 1.0684839521247584e-05,
      "loss": 0.05233001708984375,
      "step": 4595
    },
    {
      "epoch": 0.6213434727503169,
      "grad_norm": 0.3394359350204468,
      "learning_rate": 1.0678284874181234e-05,
      "loss": 0.08089494705200195,
      "step": 4596
    },
    {
      "epoch": 0.6214786649767639,
      "grad_norm": 0.22854389250278473,
      "learning_rate": 1.0671731126919028e-05,
      "loss": 0.05193519592285156,
      "step": 4597
    },
    {
      "epoch": 0.6216138572032108,
      "grad_norm": 0.20391437411308289,
      "learning_rate": 1.066517828082548e-05,
      "loss": 0.05152487754821777,
      "step": 4598
    },
    {
      "epoch": 0.6217490494296578,
      "grad_norm": 0.2641715705394745,
      "learning_rate": 1.0658626337264926e-05,
      "loss": 0.060117244720458984,
      "step": 4599
    },
    {
      "epoch": 0.6218842416561048,
      "grad_norm": 0.31620243191719055,
      "learning_rate": 1.0652075297601518e-05,
      "loss": 0.04932093620300293,
      "step": 4600
    },
    {
      "epoch": 0.6220194338825518,
      "grad_norm": 0.3243793547153473,
      "learning_rate": 1.0645525163199222e-05,
      "loss": 0.0713043212890625,
      "step": 4601
    },
    {
      "epoch": 0.6221546261089987,
      "grad_norm": 0.27476295828819275,
      "learning_rate": 1.063897593542181e-05,
      "loss": 0.06162452697753906,
      "step": 4602
    },
    {
      "epoch": 0.6222898183354457,
      "grad_norm": 0.39935553073883057,
      "learning_rate": 1.0632427615632864e-05,
      "loss": 0.06130671501159668,
      "step": 4603
    },
    {
      "epoch": 0.6224250105618927,
      "grad_norm": 0.22969891130924225,
      "learning_rate": 1.0625880205195776e-05,
      "loss": 0.04492330551147461,
      "step": 4604
    },
    {
      "epoch": 0.6225602027883397,
      "grad_norm": 0.29548484086990356,
      "learning_rate": 1.0619333705473754e-05,
      "loss": 0.06001782417297363,
      "step": 4605
    },
    {
      "epoch": 0.6226953950147867,
      "grad_norm": 0.631999135017395,
      "learning_rate": 1.0612788117829821e-05,
      "loss": 0.08805418014526367,
      "step": 4606
    },
    {
      "epoch": 0.6228305872412336,
      "grad_norm": 0.35514265298843384,
      "learning_rate": 1.0606243443626792e-05,
      "loss": 0.07119536399841309,
      "step": 4607
    },
    {
      "epoch": 0.6229657794676806,
      "grad_norm": 0.1862819939851761,
      "learning_rate": 1.0599699684227313e-05,
      "loss": 0.03358948230743408,
      "step": 4608
    },
    {
      "epoch": 0.6231009716941276,
      "grad_norm": 0.30135810375213623,
      "learning_rate": 1.0593156840993818e-05,
      "loss": 0.04882550239562988,
      "step": 4609
    },
    {
      "epoch": 0.6232361639205746,
      "grad_norm": 0.17869390547275543,
      "learning_rate": 1.0586614915288571e-05,
      "loss": 0.03380930423736572,
      "step": 4610
    },
    {
      "epoch": 0.6233713561470215,
      "grad_norm": 0.2399933636188507,
      "learning_rate": 1.0580073908473641e-05,
      "loss": 0.06104230880737305,
      "step": 4611
    },
    {
      "epoch": 0.6235065483734685,
      "grad_norm": 0.20708365738391876,
      "learning_rate": 1.0573533821910885e-05,
      "loss": 0.04581332206726074,
      "step": 4612
    },
    {
      "epoch": 0.6236417405999155,
      "grad_norm": 0.49541765451431274,
      "learning_rate": 1.0566994656961997e-05,
      "loss": 0.07208776473999023,
      "step": 4613
    },
    {
      "epoch": 0.6237769328263625,
      "grad_norm": 0.3167388141155243,
      "learning_rate": 1.0560456414988456e-05,
      "loss": 0.0721883773803711,
      "step": 4614
    },
    {
      "epoch": 0.6239121250528095,
      "grad_norm": 0.2761615812778473,
      "learning_rate": 1.0553919097351564e-05,
      "loss": 0.06961297988891602,
      "step": 4615
    },
    {
      "epoch": 0.6240473172792564,
      "grad_norm": 0.4039650559425354,
      "learning_rate": 1.0547382705412434e-05,
      "loss": 0.07571643590927124,
      "step": 4616
    },
    {
      "epoch": 0.6241825095057034,
      "grad_norm": 0.7439659833908081,
      "learning_rate": 1.054084724053196e-05,
      "loss": 0.13332080841064453,
      "step": 4617
    },
    {
      "epoch": 0.6243177017321504,
      "grad_norm": 0.26782482862472534,
      "learning_rate": 1.0534312704070875e-05,
      "loss": 0.06744956970214844,
      "step": 4618
    },
    {
      "epoch": 0.6244528939585974,
      "grad_norm": 0.32765668630599976,
      "learning_rate": 1.0527779097389695e-05,
      "loss": 0.07637345790863037,
      "step": 4619
    },
    {
      "epoch": 0.6245880861850444,
      "grad_norm": 0.29157114028930664,
      "learning_rate": 1.0521246421848762e-05,
      "loss": 0.08211612701416016,
      "step": 4620
    },
    {
      "epoch": 0.6247232784114913,
      "grad_norm": 0.7886411547660828,
      "learning_rate": 1.0514714678808202e-05,
      "loss": 0.10589981079101562,
      "step": 4621
    },
    {
      "epoch": 0.6248584706379383,
      "grad_norm": 0.22971338033676147,
      "learning_rate": 1.0508183869627962e-05,
      "loss": 0.0625,
      "step": 4622
    },
    {
      "epoch": 0.6249936628643853,
      "grad_norm": 0.2727302610874176,
      "learning_rate": 1.0501653995667798e-05,
      "loss": 0.07536149024963379,
      "step": 4623
    },
    {
      "epoch": 0.6251288550908323,
      "grad_norm": 0.42857256531715393,
      "learning_rate": 1.0495125058287258e-05,
      "loss": 0.09780502319335938,
      "step": 4624
    },
    {
      "epoch": 0.6252640473172792,
      "grad_norm": 0.7165300846099854,
      "learning_rate": 1.0488597058845708e-05,
      "loss": 0.08171486854553223,
      "step": 4625
    },
    {
      "epoch": 0.6253992395437262,
      "grad_norm": 0.41147565841674805,
      "learning_rate": 1.0482069998702304e-05,
      "loss": 0.08458304405212402,
      "step": 4626
    },
    {
      "epoch": 0.6255344317701732,
      "grad_norm": 0.2602044939994812,
      "learning_rate": 1.0475543879216017e-05,
      "loss": 0.05593538284301758,
      "step": 4627
    },
    {
      "epoch": 0.6256696239966202,
      "grad_norm": 0.3791307508945465,
      "learning_rate": 1.0469018701745626e-05,
      "loss": 0.08008337020874023,
      "step": 4628
    },
    {
      "epoch": 0.6258048162230672,
      "grad_norm": 0.2977467179298401,
      "learning_rate": 1.0462494467649704e-05,
      "loss": 0.07937431335449219,
      "step": 4629
    },
    {
      "epoch": 0.6259400084495141,
      "grad_norm": 0.39518147706985474,
      "learning_rate": 1.045597117828663e-05,
      "loss": 0.05633068084716797,
      "step": 4630
    },
    {
      "epoch": 0.6260752006759611,
      "grad_norm": 0.3023243844509125,
      "learning_rate": 1.0449448835014586e-05,
      "loss": 0.06179237365722656,
      "step": 4631
    },
    {
      "epoch": 0.6262103929024081,
      "grad_norm": 0.3019101023674011,
      "learning_rate": 1.044292743919156e-05,
      "loss": 0.049596309661865234,
      "step": 4632
    },
    {
      "epoch": 0.6263455851288551,
      "grad_norm": 0.43353164196014404,
      "learning_rate": 1.0436406992175343e-05,
      "loss": 0.06688785552978516,
      "step": 4633
    },
    {
      "epoch": 0.626480777355302,
      "grad_norm": 0.2690771222114563,
      "learning_rate": 1.0429887495323532e-05,
      "loss": 0.04748481512069702,
      "step": 4634
    },
    {
      "epoch": 0.626615969581749,
      "grad_norm": 0.7041105031967163,
      "learning_rate": 1.0423368949993512e-05,
      "loss": 0.07501733303070068,
      "step": 4635
    },
    {
      "epoch": 0.626751161808196,
      "grad_norm": 0.35685858130455017,
      "learning_rate": 1.041685135754248e-05,
      "loss": 0.08731919527053833,
      "step": 4636
    },
    {
      "epoch": 0.626886354034643,
      "grad_norm": 0.20517200231552124,
      "learning_rate": 1.0410334719327435e-05,
      "loss": 0.05757880210876465,
      "step": 4637
    },
    {
      "epoch": 0.62702154626109,
      "grad_norm": 0.2358854115009308,
      "learning_rate": 1.0403819036705177e-05,
      "loss": 0.06907224655151367,
      "step": 4638
    },
    {
      "epoch": 0.627156738487537,
      "grad_norm": 0.40448957681655884,
      "learning_rate": 1.0397304311032311e-05,
      "loss": 0.06854289770126343,
      "step": 4639
    },
    {
      "epoch": 0.6272919307139839,
      "grad_norm": 0.40503883361816406,
      "learning_rate": 1.039079054366523e-05,
      "loss": 0.07426023483276367,
      "step": 4640
    },
    {
      "epoch": 0.6274271229404309,
      "grad_norm": 0.49728554487228394,
      "learning_rate": 1.0384277735960133e-05,
      "loss": 0.07779121398925781,
      "step": 4641
    },
    {
      "epoch": 0.6275623151668779,
      "grad_norm": 0.5770511627197266,
      "learning_rate": 1.0377765889273025e-05,
      "loss": 0.06285107135772705,
      "step": 4642
    },
    {
      "epoch": 0.6276975073933249,
      "grad_norm": 0.2105005979537964,
      "learning_rate": 1.0371255004959715e-05,
      "loss": 0.062136173248291016,
      "step": 4643
    },
    {
      "epoch": 0.6278326996197718,
      "grad_norm": 0.29505887627601624,
      "learning_rate": 1.036474508437579e-05,
      "loss": 0.05975842475891113,
      "step": 4644
    },
    {
      "epoch": 0.6279678918462188,
      "grad_norm": 0.2482919991016388,
      "learning_rate": 1.035823612887666e-05,
      "loss": 0.05152767896652222,
      "step": 4645
    },
    {
      "epoch": 0.6281030840726658,
      "grad_norm": 0.24198034405708313,
      "learning_rate": 1.0351728139817517e-05,
      "loss": 0.037592172622680664,
      "step": 4646
    },
    {
      "epoch": 0.6282382762991128,
      "grad_norm": 0.26556435227394104,
      "learning_rate": 1.0345221118553362e-05,
      "loss": 0.0791773796081543,
      "step": 4647
    },
    {
      "epoch": 0.6283734685255598,
      "grad_norm": 0.22227388620376587,
      "learning_rate": 1.0338715066439002e-05,
      "loss": 0.06063961982727051,
      "step": 4648
    },
    {
      "epoch": 0.6285086607520067,
      "grad_norm": 0.4060885012149811,
      "learning_rate": 1.0332209984829013e-05,
      "loss": 0.08052146434783936,
      "step": 4649
    },
    {
      "epoch": 0.6286438529784537,
      "grad_norm": 0.2979709804058075,
      "learning_rate": 1.03257058750778e-05,
      "loss": 0.062424659729003906,
      "step": 4650
    },
    {
      "epoch": 0.6287790452049007,
      "grad_norm": 0.1807127147912979,
      "learning_rate": 1.0319202738539548e-05,
      "loss": 0.050330400466918945,
      "step": 4651
    },
    {
      "epoch": 0.6289142374313477,
      "grad_norm": 0.28656843304634094,
      "learning_rate": 1.0312700576568253e-05,
      "loss": 0.06861495971679688,
      "step": 4652
    },
    {
      "epoch": 0.6290494296577946,
      "grad_norm": 0.29121947288513184,
      "learning_rate": 1.0306199390517688e-05,
      "loss": 0.05763101577758789,
      "step": 4653
    },
    {
      "epoch": 0.6291846218842416,
      "grad_norm": 0.23022115230560303,
      "learning_rate": 1.0299699181741439e-05,
      "loss": 0.059467315673828125,
      "step": 4654
    },
    {
      "epoch": 0.6293198141106886,
      "grad_norm": 0.3273932635784149,
      "learning_rate": 1.0293199951592889e-05,
      "loss": 0.07540702819824219,
      "step": 4655
    },
    {
      "epoch": 0.6294550063371356,
      "grad_norm": 0.4194307029247284,
      "learning_rate": 1.0286701701425206e-05,
      "loss": 0.09010553359985352,
      "step": 4656
    },
    {
      "epoch": 0.6295901985635826,
      "grad_norm": 0.43677449226379395,
      "learning_rate": 1.0280204432591369e-05,
      "loss": 0.10966968536376953,
      "step": 4657
    },
    {
      "epoch": 0.6297253907900295,
      "grad_norm": 0.24091710150241852,
      "learning_rate": 1.0273708146444133e-05,
      "loss": 0.07066917419433594,
      "step": 4658
    },
    {
      "epoch": 0.6298605830164765,
      "grad_norm": 0.20182935893535614,
      "learning_rate": 1.0267212844336062e-05,
      "loss": 0.054204583168029785,
      "step": 4659
    },
    {
      "epoch": 0.6299957752429235,
      "grad_norm": 0.40133270621299744,
      "learning_rate": 1.026071852761952e-05,
      "loss": 0.06823158264160156,
      "step": 4660
    },
    {
      "epoch": 0.6301309674693705,
      "grad_norm": 0.35685959458351135,
      "learning_rate": 1.025422519764665e-05,
      "loss": 0.08153486251831055,
      "step": 4661
    },
    {
      "epoch": 0.6302661596958175,
      "grad_norm": 0.16551874577999115,
      "learning_rate": 1.024773285576941e-05,
      "loss": 0.05325794219970703,
      "step": 4662
    },
    {
      "epoch": 0.6304013519222644,
      "grad_norm": 0.18010757863521576,
      "learning_rate": 1.0241241503339524e-05,
      "loss": 0.044170379638671875,
      "step": 4663
    },
    {
      "epoch": 0.6305365441487114,
      "grad_norm": 0.29997074604034424,
      "learning_rate": 1.023475114170853e-05,
      "loss": 0.07453131675720215,
      "step": 4664
    },
    {
      "epoch": 0.6306717363751584,
      "grad_norm": 0.2977597415447235,
      "learning_rate": 1.0228261772227768e-05,
      "loss": 0.06371140480041504,
      "step": 4665
    },
    {
      "epoch": 0.6308069286016054,
      "grad_norm": 0.32815954089164734,
      "learning_rate": 1.0221773396248349e-05,
      "loss": 0.07728123664855957,
      "step": 4666
    },
    {
      "epoch": 0.6309421208280523,
      "grad_norm": 0.3637869358062744,
      "learning_rate": 1.021528601512119e-05,
      "loss": 0.06687641143798828,
      "step": 4667
    },
    {
      "epoch": 0.6310773130544993,
      "grad_norm": 0.2982170879840851,
      "learning_rate": 1.0208799630196994e-05,
      "loss": 0.07450008392333984,
      "step": 4668
    },
    {
      "epoch": 0.6312125052809463,
      "grad_norm": 0.2322484850883484,
      "learning_rate": 1.0202314242826264e-05,
      "loss": 0.07700347900390625,
      "step": 4669
    },
    {
      "epoch": 0.6313476975073933,
      "grad_norm": 0.2507709562778473,
      "learning_rate": 1.0195829854359299e-05,
      "loss": 0.06022167205810547,
      "step": 4670
    },
    {
      "epoch": 0.6314828897338403,
      "grad_norm": 0.25489166378974915,
      "learning_rate": 1.0189346466146175e-05,
      "loss": 0.06063413619995117,
      "step": 4671
    },
    {
      "epoch": 0.6316180819602872,
      "grad_norm": 0.2549836337566376,
      "learning_rate": 1.018286407953677e-05,
      "loss": 0.06438636779785156,
      "step": 4672
    },
    {
      "epoch": 0.6317532741867342,
      "grad_norm": 0.22776448726654053,
      "learning_rate": 1.017638269588075e-05,
      "loss": 0.05417680740356445,
      "step": 4673
    },
    {
      "epoch": 0.6318884664131812,
      "grad_norm": 0.2738092541694641,
      "learning_rate": 1.0169902316527575e-05,
      "loss": 0.04232466220855713,
      "step": 4674
    },
    {
      "epoch": 0.6320236586396282,
      "grad_norm": 0.3591974377632141,
      "learning_rate": 1.0163422942826502e-05,
      "loss": 0.07692217826843262,
      "step": 4675
    },
    {
      "epoch": 0.6321588508660752,
      "grad_norm": 0.2739556133747101,
      "learning_rate": 1.0156944576126555e-05,
      "loss": 0.06323790550231934,
      "step": 4676
    },
    {
      "epoch": 0.6322940430925221,
      "grad_norm": 0.32111856341362,
      "learning_rate": 1.0150467217776579e-05,
      "loss": 0.06371116638183594,
      "step": 4677
    },
    {
      "epoch": 0.6324292353189691,
      "grad_norm": 0.16997277736663818,
      "learning_rate": 1.0143990869125185e-05,
      "loss": 0.03658187389373779,
      "step": 4678
    },
    {
      "epoch": 0.6325644275454161,
      "grad_norm": 0.40965399146080017,
      "learning_rate": 1.013751553152079e-05,
      "loss": 0.07799696922302246,
      "step": 4679
    },
    {
      "epoch": 0.6326996197718631,
      "grad_norm": 0.488856703042984,
      "learning_rate": 1.0131041206311594e-05,
      "loss": 0.07051825523376465,
      "step": 4680
    },
    {
      "epoch": 0.63283481199831,
      "grad_norm": 0.3490300476551056,
      "learning_rate": 1.0124567894845578e-05,
      "loss": 0.06315314769744873,
      "step": 4681
    },
    {
      "epoch": 0.632970004224757,
      "grad_norm": 0.25645068287849426,
      "learning_rate": 1.0118095598470528e-05,
      "loss": 0.045945167541503906,
      "step": 4682
    },
    {
      "epoch": 0.6331051964512041,
      "grad_norm": 0.35110464692115784,
      "learning_rate": 1.0111624318534006e-05,
      "loss": 0.07792937755584717,
      "step": 4683
    },
    {
      "epoch": 0.6332403886776511,
      "grad_norm": 0.2600908875465393,
      "learning_rate": 1.0105154056383377e-05,
      "loss": 0.05178546905517578,
      "step": 4684
    },
    {
      "epoch": 0.6333755809040981,
      "grad_norm": 0.5623437166213989,
      "learning_rate": 1.0098684813365764e-05,
      "loss": 0.09689855575561523,
      "step": 4685
    },
    {
      "epoch": 0.633510773130545,
      "grad_norm": 0.2899349331855774,
      "learning_rate": 1.0092216590828115e-05,
      "loss": 0.06019878387451172,
      "step": 4686
    },
    {
      "epoch": 0.633645965356992,
      "grad_norm": 0.3198620676994324,
      "learning_rate": 1.0085749390117146e-05,
      "loss": 0.07785367965698242,
      "step": 4687
    },
    {
      "epoch": 0.633781157583439,
      "grad_norm": 0.20298810303211212,
      "learning_rate": 1.0079283212579354e-05,
      "loss": 0.05445504188537598,
      "step": 4688
    },
    {
      "epoch": 0.633916349809886,
      "grad_norm": 0.2476663440465927,
      "learning_rate": 1.0072818059561045e-05,
      "loss": 0.07100027799606323,
      "step": 4689
    },
    {
      "epoch": 0.634051542036333,
      "grad_norm": 0.3452564477920532,
      "learning_rate": 1.0066353932408285e-05,
      "loss": 0.07155418395996094,
      "step": 4690
    },
    {
      "epoch": 0.6341867342627799,
      "grad_norm": 0.38678160309791565,
      "learning_rate": 1.0059890832466948e-05,
      "loss": 0.06813859939575195,
      "step": 4691
    },
    {
      "epoch": 0.6343219264892269,
      "grad_norm": 0.2522895336151123,
      "learning_rate": 1.0053428761082684e-05,
      "loss": 0.04723095893859863,
      "step": 4692
    },
    {
      "epoch": 0.6344571187156739,
      "grad_norm": 0.1959107518196106,
      "learning_rate": 1.0046967719600927e-05,
      "loss": 0.04604077339172363,
      "step": 4693
    },
    {
      "epoch": 0.6345923109421209,
      "grad_norm": 0.15962094068527222,
      "learning_rate": 1.0040507709366912e-05,
      "loss": 0.040190696716308594,
      "step": 4694
    },
    {
      "epoch": 0.6347275031685679,
      "grad_norm": 0.30821219086647034,
      "learning_rate": 1.0034048731725631e-05,
      "loss": 0.08436393737792969,
      "step": 4695
    },
    {
      "epoch": 0.6348626953950148,
      "grad_norm": 0.22390079498291016,
      "learning_rate": 1.0027590788021886e-05,
      "loss": 0.05712532997131348,
      "step": 4696
    },
    {
      "epoch": 0.6349978876214618,
      "grad_norm": 0.28145474195480347,
      "learning_rate": 1.0021133879600258e-05,
      "loss": 0.07906675338745117,
      "step": 4697
    },
    {
      "epoch": 0.6351330798479088,
      "grad_norm": 0.23368291556835175,
      "learning_rate": 1.0014678007805108e-05,
      "loss": 0.042389869689941406,
      "step": 4698
    },
    {
      "epoch": 0.6352682720743558,
      "grad_norm": 0.2637837529182434,
      "learning_rate": 1.0008223173980579e-05,
      "loss": 0.050539493560791016,
      "step": 4699
    },
    {
      "epoch": 0.6354034643008027,
      "grad_norm": 0.3639850616455078,
      "learning_rate": 1.0001769379470604e-05,
      "loss": 0.058909863233566284,
      "step": 4700
    },
    {
      "epoch": 0.6355386565272497,
      "grad_norm": 0.4069882035255432,
      "learning_rate": 9.995316625618898e-06,
      "loss": 0.06889724731445312,
      "step": 4701
    },
    {
      "epoch": 0.6356738487536967,
      "grad_norm": 0.45420369505882263,
      "learning_rate": 9.988864913768962e-06,
      "loss": 0.06771183013916016,
      "step": 4702
    },
    {
      "epoch": 0.6358090409801437,
      "grad_norm": 0.1920868158340454,
      "learning_rate": 9.982414245264071e-06,
      "loss": 0.03780162334442139,
      "step": 4703
    },
    {
      "epoch": 0.6359442332065907,
      "grad_norm": 0.24480357766151428,
      "learning_rate": 9.975964621447293e-06,
      "loss": 0.04808378219604492,
      "step": 4704
    },
    {
      "epoch": 0.6360794254330376,
      "grad_norm": 0.3678262531757355,
      "learning_rate": 9.96951604366147e-06,
      "loss": 0.0884695053100586,
      "step": 4705
    },
    {
      "epoch": 0.6362146176594846,
      "grad_norm": 0.2896016836166382,
      "learning_rate": 9.963068513249233e-06,
      "loss": 0.06454920768737793,
      "step": 4706
    },
    {
      "epoch": 0.6363498098859316,
      "grad_norm": 0.23731695115566254,
      "learning_rate": 9.956622031552996e-06,
      "loss": 0.04004871845245361,
      "step": 4707
    },
    {
      "epoch": 0.6364850021123786,
      "grad_norm": 0.25214889645576477,
      "learning_rate": 9.950176599914942e-06,
      "loss": 0.05359506607055664,
      "step": 4708
    },
    {
      "epoch": 0.6366201943388256,
      "grad_norm": 0.2975670397281647,
      "learning_rate": 9.943732219677048e-06,
      "loss": 0.06443309783935547,
      "step": 4709
    },
    {
      "epoch": 0.6367553865652725,
      "grad_norm": 0.19786369800567627,
      "learning_rate": 9.93728889218107e-06,
      "loss": 0.06727170944213867,
      "step": 4710
    },
    {
      "epoch": 0.6368905787917195,
      "grad_norm": 0.2943669557571411,
      "learning_rate": 9.930846618768543e-06,
      "loss": 0.06118488311767578,
      "step": 4711
    },
    {
      "epoch": 0.6370257710181665,
      "grad_norm": 0.2945265471935272,
      "learning_rate": 9.924405400780784e-06,
      "loss": 0.0697164535522461,
      "step": 4712
    },
    {
      "epoch": 0.6371609632446135,
      "grad_norm": 0.2885686457157135,
      "learning_rate": 9.917965239558885e-06,
      "loss": 0.06333780288696289,
      "step": 4713
    },
    {
      "epoch": 0.6372961554710604,
      "grad_norm": 0.4006820619106293,
      "learning_rate": 9.911526136443726e-06,
      "loss": 0.059378623962402344,
      "step": 4714
    },
    {
      "epoch": 0.6374313476975074,
      "grad_norm": 0.186470627784729,
      "learning_rate": 9.905088092775956e-06,
      "loss": 0.04387474060058594,
      "step": 4715
    },
    {
      "epoch": 0.6375665399239544,
      "grad_norm": 0.17098212242126465,
      "learning_rate": 9.898651109896015e-06,
      "loss": 0.04893779754638672,
      "step": 4716
    },
    {
      "epoch": 0.6377017321504014,
      "grad_norm": 0.3996959626674652,
      "learning_rate": 9.892215189144123e-06,
      "loss": 0.05804312229156494,
      "step": 4717
    },
    {
      "epoch": 0.6378369243768484,
      "grad_norm": 0.26427116990089417,
      "learning_rate": 9.88578033186026e-06,
      "loss": 0.04778647422790527,
      "step": 4718
    },
    {
      "epoch": 0.6379721166032953,
      "grad_norm": 0.31532660126686096,
      "learning_rate": 9.879346539384207e-06,
      "loss": 0.06215834617614746,
      "step": 4719
    },
    {
      "epoch": 0.6381073088297423,
      "grad_norm": 0.3271363377571106,
      "learning_rate": 9.87291381305551e-06,
      "loss": 0.07738649845123291,
      "step": 4720
    },
    {
      "epoch": 0.6382425010561893,
      "grad_norm": 0.4354500472545624,
      "learning_rate": 9.866482154213502e-06,
      "loss": 0.06219172477722168,
      "step": 4721
    },
    {
      "epoch": 0.6383776932826363,
      "grad_norm": 0.5227985978126526,
      "learning_rate": 9.86005156419728e-06,
      "loss": 0.07993030548095703,
      "step": 4722
    },
    {
      "epoch": 0.6385128855090833,
      "grad_norm": 0.45466184616088867,
      "learning_rate": 9.853622044345732e-06,
      "loss": 0.09350919723510742,
      "step": 4723
    },
    {
      "epoch": 0.6386480777355302,
      "grad_norm": 0.2696090042591095,
      "learning_rate": 9.847193595997522e-06,
      "loss": 0.05406975746154785,
      "step": 4724
    },
    {
      "epoch": 0.6387832699619772,
      "grad_norm": 0.469687819480896,
      "learning_rate": 9.840766220491078e-06,
      "loss": 0.08649110794067383,
      "step": 4725
    },
    {
      "epoch": 0.6389184621884242,
      "grad_norm": 0.32442426681518555,
      "learning_rate": 9.834339919164625e-06,
      "loss": 0.04985618591308594,
      "step": 4726
    },
    {
      "epoch": 0.6390536544148712,
      "grad_norm": 0.4006844162940979,
      "learning_rate": 9.827914693356145e-06,
      "loss": 0.07015800476074219,
      "step": 4727
    },
    {
      "epoch": 0.6391888466413181,
      "grad_norm": 0.49044305086135864,
      "learning_rate": 9.821490544403403e-06,
      "loss": 0.07341480255126953,
      "step": 4728
    },
    {
      "epoch": 0.6393240388677651,
      "grad_norm": 0.17260241508483887,
      "learning_rate": 9.815067473643951e-06,
      "loss": 0.041432857513427734,
      "step": 4729
    },
    {
      "epoch": 0.6394592310942121,
      "grad_norm": 0.33676180243492126,
      "learning_rate": 9.808645482415097e-06,
      "loss": 0.06645870208740234,
      "step": 4730
    },
    {
      "epoch": 0.6395944233206591,
      "grad_norm": 0.4269150495529175,
      "learning_rate": 9.80222457205394e-06,
      "loss": 0.06735152006149292,
      "step": 4731
    },
    {
      "epoch": 0.6397296155471061,
      "grad_norm": 0.38384002447128296,
      "learning_rate": 9.795804743897341e-06,
      "loss": 0.08746576309204102,
      "step": 4732
    },
    {
      "epoch": 0.639864807773553,
      "grad_norm": 0.3227181136608124,
      "learning_rate": 9.789385999281948e-06,
      "loss": 0.06382691860198975,
      "step": 4733
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.14987395703792572,
      "learning_rate": 9.782968339544179e-06,
      "loss": 0.03530263900756836,
      "step": 4734
    },
    {
      "epoch": 0.640135192226447,
      "grad_norm": 0.6400836110115051,
      "learning_rate": 9.776551766020219e-06,
      "loss": 0.07858467102050781,
      "step": 4735
    },
    {
      "epoch": 0.640270384452894,
      "grad_norm": 0.16940130293369293,
      "learning_rate": 9.77013628004604e-06,
      "loss": 0.031525999307632446,
      "step": 4736
    },
    {
      "epoch": 0.640405576679341,
      "grad_norm": 0.23942354321479797,
      "learning_rate": 9.763721882957371e-06,
      "loss": 0.051934123039245605,
      "step": 4737
    },
    {
      "epoch": 0.6405407689057879,
      "grad_norm": 0.27816030383110046,
      "learning_rate": 9.757308576089732e-06,
      "loss": 0.05612468719482422,
      "step": 4738
    },
    {
      "epoch": 0.6406759611322349,
      "grad_norm": 0.2609564960002899,
      "learning_rate": 9.750896360778404e-06,
      "loss": 0.05990457534790039,
      "step": 4739
    },
    {
      "epoch": 0.6408111533586819,
      "grad_norm": 0.4077971577644348,
      "learning_rate": 9.744485238358448e-06,
      "loss": 0.08354783058166504,
      "step": 4740
    },
    {
      "epoch": 0.6409463455851289,
      "grad_norm": 0.17238418757915497,
      "learning_rate": 9.73807521016469e-06,
      "loss": 0.04747796058654785,
      "step": 4741
    },
    {
      "epoch": 0.6410815378115758,
      "grad_norm": 0.18301935493946075,
      "learning_rate": 9.731666277531732e-06,
      "loss": 0.048253536224365234,
      "step": 4742
    },
    {
      "epoch": 0.6412167300380228,
      "grad_norm": 0.4359232187271118,
      "learning_rate": 9.725258441793947e-06,
      "loss": 0.07181811332702637,
      "step": 4743
    },
    {
      "epoch": 0.6413519222644698,
      "grad_norm": 0.23843605816364288,
      "learning_rate": 9.71885170428549e-06,
      "loss": 0.052263736724853516,
      "step": 4744
    },
    {
      "epoch": 0.6414871144909168,
      "grad_norm": 0.20009545981884003,
      "learning_rate": 9.712446066340265e-06,
      "loss": 0.03910231590270996,
      "step": 4745
    },
    {
      "epoch": 0.6416223067173638,
      "grad_norm": 0.25603142380714417,
      "learning_rate": 9.70604152929197e-06,
      "loss": 0.060760498046875,
      "step": 4746
    },
    {
      "epoch": 0.6417574989438107,
      "grad_norm": 0.25704440474510193,
      "learning_rate": 9.699638094474054e-06,
      "loss": 0.0583195686340332,
      "step": 4747
    },
    {
      "epoch": 0.6418926911702577,
      "grad_norm": 0.31259119510650635,
      "learning_rate": 9.693235763219752e-06,
      "loss": 0.06410646438598633,
      "step": 4748
    },
    {
      "epoch": 0.6420278833967047,
      "grad_norm": 0.26537519693374634,
      "learning_rate": 9.68683453686207e-06,
      "loss": 0.059996604919433594,
      "step": 4749
    },
    {
      "epoch": 0.6421630756231517,
      "grad_norm": 0.4195689558982849,
      "learning_rate": 9.680434416733763e-06,
      "loss": 0.06110048294067383,
      "step": 4750
    },
    {
      "epoch": 0.6422982678495986,
      "grad_norm": 0.2629587948322296,
      "learning_rate": 9.674035404167381e-06,
      "loss": 0.04708099365234375,
      "step": 4751
    },
    {
      "epoch": 0.6424334600760456,
      "grad_norm": 0.43677568435668945,
      "learning_rate": 9.66763750049523e-06,
      "loss": 0.07480764389038086,
      "step": 4752
    },
    {
      "epoch": 0.6425686523024926,
      "grad_norm": 0.3900323212146759,
      "learning_rate": 9.66124070704939e-06,
      "loss": 0.04446077346801758,
      "step": 4753
    },
    {
      "epoch": 0.6427038445289396,
      "grad_norm": 0.39845535159111023,
      "learning_rate": 9.654845025161699e-06,
      "loss": 0.050692081451416016,
      "step": 4754
    },
    {
      "epoch": 0.6428390367553866,
      "grad_norm": 0.44947630167007446,
      "learning_rate": 9.648450456163777e-06,
      "loss": 0.0677117109298706,
      "step": 4755
    },
    {
      "epoch": 0.6429742289818335,
      "grad_norm": 0.3863936960697174,
      "learning_rate": 9.64205700138701e-06,
      "loss": 0.06791448593139648,
      "step": 4756
    },
    {
      "epoch": 0.6431094212082805,
      "grad_norm": 0.21248199045658112,
      "learning_rate": 9.635664662162548e-06,
      "loss": 0.06186962127685547,
      "step": 4757
    },
    {
      "epoch": 0.6432446134347275,
      "grad_norm": 0.6182746887207031,
      "learning_rate": 9.629273439821315e-06,
      "loss": 0.09067821502685547,
      "step": 4758
    },
    {
      "epoch": 0.6433798056611745,
      "grad_norm": 0.568256676197052,
      "learning_rate": 9.622883335693984e-06,
      "loss": 0.09613311290740967,
      "step": 4759
    },
    {
      "epoch": 0.6435149978876215,
      "grad_norm": 0.20267054438591003,
      "learning_rate": 9.616494351111017e-06,
      "loss": 0.05497455596923828,
      "step": 4760
    },
    {
      "epoch": 0.6436501901140684,
      "grad_norm": 0.5923112630844116,
      "learning_rate": 9.610106487402637e-06,
      "loss": 0.10086488723754883,
      "step": 4761
    },
    {
      "epoch": 0.6437853823405154,
      "grad_norm": 0.31522035598754883,
      "learning_rate": 9.603719745898826e-06,
      "loss": 0.08345794677734375,
      "step": 4762
    },
    {
      "epoch": 0.6439205745669624,
      "grad_norm": 0.3246569037437439,
      "learning_rate": 9.597334127929346e-06,
      "loss": 0.050672054290771484,
      "step": 4763
    },
    {
      "epoch": 0.6440557667934094,
      "grad_norm": 0.22075532376766205,
      "learning_rate": 9.590949634823707e-06,
      "loss": 0.05805182456970215,
      "step": 4764
    },
    {
      "epoch": 0.6441909590198563,
      "grad_norm": 0.4642619490623474,
      "learning_rate": 9.584566267911198e-06,
      "loss": 0.050574302673339844,
      "step": 4765
    },
    {
      "epoch": 0.6443261512463033,
      "grad_norm": 0.3408973515033722,
      "learning_rate": 9.578184028520874e-06,
      "loss": 0.0832815170288086,
      "step": 4766
    },
    {
      "epoch": 0.6444613434727503,
      "grad_norm": 0.22710277140140533,
      "learning_rate": 9.571802917981548e-06,
      "loss": 0.039969682693481445,
      "step": 4767
    },
    {
      "epoch": 0.6445965356991973,
      "grad_norm": 0.3381836414337158,
      "learning_rate": 9.565422937621798e-06,
      "loss": 0.0602116584777832,
      "step": 4768
    },
    {
      "epoch": 0.6447317279256443,
      "grad_norm": 0.2356148213148117,
      "learning_rate": 9.559044088769971e-06,
      "loss": 0.05148506164550781,
      "step": 4769
    },
    {
      "epoch": 0.6448669201520912,
      "grad_norm": 0.20472785830497742,
      "learning_rate": 9.552666372754182e-06,
      "loss": 0.04103350639343262,
      "step": 4770
    },
    {
      "epoch": 0.6450021123785382,
      "grad_norm": 0.20671877264976501,
      "learning_rate": 9.546289790902307e-06,
      "loss": 0.04780244827270508,
      "step": 4771
    },
    {
      "epoch": 0.6451373046049852,
      "grad_norm": 0.28850510716438293,
      "learning_rate": 9.539914344541976e-06,
      "loss": 0.040769219398498535,
      "step": 4772
    },
    {
      "epoch": 0.6452724968314322,
      "grad_norm": 0.36963313817977905,
      "learning_rate": 9.533540035000598e-06,
      "loss": 0.05578446388244629,
      "step": 4773
    },
    {
      "epoch": 0.6454076890578792,
      "grad_norm": 0.8242632150650024,
      "learning_rate": 9.52716686360533e-06,
      "loss": 0.09585762023925781,
      "step": 4774
    },
    {
      "epoch": 0.6455428812843261,
      "grad_norm": 0.26632484793663025,
      "learning_rate": 9.520794831683108e-06,
      "loss": 0.05351734161376953,
      "step": 4775
    },
    {
      "epoch": 0.6456780735107731,
      "grad_norm": 0.3278503715991974,
      "learning_rate": 9.514423940560627e-06,
      "loss": 0.041861534118652344,
      "step": 4776
    },
    {
      "epoch": 0.6458132657372201,
      "grad_norm": 0.2619352638721466,
      "learning_rate": 9.508054191564326e-06,
      "loss": 0.06525230407714844,
      "step": 4777
    },
    {
      "epoch": 0.6459484579636671,
      "grad_norm": 0.3093946874141693,
      "learning_rate": 9.501685586020434e-06,
      "loss": 0.06375670433044434,
      "step": 4778
    },
    {
      "epoch": 0.646083650190114,
      "grad_norm": 0.32293379306793213,
      "learning_rate": 9.495318125254919e-06,
      "loss": 0.07803821563720703,
      "step": 4779
    },
    {
      "epoch": 0.646218842416561,
      "grad_norm": 0.32007893919944763,
      "learning_rate": 9.488951810593527e-06,
      "loss": 0.05273151397705078,
      "step": 4780
    },
    {
      "epoch": 0.646354034643008,
      "grad_norm": 0.21784581243991852,
      "learning_rate": 9.48258664336176e-06,
      "loss": 0.04521918296813965,
      "step": 4781
    },
    {
      "epoch": 0.646489226869455,
      "grad_norm": 0.3079139292240143,
      "learning_rate": 9.476222624884873e-06,
      "loss": 0.05839395523071289,
      "step": 4782
    },
    {
      "epoch": 0.646624419095902,
      "grad_norm": 0.335799902677536,
      "learning_rate": 9.469859756487893e-06,
      "loss": 0.06855034828186035,
      "step": 4783
    },
    {
      "epoch": 0.6467596113223489,
      "grad_norm": 0.19430774450302124,
      "learning_rate": 9.463498039495598e-06,
      "loss": 0.04897332191467285,
      "step": 4784
    },
    {
      "epoch": 0.6468948035487959,
      "grad_norm": 0.33547085523605347,
      "learning_rate": 9.457137475232537e-06,
      "loss": 0.06531786918640137,
      "step": 4785
    },
    {
      "epoch": 0.6470299957752429,
      "grad_norm": 0.5125856995582581,
      "learning_rate": 9.450778065023019e-06,
      "loss": 0.07713031768798828,
      "step": 4786
    },
    {
      "epoch": 0.6471651880016899,
      "grad_norm": 0.32903537154197693,
      "learning_rate": 9.444419810191091e-06,
      "loss": 0.0629720687866211,
      "step": 4787
    },
    {
      "epoch": 0.6473003802281369,
      "grad_norm": 0.34005698561668396,
      "learning_rate": 9.43806271206059e-06,
      "loss": 0.062456607818603516,
      "step": 4788
    },
    {
      "epoch": 0.6474355724545838,
      "grad_norm": 0.35717150568962097,
      "learning_rate": 9.431706771955089e-06,
      "loss": 0.0530930757522583,
      "step": 4789
    },
    {
      "epoch": 0.6475707646810308,
      "grad_norm": 0.17114569246768951,
      "learning_rate": 9.425351991197937e-06,
      "loss": 0.045818328857421875,
      "step": 4790
    },
    {
      "epoch": 0.6477059569074778,
      "grad_norm": 0.2643604278564453,
      "learning_rate": 9.418998371112221e-06,
      "loss": 0.057444095611572266,
      "step": 4791
    },
    {
      "epoch": 0.6478411491339248,
      "grad_norm": 0.17470473051071167,
      "learning_rate": 9.412645913020807e-06,
      "loss": 0.04263979196548462,
      "step": 4792
    },
    {
      "epoch": 0.6479763413603717,
      "grad_norm": 0.25354933738708496,
      "learning_rate": 9.406294618246313e-06,
      "loss": 0.07153701782226562,
      "step": 4793
    },
    {
      "epoch": 0.6481115335868187,
      "grad_norm": 0.3025719225406647,
      "learning_rate": 9.399944488111103e-06,
      "loss": 0.05858159065246582,
      "step": 4794
    },
    {
      "epoch": 0.6482467258132657,
      "grad_norm": 0.2793440520763397,
      "learning_rate": 9.39359552393732e-06,
      "loss": 0.0573887825012207,
      "step": 4795
    },
    {
      "epoch": 0.6483819180397127,
      "grad_norm": 0.2921557128429413,
      "learning_rate": 9.387247727046845e-06,
      "loss": 0.04311037063598633,
      "step": 4796
    },
    {
      "epoch": 0.6485171102661597,
      "grad_norm": 0.2962039113044739,
      "learning_rate": 9.380901098761319e-06,
      "loss": 0.07325172424316406,
      "step": 4797
    },
    {
      "epoch": 0.6486523024926066,
      "grad_norm": 0.2610384225845337,
      "learning_rate": 9.374555640402153e-06,
      "loss": 0.03996849060058594,
      "step": 4798
    },
    {
      "epoch": 0.6487874947190536,
      "grad_norm": 0.34995609521865845,
      "learning_rate": 9.368211353290503e-06,
      "loss": 0.0531916618347168,
      "step": 4799
    },
    {
      "epoch": 0.6489226869455006,
      "grad_norm": 0.5493077635765076,
      "learning_rate": 9.36186823874728e-06,
      "loss": 0.0944375991821289,
      "step": 4800
    },
    {
      "epoch": 0.6490578791719476,
      "grad_norm": 0.147910937666893,
      "learning_rate": 9.355526298093152e-06,
      "loss": 0.04943704605102539,
      "step": 4801
    },
    {
      "epoch": 0.6491930713983946,
      "grad_norm": 0.5425330400466919,
      "learning_rate": 9.34918553264855e-06,
      "loss": 0.08262139558792114,
      "step": 4802
    },
    {
      "epoch": 0.6493282636248415,
      "grad_norm": 0.3683369755744934,
      "learning_rate": 9.342845943733658e-06,
      "loss": 0.0776968002319336,
      "step": 4803
    },
    {
      "epoch": 0.6494634558512885,
      "grad_norm": 0.3404667377471924,
      "learning_rate": 9.336507532668407e-06,
      "loss": 0.07655715942382812,
      "step": 4804
    },
    {
      "epoch": 0.6495986480777355,
      "grad_norm": 0.20939965546131134,
      "learning_rate": 9.33017030077249e-06,
      "loss": 0.04820966720581055,
      "step": 4805
    },
    {
      "epoch": 0.6497338403041825,
      "grad_norm": 0.570598304271698,
      "learning_rate": 9.323834249365346e-06,
      "loss": 0.07071280479431152,
      "step": 4806
    },
    {
      "epoch": 0.6498690325306294,
      "grad_norm": 0.1624952256679535,
      "learning_rate": 9.317499379766183e-06,
      "loss": 0.03916573524475098,
      "step": 4807
    },
    {
      "epoch": 0.6500042247570764,
      "grad_norm": 0.22451962530612946,
      "learning_rate": 9.311165693293954e-06,
      "loss": 0.05844736099243164,
      "step": 4808
    },
    {
      "epoch": 0.6501394169835234,
      "grad_norm": 0.24295949935913086,
      "learning_rate": 9.304833191267364e-06,
      "loss": 0.05803394317626953,
      "step": 4809
    },
    {
      "epoch": 0.6502746092099704,
      "grad_norm": 0.3330073356628418,
      "learning_rate": 9.298501875004874e-06,
      "loss": 0.07588577270507812,
      "step": 4810
    },
    {
      "epoch": 0.6504098014364174,
      "grad_norm": 0.2678002715110779,
      "learning_rate": 9.292171745824695e-06,
      "loss": 0.06055879592895508,
      "step": 4811
    },
    {
      "epoch": 0.6505449936628643,
      "grad_norm": 0.2708369195461273,
      "learning_rate": 9.285842805044797e-06,
      "loss": 0.06868457794189453,
      "step": 4812
    },
    {
      "epoch": 0.6506801858893113,
      "grad_norm": 0.19736354053020477,
      "learning_rate": 9.279515053982905e-06,
      "loss": 0.04729270935058594,
      "step": 4813
    },
    {
      "epoch": 0.6508153781157583,
      "grad_norm": 0.3077872693538666,
      "learning_rate": 9.273188493956476e-06,
      "loss": 0.056958556175231934,
      "step": 4814
    },
    {
      "epoch": 0.6509505703422053,
      "grad_norm": 0.362509548664093,
      "learning_rate": 9.266863126282746e-06,
      "loss": 0.05745649337768555,
      "step": 4815
    },
    {
      "epoch": 0.6510857625686522,
      "grad_norm": 0.343902587890625,
      "learning_rate": 9.260538952278683e-06,
      "loss": 0.07398080825805664,
      "step": 4816
    },
    {
      "epoch": 0.6512209547950992,
      "grad_norm": 0.19427059590816498,
      "learning_rate": 9.254215973261014e-06,
      "loss": 0.045670509338378906,
      "step": 4817
    },
    {
      "epoch": 0.6513561470215462,
      "grad_norm": 0.26505959033966064,
      "learning_rate": 9.247894190546228e-06,
      "loss": 0.05563807487487793,
      "step": 4818
    },
    {
      "epoch": 0.6514913392479933,
      "grad_norm": 0.21521027386188507,
      "learning_rate": 9.241573605450539e-06,
      "loss": 0.04488945007324219,
      "step": 4819
    },
    {
      "epoch": 0.6516265314744403,
      "grad_norm": 0.23863205313682556,
      "learning_rate": 9.235254219289937e-06,
      "loss": 0.05624032020568848,
      "step": 4820
    },
    {
      "epoch": 0.6517617237008873,
      "grad_norm": 0.31005093455314636,
      "learning_rate": 9.228936033380143e-06,
      "loss": 0.05953836441040039,
      "step": 4821
    },
    {
      "epoch": 0.6518969159273342,
      "grad_norm": 0.6588156223297119,
      "learning_rate": 9.222619049036649e-06,
      "loss": 0.09979248046875,
      "step": 4822
    },
    {
      "epoch": 0.6520321081537812,
      "grad_norm": 0.33989521861076355,
      "learning_rate": 9.216303267574674e-06,
      "loss": 0.09782075881958008,
      "step": 4823
    },
    {
      "epoch": 0.6521673003802282,
      "grad_norm": 0.2392580211162567,
      "learning_rate": 9.209988690309198e-06,
      "loss": 0.061174869537353516,
      "step": 4824
    },
    {
      "epoch": 0.6523024926066752,
      "grad_norm": 0.28097352385520935,
      "learning_rate": 9.203675318554956e-06,
      "loss": 0.05446434020996094,
      "step": 4825
    },
    {
      "epoch": 0.6524376848331221,
      "grad_norm": 0.30071333050727844,
      "learning_rate": 9.19736315362642e-06,
      "loss": 0.07001447677612305,
      "step": 4826
    },
    {
      "epoch": 0.6525728770595691,
      "grad_norm": 0.27183136343955994,
      "learning_rate": 9.191052196837825e-06,
      "loss": 0.07165145874023438,
      "step": 4827
    },
    {
      "epoch": 0.6527080692860161,
      "grad_norm": 0.23894578218460083,
      "learning_rate": 9.184742449503135e-06,
      "loss": 0.04506540298461914,
      "step": 4828
    },
    {
      "epoch": 0.6528432615124631,
      "grad_norm": 0.4670444428920746,
      "learning_rate": 9.178433912936077e-06,
      "loss": 0.07266426086425781,
      "step": 4829
    },
    {
      "epoch": 0.6529784537389101,
      "grad_norm": 0.2888253629207611,
      "learning_rate": 9.172126588450125e-06,
      "loss": 0.08974266052246094,
      "step": 4830
    },
    {
      "epoch": 0.653113645965357,
      "grad_norm": 0.23133684694766998,
      "learning_rate": 9.165820477358491e-06,
      "loss": 0.05648946762084961,
      "step": 4831
    },
    {
      "epoch": 0.653248838191804,
      "grad_norm": 0.43320906162261963,
      "learning_rate": 9.159515580974154e-06,
      "loss": 0.08557617664337158,
      "step": 4832
    },
    {
      "epoch": 0.653384030418251,
      "grad_norm": 0.26697322726249695,
      "learning_rate": 9.15321190060981e-06,
      "loss": 0.06428468227386475,
      "step": 4833
    },
    {
      "epoch": 0.653519222644698,
      "grad_norm": 0.4658189117908478,
      "learning_rate": 9.14690943757793e-06,
      "loss": 0.08957171440124512,
      "step": 4834
    },
    {
      "epoch": 0.653654414871145,
      "grad_norm": 0.27478429675102234,
      "learning_rate": 9.14060819319072e-06,
      "loss": 0.07382440567016602,
      "step": 4835
    },
    {
      "epoch": 0.6537896070975919,
      "grad_norm": 0.24716217815876007,
      "learning_rate": 9.134308168760127e-06,
      "loss": 0.05063152313232422,
      "step": 4836
    },
    {
      "epoch": 0.6539247993240389,
      "grad_norm": 0.4276491403579712,
      "learning_rate": 9.128009365597854e-06,
      "loss": 0.062435150146484375,
      "step": 4837
    },
    {
      "epoch": 0.6540599915504859,
      "grad_norm": 0.5603787302970886,
      "learning_rate": 9.121711785015342e-06,
      "loss": 0.06441593170166016,
      "step": 4838
    },
    {
      "epoch": 0.6541951837769329,
      "grad_norm": 0.31718993186950684,
      "learning_rate": 9.115415428323787e-06,
      "loss": 0.07912254333496094,
      "step": 4839
    },
    {
      "epoch": 0.6543303760033798,
      "grad_norm": 0.48668864369392395,
      "learning_rate": 9.109120296834118e-06,
      "loss": 0.07065677642822266,
      "step": 4840
    },
    {
      "epoch": 0.6544655682298268,
      "grad_norm": 0.26175540685653687,
      "learning_rate": 9.10282639185702e-06,
      "loss": 0.06042909622192383,
      "step": 4841
    },
    {
      "epoch": 0.6546007604562738,
      "grad_norm": 0.24747289717197418,
      "learning_rate": 9.096533714702913e-06,
      "loss": 0.05627012252807617,
      "step": 4842
    },
    {
      "epoch": 0.6547359526827208,
      "grad_norm": 0.21935872733592987,
      "learning_rate": 9.090242266681967e-06,
      "loss": 0.0528559684753418,
      "step": 4843
    },
    {
      "epoch": 0.6548711449091678,
      "grad_norm": 0.5683515667915344,
      "learning_rate": 9.083952049104094e-06,
      "loss": 0.06490063667297363,
      "step": 4844
    },
    {
      "epoch": 0.6550063371356147,
      "grad_norm": 0.23414266109466553,
      "learning_rate": 9.07766306327896e-06,
      "loss": 0.0586695671081543,
      "step": 4845
    },
    {
      "epoch": 0.6551415293620617,
      "grad_norm": 0.3659723401069641,
      "learning_rate": 9.071375310515949e-06,
      "loss": 0.06589281558990479,
      "step": 4846
    },
    {
      "epoch": 0.6552767215885087,
      "grad_norm": 0.48043200373649597,
      "learning_rate": 9.065088792124219e-06,
      "loss": 0.06490880250930786,
      "step": 4847
    },
    {
      "epoch": 0.6554119138149557,
      "grad_norm": 0.3887607455253601,
      "learning_rate": 9.058803509412647e-06,
      "loss": 0.07252264022827148,
      "step": 4848
    },
    {
      "epoch": 0.6555471060414026,
      "grad_norm": 0.328997939825058,
      "learning_rate": 9.05251946368987e-06,
      "loss": 0.057118892669677734,
      "step": 4849
    },
    {
      "epoch": 0.6556822982678496,
      "grad_norm": 0.5450140237808228,
      "learning_rate": 9.046236656264258e-06,
      "loss": 0.05856800079345703,
      "step": 4850
    },
    {
      "epoch": 0.6558174904942966,
      "grad_norm": 0.3553744852542877,
      "learning_rate": 9.03995508844392e-06,
      "loss": 0.08736777305603027,
      "step": 4851
    },
    {
      "epoch": 0.6559526827207436,
      "grad_norm": 0.1743435114622116,
      "learning_rate": 9.033674761536718e-06,
      "loss": 0.040869712829589844,
      "step": 4852
    },
    {
      "epoch": 0.6560878749471906,
      "grad_norm": 0.2205936163663864,
      "learning_rate": 9.027395676850244e-06,
      "loss": 0.04245615005493164,
      "step": 4853
    },
    {
      "epoch": 0.6562230671736375,
      "grad_norm": 0.38957759737968445,
      "learning_rate": 9.02111783569184e-06,
      "loss": 0.07975459098815918,
      "step": 4854
    },
    {
      "epoch": 0.6563582594000845,
      "grad_norm": 0.14929905533790588,
      "learning_rate": 9.014841239368591e-06,
      "loss": 0.03610038757324219,
      "step": 4855
    },
    {
      "epoch": 0.6564934516265315,
      "grad_norm": 0.33773738145828247,
      "learning_rate": 9.008565889187308e-06,
      "loss": 0.0732421875,
      "step": 4856
    },
    {
      "epoch": 0.6566286438529785,
      "grad_norm": 0.35000213980674744,
      "learning_rate": 9.00229178645456e-06,
      "loss": 0.08603858947753906,
      "step": 4857
    },
    {
      "epoch": 0.6567638360794255,
      "grad_norm": 0.3118942081928253,
      "learning_rate": 8.996018932476641e-06,
      "loss": 0.06601238250732422,
      "step": 4858
    },
    {
      "epoch": 0.6568990283058724,
      "grad_norm": 0.3083891272544861,
      "learning_rate": 8.989747328559606e-06,
      "loss": 0.060178279876708984,
      "step": 4859
    },
    {
      "epoch": 0.6570342205323194,
      "grad_norm": 0.3511293828487396,
      "learning_rate": 8.98347697600922e-06,
      "loss": 0.07991886138916016,
      "step": 4860
    },
    {
      "epoch": 0.6571694127587664,
      "grad_norm": 0.34385982155799866,
      "learning_rate": 8.977207876131013e-06,
      "loss": 0.050768136978149414,
      "step": 4861
    },
    {
      "epoch": 0.6573046049852134,
      "grad_norm": 0.3257488012313843,
      "learning_rate": 8.970940030230245e-06,
      "loss": 0.06351375579833984,
      "step": 4862
    },
    {
      "epoch": 0.6574397972116603,
      "grad_norm": 0.23448587954044342,
      "learning_rate": 8.96467343961191e-06,
      "loss": 0.053433895111083984,
      "step": 4863
    },
    {
      "epoch": 0.6575749894381073,
      "grad_norm": 0.29061970114707947,
      "learning_rate": 8.958408105580759e-06,
      "loss": 0.06228160858154297,
      "step": 4864
    },
    {
      "epoch": 0.6577101816645543,
      "grad_norm": 0.33464333415031433,
      "learning_rate": 8.952144029441248e-06,
      "loss": 0.07623767852783203,
      "step": 4865
    },
    {
      "epoch": 0.6578453738910013,
      "grad_norm": 0.26314496994018555,
      "learning_rate": 8.945881212497603e-06,
      "loss": 0.05088233947753906,
      "step": 4866
    },
    {
      "epoch": 0.6579805661174483,
      "grad_norm": 0.4107857346534729,
      "learning_rate": 8.939619656053777e-06,
      "loss": 0.06191253662109375,
      "step": 4867
    },
    {
      "epoch": 0.6581157583438952,
      "grad_norm": 0.1694977730512619,
      "learning_rate": 8.933359361413456e-06,
      "loss": 0.0455777645111084,
      "step": 4868
    },
    {
      "epoch": 0.6582509505703422,
      "grad_norm": 0.15742100775241852,
      "learning_rate": 8.92710032988007e-06,
      "loss": 0.040313780307769775,
      "step": 4869
    },
    {
      "epoch": 0.6583861427967892,
      "grad_norm": 0.27373185753822327,
      "learning_rate": 8.920842562756773e-06,
      "loss": 0.06558692455291748,
      "step": 4870
    },
    {
      "epoch": 0.6585213350232362,
      "grad_norm": 0.27360764145851135,
      "learning_rate": 8.914586061346474e-06,
      "loss": 0.058560848236083984,
      "step": 4871
    },
    {
      "epoch": 0.6586565272496832,
      "grad_norm": 0.41100335121154785,
      "learning_rate": 8.908330826951811e-06,
      "loss": 0.06381797790527344,
      "step": 4872
    },
    {
      "epoch": 0.6587917194761301,
      "grad_norm": 0.403158038854599,
      "learning_rate": 8.902076860875155e-06,
      "loss": 0.08599472045898438,
      "step": 4873
    },
    {
      "epoch": 0.6589269117025771,
      "grad_norm": 0.2476569265127182,
      "learning_rate": 8.895824164418615e-06,
      "loss": 0.06425809860229492,
      "step": 4874
    },
    {
      "epoch": 0.6590621039290241,
      "grad_norm": 0.2766430675983429,
      "learning_rate": 8.889572738884033e-06,
      "loss": 0.04670858383178711,
      "step": 4875
    },
    {
      "epoch": 0.6591972961554711,
      "grad_norm": 0.20553795993328094,
      "learning_rate": 8.88332258557299e-06,
      "loss": 0.027480363845825195,
      "step": 4876
    },
    {
      "epoch": 0.659332488381918,
      "grad_norm": 0.2489897906780243,
      "learning_rate": 8.877073705786806e-06,
      "loss": 0.05231022834777832,
      "step": 4877
    },
    {
      "epoch": 0.659467680608365,
      "grad_norm": 0.589600682258606,
      "learning_rate": 8.870826100826527e-06,
      "loss": 0.07990264892578125,
      "step": 4878
    },
    {
      "epoch": 0.659602872834812,
      "grad_norm": 0.3277297019958496,
      "learning_rate": 8.86457977199294e-06,
      "loss": 0.07029342651367188,
      "step": 4879
    },
    {
      "epoch": 0.659738065061259,
      "grad_norm": 0.2849866449832916,
      "learning_rate": 8.85833472058656e-06,
      "loss": 0.05579376220703125,
      "step": 4880
    },
    {
      "epoch": 0.659873257287706,
      "grad_norm": 0.2331986129283905,
      "learning_rate": 8.852090947907643e-06,
      "loss": 0.04408740997314453,
      "step": 4881
    },
    {
      "epoch": 0.6600084495141529,
      "grad_norm": 0.46188732981681824,
      "learning_rate": 8.84584845525618e-06,
      "loss": 0.08914899826049805,
      "step": 4882
    },
    {
      "epoch": 0.6601436417405999,
      "grad_norm": 0.18672581017017365,
      "learning_rate": 8.83960724393188e-06,
      "loss": 0.044676899909973145,
      "step": 4883
    },
    {
      "epoch": 0.6602788339670469,
      "grad_norm": 0.368912011384964,
      "learning_rate": 8.833367315234206e-06,
      "loss": 0.07218313217163086,
      "step": 4884
    },
    {
      "epoch": 0.6604140261934939,
      "grad_norm": 0.16541294753551483,
      "learning_rate": 8.82712867046234e-06,
      "loss": 0.0427403450012207,
      "step": 4885
    },
    {
      "epoch": 0.6605492184199409,
      "grad_norm": 0.34060078859329224,
      "learning_rate": 8.820891310915203e-06,
      "loss": 0.04632425308227539,
      "step": 4886
    },
    {
      "epoch": 0.6606844106463878,
      "grad_norm": 0.34567373991012573,
      "learning_rate": 8.81465523789145e-06,
      "loss": 0.06366896629333496,
      "step": 4887
    },
    {
      "epoch": 0.6608196028728348,
      "grad_norm": 0.2853766679763794,
      "learning_rate": 8.808420452689455e-06,
      "loss": 0.043752312660217285,
      "step": 4888
    },
    {
      "epoch": 0.6609547950992818,
      "grad_norm": 0.46743839979171753,
      "learning_rate": 8.802186956607344e-06,
      "loss": 0.08290624618530273,
      "step": 4889
    },
    {
      "epoch": 0.6610899873257288,
      "grad_norm": 0.38309580087661743,
      "learning_rate": 8.795954750942954e-06,
      "loss": 0.08163714408874512,
      "step": 4890
    },
    {
      "epoch": 0.6612251795521757,
      "grad_norm": 0.3507712185382843,
      "learning_rate": 8.789723836993878e-06,
      "loss": 0.08322834968566895,
      "step": 4891
    },
    {
      "epoch": 0.6613603717786227,
      "grad_norm": 0.3610805571079254,
      "learning_rate": 8.783494216057407e-06,
      "loss": 0.06266021728515625,
      "step": 4892
    },
    {
      "epoch": 0.6614955640050697,
      "grad_norm": 0.5142252445220947,
      "learning_rate": 8.777265889430593e-06,
      "loss": 0.08034348487854004,
      "step": 4893
    },
    {
      "epoch": 0.6616307562315167,
      "grad_norm": 0.25728100538253784,
      "learning_rate": 8.771038858410206e-06,
      "loss": 0.05245685577392578,
      "step": 4894
    },
    {
      "epoch": 0.6617659484579637,
      "grad_norm": 0.3067833483219147,
      "learning_rate": 8.764813124292744e-06,
      "loss": 0.07212018966674805,
      "step": 4895
    },
    {
      "epoch": 0.6619011406844106,
      "grad_norm": 0.30051174759864807,
      "learning_rate": 8.758588688374445e-06,
      "loss": 0.055178165435791016,
      "step": 4896
    },
    {
      "epoch": 0.6620363329108576,
      "grad_norm": 0.43290379643440247,
      "learning_rate": 8.752365551951262e-06,
      "loss": 0.09206390380859375,
      "step": 4897
    },
    {
      "epoch": 0.6621715251373046,
      "grad_norm": 0.4254007339477539,
      "learning_rate": 8.74614371631888e-06,
      "loss": 0.059029579162597656,
      "step": 4898
    },
    {
      "epoch": 0.6623067173637516,
      "grad_norm": 0.4177686870098114,
      "learning_rate": 8.739923182772732e-06,
      "loss": 0.06552982330322266,
      "step": 4899
    },
    {
      "epoch": 0.6624419095901986,
      "grad_norm": 0.43566298484802246,
      "learning_rate": 8.733703952607956e-06,
      "loss": 0.09009742736816406,
      "step": 4900
    },
    {
      "epoch": 0.6625771018166455,
      "grad_norm": 0.28038373589515686,
      "learning_rate": 8.727486027119443e-06,
      "loss": 0.08281230926513672,
      "step": 4901
    },
    {
      "epoch": 0.6627122940430925,
      "grad_norm": 0.37428146600723267,
      "learning_rate": 8.721269407601783e-06,
      "loss": 0.06670522689819336,
      "step": 4902
    },
    {
      "epoch": 0.6628474862695395,
      "grad_norm": 0.22457443177700043,
      "learning_rate": 8.71505409534931e-06,
      "loss": 0.05708813667297363,
      "step": 4903
    },
    {
      "epoch": 0.6629826784959865,
      "grad_norm": 0.26721227169036865,
      "learning_rate": 8.708840091656093e-06,
      "loss": 0.07717132568359375,
      "step": 4904
    },
    {
      "epoch": 0.6631178707224334,
      "grad_norm": 0.2777976393699646,
      "learning_rate": 8.70262739781592e-06,
      "loss": 0.054625749588012695,
      "step": 4905
    },
    {
      "epoch": 0.6632530629488804,
      "grad_norm": 0.18250279128551483,
      "learning_rate": 8.696416015122302e-06,
      "loss": 0.0447998046875,
      "step": 4906
    },
    {
      "epoch": 0.6633882551753274,
      "grad_norm": 0.32662713527679443,
      "learning_rate": 8.690205944868487e-06,
      "loss": 0.07663249969482422,
      "step": 4907
    },
    {
      "epoch": 0.6635234474017744,
      "grad_norm": 0.302788108587265,
      "learning_rate": 8.683997188347436e-06,
      "loss": 0.05095010995864868,
      "step": 4908
    },
    {
      "epoch": 0.6636586396282214,
      "grad_norm": 0.5169642567634583,
      "learning_rate": 8.677789746851855e-06,
      "loss": 0.07211065292358398,
      "step": 4909
    },
    {
      "epoch": 0.6637938318546683,
      "grad_norm": 0.38796767592430115,
      "learning_rate": 8.671583621674167e-06,
      "loss": 0.058159828186035156,
      "step": 4910
    },
    {
      "epoch": 0.6639290240811153,
      "grad_norm": 0.20360036194324493,
      "learning_rate": 8.665378814106512e-06,
      "loss": 0.05255138874053955,
      "step": 4911
    },
    {
      "epoch": 0.6640642163075623,
      "grad_norm": 0.16078992187976837,
      "learning_rate": 8.65917532544077e-06,
      "loss": 0.05150127410888672,
      "step": 4912
    },
    {
      "epoch": 0.6641994085340093,
      "grad_norm": 0.2832340896129608,
      "learning_rate": 8.652973156968532e-06,
      "loss": 0.05080914497375488,
      "step": 4913
    },
    {
      "epoch": 0.6643346007604563,
      "grad_norm": 0.31240326166152954,
      "learning_rate": 8.646772309981141e-06,
      "loss": 0.06571626663208008,
      "step": 4914
    },
    {
      "epoch": 0.6644697929869032,
      "grad_norm": 0.2230207473039627,
      "learning_rate": 8.640572785769624e-06,
      "loss": 0.059484243392944336,
      "step": 4915
    },
    {
      "epoch": 0.6646049852133502,
      "grad_norm": 0.21061892807483673,
      "learning_rate": 8.63437458562477e-06,
      "loss": 0.06384515762329102,
      "step": 4916
    },
    {
      "epoch": 0.6647401774397972,
      "grad_norm": 0.49494194984436035,
      "learning_rate": 8.628177710837068e-06,
      "loss": 0.054257869720458984,
      "step": 4917
    },
    {
      "epoch": 0.6648753696662442,
      "grad_norm": 0.3318653106689453,
      "learning_rate": 8.621982162696752e-06,
      "loss": 0.05859947204589844,
      "step": 4918
    },
    {
      "epoch": 0.6650105618926911,
      "grad_norm": 0.2751937806606293,
      "learning_rate": 8.615787942493766e-06,
      "loss": 0.062012553215026855,
      "step": 4919
    },
    {
      "epoch": 0.6651457541191381,
      "grad_norm": 0.2703063488006592,
      "learning_rate": 8.609595051517765e-06,
      "loss": 0.05864429473876953,
      "step": 4920
    },
    {
      "epoch": 0.6652809463455851,
      "grad_norm": 0.3056776523590088,
      "learning_rate": 8.603403491058157e-06,
      "loss": 0.08215045928955078,
      "step": 4921
    },
    {
      "epoch": 0.6654161385720321,
      "grad_norm": 0.21935805678367615,
      "learning_rate": 8.597213262404046e-06,
      "loss": 0.04139423370361328,
      "step": 4922
    },
    {
      "epoch": 0.6655513307984791,
      "grad_norm": 0.4530940651893616,
      "learning_rate": 8.591024366844291e-06,
      "loss": 0.11154651641845703,
      "step": 4923
    },
    {
      "epoch": 0.665686523024926,
      "grad_norm": 0.3644537329673767,
      "learning_rate": 8.584836805667434e-06,
      "loss": 0.07296013832092285,
      "step": 4924
    },
    {
      "epoch": 0.665821715251373,
      "grad_norm": 1.0423619747161865,
      "learning_rate": 8.578650580161754e-06,
      "loss": 0.09050202369689941,
      "step": 4925
    },
    {
      "epoch": 0.66595690747782,
      "grad_norm": 0.4302298426628113,
      "learning_rate": 8.572465691615275e-06,
      "loss": 0.086334228515625,
      "step": 4926
    },
    {
      "epoch": 0.666092099704267,
      "grad_norm": 0.16452211141586304,
      "learning_rate": 8.56628214131571e-06,
      "loss": 0.03101372718811035,
      "step": 4927
    },
    {
      "epoch": 0.666227291930714,
      "grad_norm": 0.2247740477323532,
      "learning_rate": 8.560099930550523e-06,
      "loss": 0.06033158302307129,
      "step": 4928
    },
    {
      "epoch": 0.6663624841571609,
      "grad_norm": 0.3494688868522644,
      "learning_rate": 8.553919060606866e-06,
      "loss": 0.06897926330566406,
      "step": 4929
    },
    {
      "epoch": 0.6664976763836079,
      "grad_norm": 0.22867992520332336,
      "learning_rate": 8.54773953277163e-06,
      "loss": 0.0646820068359375,
      "step": 4930
    },
    {
      "epoch": 0.6666328686100549,
      "grad_norm": 0.2649802565574646,
      "learning_rate": 8.541561348331433e-06,
      "loss": 0.04089951515197754,
      "step": 4931
    },
    {
      "epoch": 0.6667680608365019,
      "grad_norm": 0.23530559241771698,
      "learning_rate": 8.535384508572603e-06,
      "loss": 0.05472928285598755,
      "step": 4932
    },
    {
      "epoch": 0.6669032530629488,
      "grad_norm": 0.30980250239372253,
      "learning_rate": 8.529209014781202e-06,
      "loss": 0.08345746994018555,
      "step": 4933
    },
    {
      "epoch": 0.6670384452893958,
      "grad_norm": 0.3073015809059143,
      "learning_rate": 8.523034868242984e-06,
      "loss": 0.07625198364257812,
      "step": 4934
    },
    {
      "epoch": 0.6671736375158428,
      "grad_norm": 0.3319665193557739,
      "learning_rate": 8.51686207024344e-06,
      "loss": 0.05214571952819824,
      "step": 4935
    },
    {
      "epoch": 0.6673088297422898,
      "grad_norm": 0.4453583061695099,
      "learning_rate": 8.510690622067792e-06,
      "loss": 0.06069660186767578,
      "step": 4936
    },
    {
      "epoch": 0.6674440219687368,
      "grad_norm": 0.25854286551475525,
      "learning_rate": 8.50452052500096e-06,
      "loss": 0.0675954818725586,
      "step": 4937
    },
    {
      "epoch": 0.6675792141951837,
      "grad_norm": 0.35607650876045227,
      "learning_rate": 8.498351780327594e-06,
      "loss": 0.059325218200683594,
      "step": 4938
    },
    {
      "epoch": 0.6677144064216307,
      "grad_norm": 0.6138937473297119,
      "learning_rate": 8.492184389332061e-06,
      "loss": 0.08625459671020508,
      "step": 4939
    },
    {
      "epoch": 0.6678495986480777,
      "grad_norm": 0.33962374925613403,
      "learning_rate": 8.486018353298432e-06,
      "loss": 0.05710101127624512,
      "step": 4940
    },
    {
      "epoch": 0.6679847908745247,
      "grad_norm": 0.3117373585700989,
      "learning_rate": 8.479853673510528e-06,
      "loss": 0.0627889633178711,
      "step": 4941
    },
    {
      "epoch": 0.6681199831009716,
      "grad_norm": 0.3765493333339691,
      "learning_rate": 8.473690351251855e-06,
      "loss": 0.08568382263183594,
      "step": 4942
    },
    {
      "epoch": 0.6682551753274186,
      "grad_norm": 0.2630918025970459,
      "learning_rate": 8.467528387805656e-06,
      "loss": 0.04690957069396973,
      "step": 4943
    },
    {
      "epoch": 0.6683903675538656,
      "grad_norm": 0.29838240146636963,
      "learning_rate": 8.461367784454881e-06,
      "loss": 0.06570672988891602,
      "step": 4944
    },
    {
      "epoch": 0.6685255597803126,
      "grad_norm": 0.34062016010284424,
      "learning_rate": 8.455208542482195e-06,
      "loss": 0.06801176071166992,
      "step": 4945
    },
    {
      "epoch": 0.6686607520067596,
      "grad_norm": 0.1843264400959015,
      "learning_rate": 8.449050663170004e-06,
      "loss": 0.03787040710449219,
      "step": 4946
    },
    {
      "epoch": 0.6687959442332065,
      "grad_norm": 0.1837097555398941,
      "learning_rate": 8.442894147800387e-06,
      "loss": 0.0482785701751709,
      "step": 4947
    },
    {
      "epoch": 0.6689311364596535,
      "grad_norm": 0.2825604975223541,
      "learning_rate": 8.436738997655184e-06,
      "loss": 0.06084108352661133,
      "step": 4948
    },
    {
      "epoch": 0.6690663286861005,
      "grad_norm": 0.35577651858329773,
      "learning_rate": 8.430585214015918e-06,
      "loss": 0.06685876846313477,
      "step": 4949
    },
    {
      "epoch": 0.6692015209125475,
      "grad_norm": 0.24625088274478912,
      "learning_rate": 8.424432798163838e-06,
      "loss": 0.05525350570678711,
      "step": 4950
    },
    {
      "epoch": 0.6693367131389945,
      "grad_norm": 0.1947268694639206,
      "learning_rate": 8.418281751379926e-06,
      "loss": 0.0492253303527832,
      "step": 4951
    },
    {
      "epoch": 0.6694719053654414,
      "grad_norm": 0.32005617022514343,
      "learning_rate": 8.41213207494484e-06,
      "loss": 0.060690879821777344,
      "step": 4952
    },
    {
      "epoch": 0.6696070975918884,
      "grad_norm": 0.1298406720161438,
      "learning_rate": 8.405983770138992e-06,
      "loss": 0.03653430938720703,
      "step": 4953
    },
    {
      "epoch": 0.6697422898183355,
      "grad_norm": 0.17876555025577545,
      "learning_rate": 8.399836838242479e-06,
      "loss": 0.0449223518371582,
      "step": 4954
    },
    {
      "epoch": 0.6698774820447825,
      "grad_norm": 0.34702861309051514,
      "learning_rate": 8.393691280535143e-06,
      "loss": 0.07400703430175781,
      "step": 4955
    },
    {
      "epoch": 0.6700126742712295,
      "grad_norm": 0.21426530182361603,
      "learning_rate": 8.387547098296516e-06,
      "loss": 0.05273914337158203,
      "step": 4956
    },
    {
      "epoch": 0.6701478664976764,
      "grad_norm": 0.24043278396129608,
      "learning_rate": 8.38140429280583e-06,
      "loss": 0.05688631534576416,
      "step": 4957
    },
    {
      "epoch": 0.6702830587241234,
      "grad_norm": 0.5707756280899048,
      "learning_rate": 8.375262865342073e-06,
      "loss": 0.0760960578918457,
      "step": 4958
    },
    {
      "epoch": 0.6704182509505704,
      "grad_norm": 0.2311011552810669,
      "learning_rate": 8.36912281718391e-06,
      "loss": 0.061896324157714844,
      "step": 4959
    },
    {
      "epoch": 0.6705534431770174,
      "grad_norm": 0.23181913793087006,
      "learning_rate": 8.362984149609748e-06,
      "loss": 0.0648641586303711,
      "step": 4960
    },
    {
      "epoch": 0.6706886354034644,
      "grad_norm": 0.47086256742477417,
      "learning_rate": 8.356846863897672e-06,
      "loss": 0.06805562973022461,
      "step": 4961
    },
    {
      "epoch": 0.6708238276299113,
      "grad_norm": 0.3221588730812073,
      "learning_rate": 8.350710961325498e-06,
      "loss": 0.06921911239624023,
      "step": 4962
    },
    {
      "epoch": 0.6709590198563583,
      "grad_norm": 0.19285617768764496,
      "learning_rate": 8.344576443170768e-06,
      "loss": 0.040163516998291016,
      "step": 4963
    },
    {
      "epoch": 0.6710942120828053,
      "grad_norm": 0.46340784430503845,
      "learning_rate": 8.338443310710708e-06,
      "loss": 0.07517766952514648,
      "step": 4964
    },
    {
      "epoch": 0.6712294043092523,
      "grad_norm": 0.3034537732601166,
      "learning_rate": 8.332311565222284e-06,
      "loss": 0.0892934799194336,
      "step": 4965
    },
    {
      "epoch": 0.6713645965356992,
      "grad_norm": 0.5148158669471741,
      "learning_rate": 8.326181207982145e-06,
      "loss": 0.0970768928527832,
      "step": 4966
    },
    {
      "epoch": 0.6714997887621462,
      "grad_norm": 0.2379639893770218,
      "learning_rate": 8.32005224026666e-06,
      "loss": 0.05786752700805664,
      "step": 4967
    },
    {
      "epoch": 0.6716349809885932,
      "grad_norm": 0.5114599466323853,
      "learning_rate": 8.313924663351927e-06,
      "loss": 0.0935206413269043,
      "step": 4968
    },
    {
      "epoch": 0.6717701732150402,
      "grad_norm": 0.31471630930900574,
      "learning_rate": 8.307798478513733e-06,
      "loss": 0.0633854866027832,
      "step": 4969
    },
    {
      "epoch": 0.6719053654414872,
      "grad_norm": 0.3404069244861603,
      "learning_rate": 8.301673687027583e-06,
      "loss": 0.050209999084472656,
      "step": 4970
    },
    {
      "epoch": 0.6720405576679341,
      "grad_norm": 0.12643036246299744,
      "learning_rate": 8.295550290168692e-06,
      "loss": 0.027638673782348633,
      "step": 4971
    },
    {
      "epoch": 0.6721757498943811,
      "grad_norm": 0.2427472323179245,
      "learning_rate": 8.289428289211977e-06,
      "loss": 0.05264449119567871,
      "step": 4972
    },
    {
      "epoch": 0.6723109421208281,
      "grad_norm": 0.2014956772327423,
      "learning_rate": 8.283307685432083e-06,
      "loss": 0.03870415687561035,
      "step": 4973
    },
    {
      "epoch": 0.6724461343472751,
      "grad_norm": 0.24342061579227448,
      "learning_rate": 8.277188480103348e-06,
      "loss": 0.06587362289428711,
      "step": 4974
    },
    {
      "epoch": 0.672581326573722,
      "grad_norm": 0.5443671941757202,
      "learning_rate": 8.271070674499821e-06,
      "loss": 0.07341784238815308,
      "step": 4975
    },
    {
      "epoch": 0.672716518800169,
      "grad_norm": 0.3734854757785797,
      "learning_rate": 8.264954269895262e-06,
      "loss": 0.058185577392578125,
      "step": 4976
    },
    {
      "epoch": 0.672851711026616,
      "grad_norm": 0.29039379954338074,
      "learning_rate": 8.258839267563134e-06,
      "loss": 0.07250142097473145,
      "step": 4977
    },
    {
      "epoch": 0.672986903253063,
      "grad_norm": 0.1804579645395279,
      "learning_rate": 8.252725668776623e-06,
      "loss": 0.04205751419067383,
      "step": 4978
    },
    {
      "epoch": 0.67312209547951,
      "grad_norm": 0.35668709874153137,
      "learning_rate": 8.24661347480861e-06,
      "loss": 0.059288978576660156,
      "step": 4979
    },
    {
      "epoch": 0.6732572877059569,
      "grad_norm": 0.6256442666053772,
      "learning_rate": 8.240502686931686e-06,
      "loss": 0.0910944938659668,
      "step": 4980
    },
    {
      "epoch": 0.6733924799324039,
      "grad_norm": 0.36551544070243835,
      "learning_rate": 8.234393306418148e-06,
      "loss": 0.07981348037719727,
      "step": 4981
    },
    {
      "epoch": 0.6735276721588509,
      "grad_norm": 0.2094249129295349,
      "learning_rate": 8.228285334539995e-06,
      "loss": 0.05939030647277832,
      "step": 4982
    },
    {
      "epoch": 0.6736628643852979,
      "grad_norm": 0.33449500799179077,
      "learning_rate": 8.22217877256896e-06,
      "loss": 0.0742940902709961,
      "step": 4983
    },
    {
      "epoch": 0.6737980566117449,
      "grad_norm": 0.2808586657047272,
      "learning_rate": 8.216073621776436e-06,
      "loss": 0.055348336696624756,
      "step": 4984
    },
    {
      "epoch": 0.6739332488381918,
      "grad_norm": 0.3043719530105591,
      "learning_rate": 8.209969883433566e-06,
      "loss": 0.056771039962768555,
      "step": 4985
    },
    {
      "epoch": 0.6740684410646388,
      "grad_norm": 0.3775617480278015,
      "learning_rate": 8.203867558811177e-06,
      "loss": 0.05510234832763672,
      "step": 4986
    },
    {
      "epoch": 0.6742036332910858,
      "grad_norm": 0.5224157571792603,
      "learning_rate": 8.197766649179795e-06,
      "loss": 0.10012221336364746,
      "step": 4987
    },
    {
      "epoch": 0.6743388255175328,
      "grad_norm": 0.34463098645210266,
      "learning_rate": 8.191667155809684e-06,
      "loss": 0.06544208526611328,
      "step": 4988
    },
    {
      "epoch": 0.6744740177439797,
      "grad_norm": 0.21582862734794617,
      "learning_rate": 8.185569079970764e-06,
      "loss": 0.04866981506347656,
      "step": 4989
    },
    {
      "epoch": 0.6746092099704267,
      "grad_norm": 0.8155859708786011,
      "learning_rate": 8.179472422932709e-06,
      "loss": 0.062165021896362305,
      "step": 4990
    },
    {
      "epoch": 0.6747444021968737,
      "grad_norm": 0.3104049861431122,
      "learning_rate": 8.17337718596486e-06,
      "loss": 0.052139997482299805,
      "step": 4991
    },
    {
      "epoch": 0.6748795944233207,
      "grad_norm": 0.3490195572376251,
      "learning_rate": 8.167283370336295e-06,
      "loss": 0.06679701805114746,
      "step": 4992
    },
    {
      "epoch": 0.6750147866497677,
      "grad_norm": 0.3360089957714081,
      "learning_rate": 8.161190977315766e-06,
      "loss": 0.06595468521118164,
      "step": 4993
    },
    {
      "epoch": 0.6751499788762146,
      "grad_norm": 0.18689323961734772,
      "learning_rate": 8.155100008171736e-06,
      "loss": 0.051087379455566406,
      "step": 4994
    },
    {
      "epoch": 0.6752851711026616,
      "grad_norm": 0.23811936378479004,
      "learning_rate": 8.149010464172392e-06,
      "loss": 0.048638343811035156,
      "step": 4995
    },
    {
      "epoch": 0.6754203633291086,
      "grad_norm": 0.24320784211158752,
      "learning_rate": 8.142922346585597e-06,
      "loss": 0.05586671829223633,
      "step": 4996
    },
    {
      "epoch": 0.6755555555555556,
      "grad_norm": 0.43750205636024475,
      "learning_rate": 8.13683565667895e-06,
      "loss": 0.09529876708984375,
      "step": 4997
    },
    {
      "epoch": 0.6756907477820026,
      "grad_norm": 0.23445385694503784,
      "learning_rate": 8.13075039571971e-06,
      "loss": 0.05383414030075073,
      "step": 4998
    },
    {
      "epoch": 0.6758259400084495,
      "grad_norm": 0.5510262846946716,
      "learning_rate": 8.124666564974864e-06,
      "loss": 0.06421995162963867,
      "step": 4999
    },
    {
      "epoch": 0.6759611322348965,
      "grad_norm": 0.40528297424316406,
      "learning_rate": 8.11858416571111e-06,
      "loss": 0.08857870101928711,
      "step": 5000
    },
    {
      "epoch": 0.6760963244613435,
      "grad_norm": 0.1624949872493744,
      "learning_rate": 8.112503199194821e-06,
      "loss": 0.05199551582336426,
      "step": 5001
    },
    {
      "epoch": 0.6762315166877905,
      "grad_norm": 0.26256316900253296,
      "learning_rate": 8.106423666692108e-06,
      "loss": 0.04989594221115112,
      "step": 5002
    },
    {
      "epoch": 0.6763667089142374,
      "grad_norm": 0.18374936282634735,
      "learning_rate": 8.100345569468742e-06,
      "loss": 0.05907416343688965,
      "step": 5003
    },
    {
      "epoch": 0.6765019011406844,
      "grad_norm": 0.4773852527141571,
      "learning_rate": 8.094268908790215e-06,
      "loss": 0.08037519454956055,
      "step": 5004
    },
    {
      "epoch": 0.6766370933671314,
      "grad_norm": 0.2970427870750427,
      "learning_rate": 8.088193685921733e-06,
      "loss": 0.0757899284362793,
      "step": 5005
    },
    {
      "epoch": 0.6767722855935784,
      "grad_norm": 0.2060684859752655,
      "learning_rate": 8.082119902128185e-06,
      "loss": 0.04117774963378906,
      "step": 5006
    },
    {
      "epoch": 0.6769074778200254,
      "grad_norm": 0.21486613154411316,
      "learning_rate": 8.076047558674164e-06,
      "loss": 0.05521678924560547,
      "step": 5007
    },
    {
      "epoch": 0.6770426700464723,
      "grad_norm": 0.4274241626262665,
      "learning_rate": 8.069976656823964e-06,
      "loss": 0.08288079500198364,
      "step": 5008
    },
    {
      "epoch": 0.6771778622729193,
      "grad_norm": 0.2574945092201233,
      "learning_rate": 8.063907197841574e-06,
      "loss": 0.056850433349609375,
      "step": 5009
    },
    {
      "epoch": 0.6773130544993663,
      "grad_norm": 0.21665945649147034,
      "learning_rate": 8.057839182990698e-06,
      "loss": 0.041300177574157715,
      "step": 5010
    },
    {
      "epoch": 0.6774482467258133,
      "grad_norm": 0.18303847312927246,
      "learning_rate": 8.051772613534725e-06,
      "loss": 0.0461878776550293,
      "step": 5011
    },
    {
      "epoch": 0.6775834389522603,
      "grad_norm": 0.21712854504585266,
      "learning_rate": 8.045707490736745e-06,
      "loss": 0.05544382333755493,
      "step": 5012
    },
    {
      "epoch": 0.6777186311787072,
      "grad_norm": 0.6855584979057312,
      "learning_rate": 8.039643815859552e-06,
      "loss": 0.10220527648925781,
      "step": 5013
    },
    {
      "epoch": 0.6778538234051542,
      "grad_norm": 0.5897433161735535,
      "learning_rate": 8.033581590165627e-06,
      "loss": 0.08015584945678711,
      "step": 5014
    },
    {
      "epoch": 0.6779890156316012,
      "grad_norm": 0.5282954573631287,
      "learning_rate": 8.027520814917175e-06,
      "loss": 0.09360551834106445,
      "step": 5015
    },
    {
      "epoch": 0.6781242078580482,
      "grad_norm": 0.32223278284072876,
      "learning_rate": 8.021461491376064e-06,
      "loss": 0.06849193572998047,
      "step": 5016
    },
    {
      "epoch": 0.6782594000844951,
      "grad_norm": 0.2557280659675598,
      "learning_rate": 8.015403620803885e-06,
      "loss": 0.05202054977416992,
      "step": 5017
    },
    {
      "epoch": 0.6783945923109421,
      "grad_norm": 0.5934728980064392,
      "learning_rate": 8.009347204461922e-06,
      "loss": 0.1055903434753418,
      "step": 5018
    },
    {
      "epoch": 0.6785297845373891,
      "grad_norm": 0.3711104393005371,
      "learning_rate": 8.003292243611143e-06,
      "loss": 0.07085001468658447,
      "step": 5019
    },
    {
      "epoch": 0.6786649767638361,
      "grad_norm": 0.2455320805311203,
      "learning_rate": 7.99723873951224e-06,
      "loss": 0.055066585540771484,
      "step": 5020
    },
    {
      "epoch": 0.6788001689902831,
      "grad_norm": 0.3658488690853119,
      "learning_rate": 7.991186693425563e-06,
      "loss": 0.06515157222747803,
      "step": 5021
    },
    {
      "epoch": 0.67893536121673,
      "grad_norm": 0.27166593074798584,
      "learning_rate": 7.9851361066112e-06,
      "loss": 0.060401201248168945,
      "step": 5022
    },
    {
      "epoch": 0.679070553443177,
      "grad_norm": 0.17168448865413666,
      "learning_rate": 7.979086980328907e-06,
      "loss": 0.03423810005187988,
      "step": 5023
    },
    {
      "epoch": 0.679205745669624,
      "grad_norm": 0.15034841001033783,
      "learning_rate": 7.973039315838137e-06,
      "loss": 0.03233814239501953,
      "step": 5024
    },
    {
      "epoch": 0.679340937896071,
      "grad_norm": 0.39001956582069397,
      "learning_rate": 7.966993114398067e-06,
      "loss": 0.07499301433563232,
      "step": 5025
    },
    {
      "epoch": 0.679476130122518,
      "grad_norm": 0.2622569799423218,
      "learning_rate": 7.960948377267524e-06,
      "loss": 0.0768585205078125,
      "step": 5026
    },
    {
      "epoch": 0.6796113223489649,
      "grad_norm": 0.31650036573410034,
      "learning_rate": 7.954905105705071e-06,
      "loss": 0.05341529846191406,
      "step": 5027
    },
    {
      "epoch": 0.6797465145754119,
      "grad_norm": 0.2731419801712036,
      "learning_rate": 7.948863300968938e-06,
      "loss": 0.060317277908325195,
      "step": 5028
    },
    {
      "epoch": 0.6798817068018589,
      "grad_norm": 0.3227729797363281,
      "learning_rate": 7.942822964317078e-06,
      "loss": 0.06663703918457031,
      "step": 5029
    },
    {
      "epoch": 0.6800168990283059,
      "grad_norm": 0.4316539168357849,
      "learning_rate": 7.936784097007105e-06,
      "loss": 0.07207369804382324,
      "step": 5030
    },
    {
      "epoch": 0.6801520912547528,
      "grad_norm": 0.35629063844680786,
      "learning_rate": 7.930746700296344e-06,
      "loss": 0.06932616233825684,
      "step": 5031
    },
    {
      "epoch": 0.6802872834811998,
      "grad_norm": 0.24202732741832733,
      "learning_rate": 7.924710775441822e-06,
      "loss": 0.057445526123046875,
      "step": 5032
    },
    {
      "epoch": 0.6804224757076468,
      "grad_norm": 0.25883346796035767,
      "learning_rate": 7.918676323700241e-06,
      "loss": 0.053176820278167725,
      "step": 5033
    },
    {
      "epoch": 0.6805576679340938,
      "grad_norm": 0.215104877948761,
      "learning_rate": 7.912643346328023e-06,
      "loss": 0.036556124687194824,
      "step": 5034
    },
    {
      "epoch": 0.6806928601605408,
      "grad_norm": 0.3205788731575012,
      "learning_rate": 7.906611844581251e-06,
      "loss": 0.04453611373901367,
      "step": 5035
    },
    {
      "epoch": 0.6808280523869877,
      "grad_norm": 0.33966442942619324,
      "learning_rate": 7.900581819715713e-06,
      "loss": 0.06298494338989258,
      "step": 5036
    },
    {
      "epoch": 0.6809632446134347,
      "grad_norm": 0.20713776350021362,
      "learning_rate": 7.894553272986901e-06,
      "loss": 0.06315851211547852,
      "step": 5037
    },
    {
      "epoch": 0.6810984368398817,
      "grad_norm": 0.31027498841285706,
      "learning_rate": 7.888526205649993e-06,
      "loss": 0.06762075424194336,
      "step": 5038
    },
    {
      "epoch": 0.6812336290663287,
      "grad_norm": 0.3930688202381134,
      "learning_rate": 7.882500618959849e-06,
      "loss": 0.07503199577331543,
      "step": 5039
    },
    {
      "epoch": 0.6813688212927757,
      "grad_norm": 0.2407899796962738,
      "learning_rate": 7.876476514171033e-06,
      "loss": 0.05850982666015625,
      "step": 5040
    },
    {
      "epoch": 0.6815040135192226,
      "grad_norm": 0.3437891900539398,
      "learning_rate": 7.870453892537788e-06,
      "loss": 0.05692625045776367,
      "step": 5041
    },
    {
      "epoch": 0.6816392057456696,
      "grad_norm": 0.2352464497089386,
      "learning_rate": 7.864432755314068e-06,
      "loss": 0.054657936096191406,
      "step": 5042
    },
    {
      "epoch": 0.6817743979721166,
      "grad_norm": 0.29077890515327454,
      "learning_rate": 7.858413103753499e-06,
      "loss": 0.05873918533325195,
      "step": 5043
    },
    {
      "epoch": 0.6819095901985636,
      "grad_norm": 0.6263427734375,
      "learning_rate": 7.852394939109408e-06,
      "loss": 0.06770062446594238,
      "step": 5044
    },
    {
      "epoch": 0.6820447824250105,
      "grad_norm": 0.29261234402656555,
      "learning_rate": 7.846378262634803e-06,
      "loss": 0.0538640022277832,
      "step": 5045
    },
    {
      "epoch": 0.6821799746514575,
      "grad_norm": 0.18424251675605774,
      "learning_rate": 7.840363075582385e-06,
      "loss": 0.04834175109863281,
      "step": 5046
    },
    {
      "epoch": 0.6823151668779045,
      "grad_norm": 0.3204640746116638,
      "learning_rate": 7.834349379204565e-06,
      "loss": 0.06139826774597168,
      "step": 5047
    },
    {
      "epoch": 0.6824503591043515,
      "grad_norm": 0.2606144845485687,
      "learning_rate": 7.828337174753411e-06,
      "loss": 0.0481867790222168,
      "step": 5048
    },
    {
      "epoch": 0.6825855513307985,
      "grad_norm": 0.2833598256111145,
      "learning_rate": 7.822326463480703e-06,
      "loss": 0.07135415077209473,
      "step": 5049
    },
    {
      "epoch": 0.6827207435572454,
      "grad_norm": 0.37110117077827454,
      "learning_rate": 7.816317246637901e-06,
      "loss": 0.06330126523971558,
      "step": 5050
    },
    {
      "epoch": 0.6828559357836924,
      "grad_norm": 0.28293147683143616,
      "learning_rate": 7.810309525476152e-06,
      "loss": 0.07352352142333984,
      "step": 5051
    },
    {
      "epoch": 0.6829911280101394,
      "grad_norm": 0.15649206936359406,
      "learning_rate": 7.804303301246311e-06,
      "loss": 0.036525845527648926,
      "step": 5052
    },
    {
      "epoch": 0.6831263202365864,
      "grad_norm": 0.23132957518100739,
      "learning_rate": 7.798298575198884e-06,
      "loss": 0.05054187774658203,
      "step": 5053
    },
    {
      "epoch": 0.6832615124630333,
      "grad_norm": 0.3829394280910492,
      "learning_rate": 7.792295348584103e-06,
      "loss": 0.05855894088745117,
      "step": 5054
    },
    {
      "epoch": 0.6833967046894803,
      "grad_norm": 0.39166468381881714,
      "learning_rate": 7.786293622651866e-06,
      "loss": 0.07469868659973145,
      "step": 5055
    },
    {
      "epoch": 0.6835318969159273,
      "grad_norm": 0.21898894011974335,
      "learning_rate": 7.78029339865176e-06,
      "loss": 0.06132245063781738,
      "step": 5056
    },
    {
      "epoch": 0.6836670891423743,
      "grad_norm": 0.29900994896888733,
      "learning_rate": 7.774294677833078e-06,
      "loss": 0.05816006660461426,
      "step": 5057
    },
    {
      "epoch": 0.6838022813688213,
      "grad_norm": 0.48562493920326233,
      "learning_rate": 7.768297461444766e-06,
      "loss": 0.09121572971343994,
      "step": 5058
    },
    {
      "epoch": 0.6839374735952682,
      "grad_norm": 0.2602877914905548,
      "learning_rate": 7.762301750735494e-06,
      "loss": 0.06448173522949219,
      "step": 5059
    },
    {
      "epoch": 0.6840726658217152,
      "grad_norm": 0.5472441911697388,
      "learning_rate": 7.756307546953592e-06,
      "loss": 0.061374664306640625,
      "step": 5060
    },
    {
      "epoch": 0.6842078580481622,
      "grad_norm": 0.2805687487125397,
      "learning_rate": 7.750314851347087e-06,
      "loss": 0.06966114044189453,
      "step": 5061
    },
    {
      "epoch": 0.6843430502746092,
      "grad_norm": 0.31316593289375305,
      "learning_rate": 7.74432366516369e-06,
      "loss": 0.08157825469970703,
      "step": 5062
    },
    {
      "epoch": 0.6844782425010562,
      "grad_norm": 0.27892011404037476,
      "learning_rate": 7.738333989650794e-06,
      "loss": 0.0483742356300354,
      "step": 5063
    },
    {
      "epoch": 0.6846134347275031,
      "grad_norm": 0.21512150764465332,
      "learning_rate": 7.732345826055487e-06,
      "loss": 0.04469776153564453,
      "step": 5064
    },
    {
      "epoch": 0.6847486269539501,
      "grad_norm": 0.24923019111156464,
      "learning_rate": 7.726359175624537e-06,
      "loss": 0.05147719383239746,
      "step": 5065
    },
    {
      "epoch": 0.6848838191803971,
      "grad_norm": 0.37343448400497437,
      "learning_rate": 7.720374039604395e-06,
      "loss": 0.07387542724609375,
      "step": 5066
    },
    {
      "epoch": 0.6850190114068441,
      "grad_norm": 0.21732354164123535,
      "learning_rate": 7.714390419241198e-06,
      "loss": 0.05731576681137085,
      "step": 5067
    },
    {
      "epoch": 0.685154203633291,
      "grad_norm": 0.42002561688423157,
      "learning_rate": 7.70840831578076e-06,
      "loss": 0.08342552185058594,
      "step": 5068
    },
    {
      "epoch": 0.685289395859738,
      "grad_norm": 0.18823859095573425,
      "learning_rate": 7.702427730468601e-06,
      "loss": 0.0579681396484375,
      "step": 5069
    },
    {
      "epoch": 0.685424588086185,
      "grad_norm": 0.2545880079269409,
      "learning_rate": 7.696448664549898e-06,
      "loss": 0.05961036682128906,
      "step": 5070
    },
    {
      "epoch": 0.685559780312632,
      "grad_norm": 0.2552105486392975,
      "learning_rate": 7.690471119269541e-06,
      "loss": 0.05908393859863281,
      "step": 5071
    },
    {
      "epoch": 0.685694972539079,
      "grad_norm": 0.19240762293338776,
      "learning_rate": 7.684495095872073e-06,
      "loss": 0.04995250701904297,
      "step": 5072
    },
    {
      "epoch": 0.6858301647655259,
      "grad_norm": 0.5534252524375916,
      "learning_rate": 7.678520595601728e-06,
      "loss": 0.08865594863891602,
      "step": 5073
    },
    {
      "epoch": 0.6859653569919729,
      "grad_norm": 0.2804965078830719,
      "learning_rate": 7.672547619702445e-06,
      "loss": 0.05677509307861328,
      "step": 5074
    },
    {
      "epoch": 0.6861005492184199,
      "grad_norm": 0.26258593797683716,
      "learning_rate": 7.666576169417823e-06,
      "loss": 0.0659494400024414,
      "step": 5075
    },
    {
      "epoch": 0.6862357414448669,
      "grad_norm": 0.21473030745983124,
      "learning_rate": 7.660606245991147e-06,
      "loss": 0.04909515380859375,
      "step": 5076
    },
    {
      "epoch": 0.6863709336713139,
      "grad_norm": 0.8554005026817322,
      "learning_rate": 7.654637850665393e-06,
      "loss": 0.10103797912597656,
      "step": 5077
    },
    {
      "epoch": 0.6865061258977608,
      "grad_norm": 0.32204896211624146,
      "learning_rate": 7.648670984683199e-06,
      "loss": 0.07646989822387695,
      "step": 5078
    },
    {
      "epoch": 0.6866413181242078,
      "grad_norm": 0.3911975920200348,
      "learning_rate": 7.642705649286916e-06,
      "loss": 0.057561635971069336,
      "step": 5079
    },
    {
      "epoch": 0.6867765103506548,
      "grad_norm": 0.15990619361400604,
      "learning_rate": 7.63674184571855e-06,
      "loss": 0.0347064733505249,
      "step": 5080
    },
    {
      "epoch": 0.6869117025771018,
      "grad_norm": 0.3655719757080078,
      "learning_rate": 7.630779575219797e-06,
      "loss": 0.06616973876953125,
      "step": 5081
    },
    {
      "epoch": 0.6870468948035487,
      "grad_norm": 0.13986630737781525,
      "learning_rate": 7.6248188390320344e-06,
      "loss": 0.03964686393737793,
      "step": 5082
    },
    {
      "epoch": 0.6871820870299957,
      "grad_norm": 0.32323238253593445,
      "learning_rate": 7.6188596383963135e-06,
      "loss": 0.06238126754760742,
      "step": 5083
    },
    {
      "epoch": 0.6873172792564427,
      "grad_norm": 0.2665664553642273,
      "learning_rate": 7.612901974553388e-06,
      "loss": 0.06083226203918457,
      "step": 5084
    },
    {
      "epoch": 0.6874524714828897,
      "grad_norm": 0.23771488666534424,
      "learning_rate": 7.606945848743653e-06,
      "loss": 0.04768180847167969,
      "step": 5085
    },
    {
      "epoch": 0.6875876637093367,
      "grad_norm": 0.18521255254745483,
      "learning_rate": 7.600991262207221e-06,
      "loss": 0.05417442321777344,
      "step": 5086
    },
    {
      "epoch": 0.6877228559357836,
      "grad_norm": 0.2142123281955719,
      "learning_rate": 7.595038216183867e-06,
      "loss": 0.059854984283447266,
      "step": 5087
    },
    {
      "epoch": 0.6878580481622306,
      "grad_norm": 0.15788207948207855,
      "learning_rate": 7.589086711913037e-06,
      "loss": 0.04033863544464111,
      "step": 5088
    },
    {
      "epoch": 0.6879932403886776,
      "grad_norm": 0.1700204759836197,
      "learning_rate": 7.583136750633885e-06,
      "loss": 0.04038369655609131,
      "step": 5089
    },
    {
      "epoch": 0.6881284326151247,
      "grad_norm": 0.26140329241752625,
      "learning_rate": 7.577188333585202e-06,
      "loss": 0.07221794128417969,
      "step": 5090
    },
    {
      "epoch": 0.6882636248415717,
      "grad_norm": 0.32084646821022034,
      "learning_rate": 7.5712414620054975e-06,
      "loss": 0.0688483715057373,
      "step": 5091
    },
    {
      "epoch": 0.6883988170680186,
      "grad_norm": 0.45135194063186646,
      "learning_rate": 7.565296137132935e-06,
      "loss": 0.0956869125366211,
      "step": 5092
    },
    {
      "epoch": 0.6885340092944656,
      "grad_norm": 0.37323784828186035,
      "learning_rate": 7.559352360205357e-06,
      "loss": 0.08292341232299805,
      "step": 5093
    },
    {
      "epoch": 0.6886692015209126,
      "grad_norm": 0.34570854902267456,
      "learning_rate": 7.553410132460308e-06,
      "loss": 0.06216931343078613,
      "step": 5094
    },
    {
      "epoch": 0.6888043937473596,
      "grad_norm": 0.5107104778289795,
      "learning_rate": 7.547469455134968e-06,
      "loss": 0.07967972755432129,
      "step": 5095
    },
    {
      "epoch": 0.6889395859738066,
      "grad_norm": 0.4013323485851288,
      "learning_rate": 7.541530329466236e-06,
      "loss": 0.09386730194091797,
      "step": 5096
    },
    {
      "epoch": 0.6890747782002535,
      "grad_norm": 0.46993210911750793,
      "learning_rate": 7.535592756690661e-06,
      "loss": 0.06214261054992676,
      "step": 5097
    },
    {
      "epoch": 0.6892099704267005,
      "grad_norm": 0.1623968631029129,
      "learning_rate": 7.52965673804448e-06,
      "loss": 0.03421461582183838,
      "step": 5098
    },
    {
      "epoch": 0.6893451626531475,
      "grad_norm": 0.23661763966083527,
      "learning_rate": 7.5237222747636025e-06,
      "loss": 0.05874800682067871,
      "step": 5099
    },
    {
      "epoch": 0.6894803548795945,
      "grad_norm": 0.33158764243125916,
      "learning_rate": 7.517789368083611e-06,
      "loss": 0.07294607162475586,
      "step": 5100
    },
    {
      "epoch": 0.6896155471060414,
      "grad_norm": 0.272876501083374,
      "learning_rate": 7.511858019239778e-06,
      "loss": 0.05669736862182617,
      "step": 5101
    },
    {
      "epoch": 0.6897507393324884,
      "grad_norm": 0.4327941834926605,
      "learning_rate": 7.505928229467038e-06,
      "loss": 0.07514095306396484,
      "step": 5102
    },
    {
      "epoch": 0.6898859315589354,
      "grad_norm": 0.333072692155838,
      "learning_rate": 7.500000000000004e-06,
      "loss": 0.049433231353759766,
      "step": 5103
    },
    {
      "epoch": 0.6900211237853824,
      "grad_norm": 0.3539844751358032,
      "learning_rate": 7.494073332072963e-06,
      "loss": 0.0708761215209961,
      "step": 5104
    },
    {
      "epoch": 0.6901563160118294,
      "grad_norm": 0.38296326994895935,
      "learning_rate": 7.488148226919877e-06,
      "loss": 0.06740236282348633,
      "step": 5105
    },
    {
      "epoch": 0.6902915082382763,
      "grad_norm": 0.2936798930168152,
      "learning_rate": 7.482224685774393e-06,
      "loss": 0.045949459075927734,
      "step": 5106
    },
    {
      "epoch": 0.6904267004647233,
      "grad_norm": 0.4355970323085785,
      "learning_rate": 7.4763027098698184e-06,
      "loss": 0.08402824401855469,
      "step": 5107
    },
    {
      "epoch": 0.6905618926911703,
      "grad_norm": 0.35099583864212036,
      "learning_rate": 7.470382300439143e-06,
      "loss": 0.07074165344238281,
      "step": 5108
    },
    {
      "epoch": 0.6906970849176173,
      "grad_norm": 0.2727399468421936,
      "learning_rate": 7.4644634587150225e-06,
      "loss": 0.06038612127304077,
      "step": 5109
    },
    {
      "epoch": 0.6908322771440643,
      "grad_norm": 0.1744815707206726,
      "learning_rate": 7.4585461859297906e-06,
      "loss": 0.04171609878540039,
      "step": 5110
    },
    {
      "epoch": 0.6909674693705112,
      "grad_norm": 0.3430871069431305,
      "learning_rate": 7.452630483315463e-06,
      "loss": 0.06779289245605469,
      "step": 5111
    },
    {
      "epoch": 0.6911026615969582,
      "grad_norm": 0.27155163884162903,
      "learning_rate": 7.4467163521037186e-06,
      "loss": 0.08106803894042969,
      "step": 5112
    },
    {
      "epoch": 0.6912378538234052,
      "grad_norm": 0.3969530761241913,
      "learning_rate": 7.440803793525907e-06,
      "loss": 0.09438371658325195,
      "step": 5113
    },
    {
      "epoch": 0.6913730460498522,
      "grad_norm": 0.2457677572965622,
      "learning_rate": 7.434892808813056e-06,
      "loss": 0.05474114418029785,
      "step": 5114
    },
    {
      "epoch": 0.6915082382762991,
      "grad_norm": 0.511020302772522,
      "learning_rate": 7.42898339919586e-06,
      "loss": 0.06138277053833008,
      "step": 5115
    },
    {
      "epoch": 0.6916434305027461,
      "grad_norm": 0.2703227698802948,
      "learning_rate": 7.423075565904698e-06,
      "loss": 0.0662994384765625,
      "step": 5116
    },
    {
      "epoch": 0.6917786227291931,
      "grad_norm": 0.34211331605911255,
      "learning_rate": 7.417169310169609e-06,
      "loss": 0.0668785572052002,
      "step": 5117
    },
    {
      "epoch": 0.6919138149556401,
      "grad_norm": 0.22787807881832123,
      "learning_rate": 7.411264633220305e-06,
      "loss": 0.04981422424316406,
      "step": 5118
    },
    {
      "epoch": 0.6920490071820871,
      "grad_norm": 0.3589753210544586,
      "learning_rate": 7.405361536286174e-06,
      "loss": 0.0652472972869873,
      "step": 5119
    },
    {
      "epoch": 0.692184199408534,
      "grad_norm": 0.29489415884017944,
      "learning_rate": 7.399460020596266e-06,
      "loss": 0.056148529052734375,
      "step": 5120
    },
    {
      "epoch": 0.692319391634981,
      "grad_norm": 0.4208012819290161,
      "learning_rate": 7.393560087379322e-06,
      "loss": 0.09299284219741821,
      "step": 5121
    },
    {
      "epoch": 0.692454583861428,
      "grad_norm": 0.2477555274963379,
      "learning_rate": 7.3876617378637195e-06,
      "loss": 0.05971670150756836,
      "step": 5122
    },
    {
      "epoch": 0.692589776087875,
      "grad_norm": 0.3015389144420624,
      "learning_rate": 7.381764973277543e-06,
      "loss": 0.05284738540649414,
      "step": 5123
    },
    {
      "epoch": 0.692724968314322,
      "grad_norm": 0.4995321035385132,
      "learning_rate": 7.375869794848525e-06,
      "loss": 0.07973992824554443,
      "step": 5124
    },
    {
      "epoch": 0.6928601605407689,
      "grad_norm": 0.924176812171936,
      "learning_rate": 7.3699762038040654e-06,
      "loss": 0.06792068481445312,
      "step": 5125
    },
    {
      "epoch": 0.6929953527672159,
      "grad_norm": 0.29413217306137085,
      "learning_rate": 7.364084201371261e-06,
      "loss": 0.07237696647644043,
      "step": 5126
    },
    {
      "epoch": 0.6931305449936629,
      "grad_norm": 0.4069044589996338,
      "learning_rate": 7.3581937887768334e-06,
      "loss": 0.10204553604125977,
      "step": 5127
    },
    {
      "epoch": 0.6932657372201099,
      "grad_norm": 0.4495379328727722,
      "learning_rate": 7.352304967247217e-06,
      "loss": 0.05169111490249634,
      "step": 5128
    },
    {
      "epoch": 0.6934009294465568,
      "grad_norm": 0.22065100073814392,
      "learning_rate": 7.346417738008487e-06,
      "loss": 0.046114444732666016,
      "step": 5129
    },
    {
      "epoch": 0.6935361216730038,
      "grad_norm": 0.31105419993400574,
      "learning_rate": 7.340532102286399e-06,
      "loss": 0.07995033264160156,
      "step": 5130
    },
    {
      "epoch": 0.6936713138994508,
      "grad_norm": 0.24137845635414124,
      "learning_rate": 7.3346480613063725e-06,
      "loss": 0.04705667495727539,
      "step": 5131
    },
    {
      "epoch": 0.6938065061258978,
      "grad_norm": 0.2521708607673645,
      "learning_rate": 7.328765616293491e-06,
      "loss": 0.06670522689819336,
      "step": 5132
    },
    {
      "epoch": 0.6939416983523448,
      "grad_norm": 0.3369907736778259,
      "learning_rate": 7.322884768472521e-06,
      "loss": 0.06294393539428711,
      "step": 5133
    },
    {
      "epoch": 0.6940768905787917,
      "grad_norm": 0.5765347480773926,
      "learning_rate": 7.317005519067881e-06,
      "loss": 0.08425617218017578,
      "step": 5134
    },
    {
      "epoch": 0.6942120828052387,
      "grad_norm": 0.3203013241291046,
      "learning_rate": 7.311127869303665e-06,
      "loss": 0.0680396556854248,
      "step": 5135
    },
    {
      "epoch": 0.6943472750316857,
      "grad_norm": 0.302909791469574,
      "learning_rate": 7.305251820403628e-06,
      "loss": 0.06524085998535156,
      "step": 5136
    },
    {
      "epoch": 0.6944824672581327,
      "grad_norm": 0.4749617874622345,
      "learning_rate": 7.299377373591188e-06,
      "loss": 0.05928301811218262,
      "step": 5137
    },
    {
      "epoch": 0.6946176594845797,
      "grad_norm": 0.5682387351989746,
      "learning_rate": 7.29350453008945e-06,
      "loss": 0.09905683994293213,
      "step": 5138
    },
    {
      "epoch": 0.6947528517110266,
      "grad_norm": 0.35577449202537537,
      "learning_rate": 7.287633291121166e-06,
      "loss": 0.06443309783935547,
      "step": 5139
    },
    {
      "epoch": 0.6948880439374736,
      "grad_norm": 0.2810805141925812,
      "learning_rate": 7.281763657908756e-06,
      "loss": 0.05539989471435547,
      "step": 5140
    },
    {
      "epoch": 0.6950232361639206,
      "grad_norm": 0.26290300488471985,
      "learning_rate": 7.275895631674313e-06,
      "loss": 0.05199289321899414,
      "step": 5141
    },
    {
      "epoch": 0.6951584283903676,
      "grad_norm": 0.17403098940849304,
      "learning_rate": 7.2700292136395826e-06,
      "loss": 0.03617501258850098,
      "step": 5142
    },
    {
      "epoch": 0.6952936206168145,
      "grad_norm": 0.2667497992515564,
      "learning_rate": 7.264164405025997e-06,
      "loss": 0.06101787090301514,
      "step": 5143
    },
    {
      "epoch": 0.6954288128432615,
      "grad_norm": 0.36639004945755005,
      "learning_rate": 7.2583012070546364e-06,
      "loss": 0.09653186798095703,
      "step": 5144
    },
    {
      "epoch": 0.6955640050697085,
      "grad_norm": 0.23468759655952454,
      "learning_rate": 7.252439620946247e-06,
      "loss": 0.05823396146297455,
      "step": 5145
    },
    {
      "epoch": 0.6956991972961555,
      "grad_norm": 0.26787322759628296,
      "learning_rate": 7.246579647921243e-06,
      "loss": 0.06267523765563965,
      "step": 5146
    },
    {
      "epoch": 0.6958343895226025,
      "grad_norm": 0.4143868684768677,
      "learning_rate": 7.240721289199699e-06,
      "loss": 0.061948537826538086,
      "step": 5147
    },
    {
      "epoch": 0.6959695817490494,
      "grad_norm": 0.5479666590690613,
      "learning_rate": 7.234864546001364e-06,
      "loss": 0.07314252853393555,
      "step": 5148
    },
    {
      "epoch": 0.6961047739754964,
      "grad_norm": 0.509655237197876,
      "learning_rate": 7.229009419545638e-06,
      "loss": 0.05440187454223633,
      "step": 5149
    },
    {
      "epoch": 0.6962399662019434,
      "grad_norm": 0.2680950164794922,
      "learning_rate": 7.223155911051593e-06,
      "loss": 0.04847145080566406,
      "step": 5150
    },
    {
      "epoch": 0.6963751584283904,
      "grad_norm": 0.3453781008720398,
      "learning_rate": 7.2173040217379575e-06,
      "loss": 0.0657196044921875,
      "step": 5151
    },
    {
      "epoch": 0.6965103506548374,
      "grad_norm": 0.1617542803287506,
      "learning_rate": 7.211453752823122e-06,
      "loss": 0.04308795928955078,
      "step": 5152
    },
    {
      "epoch": 0.6966455428812843,
      "grad_norm": 0.13926690816879272,
      "learning_rate": 7.205605105525161e-06,
      "loss": 0.03722953796386719,
      "step": 5153
    },
    {
      "epoch": 0.6967807351077313,
      "grad_norm": 0.35108423233032227,
      "learning_rate": 7.19975808106177e-06,
      "loss": 0.07530021667480469,
      "step": 5154
    },
    {
      "epoch": 0.6969159273341783,
      "grad_norm": 0.25188055634498596,
      "learning_rate": 7.193912680650346e-06,
      "loss": 0.06133103370666504,
      "step": 5155
    },
    {
      "epoch": 0.6970511195606253,
      "grad_norm": 0.26292234659194946,
      "learning_rate": 7.188068905507931e-06,
      "loss": 0.051703691482543945,
      "step": 5156
    },
    {
      "epoch": 0.6971863117870722,
      "grad_norm": 0.48832032084465027,
      "learning_rate": 7.182226756851223e-06,
      "loss": 0.0809781551361084,
      "step": 5157
    },
    {
      "epoch": 0.6973215040135192,
      "grad_norm": 2.010105848312378,
      "learning_rate": 7.176386235896603e-06,
      "loss": 0.06756019592285156,
      "step": 5158
    },
    {
      "epoch": 0.6974566962399662,
      "grad_norm": 0.16677618026733398,
      "learning_rate": 7.170547343860079e-06,
      "loss": 0.046690940856933594,
      "step": 5159
    },
    {
      "epoch": 0.6975918884664132,
      "grad_norm": 0.3117125928401947,
      "learning_rate": 7.164710081957355e-06,
      "loss": 0.07377815246582031,
      "step": 5160
    },
    {
      "epoch": 0.6977270806928602,
      "grad_norm": 0.2945198714733124,
      "learning_rate": 7.158874451403777e-06,
      "loss": 0.04975748062133789,
      "step": 5161
    },
    {
      "epoch": 0.6978622729193071,
      "grad_norm": 0.4614226520061493,
      "learning_rate": 7.15304045341435e-06,
      "loss": 0.08585166931152344,
      "step": 5162
    },
    {
      "epoch": 0.6979974651457541,
      "grad_norm": 0.30252423882484436,
      "learning_rate": 7.147208089203745e-06,
      "loss": 0.05503225326538086,
      "step": 5163
    },
    {
      "epoch": 0.6981326573722011,
      "grad_norm": 0.402742475271225,
      "learning_rate": 7.141377359986288e-06,
      "loss": 0.06532526016235352,
      "step": 5164
    },
    {
      "epoch": 0.6982678495986481,
      "grad_norm": 0.28376224637031555,
      "learning_rate": 7.135548266975978e-06,
      "loss": 0.0535738468170166,
      "step": 5165
    },
    {
      "epoch": 0.698403041825095,
      "grad_norm": 0.38947048783302307,
      "learning_rate": 7.129720811386456e-06,
      "loss": 0.06785106658935547,
      "step": 5166
    },
    {
      "epoch": 0.698538234051542,
      "grad_norm": 0.22312481701374054,
      "learning_rate": 7.12389499443103e-06,
      "loss": 0.06464958190917969,
      "step": 5167
    },
    {
      "epoch": 0.698673426277989,
      "grad_norm": 0.46874937415122986,
      "learning_rate": 7.118070817322668e-06,
      "loss": 0.08054065704345703,
      "step": 5168
    },
    {
      "epoch": 0.698808618504436,
      "grad_norm": 0.2867968678474426,
      "learning_rate": 7.1122482812739885e-06,
      "loss": 0.06430935859680176,
      "step": 5169
    },
    {
      "epoch": 0.698943810730883,
      "grad_norm": 0.44140592217445374,
      "learning_rate": 7.106427387497283e-06,
      "loss": 0.07148480415344238,
      "step": 5170
    },
    {
      "epoch": 0.6990790029573299,
      "grad_norm": 0.3615421652793884,
      "learning_rate": 7.10060813720449e-06,
      "loss": 0.07390451431274414,
      "step": 5171
    },
    {
      "epoch": 0.6992141951837769,
      "grad_norm": 0.3820439875125885,
      "learning_rate": 7.094790531607207e-06,
      "loss": 0.08645200729370117,
      "step": 5172
    },
    {
      "epoch": 0.6993493874102239,
      "grad_norm": 0.3789469301700592,
      "learning_rate": 7.088974571916692e-06,
      "loss": 0.07278966903686523,
      "step": 5173
    },
    {
      "epoch": 0.6994845796366709,
      "grad_norm": 0.2615392208099365,
      "learning_rate": 7.0831602593438515e-06,
      "loss": 0.04927694797515869,
      "step": 5174
    },
    {
      "epoch": 0.6996197718631179,
      "grad_norm": 0.20419496297836304,
      "learning_rate": 7.077347595099269e-06,
      "loss": 0.047860145568847656,
      "step": 5175
    },
    {
      "epoch": 0.6997549640895648,
      "grad_norm": 0.26666179299354553,
      "learning_rate": 7.071536580393166e-06,
      "loss": 0.05436229705810547,
      "step": 5176
    },
    {
      "epoch": 0.6998901563160118,
      "grad_norm": 0.2581121623516083,
      "learning_rate": 7.065727216435426e-06,
      "loss": 0.04289531707763672,
      "step": 5177
    },
    {
      "epoch": 0.7000253485424588,
      "grad_norm": 0.39488500356674194,
      "learning_rate": 7.05991950443559e-06,
      "loss": 0.08115530014038086,
      "step": 5178
    },
    {
      "epoch": 0.7001605407689058,
      "grad_norm": 0.3415122330188751,
      "learning_rate": 7.05411344560285e-06,
      "loss": 0.05090904235839844,
      "step": 5179
    },
    {
      "epoch": 0.7002957329953527,
      "grad_norm": 0.4603601098060608,
      "learning_rate": 7.048309041146069e-06,
      "loss": 0.07886600494384766,
      "step": 5180
    },
    {
      "epoch": 0.7004309252217997,
      "grad_norm": 0.2843720614910126,
      "learning_rate": 7.0425062922737495e-06,
      "loss": 0.06193208694458008,
      "step": 5181
    },
    {
      "epoch": 0.7005661174482467,
      "grad_norm": 0.4288102090358734,
      "learning_rate": 7.036705200194053e-06,
      "loss": 0.06223297119140625,
      "step": 5182
    },
    {
      "epoch": 0.7007013096746937,
      "grad_norm": 0.6508801579475403,
      "learning_rate": 7.0309057661148e-06,
      "loss": 0.09417963027954102,
      "step": 5183
    },
    {
      "epoch": 0.7008365019011407,
      "grad_norm": 0.28302961587905884,
      "learning_rate": 7.0251079912434565e-06,
      "loss": 0.0530397891998291,
      "step": 5184
    },
    {
      "epoch": 0.7009716941275876,
      "grad_norm": 0.3496130406856537,
      "learning_rate": 7.019311876787169e-06,
      "loss": 0.0873725414276123,
      "step": 5185
    },
    {
      "epoch": 0.7011068863540346,
      "grad_norm": 0.17822697758674622,
      "learning_rate": 7.013517423952696e-06,
      "loss": 0.0391998291015625,
      "step": 5186
    },
    {
      "epoch": 0.7012420785804816,
      "grad_norm": 0.4951348602771759,
      "learning_rate": 7.0077246339464904e-06,
      "loss": 0.1033172607421875,
      "step": 5187
    },
    {
      "epoch": 0.7013772708069286,
      "grad_norm": 0.31052571535110474,
      "learning_rate": 7.001933507974635e-06,
      "loss": 0.0532526969909668,
      "step": 5188
    },
    {
      "epoch": 0.7015124630333756,
      "grad_norm": 0.2784779667854309,
      "learning_rate": 6.996144047242868e-06,
      "loss": 0.057559967041015625,
      "step": 5189
    },
    {
      "epoch": 0.7016476552598225,
      "grad_norm": 0.2698589265346527,
      "learning_rate": 6.9903562529566044e-06,
      "loss": 0.05294942855834961,
      "step": 5190
    },
    {
      "epoch": 0.7017828474862695,
      "grad_norm": 0.5403467416763306,
      "learning_rate": 6.984570126320869e-06,
      "loss": 0.07762980461120605,
      "step": 5191
    },
    {
      "epoch": 0.7019180397127165,
      "grad_norm": 0.3751070201396942,
      "learning_rate": 6.978785668540384e-06,
      "loss": 0.07057070732116699,
      "step": 5192
    },
    {
      "epoch": 0.7020532319391635,
      "grad_norm": 0.5298959612846375,
      "learning_rate": 6.973002880819496e-06,
      "loss": 0.08171343803405762,
      "step": 5193
    },
    {
      "epoch": 0.7021884241656104,
      "grad_norm": 0.4307243525981903,
      "learning_rate": 6.96722176436221e-06,
      "loss": 0.08375358581542969,
      "step": 5194
    },
    {
      "epoch": 0.7023236163920574,
      "grad_norm": 0.2923785448074341,
      "learning_rate": 6.9614423203721975e-06,
      "loss": 0.0440557599067688,
      "step": 5195
    },
    {
      "epoch": 0.7024588086185044,
      "grad_norm": 0.3036232590675354,
      "learning_rate": 6.955664550052749e-06,
      "loss": 0.07849550247192383,
      "step": 5196
    },
    {
      "epoch": 0.7025940008449514,
      "grad_norm": 0.5459083914756775,
      "learning_rate": 6.949888454606847e-06,
      "loss": 0.08972203731536865,
      "step": 5197
    },
    {
      "epoch": 0.7027291930713984,
      "grad_norm": 0.3295241594314575,
      "learning_rate": 6.944114035237095e-06,
      "loss": 0.07895088195800781,
      "step": 5198
    },
    {
      "epoch": 0.7028643852978453,
      "grad_norm": 0.2893649935722351,
      "learning_rate": 6.93834129314576e-06,
      "loss": 0.0626974105834961,
      "step": 5199
    },
    {
      "epoch": 0.7029995775242923,
      "grad_norm": 0.22155827283859253,
      "learning_rate": 6.932570229534759e-06,
      "loss": 0.061110734939575195,
      "step": 5200
    },
    {
      "epoch": 0.7031347697507393,
      "grad_norm": 0.4836445152759552,
      "learning_rate": 6.9268008456056505e-06,
      "loss": 0.07233667373657227,
      "step": 5201
    },
    {
      "epoch": 0.7032699619771863,
      "grad_norm": 0.2334105670452118,
      "learning_rate": 6.921033142559664e-06,
      "loss": 0.05377340316772461,
      "step": 5202
    },
    {
      "epoch": 0.7034051542036333,
      "grad_norm": 0.39910081028938293,
      "learning_rate": 6.915267121597659e-06,
      "loss": 0.06854391098022461,
      "step": 5203
    },
    {
      "epoch": 0.7035403464300802,
      "grad_norm": 0.21269190311431885,
      "learning_rate": 6.909502783920153e-06,
      "loss": 0.04930448532104492,
      "step": 5204
    },
    {
      "epoch": 0.7036755386565272,
      "grad_norm": 0.23991501331329346,
      "learning_rate": 6.903740130727312e-06,
      "loss": 0.06479740142822266,
      "step": 5205
    },
    {
      "epoch": 0.7038107308829742,
      "grad_norm": 0.18447642028331757,
      "learning_rate": 6.8979791632189425e-06,
      "loss": 0.042063236236572266,
      "step": 5206
    },
    {
      "epoch": 0.7039459231094212,
      "grad_norm": 0.3013533055782318,
      "learning_rate": 6.892219882594523e-06,
      "loss": 0.05243110656738281,
      "step": 5207
    },
    {
      "epoch": 0.7040811153358681,
      "grad_norm": 0.28287580609321594,
      "learning_rate": 6.886462290053159e-06,
      "loss": 0.05308115482330322,
      "step": 5208
    },
    {
      "epoch": 0.7042163075623151,
      "grad_norm": 0.12992000579833984,
      "learning_rate": 6.880706386793614e-06,
      "loss": 0.03340864181518555,
      "step": 5209
    },
    {
      "epoch": 0.7043514997887621,
      "grad_norm": 0.2752667963504791,
      "learning_rate": 6.874952174014298e-06,
      "loss": 0.0632176399230957,
      "step": 5210
    },
    {
      "epoch": 0.7044866920152091,
      "grad_norm": 0.2495708018541336,
      "learning_rate": 6.8691996529132585e-06,
      "loss": 0.04654431343078613,
      "step": 5211
    },
    {
      "epoch": 0.7046218842416561,
      "grad_norm": 0.4489001929759979,
      "learning_rate": 6.863448824688217e-06,
      "loss": 0.07932806015014648,
      "step": 5212
    },
    {
      "epoch": 0.704757076468103,
      "grad_norm": 0.45992836356163025,
      "learning_rate": 6.857699690536521e-06,
      "loss": 0.08535480499267578,
      "step": 5213
    },
    {
      "epoch": 0.70489226869455,
      "grad_norm": 0.41505518555641174,
      "learning_rate": 6.8519522516551685e-06,
      "loss": 0.07698339223861694,
      "step": 5214
    },
    {
      "epoch": 0.705027460920997,
      "grad_norm": 0.25965675711631775,
      "learning_rate": 6.846206509240807e-06,
      "loss": 0.06387710571289062,
      "step": 5215
    },
    {
      "epoch": 0.705162653147444,
      "grad_norm": 0.28890877962112427,
      "learning_rate": 6.840462464489726e-06,
      "loss": 0.09001302719116211,
      "step": 5216
    },
    {
      "epoch": 0.705297845373891,
      "grad_norm": 0.19721145927906036,
      "learning_rate": 6.834720118597879e-06,
      "loss": 0.05830240249633789,
      "step": 5217
    },
    {
      "epoch": 0.7054330376003379,
      "grad_norm": 0.49042749404907227,
      "learning_rate": 6.828979472760846e-06,
      "loss": 0.07706403732299805,
      "step": 5218
    },
    {
      "epoch": 0.7055682298267849,
      "grad_norm": 0.24993939697742462,
      "learning_rate": 6.823240528173858e-06,
      "loss": 0.047280073165893555,
      "step": 5219
    },
    {
      "epoch": 0.7057034220532319,
      "grad_norm": 0.32656624913215637,
      "learning_rate": 6.817503286031797e-06,
      "loss": 0.06964492797851562,
      "step": 5220
    },
    {
      "epoch": 0.7058386142796789,
      "grad_norm": 0.3378008306026459,
      "learning_rate": 6.811767747529181e-06,
      "loss": 0.07908868789672852,
      "step": 5221
    },
    {
      "epoch": 0.7059738065061258,
      "grad_norm": 0.5229734778404236,
      "learning_rate": 6.806033913860195e-06,
      "loss": 0.08921527862548828,
      "step": 5222
    },
    {
      "epoch": 0.7061089987325728,
      "grad_norm": 0.3965039849281311,
      "learning_rate": 6.800301786218634e-06,
      "loss": 0.07672739028930664,
      "step": 5223
    },
    {
      "epoch": 0.7062441909590198,
      "grad_norm": 0.46000194549560547,
      "learning_rate": 6.794571365797971e-06,
      "loss": 0.07695484161376953,
      "step": 5224
    },
    {
      "epoch": 0.7063793831854668,
      "grad_norm": 0.2436547726392746,
      "learning_rate": 6.788842653791308e-06,
      "loss": 0.0603257417678833,
      "step": 5225
    },
    {
      "epoch": 0.7065145754119139,
      "grad_norm": 0.3595168888568878,
      "learning_rate": 6.7831156513913864e-06,
      "loss": 0.05218696594238281,
      "step": 5226
    },
    {
      "epoch": 0.7066497676383608,
      "grad_norm": 0.32203981280326843,
      "learning_rate": 6.777390359790614e-06,
      "loss": 0.06536483764648438,
      "step": 5227
    },
    {
      "epoch": 0.7067849598648078,
      "grad_norm": 0.30787691473960876,
      "learning_rate": 6.771666780181004e-06,
      "loss": 0.07740449905395508,
      "step": 5228
    },
    {
      "epoch": 0.7069201520912548,
      "grad_norm": 0.1736706644296646,
      "learning_rate": 6.765944913754258e-06,
      "loss": 0.03533220291137695,
      "step": 5229
    },
    {
      "epoch": 0.7070553443177018,
      "grad_norm": 0.3201066851615906,
      "learning_rate": 6.7602247617016885e-06,
      "loss": 0.08419656753540039,
      "step": 5230
    },
    {
      "epoch": 0.7071905365441488,
      "grad_norm": 0.3639947175979614,
      "learning_rate": 6.754506325214265e-06,
      "loss": 0.07221412658691406,
      "step": 5231
    },
    {
      "epoch": 0.7073257287705957,
      "grad_norm": 0.41238322854042053,
      "learning_rate": 6.748789605482593e-06,
      "loss": 0.09039163589477539,
      "step": 5232
    },
    {
      "epoch": 0.7074609209970427,
      "grad_norm": 0.21321308612823486,
      "learning_rate": 6.743074603696922e-06,
      "loss": 0.037832021713256836,
      "step": 5233
    },
    {
      "epoch": 0.7075961132234897,
      "grad_norm": 0.2813952565193176,
      "learning_rate": 6.737361321047155e-06,
      "loss": 0.05194282531738281,
      "step": 5234
    },
    {
      "epoch": 0.7077313054499367,
      "grad_norm": 0.4473142623901367,
      "learning_rate": 6.731649758722823e-06,
      "loss": 0.09396696090698242,
      "step": 5235
    },
    {
      "epoch": 0.7078664976763837,
      "grad_norm": 0.1799984872341156,
      "learning_rate": 6.725939917913102e-06,
      "loss": 0.043128252029418945,
      "step": 5236
    },
    {
      "epoch": 0.7080016899028306,
      "grad_norm": 0.2964251637458801,
      "learning_rate": 6.720231799806814e-06,
      "loss": 0.06183105707168579,
      "step": 5237
    },
    {
      "epoch": 0.7081368821292776,
      "grad_norm": 0.16865837574005127,
      "learning_rate": 6.7145254055924136e-06,
      "loss": 0.04227583110332489,
      "step": 5238
    },
    {
      "epoch": 0.7082720743557246,
      "grad_norm": 0.19701269268989563,
      "learning_rate": 6.70882073645801e-06,
      "loss": 0.04064035415649414,
      "step": 5239
    },
    {
      "epoch": 0.7084072665821716,
      "grad_norm": 0.13464172184467316,
      "learning_rate": 6.703117793591346e-06,
      "loss": 0.03891634941101074,
      "step": 5240
    },
    {
      "epoch": 0.7085424588086185,
      "grad_norm": 0.29263293743133545,
      "learning_rate": 6.6974165781798e-06,
      "loss": 0.0652322769165039,
      "step": 5241
    },
    {
      "epoch": 0.7086776510350655,
      "grad_norm": 0.18063101172447205,
      "learning_rate": 6.691717091410398e-06,
      "loss": 0.04186379909515381,
      "step": 5242
    },
    {
      "epoch": 0.7088128432615125,
      "grad_norm": 0.16540196537971497,
      "learning_rate": 6.686019334469797e-06,
      "loss": 0.038457632064819336,
      "step": 5243
    },
    {
      "epoch": 0.7089480354879595,
      "grad_norm": 0.3550252318382263,
      "learning_rate": 6.680323308544312e-06,
      "loss": 0.07416820526123047,
      "step": 5244
    },
    {
      "epoch": 0.7090832277144065,
      "grad_norm": 0.41331881284713745,
      "learning_rate": 6.674629014819879e-06,
      "loss": 0.06978464126586914,
      "step": 5245
    },
    {
      "epoch": 0.7092184199408534,
      "grad_norm": 0.28112414479255676,
      "learning_rate": 6.668936454482082e-06,
      "loss": 0.060790300369262695,
      "step": 5246
    },
    {
      "epoch": 0.7093536121673004,
      "grad_norm": 0.2451619654893875,
      "learning_rate": 6.6632456287161426e-06,
      "loss": 0.0528407096862793,
      "step": 5247
    },
    {
      "epoch": 0.7094888043937474,
      "grad_norm": 0.28554636240005493,
      "learning_rate": 6.657556538706914e-06,
      "loss": 0.06327342987060547,
      "step": 5248
    },
    {
      "epoch": 0.7096239966201944,
      "grad_norm": 0.20135733485221863,
      "learning_rate": 6.651869185638907e-06,
      "loss": 0.05432701110839844,
      "step": 5249
    },
    {
      "epoch": 0.7097591888466414,
      "grad_norm": 0.330646276473999,
      "learning_rate": 6.646183570696253e-06,
      "loss": 0.06869697570800781,
      "step": 5250
    },
    {
      "epoch": 0.7098943810730883,
      "grad_norm": 0.19352610409259796,
      "learning_rate": 6.6404996950627275e-06,
      "loss": 0.04934835433959961,
      "step": 5251
    },
    {
      "epoch": 0.7100295732995353,
      "grad_norm": 0.32562071084976196,
      "learning_rate": 6.634817559921744e-06,
      "loss": 0.08568143844604492,
      "step": 5252
    },
    {
      "epoch": 0.7101647655259823,
      "grad_norm": 0.14758622646331787,
      "learning_rate": 6.629137166456348e-06,
      "loss": 0.038468360900878906,
      "step": 5253
    },
    {
      "epoch": 0.7102999577524293,
      "grad_norm": 0.43937766551971436,
      "learning_rate": 6.623458515849244e-06,
      "loss": 0.07604408264160156,
      "step": 5254
    },
    {
      "epoch": 0.7104351499788762,
      "grad_norm": 0.5796530246734619,
      "learning_rate": 6.6177816092827354e-06,
      "loss": 0.08571100234985352,
      "step": 5255
    },
    {
      "epoch": 0.7105703422053232,
      "grad_norm": 0.5068479180335999,
      "learning_rate": 6.6121064479388e-06,
      "loss": 0.07922065258026123,
      "step": 5256
    },
    {
      "epoch": 0.7107055344317702,
      "grad_norm": 0.25678953528404236,
      "learning_rate": 6.606433032999031e-06,
      "loss": 0.06104302406311035,
      "step": 5257
    },
    {
      "epoch": 0.7108407266582172,
      "grad_norm": 0.4827907383441925,
      "learning_rate": 6.60076136564466e-06,
      "loss": 0.05416154861450195,
      "step": 5258
    },
    {
      "epoch": 0.7109759188846642,
      "grad_norm": 0.45612362027168274,
      "learning_rate": 6.595091447056574e-06,
      "loss": 0.07600259780883789,
      "step": 5259
    },
    {
      "epoch": 0.7111111111111111,
      "grad_norm": 0.3709104061126709,
      "learning_rate": 6.589423278415259e-06,
      "loss": 0.0860128402709961,
      "step": 5260
    },
    {
      "epoch": 0.7112463033375581,
      "grad_norm": 0.2493404597043991,
      "learning_rate": 6.583756860900872e-06,
      "loss": 0.05909276008605957,
      "step": 5261
    },
    {
      "epoch": 0.7113814955640051,
      "grad_norm": 0.2866962254047394,
      "learning_rate": 6.578092195693187e-06,
      "loss": 0.05163431167602539,
      "step": 5262
    },
    {
      "epoch": 0.7115166877904521,
      "grad_norm": 0.6155808568000793,
      "learning_rate": 6.572429283971614e-06,
      "loss": 0.0694810152053833,
      "step": 5263
    },
    {
      "epoch": 0.711651880016899,
      "grad_norm": 0.2943473160266876,
      "learning_rate": 6.566768126915215e-06,
      "loss": 0.05535316467285156,
      "step": 5264
    },
    {
      "epoch": 0.711787072243346,
      "grad_norm": 0.15070411562919617,
      "learning_rate": 6.561108725702653e-06,
      "loss": 0.038384437561035156,
      "step": 5265
    },
    {
      "epoch": 0.711922264469793,
      "grad_norm": 0.4607013463973999,
      "learning_rate": 6.555451081512262e-06,
      "loss": 0.07212257385253906,
      "step": 5266
    },
    {
      "epoch": 0.71205745669624,
      "grad_norm": 0.4438392221927643,
      "learning_rate": 6.549795195521988e-06,
      "loss": 0.09726285934448242,
      "step": 5267
    },
    {
      "epoch": 0.712192648922687,
      "grad_norm": 0.20698747038841248,
      "learning_rate": 6.544141068909416e-06,
      "loss": 0.05082368850708008,
      "step": 5268
    },
    {
      "epoch": 0.712327841149134,
      "grad_norm": 0.3001084625720978,
      "learning_rate": 6.5384887028517645e-06,
      "loss": 0.0547177791595459,
      "step": 5269
    },
    {
      "epoch": 0.7124630333755809,
      "grad_norm": 0.40025028586387634,
      "learning_rate": 6.532838098525883e-06,
      "loss": 0.08511734008789062,
      "step": 5270
    },
    {
      "epoch": 0.7125982256020279,
      "grad_norm": 0.35512030124664307,
      "learning_rate": 6.5271892571082655e-06,
      "loss": 0.06788158416748047,
      "step": 5271
    },
    {
      "epoch": 0.7127334178284749,
      "grad_norm": 0.46489977836608887,
      "learning_rate": 6.521542179775029e-06,
      "loss": 0.06070709228515625,
      "step": 5272
    },
    {
      "epoch": 0.7128686100549219,
      "grad_norm": 0.2922933101654053,
      "learning_rate": 6.515896867701924e-06,
      "loss": 0.05234217643737793,
      "step": 5273
    },
    {
      "epoch": 0.7130038022813688,
      "grad_norm": 0.2974027991294861,
      "learning_rate": 6.510253322064333e-06,
      "loss": 0.05164146423339844,
      "step": 5274
    },
    {
      "epoch": 0.7131389945078158,
      "grad_norm": 0.7776655554771423,
      "learning_rate": 6.504611544037267e-06,
      "loss": 0.11277198791503906,
      "step": 5275
    },
    {
      "epoch": 0.7132741867342628,
      "grad_norm": 0.33653420209884644,
      "learning_rate": 6.498971534795387e-06,
      "loss": 0.0775156021118164,
      "step": 5276
    },
    {
      "epoch": 0.7134093789607098,
      "grad_norm": 0.19908814132213593,
      "learning_rate": 6.493333295512965e-06,
      "loss": 0.04477423429489136,
      "step": 5277
    },
    {
      "epoch": 0.7135445711871568,
      "grad_norm": 0.25313863158226013,
      "learning_rate": 6.487696827363916e-06,
      "loss": 0.052074432373046875,
      "step": 5278
    },
    {
      "epoch": 0.7136797634136037,
      "grad_norm": 0.3654668629169464,
      "learning_rate": 6.48206213152178e-06,
      "loss": 0.061371803283691406,
      "step": 5279
    },
    {
      "epoch": 0.7138149556400507,
      "grad_norm": 0.2771454453468323,
      "learning_rate": 6.476429209159725e-06,
      "loss": 0.07045698165893555,
      "step": 5280
    },
    {
      "epoch": 0.7139501478664977,
      "grad_norm": 0.24215292930603027,
      "learning_rate": 6.470798061450568e-06,
      "loss": 0.052469611167907715,
      "step": 5281
    },
    {
      "epoch": 0.7140853400929447,
      "grad_norm": 0.2678535282611847,
      "learning_rate": 6.465168689566738e-06,
      "loss": 0.056267738342285156,
      "step": 5282
    },
    {
      "epoch": 0.7142205323193916,
      "grad_norm": 0.4750436246395111,
      "learning_rate": 6.4595410946803e-06,
      "loss": 0.08125638961791992,
      "step": 5283
    },
    {
      "epoch": 0.7143557245458386,
      "grad_norm": 0.6535443663597107,
      "learning_rate": 6.453915277962948e-06,
      "loss": 0.0846710205078125,
      "step": 5284
    },
    {
      "epoch": 0.7144909167722856,
      "grad_norm": 0.31681588292121887,
      "learning_rate": 6.4482912405860055e-06,
      "loss": 0.07419633865356445,
      "step": 5285
    },
    {
      "epoch": 0.7146261089987326,
      "grad_norm": 0.5306790471076965,
      "learning_rate": 6.442668983720434e-06,
      "loss": 0.07944530248641968,
      "step": 5286
    },
    {
      "epoch": 0.7147613012251796,
      "grad_norm": 0.42185264825820923,
      "learning_rate": 6.437048508536813e-06,
      "loss": 0.07938385009765625,
      "step": 5287
    },
    {
      "epoch": 0.7148964934516265,
      "grad_norm": 0.16323237121105194,
      "learning_rate": 6.431429816205357e-06,
      "loss": 0.03561758995056152,
      "step": 5288
    },
    {
      "epoch": 0.7150316856780735,
      "grad_norm": 0.29352566599845886,
      "learning_rate": 6.425812907895904e-06,
      "loss": 0.05605483055114746,
      "step": 5289
    },
    {
      "epoch": 0.7151668779045205,
      "grad_norm": 0.25459399819374084,
      "learning_rate": 6.420197784777925e-06,
      "loss": 0.054135799407958984,
      "step": 5290
    },
    {
      "epoch": 0.7153020701309675,
      "grad_norm": 0.4943426549434662,
      "learning_rate": 6.414584448020528e-06,
      "loss": 0.08211421966552734,
      "step": 5291
    },
    {
      "epoch": 0.7154372623574144,
      "grad_norm": 0.21288572251796722,
      "learning_rate": 6.408972898792423e-06,
      "loss": 0.06739640235900879,
      "step": 5292
    },
    {
      "epoch": 0.7155724545838614,
      "grad_norm": 0.18440675735473633,
      "learning_rate": 6.4033631382619766e-06,
      "loss": 0.04372048377990723,
      "step": 5293
    },
    {
      "epoch": 0.7157076468103084,
      "grad_norm": 0.3132110834121704,
      "learning_rate": 6.397755167597171e-06,
      "loss": 0.07593369483947754,
      "step": 5294
    },
    {
      "epoch": 0.7158428390367554,
      "grad_norm": 0.33786335587501526,
      "learning_rate": 6.392148987965603e-06,
      "loss": 0.05293989181518555,
      "step": 5295
    },
    {
      "epoch": 0.7159780312632024,
      "grad_norm": 0.46040597558021545,
      "learning_rate": 6.386544600534532e-06,
      "loss": 0.08098006248474121,
      "step": 5296
    },
    {
      "epoch": 0.7161132234896493,
      "grad_norm": 0.1991870105266571,
      "learning_rate": 6.3809420064707965e-06,
      "loss": 0.05335187911987305,
      "step": 5297
    },
    {
      "epoch": 0.7162484157160963,
      "grad_norm": 0.3220804035663605,
      "learning_rate": 6.375341206940902e-06,
      "loss": 0.04731142520904541,
      "step": 5298
    },
    {
      "epoch": 0.7163836079425433,
      "grad_norm": 0.9213958978652954,
      "learning_rate": 6.369742203110962e-06,
      "loss": 0.07104718685150146,
      "step": 5299
    },
    {
      "epoch": 0.7165188001689903,
      "grad_norm": 0.44197481870651245,
      "learning_rate": 6.364144996146716e-06,
      "loss": 0.05142831802368164,
      "step": 5300
    },
    {
      "epoch": 0.7166539923954373,
      "grad_norm": 0.32369789481163025,
      "learning_rate": 6.358549587213534e-06,
      "loss": 0.06956243515014648,
      "step": 5301
    },
    {
      "epoch": 0.7167891846218842,
      "grad_norm": 0.30617251992225647,
      "learning_rate": 6.352955977476405e-06,
      "loss": 0.05008745193481445,
      "step": 5302
    },
    {
      "epoch": 0.7169243768483312,
      "grad_norm": 0.31715476512908936,
      "learning_rate": 6.347364168099959e-06,
      "loss": 0.060420989990234375,
      "step": 5303
    },
    {
      "epoch": 0.7170595690747782,
      "grad_norm": 0.19058825075626373,
      "learning_rate": 6.341774160248435e-06,
      "loss": 0.03251791000366211,
      "step": 5304
    },
    {
      "epoch": 0.7171947613012252,
      "grad_norm": 0.38277342915534973,
      "learning_rate": 6.3361859550857e-06,
      "loss": 0.07580947875976562,
      "step": 5305
    },
    {
      "epoch": 0.7173299535276721,
      "grad_norm": 0.40562063455581665,
      "learning_rate": 6.330599553775252e-06,
      "loss": 0.07514238357543945,
      "step": 5306
    },
    {
      "epoch": 0.7174651457541191,
      "grad_norm": 0.4485914409160614,
      "learning_rate": 6.325014957480203e-06,
      "loss": 0.09454870223999023,
      "step": 5307
    },
    {
      "epoch": 0.7176003379805661,
      "grad_norm": 0.4483773410320282,
      "learning_rate": 6.319432167363305e-06,
      "loss": 0.09784317016601562,
      "step": 5308
    },
    {
      "epoch": 0.7177355302070131,
      "grad_norm": 0.32551896572113037,
      "learning_rate": 6.313851184586918e-06,
      "loss": 0.04758763313293457,
      "step": 5309
    },
    {
      "epoch": 0.7178707224334601,
      "grad_norm": 0.7177479267120361,
      "learning_rate": 6.308272010313037e-06,
      "loss": 0.09788131713867188,
      "step": 5310
    },
    {
      "epoch": 0.718005914659907,
      "grad_norm": 0.29879510402679443,
      "learning_rate": 6.302694645703273e-06,
      "loss": 0.0753021240234375,
      "step": 5311
    },
    {
      "epoch": 0.718141106886354,
      "grad_norm": 0.44537603855133057,
      "learning_rate": 6.297119091918857e-06,
      "loss": 0.08078575134277344,
      "step": 5312
    },
    {
      "epoch": 0.718276299112801,
      "grad_norm": 0.3085884749889374,
      "learning_rate": 6.2915453501206634e-06,
      "loss": 0.05964779853820801,
      "step": 5313
    },
    {
      "epoch": 0.718411491339248,
      "grad_norm": 0.33680471777915955,
      "learning_rate": 6.285973421469166e-06,
      "loss": 0.08896994590759277,
      "step": 5314
    },
    {
      "epoch": 0.718546683565695,
      "grad_norm": 0.2171669453382492,
      "learning_rate": 6.28040330712447e-06,
      "loss": 0.05547380447387695,
      "step": 5315
    },
    {
      "epoch": 0.7186818757921419,
      "grad_norm": 0.17646951973438263,
      "learning_rate": 6.274835008246304e-06,
      "loss": 0.05302882194519043,
      "step": 5316
    },
    {
      "epoch": 0.7188170680185889,
      "grad_norm": 0.32431310415267944,
      "learning_rate": 6.269268525994013e-06,
      "loss": 0.07574915885925293,
      "step": 5317
    },
    {
      "epoch": 0.7189522602450359,
      "grad_norm": 0.2873489260673523,
      "learning_rate": 6.263703861526578e-06,
      "loss": 0.04230654239654541,
      "step": 5318
    },
    {
      "epoch": 0.7190874524714829,
      "grad_norm": 0.33394187688827515,
      "learning_rate": 6.258141016002587e-06,
      "loss": 0.07281017303466797,
      "step": 5319
    },
    {
      "epoch": 0.7192226446979298,
      "grad_norm": 0.460060179233551,
      "learning_rate": 6.252579990580254e-06,
      "loss": 0.08744525909423828,
      "step": 5320
    },
    {
      "epoch": 0.7193578369243768,
      "grad_norm": 0.3462483882904053,
      "learning_rate": 6.247020786417412e-06,
      "loss": 0.04702150821685791,
      "step": 5321
    },
    {
      "epoch": 0.7194930291508238,
      "grad_norm": 0.1926223337650299,
      "learning_rate": 6.241463404671516e-06,
      "loss": 0.0530247688293457,
      "step": 5322
    },
    {
      "epoch": 0.7196282213772708,
      "grad_norm": 0.3362235724925995,
      "learning_rate": 6.235907846499655e-06,
      "loss": 0.06870841979980469,
      "step": 5323
    },
    {
      "epoch": 0.7197634136037178,
      "grad_norm": 0.25347286462783813,
      "learning_rate": 6.230354113058505e-06,
      "loss": 0.063079833984375,
      "step": 5324
    },
    {
      "epoch": 0.7198986058301647,
      "grad_norm": 0.2189210206270218,
      "learning_rate": 6.2248022055044e-06,
      "loss": 0.04768264293670654,
      "step": 5325
    },
    {
      "epoch": 0.7200337980566117,
      "grad_norm": 0.30733242630958557,
      "learning_rate": 6.219252124993271e-06,
      "loss": 0.05814337730407715,
      "step": 5326
    },
    {
      "epoch": 0.7201689902830587,
      "grad_norm": 0.29368776082992554,
      "learning_rate": 6.213703872680668e-06,
      "loss": 0.07378196716308594,
      "step": 5327
    },
    {
      "epoch": 0.7203041825095057,
      "grad_norm": 0.3042683005332947,
      "learning_rate": 6.208157449721785e-06,
      "loss": 0.06443142890930176,
      "step": 5328
    },
    {
      "epoch": 0.7204393747359527,
      "grad_norm": 0.42213401198387146,
      "learning_rate": 6.202612857271393e-06,
      "loss": 0.07448077201843262,
      "step": 5329
    },
    {
      "epoch": 0.7205745669623996,
      "grad_norm": 0.41049185395240784,
      "learning_rate": 6.197070096483923e-06,
      "loss": 0.06332707405090332,
      "step": 5330
    },
    {
      "epoch": 0.7207097591888466,
      "grad_norm": 0.31218448281288147,
      "learning_rate": 6.191529168513403e-06,
      "loss": 0.06288552284240723,
      "step": 5331
    },
    {
      "epoch": 0.7208449514152936,
      "grad_norm": 0.2701214849948883,
      "learning_rate": 6.1859900745134755e-06,
      "loss": 0.057013511657714844,
      "step": 5332
    },
    {
      "epoch": 0.7209801436417406,
      "grad_norm": 0.2745994031429291,
      "learning_rate": 6.180452815637429e-06,
      "loss": 0.05720233917236328,
      "step": 5333
    },
    {
      "epoch": 0.7211153358681875,
      "grad_norm": 0.39080315828323364,
      "learning_rate": 6.174917393038126e-06,
      "loss": 0.06616663932800293,
      "step": 5334
    },
    {
      "epoch": 0.7212505280946345,
      "grad_norm": 0.3375305235385895,
      "learning_rate": 6.169383807868088e-06,
      "loss": 0.083038330078125,
      "step": 5335
    },
    {
      "epoch": 0.7213857203210815,
      "grad_norm": 0.2510044574737549,
      "learning_rate": 6.163852061279432e-06,
      "loss": 0.053519248962402344,
      "step": 5336
    },
    {
      "epoch": 0.7215209125475285,
      "grad_norm": 0.2927513122558594,
      "learning_rate": 6.158322154423897e-06,
      "loss": 0.04729747772216797,
      "step": 5337
    },
    {
      "epoch": 0.7216561047739755,
      "grad_norm": 0.5078144073486328,
      "learning_rate": 6.15279408845284e-06,
      "loss": 0.06560850143432617,
      "step": 5338
    },
    {
      "epoch": 0.7217912970004224,
      "grad_norm": 0.3834516704082489,
      "learning_rate": 6.147267864517226e-06,
      "loss": 0.05035281181335449,
      "step": 5339
    },
    {
      "epoch": 0.7219264892268694,
      "grad_norm": 0.3078523576259613,
      "learning_rate": 6.141743483767658e-06,
      "loss": 0.0785684585571289,
      "step": 5340
    },
    {
      "epoch": 0.7220616814533164,
      "grad_norm": 0.5647321343421936,
      "learning_rate": 6.136220947354333e-06,
      "loss": 0.11573410034179688,
      "step": 5341
    },
    {
      "epoch": 0.7221968736797634,
      "grad_norm": 0.3144357204437256,
      "learning_rate": 6.130700256427075e-06,
      "loss": 0.07591438293457031,
      "step": 5342
    },
    {
      "epoch": 0.7223320659062104,
      "grad_norm": 0.34056612849235535,
      "learning_rate": 6.1251814121353204e-06,
      "loss": 0.06491851806640625,
      "step": 5343
    },
    {
      "epoch": 0.7224672581326573,
      "grad_norm": 0.18165573477745056,
      "learning_rate": 6.1196644156281175e-06,
      "loss": 0.043727874755859375,
      "step": 5344
    },
    {
      "epoch": 0.7226024503591043,
      "grad_norm": 0.34186694025993347,
      "learning_rate": 6.114149268054143e-06,
      "loss": 0.07252335548400879,
      "step": 5345
    },
    {
      "epoch": 0.7227376425855513,
      "grad_norm": 0.2078217715024948,
      "learning_rate": 6.108635970561679e-06,
      "loss": 0.0397036075592041,
      "step": 5346
    },
    {
      "epoch": 0.7228728348119983,
      "grad_norm": 0.5245160460472107,
      "learning_rate": 6.103124524298617e-06,
      "loss": 0.09609150886535645,
      "step": 5347
    },
    {
      "epoch": 0.7230080270384452,
      "grad_norm": 0.4445883333683014,
      "learning_rate": 6.097614930412475e-06,
      "loss": 0.07157659530639648,
      "step": 5348
    },
    {
      "epoch": 0.7231432192648922,
      "grad_norm": 0.2465265393257141,
      "learning_rate": 6.092107190050371e-06,
      "loss": 0.04505133628845215,
      "step": 5349
    },
    {
      "epoch": 0.7232784114913392,
      "grad_norm": 0.60282963514328,
      "learning_rate": 6.086601304359059e-06,
      "loss": 0.07806289196014404,
      "step": 5350
    },
    {
      "epoch": 0.7234136037177862,
      "grad_norm": 0.1989317238330841,
      "learning_rate": 6.081097274484887e-06,
      "loss": 0.058745503425598145,
      "step": 5351
    },
    {
      "epoch": 0.7235487959442332,
      "grad_norm": 0.28250420093536377,
      "learning_rate": 6.075595101573825e-06,
      "loss": 0.043283939361572266,
      "step": 5352
    },
    {
      "epoch": 0.7236839881706801,
      "grad_norm": 0.452950119972229,
      "learning_rate": 6.070094786771451e-06,
      "loss": 0.07983648777008057,
      "step": 5353
    },
    {
      "epoch": 0.7238191803971271,
      "grad_norm": 0.23428699374198914,
      "learning_rate": 6.06459633122296e-06,
      "loss": 0.046811580657958984,
      "step": 5354
    },
    {
      "epoch": 0.7239543726235741,
      "grad_norm": 0.2537497282028198,
      "learning_rate": 6.059099736073166e-06,
      "loss": 0.07736682891845703,
      "step": 5355
    },
    {
      "epoch": 0.7240895648500211,
      "grad_norm": 0.38944104313850403,
      "learning_rate": 6.0536050024664865e-06,
      "loss": 0.05591392517089844,
      "step": 5356
    },
    {
      "epoch": 0.724224757076468,
      "grad_norm": 0.35275816917419434,
      "learning_rate": 6.048112131546953e-06,
      "loss": 0.0651397705078125,
      "step": 5357
    },
    {
      "epoch": 0.724359949302915,
      "grad_norm": 0.3773675858974457,
      "learning_rate": 6.0426211244582105e-06,
      "loss": 0.07550448179244995,
      "step": 5358
    },
    {
      "epoch": 0.724495141529362,
      "grad_norm": 0.17066816985607147,
      "learning_rate": 6.03713198234351e-06,
      "loss": 0.04315590858459473,
      "step": 5359
    },
    {
      "epoch": 0.724630333755809,
      "grad_norm": 0.27874502539634705,
      "learning_rate": 6.0316447063457395e-06,
      "loss": 0.0776357650756836,
      "step": 5360
    },
    {
      "epoch": 0.724765525982256,
      "grad_norm": 0.285234659910202,
      "learning_rate": 6.026159297607356e-06,
      "loss": 0.06747627258300781,
      "step": 5361
    },
    {
      "epoch": 0.724900718208703,
      "grad_norm": 0.25053516030311584,
      "learning_rate": 6.020675757270466e-06,
      "loss": 0.05825948715209961,
      "step": 5362
    },
    {
      "epoch": 0.72503591043515,
      "grad_norm": 0.3004065155982971,
      "learning_rate": 6.015194086476766e-06,
      "loss": 0.043872833251953125,
      "step": 5363
    },
    {
      "epoch": 0.725171102661597,
      "grad_norm": 0.33169230818748474,
      "learning_rate": 6.009714286367565e-06,
      "loss": 0.06735634803771973,
      "step": 5364
    },
    {
      "epoch": 0.725306294888044,
      "grad_norm": 0.40811291337013245,
      "learning_rate": 6.004236358083802e-06,
      "loss": 0.07512474060058594,
      "step": 5365
    },
    {
      "epoch": 0.725441487114491,
      "grad_norm": 0.35243967175483704,
      "learning_rate": 5.998760302765989e-06,
      "loss": 0.06746244430541992,
      "step": 5366
    },
    {
      "epoch": 0.725576679340938,
      "grad_norm": 0.3048992156982422,
      "learning_rate": 5.993286121554289e-06,
      "loss": 0.05666673183441162,
      "step": 5367
    },
    {
      "epoch": 0.7257118715673849,
      "grad_norm": 0.3580392003059387,
      "learning_rate": 5.987813815588447e-06,
      "loss": 0.0666208267211914,
      "step": 5368
    },
    {
      "epoch": 0.7258470637938319,
      "grad_norm": 0.37930622696876526,
      "learning_rate": 5.982343386007827e-06,
      "loss": 0.08964824676513672,
      "step": 5369
    },
    {
      "epoch": 0.7259822560202789,
      "grad_norm": 0.2211666852235794,
      "learning_rate": 5.976874833951404e-06,
      "loss": 0.050545692443847656,
      "step": 5370
    },
    {
      "epoch": 0.7261174482467259,
      "grad_norm": 0.24112331867218018,
      "learning_rate": 5.971408160557751e-06,
      "loss": 0.05179929733276367,
      "step": 5371
    },
    {
      "epoch": 0.7262526404731728,
      "grad_norm": 0.3979681432247162,
      "learning_rate": 5.965943366965069e-06,
      "loss": 0.08137750625610352,
      "step": 5372
    },
    {
      "epoch": 0.7263878326996198,
      "grad_norm": 0.21101003885269165,
      "learning_rate": 5.960480454311155e-06,
      "loss": 0.05811119079589844,
      "step": 5373
    },
    {
      "epoch": 0.7265230249260668,
      "grad_norm": 0.32828155159950256,
      "learning_rate": 5.955019423733416e-06,
      "loss": 0.08378124237060547,
      "step": 5374
    },
    {
      "epoch": 0.7266582171525138,
      "grad_norm": 0.4171237349510193,
      "learning_rate": 5.949560276368866e-06,
      "loss": 0.09240245819091797,
      "step": 5375
    },
    {
      "epoch": 0.7267934093789608,
      "grad_norm": 0.24032852053642273,
      "learning_rate": 5.9441030133541235e-06,
      "loss": 0.07701969146728516,
      "step": 5376
    },
    {
      "epoch": 0.7269286016054077,
      "grad_norm": 0.24201998114585876,
      "learning_rate": 5.938647635825432e-06,
      "loss": 0.05319392681121826,
      "step": 5377
    },
    {
      "epoch": 0.7270637938318547,
      "grad_norm": 0.332367867231369,
      "learning_rate": 5.933194144918623e-06,
      "loss": 0.060341596603393555,
      "step": 5378
    },
    {
      "epoch": 0.7271989860583017,
      "grad_norm": 0.3521791100502014,
      "learning_rate": 5.927742541769142e-06,
      "loss": 0.04864215850830078,
      "step": 5379
    },
    {
      "epoch": 0.7273341782847487,
      "grad_norm": 0.35975217819213867,
      "learning_rate": 5.9222928275120445e-06,
      "loss": 0.05352163314819336,
      "step": 5380
    },
    {
      "epoch": 0.7274693705111956,
      "grad_norm": 0.2611422836780548,
      "learning_rate": 5.916845003281983e-06,
      "loss": 0.061258792877197266,
      "step": 5381
    },
    {
      "epoch": 0.7276045627376426,
      "grad_norm": 0.3710670471191406,
      "learning_rate": 5.911399070213234e-06,
      "loss": 0.07865190505981445,
      "step": 5382
    },
    {
      "epoch": 0.7277397549640896,
      "grad_norm": 0.25374647974967957,
      "learning_rate": 5.905955029439665e-06,
      "loss": 0.06693243980407715,
      "step": 5383
    },
    {
      "epoch": 0.7278749471905366,
      "grad_norm": 0.688390851020813,
      "learning_rate": 5.900512882094754e-06,
      "loss": 0.10462188720703125,
      "step": 5384
    },
    {
      "epoch": 0.7280101394169836,
      "grad_norm": 0.3552984595298767,
      "learning_rate": 5.8950726293115855e-06,
      "loss": 0.059075117111206055,
      "step": 5385
    },
    {
      "epoch": 0.7281453316434305,
      "grad_norm": 0.28323695063591003,
      "learning_rate": 5.889634272222844e-06,
      "loss": 0.07794666290283203,
      "step": 5386
    },
    {
      "epoch": 0.7282805238698775,
      "grad_norm": 0.2585649788379669,
      "learning_rate": 5.8841978119608345e-06,
      "loss": 0.04220008850097656,
      "step": 5387
    },
    {
      "epoch": 0.7284157160963245,
      "grad_norm": 0.27961984276771545,
      "learning_rate": 5.878763249657452e-06,
      "loss": 0.050080060958862305,
      "step": 5388
    },
    {
      "epoch": 0.7285509083227715,
      "grad_norm": 0.30263832211494446,
      "learning_rate": 5.873330586444202e-06,
      "loss": 0.05263853073120117,
      "step": 5389
    },
    {
      "epoch": 0.7286861005492185,
      "grad_norm": 0.2795065939426422,
      "learning_rate": 5.867899823452193e-06,
      "loss": 0.0766400694847107,
      "step": 5390
    },
    {
      "epoch": 0.7288212927756654,
      "grad_norm": 0.3725234270095825,
      "learning_rate": 5.862470961812133e-06,
      "loss": 0.057196617126464844,
      "step": 5391
    },
    {
      "epoch": 0.7289564850021124,
      "grad_norm": 0.30894455313682556,
      "learning_rate": 5.857044002654357e-06,
      "loss": 0.06757664680480957,
      "step": 5392
    },
    {
      "epoch": 0.7290916772285594,
      "grad_norm": 0.22640785574913025,
      "learning_rate": 5.851618947108764e-06,
      "loss": 0.04272150993347168,
      "step": 5393
    },
    {
      "epoch": 0.7292268694550064,
      "grad_norm": 0.23192906379699707,
      "learning_rate": 5.8461957963048984e-06,
      "loss": 0.06345391273498535,
      "step": 5394
    },
    {
      "epoch": 0.7293620616814533,
      "grad_norm": 0.24933519959449768,
      "learning_rate": 5.840774551371882e-06,
      "loss": 0.07121109962463379,
      "step": 5395
    },
    {
      "epoch": 0.7294972539079003,
      "grad_norm": 0.3676450848579407,
      "learning_rate": 5.8353552134384405e-06,
      "loss": 0.07812178134918213,
      "step": 5396
    },
    {
      "epoch": 0.7296324461343473,
      "grad_norm": 0.32973673939704895,
      "learning_rate": 5.829937783632926e-06,
      "loss": 0.076987624168396,
      "step": 5397
    },
    {
      "epoch": 0.7297676383607943,
      "grad_norm": 0.5101771354675293,
      "learning_rate": 5.824522263083256e-06,
      "loss": 0.06301406025886536,
      "step": 5398
    },
    {
      "epoch": 0.7299028305872413,
      "grad_norm": 0.4950540065765381,
      "learning_rate": 5.8191086529169855e-06,
      "loss": 0.051572561264038086,
      "step": 5399
    },
    {
      "epoch": 0.7300380228136882,
      "grad_norm": 0.3083554208278656,
      "learning_rate": 5.813696954261253e-06,
      "loss": 0.06078815460205078,
      "step": 5400
    },
    {
      "epoch": 0.7301732150401352,
      "grad_norm": 0.323513001203537,
      "learning_rate": 5.8082871682428e-06,
      "loss": 0.07897329330444336,
      "step": 5401
    },
    {
      "epoch": 0.7303084072665822,
      "grad_norm": 0.2907189726829529,
      "learning_rate": 5.802879295987975e-06,
      "loss": 0.057209014892578125,
      "step": 5402
    },
    {
      "epoch": 0.7304435994930292,
      "grad_norm": 0.35615667700767517,
      "learning_rate": 5.797473338622722e-06,
      "loss": 0.06593179702758789,
      "step": 5403
    },
    {
      "epoch": 0.7305787917194762,
      "grad_norm": 0.3863239586353302,
      "learning_rate": 5.792069297272599e-06,
      "loss": 0.08330130577087402,
      "step": 5404
    },
    {
      "epoch": 0.7307139839459231,
      "grad_norm": 0.13111256062984467,
      "learning_rate": 5.7866671730627485e-06,
      "loss": 0.044106483459472656,
      "step": 5405
    },
    {
      "epoch": 0.7308491761723701,
      "grad_norm": 0.47994399070739746,
      "learning_rate": 5.781266967117925e-06,
      "loss": 0.07491302490234375,
      "step": 5406
    },
    {
      "epoch": 0.7309843683988171,
      "grad_norm": 0.3283345103263855,
      "learning_rate": 5.7758686805624815e-06,
      "loss": 0.0642390251159668,
      "step": 5407
    },
    {
      "epoch": 0.7311195606252641,
      "grad_norm": 0.26693981885910034,
      "learning_rate": 5.7704723145203605e-06,
      "loss": 0.07425355911254883,
      "step": 5408
    },
    {
      "epoch": 0.731254752851711,
      "grad_norm": 0.28047069907188416,
      "learning_rate": 5.765077870115126e-06,
      "loss": 0.05912092328071594,
      "step": 5409
    },
    {
      "epoch": 0.731389945078158,
      "grad_norm": 0.3144618272781372,
      "learning_rate": 5.759685348469928e-06,
      "loss": 0.07192420959472656,
      "step": 5410
    },
    {
      "epoch": 0.731525137304605,
      "grad_norm": 0.2807704508304596,
      "learning_rate": 5.754294750707514e-06,
      "loss": 0.05621659755706787,
      "step": 5411
    },
    {
      "epoch": 0.731660329531052,
      "grad_norm": 0.19868092238903046,
      "learning_rate": 5.748906077950237e-06,
      "loss": 0.04578137397766113,
      "step": 5412
    },
    {
      "epoch": 0.731795521757499,
      "grad_norm": 0.3376769423484802,
      "learning_rate": 5.743519331320042e-06,
      "loss": 0.0765380859375,
      "step": 5413
    },
    {
      "epoch": 0.7319307139839459,
      "grad_norm": 0.2248193621635437,
      "learning_rate": 5.73813451193849e-06,
      "loss": 0.058924734592437744,
      "step": 5414
    },
    {
      "epoch": 0.7320659062103929,
      "grad_norm": 0.3640033006668091,
      "learning_rate": 5.7327516209267225e-06,
      "loss": 0.08541679382324219,
      "step": 5415
    },
    {
      "epoch": 0.7322010984368399,
      "grad_norm": 0.18708643317222595,
      "learning_rate": 5.727370659405486e-06,
      "loss": 0.046230435371398926,
      "step": 5416
    },
    {
      "epoch": 0.7323362906632869,
      "grad_norm": 0.22921724617481232,
      "learning_rate": 5.7219916284951265e-06,
      "loss": 0.07297617197036743,
      "step": 5417
    },
    {
      "epoch": 0.7324714828897338,
      "grad_norm": 0.40217429399490356,
      "learning_rate": 5.716614529315582e-06,
      "loss": 0.0736227035522461,
      "step": 5418
    },
    {
      "epoch": 0.7326066751161808,
      "grad_norm": 0.40664488077163696,
      "learning_rate": 5.711239362986401e-06,
      "loss": 0.06919479370117188,
      "step": 5419
    },
    {
      "epoch": 0.7327418673426278,
      "grad_norm": 0.19880792498588562,
      "learning_rate": 5.705866130626719e-06,
      "loss": 0.03999662399291992,
      "step": 5420
    },
    {
      "epoch": 0.7328770595690748,
      "grad_norm": 0.3941669464111328,
      "learning_rate": 5.700494833355271e-06,
      "loss": 0.08850431442260742,
      "step": 5421
    },
    {
      "epoch": 0.7330122517955218,
      "grad_norm": 0.3398006558418274,
      "learning_rate": 5.69512547229039e-06,
      "loss": 0.0892648696899414,
      "step": 5422
    },
    {
      "epoch": 0.7331474440219687,
      "grad_norm": 0.4490032494068146,
      "learning_rate": 5.689758048550001e-06,
      "loss": 0.08185863494873047,
      "step": 5423
    },
    {
      "epoch": 0.7332826362484157,
      "grad_norm": 0.23250049352645874,
      "learning_rate": 5.684392563251644e-06,
      "loss": 0.04620325565338135,
      "step": 5424
    },
    {
      "epoch": 0.7334178284748627,
      "grad_norm": 0.3230533003807068,
      "learning_rate": 5.679029017512422e-06,
      "loss": 0.07272601127624512,
      "step": 5425
    },
    {
      "epoch": 0.7335530207013097,
      "grad_norm": 0.465389609336853,
      "learning_rate": 5.6736674124490684e-06,
      "loss": 0.10321903228759766,
      "step": 5426
    },
    {
      "epoch": 0.7336882129277567,
      "grad_norm": 0.2790832221508026,
      "learning_rate": 5.6683077491778935e-06,
      "loss": 0.08725085854530334,
      "step": 5427
    },
    {
      "epoch": 0.7338234051542036,
      "grad_norm": 0.14180271327495575,
      "learning_rate": 5.6629500288148044e-06,
      "loss": 0.034332990646362305,
      "step": 5428
    },
    {
      "epoch": 0.7339585973806506,
      "grad_norm": 0.197455495595932,
      "learning_rate": 5.657594252475319e-06,
      "loss": 0.03261923789978027,
      "step": 5429
    },
    {
      "epoch": 0.7340937896070976,
      "grad_norm": 0.33095598220825195,
      "learning_rate": 5.652240421274521e-06,
      "loss": 0.06939244270324707,
      "step": 5430
    },
    {
      "epoch": 0.7342289818335446,
      "grad_norm": 0.27738887071609497,
      "learning_rate": 5.646888536327121e-06,
      "loss": 0.08073282241821289,
      "step": 5431
    },
    {
      "epoch": 0.7343641740599915,
      "grad_norm": 0.26231980323791504,
      "learning_rate": 5.641538598747403e-06,
      "loss": 0.059847116470336914,
      "step": 5432
    },
    {
      "epoch": 0.7344993662864385,
      "grad_norm": 0.2408897876739502,
      "learning_rate": 5.6361906096492495e-06,
      "loss": 0.05955934524536133,
      "step": 5433
    },
    {
      "epoch": 0.7346345585128855,
      "grad_norm": 0.1614508181810379,
      "learning_rate": 5.630844570146157e-06,
      "loss": 0.03942972421646118,
      "step": 5434
    },
    {
      "epoch": 0.7347697507393325,
      "grad_norm": 0.22522589564323425,
      "learning_rate": 5.625500481351176e-06,
      "loss": 0.052280426025390625,
      "step": 5435
    },
    {
      "epoch": 0.7349049429657795,
      "grad_norm": 0.313833624124527,
      "learning_rate": 5.6201583443769895e-06,
      "loss": 0.06171607971191406,
      "step": 5436
    },
    {
      "epoch": 0.7350401351922264,
      "grad_norm": 0.36819571256637573,
      "learning_rate": 5.614818160335857e-06,
      "loss": 0.06548786163330078,
      "step": 5437
    },
    {
      "epoch": 0.7351753274186734,
      "grad_norm": 0.5532251596450806,
      "learning_rate": 5.6094799303396315e-06,
      "loss": 0.08300328254699707,
      "step": 5438
    },
    {
      "epoch": 0.7353105196451204,
      "grad_norm": 0.30740243196487427,
      "learning_rate": 5.6041436554997595e-06,
      "loss": 0.06314802169799805,
      "step": 5439
    },
    {
      "epoch": 0.7354457118715674,
      "grad_norm": 0.11969193816184998,
      "learning_rate": 5.598809336927278e-06,
      "loss": 0.03349733352661133,
      "step": 5440
    },
    {
      "epoch": 0.7355809040980144,
      "grad_norm": 0.4020923376083374,
      "learning_rate": 5.5934769757328325e-06,
      "loss": 0.0655984878540039,
      "step": 5441
    },
    {
      "epoch": 0.7357160963244613,
      "grad_norm": 0.4389861822128296,
      "learning_rate": 5.588146573026642e-06,
      "loss": 0.07568025588989258,
      "step": 5442
    },
    {
      "epoch": 0.7358512885509083,
      "grad_norm": 0.2634657323360443,
      "learning_rate": 5.582818129918525e-06,
      "loss": 0.0838460922241211,
      "step": 5443
    },
    {
      "epoch": 0.7359864807773553,
      "grad_norm": 0.5413923263549805,
      "learning_rate": 5.5774916475178915e-06,
      "loss": 0.09250211715698242,
      "step": 5444
    },
    {
      "epoch": 0.7361216730038023,
      "grad_norm": 0.46186718344688416,
      "learning_rate": 5.572167126933738e-06,
      "loss": 0.06212717294692993,
      "step": 5445
    },
    {
      "epoch": 0.7362568652302492,
      "grad_norm": 0.41852033138275146,
      "learning_rate": 5.566844569274669e-06,
      "loss": 0.09307289123535156,
      "step": 5446
    },
    {
      "epoch": 0.7363920574566962,
      "grad_norm": 0.2844407260417938,
      "learning_rate": 5.5615239756488665e-06,
      "loss": 0.07479667663574219,
      "step": 5447
    },
    {
      "epoch": 0.7365272496831432,
      "grad_norm": 0.42158037424087524,
      "learning_rate": 5.556205347164104e-06,
      "loss": 0.0788877010345459,
      "step": 5448
    },
    {
      "epoch": 0.7366624419095902,
      "grad_norm": 0.2536088526248932,
      "learning_rate": 5.550888684927746e-06,
      "loss": 0.052857398986816406,
      "step": 5449
    },
    {
      "epoch": 0.7367976341360372,
      "grad_norm": 0.6148597002029419,
      "learning_rate": 5.545573990046752e-06,
      "loss": 0.08426094055175781,
      "step": 5450
    },
    {
      "epoch": 0.7369328263624841,
      "grad_norm": 0.44636231660842896,
      "learning_rate": 5.540261263627672e-06,
      "loss": 0.05566573143005371,
      "step": 5451
    },
    {
      "epoch": 0.7370680185889311,
      "grad_norm": 0.3422273099422455,
      "learning_rate": 5.534950506776644e-06,
      "loss": 0.07265329360961914,
      "step": 5452
    },
    {
      "epoch": 0.7372032108153781,
      "grad_norm": 0.34439584612846375,
      "learning_rate": 5.529641720599393e-06,
      "loss": 0.06811285018920898,
      "step": 5453
    },
    {
      "epoch": 0.7373384030418251,
      "grad_norm": 0.5353866815567017,
      "learning_rate": 5.52433490620124e-06,
      "loss": 0.08996915817260742,
      "step": 5454
    },
    {
      "epoch": 0.737473595268272,
      "grad_norm": 0.24170364439487457,
      "learning_rate": 5.519030064687082e-06,
      "loss": 0.057979583740234375,
      "step": 5455
    },
    {
      "epoch": 0.737608787494719,
      "grad_norm": 0.35214370489120483,
      "learning_rate": 5.51372719716143e-06,
      "loss": 0.06314373016357422,
      "step": 5456
    },
    {
      "epoch": 0.737743979721166,
      "grad_norm": 0.38871315121650696,
      "learning_rate": 5.508426304728363e-06,
      "loss": 0.0693821907043457,
      "step": 5457
    },
    {
      "epoch": 0.737879171947613,
      "grad_norm": 0.31814950704574585,
      "learning_rate": 5.503127388491552e-06,
      "loss": 0.06226396560668945,
      "step": 5458
    },
    {
      "epoch": 0.73801436417406,
      "grad_norm": 0.4149168133735657,
      "learning_rate": 5.497830449554266e-06,
      "loss": 0.08545207977294922,
      "step": 5459
    },
    {
      "epoch": 0.738149556400507,
      "grad_norm": 0.4107033908367157,
      "learning_rate": 5.492535489019344e-06,
      "loss": 0.057828426361083984,
      "step": 5460
    },
    {
      "epoch": 0.7382847486269539,
      "grad_norm": 0.545863151550293,
      "learning_rate": 5.4872425079892454e-06,
      "loss": 0.08362340927124023,
      "step": 5461
    },
    {
      "epoch": 0.7384199408534009,
      "grad_norm": 0.5547133088111877,
      "learning_rate": 5.481951507565973e-06,
      "loss": 0.08692812919616699,
      "step": 5462
    },
    {
      "epoch": 0.7385551330798479,
      "grad_norm": 0.40846455097198486,
      "learning_rate": 5.476662488851159e-06,
      "loss": 0.06932711601257324,
      "step": 5463
    },
    {
      "epoch": 0.7386903253062949,
      "grad_norm": 0.18166811764240265,
      "learning_rate": 5.471375452946e-06,
      "loss": 0.05360269546508789,
      "step": 5464
    },
    {
      "epoch": 0.7388255175327418,
      "grad_norm": 0.3002277612686157,
      "learning_rate": 5.466090400951279e-06,
      "loss": 0.0629281997680664,
      "step": 5465
    },
    {
      "epoch": 0.7389607097591888,
      "grad_norm": 0.2697925269603729,
      "learning_rate": 5.460807333967387e-06,
      "loss": 0.0595545768737793,
      "step": 5466
    },
    {
      "epoch": 0.7390959019856358,
      "grad_norm": 0.3708248734474182,
      "learning_rate": 5.455526253094267e-06,
      "loss": 0.06967902183532715,
      "step": 5467
    },
    {
      "epoch": 0.7392310942120828,
      "grad_norm": 0.29452773928642273,
      "learning_rate": 5.450247159431486e-06,
      "loss": 0.03252696990966797,
      "step": 5468
    },
    {
      "epoch": 0.7393662864385298,
      "grad_norm": 0.4548581540584564,
      "learning_rate": 5.44497005407817e-06,
      "loss": 0.08553218841552734,
      "step": 5469
    },
    {
      "epoch": 0.7395014786649767,
      "grad_norm": 0.3221522867679596,
      "learning_rate": 5.439694938133042e-06,
      "loss": 0.09247910976409912,
      "step": 5470
    },
    {
      "epoch": 0.7396366708914237,
      "grad_norm": 0.7349978685379028,
      "learning_rate": 5.434421812694409e-06,
      "loss": 0.08335566520690918,
      "step": 5471
    },
    {
      "epoch": 0.7397718631178707,
      "grad_norm": 0.32768312096595764,
      "learning_rate": 5.4291506788601624e-06,
      "loss": 0.04716801643371582,
      "step": 5472
    },
    {
      "epoch": 0.7399070553443177,
      "grad_norm": 0.4047926664352417,
      "learning_rate": 5.423881537727785e-06,
      "loss": 0.08875751495361328,
      "step": 5473
    },
    {
      "epoch": 0.7400422475707646,
      "grad_norm": 0.302165687084198,
      "learning_rate": 5.418614390394338e-06,
      "loss": 0.05476647615432739,
      "step": 5474
    },
    {
      "epoch": 0.7401774397972116,
      "grad_norm": 0.6546356678009033,
      "learning_rate": 5.413349237956469e-06,
      "loss": 0.10552167892456055,
      "step": 5475
    },
    {
      "epoch": 0.7403126320236586,
      "grad_norm": 0.586347222328186,
      "learning_rate": 5.4080860815104125e-06,
      "loss": 0.10285234451293945,
      "step": 5476
    },
    {
      "epoch": 0.7404478242501056,
      "grad_norm": 0.3252721428871155,
      "learning_rate": 5.402824922151977e-06,
      "loss": 0.05477714538574219,
      "step": 5477
    },
    {
      "epoch": 0.7405830164765526,
      "grad_norm": 0.30552777647972107,
      "learning_rate": 5.397565760976577e-06,
      "loss": 0.0667119026184082,
      "step": 5478
    },
    {
      "epoch": 0.7407182087029995,
      "grad_norm": 0.40809082984924316,
      "learning_rate": 5.392308599079193e-06,
      "loss": 0.07940340042114258,
      "step": 5479
    },
    {
      "epoch": 0.7408534009294465,
      "grad_norm": 0.26088157296180725,
      "learning_rate": 5.3870534375543916e-06,
      "loss": 0.0644269585609436,
      "step": 5480
    },
    {
      "epoch": 0.7409885931558935,
      "grad_norm": 0.3476581275463104,
      "learning_rate": 5.381800277496328e-06,
      "loss": 0.0823659896850586,
      "step": 5481
    },
    {
      "epoch": 0.7411237853823405,
      "grad_norm": 0.35466960072517395,
      "learning_rate": 5.376549119998731e-06,
      "loss": 0.06915223598480225,
      "step": 5482
    },
    {
      "epoch": 0.7412589776087875,
      "grad_norm": 0.3833767771720886,
      "learning_rate": 5.3712999661549314e-06,
      "loss": 0.07703423500061035,
      "step": 5483
    },
    {
      "epoch": 0.7413941698352344,
      "grad_norm": 0.36239394545555115,
      "learning_rate": 5.366052817057826e-06,
      "loss": 0.06222820281982422,
      "step": 5484
    },
    {
      "epoch": 0.7415293620616814,
      "grad_norm": 0.40474584698677063,
      "learning_rate": 5.360807673799899e-06,
      "loss": 0.06822347640991211,
      "step": 5485
    },
    {
      "epoch": 0.7416645542881284,
      "grad_norm": 0.3074905276298523,
      "learning_rate": 5.355564537473214e-06,
      "loss": 0.07065606117248535,
      "step": 5486
    },
    {
      "epoch": 0.7417997465145754,
      "grad_norm": 0.3259347081184387,
      "learning_rate": 5.35032340916942e-06,
      "loss": 0.07717752456665039,
      "step": 5487
    },
    {
      "epoch": 0.7419349387410223,
      "grad_norm": 0.3179175555706024,
      "learning_rate": 5.345084289979755e-06,
      "loss": 0.05710721015930176,
      "step": 5488
    },
    {
      "epoch": 0.7420701309674693,
      "grad_norm": 0.30644139647483826,
      "learning_rate": 5.339847180995026e-06,
      "loss": 0.04585671424865723,
      "step": 5489
    },
    {
      "epoch": 0.7422053231939163,
      "grad_norm": 0.25662174820899963,
      "learning_rate": 5.33461208330563e-06,
      "loss": 0.04766833782196045,
      "step": 5490
    },
    {
      "epoch": 0.7423405154203633,
      "grad_norm": 0.3106755018234253,
      "learning_rate": 5.32937899800154e-06,
      "loss": 0.0757284164428711,
      "step": 5491
    },
    {
      "epoch": 0.7424757076468103,
      "grad_norm": 0.322922945022583,
      "learning_rate": 5.324147926172307e-06,
      "loss": 0.07869482040405273,
      "step": 5492
    },
    {
      "epoch": 0.7426108998732572,
      "grad_norm": 0.26171329617500305,
      "learning_rate": 5.318918868907084e-06,
      "loss": 0.06150388717651367,
      "step": 5493
    },
    {
      "epoch": 0.7427460920997042,
      "grad_norm": 0.21698588132858276,
      "learning_rate": 5.313691827294568e-06,
      "loss": 0.05281102657318115,
      "step": 5494
    },
    {
      "epoch": 0.7428812843261512,
      "grad_norm": 0.25994858145713806,
      "learning_rate": 5.308466802423072e-06,
      "loss": 0.05610466003417969,
      "step": 5495
    },
    {
      "epoch": 0.7430164765525982,
      "grad_norm": 0.3296898305416107,
      "learning_rate": 5.303243795380471e-06,
      "loss": 0.06893157958984375,
      "step": 5496
    },
    {
      "epoch": 0.7431516687790453,
      "grad_norm": 0.46212247014045715,
      "learning_rate": 5.298022807254215e-06,
      "loss": 0.07411670684814453,
      "step": 5497
    },
    {
      "epoch": 0.7432868610054922,
      "grad_norm": 0.24345175921916962,
      "learning_rate": 5.292803839131358e-06,
      "loss": 0.05167675018310547,
      "step": 5498
    },
    {
      "epoch": 0.7434220532319392,
      "grad_norm": 0.29237526655197144,
      "learning_rate": 5.287586892098496e-06,
      "loss": 0.06631326675415039,
      "step": 5499
    },
    {
      "epoch": 0.7435572454583862,
      "grad_norm": 0.3614571690559387,
      "learning_rate": 5.282371967241842e-06,
      "loss": 0.09359931945800781,
      "step": 5500
    },
    {
      "epoch": 0.7436924376848332,
      "grad_norm": 0.2692449986934662,
      "learning_rate": 5.277159065647164e-06,
      "loss": 0.06752258539199829,
      "step": 5501
    },
    {
      "epoch": 0.7438276299112802,
      "grad_norm": 1.1080949306488037,
      "learning_rate": 5.271948188399814e-06,
      "loss": 0.09859180450439453,
      "step": 5502
    },
    {
      "epoch": 0.7439628221377271,
      "grad_norm": 0.351872980594635,
      "learning_rate": 5.266739336584735e-06,
      "loss": 0.07297849655151367,
      "step": 5503
    },
    {
      "epoch": 0.7440980143641741,
      "grad_norm": 0.46755194664001465,
      "learning_rate": 5.261532511286422e-06,
      "loss": 0.07714438438415527,
      "step": 5504
    },
    {
      "epoch": 0.7442332065906211,
      "grad_norm": 0.4388406276702881,
      "learning_rate": 5.256327713588977e-06,
      "loss": 0.0654829740524292,
      "step": 5505
    },
    {
      "epoch": 0.7443683988170681,
      "grad_norm": 0.6473273634910583,
      "learning_rate": 5.25112494457606e-06,
      "loss": 0.07860946655273438,
      "step": 5506
    },
    {
      "epoch": 0.744503591043515,
      "grad_norm": 0.5071631073951721,
      "learning_rate": 5.245924205330919e-06,
      "loss": 0.07981467247009277,
      "step": 5507
    },
    {
      "epoch": 0.744638783269962,
      "grad_norm": 0.2399771809577942,
      "learning_rate": 5.240725496936373e-06,
      "loss": 0.0605168342590332,
      "step": 5508
    },
    {
      "epoch": 0.744773975496409,
      "grad_norm": 0.2920646369457245,
      "learning_rate": 5.2355288204748145e-06,
      "loss": 0.062044739723205566,
      "step": 5509
    },
    {
      "epoch": 0.744909167722856,
      "grad_norm": 0.4028262197971344,
      "learning_rate": 5.230334177028233e-06,
      "loss": 0.06833958625793457,
      "step": 5510
    },
    {
      "epoch": 0.745044359949303,
      "grad_norm": 0.4029131233692169,
      "learning_rate": 5.2251415676781726e-06,
      "loss": 0.07670450210571289,
      "step": 5511
    },
    {
      "epoch": 0.7451795521757499,
      "grad_norm": 0.3253644108772278,
      "learning_rate": 5.2199509935057655e-06,
      "loss": 0.058380842208862305,
      "step": 5512
    },
    {
      "epoch": 0.7453147444021969,
      "grad_norm": 0.1138840913772583,
      "learning_rate": 5.214762455591713e-06,
      "loss": 0.028551578521728516,
      "step": 5513
    },
    {
      "epoch": 0.7454499366286439,
      "grad_norm": 0.2654602825641632,
      "learning_rate": 5.209575955016295e-06,
      "loss": 0.07141876220703125,
      "step": 5514
    },
    {
      "epoch": 0.7455851288550909,
      "grad_norm": 0.2957044839859009,
      "learning_rate": 5.204391492859377e-06,
      "loss": 0.06686973571777344,
      "step": 5515
    },
    {
      "epoch": 0.7457203210815379,
      "grad_norm": 0.3932427167892456,
      "learning_rate": 5.199209070200388e-06,
      "loss": 0.07320308685302734,
      "step": 5516
    },
    {
      "epoch": 0.7458555133079848,
      "grad_norm": 0.19216471910476685,
      "learning_rate": 5.194028688118332e-06,
      "loss": 0.04555559158325195,
      "step": 5517
    },
    {
      "epoch": 0.7459907055344318,
      "grad_norm": 0.23753167688846588,
      "learning_rate": 5.188850347691797e-06,
      "loss": 0.05741691589355469,
      "step": 5518
    },
    {
      "epoch": 0.7461258977608788,
      "grad_norm": 0.4030016362667084,
      "learning_rate": 5.183674049998934e-06,
      "loss": 0.07119739055633545,
      "step": 5519
    },
    {
      "epoch": 0.7462610899873258,
      "grad_norm": 0.29846253991127014,
      "learning_rate": 5.178499796117485e-06,
      "loss": 0.06371355056762695,
      "step": 5520
    },
    {
      "epoch": 0.7463962822137727,
      "grad_norm": 0.25891730189323425,
      "learning_rate": 5.173327587124753e-06,
      "loss": 0.06730508804321289,
      "step": 5521
    },
    {
      "epoch": 0.7465314744402197,
      "grad_norm": 0.35525593161582947,
      "learning_rate": 5.16815742409762e-06,
      "loss": 0.08658409118652344,
      "step": 5522
    },
    {
      "epoch": 0.7466666666666667,
      "grad_norm": 0.19685225188732147,
      "learning_rate": 5.16298930811254e-06,
      "loss": 0.05870342254638672,
      "step": 5523
    },
    {
      "epoch": 0.7468018588931137,
      "grad_norm": 0.3284025192260742,
      "learning_rate": 5.15782324024554e-06,
      "loss": 0.06280958652496338,
      "step": 5524
    },
    {
      "epoch": 0.7469370511195607,
      "grad_norm": 0.3075641393661499,
      "learning_rate": 5.152659221572231e-06,
      "loss": 0.05977821350097656,
      "step": 5525
    },
    {
      "epoch": 0.7470722433460076,
      "grad_norm": 0.25582441687583923,
      "learning_rate": 5.147497253167784e-06,
      "loss": 0.057166099548339844,
      "step": 5526
    },
    {
      "epoch": 0.7472074355724546,
      "grad_norm": 0.5908964276313782,
      "learning_rate": 5.142337336106948e-06,
      "loss": 0.09470438957214355,
      "step": 5527
    },
    {
      "epoch": 0.7473426277989016,
      "grad_norm": 0.3191787004470825,
      "learning_rate": 5.137179471464047e-06,
      "loss": 0.06888628005981445,
      "step": 5528
    },
    {
      "epoch": 0.7474778200253486,
      "grad_norm": 0.32175543904304504,
      "learning_rate": 5.13202366031297e-06,
      "loss": 0.06878399848937988,
      "step": 5529
    },
    {
      "epoch": 0.7476130122517956,
      "grad_norm": 0.3113773763179779,
      "learning_rate": 5.1268699037272e-06,
      "loss": 0.06818485260009766,
      "step": 5530
    },
    {
      "epoch": 0.7477482044782425,
      "grad_norm": 0.2162005603313446,
      "learning_rate": 5.121718202779756e-06,
      "loss": 0.05885887145996094,
      "step": 5531
    },
    {
      "epoch": 0.7478833967046895,
      "grad_norm": 0.4877583086490631,
      "learning_rate": 5.116568558543264e-06,
      "loss": 0.09424686431884766,
      "step": 5532
    },
    {
      "epoch": 0.7480185889311365,
      "grad_norm": 0.31521716713905334,
      "learning_rate": 5.1114209720899025e-06,
      "loss": 0.06360030174255371,
      "step": 5533
    },
    {
      "epoch": 0.7481537811575835,
      "grad_norm": 0.12664315104484558,
      "learning_rate": 5.106275444491423e-06,
      "loss": 0.025168538093566895,
      "step": 5534
    },
    {
      "epoch": 0.7482889733840304,
      "grad_norm": 0.36357641220092773,
      "learning_rate": 5.101131976819165e-06,
      "loss": 0.07224273681640625,
      "step": 5535
    },
    {
      "epoch": 0.7484241656104774,
      "grad_norm": 0.2043297439813614,
      "learning_rate": 5.095990570144008e-06,
      "loss": 0.054151058197021484,
      "step": 5536
    },
    {
      "epoch": 0.7485593578369244,
      "grad_norm": 0.2542010247707367,
      "learning_rate": 5.090851225536432e-06,
      "loss": 0.037700653076171875,
      "step": 5537
    },
    {
      "epoch": 0.7486945500633714,
      "grad_norm": 0.13683252036571503,
      "learning_rate": 5.085713944066474e-06,
      "loss": 0.0431215763092041,
      "step": 5538
    },
    {
      "epoch": 0.7488297422898184,
      "grad_norm": 0.3221246004104614,
      "learning_rate": 5.080578726803741e-06,
      "loss": 0.059185028076171875,
      "step": 5539
    },
    {
      "epoch": 0.7489649345162653,
      "grad_norm": 0.2807500660419464,
      "learning_rate": 5.075445574817415e-06,
      "loss": 0.07166194915771484,
      "step": 5540
    },
    {
      "epoch": 0.7491001267427123,
      "grad_norm": 0.4837017357349396,
      "learning_rate": 5.07031448917624e-06,
      "loss": 0.07048225402832031,
      "step": 5541
    },
    {
      "epoch": 0.7492353189691593,
      "grad_norm": 0.30621686577796936,
      "learning_rate": 5.065185470948544e-06,
      "loss": 0.07455205917358398,
      "step": 5542
    },
    {
      "epoch": 0.7493705111956063,
      "grad_norm": 0.5735632181167603,
      "learning_rate": 5.060058521202211e-06,
      "loss": 0.07740139961242676,
      "step": 5543
    },
    {
      "epoch": 0.7495057034220532,
      "grad_norm": 0.38835254311561584,
      "learning_rate": 5.054933641004703e-06,
      "loss": 0.08188867568969727,
      "step": 5544
    },
    {
      "epoch": 0.7496408956485002,
      "grad_norm": 0.3186468183994293,
      "learning_rate": 5.0498108314230425e-06,
      "loss": 0.07472836971282959,
      "step": 5545
    },
    {
      "epoch": 0.7497760878749472,
      "grad_norm": 0.23334920406341553,
      "learning_rate": 5.044690093523823e-06,
      "loss": 0.049179792404174805,
      "step": 5546
    },
    {
      "epoch": 0.7499112801013942,
      "grad_norm": 0.2741117775440216,
      "learning_rate": 5.039571428373219e-06,
      "loss": 0.05063951015472412,
      "step": 5547
    },
    {
      "epoch": 0.7500464723278412,
      "grad_norm": 0.3651045560836792,
      "learning_rate": 5.034454837036959e-06,
      "loss": 0.06481456756591797,
      "step": 5548
    },
    {
      "epoch": 0.7501816645542881,
      "grad_norm": 0.29171857237815857,
      "learning_rate": 5.0293403205803455e-06,
      "loss": 0.06099820137023926,
      "step": 5549
    },
    {
      "epoch": 0.7503168567807351,
      "grad_norm": 0.33192285895347595,
      "learning_rate": 5.024227880068247e-06,
      "loss": 0.058637142181396484,
      "step": 5550
    },
    {
      "epoch": 0.7504520490071821,
      "grad_norm": 0.2700657248497009,
      "learning_rate": 5.019117516565096e-06,
      "loss": 0.04500782489776611,
      "step": 5551
    },
    {
      "epoch": 0.7505872412336291,
      "grad_norm": 0.23167628049850464,
      "learning_rate": 5.014009231134908e-06,
      "loss": 0.0492253303527832,
      "step": 5552
    },
    {
      "epoch": 0.750722433460076,
      "grad_norm": 0.3039982318878174,
      "learning_rate": 5.008903024841248e-06,
      "loss": 0.06557083129882812,
      "step": 5553
    },
    {
      "epoch": 0.750857625686523,
      "grad_norm": 0.5059700608253479,
      "learning_rate": 5.0037988987472595e-06,
      "loss": 0.07209920883178711,
      "step": 5554
    },
    {
      "epoch": 0.75099281791297,
      "grad_norm": 0.3052729070186615,
      "learning_rate": 4.998696853915646e-06,
      "loss": 0.059540748596191406,
      "step": 5555
    },
    {
      "epoch": 0.751128010139417,
      "grad_norm": 0.3321710526943207,
      "learning_rate": 4.993596891408676e-06,
      "loss": 0.07263684272766113,
      "step": 5556
    },
    {
      "epoch": 0.751263202365864,
      "grad_norm": 0.47605693340301514,
      "learning_rate": 4.988499012288198e-06,
      "loss": 0.0726243257522583,
      "step": 5557
    },
    {
      "epoch": 0.751398394592311,
      "grad_norm": 0.11801926791667938,
      "learning_rate": 4.983403217615614e-06,
      "loss": 0.033270835876464844,
      "step": 5558
    },
    {
      "epoch": 0.7515335868187579,
      "grad_norm": 0.5052714943885803,
      "learning_rate": 4.978309508451896e-06,
      "loss": 0.09722709655761719,
      "step": 5559
    },
    {
      "epoch": 0.7516687790452049,
      "grad_norm": 0.25686025619506836,
      "learning_rate": 4.973217885857578e-06,
      "loss": 0.053447723388671875,
      "step": 5560
    },
    {
      "epoch": 0.7518039712716519,
      "grad_norm": 0.26720160245895386,
      "learning_rate": 4.968128350892763e-06,
      "loss": 0.058026671409606934,
      "step": 5561
    },
    {
      "epoch": 0.7519391634980989,
      "grad_norm": 0.34509679675102234,
      "learning_rate": 4.963040904617131e-06,
      "loss": 0.0823512077331543,
      "step": 5562
    },
    {
      "epoch": 0.7520743557245458,
      "grad_norm": 0.2777736485004425,
      "learning_rate": 4.9579555480898955e-06,
      "loss": 0.0454862117767334,
      "step": 5563
    },
    {
      "epoch": 0.7522095479509928,
      "grad_norm": 0.22768519818782806,
      "learning_rate": 4.952872282369873e-06,
      "loss": 0.04216265678405762,
      "step": 5564
    },
    {
      "epoch": 0.7523447401774398,
      "grad_norm": 0.8205633759498596,
      "learning_rate": 4.947791108515417e-06,
      "loss": 0.04407358169555664,
      "step": 5565
    },
    {
      "epoch": 0.7524799324038868,
      "grad_norm": 0.4990426301956177,
      "learning_rate": 4.942712027584453e-06,
      "loss": 0.06133723258972168,
      "step": 5566
    },
    {
      "epoch": 0.7526151246303338,
      "grad_norm": 0.3296782970428467,
      "learning_rate": 4.937635040634485e-06,
      "loss": 0.07139110565185547,
      "step": 5567
    },
    {
      "epoch": 0.7527503168567807,
      "grad_norm": 0.25605350732803345,
      "learning_rate": 4.9325601487225545e-06,
      "loss": 0.05712485313415527,
      "step": 5568
    },
    {
      "epoch": 0.7528855090832277,
      "grad_norm": 0.4864124357700348,
      "learning_rate": 4.927487352905289e-06,
      "loss": 0.08876609802246094,
      "step": 5569
    },
    {
      "epoch": 0.7530207013096747,
      "grad_norm": 0.26311105489730835,
      "learning_rate": 4.92241665423887e-06,
      "loss": 0.04821968078613281,
      "step": 5570
    },
    {
      "epoch": 0.7531558935361217,
      "grad_norm": 0.38874098658561707,
      "learning_rate": 4.917348053779039e-06,
      "loss": 0.06451725959777832,
      "step": 5571
    },
    {
      "epoch": 0.7532910857625686,
      "grad_norm": 0.262574702501297,
      "learning_rate": 4.912281552581122e-06,
      "loss": 0.05056285858154297,
      "step": 5572
    },
    {
      "epoch": 0.7534262779890156,
      "grad_norm": 0.36956506967544556,
      "learning_rate": 4.907217151699969e-06,
      "loss": 0.05255019664764404,
      "step": 5573
    },
    {
      "epoch": 0.7535614702154626,
      "grad_norm": 0.1910557746887207,
      "learning_rate": 4.9021548521900305e-06,
      "loss": 0.03656738996505737,
      "step": 5574
    },
    {
      "epoch": 0.7536966624419096,
      "grad_norm": 0.3456321954727173,
      "learning_rate": 4.8970946551053005e-06,
      "loss": 0.0818791389465332,
      "step": 5575
    },
    {
      "epoch": 0.7538318546683566,
      "grad_norm": 0.3904494047164917,
      "learning_rate": 4.892036561499339e-06,
      "loss": 0.06348228454589844,
      "step": 5576
    },
    {
      "epoch": 0.7539670468948035,
      "grad_norm": 0.21301542222499847,
      "learning_rate": 4.8869805724252675e-06,
      "loss": 0.06483745574951172,
      "step": 5577
    },
    {
      "epoch": 0.7541022391212505,
      "grad_norm": 0.403944730758667,
      "learning_rate": 4.8819266889357665e-06,
      "loss": 0.06790876388549805,
      "step": 5578
    },
    {
      "epoch": 0.7542374313476975,
      "grad_norm": 0.5573452711105347,
      "learning_rate": 4.876874912083088e-06,
      "loss": 0.06965208053588867,
      "step": 5579
    },
    {
      "epoch": 0.7543726235741445,
      "grad_norm": 0.6763182282447815,
      "learning_rate": 4.871825242919037e-06,
      "loss": 0.09247112274169922,
      "step": 5580
    },
    {
      "epoch": 0.7545078158005915,
      "grad_norm": 0.19230929017066956,
      "learning_rate": 4.866777682494978e-06,
      "loss": 0.05852961540222168,
      "step": 5581
    },
    {
      "epoch": 0.7546430080270384,
      "grad_norm": 0.33645913004875183,
      "learning_rate": 4.861732231861845e-06,
      "loss": 0.07019758224487305,
      "step": 5582
    },
    {
      "epoch": 0.7547782002534854,
      "grad_norm": 0.1876869797706604,
      "learning_rate": 4.85668889207012e-06,
      "loss": 0.0445864200592041,
      "step": 5583
    },
    {
      "epoch": 0.7549133924799324,
      "grad_norm": 0.3423718810081482,
      "learning_rate": 4.851647664169862e-06,
      "loss": 0.06366348266601562,
      "step": 5584
    },
    {
      "epoch": 0.7550485847063794,
      "grad_norm": 0.40339142084121704,
      "learning_rate": 4.846608549210679e-06,
      "loss": 0.08297139406204224,
      "step": 5585
    },
    {
      "epoch": 0.7551837769328263,
      "grad_norm": 0.38900959491729736,
      "learning_rate": 4.841571548241741e-06,
      "loss": 0.04147911071777344,
      "step": 5586
    },
    {
      "epoch": 0.7553189691592733,
      "grad_norm": 0.27130749821662903,
      "learning_rate": 4.836536662311777e-06,
      "loss": 0.06513595581054688,
      "step": 5587
    },
    {
      "epoch": 0.7554541613857203,
      "grad_norm": 0.3792252540588379,
      "learning_rate": 4.8315038924690745e-06,
      "loss": 0.06888812780380249,
      "step": 5588
    },
    {
      "epoch": 0.7555893536121673,
      "grad_norm": 0.5799320936203003,
      "learning_rate": 4.82647323976149e-06,
      "loss": 0.10873842239379883,
      "step": 5589
    },
    {
      "epoch": 0.7557245458386143,
      "grad_norm": 0.19998204708099365,
      "learning_rate": 4.821444705236429e-06,
      "loss": 0.04957246780395508,
      "step": 5590
    },
    {
      "epoch": 0.7558597380650612,
      "grad_norm": 0.2670244574546814,
      "learning_rate": 4.81641828994086e-06,
      "loss": 0.05125004053115845,
      "step": 5591
    },
    {
      "epoch": 0.7559949302915082,
      "grad_norm": 0.28263622522354126,
      "learning_rate": 4.811393994921308e-06,
      "loss": 0.0540766716003418,
      "step": 5592
    },
    {
      "epoch": 0.7561301225179552,
      "grad_norm": 0.6768190860748291,
      "learning_rate": 4.806371821223854e-06,
      "loss": 0.08544349670410156,
      "step": 5593
    },
    {
      "epoch": 0.7562653147444022,
      "grad_norm": 0.210227370262146,
      "learning_rate": 4.801351769894151e-06,
      "loss": 0.039162009954452515,
      "step": 5594
    },
    {
      "epoch": 0.7564005069708492,
      "grad_norm": 0.25639989972114563,
      "learning_rate": 4.796333841977394e-06,
      "loss": 0.05685591697692871,
      "step": 5595
    },
    {
      "epoch": 0.7565356991972961,
      "grad_norm": 0.3441276550292969,
      "learning_rate": 4.791318038518345e-06,
      "loss": 0.0647817850112915,
      "step": 5596
    },
    {
      "epoch": 0.7566708914237431,
      "grad_norm": 0.27997344732284546,
      "learning_rate": 4.7863043605613185e-06,
      "loss": 0.054930925369262695,
      "step": 5597
    },
    {
      "epoch": 0.7568060836501901,
      "grad_norm": 0.1451415866613388,
      "learning_rate": 4.7812928091501865e-06,
      "loss": 0.05081653594970703,
      "step": 5598
    },
    {
      "epoch": 0.7569412758766371,
      "grad_norm": 0.2508980333805084,
      "learning_rate": 4.7762833853283935e-06,
      "loss": 0.04594683647155762,
      "step": 5599
    },
    {
      "epoch": 0.757076468103084,
      "grad_norm": 0.38984835147857666,
      "learning_rate": 4.77127609013891e-06,
      "loss": 0.07874011993408203,
      "step": 5600
    },
    {
      "epoch": 0.757211660329531,
      "grad_norm": 0.3819902241230011,
      "learning_rate": 4.766270924624295e-06,
      "loss": 0.08634424209594727,
      "step": 5601
    },
    {
      "epoch": 0.757346852555978,
      "grad_norm": 0.44710078835487366,
      "learning_rate": 4.761267889826647e-06,
      "loss": 0.07189011573791504,
      "step": 5602
    },
    {
      "epoch": 0.757482044782425,
      "grad_norm": 0.6716780662536621,
      "learning_rate": 4.756266986787619e-06,
      "loss": 0.07186508178710938,
      "step": 5603
    },
    {
      "epoch": 0.757617237008872,
      "grad_norm": 0.20318977534770966,
      "learning_rate": 4.751268216548439e-06,
      "loss": 0.04666495323181152,
      "step": 5604
    },
    {
      "epoch": 0.7577524292353189,
      "grad_norm": 0.29602131247520447,
      "learning_rate": 4.746271580149861e-06,
      "loss": 0.0380130410194397,
      "step": 5605
    },
    {
      "epoch": 0.7578876214617659,
      "grad_norm": 0.4363838732242584,
      "learning_rate": 4.7412770786322244e-06,
      "loss": 0.06917369365692139,
      "step": 5606
    },
    {
      "epoch": 0.7580228136882129,
      "grad_norm": 0.23910731077194214,
      "learning_rate": 4.736284713035406e-06,
      "loss": 0.04737520217895508,
      "step": 5607
    },
    {
      "epoch": 0.7581580059146599,
      "grad_norm": 0.19320856034755707,
      "learning_rate": 4.731294484398843e-06,
      "loss": 0.049562811851501465,
      "step": 5608
    },
    {
      "epoch": 0.7582931981411069,
      "grad_norm": 0.30218300223350525,
      "learning_rate": 4.726306393761526e-06,
      "loss": 0.056662678718566895,
      "step": 5609
    },
    {
      "epoch": 0.7584283903675538,
      "grad_norm": 0.22269980609416962,
      "learning_rate": 4.721320442162001e-06,
      "loss": 0.06302642822265625,
      "step": 5610
    },
    {
      "epoch": 0.7585635825940008,
      "grad_norm": 0.373076468706131,
      "learning_rate": 4.716336630638378e-06,
      "loss": 0.05855274200439453,
      "step": 5611
    },
    {
      "epoch": 0.7586987748204478,
      "grad_norm": 0.3565768897533417,
      "learning_rate": 4.711354960228306e-06,
      "loss": 0.08982467651367188,
      "step": 5612
    },
    {
      "epoch": 0.7588339670468948,
      "grad_norm": 0.1863771378993988,
      "learning_rate": 4.706375431968998e-06,
      "loss": 0.04405832290649414,
      "step": 5613
    },
    {
      "epoch": 0.7589691592733417,
      "grad_norm": 0.4505642354488373,
      "learning_rate": 4.701398046897218e-06,
      "loss": 0.06322681903839111,
      "step": 5614
    },
    {
      "epoch": 0.7591043514997887,
      "grad_norm": 0.17876534163951874,
      "learning_rate": 4.696422806049277e-06,
      "loss": 0.05791473388671875,
      "step": 5615
    },
    {
      "epoch": 0.7592395437262357,
      "grad_norm": 0.1927298903465271,
      "learning_rate": 4.69144971046106e-06,
      "loss": 0.03877538442611694,
      "step": 5616
    },
    {
      "epoch": 0.7593747359526827,
      "grad_norm": 0.23871183395385742,
      "learning_rate": 4.686478761167984e-06,
      "loss": 0.06244182586669922,
      "step": 5617
    },
    {
      "epoch": 0.7595099281791297,
      "grad_norm": 0.29319432377815247,
      "learning_rate": 4.681509959205028e-06,
      "loss": 0.07445716857910156,
      "step": 5618
    },
    {
      "epoch": 0.7596451204055766,
      "grad_norm": 0.32316696643829346,
      "learning_rate": 4.676543305606724e-06,
      "loss": 0.047013282775878906,
      "step": 5619
    },
    {
      "epoch": 0.7597803126320236,
      "grad_norm": 0.24482862651348114,
      "learning_rate": 4.67157880140715e-06,
      "loss": 0.056181907653808594,
      "step": 5620
    },
    {
      "epoch": 0.7599155048584706,
      "grad_norm": 0.3955845236778259,
      "learning_rate": 4.666616447639952e-06,
      "loss": 0.06490731239318848,
      "step": 5621
    },
    {
      "epoch": 0.7600506970849176,
      "grad_norm": 0.3440779149532318,
      "learning_rate": 4.661656245338314e-06,
      "loss": 0.06698846817016602,
      "step": 5622
    },
    {
      "epoch": 0.7601858893113645,
      "grad_norm": 0.6869900226593018,
      "learning_rate": 4.656698195534978e-06,
      "loss": 0.07489728927612305,
      "step": 5623
    },
    {
      "epoch": 0.7603210815378115,
      "grad_norm": 0.32057246565818787,
      "learning_rate": 4.651742299262233e-06,
      "loss": 0.058403968811035156,
      "step": 5624
    },
    {
      "epoch": 0.7604562737642585,
      "grad_norm": 0.2266540676355362,
      "learning_rate": 4.646788557551921e-06,
      "loss": 0.04056048393249512,
      "step": 5625
    },
    {
      "epoch": 0.7605914659907055,
      "grad_norm": 0.3709670305252075,
      "learning_rate": 4.641836971435445e-06,
      "loss": 0.08048868179321289,
      "step": 5626
    },
    {
      "epoch": 0.7607266582171525,
      "grad_norm": 0.2754816710948944,
      "learning_rate": 4.63688754194375e-06,
      "loss": 0.0615391731262207,
      "step": 5627
    },
    {
      "epoch": 0.7608618504435994,
      "grad_norm": 0.22325558960437775,
      "learning_rate": 4.6319402701073295e-06,
      "loss": 0.06150674819946289,
      "step": 5628
    },
    {
      "epoch": 0.7609970426700464,
      "grad_norm": 0.35540273785591125,
      "learning_rate": 4.6269951569562355e-06,
      "loss": 0.057094573974609375,
      "step": 5629
    },
    {
      "epoch": 0.7611322348964934,
      "grad_norm": 0.33841997385025024,
      "learning_rate": 4.622052203520061e-06,
      "loss": 0.06940567493438721,
      "step": 5630
    },
    {
      "epoch": 0.7612674271229404,
      "grad_norm": 0.27616506814956665,
      "learning_rate": 4.617111410827968e-06,
      "loss": 0.060190439224243164,
      "step": 5631
    },
    {
      "epoch": 0.7614026193493874,
      "grad_norm": 0.44101110100746155,
      "learning_rate": 4.612172779908639e-06,
      "loss": 0.053999900817871094,
      "step": 5632
    },
    {
      "epoch": 0.7615378115758344,
      "grad_norm": 0.43242114782333374,
      "learning_rate": 4.607236311790335e-06,
      "loss": 0.08671188354492188,
      "step": 5633
    },
    {
      "epoch": 0.7616730038022814,
      "grad_norm": 0.518928050994873,
      "learning_rate": 4.602302007500854e-06,
      "loss": 0.07594013214111328,
      "step": 5634
    },
    {
      "epoch": 0.7618081960287284,
      "grad_norm": 0.2190275490283966,
      "learning_rate": 4.597369868067537e-06,
      "loss": 0.047338008880615234,
      "step": 5635
    },
    {
      "epoch": 0.7619433882551754,
      "grad_norm": 0.24892202019691467,
      "learning_rate": 4.592439894517296e-06,
      "loss": 0.055782198905944824,
      "step": 5636
    },
    {
      "epoch": 0.7620785804816224,
      "grad_norm": 0.21689964830875397,
      "learning_rate": 4.587512087876559e-06,
      "loss": 0.05381143093109131,
      "step": 5637
    },
    {
      "epoch": 0.7622137727080693,
      "grad_norm": 0.36855658888816833,
      "learning_rate": 4.582586449171336e-06,
      "loss": 0.05515718460083008,
      "step": 5638
    },
    {
      "epoch": 0.7623489649345163,
      "grad_norm": 0.22664356231689453,
      "learning_rate": 4.577662979427168e-06,
      "loss": 0.04027509689331055,
      "step": 5639
    },
    {
      "epoch": 0.7624841571609633,
      "grad_norm": 0.5098745226860046,
      "learning_rate": 4.572741679669147e-06,
      "loss": 0.0943603515625,
      "step": 5640
    },
    {
      "epoch": 0.7626193493874103,
      "grad_norm": 0.2838180959224701,
      "learning_rate": 4.567822550921912e-06,
      "loss": 0.0681074857711792,
      "step": 5641
    },
    {
      "epoch": 0.7627545416138573,
      "grad_norm": 0.28811103105545044,
      "learning_rate": 4.562905594209647e-06,
      "loss": 0.05295705795288086,
      "step": 5642
    },
    {
      "epoch": 0.7628897338403042,
      "grad_norm": 0.4805182218551636,
      "learning_rate": 4.557990810556102e-06,
      "loss": 0.08113241195678711,
      "step": 5643
    },
    {
      "epoch": 0.7630249260667512,
      "grad_norm": 0.26393046975135803,
      "learning_rate": 4.553078200984553e-06,
      "loss": 0.0574946403503418,
      "step": 5644
    },
    {
      "epoch": 0.7631601182931982,
      "grad_norm": 0.41263237595558167,
      "learning_rate": 4.548167766517832e-06,
      "loss": 0.08613967895507812,
      "step": 5645
    },
    {
      "epoch": 0.7632953105196452,
      "grad_norm": 0.2544111907482147,
      "learning_rate": 4.543259508178318e-06,
      "loss": 0.05940127372741699,
      "step": 5646
    },
    {
      "epoch": 0.7634305027460921,
      "grad_norm": 0.21949402987957,
      "learning_rate": 4.538353426987931e-06,
      "loss": 0.05451488494873047,
      "step": 5647
    },
    {
      "epoch": 0.7635656949725391,
      "grad_norm": 0.30404141545295715,
      "learning_rate": 4.533449523968154e-06,
      "loss": 0.07459366321563721,
      "step": 5648
    },
    {
      "epoch": 0.7637008871989861,
      "grad_norm": 0.26245078444480896,
      "learning_rate": 4.528547800140001e-06,
      "loss": 0.05554288625717163,
      "step": 5649
    },
    {
      "epoch": 0.7638360794254331,
      "grad_norm": 0.3203220069408417,
      "learning_rate": 4.523648256524037e-06,
      "loss": 0.05026090145111084,
      "step": 5650
    },
    {
      "epoch": 0.7639712716518801,
      "grad_norm": 0.5115206837654114,
      "learning_rate": 4.518750894140372e-06,
      "loss": 0.09605836868286133,
      "step": 5651
    },
    {
      "epoch": 0.764106463878327,
      "grad_norm": 0.30319446325302124,
      "learning_rate": 4.513855714008659e-06,
      "loss": 0.07964706420898438,
      "step": 5652
    },
    {
      "epoch": 0.764241656104774,
      "grad_norm": 0.3309054970741272,
      "learning_rate": 4.508962717148111e-06,
      "loss": 0.060248732566833496,
      "step": 5653
    },
    {
      "epoch": 0.764376848331221,
      "grad_norm": 0.3443457782268524,
      "learning_rate": 4.504071904577469e-06,
      "loss": 0.07103848457336426,
      "step": 5654
    },
    {
      "epoch": 0.764512040557668,
      "grad_norm": 0.20302394032478333,
      "learning_rate": 4.499183277315027e-06,
      "loss": 0.040272653102874756,
      "step": 5655
    },
    {
      "epoch": 0.764647232784115,
      "grad_norm": 0.1711495816707611,
      "learning_rate": 4.494296836378625e-06,
      "loss": 0.04271411895751953,
      "step": 5656
    },
    {
      "epoch": 0.7647824250105619,
      "grad_norm": 0.3251838982105255,
      "learning_rate": 4.4894125827856415e-06,
      "loss": 0.0497438907623291,
      "step": 5657
    },
    {
      "epoch": 0.7649176172370089,
      "grad_norm": 0.26741740107536316,
      "learning_rate": 4.4845305175530105e-06,
      "loss": 0.053354740142822266,
      "step": 5658
    },
    {
      "epoch": 0.7650528094634559,
      "grad_norm": 0.3099420964717865,
      "learning_rate": 4.479650641697201e-06,
      "loss": 0.0755605697631836,
      "step": 5659
    },
    {
      "epoch": 0.7651880016899029,
      "grad_norm": 0.36197108030319214,
      "learning_rate": 4.4747729562342305e-06,
      "loss": 0.07172966003417969,
      "step": 5660
    },
    {
      "epoch": 0.7653231939163498,
      "grad_norm": 0.39862021803855896,
      "learning_rate": 4.469897462179656e-06,
      "loss": 0.06305938959121704,
      "step": 5661
    },
    {
      "epoch": 0.7654583861427968,
      "grad_norm": 0.4498077630996704,
      "learning_rate": 4.46502416054858e-06,
      "loss": 0.0718998908996582,
      "step": 5662
    },
    {
      "epoch": 0.7655935783692438,
      "grad_norm": 0.38541167974472046,
      "learning_rate": 4.460153052355663e-06,
      "loss": 0.06909990310668945,
      "step": 5663
    },
    {
      "epoch": 0.7657287705956908,
      "grad_norm": 0.2797292470932007,
      "learning_rate": 4.455284138615074e-06,
      "loss": 0.05756568908691406,
      "step": 5664
    },
    {
      "epoch": 0.7658639628221378,
      "grad_norm": 0.19385021924972534,
      "learning_rate": 4.4504174203405656e-06,
      "loss": 0.05148935317993164,
      "step": 5665
    },
    {
      "epoch": 0.7659991550485847,
      "grad_norm": 0.30873343348503113,
      "learning_rate": 4.445552898545407e-06,
      "loss": 0.0481414794921875,
      "step": 5666
    },
    {
      "epoch": 0.7661343472750317,
      "grad_norm": 0.21540817618370056,
      "learning_rate": 4.440690574242413e-06,
      "loss": 0.052638351917266846,
      "step": 5667
    },
    {
      "epoch": 0.7662695395014787,
      "grad_norm": 0.30132579803466797,
      "learning_rate": 4.435830448443961e-06,
      "loss": 0.05691814422607422,
      "step": 5668
    },
    {
      "epoch": 0.7664047317279257,
      "grad_norm": 0.37630772590637207,
      "learning_rate": 4.430972522161934e-06,
      "loss": 0.07995986938476562,
      "step": 5669
    },
    {
      "epoch": 0.7665399239543726,
      "grad_norm": 0.29022759199142456,
      "learning_rate": 4.426116796407794e-06,
      "loss": 0.06147289276123047,
      "step": 5670
    },
    {
      "epoch": 0.7666751161808196,
      "grad_norm": 0.28281182050704956,
      "learning_rate": 4.421263272192523e-06,
      "loss": 0.06667089462280273,
      "step": 5671
    },
    {
      "epoch": 0.7668103084072666,
      "grad_norm": 0.36075347661972046,
      "learning_rate": 4.416411950526648e-06,
      "loss": 0.06223917007446289,
      "step": 5672
    },
    {
      "epoch": 0.7669455006337136,
      "grad_norm": 0.24650578200817108,
      "learning_rate": 4.411562832420252e-06,
      "loss": 0.055058956146240234,
      "step": 5673
    },
    {
      "epoch": 0.7670806928601606,
      "grad_norm": 0.3108918070793152,
      "learning_rate": 4.406715918882929e-06,
      "loss": 0.059282779693603516,
      "step": 5674
    },
    {
      "epoch": 0.7672158850866075,
      "grad_norm": 0.2328052818775177,
      "learning_rate": 4.4018712109238475e-06,
      "loss": 0.05899167060852051,
      "step": 5675
    },
    {
      "epoch": 0.7673510773130545,
      "grad_norm": 0.4828590452671051,
      "learning_rate": 4.3970287095516965e-06,
      "loss": 0.052896976470947266,
      "step": 5676
    },
    {
      "epoch": 0.7674862695395015,
      "grad_norm": 0.2692180573940277,
      "learning_rate": 4.39218841577471e-06,
      "loss": 0.0655522346496582,
      "step": 5677
    },
    {
      "epoch": 0.7676214617659485,
      "grad_norm": 0.44871872663497925,
      "learning_rate": 4.387350330600662e-06,
      "loss": 0.07520866394042969,
      "step": 5678
    },
    {
      "epoch": 0.7677566539923955,
      "grad_norm": 0.3476618826389313,
      "learning_rate": 4.382514455036864e-06,
      "loss": 0.06399822235107422,
      "step": 5679
    },
    {
      "epoch": 0.7678918462188424,
      "grad_norm": 0.19479790329933167,
      "learning_rate": 4.377680790090182e-06,
      "loss": 0.05185079574584961,
      "step": 5680
    },
    {
      "epoch": 0.7680270384452894,
      "grad_norm": 0.2459198385477066,
      "learning_rate": 4.372849336767004e-06,
      "loss": 0.06309765577316284,
      "step": 5681
    },
    {
      "epoch": 0.7681622306717364,
      "grad_norm": 0.34146684408187866,
      "learning_rate": 4.3680200960732645e-06,
      "loss": 0.07272100448608398,
      "step": 5682
    },
    {
      "epoch": 0.7682974228981834,
      "grad_norm": 0.3710181713104248,
      "learning_rate": 4.363193069014439e-06,
      "loss": 0.0505526065826416,
      "step": 5683
    },
    {
      "epoch": 0.7684326151246303,
      "grad_norm": 0.4271155893802643,
      "learning_rate": 4.3583682565955325e-06,
      "loss": 0.09519720077514648,
      "step": 5684
    },
    {
      "epoch": 0.7685678073510773,
      "grad_norm": 0.3301430940628052,
      "learning_rate": 4.3535456598211074e-06,
      "loss": 0.061621665954589844,
      "step": 5685
    },
    {
      "epoch": 0.7687029995775243,
      "grad_norm": 0.3569306433200836,
      "learning_rate": 4.348725279695251e-06,
      "loss": 0.0776815414428711,
      "step": 5686
    },
    {
      "epoch": 0.7688381918039713,
      "grad_norm": 0.19864505529403687,
      "learning_rate": 4.343907117221591e-06,
      "loss": 0.03427314758300781,
      "step": 5687
    },
    {
      "epoch": 0.7689733840304183,
      "grad_norm": 0.1509825736284256,
      "learning_rate": 4.339091173403294e-06,
      "loss": 0.041939735412597656,
      "step": 5688
    },
    {
      "epoch": 0.7691085762568652,
      "grad_norm": 0.20300842821598053,
      "learning_rate": 4.334277449243061e-06,
      "loss": 0.041044920682907104,
      "step": 5689
    },
    {
      "epoch": 0.7692437684833122,
      "grad_norm": 0.20278188586235046,
      "learning_rate": 4.329465945743144e-06,
      "loss": 0.048657894134521484,
      "step": 5690
    },
    {
      "epoch": 0.7693789607097592,
      "grad_norm": 0.41579997539520264,
      "learning_rate": 4.32465666390532e-06,
      "loss": 0.0794367790222168,
      "step": 5691
    },
    {
      "epoch": 0.7695141529362062,
      "grad_norm": 0.4104713797569275,
      "learning_rate": 4.319849604730905e-06,
      "loss": 0.06893670558929443,
      "step": 5692
    },
    {
      "epoch": 0.7696493451626532,
      "grad_norm": 0.2362470179796219,
      "learning_rate": 4.315044769220758e-06,
      "loss": 0.06392478942871094,
      "step": 5693
    },
    {
      "epoch": 0.7697845373891001,
      "grad_norm": 0.3901192545890808,
      "learning_rate": 4.310242158375264e-06,
      "loss": 0.06928801536560059,
      "step": 5694
    },
    {
      "epoch": 0.7699197296155471,
      "grad_norm": 0.19646479189395905,
      "learning_rate": 4.30544177319436e-06,
      "loss": 0.04744386672973633,
      "step": 5695
    },
    {
      "epoch": 0.7700549218419941,
      "grad_norm": 0.3500657081604004,
      "learning_rate": 4.300643614677511e-06,
      "loss": 0.05603682994842529,
      "step": 5696
    },
    {
      "epoch": 0.7701901140684411,
      "grad_norm": 0.3439147472381592,
      "learning_rate": 4.2958476838237165e-06,
      "loss": 0.06445038318634033,
      "step": 5697
    },
    {
      "epoch": 0.770325306294888,
      "grad_norm": 0.2937512695789337,
      "learning_rate": 4.2910539816315166e-06,
      "loss": 0.06619882583618164,
      "step": 5698
    },
    {
      "epoch": 0.770460498521335,
      "grad_norm": 0.1523667722940445,
      "learning_rate": 4.286262509098979e-06,
      "loss": 0.043492794036865234,
      "step": 5699
    },
    {
      "epoch": 0.770595690747782,
      "grad_norm": 0.44285938143730164,
      "learning_rate": 4.28147326722373e-06,
      "loss": 0.09971809387207031,
      "step": 5700
    },
    {
      "epoch": 0.770730882974229,
      "grad_norm": 0.17285388708114624,
      "learning_rate": 4.2766862570028965e-06,
      "loss": 0.04359164834022522,
      "step": 5701
    },
    {
      "epoch": 0.770866075200676,
      "grad_norm": 0.21552710235118866,
      "learning_rate": 4.2719014794331715e-06,
      "loss": 0.05307817459106445,
      "step": 5702
    },
    {
      "epoch": 0.7710012674271229,
      "grad_norm": 0.2892248034477234,
      "learning_rate": 4.267118935510767e-06,
      "loss": 0.0508270263671875,
      "step": 5703
    },
    {
      "epoch": 0.7711364596535699,
      "grad_norm": 0.2051730453968048,
      "learning_rate": 4.2623386262314306e-06,
      "loss": 0.0408627986907959,
      "step": 5704
    },
    {
      "epoch": 0.7712716518800169,
      "grad_norm": 0.24431021511554718,
      "learning_rate": 4.257560552590461e-06,
      "loss": 0.05776238441467285,
      "step": 5705
    },
    {
      "epoch": 0.7714068441064639,
      "grad_norm": 0.49438828229904175,
      "learning_rate": 4.252784715582661e-06,
      "loss": 0.053160011768341064,
      "step": 5706
    },
    {
      "epoch": 0.7715420363329109,
      "grad_norm": 0.46312135457992554,
      "learning_rate": 4.2480111162024e-06,
      "loss": 0.060565948486328125,
      "step": 5707
    },
    {
      "epoch": 0.7716772285593578,
      "grad_norm": 0.3173055350780487,
      "learning_rate": 4.243239755443561e-06,
      "loss": 0.0741962194442749,
      "step": 5708
    },
    {
      "epoch": 0.7718124207858048,
      "grad_norm": 0.3241714537143707,
      "learning_rate": 4.238470634299567e-06,
      "loss": 0.06093263626098633,
      "step": 5709
    },
    {
      "epoch": 0.7719476130122518,
      "grad_norm": 0.41119104623794556,
      "learning_rate": 4.233703753763375e-06,
      "loss": 0.08883190155029297,
      "step": 5710
    },
    {
      "epoch": 0.7720828052386988,
      "grad_norm": 0.3727583587169647,
      "learning_rate": 4.228939114827469e-06,
      "loss": 0.08051329851150513,
      "step": 5711
    },
    {
      "epoch": 0.7722179974651457,
      "grad_norm": 0.22532910108566284,
      "learning_rate": 4.224176718483881e-06,
      "loss": 0.04881012439727783,
      "step": 5712
    },
    {
      "epoch": 0.7723531896915927,
      "grad_norm": 0.199547678232193,
      "learning_rate": 4.219416565724165e-06,
      "loss": 0.03948569297790527,
      "step": 5713
    },
    {
      "epoch": 0.7724883819180397,
      "grad_norm": 0.3262271285057068,
      "learning_rate": 4.21465865753941e-06,
      "loss": 0.058938026428222656,
      "step": 5714
    },
    {
      "epoch": 0.7726235741444867,
      "grad_norm": 0.3589162528514862,
      "learning_rate": 4.209902994920236e-06,
      "loss": 0.07290387153625488,
      "step": 5715
    },
    {
      "epoch": 0.7727587663709337,
      "grad_norm": 0.45577287673950195,
      "learning_rate": 4.205149578856794e-06,
      "loss": 0.07767486572265625,
      "step": 5716
    },
    {
      "epoch": 0.7728939585973806,
      "grad_norm": 0.3757721483707428,
      "learning_rate": 4.200398410338779e-06,
      "loss": 0.0785365104675293,
      "step": 5717
    },
    {
      "epoch": 0.7730291508238276,
      "grad_norm": 0.38008055090904236,
      "learning_rate": 4.1956494903554056e-06,
      "loss": 0.08467793464660645,
      "step": 5718
    },
    {
      "epoch": 0.7731643430502746,
      "grad_norm": 0.2644762396812439,
      "learning_rate": 4.190902819895425e-06,
      "loss": 0.05213308334350586,
      "step": 5719
    },
    {
      "epoch": 0.7732995352767216,
      "grad_norm": 0.29041609168052673,
      "learning_rate": 4.186158399947118e-06,
      "loss": 0.07232236862182617,
      "step": 5720
    },
    {
      "epoch": 0.7734347275031686,
      "grad_norm": 0.2394200712442398,
      "learning_rate": 4.181416231498292e-06,
      "loss": 0.05535078048706055,
      "step": 5721
    },
    {
      "epoch": 0.7735699197296155,
      "grad_norm": 0.33892616629600525,
      "learning_rate": 4.176676315536306e-06,
      "loss": 0.06801414489746094,
      "step": 5722
    },
    {
      "epoch": 0.7737051119560625,
      "grad_norm": 0.3413313329219818,
      "learning_rate": 4.171938653048027e-06,
      "loss": 0.06963539123535156,
      "step": 5723
    },
    {
      "epoch": 0.7738403041825095,
      "grad_norm": 0.39870211482048035,
      "learning_rate": 4.1672032450198616e-06,
      "loss": 0.0765681266784668,
      "step": 5724
    },
    {
      "epoch": 0.7739754964089565,
      "grad_norm": 0.7043061852455139,
      "learning_rate": 4.16247009243775e-06,
      "loss": 0.10092782974243164,
      "step": 5725
    },
    {
      "epoch": 0.7741106886354034,
      "grad_norm": 0.36140236258506775,
      "learning_rate": 4.1577391962871504e-06,
      "loss": 0.07085418701171875,
      "step": 5726
    },
    {
      "epoch": 0.7742458808618504,
      "grad_norm": 0.38208651542663574,
      "learning_rate": 4.153010557553076e-06,
      "loss": 0.07592105865478516,
      "step": 5727
    },
    {
      "epoch": 0.7743810730882974,
      "grad_norm": 0.1892228126525879,
      "learning_rate": 4.148284177220045e-06,
      "loss": 0.04671287536621094,
      "step": 5728
    },
    {
      "epoch": 0.7745162653147444,
      "grad_norm": 0.28675931692123413,
      "learning_rate": 4.143560056272117e-06,
      "loss": 0.07384371757507324,
      "step": 5729
    },
    {
      "epoch": 0.7746514575411914,
      "grad_norm": 0.30072644352912903,
      "learning_rate": 4.1388381956928796e-06,
      "loss": 0.05128073692321777,
      "step": 5730
    },
    {
      "epoch": 0.7747866497676383,
      "grad_norm": 0.284824401140213,
      "learning_rate": 4.134118596465443e-06,
      "loss": 0.0562664270401001,
      "step": 5731
    },
    {
      "epoch": 0.7749218419940853,
      "grad_norm": 0.416551411151886,
      "learning_rate": 4.1294012595724675e-06,
      "loss": 0.06118416786193848,
      "step": 5732
    },
    {
      "epoch": 0.7750570342205323,
      "grad_norm": 0.26712268590927124,
      "learning_rate": 4.1246861859961114e-06,
      "loss": 0.07384276390075684,
      "step": 5733
    },
    {
      "epoch": 0.7751922264469793,
      "grad_norm": 0.3755200207233429,
      "learning_rate": 4.119973376718089e-06,
      "loss": 0.07685661315917969,
      "step": 5734
    },
    {
      "epoch": 0.7753274186734262,
      "grad_norm": 0.3086282014846802,
      "learning_rate": 4.115262832719628e-06,
      "loss": 0.04568958282470703,
      "step": 5735
    },
    {
      "epoch": 0.7754626108998732,
      "grad_norm": 0.36932235956192017,
      "learning_rate": 4.110554554981486e-06,
      "loss": 0.06133699417114258,
      "step": 5736
    },
    {
      "epoch": 0.7755978031263202,
      "grad_norm": 0.1857132464647293,
      "learning_rate": 4.1058485444839655e-06,
      "loss": 0.04334378242492676,
      "step": 5737
    },
    {
      "epoch": 0.7757329953527672,
      "grad_norm": 0.3986019790172577,
      "learning_rate": 4.101144802206862e-06,
      "loss": 0.07707595825195312,
      "step": 5738
    },
    {
      "epoch": 0.7758681875792142,
      "grad_norm": 0.19556692242622375,
      "learning_rate": 4.096443329129535e-06,
      "loss": 0.0461277961730957,
      "step": 5739
    },
    {
      "epoch": 0.7760033798056611,
      "grad_norm": 0.3307970464229584,
      "learning_rate": 4.091744126230853e-06,
      "loss": 0.0712122917175293,
      "step": 5740
    },
    {
      "epoch": 0.7761385720321081,
      "grad_norm": 0.4119226634502411,
      "learning_rate": 4.08704719448921e-06,
      "loss": 0.07254886627197266,
      "step": 5741
    },
    {
      "epoch": 0.7762737642585551,
      "grad_norm": 0.1848524808883667,
      "learning_rate": 4.082352534882543e-06,
      "loss": 0.03823113441467285,
      "step": 5742
    },
    {
      "epoch": 0.7764089564850021,
      "grad_norm": 0.2473663091659546,
      "learning_rate": 4.07766014838829e-06,
      "loss": 0.061011314392089844,
      "step": 5743
    },
    {
      "epoch": 0.7765441487114491,
      "grad_norm": 0.2692262530326843,
      "learning_rate": 4.072970035983443e-06,
      "loss": 0.05391550064086914,
      "step": 5744
    },
    {
      "epoch": 0.776679340937896,
      "grad_norm": 0.3563028872013092,
      "learning_rate": 4.068282198644505e-06,
      "loss": 0.08204364776611328,
      "step": 5745
    },
    {
      "epoch": 0.776814533164343,
      "grad_norm": 0.34030359983444214,
      "learning_rate": 4.06359663734751e-06,
      "loss": 0.04147148132324219,
      "step": 5746
    },
    {
      "epoch": 0.77694972539079,
      "grad_norm": 0.6173545122146606,
      "learning_rate": 4.058913353068013e-06,
      "loss": 0.08217430114746094,
      "step": 5747
    },
    {
      "epoch": 0.777084917617237,
      "grad_norm": 0.3259998559951782,
      "learning_rate": 4.0542323467810985e-06,
      "loss": 0.06691491603851318,
      "step": 5748
    },
    {
      "epoch": 0.777220109843684,
      "grad_norm": 0.28305602073669434,
      "learning_rate": 4.049553619461381e-06,
      "loss": 0.03863120079040527,
      "step": 5749
    },
    {
      "epoch": 0.7773553020701309,
      "grad_norm": 0.232561394572258,
      "learning_rate": 4.044877172082997e-06,
      "loss": 0.05492901802062988,
      "step": 5750
    },
    {
      "epoch": 0.7774904942965779,
      "grad_norm": 0.37049171328544617,
      "learning_rate": 4.040203005619604e-06,
      "loss": 0.06410980224609375,
      "step": 5751
    },
    {
      "epoch": 0.7776256865230249,
      "grad_norm": 0.24890120327472687,
      "learning_rate": 4.035531121044392e-06,
      "loss": 0.07123613357543945,
      "step": 5752
    },
    {
      "epoch": 0.7777608787494719,
      "grad_norm": 0.2749079763889313,
      "learning_rate": 4.030861519330065e-06,
      "loss": 0.06884384155273438,
      "step": 5753
    },
    {
      "epoch": 0.7778960709759188,
      "grad_norm": 0.5079923868179321,
      "learning_rate": 4.026194201448868e-06,
      "loss": 0.08815860748291016,
      "step": 5754
    },
    {
      "epoch": 0.7780312632023658,
      "grad_norm": 0.46686694025993347,
      "learning_rate": 4.021529168372558e-06,
      "loss": 0.0786428451538086,
      "step": 5755
    },
    {
      "epoch": 0.7781664554288128,
      "grad_norm": 0.2763957977294922,
      "learning_rate": 4.01686642107242e-06,
      "loss": 0.05726909637451172,
      "step": 5756
    },
    {
      "epoch": 0.7783016476552598,
      "grad_norm": 0.29116520285606384,
      "learning_rate": 4.0122059605192624e-06,
      "loss": 0.06893444061279297,
      "step": 5757
    },
    {
      "epoch": 0.7784368398817068,
      "grad_norm": 0.19644325971603394,
      "learning_rate": 4.007547787683412e-06,
      "loss": 0.04460024833679199,
      "step": 5758
    },
    {
      "epoch": 0.7785720321081537,
      "grad_norm": 0.19754086434841156,
      "learning_rate": 4.002891903534736e-06,
      "loss": 0.05706286430358887,
      "step": 5759
    },
    {
      "epoch": 0.7787072243346007,
      "grad_norm": 0.5186466574668884,
      "learning_rate": 3.998238309042611e-06,
      "loss": 0.096160888671875,
      "step": 5760
    },
    {
      "epoch": 0.7788424165610477,
      "grad_norm": 0.30563274025917053,
      "learning_rate": 3.993587005175937e-06,
      "loss": 0.06788778305053711,
      "step": 5761
    },
    {
      "epoch": 0.7789776087874947,
      "grad_norm": 0.2985718846321106,
      "learning_rate": 3.988937992903144e-06,
      "loss": 0.07439279556274414,
      "step": 5762
    },
    {
      "epoch": 0.7791128010139416,
      "grad_norm": 0.3044978678226471,
      "learning_rate": 3.9842912731921716e-06,
      "loss": 0.06938791275024414,
      "step": 5763
    },
    {
      "epoch": 0.7792479932403886,
      "grad_norm": 0.1739664226770401,
      "learning_rate": 3.979646847010506e-06,
      "loss": 0.05845212936401367,
      "step": 5764
    },
    {
      "epoch": 0.7793831854668356,
      "grad_norm": 0.4711625576019287,
      "learning_rate": 3.975004715325134e-06,
      "loss": 0.08249151706695557,
      "step": 5765
    },
    {
      "epoch": 0.7795183776932826,
      "grad_norm": 0.28729113936424255,
      "learning_rate": 3.970364879102572e-06,
      "loss": 0.06833052635192871,
      "step": 5766
    },
    {
      "epoch": 0.7796535699197296,
      "grad_norm": 0.3286392390727997,
      "learning_rate": 3.96572733930886e-06,
      "loss": 0.058071136474609375,
      "step": 5767
    },
    {
      "epoch": 0.7797887621461765,
      "grad_norm": 0.43296244740486145,
      "learning_rate": 3.961092096909552e-06,
      "loss": 0.06349897384643555,
      "step": 5768
    },
    {
      "epoch": 0.7799239543726236,
      "grad_norm": 0.28133082389831543,
      "learning_rate": 3.9564591528697455e-06,
      "loss": 0.046489715576171875,
      "step": 5769
    },
    {
      "epoch": 0.7800591465990706,
      "grad_norm": 0.44722869992256165,
      "learning_rate": 3.9518285081540275e-06,
      "loss": 0.0754251480102539,
      "step": 5770
    },
    {
      "epoch": 0.7801943388255176,
      "grad_norm": 0.40870973467826843,
      "learning_rate": 3.947200163726534e-06,
      "loss": 0.06529468297958374,
      "step": 5771
    },
    {
      "epoch": 0.7803295310519646,
      "grad_norm": 0.14248031377792358,
      "learning_rate": 3.9425741205509055e-06,
      "loss": 0.033555030822753906,
      "step": 5772
    },
    {
      "epoch": 0.7804647232784115,
      "grad_norm": 0.3367745876312256,
      "learning_rate": 3.9379503795903065e-06,
      "loss": 0.06927061080932617,
      "step": 5773
    },
    {
      "epoch": 0.7805999155048585,
      "grad_norm": 0.29812562465667725,
      "learning_rate": 3.933328941807439e-06,
      "loss": 0.06696891784667969,
      "step": 5774
    },
    {
      "epoch": 0.7807351077313055,
      "grad_norm": 0.349993497133255,
      "learning_rate": 3.928709808164491e-06,
      "loss": 0.0759577751159668,
      "step": 5775
    },
    {
      "epoch": 0.7808702999577525,
      "grad_norm": 0.36581605672836304,
      "learning_rate": 3.924092979623203e-06,
      "loss": 0.08065176010131836,
      "step": 5776
    },
    {
      "epoch": 0.7810054921841995,
      "grad_norm": 0.21663056313991547,
      "learning_rate": 3.919478457144824e-06,
      "loss": 0.047911882400512695,
      "step": 5777
    },
    {
      "epoch": 0.7811406844106464,
      "grad_norm": 0.3258569836616516,
      "learning_rate": 3.914866241690115e-06,
      "loss": 0.06820964813232422,
      "step": 5778
    },
    {
      "epoch": 0.7812758766370934,
      "grad_norm": 0.2599864900112152,
      "learning_rate": 3.9102563342193695e-06,
      "loss": 0.044528961181640625,
      "step": 5779
    },
    {
      "epoch": 0.7814110688635404,
      "grad_norm": 0.2585674226284027,
      "learning_rate": 3.905648735692389e-06,
      "loss": 0.04642355442047119,
      "step": 5780
    },
    {
      "epoch": 0.7815462610899874,
      "grad_norm": 0.35366174578666687,
      "learning_rate": 3.901043447068508e-06,
      "loss": 0.06553888320922852,
      "step": 5781
    },
    {
      "epoch": 0.7816814533164343,
      "grad_norm": 0.37189918756484985,
      "learning_rate": 3.896440469306567e-06,
      "loss": 0.08901119232177734,
      "step": 5782
    },
    {
      "epoch": 0.7818166455428813,
      "grad_norm": 0.20578940212726593,
      "learning_rate": 3.891839803364934e-06,
      "loss": 0.04856395721435547,
      "step": 5783
    },
    {
      "epoch": 0.7819518377693283,
      "grad_norm": 0.17473185062408447,
      "learning_rate": 3.887241450201487e-06,
      "loss": 0.03539228439331055,
      "step": 5784
    },
    {
      "epoch": 0.7820870299957753,
      "grad_norm": 0.30684182047843933,
      "learning_rate": 3.882645410773629e-06,
      "loss": 0.06535720825195312,
      "step": 5785
    },
    {
      "epoch": 0.7822222222222223,
      "grad_norm": 0.22718492150306702,
      "learning_rate": 3.878051686038284e-06,
      "loss": 0.04875516891479492,
      "step": 5786
    },
    {
      "epoch": 0.7823574144486692,
      "grad_norm": 0.27810609340667725,
      "learning_rate": 3.873460276951889e-06,
      "loss": 0.047110557556152344,
      "step": 5787
    },
    {
      "epoch": 0.7824926066751162,
      "grad_norm": 0.3875526189804077,
      "learning_rate": 3.868871184470397e-06,
      "loss": 0.06734180450439453,
      "step": 5788
    },
    {
      "epoch": 0.7826277989015632,
      "grad_norm": 0.36902740597724915,
      "learning_rate": 3.864284409549282e-06,
      "loss": 0.08010327816009521,
      "step": 5789
    },
    {
      "epoch": 0.7827629911280102,
      "grad_norm": 0.2538996636867523,
      "learning_rate": 3.859699953143532e-06,
      "loss": 0.05434715747833252,
      "step": 5790
    },
    {
      "epoch": 0.7828981833544572,
      "grad_norm": 0.21697865426540375,
      "learning_rate": 3.855117816207665e-06,
      "loss": 0.04898500442504883,
      "step": 5791
    },
    {
      "epoch": 0.7830333755809041,
      "grad_norm": 0.17005237936973572,
      "learning_rate": 3.850537999695699e-06,
      "loss": 0.04483509063720703,
      "step": 5792
    },
    {
      "epoch": 0.7831685678073511,
      "grad_norm": 0.2478007972240448,
      "learning_rate": 3.845960504561179e-06,
      "loss": 0.05867290496826172,
      "step": 5793
    },
    {
      "epoch": 0.7833037600337981,
      "grad_norm": 0.219625785946846,
      "learning_rate": 3.841385331757161e-06,
      "loss": 0.04551362991333008,
      "step": 5794
    },
    {
      "epoch": 0.7834389522602451,
      "grad_norm": 0.26169317960739136,
      "learning_rate": 3.8368124822362184e-06,
      "loss": 0.06520521640777588,
      "step": 5795
    },
    {
      "epoch": 0.783574144486692,
      "grad_norm": 0.35230061411857605,
      "learning_rate": 3.832241956950449e-06,
      "loss": 0.057332515716552734,
      "step": 5796
    },
    {
      "epoch": 0.783709336713139,
      "grad_norm": 0.32186058163642883,
      "learning_rate": 3.82767375685146e-06,
      "loss": 0.07688665390014648,
      "step": 5797
    },
    {
      "epoch": 0.783844528939586,
      "grad_norm": 0.5881981253623962,
      "learning_rate": 3.823107882890373e-06,
      "loss": 0.09408235549926758,
      "step": 5798
    },
    {
      "epoch": 0.783979721166033,
      "grad_norm": 0.29914772510528564,
      "learning_rate": 3.8185443360178265e-06,
      "loss": 0.05277276039123535,
      "step": 5799
    },
    {
      "epoch": 0.78411491339248,
      "grad_norm": 0.2215801328420639,
      "learning_rate": 3.813983117183973e-06,
      "loss": 0.049486637115478516,
      "step": 5800
    },
    {
      "epoch": 0.7842501056189269,
      "grad_norm": 0.35752561688423157,
      "learning_rate": 3.8094242273384932e-06,
      "loss": 0.0811152458190918,
      "step": 5801
    },
    {
      "epoch": 0.7843852978453739,
      "grad_norm": 0.42034009099006653,
      "learning_rate": 3.804867667430555e-06,
      "loss": 0.0776982307434082,
      "step": 5802
    },
    {
      "epoch": 0.7845204900718209,
      "grad_norm": 0.18439221382141113,
      "learning_rate": 3.800313438408874e-06,
      "loss": 0.04037666320800781,
      "step": 5803
    },
    {
      "epoch": 0.7846556822982679,
      "grad_norm": 0.3341442942619324,
      "learning_rate": 3.7957615412216582e-06,
      "loss": 0.07306098937988281,
      "step": 5804
    },
    {
      "epoch": 0.7847908745247149,
      "grad_norm": 0.4744199216365814,
      "learning_rate": 3.791211976816634e-06,
      "loss": 0.0776371955871582,
      "step": 5805
    },
    {
      "epoch": 0.7849260667511618,
      "grad_norm": 0.4500978887081146,
      "learning_rate": 3.786664746141057e-06,
      "loss": 0.06372880935668945,
      "step": 5806
    },
    {
      "epoch": 0.7850612589776088,
      "grad_norm": 0.3885696232318878,
      "learning_rate": 3.782119850141669e-06,
      "loss": 0.060495853424072266,
      "step": 5807
    },
    {
      "epoch": 0.7851964512040558,
      "grad_norm": 0.7141410112380981,
      "learning_rate": 3.777577289764752e-06,
      "loss": 0.08289527893066406,
      "step": 5808
    },
    {
      "epoch": 0.7853316434305028,
      "grad_norm": 0.3788013756275177,
      "learning_rate": 3.7730370659560904e-06,
      "loss": 0.05214107036590576,
      "step": 5809
    },
    {
      "epoch": 0.7854668356569497,
      "grad_norm": 0.28666380047798157,
      "learning_rate": 3.7684991796609746e-06,
      "loss": 0.05033731460571289,
      "step": 5810
    },
    {
      "epoch": 0.7856020278833967,
      "grad_norm": 0.21786850690841675,
      "learning_rate": 3.7639636318242344e-06,
      "loss": 0.05078268051147461,
      "step": 5811
    },
    {
      "epoch": 0.7857372201098437,
      "grad_norm": 0.34412282705307007,
      "learning_rate": 3.7594304233901738e-06,
      "loss": 0.08062267303466797,
      "step": 5812
    },
    {
      "epoch": 0.7858724123362907,
      "grad_norm": 0.22413675487041473,
      "learning_rate": 3.754899555302645e-06,
      "loss": 0.03797823190689087,
      "step": 5813
    },
    {
      "epoch": 0.7860076045627377,
      "grad_norm": 0.19559744000434875,
      "learning_rate": 3.7503710285049964e-06,
      "loss": 0.04233598709106445,
      "step": 5814
    },
    {
      "epoch": 0.7861427967891846,
      "grad_norm": 0.32189080119132996,
      "learning_rate": 3.7458448439400888e-06,
      "loss": 0.061079978942871094,
      "step": 5815
    },
    {
      "epoch": 0.7862779890156316,
      "grad_norm": 0.1955747902393341,
      "learning_rate": 3.7413210025502985e-06,
      "loss": 0.03945612907409668,
      "step": 5816
    },
    {
      "epoch": 0.7864131812420786,
      "grad_norm": 0.3130621016025543,
      "learning_rate": 3.7367995052775123e-06,
      "loss": 0.05901479721069336,
      "step": 5817
    },
    {
      "epoch": 0.7865483734685256,
      "grad_norm": 0.19198979437351227,
      "learning_rate": 3.732280353063133e-06,
      "loss": 0.05481517314910889,
      "step": 5818
    },
    {
      "epoch": 0.7866835656949726,
      "grad_norm": 0.46729937195777893,
      "learning_rate": 3.727763546848074e-06,
      "loss": 0.06125986576080322,
      "step": 5819
    },
    {
      "epoch": 0.7868187579214195,
      "grad_norm": 0.24033065140247345,
      "learning_rate": 3.7232490875727544e-06,
      "loss": 0.04735279083251953,
      "step": 5820
    },
    {
      "epoch": 0.7869539501478665,
      "grad_norm": 0.3184490501880646,
      "learning_rate": 3.718736976177108e-06,
      "loss": 0.07044410705566406,
      "step": 5821
    },
    {
      "epoch": 0.7870891423743135,
      "grad_norm": 0.24775823950767517,
      "learning_rate": 3.71422721360058e-06,
      "loss": 0.055655479431152344,
      "step": 5822
    },
    {
      "epoch": 0.7872243346007605,
      "grad_norm": 0.18539845943450928,
      "learning_rate": 3.709719800782133e-06,
      "loss": 0.04623746871948242,
      "step": 5823
    },
    {
      "epoch": 0.7873595268272074,
      "grad_norm": 0.47386887669563293,
      "learning_rate": 3.7052147386602304e-06,
      "loss": 0.0531386137008667,
      "step": 5824
    },
    {
      "epoch": 0.7874947190536544,
      "grad_norm": 0.2164570838212967,
      "learning_rate": 3.700712028172851e-06,
      "loss": 0.04325532913208008,
      "step": 5825
    },
    {
      "epoch": 0.7876299112801014,
      "grad_norm": 0.29097384214401245,
      "learning_rate": 3.696211670257481e-06,
      "loss": 0.06789112091064453,
      "step": 5826
    },
    {
      "epoch": 0.7877651035065484,
      "grad_norm": 0.5457591414451599,
      "learning_rate": 3.691713665851117e-06,
      "loss": 0.08944034576416016,
      "step": 5827
    },
    {
      "epoch": 0.7879002957329954,
      "grad_norm": 0.3387489914894104,
      "learning_rate": 3.6872180158902764e-06,
      "loss": 0.0661458969116211,
      "step": 5828
    },
    {
      "epoch": 0.7880354879594423,
      "grad_norm": 0.5111722946166992,
      "learning_rate": 3.6827247213109705e-06,
      "loss": 0.060814619064331055,
      "step": 5829
    },
    {
      "epoch": 0.7881706801858893,
      "grad_norm": 0.28169646859169006,
      "learning_rate": 3.6782337830487294e-06,
      "loss": 0.05659973621368408,
      "step": 5830
    },
    {
      "epoch": 0.7883058724123363,
      "grad_norm": 0.3352935314178467,
      "learning_rate": 3.6737452020385886e-06,
      "loss": 0.05640769004821777,
      "step": 5831
    },
    {
      "epoch": 0.7884410646387833,
      "grad_norm": 0.20845596492290497,
      "learning_rate": 3.6692589792150923e-06,
      "loss": 0.04909205436706543,
      "step": 5832
    },
    {
      "epoch": 0.7885762568652303,
      "grad_norm": 0.20986799895763397,
      "learning_rate": 3.6647751155123026e-06,
      "loss": 0.05309581756591797,
      "step": 5833
    },
    {
      "epoch": 0.7887114490916772,
      "grad_norm": 0.1778966784477234,
      "learning_rate": 3.660293611863782e-06,
      "loss": 0.050807952880859375,
      "step": 5834
    },
    {
      "epoch": 0.7888466413181242,
      "grad_norm": 0.473964124917984,
      "learning_rate": 3.655814469202602e-06,
      "loss": 0.0950632095336914,
      "step": 5835
    },
    {
      "epoch": 0.7889818335445712,
      "grad_norm": 0.29631367325782776,
      "learning_rate": 3.6513376884613446e-06,
      "loss": 0.07864713668823242,
      "step": 5836
    },
    {
      "epoch": 0.7891170257710182,
      "grad_norm": 0.25132322311401367,
      "learning_rate": 3.6468632705720934e-06,
      "loss": 0.05677342414855957,
      "step": 5837
    },
    {
      "epoch": 0.7892522179974651,
      "grad_norm": 0.39402106404304504,
      "learning_rate": 3.6423912164664606e-06,
      "loss": 0.06469893455505371,
      "step": 5838
    },
    {
      "epoch": 0.7893874102239121,
      "grad_norm": 0.18110719323158264,
      "learning_rate": 3.637921527075534e-06,
      "loss": 0.04526472091674805,
      "step": 5839
    },
    {
      "epoch": 0.7895226024503591,
      "grad_norm": 0.31015798449516296,
      "learning_rate": 3.63345420332994e-06,
      "loss": 0.06031608581542969,
      "step": 5840
    },
    {
      "epoch": 0.7896577946768061,
      "grad_norm": 0.2274787276983261,
      "learning_rate": 3.628989246159795e-06,
      "loss": 0.06720209121704102,
      "step": 5841
    },
    {
      "epoch": 0.7897929869032531,
      "grad_norm": 0.42186152935028076,
      "learning_rate": 3.6245266564947205e-06,
      "loss": 0.06429600715637207,
      "step": 5842
    },
    {
      "epoch": 0.7899281791297,
      "grad_norm": 0.2879282534122467,
      "learning_rate": 3.620066435263868e-06,
      "loss": 0.06023454666137695,
      "step": 5843
    },
    {
      "epoch": 0.790063371356147,
      "grad_norm": 0.5149502158164978,
      "learning_rate": 3.6156085833958596e-06,
      "loss": 0.06293487548828125,
      "step": 5844
    },
    {
      "epoch": 0.790198563582594,
      "grad_norm": 0.3775785565376282,
      "learning_rate": 3.6111531018188584e-06,
      "loss": 0.07205677032470703,
      "step": 5845
    },
    {
      "epoch": 0.790333755809041,
      "grad_norm": 0.24140354990959167,
      "learning_rate": 3.606699991460513e-06,
      "loss": 0.046868324279785156,
      "step": 5846
    },
    {
      "epoch": 0.790468948035488,
      "grad_norm": 0.4178244173526764,
      "learning_rate": 3.602249253247986e-06,
      "loss": 0.08034706115722656,
      "step": 5847
    },
    {
      "epoch": 0.7906041402619349,
      "grad_norm": 0.2813854217529297,
      "learning_rate": 3.5978008881079445e-06,
      "loss": 0.05697774887084961,
      "step": 5848
    },
    {
      "epoch": 0.7907393324883819,
      "grad_norm": 0.21201984584331512,
      "learning_rate": 3.5933548969665587e-06,
      "loss": 0.038930654525756836,
      "step": 5849
    },
    {
      "epoch": 0.7908745247148289,
      "grad_norm": 0.18337924778461456,
      "learning_rate": 3.5889112807495152e-06,
      "loss": 0.04210853576660156,
      "step": 5850
    },
    {
      "epoch": 0.7910097169412759,
      "grad_norm": 0.3261737525463104,
      "learning_rate": 3.5844700403819935e-06,
      "loss": 0.06548404693603516,
      "step": 5851
    },
    {
      "epoch": 0.7911449091677228,
      "grad_norm": 0.2853728234767914,
      "learning_rate": 3.5800311767886847e-06,
      "loss": 0.05063962936401367,
      "step": 5852
    },
    {
      "epoch": 0.7912801013941698,
      "grad_norm": 0.31865543127059937,
      "learning_rate": 3.575594690893784e-06,
      "loss": 0.07824563980102539,
      "step": 5853
    },
    {
      "epoch": 0.7914152936206168,
      "grad_norm": 0.30337998270988464,
      "learning_rate": 3.5711605836209853e-06,
      "loss": 0.05076479911804199,
      "step": 5854
    },
    {
      "epoch": 0.7915504858470638,
      "grad_norm": 0.3420272469520569,
      "learning_rate": 3.566728855893505e-06,
      "loss": 0.07276201248168945,
      "step": 5855
    },
    {
      "epoch": 0.7916856780735108,
      "grad_norm": 0.28453660011291504,
      "learning_rate": 3.5622995086340466e-06,
      "loss": 0.07154130935668945,
      "step": 5856
    },
    {
      "epoch": 0.7918208702999577,
      "grad_norm": 0.2686252295970917,
      "learning_rate": 3.5578725427648233e-06,
      "loss": 0.05756998062133789,
      "step": 5857
    },
    {
      "epoch": 0.7919560625264047,
      "grad_norm": 0.20000793039798737,
      "learning_rate": 3.553447959207553e-06,
      "loss": 0.04842329025268555,
      "step": 5858
    },
    {
      "epoch": 0.7920912547528517,
      "grad_norm": 0.4091678857803345,
      "learning_rate": 3.5490257588834552e-06,
      "loss": 0.07534807920455933,
      "step": 5859
    },
    {
      "epoch": 0.7922264469792987,
      "grad_norm": 0.48165163397789,
      "learning_rate": 3.5446059427132615e-06,
      "loss": 0.06319570541381836,
      "step": 5860
    },
    {
      "epoch": 0.7923616392057456,
      "grad_norm": 0.32107535004615784,
      "learning_rate": 3.5401885116171977e-06,
      "loss": 0.06496143341064453,
      "step": 5861
    },
    {
      "epoch": 0.7924968314321926,
      "grad_norm": 0.34414857625961304,
      "learning_rate": 3.5357734665149983e-06,
      "loss": 0.060967087745666504,
      "step": 5862
    },
    {
      "epoch": 0.7926320236586396,
      "grad_norm": 0.23059622943401337,
      "learning_rate": 3.5313608083258975e-06,
      "loss": 0.04185652732849121,
      "step": 5863
    },
    {
      "epoch": 0.7927672158850866,
      "grad_norm": 0.38646480441093445,
      "learning_rate": 3.526950537968629e-06,
      "loss": 0.07922554016113281,
      "step": 5864
    },
    {
      "epoch": 0.7929024081115336,
      "grad_norm": 0.28850066661834717,
      "learning_rate": 3.5225426563614466e-06,
      "loss": 0.05431365966796875,
      "step": 5865
    },
    {
      "epoch": 0.7930376003379805,
      "grad_norm": 0.12900696694850922,
      "learning_rate": 3.518137164422088e-06,
      "loss": 0.03278541564941406,
      "step": 5866
    },
    {
      "epoch": 0.7931727925644275,
      "grad_norm": 0.35914337635040283,
      "learning_rate": 3.513734063067799e-06,
      "loss": 0.0921776294708252,
      "step": 5867
    },
    {
      "epoch": 0.7933079847908745,
      "grad_norm": 0.14453966915607452,
      "learning_rate": 3.5093333532153316e-06,
      "loss": 0.03630661964416504,
      "step": 5868
    },
    {
      "epoch": 0.7934431770173215,
      "grad_norm": 0.2894953489303589,
      "learning_rate": 3.504935035780931e-06,
      "loss": 0.06284189224243164,
      "step": 5869
    },
    {
      "epoch": 0.7935783692437685,
      "grad_norm": 0.30163708329200745,
      "learning_rate": 3.500539111680364e-06,
      "loss": 0.07058584690093994,
      "step": 5870
    },
    {
      "epoch": 0.7937135614702154,
      "grad_norm": 0.524236798286438,
      "learning_rate": 3.4961455818288683e-06,
      "loss": 0.09031009674072266,
      "step": 5871
    },
    {
      "epoch": 0.7938487536966624,
      "grad_norm": 0.23061595857143402,
      "learning_rate": 3.491754447141212e-06,
      "loss": 0.04869508743286133,
      "step": 5872
    },
    {
      "epoch": 0.7939839459231094,
      "grad_norm": 0.3281051814556122,
      "learning_rate": 3.4873657085316504e-06,
      "loss": 0.06250953674316406,
      "step": 5873
    },
    {
      "epoch": 0.7941191381495564,
      "grad_norm": 0.33008918166160583,
      "learning_rate": 3.482979366913935e-06,
      "loss": 0.06384420394897461,
      "step": 5874
    },
    {
      "epoch": 0.7942543303760033,
      "grad_norm": 0.4086730182170868,
      "learning_rate": 3.4785954232013423e-06,
      "loss": 0.0685889720916748,
      "step": 5875
    },
    {
      "epoch": 0.7943895226024503,
      "grad_norm": 0.3937242925167084,
      "learning_rate": 3.4742138783066122e-06,
      "loss": 0.08018016815185547,
      "step": 5876
    },
    {
      "epoch": 0.7945247148288973,
      "grad_norm": 0.5728943943977356,
      "learning_rate": 3.4698347331420206e-06,
      "loss": 0.08187556266784668,
      "step": 5877
    },
    {
      "epoch": 0.7946599070553443,
      "grad_norm": 0.2489577978849411,
      "learning_rate": 3.4654579886193223e-06,
      "loss": 0.061530113220214844,
      "step": 5878
    },
    {
      "epoch": 0.7947950992817913,
      "grad_norm": 0.29962947964668274,
      "learning_rate": 3.461083645649782e-06,
      "loss": 0.061600685119628906,
      "step": 5879
    },
    {
      "epoch": 0.7949302915082382,
      "grad_norm": 0.4102258086204529,
      "learning_rate": 3.4567117051441594e-06,
      "loss": 0.05533194541931152,
      "step": 5880
    },
    {
      "epoch": 0.7950654837346852,
      "grad_norm": 0.16216807067394257,
      "learning_rate": 3.4523421680127115e-06,
      "loss": 0.03956878185272217,
      "step": 5881
    },
    {
      "epoch": 0.7952006759611322,
      "grad_norm": 0.4279293119907379,
      "learning_rate": 3.447975035165209e-06,
      "loss": 0.07097578048706055,
      "step": 5882
    },
    {
      "epoch": 0.7953358681875792,
      "grad_norm": 0.3215303122997284,
      "learning_rate": 3.4436103075109076e-06,
      "loss": 0.05742359161376953,
      "step": 5883
    },
    {
      "epoch": 0.7954710604140262,
      "grad_norm": 0.34310388565063477,
      "learning_rate": 3.4392479859585642e-06,
      "loss": 0.07660746574401855,
      "step": 5884
    },
    {
      "epoch": 0.7956062526404731,
      "grad_norm": 0.36039555072784424,
      "learning_rate": 3.4348880714164416e-06,
      "loss": 0.06647872924804688,
      "step": 5885
    },
    {
      "epoch": 0.7957414448669201,
      "grad_norm": 0.2750612795352936,
      "learning_rate": 3.430530564792289e-06,
      "loss": 0.056557655334472656,
      "step": 5886
    },
    {
      "epoch": 0.7958766370933671,
      "grad_norm": 0.5428061485290527,
      "learning_rate": 3.426175466993374e-06,
      "loss": 0.07390868663787842,
      "step": 5887
    },
    {
      "epoch": 0.7960118293198141,
      "grad_norm": 0.2898423969745636,
      "learning_rate": 3.4218227789264468e-06,
      "loss": 0.05636954307556152,
      "step": 5888
    },
    {
      "epoch": 0.796147021546261,
      "grad_norm": 0.21055658161640167,
      "learning_rate": 3.417472501497758e-06,
      "loss": 0.04013681411743164,
      "step": 5889
    },
    {
      "epoch": 0.796282213772708,
      "grad_norm": 0.6051457524299622,
      "learning_rate": 3.413124635613061e-06,
      "loss": 0.07826828956604004,
      "step": 5890
    },
    {
      "epoch": 0.796417405999155,
      "grad_norm": 0.2119661122560501,
      "learning_rate": 3.4087791821775986e-06,
      "loss": 0.05274820327758789,
      "step": 5891
    },
    {
      "epoch": 0.796552598225602,
      "grad_norm": 0.3171161711215973,
      "learning_rate": 3.4044361420961285e-06,
      "loss": 0.06368637084960938,
      "step": 5892
    },
    {
      "epoch": 0.796687790452049,
      "grad_norm": 0.1474030613899231,
      "learning_rate": 3.4000955162728866e-06,
      "loss": 0.041254520416259766,
      "step": 5893
    },
    {
      "epoch": 0.7968229826784959,
      "grad_norm": 0.30869102478027344,
      "learning_rate": 3.3957573056116164e-06,
      "loss": 0.06182575225830078,
      "step": 5894
    },
    {
      "epoch": 0.7969581749049429,
      "grad_norm": 0.26358070969581604,
      "learning_rate": 3.391421511015558e-06,
      "loss": 0.04130411148071289,
      "step": 5895
    },
    {
      "epoch": 0.7970933671313899,
      "grad_norm": 0.22798016667366028,
      "learning_rate": 3.38708813338744e-06,
      "loss": 0.041803598403930664,
      "step": 5896
    },
    {
      "epoch": 0.7972285593578369,
      "grad_norm": 0.49194031953811646,
      "learning_rate": 3.382757173629506e-06,
      "loss": 0.06548714637756348,
      "step": 5897
    },
    {
      "epoch": 0.7973637515842839,
      "grad_norm": 0.2816850543022156,
      "learning_rate": 3.378428632643478e-06,
      "loss": 0.07463455200195312,
      "step": 5898
    },
    {
      "epoch": 0.7974989438107308,
      "grad_norm": 0.17052793502807617,
      "learning_rate": 3.3741025113305825e-06,
      "loss": 0.04622983932495117,
      "step": 5899
    },
    {
      "epoch": 0.7976341360371778,
      "grad_norm": 0.23189640045166016,
      "learning_rate": 3.369778810591541e-06,
      "loss": 0.061908721923828125,
      "step": 5900
    },
    {
      "epoch": 0.7977693282636248,
      "grad_norm": 0.2454507201910019,
      "learning_rate": 3.3654575313265664e-06,
      "loss": 0.05745649337768555,
      "step": 5901
    },
    {
      "epoch": 0.7979045204900718,
      "grad_norm": 0.4106588363647461,
      "learning_rate": 3.361138674435386e-06,
      "loss": 0.05560147762298584,
      "step": 5902
    },
    {
      "epoch": 0.7980397127165187,
      "grad_norm": 0.29690682888031006,
      "learning_rate": 3.35682224081719e-06,
      "loss": 0.06985664367675781,
      "step": 5903
    },
    {
      "epoch": 0.7981749049429658,
      "grad_norm": 0.2685743272304535,
      "learning_rate": 3.352508231370699e-06,
      "loss": 0.059015750885009766,
      "step": 5904
    },
    {
      "epoch": 0.7983100971694128,
      "grad_norm": 0.20073962211608887,
      "learning_rate": 3.3481966469941044e-06,
      "loss": 0.03816652297973633,
      "step": 5905
    },
    {
      "epoch": 0.7984452893958598,
      "grad_norm": 0.3108757436275482,
      "learning_rate": 3.3438874885850984e-06,
      "loss": 0.07769346237182617,
      "step": 5906
    },
    {
      "epoch": 0.7985804816223068,
      "grad_norm": 0.2990017533302307,
      "learning_rate": 3.3395807570408847e-06,
      "loss": 0.06526756286621094,
      "step": 5907
    },
    {
      "epoch": 0.7987156738487537,
      "grad_norm": 0.609994649887085,
      "learning_rate": 3.33527645325813e-06,
      "loss": 0.07463288307189941,
      "step": 5908
    },
    {
      "epoch": 0.7988508660752007,
      "grad_norm": 0.456243097782135,
      "learning_rate": 3.3309745781330247e-06,
      "loss": 0.06634783744812012,
      "step": 5909
    },
    {
      "epoch": 0.7989860583016477,
      "grad_norm": 0.24894843995571136,
      "learning_rate": 3.32667513256124e-06,
      "loss": 0.056507110595703125,
      "step": 5910
    },
    {
      "epoch": 0.7991212505280947,
      "grad_norm": 0.20762842893600464,
      "learning_rate": 3.3223781174379375e-06,
      "loss": 0.04933905601501465,
      "step": 5911
    },
    {
      "epoch": 0.7992564427545417,
      "grad_norm": 0.5355972647666931,
      "learning_rate": 3.3180835336577917e-06,
      "loss": 0.06595611572265625,
      "step": 5912
    },
    {
      "epoch": 0.7993916349809886,
      "grad_norm": 0.23957689106464386,
      "learning_rate": 3.313791382114943e-06,
      "loss": 0.0684194564819336,
      "step": 5913
    },
    {
      "epoch": 0.7995268272074356,
      "grad_norm": 0.3925321400165558,
      "learning_rate": 3.3095016637030505e-06,
      "loss": 0.09416913986206055,
      "step": 5914
    },
    {
      "epoch": 0.7996620194338826,
      "grad_norm": 0.23857423663139343,
      "learning_rate": 3.3052143793152524e-06,
      "loss": 0.053450584411621094,
      "step": 5915
    },
    {
      "epoch": 0.7997972116603296,
      "grad_norm": 0.2391202598810196,
      "learning_rate": 3.3009295298441855e-06,
      "loss": 0.049214839935302734,
      "step": 5916
    },
    {
      "epoch": 0.7999324038867766,
      "grad_norm": 0.32320865988731384,
      "learning_rate": 3.2966471161819767e-06,
      "loss": 0.05562543869018555,
      "step": 5917
    },
    {
      "epoch": 0.8000675961132235,
      "grad_norm": 0.29675161838531494,
      "learning_rate": 3.292367139220246e-06,
      "loss": 0.057188987731933594,
      "step": 5918
    },
    {
      "epoch": 0.8002027883396705,
      "grad_norm": 0.18438327312469482,
      "learning_rate": 3.288089599850112e-06,
      "loss": 0.03248238563537598,
      "step": 5919
    },
    {
      "epoch": 0.8003379805661175,
      "grad_norm": 0.36449599266052246,
      "learning_rate": 3.2838144989621795e-06,
      "loss": 0.06055879592895508,
      "step": 5920
    },
    {
      "epoch": 0.8004731727925645,
      "grad_norm": 0.42434123158454895,
      "learning_rate": 3.2795418374465458e-06,
      "loss": 0.06690025329589844,
      "step": 5921
    },
    {
      "epoch": 0.8006083650190114,
      "grad_norm": 0.308083176612854,
      "learning_rate": 3.275271616192803e-06,
      "loss": 0.05982637405395508,
      "step": 5922
    },
    {
      "epoch": 0.8007435572454584,
      "grad_norm": 0.30605998635292053,
      "learning_rate": 3.2710038360900303e-06,
      "loss": 0.06424713134765625,
      "step": 5923
    },
    {
      "epoch": 0.8008787494719054,
      "grad_norm": 0.23421098291873932,
      "learning_rate": 3.266738498026808e-06,
      "loss": 0.059783935546875,
      "step": 5924
    },
    {
      "epoch": 0.8010139416983524,
      "grad_norm": 0.23105604946613312,
      "learning_rate": 3.2624756028912005e-06,
      "loss": 0.05173146724700928,
      "step": 5925
    },
    {
      "epoch": 0.8011491339247994,
      "grad_norm": 0.4146019518375397,
      "learning_rate": 3.2582151515707655e-06,
      "loss": 0.076080322265625,
      "step": 5926
    },
    {
      "epoch": 0.8012843261512463,
      "grad_norm": 0.3450862467288971,
      "learning_rate": 3.253957144952551e-06,
      "loss": 0.07610392570495605,
      "step": 5927
    },
    {
      "epoch": 0.8014195183776933,
      "grad_norm": 0.3583201766014099,
      "learning_rate": 3.249701583923091e-06,
      "loss": 0.07472729682922363,
      "step": 5928
    },
    {
      "epoch": 0.8015547106041403,
      "grad_norm": 0.395272821187973,
      "learning_rate": 3.2454484693684257e-06,
      "loss": 0.07549095153808594,
      "step": 5929
    },
    {
      "epoch": 0.8016899028305873,
      "grad_norm": 0.2994285821914673,
      "learning_rate": 3.2411978021740727e-06,
      "loss": 0.04777020215988159,
      "step": 5930
    },
    {
      "epoch": 0.8018250950570343,
      "grad_norm": 0.25583797693252563,
      "learning_rate": 3.2369495832250434e-06,
      "loss": 0.0537114143371582,
      "step": 5931
    },
    {
      "epoch": 0.8019602872834812,
      "grad_norm": 0.28064626455307007,
      "learning_rate": 3.2327038134058378e-06,
      "loss": 0.05535268783569336,
      "step": 5932
    },
    {
      "epoch": 0.8020954795099282,
      "grad_norm": 0.28295469284057617,
      "learning_rate": 3.228460493600446e-06,
      "loss": 0.07952594757080078,
      "step": 5933
    },
    {
      "epoch": 0.8022306717363752,
      "grad_norm": 0.3977964520454407,
      "learning_rate": 3.2242196246923554e-06,
      "loss": 0.0649835467338562,
      "step": 5934
    },
    {
      "epoch": 0.8023658639628222,
      "grad_norm": 0.37444525957107544,
      "learning_rate": 3.2199812075645375e-06,
      "loss": 0.08574485778808594,
      "step": 5935
    },
    {
      "epoch": 0.8025010561892691,
      "grad_norm": 0.19850070774555206,
      "learning_rate": 3.2157452430994487e-06,
      "loss": 0.04955482482910156,
      "step": 5936
    },
    {
      "epoch": 0.8026362484157161,
      "grad_norm": 0.62302565574646,
      "learning_rate": 3.2115117321790427e-06,
      "loss": 0.10436153411865234,
      "step": 5937
    },
    {
      "epoch": 0.8027714406421631,
      "grad_norm": 0.4690047800540924,
      "learning_rate": 3.207280675684754e-06,
      "loss": 0.06291484832763672,
      "step": 5938
    },
    {
      "epoch": 0.8029066328686101,
      "grad_norm": 0.3041515648365021,
      "learning_rate": 3.203052074497523e-06,
      "loss": 0.05617856979370117,
      "step": 5939
    },
    {
      "epoch": 0.8030418250950571,
      "grad_norm": 0.25223445892333984,
      "learning_rate": 3.198825929497752e-06,
      "loss": 0.049494028091430664,
      "step": 5940
    },
    {
      "epoch": 0.803177017321504,
      "grad_norm": 0.2859956920146942,
      "learning_rate": 3.194602241565357e-06,
      "loss": 0.053922057151794434,
      "step": 5941
    },
    {
      "epoch": 0.803312209547951,
      "grad_norm": 0.5730636715888977,
      "learning_rate": 3.1903810115797282e-06,
      "loss": 0.053530335426330566,
      "step": 5942
    },
    {
      "epoch": 0.803447401774398,
      "grad_norm": 0.24570654332637787,
      "learning_rate": 3.1861622404197475e-06,
      "loss": 0.04247868061065674,
      "step": 5943
    },
    {
      "epoch": 0.803582594000845,
      "grad_norm": 0.22582951188087463,
      "learning_rate": 3.181945928963794e-06,
      "loss": 0.05179309844970703,
      "step": 5944
    },
    {
      "epoch": 0.803717786227292,
      "grad_norm": 0.24250128865242004,
      "learning_rate": 3.1777320780897124e-06,
      "loss": 0.05829143524169922,
      "step": 5945
    },
    {
      "epoch": 0.8038529784537389,
      "grad_norm": 0.2799808979034424,
      "learning_rate": 3.1735206886748602e-06,
      "loss": 0.05966770648956299,
      "step": 5946
    },
    {
      "epoch": 0.8039881706801859,
      "grad_norm": 0.3474852442741394,
      "learning_rate": 3.1693117615960665e-06,
      "loss": 0.06831169128417969,
      "step": 5947
    },
    {
      "epoch": 0.8041233629066329,
      "grad_norm": 0.4207124710083008,
      "learning_rate": 3.1651052977296537e-06,
      "loss": 0.09029388427734375,
      "step": 5948
    },
    {
      "epoch": 0.8042585551330799,
      "grad_norm": 0.2723088264465332,
      "learning_rate": 3.1609012979514273e-06,
      "loss": 0.0634164810180664,
      "step": 5949
    },
    {
      "epoch": 0.8043937473595268,
      "grad_norm": 0.41889747977256775,
      "learning_rate": 3.156699763136683e-06,
      "loss": 0.08170890808105469,
      "step": 5950
    },
    {
      "epoch": 0.8045289395859738,
      "grad_norm": 0.29862862825393677,
      "learning_rate": 3.152500694160207e-06,
      "loss": 0.06790554523468018,
      "step": 5951
    },
    {
      "epoch": 0.8046641318124208,
      "grad_norm": 0.5785182118415833,
      "learning_rate": 3.148304091896265e-06,
      "loss": 0.07030558586120605,
      "step": 5952
    },
    {
      "epoch": 0.8047993240388678,
      "grad_norm": 0.3827217221260071,
      "learning_rate": 3.144109957218612e-06,
      "loss": 0.06324422359466553,
      "step": 5953
    },
    {
      "epoch": 0.8049345162653148,
      "grad_norm": 0.4175521433353424,
      "learning_rate": 3.1399182910004893e-06,
      "loss": 0.08026123046875,
      "step": 5954
    },
    {
      "epoch": 0.8050697084917617,
      "grad_norm": 0.28439274430274963,
      "learning_rate": 3.1357290941146215e-06,
      "loss": 0.04622602462768555,
      "step": 5955
    },
    {
      "epoch": 0.8052049007182087,
      "grad_norm": 0.2630952000617981,
      "learning_rate": 3.1315423674332265e-06,
      "loss": 0.06127119064331055,
      "step": 5956
    },
    {
      "epoch": 0.8053400929446557,
      "grad_norm": 0.2002151608467102,
      "learning_rate": 3.127358111828002e-06,
      "loss": 0.05383944511413574,
      "step": 5957
    },
    {
      "epoch": 0.8054752851711027,
      "grad_norm": 0.1441686749458313,
      "learning_rate": 3.123176328170131e-06,
      "loss": 0.04285383224487305,
      "step": 5958
    },
    {
      "epoch": 0.8056104773975497,
      "grad_norm": 0.31765300035476685,
      "learning_rate": 3.1189970173302816e-06,
      "loss": 0.07411348819732666,
      "step": 5959
    },
    {
      "epoch": 0.8057456696239966,
      "grad_norm": 0.2846148908138275,
      "learning_rate": 3.1148201801786085e-06,
      "loss": 0.07054901123046875,
      "step": 5960
    },
    {
      "epoch": 0.8058808618504436,
      "grad_norm": 0.24253971874713898,
      "learning_rate": 3.1106458175847572e-06,
      "loss": 0.045412540435791016,
      "step": 5961
    },
    {
      "epoch": 0.8060160540768906,
      "grad_norm": 0.11731071770191193,
      "learning_rate": 3.106473930417848e-06,
      "loss": 0.03174871206283569,
      "step": 5962
    },
    {
      "epoch": 0.8061512463033376,
      "grad_norm": 0.29594147205352783,
      "learning_rate": 3.1023045195464903e-06,
      "loss": 0.051989197731018066,
      "step": 5963
    },
    {
      "epoch": 0.8062864385297845,
      "grad_norm": 0.22837471961975098,
      "learning_rate": 3.098137585838779e-06,
      "loss": 0.05999565124511719,
      "step": 5964
    },
    {
      "epoch": 0.8064216307562315,
      "grad_norm": 0.13394711911678314,
      "learning_rate": 3.093973130162286e-06,
      "loss": 0.031126737594604492,
      "step": 5965
    },
    {
      "epoch": 0.8065568229826785,
      "grad_norm": 0.28221338987350464,
      "learning_rate": 3.089811153384083e-06,
      "loss": 0.04995131492614746,
      "step": 5966
    },
    {
      "epoch": 0.8066920152091255,
      "grad_norm": 0.2014472335577011,
      "learning_rate": 3.08565165637071e-06,
      "loss": 0.05839824676513672,
      "step": 5967
    },
    {
      "epoch": 0.8068272074355725,
      "grad_norm": 0.18108515441417694,
      "learning_rate": 3.081494639988196e-06,
      "loss": 0.045146942138671875,
      "step": 5968
    },
    {
      "epoch": 0.8069623996620194,
      "grad_norm": 0.22258733212947845,
      "learning_rate": 3.077340105102057e-06,
      "loss": 0.0531005859375,
      "step": 5969
    },
    {
      "epoch": 0.8070975918884664,
      "grad_norm": 0.2183443009853363,
      "learning_rate": 3.0731880525772817e-06,
      "loss": 0.05909299850463867,
      "step": 5970
    },
    {
      "epoch": 0.8072327841149134,
      "grad_norm": 0.23940695822238922,
      "learning_rate": 3.069038483278364e-06,
      "loss": 0.056195735931396484,
      "step": 5971
    },
    {
      "epoch": 0.8073679763413604,
      "grad_norm": 0.3401727080345154,
      "learning_rate": 3.0648913980692505e-06,
      "loss": 0.0833749771118164,
      "step": 5972
    },
    {
      "epoch": 0.8075031685678074,
      "grad_norm": 0.30367034673690796,
      "learning_rate": 3.0607467978133985e-06,
      "loss": 0.05263376235961914,
      "step": 5973
    },
    {
      "epoch": 0.8076383607942543,
      "grad_norm": 0.33931174874305725,
      "learning_rate": 3.0566046833737294e-06,
      "loss": 0.07328853011131287,
      "step": 5974
    },
    {
      "epoch": 0.8077735530207013,
      "grad_norm": 0.3551097810268402,
      "learning_rate": 3.0524650556126517e-06,
      "loss": 0.07681083679199219,
      "step": 5975
    },
    {
      "epoch": 0.8079087452471483,
      "grad_norm": 0.13805121183395386,
      "learning_rate": 3.048327915392069e-06,
      "loss": 0.03244829177856445,
      "step": 5976
    },
    {
      "epoch": 0.8080439374735953,
      "grad_norm": 0.33022773265838623,
      "learning_rate": 3.044193263573341e-06,
      "loss": 0.0790395736694336,
      "step": 5977
    },
    {
      "epoch": 0.8081791297000422,
      "grad_norm": 0.15861350297927856,
      "learning_rate": 3.0400611010173355e-06,
      "loss": 0.04596138000488281,
      "step": 5978
    },
    {
      "epoch": 0.8083143219264892,
      "grad_norm": 0.32104912400245667,
      "learning_rate": 3.0359314285843863e-06,
      "loss": 0.07611799240112305,
      "step": 5979
    },
    {
      "epoch": 0.8084495141529362,
      "grad_norm": 0.3161945641040802,
      "learning_rate": 3.0318042471343104e-06,
      "loss": 0.07168233394622803,
      "step": 5980
    },
    {
      "epoch": 0.8085847063793832,
      "grad_norm": 0.3164866268634796,
      "learning_rate": 3.027679557526422e-06,
      "loss": 0.07014846801757812,
      "step": 5981
    },
    {
      "epoch": 0.8087198986058302,
      "grad_norm": 0.4257417917251587,
      "learning_rate": 3.0235573606194844e-06,
      "loss": 0.0629739761352539,
      "step": 5982
    },
    {
      "epoch": 0.8088550908322771,
      "grad_norm": 0.5392917990684509,
      "learning_rate": 3.0194376572717743e-06,
      "loss": 0.09183883666992188,
      "step": 5983
    },
    {
      "epoch": 0.8089902830587241,
      "grad_norm": 0.20555761456489563,
      "learning_rate": 3.0153204483410318e-06,
      "loss": 0.05644631385803223,
      "step": 5984
    },
    {
      "epoch": 0.8091254752851711,
      "grad_norm": 0.4321295917034149,
      "learning_rate": 3.0112057346844834e-06,
      "loss": 0.08102297782897949,
      "step": 5985
    },
    {
      "epoch": 0.8092606675116181,
      "grad_norm": 0.2864816188812256,
      "learning_rate": 3.007093517158832e-06,
      "loss": 0.07088708877563477,
      "step": 5986
    },
    {
      "epoch": 0.809395859738065,
      "grad_norm": 0.27273401618003845,
      "learning_rate": 3.002983796620261e-06,
      "loss": 0.053546905517578125,
      "step": 5987
    },
    {
      "epoch": 0.809531051964512,
      "grad_norm": 0.5449060201644897,
      "learning_rate": 2.9988765739244427e-06,
      "loss": 0.0780942440032959,
      "step": 5988
    },
    {
      "epoch": 0.809666244190959,
      "grad_norm": 0.3616132438182831,
      "learning_rate": 2.9947718499265197e-06,
      "loss": 0.08637809753417969,
      "step": 5989
    },
    {
      "epoch": 0.809801436417406,
      "grad_norm": 0.17864102125167847,
      "learning_rate": 2.9906696254811184e-06,
      "loss": 0.046005964279174805,
      "step": 5990
    },
    {
      "epoch": 0.809936628643853,
      "grad_norm": 0.384751558303833,
      "learning_rate": 2.9865699014423404e-06,
      "loss": 0.049784183502197266,
      "step": 5991
    },
    {
      "epoch": 0.8100718208702999,
      "grad_norm": 0.22934868931770325,
      "learning_rate": 2.9824726786637698e-06,
      "loss": 0.050343096256256104,
      "step": 5992
    },
    {
      "epoch": 0.8102070130967469,
      "grad_norm": 0.48752734065055847,
      "learning_rate": 2.978377957998477e-06,
      "loss": 0.09496164321899414,
      "step": 5993
    },
    {
      "epoch": 0.8103422053231939,
      "grad_norm": 0.27358347177505493,
      "learning_rate": 2.974285740299001e-06,
      "loss": 0.0584406852722168,
      "step": 5994
    },
    {
      "epoch": 0.8104773975496409,
      "grad_norm": 0.3991025388240814,
      "learning_rate": 2.9701960264173612e-06,
      "loss": 0.0663614273071289,
      "step": 5995
    },
    {
      "epoch": 0.8106125897760879,
      "grad_norm": 0.33150970935821533,
      "learning_rate": 2.96610881720506e-06,
      "loss": 0.05559682846069336,
      "step": 5996
    },
    {
      "epoch": 0.8107477820025348,
      "grad_norm": 0.31655383110046387,
      "learning_rate": 2.9620241135130715e-06,
      "loss": 0.0678856372833252,
      "step": 5997
    },
    {
      "epoch": 0.8108829742289818,
      "grad_norm": 0.6779840588569641,
      "learning_rate": 2.9579419161918607e-06,
      "loss": 0.08355236053466797,
      "step": 5998
    },
    {
      "epoch": 0.8110181664554288,
      "grad_norm": 0.19206367433071136,
      "learning_rate": 2.9538622260913595e-06,
      "loss": 0.04270792007446289,
      "step": 5999
    },
    {
      "epoch": 0.8111533586818758,
      "grad_norm": 0.3747929036617279,
      "learning_rate": 2.9497850440609814e-06,
      "loss": 0.06484103202819824,
      "step": 6000
    },
    {
      "epoch": 0.8112885509083227,
      "grad_norm": 0.2159186452627182,
      "learning_rate": 2.945710370949616e-06,
      "loss": 0.05137217044830322,
      "step": 6001
    },
    {
      "epoch": 0.8114237431347697,
      "grad_norm": 0.19466176629066467,
      "learning_rate": 2.941638207605629e-06,
      "loss": 0.03656339645385742,
      "step": 6002
    },
    {
      "epoch": 0.8115589353612167,
      "grad_norm": 0.36241474747657776,
      "learning_rate": 2.937568554876873e-06,
      "loss": 0.08582782745361328,
      "step": 6003
    },
    {
      "epoch": 0.8116941275876637,
      "grad_norm": 0.2982935607433319,
      "learning_rate": 2.9335014136106704e-06,
      "loss": 0.06539297103881836,
      "step": 6004
    },
    {
      "epoch": 0.8118293198141107,
      "grad_norm": 0.33213871717453003,
      "learning_rate": 2.929436784653818e-06,
      "loss": 0.059813737869262695,
      "step": 6005
    },
    {
      "epoch": 0.8119645120405576,
      "grad_norm": 0.3062089681625366,
      "learning_rate": 2.925374668852597e-06,
      "loss": 0.06751632690429688,
      "step": 6006
    },
    {
      "epoch": 0.8120997042670046,
      "grad_norm": 0.19817759096622467,
      "learning_rate": 2.921315067052754e-06,
      "loss": 0.05138111114501953,
      "step": 6007
    },
    {
      "epoch": 0.8122348964934516,
      "grad_norm": 0.2660796046257019,
      "learning_rate": 2.917257980099535e-06,
      "loss": 0.057718753814697266,
      "step": 6008
    },
    {
      "epoch": 0.8123700887198986,
      "grad_norm": 0.17224065959453583,
      "learning_rate": 2.913203408837629e-06,
      "loss": 0.05161905288696289,
      "step": 6009
    },
    {
      "epoch": 0.8125052809463456,
      "grad_norm": 0.1696305274963379,
      "learning_rate": 2.909151354111232e-06,
      "loss": 0.04230022430419922,
      "step": 6010
    },
    {
      "epoch": 0.8126404731727925,
      "grad_norm": 0.3935920298099518,
      "learning_rate": 2.905101816763998e-06,
      "loss": 0.07367944717407227,
      "step": 6011
    },
    {
      "epoch": 0.8127756653992395,
      "grad_norm": 0.1789814680814743,
      "learning_rate": 2.9010547976390617e-06,
      "loss": 0.04308140277862549,
      "step": 6012
    },
    {
      "epoch": 0.8129108576256865,
      "grad_norm": 0.2439608871936798,
      "learning_rate": 2.897010297579042e-06,
      "loss": 0.06258678436279297,
      "step": 6013
    },
    {
      "epoch": 0.8130460498521335,
      "grad_norm": 0.25089651346206665,
      "learning_rate": 2.8929683174260133e-06,
      "loss": 0.05801820755004883,
      "step": 6014
    },
    {
      "epoch": 0.8131812420785804,
      "grad_norm": 0.2685256600379944,
      "learning_rate": 2.8889288580215467e-06,
      "loss": 0.035607337951660156,
      "step": 6015
    },
    {
      "epoch": 0.8133164343050274,
      "grad_norm": 0.28391289710998535,
      "learning_rate": 2.8848919202066752e-06,
      "loss": 0.07107210159301758,
      "step": 6016
    },
    {
      "epoch": 0.8134516265314744,
      "grad_norm": 0.7301496863365173,
      "learning_rate": 2.8808575048219123e-06,
      "loss": 0.12547564506530762,
      "step": 6017
    },
    {
      "epoch": 0.8135868187579214,
      "grad_norm": 0.2270822525024414,
      "learning_rate": 2.8768256127072436e-06,
      "loss": 0.04033780097961426,
      "step": 6018
    },
    {
      "epoch": 0.8137220109843684,
      "grad_norm": 0.2530202269554138,
      "learning_rate": 2.872796244702128e-06,
      "loss": 0.052034854888916016,
      "step": 6019
    },
    {
      "epoch": 0.8138572032108153,
      "grad_norm": 0.45868924260139465,
      "learning_rate": 2.8687694016455075e-06,
      "loss": 0.09524834156036377,
      "step": 6020
    },
    {
      "epoch": 0.8139923954372623,
      "grad_norm": 0.2891275882720947,
      "learning_rate": 2.86474508437579e-06,
      "loss": 0.05807161331176758,
      "step": 6021
    },
    {
      "epoch": 0.8141275876637093,
      "grad_norm": 0.19792482256889343,
      "learning_rate": 2.8607232937308587e-06,
      "loss": 0.04277324676513672,
      "step": 6022
    },
    {
      "epoch": 0.8142627798901563,
      "grad_norm": 0.5736647844314575,
      "learning_rate": 2.856704030548072e-06,
      "loss": 0.0836835503578186,
      "step": 6023
    },
    {
      "epoch": 0.8143979721166033,
      "grad_norm": 0.1461063027381897,
      "learning_rate": 2.8526872956642568e-06,
      "loss": 0.03426766395568848,
      "step": 6024
    },
    {
      "epoch": 0.8145331643430502,
      "grad_norm": 0.2663962244987488,
      "learning_rate": 2.84867308991573e-06,
      "loss": 0.04257082939147949,
      "step": 6025
    },
    {
      "epoch": 0.8146683565694972,
      "grad_norm": 0.34636518359184265,
      "learning_rate": 2.8446614141382638e-06,
      "loss": 0.08236312866210938,
      "step": 6026
    },
    {
      "epoch": 0.8148035487959442,
      "grad_norm": 0.3127804100513458,
      "learning_rate": 2.8406522691671104e-06,
      "loss": 0.068572998046875,
      "step": 6027
    },
    {
      "epoch": 0.8149387410223912,
      "grad_norm": 0.2658383548259735,
      "learning_rate": 2.8366456558369975e-06,
      "loss": 0.061776161193847656,
      "step": 6028
    },
    {
      "epoch": 0.8150739332488381,
      "grad_norm": 0.1475418657064438,
      "learning_rate": 2.8326415749821186e-06,
      "loss": 0.044802188873291016,
      "step": 6029
    },
    {
      "epoch": 0.8152091254752851,
      "grad_norm": 0.32792845368385315,
      "learning_rate": 2.828640027436151e-06,
      "loss": 0.06208324432373047,
      "step": 6030
    },
    {
      "epoch": 0.8153443177017321,
      "grad_norm": 0.30077195167541504,
      "learning_rate": 2.824641014032235e-06,
      "loss": 0.056476593017578125,
      "step": 6031
    },
    {
      "epoch": 0.8154795099281791,
      "grad_norm": 0.21988511085510254,
      "learning_rate": 2.820644535602987e-06,
      "loss": 0.05349874496459961,
      "step": 6032
    },
    {
      "epoch": 0.8156147021546261,
      "grad_norm": 0.28167712688446045,
      "learning_rate": 2.8166505929804953e-06,
      "loss": 0.07236099243164062,
      "step": 6033
    },
    {
      "epoch": 0.815749894381073,
      "grad_norm": 0.2284173220396042,
      "learning_rate": 2.8126591869963163e-06,
      "loss": 0.03714609146118164,
      "step": 6034
    },
    {
      "epoch": 0.81588508660752,
      "grad_norm": 0.3360145688056946,
      "learning_rate": 2.8086703184814887e-06,
      "loss": 0.07312464714050293,
      "step": 6035
    },
    {
      "epoch": 0.816020278833967,
      "grad_norm": 0.2188270539045334,
      "learning_rate": 2.8046839882665134e-06,
      "loss": 0.050281524658203125,
      "step": 6036
    },
    {
      "epoch": 0.816155471060414,
      "grad_norm": 0.4058283567428589,
      "learning_rate": 2.800700197181364e-06,
      "loss": 0.08154439926147461,
      "step": 6037
    },
    {
      "epoch": 0.816290663286861,
      "grad_norm": 0.40668433904647827,
      "learning_rate": 2.7967189460554876e-06,
      "loss": 0.06850337982177734,
      "step": 6038
    },
    {
      "epoch": 0.8164258555133079,
      "grad_norm": 0.46174219250679016,
      "learning_rate": 2.792740235717801e-06,
      "loss": 0.08393049240112305,
      "step": 6039
    },
    {
      "epoch": 0.816561047739755,
      "grad_norm": 0.2694415748119354,
      "learning_rate": 2.7887640669967e-06,
      "loss": 0.06118583679199219,
      "step": 6040
    },
    {
      "epoch": 0.816696239966202,
      "grad_norm": 0.28720611333847046,
      "learning_rate": 2.7847904407200327e-06,
      "loss": 0.07139301300048828,
      "step": 6041
    },
    {
      "epoch": 0.816831432192649,
      "grad_norm": 0.3160383105278015,
      "learning_rate": 2.7808193577151363e-06,
      "loss": 0.061397552490234375,
      "step": 6042
    },
    {
      "epoch": 0.816966624419096,
      "grad_norm": 0.2736820876598358,
      "learning_rate": 2.776850818808812e-06,
      "loss": 0.06472396850585938,
      "step": 6043
    },
    {
      "epoch": 0.8171018166455429,
      "grad_norm": 0.2239132821559906,
      "learning_rate": 2.772884824827325e-06,
      "loss": 0.0593411922454834,
      "step": 6044
    },
    {
      "epoch": 0.8172370088719899,
      "grad_norm": 0.21632635593414307,
      "learning_rate": 2.768921376596429e-06,
      "loss": 0.04656839370727539,
      "step": 6045
    },
    {
      "epoch": 0.8173722010984369,
      "grad_norm": 0.3758517801761627,
      "learning_rate": 2.7649604749413176e-06,
      "loss": 0.07829809188842773,
      "step": 6046
    },
    {
      "epoch": 0.8175073933248839,
      "grad_norm": 0.3210703432559967,
      "learning_rate": 2.7610021206866837e-06,
      "loss": 0.06727910041809082,
      "step": 6047
    },
    {
      "epoch": 0.8176425855513308,
      "grad_norm": 0.3311633765697479,
      "learning_rate": 2.757046314656676e-06,
      "loss": 0.08884382247924805,
      "step": 6048
    },
    {
      "epoch": 0.8177777777777778,
      "grad_norm": 0.5232706665992737,
      "learning_rate": 2.753093057674909e-06,
      "loss": 0.08345842361450195,
      "step": 6049
    },
    {
      "epoch": 0.8179129700042248,
      "grad_norm": 0.30730652809143066,
      "learning_rate": 2.749142350564483e-06,
      "loss": 0.061348915100097656,
      "step": 6050
    },
    {
      "epoch": 0.8180481622306718,
      "grad_norm": 0.3373664617538452,
      "learning_rate": 2.7451941941479414e-06,
      "loss": 0.05693459510803223,
      "step": 6051
    },
    {
      "epoch": 0.8181833544571188,
      "grad_norm": 0.5161705017089844,
      "learning_rate": 2.741248589247323e-06,
      "loss": 0.07698631286621094,
      "step": 6052
    },
    {
      "epoch": 0.8183185466835657,
      "grad_norm": 0.2258588671684265,
      "learning_rate": 2.73730553668412e-06,
      "loss": 0.04509568214416504,
      "step": 6053
    },
    {
      "epoch": 0.8184537389100127,
      "grad_norm": 0.3730873763561249,
      "learning_rate": 2.7333650372792978e-06,
      "loss": 0.0525355339050293,
      "step": 6054
    },
    {
      "epoch": 0.8185889311364597,
      "grad_norm": 0.17569658160209656,
      "learning_rate": 2.7294270918532876e-06,
      "loss": 0.042198896408081055,
      "step": 6055
    },
    {
      "epoch": 0.8187241233629067,
      "grad_norm": 0.20935864746570587,
      "learning_rate": 2.7254917012259882e-06,
      "loss": 0.03873276710510254,
      "step": 6056
    },
    {
      "epoch": 0.8188593155893537,
      "grad_norm": 0.28307756781578064,
      "learning_rate": 2.721558866216776e-06,
      "loss": 0.050560712814331055,
      "step": 6057
    },
    {
      "epoch": 0.8189945078158006,
      "grad_norm": 0.5794156789779663,
      "learning_rate": 2.7176285876444846e-06,
      "loss": 0.11113327741622925,
      "step": 6058
    },
    {
      "epoch": 0.8191297000422476,
      "grad_norm": 0.313987672328949,
      "learning_rate": 2.713700866327417e-06,
      "loss": 0.0690147876739502,
      "step": 6059
    },
    {
      "epoch": 0.8192648922686946,
      "grad_norm": 0.21553422510623932,
      "learning_rate": 2.7097757030833497e-06,
      "loss": 0.043886661529541016,
      "step": 6060
    },
    {
      "epoch": 0.8194000844951416,
      "grad_norm": 0.3039560317993164,
      "learning_rate": 2.705853098729517e-06,
      "loss": 0.06674432754516602,
      "step": 6061
    },
    {
      "epoch": 0.8195352767215885,
      "grad_norm": 0.21206343173980713,
      "learning_rate": 2.7019330540826325e-06,
      "loss": 0.038247108459472656,
      "step": 6062
    },
    {
      "epoch": 0.8196704689480355,
      "grad_norm": 0.3314515948295593,
      "learning_rate": 2.6980155699588666e-06,
      "loss": 0.07146835327148438,
      "step": 6063
    },
    {
      "epoch": 0.8198056611744825,
      "grad_norm": 0.3042283058166504,
      "learning_rate": 2.6941006471738633e-06,
      "loss": 0.07489490509033203,
      "step": 6064
    },
    {
      "epoch": 0.8199408534009295,
      "grad_norm": 0.22345556318759918,
      "learning_rate": 2.690188286542726e-06,
      "loss": 0.04886627197265625,
      "step": 6065
    },
    {
      "epoch": 0.8200760456273765,
      "grad_norm": 0.22376970946788788,
      "learning_rate": 2.686278488880029e-06,
      "loss": 0.04528689384460449,
      "step": 6066
    },
    {
      "epoch": 0.8202112378538234,
      "grad_norm": 0.33989548683166504,
      "learning_rate": 2.6823712549998187e-06,
      "loss": 0.0739513635635376,
      "step": 6067
    },
    {
      "epoch": 0.8203464300802704,
      "grad_norm": 0.5353378057479858,
      "learning_rate": 2.678466585715599e-06,
      "loss": 0.06407666206359863,
      "step": 6068
    },
    {
      "epoch": 0.8204816223067174,
      "grad_norm": 0.4994027316570282,
      "learning_rate": 2.6745644818403426e-06,
      "loss": 0.09122371673583984,
      "step": 6069
    },
    {
      "epoch": 0.8206168145331644,
      "grad_norm": 0.5354039072990417,
      "learning_rate": 2.6706649441864883e-06,
      "loss": 0.08888816833496094,
      "step": 6070
    },
    {
      "epoch": 0.8207520067596114,
      "grad_norm": 0.23649027943611145,
      "learning_rate": 2.666767973565937e-06,
      "loss": 0.05955362319946289,
      "step": 6071
    },
    {
      "epoch": 0.8208871989860583,
      "grad_norm": 0.4597260355949402,
      "learning_rate": 2.6628735707900653e-06,
      "loss": 0.07512426376342773,
      "step": 6072
    },
    {
      "epoch": 0.8210223912125053,
      "grad_norm": 0.39809486269950867,
      "learning_rate": 2.658981736669707e-06,
      "loss": 0.10382843017578125,
      "step": 6073
    },
    {
      "epoch": 0.8211575834389523,
      "grad_norm": 0.3359104096889496,
      "learning_rate": 2.655092472015161e-06,
      "loss": 0.0579218864440918,
      "step": 6074
    },
    {
      "epoch": 0.8212927756653993,
      "grad_norm": 0.4085558354854584,
      "learning_rate": 2.6512057776361935e-06,
      "loss": 0.0943596363067627,
      "step": 6075
    },
    {
      "epoch": 0.8214279678918462,
      "grad_norm": 0.4857843220233917,
      "learning_rate": 2.64732165434203e-06,
      "loss": 0.0741429328918457,
      "step": 6076
    },
    {
      "epoch": 0.8215631601182932,
      "grad_norm": 0.2855377495288849,
      "learning_rate": 2.6434401029413792e-06,
      "loss": 0.06378507614135742,
      "step": 6077
    },
    {
      "epoch": 0.8216983523447402,
      "grad_norm": 0.5213587880134583,
      "learning_rate": 2.639561124242385e-06,
      "loss": 0.08265399932861328,
      "step": 6078
    },
    {
      "epoch": 0.8218335445711872,
      "grad_norm": 0.28476154804229736,
      "learning_rate": 2.635684719052682e-06,
      "loss": 0.04998493194580078,
      "step": 6079
    },
    {
      "epoch": 0.8219687367976342,
      "grad_norm": 0.2014065831899643,
      "learning_rate": 2.631810888179355e-06,
      "loss": 0.04931068420410156,
      "step": 6080
    },
    {
      "epoch": 0.8221039290240811,
      "grad_norm": 0.2453290820121765,
      "learning_rate": 2.627939632428952e-06,
      "loss": 0.04505336284637451,
      "step": 6081
    },
    {
      "epoch": 0.8222391212505281,
      "grad_norm": 0.360598623752594,
      "learning_rate": 2.624070952607502e-06,
      "loss": 0.0605320930480957,
      "step": 6082
    },
    {
      "epoch": 0.8223743134769751,
      "grad_norm": 0.26906445622444153,
      "learning_rate": 2.620204849520468e-06,
      "loss": 0.07411766052246094,
      "step": 6083
    },
    {
      "epoch": 0.8225095057034221,
      "grad_norm": 0.29481276869773865,
      "learning_rate": 2.616341323972806e-06,
      "loss": 0.0637279748916626,
      "step": 6084
    },
    {
      "epoch": 0.822644697929869,
      "grad_norm": 0.25700411200523376,
      "learning_rate": 2.612480376768917e-06,
      "loss": 0.06525135040283203,
      "step": 6085
    },
    {
      "epoch": 0.822779890156316,
      "grad_norm": 0.14568747580051422,
      "learning_rate": 2.608622008712672e-06,
      "loss": 0.02857351303100586,
      "step": 6086
    },
    {
      "epoch": 0.822915082382763,
      "grad_norm": 0.4644869565963745,
      "learning_rate": 2.6047662206074034e-06,
      "loss": 0.08104497194290161,
      "step": 6087
    },
    {
      "epoch": 0.82305027460921,
      "grad_norm": 0.28576111793518066,
      "learning_rate": 2.600913013255904e-06,
      "loss": 0.05170249938964844,
      "step": 6088
    },
    {
      "epoch": 0.823185466835657,
      "grad_norm": 0.23014803230762482,
      "learning_rate": 2.59706238746044e-06,
      "loss": 0.051761627197265625,
      "step": 6089
    },
    {
      "epoch": 0.8233206590621039,
      "grad_norm": 0.4032961130142212,
      "learning_rate": 2.593214344022725e-06,
      "loss": 0.09865593910217285,
      "step": 6090
    },
    {
      "epoch": 0.8234558512885509,
      "grad_norm": 0.5228615999221802,
      "learning_rate": 2.5893688837439474e-06,
      "loss": 0.09393692016601562,
      "step": 6091
    },
    {
      "epoch": 0.8235910435149979,
      "grad_norm": 0.3720521628856659,
      "learning_rate": 2.5855260074247473e-06,
      "loss": 0.045456886291503906,
      "step": 6092
    },
    {
      "epoch": 0.8237262357414449,
      "grad_norm": 0.27444711327552795,
      "learning_rate": 2.581685715865232e-06,
      "loss": 0.07110810279846191,
      "step": 6093
    },
    {
      "epoch": 0.8238614279678919,
      "grad_norm": 0.38014987111091614,
      "learning_rate": 2.5778480098649766e-06,
      "loss": 0.10168582201004028,
      "step": 6094
    },
    {
      "epoch": 0.8239966201943388,
      "grad_norm": 0.2758583128452301,
      "learning_rate": 2.5740128902230087e-06,
      "loss": 0.07007026672363281,
      "step": 6095
    },
    {
      "epoch": 0.8241318124207858,
      "grad_norm": 0.22813992202281952,
      "learning_rate": 2.5701803577378214e-06,
      "loss": 0.05235719680786133,
      "step": 6096
    },
    {
      "epoch": 0.8242670046472328,
      "grad_norm": 0.42385274171829224,
      "learning_rate": 2.566350413207366e-06,
      "loss": 0.08513689041137695,
      "step": 6097
    },
    {
      "epoch": 0.8244021968736798,
      "grad_norm": 0.3720324635505676,
      "learning_rate": 2.5625230574290554e-06,
      "loss": 0.07318711280822754,
      "step": 6098
    },
    {
      "epoch": 0.8245373891001267,
      "grad_norm": 0.33294564485549927,
      "learning_rate": 2.558698291199773e-06,
      "loss": 0.056952476501464844,
      "step": 6099
    },
    {
      "epoch": 0.8246725813265737,
      "grad_norm": 0.3401033580303192,
      "learning_rate": 2.5548761153158524e-06,
      "loss": 0.06594085693359375,
      "step": 6100
    },
    {
      "epoch": 0.8248077735530207,
      "grad_norm": 0.255386620759964,
      "learning_rate": 2.55105653057309e-06,
      "loss": 0.05640685558319092,
      "step": 6101
    },
    {
      "epoch": 0.8249429657794677,
      "grad_norm": 0.41073131561279297,
      "learning_rate": 2.547239537766743e-06,
      "loss": 0.07777738571166992,
      "step": 6102
    },
    {
      "epoch": 0.8250781580059147,
      "grad_norm": 0.2563095986843109,
      "learning_rate": 2.543425137691526e-06,
      "loss": 0.053259849548339844,
      "step": 6103
    },
    {
      "epoch": 0.8252133502323616,
      "grad_norm": 0.40238744020462036,
      "learning_rate": 2.5396133311416264e-06,
      "loss": 0.07820796966552734,
      "step": 6104
    },
    {
      "epoch": 0.8253485424588086,
      "grad_norm": 0.4039735794067383,
      "learning_rate": 2.5358041189106784e-06,
      "loss": 0.08084726333618164,
      "step": 6105
    },
    {
      "epoch": 0.8254837346852556,
      "grad_norm": 0.28965792059898376,
      "learning_rate": 2.531997501791779e-06,
      "loss": 0.04353666305541992,
      "step": 6106
    },
    {
      "epoch": 0.8256189269117026,
      "grad_norm": 0.4189249575138092,
      "learning_rate": 2.528193480577489e-06,
      "loss": 0.04222249984741211,
      "step": 6107
    },
    {
      "epoch": 0.8257541191381496,
      "grad_norm": 0.22650189697742462,
      "learning_rate": 2.5243920560598186e-06,
      "loss": 0.04802274703979492,
      "step": 6108
    },
    {
      "epoch": 0.8258893113645965,
      "grad_norm": 0.3336423933506012,
      "learning_rate": 2.5205932290302598e-06,
      "loss": 0.07125353813171387,
      "step": 6109
    },
    {
      "epoch": 0.8260245035910435,
      "grad_norm": 0.18595555424690247,
      "learning_rate": 2.516797000279729e-06,
      "loss": 0.03815329074859619,
      "step": 6110
    },
    {
      "epoch": 0.8261596958174905,
      "grad_norm": 0.3717397451400757,
      "learning_rate": 2.513003370598637e-06,
      "loss": 0.08181643486022949,
      "step": 6111
    },
    {
      "epoch": 0.8262948880439375,
      "grad_norm": 0.3404862880706787,
      "learning_rate": 2.509212340776832e-06,
      "loss": 0.0374302864074707,
      "step": 6112
    },
    {
      "epoch": 0.8264300802703844,
      "grad_norm": 0.28415605425834656,
      "learning_rate": 2.505423911603622e-06,
      "loss": 0.05466151237487793,
      "step": 6113
    },
    {
      "epoch": 0.8265652724968314,
      "grad_norm": 0.19801393151283264,
      "learning_rate": 2.501638083867789e-06,
      "loss": 0.047365665435791016,
      "step": 6114
    },
    {
      "epoch": 0.8267004647232784,
      "grad_norm": 0.24348372220993042,
      "learning_rate": 2.497854858357552e-06,
      "loss": 0.06348276138305664,
      "step": 6115
    },
    {
      "epoch": 0.8268356569497254,
      "grad_norm": 0.2263336181640625,
      "learning_rate": 2.494074235860604e-06,
      "loss": 0.055605411529541016,
      "step": 6116
    },
    {
      "epoch": 0.8269708491761724,
      "grad_norm": 0.192909374833107,
      "learning_rate": 2.4902962171640913e-06,
      "loss": 0.04351067543029785,
      "step": 6117
    },
    {
      "epoch": 0.8271060414026193,
      "grad_norm": 0.3105994462966919,
      "learning_rate": 2.4865208030546167e-06,
      "loss": 0.06407403945922852,
      "step": 6118
    },
    {
      "epoch": 0.8272412336290663,
      "grad_norm": 0.28907307982444763,
      "learning_rate": 2.482747994318239e-06,
      "loss": 0.04296064376831055,
      "step": 6119
    },
    {
      "epoch": 0.8273764258555133,
      "grad_norm": 0.17278222739696503,
      "learning_rate": 2.478977791740477e-06,
      "loss": 0.03933072090148926,
      "step": 6120
    },
    {
      "epoch": 0.8275116180819603,
      "grad_norm": 0.4144207239151001,
      "learning_rate": 2.475210196106313e-06,
      "loss": 0.08549785614013672,
      "step": 6121
    },
    {
      "epoch": 0.8276468103084073,
      "grad_norm": 0.26255449652671814,
      "learning_rate": 2.4714452082001753e-06,
      "loss": 0.05323374271392822,
      "step": 6122
    },
    {
      "epoch": 0.8277820025348542,
      "grad_norm": 0.36885082721710205,
      "learning_rate": 2.467682828805956e-06,
      "loss": 0.07731294631958008,
      "step": 6123
    },
    {
      "epoch": 0.8279171947613012,
      "grad_norm": 0.31515181064605713,
      "learning_rate": 2.4639230587070017e-06,
      "loss": 0.07038754224777222,
      "step": 6124
    },
    {
      "epoch": 0.8280523869877482,
      "grad_norm": 0.3263688385486603,
      "learning_rate": 2.460165898686114e-06,
      "loss": 0.08073592185974121,
      "step": 6125
    },
    {
      "epoch": 0.8281875792141952,
      "grad_norm": 0.3848995566368103,
      "learning_rate": 2.4564113495255597e-06,
      "loss": 0.0795893669128418,
      "step": 6126
    },
    {
      "epoch": 0.8283227714406421,
      "grad_norm": 0.28529009222984314,
      "learning_rate": 2.4526594120070545e-06,
      "loss": 0.06337499618530273,
      "step": 6127
    },
    {
      "epoch": 0.8284579636670891,
      "grad_norm": 0.322240948677063,
      "learning_rate": 2.4489100869117686e-06,
      "loss": 0.06416690349578857,
      "step": 6128
    },
    {
      "epoch": 0.8285931558935361,
      "grad_norm": 0.4755936563014984,
      "learning_rate": 2.4451633750203344e-06,
      "loss": 0.09140324592590332,
      "step": 6129
    },
    {
      "epoch": 0.8287283481199831,
      "grad_norm": 0.29732969403266907,
      "learning_rate": 2.441419277112831e-06,
      "loss": 0.0648496150970459,
      "step": 6130
    },
    {
      "epoch": 0.8288635403464301,
      "grad_norm": 0.23001553118228912,
      "learning_rate": 2.4376777939688107e-06,
      "loss": 0.06379508972167969,
      "step": 6131
    },
    {
      "epoch": 0.828998732572877,
      "grad_norm": 0.31154295802116394,
      "learning_rate": 2.4339389263672625e-06,
      "loss": 0.054573774337768555,
      "step": 6132
    },
    {
      "epoch": 0.829133924799324,
      "grad_norm": 0.19406574964523315,
      "learning_rate": 2.4302026750866406e-06,
      "loss": 0.03343939781188965,
      "step": 6133
    },
    {
      "epoch": 0.829269117025771,
      "grad_norm": 0.2403125911951065,
      "learning_rate": 2.4264690409048517e-06,
      "loss": 0.04829525947570801,
      "step": 6134
    },
    {
      "epoch": 0.829404309252218,
      "grad_norm": 0.34234941005706787,
      "learning_rate": 2.4227380245992555e-06,
      "loss": 0.06604862213134766,
      "step": 6135
    },
    {
      "epoch": 0.829539501478665,
      "grad_norm": 0.22851765155792236,
      "learning_rate": 2.4190096269466767e-06,
      "loss": 0.0566706657409668,
      "step": 6136
    },
    {
      "epoch": 0.8296746937051119,
      "grad_norm": 0.3730808198451996,
      "learning_rate": 2.415283848723383e-06,
      "loss": 0.06548690795898438,
      "step": 6137
    },
    {
      "epoch": 0.8298098859315589,
      "grad_norm": 0.28897351026535034,
      "learning_rate": 2.411560690705101e-06,
      "loss": 0.07234847545623779,
      "step": 6138
    },
    {
      "epoch": 0.8299450781580059,
      "grad_norm": 0.3524039685726166,
      "learning_rate": 2.4078401536670146e-06,
      "loss": 0.07749509811401367,
      "step": 6139
    },
    {
      "epoch": 0.8300802703844529,
      "grad_norm": 0.2997666597366333,
      "learning_rate": 2.4041222383837538e-06,
      "loss": 0.06832456588745117,
      "step": 6140
    },
    {
      "epoch": 0.8302154626108998,
      "grad_norm": 0.3855161666870117,
      "learning_rate": 2.400406945629418e-06,
      "loss": 0.06428265571594238,
      "step": 6141
    },
    {
      "epoch": 0.8303506548373468,
      "grad_norm": 0.159975066781044,
      "learning_rate": 2.3966942761775396e-06,
      "loss": 0.03423929214477539,
      "step": 6142
    },
    {
      "epoch": 0.8304858470637938,
      "grad_norm": 0.1802024245262146,
      "learning_rate": 2.3929842308011263e-06,
      "loss": 0.03900432586669922,
      "step": 6143
    },
    {
      "epoch": 0.8306210392902408,
      "grad_norm": 0.3452477753162384,
      "learning_rate": 2.3892768102726236e-06,
      "loss": 0.07439923286437988,
      "step": 6144
    },
    {
      "epoch": 0.8307562315166878,
      "grad_norm": 0.4625161588191986,
      "learning_rate": 2.3855720153639344e-06,
      "loss": 0.07840752601623535,
      "step": 6145
    },
    {
      "epoch": 0.8308914237431347,
      "grad_norm": 0.35174861550331116,
      "learning_rate": 2.381869846846428e-06,
      "loss": 0.06035876274108887,
      "step": 6146
    },
    {
      "epoch": 0.8310266159695817,
      "grad_norm": 0.36810293793678284,
      "learning_rate": 2.3781703054908993e-06,
      "loss": 0.05990767478942871,
      "step": 6147
    },
    {
      "epoch": 0.8311618081960287,
      "grad_norm": 0.46127569675445557,
      "learning_rate": 2.374473392067624e-06,
      "loss": 0.0831599235534668,
      "step": 6148
    },
    {
      "epoch": 0.8312970004224757,
      "grad_norm": 0.2868509590625763,
      "learning_rate": 2.370779107346317e-06,
      "loss": 0.047698140144348145,
      "step": 6149
    },
    {
      "epoch": 0.8314321926489227,
      "grad_norm": 0.20161646604537964,
      "learning_rate": 2.3670874520961437e-06,
      "loss": 0.04446291923522949,
      "step": 6150
    },
    {
      "epoch": 0.8315673848753696,
      "grad_norm": 0.19508378207683563,
      "learning_rate": 2.3633984270857367e-06,
      "loss": 0.043001532554626465,
      "step": 6151
    },
    {
      "epoch": 0.8317025771018166,
      "grad_norm": 0.4475431740283966,
      "learning_rate": 2.359712033083156e-06,
      "loss": 0.061325669288635254,
      "step": 6152
    },
    {
      "epoch": 0.8318377693282636,
      "grad_norm": 0.21206475794315338,
      "learning_rate": 2.35602827085594e-06,
      "loss": 0.050330162048339844,
      "step": 6153
    },
    {
      "epoch": 0.8319729615547106,
      "grad_norm": 0.5338788628578186,
      "learning_rate": 2.3523471411710644e-06,
      "loss": 0.08822345733642578,
      "step": 6154
    },
    {
      "epoch": 0.8321081537811575,
      "grad_norm": 0.3041779696941376,
      "learning_rate": 2.3486686447949585e-06,
      "loss": 0.062389373779296875,
      "step": 6155
    },
    {
      "epoch": 0.8322433460076045,
      "grad_norm": 0.5393385291099548,
      "learning_rate": 2.3449927824935075e-06,
      "loss": 0.0924372673034668,
      "step": 6156
    },
    {
      "epoch": 0.8323785382340515,
      "grad_norm": 0.32297420501708984,
      "learning_rate": 2.3413195550320393e-06,
      "loss": 0.0675058364868164,
      "step": 6157
    },
    {
      "epoch": 0.8325137304604985,
      "grad_norm": 0.25539419054985046,
      "learning_rate": 2.3376489631753474e-06,
      "loss": 0.06180715560913086,
      "step": 6158
    },
    {
      "epoch": 0.8326489226869455,
      "grad_norm": 0.2430296391248703,
      "learning_rate": 2.3339810076876665e-06,
      "loss": 0.0410078763961792,
      "step": 6159
    },
    {
      "epoch": 0.8327841149133924,
      "grad_norm": 0.30485233664512634,
      "learning_rate": 2.3303156893326815e-06,
      "loss": 0.060146331787109375,
      "step": 6160
    },
    {
      "epoch": 0.8329193071398394,
      "grad_norm": 0.3537285625934601,
      "learning_rate": 2.326653008873535e-06,
      "loss": 0.09344768524169922,
      "step": 6161
    },
    {
      "epoch": 0.8330544993662864,
      "grad_norm": 0.19936206936836243,
      "learning_rate": 2.3229929670728085e-06,
      "loss": 0.05114603042602539,
      "step": 6162
    },
    {
      "epoch": 0.8331896915927334,
      "grad_norm": 0.5031552314758301,
      "learning_rate": 2.319335564692554e-06,
      "loss": 0.07624435424804688,
      "step": 6163
    },
    {
      "epoch": 0.8333248838191804,
      "grad_norm": 0.202814981341362,
      "learning_rate": 2.315680802494256e-06,
      "loss": 0.045780181884765625,
      "step": 6164
    },
    {
      "epoch": 0.8334600760456273,
      "grad_norm": 0.25232431292533875,
      "learning_rate": 2.312028681238856e-06,
      "loss": 0.07368946075439453,
      "step": 6165
    },
    {
      "epoch": 0.8335952682720743,
      "grad_norm": 0.4809962213039398,
      "learning_rate": 2.3083792016867434e-06,
      "loss": 0.06068849563598633,
      "step": 6166
    },
    {
      "epoch": 0.8337304604985213,
      "grad_norm": 0.32066455483436584,
      "learning_rate": 2.304732364597759e-06,
      "loss": 0.036364734172821045,
      "step": 6167
    },
    {
      "epoch": 0.8338656527249683,
      "grad_norm": 0.328524649143219,
      "learning_rate": 2.3010881707311994e-06,
      "loss": 0.07408523559570312,
      "step": 6168
    },
    {
      "epoch": 0.8340008449514152,
      "grad_norm": 0.40732115507125854,
      "learning_rate": 2.2974466208458017e-06,
      "loss": 0.048784077167510986,
      "step": 6169
    },
    {
      "epoch": 0.8341360371778622,
      "grad_norm": 0.34277865290641785,
      "learning_rate": 2.293807715699755e-06,
      "loss": 0.08569717407226562,
      "step": 6170
    },
    {
      "epoch": 0.8342712294043092,
      "grad_norm": 0.2897603511810303,
      "learning_rate": 2.2901714560507e-06,
      "loss": 0.08003807067871094,
      "step": 6171
    },
    {
      "epoch": 0.8344064216307562,
      "grad_norm": 0.3742716312408447,
      "learning_rate": 2.286537842655722e-06,
      "loss": 0.06242990493774414,
      "step": 6172
    },
    {
      "epoch": 0.8345416138572032,
      "grad_norm": 0.44942155480384827,
      "learning_rate": 2.2829068762713633e-06,
      "loss": 0.05798053741455078,
      "step": 6173
    },
    {
      "epoch": 0.8346768060836501,
      "grad_norm": 0.45306596159935,
      "learning_rate": 2.279278557653611e-06,
      "loss": 0.0857686996459961,
      "step": 6174
    },
    {
      "epoch": 0.8348119983100971,
      "grad_norm": 0.4919930696487427,
      "learning_rate": 2.2756528875578965e-06,
      "loss": 0.08963966369628906,
      "step": 6175
    },
    {
      "epoch": 0.8349471905365442,
      "grad_norm": 0.18259121477603912,
      "learning_rate": 2.2720298667391067e-06,
      "loss": 0.03711986541748047,
      "step": 6176
    },
    {
      "epoch": 0.8350823827629912,
      "grad_norm": 0.45871835947036743,
      "learning_rate": 2.268409495951568e-06,
      "loss": 0.08042430877685547,
      "step": 6177
    },
    {
      "epoch": 0.8352175749894382,
      "grad_norm": 0.26225897669792175,
      "learning_rate": 2.2647917759490723e-06,
      "loss": 0.05420804023742676,
      "step": 6178
    },
    {
      "epoch": 0.8353527672158851,
      "grad_norm": 0.13552093505859375,
      "learning_rate": 2.261176707484834e-06,
      "loss": 0.03930556774139404,
      "step": 6179
    },
    {
      "epoch": 0.8354879594423321,
      "grad_norm": 0.232827827334404,
      "learning_rate": 2.2575642913115408e-06,
      "loss": 0.06036567687988281,
      "step": 6180
    },
    {
      "epoch": 0.8356231516687791,
      "grad_norm": 0.19267582893371582,
      "learning_rate": 2.253954528181313e-06,
      "loss": 0.05502462387084961,
      "step": 6181
    },
    {
      "epoch": 0.8357583438952261,
      "grad_norm": 0.46230658888816833,
      "learning_rate": 2.2503474188457206e-06,
      "loss": 0.09131145477294922,
      "step": 6182
    },
    {
      "epoch": 0.835893536121673,
      "grad_norm": 0.29804858565330505,
      "learning_rate": 2.2467429640557903e-06,
      "loss": 0.04694533348083496,
      "step": 6183
    },
    {
      "epoch": 0.83602872834812,
      "grad_norm": 0.42065635323524475,
      "learning_rate": 2.2431411645619776e-06,
      "loss": 0.056189775466918945,
      "step": 6184
    },
    {
      "epoch": 0.836163920574567,
      "grad_norm": 0.3421494662761688,
      "learning_rate": 2.239542021114205e-06,
      "loss": 0.06933164596557617,
      "step": 6185
    },
    {
      "epoch": 0.836299112801014,
      "grad_norm": 0.38711124658584595,
      "learning_rate": 2.2359455344618306e-06,
      "loss": 0.08873653411865234,
      "step": 6186
    },
    {
      "epoch": 0.836434305027461,
      "grad_norm": 0.3926772475242615,
      "learning_rate": 2.232351705353663e-06,
      "loss": 0.08452796936035156,
      "step": 6187
    },
    {
      "epoch": 0.836569497253908,
      "grad_norm": 0.35300230979919434,
      "learning_rate": 2.228760534537955e-06,
      "loss": 0.08152008056640625,
      "step": 6188
    },
    {
      "epoch": 0.8367046894803549,
      "grad_norm": 0.27519428730010986,
      "learning_rate": 2.2251720227624044e-06,
      "loss": 0.06858253479003906,
      "step": 6189
    },
    {
      "epoch": 0.8368398817068019,
      "grad_norm": 0.4944668114185333,
      "learning_rate": 2.2215861707741666e-06,
      "loss": 0.09633874893188477,
      "step": 6190
    },
    {
      "epoch": 0.8369750739332489,
      "grad_norm": 0.348843514919281,
      "learning_rate": 2.2180029793198313e-06,
      "loss": 0.059391021728515625,
      "step": 6191
    },
    {
      "epoch": 0.8371102661596959,
      "grad_norm": 0.3249976933002472,
      "learning_rate": 2.2144224491454363e-06,
      "loss": 0.06228768825531006,
      "step": 6192
    },
    {
      "epoch": 0.8372454583861428,
      "grad_norm": 0.4631642997264862,
      "learning_rate": 2.2108445809964695e-06,
      "loss": 0.06086874008178711,
      "step": 6193
    },
    {
      "epoch": 0.8373806506125898,
      "grad_norm": 0.32808300852775574,
      "learning_rate": 2.2072693756178567e-06,
      "loss": 0.07857704162597656,
      "step": 6194
    },
    {
      "epoch": 0.8375158428390368,
      "grad_norm": 0.28030431270599365,
      "learning_rate": 2.203696833753983e-06,
      "loss": 0.06250762939453125,
      "step": 6195
    },
    {
      "epoch": 0.8376510350654838,
      "grad_norm": 0.31675535440444946,
      "learning_rate": 2.200126956148668e-06,
      "loss": 0.05798828601837158,
      "step": 6196
    },
    {
      "epoch": 0.8377862272919308,
      "grad_norm": 0.33596113324165344,
      "learning_rate": 2.196559743545177e-06,
      "loss": 0.0630028247833252,
      "step": 6197
    },
    {
      "epoch": 0.8379214195183777,
      "grad_norm": 0.23111020028591156,
      "learning_rate": 2.1929951966862233e-06,
      "loss": 0.05372738838195801,
      "step": 6198
    },
    {
      "epoch": 0.8380566117448247,
      "grad_norm": 0.45794913172721863,
      "learning_rate": 2.1894333163139607e-06,
      "loss": 0.0893559455871582,
      "step": 6199
    },
    {
      "epoch": 0.8381918039712717,
      "grad_norm": 0.26923510432243347,
      "learning_rate": 2.1858741031700015e-06,
      "loss": 0.05803251266479492,
      "step": 6200
    },
    {
      "epoch": 0.8383269961977187,
      "grad_norm": 0.16260357201099396,
      "learning_rate": 2.1823175579953856e-06,
      "loss": 0.03156423568725586,
      "step": 6201
    },
    {
      "epoch": 0.8384621884241656,
      "grad_norm": 0.4199337959289551,
      "learning_rate": 2.1787636815306065e-06,
      "loss": 0.08827757835388184,
      "step": 6202
    },
    {
      "epoch": 0.8385973806506126,
      "grad_norm": 0.47525399923324585,
      "learning_rate": 2.1752124745156005e-06,
      "loss": 0.07206487655639648,
      "step": 6203
    },
    {
      "epoch": 0.8387325728770596,
      "grad_norm": 0.32013580203056335,
      "learning_rate": 2.171663937689744e-06,
      "loss": 0.06802785396575928,
      "step": 6204
    },
    {
      "epoch": 0.8388677651035066,
      "grad_norm": 0.23136280477046967,
      "learning_rate": 2.168118071791868e-06,
      "loss": 0.041516780853271484,
      "step": 6205
    },
    {
      "epoch": 0.8390029573299536,
      "grad_norm": 0.37992942333221436,
      "learning_rate": 2.164574877560237e-06,
      "loss": 0.08926022052764893,
      "step": 6206
    },
    {
      "epoch": 0.8391381495564005,
      "grad_norm": 0.40065091848373413,
      "learning_rate": 2.161034355732564e-06,
      "loss": 0.09440803527832031,
      "step": 6207
    },
    {
      "epoch": 0.8392733417828475,
      "grad_norm": 0.3198845088481903,
      "learning_rate": 2.1574965070460047e-06,
      "loss": 0.06119346618652344,
      "step": 6208
    },
    {
      "epoch": 0.8394085340092945,
      "grad_norm": 0.391506165266037,
      "learning_rate": 2.1539613322371527e-06,
      "loss": 0.05940985679626465,
      "step": 6209
    },
    {
      "epoch": 0.8395437262357415,
      "grad_norm": 0.368939071893692,
      "learning_rate": 2.1504288320420613e-06,
      "loss": 0.08240604400634766,
      "step": 6210
    },
    {
      "epoch": 0.8396789184621885,
      "grad_norm": 0.2728017270565033,
      "learning_rate": 2.1468990071962038e-06,
      "loss": 0.05499708652496338,
      "step": 6211
    },
    {
      "epoch": 0.8398141106886354,
      "grad_norm": 0.2578313648700714,
      "learning_rate": 2.143371858434515e-06,
      "loss": 0.05638861656188965,
      "step": 6212
    },
    {
      "epoch": 0.8399493029150824,
      "grad_norm": 0.379812091588974,
      "learning_rate": 2.139847386491367e-06,
      "loss": 0.0633387565612793,
      "step": 6213
    },
    {
      "epoch": 0.8400844951415294,
      "grad_norm": 0.3931023180484772,
      "learning_rate": 2.1363255921005685e-06,
      "loss": 0.08740377426147461,
      "step": 6214
    },
    {
      "epoch": 0.8402196873679764,
      "grad_norm": 0.2514142692089081,
      "learning_rate": 2.1328064759953853e-06,
      "loss": 0.0461575984954834,
      "step": 6215
    },
    {
      "epoch": 0.8403548795944233,
      "grad_norm": 0.16072365641593933,
      "learning_rate": 2.129290038908504e-06,
      "loss": 0.03383278846740723,
      "step": 6216
    },
    {
      "epoch": 0.8404900718208703,
      "grad_norm": 0.2062246948480606,
      "learning_rate": 2.1257762815720745e-06,
      "loss": 0.046831727027893066,
      "step": 6217
    },
    {
      "epoch": 0.8406252640473173,
      "grad_norm": 0.4210064709186554,
      "learning_rate": 2.122265204717678e-06,
      "loss": 0.07840967178344727,
      "step": 6218
    },
    {
      "epoch": 0.8407604562737643,
      "grad_norm": 0.36550840735435486,
      "learning_rate": 2.1187568090763328e-06,
      "loss": 0.06644487380981445,
      "step": 6219
    },
    {
      "epoch": 0.8408956485002113,
      "grad_norm": 0.20956704020500183,
      "learning_rate": 2.1152510953785196e-06,
      "loss": 0.0396723747253418,
      "step": 6220
    },
    {
      "epoch": 0.8410308407266582,
      "grad_norm": 0.22913040220737457,
      "learning_rate": 2.1117480643541304e-06,
      "loss": 0.05521726608276367,
      "step": 6221
    },
    {
      "epoch": 0.8411660329531052,
      "grad_norm": 0.34781816601753235,
      "learning_rate": 2.1082477167325275e-06,
      "loss": 0.05765199661254883,
      "step": 6222
    },
    {
      "epoch": 0.8413012251795522,
      "grad_norm": 0.32238680124282837,
      "learning_rate": 2.1047500532424968e-06,
      "loss": 0.06810712814331055,
      "step": 6223
    },
    {
      "epoch": 0.8414364174059992,
      "grad_norm": 0.2920459806919098,
      "learning_rate": 2.1012550746122705e-06,
      "loss": 0.07100939750671387,
      "step": 6224
    },
    {
      "epoch": 0.8415716096324461,
      "grad_norm": 0.35092565417289734,
      "learning_rate": 2.0977627815695217e-06,
      "loss": 0.07351398468017578,
      "step": 6225
    },
    {
      "epoch": 0.8417068018588931,
      "grad_norm": 0.2670595645904541,
      "learning_rate": 2.094273174841362e-06,
      "loss": 0.04958820343017578,
      "step": 6226
    },
    {
      "epoch": 0.8418419940853401,
      "grad_norm": 0.38839006423950195,
      "learning_rate": 2.0907862551543516e-06,
      "loss": 0.058518409729003906,
      "step": 6227
    },
    {
      "epoch": 0.8419771863117871,
      "grad_norm": 0.3022480010986328,
      "learning_rate": 2.087302023234485e-06,
      "loss": 0.06590914726257324,
      "step": 6228
    },
    {
      "epoch": 0.8421123785382341,
      "grad_norm": 0.1451275497674942,
      "learning_rate": 2.083820479807194e-06,
      "loss": 0.031537652015686035,
      "step": 6229
    },
    {
      "epoch": 0.842247570764681,
      "grad_norm": 0.3312985301017761,
      "learning_rate": 2.0803416255973585e-06,
      "loss": 0.08653116226196289,
      "step": 6230
    },
    {
      "epoch": 0.842382762991128,
      "grad_norm": 0.3806116282939911,
      "learning_rate": 2.0768654613292887e-06,
      "loss": 0.06750082969665527,
      "step": 6231
    },
    {
      "epoch": 0.842517955217575,
      "grad_norm": 0.6721171140670776,
      "learning_rate": 2.0733919877267477e-06,
      "loss": 0.08948570489883423,
      "step": 6232
    },
    {
      "epoch": 0.842653147444022,
      "grad_norm": 0.2881007194519043,
      "learning_rate": 2.0699212055129268e-06,
      "loss": 0.05344271659851074,
      "step": 6233
    },
    {
      "epoch": 0.842788339670469,
      "grad_norm": 0.3937076926231384,
      "learning_rate": 2.066453115410463e-06,
      "loss": 0.05979776382446289,
      "step": 6234
    },
    {
      "epoch": 0.8429235318969159,
      "grad_norm": 0.6550187468528748,
      "learning_rate": 2.062987718141431e-06,
      "loss": 0.09293651580810547,
      "step": 6235
    },
    {
      "epoch": 0.8430587241233629,
      "grad_norm": 0.3156546652317047,
      "learning_rate": 2.0595250144273423e-06,
      "loss": 0.054172515869140625,
      "step": 6236
    },
    {
      "epoch": 0.8431939163498099,
      "grad_norm": 0.2758477032184601,
      "learning_rate": 2.056065004989155e-06,
      "loss": 0.06659078598022461,
      "step": 6237
    },
    {
      "epoch": 0.8433291085762569,
      "grad_norm": 0.25198909640312195,
      "learning_rate": 2.0526076905472585e-06,
      "loss": 0.05923342704772949,
      "step": 6238
    },
    {
      "epoch": 0.8434643008027038,
      "grad_norm": 0.49366337060928345,
      "learning_rate": 2.0491530718214855e-06,
      "loss": 0.06829285621643066,
      "step": 6239
    },
    {
      "epoch": 0.8435994930291508,
      "grad_norm": 0.2753165364265442,
      "learning_rate": 2.0457011495311045e-06,
      "loss": 0.05621814727783203,
      "step": 6240
    },
    {
      "epoch": 0.8437346852555978,
      "grad_norm": 0.2757529616355896,
      "learning_rate": 2.0422519243948232e-06,
      "loss": 0.04640340805053711,
      "step": 6241
    },
    {
      "epoch": 0.8438698774820448,
      "grad_norm": 0.4832799434661865,
      "learning_rate": 2.0388053971307927e-06,
      "loss": 0.09984779357910156,
      "step": 6242
    },
    {
      "epoch": 0.8440050697084918,
      "grad_norm": 0.22580106556415558,
      "learning_rate": 2.0353615684565956e-06,
      "loss": 0.041429996490478516,
      "step": 6243
    },
    {
      "epoch": 0.8441402619349387,
      "grad_norm": 0.29000264406204224,
      "learning_rate": 2.0319204390892566e-06,
      "loss": 0.05859375,
      "step": 6244
    },
    {
      "epoch": 0.8442754541613857,
      "grad_norm": 0.15063822269439697,
      "learning_rate": 2.0284820097452374e-06,
      "loss": 0.03659510612487793,
      "step": 6245
    },
    {
      "epoch": 0.8444106463878327,
      "grad_norm": 0.40340688824653625,
      "learning_rate": 2.02504628114043e-06,
      "loss": 0.05350208282470703,
      "step": 6246
    },
    {
      "epoch": 0.8445458386142797,
      "grad_norm": 0.1933862268924713,
      "learning_rate": 2.0216132539901865e-06,
      "loss": 0.04899430274963379,
      "step": 6247
    },
    {
      "epoch": 0.8446810308407267,
      "grad_norm": 0.2647876739501953,
      "learning_rate": 2.0181829290092663e-06,
      "loss": 0.05927896499633789,
      "step": 6248
    },
    {
      "epoch": 0.8448162230671736,
      "grad_norm": 0.21264617145061493,
      "learning_rate": 2.014755306911891e-06,
      "loss": 0.046753883361816406,
      "step": 6249
    },
    {
      "epoch": 0.8449514152936206,
      "grad_norm": 0.3327547609806061,
      "learning_rate": 2.0113303884117057e-06,
      "loss": 0.04807734489440918,
      "step": 6250
    },
    {
      "epoch": 0.8450866075200676,
      "grad_norm": 0.2826550602912903,
      "learning_rate": 2.0079081742217957e-06,
      "loss": 0.044869422912597656,
      "step": 6251
    },
    {
      "epoch": 0.8452217997465146,
      "grad_norm": 0.2259238064289093,
      "learning_rate": 2.0044886650546915e-06,
      "loss": 0.048828125,
      "step": 6252
    },
    {
      "epoch": 0.8453569919729615,
      "grad_norm": 0.2367219179868698,
      "learning_rate": 2.0010718616223406e-06,
      "loss": 0.057765960693359375,
      "step": 6253
    },
    {
      "epoch": 0.8454921841994085,
      "grad_norm": 0.18984411656856537,
      "learning_rate": 1.9976577646361514e-06,
      "loss": 0.0418243408203125,
      "step": 6254
    },
    {
      "epoch": 0.8456273764258555,
      "grad_norm": 0.4283030927181244,
      "learning_rate": 1.994246374806953e-06,
      "loss": 0.0626521110534668,
      "step": 6255
    },
    {
      "epoch": 0.8457625686523025,
      "grad_norm": 0.4187628924846649,
      "learning_rate": 1.9908376928450128e-06,
      "loss": 0.07620739936828613,
      "step": 6256
    },
    {
      "epoch": 0.8458977608787495,
      "grad_norm": 0.1862669140100479,
      "learning_rate": 1.987431719460039e-06,
      "loss": 0.04925107955932617,
      "step": 6257
    },
    {
      "epoch": 0.8460329531051964,
      "grad_norm": 0.17124955356121063,
      "learning_rate": 1.9840284553611706e-06,
      "loss": 0.047887325286865234,
      "step": 6258
    },
    {
      "epoch": 0.8461681453316434,
      "grad_norm": 0.3132363259792328,
      "learning_rate": 1.980627901256989e-06,
      "loss": 0.06468343734741211,
      "step": 6259
    },
    {
      "epoch": 0.8463033375580904,
      "grad_norm": 0.31490230560302734,
      "learning_rate": 1.9772300578555062e-06,
      "loss": 0.05545806884765625,
      "step": 6260
    },
    {
      "epoch": 0.8464385297845374,
      "grad_norm": 0.3261524438858032,
      "learning_rate": 1.973834925864172e-06,
      "loss": 0.0727154016494751,
      "step": 6261
    },
    {
      "epoch": 0.8465737220109844,
      "grad_norm": 0.31746163964271545,
      "learning_rate": 1.97044250598987e-06,
      "loss": 0.0643925666809082,
      "step": 6262
    },
    {
      "epoch": 0.8467089142374313,
      "grad_norm": 0.5831186175346375,
      "learning_rate": 1.9670527989389177e-06,
      "loss": 0.09212028980255127,
      "step": 6263
    },
    {
      "epoch": 0.8468441064638783,
      "grad_norm": 0.6222190260887146,
      "learning_rate": 1.9636658054170747e-06,
      "loss": 0.07861185073852539,
      "step": 6264
    },
    {
      "epoch": 0.8469792986903253,
      "grad_norm": 0.26287853717803955,
      "learning_rate": 1.960281526129531e-06,
      "loss": 0.06790542602539062,
      "step": 6265
    },
    {
      "epoch": 0.8471144909167723,
      "grad_norm": 0.36416393518447876,
      "learning_rate": 1.9568999617809077e-06,
      "loss": 0.059990644454956055,
      "step": 6266
    },
    {
      "epoch": 0.8472496831432192,
      "grad_norm": 0.26137810945510864,
      "learning_rate": 1.9535211130752676e-06,
      "loss": 0.055304527282714844,
      "step": 6267
    },
    {
      "epoch": 0.8473848753696662,
      "grad_norm": 0.30726176500320435,
      "learning_rate": 1.950144980716101e-06,
      "loss": 0.06860160827636719,
      "step": 6268
    },
    {
      "epoch": 0.8475200675961132,
      "grad_norm": 0.34413042664527893,
      "learning_rate": 1.9467715654063444e-06,
      "loss": 0.04716300964355469,
      "step": 6269
    },
    {
      "epoch": 0.8476552598225602,
      "grad_norm": 0.3800031244754791,
      "learning_rate": 1.9434008678483532e-06,
      "loss": 0.06565046310424805,
      "step": 6270
    },
    {
      "epoch": 0.8477904520490072,
      "grad_norm": 0.30992594361305237,
      "learning_rate": 1.9400328887439295e-06,
      "loss": 0.0719594955444336,
      "step": 6271
    },
    {
      "epoch": 0.8479256442754541,
      "grad_norm": 0.2748137414455414,
      "learning_rate": 1.9366676287943038e-06,
      "loss": 0.05409574508666992,
      "step": 6272
    },
    {
      "epoch": 0.8480608365019011,
      "grad_norm": 0.2654644846916199,
      "learning_rate": 1.9333050887001337e-06,
      "loss": 0.05579793453216553,
      "step": 6273
    },
    {
      "epoch": 0.8481960287283481,
      "grad_norm": 0.2429201453924179,
      "learning_rate": 1.9299452691615293e-06,
      "loss": 0.0350804328918457,
      "step": 6274
    },
    {
      "epoch": 0.8483312209547951,
      "grad_norm": 0.22114381194114685,
      "learning_rate": 1.9265881708780182e-06,
      "loss": 0.057952880859375,
      "step": 6275
    },
    {
      "epoch": 0.848466413181242,
      "grad_norm": 0.3228522837162018,
      "learning_rate": 1.9232337945485657e-06,
      "loss": 0.06509852409362793,
      "step": 6276
    },
    {
      "epoch": 0.848601605407689,
      "grad_norm": 0.3310171663761139,
      "learning_rate": 1.91988214087157e-06,
      "loss": 0.07742607593536377,
      "step": 6277
    },
    {
      "epoch": 0.848736797634136,
      "grad_norm": 0.28843954205513,
      "learning_rate": 1.9165332105448613e-06,
      "loss": 0.08643460273742676,
      "step": 6278
    },
    {
      "epoch": 0.848871989860583,
      "grad_norm": 0.3420291244983673,
      "learning_rate": 1.913187004265715e-06,
      "loss": 0.07580137252807617,
      "step": 6279
    },
    {
      "epoch": 0.84900718208703,
      "grad_norm": 0.19976767897605896,
      "learning_rate": 1.909843522730814e-06,
      "loss": 0.05601692199707031,
      "step": 6280
    },
    {
      "epoch": 0.8491423743134769,
      "grad_norm": 0.2512938380241394,
      "learning_rate": 1.9065027666363017e-06,
      "loss": 0.048183441162109375,
      "step": 6281
    },
    {
      "epoch": 0.8492775665399239,
      "grad_norm": 0.5086610913276672,
      "learning_rate": 1.903164736677736e-06,
      "loss": 0.07693183422088623,
      "step": 6282
    },
    {
      "epoch": 0.8494127587663709,
      "grad_norm": 0.21472832560539246,
      "learning_rate": 1.8998294335501082e-06,
      "loss": 0.049195051193237305,
      "step": 6283
    },
    {
      "epoch": 0.8495479509928179,
      "grad_norm": 0.4109760522842407,
      "learning_rate": 1.8964968579478592e-06,
      "loss": 0.0492548942565918,
      "step": 6284
    },
    {
      "epoch": 0.8496831432192649,
      "grad_norm": 0.4223787784576416,
      "learning_rate": 1.893167010564834e-06,
      "loss": 0.04801177978515625,
      "step": 6285
    },
    {
      "epoch": 0.8498183354457118,
      "grad_norm": 0.40585052967071533,
      "learning_rate": 1.8898398920943349e-06,
      "loss": 0.08501172065734863,
      "step": 6286
    },
    {
      "epoch": 0.8499535276721588,
      "grad_norm": 0.3826312720775604,
      "learning_rate": 1.886515503229081e-06,
      "loss": 0.06133604049682617,
      "step": 6287
    },
    {
      "epoch": 0.8500887198986058,
      "grad_norm": 0.3597368597984314,
      "learning_rate": 1.8831938446612269e-06,
      "loss": 0.047585248947143555,
      "step": 6288
    },
    {
      "epoch": 0.8502239121250528,
      "grad_norm": 0.3577701151371002,
      "learning_rate": 1.8798749170823676e-06,
      "loss": 0.046039581298828125,
      "step": 6289
    },
    {
      "epoch": 0.8503591043514998,
      "grad_norm": 0.28235092759132385,
      "learning_rate": 1.8765587211835089e-06,
      "loss": 0.05596637725830078,
      "step": 6290
    },
    {
      "epoch": 0.8504942965779467,
      "grad_norm": 0.362808495759964,
      "learning_rate": 1.8732452576551102e-06,
      "loss": 0.0826263427734375,
      "step": 6291
    },
    {
      "epoch": 0.8506294888043937,
      "grad_norm": 0.40157967805862427,
      "learning_rate": 1.8699345271870493e-06,
      "loss": 0.06565523147583008,
      "step": 6292
    },
    {
      "epoch": 0.8507646810308407,
      "grad_norm": 0.24076251685619354,
      "learning_rate": 1.8666265304686387e-06,
      "loss": 0.0528717041015625,
      "step": 6293
    },
    {
      "epoch": 0.8508998732572877,
      "grad_norm": 0.3019689619541168,
      "learning_rate": 1.8633212681886203e-06,
      "loss": 0.049294471740722656,
      "step": 6294
    },
    {
      "epoch": 0.8510350654837346,
      "grad_norm": 0.49325188994407654,
      "learning_rate": 1.8600187410351621e-06,
      "loss": 0.07761192321777344,
      "step": 6295
    },
    {
      "epoch": 0.8511702577101816,
      "grad_norm": 0.1678040772676468,
      "learning_rate": 1.8567189496958776e-06,
      "loss": 0.03497183322906494,
      "step": 6296
    },
    {
      "epoch": 0.8513054499366286,
      "grad_norm": 0.22160252928733826,
      "learning_rate": 1.853421894857797e-06,
      "loss": 0.042299747467041016,
      "step": 6297
    },
    {
      "epoch": 0.8514406421630756,
      "grad_norm": 0.2647058665752411,
      "learning_rate": 1.8501275772073827e-06,
      "loss": 0.06014204025268555,
      "step": 6298
    },
    {
      "epoch": 0.8515758343895226,
      "grad_norm": 0.20429734885692596,
      "learning_rate": 1.8468359974305315e-06,
      "loss": 0.048209190368652344,
      "step": 6299
    },
    {
      "epoch": 0.8517110266159695,
      "grad_norm": 0.38039442896842957,
      "learning_rate": 1.8435471562125633e-06,
      "loss": 0.04333961009979248,
      "step": 6300
    },
    {
      "epoch": 0.8518462188424165,
      "grad_norm": 0.22751489281654358,
      "learning_rate": 1.8402610542382386e-06,
      "loss": 0.043682098388671875,
      "step": 6301
    },
    {
      "epoch": 0.8519814110688635,
      "grad_norm": 0.41948744654655457,
      "learning_rate": 1.836977692191742e-06,
      "loss": 0.0705876350402832,
      "step": 6302
    },
    {
      "epoch": 0.8521166032953105,
      "grad_norm": 0.3832312524318695,
      "learning_rate": 1.8336970707566781e-06,
      "loss": 0.07243871688842773,
      "step": 6303
    },
    {
      "epoch": 0.8522517955217574,
      "grad_norm": 0.3912931978702545,
      "learning_rate": 1.8304191906160973e-06,
      "loss": 0.06734943389892578,
      "step": 6304
    },
    {
      "epoch": 0.8523869877482044,
      "grad_norm": 0.3388195037841797,
      "learning_rate": 1.8271440524524668e-06,
      "loss": 0.06375408172607422,
      "step": 6305
    },
    {
      "epoch": 0.8525221799746514,
      "grad_norm": 0.9208351373672485,
      "learning_rate": 1.8238716569476949e-06,
      "loss": 0.08703303337097168,
      "step": 6306
    },
    {
      "epoch": 0.8526573722010984,
      "grad_norm": 0.5578576922416687,
      "learning_rate": 1.8206020047831078e-06,
      "loss": 0.06214499473571777,
      "step": 6307
    },
    {
      "epoch": 0.8527925644275454,
      "grad_norm": 0.20516066253185272,
      "learning_rate": 1.8173350966394648e-06,
      "loss": 0.04634857177734375,
      "step": 6308
    },
    {
      "epoch": 0.8529277566539923,
      "grad_norm": 0.3526214361190796,
      "learning_rate": 1.8140709331969513e-06,
      "loss": 0.06889486312866211,
      "step": 6309
    },
    {
      "epoch": 0.8530629488804393,
      "grad_norm": 0.5277717709541321,
      "learning_rate": 1.810809515135184e-06,
      "loss": 0.07633543014526367,
      "step": 6310
    },
    {
      "epoch": 0.8531981411068863,
      "grad_norm": 0.36538299918174744,
      "learning_rate": 1.8075508431332111e-06,
      "loss": 0.0766000747680664,
      "step": 6311
    },
    {
      "epoch": 0.8533333333333334,
      "grad_norm": 0.16320344805717468,
      "learning_rate": 1.8042949178695034e-06,
      "loss": 0.03675508499145508,
      "step": 6312
    },
    {
      "epoch": 0.8534685255597804,
      "grad_norm": 0.29728835821151733,
      "learning_rate": 1.8010417400219636e-06,
      "loss": 0.0659797191619873,
      "step": 6313
    },
    {
      "epoch": 0.8536037177862273,
      "grad_norm": 0.35085615515708923,
      "learning_rate": 1.7977913102679167e-06,
      "loss": 0.0602877140045166,
      "step": 6314
    },
    {
      "epoch": 0.8537389100126743,
      "grad_norm": 0.33504942059516907,
      "learning_rate": 1.7945436292841193e-06,
      "loss": 0.07447564601898193,
      "step": 6315
    },
    {
      "epoch": 0.8538741022391213,
      "grad_norm": 0.24959515035152435,
      "learning_rate": 1.791298697746766e-06,
      "loss": 0.052007198333740234,
      "step": 6316
    },
    {
      "epoch": 0.8540092944655683,
      "grad_norm": 0.36414414644241333,
      "learning_rate": 1.7880565163314545e-06,
      "loss": 0.06575441360473633,
      "step": 6317
    },
    {
      "epoch": 0.8541444866920153,
      "grad_norm": 0.2594071626663208,
      "learning_rate": 1.784817085713233e-06,
      "loss": 0.05369305610656738,
      "step": 6318
    },
    {
      "epoch": 0.8542796789184622,
      "grad_norm": 0.18506945669651031,
      "learning_rate": 1.7815804065665669e-06,
      "loss": 0.04692578315734863,
      "step": 6319
    },
    {
      "epoch": 0.8544148711449092,
      "grad_norm": 0.5307829976081848,
      "learning_rate": 1.778346479565346e-06,
      "loss": 0.07484650611877441,
      "step": 6320
    },
    {
      "epoch": 0.8545500633713562,
      "grad_norm": 0.4423481523990631,
      "learning_rate": 1.7751153053829011e-06,
      "loss": 0.0657958984375,
      "step": 6321
    },
    {
      "epoch": 0.8546852555978032,
      "grad_norm": 0.9577394127845764,
      "learning_rate": 1.7718868846919662e-06,
      "loss": 0.07073616981506348,
      "step": 6322
    },
    {
      "epoch": 0.8548204478242502,
      "grad_norm": 0.32180920243263245,
      "learning_rate": 1.7686612181647266e-06,
      "loss": 0.0552525520324707,
      "step": 6323
    },
    {
      "epoch": 0.8549556400506971,
      "grad_norm": 0.29027029871940613,
      "learning_rate": 1.7654383064727802e-06,
      "loss": 0.05681467056274414,
      "step": 6324
    },
    {
      "epoch": 0.8550908322771441,
      "grad_norm": 0.20261068642139435,
      "learning_rate": 1.762218150287152e-06,
      "loss": 0.04305607080459595,
      "step": 6325
    },
    {
      "epoch": 0.8552260245035911,
      "grad_norm": 0.4485626518726349,
      "learning_rate": 1.759000750278299e-06,
      "loss": 0.07387542724609375,
      "step": 6326
    },
    {
      "epoch": 0.8553612167300381,
      "grad_norm": 0.3260519504547119,
      "learning_rate": 1.7557861071160953e-06,
      "loss": 0.07190752029418945,
      "step": 6327
    },
    {
      "epoch": 0.855496408956485,
      "grad_norm": 0.11344873160123825,
      "learning_rate": 1.7525742214698538e-06,
      "loss": 0.028392374515533447,
      "step": 6328
    },
    {
      "epoch": 0.855631601182932,
      "grad_norm": 0.21213459968566895,
      "learning_rate": 1.7493650940083045e-06,
      "loss": 0.043506622314453125,
      "step": 6329
    },
    {
      "epoch": 0.855766793409379,
      "grad_norm": 0.17933131754398346,
      "learning_rate": 1.746158725399603e-06,
      "loss": 0.04773974418640137,
      "step": 6330
    },
    {
      "epoch": 0.855901985635826,
      "grad_norm": 0.24295440316200256,
      "learning_rate": 1.7429551163113322e-06,
      "loss": 0.06258130073547363,
      "step": 6331
    },
    {
      "epoch": 0.856037177862273,
      "grad_norm": 0.3535444438457489,
      "learning_rate": 1.7397542674105e-06,
      "loss": 0.07405781745910645,
      "step": 6332
    },
    {
      "epoch": 0.8561723700887199,
      "grad_norm": 0.22392292320728302,
      "learning_rate": 1.7365561793635431e-06,
      "loss": 0.056971073150634766,
      "step": 6333
    },
    {
      "epoch": 0.8563075623151669,
      "grad_norm": 0.4279118478298187,
      "learning_rate": 1.7333608528363227e-06,
      "loss": 0.07200860977172852,
      "step": 6334
    },
    {
      "epoch": 0.8564427545416139,
      "grad_norm": 0.2679310739040375,
      "learning_rate": 1.7301682884941128e-06,
      "loss": 0.06176877021789551,
      "step": 6335
    },
    {
      "epoch": 0.8565779467680609,
      "grad_norm": 0.2931089699268341,
      "learning_rate": 1.726978487001632e-06,
      "loss": 0.055226802825927734,
      "step": 6336
    },
    {
      "epoch": 0.8567131389945079,
      "grad_norm": 0.36156946420669556,
      "learning_rate": 1.7237914490230072e-06,
      "loss": 0.09280610084533691,
      "step": 6337
    },
    {
      "epoch": 0.8568483312209548,
      "grad_norm": 0.33498939871788025,
      "learning_rate": 1.7206071752218027e-06,
      "loss": 0.05925607681274414,
      "step": 6338
    },
    {
      "epoch": 0.8569835234474018,
      "grad_norm": 0.2283213585615158,
      "learning_rate": 1.7174256662610032e-06,
      "loss": 0.042500972747802734,
      "step": 6339
    },
    {
      "epoch": 0.8571187156738488,
      "grad_norm": 0.39317595958709717,
      "learning_rate": 1.714246922803004e-06,
      "loss": 0.07482242584228516,
      "step": 6340
    },
    {
      "epoch": 0.8572539079002958,
      "grad_norm": 0.4603132903575897,
      "learning_rate": 1.7110709455096468e-06,
      "loss": 0.04822203516960144,
      "step": 6341
    },
    {
      "epoch": 0.8573891001267427,
      "grad_norm": 0.5118915438652039,
      "learning_rate": 1.7078977350421815e-06,
      "loss": 0.06947517395019531,
      "step": 6342
    },
    {
      "epoch": 0.8575242923531897,
      "grad_norm": 0.22298239171504974,
      "learning_rate": 1.7047272920612926e-06,
      "loss": 0.04576289653778076,
      "step": 6343
    },
    {
      "epoch": 0.8576594845796367,
      "grad_norm": 0.2964182198047638,
      "learning_rate": 1.7015596172270841e-06,
      "loss": 0.06253862380981445,
      "step": 6344
    },
    {
      "epoch": 0.8577946768060837,
      "grad_norm": 0.3406200408935547,
      "learning_rate": 1.6983947111990717e-06,
      "loss": 0.0810537338256836,
      "step": 6345
    },
    {
      "epoch": 0.8579298690325307,
      "grad_norm": 0.689912736415863,
      "learning_rate": 1.695232574636218e-06,
      "loss": 0.08181846141815186,
      "step": 6346
    },
    {
      "epoch": 0.8580650612589776,
      "grad_norm": 0.28458845615386963,
      "learning_rate": 1.6920732081968882e-06,
      "loss": 0.0625309944152832,
      "step": 6347
    },
    {
      "epoch": 0.8582002534854246,
      "grad_norm": 0.29292938113212585,
      "learning_rate": 1.6889166125388878e-06,
      "loss": 0.056079864501953125,
      "step": 6348
    },
    {
      "epoch": 0.8583354457118716,
      "grad_norm": 0.38736993074417114,
      "learning_rate": 1.6857627883194277e-06,
      "loss": 0.06703901290893555,
      "step": 6349
    },
    {
      "epoch": 0.8584706379383186,
      "grad_norm": 0.5575077533721924,
      "learning_rate": 1.6826117361951577e-06,
      "loss": 0.08139348030090332,
      "step": 6350
    },
    {
      "epoch": 0.8586058301647655,
      "grad_norm": 0.21039235591888428,
      "learning_rate": 1.6794634568221412e-06,
      "loss": 0.05111062526702881,
      "step": 6351
    },
    {
      "epoch": 0.8587410223912125,
      "grad_norm": 0.2495880275964737,
      "learning_rate": 1.676317950855864e-06,
      "loss": 0.07018160820007324,
      "step": 6352
    },
    {
      "epoch": 0.8588762146176595,
      "grad_norm": 0.28126993775367737,
      "learning_rate": 1.6731752189512456e-06,
      "loss": 0.04293978214263916,
      "step": 6353
    },
    {
      "epoch": 0.8590114068441065,
      "grad_norm": 0.29608556628227234,
      "learning_rate": 1.6700352617626092e-06,
      "loss": 0.06672406196594238,
      "step": 6354
    },
    {
      "epoch": 0.8591465990705535,
      "grad_norm": 0.19749513268470764,
      "learning_rate": 1.6668980799437167e-06,
      "loss": 0.044228553771972656,
      "step": 6355
    },
    {
      "epoch": 0.8592817912970004,
      "grad_norm": 0.227354034781456,
      "learning_rate": 1.6637636741477458e-06,
      "loss": 0.041936516761779785,
      "step": 6356
    },
    {
      "epoch": 0.8594169835234474,
      "grad_norm": 0.2797456681728363,
      "learning_rate": 1.6606320450272943e-06,
      "loss": 0.06017112731933594,
      "step": 6357
    },
    {
      "epoch": 0.8595521757498944,
      "grad_norm": 0.2043454349040985,
      "learning_rate": 1.657503193234386e-06,
      "loss": 0.04774284362792969,
      "step": 6358
    },
    {
      "epoch": 0.8596873679763414,
      "grad_norm": 0.25736096501350403,
      "learning_rate": 1.654377119420461e-06,
      "loss": 0.05237627029418945,
      "step": 6359
    },
    {
      "epoch": 0.8598225602027884,
      "grad_norm": 0.2511076033115387,
      "learning_rate": 1.6512538242363889e-06,
      "loss": 0.04599905014038086,
      "step": 6360
    },
    {
      "epoch": 0.8599577524292353,
      "grad_norm": 0.35518598556518555,
      "learning_rate": 1.6481333083324563e-06,
      "loss": 0.07319915294647217,
      "step": 6361
    },
    {
      "epoch": 0.8600929446556823,
      "grad_norm": 0.2266836315393448,
      "learning_rate": 1.6450155723583698e-06,
      "loss": 0.04404246807098389,
      "step": 6362
    },
    {
      "epoch": 0.8602281368821293,
      "grad_norm": 0.25336751341819763,
      "learning_rate": 1.6419006169632573e-06,
      "loss": 0.04238080978393555,
      "step": 6363
    },
    {
      "epoch": 0.8603633291085763,
      "grad_norm": 0.47562745213508606,
      "learning_rate": 1.638788442795668e-06,
      "loss": 0.07038497924804688,
      "step": 6364
    },
    {
      "epoch": 0.8604985213350232,
      "grad_norm": 0.44237881898880005,
      "learning_rate": 1.6356790505035785e-06,
      "loss": 0.0656578540802002,
      "step": 6365
    },
    {
      "epoch": 0.8606337135614702,
      "grad_norm": 0.34531697630882263,
      "learning_rate": 1.6325724407343795e-06,
      "loss": 0.07509088516235352,
      "step": 6366
    },
    {
      "epoch": 0.8607689057879172,
      "grad_norm": 0.29616063833236694,
      "learning_rate": 1.6294686141348801e-06,
      "loss": 0.06722640991210938,
      "step": 6367
    },
    {
      "epoch": 0.8609040980143642,
      "grad_norm": 0.47764697670936584,
      "learning_rate": 1.626367571351317e-06,
      "loss": 0.07256317138671875,
      "step": 6368
    },
    {
      "epoch": 0.8610392902408112,
      "grad_norm": 0.3566100001335144,
      "learning_rate": 1.6232693130293386e-06,
      "loss": 0.0624079704284668,
      "step": 6369
    },
    {
      "epoch": 0.8611744824672581,
      "grad_norm": 0.3481585681438446,
      "learning_rate": 1.6201738398140254e-06,
      "loss": 0.06539058685302734,
      "step": 6370
    },
    {
      "epoch": 0.8613096746937051,
      "grad_norm": 0.31021958589553833,
      "learning_rate": 1.6170811523498718e-06,
      "loss": 0.05242288112640381,
      "step": 6371
    },
    {
      "epoch": 0.8614448669201521,
      "grad_norm": 0.26138874888420105,
      "learning_rate": 1.613991251280783e-06,
      "loss": 0.06061601638793945,
      "step": 6372
    },
    {
      "epoch": 0.8615800591465991,
      "grad_norm": 0.36094316840171814,
      "learning_rate": 1.6109041372501028e-06,
      "loss": 0.06380867958068848,
      "step": 6373
    },
    {
      "epoch": 0.861715251373046,
      "grad_norm": 0.33734700083732605,
      "learning_rate": 1.6078198109005766e-06,
      "loss": 0.061342716217041016,
      "step": 6374
    },
    {
      "epoch": 0.861850443599493,
      "grad_norm": 0.35576608777046204,
      "learning_rate": 1.6047382728743843e-06,
      "loss": 0.08353328704833984,
      "step": 6375
    },
    {
      "epoch": 0.86198563582594,
      "grad_norm": 0.3268064856529236,
      "learning_rate": 1.6016595238131176e-06,
      "loss": 0.06797528266906738,
      "step": 6376
    },
    {
      "epoch": 0.862120828052387,
      "grad_norm": 0.34670352935791016,
      "learning_rate": 1.5985835643577824e-06,
      "loss": 0.05814170837402344,
      "step": 6377
    },
    {
      "epoch": 0.862256020278834,
      "grad_norm": 0.3909570872783661,
      "learning_rate": 1.5955103951488177e-06,
      "loss": 0.0650014877319336,
      "step": 6378
    },
    {
      "epoch": 0.862391212505281,
      "grad_norm": 0.5527427196502686,
      "learning_rate": 1.5924400168260666e-06,
      "loss": 0.07494068145751953,
      "step": 6379
    },
    {
      "epoch": 0.8625264047317279,
      "grad_norm": 0.24005034565925598,
      "learning_rate": 1.5893724300288064e-06,
      "loss": 0.05875349044799805,
      "step": 6380
    },
    {
      "epoch": 0.8626615969581749,
      "grad_norm": 0.2700231671333313,
      "learning_rate": 1.5863076353957196e-06,
      "loss": 0.06253039836883545,
      "step": 6381
    },
    {
      "epoch": 0.8627967891846219,
      "grad_norm": 0.3574848473072052,
      "learning_rate": 1.5832456335649104e-06,
      "loss": 0.07944250106811523,
      "step": 6382
    },
    {
      "epoch": 0.8629319814110689,
      "grad_norm": 0.2679492235183716,
      "learning_rate": 1.580186425173909e-06,
      "loss": 0.06584036350250244,
      "step": 6383
    },
    {
      "epoch": 0.8630671736375158,
      "grad_norm": 0.25383493304252625,
      "learning_rate": 1.5771300108596543e-06,
      "loss": 0.04120588302612305,
      "step": 6384
    },
    {
      "epoch": 0.8632023658639628,
      "grad_norm": 0.2823106050491333,
      "learning_rate": 1.5740763912585171e-06,
      "loss": 0.0635061264038086,
      "step": 6385
    },
    {
      "epoch": 0.8633375580904098,
      "grad_norm": 0.23393428325653076,
      "learning_rate": 1.5710255670062657e-06,
      "loss": 0.06041669845581055,
      "step": 6386
    },
    {
      "epoch": 0.8634727503168568,
      "grad_norm": 0.6842442154884338,
      "learning_rate": 1.567977538738105e-06,
      "loss": 0.08539104461669922,
      "step": 6387
    },
    {
      "epoch": 0.8636079425433038,
      "grad_norm": 0.22804664075374603,
      "learning_rate": 1.5649323070886494e-06,
      "loss": 0.0572429895401001,
      "step": 6388
    },
    {
      "epoch": 0.8637431347697507,
      "grad_norm": 0.2404095083475113,
      "learning_rate": 1.5618898726919284e-06,
      "loss": 0.05065178871154785,
      "step": 6389
    },
    {
      "epoch": 0.8638783269961977,
      "grad_norm": 0.19856587052345276,
      "learning_rate": 1.5588502361814032e-06,
      "loss": 0.04634976387023926,
      "step": 6390
    },
    {
      "epoch": 0.8640135192226447,
      "grad_norm": 0.3801005780696869,
      "learning_rate": 1.5558133981899314e-06,
      "loss": 0.07269001007080078,
      "step": 6391
    },
    {
      "epoch": 0.8641487114490917,
      "grad_norm": 0.5373231172561646,
      "learning_rate": 1.5527793593498053e-06,
      "loss": 0.08079099655151367,
      "step": 6392
    },
    {
      "epoch": 0.8642839036755386,
      "grad_norm": 0.22107817232608795,
      "learning_rate": 1.5497481202927244e-06,
      "loss": 0.045655250549316406,
      "step": 6393
    },
    {
      "epoch": 0.8644190959019856,
      "grad_norm": 0.3428370952606201,
      "learning_rate": 1.5467196816498107e-06,
      "loss": 0.07546043395996094,
      "step": 6394
    },
    {
      "epoch": 0.8645542881284326,
      "grad_norm": 0.19990627467632294,
      "learning_rate": 1.5436940440516018e-06,
      "loss": 0.039630770683288574,
      "step": 6395
    },
    {
      "epoch": 0.8646894803548796,
      "grad_norm": 0.3012692630290985,
      "learning_rate": 1.5406712081280484e-06,
      "loss": 0.07137680053710938,
      "step": 6396
    },
    {
      "epoch": 0.8648246725813266,
      "grad_norm": 0.29851406812667847,
      "learning_rate": 1.5376511745085254e-06,
      "loss": 0.06910943984985352,
      "step": 6397
    },
    {
      "epoch": 0.8649598648077735,
      "grad_norm": 0.3407972455024719,
      "learning_rate": 1.5346339438218181e-06,
      "loss": 0.07711267471313477,
      "step": 6398
    },
    {
      "epoch": 0.8650950570342205,
      "grad_norm": 0.24595056474208832,
      "learning_rate": 1.5316195166961295e-06,
      "loss": 0.04909813404083252,
      "step": 6399
    },
    {
      "epoch": 0.8652302492606675,
      "grad_norm": 0.2624797523021698,
      "learning_rate": 1.5286078937590802e-06,
      "loss": 0.05071711540222168,
      "step": 6400
    },
    {
      "epoch": 0.8653654414871145,
      "grad_norm": 0.2919888496398926,
      "learning_rate": 1.5255990756377025e-06,
      "loss": 0.06617045402526855,
      "step": 6401
    },
    {
      "epoch": 0.8655006337135615,
      "grad_norm": 0.2081049531698227,
      "learning_rate": 1.5225930629584534e-06,
      "loss": 0.04576921463012695,
      "step": 6402
    },
    {
      "epoch": 0.8656358259400084,
      "grad_norm": 0.6203389167785645,
      "learning_rate": 1.5195898563472038e-06,
      "loss": 0.08098185062408447,
      "step": 6403
    },
    {
      "epoch": 0.8657710181664554,
      "grad_norm": 0.30733442306518555,
      "learning_rate": 1.5165894564292254e-06,
      "loss": 0.05830669403076172,
      "step": 6404
    },
    {
      "epoch": 0.8659062103929024,
      "grad_norm": 0.25245144963264465,
      "learning_rate": 1.5135918638292269e-06,
      "loss": 0.05776369571685791,
      "step": 6405
    },
    {
      "epoch": 0.8660414026193494,
      "grad_norm": 0.3321148157119751,
      "learning_rate": 1.5105970791713186e-06,
      "loss": 0.06632804870605469,
      "step": 6406
    },
    {
      "epoch": 0.8661765948457963,
      "grad_norm": 0.4239051043987274,
      "learning_rate": 1.5076051030790355e-06,
      "loss": 0.09187984466552734,
      "step": 6407
    },
    {
      "epoch": 0.8663117870722433,
      "grad_norm": 0.26415812969207764,
      "learning_rate": 1.5046159361753226e-06,
      "loss": 0.06444501876831055,
      "step": 6408
    },
    {
      "epoch": 0.8664469792986903,
      "grad_norm": 0.2414625734090805,
      "learning_rate": 1.5016295790825336e-06,
      "loss": 0.061753273010253906,
      "step": 6409
    },
    {
      "epoch": 0.8665821715251373,
      "grad_norm": 0.2623870074748993,
      "learning_rate": 1.4986460324224493e-06,
      "loss": 0.06393909454345703,
      "step": 6410
    },
    {
      "epoch": 0.8667173637515843,
      "grad_norm": 0.30190572142601013,
      "learning_rate": 1.4956652968162582e-06,
      "loss": 0.05830419063568115,
      "step": 6411
    },
    {
      "epoch": 0.8668525559780312,
      "grad_norm": 0.28045520186424255,
      "learning_rate": 1.492687372884567e-06,
      "loss": 0.06609934568405151,
      "step": 6412
    },
    {
      "epoch": 0.8669877482044782,
      "grad_norm": 0.3405071794986725,
      "learning_rate": 1.4897122612473978e-06,
      "loss": 0.0882425308227539,
      "step": 6413
    },
    {
      "epoch": 0.8671229404309252,
      "grad_norm": 0.2653646469116211,
      "learning_rate": 1.4867399625241772e-06,
      "loss": 0.06584668159484863,
      "step": 6414
    },
    {
      "epoch": 0.8672581326573722,
      "grad_norm": 0.2610979676246643,
      "learning_rate": 1.4837704773337602e-06,
      "loss": 0.05601084232330322,
      "step": 6415
    },
    {
      "epoch": 0.8673933248838192,
      "grad_norm": 0.20413154363632202,
      "learning_rate": 1.4808038062944036e-06,
      "loss": 0.05117321014404297,
      "step": 6416
    },
    {
      "epoch": 0.8675285171102661,
      "grad_norm": 0.35246792435646057,
      "learning_rate": 1.4778399500237933e-06,
      "loss": 0.062796950340271,
      "step": 6417
    },
    {
      "epoch": 0.8676637093367131,
      "grad_norm": 0.26034027338027954,
      "learning_rate": 1.4748789091390124e-06,
      "loss": 0.04463851451873779,
      "step": 6418
    },
    {
      "epoch": 0.8677989015631601,
      "grad_norm": 0.28708788752555847,
      "learning_rate": 1.471920684256563e-06,
      "loss": 0.06418430805206299,
      "step": 6419
    },
    {
      "epoch": 0.8679340937896071,
      "grad_norm": 0.2791190445423126,
      "learning_rate": 1.4689652759923721e-06,
      "loss": 0.04939424991607666,
      "step": 6420
    },
    {
      "epoch": 0.868069286016054,
      "grad_norm": 0.45456793904304504,
      "learning_rate": 1.4660126849617645e-06,
      "loss": 0.06717270612716675,
      "step": 6421
    },
    {
      "epoch": 0.868204478242501,
      "grad_norm": 0.25810715556144714,
      "learning_rate": 1.4630629117794914e-06,
      "loss": 0.06435108184814453,
      "step": 6422
    },
    {
      "epoch": 0.868339670468948,
      "grad_norm": 0.2916312515735626,
      "learning_rate": 1.4601159570597033e-06,
      "loss": 0.06496095657348633,
      "step": 6423
    },
    {
      "epoch": 0.868474862695395,
      "grad_norm": 0.30769601464271545,
      "learning_rate": 1.4571718214159795e-06,
      "loss": 0.06269550323486328,
      "step": 6424
    },
    {
      "epoch": 0.868610054921842,
      "grad_norm": 0.3282182812690735,
      "learning_rate": 1.454230505461303e-06,
      "loss": 0.058388471603393555,
      "step": 6425
    },
    {
      "epoch": 0.8687452471482889,
      "grad_norm": 0.41010263562202454,
      "learning_rate": 1.4512920098080672e-06,
      "loss": 0.08212709426879883,
      "step": 6426
    },
    {
      "epoch": 0.8688804393747359,
      "grad_norm": 0.31919345259666443,
      "learning_rate": 1.4483563350680878e-06,
      "loss": 0.059348106384277344,
      "step": 6427
    },
    {
      "epoch": 0.8690156316011829,
      "grad_norm": 0.41883012652397156,
      "learning_rate": 1.4454234818525824e-06,
      "loss": 0.09795856475830078,
      "step": 6428
    },
    {
      "epoch": 0.8691508238276299,
      "grad_norm": 0.2800425887107849,
      "learning_rate": 1.4424934507721926e-06,
      "loss": 0.0751805305480957,
      "step": 6429
    },
    {
      "epoch": 0.8692860160540768,
      "grad_norm": 0.3102518618106842,
      "learning_rate": 1.4395662424369622e-06,
      "loss": 0.056693315505981445,
      "step": 6430
    },
    {
      "epoch": 0.8694212082805238,
      "grad_norm": 0.33799704909324646,
      "learning_rate": 1.436641857456355e-06,
      "loss": 0.054157257080078125,
      "step": 6431
    },
    {
      "epoch": 0.8695564005069708,
      "grad_norm": 0.2771012485027313,
      "learning_rate": 1.4337202964392409e-06,
      "loss": 0.05644416809082031,
      "step": 6432
    },
    {
      "epoch": 0.8696915927334178,
      "grad_norm": 0.26662665605545044,
      "learning_rate": 1.4308015599939033e-06,
      "loss": 0.04773449897766113,
      "step": 6433
    },
    {
      "epoch": 0.8698267849598648,
      "grad_norm": 0.36784619092941284,
      "learning_rate": 1.4278856487280428e-06,
      "loss": 0.09210872650146484,
      "step": 6434
    },
    {
      "epoch": 0.8699619771863117,
      "grad_norm": 0.18432922661304474,
      "learning_rate": 1.4249725632487653e-06,
      "loss": 0.04014027118682861,
      "step": 6435
    },
    {
      "epoch": 0.8700971694127587,
      "grad_norm": 0.18665963411331177,
      "learning_rate": 1.4220623041625924e-06,
      "loss": 0.030408620834350586,
      "step": 6436
    },
    {
      "epoch": 0.8702323616392057,
      "grad_norm": 0.47944337129592896,
      "learning_rate": 1.4191548720754527e-06,
      "loss": 0.07666683197021484,
      "step": 6437
    },
    {
      "epoch": 0.8703675538656527,
      "grad_norm": 0.18948031961917877,
      "learning_rate": 1.4162502675926887e-06,
      "loss": 0.04884648323059082,
      "step": 6438
    },
    {
      "epoch": 0.8705027460920997,
      "grad_norm": 0.3038956820964813,
      "learning_rate": 1.4133484913190596e-06,
      "loss": 0.06389188766479492,
      "step": 6439
    },
    {
      "epoch": 0.8706379383185466,
      "grad_norm": 0.1291380524635315,
      "learning_rate": 1.4104495438587295e-06,
      "loss": 0.03372645378112793,
      "step": 6440
    },
    {
      "epoch": 0.8707731305449936,
      "grad_norm": 0.1952275037765503,
      "learning_rate": 1.4075534258152667e-06,
      "loss": 0.04088759422302246,
      "step": 6441
    },
    {
      "epoch": 0.8709083227714406,
      "grad_norm": 0.2805889844894409,
      "learning_rate": 1.4046601377916673e-06,
      "loss": 0.07518148422241211,
      "step": 6442
    },
    {
      "epoch": 0.8710435149978876,
      "grad_norm": 0.23121748864650726,
      "learning_rate": 1.4017696803903246e-06,
      "loss": 0.054850101470947266,
      "step": 6443
    },
    {
      "epoch": 0.8711787072243345,
      "grad_norm": 0.2672847509384155,
      "learning_rate": 1.3988820542130504e-06,
      "loss": 0.058417320251464844,
      "step": 6444
    },
    {
      "epoch": 0.8713138994507815,
      "grad_norm": 0.4756321310997009,
      "learning_rate": 1.395997259861067e-06,
      "loss": 0.05745410919189453,
      "step": 6445
    },
    {
      "epoch": 0.8714490916772285,
      "grad_norm": 0.113944873213768,
      "learning_rate": 1.3931152979349926e-06,
      "loss": 0.030443429946899414,
      "step": 6446
    },
    {
      "epoch": 0.8715842839036756,
      "grad_norm": 0.37563037872314453,
      "learning_rate": 1.3902361690348769e-06,
      "loss": 0.07578229904174805,
      "step": 6447
    },
    {
      "epoch": 0.8717194761301226,
      "grad_norm": 0.4331814646720886,
      "learning_rate": 1.3873598737601639e-06,
      "loss": 0.1031179428100586,
      "step": 6448
    },
    {
      "epoch": 0.8718546683565696,
      "grad_norm": 0.16388234496116638,
      "learning_rate": 1.3844864127097229e-06,
      "loss": 0.04517841339111328,
      "step": 6449
    },
    {
      "epoch": 0.8719898605830165,
      "grad_norm": 0.44739118218421936,
      "learning_rate": 1.3816157864818151e-06,
      "loss": 0.06436538696289062,
      "step": 6450
    },
    {
      "epoch": 0.8721250528094635,
      "grad_norm": 0.2669844627380371,
      "learning_rate": 1.3787479956741194e-06,
      "loss": 0.08227872848510742,
      "step": 6451
    },
    {
      "epoch": 0.8722602450359105,
      "grad_norm": 0.24546951055526733,
      "learning_rate": 1.3758830408837314e-06,
      "loss": 0.04765129089355469,
      "step": 6452
    },
    {
      "epoch": 0.8723954372623575,
      "grad_norm": 0.2474411427974701,
      "learning_rate": 1.3730209227071439e-06,
      "loss": 0.04482245445251465,
      "step": 6453
    },
    {
      "epoch": 0.8725306294888044,
      "grad_norm": 1.1443266868591309,
      "learning_rate": 1.3701616417402734e-06,
      "loss": 0.07504284381866455,
      "step": 6454
    },
    {
      "epoch": 0.8726658217152514,
      "grad_norm": 0.37688857316970825,
      "learning_rate": 1.367305198578429e-06,
      "loss": 0.06220531463623047,
      "step": 6455
    },
    {
      "epoch": 0.8728010139416984,
      "grad_norm": 0.3509252965450287,
      "learning_rate": 1.36445159381634e-06,
      "loss": 0.07417893409729004,
      "step": 6456
    },
    {
      "epoch": 0.8729362061681454,
      "grad_norm": 0.3079795837402344,
      "learning_rate": 1.361600828048144e-06,
      "loss": 0.0628666877746582,
      "step": 6457
    },
    {
      "epoch": 0.8730713983945924,
      "grad_norm": 0.2026120275259018,
      "learning_rate": 1.3587529018673816e-06,
      "loss": 0.05359983444213867,
      "step": 6458
    },
    {
      "epoch": 0.8732065906210393,
      "grad_norm": 0.1779441386461258,
      "learning_rate": 1.3559078158670152e-06,
      "loss": 0.03676795959472656,
      "step": 6459
    },
    {
      "epoch": 0.8733417828474863,
      "grad_norm": 0.28361883759498596,
      "learning_rate": 1.353065570639394e-06,
      "loss": 0.06538915634155273,
      "step": 6460
    },
    {
      "epoch": 0.8734769750739333,
      "grad_norm": 0.4011971354484558,
      "learning_rate": 1.3502261667763e-06,
      "loss": 0.06628942489624023,
      "step": 6461
    },
    {
      "epoch": 0.8736121673003803,
      "grad_norm": 0.2483941912651062,
      "learning_rate": 1.3473896048689067e-06,
      "loss": 0.06180381774902344,
      "step": 6462
    },
    {
      "epoch": 0.8737473595268272,
      "grad_norm": 0.2961508631706238,
      "learning_rate": 1.3445558855078017e-06,
      "loss": 0.0854034423828125,
      "step": 6463
    },
    {
      "epoch": 0.8738825517532742,
      "grad_norm": 0.3669814467430115,
      "learning_rate": 1.3417250092829814e-06,
      "loss": 0.05763888359069824,
      "step": 6464
    },
    {
      "epoch": 0.8740177439797212,
      "grad_norm": 0.18326736986637115,
      "learning_rate": 1.338896976783846e-06,
      "loss": 0.04211020469665527,
      "step": 6465
    },
    {
      "epoch": 0.8741529362061682,
      "grad_norm": 0.31272953748703003,
      "learning_rate": 1.336071788599213e-06,
      "loss": 0.04508852958679199,
      "step": 6466
    },
    {
      "epoch": 0.8742881284326152,
      "grad_norm": 0.18777328729629517,
      "learning_rate": 1.3332494453172982e-06,
      "loss": 0.04358077049255371,
      "step": 6467
    },
    {
      "epoch": 0.8744233206590621,
      "grad_norm": 0.6595178246498108,
      "learning_rate": 1.3304299475257287e-06,
      "loss": 0.1093130111694336,
      "step": 6468
    },
    {
      "epoch": 0.8745585128855091,
      "grad_norm": 0.2750481069087982,
      "learning_rate": 1.3276132958115394e-06,
      "loss": 0.06076812744140625,
      "step": 6469
    },
    {
      "epoch": 0.8746937051119561,
      "grad_norm": 0.1557319164276123,
      "learning_rate": 1.32479949076117e-06,
      "loss": 0.047522544860839844,
      "step": 6470
    },
    {
      "epoch": 0.8748288973384031,
      "grad_norm": 0.2280968874692917,
      "learning_rate": 1.3219885329604747e-06,
      "loss": 0.057415008544921875,
      "step": 6471
    },
    {
      "epoch": 0.8749640895648501,
      "grad_norm": 0.2835663855075836,
      "learning_rate": 1.319180422994709e-06,
      "loss": 0.06455183029174805,
      "step": 6472
    },
    {
      "epoch": 0.875099281791297,
      "grad_norm": 0.3616994321346283,
      "learning_rate": 1.3163751614485287e-06,
      "loss": 0.053040504455566406,
      "step": 6473
    },
    {
      "epoch": 0.875234474017744,
      "grad_norm": 0.4216803014278412,
      "learning_rate": 1.3135727489060113e-06,
      "loss": 0.07476139068603516,
      "step": 6474
    },
    {
      "epoch": 0.875369666244191,
      "grad_norm": 0.49317875504493713,
      "learning_rate": 1.3107731859506317e-06,
      "loss": 0.06447434425354004,
      "step": 6475
    },
    {
      "epoch": 0.875504858470638,
      "grad_norm": 0.2787105441093445,
      "learning_rate": 1.3079764731652772e-06,
      "loss": 0.05824148654937744,
      "step": 6476
    },
    {
      "epoch": 0.875640050697085,
      "grad_norm": 0.3161103427410126,
      "learning_rate": 1.3051826111322368e-06,
      "loss": 0.07431697845458984,
      "step": 6477
    },
    {
      "epoch": 0.8757752429235319,
      "grad_norm": 0.2119627445936203,
      "learning_rate": 1.3023916004332021e-06,
      "loss": 0.054253339767456055,
      "step": 6478
    },
    {
      "epoch": 0.8759104351499789,
      "grad_norm": 0.35281893610954285,
      "learning_rate": 1.2996034416492847e-06,
      "loss": 0.0543445348739624,
      "step": 6479
    },
    {
      "epoch": 0.8760456273764259,
      "grad_norm": 0.34522977471351624,
      "learning_rate": 1.2968181353609854e-06,
      "loss": 0.06467533111572266,
      "step": 6480
    },
    {
      "epoch": 0.8761808196028729,
      "grad_norm": 0.26410940289497375,
      "learning_rate": 1.2940356821482285e-06,
      "loss": 0.0499957799911499,
      "step": 6481
    },
    {
      "epoch": 0.8763160118293198,
      "grad_norm": 0.29039788246154785,
      "learning_rate": 1.291256082590334e-06,
      "loss": 0.06514358520507812,
      "step": 6482
    },
    {
      "epoch": 0.8764512040557668,
      "grad_norm": 0.25613754987716675,
      "learning_rate": 1.2884793372660208e-06,
      "loss": 0.07200336456298828,
      "step": 6483
    },
    {
      "epoch": 0.8765863962822138,
      "grad_norm": 0.2594621777534485,
      "learning_rate": 1.285705446753433e-06,
      "loss": 0.04393339157104492,
      "step": 6484
    },
    {
      "epoch": 0.8767215885086608,
      "grad_norm": 0.25352996587753296,
      "learning_rate": 1.2829344116301e-06,
      "loss": 0.05715656280517578,
      "step": 6485
    },
    {
      "epoch": 0.8768567807351078,
      "grad_norm": 0.48465704917907715,
      "learning_rate": 1.2801662324729774e-06,
      "loss": 0.09092998504638672,
      "step": 6486
    },
    {
      "epoch": 0.8769919729615547,
      "grad_norm": 0.4309283494949341,
      "learning_rate": 1.2774009098584055e-06,
      "loss": 0.07410907745361328,
      "step": 6487
    },
    {
      "epoch": 0.8771271651880017,
      "grad_norm": 0.2961377799510956,
      "learning_rate": 1.274638444362139e-06,
      "loss": 0.06084012985229492,
      "step": 6488
    },
    {
      "epoch": 0.8772623574144487,
      "grad_norm": 0.25042974948883057,
      "learning_rate": 1.2718788365593443e-06,
      "loss": 0.0670318603515625,
      "step": 6489
    },
    {
      "epoch": 0.8773975496408957,
      "grad_norm": 0.28116375207901,
      "learning_rate": 1.26912208702458e-06,
      "loss": 0.04669928550720215,
      "step": 6490
    },
    {
      "epoch": 0.8775327418673426,
      "grad_norm": 0.2501729726791382,
      "learning_rate": 1.2663681963318242e-06,
      "loss": 0.04172492027282715,
      "step": 6491
    },
    {
      "epoch": 0.8776679340937896,
      "grad_norm": 0.28565719723701477,
      "learning_rate": 1.2636171650544443e-06,
      "loss": 0.04670226573944092,
      "step": 6492
    },
    {
      "epoch": 0.8778031263202366,
      "grad_norm": 0.261688232421875,
      "learning_rate": 1.260868993765219e-06,
      "loss": 0.05399513244628906,
      "step": 6493
    },
    {
      "epoch": 0.8779383185466836,
      "grad_norm": 0.27596351504325867,
      "learning_rate": 1.258123683036339e-06,
      "loss": 0.05591607093811035,
      "step": 6494
    },
    {
      "epoch": 0.8780735107731306,
      "grad_norm": 0.290924072265625,
      "learning_rate": 1.2553812334393872e-06,
      "loss": 0.06591248512268066,
      "step": 6495
    },
    {
      "epoch": 0.8782087029995775,
      "grad_norm": 0.3651522994041443,
      "learning_rate": 1.2526416455453582e-06,
      "loss": 0.0653609037399292,
      "step": 6496
    },
    {
      "epoch": 0.8783438952260245,
      "grad_norm": 0.38874951004981995,
      "learning_rate": 1.249904919924646e-06,
      "loss": 0.06723189353942871,
      "step": 6497
    },
    {
      "epoch": 0.8784790874524715,
      "grad_norm": 0.309001624584198,
      "learning_rate": 1.2471710571470579e-06,
      "loss": 0.058152854442596436,
      "step": 6498
    },
    {
      "epoch": 0.8786142796789185,
      "grad_norm": 0.20150302350521088,
      "learning_rate": 1.2444400577817922e-06,
      "loss": 0.043239593505859375,
      "step": 6499
    },
    {
      "epoch": 0.8787494719053655,
      "grad_norm": 0.32289421558380127,
      "learning_rate": 1.2417119223974621e-06,
      "loss": 0.06432437896728516,
      "step": 6500
    },
    {
      "epoch": 0.8788846641318124,
      "grad_norm": 0.3440784811973572,
      "learning_rate": 1.2389866515620768e-06,
      "loss": 0.04827141761779785,
      "step": 6501
    },
    {
      "epoch": 0.8790198563582594,
      "grad_norm": 0.337507039308548,
      "learning_rate": 1.2362642458430505e-06,
      "loss": 0.056401729583740234,
      "step": 6502
    },
    {
      "epoch": 0.8791550485847064,
      "grad_norm": 0.30003872513771057,
      "learning_rate": 1.2335447058072103e-06,
      "loss": 0.0617830753326416,
      "step": 6503
    },
    {
      "epoch": 0.8792902408111534,
      "grad_norm": 0.2757251560688019,
      "learning_rate": 1.230828032020771e-06,
      "loss": 0.05974173545837402,
      "step": 6504
    },
    {
      "epoch": 0.8794254330376003,
      "grad_norm": 0.3699774146080017,
      "learning_rate": 1.2281142250493638e-06,
      "loss": 0.06078982353210449,
      "step": 6505
    },
    {
      "epoch": 0.8795606252640473,
      "grad_norm": 0.33825087547302246,
      "learning_rate": 1.225403285458015e-06,
      "loss": 0.06746101379394531,
      "step": 6506
    },
    {
      "epoch": 0.8796958174904943,
      "grad_norm": 0.15928079187870026,
      "learning_rate": 1.2226952138111546e-06,
      "loss": 0.0419154167175293,
      "step": 6507
    },
    {
      "epoch": 0.8798310097169413,
      "grad_norm": 0.20003150403499603,
      "learning_rate": 1.219990010672622e-06,
      "loss": 0.04114890098571777,
      "step": 6508
    },
    {
      "epoch": 0.8799662019433883,
      "grad_norm": 0.20392554998397827,
      "learning_rate": 1.2172876766056562e-06,
      "loss": 0.053170204162597656,
      "step": 6509
    },
    {
      "epoch": 0.8801013941698352,
      "grad_norm": 0.17940299212932587,
      "learning_rate": 1.2145882121728906e-06,
      "loss": 0.04723215103149414,
      "step": 6510
    },
    {
      "epoch": 0.8802365863962822,
      "grad_norm": 0.2120024710893631,
      "learning_rate": 1.2118916179363727e-06,
      "loss": 0.03814142942428589,
      "step": 6511
    },
    {
      "epoch": 0.8803717786227292,
      "grad_norm": 0.2852402329444885,
      "learning_rate": 1.209197894457546e-06,
      "loss": 0.059386253356933594,
      "step": 6512
    },
    {
      "epoch": 0.8805069708491762,
      "grad_norm": 0.24861997365951538,
      "learning_rate": 1.2065070422972606e-06,
      "loss": 0.05455970764160156,
      "step": 6513
    },
    {
      "epoch": 0.8806421630756232,
      "grad_norm": 0.398517906665802,
      "learning_rate": 1.2038190620157685e-06,
      "loss": 0.10115814208984375,
      "step": 6514
    },
    {
      "epoch": 0.8807773553020701,
      "grad_norm": 0.37528836727142334,
      "learning_rate": 1.2011339541727117e-06,
      "loss": 0.07947158813476562,
      "step": 6515
    },
    {
      "epoch": 0.8809125475285171,
      "grad_norm": 0.1211782917380333,
      "learning_rate": 1.198451719327155e-06,
      "loss": 0.033257246017456055,
      "step": 6516
    },
    {
      "epoch": 0.8810477397549641,
      "grad_norm": 0.4164644181728363,
      "learning_rate": 1.1957723580375447e-06,
      "loss": 0.0876150131225586,
      "step": 6517
    },
    {
      "epoch": 0.8811829319814111,
      "grad_norm": 0.5009216666221619,
      "learning_rate": 1.193095870861748e-06,
      "loss": 0.08709907531738281,
      "step": 6518
    },
    {
      "epoch": 0.881318124207858,
      "grad_norm": 0.21098342537879944,
      "learning_rate": 1.1904222583570156e-06,
      "loss": 0.04839920997619629,
      "step": 6519
    },
    {
      "epoch": 0.881453316434305,
      "grad_norm": 0.40740567445755005,
      "learning_rate": 1.1877515210800077e-06,
      "loss": 0.07437682151794434,
      "step": 6520
    },
    {
      "epoch": 0.881588508660752,
      "grad_norm": 0.296480655670166,
      "learning_rate": 1.1850836595867925e-06,
      "loss": 0.059606075286865234,
      "step": 6521
    },
    {
      "epoch": 0.881723700887199,
      "grad_norm": 0.4370913505554199,
      "learning_rate": 1.1824186744328259e-06,
      "loss": 0.08585071563720703,
      "step": 6522
    },
    {
      "epoch": 0.881858893113646,
      "grad_norm": 0.28805118799209595,
      "learning_rate": 1.179756566172982e-06,
      "loss": 0.05884349346160889,
      "step": 6523
    },
    {
      "epoch": 0.8819940853400929,
      "grad_norm": 0.3825747072696686,
      "learning_rate": 1.177097335361516e-06,
      "loss": 0.08362960815429688,
      "step": 6524
    },
    {
      "epoch": 0.8821292775665399,
      "grad_norm": 0.2790987193584442,
      "learning_rate": 1.1744409825520969e-06,
      "loss": 0.04493391513824463,
      "step": 6525
    },
    {
      "epoch": 0.8822644697929869,
      "grad_norm": 0.26410335302352905,
      "learning_rate": 1.171787508297792e-06,
      "loss": 0.056577205657958984,
      "step": 6526
    },
    {
      "epoch": 0.8823996620194339,
      "grad_norm": 0.36177676916122437,
      "learning_rate": 1.1691369131510676e-06,
      "loss": 0.06201958656311035,
      "step": 6527
    },
    {
      "epoch": 0.8825348542458809,
      "grad_norm": 0.21636635065078735,
      "learning_rate": 1.1664891976637992e-06,
      "loss": 0.04668569564819336,
      "step": 6528
    },
    {
      "epoch": 0.8826700464723278,
      "grad_norm": 0.49301567673683167,
      "learning_rate": 1.1638443623872442e-06,
      "loss": 0.09611129760742188,
      "step": 6529
    },
    {
      "epoch": 0.8828052386987748,
      "grad_norm": 0.1500963568687439,
      "learning_rate": 1.1612024078720752e-06,
      "loss": 0.035099148750305176,
      "step": 6530
    },
    {
      "epoch": 0.8829404309252218,
      "grad_norm": 0.45520225167274475,
      "learning_rate": 1.1585633346683655e-06,
      "loss": 0.06321001052856445,
      "step": 6531
    },
    {
      "epoch": 0.8830756231516688,
      "grad_norm": 0.22220589220523834,
      "learning_rate": 1.155927143325579e-06,
      "loss": 0.05169510841369629,
      "step": 6532
    },
    {
      "epoch": 0.8832108153781157,
      "grad_norm": 0.32102254033088684,
      "learning_rate": 1.1532938343925887e-06,
      "loss": 0.059134721755981445,
      "step": 6533
    },
    {
      "epoch": 0.8833460076045627,
      "grad_norm": 0.37606751918792725,
      "learning_rate": 1.1506634084176587e-06,
      "loss": 0.07471990585327148,
      "step": 6534
    },
    {
      "epoch": 0.8834811998310097,
      "grad_norm": 0.30589964985847473,
      "learning_rate": 1.148035865948463e-06,
      "loss": 0.03421425819396973,
      "step": 6535
    },
    {
      "epoch": 0.8836163920574567,
      "grad_norm": 0.2541763186454773,
      "learning_rate": 1.1454112075320688e-06,
      "loss": 0.06158638000488281,
      "step": 6536
    },
    {
      "epoch": 0.8837515842839037,
      "grad_norm": 0.31818345189094543,
      "learning_rate": 1.1427894337149426e-06,
      "loss": 0.03979086875915527,
      "step": 6537
    },
    {
      "epoch": 0.8838867765103506,
      "grad_norm": 0.1632911115884781,
      "learning_rate": 1.1401705450429506e-06,
      "loss": 0.04642629623413086,
      "step": 6538
    },
    {
      "epoch": 0.8840219687367976,
      "grad_norm": 0.6135344505310059,
      "learning_rate": 1.1375545420613586e-06,
      "loss": 0.07690876722335815,
      "step": 6539
    },
    {
      "epoch": 0.8841571609632446,
      "grad_norm": 0.30065006017684937,
      "learning_rate": 1.1349414253148377e-06,
      "loss": 0.06247854232788086,
      "step": 6540
    },
    {
      "epoch": 0.8842923531896916,
      "grad_norm": 0.24106524884700775,
      "learning_rate": 1.1323311953474524e-06,
      "loss": 0.061649322509765625,
      "step": 6541
    },
    {
      "epoch": 0.8844275454161385,
      "grad_norm": 0.26848623156547546,
      "learning_rate": 1.1297238527026582e-06,
      "loss": 0.06455421447753906,
      "step": 6542
    },
    {
      "epoch": 0.8845627376425855,
      "grad_norm": 0.20980583131313324,
      "learning_rate": 1.1271193979233258e-06,
      "loss": 0.039904117584228516,
      "step": 6543
    },
    {
      "epoch": 0.8846979298690325,
      "grad_norm": 0.35986241698265076,
      "learning_rate": 1.1245178315517113e-06,
      "loss": 0.0874481201171875,
      "step": 6544
    },
    {
      "epoch": 0.8848331220954795,
      "grad_norm": 0.251010924577713,
      "learning_rate": 1.1219191541294798e-06,
      "loss": 0.044727861881256104,
      "step": 6545
    },
    {
      "epoch": 0.8849683143219265,
      "grad_norm": 0.156352698802948,
      "learning_rate": 1.1193233661976887e-06,
      "loss": 0.04927706718444824,
      "step": 6546
    },
    {
      "epoch": 0.8851035065483734,
      "grad_norm": 0.26124900579452515,
      "learning_rate": 1.1167304682967904e-06,
      "loss": 0.0682673454284668,
      "step": 6547
    },
    {
      "epoch": 0.8852386987748204,
      "grad_norm": 0.32066163420677185,
      "learning_rate": 1.114140460966645e-06,
      "loss": 0.06539630889892578,
      "step": 6548
    },
    {
      "epoch": 0.8853738910012674,
      "grad_norm": 0.3091585040092468,
      "learning_rate": 1.111553344746501e-06,
      "loss": 0.05526590347290039,
      "step": 6549
    },
    {
      "epoch": 0.8855090832277144,
      "grad_norm": 0.22207492589950562,
      "learning_rate": 1.1089691201750174e-06,
      "loss": 0.04162168502807617,
      "step": 6550
    },
    {
      "epoch": 0.8856442754541614,
      "grad_norm": 0.2518909275531769,
      "learning_rate": 1.106387787790239e-06,
      "loss": 0.053722500801086426,
      "step": 6551
    },
    {
      "epoch": 0.8857794676806083,
      "grad_norm": 0.43330979347229004,
      "learning_rate": 1.1038093481296091e-06,
      "loss": 0.09711885452270508,
      "step": 6552
    },
    {
      "epoch": 0.8859146599070553,
      "grad_norm": 0.32390764355659485,
      "learning_rate": 1.10123380172998e-06,
      "loss": 0.06100893020629883,
      "step": 6553
    },
    {
      "epoch": 0.8860498521335023,
      "grad_norm": 2.1229729652404785,
      "learning_rate": 1.098661149127586e-06,
      "loss": 0.04858589172363281,
      "step": 6554
    },
    {
      "epoch": 0.8861850443599493,
      "grad_norm": 0.20098699629306793,
      "learning_rate": 1.0960913908580788e-06,
      "loss": 0.03773832321166992,
      "step": 6555
    },
    {
      "epoch": 0.8863202365863962,
      "grad_norm": 0.2583629786968231,
      "learning_rate": 1.0935245274564852e-06,
      "loss": 0.06451964378356934,
      "step": 6556
    },
    {
      "epoch": 0.8864554288128432,
      "grad_norm": 0.3037578761577606,
      "learning_rate": 1.0909605594572413e-06,
      "loss": 0.06290674209594727,
      "step": 6557
    },
    {
      "epoch": 0.8865906210392902,
      "grad_norm": 0.31126680970191956,
      "learning_rate": 1.0883994873941816e-06,
      "loss": 0.052382588386535645,
      "step": 6558
    },
    {
      "epoch": 0.8867258132657372,
      "grad_norm": 0.2369667887687683,
      "learning_rate": 1.0858413118005345e-06,
      "loss": 0.06488275527954102,
      "step": 6559
    },
    {
      "epoch": 0.8868610054921842,
      "grad_norm": 0.1917002946138382,
      "learning_rate": 1.0832860332089288e-06,
      "loss": 0.03648841381072998,
      "step": 6560
    },
    {
      "epoch": 0.8869961977186311,
      "grad_norm": 0.3454590141773224,
      "learning_rate": 1.0807336521513828e-06,
      "loss": 0.0711512565612793,
      "step": 6561
    },
    {
      "epoch": 0.8871313899450781,
      "grad_norm": 0.24861523509025574,
      "learning_rate": 1.0781841691593142e-06,
      "loss": 0.05977523326873779,
      "step": 6562
    },
    {
      "epoch": 0.8872665821715251,
      "grad_norm": 0.4173514246940613,
      "learning_rate": 1.0756375847635435e-06,
      "loss": 0.06826353073120117,
      "step": 6563
    },
    {
      "epoch": 0.8874017743979721,
      "grad_norm": 0.20597350597381592,
      "learning_rate": 1.0730938994942818e-06,
      "loss": 0.03624802827835083,
      "step": 6564
    },
    {
      "epoch": 0.887536966624419,
      "grad_norm": 0.21884912252426147,
      "learning_rate": 1.070553113881137e-06,
      "loss": 0.04353678226470947,
      "step": 6565
    },
    {
      "epoch": 0.887672158850866,
      "grad_norm": 0.6119565963745117,
      "learning_rate": 1.0680152284531158e-06,
      "loss": 0.06782770156860352,
      "step": 6566
    },
    {
      "epoch": 0.887807351077313,
      "grad_norm": 0.5683345794677734,
      "learning_rate": 1.0654802437386157e-06,
      "loss": 0.08216142654418945,
      "step": 6567
    },
    {
      "epoch": 0.88794254330376,
      "grad_norm": 0.290649950504303,
      "learning_rate": 1.062948160265438e-06,
      "loss": 0.062177419662475586,
      "step": 6568
    },
    {
      "epoch": 0.888077735530207,
      "grad_norm": 0.3959812521934509,
      "learning_rate": 1.0604189785607772e-06,
      "loss": 0.0941762924194336,
      "step": 6569
    },
    {
      "epoch": 0.888212927756654,
      "grad_norm": 0.3331816792488098,
      "learning_rate": 1.0578926991512171e-06,
      "loss": 0.0622258186340332,
      "step": 6570
    },
    {
      "epoch": 0.8883481199831009,
      "grad_norm": 0.25859034061431885,
      "learning_rate": 1.0553693225627458e-06,
      "loss": 0.05439615249633789,
      "step": 6571
    },
    {
      "epoch": 0.8884833122095479,
      "grad_norm": 0.33288052678108215,
      "learning_rate": 1.0528488493207444e-06,
      "loss": 0.07359981536865234,
      "step": 6572
    },
    {
      "epoch": 0.8886185044359949,
      "grad_norm": 0.31173184514045715,
      "learning_rate": 1.0503312799499898e-06,
      "loss": 0.057944297790527344,
      "step": 6573
    },
    {
      "epoch": 0.8887536966624419,
      "grad_norm": 0.20959161221981049,
      "learning_rate": 1.0478166149746476e-06,
      "loss": 0.050884902477264404,
      "step": 6574
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 0.4633617401123047,
      "learning_rate": 1.0453048549182892e-06,
      "loss": 0.07828986644744873,
      "step": 6575
    },
    {
      "epoch": 0.8890240811153358,
      "grad_norm": 0.347409725189209,
      "learning_rate": 1.0427960003038744e-06,
      "loss": 0.048555612564086914,
      "step": 6576
    },
    {
      "epoch": 0.8891592733417828,
      "grad_norm": 0.22929781675338745,
      "learning_rate": 1.040290051653764e-06,
      "loss": 0.03709220886230469,
      "step": 6577
    },
    {
      "epoch": 0.8892944655682298,
      "grad_norm": 0.3411514163017273,
      "learning_rate": 1.0377870094897085e-06,
      "loss": 0.07570827007293701,
      "step": 6578
    },
    {
      "epoch": 0.8894296577946768,
      "grad_norm": 0.32179558277130127,
      "learning_rate": 1.0352868743328497e-06,
      "loss": 0.052057743072509766,
      "step": 6579
    },
    {
      "epoch": 0.8895648500211237,
      "grad_norm": 0.36982619762420654,
      "learning_rate": 1.032789646703733e-06,
      "loss": 0.05653858184814453,
      "step": 6580
    },
    {
      "epoch": 0.8897000422475707,
      "grad_norm": 0.23097890615463257,
      "learning_rate": 1.0302953271222938e-06,
      "loss": 0.051781654357910156,
      "step": 6581
    },
    {
      "epoch": 0.8898352344740177,
      "grad_norm": 0.5853762626647949,
      "learning_rate": 1.0278039161078634e-06,
      "loss": 0.09038478136062622,
      "step": 6582
    },
    {
      "epoch": 0.8899704267004648,
      "grad_norm": 0.2296319156885147,
      "learning_rate": 1.0253154141791705e-06,
      "loss": 0.05182623863220215,
      "step": 6583
    },
    {
      "epoch": 0.8901056189269118,
      "grad_norm": 0.1901443749666214,
      "learning_rate": 1.0228298218543253e-06,
      "loss": 0.03893446922302246,
      "step": 6584
    },
    {
      "epoch": 0.8902408111533587,
      "grad_norm": 0.1707037389278412,
      "learning_rate": 1.020347139650849e-06,
      "loss": 0.04085683822631836,
      "step": 6585
    },
    {
      "epoch": 0.8903760033798057,
      "grad_norm": 0.29270249605178833,
      "learning_rate": 1.0178673680856448e-06,
      "loss": 0.07581907510757446,
      "step": 6586
    },
    {
      "epoch": 0.8905111956062527,
      "grad_norm": 0.3180233836174011,
      "learning_rate": 1.0153905076750196e-06,
      "loss": 0.05968618392944336,
      "step": 6587
    },
    {
      "epoch": 0.8906463878326997,
      "grad_norm": 0.2173004299402237,
      "learning_rate": 1.0129165589346644e-06,
      "loss": 0.0568547248840332,
      "step": 6588
    },
    {
      "epoch": 0.8907815800591466,
      "grad_norm": 0.3191970884799957,
      "learning_rate": 1.0104455223796688e-06,
      "loss": 0.050228118896484375,
      "step": 6589
    },
    {
      "epoch": 0.8909167722855936,
      "grad_norm": 0.31993502378463745,
      "learning_rate": 1.0079773985245178e-06,
      "loss": 0.06128692626953125,
      "step": 6590
    },
    {
      "epoch": 0.8910519645120406,
      "grad_norm": 0.35776931047439575,
      "learning_rate": 1.0055121878830837e-06,
      "loss": 0.08691418170928955,
      "step": 6591
    },
    {
      "epoch": 0.8911871567384876,
      "grad_norm": 0.35664451122283936,
      "learning_rate": 1.0030498909686458e-06,
      "loss": 0.07600116729736328,
      "step": 6592
    },
    {
      "epoch": 0.8913223489649346,
      "grad_norm": 0.21843941509723663,
      "learning_rate": 1.0005905082938593e-06,
      "loss": 0.06284713745117188,
      "step": 6593
    },
    {
      "epoch": 0.8914575411913815,
      "grad_norm": 0.2739088237285614,
      "learning_rate": 9.981340403707794e-07,
      "loss": 0.058686017990112305,
      "step": 6594
    },
    {
      "epoch": 0.8915927334178285,
      "grad_norm": 0.1872408241033554,
      "learning_rate": 9.956804877108638e-07,
      "loss": 0.029108881950378418,
      "step": 6595
    },
    {
      "epoch": 0.8917279256442755,
      "grad_norm": 0.2940690815448761,
      "learning_rate": 9.932298508249488e-07,
      "loss": 0.06449627876281738,
      "step": 6596
    },
    {
      "epoch": 0.8918631178707225,
      "grad_norm": 0.35881543159484863,
      "learning_rate": 9.907821302232729e-07,
      "loss": 0.06825017929077148,
      "step": 6597
    },
    {
      "epoch": 0.8919983100971695,
      "grad_norm": 0.19285565614700317,
      "learning_rate": 9.883373264154633e-07,
      "loss": 0.048618316650390625,
      "step": 6598
    },
    {
      "epoch": 0.8921335023236164,
      "grad_norm": 0.23325026035308838,
      "learning_rate": 9.858954399105397e-07,
      "loss": 0.06113481521606445,
      "step": 6599
    },
    {
      "epoch": 0.8922686945500634,
      "grad_norm": 0.5605015158653259,
      "learning_rate": 9.834564712169202e-07,
      "loss": 0.08972692489624023,
      "step": 6600
    },
    {
      "epoch": 0.8924038867765104,
      "grad_norm": 0.34347522258758545,
      "learning_rate": 9.81020420842409e-07,
      "loss": 0.07176446914672852,
      "step": 6601
    },
    {
      "epoch": 0.8925390790029574,
      "grad_norm": 0.27367064356803894,
      "learning_rate": 9.785872892942033e-07,
      "loss": 0.07134056091308594,
      "step": 6602
    },
    {
      "epoch": 0.8926742712294043,
      "grad_norm": 0.25994497537612915,
      "learning_rate": 9.761570770788964e-07,
      "loss": 0.06648492813110352,
      "step": 6603
    },
    {
      "epoch": 0.8928094634558513,
      "grad_norm": 0.5097453594207764,
      "learning_rate": 9.737297847024685e-07,
      "loss": 0.07680559158325195,
      "step": 6604
    },
    {
      "epoch": 0.8929446556822983,
      "grad_norm": 0.288347989320755,
      "learning_rate": 9.713054126702968e-07,
      "loss": 0.06078195571899414,
      "step": 6605
    },
    {
      "epoch": 0.8930798479087453,
      "grad_norm": 0.2569486200809479,
      "learning_rate": 9.688839614871497e-07,
      "loss": 0.05570650100708008,
      "step": 6606
    },
    {
      "epoch": 0.8932150401351923,
      "grad_norm": 0.29454556107521057,
      "learning_rate": 9.664654316571852e-07,
      "loss": 0.07751941680908203,
      "step": 6607
    },
    {
      "epoch": 0.8933502323616392,
      "grad_norm": 0.39530429244041443,
      "learning_rate": 9.640498236839507e-07,
      "loss": 0.08114385604858398,
      "step": 6608
    },
    {
      "epoch": 0.8934854245880862,
      "grad_norm": 0.2667555510997772,
      "learning_rate": 9.616371380703953e-07,
      "loss": 0.0582427978515625,
      "step": 6609
    },
    {
      "epoch": 0.8936206168145332,
      "grad_norm": 0.17970451712608337,
      "learning_rate": 9.592273753188507e-07,
      "loss": 0.03761601448059082,
      "step": 6610
    },
    {
      "epoch": 0.8937558090409802,
      "grad_norm": 0.31466150283813477,
      "learning_rate": 9.568205359310372e-07,
      "loss": 0.06345224380493164,
      "step": 6611
    },
    {
      "epoch": 0.8938910012674272,
      "grad_norm": 0.5389094352722168,
      "learning_rate": 9.544166204080772e-07,
      "loss": 0.07818794250488281,
      "step": 6612
    },
    {
      "epoch": 0.8940261934938741,
      "grad_norm": 0.23919075727462769,
      "learning_rate": 9.520156292504739e-07,
      "loss": 0.04798460006713867,
      "step": 6613
    },
    {
      "epoch": 0.8941613857203211,
      "grad_norm": 0.46265608072280884,
      "learning_rate": 9.496175629581322e-07,
      "loss": 0.07718026638031006,
      "step": 6614
    },
    {
      "epoch": 0.8942965779467681,
      "grad_norm": 0.39650270342826843,
      "learning_rate": 9.472224220303427e-07,
      "loss": 0.0720372200012207,
      "step": 6615
    },
    {
      "epoch": 0.8944317701732151,
      "grad_norm": 0.1982034146785736,
      "learning_rate": 9.448302069657799e-07,
      "loss": 0.04893851280212402,
      "step": 6616
    },
    {
      "epoch": 0.894566962399662,
      "grad_norm": 0.35795077681541443,
      "learning_rate": 9.424409182625205e-07,
      "loss": 0.056674957275390625,
      "step": 6617
    },
    {
      "epoch": 0.894702154626109,
      "grad_norm": 0.3393133878707886,
      "learning_rate": 9.40054556418023e-07,
      "loss": 0.08192062377929688,
      "step": 6618
    },
    {
      "epoch": 0.894837346852556,
      "grad_norm": 0.4634474217891693,
      "learning_rate": 9.376711219291483e-07,
      "loss": 0.08518218994140625,
      "step": 6619
    },
    {
      "epoch": 0.894972539079003,
      "grad_norm": 0.1800144761800766,
      "learning_rate": 9.352906152921348e-07,
      "loss": 0.049013614654541016,
      "step": 6620
    },
    {
      "epoch": 0.89510773130545,
      "grad_norm": 0.2422923594713211,
      "learning_rate": 9.32913037002614e-07,
      "loss": 0.07240152359008789,
      "step": 6621
    },
    {
      "epoch": 0.8952429235318969,
      "grad_norm": 0.2702312171459198,
      "learning_rate": 9.30538387555615e-07,
      "loss": 0.03886592388153076,
      "step": 6622
    },
    {
      "epoch": 0.8953781157583439,
      "grad_norm": 0.27723395824432373,
      "learning_rate": 9.281666674455508e-07,
      "loss": 0.07527542114257812,
      "step": 6623
    },
    {
      "epoch": 0.8955133079847909,
      "grad_norm": 0.23852206766605377,
      "learning_rate": 9.257978771662295e-07,
      "loss": 0.0500791072845459,
      "step": 6624
    },
    {
      "epoch": 0.8956485002112379,
      "grad_norm": 0.296090692281723,
      "learning_rate": 9.234320172108418e-07,
      "loss": 0.07099008560180664,
      "step": 6625
    },
    {
      "epoch": 0.8957836924376849,
      "grad_norm": 0.2960122227668762,
      "learning_rate": 9.210690880719719e-07,
      "loss": 0.07332897186279297,
      "step": 6626
    },
    {
      "epoch": 0.8959188846641318,
      "grad_norm": 0.33863869309425354,
      "learning_rate": 9.187090902415962e-07,
      "loss": 0.07859325408935547,
      "step": 6627
    },
    {
      "epoch": 0.8960540768905788,
      "grad_norm": 0.3466472029685974,
      "learning_rate": 9.163520242110784e-07,
      "loss": 0.06715989112854004,
      "step": 6628
    },
    {
      "epoch": 0.8961892691170258,
      "grad_norm": 0.32666298747062683,
      "learning_rate": 9.13997890471176e-07,
      "loss": 0.060756683349609375,
      "step": 6629
    },
    {
      "epoch": 0.8963244613434728,
      "grad_norm": 0.26205384731292725,
      "learning_rate": 9.116466895120251e-07,
      "loss": 0.05854201316833496,
      "step": 6630
    },
    {
      "epoch": 0.8964596535699197,
      "grad_norm": 0.2584305703639984,
      "learning_rate": 9.092984218231609e-07,
      "loss": 0.05685114860534668,
      "step": 6631
    },
    {
      "epoch": 0.8965948457963667,
      "grad_norm": 0.607467770576477,
      "learning_rate": 9.069530878935072e-07,
      "loss": 0.08487892150878906,
      "step": 6632
    },
    {
      "epoch": 0.8967300380228137,
      "grad_norm": 0.3411499857902527,
      "learning_rate": 9.046106882113753e-07,
      "loss": 0.059922218322753906,
      "step": 6633
    },
    {
      "epoch": 0.8968652302492607,
      "grad_norm": 0.21270091831684113,
      "learning_rate": 9.022712232644631e-07,
      "loss": 0.05106949806213379,
      "step": 6634
    },
    {
      "epoch": 0.8970004224757077,
      "grad_norm": 0.32793837785720825,
      "learning_rate": 8.999346935398611e-07,
      "loss": 0.048998117446899414,
      "step": 6635
    },
    {
      "epoch": 0.8971356147021546,
      "grad_norm": 0.43046560883522034,
      "learning_rate": 8.976010995240436e-07,
      "loss": 0.0726022720336914,
      "step": 6636
    },
    {
      "epoch": 0.8972708069286016,
      "grad_norm": 0.4204343259334564,
      "learning_rate": 8.952704417028818e-07,
      "loss": 0.05670487880706787,
      "step": 6637
    },
    {
      "epoch": 0.8974059991550486,
      "grad_norm": 0.17617012560367584,
      "learning_rate": 8.929427205616308e-07,
      "loss": 0.0431976318359375,
      "step": 6638
    },
    {
      "epoch": 0.8975411913814956,
      "grad_norm": 0.5180122256278992,
      "learning_rate": 8.906179365849332e-07,
      "loss": 0.09676170349121094,
      "step": 6639
    },
    {
      "epoch": 0.8976763836079426,
      "grad_norm": 0.34688785672187805,
      "learning_rate": 8.882960902568216e-07,
      "loss": 0.0789722204208374,
      "step": 6640
    },
    {
      "epoch": 0.8978115758343895,
      "grad_norm": 0.5823962092399597,
      "learning_rate": 8.85977182060716e-07,
      "loss": 0.08455848693847656,
      "step": 6641
    },
    {
      "epoch": 0.8979467680608365,
      "grad_norm": 0.31170040369033813,
      "learning_rate": 8.836612124794285e-07,
      "loss": 0.06884980201721191,
      "step": 6642
    },
    {
      "epoch": 0.8980819602872835,
      "grad_norm": 0.1792592853307724,
      "learning_rate": 8.813481819951502e-07,
      "loss": 0.0627431869506836,
      "step": 6643
    },
    {
      "epoch": 0.8982171525137305,
      "grad_norm": 0.29300495982170105,
      "learning_rate": 8.790380910894724e-07,
      "loss": 0.05310249328613281,
      "step": 6644
    },
    {
      "epoch": 0.8983523447401774,
      "grad_norm": 0.38170912861824036,
      "learning_rate": 8.767309402433671e-07,
      "loss": 0.06711244583129883,
      "step": 6645
    },
    {
      "epoch": 0.8984875369666244,
      "grad_norm": 0.19290301203727722,
      "learning_rate": 8.744267299371917e-07,
      "loss": 0.0653371810913086,
      "step": 6646
    },
    {
      "epoch": 0.8986227291930714,
      "grad_norm": 0.5893896818161011,
      "learning_rate": 8.721254606507023e-07,
      "loss": 0.10550880432128906,
      "step": 6647
    },
    {
      "epoch": 0.8987579214195184,
      "grad_norm": 0.3628849685192108,
      "learning_rate": 8.698271328630275e-07,
      "loss": 0.05511760711669922,
      "step": 6648
    },
    {
      "epoch": 0.8988931136459654,
      "grad_norm": 0.38034719228744507,
      "learning_rate": 8.675317470526961e-07,
      "loss": 0.05885505676269531,
      "step": 6649
    },
    {
      "epoch": 0.8990283058724123,
      "grad_norm": 0.5596439838409424,
      "learning_rate": 8.652393036976159e-07,
      "loss": 0.09777212142944336,
      "step": 6650
    },
    {
      "epoch": 0.8991634980988593,
      "grad_norm": 0.2142232209444046,
      "learning_rate": 8.629498032750916e-07,
      "loss": 0.04836559295654297,
      "step": 6651
    },
    {
      "epoch": 0.8992986903253063,
      "grad_norm": 0.17117846012115479,
      "learning_rate": 8.606632462618069e-07,
      "loss": 0.04703378677368164,
      "step": 6652
    },
    {
      "epoch": 0.8994338825517533,
      "grad_norm": 0.29976606369018555,
      "learning_rate": 8.583796331338311e-07,
      "loss": 0.062484025955200195,
      "step": 6653
    },
    {
      "epoch": 0.8995690747782003,
      "grad_norm": 0.6134013533592224,
      "learning_rate": 8.560989643666306e-07,
      "loss": 0.09510993957519531,
      "step": 6654
    },
    {
      "epoch": 0.8997042670046472,
      "grad_norm": 0.39132779836654663,
      "learning_rate": 8.538212404350471e-07,
      "loss": 0.06367635726928711,
      "step": 6655
    },
    {
      "epoch": 0.8998394592310942,
      "grad_norm": 0.25858405232429504,
      "learning_rate": 8.515464618133228e-07,
      "loss": 0.0593571662902832,
      "step": 6656
    },
    {
      "epoch": 0.8999746514575412,
      "grad_norm": 0.2673133909702301,
      "learning_rate": 8.492746289750725e-07,
      "loss": 0.06683647632598877,
      "step": 6657
    },
    {
      "epoch": 0.9001098436839882,
      "grad_norm": 0.1790904849767685,
      "learning_rate": 8.470057423933026e-07,
      "loss": 0.044456422328948975,
      "step": 6658
    },
    {
      "epoch": 0.9002450359104351,
      "grad_norm": 0.39251089096069336,
      "learning_rate": 8.447398025404118e-07,
      "loss": 0.06880021095275879,
      "step": 6659
    },
    {
      "epoch": 0.9003802281368821,
      "grad_norm": 0.2714976668357849,
      "learning_rate": 8.42476809888178e-07,
      "loss": 0.053466737270355225,
      "step": 6660
    },
    {
      "epoch": 0.9005154203633291,
      "grad_norm": 0.3734161853790283,
      "learning_rate": 8.402167649077725e-07,
      "loss": 0.07591962814331055,
      "step": 6661
    },
    {
      "epoch": 0.9006506125897761,
      "grad_norm": 0.27058759331703186,
      "learning_rate": 8.379596680697454e-07,
      "loss": 0.07022619247436523,
      "step": 6662
    },
    {
      "epoch": 0.9007858048162231,
      "grad_norm": 0.474793016910553,
      "learning_rate": 8.357055198440328e-07,
      "loss": 0.0609898567199707,
      "step": 6663
    },
    {
      "epoch": 0.90092099704267,
      "grad_norm": 0.5272277593612671,
      "learning_rate": 8.334543206999673e-07,
      "loss": 0.10007524490356445,
      "step": 6664
    },
    {
      "epoch": 0.901056189269117,
      "grad_norm": 0.5144278407096863,
      "learning_rate": 8.312060711062558e-07,
      "loss": 0.1001276969909668,
      "step": 6665
    },
    {
      "epoch": 0.901191381495564,
      "grad_norm": 0.4902416169643402,
      "learning_rate": 8.289607715309988e-07,
      "loss": 0.09573745727539062,
      "step": 6666
    },
    {
      "epoch": 0.901326573722011,
      "grad_norm": 0.22319334745407104,
      "learning_rate": 8.267184224416791e-07,
      "loss": 0.04765605926513672,
      "step": 6667
    },
    {
      "epoch": 0.901461765948458,
      "grad_norm": 0.3537706434726715,
      "learning_rate": 8.244790243051614e-07,
      "loss": 0.06658744812011719,
      "step": 6668
    },
    {
      "epoch": 0.9015969581749049,
      "grad_norm": 0.35011163353919983,
      "learning_rate": 8.222425775877079e-07,
      "loss": 0.08805465698242188,
      "step": 6669
    },
    {
      "epoch": 0.9017321504013519,
      "grad_norm": 0.44920170307159424,
      "learning_rate": 8.200090827549527e-07,
      "loss": 0.06653451919555664,
      "step": 6670
    },
    {
      "epoch": 0.9018673426277989,
      "grad_norm": 0.3928013741970062,
      "learning_rate": 8.17778540271924e-07,
      "loss": 0.05397987365722656,
      "step": 6671
    },
    {
      "epoch": 0.9020025348542459,
      "grad_norm": 0.24984316527843475,
      "learning_rate": 8.155509506030334e-07,
      "loss": 0.054976463317871094,
      "step": 6672
    },
    {
      "epoch": 0.9021377270806928,
      "grad_norm": 0.18410629034042358,
      "learning_rate": 8.133263142120717e-07,
      "loss": 0.030864715576171875,
      "step": 6673
    },
    {
      "epoch": 0.9022729193071398,
      "grad_norm": 0.37082603573799133,
      "learning_rate": 8.111046315622284e-07,
      "loss": 0.07468032836914062,
      "step": 6674
    },
    {
      "epoch": 0.9024081115335868,
      "grad_norm": 0.26319846510887146,
      "learning_rate": 8.088859031160633e-07,
      "loss": 0.04930877685546875,
      "step": 6675
    },
    {
      "epoch": 0.9025433037600338,
      "grad_norm": 0.3214379847049713,
      "learning_rate": 8.066701293355288e-07,
      "loss": 0.056827664375305176,
      "step": 6676
    },
    {
      "epoch": 0.9026784959864808,
      "grad_norm": 0.2739381194114685,
      "learning_rate": 8.044573106819625e-07,
      "loss": 0.07780933380126953,
      "step": 6677
    },
    {
      "epoch": 0.9028136882129277,
      "grad_norm": 0.27685046195983887,
      "learning_rate": 8.022474476160824e-07,
      "loss": 0.04792189598083496,
      "step": 6678
    },
    {
      "epoch": 0.9029488804393747,
      "grad_norm": 0.2327430546283722,
      "learning_rate": 8.000405405979988e-07,
      "loss": 0.043967247009277344,
      "step": 6679
    },
    {
      "epoch": 0.9030840726658217,
      "grad_norm": 0.4293173849582672,
      "learning_rate": 7.978365900871943e-07,
      "loss": 0.07202434539794922,
      "step": 6680
    },
    {
      "epoch": 0.9032192648922687,
      "grad_norm": 0.3255581855773926,
      "learning_rate": 7.956355965425482e-07,
      "loss": 0.04250144958496094,
      "step": 6681
    },
    {
      "epoch": 0.9033544571187156,
      "grad_norm": 0.41620928049087524,
      "learning_rate": 7.934375604223193e-07,
      "loss": 0.09491443634033203,
      "step": 6682
    },
    {
      "epoch": 0.9034896493451626,
      "grad_norm": 0.2380094975233078,
      "learning_rate": 7.912424821841463e-07,
      "loss": 0.04782736301422119,
      "step": 6683
    },
    {
      "epoch": 0.9036248415716096,
      "grad_norm": 0.1913856416940689,
      "learning_rate": 7.89050362285062e-07,
      "loss": 0.04988408088684082,
      "step": 6684
    },
    {
      "epoch": 0.9037600337980566,
      "grad_norm": 0.22582587599754333,
      "learning_rate": 7.868612011814713e-07,
      "loss": 0.047875404357910156,
      "step": 6685
    },
    {
      "epoch": 0.9038952260245036,
      "grad_norm": 0.28033584356307983,
      "learning_rate": 7.846749993291746e-07,
      "loss": 0.05279040336608887,
      "step": 6686
    },
    {
      "epoch": 0.9040304182509505,
      "grad_norm": 0.31647419929504395,
      "learning_rate": 7.824917571833445e-07,
      "loss": 0.05295419692993164,
      "step": 6687
    },
    {
      "epoch": 0.9041656104773975,
      "grad_norm": 0.3373316526412964,
      "learning_rate": 7.80311475198554e-07,
      "loss": 0.04968404769897461,
      "step": 6688
    },
    {
      "epoch": 0.9043008027038445,
      "grad_norm": 0.22313278913497925,
      "learning_rate": 7.781341538287384e-07,
      "loss": 0.05129897594451904,
      "step": 6689
    },
    {
      "epoch": 0.9044359949302915,
      "grad_norm": 0.4062902331352234,
      "learning_rate": 7.759597935272316e-07,
      "loss": 0.06038779020309448,
      "step": 6690
    },
    {
      "epoch": 0.9045711871567385,
      "grad_norm": 0.4131912887096405,
      "learning_rate": 7.7378839474675e-07,
      "loss": 0.05027008056640625,
      "step": 6691
    },
    {
      "epoch": 0.9047063793831854,
      "grad_norm": 0.3348398804664612,
      "learning_rate": 7.716199579393851e-07,
      "loss": 0.06008744239807129,
      "step": 6692
    },
    {
      "epoch": 0.9048415716096324,
      "grad_norm": 0.2865051329135895,
      "learning_rate": 7.694544835566259e-07,
      "loss": 0.05042457580566406,
      "step": 6693
    },
    {
      "epoch": 0.9049767638360794,
      "grad_norm": 0.2408238798379898,
      "learning_rate": 7.672919720493249e-07,
      "loss": 0.06216144561767578,
      "step": 6694
    },
    {
      "epoch": 0.9051119560625264,
      "grad_norm": 0.2508491575717926,
      "learning_rate": 7.651324238677338e-07,
      "loss": 0.054028868675231934,
      "step": 6695
    },
    {
      "epoch": 0.9052471482889733,
      "grad_norm": 0.2306889146566391,
      "learning_rate": 7.629758394614828e-07,
      "loss": 0.05232954025268555,
      "step": 6696
    },
    {
      "epoch": 0.9053823405154203,
      "grad_norm": 0.2982030510902405,
      "learning_rate": 7.608222192795794e-07,
      "loss": 0.05411696434020996,
      "step": 6697
    },
    {
      "epoch": 0.9055175327418673,
      "grad_norm": 0.27433010935783386,
      "learning_rate": 7.586715637704284e-07,
      "loss": 0.05402827262878418,
      "step": 6698
    },
    {
      "epoch": 0.9056527249683143,
      "grad_norm": 0.36263376474380493,
      "learning_rate": 7.565238733817998e-07,
      "loss": 0.061466217041015625,
      "step": 6699
    },
    {
      "epoch": 0.9057879171947613,
      "grad_norm": 0.24579842388629913,
      "learning_rate": 7.543791485608542e-07,
      "loss": 0.07651901245117188,
      "step": 6700
    },
    {
      "epoch": 0.9059231094212082,
      "grad_norm": 0.2787382900714874,
      "learning_rate": 7.52237389754138e-07,
      "loss": 0.07133591175079346,
      "step": 6701
    },
    {
      "epoch": 0.9060583016476552,
      "grad_norm": 0.21613316237926483,
      "learning_rate": 7.500985974075758e-07,
      "loss": 0.04439842700958252,
      "step": 6702
    },
    {
      "epoch": 0.9061934938741022,
      "grad_norm": 0.2324300855398178,
      "learning_rate": 7.479627719664767e-07,
      "loss": 0.0670309066772461,
      "step": 6703
    },
    {
      "epoch": 0.9063286861005492,
      "grad_norm": 0.21317701041698456,
      "learning_rate": 7.458299138755281e-07,
      "loss": 0.05440521240234375,
      "step": 6704
    },
    {
      "epoch": 0.9064638783269962,
      "grad_norm": 0.2382166087627411,
      "learning_rate": 7.437000235788033e-07,
      "loss": 0.049439430236816406,
      "step": 6705
    },
    {
      "epoch": 0.9065990705534431,
      "grad_norm": 0.35315731167793274,
      "learning_rate": 7.415731015197575e-07,
      "loss": 0.06222355365753174,
      "step": 6706
    },
    {
      "epoch": 0.9067342627798901,
      "grad_norm": 0.4295409619808197,
      "learning_rate": 7.39449148141228e-07,
      "loss": 0.046143293380737305,
      "step": 6707
    },
    {
      "epoch": 0.9068694550063371,
      "grad_norm": 0.24091030657291412,
      "learning_rate": 7.373281638854329e-07,
      "loss": 0.0615234375,
      "step": 6708
    },
    {
      "epoch": 0.9070046472327841,
      "grad_norm": 0.29950758814811707,
      "learning_rate": 7.352101491939722e-07,
      "loss": 0.06075394153594971,
      "step": 6709
    },
    {
      "epoch": 0.907139839459231,
      "grad_norm": 0.2839188575744629,
      "learning_rate": 7.330951045078249e-07,
      "loss": 0.05700480937957764,
      "step": 6710
    },
    {
      "epoch": 0.907275031685678,
      "grad_norm": 0.4149356782436371,
      "learning_rate": 7.309830302673621e-07,
      "loss": 0.06901121139526367,
      "step": 6711
    },
    {
      "epoch": 0.907410223912125,
      "grad_norm": 0.17082051932811737,
      "learning_rate": 7.288739269123184e-07,
      "loss": 0.046177804470062256,
      "step": 6712
    },
    {
      "epoch": 0.907545416138572,
      "grad_norm": 0.4979169964790344,
      "learning_rate": 7.267677948818296e-07,
      "loss": 0.06458663940429688,
      "step": 6713
    },
    {
      "epoch": 0.907680608365019,
      "grad_norm": 0.17504222691059113,
      "learning_rate": 7.246646346143997e-07,
      "loss": 0.033037662506103516,
      "step": 6714
    },
    {
      "epoch": 0.9078158005914659,
      "grad_norm": 0.23155361413955688,
      "learning_rate": 7.225644465479153e-07,
      "loss": 0.03666114807128906,
      "step": 6715
    },
    {
      "epoch": 0.9079509928179129,
      "grad_norm": 0.4524311125278473,
      "learning_rate": 7.204672311196547e-07,
      "loss": 0.06278514862060547,
      "step": 6716
    },
    {
      "epoch": 0.9080861850443599,
      "grad_norm": 0.1646474003791809,
      "learning_rate": 7.183729887662604e-07,
      "loss": 0.033927321434020996,
      "step": 6717
    },
    {
      "epoch": 0.9082213772708069,
      "grad_norm": 0.2800999879837036,
      "learning_rate": 7.162817199237703e-07,
      "loss": 0.06309652328491211,
      "step": 6718
    },
    {
      "epoch": 0.908356569497254,
      "grad_norm": 0.2834169268608093,
      "learning_rate": 7.141934250275978e-07,
      "loss": 0.06281876564025879,
      "step": 6719
    },
    {
      "epoch": 0.9084917617237009,
      "grad_norm": 0.3598640263080597,
      "learning_rate": 7.121081045125316e-07,
      "loss": 0.08004379272460938,
      "step": 6720
    },
    {
      "epoch": 0.9086269539501479,
      "grad_norm": 0.38457831740379333,
      "learning_rate": 7.100257588127545e-07,
      "loss": 0.0518949031829834,
      "step": 6721
    },
    {
      "epoch": 0.9087621461765949,
      "grad_norm": 0.6134139895439148,
      "learning_rate": 7.079463883618148e-07,
      "loss": 0.0683935284614563,
      "step": 6722
    },
    {
      "epoch": 0.9088973384030419,
      "grad_norm": 0.2467769980430603,
      "learning_rate": 7.058699935926527e-07,
      "loss": 0.05888652801513672,
      "step": 6723
    },
    {
      "epoch": 0.9090325306294889,
      "grad_norm": 0.2695505917072296,
      "learning_rate": 7.037965749375808e-07,
      "loss": 0.05156731605529785,
      "step": 6724
    },
    {
      "epoch": 0.9091677228559358,
      "grad_norm": 0.21168826520442963,
      "learning_rate": 7.017261328283037e-07,
      "loss": 0.0583263635635376,
      "step": 6725
    },
    {
      "epoch": 0.9093029150823828,
      "grad_norm": 0.2085266411304474,
      "learning_rate": 6.996586676958916e-07,
      "loss": 0.056255340576171875,
      "step": 6726
    },
    {
      "epoch": 0.9094381073088298,
      "grad_norm": 0.3592725396156311,
      "learning_rate": 6.975941799708019e-07,
      "loss": 0.06098222732543945,
      "step": 6727
    },
    {
      "epoch": 0.9095732995352768,
      "grad_norm": 0.1824830025434494,
      "learning_rate": 6.955326700828757e-07,
      "loss": 0.05001598596572876,
      "step": 6728
    },
    {
      "epoch": 0.9097084917617237,
      "grad_norm": 0.4089336097240448,
      "learning_rate": 6.934741384613246e-07,
      "loss": 0.053939104080200195,
      "step": 6729
    },
    {
      "epoch": 0.9098436839881707,
      "grad_norm": 0.2246934175491333,
      "learning_rate": 6.91418585534756e-07,
      "loss": 0.06101179122924805,
      "step": 6730
    },
    {
      "epoch": 0.9099788762146177,
      "grad_norm": 0.41654840111732483,
      "learning_rate": 6.893660117311373e-07,
      "loss": 0.08330154418945312,
      "step": 6731
    },
    {
      "epoch": 0.9101140684410647,
      "grad_norm": 0.41293761134147644,
      "learning_rate": 6.873164174778252e-07,
      "loss": 0.06404829025268555,
      "step": 6732
    },
    {
      "epoch": 0.9102492606675117,
      "grad_norm": 0.30583998560905457,
      "learning_rate": 6.852698032015631e-07,
      "loss": 0.048659563064575195,
      "step": 6733
    },
    {
      "epoch": 0.9103844528939586,
      "grad_norm": 0.24714858829975128,
      "learning_rate": 6.832261693284636e-07,
      "loss": 0.039984941482543945,
      "step": 6734
    },
    {
      "epoch": 0.9105196451204056,
      "grad_norm": 0.31316837668418884,
      "learning_rate": 6.811855162840214e-07,
      "loss": 0.061365604400634766,
      "step": 6735
    },
    {
      "epoch": 0.9106548373468526,
      "grad_norm": 0.399715781211853,
      "learning_rate": 6.791478444931132e-07,
      "loss": 0.07841825485229492,
      "step": 6736
    },
    {
      "epoch": 0.9107900295732996,
      "grad_norm": 0.302395761013031,
      "learning_rate": 6.77113154379988e-07,
      "loss": 0.0564122200012207,
      "step": 6737
    },
    {
      "epoch": 0.9109252217997466,
      "grad_norm": 0.38264378905296326,
      "learning_rate": 6.75081446368287e-07,
      "loss": 0.07637691497802734,
      "step": 6738
    },
    {
      "epoch": 0.9110604140261935,
      "grad_norm": 0.35266685485839844,
      "learning_rate": 6.730527208810166e-07,
      "loss": 0.06119513511657715,
      "step": 6739
    },
    {
      "epoch": 0.9111956062526405,
      "grad_norm": 0.25630560517311096,
      "learning_rate": 6.710269783405709e-07,
      "loss": 0.06500899791717529,
      "step": 6740
    },
    {
      "epoch": 0.9113307984790875,
      "grad_norm": 0.318916380405426,
      "learning_rate": 6.690042191687206e-07,
      "loss": 0.061501502990722656,
      "step": 6741
    },
    {
      "epoch": 0.9114659907055345,
      "grad_norm": 0.29961079359054565,
      "learning_rate": 6.669844437866124e-07,
      "loss": 0.060685157775878906,
      "step": 6742
    },
    {
      "epoch": 0.9116011829319814,
      "grad_norm": 0.2267502248287201,
      "learning_rate": 6.649676526147764e-07,
      "loss": 0.05499839782714844,
      "step": 6743
    },
    {
      "epoch": 0.9117363751584284,
      "grad_norm": 0.2665746212005615,
      "learning_rate": 6.629538460731199e-07,
      "loss": 0.07109498977661133,
      "step": 6744
    },
    {
      "epoch": 0.9118715673848754,
      "grad_norm": 0.18866756558418274,
      "learning_rate": 6.609430245809261e-07,
      "loss": 0.05088448524475098,
      "step": 6745
    },
    {
      "epoch": 0.9120067596113224,
      "grad_norm": 0.30700844526290894,
      "learning_rate": 6.589351885568617e-07,
      "loss": 0.055596351623535156,
      "step": 6746
    },
    {
      "epoch": 0.9121419518377694,
      "grad_norm": 0.314453661441803,
      "learning_rate": 6.569303384189624e-07,
      "loss": 0.08609580993652344,
      "step": 6747
    },
    {
      "epoch": 0.9122771440642163,
      "grad_norm": 0.23913970589637756,
      "learning_rate": 6.54928474584659e-07,
      "loss": 0.07298660278320312,
      "step": 6748
    },
    {
      "epoch": 0.9124123362906633,
      "grad_norm": 0.36009082198143005,
      "learning_rate": 6.5292959747074e-07,
      "loss": 0.09228992462158203,
      "step": 6749
    },
    {
      "epoch": 0.9125475285171103,
      "grad_norm": 0.28268963098526,
      "learning_rate": 6.509337074933891e-07,
      "loss": 0.07158422470092773,
      "step": 6750
    },
    {
      "epoch": 0.9126827207435573,
      "grad_norm": 0.22034013271331787,
      "learning_rate": 6.489408050681589e-07,
      "loss": 0.0398256778717041,
      "step": 6751
    },
    {
      "epoch": 0.9128179129700043,
      "grad_norm": 0.2606247067451477,
      "learning_rate": 6.469508906099792e-07,
      "loss": 0.05727100372314453,
      "step": 6752
    },
    {
      "epoch": 0.9129531051964512,
      "grad_norm": 0.25070828199386597,
      "learning_rate": 6.449639645331684e-07,
      "loss": 0.05663037300109863,
      "step": 6753
    },
    {
      "epoch": 0.9130882974228982,
      "grad_norm": 0.5489130020141602,
      "learning_rate": 6.429800272514058e-07,
      "loss": 0.08507394790649414,
      "step": 6754
    },
    {
      "epoch": 0.9132234896493452,
      "grad_norm": 0.19063083827495575,
      "learning_rate": 6.409990791777659e-07,
      "loss": 0.03476130962371826,
      "step": 6755
    },
    {
      "epoch": 0.9133586818757922,
      "grad_norm": 0.27534112334251404,
      "learning_rate": 6.390211207246888e-07,
      "loss": 0.06719374656677246,
      "step": 6756
    },
    {
      "epoch": 0.9134938741022391,
      "grad_norm": 0.2823287844657898,
      "learning_rate": 6.370461523039967e-07,
      "loss": 0.07176494598388672,
      "step": 6757
    },
    {
      "epoch": 0.9136290663286861,
      "grad_norm": 0.20179909467697144,
      "learning_rate": 6.350741743268873e-07,
      "loss": 0.05219745635986328,
      "step": 6758
    },
    {
      "epoch": 0.9137642585551331,
      "grad_norm": 0.33456456661224365,
      "learning_rate": 6.331051872039373e-07,
      "loss": 0.06334924697875977,
      "step": 6759
    },
    {
      "epoch": 0.9138994507815801,
      "grad_norm": 0.32684987783432007,
      "learning_rate": 6.31139191345102e-07,
      "loss": 0.03999185562133789,
      "step": 6760
    },
    {
      "epoch": 0.9140346430080271,
      "grad_norm": 0.3114374577999115,
      "learning_rate": 6.291761871597091e-07,
      "loss": 0.07558441162109375,
      "step": 6761
    },
    {
      "epoch": 0.914169835234474,
      "grad_norm": 0.33083027601242065,
      "learning_rate": 6.272161750564731e-07,
      "loss": 0.05785322189331055,
      "step": 6762
    },
    {
      "epoch": 0.914305027460921,
      "grad_norm": 0.48567214608192444,
      "learning_rate": 6.252591554434728e-07,
      "loss": 0.06440496444702148,
      "step": 6763
    },
    {
      "epoch": 0.914440219687368,
      "grad_norm": 0.3678141236305237,
      "learning_rate": 6.233051287281688e-07,
      "loss": 0.08142662048339844,
      "step": 6764
    },
    {
      "epoch": 0.914575411913815,
      "grad_norm": 0.7030377388000488,
      "learning_rate": 6.213540953174057e-07,
      "loss": 0.07967948913574219,
      "step": 6765
    },
    {
      "epoch": 0.914710604140262,
      "grad_norm": 0.4859228730201721,
      "learning_rate": 6.194060556173953e-07,
      "loss": 0.06463241577148438,
      "step": 6766
    },
    {
      "epoch": 0.9148457963667089,
      "grad_norm": 0.2780775725841522,
      "learning_rate": 6.17461010033733e-07,
      "loss": 0.048812687397003174,
      "step": 6767
    },
    {
      "epoch": 0.9149809885931559,
      "grad_norm": 0.22061297297477722,
      "learning_rate": 6.155189589713833e-07,
      "loss": 0.047280073165893555,
      "step": 6768
    },
    {
      "epoch": 0.9151161808196029,
      "grad_norm": 0.2835690379142761,
      "learning_rate": 6.135799028346928e-07,
      "loss": 0.0580897331237793,
      "step": 6769
    },
    {
      "epoch": 0.9152513730460499,
      "grad_norm": 0.28901752829551697,
      "learning_rate": 6.116438420273868e-07,
      "loss": 0.06691884994506836,
      "step": 6770
    },
    {
      "epoch": 0.9153865652724968,
      "grad_norm": 0.3729507625102997,
      "learning_rate": 6.097107769525595e-07,
      "loss": 0.049791932106018066,
      "step": 6771
    },
    {
      "epoch": 0.9155217574989438,
      "grad_norm": 0.3505827486515045,
      "learning_rate": 6.077807080126873e-07,
      "loss": 0.0691225528717041,
      "step": 6772
    },
    {
      "epoch": 0.9156569497253908,
      "grad_norm": 0.2565617561340332,
      "learning_rate": 6.058536356096206e-07,
      "loss": 0.0696115493774414,
      "step": 6773
    },
    {
      "epoch": 0.9157921419518378,
      "grad_norm": 0.4454917013645172,
      "learning_rate": 6.039295601445833e-07,
      "loss": 0.07009410858154297,
      "step": 6774
    },
    {
      "epoch": 0.9159273341782848,
      "grad_norm": 0.34154823422431946,
      "learning_rate": 6.020084820181831e-07,
      "loss": 0.06250905990600586,
      "step": 6775
    },
    {
      "epoch": 0.9160625264047317,
      "grad_norm": 0.20340220630168915,
      "learning_rate": 6.000904016303971e-07,
      "loss": 0.04442262649536133,
      "step": 6776
    },
    {
      "epoch": 0.9161977186311787,
      "grad_norm": 0.2908305823802948,
      "learning_rate": 5.981753193805789e-07,
      "loss": 0.05586540699005127,
      "step": 6777
    },
    {
      "epoch": 0.9163329108576257,
      "grad_norm": 0.22283963859081268,
      "learning_rate": 5.962632356674597e-07,
      "loss": 0.04918527603149414,
      "step": 6778
    },
    {
      "epoch": 0.9164681030840727,
      "grad_norm": 0.2516134977340698,
      "learning_rate": 5.94354150889141e-07,
      "loss": 0.05009293556213379,
      "step": 6779
    },
    {
      "epoch": 0.9166032953105197,
      "grad_norm": 0.23964817821979523,
      "learning_rate": 5.924480654431147e-07,
      "loss": 0.04578065872192383,
      "step": 6780
    },
    {
      "epoch": 0.9167384875369666,
      "grad_norm": 0.42407140135765076,
      "learning_rate": 5.905449797262252e-07,
      "loss": 0.06870126724243164,
      "step": 6781
    },
    {
      "epoch": 0.9168736797634136,
      "grad_norm": 0.27598291635513306,
      "learning_rate": 5.886448941347156e-07,
      "loss": 0.04606819152832031,
      "step": 6782
    },
    {
      "epoch": 0.9170088719898606,
      "grad_norm": 0.3802034556865692,
      "learning_rate": 5.867478090641892e-07,
      "loss": 0.0780797004699707,
      "step": 6783
    },
    {
      "epoch": 0.9171440642163076,
      "grad_norm": 0.23938871920108795,
      "learning_rate": 5.848537249096269e-07,
      "loss": 0.05454683303833008,
      "step": 6784
    },
    {
      "epoch": 0.9172792564427545,
      "grad_norm": 0.33651578426361084,
      "learning_rate": 5.829626420653949e-07,
      "loss": 0.07274627685546875,
      "step": 6785
    },
    {
      "epoch": 0.9174144486692015,
      "grad_norm": 0.3467169404029846,
      "learning_rate": 5.810745609252166e-07,
      "loss": 0.0642080307006836,
      "step": 6786
    },
    {
      "epoch": 0.9175496408956485,
      "grad_norm": 0.25199273228645325,
      "learning_rate": 5.791894818822091e-07,
      "loss": 0.05392313003540039,
      "step": 6787
    },
    {
      "epoch": 0.9176848331220955,
      "grad_norm": 0.3076941967010498,
      "learning_rate": 5.773074053288519e-07,
      "loss": 0.06535136699676514,
      "step": 6788
    },
    {
      "epoch": 0.9178200253485425,
      "grad_norm": 0.40538471937179565,
      "learning_rate": 5.75428331657003e-07,
      "loss": 0.05321621894836426,
      "step": 6789
    },
    {
      "epoch": 0.9179552175749894,
      "grad_norm": 0.16008977591991425,
      "learning_rate": 5.735522612578998e-07,
      "loss": 0.031260907649993896,
      "step": 6790
    },
    {
      "epoch": 0.9180904098014364,
      "grad_norm": 0.3699582815170288,
      "learning_rate": 5.716791945221444e-07,
      "loss": 0.06197977066040039,
      "step": 6791
    },
    {
      "epoch": 0.9182256020278834,
      "grad_norm": 0.20649270713329315,
      "learning_rate": 5.698091318397219e-07,
      "loss": 0.047204017639160156,
      "step": 6792
    },
    {
      "epoch": 0.9183607942543304,
      "grad_norm": 0.3541818857192993,
      "learning_rate": 5.679420735999908e-07,
      "loss": 0.0786905288696289,
      "step": 6793
    },
    {
      "epoch": 0.9184959864807773,
      "grad_norm": 0.18114563822746277,
      "learning_rate": 5.660780201916799e-07,
      "loss": 0.038895606994628906,
      "step": 6794
    },
    {
      "epoch": 0.9186311787072243,
      "grad_norm": 0.3154070973396301,
      "learning_rate": 5.642169720028973e-07,
      "loss": 0.08837461471557617,
      "step": 6795
    },
    {
      "epoch": 0.9187663709336713,
      "grad_norm": 0.20202623307704926,
      "learning_rate": 5.623589294211196e-07,
      "loss": 0.04513823986053467,
      "step": 6796
    },
    {
      "epoch": 0.9189015631601183,
      "grad_norm": 0.5871425271034241,
      "learning_rate": 5.605038928332057e-07,
      "loss": 0.09560298919677734,
      "step": 6797
    },
    {
      "epoch": 0.9190367553865653,
      "grad_norm": 0.47953614592552185,
      "learning_rate": 5.586518626253817e-07,
      "loss": 0.06708073616027832,
      "step": 6798
    },
    {
      "epoch": 0.9191719476130122,
      "grad_norm": 0.25075042247772217,
      "learning_rate": 5.568028391832524e-07,
      "loss": 0.053061485290527344,
      "step": 6799
    },
    {
      "epoch": 0.9193071398394592,
      "grad_norm": 0.33691227436065674,
      "learning_rate": 5.549568228917917e-07,
      "loss": 0.06907331943511963,
      "step": 6800
    },
    {
      "epoch": 0.9194423320659062,
      "grad_norm": 0.5234549045562744,
      "learning_rate": 5.531138141353486e-07,
      "loss": 0.06320500373840332,
      "step": 6801
    },
    {
      "epoch": 0.9195775242923532,
      "grad_norm": 0.35388708114624023,
      "learning_rate": 5.512738132976514e-07,
      "loss": 0.06179189682006836,
      "step": 6802
    },
    {
      "epoch": 0.9197127165188002,
      "grad_norm": 0.2902572453022003,
      "learning_rate": 5.49436820761795e-07,
      "loss": 0.05215871334075928,
      "step": 6803
    },
    {
      "epoch": 0.9198479087452471,
      "grad_norm": 0.3174911141395569,
      "learning_rate": 5.476028369102537e-07,
      "loss": 0.05806374549865723,
      "step": 6804
    },
    {
      "epoch": 0.9199831009716941,
      "grad_norm": 0.18217256665229797,
      "learning_rate": 5.45771862124872e-07,
      "loss": 0.03731536865234375,
      "step": 6805
    },
    {
      "epoch": 0.9201182931981411,
      "grad_norm": 0.43400830030441284,
      "learning_rate": 5.439438967868649e-07,
      "loss": 0.08292078971862793,
      "step": 6806
    },
    {
      "epoch": 0.9202534854245881,
      "grad_norm": 0.3415291905403137,
      "learning_rate": 5.421189412768296e-07,
      "loss": 0.05956315994262695,
      "step": 6807
    },
    {
      "epoch": 0.920388677651035,
      "grad_norm": 0.2812170386314392,
      "learning_rate": 5.402969959747306e-07,
      "loss": 0.06839466094970703,
      "step": 6808
    },
    {
      "epoch": 0.920523869877482,
      "grad_norm": 0.2639061212539673,
      "learning_rate": 5.384780612599044e-07,
      "loss": 0.06360101699829102,
      "step": 6809
    },
    {
      "epoch": 0.920659062103929,
      "grad_norm": 0.2789705693721771,
      "learning_rate": 5.366621375110647e-07,
      "loss": 0.04472571611404419,
      "step": 6810
    },
    {
      "epoch": 0.920794254330376,
      "grad_norm": 0.21159721910953522,
      "learning_rate": 5.348492251062942e-07,
      "loss": 0.05944252014160156,
      "step": 6811
    },
    {
      "epoch": 0.920929446556823,
      "grad_norm": 0.306159108877182,
      "learning_rate": 5.330393244230558e-07,
      "loss": 0.06237149238586426,
      "step": 6812
    },
    {
      "epoch": 0.9210646387832699,
      "grad_norm": 0.4733816981315613,
      "learning_rate": 5.312324358381731e-07,
      "loss": 0.09130477905273438,
      "step": 6813
    },
    {
      "epoch": 0.9211998310097169,
      "grad_norm": 0.2981599271297455,
      "learning_rate": 5.29428559727857e-07,
      "loss": 0.06065654754638672,
      "step": 6814
    },
    {
      "epoch": 0.9213350232361639,
      "grad_norm": 0.3937837779521942,
      "learning_rate": 5.276276964676802e-07,
      "loss": 0.07400846481323242,
      "step": 6815
    },
    {
      "epoch": 0.9214702154626109,
      "grad_norm": 0.389931857585907,
      "learning_rate": 5.258298464325928e-07,
      "loss": 0.0682833194732666,
      "step": 6816
    },
    {
      "epoch": 0.9216054076890579,
      "grad_norm": 0.2545563578605652,
      "learning_rate": 5.240350099969204e-07,
      "loss": 0.06184971332550049,
      "step": 6817
    },
    {
      "epoch": 0.9217405999155048,
      "grad_norm": 0.2999067008495331,
      "learning_rate": 5.222431875343492e-07,
      "loss": 0.061218976974487305,
      "step": 6818
    },
    {
      "epoch": 0.9218757921419518,
      "grad_norm": 0.18707145750522614,
      "learning_rate": 5.204543794179539e-07,
      "loss": 0.04723691940307617,
      "step": 6819
    },
    {
      "epoch": 0.9220109843683988,
      "grad_norm": 0.2815069854259491,
      "learning_rate": 5.186685860201717e-07,
      "loss": 0.05034446716308594,
      "step": 6820
    },
    {
      "epoch": 0.9221461765948458,
      "grad_norm": 0.23036843538284302,
      "learning_rate": 5.16885807712812e-07,
      "loss": 0.04822182655334473,
      "step": 6821
    },
    {
      "epoch": 0.9222813688212927,
      "grad_norm": 0.11852409690618515,
      "learning_rate": 5.151060448670625e-07,
      "loss": 0.03097057342529297,
      "step": 6822
    },
    {
      "epoch": 0.9224165610477397,
      "grad_norm": 0.257041335105896,
      "learning_rate": 5.133292978534754e-07,
      "loss": 0.05121743679046631,
      "step": 6823
    },
    {
      "epoch": 0.9225517532741867,
      "grad_norm": 0.478397935628891,
      "learning_rate": 5.115555670419814e-07,
      "loss": 0.06816601753234863,
      "step": 6824
    },
    {
      "epoch": 0.9226869455006337,
      "grad_norm": 0.43804439902305603,
      "learning_rate": 5.097848528018817e-07,
      "loss": 0.06466591358184814,
      "step": 6825
    },
    {
      "epoch": 0.9228221377270807,
      "grad_norm": 0.2722177803516388,
      "learning_rate": 5.080171555018448e-07,
      "loss": 0.06040811538696289,
      "step": 6826
    },
    {
      "epoch": 0.9229573299535276,
      "grad_norm": 0.39246806502342224,
      "learning_rate": 5.06252475509918e-07,
      "loss": 0.0710756778717041,
      "step": 6827
    },
    {
      "epoch": 0.9230925221799746,
      "grad_norm": 0.34272870421409607,
      "learning_rate": 5.044908131935139e-07,
      "loss": 0.09210443496704102,
      "step": 6828
    },
    {
      "epoch": 0.9232277144064216,
      "grad_norm": 0.4193495213985443,
      "learning_rate": 5.027321689194242e-07,
      "loss": 0.06343269348144531,
      "step": 6829
    },
    {
      "epoch": 0.9233629066328686,
      "grad_norm": 0.3049297034740448,
      "learning_rate": 5.009765430538061e-07,
      "loss": 0.06045010685920715,
      "step": 6830
    },
    {
      "epoch": 0.9234980988593156,
      "grad_norm": 0.29091471433639526,
      "learning_rate": 4.992239359621886e-07,
      "loss": 0.05863833427429199,
      "step": 6831
    },
    {
      "epoch": 0.9236332910857625,
      "grad_norm": 0.1579255610704422,
      "learning_rate": 4.974743480094767e-07,
      "loss": 0.04657268524169922,
      "step": 6832
    },
    {
      "epoch": 0.9237684833122095,
      "grad_norm": 0.3201614320278168,
      "learning_rate": 4.957277795599407e-07,
      "loss": 0.04777860641479492,
      "step": 6833
    },
    {
      "epoch": 0.9239036755386565,
      "grad_norm": 0.17296139895915985,
      "learning_rate": 4.93984230977228e-07,
      "loss": 0.03850138187408447,
      "step": 6834
    },
    {
      "epoch": 0.9240388677651035,
      "grad_norm": 0.24672360718250275,
      "learning_rate": 4.922437026243531e-07,
      "loss": 0.06394791603088379,
      "step": 6835
    },
    {
      "epoch": 0.9241740599915504,
      "grad_norm": 0.3869103193283081,
      "learning_rate": 4.905061948637063e-07,
      "loss": 0.09241247177124023,
      "step": 6836
    },
    {
      "epoch": 0.9243092522179974,
      "grad_norm": 0.196123406291008,
      "learning_rate": 4.887717080570431e-07,
      "loss": 0.04959821701049805,
      "step": 6837
    },
    {
      "epoch": 0.9244444444444444,
      "grad_norm": 0.29419219493865967,
      "learning_rate": 4.870402425654913e-07,
      "loss": 0.05415534973144531,
      "step": 6838
    },
    {
      "epoch": 0.9245796366708914,
      "grad_norm": 0.251487135887146,
      "learning_rate": 4.853117987495542e-07,
      "loss": 0.06594061851501465,
      "step": 6839
    },
    {
      "epoch": 0.9247148288973384,
      "grad_norm": 0.32523199915885925,
      "learning_rate": 4.83586376969104e-07,
      "loss": 0.07776308059692383,
      "step": 6840
    },
    {
      "epoch": 0.9248500211237853,
      "grad_norm": 0.30944687128067017,
      "learning_rate": 4.818639775833816e-07,
      "loss": 0.07084798812866211,
      "step": 6841
    },
    {
      "epoch": 0.9249852133502323,
      "grad_norm": 0.25586116313934326,
      "learning_rate": 4.801446009509969e-07,
      "loss": 0.05023503303527832,
      "step": 6842
    },
    {
      "epoch": 0.9251204055766793,
      "grad_norm": 0.3534484803676605,
      "learning_rate": 4.784282474299367e-07,
      "loss": 0.06169772148132324,
      "step": 6843
    },
    {
      "epoch": 0.9252555978031263,
      "grad_norm": 0.25079667568206787,
      "learning_rate": 4.767149173775537e-07,
      "loss": 0.05755043029785156,
      "step": 6844
    },
    {
      "epoch": 0.9253907900295733,
      "grad_norm": 0.45221132040023804,
      "learning_rate": 4.750046111505724e-07,
      "loss": 0.07118082046508789,
      "step": 6845
    },
    {
      "epoch": 0.9255259822560202,
      "grad_norm": 0.3971695601940155,
      "learning_rate": 4.732973291050896e-07,
      "loss": 0.0812540054321289,
      "step": 6846
    },
    {
      "epoch": 0.9256611744824672,
      "grad_norm": 0.2767832279205322,
      "learning_rate": 4.7159307159656607e-07,
      "loss": 0.05943197011947632,
      "step": 6847
    },
    {
      "epoch": 0.9257963667089142,
      "grad_norm": 0.1832389086484909,
      "learning_rate": 4.6989183897983954e-07,
      "loss": 0.05013716220855713,
      "step": 6848
    },
    {
      "epoch": 0.9259315589353612,
      "grad_norm": 0.2647983431816101,
      "learning_rate": 4.681936316091201e-07,
      "loss": 0.054838478565216064,
      "step": 6849
    },
    {
      "epoch": 0.9260667511618081,
      "grad_norm": 0.26054203510284424,
      "learning_rate": 4.664984498379765e-07,
      "loss": 0.05982637405395508,
      "step": 6850
    },
    {
      "epoch": 0.9262019433882551,
      "grad_norm": 0.3227074146270752,
      "learning_rate": 4.6480629401935814e-07,
      "loss": 0.05984306335449219,
      "step": 6851
    },
    {
      "epoch": 0.9263371356147021,
      "grad_norm": 0.30183130502700806,
      "learning_rate": 4.631171645055815e-07,
      "loss": 0.07209181785583496,
      "step": 6852
    },
    {
      "epoch": 0.9264723278411491,
      "grad_norm": 0.2954559624195099,
      "learning_rate": 4.614310616483286e-07,
      "loss": 0.04852783679962158,
      "step": 6853
    },
    {
      "epoch": 0.9266075200675962,
      "grad_norm": 0.23906128108501434,
      "learning_rate": 4.5974798579866193e-07,
      "loss": 0.06581377983093262,
      "step": 6854
    },
    {
      "epoch": 0.9267427122940431,
      "grad_norm": 0.4055463969707489,
      "learning_rate": 4.580679373069996e-07,
      "loss": 0.08595585823059082,
      "step": 6855
    },
    {
      "epoch": 0.9268779045204901,
      "grad_norm": 0.5272423624992371,
      "learning_rate": 4.5639091652314e-07,
      "loss": 0.09612560272216797,
      "step": 6856
    },
    {
      "epoch": 0.9270130967469371,
      "grad_norm": 0.28300583362579346,
      "learning_rate": 4.54716923796249e-07,
      "loss": 0.06563735008239746,
      "step": 6857
    },
    {
      "epoch": 0.9271482889733841,
      "grad_norm": 0.2970777750015259,
      "learning_rate": 4.5304595947485927e-07,
      "loss": 0.04386281967163086,
      "step": 6858
    },
    {
      "epoch": 0.9272834811998311,
      "grad_norm": 0.3516584038734436,
      "learning_rate": 4.5137802390687433e-07,
      "loss": 0.05901622772216797,
      "step": 6859
    },
    {
      "epoch": 0.927418673426278,
      "grad_norm": 0.23541265726089478,
      "learning_rate": 4.497131174395663e-07,
      "loss": 0.056168556213378906,
      "step": 6860
    },
    {
      "epoch": 0.927553865652725,
      "grad_norm": 0.29414865374565125,
      "learning_rate": 4.4805124041957967e-07,
      "loss": 0.05954241752624512,
      "step": 6861
    },
    {
      "epoch": 0.927689057879172,
      "grad_norm": 0.36588066816329956,
      "learning_rate": 4.463923931929259e-07,
      "loss": 0.0796670913696289,
      "step": 6862
    },
    {
      "epoch": 0.927824250105619,
      "grad_norm": 0.2690092623233795,
      "learning_rate": 4.4473657610498377e-07,
      "loss": 0.05303668975830078,
      "step": 6863
    },
    {
      "epoch": 0.927959442332066,
      "grad_norm": 0.36585530638694763,
      "learning_rate": 4.430837895005058e-07,
      "loss": 0.05746293067932129,
      "step": 6864
    },
    {
      "epoch": 0.9280946345585129,
      "grad_norm": 0.20604927837848663,
      "learning_rate": 4.4143403372360836e-07,
      "loss": 0.039200663566589355,
      "step": 6865
    },
    {
      "epoch": 0.9282298267849599,
      "grad_norm": 0.3721497356891632,
      "learning_rate": 4.3978730911778176e-07,
      "loss": 0.07439613342285156,
      "step": 6866
    },
    {
      "epoch": 0.9283650190114069,
      "grad_norm": 0.2101733386516571,
      "learning_rate": 4.381436160258834e-07,
      "loss": 0.058396339416503906,
      "step": 6867
    },
    {
      "epoch": 0.9285002112378539,
      "grad_norm": 0.43266987800598145,
      "learning_rate": 4.3650295479013615e-07,
      "loss": 0.09219121932983398,
      "step": 6868
    },
    {
      "epoch": 0.9286354034643008,
      "grad_norm": 0.4713047444820404,
      "learning_rate": 4.348653257521351e-07,
      "loss": 0.07306671142578125,
      "step": 6869
    },
    {
      "epoch": 0.9287705956907478,
      "grad_norm": 0.24050207436084747,
      "learning_rate": 4.332307292528442e-07,
      "loss": 0.058725953102111816,
      "step": 6870
    },
    {
      "epoch": 0.9289057879171948,
      "grad_norm": 0.4709692597389221,
      "learning_rate": 4.315991656325946e-07,
      "loss": 0.0814213752746582,
      "step": 6871
    },
    {
      "epoch": 0.9290409801436418,
      "grad_norm": 0.22877316176891327,
      "learning_rate": 4.299706352310895e-07,
      "loss": 0.058759212493896484,
      "step": 6872
    },
    {
      "epoch": 0.9291761723700888,
      "grad_norm": 0.4701555371284485,
      "learning_rate": 4.283451383873926e-07,
      "loss": 0.08330810070037842,
      "step": 6873
    },
    {
      "epoch": 0.9293113645965357,
      "grad_norm": 0.35321491956710815,
      "learning_rate": 4.26722675439945e-07,
      "loss": 0.05710732936859131,
      "step": 6874
    },
    {
      "epoch": 0.9294465568229827,
      "grad_norm": 0.2343061864376068,
      "learning_rate": 4.251032467265481e-07,
      "loss": 0.06146240234375,
      "step": 6875
    },
    {
      "epoch": 0.9295817490494297,
      "grad_norm": 0.3130585849285126,
      "learning_rate": 4.234868525843805e-07,
      "loss": 0.06228923797607422,
      "step": 6876
    },
    {
      "epoch": 0.9297169412758767,
      "grad_norm": 0.44318705797195435,
      "learning_rate": 4.218734933499796e-07,
      "loss": 0.08170604705810547,
      "step": 6877
    },
    {
      "epoch": 0.9298521335023237,
      "grad_norm": 0.16497454047203064,
      "learning_rate": 4.202631693592601e-07,
      "loss": 0.036772727966308594,
      "step": 6878
    },
    {
      "epoch": 0.9299873257287706,
      "grad_norm": 0.2680743932723999,
      "learning_rate": 4.186558809474955e-07,
      "loss": 0.07556962966918945,
      "step": 6879
    },
    {
      "epoch": 0.9301225179552176,
      "grad_norm": 0.26797208189964294,
      "learning_rate": 4.170516284493331e-07,
      "loss": 0.05643129348754883,
      "step": 6880
    },
    {
      "epoch": 0.9302577101816646,
      "grad_norm": 0.2170102447271347,
      "learning_rate": 4.1545041219879063e-07,
      "loss": 0.049879372119903564,
      "step": 6881
    },
    {
      "epoch": 0.9303929024081116,
      "grad_norm": 0.3908720314502716,
      "learning_rate": 4.138522325292432e-07,
      "loss": 0.05023479461669922,
      "step": 6882
    },
    {
      "epoch": 0.9305280946345585,
      "grad_norm": 0.3174516558647156,
      "learning_rate": 4.1225708977344457e-07,
      "loss": 0.061882972717285156,
      "step": 6883
    },
    {
      "epoch": 0.9306632868610055,
      "grad_norm": 0.26105132699012756,
      "learning_rate": 4.106649842635124e-07,
      "loss": 0.06191539764404297,
      "step": 6884
    },
    {
      "epoch": 0.9307984790874525,
      "grad_norm": 0.24962367117404938,
      "learning_rate": 4.090759163309282e-07,
      "loss": 0.07141685485839844,
      "step": 6885
    },
    {
      "epoch": 0.9309336713138995,
      "grad_norm": 0.31625574827194214,
      "learning_rate": 4.07489886306549e-07,
      "loss": 0.05345571041107178,
      "step": 6886
    },
    {
      "epoch": 0.9310688635403465,
      "grad_norm": 0.1707090139389038,
      "learning_rate": 4.059068945205907e-07,
      "loss": 0.038506269454956055,
      "step": 6887
    },
    {
      "epoch": 0.9312040557667934,
      "grad_norm": 0.17831982672214508,
      "learning_rate": 4.043269413026429e-07,
      "loss": 0.04871511459350586,
      "step": 6888
    },
    {
      "epoch": 0.9313392479932404,
      "grad_norm": 0.22578223049640656,
      "learning_rate": 4.027500269816592e-07,
      "loss": 0.0528789758682251,
      "step": 6889
    },
    {
      "epoch": 0.9314744402196874,
      "grad_norm": 0.20749430358409882,
      "learning_rate": 4.011761518859619e-07,
      "loss": 0.051172733306884766,
      "step": 6890
    },
    {
      "epoch": 0.9316096324461344,
      "grad_norm": 0.25292253494262695,
      "learning_rate": 3.996053163432406e-07,
      "loss": 0.05469846725463867,
      "step": 6891
    },
    {
      "epoch": 0.9317448246725814,
      "grad_norm": 0.22334545850753784,
      "learning_rate": 3.980375206805503e-07,
      "loss": 0.05817916989326477,
      "step": 6892
    },
    {
      "epoch": 0.9318800168990283,
      "grad_norm": 0.27013280987739563,
      "learning_rate": 3.9647276522431664e-07,
      "loss": 0.054511189460754395,
      "step": 6893
    },
    {
      "epoch": 0.9320152091254753,
      "grad_norm": 0.6775422692298889,
      "learning_rate": 3.949110503003289e-07,
      "loss": 0.08910083770751953,
      "step": 6894
    },
    {
      "epoch": 0.9321504013519223,
      "grad_norm": 0.21085061132907867,
      "learning_rate": 3.9335237623374377e-07,
      "loss": 0.05225324630737305,
      "step": 6895
    },
    {
      "epoch": 0.9322855935783693,
      "grad_norm": 0.4456910192966461,
      "learning_rate": 3.917967433490849e-07,
      "loss": 0.0707392692565918,
      "step": 6896
    },
    {
      "epoch": 0.9324207858048162,
      "grad_norm": 0.2432425171136856,
      "learning_rate": 3.902441519702449e-07,
      "loss": 0.06799256801605225,
      "step": 6897
    },
    {
      "epoch": 0.9325559780312632,
      "grad_norm": 0.3617327809333801,
      "learning_rate": 3.886946024204818e-07,
      "loss": 0.051616370677948,
      "step": 6898
    },
    {
      "epoch": 0.9326911702577102,
      "grad_norm": 0.9337500333786011,
      "learning_rate": 3.871480950224193e-07,
      "loss": 0.08326995372772217,
      "step": 6899
    },
    {
      "epoch": 0.9328263624841572,
      "grad_norm": 0.46419233083724976,
      "learning_rate": 3.856046300980498e-07,
      "loss": 0.06556767225265503,
      "step": 6900
    },
    {
      "epoch": 0.9329615547106042,
      "grad_norm": 0.17622384428977966,
      "learning_rate": 3.8406420796872953e-07,
      "loss": 0.0465550422668457,
      "step": 6901
    },
    {
      "epoch": 0.9330967469370511,
      "grad_norm": 0.36447086930274963,
      "learning_rate": 3.825268289551803e-07,
      "loss": 0.060393333435058594,
      "step": 6902
    },
    {
      "epoch": 0.9332319391634981,
      "grad_norm": 0.4901309609413147,
      "learning_rate": 3.8099249337749777e-07,
      "loss": 0.09293699264526367,
      "step": 6903
    },
    {
      "epoch": 0.9333671313899451,
      "grad_norm": 0.2701214849948883,
      "learning_rate": 3.7946120155513465e-07,
      "loss": 0.04080629348754883,
      "step": 6904
    },
    {
      "epoch": 0.9335023236163921,
      "grad_norm": 0.2428189516067505,
      "learning_rate": 3.7793295380691595e-07,
      "loss": 0.04715776443481445,
      "step": 6905
    },
    {
      "epoch": 0.933637515842839,
      "grad_norm": 0.4616331458091736,
      "learning_rate": 3.7640775045103214e-07,
      "loss": 0.07706832885742188,
      "step": 6906
    },
    {
      "epoch": 0.933772708069286,
      "grad_norm": 0.23217806220054626,
      "learning_rate": 3.7488559180503423e-07,
      "loss": 0.05143463611602783,
      "step": 6907
    },
    {
      "epoch": 0.933907900295733,
      "grad_norm": 0.33055245876312256,
      "learning_rate": 3.7336647818584866e-07,
      "loss": 0.049836158752441406,
      "step": 6908
    },
    {
      "epoch": 0.93404309252218,
      "grad_norm": 0.24888643622398376,
      "learning_rate": 3.718504099097625e-07,
      "loss": 0.03773140907287598,
      "step": 6909
    },
    {
      "epoch": 0.934178284748627,
      "grad_norm": 0.28889983892440796,
      "learning_rate": 3.703373872924265e-07,
      "loss": 0.05840158462524414,
      "step": 6910
    },
    {
      "epoch": 0.9343134769750739,
      "grad_norm": 0.29102545976638794,
      "learning_rate": 3.688274106488604e-07,
      "loss": 0.0663900375366211,
      "step": 6911
    },
    {
      "epoch": 0.9344486692015209,
      "grad_norm": 0.39317840337753296,
      "learning_rate": 3.67320480293451e-07,
      "loss": 0.0639181137084961,
      "step": 6912
    },
    {
      "epoch": 0.9345838614279679,
      "grad_norm": 0.35694122314453125,
      "learning_rate": 3.6581659653994736e-07,
      "loss": 0.06257116794586182,
      "step": 6913
    },
    {
      "epoch": 0.9347190536544149,
      "grad_norm": 0.42753785848617554,
      "learning_rate": 3.64315759701469e-07,
      "loss": 0.06165933609008789,
      "step": 6914
    },
    {
      "epoch": 0.9348542458808619,
      "grad_norm": 0.2149232029914856,
      "learning_rate": 3.6281797009049765e-07,
      "loss": 0.04876965284347534,
      "step": 6915
    },
    {
      "epoch": 0.9349894381073088,
      "grad_norm": 0.33355018496513367,
      "learning_rate": 3.613232280188772e-07,
      "loss": 0.07210326194763184,
      "step": 6916
    },
    {
      "epoch": 0.9351246303337558,
      "grad_norm": 0.24563607573509216,
      "learning_rate": 3.5983153379782363e-07,
      "loss": 0.04275798797607422,
      "step": 6917
    },
    {
      "epoch": 0.9352598225602028,
      "grad_norm": 0.3190612494945526,
      "learning_rate": 3.5834288773791854e-07,
      "loss": 0.055036067962646484,
      "step": 6918
    },
    {
      "epoch": 0.9353950147866498,
      "grad_norm": 0.6189736723899841,
      "learning_rate": 3.568572901491007e-07,
      "loss": 0.07856535911560059,
      "step": 6919
    },
    {
      "epoch": 0.9355302070130967,
      "grad_norm": 0.5020914673805237,
      "learning_rate": 3.553747413406827e-07,
      "loss": 0.07719254493713379,
      "step": 6920
    },
    {
      "epoch": 0.9356653992395437,
      "grad_norm": 0.3558565378189087,
      "learning_rate": 3.538952416213376e-07,
      "loss": 0.06122446060180664,
      "step": 6921
    },
    {
      "epoch": 0.9358005914659907,
      "grad_norm": 0.3230915665626526,
      "learning_rate": 3.524187912991056e-07,
      "loss": 0.07481193542480469,
      "step": 6922
    },
    {
      "epoch": 0.9359357836924377,
      "grad_norm": 0.22972339391708374,
      "learning_rate": 3.5094539068139254e-07,
      "loss": 0.04806637763977051,
      "step": 6923
    },
    {
      "epoch": 0.9360709759188847,
      "grad_norm": 0.24116207659244537,
      "learning_rate": 3.494750400749663e-07,
      "loss": 0.036308228969573975,
      "step": 6924
    },
    {
      "epoch": 0.9362061681453316,
      "grad_norm": 0.30507636070251465,
      "learning_rate": 3.480077397859638e-07,
      "loss": 0.05286884307861328,
      "step": 6925
    },
    {
      "epoch": 0.9363413603717786,
      "grad_norm": 0.2481544464826584,
      "learning_rate": 3.4654349011988384e-07,
      "loss": 0.06706428527832031,
      "step": 6926
    },
    {
      "epoch": 0.9364765525982256,
      "grad_norm": 0.5513602495193481,
      "learning_rate": 3.4508229138159095e-07,
      "loss": 0.07376748323440552,
      "step": 6927
    },
    {
      "epoch": 0.9366117448246726,
      "grad_norm": 0.27851566672325134,
      "learning_rate": 3.4362414387531516e-07,
      "loss": 0.04263782501220703,
      "step": 6928
    },
    {
      "epoch": 0.9367469370511196,
      "grad_norm": 0.3507442772388458,
      "learning_rate": 3.4216904790464854e-07,
      "loss": 0.0760190486907959,
      "step": 6929
    },
    {
      "epoch": 0.9368821292775665,
      "grad_norm": 0.38494110107421875,
      "learning_rate": 3.407170037725521e-07,
      "loss": 0.08167338371276855,
      "step": 6930
    },
    {
      "epoch": 0.9370173215040135,
      "grad_norm": 0.3074328303337097,
      "learning_rate": 3.3926801178134737e-07,
      "loss": 0.07122206687927246,
      "step": 6931
    },
    {
      "epoch": 0.9371525137304605,
      "grad_norm": 0.2962808310985565,
      "learning_rate": 3.3782207223272467e-07,
      "loss": 0.04630577564239502,
      "step": 6932
    },
    {
      "epoch": 0.9372877059569075,
      "grad_norm": 0.3505702018737793,
      "learning_rate": 3.363791854277348e-07,
      "loss": 0.06173086166381836,
      "step": 6933
    },
    {
      "epoch": 0.9374228981833544,
      "grad_norm": 0.4292900562286377,
      "learning_rate": 3.349393516667926e-07,
      "loss": 0.07289457321166992,
      "step": 6934
    },
    {
      "epoch": 0.9375580904098014,
      "grad_norm": 0.22617672383785248,
      "learning_rate": 3.335025712496814e-07,
      "loss": 0.051015377044677734,
      "step": 6935
    },
    {
      "epoch": 0.9376932826362484,
      "grad_norm": 0.3592321276664734,
      "learning_rate": 3.320688444755471e-07,
      "loss": 0.07427334785461426,
      "step": 6936
    },
    {
      "epoch": 0.9378284748626954,
      "grad_norm": 0.33925876021385193,
      "learning_rate": 3.306381716428991e-07,
      "loss": 0.08819043636322021,
      "step": 6937
    },
    {
      "epoch": 0.9379636670891424,
      "grad_norm": 0.3518395721912384,
      "learning_rate": 3.2921055304960925e-07,
      "loss": 0.07595157623291016,
      "step": 6938
    },
    {
      "epoch": 0.9380988593155893,
      "grad_norm": 0.32647407054901123,
      "learning_rate": 3.277859889929147e-07,
      "loss": 0.0714254379272461,
      "step": 6939
    },
    {
      "epoch": 0.9382340515420363,
      "grad_norm": 0.18644031882286072,
      "learning_rate": 3.263644797694215e-07,
      "loss": 0.04388880729675293,
      "step": 6940
    },
    {
      "epoch": 0.9383692437684833,
      "grad_norm": 0.14926472306251526,
      "learning_rate": 3.2494602567509303e-07,
      "loss": 0.035822153091430664,
      "step": 6941
    },
    {
      "epoch": 0.9385044359949303,
      "grad_norm": 0.3328128457069397,
      "learning_rate": 3.2353062700525794e-07,
      "loss": 0.08460092544555664,
      "step": 6942
    },
    {
      "epoch": 0.9386396282213773,
      "grad_norm": 0.22114552557468414,
      "learning_rate": 3.221182840546122e-07,
      "loss": 0.055501580238342285,
      "step": 6943
    },
    {
      "epoch": 0.9387748204478242,
      "grad_norm": 0.15361034870147705,
      "learning_rate": 3.207089971172089e-07,
      "loss": 0.040936946868896484,
      "step": 6944
    },
    {
      "epoch": 0.9389100126742712,
      "grad_norm": 0.2066817581653595,
      "learning_rate": 3.1930276648647504e-07,
      "loss": 0.05524873733520508,
      "step": 6945
    },
    {
      "epoch": 0.9390452049007182,
      "grad_norm": 0.26075878739356995,
      "learning_rate": 3.178995924551914e-07,
      "loss": 0.050370216369628906,
      "step": 6946
    },
    {
      "epoch": 0.9391803971271652,
      "grad_norm": 0.2528645098209381,
      "learning_rate": 3.164994753155059e-07,
      "loss": 0.05456829071044922,
      "step": 6947
    },
    {
      "epoch": 0.9393155893536121,
      "grad_norm": 0.29967832565307617,
      "learning_rate": 3.1510241535893215e-07,
      "loss": 0.07141304016113281,
      "step": 6948
    },
    {
      "epoch": 0.9394507815800591,
      "grad_norm": 0.28389590978622437,
      "learning_rate": 3.1370841287634567e-07,
      "loss": 0.05772805213928223,
      "step": 6949
    },
    {
      "epoch": 0.9395859738065061,
      "grad_norm": 0.32096779346466064,
      "learning_rate": 3.1231746815798436e-07,
      "loss": 0.06969261169433594,
      "step": 6950
    },
    {
      "epoch": 0.9397211660329531,
      "grad_norm": 0.20565739274024963,
      "learning_rate": 3.1092958149344985e-07,
      "loss": 0.04460740089416504,
      "step": 6951
    },
    {
      "epoch": 0.9398563582594001,
      "grad_norm": 0.18816794455051422,
      "learning_rate": 3.095447531717077e-07,
      "loss": 0.040885329246520996,
      "step": 6952
    },
    {
      "epoch": 0.939991550485847,
      "grad_norm": 0.524341881275177,
      "learning_rate": 3.08162983481089e-07,
      "loss": 0.08219385147094727,
      "step": 6953
    },
    {
      "epoch": 0.940126742712294,
      "grad_norm": 0.14859233796596527,
      "learning_rate": 3.067842727092801e-07,
      "loss": 0.03764939308166504,
      "step": 6954
    },
    {
      "epoch": 0.940261934938741,
      "grad_norm": 0.22179381549358368,
      "learning_rate": 3.0540862114334323e-07,
      "loss": 0.050228118896484375,
      "step": 6955
    },
    {
      "epoch": 0.940397127165188,
      "grad_norm": 0.2436661720275879,
      "learning_rate": 3.0403602906969086e-07,
      "loss": 0.04270893335342407,
      "step": 6956
    },
    {
      "epoch": 0.940532319391635,
      "grad_norm": 0.2125399112701416,
      "learning_rate": 3.0266649677410605e-07,
      "loss": 0.06122398376464844,
      "step": 6957
    },
    {
      "epoch": 0.9406675116180819,
      "grad_norm": 0.20838016271591187,
      "learning_rate": 3.0130002454173243e-07,
      "loss": 0.0460360050201416,
      "step": 6958
    },
    {
      "epoch": 0.9408027038445289,
      "grad_norm": 0.34980130195617676,
      "learning_rate": 2.9993661265707407e-07,
      "loss": 0.07094597816467285,
      "step": 6959
    },
    {
      "epoch": 0.9409378960709759,
      "grad_norm": 0.364028662443161,
      "learning_rate": 2.985762614040072e-07,
      "loss": 0.06134223937988281,
      "step": 6960
    },
    {
      "epoch": 0.9410730882974229,
      "grad_norm": 0.21083399653434753,
      "learning_rate": 2.972189710657586e-07,
      "loss": 0.04069042205810547,
      "step": 6961
    },
    {
      "epoch": 0.9412082805238698,
      "grad_norm": 0.16762679815292358,
      "learning_rate": 2.958647419249255e-07,
      "loss": 0.03345155715942383,
      "step": 6962
    },
    {
      "epoch": 0.9413434727503168,
      "grad_norm": 0.26082301139831543,
      "learning_rate": 2.9451357426346415e-07,
      "loss": 0.05316615104675293,
      "step": 6963
    },
    {
      "epoch": 0.9414786649767638,
      "grad_norm": 0.2976631224155426,
      "learning_rate": 2.9316546836269776e-07,
      "loss": 0.06584155559539795,
      "step": 6964
    },
    {
      "epoch": 0.9416138572032108,
      "grad_norm": 0.5661160349845886,
      "learning_rate": 2.9182042450330516e-07,
      "loss": 0.08023929595947266,
      "step": 6965
    },
    {
      "epoch": 0.9417490494296578,
      "grad_norm": 0.3103455603122711,
      "learning_rate": 2.9047844296533397e-07,
      "loss": 0.07018375396728516,
      "step": 6966
    },
    {
      "epoch": 0.9418842416561047,
      "grad_norm": 0.3112802803516388,
      "learning_rate": 2.8913952402819246e-07,
      "loss": 0.06319558620452881,
      "step": 6967
    },
    {
      "epoch": 0.9420194338825517,
      "grad_norm": 0.4048909842967987,
      "learning_rate": 2.878036679706492e-07,
      "loss": 0.08551788330078125,
      "step": 6968
    },
    {
      "epoch": 0.9421546261089987,
      "grad_norm": 0.37367865443229675,
      "learning_rate": 2.8647087507083837e-07,
      "loss": 0.0600590705871582,
      "step": 6969
    },
    {
      "epoch": 0.9422898183354457,
      "grad_norm": 0.567959725856781,
      "learning_rate": 2.8514114560625303e-07,
      "loss": 0.08487677574157715,
      "step": 6970
    },
    {
      "epoch": 0.9424250105618927,
      "grad_norm": 0.38909992575645447,
      "learning_rate": 2.8381447985375007e-07,
      "loss": 0.07983970642089844,
      "step": 6971
    },
    {
      "epoch": 0.9425602027883396,
      "grad_norm": 0.27797582745552063,
      "learning_rate": 2.8249087808954853e-07,
      "loss": 0.07405376434326172,
      "step": 6972
    },
    {
      "epoch": 0.9426953950147866,
      "grad_norm": 0.3441482186317444,
      "learning_rate": 2.811703405892296e-07,
      "loss": 0.08803129196166992,
      "step": 6973
    },
    {
      "epoch": 0.9428305872412336,
      "grad_norm": 0.21144941449165344,
      "learning_rate": 2.798528676277368e-07,
      "loss": 0.04326224327087402,
      "step": 6974
    },
    {
      "epoch": 0.9429657794676806,
      "grad_norm": 0.5733159184455872,
      "learning_rate": 2.785384594793738e-07,
      "loss": 0.061722517013549805,
      "step": 6975
    },
    {
      "epoch": 0.9431009716941275,
      "grad_norm": 0.4013594388961792,
      "learning_rate": 2.772271164178086e-07,
      "loss": 0.07347822189331055,
      "step": 6976
    },
    {
      "epoch": 0.9432361639205745,
      "grad_norm": 0.31069520115852356,
      "learning_rate": 2.759188387160677e-07,
      "loss": 0.07013297080993652,
      "step": 6977
    },
    {
      "epoch": 0.9433713561470215,
      "grad_norm": 0.37476569414138794,
      "learning_rate": 2.746136266465449e-07,
      "loss": 0.06091737747192383,
      "step": 6978
    },
    {
      "epoch": 0.9435065483734685,
      "grad_norm": 0.40252289175987244,
      "learning_rate": 2.7331148048098943e-07,
      "loss": 0.05612373352050781,
      "step": 6979
    },
    {
      "epoch": 0.9436417405999155,
      "grad_norm": 0.7355493307113647,
      "learning_rate": 2.7201240049051613e-07,
      "loss": 0.062238216400146484,
      "step": 6980
    },
    {
      "epoch": 0.9437769328263624,
      "grad_norm": 0.5777159333229065,
      "learning_rate": 2.707163869455986e-07,
      "loss": 0.0664830207824707,
      "step": 6981
    },
    {
      "epoch": 0.9439121250528094,
      "grad_norm": 0.28528687357902527,
      "learning_rate": 2.694234401160778e-07,
      "loss": 0.07567930221557617,
      "step": 6982
    },
    {
      "epoch": 0.9440473172792564,
      "grad_norm": 0.2785634398460388,
      "learning_rate": 2.6813356027114986e-07,
      "loss": 0.06937122344970703,
      "step": 6983
    },
    {
      "epoch": 0.9441825095057034,
      "grad_norm": 0.21990138292312622,
      "learning_rate": 2.6684674767937346e-07,
      "loss": 0.05024909973144531,
      "step": 6984
    },
    {
      "epoch": 0.9443177017321503,
      "grad_norm": 0.3015028238296509,
      "learning_rate": 2.655630026086708e-07,
      "loss": 0.08124089241027832,
      "step": 6985
    },
    {
      "epoch": 0.9444528939585973,
      "grad_norm": 0.2186887264251709,
      "learning_rate": 2.642823253263249e-07,
      "loss": 0.0534210205078125,
      "step": 6986
    },
    {
      "epoch": 0.9445880861850443,
      "grad_norm": 0.1894114762544632,
      "learning_rate": 2.630047160989807e-07,
      "loss": 0.04532814025878906,
      "step": 6987
    },
    {
      "epoch": 0.9447232784114913,
      "grad_norm": 0.41836366057395935,
      "learning_rate": 2.6173017519263875e-07,
      "loss": 0.07275712490081787,
      "step": 6988
    },
    {
      "epoch": 0.9448584706379383,
      "grad_norm": 0.3209652304649353,
      "learning_rate": 2.6045870287267014e-07,
      "loss": 0.035703182220458984,
      "step": 6989
    },
    {
      "epoch": 0.9449936628643854,
      "grad_norm": 0.451030969619751,
      "learning_rate": 2.5919029940380147e-07,
      "loss": 0.11251354217529297,
      "step": 6990
    },
    {
      "epoch": 0.9451288550908323,
      "grad_norm": 0.3431937098503113,
      "learning_rate": 2.5792496505011807e-07,
      "loss": 0.056937217712402344,
      "step": 6991
    },
    {
      "epoch": 0.9452640473172793,
      "grad_norm": 0.2512618601322174,
      "learning_rate": 2.5666270007507266e-07,
      "loss": 0.06287622451782227,
      "step": 6992
    },
    {
      "epoch": 0.9453992395437263,
      "grad_norm": 0.21283449232578278,
      "learning_rate": 2.5540350474147324e-07,
      "loss": 0.048261940479278564,
      "step": 6993
    },
    {
      "epoch": 0.9455344317701733,
      "grad_norm": 0.14424695074558258,
      "learning_rate": 2.5414737931149346e-07,
      "loss": 0.03175187110900879,
      "step": 6994
    },
    {
      "epoch": 0.9456696239966202,
      "grad_norm": 0.20185905694961548,
      "learning_rate": 2.5289432404666246e-07,
      "loss": 0.0544891357421875,
      "step": 6995
    },
    {
      "epoch": 0.9458048162230672,
      "grad_norm": 0.40715834498405457,
      "learning_rate": 2.5164433920787487e-07,
      "loss": 0.08652234077453613,
      "step": 6996
    },
    {
      "epoch": 0.9459400084495142,
      "grad_norm": 0.34804922342300415,
      "learning_rate": 2.503974250553842e-07,
      "loss": 0.07156586647033691,
      "step": 6997
    },
    {
      "epoch": 0.9460752006759612,
      "grad_norm": 0.3125251829624176,
      "learning_rate": 2.491535818488011e-07,
      "loss": 0.054769039154052734,
      "step": 6998
    },
    {
      "epoch": 0.9462103929024082,
      "grad_norm": 0.16978426277637482,
      "learning_rate": 2.479128098471067e-07,
      "loss": 0.05221843719482422,
      "step": 6999
    },
    {
      "epoch": 0.9463455851288551,
      "grad_norm": 0.2932808995246887,
      "learning_rate": 2.466751093086328e-07,
      "loss": 0.05717134475708008,
      "step": 7000
    },
    {
      "epoch": 0.9464807773553021,
      "grad_norm": 0.24437984824180603,
      "learning_rate": 2.454404804910748e-07,
      "loss": 0.06388866901397705,
      "step": 7001
    },
    {
      "epoch": 0.9466159695817491,
      "grad_norm": 0.34393393993377686,
      "learning_rate": 2.442089236514888e-07,
      "loss": 0.06814050674438477,
      "step": 7002
    },
    {
      "epoch": 0.9467511618081961,
      "grad_norm": 0.21317288279533386,
      "learning_rate": 2.429804390462931e-07,
      "loss": 0.046701908111572266,
      "step": 7003
    },
    {
      "epoch": 0.946886354034643,
      "grad_norm": 0.39425522089004517,
      "learning_rate": 2.4175502693126293e-07,
      "loss": 0.08556795120239258,
      "step": 7004
    },
    {
      "epoch": 0.94702154626109,
      "grad_norm": 0.34288930892944336,
      "learning_rate": 2.4053268756153933e-07,
      "loss": 0.05003559589385986,
      "step": 7005
    },
    {
      "epoch": 0.947156738487537,
      "grad_norm": 0.3130124807357788,
      "learning_rate": 2.393134211916154e-07,
      "loss": 0.08249187469482422,
      "step": 7006
    },
    {
      "epoch": 0.947291930713984,
      "grad_norm": 0.43975111842155457,
      "learning_rate": 2.3809722807535128e-07,
      "loss": 0.07442378997802734,
      "step": 7007
    },
    {
      "epoch": 0.947427122940431,
      "grad_norm": 0.2952204644680023,
      "learning_rate": 2.3688410846596287e-07,
      "loss": 0.07238435745239258,
      "step": 7008
    },
    {
      "epoch": 0.9475623151668779,
      "grad_norm": 0.1816512793302536,
      "learning_rate": 2.3567406261603143e-07,
      "loss": 0.043540239334106445,
      "step": 7009
    },
    {
      "epoch": 0.9476975073933249,
      "grad_norm": 0.3818410336971283,
      "learning_rate": 2.3446709077749206e-07,
      "loss": 0.06792497634887695,
      "step": 7010
    },
    {
      "epoch": 0.9478326996197719,
      "grad_norm": 0.3031005561351776,
      "learning_rate": 2.3326319320164546e-07,
      "loss": 0.06437397003173828,
      "step": 7011
    },
    {
      "epoch": 0.9479678918462189,
      "grad_norm": 0.28739050030708313,
      "learning_rate": 2.320623701391461e-07,
      "loss": 0.054616451263427734,
      "step": 7012
    },
    {
      "epoch": 0.9481030840726659,
      "grad_norm": 0.2807123363018036,
      "learning_rate": 2.30864621840014e-07,
      "loss": 0.05580496788024902,
      "step": 7013
    },
    {
      "epoch": 0.9482382762991128,
      "grad_norm": 0.35229331254959106,
      "learning_rate": 2.2966994855362633e-07,
      "loss": 0.060595035552978516,
      "step": 7014
    },
    {
      "epoch": 0.9483734685255598,
      "grad_norm": 0.38257506489753723,
      "learning_rate": 2.2847835052872079e-07,
      "loss": 0.08170199394226074,
      "step": 7015
    },
    {
      "epoch": 0.9485086607520068,
      "grad_norm": 0.23198705911636353,
      "learning_rate": 2.2728982801339392e-07,
      "loss": 0.049857258796691895,
      "step": 7016
    },
    {
      "epoch": 0.9486438529784538,
      "grad_norm": 0.5656779408454895,
      "learning_rate": 2.261043812551028e-07,
      "loss": 0.07619524002075195,
      "step": 7017
    },
    {
      "epoch": 0.9487790452049008,
      "grad_norm": 0.34326377511024475,
      "learning_rate": 2.249220105006633e-07,
      "loss": 0.0723569393157959,
      "step": 7018
    },
    {
      "epoch": 0.9489142374313477,
      "grad_norm": 0.872576892375946,
      "learning_rate": 2.2374271599625185e-07,
      "loss": 0.06780195236206055,
      "step": 7019
    },
    {
      "epoch": 0.9490494296577947,
      "grad_norm": 0.3512718975543976,
      "learning_rate": 2.2256649798740204e-07,
      "loss": 0.07598590850830078,
      "step": 7020
    },
    {
      "epoch": 0.9491846218842417,
      "grad_norm": 0.20659223198890686,
      "learning_rate": 2.2139335671901294e-07,
      "loss": 0.048117876052856445,
      "step": 7021
    },
    {
      "epoch": 0.9493198141106887,
      "grad_norm": 0.4059467613697052,
      "learning_rate": 2.2022329243533422e-07,
      "loss": 0.07088565826416016,
      "step": 7022
    },
    {
      "epoch": 0.9494550063371356,
      "grad_norm": 0.25918686389923096,
      "learning_rate": 2.19056305379981e-07,
      "loss": 0.0582733154296875,
      "step": 7023
    },
    {
      "epoch": 0.9495901985635826,
      "grad_norm": 0.4293784201145172,
      "learning_rate": 2.178923957959289e-07,
      "loss": 0.08427047729492188,
      "step": 7024
    },
    {
      "epoch": 0.9497253907900296,
      "grad_norm": 0.3285316526889801,
      "learning_rate": 2.1673156392550408e-07,
      "loss": 0.07593441009521484,
      "step": 7025
    },
    {
      "epoch": 0.9498605830164766,
      "grad_norm": 0.21758310496807098,
      "learning_rate": 2.155738100104049e-07,
      "loss": 0.052509307861328125,
      "step": 7026
    },
    {
      "epoch": 0.9499957752429236,
      "grad_norm": 0.45218056440353394,
      "learning_rate": 2.1441913429167682e-07,
      "loss": 0.08870363235473633,
      "step": 7027
    },
    {
      "epoch": 0.9501309674693705,
      "grad_norm": 0.24682217836380005,
      "learning_rate": 2.1326753700973256e-07,
      "loss": 0.05330920219421387,
      "step": 7028
    },
    {
      "epoch": 0.9502661596958175,
      "grad_norm": 0.22413359582424164,
      "learning_rate": 2.1211901840434034e-07,
      "loss": 0.050739288330078125,
      "step": 7029
    },
    {
      "epoch": 0.9504013519222645,
      "grad_norm": 0.3305968940258026,
      "learning_rate": 2.1097357871462386e-07,
      "loss": 0.07822704315185547,
      "step": 7030
    },
    {
      "epoch": 0.9505365441487115,
      "grad_norm": 0.6246079206466675,
      "learning_rate": 2.098312181790757e-07,
      "loss": 0.10966968536376953,
      "step": 7031
    },
    {
      "epoch": 0.9506717363751584,
      "grad_norm": 0.20123112201690674,
      "learning_rate": 2.086919370355389e-07,
      "loss": 0.0447077751159668,
      "step": 7032
    },
    {
      "epoch": 0.9508069286016054,
      "grad_norm": 0.255611777305603,
      "learning_rate": 2.075557355212171e-07,
      "loss": 0.052898406982421875,
      "step": 7033
    },
    {
      "epoch": 0.9509421208280524,
      "grad_norm": 0.3235122263431549,
      "learning_rate": 2.0642261387267268e-07,
      "loss": 0.0729207992553711,
      "step": 7034
    },
    {
      "epoch": 0.9510773130544994,
      "grad_norm": 0.28758180141448975,
      "learning_rate": 2.0529257232583033e-07,
      "loss": 0.08346271514892578,
      "step": 7035
    },
    {
      "epoch": 0.9512125052809464,
      "grad_norm": 0.4493849277496338,
      "learning_rate": 2.0416561111596844e-07,
      "loss": 0.0783393383026123,
      "step": 7036
    },
    {
      "epoch": 0.9513476975073933,
      "grad_norm": 0.3834860920906067,
      "learning_rate": 2.0304173047772933e-07,
      "loss": 0.08037376403808594,
      "step": 7037
    },
    {
      "epoch": 0.9514828897338403,
      "grad_norm": 0.4030027389526367,
      "learning_rate": 2.0192093064510753e-07,
      "loss": 0.07949918508529663,
      "step": 7038
    },
    {
      "epoch": 0.9516180819602873,
      "grad_norm": 0.2951275408267975,
      "learning_rate": 2.0080321185146134e-07,
      "loss": 0.06663179397583008,
      "step": 7039
    },
    {
      "epoch": 0.9517532741867343,
      "grad_norm": 0.3480106592178345,
      "learning_rate": 1.996885743295046e-07,
      "loss": 0.0656588077545166,
      "step": 7040
    },
    {
      "epoch": 0.9518884664131813,
      "grad_norm": 0.2948501706123352,
      "learning_rate": 1.985770183113117e-07,
      "loss": 0.07940053939819336,
      "step": 7041
    },
    {
      "epoch": 0.9520236586396282,
      "grad_norm": 0.40841129422187805,
      "learning_rate": 1.9746854402831583e-07,
      "loss": 0.06979656219482422,
      "step": 7042
    },
    {
      "epoch": 0.9521588508660752,
      "grad_norm": 0.2220703512430191,
      "learning_rate": 1.963631517113057e-07,
      "loss": 0.036959171295166016,
      "step": 7043
    },
    {
      "epoch": 0.9522940430925222,
      "grad_norm": 0.4347338378429413,
      "learning_rate": 1.952608415904289e-07,
      "loss": 0.07242012023925781,
      "step": 7044
    },
    {
      "epoch": 0.9524292353189692,
      "grad_norm": 0.23542477190494537,
      "learning_rate": 1.9416161389519348e-07,
      "loss": 0.05971503257751465,
      "step": 7045
    },
    {
      "epoch": 0.9525644275454161,
      "grad_norm": 0.3125111758708954,
      "learning_rate": 1.9306546885446475e-07,
      "loss": 0.06570959091186523,
      "step": 7046
    },
    {
      "epoch": 0.9526996197718631,
      "grad_norm": 0.3278358280658722,
      "learning_rate": 1.919724066964651e-07,
      "loss": 0.05412447452545166,
      "step": 7047
    },
    {
      "epoch": 0.9528348119983101,
      "grad_norm": 0.3037605285644531,
      "learning_rate": 1.908824276487775e-07,
      "loss": 0.05943584442138672,
      "step": 7048
    },
    {
      "epoch": 0.9529700042247571,
      "grad_norm": 0.3124946057796478,
      "learning_rate": 1.8979553193833876e-07,
      "loss": 0.06500482559204102,
      "step": 7049
    },
    {
      "epoch": 0.9531051964512041,
      "grad_norm": 0.2885887026786804,
      "learning_rate": 1.8871171979144786e-07,
      "loss": 0.05474728345870972,
      "step": 7050
    },
    {
      "epoch": 0.953240388677651,
      "grad_norm": 0.2927285432815552,
      "learning_rate": 1.8763099143376262e-07,
      "loss": 0.06365370750427246,
      "step": 7051
    },
    {
      "epoch": 0.953375580904098,
      "grad_norm": 0.40386590361595154,
      "learning_rate": 1.8655334709029303e-07,
      "loss": 0.08675789833068848,
      "step": 7052
    },
    {
      "epoch": 0.953510773130545,
      "grad_norm": 0.19471007585525513,
      "learning_rate": 1.8547878698541132e-07,
      "loss": 0.04690289497375488,
      "step": 7053
    },
    {
      "epoch": 0.953645965356992,
      "grad_norm": 0.29762133955955505,
      "learning_rate": 1.8440731134284684e-07,
      "loss": 0.04564952850341797,
      "step": 7054
    },
    {
      "epoch": 0.953781157583439,
      "grad_norm": 0.341513991355896,
      "learning_rate": 1.833389203856861e-07,
      "loss": 0.07873868942260742,
      "step": 7055
    },
    {
      "epoch": 0.9539163498098859,
      "grad_norm": 0.24388428032398224,
      "learning_rate": 1.8227361433637625e-07,
      "loss": 0.06833457946777344,
      "step": 7056
    },
    {
      "epoch": 0.9540515420363329,
      "grad_norm": 0.2513248026371002,
      "learning_rate": 1.812113934167148e-07,
      "loss": 0.07186603546142578,
      "step": 7057
    },
    {
      "epoch": 0.9541867342627799,
      "grad_norm": 0.263601154088974,
      "learning_rate": 1.8015225784786483e-07,
      "loss": 0.06045389175415039,
      "step": 7058
    },
    {
      "epoch": 0.9543219264892269,
      "grad_norm": 0.34621843695640564,
      "learning_rate": 1.7909620785034663e-07,
      "loss": 0.07863807678222656,
      "step": 7059
    },
    {
      "epoch": 0.9544571187156738,
      "grad_norm": 0.24743366241455078,
      "learning_rate": 1.7804324364402936e-07,
      "loss": 0.056606292724609375,
      "step": 7060
    },
    {
      "epoch": 0.9545923109421208,
      "grad_norm": 0.3017016053199768,
      "learning_rate": 1.769933654481526e-07,
      "loss": 0.07212352752685547,
      "step": 7061
    },
    {
      "epoch": 0.9547275031685678,
      "grad_norm": 0.2539996802806854,
      "learning_rate": 1.7594657348129984e-07,
      "loss": 0.06033825874328613,
      "step": 7062
    },
    {
      "epoch": 0.9548626953950148,
      "grad_norm": 0.24367156624794006,
      "learning_rate": 1.749028679614234e-07,
      "loss": 0.0654909610748291,
      "step": 7063
    },
    {
      "epoch": 0.9549978876214618,
      "grad_norm": 0.501304566860199,
      "learning_rate": 1.7386224910582615e-07,
      "loss": 0.0479697585105896,
      "step": 7064
    },
    {
      "epoch": 0.9551330798479087,
      "grad_norm": 0.21164806187152863,
      "learning_rate": 1.728247171311731e-07,
      "loss": 0.05570030212402344,
      "step": 7065
    },
    {
      "epoch": 0.9552682720743557,
      "grad_norm": 0.20658764243125916,
      "learning_rate": 1.7179027225348142e-07,
      "loss": 0.05929446220397949,
      "step": 7066
    },
    {
      "epoch": 0.9554034643008027,
      "grad_norm": 0.2435864955186844,
      "learning_rate": 1.7075891468812722e-07,
      "loss": 0.05292034149169922,
      "step": 7067
    },
    {
      "epoch": 0.9555386565272497,
      "grad_norm": 0.2956370711326599,
      "learning_rate": 1.69730644649847e-07,
      "loss": 0.0629395842552185,
      "step": 7068
    },
    {
      "epoch": 0.9556738487536967,
      "grad_norm": 0.30187246203422546,
      "learning_rate": 1.687054623527312e-07,
      "loss": 0.0683140754699707,
      "step": 7069
    },
    {
      "epoch": 0.9558090409801436,
      "grad_norm": 0.1824604570865631,
      "learning_rate": 1.676833680102291e-07,
      "loss": 0.03387176990509033,
      "step": 7070
    },
    {
      "epoch": 0.9559442332065906,
      "grad_norm": 0.15600618720054626,
      "learning_rate": 1.6666436183514378e-07,
      "loss": 0.03223252296447754,
      "step": 7071
    },
    {
      "epoch": 0.9560794254330376,
      "grad_norm": 0.21348808705806732,
      "learning_rate": 1.6564844403964053e-07,
      "loss": 0.05842113494873047,
      "step": 7072
    },
    {
      "epoch": 0.9562146176594846,
      "grad_norm": 0.22913876175880432,
      "learning_rate": 1.6463561483523682e-07,
      "loss": 0.04892158508300781,
      "step": 7073
    },
    {
      "epoch": 0.9563498098859315,
      "grad_norm": 0.279837429523468,
      "learning_rate": 1.6362587443281063e-07,
      "loss": 0.05359315872192383,
      "step": 7074
    },
    {
      "epoch": 0.9564850021123785,
      "grad_norm": 0.40154993534088135,
      "learning_rate": 1.626192230425938e-07,
      "loss": 0.07887077331542969,
      "step": 7075
    },
    {
      "epoch": 0.9566201943388255,
      "grad_norm": 0.43644243478775024,
      "learning_rate": 1.6161566087417868e-07,
      "loss": 0.08739185333251953,
      "step": 7076
    },
    {
      "epoch": 0.9567553865652725,
      "grad_norm": 0.2820413410663605,
      "learning_rate": 1.6061518813650977e-07,
      "loss": 0.06787729263305664,
      "step": 7077
    },
    {
      "epoch": 0.9568905787917195,
      "grad_norm": 0.27579060196876526,
      "learning_rate": 1.5961780503789215e-07,
      "loss": 0.050455570220947266,
      "step": 7078
    },
    {
      "epoch": 0.9570257710181664,
      "grad_norm": 0.2353002429008484,
      "learning_rate": 1.5862351178598633e-07,
      "loss": 0.055228233337402344,
      "step": 7079
    },
    {
      "epoch": 0.9571609632446134,
      "grad_norm": 0.22778448462486267,
      "learning_rate": 1.5763230858781008e-07,
      "loss": 0.04465299844741821,
      "step": 7080
    },
    {
      "epoch": 0.9572961554710604,
      "grad_norm": 0.3392725884914398,
      "learning_rate": 1.5664419564973497e-07,
      "loss": 0.06159639358520508,
      "step": 7081
    },
    {
      "epoch": 0.9574313476975074,
      "grad_norm": 0.31525975465774536,
      "learning_rate": 1.5565917317749146e-07,
      "loss": 0.07815194129943848,
      "step": 7082
    },
    {
      "epoch": 0.9575665399239544,
      "grad_norm": 0.28295913338661194,
      "learning_rate": 1.5467724137617046e-07,
      "loss": 0.06441307067871094,
      "step": 7083
    },
    {
      "epoch": 0.9577017321504013,
      "grad_norm": 0.30150237679481506,
      "learning_rate": 1.5369840045021178e-07,
      "loss": 0.06901931762695312,
      "step": 7084
    },
    {
      "epoch": 0.9578369243768483,
      "grad_norm": 0.2961951792240143,
      "learning_rate": 1.5272265060341572e-07,
      "loss": 0.06909561157226562,
      "step": 7085
    },
    {
      "epoch": 0.9579721166032953,
      "grad_norm": 0.18949323892593384,
      "learning_rate": 1.517499920389398e-07,
      "loss": 0.04059183597564697,
      "step": 7086
    },
    {
      "epoch": 0.9581073088297423,
      "grad_norm": 0.373098760843277,
      "learning_rate": 1.5078042495929534e-07,
      "loss": 0.07132530212402344,
      "step": 7087
    },
    {
      "epoch": 0.9582425010561892,
      "grad_norm": 0.3374219536781311,
      "learning_rate": 1.498139495663542e-07,
      "loss": 0.06792446970939636,
      "step": 7088
    },
    {
      "epoch": 0.9583776932826362,
      "grad_norm": 0.46853941679000854,
      "learning_rate": 1.4885056606133707e-07,
      "loss": 0.08032035827636719,
      "step": 7089
    },
    {
      "epoch": 0.9585128855090832,
      "grad_norm": 0.1993354856967926,
      "learning_rate": 1.478902746448302e-07,
      "loss": 0.04134368896484375,
      "step": 7090
    },
    {
      "epoch": 0.9586480777355302,
      "grad_norm": 0.34106630086898804,
      "learning_rate": 1.469330755167686e-07,
      "loss": 0.06944561004638672,
      "step": 7091
    },
    {
      "epoch": 0.9587832699619772,
      "grad_norm": 0.3594677150249481,
      "learning_rate": 1.4597896887644458e-07,
      "loss": 0.06741833686828613,
      "step": 7092
    },
    {
      "epoch": 0.9589184621884241,
      "grad_norm": 0.3508850038051605,
      "learning_rate": 1.4502795492251418e-07,
      "loss": 0.0678098201751709,
      "step": 7093
    },
    {
      "epoch": 0.9590536544148711,
      "grad_norm": 0.16216158866882324,
      "learning_rate": 1.4408003385297742e-07,
      "loss": 0.03735470771789551,
      "step": 7094
    },
    {
      "epoch": 0.9591888466413181,
      "grad_norm": 0.2320837825536728,
      "learning_rate": 1.4313520586519968e-07,
      "loss": 0.04517108201980591,
      "step": 7095
    },
    {
      "epoch": 0.9593240388677651,
      "grad_norm": 0.2104930579662323,
      "learning_rate": 1.4219347115589863e-07,
      "loss": 0.04501771926879883,
      "step": 7096
    },
    {
      "epoch": 0.959459231094212,
      "grad_norm": 0.21901686489582062,
      "learning_rate": 1.4125482992114914e-07,
      "loss": 0.051828861236572266,
      "step": 7097
    },
    {
      "epoch": 0.959594423320659,
      "grad_norm": 0.2667662501335144,
      "learning_rate": 1.403192823563798e-07,
      "loss": 0.05644547939300537,
      "step": 7098
    },
    {
      "epoch": 0.959729615547106,
      "grad_norm": 0.33934083580970764,
      "learning_rate": 1.3938682865637654e-07,
      "loss": 0.07100439071655273,
      "step": 7099
    },
    {
      "epoch": 0.959864807773553,
      "grad_norm": 0.5984017848968506,
      "learning_rate": 1.38457469015284e-07,
      "loss": 0.07280826568603516,
      "step": 7100
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2662726640701294,
      "learning_rate": 1.3753120362659576e-07,
      "loss": 0.05173593759536743,
      "step": 7101
    },
    {
      "epoch": 0.9601351922264469,
      "grad_norm": 0.3209396004676819,
      "learning_rate": 1.3660803268316925e-07,
      "loss": 0.06452083587646484,
      "step": 7102
    },
    {
      "epoch": 0.9602703844528939,
      "grad_norm": 0.23695382475852966,
      "learning_rate": 1.3568795637721065e-07,
      "loss": 0.058144211769104004,
      "step": 7103
    },
    {
      "epoch": 0.9604055766793409,
      "grad_norm": 0.3340555429458618,
      "learning_rate": 1.347709749002851e-07,
      "loss": 0.06425333023071289,
      "step": 7104
    },
    {
      "epoch": 0.9605407689057879,
      "grad_norm": 0.37709692120552063,
      "learning_rate": 1.338570884433149e-07,
      "loss": 0.07504653930664062,
      "step": 7105
    },
    {
      "epoch": 0.9606759611322349,
      "grad_norm": 0.44086870551109314,
      "learning_rate": 1.3294629719657448e-07,
      "loss": 0.09051513671875,
      "step": 7106
    },
    {
      "epoch": 0.9608111533586818,
      "grad_norm": 0.27949607372283936,
      "learning_rate": 1.3203860134969548e-07,
      "loss": 0.07896995544433594,
      "step": 7107
    },
    {
      "epoch": 0.9609463455851288,
      "grad_norm": 0.18890832364559174,
      "learning_rate": 1.3113400109166508e-07,
      "loss": 0.04881572723388672,
      "step": 7108
    },
    {
      "epoch": 0.9610815378115758,
      "grad_norm": 0.7075672149658203,
      "learning_rate": 1.3023249661082592e-07,
      "loss": 0.10370111465454102,
      "step": 7109
    },
    {
      "epoch": 0.9612167300380228,
      "grad_norm": 0.38277676701545715,
      "learning_rate": 1.2933408809487623e-07,
      "loss": 0.08023452758789062,
      "step": 7110
    },
    {
      "epoch": 0.9613519222644697,
      "grad_norm": 0.35066574811935425,
      "learning_rate": 1.2843877573086972e-07,
      "loss": 0.0773056149482727,
      "step": 7111
    },
    {
      "epoch": 0.9614871144909167,
      "grad_norm": 0.17431510984897614,
      "learning_rate": 1.2754655970521556e-07,
      "loss": 0.03415632247924805,
      "step": 7112
    },
    {
      "epoch": 0.9616223067173637,
      "grad_norm": 0.32355645298957825,
      "learning_rate": 1.2665744020367686e-07,
      "loss": 0.07684391736984253,
      "step": 7113
    },
    {
      "epoch": 0.9617574989438107,
      "grad_norm": 0.2118271142244339,
      "learning_rate": 1.2577141741137388e-07,
      "loss": 0.053942203521728516,
      "step": 7114
    },
    {
      "epoch": 0.9618926911702577,
      "grad_norm": 0.4489447772502899,
      "learning_rate": 1.248884915127807e-07,
      "loss": 0.07523274421691895,
      "step": 7115
    },
    {
      "epoch": 0.9620278833967046,
      "grad_norm": 0.13798269629478455,
      "learning_rate": 1.2400866269172694e-07,
      "loss": 0.038768768310546875,
      "step": 7116
    },
    {
      "epoch": 0.9621630756231516,
      "grad_norm": 0.56400465965271,
      "learning_rate": 1.2313193113139777e-07,
      "loss": 0.07834219932556152,
      "step": 7117
    },
    {
      "epoch": 0.9622982678495986,
      "grad_norm": 0.22872433066368103,
      "learning_rate": 1.2225829701433545e-07,
      "loss": 0.04670572280883789,
      "step": 7118
    },
    {
      "epoch": 0.9624334600760456,
      "grad_norm": 0.18286460638046265,
      "learning_rate": 1.2138776052243116e-07,
      "loss": 0.041526854038238525,
      "step": 7119
    },
    {
      "epoch": 0.9625686523024926,
      "grad_norm": 0.40548887848854065,
      "learning_rate": 1.2052032183693996e-07,
      "loss": 0.07451128959655762,
      "step": 7120
    },
    {
      "epoch": 0.9627038445289395,
      "grad_norm": 0.43057578802108765,
      "learning_rate": 1.196559811384623e-07,
      "loss": 0.09782171249389648,
      "step": 7121
    },
    {
      "epoch": 0.9628390367553865,
      "grad_norm": 0.3550565540790558,
      "learning_rate": 1.1879473860696266e-07,
      "loss": 0.06103086471557617,
      "step": 7122
    },
    {
      "epoch": 0.9629742289818335,
      "grad_norm": 0.5130022168159485,
      "learning_rate": 1.179365944217542e-07,
      "loss": 0.09292316436767578,
      "step": 7123
    },
    {
      "epoch": 0.9631094212082805,
      "grad_norm": 0.20128019154071808,
      "learning_rate": 1.1708154876150735e-07,
      "loss": 0.05059695243835449,
      "step": 7124
    },
    {
      "epoch": 0.9632446134347274,
      "grad_norm": 0.16424371302127838,
      "learning_rate": 1.1622960180424801e-07,
      "loss": 0.0390017032623291,
      "step": 7125
    },
    {
      "epoch": 0.9633798056611745,
      "grad_norm": 0.3007664978504181,
      "learning_rate": 1.1538075372735435e-07,
      "loss": 0.06849813461303711,
      "step": 7126
    },
    {
      "epoch": 0.9635149978876215,
      "grad_norm": 0.33393779397010803,
      "learning_rate": 1.1453500470756328e-07,
      "loss": 0.0654149055480957,
      "step": 7127
    },
    {
      "epoch": 0.9636501901140685,
      "grad_norm": 0.36746785044670105,
      "learning_rate": 1.1369235492096397e-07,
      "loss": 0.07008492946624756,
      "step": 7128
    },
    {
      "epoch": 0.9637853823405155,
      "grad_norm": 0.24565951526165009,
      "learning_rate": 1.1285280454299774e-07,
      "loss": 0.05627632141113281,
      "step": 7129
    },
    {
      "epoch": 0.9639205745669625,
      "grad_norm": 0.2792106568813324,
      "learning_rate": 1.1201635374846808e-07,
      "loss": 0.04860877990722656,
      "step": 7130
    },
    {
      "epoch": 0.9640557667934094,
      "grad_norm": 0.2769196033477783,
      "learning_rate": 1.1118300271152404e-07,
      "loss": 0.05370759963989258,
      "step": 7131
    },
    {
      "epoch": 0.9641909590198564,
      "grad_norm": 0.32462385296821594,
      "learning_rate": 1.1035275160567682e-07,
      "loss": 0.050774216651916504,
      "step": 7132
    },
    {
      "epoch": 0.9643261512463034,
      "grad_norm": 0.29091402888298035,
      "learning_rate": 1.0952560060378813e-07,
      "loss": 0.06504011154174805,
      "step": 7133
    },
    {
      "epoch": 0.9644613434727504,
      "grad_norm": 0.24630087614059448,
      "learning_rate": 1.0870154987807523e-07,
      "loss": 0.059615135192871094,
      "step": 7134
    },
    {
      "epoch": 0.9645965356991973,
      "grad_norm": 0.36789265275001526,
      "learning_rate": 1.0788059960010921e-07,
      "loss": 0.06681013107299805,
      "step": 7135
    },
    {
      "epoch": 0.9647317279256443,
      "grad_norm": 0.5074768662452698,
      "learning_rate": 1.0706274994081499e-07,
      "loss": 0.0800933837890625,
      "step": 7136
    },
    {
      "epoch": 0.9648669201520913,
      "grad_norm": 0.3125888407230377,
      "learning_rate": 1.0624800107047805e-07,
      "loss": 0.05780029296875,
      "step": 7137
    },
    {
      "epoch": 0.9650021123785383,
      "grad_norm": 0.24129152297973633,
      "learning_rate": 1.0543635315872934e-07,
      "loss": 0.05544567108154297,
      "step": 7138
    },
    {
      "epoch": 0.9651373046049853,
      "grad_norm": 0.22976483404636383,
      "learning_rate": 1.0462780637455871e-07,
      "loss": 0.05627632141113281,
      "step": 7139
    },
    {
      "epoch": 0.9652724968314322,
      "grad_norm": 0.36747923493385315,
      "learning_rate": 1.0382236088631148e-07,
      "loss": 0.07140541076660156,
      "step": 7140
    },
    {
      "epoch": 0.9654076890578792,
      "grad_norm": 0.37640753388404846,
      "learning_rate": 1.0302001686168349e-07,
      "loss": 0.07625985145568848,
      "step": 7141
    },
    {
      "epoch": 0.9655428812843262,
      "grad_norm": 0.3509242832660675,
      "learning_rate": 1.0222077446772949e-07,
      "loss": 0.0732278823852539,
      "step": 7142
    },
    {
      "epoch": 0.9656780735107732,
      "grad_norm": 0.3466155230998993,
      "learning_rate": 1.0142463387085465e-07,
      "loss": 0.05558037757873535,
      "step": 7143
    },
    {
      "epoch": 0.9658132657372202,
      "grad_norm": 0.36956116557121277,
      "learning_rate": 1.0063159523682142e-07,
      "loss": 0.06895875930786133,
      "step": 7144
    },
    {
      "epoch": 0.9659484579636671,
      "grad_norm": 0.15760314464569092,
      "learning_rate": 9.984165873074102e-08,
      "loss": 0.03542661666870117,
      "step": 7145
    },
    {
      "epoch": 0.9660836501901141,
      "grad_norm": 0.2602253258228302,
      "learning_rate": 9.905482451708526e-08,
      "loss": 0.06183439493179321,
      "step": 7146
    },
    {
      "epoch": 0.9662188424165611,
      "grad_norm": 0.32091230154037476,
      "learning_rate": 9.827109275967638e-08,
      "loss": 0.06027671694755554,
      "step": 7147
    },
    {
      "epoch": 0.9663540346430081,
      "grad_norm": 0.8618775010108948,
      "learning_rate": 9.749046362169223e-08,
      "loss": 0.08211898803710938,
      "step": 7148
    },
    {
      "epoch": 0.966489226869455,
      "grad_norm": 0.18240094184875488,
      "learning_rate": 9.671293726566443e-08,
      "loss": 0.042287349700927734,
      "step": 7149
    },
    {
      "epoch": 0.966624419095902,
      "grad_norm": 0.22764787077903748,
      "learning_rate": 9.593851385347518e-08,
      "loss": 0.05000448226928711,
      "step": 7150
    },
    {
      "epoch": 0.966759611322349,
      "grad_norm": 0.22367732226848602,
      "learning_rate": 9.516719354636716e-08,
      "loss": 0.04016005992889404,
      "step": 7151
    },
    {
      "epoch": 0.966894803548796,
      "grad_norm": 0.16878516972064972,
      "learning_rate": 9.439897650493024e-08,
      "loss": 0.04114675521850586,
      "step": 7152
    },
    {
      "epoch": 0.967029995775243,
      "grad_norm": 0.41311609745025635,
      "learning_rate": 9.363386288911313e-08,
      "loss": 0.07525777816772461,
      "step": 7153
    },
    {
      "epoch": 0.9671651880016899,
      "grad_norm": 0.48723217844963074,
      "learning_rate": 9.287185285821675e-08,
      "loss": 0.06959009170532227,
      "step": 7154
    },
    {
      "epoch": 0.9673003802281369,
      "grad_norm": 0.3045305907726288,
      "learning_rate": 9.211294657089587e-08,
      "loss": 0.07018661499023438,
      "step": 7155
    },
    {
      "epoch": 0.9674355724545839,
      "grad_norm": 0.39163950085639954,
      "learning_rate": 9.135714418515573e-08,
      "loss": 0.07484912872314453,
      "step": 7156
    },
    {
      "epoch": 0.9675707646810309,
      "grad_norm": 0.2506008744239807,
      "learning_rate": 9.060444585836381e-08,
      "loss": 0.0506129264831543,
      "step": 7157
    },
    {
      "epoch": 0.9677059569074778,
      "grad_norm": 0.3191521167755127,
      "learning_rate": 8.985485174722974e-08,
      "loss": 0.07621145248413086,
      "step": 7158
    },
    {
      "epoch": 0.9678411491339248,
      "grad_norm": 0.25807011127471924,
      "learning_rate": 8.910836200782868e-08,
      "loss": 0.053945839405059814,
      "step": 7159
    },
    {
      "epoch": 0.9679763413603718,
      "grad_norm": 0.15680165588855743,
      "learning_rate": 8.836497679557964e-08,
      "loss": 0.04192084074020386,
      "step": 7160
    },
    {
      "epoch": 0.9681115335868188,
      "grad_norm": 0.3204350769519806,
      "learning_rate": 8.762469626526048e-08,
      "loss": 0.06955146789550781,
      "step": 7161
    },
    {
      "epoch": 0.9682467258132658,
      "grad_norm": 0.16617003083229065,
      "learning_rate": 8.688752057100457e-08,
      "loss": 0.041983604431152344,
      "step": 7162
    },
    {
      "epoch": 0.9683819180397127,
      "grad_norm": 0.29832717776298523,
      "learning_rate": 8.615344986629082e-08,
      "loss": 0.06909561157226562,
      "step": 7163
    },
    {
      "epoch": 0.9685171102661597,
      "grad_norm": 0.24804559350013733,
      "learning_rate": 8.542248430396027e-08,
      "loss": 0.05909538269042969,
      "step": 7164
    },
    {
      "epoch": 0.9686523024926067,
      "grad_norm": 0.26386067271232605,
      "learning_rate": 8.469462403620287e-08,
      "loss": 0.06704902648925781,
      "step": 7165
    },
    {
      "epoch": 0.9687874947190537,
      "grad_norm": 0.2206057906150818,
      "learning_rate": 8.39698692145624e-08,
      "loss": 0.04303264617919922,
      "step": 7166
    },
    {
      "epoch": 0.9689226869455007,
      "grad_norm": 0.4482867121696472,
      "learning_rate": 8.324821998993648e-08,
      "loss": 0.08037328720092773,
      "step": 7167
    },
    {
      "epoch": 0.9690578791719476,
      "grad_norm": 0.28336599469184875,
      "learning_rate": 8.252967651257826e-08,
      "loss": 0.06237316131591797,
      "step": 7168
    },
    {
      "epoch": 0.9691930713983946,
      "grad_norm": 0.31158414483070374,
      "learning_rate": 8.181423893208973e-08,
      "loss": 0.05685222148895264,
      "step": 7169
    },
    {
      "epoch": 0.9693282636248416,
      "grad_norm": 0.4446757733821869,
      "learning_rate": 8.110190739743172e-08,
      "loss": 0.0708761215209961,
      "step": 7170
    },
    {
      "epoch": 0.9694634558512886,
      "grad_norm": 0.2637402415275574,
      "learning_rate": 8.03926820569123e-08,
      "loss": 0.06328153610229492,
      "step": 7171
    },
    {
      "epoch": 0.9695986480777355,
      "grad_norm": 0.34525594115257263,
      "learning_rate": 7.968656305819833e-08,
      "loss": 0.0838625431060791,
      "step": 7172
    },
    {
      "epoch": 0.9697338403041825,
      "grad_norm": 0.34984028339385986,
      "learning_rate": 7.898355054830719e-08,
      "loss": 0.06069469451904297,
      "step": 7173
    },
    {
      "epoch": 0.9698690325306295,
      "grad_norm": 0.21088449656963348,
      "learning_rate": 7.828364467360849e-08,
      "loss": 0.04669666290283203,
      "step": 7174
    },
    {
      "epoch": 0.9700042247570765,
      "grad_norm": 0.35954582691192627,
      "learning_rate": 7.758684557982731e-08,
      "loss": 0.06922340393066406,
      "step": 7175
    },
    {
      "epoch": 0.9701394169835235,
      "grad_norm": 0.4271712899208069,
      "learning_rate": 7.689315341204262e-08,
      "loss": 0.08757305145263672,
      "step": 7176
    },
    {
      "epoch": 0.9702746092099704,
      "grad_norm": 0.3467499613761902,
      "learning_rate": 7.62025683146822e-08,
      "loss": 0.05325937271118164,
      "step": 7177
    },
    {
      "epoch": 0.9704098014364174,
      "grad_norm": 0.23939074575901031,
      "learning_rate": 7.551509043152937e-08,
      "loss": 0.061994075775146484,
      "step": 7178
    },
    {
      "epoch": 0.9705449936628644,
      "grad_norm": 0.2400912642478943,
      "learning_rate": 7.483071990572132e-08,
      "loss": 0.05511891841888428,
      "step": 7179
    },
    {
      "epoch": 0.9706801858893114,
      "grad_norm": 0.23460254073143005,
      "learning_rate": 7.414945687975072e-08,
      "loss": 0.03956270217895508,
      "step": 7180
    },
    {
      "epoch": 0.9708153781157584,
      "grad_norm": 0.25749945640563965,
      "learning_rate": 7.347130149545578e-08,
      "loss": 0.05767488479614258,
      "step": 7181
    },
    {
      "epoch": 0.9709505703422053,
      "grad_norm": 0.32734954357147217,
      "learning_rate": 7.279625389403355e-08,
      "loss": 0.06890296936035156,
      "step": 7182
    },
    {
      "epoch": 0.9710857625686523,
      "grad_norm": 0.2572149932384491,
      "learning_rate": 7.212431421603327e-08,
      "loss": 0.0688391923904419,
      "step": 7183
    },
    {
      "epoch": 0.9712209547950993,
      "grad_norm": 0.3380451202392578,
      "learning_rate": 7.145548260135638e-08,
      "loss": 0.07073593139648438,
      "step": 7184
    },
    {
      "epoch": 0.9713561470215463,
      "grad_norm": 0.463604211807251,
      "learning_rate": 7.078975918925645e-08,
      "loss": 0.08678054809570312,
      "step": 7185
    },
    {
      "epoch": 0.9714913392479932,
      "grad_norm": 0.320751816034317,
      "learning_rate": 7.012714411834098e-08,
      "loss": 0.08072137832641602,
      "step": 7186
    },
    {
      "epoch": 0.9716265314744402,
      "grad_norm": 0.2804392874240875,
      "learning_rate": 6.946763752656959e-08,
      "loss": 0.06387472152709961,
      "step": 7187
    },
    {
      "epoch": 0.9717617237008872,
      "grad_norm": 0.3662104606628418,
      "learning_rate": 6.881123955125579e-08,
      "loss": 0.07719039916992188,
      "step": 7188
    },
    {
      "epoch": 0.9718969159273342,
      "grad_norm": 0.19144761562347412,
      "learning_rate": 6.815795032906524e-08,
      "loss": 0.050726890563964844,
      "step": 7189
    },
    {
      "epoch": 0.9720321081537812,
      "grad_norm": 0.19675759971141815,
      "learning_rate": 6.750776999601415e-08,
      "loss": 0.04305756092071533,
      "step": 7190
    },
    {
      "epoch": 0.9721673003802281,
      "grad_norm": 0.45745009183883667,
      "learning_rate": 6.68606986874759e-08,
      "loss": 0.07642531394958496,
      "step": 7191
    },
    {
      "epoch": 0.9723024926066751,
      "grad_norm": 0.23230880498886108,
      "learning_rate": 6.62167365381744e-08,
      "loss": 0.05736708641052246,
      "step": 7192
    },
    {
      "epoch": 0.9724376848331221,
      "grad_norm": 0.29215314984321594,
      "learning_rate": 6.557588368218237e-08,
      "loss": 0.06390190124511719,
      "step": 7193
    },
    {
      "epoch": 0.9725728770595691,
      "grad_norm": 0.2768566906452179,
      "learning_rate": 6.493814025293476e-08,
      "loss": 0.06568622589111328,
      "step": 7194
    },
    {
      "epoch": 0.972708069286016,
      "grad_norm": 0.3239062428474426,
      "learning_rate": 6.430350638320704e-08,
      "loss": 0.07325363159179688,
      "step": 7195
    },
    {
      "epoch": 0.972843261512463,
      "grad_norm": 0.5337064862251282,
      "learning_rate": 6.367198220513848e-08,
      "loss": 0.11066055297851562,
      "step": 7196
    },
    {
      "epoch": 0.97297845373891,
      "grad_norm": 0.23224112391471863,
      "learning_rate": 6.304356785021226e-08,
      "loss": 0.05916738510131836,
      "step": 7197
    },
    {
      "epoch": 0.973113645965357,
      "grad_norm": 0.5669562816619873,
      "learning_rate": 6.241826344926704e-08,
      "loss": 0.08518695831298828,
      "step": 7198
    },
    {
      "epoch": 0.973248838191804,
      "grad_norm": 0.28966808319091797,
      "learning_rate": 6.17960691324987e-08,
      "loss": 0.0686178207397461,
      "step": 7199
    },
    {
      "epoch": 0.973384030418251,
      "grad_norm": 0.45771700143814087,
      "learning_rate": 6.117698502944857e-08,
      "loss": 0.07668817043304443,
      "step": 7200
    },
    {
      "epoch": 0.9735192226446979,
      "grad_norm": 0.20307712256908417,
      "learning_rate": 6.056101126901358e-08,
      "loss": 0.05368947982788086,
      "step": 7201
    },
    {
      "epoch": 0.9736544148711449,
      "grad_norm": 0.5064142346382141,
      "learning_rate": 5.994814797944281e-08,
      "loss": 0.07696700096130371,
      "step": 7202
    },
    {
      "epoch": 0.9737896070975919,
      "grad_norm": 0.15008561313152313,
      "learning_rate": 5.933839528833751e-08,
      "loss": 0.023941993713378906,
      "step": 7203
    },
    {
      "epoch": 0.9739247993240389,
      "grad_norm": 0.17597651481628418,
      "learning_rate": 5.873175332265279e-08,
      "loss": 0.03881549835205078,
      "step": 7204
    },
    {
      "epoch": 0.9740599915504858,
      "grad_norm": 0.46539902687072754,
      "learning_rate": 5.812822220869096e-08,
      "loss": 0.08562612533569336,
      "step": 7205
    },
    {
      "epoch": 0.9741951837769328,
      "grad_norm": 0.36038830876350403,
      "learning_rate": 5.752780207211483e-08,
      "loss": 0.07819366455078125,
      "step": 7206
    },
    {
      "epoch": 0.9743303760033798,
      "grad_norm": 0.21653492748737335,
      "learning_rate": 5.693049303793274e-08,
      "loss": 0.04432499408721924,
      "step": 7207
    },
    {
      "epoch": 0.9744655682298268,
      "grad_norm": 0.4385357201099396,
      "learning_rate": 5.6336295230508536e-08,
      "loss": 0.05430912971496582,
      "step": 7208
    },
    {
      "epoch": 0.9746007604562738,
      "grad_norm": 0.3062569797039032,
      "learning_rate": 5.5745208773558266e-08,
      "loss": 0.06261181831359863,
      "step": 7209
    },
    {
      "epoch": 0.9747359526827207,
      "grad_norm": 0.288725882768631,
      "learning_rate": 5.515723379014681e-08,
      "loss": 0.056552886962890625,
      "step": 7210
    },
    {
      "epoch": 0.9748711449091677,
      "grad_norm": 0.2708200216293335,
      "learning_rate": 5.4572370402694583e-08,
      "loss": 0.057570457458496094,
      "step": 7211
    },
    {
      "epoch": 0.9750063371356147,
      "grad_norm": 0.48602837324142456,
      "learning_rate": 5.399061873297417e-08,
      "loss": 0.0813295841217041,
      "step": 7212
    },
    {
      "epoch": 0.9751415293620617,
      "grad_norm": 0.2376367151737213,
      "learning_rate": 5.341197890210869e-08,
      "loss": 0.060228586196899414,
      "step": 7213
    },
    {
      "epoch": 0.9752767215885086,
      "grad_norm": 0.26292112469673157,
      "learning_rate": 5.283645103057344e-08,
      "loss": 0.04542899131774902,
      "step": 7214
    },
    {
      "epoch": 0.9754119138149556,
      "grad_norm": 0.2552328407764435,
      "learning_rate": 5.226403523819756e-08,
      "loss": 0.062427520751953125,
      "step": 7215
    },
    {
      "epoch": 0.9755471060414026,
      "grad_norm": 0.3440481424331665,
      "learning_rate": 5.169473164416072e-08,
      "loss": 0.08122038841247559,
      "step": 7216
    },
    {
      "epoch": 0.9756822982678496,
      "grad_norm": 0.1281677782535553,
      "learning_rate": 5.112854036699477e-08,
      "loss": 0.039642333984375,
      "step": 7217
    },
    {
      "epoch": 0.9758174904942966,
      "grad_norm": 0.30934494733810425,
      "learning_rate": 5.0565461524583745e-08,
      "loss": 0.0696796178817749,
      "step": 7218
    },
    {
      "epoch": 0.9759526827207435,
      "grad_norm": 0.3011772334575653,
      "learning_rate": 5.0005495234163865e-08,
      "loss": 0.05882906913757324,
      "step": 7219
    },
    {
      "epoch": 0.9760878749471905,
      "grad_norm": 0.2965177297592163,
      "learning_rate": 4.9448641612321874e-08,
      "loss": 0.0741887092590332,
      "step": 7220
    },
    {
      "epoch": 0.9762230671736375,
      "grad_norm": 0.1637433022260666,
      "learning_rate": 4.889490077500003e-08,
      "loss": 0.042545318603515625,
      "step": 7221
    },
    {
      "epoch": 0.9763582594000845,
      "grad_norm": 0.49014654755592346,
      "learning_rate": 4.8344272837489434e-08,
      "loss": 0.07047206163406372,
      "step": 7222
    },
    {
      "epoch": 0.9764934516265315,
      "grad_norm": 0.42014312744140625,
      "learning_rate": 4.779675791443172e-08,
      "loss": 0.06003129482269287,
      "step": 7223
    },
    {
      "epoch": 0.9766286438529784,
      "grad_norm": 0.4761996865272522,
      "learning_rate": 4.72523561198257e-08,
      "loss": 0.07059383392333984,
      "step": 7224
    },
    {
      "epoch": 0.9767638360794254,
      "grad_norm": 0.24438482522964478,
      "learning_rate": 4.6711067567014044e-08,
      "loss": 0.06229877471923828,
      "step": 7225
    },
    {
      "epoch": 0.9768990283058724,
      "grad_norm": 0.26329684257507324,
      "learning_rate": 4.6172892368701595e-08,
      "loss": 0.06503629684448242,
      "step": 7226
    },
    {
      "epoch": 0.9770342205323194,
      "grad_norm": 0.2561330795288086,
      "learning_rate": 4.5637830636935385e-08,
      "loss": 0.048719167709350586,
      "step": 7227
    },
    {
      "epoch": 0.9771694127587663,
      "grad_norm": 0.3064153492450714,
      "learning_rate": 4.5105882483119643e-08,
      "loss": 0.055995821952819824,
      "step": 7228
    },
    {
      "epoch": 0.9773046049852133,
      "grad_norm": 0.3868359625339508,
      "learning_rate": 4.4577048018007436e-08,
      "loss": 0.06254005432128906,
      "step": 7229
    },
    {
      "epoch": 0.9774397972116603,
      "grad_norm": 0.24148406088352203,
      "learning_rate": 4.405132735170569e-08,
      "loss": 0.050814270973205566,
      "step": 7230
    },
    {
      "epoch": 0.9775749894381073,
      "grad_norm": 0.2694374918937683,
      "learning_rate": 4.3528720593675184e-08,
      "loss": 0.052547454833984375,
      "step": 7231
    },
    {
      "epoch": 0.9777101816645543,
      "grad_norm": 0.2057885378599167,
      "learning_rate": 4.300922785271888e-08,
      "loss": 0.048792481422424316,
      "step": 7232
    },
    {
      "epoch": 0.9778453738910012,
      "grad_norm": 0.18473146855831146,
      "learning_rate": 4.249284923700358e-08,
      "loss": 0.0374605655670166,
      "step": 7233
    },
    {
      "epoch": 0.9779805661174482,
      "grad_norm": 0.2718767523765564,
      "learning_rate": 4.197958485404163e-08,
      "loss": 0.059324026107788086,
      "step": 7234
    },
    {
      "epoch": 0.9781157583438952,
      "grad_norm": 0.2652534544467926,
      "learning_rate": 4.1469434810694206e-08,
      "loss": 0.0555729866027832,
      "step": 7235
    },
    {
      "epoch": 0.9782509505703422,
      "grad_norm": 0.6530328989028931,
      "learning_rate": 4.096239921317968e-08,
      "loss": 0.08185791969299316,
      "step": 7236
    },
    {
      "epoch": 0.9783861427967891,
      "grad_norm": 0.23014208674430847,
      "learning_rate": 4.045847816706361e-08,
      "loss": 0.04120290279388428,
      "step": 7237
    },
    {
      "epoch": 0.9785213350232361,
      "grad_norm": 0.2994944453239441,
      "learning_rate": 3.9957671777268724e-08,
      "loss": 0.07490706443786621,
      "step": 7238
    },
    {
      "epoch": 0.9786565272496831,
      "grad_norm": 0.463821679353714,
      "learning_rate": 3.945998014806163e-08,
      "loss": 0.08352804183959961,
      "step": 7239
    },
    {
      "epoch": 0.9787917194761301,
      "grad_norm": 0.3556688725948334,
      "learning_rate": 3.896540338306609e-08,
      "loss": 0.07006645202636719,
      "step": 7240
    },
    {
      "epoch": 0.9789269117025771,
      "grad_norm": 0.26082634925842285,
      "learning_rate": 3.847394158525641e-08,
      "loss": 0.06986570358276367,
      "step": 7241
    },
    {
      "epoch": 0.979062103929024,
      "grad_norm": 0.2542496621608734,
      "learning_rate": 3.798559485695574e-08,
      "loss": 0.07001972198486328,
      "step": 7242
    },
    {
      "epoch": 0.979197296155471,
      "grad_norm": 0.26027971506118774,
      "learning_rate": 3.7500363299842746e-08,
      "loss": 0.05592823028564453,
      "step": 7243
    },
    {
      "epoch": 0.979332488381918,
      "grad_norm": 0.16457578539848328,
      "learning_rate": 3.701824701494327e-08,
      "loss": 0.046762943267822266,
      "step": 7244
    },
    {
      "epoch": 0.979467680608365,
      "grad_norm": 0.2241346538066864,
      "learning_rate": 3.653924610263703e-08,
      "loss": 0.055150508880615234,
      "step": 7245
    },
    {
      "epoch": 0.979602872834812,
      "grad_norm": 0.2701142132282257,
      "learning_rate": 3.6063360662654255e-08,
      "loss": 0.05116415023803711,
      "step": 7246
    },
    {
      "epoch": 0.9797380650612589,
      "grad_norm": 0.3890156149864197,
      "learning_rate": 3.559059079407734e-08,
      "loss": 0.05099201202392578,
      "step": 7247
    },
    {
      "epoch": 0.9798732572877059,
      "grad_norm": 0.39969396591186523,
      "learning_rate": 3.512093659533922e-08,
      "loss": 0.08316755294799805,
      "step": 7248
    },
    {
      "epoch": 0.9800084495141529,
      "grad_norm": 0.33844202756881714,
      "learning_rate": 3.4654398164225e-08,
      "loss": 0.058484554290771484,
      "step": 7249
    },
    {
      "epoch": 0.9801436417405999,
      "grad_norm": 0.29372870922088623,
      "learning_rate": 3.4190975597870325e-08,
      "loss": 0.07604849338531494,
      "step": 7250
    },
    {
      "epoch": 0.9802788339670468,
      "grad_norm": 0.24319076538085938,
      "learning_rate": 3.373066899276134e-08,
      "loss": 0.05558300018310547,
      "step": 7251
    },
    {
      "epoch": 0.9804140261934938,
      "grad_norm": 0.3392714858055115,
      "learning_rate": 3.3273478444736386e-08,
      "loss": 0.05737876892089844,
      "step": 7252
    },
    {
      "epoch": 0.9805492184199408,
      "grad_norm": 0.2648780643939972,
      "learning_rate": 3.281940404898764e-08,
      "loss": 0.05582237243652344,
      "step": 7253
    },
    {
      "epoch": 0.9806844106463878,
      "grad_norm": 0.31967294216156006,
      "learning_rate": 3.236844590005117e-08,
      "loss": 0.06507253646850586,
      "step": 7254
    },
    {
      "epoch": 0.9808196028728348,
      "grad_norm": 0.7116628289222717,
      "learning_rate": 3.192060409182351e-08,
      "loss": 0.09038543701171875,
      "step": 7255
    },
    {
      "epoch": 0.9809547950992817,
      "grad_norm": 0.28085264563560486,
      "learning_rate": 3.147587871754509e-08,
      "loss": 0.06172943115234375,
      "step": 7256
    },
    {
      "epoch": 0.9810899873257287,
      "grad_norm": 0.4066506326198578,
      "learning_rate": 3.1034269869810174e-08,
      "loss": 0.06013298034667969,
      "step": 7257
    },
    {
      "epoch": 0.9812251795521757,
      "grad_norm": 0.47677507996559143,
      "learning_rate": 3.05957776405652e-08,
      "loss": 0.06224295496940613,
      "step": 7258
    },
    {
      "epoch": 0.9813603717786227,
      "grad_norm": 0.2198784202337265,
      "learning_rate": 3.016040212110549e-08,
      "loss": 0.04728579521179199,
      "step": 7259
    },
    {
      "epoch": 0.9814955640050697,
      "grad_norm": 0.7776672840118408,
      "learning_rate": 2.9728143402078522e-08,
      "loss": 0.09712910652160645,
      "step": 7260
    },
    {
      "epoch": 0.9816307562315166,
      "grad_norm": 0.27797162532806396,
      "learning_rate": 2.9299001573483975e-08,
      "loss": 0.05040121078491211,
      "step": 7261
    },
    {
      "epoch": 0.9817659484579637,
      "grad_norm": 0.29053762555122375,
      "learning_rate": 2.8872976724670375e-08,
      "loss": 0.07041716575622559,
      "step": 7262
    },
    {
      "epoch": 0.9819011406844107,
      "grad_norm": 0.24438482522964478,
      "learning_rate": 2.8450068944338436e-08,
      "loss": 0.04060029983520508,
      "step": 7263
    },
    {
      "epoch": 0.9820363329108577,
      "grad_norm": 0.2871078550815582,
      "learning_rate": 2.803027832054106e-08,
      "loss": 0.06870365142822266,
      "step": 7264
    },
    {
      "epoch": 0.9821715251373047,
      "grad_norm": 0.2679024934768677,
      "learning_rate": 2.7613604940679995e-08,
      "loss": 0.05549955368041992,
      "step": 7265
    },
    {
      "epoch": 0.9823067173637516,
      "grad_norm": 0.32359033823013306,
      "learning_rate": 2.7200048891509176e-08,
      "loss": 0.06971162557601929,
      "step": 7266
    },
    {
      "epoch": 0.9824419095901986,
      "grad_norm": 0.35307279229164124,
      "learning_rate": 2.67896102591314e-08,
      "loss": 0.06990706920623779,
      "step": 7267
    },
    {
      "epoch": 0.9825771018166456,
      "grad_norm": 0.2598518431186676,
      "learning_rate": 2.6382289129004978e-08,
      "loss": 0.060251712799072266,
      "step": 7268
    },
    {
      "epoch": 0.9827122940430926,
      "grad_norm": 0.4024988114833832,
      "learning_rate": 2.5978085585935395e-08,
      "loss": 0.06672167778015137,
      "step": 7269
    },
    {
      "epoch": 0.9828474862695395,
      "grad_norm": 0.19556914269924164,
      "learning_rate": 2.5576999714078676e-08,
      "loss": 0.05283021926879883,
      "step": 7270
    },
    {
      "epoch": 0.9829826784959865,
      "grad_norm": 0.12614038586616516,
      "learning_rate": 2.517903159694468e-08,
      "loss": 0.03987550735473633,
      "step": 7271
    },
    {
      "epoch": 0.9831178707224335,
      "grad_norm": 0.49008631706237793,
      "learning_rate": 2.4784181317390465e-08,
      "loss": 0.07659709453582764,
      "step": 7272
    },
    {
      "epoch": 0.9832530629488805,
      "grad_norm": 0.3741593360900879,
      "learning_rate": 2.4392448957628598e-08,
      "loss": 0.08503150939941406,
      "step": 7273
    },
    {
      "epoch": 0.9833882551753275,
      "grad_norm": 0.2883245050907135,
      "learning_rate": 2.4003834599217177e-08,
      "loss": 0.06562232971191406,
      "step": 7274
    },
    {
      "epoch": 0.9835234474017744,
      "grad_norm": 0.20501579344272614,
      "learning_rate": 2.3618338323071474e-08,
      "loss": 0.03697967529296875,
      "step": 7275
    },
    {
      "epoch": 0.9836586396282214,
      "grad_norm": 0.10614482313394547,
      "learning_rate": 2.3235960209448958e-08,
      "loss": 0.022292613983154297,
      "step": 7276
    },
    {
      "epoch": 0.9837938318546684,
      "grad_norm": 0.2681671977043152,
      "learning_rate": 2.2856700337967606e-08,
      "loss": 0.06901407241821289,
      "step": 7277
    },
    {
      "epoch": 0.9839290240811154,
      "grad_norm": 0.35393616557121277,
      "learning_rate": 2.2480558787587592e-08,
      "loss": 0.05017447471618652,
      "step": 7278
    },
    {
      "epoch": 0.9840642163075624,
      "grad_norm": 0.15665088593959808,
      "learning_rate": 2.2107535636626263e-08,
      "loss": 0.04813027381896973,
      "step": 7279
    },
    {
      "epoch": 0.9841994085340093,
      "grad_norm": 0.39157634973526,
      "learning_rate": 2.1737630962746502e-08,
      "loss": 0.05547666549682617,
      "step": 7280
    },
    {
      "epoch": 0.9843346007604563,
      "grad_norm": 0.2564446032047272,
      "learning_rate": 2.1370844842966696e-08,
      "loss": 0.051375389099121094,
      "step": 7281
    },
    {
      "epoch": 0.9844697929869033,
      "grad_norm": 0.1833972930908203,
      "learning_rate": 2.100717735365243e-08,
      "loss": 0.03956317901611328,
      "step": 7282
    },
    {
      "epoch": 0.9846049852133503,
      "grad_norm": 0.19278490543365479,
      "learning_rate": 2.0646628570521464e-08,
      "loss": 0.05518698692321777,
      "step": 7283
    },
    {
      "epoch": 0.9847401774397972,
      "grad_norm": 0.6498562693595886,
      "learning_rate": 2.028919856864375e-08,
      "loss": 0.08387279510498047,
      "step": 7284
    },
    {
      "epoch": 0.9848753696662442,
      "grad_norm": 0.18656525015830994,
      "learning_rate": 1.9934887422434766e-08,
      "loss": 0.04199361801147461,
      "step": 7285
    },
    {
      "epoch": 0.9850105618926912,
      "grad_norm": 0.3669340908527374,
      "learning_rate": 1.9583695205665496e-08,
      "loss": 0.07400131225585938,
      "step": 7286
    },
    {
      "epoch": 0.9851457541191382,
      "grad_norm": 0.16416636109352112,
      "learning_rate": 1.9235621991457454e-08,
      "loss": 0.03388023376464844,
      "step": 7287
    },
    {
      "epoch": 0.9852809463455852,
      "grad_norm": 0.238312765955925,
      "learning_rate": 1.889066785227933e-08,
      "loss": 0.05614876747131348,
      "step": 7288
    },
    {
      "epoch": 0.9854161385720321,
      "grad_norm": 0.333103746175766,
      "learning_rate": 1.854883285995368e-08,
      "loss": 0.06418037414550781,
      "step": 7289
    },
    {
      "epoch": 0.9855513307984791,
      "grad_norm": 0.4479035437107086,
      "learning_rate": 1.8210117085651902e-08,
      "loss": 0.07337713241577148,
      "step": 7290
    },
    {
      "epoch": 0.9856865230249261,
      "grad_norm": 0.21257217228412628,
      "learning_rate": 1.7874520599894252e-08,
      "loss": 0.04496932029724121,
      "step": 7291
    },
    {
      "epoch": 0.9858217152513731,
      "grad_norm": 0.42596161365509033,
      "learning_rate": 1.7542043472558166e-08,
      "loss": 0.07498455047607422,
      "step": 7292
    },
    {
      "epoch": 0.98595690747782,
      "grad_norm": 0.1502128541469574,
      "learning_rate": 1.7212685772864945e-08,
      "loss": 0.04128670692443848,
      "step": 7293
    },
    {
      "epoch": 0.986092099704267,
      "grad_norm": 0.35827237367630005,
      "learning_rate": 1.68864475693864e-08,
      "loss": 0.056105613708496094,
      "step": 7294
    },
    {
      "epoch": 0.986227291930714,
      "grad_norm": 0.4056648910045624,
      "learning_rate": 1.6563328930051526e-08,
      "loss": 0.08824682235717773,
      "step": 7295
    },
    {
      "epoch": 0.986362484157161,
      "grad_norm": 0.26973649859428406,
      "learning_rate": 1.624332992213151e-08,
      "loss": 0.06273508071899414,
      "step": 7296
    },
    {
      "epoch": 0.986497676383608,
      "grad_norm": 0.38383224606513977,
      "learning_rate": 1.5926450612254728e-08,
      "loss": 0.0930633544921875,
      "step": 7297
    },
    {
      "epoch": 0.986632868610055,
      "grad_norm": 0.27624213695526123,
      "learning_rate": 1.5612691066395068e-08,
      "loss": 0.06635808944702148,
      "step": 7298
    },
    {
      "epoch": 0.9867680608365019,
      "grad_norm": 0.1757369339466095,
      "learning_rate": 1.530205134987861e-08,
      "loss": 0.03658497333526611,
      "step": 7299
    },
    {
      "epoch": 0.9869032530629489,
      "grad_norm": 0.27076634764671326,
      "learning_rate": 1.499453152738528e-08,
      "loss": 0.0631113052368164,
      "step": 7300
    },
    {
      "epoch": 0.9870384452893959,
      "grad_norm": 0.3757174015045166,
      "learning_rate": 1.4690131662938866e-08,
      "loss": 0.0719306468963623,
      "step": 7301
    },
    {
      "epoch": 0.9871736375158429,
      "grad_norm": 0.25306931138038635,
      "learning_rate": 1.438885181991867e-08,
      "loss": 0.06742358207702637,
      "step": 7302
    },
    {
      "epoch": 0.9873088297422898,
      "grad_norm": 0.21392711997032166,
      "learning_rate": 1.4090692061052846e-08,
      "loss": 0.04740428924560547,
      "step": 7303
    },
    {
      "epoch": 0.9874440219687368,
      "grad_norm": 0.4029611051082611,
      "learning_rate": 1.3795652448420071e-08,
      "loss": 0.06776756048202515,
      "step": 7304
    },
    {
      "epoch": 0.9875792141951838,
      "grad_norm": 0.3412213623523712,
      "learning_rate": 1.3503733043447874e-08,
      "loss": 0.04248058795928955,
      "step": 7305
    },
    {
      "epoch": 0.9877144064216308,
      "grad_norm": 0.5584572553634644,
      "learning_rate": 1.3214933906915971e-08,
      "loss": 0.08418989181518555,
      "step": 7306
    },
    {
      "epoch": 0.9878495986480778,
      "grad_norm": 0.20628269016742706,
      "learning_rate": 1.2929255098954596e-08,
      "loss": 0.05633211135864258,
      "step": 7307
    },
    {
      "epoch": 0.9879847908745247,
      "grad_norm": 0.1590343862771988,
      "learning_rate": 1.2646696679042835e-08,
      "loss": 0.040009498596191406,
      "step": 7308
    },
    {
      "epoch": 0.9881199831009717,
      "grad_norm": 0.28296276926994324,
      "learning_rate": 1.2367258706010298e-08,
      "loss": 0.07425785064697266,
      "step": 7309
    },
    {
      "epoch": 0.9882551753274187,
      "grad_norm": 0.27983108162879944,
      "learning_rate": 1.2090941238040443e-08,
      "loss": 0.05311179161071777,
      "step": 7310
    },
    {
      "epoch": 0.9883903675538657,
      "grad_norm": 0.4429514706134796,
      "learning_rate": 1.1817744332660584e-08,
      "loss": 0.08093523979187012,
      "step": 7311
    },
    {
      "epoch": 0.9885255597803126,
      "grad_norm": 0.41722482442855835,
      "learning_rate": 1.1547668046751891e-08,
      "loss": 0.06028485298156738,
      "step": 7312
    },
    {
      "epoch": 0.9886607520067596,
      "grad_norm": 0.5115593671798706,
      "learning_rate": 1.1280712436549379e-08,
      "loss": 0.0912933349609375,
      "step": 7313
    },
    {
      "epoch": 0.9887959442332066,
      "grad_norm": 0.4645446538925171,
      "learning_rate": 1.1016877557630257e-08,
      "loss": 0.056313514709472656,
      "step": 7314
    },
    {
      "epoch": 0.9889311364596536,
      "grad_norm": 0.31187713146209717,
      "learning_rate": 1.0756163464928915e-08,
      "loss": 0.05369257926940918,
      "step": 7315
    },
    {
      "epoch": 0.9890663286861006,
      "grad_norm": 0.1749580204486847,
      "learning_rate": 1.0498570212726932e-08,
      "loss": 0.03262472152709961,
      "step": 7316
    },
    {
      "epoch": 0.9892015209125475,
      "grad_norm": 0.27190452814102173,
      "learning_rate": 1.024409785465641e-08,
      "loss": 0.07194137573242188,
      "step": 7317
    },
    {
      "epoch": 0.9893367131389945,
      "grad_norm": 0.27393752336502075,
      "learning_rate": 9.992746443699962e-09,
      "loss": 0.057901859283447266,
      "step": 7318
    },
    {
      "epoch": 0.9894719053654415,
      "grad_norm": 0.29374632239341736,
      "learning_rate": 9.744516032190731e-09,
      "loss": 0.05480337142944336,
      "step": 7319
    },
    {
      "epoch": 0.9896070975918885,
      "grad_norm": 0.204203262925148,
      "learning_rate": 9.499406671809041e-09,
      "loss": 0.045647382736206055,
      "step": 7320
    },
    {
      "epoch": 0.9897422898183355,
      "grad_norm": 0.3786998987197876,
      "learning_rate": 9.2574184135924e-09,
      "loss": 0.06525158882141113,
      "step": 7321
    },
    {
      "epoch": 0.9898774820447824,
      "grad_norm": 0.28615784645080566,
      "learning_rate": 9.018551307920508e-09,
      "loss": 0.06687963008880615,
      "step": 7322
    },
    {
      "epoch": 0.9900126742712294,
      "grad_norm": 0.33808377385139465,
      "learning_rate": 8.782805404526917e-09,
      "loss": 0.059148550033569336,
      "step": 7323
    },
    {
      "epoch": 0.9901478664976764,
      "grad_norm": 0.2744404077529907,
      "learning_rate": 8.55018075249736e-09,
      "loss": 0.05714988708496094,
      "step": 7324
    },
    {
      "epoch": 0.9902830587241234,
      "grad_norm": 0.18194349110126495,
      "learning_rate": 8.320677400264764e-09,
      "loss": 0.041713714599609375,
      "step": 7325
    },
    {
      "epoch": 0.9904182509505703,
      "grad_norm": 0.29345688223838806,
      "learning_rate": 8.094295395610906e-09,
      "loss": 0.05753040313720703,
      "step": 7326
    },
    {
      "epoch": 0.9905534431770173,
      "grad_norm": 0.1978025585412979,
      "learning_rate": 7.87103478567308e-09,
      "loss": 0.05142700672149658,
      "step": 7327
    },
    {
      "epoch": 0.9906886354034643,
      "grad_norm": 0.33782291412353516,
      "learning_rate": 7.65089561693244e-09,
      "loss": 0.06603479385375977,
      "step": 7328
    },
    {
      "epoch": 0.9908238276299113,
      "grad_norm": 0.31460145115852356,
      "learning_rate": 7.433877935225652e-09,
      "loss": 0.08867168426513672,
      "step": 7329
    },
    {
      "epoch": 0.9909590198563583,
      "grad_norm": 0.5301055908203125,
      "learning_rate": 7.219981785733243e-09,
      "loss": 0.10008573532104492,
      "step": 7330
    },
    {
      "epoch": 0.9910942120828052,
      "grad_norm": 0.27034464478492737,
      "learning_rate": 7.009207212992919e-09,
      "loss": 0.05199456214904785,
      "step": 7331
    },
    {
      "epoch": 0.9912294043092522,
      "grad_norm": 0.3467728793621063,
      "learning_rate": 6.801554260889575e-09,
      "loss": 0.06944608688354492,
      "step": 7332
    },
    {
      "epoch": 0.9913645965356992,
      "grad_norm": 0.23734015226364136,
      "learning_rate": 6.5970229726552976e-09,
      "loss": 0.061892032623291016,
      "step": 7333
    },
    {
      "epoch": 0.9914997887621462,
      "grad_norm": 0.2939838767051697,
      "learning_rate": 6.3956133908743556e-09,
      "loss": 0.06702494621276855,
      "step": 7334
    },
    {
      "epoch": 0.9916349809885932,
      "grad_norm": 0.46972307562828064,
      "learning_rate": 6.197325557483202e-09,
      "loss": 0.08815503120422363,
      "step": 7335
    },
    {
      "epoch": 0.9917701732150401,
      "grad_norm": 0.3913094401359558,
      "learning_rate": 6.002159513765482e-09,
      "loss": 0.07429647445678711,
      "step": 7336
    },
    {
      "epoch": 0.9919053654414871,
      "grad_norm": 0.25788867473602295,
      "learning_rate": 5.810115300355357e-09,
      "loss": 0.05698126554489136,
      "step": 7337
    },
    {
      "epoch": 0.9920405576679341,
      "grad_norm": 0.29297149181365967,
      "learning_rate": 5.621192957239174e-09,
      "loss": 0.06451702117919922,
      "step": 7338
    },
    {
      "epoch": 0.9921757498943811,
      "grad_norm": 0.20773686468601227,
      "learning_rate": 5.435392523748806e-09,
      "loss": 0.06315559148788452,
      "step": 7339
    },
    {
      "epoch": 0.992310942120828,
      "grad_norm": 0.2527959644794464,
      "learning_rate": 5.252714038571638e-09,
      "loss": 0.0677328109741211,
      "step": 7340
    },
    {
      "epoch": 0.992446134347275,
      "grad_norm": 0.24812406301498413,
      "learning_rate": 5.073157539742246e-09,
      "loss": 0.06432628631591797,
      "step": 7341
    },
    {
      "epoch": 0.992581326573722,
      "grad_norm": 0.22527673840522766,
      "learning_rate": 4.896723064642394e-09,
      "loss": 0.05236053466796875,
      "step": 7342
    },
    {
      "epoch": 0.992716518800169,
      "grad_norm": 0.7520393133163452,
      "learning_rate": 4.723410650009363e-09,
      "loss": 0.07049024105072021,
      "step": 7343
    },
    {
      "epoch": 0.992851711026616,
      "grad_norm": 0.2114769071340561,
      "learning_rate": 4.553220331925956e-09,
      "loss": 0.047716617584228516,
      "step": 7344
    },
    {
      "epoch": 0.9929869032530629,
      "grad_norm": 0.21498046815395355,
      "learning_rate": 4.38615214582716e-09,
      "loss": 0.05808830261230469,
      "step": 7345
    },
    {
      "epoch": 0.9931220954795099,
      "grad_norm": 0.42126530408859253,
      "learning_rate": 4.2222061265001496e-09,
      "loss": 0.04510009288787842,
      "step": 7346
    },
    {
      "epoch": 0.9932572877059569,
      "grad_norm": 0.23278068006038666,
      "learning_rate": 4.0613823080742905e-09,
      "loss": 0.04508709907531738,
      "step": 7347
    },
    {
      "epoch": 0.9933924799324039,
      "grad_norm": 0.4372054934501648,
      "learning_rate": 3.903680724037795e-09,
      "loss": 0.06105327606201172,
      "step": 7348
    },
    {
      "epoch": 0.9935276721588508,
      "grad_norm": 0.15601420402526855,
      "learning_rate": 3.749101407224398e-09,
      "loss": 0.03406357765197754,
      "step": 7349
    },
    {
      "epoch": 0.9936628643852978,
      "grad_norm": 0.2650086283683777,
      "learning_rate": 3.597644389818355e-09,
      "loss": 0.0564914345741272,
      "step": 7350
    },
    {
      "epoch": 0.9937980566117448,
      "grad_norm": 0.24166135489940643,
      "learning_rate": 3.4493097033527767e-09,
      "loss": 0.055477142333984375,
      "step": 7351
    },
    {
      "epoch": 0.9939332488381918,
      "grad_norm": 0.3690648078918457,
      "learning_rate": 3.3040973787112904e-09,
      "loss": 0.0753936767578125,
      "step": 7352
    },
    {
      "epoch": 0.9940684410646388,
      "grad_norm": 0.2443806380033493,
      "learning_rate": 3.162007446129711e-09,
      "loss": 0.05772757530212402,
      "step": 7353
    },
    {
      "epoch": 0.9942036332910857,
      "grad_norm": 0.29930561780929565,
      "learning_rate": 3.023039935191041e-09,
      "loss": 0.05570673942565918,
      "step": 7354
    },
    {
      "epoch": 0.9943388255175327,
      "grad_norm": 0.3058774173259735,
      "learning_rate": 2.887194874830468e-09,
      "loss": 0.07515096664428711,
      "step": 7355
    },
    {
      "epoch": 0.9944740177439797,
      "grad_norm": 0.21760514378547668,
      "learning_rate": 2.7544722933287026e-09,
      "loss": 0.051554203033447266,
      "step": 7356
    },
    {
      "epoch": 0.9946092099704267,
      "grad_norm": 0.3237847685813904,
      "learning_rate": 2.6248722183203066e-09,
      "loss": 0.06708765029907227,
      "step": 7357
    },
    {
      "epoch": 0.9947444021968737,
      "grad_norm": 0.24604564905166626,
      "learning_rate": 2.498394676790361e-09,
      "loss": 0.04879283905029297,
      "step": 7358
    },
    {
      "epoch": 0.9948795944233206,
      "grad_norm": 0.43226322531700134,
      "learning_rate": 2.375039695071135e-09,
      "loss": 0.08555889129638672,
      "step": 7359
    },
    {
      "epoch": 0.9950147866497676,
      "grad_norm": 0.3227103352546692,
      "learning_rate": 2.2548072988454184e-09,
      "loss": 0.07320523262023926,
      "step": 7360
    },
    {
      "epoch": 0.9951499788762146,
      "grad_norm": 0.5129017233848572,
      "learning_rate": 2.1376975131465194e-09,
      "loss": 0.09375405311584473,
      "step": 7361
    },
    {
      "epoch": 0.9952851711026616,
      "grad_norm": 0.35747861862182617,
      "learning_rate": 2.023710362356601e-09,
      "loss": 0.06770801544189453,
      "step": 7362
    },
    {
      "epoch": 0.9954203633291085,
      "grad_norm": 0.2408422976732254,
      "learning_rate": 1.9128458702100117e-09,
      "loss": 0.037334144115448,
      "step": 7363
    },
    {
      "epoch": 0.9955555555555555,
      "grad_norm": 0.3053073287010193,
      "learning_rate": 1.8051040597882873e-09,
      "loss": 0.05077546834945679,
      "step": 7364
    },
    {
      "epoch": 0.9956907477820025,
      "grad_norm": 0.23652294278144836,
      "learning_rate": 1.70048495352515e-09,
      "loss": 0.04577064514160156,
      "step": 7365
    },
    {
      "epoch": 0.9958259400084495,
      "grad_norm": 0.3810080885887146,
      "learning_rate": 1.5989885731998443e-09,
      "loss": 0.07359600067138672,
      "step": 7366
    },
    {
      "epoch": 0.9959611322348965,
      "grad_norm": 0.2785953879356384,
      "learning_rate": 1.5006149399487966e-09,
      "loss": 0.05605125427246094,
      "step": 7367
    },
    {
      "epoch": 0.9960963244613434,
      "grad_norm": 0.30559104681015015,
      "learning_rate": 1.4053640742489604e-09,
      "loss": 0.049391746520996094,
      "step": 7368
    },
    {
      "epoch": 0.9962315166877904,
      "grad_norm": 0.21644267439842224,
      "learning_rate": 1.3132359959361351e-09,
      "loss": 0.04492950439453125,
      "step": 7369
    },
    {
      "epoch": 0.9963667089142374,
      "grad_norm": 0.15193523466587067,
      "learning_rate": 1.2242307241899787e-09,
      "loss": 0.0377657413482666,
      "step": 7370
    },
    {
      "epoch": 0.9965019011406844,
      "grad_norm": 0.6023728847503662,
      "learning_rate": 1.1383482775406685e-09,
      "loss": 0.07380247116088867,
      "step": 7371
    },
    {
      "epoch": 0.9966370933671314,
      "grad_norm": 0.2014322578907013,
      "learning_rate": 1.0555886738738973e-09,
      "loss": 0.04391944408416748,
      "step": 7372
    },
    {
      "epoch": 0.9967722855935783,
      "grad_norm": 0.2580380439758301,
      "learning_rate": 9.75951930415886e-10,
      "loss": 0.06332147121429443,
      "step": 7373
    },
    {
      "epoch": 0.9969074778200253,
      "grad_norm": 0.2405771017074585,
      "learning_rate": 8.994380637483701e-10,
      "loss": 0.05062103271484375,
      "step": 7374
    },
    {
      "epoch": 0.9970426700464723,
      "grad_norm": 0.25464409589767456,
      "learning_rate": 8.260470898036054e-10,
      "loss": 0.0672445297241211,
      "step": 7375
    },
    {
      "epoch": 0.9971778622729193,
      "grad_norm": 0.24049629271030426,
      "learning_rate": 7.557790238627016e-10,
      "loss": 0.03889155387878418,
      "step": 7376
    },
    {
      "epoch": 0.9973130544993662,
      "grad_norm": 0.378265380859375,
      "learning_rate": 6.886338805522918e-10,
      "loss": 0.06474447250366211,
      "step": 7377
    },
    {
      "epoch": 0.9974482467258132,
      "grad_norm": 0.35563474893569946,
      "learning_rate": 6.246116738561903e-10,
      "loss": 0.06737220287322998,
      "step": 7378
    },
    {
      "epoch": 0.9975834389522602,
      "grad_norm": 0.40668416023254395,
      "learning_rate": 5.637124171004038e-10,
      "loss": 0.08114171028137207,
      "step": 7379
    },
    {
      "epoch": 0.9977186311787072,
      "grad_norm": 0.2842503786087036,
      "learning_rate": 5.059361229681203e-10,
      "loss": 0.06357669830322266,
      "step": 7380
    },
    {
      "epoch": 0.9978538234051542,
      "grad_norm": 0.30374807119369507,
      "learning_rate": 4.5128280348638583e-10,
      "loss": 0.07045936584472656,
      "step": 7381
    },
    {
      "epoch": 0.9979890156316011,
      "grad_norm": 0.4822167158126831,
      "learning_rate": 3.9975247003443127e-10,
      "loss": 0.08537006378173828,
      "step": 7382
    },
    {
      "epoch": 0.9981242078580481,
      "grad_norm": 0.3208787739276886,
      "learning_rate": 3.51345133342007e-10,
      "loss": 0.05112278461456299,
      "step": 7383
    },
    {
      "epoch": 0.9982594000844951,
      "grad_norm": 0.43507885932922363,
      "learning_rate": 3.060608034877177e-10,
      "loss": 0.05134916305541992,
      "step": 7384
    },
    {
      "epoch": 0.9983945923109421,
      "grad_norm": 0.32505616545677185,
      "learning_rate": 2.638994898990221e-10,
      "loss": 0.07700061798095703,
      "step": 7385
    },
    {
      "epoch": 0.998529784537389,
      "grad_norm": 0.22194541990756989,
      "learning_rate": 2.2486120135556398e-10,
      "loss": 0.05081915855407715,
      "step": 7386
    },
    {
      "epoch": 0.998664976763836,
      "grad_norm": 0.35013654828071594,
      "learning_rate": 1.889459459841758e-10,
      "loss": 0.060456275939941406,
      "step": 7387
    },
    {
      "epoch": 0.998800168990283,
      "grad_norm": 0.26419302821159363,
      "learning_rate": 1.56153731263875e-10,
      "loss": 0.04935884475708008,
      "step": 7388
    },
    {
      "epoch": 0.99893536121673,
      "grad_norm": 0.28081369400024414,
      "learning_rate": 1.2648456402086784e-10,
      "loss": 0.055562734603881836,
      "step": 7389
    },
    {
      "epoch": 0.999070553443177,
      "grad_norm": 0.252122163772583,
      "learning_rate": 9.99384504318801e-11,
      "loss": 0.05839800834655762,
      "step": 7390
    },
    {
      "epoch": 0.999205745669624,
      "grad_norm": 0.4285535514354706,
      "learning_rate": 7.651539602582247e-11,
      "loss": 0.055081844329833984,
      "step": 7391
    },
    {
      "epoch": 0.9993409378960709,
      "grad_norm": 0.3640401363372803,
      "learning_rate": 5.6215405678794464e-11,
      "loss": 0.07813167572021484,
      "step": 7392
    },
    {
      "epoch": 0.9994761301225179,
      "grad_norm": 0.3815842270851135,
      "learning_rate": 3.9038483615749795e-11,
      "loss": 0.0639352798461914,
      "step": 7393
    },
    {
      "epoch": 0.9996113223489649,
      "grad_norm": 0.22268462181091309,
      "learning_rate": 2.4984633415492398e-11,
      "loss": 0.05288577079772949,
      "step": 7394
    },
    {
      "epoch": 0.9997465145754119,
      "grad_norm": 0.2326449155807495,
      "learning_rate": 1.4053858004015041e-11,
      "loss": 0.058682918548583984,
      "step": 7395
    },
    {
      "epoch": 0.9998817068018588,
      "grad_norm": 0.460287481546402,
      "learning_rate": 6.246159654499373e-12,
      "loss": 0.0901026725769043,
      "step": 7396
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.29362422227859497,
      "learning_rate": 1.561539995642569e-12,
      "loss": 0.06835460662841797,
      "step": 7397
    },
    {
      "epoch": 1.0,
      "step": 7397,
      "total_flos": 1.847331328879619e+20,
      "train_loss": 0.0730548164313333,
      "train_runtime": 153923.9428,
      "train_samples_per_second": 12.302,
      "train_steps_per_second": 0.048
    }
  ],
  "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": 1.847331328879619e+20,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}